Stephen A. Marglin - Raising Keynes - A Twenty-First-Century General Theory-Harvard University Press (2021)
Uploaded by
PauloStephen A. Marglin - Raising Keynes - A Twenty-First-Century General Theory-Harvard University Press (2021)
Uploaded by
Paulor aising keynes
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Raising Keynes
A Twenty-First-Century General Theory
Stephen A. Marglin
Copyright © 2021. Harvard University Press. All rights reserved.
cambridge, massachusetts
london, engl and
2021
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021 by Stephen A. Marglin
All rights reserved
Printed in the United States of America
First printing
Cover design by Tim Jones
9780674246201 (EPUB)
9780674246225 (PDF)
Te Library of Congress has cataloged the printed edition as follows:
Names: Marglin, Stephen A., author.
Title: Raising Keynes : a twenty-frst-century general theory / Stephen A. Marglin.
Description: Cambridge, Massachusetts : Harvard University Press, 2021. |
Includes bibliographical references and index. |
Identifers: LCCN 2020019316 | ISBN 9780674971028 (cloth)
Subjects: LCSH: Keynes, John Maynard, 1883–1946. | Keynesian economics. |
Industries—Self-regulation. | Economics, Mathematical.
Classifcation: LCC HB99.7 .M344 2021 | DDC 330.15/6—dc23
LC record available at https://lccn.loc.gov/2020019316
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
For Emmanuelle, Suzanne, Oriah, Jordan,
Micah, Nasia, Mira, Yael, Noah, and Gabriel,
the next generation
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Contents
Notation xi
Prologue: What Is Tis Book About? 1
part i Background: The Rise and Fall
1 Introduction: Is Tis Resurrection Necessary? 17
2 What Were Tey Tinking? Economics Before
Te General Teory 50
part ii Keynes Defeated: Static Models and the Critics
3 Te Determination of Output and Employment: First and
Second Passes at Equilibrium 77
Appendix 1: Keynes’s Defnition(s) of Unemployment 109
Appendix 2: Do Interest Rates Adjust Saving and Investment? 110
Mathematical Appendix 119
4 Equilibrium with a Given Money Supply:
Copyright © 2021. Harvard University Press. All rights reserved.
Critical Perspectives on the Second-Pass Model 122
Mathematical Appendix 154
part iii Keynes Vindicated: A Theory of Real-Time Changes
5 Te Price Mechanism: Gospels According to Marshall
and Walras 159
Mathematical Appendix 187
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
viii Contents
6 Te General Teory without Rigid Prices and Wages 191
Appendix: A Brief History of Stationary Real-Price Equilibria 214
Mathematical Appendix 219
7 Dynamics vs. Statics: Can the Economy Get from the Here of
Unemployment to the Tere of Full Employment? 222
Mathematical Appendix 255
8 A Dose of Reality: Te Evidence of the Great Depression 261
Appendix: Milton Friedman and Anna Schwartz on
What Made the Depression Great 293
Mathematical Appendix 300
part iv Building Blocks
9 Consumption and Saving 307
Mathematical Appendix 338
10 Investment 340
Mathematical Appendix 366
11 Te Teory of Interest, I: Liquidity Preference in a World
of Money and Bonds 376
Appendix: Bond Coupons as Insurance against Price Declines 404
Mathematical Appendix 407
12 Te Teory of Interest, II: Liquidity Preference as a
Teory of Spreads 409
Mathematical Appendix 432
Copyright © 2021. Harvard University Press. All rights reserved.
Empirical Appendix: What Do the Data Say? 457
13 Taking Money Seriously 497
Mathematical Appendix 524
part v Fiscal Policy in Theory and Practice
14 Functional Finance and the Stabilization of Aggregate Demand 529
15 Did the Obama Stimulus Work? 544
Empirical Appendix: Regressions and Teir Discontents 572
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Contents ix
16 Functional Finance and the Composition of
Aggregate Demand 594
Appendix 1: Sound Finance as Starving the Beast 627
Appendix 2: Te Empirics of Debt Sustainability 628
Appendix 3: Are Government Bonds Private Wealth?
And What Diference Does It Make for the Sustainability
of the Debt? 631
Mathematical Appendix 639
part vi Keynes in the Long Run
17 First Steps into the Long Run: Harrod, Domar, Solow,
and Robinson 647
Appendix: Inventory Accumulation as a Brake on Output 673
18 Keynes in the Long Run: A Teory of Wages, Prices,
and Employment 675
Mathematical Appendix 713
19 Infation and Employment Empirics in the Keynesian
Long Run 754
Epilogue: Attack Tem in Teir Citadel 773
notes 797
references 857
acknowledgments 881
index 883
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notation
Chapter 1
P = Nominal price of output
W = Nominal wage rate
P/W = Real price, the ratio of the nominal price of output to the nominal wage
Y = Output = Income (in real terms)
W/P = Real wage, the ratio of the nominal wage to the nominal price of output
L = Labor
Chapter 2
E = Expenditure (real)
= Interest rate
I = Investment
Copyright © 2021. Harvard University Press. All rights reserved.
Chapter 3
I( ) = Investment as a function of the (hurdle) rate of interest = ID, desired invest-
ment
c = MPC = Marginal propensity to consume
C = Consumption (real)
CD = Desired consumption
MEC = Marginal efficiency of capital = rate of interest associated with any given level
of desired investment
M1 = Transactions demand for money
M1 = Supply of money for transactions
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
xii notation
= Ratio of transactions demand for money to nominal value of output
A = Wealth (nominal)
M2 = Asset demand for money
M 2 = Supply of money in market for fnancial assets
= Ratio of asset demand for money to nominal value of wealth
K = Capital stock (real)
B = Stock of bonds in the hands of the public
PB = Price of bonds
R = Bond coupon = Periodic payment due until bond is redeemed
M = Total demand for money
M = Money supply
V = Income velocity of money = 1/ if we ignore the asset demand for money
M/P = Money supply (real)
Y = F(K, L) = Output as function of capital and labor inputs (production function)
FL ≡ ∂Y/∂L = Marginal productivity of labor
L = L(P/W) = Supply of labor in terms of time worked
LS(P/W) = F(K, L(P/W)) = Supply of labor in terms of output produced
GS(P/W) = Supply of goods
Derivatives indicated by ′
Chapter 5
Q = Quantity
D(P) = Quantity demanded as a function of price
S(P) = Quantity demanded as a function of price
X = Rate of change of the variable X with respect to time
Copyright © 2021. Harvard University Press. All rights reserved.
S−1(Q) = Supply price associated with the quantity Q = Marginal cost of Q
D−1(Q) = Demand price associated with the quantity Q
ED = Desired expenditure
i = Speed of adjustment
Chapter 7
WHEAT = Own rate of interest for wheat
OIL = Own rate of interest for oil
W/O = Rate of change of the price of wheat relative to the price of oil
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
notation xiii
= Rate of change of the price level
REAL = Real rate of interest
NOMINAL = Nominal rate of interest
RB = Real interest rate on bonds
RM = Real interest rate on money
NB = Nominal interest rate on bonds
NM = Nominal interest rate on money
c1 = Propensity of wealth holders (creditors) to consume out of income
c2 = Propensity of businessmen (debtors) to consume out of income
s1 = Propensity of wealth holders (creditors) to save out of income
s2 = Propensity of businessmen (debtors) to save out of income
a = Propensity to consume out of wealth
a1 = Propensity of wealth holders (creditors) to consume out of wealth
a2 = Propensity of debtors to consume out of (negative) wealth
i = Conditions on elements of Jacobian matrix required for stability
i = Proportion of income received by creditors (i = 1) and debtors (i = 2)
Chapter 8
Yi = Income of Type i agents, i = 1 (creditors), = 2 (debtors)
YI = Industrial output = Industrial Income
YA = Agricultural output
Y = Agricultural income
A
YAS = Agricultural output for crops consumed domestically and exported
YA0 , YA1 = Part of YAS consumed domestically (in Figure 8.10)
CI = Consumption of industrial output
Copyright © 2021. Harvard University Press. All rights reserved.
ij = Propensity to consume goods produced in sector j from income resulting from
production in sector i, i = A (agriculture), = I (industry)
Ii = Investment demand in sector i
Pi = Nominal price of goods produced in sector i
Chapter 9
YFE = Y(K, LS(P/W)), Output at full employment (labor supply is assumed to be in
dependent of P/W in Figure 9.4)
CD FE = Desired consumption at full employment
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
xiv notation
S = Aggregate saving
= Total profts
r = /K = Rate of p
roft
s = Propensity to save out of p
rofts
sw = Propensity to save out of wages
Chapter 10
q = quasi-rent on investment, the annual fow of returns net of operating costs other
than interest
= Subjective probability of being able to utilize additional capacity
= Functional relationship between investment demand and the annual return on
investment. Te general functional form is I = (q − )
(q − ) = ((Y)[Y/K − (P/W)−1L/K] − ) is the functional form of investment de-
mand for capital widening
(q − ) = ((P/W)−1 MPK/MPL − ) is the functional form of investment demand
for capital deepening
1 = Coef
ficient of K in constant-elasticity-of-substitution production function
2 = Coef
ficient of L in constant-elasticity-of-substitution production function
= 1/(1 − ) = Elasticity of substitution between capital and labor
Chapter 11
Y = Speed of adjustment of output (income) when desired investment and desired
saving are diferent
= Speed of adjustment of the interest rate when desired investment and desired
saving are diferent
DA = Demand for fnancial assets corresponding to the capital stock
Copyright © 2021. Harvard University Press. All rights reserved.
SA = Supply of fnancial assets corresponding to the capital stock
s = Short-term interest rate, assumed to be the rate on 3-month Treasury bills in
empirical analysis
*s = Normal short-term rate of interest
coup = Coupon yield = ratio of periodic payment to bond price
mat = Yield to maturity = rate of discount which makes the stream of bond payments
(including redemption value) equal to the current bond price
hold = Holding yield = sum of coupon yield plus the percentage change in bond price
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
notation xv
Chapter 12
M = Money in asset portfolios
M* = Demand for money or bills in asset portfolios (in the mathematical appendix)
M = Supply of money, endowment to agent
E(P B)/PB = Expected value of the bond-price change (in continuous time) as a per-
centage of the price, abbreviated to E(P)/P in the empirical appendix
E(PB)/PB = Expected value of the bond-price change (in discrete time) as a percent-
age of the price
E(PB,t) = Demand price at time t of bonds on the part of agents characterized by nor-
mal reversion and risk neutrality
E( coup,t) = Coupon yield at time t assuming bond price is equal to the demand price
BCORP = Quantity of corporate bonds demanded
B CORP = Endowment of corporate bonds
PCORP = Price of corporate bond
CORP = Yield on corporate bond
P = Bond price (in the empirical appendix)
mat(m,t) = Yield to maturity of a bond with maturity m at time t
hold(m,t) = Holding yield on a bond of maturity m at time t
mat(∞) = Limiting yield to maturity as bond maturity increases without limit
*s = Normal short-term rate of interest
s = Short-term rate of interest
= Speed of adjustment of short-term rate of interest to normal short-term rate
B = Rate of change of bond price
(m) = Risk premium as a function of the time until the bond matures
m
∫
m −1 0
α(τ)d τ = average risk premium for bonds with maturities ranging from 0 to m
Copyright © 2021. Harvard University Press. All rights reserved.
R*
s = Real normal short-term interest rate
ˆ Rs ( ) = Estimated real short-term interest rate at time
mCORP = Yield to maturity on corporate bond
U1 = Marginal utility of (expected) wealth
U2 = Marginal utility of holding bills (money)
R
mat = Real yield to maturity
X̂ = Estimated value of X
B* = Demand for bonds
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
xvi notation
n = Number of agents, each having the same endowment of bills (M)and the same
endowment of bonds (B)
m(PB, s) = Number of agents desiring to hold only bills, as a function of the price of
bonds and the bill rate
Chapter 13
= Elasticity of transactions demand with respect to s
f = bank fee per conversion of bills to money (and vice versa), assumed to be inde
pendent of transaction size
m = number of conversions between bills and money by agents who hold bills to
earn interest but need money to settle accounts
TC = total cost of holding money including foregone interest on bills
Chapter 14
CA = Additional consumption demand from wealth
YFE = Aggregate demand required for full employment, which is assumed to be a
fxed level of output, in efect assuming a vertical labor-supply schedule
G = Government expenditure on the purchase of goods and services
t = Tax rate, assumed to be a constant fraction of income
Chapter 15
CO = Te fraction of new spending which displaces existing production
m = 1 − CO = Te fraction of new spending that results in new production
v = Te frst-round MPC, the fraction of income spent by the direct benefciaries of
government transfers or tax cuts
g = Te rate of growth of output and income
Copyright © 2021. Harvard University Press. All rights reserved.
= Te propensity to consume out of permanent income
YP = Permanent income
= Rate of growth of permanent income
= Speed of adjustment of permanent income to actual income
E = Transfer payments by states, chiefy Medicaid
R = Non-ARRA (American Recovery and Reinvestment Act) state revenues
A = Transfers from federal government to states under ARRA
L = R + A − (G + E) = Saving (negative sign means dissaving) by states
O = G + E = State outlays
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
notation xvii
N = Number of observations
z = Ratio of estimated regression coef
ficient to standard error of the estimate
Chapter 16
X* = Optimal value of X = C, I, G
= Y/YFE = Ratio of output to full-employment output
t0 = Upper limit on tax rate
t1 = Tax rate consistent with the specifed level of consumption and full employment
h = Hurdle rate for investment, assumed fxed by central bank
D = Government debt in nominal terms
P = Price level
d = Ratio of the primary defcit to GDP (the primary defcit leaves out interest on the
debt)
= Rate of interest on government debt (conceptually distinct from the hurdle rate
h; in practice an average of rates on debts of diferent maturities)
g = Rate of growth of nominal GDP
= D/PY = Ratio of government debt to nominal GDP
Chapter 17
g = Rate of growth of real output
v = K/Y = Incremental capital:output ratio
s = Propensity to save, assumed to be constant
gw = s/v = (for Harrod) Te warranted rate of growth of output, the value of g for
which ID = SD
gn = N/N = Rate of growth of population and labor force
Copyright © 2021. Harvard University Press. All rights reserved.
gw = ID/ID = (for Domar) Te change in desired investment as a percentage of de-
sired investment
s = SD/Y = (for Domar) Ratio of change in desired saving to change in income
v = K/Y = (for Domar, as for Harrod) Te incremental capital:output ratio, but it
is determined diferently. For Domar it is the multiplier process that fxes the
value of Y
Y = F(K, L) = Output as a function of capital and labor inputs, assumed to exhibit
constant returns to scale
k = K/L = Ratio of capital to labor input
f(k) ≡ F(K/L, 1) = Output per unit of labor as a function of the capital:labor ratio
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
xviii notation
f(l) ≡ F(1, L/K) = Output per unit of capital as a function of the labor:capital ratio
ID = ( h) = Investment demand per unit of capital as a function of the hurdle rate of
interest (in the long run, with l as the state variable rather than Y)
SD = Desired saving per unit of capital (in the long run, with l as the state variable
rather than Y)
LS/N = (P/W) = Ratio of labor supply to population as a function of the real price
g wi = (for Robinson) Warranted rate of growth of capital stock, i = 0, 1
rwi = (for Robinson) Rate of p
roft at the warranted rate of growth, rwi corresponding
i
to g = g w , i = 0, 1
Chapter 18
ID = Investment demand per unit of capital (no subscript in mathematical appendix)
SD = Desired saving per unit of capital (no subscript in mathematical appendix)
E = energy
Y = F(K, L, E) = Production function with three arguments, capital, labor, and energy
PE = Nominal price of energy
= PE/P = Real price of energy
l = L/K = Labor per unit of capital
e = E/K = Energy per unit of capital
Y/K = F(1, l, e) = f(l, e) = Output per unit of capital as a function of l and e
h(l, ) = max f (l , e) − e = f ( l , e(l , ) ) − e(l , ), where e(l, ) is the amount of
e
energy per unit of capital which, for given l and , maximizes the expression
f(l, e) − e
hi = ∂h/∂i, i = l, e,
hij = ∂2y/∂j∂i, i, j = l, e,
ei = ∂e(l, )/∂i, i = l,
r = /K = f (l, e) −(P/W)−1 l − e = h(l, ) − (P/W)−1 l = Rate of p
Copyright © 2021. Harvard University Press. All rights reserved.
roft
MPL = hl = Marginal product of labor
MPK = h − hll = Marginal product of capital
Xi = ∂X/∂i, X = I, S; i = P/W, l,
sw = Propensity to save of workers
K = Capital stock owned by capitalists
= K/K = Te fraction of the capital stock owned by capitalists
L = Employment
= L/LS = Employment as a fraction of the labor supply
n = Te sum of the population growth rate and a normal rate of immigration into the
capitalist sector of the economy at full employment
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
r aising keynes
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
prologue
What Is This Book About?
Te orthodox equilibrium theory of economics has assumed, or has
at least not denied, that there are natural forces tending to bring the
volume of the community’s output, and hence its real income, back to
the optimum level whenever temporary forces have led it to depart
from this level. But . . . the equilibrium level towards which output
tends to return afer temporary disturbances is not necessarily the
optimum level, but depends on the strength of the forces in the
community which tend towards saving.
—john maynard keynes
Te most shocking view in Te General Teory was the allegation that
economic equilibrium need not produce full employment. Econo-
mists like Schumpeter found this to be simply incredible. Smith’s
Invisible Hand was brought under direct attack. Tis was revolution,
not evolution.
—paul samuelson
Tis book presents a macroeconomic theory that builds on the central vision
of Keynes’s General Teory of Employment, Interest and Money, namely, that a
capitalist economy is not self-regulating; there is no endogenous mechanism,
Copyright © 2021. Harvard University Press. All rights reserved.
no invisible hand, that guarantees full employment even in the most favorable
case of a competitive economy in which there are neither frictions nor imper-
fections. In Keynes’s vision, aggregate demand is a determinant of how the
level of output and employment adjusts to a variety of shocks. In my render-
ing of Keynes, the adjustment mechanism not only determines the trajec-
tory of the economy outside of equilibrium it also determines the equilibrium
itself.
And none of the putative guarantors of full employment operate in real
time as they do in a static comparison of equilibrium positions—not a reduc-
tion of the cost of production via a decrease in the money wage; not a reduc-
tion of the price level and a consequent decrease in the interest rate and an
1
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 Raising Keynes
increase in investment, via a decrease in the transactions demand for money;
not an increase in the real purchasing power of money and an associated in-
crease in consumption demand consequent upon a fall in the price level. In
real time these mechanisms are more likely to contribute to unemployment
than to cure it.
Why is it important to elaborate a theory based on the centrality of the ad-
justment process in the context of a competitive economy? Why not continue
to rely on mainstream theories based on imperfections and frictions, espe-
cially since we know that frictions and imperfections are endemic?
Te frst reason is that the focus on imperfections and frictions leads natu-
rally to a focus on policies that will remove these imperfections and frictions,
policies that will make the economy more closely resemble the textbook
model of perfect competition—and a rejection of policies that adjust aggre-
gate demand to the full-employment potential of the economy. Tis is not
hypothetical: the push toward deregulation at the end of the twentieth cen-
tury, particularly fnancial deregulation, contributed to the fnancial crisis of
2008 and the ensuing Great Recession. And deregulation was promoted on
the basis of the idea that regulation above and beyond the self-regulation of
the market gets in the way of good economic performance. Te hostility to
aggregate-demand management stems in large part from a failure to under-
stand the argument why aggregate demand matters.
Te second reason for assuming an idealized economy is that a focus
on imperfections and frictions limits the applicability of aggregate demand to
the short run; New Keynesians, like Monetarists, New Classicals, and Real
Business-Cycle theorists, regard the long run as a period in which time dis-
solves whatever warts might mar capitalism in the short run, particularly
price and wage rigidities. By situating the argument in a perfectly competitive
economy, it becomes possible to address the question of whether there is a
long-run trade-of between price stability and growth built into the structure
of a market economy. (Te answer is yes, there is.)
Copyright © 2021. Harvard University Press. All rights reserved.
Making the argument that unemployment can exist even under conditions
of perfect competition requires us to rescue Keynes’s insight about aggre-
gate demand from his friends as well as from his foes, for the friends (the
Keynesians of Axel Leijonhufvud’s Keynesian Economics and the Economics of
Keynes) bear much of the responsibility for the distortion of Keynes’s mes-
sage. Friendly criticism of Te General Teory, following the lead of Franco
Modigliani (1944), attributed unemployment to nominal-wage rigidity, argu-
ing that wage fexibility would make the rate of interest and investment de-
mand a mechanism for adjusting aggregate demand to the needs of full em-
ployment. Less friendly criticism from the likes of Gottfried Haberler (1939,
1941) and Arthur Pigou (1943) made the “real-balance” efect, the adjustment
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Prologue 3
of the purchasing power of money and consumption demand, the mechanism
for doing the heavy lifing for restoring full employment afer a shock to de-
mand. For Milton Friedman (1970) the real-balance efect was the killer argu-
ment against Te General Teory.
Without some exegesis of both Te General Teory and the dispute that fol-
lowed about the nature of Keynes’s message, it is impossible to understand the
real lesson of the dispute, namely, the limitations of comparative statics for
describing and assessing how a capitalist economy works. Keynes and his crit-
ics were talking past each other, in the sense that the critics relied on static
models in which “change” involves a comparison of equilibrium positions
that exist outside of time, while Keynes’s idea of change is a movement in real
time. Te critics were right in the context of comparative statics; Keynes was
right in the context of real-time changes, and right about the consequent need
for dynamic models.
Tis is not an exercise in the history of thought. I have neither the training
nor the inclination for such an exercise. Nor do I attempt to provide insight as
to what Keynes “really” meant, an enterprise I regard as a search for the holy
grail. I can hope that this is the book that Keynes would have written if he had
had eighty years to think about his project, to bone up on mathematics un-
available to him in the 1930s, to digest the criticisms that were ofered on all
sides, and, not least, to observe the changes in the economy that have taken
place since Te General Teory was published. But I intend the book to stand
on its own, whatever Keynes might have thought of it.
While sharing Keynes’s vision of how capitalism works, this book recog-
nizes and attempts to fll major gaps in his argument, gaps that ultimately led
to the eclipse, indeed the demise, of Te General Teory. Tis requires fve
important innovations:
1. Tis book provides a dynamic model that concretizes Keynes’s intu-
ition about how capitalism operates lef to itself: even with fexible
Copyright © 2021. Harvard University Press. All rights reserved.
wages and prices, a capitalist economy normally does not move to a
full-employment equilibrium.
2. I extend the Keynesian vision to the long period, in which neither the
capital stock nor the labor force is fxed. If there is a long-run ten-
dency to full employment, it is because in the long run the labor force
is endogenous. Te central role of aggregate demand is not limited to
the short period, as most present-day Keynesians believe. In the long
run, as in the short run, there is a trade-of between price stability and
employment. Farewell to the vertical Phillips curve.
3. Rightly understood, Keynes’s theory of liquidity preference provides a
theory of spreads between various interest rates, while leaving the level
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 Raising Keynes
of interest rates indeterminate. Building on Keynes, I provide a coher-
ent theory of interest in which it is only when an institution like a cen-
tral bank anchors the spread that the interest rate becomes determi-
nate.
4. I relate the theory of employment and interest to a theory of money. A
theory of money is necessary for the theory of interest and employ-
ment, which makes its absence from Te General Teory all the more
surprising—even though Keynes’s complete title is Te General Teory
of Employment, Interest and Money.
5. Generalizing Abba Lerner’s theory of functional fnance, this book
provides a framework for fscal policy that takes account not only of
the choice facing governments with respect to the level of employment
but also the choices about the composition of aggregate demand, how
much private consumption and investment, how much infrastructure,
education, health care, and the like, goods which the government typi-
cally has a preponderant role in providing. It is this second set of is-
sues that raises the issue of the public debt, but the real issue is very
diferent from the noise that has substituted for reasoned political
debate.
Because so much of this book is devoted to the elaboration and criticism of
models, I should perhaps say something up front about what models are good
for—and their limits. For me (and, I suspect, for Keynes, but no matter) mod-
els are ways of organizing one’s thinking, not machines for prediction. At a
practical level, Keynes’s conception of capitalism provided a way of thinking
about the economy in which the greatness of the Great Depression was not
the surprise that it was for the mainstream. An updated version of Keynes’s
vision will clarify the role of aggregate demand in bringing about and pro-
longing the global recession that came in the wake of the fnancial crisis of
2008—and for thinking about the policies appropriate for the next recession,
Copyright © 2021. Harvard University Press. All rights reserved.
not to mention policies for demand management for growth.
Indeed, Keynes provided not only diagnosis but medicine, a framework for
organizing appropriate policy responses to economic shocks. Keynes was
more successful in this than he was in providing a new way of thinking about
how the economy works because the policy message survived the distortions
of Te General Teory perpetrated by friends as well as enemies. For a variety
of reasons, the policy message could be, and was, acted upon more efectively
in the United States than in Europe afer the fall of Lehman Brothers in Sep-
tember 2008. To be sure, the United States fscal response would have been
more successful if Keynes’s way of thinking about the economy had prevailed
so that there was a general receptivity to activist fscal policy. And the Euro-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Prologue 5
pean monetary response would have been more efective if Keynes rather
than Friedman had been the guiding light of the European Central Bank.
Te problem is not that macroeconomists failed to predict the fnancial
crisis. I don’t believe that any model inspired by Keynes’s vision, certainly not
my models, would have done any better. Te failure of macroeconomics is
that it provided no plausible way of understanding the recession that fol-
lowed—not for nothing called the Great Recession. Here the contrast with a
macroeconomics building on Keynes is dramatic.
I do not deny the technical strides in mainstream macroeconomics since
Keynes’s day. But the gain in technical sophistication has been achieved at a
steep price. Te currently dominant approach to macroeconomic modeling,
so-called dynamic, stochastic, general-equilibrium (DSGE), obscures key in-
sights of Te General Teory, in particular that aggregate demand matters
even in the absence of frictions and rigidities. To make progress in macroeco-
nomics requires us to take a step backward in order to undo the damage done
by an “equilibrium discipline” (Lucas 1977) that identifes equilibrium with
market clearing, a key feature of the DSGE approach in all its avatars.
The Plan of the Book
Tis book is not an easy read. It necessarily contains sustained and detailed
theoretical argument, and parts of the argument are accessible only with some
mathematics and econometrics. But these more technical parts are set apart
from the main text in appendices and serve to supplement a self-contained
narrative that is accessible to readers with a minimal background in econom-
ics and even less background in math.
Part I, “Background: Te Rise and Fall,” provides an overview of the history
of the central message of Te General Teory, that aggregate demand is a sepa-
rate force infuencing output and employment. It also sketches the views that
Keynes opposes, and why these views had no place for aggregate demand in
Copyright © 2021. Harvard University Press. All rights reserved.
their picture of the economy.
Chapter 1 lays out the strategy and structure of Keynes’s argument and pro-
vides an overview of the theoretical criticisms that emerged in the decade af-
ter publication. In addition to the professional critique of Keynes, there was a
political reaction, a reading of Keynes as the useful idiot of a communist
conspiracy that was supposed to be undermining capitalism from within at
the same time as the masters of this conspiracy in the Kremlin were extending
the borders of the Soviet empire in Europe and Asia. Tese criticisms had lit-
tle to do with the theoretical opposition to Keynes, but political opposition
shaped how Keynes’s message was received by the public. Finally, events—
frst proft squeeze in the late 1960s, then the oil shocks of the 1970s—con-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 Raising Keynes
spired with theoretical critique and political opposition to derail Keynes alto-
gether. Tere was by this time a lot of accumulated bath water that deserved
to be thrown out the window; alas, the baby was also thrown out.
Chapter 2 examines the mainstream economics that formed the context for
Te General Teory. Te focus is on why the economics profession had no
need for aggregate demand, thinking that wage and price fexibility were not
only necessary but also suf ficient to guarantee full employment, and how this
view has changed—and how it has not changed—over time.
Part II, “Keynes Defeated: Static Models and the Critics,” lays out two static
models that are more or less explicit in Te General Teory; chapter 3 elabo-
rates a “frst-pass” model based on a given interest rate and a “second-pass”
model based on a given money supply. Te frst shows how the interest rate
connects fnancial markets to the real economy of output and employment,
and it provides a vehicle for arguing the scope and limits of monetary policy.
Te second-pass model is intended to show the limits of the invisible hand in
guiding the economy to full employment; it replaces a given interest rate with
a given money supply and allows the interest rate to be determined by eco-
nomic forces. In terms of the logical progression of the argument for what ails
capitalism and how to fx it, the second model should precede the frst, but
the frst-pass model makes it easier to understand how the parts of Keynes’s
theory intertwine and provides a basis for understanding the policy implica-
tions of Te General Teory.
Chapter 4 presents the arguments of the early critics who honed in on an
assumption Keynes considered mere scafolding, to be torn down afer the
edifce of Te General Teory was built: the assumption of rigid money wages.
Franco Modigliani argued in terms of the responsiveness of the interest rate
and investment to the level of wages (through the impact of money wages on
the price level), and Gottfried Haberler and Arthur Pigou argued in terms of
the impact of prices on consumption via the real-balance efect. Analyzing
diferent parts of Keynes’s theoretical apparatus, Modigliani and Haberler
Copyright © 2021. Harvard University Press. All rights reserved.
(and Pigou) reached mutually reinforcing conclusions, namely, that even on
Keynes’s own terms, his pessimistic conclusion about the possibility of cap
italism to sustain full employment relied critically on the assumption of wage
rigidity. Keynes, no doubt kicking and screaming in his grave, was reduced to
a sophisticated theorist of sand in the wheels, in this sense no diferent from
the myriad economists who had argued that the problems of capitalism are
rigidities, frictions, and imperfections, and can be cured by remaking the
economy in the image of textbook accounts of perfect competition.
All this is prologue, but essential prologue, for it establishes the need for
addressing real-time changes in a more appropriate framework than the
comparative-statics analysis in which both the models of Te General Teory
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Prologue 7
and the models of the critics were cast. Part III, “Keynes Vindicated: A Teory
of Real-Time Changes,” sets out a “third-pass” model that starts from the ad-
justments economic actors make outside of equilibrium. Tis model deter-
mines equilibrium from these disequilibrium actions. Te novelty of my ap-
proach is its reliance on the signals—plural—that producers receive outside of
equilibrium and how these signals are processed. In contrast with the single
signal, namely, price, that guides production at a market-clearing equilib-
rium, away from equilibrium producers receive two signals, a price signal that
contains information about the proftability of production at the margin and a
quantity signal that contains information about how fast goods are moving of
(or piling up on) their shelves. And, unlike the situation of competitive pro-
ducers at a market-clearing equilibrium, producers have two decisions to
make, one on how much to change the price they charge, the other on how
much to change the quantity they produce. How the signals are processed
determines the trajectory of the economy out of equilibrium—not only the
disequilibrium trajectory: the trajectory determines the equilibrium.
An important characteristic of the equilibrium is that while, absent techno-
logical progress, the real price, P/W (the inverse of the real wage), does not
change over time, both the numerator and the denominator are in fux. In a
depression setting, money wages and money prices are both falling. In the
comparative-statics context of chapter 4, lower wages and prices are key both
to Modigliani’s argument that investment demand would increase suf ficiently
to get the economy to full employment (barring a liquidity trap or completely
inelastic investment demand) and to Haberler’s and Pigou’s argument that
consumption demand would rise to the occasion. In real time, these results
no longer hold. Modigliani’s full-employment (and market-clearing) equilib-
rium may not be stable. And fractional-reserve banking may make Modigli-
ani’s argument moot because falling prices may have no impact on investment
demand. And there is an additional argument, namely, that a falling price
level in real time may actually drive up the relevant interest rates (the Fisher
Copyright © 2021. Harvard University Press. All rights reserved.
efect, afer Irving Fisher [1896]). Te Haberler–Pigou road to wealth and
prosperity fares just as badly: this road may be so littered with the corpses of
defaulting debtors that it becomes impassable.
A fair verdict has to reverse the judgment of the comparative-statics analy-
sis of chapter 4. Far from undermining Keynes’s argument about the limits of
the invisible hand, not to mention the limits of the visible hand of monetary
policy, once we shif to the real-time framework of chapter 7, the consider-
ations raised by Modigliani, as well as the considerations raised by Haberler
and Pigou, actually support the arguments of Te General Teory.
Te fnal chapter of this part of the book, chapter 8, examines data from the
Great Depression. Te idea is to test the theory elaborated in the previous
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 Raising Keynes
three chapters against the view championed by Friedman that a relatively
mild downturn morphed into the Great Depression because of bad policy
decisions by the Federal Reserve. It will probably not surprise anybody that I
come down on the side of Te General Teory.
Te experience of agriculture highlights Fisher’s (1933) debt-defation the-
ory of depression as a complement to Te General Teory. In terms of output
and employment, U.S. agriculture suffered no depression at all: 1932 was as
good as 1929. Te considerable pain suffered by farmers was due rather to a
fall in the prices they received relative to the prices they paid and to the huge
burden of real debt (the fip side of the real-balance efect) and the corre-
sponding burden of interest payments (the Fisher efect) caused by the falling
price level.
Part IV, “Building Blocks,” examines the various elements of Te General
Teory in the light of critiques that have been developed over the years. Sepa-
rate chapters are devoted to the theory of consumption and the theory of in-
vestment, and two chapters are devoted to the most novel and controversial
element, the liquidity-preference theory of interest.
Chapter 9 examines the claim that rational actors will determine today’s
consumption and saving on the basis of their long-term resources. Tis un-
derlying assumption of DSGE models builds on Modigliani’s life-cycle hy-
pothesis and Friedman’s permanent-income hypothesis. One implication of
these theories is that the marginal propensity to consume is low—zero in the
limit—when income changes are viewed as transitory. Tis result is important
for policy purposes since, were it true, it would eviscerate countercyclical fs-
cal policy intended to compensate for changes in private demand.
My view is that Modigliani and Friedman at best explain the behavior of a
small fraction of the population, upper middle-class professionals who fulfll
the psychological and cognitive prerequisites of Modigliani–Friedman opti-
mization, not to mention that this group has the fnancial cushion that makes
it feasible to optimize today’s consumption in terms of the agent’s long-term
Copyright © 2021. Harvard University Press. All rights reserved.
resources rather than in terms of her current income. For the rest of the pop-
ulation—both the 1 percent, the rich and superrich, and the vast majority of
working folk—other paradigms are more relevant.
We are actually in better theoretical shape in talking about how the 80 to 90
percent or so of the population whose lives are so insecure that rational con-
sumer choice is not rational: lagged adjustment to changes in income was
posited by both Keynes and Paul Samuelson (1943) as the explanation of why
the short-term marginal propensity to consume is likely to be smaller than
the long-term propensity. As for the 1 percent, we know very little about
the determinants of their consumption and saving decisions—except that
they are even less likely to operate according to the dictates of Modigliani–
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Prologue 9
Friedman optimization than are people who have to work for a living. Tis
chapter concludes with an analysis of the implications of taking class seriously
as a determinant of consumption and saving, as in the so-called Cambridge
saving theory, a revival of the classical view that the source of saving is the
surplus that remains afer wages and other costs are paid.
Chapter 10 deals with investment demand. As for consumption and saving,
Te General Teory ofered a highly simplifed model of investment, one that
abstracts from all its determinants except the rate of interest. Tis chapter at-
tempts to fll in some of the more important gaps, beginning with the rela-
tionship of the interest elasticity of investment demand to the durability of
projects. Tis relationship provides an important clue to why monetary policy
has been focused on residential construction, rather, than, say, on business
investment to upgrade the laptops of their employees.
A second topic of this chapter is the evolution of Keynes’s investment
model, taken over without much modifcation from Irving Fisher (1930),
to subsequent investment theories, specifcally, to so-called Q-theory devel-
oped by James Tobin (William Brainard and Tobin 1968; Tobin 1969). In the
Fisher–Keynes approach, the focus is on project evaluation, on whether or
not future revenues of particular investments justify the requisite capital out-
lay. Q-theory shifs the focus to make versus buy, to a decision whether to di-
rect capital outlays to the purchase of existing capital or to building anew. As
long as the focus is on individual projects, this shif in focus can illuminate
the dynamics of investment demand, at least in limited areas where make and
buy are real options; residential housing once again c omes to mind.
Te scope for extending Q-theory beyond individual projects to the analy-
sis of strategic decisions is much more limited. Tobin (and Keynes before
him) sought to relate investment demand to stock market valuations, but this
relies on make versus buy being an option with regard to creating new capital
or buying an existing company. I have read that Microsof did just this, having
decided to compete with Google in the search-engine business. But this ap-
Copyright © 2021. Harvard University Press. All rights reserved.
pears to be the exception to the rule: in carrying out strategic visions, new
investment and the purchase of existing companies are not generally alterna-
tives, so the level of the Dow Jones Industrial Average wouldn’t seem to be
very relevant to investment decisions.
Te major innovation of this chapter is the analysis of the impact of the real
price P/W on investment demand. Te real price has opposite efects on in-
vestment demand depending on whether the proposed investment is “capital
widening,” that is, intended to expand capacity, or “capital deepening,” in-
tended to substitute capital for labor (or other inputs). A higher real price has
a positive efect on capital-widening investment, but a negative efect when
investment is capital deepening. Tis diference is important because it plays
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 Raising Keynes
a role in answering the question of whether or not high (real) wages are good
for employment and output, a question that the emphasis on aggregate de-
mand brings to the fore even if Keynes himself paid it little attention. Tis
question is a principal concern of the long-run model developed in chapter
18. Whether or not a high real price increases investment depends on where
the economy is in the business cycle, since capital deepening will form a
larger part of total investment demand when there is considerable slack than
when the economy is bursting at the seams.
Te fnal topic in chapter 10 is the psychological aspect of investment de-
mand, why there is an irreducible element of what Keynes called “animal
spirits” in investment decisions. And why this matters.
Chapters 11 and 12 critically assess Keynes’s liquidity-preference theory of
interest. Liquidity preference is an alternative to the mainstream idea that in-
terest rates are determined by thrif and productivity, as refected in desired
saving and desired investment. Te mainstream relies on a rapid adjustment
of price (the interest rate) to bring desired investment and saving into line.
Keynes argues that because the interest rate is kept busy adjusting asset de-
mands and supplies, it will adjust with glacial slowness to diferences in de-
sired saving and investment; it is lef to output to do the heavy lifing of ad-
justing desired saving and investment.
To regard the investment-saving nexus as the site of interest-rate determi-
nation is simply a category mistake. Alas, this mistake is not just a chapter in
the history of thought. No less an economist than Larry Summers repeats the
error:
Just as the price of wheat adjusts to balance the supply of and demand for
wheat, it is natural to suppose that interest rates—the price of money—ad-
just to balance the supply of savings and the demand for investment in an
economy. (2016, p. 3)
Chapter 11 translates Keynes’s somewhat confused statement of liquidity
Copyright © 2021. Harvard University Press. All rights reserved.
preference into a coherent model. In this model, following Keynes, the only
assets are cash and bonds. Te bond yield is assumed to provide a reference
rate of interest (what I call the “hurdle rate”) for private investment, which is
in efect to assume the bonds are comparable in terms of default risk and du-
ration to the class of investment projects under consideration. Chapter 11
shows that the argument of Te General Teory, even if confused, is logically
consistent on its own premises. In this argument the hurdle rate is deter-
mined as the equilibrium rate in the market for bonds.
Te problem with this theory is not its logic but its premises. In particular,
chapter 12 focuses on Keynes’s assumption that the alternative to placing
wealth in bonds is to hold wealth as cash. Once we no longer assume that the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Prologue 11
alternative to bonds is cash, the theory outlined in chapter 11 has to be modi
fed: the hurdle rate, as the equilibrium in the bond market, will depend on
the rate of interest on the asset (say, 3-month T-bills) that is the short-term
riskless alternative to bonds.
How is the short-term rate determined? Not in the market for bonds and
bills: there are not enough degrees of freedom. In principle the dif ficulty is the
standard dif ficulty of general-equilibrium models: n goods (here n = 2) deter-
mine n − 1 relative prices. One price must be assumed, with the result that
only the spread between the two interest rates (the T-bill rate and the hurdle
rate) is determined by the market.
In a pure market economy, this leaves the hurdle rate indeterminate. In the
world in which we live, central banks resolve the indeterminacy. By fxing the
short-term rate, a central bank anchors the spread.
Te empirical appendix to chapter 12 reconciles the argument of this chap-
ter with the empirical literature that purports to fnd no confrmation of
liquidity-preference theory in actual interest-rate data. My contention is that
the vast amount of research on the so-called “expectations hypothesis” has
failed to come to grips with Keynes. Te data, I argue, actually confrm
Keynes’s view of the sources of liquidity preference and in this respect validate
the theory.
Te fnal chapter in this part of the book, chapter 13, reprises the impor-
tance of what is assumed about the nature of money. While Keynes under-
stood that, whatever its history, money was no longer tied to a commodity—
gold, silver, or cowrie shells—his theory of liquidity preference was based on
implicit acquiescence to a commodity theory of money. Keynes intended li-
quidity preference to be an alternative to the mainstream theory of interest as
the outcome of the interplay of desired investment and desired saving, and
accordingly the theory has to provide the level of the hurdle rate along with
the spread between the hurdle rate and the short-term rate. But without a
central bank to anchor the short-term rate of interest, the only way that this
Copyright © 2021. Harvard University Press. All rights reserved.
can happen is if the liquid asset in wealth portfolios is literally cash. Because
cash carries a zero rate of interest, the spread is necessarily equal to the hurdle
rate. Tis story makes sense if we are in a world of commodity money, but not
otherwise.
If we assume that money is not a commodity, but a creation of the banking
system, then the absence of a monetary authority to anchor the spread leaves
the level of the hurdle rate up in the air. Indeed, capitalism lef to its own de-
vices lacks a theory of the interest rate, which has as its counterpart the inde-
terminacy of equilibrium in the realm of employment and output. In other
words: we can calculate the hurdle rate appropriate for any level of employ-
ment, but there is no reason to think the market will oblige by selecting the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 Raising Keynes
corresponding short-term rate. With no anchor to the spread, there is a con-
tinuum of asset-market equilibria and no way to ensure that asset markets
will settle on the particular combination of long and short rates consistent
with full employment and asset-market equilibrium.
Te next section of this book, Part V, “Fiscal Policy in Teory and Practice,”
examines the main policy lesson of Te General Teory. Chapter 14 lays out
the basis for countercyclical fscal policy as a means of maintaining aggregate
demand at full-employment levels in the face of fuctuations in private de-
mand. Tis chapter explores how tax and expenditure policies combine with
the overall size of the government budget to provide stabilization of demand,
in particular, how this is largely accomplished without legislative or executive
action, operating independently of which party controls Congress or the
White House. Rather “automatic stabilization” is built into the government
budget through diferences in the way outlays and taxes respond to fuctua
tions in output and income. Tis chapter also explores the limits to automatic
stabilization, which in turn explains the need for discretionary fscal policy in
the case of severe downturns, whether the purpose of augmenting aggregate
demand is acknowledged, as in the case of the Obama stimulus, or unac-
knowledged, as in the case of the Reagan tax cuts.
Te following chapter explores the Obama stimulus in some detail, as a
case study in discretionary fscal policy. Chapter 15 provides a bridge between
the theory of the multiplier and the multiplier in practice and a detailed
analysis of the most controversial part of the Obama stimulus, the roughly
$250 billion, some one-third of the total, that went to the states. Tough sev-
eral researchers have found that these grants increased state spending signif
cantly, an infuential study by John Cogan and John Taylor (2012) argued that
the states pocketed the money, using it to shore up their balance sheets rather
than for additional spending. Tis chapter supports the view that the stimulus
did in fact stimulate state spending. Not only is the Cogan–Taylor analysis
faulty on its own terms, but complementing econometrics by interviews with
Copyright © 2021. Harvard University Press. All rights reserved.
state officials provides strong evidence that grants to states were spent rather
than hoarded.
Te fnal chapter of this part, chapter 16, extends the analysis of fscal
and monetary policy to the question of the size of government. How big the
government should be depends on how much society values the goods and
services government provides—infrastructure, military power, education—
relative to the goods provided privately—consumption and investment in
factories, machinery, residential housing. Tis choice has implications for
long-run defcits and the national debt, in contrast with countercyclical poli-
cies that can in principle balance out defcits in lean years by surpluses in fat
years. In spirit this is Abba Lerner’s “functional fnance,” in which causality
runs from decisions about government expenditures and revenues to implica-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Prologue 13
tions for defcits and debt. Te contrast is with “sound fnance,” in which bal-
anced budgets constrain the management of aggregate demand.
Part VI, “Keynes in the Long Run,” extends the analysis from the short pe-
riod, in which the capital stock is assumed to be fxed, to a longer period
characterized by the recognition that investment not only contributes to ag-
gregate demand but also shapes the capacity of the economy to supply goods
and services. Chapter 17 sketches the early attempts to translate the vision of
Te General Teory to the long run, focusing on the separate contributions of
Roy Harrod (1939, 1948) and Evsey Domar (1946, 1947), whom misunder
standing has joined at the hip in the so-called Harrod–Domar theory. Te
contribution of Robert Solow (1956) was to derail the attempt to construct a
theory of the long run in which aggregate demand plays a role analogous to
its role in the short run. Again, misunderstanding has played a big role: Solow
is widely thought to have demonstrated the irrelevance of aggregate demand
in the long run, but the fact is that he simply assumed away the whole issue—
as he himself recognized. Tis chapter concludes with an analysis of the rear-
guard action of Joan Robinson (1962) to keep aggregate demand in the long-
run picture.
Chapter 18 lays out a series of long-run models that build on Robinson. An
important theoretical innovation is to modify the standard assumption that
the labor force is exogenously given by population growth. Instead, the labor
force available to the capitalist sector of the economy is assumed to be virtu-
ally unlimited, with domestic sources such as household labor and agricul-
tural labor supplemented by immigration when kitchen and farm prove insuf
fcient.
If the supply of labor is endogenous, a substitute has to be found for the
marginal-productivity theory of wages. In the models of this chapter, unem-
ployment remains a driver of money wages, as in the short run, but the focus
of wage dynamics is on a conventional wage, determined, à la Karl Marx and
his classical predecessors, by social norms and class power.
Copyright © 2021. Harvard University Press. All rights reserved.
Coupled both with the distinction between saving responding to proft and
saving responding to income and with the distinction between capital widen-
ing and capital deepening, these models provide a framework for answering
the question of whether high real wages are good for capitalism. Te answer is
yes if the positive impact on consumption demand outweighs the negative
impact on investment demand. Te yes is likely to be more solid in slack
times, when investment is more heavily weighted toward capital deepening.
Indeed, when there is considerable slack in the economy and capital widening
is unproftable, high real wages are likely to be a positive for investment as
well as for consumption, since a high real wage increases the proftability of
capital deepening.
Chapter 19 tests the empirical implications of the model against a half cen-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
14 Raising Keynes
tury of U.S. data on infation and employment. Just as the depression-oriented
short-run model elaborated in Chapter 6 is characterized by equilibria at
which money prices and wages fall at the same percentage rate, with the result
that the real price is stationary over time, so the long-run model of chapter 18
is characterized by equilibria at which price and wage infation take place at
the same rate. Is higher infation the price of greater output and employment?
As long as the economic change is demand driven, the model implies a posi-
tive relationship between infation and employment; that is, it implies the fa-
miliar Phillips-curve result. Supply-side shocks are more complicated—de-
pending on the adjustment mechanism that is assumed—as are changes in
the conventional wage. Not to keep the reader in suspense, I conclude that the
data are consistent with the implications of the long-run models presented in
chapter 18, particularly the contention that aggregate demand matters in the
long run as well as in the short run. Tis chapter concludes with a discussion
of why infation at rates that characterized the United States over the half cen-
tury between 1970 and 2020, or even the more elevated rates that character-
ized the period between 1970 and 1990, have been generally held to be an evil
on a par with the black plague.
A short epilogue concludes the book.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. I .
Background
The Rise and Fall
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 1 .
introduction
Is This Resurrection Necessary?
I believe myself to be writing a book on economic theory which will
largely revolutionize—not, I suppose, at once but in the course of the
next ten years—the way the world thinks about economic problems.
—john maynard keynes to george bernard shaw,
january 1, 1935
Right afer Lehman Brothers went belly-up in the fall of 2008, I overheard
several younger members of my department earnestly arguing in the foyer of
the Littauer Building, the home of the Harvard Economics Department.1 Te
subject was the likely efect of the emerging fnancial crisis on the real econ-
omy, so-called—the economy of output, consumption, investment, and em-
ployment. Opinions were divided, but the consensus was clearly that the f
nancial mess should have little or no impact. Afer all, the productive capacity
of the economy was unafected by the fall of Lehman Brothers; why should
production, and hence income take a hit? And if production and income re-
mained unafected, why should employment and consumption or investment
be afected? Shades of 1930 to 1933: I daresay that when the Littauer Building
was still a gleam in the eye of Lucius Littauer, faculty and students were hav-
Copyright © 2021. Harvard University Press. All rights reserved.
ing the same debate and were reaching the same conclusions.
Well, not exactly. As 1930 turned into 1931 and 1931 into 1932, it became
less and less possible to deny the reality of the Great Depression and harder
and harder to avoid the connection between fnance and physical reality. Te
stock market had crashed in the fall of 1929, and at frst the downturn in the
real economy—which had in fact preceded the stock market crash by a few
months—seemed no worse than earlier detours on the road to ever greater
prosperity. But by 1932 the fnancial structure and the real economy were
both in shambles. In the early 1930s banks failed at double, then triple, the
rates of the 1920s, with losses to depositors reaching more than six times the
average of the 1920s (FDIC [n.d.]). Te real economy of the early 1930s was a
17
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 Raising Keynes
shadow of the Roaring Twenties. Employment decreased by more than 20
percent between 1929 and 1933, gross domestic product by more than 25
percent.2 Economists and historians continue to sort out cause and efect in
the relationship between money and fnance on the one hand and production,
income, and employment on the other and to assess the role of government
policies in alleviating or contributing to the agony of the Great Depression.
But there is an important diference in the discussion then and now: in the
early 1930s no framework existed within economics for integrating money,
fnance, and the real economy, a gap flled by John Maynard Keynes’s General
Teory of Employment, Interest, and Money only in 1936.
“Keynes” and “Keynesian” evoke and provoke. Tere is the impact on eco-
nomic theory: what was Keynes’s contribution? Was, as some have main-
tained (see for instance David Laidler [1999]), Te General Teory old wine in
new bottles? Did Keynes really have anything new to say, or were latter-day
critics right to relegate him to a formalizer of sand in the wheels? What about
policy? What were the policy implications of Te General Teory, and why did
these ideas fall into desuetude? Are there good grounds for rejecting Keynes
as a policy guru? In sum, why has Keynes gone from being the prophet of a
new economics to oblivion—and back?
Te short answer is that the critics won the day for good reasons and bad
ones. Tey had the better of the argument once it was formalized in the only
framework available to economists in the years following the publication of
Te General Teory. I demonstrate this in chapters 3 and 4.
But Keynes was aware of the need for a diferent framework, one that could
cope with real-time adjustment, even though, like Moses, he was unable
to enter the promised land. When what Joseph Schumpeter called the pre-
analytic vision behind the theory is formalized in the appropriate framework,
which this book does in chapters 5 to 7, the critiques ofered by friend and foe
turn out to support Keynes rather than to undermine him.3 Te rest of this
chapter sketches fuller answers to these questions by assessing Keynes’s con-
Copyright © 2021. Harvard University Press. All rights reserved.
tribution in Te General Teory and the reactions from both the economics
profession and the larger society. Te rest of the book flls in the sketch.
Why Was/Is The General Theory Important?
Prior to the publication of Te General Teory, Keynes had been writing for
more than fve years about the need for more activism on the part of govern-
ments, particularly his own and the government of the United States. Indeed,
Britain had never really enjoyed the pre-Depression boom, a fact that Keynes
attributed to the misguided exchange-rate policy he had opposed since the
mid-twenties. In the context of the General Election held in May 1929—well
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 19
before the U.S. stock market crash and the ensuing collapse—Keynes had ad-
vocated public works to stimulate the British economy (Keynes and H. D.
Henderson, “Can Lloyd George Do It?”, abridged as “A Programme of Expan-
sion” in Keynes 1931a, pp. 118–134).
A major goal of Te General Teory was to provide a framework for such
policy initiatives, a framework for integrating private initiative and collective
action—government intervention—to make efective use of the available re-
sources. Te General Teory was, afer all, conceived in the womb of the Great
Depression and motivated by the need to make sense of the most glaring
economic disaster in the history of capitalism.
If this had been Keynes’s only goal he most likely would have located the
argument in the specifcs of the British economy of the 1920s and 1930s, tak-
ing for granted the overwhelming importance in manufacturing and trans-
portation of large-scale monopolistic and oligopolistic industries with sub-
stantial control over prices and wages. Tis was of course in stark contrast to
the assumptions of a competitive economy, in which frms were subservient
to the market, prices and wages being set by the interaction of demand and
supply and taken by each participant as beyond his control.
Had he situated his argument in the realities of the economy of his day,
Keynes could have appealed to the theory of monopoly or the newfangled
theories of monopolistic competition, to which his own protégée, Joan Robin-
son, along with Harvard’s Edward Chamberlin, was a leading contributor. But
for the purpose of converting the economics profession to a new point of
view, the reality of the twentieth century was less important than the received
doctrines of competitive economics. Shortly afer the publication of Te Gen-
eral Teory, Keynes wrote to the Swedish economist Bertil Ohlin, “Te refer-
ence to imperfect competition is very perplexing. I cannot see how on earth it
comes in. Mrs Robinson, I may mention, read my proofs without discovering
any connection” (Keynes 1973b, p. 190).
Paradoxically, to have focused too closely on reality would have diverted
Copyright © 2021. Harvard University Press. All rights reserved.
attention from the central question of whether any new theory was necessary.
If the problem lay in monopoly or other departures from the ideal of perfect
competition, wasn’t the solution simply to make the world resemble more
closely the perfectly competitive model? Tis is the enduring mantra of main-
stream economics, as central to Adam Smith’s attack on mercantilism in the
late eighteenth century as it has been to the neoliberal project of dismantling
government “interference” and unleashing capitalism that emerged in the late
twentieth century.
In the 1930s, almost to a man (precious few economists were w omen),
economists believed that slumps were limited in duration and severity by a
self-correcting market mechanism; that slumps were salutary in purging the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
20 Raising Keynes
economy of excesses that inevitably accompanied booms; and—most impor
tant—that anything the government might do in the way of positive action
would most likely make matters worse. Te minority of economists who be-
lieved in activist government as the counterweight to a failing private sector
had no coherent framework in which to express their analysis and conclu-
sions. By situating the argument of Te General Teory in a competitive
economy, Keynes sought to convince his fellow economists that the problem
of unemployment had deeper causes than frictions and imperfections and
that attempts to address unemployment by making the economy more like
the textbook were doomed to fail.
Te basis of the idea that a market system has a built-in mechanism for
avoiding lengthy, deep slumps is that the very existence of involuntary unem-
ployment will induce an adjustment of wages, an adjustment that ends only
when the demand for labor on the part of proft-maximizing frms is equated
with the supply of labor provided by utility-maximizing households. Frictions
apart, the only obstacle to full employment recognized by orthodoxy was
the undermining of competition by the market power of some actors (trade
unions have long been a favorite bogeyman of economic orthodoxy) or the
intervention of the state (unemployment insurance and other income-support
measures of the welfare state are other bogeymen). Imperfectly competitive
labor markets may permanently prevent the wage rate from settling at a point
where there is a job for every willing worker, but the cure is implicit in the
diagnosis. More and better (read more competitive) markets were—in 1936 as
in 2020—the cure for market failure. It was literally unthinkable for an ortho-
dox economist that there could be a systemic failure of a competitive market
system, one which could lead to prolonged, profound, and painful unemploy-
ment. Any economist who thought so branded himself a Marxist or some
other stripe of heretic.
By contrast, frictions and imperfections, just as they are in 2020, were the
stuf of normal economics. So much so that by 1936, when Te General Te-
Copyright © 2021. Harvard University Press. All rights reserved.
ory was published, there was little need and equally little appetite for a new
theory of unemployment based on frictions or other market imperfections.
Only three years earlier, the “prof,” Arthur Pigou, had laid out the argument
in a well-received book called Te Teory of Unemployment.4 Nor, as I have
mentioned, was there any lack of policy recommendations for government
intervention to fght frictional unemployment, including public works.
Te confict over whether the problems of capitalism were superfcial or
systemic was not a new one; neither was it limited to academic economists.
Before World War I, three economic visions contended with each other in the
United States. Te orthodox Republican view, which came into its own in the
1920s, was that the government should stay out of the way of an economy that
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 21
had delivered, and would continue to deliver, the goods—literally. Two com-
peting visions took a less rosy view of unbridled capitalism. On the one hand
was the vision proclaimed by Teodore Roosevelt as the New Nationalism
during the electoral campaign of 1912: big business dominated the economic
landscape because of its ef ficiency, and the role of government was to provide
a countervailing power. On the other hand, against the New Nationalism,
was Woodrow Wilson’s New Freedom: big business dominated because of its
predatory power, and the role of government was to level the playing feld so
that small business had a fair chance.
Te Depression gave new point to these competing visions of the nature of
the economy and the appropriate role for the government. With a third of the
industrial labor force unemployed and a corresponding amount of capital
underutilized, the issue was no longer whether an economy dominated by
big business deployed labor and capital ef ficiently; it was now a question of
whether any sort of market economy could and would mobilize resources ef-
fectively. If the Republican vision was momentarily sidelined, echoes of the
debate between followers of the frst Roosevelt and followers of Wilson could
be heard in the debates about policies that Franklin Roosevelt should follow
to combat the Depression.
Keynes recognized that any argument for government intervention to steer
the economic ship had to be grounded in a convincing argument as to why
the ship could not steer itself. Te General Teory thus had to do more than
show how unemployment might persist in the workings of the actual econ-
omy, 1930s capitalism warts-and-all. As long as no one could be sure whether
the problem was capitalism or the warts, it was impossible to counter the
reigning orthodoxy, which held that it would be enough for the government
to remove the warts.
For Keynes nothing less than an all-out attack on the received doctrine, one
that went well beyond the superfcialities of frictions and market structure,
would do. Given the hegemony of the belief in the fundamental resilience and
Copyright © 2021. Harvard University Press. All rights reserved.
benefcence of markets, it had to be shown that even competitive markets
were defective when it came to providing jobs.
In a BBC broadcast in 1934, when he was a year away from fnishing Te
General Teory, Keynes pinpointed the fundamental diference between an
approach to the Depression based on frictions and imperfections and an ap-
proach based on more fundamental defects in the market system:
On the one side were those who believe that the existing economic system
is, in the long run, a self-adjusting system, though with creaks and groans
and jerks, and interrupted by time- lags, outside interference and mis-
takes . . .
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
22 Raising Keynes
Te strength of the self-adjusting school depends on its having behind it
almost the whole body of organized economic thinking and doctrine of the
last hundred years . . .
If the heretics on the other side of the gulf [among whom Keynes in-
cluded himself] are to demolish the forces of nineteenth century orthodoxy
. . . they must attack them in their citadel. (Keynes 1934, p. 850)
Te heretical view did not for a moment deny the existence of imperfections,
but in this view the role of monopoly and its attendant frictions was, to say
the least, exaggerated. Te problem of capitalism lay much deeper.
In short, to accomplish the constructive goal of providing a framework for
economic policy, Keynes had frst to achieve the critical goal of showing that
a capitalism resembling the idealized version of the textbooks could still fall
short of providing jobs for willing workers, could carry on indefnitely with
idle men and idle plant. For this reason Keynes eschewed appeal to theories of
monopoly, oligopoly, or monopolistic competition. His very title, emphasiz-
ing the greater generality of his theory compared to the reigning orthodoxy,
precluded the appeal to the kinds of imperfection that the orthodox readily
recognized and just as readily deplored.
Keynes saw himself, then, as faced with a twofold task, persuading his fel-
low economists to abandon old ways of thinking about how markets work
and laying out the framework for a new way of approaching the policy prob
lem of maintaining full employment. Keynes had no doubt which part was
more challenging. As he puts it in the preface to Te General Teory,
Te composition of this book has been for the author a long struggle of es-
cape, and so must the reading of it be for most readers if the author’s assault
upon them is to be successful,—a struggle of escape from habitual m odes of
thought and expression. Te ideas which are here expressed so laboriously
are extremely simple and should be obvious. Te dif ficulty lies, not in the
new ideas, but in escaping from the old ones, which ramify, for those
Copyright © 2021. Harvard University Press. All rights reserved.
brought up as most of us have been, into every corner of our minds. (p. viii)
To compound the dif ficulties which Te General Teory presented, Keynes
had in mind a revolution not only in economic policy and in economic the-
ory, but also in economic method. Tis third revolution involved nothing less
than changing the analytic framework from a static conception of equilibrium
based on equality of demand and supply to a dynamic framework based on
how the economy adjusts when not in equilibrium. Tis was not so much a
rejection of equilibrium but a reconceptualization of its meaning and role.
At the very outset of Te General Teory, in the preface, Keynes writes,
My so-called “fundamental equations” [in Keynes’s earlier book, A Treatise
on Money] were an instantaneous picture taken on the assumption of a given
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 23
output. Tey attempted to show how, assuming the given output, forces
could develop which involved a proft-disequilibrium, and thus required a
change in the level of output. But the dynamic development, as distinct from
the instantaneous picture, was lef incomplete and extremely confused. Tis
book, on the other hand has evolved into what is primarily a study of the
forces which determine changes in the scale of output and employment as a
whole. (p. vii)
However, the methodological content of that revolution was never well un-
derstood by the economics profession. Perhaps not by Keynes himself; he had
the intuition but lacked the tools to carry through this part of his program.
Te economics profession, as it has evolved over the eighty-plus years since
the publication of Te General Teory, has long since come to possess the
tools, but has never developed the intuition. Tis book attempts to fll that
gap, to marry formal models to Keynes’s intuition, to deploy the tools neces-
sary to understand what Keynes was about.
In a word, Te General Teory is methodologically based on diferences
between three concepts: the price mechanism, equilibrium, and market clear-
ing.
Te price mechanism is the process that is supposed to bring about equilib-
rium, to adjust intentions of agents so that at the end of the day (or week or
year or whatever time period might be posited) these intentions are in bal-
ance with one another.
Equilibrium describes the balance. It characterizes a situation where forces
tending to move the economy in one direction are just counterbalanced by
opposing forces.
Market clearing describes a situation where demands and supplies just
equal. It is neither a necessary nor a suf ficient condition for equilibrium.
Mainstream economics, viewed in the lens of Keynes, makes two mistakes.
First it uncritically identifes market clearing with equilibrium, when market
clearing is only one of many possible ways in which opposing forces might be
Copyright © 2021. Harvard University Press. All rights reserved.
in balance. Second, the profession emphasizes equilibrium and treats the
price mechanism, the adjustment process, as simply an adjunct to equilib-
rium. Elementary texts dismiss the price mechanism with a paragraph of
hand-waving, with a just-so story of how excess demand or excess supply is
eliminated. Graduate texts are no better. A leading text devotes fewer than ten
pages (out of one thousand) to the price mechanism, justifying this imbalance
with a frank admission:
We have, so far, carried out an extensive analysis of equilibrium equations.
A characteristic feature that distinguishes economics from other scientifc
felds is that, for us, the equations of equilibrium constitute the center of our
discipline. Other sciences, such as physics or even ecology, put compara-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
24 Raising Keynes
tively more emphasis on the determination of dynamic laws of change. In
contrast, up to now, we have hardly mentioned dynamics. Te reason, infor-
mally speaking, is that economists are good (or so we hope) at recognizing a
state of equilibrium but are poor at predicting precisely how an economy in
disequilibrium will evolve. (Mas-Colell, Whinston, and Green 1995, p. 620)
Te stakes are high: as we shall see, Keynes’s contention that a capitalist econ-
omy lef to its own devices will not gravitate to full employment cannot be
demonstrated on the basis of static equilibrium. Quite the contrary. As I indi-
cated at the outset of this chapter, on that playing feld the critics won.
Nor can anything more than a partial case be made for the limits of mone-
tary policy as a tool of economic management—though this case is particu-
larly relevant for severe downturns like the Depression or the long slump that
followed the fnancial crash of 2008. To deliver the knockout blow to the
complacent theory of a self-adjusting economy or an economy steered by
monetary policy alone, it is necessary, as Keynes intuited, that we focus on the
process of adjustment, and this is necessary whether the goal be to under-
stand the limits of laissez-faire or the limits of central banks.
Chapters 3 and 4 tell the story of what happens when the adjustment pro
cess is ignored. Keynes introduces a fxed money wage as a simplifying as-
sumption to facilitate his narrative. Franco Modigliani, a Nobel Laureate
known all his life as a Keynesian, took this feature of Keynes’s exposition and
made it central to the analysis. In Modigliani’s words,
It is usually considered as one of the most important achievements of the
Keynesian theory that it explains the consistency of economic equilibrium
with the presence of involuntary unemployment. It is, however, not suf fi
ciently recognized that, except in a limiting case to be considered later, this
result is due entirely to the assumption of “rigid wages.” (Modigliani 1944,
p. 65)
Copyright © 2021. Harvard University Press. All rights reserved.
Tus Keynes becomes a theorist of sand in the wheels: get rid of the friction of
rigid money wages and all will be well in the economy.
Modigliani demonstrates the existence of a full-employment equilibrium
in a model based on Keynes’s theoretical apparatus—without the assumption
of a fxed money wage. In efect, his analysis says that if we start from a par-
ticular level of the money wage and the outcome involves involuntary unem-
ployment, reduce the money wage and recalibrate the equilibrium. Run the
movie again, or more accurately, take another snapshot. His contention is that
if we do this repeatedly, we will fnd a money wage for which the equilibrium
is consistent with full employment.
Te important distinction is that Modigliani’s approach does not literally
imply changing the wage: it is not about change in the sense of the wage frst
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 25
being at one level, say $20 per hour, and then being reduced to $15 per hour.
Modigliani’s logic of “change” rather invites us to imagine a series of planets
that are identical in all but one respect: the level of the money wage. Te lit-
eral contention is that if we go down the list of these planets, comparing their
static equilibria, we will fnd one in which the wage meets Goldilocks, neither
too high nor too low, but just right for full employment. No change, in the
sense of movement in real time, is involved. On the Goldilocks planet, the
money wage is now, always has been, and always will be at the level necessary
for a full-employment equilibrium.
Contrast Keynes’s own approach. When he fnally drops the assumption of
a fxed money wage (chapter 19 of Te General Teory), the discussion is all
about what happens when the wage starts out at one level and then falls, per-
haps but not necessarily under the pressure of the unemployed competing for
jobs. It is all about change in the sense of adjustment. To be sure, the discus-
sion is not very satisfying, especially to the twenty-frst-century reader. Lack-
ing the tools for a formal mathematical argument, Keynes falls back on a cata-
log of the advantages and disadvantages for employment of reducing money
wages.
Te diference is between comparative statics (the study of equilibria on
diferent planets) and dynamics (the study of adjustment when the economy
starts from a disequilibrium position). It is not clear why the mainstream has
paid so little attention to dynamics. As Mas-Colell, Whinston, and Green ar-
gue, dynamics is much harder, but economists do not always shy away from
dif ficult problems.
Dif ficulties aside, one reason for avoiding these issues is that such theoriz-
ing as has been done tends to undercut rather than reinforce the basic argu-
ments of mainstream theory. All theory is by its very nature unrealistic, if for
no other reason than that it must be a map which simplifes the territory. But
the theoretical assumptions we need to make about dynamics in order for
static equilibrium to be a plausible way of characterizing an economy render
Copyright © 2021. Harvard University Press. All rights reserved.
the theoretical assumptions required for the existence of equilibrium the very
essence of realism. It is no wonder that the few studies that have addressed
the problem have sunk like stones in the sea of economic theory (for example,
F. Fisher 1983).
A second reason is the nature of the dif ficulties that economists would en-
counter were they to take dynamics seriously. Te very multiplicity of plausi-
ble adjustment mechanisms would make it necessary for economists to dirty
their hands in the messy complexities of how agents actually behave in real
life, an investigation of distinctly lower status that blurs the line between eco-
nomics, a discipline aspiring to the status of science, and anthropology, for-
ever tainted in the eyes of most economists by its reliance on interpretation.
Indeed, you can count on one hand the serious studies during my lifetime of
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
26 Raising Keynes
how agents actually set wages and prices (for example, Bewley 1999; Blinder
et al. 1998.)
A third reason for the neglect of dynamics, and I must be more tentative
here, is the comfort that economists may have drawn from Paul Samuel-
son’s “correspondence principle.” As developed by Samuelson in the 1940s,
the correspondence principle establishes a relationship between static proper-
ties of equilibrium and the dynamics of adjustment when equilibrium is dis-
turbed (Samuelson 1947). Specifcally, the correspondence principle relates
these static properties to the question of whether the equilibrium is stable or
unstable, stability characterizing the situation where disequilibrium adjust-
ment will lead back to the original equilibrium and instability the situation in
which the trajectory leads ever further away. Te analysis provides legitimacy,
at least under certain circumstances, for analyzing change without ever ad-
dressing the adjustment process, instead simply comparing static equilibria as
if they were a set of observations from diferent planets.
Te relationship between static equilibrium and dynamic adjustment has
not gone completely unnoticed in the attempt to fgure out Keynes’s message.
Samuelson’s own attempt (1947, pp. 276–283) represents one such efort, not
one, I should add, that appears to have advanced our understanding. More
promising was Don Patinkin’s magisterial Money, Interest, and Prices: An Inte-
gration of Monetary and Value Teory, the leading text for graduate students
in the heyday of Keynesian macroeconomics. Indeed, Patinkin makes dis-
equilibrium the key to Keynes’s argument that laissez-faire, even bolstered by
monetary policy, cannot be relied on to propel the economy to full employ-
ment.
But Patinkin frames the argument against laissez-faire in terms of the polit
ical unacceptability of the wage (and price) adjustments that might be neces-
sary to achieve full employment, not the diference between the existence of a
full employment equilibrium and the nonexistence of a path from the here of
unemployment to the there of full employment. (See Patinkin 1965 [1956],
Copyright © 2021. Harvard University Press. All rights reserved.
secs. XIII:3 and XIV:1.) When it comes to the efficacy of monetary policy,
Patinkin does distinguish between there and getting there, but the argument
is that getting there might mean a slow and bumpy ride. Only by stepping
outside his models does Patinkin fnd reasons—perverse behavior of expecta-
tions and redistribution of income accompanying changes in the price level—
for arguing that the problem is worse than a slow and bumpy ride: as the
Maine farmer, afer refecting on the matter for some time, told the tourist
asking directions to another town, “You can’t get there from here.” Says Patin
kin, in summary:
Keynesian economics is the economics of unemployment disequilibrium.
It argues that as a result of interest-inelasticity, on the one hand, and dis
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 27
tribution and expectation efects, on the other, the dynamic process of [a
capitalist economy lef to its own devices]—even when aided by mone-
tary policy—is unlikely to converge either smoothly or rapidly to the full-
employment equilibrium position. Indeed, if these infuences are suf ficiently
strong, they may even render this process unstable. In such a case the return
to full employment would have to await the fortunate advent of some exog-
enous force that would expand aggregate demand suf ficiently. (1965 [1956],
pp. 337–338)
Samuelson had begun to deploy the tools for analyzing disequilibrium. But by
the time the tools were adequately developed, the profession had moved on.
As Milton Friedman, another Nobel Laureate, who became the personifca-
tion of the anti-Keynes, put it in 1970 (p. 207),
Keynes’s basic challenge to the reigning theory can be summarized . . . As a
purely theoretical matter, there need not exist, even if all prices are fexible,
at long-run equilibrium position characterized by “full employment” of re-
sources . . .
[Tis] proposition can be treated summarily because it has been demon-
strated to be false. Keynes’s error consisted in neglecting the role of wealth in
the consumption function or, stated diferently, in neglecting the existence
of a desired stock of wealth as a goal motivating savings. All sorts of frictions
and rigidities may interfere with the attainment of a hypothetical long-run
equilibrium position at full employment; dynamic changes in technology,
resources, and social and economic institutions may continually change the
characteristics of that equilibrium position; but there is no fundamental
“faw in the price system” that makes unemployment the natural outcome of
a fully operative market mechanism.
Friedman notwithstanding, Keynes was right: there is a faw in the price sys-
tem. A major purpose of this book is to lay it bare.5
Copyright © 2021. Harvard University Press. All rights reserved.
Te General Teory of Employment, Interest and Money is a theory of how
monetary and real factors jointly determine real output and employment
along with the level of prices and wages. “A monetary economy,” according to
Keynes, “is essentially one in which changing views about the future are ca-
pable of infuencing the quantity of employment” (Te General Teory, p. vii).
Tis does not on its face sound like a revolutionary manifesto, but it is. Econ-
omists would not have been surprised to learn that changing views about the
future might infuence the composition of output and employment. If people
were more optimistic about the future, more inclined to put resources into
improving or expanding the capital stock, investment and saving would be
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
28 Raising Keynes
expected to increase at the expense of consumption. But the idea that chang-
ing views about the future might afect the level of output and employment
today was in fact quite novel. Tere might be frictions in the adjustment pro
cess, but the composition of output could adjust without compromising, at
least not compromising for long, the ability of a free-market system to accom-
modate the supply of labor. Te idea that views of the future might have an
impact on the current level of employment and output few in the face of the
most basic and revered lessons of economics.
Tat this could happen because of money compounded the heresy. From
David Hume on, money was considered to be a veil that obscured the work-
ings of the real economy. Te classical dichotomy put on one side “real” quan-
tities like the number of hours required to produce a ton of steel, or the quan-
tity of goods a worker received for an hour’s work, or the physical output of
the economy; the real side was separate from the side of “nominal” quantities,
like the quantity of money in circulation, the price of a ton of steel, or the
money-wage rate.6 According to the classical dichotomy, real magnitudes
were determined by other real magnitudes, nominal magnitudes by other
nominal magnitudes. (My colleagues debating the impact of the fall of the
House of Lehman on September 15, 2008, were in fact debating the classical
dichotomy.) Te so-called quantity theory of money, according to which the
quantity of money determines the price level, but not the level of production,
is a corollary of the classical dichotomy.7
All this Te General Teory rejected. Changing views about the future
would change people’s willingness to hold money, which in turn would afect
interest rates. Tat interest rates would change in response to views about the
future was all well and good: from Adam Smith on, precisely the role of the
interest rate was to equalize the desire to save with the desire to invest. But in
the received doctrine, the interest rate operated solely on the real side of the
economy. For Keynes the interest rate was two headed, one head on the
nominal side of the classical divide, infuenced by the stock of money and li-
Copyright © 2021. Harvard University Press. All rights reserved.
quidity preference, the other on the real side, infuencing investment demand,
and thereby aggregate demand. Te rate of interest afected not only the com-
position of output but also its level. Trough aggregate demand, the monetary
side of the classical divide afects output and employment. Tis is the principal
take-away of Te General Teory. Not for nothing was the full title Te Gen-
eral Teory of Employment, Interest and Money.
So the seemingly innocuous idea that the essence of a monetary economy
is that changing views about the future could afect the level of output and
employment turns out to have far-reaching implications. On the side of the-
ory, the time-honored view that—frictions apart—laissez-faire would lead to
full employment goes by the board. On the policy side, the government must
assume the role of steering the economy to full employment, and a purely
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 29
monetary policy geared to controlling the level of the short-term rate of inter-
est might fall short. On the methodological side, a new approach was called
for, one that eschewed the static equilibrium analysis based on equality of
demand and supply in favor of a dynamic focus on the process of adjustment.
The General Theory: Strategy and Structure
How to present the new ideas in a way that was not only intelligible but would
clear the mind of old ways of thinking? Te positive goal of providing a
new account of how the economy actually operates and the negative goal
of demolishing the old argument that a wart-free capitalism would provide
full employment were complementary—but hardly identical. Indeed the two
goals could and did get in the way of each other.
Te strategy of exposition Keynes a dopts is frst to lay out the critical mon-
etary and real determinants of employment and output, then to show how the
monetary and real elements of the economy are linked, fnally to remove an
important part of the scafolding needed to erect his theoretical edifce. We
shall examine this scafolding in detail beginning in chapter 3. For now a
broad-brush summary will suf fi ce. A frst pass at modeling the economy is
based on
• a fxed money wage;
• a given rate of interest;
• the relationship between the rate of interest and the amount of expendi-
ture people are willing to make on renewing and expanding the stock of
real capital (plant, equipment, infrastructure, housing);
• the relationship between people’s incomes and how much they are will-
ing to spend on consumption goods;
• and fnally, the relationship between the rate of interest and how people
allocate their wealth among various assets, including money, as well as
Copyright © 2021. Harvard University Press. All rights reserved.
the relationship between the level of output and the money required to
fnance the transactions associated with this level of economic activity—
though little attention is paid to the relationship between output and
money, the transactions demand for money, presumably because it was
well understood and accepted.
In order to determine the price level, which according to the preface to Te
General Teory, is what makes the theory general,8 Keynes appends
• a conventional supply curve based on p roft maximization, a relation-
ship between the level of prices (given the money wage) to the level of
output.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
30 Raising Keynes
Two important assumptions distinguish this frst-pass model from the revi-
sions that are introduced later in the argument. First of all is the fxed money
wage, taken as an essential element of the theory by later commentators, be-
ginning, as we have seen, with Franco Modigliani—despite Keynes’s insis-
tence that a fxed money wage is a “simplifcation, with which we shall dis-
pense later, . . . introduced solely to facilitate the exposition” (Te General
Teory, p. 27). Second is the assumption of a given rate of interest. By specify-
ing the rate of interest along with the money wage, Keynes can argue that the
level of output and employment is determined solely on the demand side,
supply entering only to determine the price level and, given the level of
money wages, the real wage as well (Te General Teory, pp. 27–32). Te link
with the monetary side of the economy at this point is therefore quite one-
sided: the given rate of interest determines the amount of money agents will
hold as an asset, and the level of output determines the amount of money re-
quired for transactions.
Several points are worth making here. First, Keynes, as I have noted, does
not bake the cake; he only provides the list of ingredients. Nowhere in Te
General Teory does Keynes present his argument in the form of a system
of equations that economists of the twenty-frst century, or indeed, econo-
mists of the younger generation in 1936, would recognize as a model. Sec-
ond, reviewers paid no attention to this model, presumably because of its
self-evidently provisional nature. Chapter 3 of the present book flls in this
frst-pass model, not out of antiquarian concern with the details of Te Gen-
eral Teory, but because the frst-pass model is the best starting point for
understanding the new policy framework that Keynes intended. Moreover,
the frst-pass model provides a basis for the methodological revolution that
Keynes also intended, replacing static equilibrium by a dynamic adjustment
process.
Afer laying out the frst-pass model, Keynes turns his attention to ex
ploring the psychological underpinnings of the key elements of the theory,
Copyright © 2021. Harvard University Press. All rights reserved.
the propensity to consume, the demand for physical investment, and the
liquidity-preference relationship between the demand for money as an asset
and the rate of interest that obtains in the money market. Having done so, he
is in a position to put these three relationships, along with the transactions-
demand for money, together in the form of a second-pass model, one that
starts not from a given rate of interest but from a given money supply.
In Keynes’s second-pass model, also laid out in chapter 3, the relationship
between the rate of interest and the amount of money that people are will-
ing to hold in cash or cash equivalents (like checking accounts and other liq-
uid assets)—the theory of liquidity preference—replaces the assumption of a
fxed interest rate. Since the overall supply of money is now fxed, the rate of
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 31
interest becomes one of the unknowns in the system of equations that encap-
sulate the underlying behavioral relationships.
Once again, Keynes tells us what goes into the model, but he does not for-
malize the model (Te General Teory, chap. 18), a pity, because he might
have discovered that the demand and supply sides of the second-pass model
are now interconnected. He simply repeats the assertion that employment
and output are determined on the demand-side alone, while dropping an as-
sumption—a given interest rate—crucial to that result. It is not true that
“if we take as given the factors specifed above [propensity to consume,
investment-demand, and liquidity preference, along with a given money wage
and a given money supply], these variables determine the national income (or
dividend) and the quantity of employment” (Te General Teory, p. 247)—
unless the price level also is taken as given. But we can’t fx the price level and
hold onto the idea that the price level is determined by the goods-supply
schedule, an assumption that Keynes makes early on, in the context of his
frst-pass model, and never abandons.
What we get from the relationships and parameters that characterize the
second-pass model is a downward sloping aggregate-demand schedule, de-
mand as a function of the price level, rather than the vertical schedule that
falls out of the frst-pass model—a big diference: once the AD schedule
is downward sloping rather than vertical, it is no longer true that the only
role for supply-side conditions is to determine the price level and the real
wage. Demand and supply—Alfred Marshall’s famous scissors—determine
both quantity and price.
Once again there is a bit of paradox in the sequel to Te General Teory.
Te second-pass model was formalized by John Hicks (1937) soon afer the
publication of Te General Teory, but in the next two decades the goods-
supply equation was dropped from Hicks’s model, and the standard represen-
tation of Keynes’s theory assumed fxed money wages and prices. By the 1960s
this modifed version of Hicks, the so-called IS-LM model, became canonical
Copyright © 2021. Harvard University Press. All rights reserved.
textbook Keynes. Missing was any recognition that with given prices and
wages the model determined one point on the entire AD schedule rather than
an equilibrium level of output and employment.
Indeed, the very language of Keynesian models from the 1950s through the
1970s is misleading: whereas the LM schedule is (correctly) presented as rep-
resenting equilibrium in fnancial markets given the price level—clearing the
money market in the jargon—the IS schedule is wrongly presented as clearing
the goods market. It does nothing of the sort. An equation characterizing
equilibrium of desired saving and desired investment, the condition of simply
being on the AD schedule, was read as if it were characterizing balance be-
tween aggregate demand and goods supply: it may be that the word “supply”
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
32 Raising Keynes
led to confusion between the supply of saving and the supply of goods (not to
mention the supply of labor). In the textbook IS-LM analysis, what Keynes
claimed made Te General Teory a general theory—that it determines out-
put, employment, and prices—had gone missing.
Te coexistence of frst-and second-pass models is one of the sources of
the confusion about the message of Te General Teory. Tere are two models
because of the dual purpose of the book, to provide both a theory of how the
fnancial and real sectors of the economy actually intertwine and a theory of
why even an idealized capitalism would fail to provide for the full use of cap
ital and labor. Te frst-pass model, precisely because it does not assume a
fxed money supply, is a better starting point for understanding how the
economy actually works, while the second-pass model is not quite up to the
job of laying bare the failure of an idealized capitalism to settle at a full-
employment equilibrium.
The Taming of the General Theory: How the
Economics Profession Dealt with the New Ideas
(or, with Friends like Modigliani . . .)
Te General Teory was initially met by incomprehension, in large part due to
the difficulty and opacity of the argument, not to mention the gaps to which I
have alluded. Indeed, Te General Teory divided the economics profession.
Older, established economists had dif ficulty with the new framework and
concepts introduced to make the case for a central role of aggregate demand,
and even more dif ficulty with the policy implications: the necessity for gov-
ernments to maintain aggregate demand in the face of a scarcity of private
demand. (Tere were notable exceptions. Alvin Hansen, once a pillar of pre-
Keynesian orthodoxy, became a leading exponent of Keynesian theory and
even more so of its implications for activist economic policy.)
Te young were equally bafed by the argument, even if more open to
Copyright © 2021. Harvard University Press. All rights reserved.
its policy implications. Samuelson describes this generational disconnect in
these terms:
Te General Teory caught most economists under the age of 35 with the
unexpected virulence of a disease frst attacking and decimating an isolated
tribe of South Sea islanders. Economists beyond 50 turned out to be quite
immune to the ailment. (1946, p. 187)
For Samuelson, money-wage rigidity coupled with involuntary unemploy-
ment was the hardest pill to swallow. Samuelson later recalled,
What I resisted in Keynes the most was the notion that there could be equi-
librium unemployment. I’d argue with Bob Bryce [a fellow graduate student
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 33
at Harvard who had attended Keynes’s lectures in Cambridge, England], and
discuss with Leontief [professor at Harvard and later winner of the Nobel
Prize for his work on input-output analysis], that frst chapter where work-
ers react diferently to an increase in money wages from the way they react
to a change in real wages that c omes from infation.
Samuelson was never persuaded. He simply put the matter to one side:
Te way I fnally convinced myself was to just stop worrying about it. I asked
myself: why do I want to refuse a paradigm that enables me to understand
the Roosevelt upturn from 1933 to 1937? It’s . . . completely untrue that the
New Deal d idn’t work until World War II came and bailed it out. Some of
the highest rates of real increase in and highest levels of plant and equip-
ment capital formation are in the period 1934 to 1937. I was content to as-
sume that there was enough rigidity in relative prices and wages to make the
Keynesian alternative to Walras operative. (Colander and Landreth 1996,
pp. 159–160)
Samuelson’s generation was not unduly troubled by the inconsistency be-
tween the standard competitive model and the models implicit in Te General
Teory. Tey were all too ready to accept that, as in Modigliani’s telling,
money-wage rigidity was not merely an expositional device, but a critical as-
sumption. In Samuelson’s words:
We [young Keynesians] always assumed the Keynesian unemployment
equilibrium foated on a substructure of administered prices and imperfect
competition. I stopped thinking about what was meant by rigid wages and
whether you could get the real wage down; I knew it was a good working
principle, a good hypothesis to explain that the real wage does not move
down indefnitely so long as there is still some unemployment. Tus I as-
sumed a disequilibrium system, in which people could not get on the supply-
of-labor curve. (Colander and Landreth 1996, pp. 160–161)
Copyright © 2021. Harvard University Press. All rights reserved.
For Modigliani, Keynes’s conclusions about the limits of the market system
depend critically on a rigid money wage that prevents the labor market from
clearing. Once the level of output is determined, the assumption of proft
maximization fxes the ratio of prices to wages, but this says nothing about
the absolute levels of prices and wages. Without money-wage rigidity there is
nothing to anchor the price level! Tis matters because the price level afects
real aggregate demand: the lower the price level, the less money is needed for
transactions at any given level of output, so—assuming the money supply is
fxed—more money spills over into asset markets. Absorbing the extra cash
requires a lower interest rate, for the amount of cash agents wish to hold var-
ies inversely with the interest rate; in its simplest form liquidity preference is a
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
34 Raising Keynes
trade-of between the higher returns available from relatively illiquid fnancial
assets and the greater security of holding wealth in the form of cash or near-
equivalents. But if the interest rate varies directly with the price level, invest-
ment demand varies in the opposite direction, since the lower the rate of
interest the more attractive are investment opportunities. Finally, higher in-
vestment demand means higher aggregate demand.
In short, once we drop the assumption of a given money wage, the level of
aggregate demand depends indirectly on the money wage via the efect of the
money wage on the price level. Te lower the money wage, the greater the
level of aggregate demand and, in consequence, the greater the level of em-
ployment and output. What then determines the money wage? Modigliani’s
answer is to assume full employment: or, more precisely, to fx the wage so
that the interest rate and investment demand are just what are required to
provide a job for every willing worker!
It may be surprising that the two models, Keynes’s and Modigliani’s, lead to
such diferent results. Afer all, doesn’t Modigliani’s model simply do what
Keynes promised: to drop the scafolding of a fxed money wage once the
building blocks of his theory were in place? In fact, the meaning of equilib-
rium is very diferent in the two models. Keynes’s second-pass equilibrium is
contingent and historical, starting from a given money wage: in chapter 19 he
asks what happens when the wage falls. Te question of what happens if the
money wage changes is, I have suggested, diferent from the question of the
existence of a market-clearing money wage (and associated price level) in a
model that abstracts from time and history. Modigliani’s equilibrium is ahis-
torical, timeless, with nothing to fx even a starting point for the evolution of
nominal wages and prices. It is, as it were, an equilibrium in the mind of God.
Equilibria in the mind of God have their uses, but it is also useful to bear in
mind their limits. In an interview with David Colander in the mid-1980s,
Paul Samuelson recounted Edwin Bidwell Wilson’s take on equilibrium (Wil-
son was a physicist whose lectures on mathematical economics Samuelson
Copyright © 2021. Harvard University Press. All rights reserved.
had attended in the 1930s):
You leave your car in the MIT parking lot overnight. Te rubber tire is a
membrane which separates the inside of the tire from the atmosphere, and
because of this stif wall there’s an equilibrium diference in pressure. Wilson
would say, “Come back a thousand years later, and that tire will be fat.” Tat
was not strict equilibrium. It’s just a very slowly adjusting disequilibrium.
(Colander and Landreth 1996, p. 163)
Modigliani’s conception of equilibrium does not invite us to examine the
process of adjustment, to investigate the consequences of moving from one
price level to another. By contrast, these consequences are crucial for Keynes,
which is why examining an economy’s trajectory starting from a given wage is
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 35
fundamentally diferent from analyzing an equilibrium in which the wage rate
is completely up for grabs, with no past and no future, existing timelessly in
the mind of God.
Te concept of a historically contingent equilibrium becomes especially
important in the context of Te General Teory because of the volatility of the
underlying behavioral relationships. An important innovation of Te General
Teory, amplifed in a response to critics in Te Quarterly Journal of Econom-
ics in 1937, is Keynes’s insistence on the role of uncertainty—as distinct from
risk—as a destabilizing factor in the economy. Keynes did not invent the dis-
tinction between risk and uncertainty, risk referring to situations in which a
probability distribution of outcomes can be deployed, uncertainty referring to
situations in which probability distributions are irrelevant, where one must
rely on some combination of hunch and convention. Uncertainty is essential
to Keynes’s argument that the underlying schedules constituting his theory
are themselves subject to sudden and violent disruption, which—given the
adjustment mechanism posited by his theory—means that a capitalist econ-
omy is subject to ups and downs in equilibrium levels of output and employ-
ment.
Te volatility of equilibrium, the property of being subject to violent dis-
turbances, is quite distinct from whether the equilibrium is stable in the sense
that there are self-correcting forces within the system that can be counted on
to guide the system to a new equilibrium if a random shock uproots the econ-
omy from its old equilibrium. It is the volatility of the underlying schedules,
perhaps more than any other consideration, that has led some of Keynes’s fol-
lowers to reject equilibrium altogether as a way of characterizing economic
outcomes.
Paul Davidson (1991) distinguishes between ergodic and nonergodic charac
terizations of the economy, that is, between economic environments charac-
terized by stable probability distributions (risk) and economic environments
characterized by the absence of stable probability distributions (uncertainty).
Copyright © 2021. Harvard University Press. All rights reserved.
For Davidson, equilibrium analysis may make sense in an ergodic environ-
ment but not otherwise, and the world we live in is otherwise. Robert Skidel-
sky’s (2009) retrospective on Keynes and Te General Teory is another ex-
ample of rejecting equilibrium as inconsistent with uncertainty.
For my own part, I do not doubt that uncertainty is an important part of
Keynes’s argument: it can lead to volatile investment demand, volatile liquid-
ity preference, and even a volatile consumption function. But this does not
persuade me that uncertainty requires us to dispense altogether with equilib-
rium. In fact, I see no alternative to equilibrium—conceived of as the poten-
tial outcome of a dynamic process—as an organizing principle for the argu-
ment of Te General Teory.
As far as the economics profession is concerned, Modigliani hit a home
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
36 Raising Keynes
run. As Samuelson made clear, the frst generation of American Keynesians
had early on accepted the idea that rigidities were central to the argument of
Te General Teory. Rigidities, like the poor, would always be with us, so if
Keynes provided a good map for fnding our way in a rigid world, that was
enough. One ought not to worry about whether Te General Teory was suf fi
ciently general.
Te frst generation, whom I have referred to as “Young Keynesians,” sought
to integrate Keynes into the mainstream canon through what Samuelson, pri-
mus inter pares, called the neoclassical synthesis. Te neoclassical synthesis,
in Samuelson’s hands, described a situation in which the government man-
aged aggregate demand to bring about full employment, mostly through fscal
policy, and thus created an environment in which the old verities and virtues
of a market system as a device for allocating resources among competing uses
came into their own.9 Active government was in this view the precondition
for the market system to function ef ficiently, at least in the short run.
Te neoclassical synthesis was undoubtedly useful in maintaining the pro-
fessional respectability of the Young Keynesians as they matured into Older
Keynesians, however much it damaged Keynes’s legacy. Had he survived the
end of World War II by more than several months—he died in April 1946—
Keynes might have said that with friends like Modigliani, who counted him-
self and was counted by others a Keynesian, he d idn’t need enemies.
How wrong he would have been! Te argument to which Milton Friedman
appeals as the Keynes slayer (1970) is not Modigliani’s, but one formulated
originally by hostile critics of Keynes, in particular by Gottfried Haberler
(1939, 1941) and independently by Pigou (1943). In brief, the “real-balance
efect” as the Haberler–Pigou argument came to be called, hinges on the
wage-price nexus. As wages vary so will prices; if wages and therefore prices
fall, the money that people hold as part of their asset portfolios will increase
in purchasing power regardless of what happens to real income. Eventually, the
holder of even a single dollar or euro, indeed of even a single yen, will become
Copyright © 2021. Harvard University Press. All rights reserved.
richer than Croesus, and her consumption demand will increase without
bound. However much aggregate demand is needed for full employment, it
will be forthcoming in a world of fexible prices and wages. As we shall see in
chapters 4 and 7, the operation of the real-balance efect is very diferent in
the comparative-statics of timeless comparisons from its operation in a world
of real-time changes. Because of its asymmetric impact on debtors, the real-
balance efect goes from being a killer argument against Te General Teory
in chapter 4 to being a pillar of support in the dynamic context of chapter 7.
As time went on, Keynes came in for treatment harsher than being reduced
to a theorist of short-run frictions. Under the banners of rational expectations
and real business cycles, the old neoclassical (“classical” in Keynes’s language)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 37
orthodoxy was revived as New Classical theory. In a competitive market sys-
tem, equilibrium, as for Pigou, is determined jointly by the goods-supply
function of proft-maximizing businesses and the labor-supply function of
utility-maximizing households. Re-outftted in the garb of fancy mathemat-
ics, this new dispensation eliminated aggregate demand in the context of cal-
culating, self-interested agents, reducing Keynes from a theorist of short-run
frictions and rigidities to a theorist of a world where most of the people were
being fooled all of the time, which is emphatically not the world of orthodox
economic theory.
Te conversion was rapid. And, as these things go, enduring. It was only
afer the fnancial crisis erupted in 2008 that New Classical theory lost its hold
on the economics profession. In Friedman’s intellectual progeny, Robert Lu-
cas, Finn Kydland, Edward Prescott, and Tomas Sargent, to name only those
who were recognized with Nobel Prizes for their contributions, economics
came full circle, to where it had been when Pigou and his ilk had argued that
unemployment was incompatible with a system of competitive markets, aris-
ing only because of the intemperate meddling of trade unions and govern-
ments.
Te confict between Keynes and orthodoxy over the ability of the invisible
hand to ensure full employment, I should emphasize, is not just an issue of
abstract theory. Te self-regulating economy is the philosophy behind the
policies of deregulation that were inficted on the economy prior to the crash
of 2008 and that bear at least some of the responsibility for the crash.10
Economists in 2020 are not of one mind about the short run. Tere are two
camps: the partisans of more or less extreme versions of New Classical theory,
for whom aggregate demand never matters, and the partisans of one or an-
other version of so-called “New Keynesian” theory, the theory embraced by
the heirs of Samuelson and his generation, for whom frictions are the core of
the argument that aggregate demand does indeed matter in the short run. Te
long run is another story: it is generally agreed that the long run is a country
Copyright © 2021. Harvard University Press. All rights reserved.
in which the writ of Keynes does not run.
Troughout the evolution of economics afer Keynes, from Samuelson and
Modigliani on, there is a thread that unites the friends and enemies of Te
General Teory: the pull toward a unifed economics that synthesizes macro
and micro, the economy as a whole with the actions of separate individu-
als. We can see this early on in Samuelson’s neoclassical synthesis (note 9
above) and in Modigliani’s and Friedman’s attempts to put the theory of con-
sumption on a rational basis (Modigliani and Brumberg 1954; M. Friedman
1957)—read, on a basis of fully informed individuals calculating and maxi-
mizing long-term utility subject to a long-term budget constraint—as against
what the profession saw as the ungrounded assumption of Keynes’s “psycho-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
38 Raising Keynes
logical law” disposing people to consume a fraction of whatever income
comes their way. (For more on the rational consumer and his discontents, see
chapter 9.)
Tis Gleichschaltung begins long before its latest and most complete mani-
festation in the guise of New Classical economics. Nor, as I have indicated, is
it limited to the avowed enemies of Keynes. Indeed, the assumption that in-
forms mainstream macroeconomics in 2020, whatever its stripe, is that macro
behavior must be grounded in the individual—not any individual, but the
“rational” individual who is formed by the microeconomic texts.
The Taming of The General Theory: The Political Response
to Keynes (Or, To Every Action . . .)
But I get ahead of my story. At the end of World War II, the economists sus-
ceptible to the blandishments of Keynes were no longer graduate students and
young instructors. Economists steeped in Te General Teory began to fll the
professorial ranks of colleges and universities and, even where not numeri-
cally preponderant, dominated in terms of energy, commitment, and the con
fdence of the bearers of a new dispensation. Samuelson himself, always a step
ahead, was already a tenured professor at MIT.
Te new prominence and even ascendance of Keynes invited reaction and
retaliation. Keynesian is a plastic word, which can mean anything from the
actual content of Te General Teory to a conception of the economy in which
the state takes responsibility for a large measure of economic activity, particu-
larly for the provision of social insurance against the loss of income because
of old age and retirement, disability, and unemployment due to a failure of
aggregate demand. Only the last of these, unemployment insurance, could be
considered Keynesian in the sense of relating to the message of Te General
Teory, though in point of fact unemployment insurance was initiated in Brit-
ain in 1911, in Italy in 1920, and in Germany in 1927—long before Keynes
Copyright © 2021. Harvard University Press. All rights reserved.
began to formulate the ideas and concepts that formed the backbone of his
magnum opus (Social Security Board 1937).
It was natural that afer the sacrifces of World War II there would emerge a
strong sentiment for an enhanced role for the government as a means of pre-
venting, to the extent possible, the calamity of depression, as well as for miti-
gating the efects of downturns to the extent prevention failed. But the senti-
ment was hardly universal, and many regretted the passing of a society based,
at least in their perception, on “rugged individualism,” with its implications of
self reliance, personal responsibility, and minimal government.
Te very idea that unemployment could be the result of a systemic failure
of demand rather than the personal choice or at least the responsibility of the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 39
individual who found himself out of work was anathema to the creed of rug-
ged individualism. Te idea that a failure of private demand might require
defcit spending, and in the worst of circumstances, defcit spending as far out
as the eye could see, compounded the original sin of Te General Teory.
And there was worse. One justifcation of economic inequality had long
been—and still is—that the rich, freed from the immediate pressures of con-
suming enough to hold body and soul together, were the source of capital ac-
cumulation, the benefts of which eventually trickled down to the lowliest
worker. Keynes himself put this justifcation forward in the opening pages of
Te Economic Consequences of the Peace, the book that in 1919 had trans-
formed him from middling civil servant—he resigned his position at the
Treasury in protest over the Treaty of Versailles—into public intellectual.
Te General Teory turned this argument on its head. In a world in which
aggregate demand was problematic, attempts to save more meant less spend-
ing on consumption, hence less output, income, and employment:
In contemporary conditions the growth of wealth, so far from being depen
dent on the abstinence of the rich, as is commonly supposed, is more likely
to be impeded by it. One of the chief social justifcations of great inequality
of wealth is, therefore, removed. (p. 373)
Tis “paradox of thrif,” as it came to be called, meant not only the loss of a
justifcation of inequality but also an argument for greater equality: if workers
and middle income folks were prepared to spend more of their income, then
transfers to them from the rich would add to aggregate demand and reduce
unemployment.
Tis was not an argument Keynes himself urged on his readers. Te Gen-
eral Teory makes no more than passing reference to the positive efects of a
more equal distribution of income on the overall propensity to consume (for
example, in chapter 22, as a counterweight to a low investment demand). In
fact, in Keynes’s own models the distribution of income between wages and
Copyright © 2021. Harvard University Press. All rights reserved.
profts is not an exogenous variable, determined outside the system, but the
consequence of the interplay of aggregate demand and aggregate supply; dis-
tribution is a thermometer, not a thermostat. Nonetheless, the idea was there
for so-called lef Keynesians to add to the reasons why more income equality
would be better than less—and one more reason for the right to oppose
Keynes and all he appeared to stand for.
In the end, however, Keynes was opposed, even vilifed, not so much be-
cause of his specifc doctrines but because he was a convenient focus for the
anxieties attendant to the transformation of the American economy, polity,
and society during a decade and a half of depression and war. Keynes was
lumped together with radical socialists and other opponents of “free enter-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
40 Raising Keynes
prise,” all seen as witting or unwitting agents of Soviet communism in its bid
for global hegemony. In this view of the world, communists were everywhere,
particularly in the government and the universities, and economists who did
not toe the free-enterprise line might be the advance party of the conspiracy,
or at least were doing the work of sofening up the country for the commu-
nists. Te indiscriminate nature of the attack on the “enemies” of capitalism is
striking: both the Veritas Foundation, which led the charge in its pamphlet,
Keynes at Harvard, and the young William Buckley, fresh from four years as
an undergraduate at Yale, Harvard’s traditional Ivy League rival (Buckley
1951), even attacked Schumpeter as a fellow traveler of Keynes and the com-
munists who would destroy the capitalist system. In point of fact, Schumpeter
was an implacable foe of Keynes and champion, albeit despairing champion,
of rugged individualism.11
Te cutting edge of the attack on Keynes was the attempt to purge college
campuses of the baneful infuence of teachers and textbooks that promulgated
the view from Te General Teory. Te campaign of alumni and other stal-
warts of rugged individualism was successful in attacking the frst text to
frame itself in the Keynesian message, Lorie Tarshis’s Elements of Economics.
Paul Samuelson was more careful and in his own words, lawyer-like (Colan-
der and Landreth 1996, p. 172), and his Economics escaped the purges.
But the economics profession had learned its lesson. I have argued that the
internal attractions of assimilating Keynes to the mainstream, stripping Te
General Teory of its most radical implications, had an intellectual appeal all
on its own. But this appeal was strengthened by the survival advantages of a
Keynesian economics that, far from reducing the mainstream to a special case
of a more general theory, reduced Te General Teory to a special case of sand
in the wheels. Teoretical accommodation served both intellectual and polit
ical purposes.
The Taming of The General Theory: Theory and Reality
Copyright © 2021. Harvard University Press. All rights reserved.
(Or, the Supply Side C
omes into Its Own)
With the passage of time there was trouble in another quarter: economic
events forced a reconsideration of the role of aggregate supply. Here Keynes’s
followers, as distinct from Keynes himself, had gone too far. In a short space
of time, the couple of decades separating the mid-1930s from the mid-1950s,
aggregate demand went from invisibility to a one-man show in the drama of
output determination.
As we have seen, in their enthusiasm for the IS-LM apparatus, Keynesian
economists lost sight of the supply side, confusing the supply of saving with
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 41
the supply of goods. Until the 1970s made it impossible to ignore the supply
side of output determination, the consensus refected in Samuelson’s neoclas-
sical synthesis was that demand management was the necessary and suf fi
cient condition for getting to full employment. As full employment was ap-
proached, the supply of labor would begin to matter, and once there, Keynes’s
economics would become irrelevant. In short, demand or supply rather than
demand and supply: up to full employment, Keynes’s frst-pass model without
any qualifcation, or the second-pass model as told by IS-LM, with fxed
prices as well as fxed wages; once full employment was attained, aggregate
demand could be ignored.
With the proft squeeze of the late 1960s, and even more so with the
oil shocks of the 1970s, the neoclassical synthesis had to be expanded to allow
a role for aggregate supply in the determination of output, regardless of
whether the economy was at full employment. In 2020, the New Keynesian
view is that both aggregate demand and aggregate supply matter in the short
run, a period defned not so much by a particular amount of calendar time,
but as the time in which rigidities constrain the adjustment of demand and
supply.
As far as the long run is concerned, the Old Keynesians surrendered before
the New Keynesians were out of diapers. In the Keynesian consensus, aggre-
gate demand does not vanish once we leave the short run, but it becomes ir-
relevant—except for the determination of the price level. As James Tobin,
who did as much as any American to develop postwar Keynesian economics,
put it, “According to the synthesis of classical and Keynesian macroeconomics
reached by 1960, Keynesian macroeconomics is short-run. It does not pre-
tend to apply to long-run growth and development” (1992, p. 392).
Tis book rejects this “Keynesian” consensus: in the story told in chapters
18 and 19, the principle idea of Te General Teory—that aggregate demand
exerts an independent infuence on economic outcomes—applies in the long
run as well as in the short run.
Copyright © 2021. Harvard University Press. All rights reserved.
Economic Policy after The General Theory
Keynes intended Te General Teory to provide the underpinnings of a new
framework for policy, but little is actually said on policy in its pages. Te pol-
icy implications of Te General Teory are nonetheless far-reaching. Te ma-
jor theoretical conclusion was that once undergoing a shock that moves it
away from full employment, an economy might remain mired in depression
indefnitely. Te economy would reach an unemployment equilibrium, equi-
librium in the sense of a position at which the forces moving the economy
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
42 Raising Keynes
one way or the other are just balanced. A full-employment equilibrium might
exist in the mind of God or his mainstream votaries, but no way exists to get
from the here of unemployment to the there of full employment. Not only
would the “self-correcting mechanism” of the market fail, but under certain
circumstances, it might be impossible for monetary policy alone to move the
economy to full employment.
Te government would have to take more direct action. Keynes does not
prescribe fscal policy, defcit spending, or any other unorthodox measure; the
most specifc he gets is a general argument for government policies to infu
ence both the demand for investment and the propensity to consume, in or-
der to equalize investment and saving at a high level of employment. To this
end Keynes speaks of the need for a “somewhat comprehensive socialization
of investment” (Te General Teory, p. 378).
But this in itself was not a new departure. Keynes had for a long time advo-
cated more government activism to address the depression, and, as I noted
earlier, even before the Great Depression began, he was an advocate of public
works as a means of creating employment—and so defcit spending.
Clarity is important here. One can fnd arguments for public works to com-
bat unemployment long before Keynes—I have found one from 1829—but
not arguments based in a framework of the economy in which unemploy-
ment is something other than a transient curse. A new framework is what Te
General Teory provided, and this framework is at the same time its major
virtue and a major cause of the opprobrium heaped on Keynes from the right.
With regard to fscal policy, it was not Keynes who drew the most radical
conclusions from Te General Teory, but Abba Lerner, a young disciple. Te
basic idea of what Lerner called “functional fnance,” in opposition to the bed-
rock prescription of “sound fnance” that the budget should be balanced at all
costs, was that government fnance should serve the goal of full employ-
ment, injecting demand into the economy when private demand is slack,
withdrawing demand when private demand is greater than the capacity of the
Copyright © 2021. Harvard University Press. All rights reserved.
economy at full employment (1941, 1943, 1944). Te point was to manipulate
aggregate demand so that it would achieve the magic of simultaneously inter-
secting with both goods supply and labor supply. Fiscal policy became not
one weapon in the arsenal of government intervention, but the only weapon
when monetary policy fell short as a means of stimulating (or, indeed, limit-
ing) aggregate demand to the necessary degree.
Countercyclical fscal policy, which is to say government expenditures
ramped up during periods of recession or depression, tax rates reduced, the
opposite in times of boom—all this was bad enough. At least there was the
prospect that over the cycle, taking good times with the bad, the government
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 43
budget might remain in balance, defcits in slumps being ofset by surpluses
in booms. But Lerner and others, especially Alvin Hansen (1938), went much
further: full employment might require a continuing and growing defcit to
ofset a continuing decline in opportunities for productive and proftable in-
vestment in the private sector. Tis vision of “secular stagnation,” buttressed
by appeals to Keynes’s own authority, held out the prospect not only of the
“euthanasia of the entrepreneur” (Keynes’s phrase) but also of a growing pub-
lic sector and, worse, a growing public debt. Te General Teory really did
lead to thinking the unthinkable.
Keynes’s infuence on policy formulation has waxed and waned; it took the
crash of 2008 to realize how profoundly, in the end, Keynes afected the polit
ical climate in which fscal policy is determined, not to mention actual poli-
cies themselves. “Well I guess everyone is a Keynesian in a foxhole,” said
Robert Lucas in October 2008, as much as anyone the father of the New Clas-
sical counterrevolution (Fox 2008a), though he quickly added, “but I don’t
think we are there yet” (2008b). Te Obama stimulus of approximately $800
billion over two years speaks volumes about the change in public opinion,
even though the efects of the stimulus are hotly contested. For all the yam-
mering of the right (and, for that matter, the center and even Obama himself)
about the unsupportable debt, educated opinion has shifed markedly since
the Great Depression, and to a large part this shif is due to the impact of Te
General Teory.
Te diference between the editorial stance of the New York Times during
the Great Depression and its editorial stance during the Great Recession is il-
luminating. During the presidential election campaign of 1932 and the run-
up to the inauguration of Roosevelt, the Times reiterated the view that the best
thing the federal government could do for the American economy was to get
its own fscal house in order. Echoing Adam Smith’s famous dictum, “What is
prudence in the conduct of every private family, can scarce be folly in that of
a great kingdom,” (1937 [1776]), editorials bemoaned the defcit. Under the
Copyright © 2021. Harvard University Press. All rights reserved.
title, “Balancing the Federal Budget,” the Times ofered this analysis:
[Te] possibility that the defcits for two years [fscal 1932 and 1933, which
is to say the period from July 1931 through June 1933] may have been un-
derestimated only strengthens the argument for increased taxation made by
[Secretary of the Treasury] Mellon. At best the Government must borrow
heavily during the next six months. . . . In these circumstances it is impor
tant for the Government to give evidence to present holders of its securities
and to those to whom it hopes to sell new issues that the budget will be bal-
anced at the earliest practicable date. As Secretary Mellon said: “Tis is es-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
44 Raising Keynes
sential not merely for maintaining unimpaired the credit of the Government
but also for reinvigorating the entire credit structure of the country.” (Janu-
ary 14, 1932, p. 20)
Just before the presidential election, the Times reiterated its stand under the
banner “Federal Budget Still Unbalanced”:
It is clear that further action must be taken when Congress reconvenes, both
to reduce costs and to provide more dependable sources of revenue. (No-
vember 4, 1932, p. 18)
Afer the election of Roosevelt, the Times trumpeted the “Sound Advice”
(this is the headline) that fnancier Bernard Baruch had ofered the Senate Fi
nance Committee:
He pointed out that the Government’s credit is good only so long as its pro-
fessed intention to balance the budget is believed. He also pointed out that
its ability to provide relief is in the long run dependent on its credit. . . .
Such testimony as Mr Baruch gave in Washington . . . supports the hope
that under President Roosevelt a new Congress, meeting in special session,
will at last move boldly to bring the Federal budget into balance and remove
all doubt of the Government’s credit. (February 16, 1933, p. 18)
On the eve of the inauguration of the new president, the advice was the
same (“Roads to Recovery”):
Te national budget was frst thrown out of balance in January 1931—
twenty-fve months ago. It has never been in balance since that time. . . .
During all this time public confdence has been disturbed by the failure of
the Government to put its fnancial afairs in order. (March 2, 1933, p. 16)
Arthur Krock, the perennial if not immortal columnist for the Times (he
retired only in the 1960s afer writing his column for more than thirty years),
Copyright © 2021. Harvard University Press. All rights reserved.
had previously been a defcit hawk, but sofened a bit afer the new president
entered of fi ce, accepting the distinction made by the Roosevelt administra-
tion between the regular budget, which Krock (and the Roosevelt people)
maintained had to be kept in balance, and the emergency budget, which re-
quired defcits in the interest of recovery. He was especially reassured by the
president’s apparent commitment to wind down the emergency budget as
soon as possible:
So determined is the President that the extraordinary budgeting shall be
prudently retired, that he has notifed Congress he will recommend a suf
fi
cient tax if that body fails to produce it. (May 19, 1933)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 45
By July, the editorial page of the Times was echoing Krock:
Te Administration argues, with much force, that it is fair to draw a distinc-
tion between charges which constitute a lasting burden on the Treasury and
emergency expenditures which can be shut of when the need for them has
passed. (July 10, 1933, p. 12)
But even sofened, the editorial position of the Times in the depths of the De-
pression is very diferent from its position as the Great Recession unfolded.
Under the title “Te Truth About the Defcit,” the Times had this to say a year
into Obama’s tenure:
Americans should be anxious, for reasons including the huge defcit. But the
cold economic truth is this: At a time of high unemployment and fragile
growth, the last thing the government should do is to slash spending. Tat
will only drive the economy into deeper trouble. (February 7, 2010)
Of course the Times, while an important voice, has no monopoly of public
opinion, and the reaction to defcit spending was building both in the United
States and abroad from the moment the ink dried on the Obama stimulus
legislation. Te G-20 summit in June 2010 revealed the deep divisions among
the governments of the world’s largest economies with regard to the proper
role of defcit spending in restoring prosperity, and the divisions are no less
profound within nations. Although there were many reasons why the Ameri-
can electorate repudiated the Democrats in the 2010 midterm elections, there
is no way that the results can be read as a ringing endorsement of the Obama
stimulus—nor for that matter the results of the 2012 elections, even though
Obama was returned to of fi ce. Keynesians would (and did) argue that the
stimulus prevented the recession from turning into a depression of the order
of the 1930s, but this argument did not resonate with the general public. Once
again, concerns for the defcit and the burden of the debt took center stage,
and these concerns, along with the political limitations of counterfactual ar-
Copyright © 2021. Harvard University Press. All rights reserved.
guments, appeared, for the moment at least, to have tipped the balance in fa-
vor of “sound fnance.” Under President Trump, the pendulum has again
swung, with former defcit hawks in the Republican Party quiet about the
defcit projections accompanying the tax cut enacted in 2017. Te important
point for purposes of this book is that the Keynesian voice is part of the de-
bate, a diference not only from the dark days of the Great Depression but also
from more recent times when Keynes’s theory was considered anachronistic
and Keynesian policies irrelevant.
Tere was more unsettling news in Te General Teory than the implica-
tions for the government defcit. John Maynard Keynes legitimized ideas that
had previously belonged to the underground of economics, chiefy the idea
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
46 Raising Keynes
that there could be too much saving. Myron Watkins, who in 1933 surveyed
the heretical responses to the Great Depression in the prestigious Quarterly
Journal of Economics, spoke for the mainstream when he characterized the
belief that more spending might promote prosperity as the hallmark of igno-
rance if not perversity. “More puerile nonsense than this would be hard to
imagine,” Watkins wrote as the Great Depression entered its fourth year (Wat-
kins 1933, p. 524).
As long as Keynes’s ideas remained in the realm of theory, none of this mat-
tered very much. Franklin Roosevelt might have been a practicing Keynesian
in countenancing budget defcits during most of the time he was president,
but in this respect he was Molière’s Monsieur Jourdain, who spoke prose for
forty years without knowing it. His budgets were unbalanced because, as Ar-
thur Krock discovered, there was no way to balance the budget and still pay
for federal programs deemed essential to recovery (and later essential to the
war efort).
Paradoxically, the frst policy use of Keynesian ideas and concepts was not
in formulating policies for recovery but in planning for World War II. Both in
Britain and in the United States, the framework of Keynes’s frst-pass model
proved serviceable in estimating the gap that had to be covered by taxation or
other means to keep aggregate demand to a level consistent with the availabil-
ity of goods afer the needs of the military and essential government services
had been met. Keynes’s pamphlet How to Pay for the War was a direct applica-
tion of the framework elaborated in Te General Teory, if to a very diferent
goal in a very diferent context.
As an end to the war became something more than a dim light on a far
shore, economists’ attention turned to planning for peace and demobilization.
Te discussion was largely in Keynes’s terms, the propensity to consume, the
demand for investment, the multiplier, all paraphernalia of Te General Te-
ory. Followers of Keynes divided sharply in their forecasts of the likely out-
come of the withdrawal of the huge stimulus to aggregate demand from war
Copyright © 2021. Harvard University Press. All rights reserved.
expenditure. Many predicted gloom and doom, even a depression of the or-
der of the 1930s. Keynes himself was more sanguine, at least about the imme-
diate postwar period, despite his sense that over the long-term investment
opportunities would dry up: he foresaw a boost to consumption demand
from the large holdings of cash, bank deposits, and government bonds on the
part of the public, coupled with the backlog of unmet needs and wants—for
housing, automobiles, and other consumer durables—that had been built up
during the war, along with the new programs of social insurance that made
saving for a rainy day or retirement less of a priority. Te discussion is inter-
esting in its own right, but the main lesson here is the extent the framework of
Te General Teory had permeated into the economist’s consciousness.12 Te
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 47
stage was set for the postwar ascendency of visions, theories, and models in-
spired by Te General Teory.
Te theoretical reaction to Keynes that culminated in the ascendency of
New Classical economics had its counterpart in economic policy. Te consen-
sus about the long run implies that demand management can afect only the
price level, the real economy being determined, as my younger colleagues in
Littauer averred on that fall day in 2008, by the real variables that operate on
aggregate supply. Even in the short run, where New Keynesians sought to
carve out a theoretical role for government intervention, it was argued that
practical limitations (see chapter 14, p. 538) made it impossible to deploy
countercyclical monetary policy efectively, and even more dif ficult to deploy
fscal policy. In any case, temporary infusions of demand through countercy-
clical monetary and fscal policy could have only a very limited impact by
virtue of their temporary nature.
Central banks around the world adopted the New Classical view to a
greater or lesser degree, with the European Central Bank the most infuential
voice for the new anti-Keynesian orthodoxy. Te ECB (n.d.) ofered the fol-
lowing rationale for making price stability the primary objective of monetary
policy:
Te natural role of monetary policy in the economy is to maintain price
stability. . . . Monetary policy can afect real activity only in the shorter
term. . . . But ultimately it can only infuence the price level in the economy
(emphasis in original).
Te ECB website amplifed:
Long-run neutrality of money
It is widely agreed that in the long run—afer all adjustments in the econ-
omy have worked through—a change in the quantity of money in the econ-
omy will be refected in a change in the general level of prices. But it will not
Copyright © 2021. Harvard University Press. All rights reserved.
induce permanent changes in real variables such as real output or unem-
ployment.
Tis general principle, referred to as “the long-run neutrality of money”,
underlies all standard macroeconomic thinking. Real income or the level
of employment are, in the long term, essentially determined by real fac-
tors, such as technology, population growth or the preferences of economic
agents.
Infation—a monetary phenomenon
In the long run a central bank can only contribute to raising the growth po-
tential of the economy by maintaining an environment of stable prices. It
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
48 Raising Keynes
cannot enhance economic growth by expanding the money supply or keep-
ing short-term interest rates at a level inconsistent with price stability. It can
only infuence the general level of prices.
Ultimately, infation is a monetary phenomenon. Prolonged periods of
high infation are typically associated with high monetary growth. While
other factors (such as variations in aggregate demand, technological changes
or commodity price shocks) can infuence price developments over shorter
horizons, over time their efects can be ofset by a change in monetary
policy.
Te ECB website made note of the sand in the wheels that may cause unem-
ployment but was unequivocal about the “hierarchy of objectives”:
In the actual implementation of monetary policy decisions aimed at main-
taining price stability, the Eurosystem should also take into account the
broader economic goals of the Community. In particular, given that mone-
tary policy can afect real activity in the shorter term, the ECB typically
should avoid generating excessive fuctuations in output and employment if
this is in line with the pursuit of its primary objective.
Post-2008 Europe is reaping the whirlwind: the continuing story about the
southern rim of the European economy is one of unemployment and stagna-
tion. It is true that following the accession of Mario Draghi, the ECB looked
for ways to stimulate the economy, but the constraints embodied in its found-
ing documents drastically impeded its eforts. And precious years were lost
before the ECB even attempted to intervene.
No greater tribute could have been paid to the power of economic theory
than the ECB’s mission statement quoted above.13 Samuelson had a point
when he observed, “Let those who will write the nation’s laws, if I can write its
textbooks” (widely cited in his obituary; see, e.g., Henderson 2009). However,
the textbook to which the ECB looked for guidance was likely to have been
Copyright © 2021. Harvard University Press. All rights reserved.
inspired more by Milton Friedman and Robert Lucas than by Keynes, or by
Samuelson for that matter.
Te chapters that follow are not an attempt to represent Keynes, though I
sometimes joke that I am channeling him. Nobody will ever know what
Keynes “really” meant, and in any case I agree with those who have argued
that Keynes’s intuition ran way ahead of his ability to set out logically tight
arguments. Tis is a book that Keynes might have written if he had had eighty
years to think about his ideas, to digest the criticisms and the changes that
were taking place as he was originally formulating his arguments, not to men-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1 • Introduction 49
tion the changes since—and if he had the tools available to us in 2020. In any
case, I am willing to assert that it is the book he should have written.
At some points I diverge from the arguments of Te General Teory. Te
complete title is Te General Teory of Employment, Interest and Money. Tere
is, I have suggested, not one theory of employment, but at least two. One—the
second-pass model—is designed to investigate and ultimately to demolish the
claim that a competitive capitalist economy is self-regulating, that competitive
capitalism lef to itself will provide jobs for all willing workers. Another—the
frst-pass model—is designed to investigate the possibilities and limitations of
using monetary policy to steer capitalism to the goal of full employment that
eludes it when lef to its own devices. Tis model sets the stage for a theory of
fscal policy—even if Te General Teory fails to provide the theory.
Tere is also a theory of interest, perhaps the most innovative of the novel-
ties of Keynes’s theory. Tis theory of interest, however, is incomplete at best;
moreover, its incompleteness contributes to the failure of Te General Teory
of Employment to carry the day in theoretical debate.
But there is no theory at all of money. What is implicitly assumed about
money is at odds with Keynes’s own stated position (in his earlier work, A
Treatise on Money), and is at the very least anachronistic if not altogether
wrong. And the absence of a theory of money is at least part of the reason why
Te General Teory of . . . Interest is incomplete (or wrong) and in turn why
Keynes’s theory of employment has had such rough sailing.
Te chapters that follow lay all this out.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 2 .
what were they thinking?
Economics Before The General Theory
With perfectly free competition among workpeople and labour
perfectly mobile . . . there will always be at work a strong tendency for
wage-rates to be so related to demand that everybody is employed.
Hence, in stable conditions every one will actually be employed. Te
implication is that such unemployment as exists at any time is due
wholly to the fact that changes in demand conditions are continually
taking place and that frictional resistances prevent the appropriate
wage adjustments from being made instantaneously.
—arthur pigou
In a perfectly fexible economy where money costs and prices quickly
adjust themselves to changed circumstances, there could be price
level fuctuations, proft fuctuations, or fuctuations in the real
income of labor, but there could not be substantial fuctuations of
employment or production. Our system rests upon the assumption of
the existence of price fexibility and cannot operate smoothly without
it. But in fact the price structure is shot through with rigidities, and
even, as in the case of tax rates and freight rates, with perverse fexi-
bilities.
—jacob viner
Copyright © 2021. Harvard University Press. All rights reserved.
It is sometimes asserted that mainstream economists had no theory of output
as a whole before Keynes. Indeed, at one point, Keynes himself appears to
have held this view (Keynes 1937a, p. 223). Te mainstream may have had the
wrong theory and lacked a clear exposition, but that is altogether diferent
from having no theory. For the mainstream a full-employment equilibrium
was the normal level of output.
Specialists on business cycles or on money and banking implicitly shared
the view that a full-employment equilibrium is the normal state of economic
activity around which business fuctuated, but they paid more attention to the
ups and downs of the economy around this normal state, which is to say, with
the causes of booms and the remedies for slumps. Rich and varied as these
50
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 51
literatures were, the question of how the normal, full-employment, level of
output is determined seems never to have been addressed! (See David Laidler
[1999] for an informative discussion of these ideas in the context of the the-
ory Keynes was to produce in Te General Teory. John Hicks [1967] provides
an insightful account of the history of monetary thought and practice).
Rather, the determination of the normal level of overall output emerged
from what has come to be called microeconomics, from the supply-demand
constructions based on Alfred Marshall’s partial equilibrium analysis, or for
the more adventurous, from the general equilibrium constructions of Léon
Walras. In microeconomics full employment was and still is the norm, the
equilibrium to which an economy gravitates, a position at which the economy
would remain if it were not disturbed by the factors that were at the heart of
the specialist literatures on cycles and money. Fluctuations were understood
to be temporary departures that need not interfere with our understanding of
the underlying tendency of the economy, at least of a perfectly competitive
economy, to provide jobs for all willing workers as well as to maximize profts
for business. Even when the possibility of deviations from full employment
was recognized, as in Keynes’s Treatise, the principal concern was with price
movements.
Keynes took Arthur Pigou’s Teory of Unemployment, published in 1933 in
the early stages of the long pregnancy leading to Te General Teory, as the
paradigm of mainstream (“classical” in Keynes’s terminology) theory. Or per-
haps it is more fair to say that Keynes creates a composite Pigou to stand for
the mainstream position, which, as Keynes himself acknowledges, is dif ficult
to pin down. But no matter. Whether or not the mainstream of Keynes’s day
held so clear a view of the determination of employment and output, there is
no question that the view Keynes attributed to Pigou is the prevailing main-
stream view in 2020. Look no further than your favorite textbook. If your text
is Mankiw’s Principles of Economics (eighth edition), check out chapter 18,
particularly fgure 4 on page 370. If you prefer Economics: Principles and Pol-
Copyright © 2021. Harvard University Press. All rights reserved.
icy by Baumol, Blinder, and Solow (fourteenth edition), read chapter 19, espe-
cially fgure 2 on page 397. Prefer Frank and Bernanke, Principles of Econom-
ics (third edition)? See chapter 14, especially the “recap” at the bottom of page
404. Finally, try Krugman and Wells, Economics (ffh edition), chapter 19,
fgure 19-6 on page 544.
Each of these texts provides its version of Figure 2.3(a) below, which is a
two-schedule summary of Pigou’s theory, one schedule depicting the supply
of goods that producers fnd proftable to provide at various prices, the other
the amount of labor that households fnd advantageous to provide at diferent
wages.1 “Prices” and “wages” here mean real magnitudes, prices relative to
wages, and wages relative to prices. Although Keynes (Te General Teory,
chap. 2) disagreed with the idea that labor supply is a function of the real
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
52 Raising Keynes
wage (or price), he had no quarrel with the idea that producers maximize
profts by equating real marginal cost to the real price—in Keynes’s terminol-
ogy the price in wage units—or fipping the equation, equating the marginal
product of labor to the real wage.
Strong assumptions are required to aggregate the myriad goods and ser
vices that are produced into a single good, “output.” But if we are willing to
swallow other assumptions that the mainstream makes in order to tell its
story about how the market guarantees a job for every willing worker—or for
that matter Keynes’s story about how it doesn’t—assumptions about aggrega-
tion shouldn’t deter us. To keep the story as simple as possible, for the most
part we ignore land and other primary resources, so that output can be taken
to be a function of the stock of capital and the fow of labor into the produc-
tive apparatus.
Te goods-supply (GS) schedule is the economy-wide real marginal-cost
schedule, as shown below in Figure 2.1, in which the vertical axis is the price
level relative to money wages, P/W, and the horizontal axis measures real
output, Y. Marginal cost is simply (W × L)/Y in terms of money, or L/Y
in real terms, that is, in terms of labor per unit of output; thus the
proft-maximizing level of output is given by P = (W × L)/Y, or P/W =
L/Y.
Te second schedule is the labor-supply (LS) schedule, determined in
P
W
Supply of goods
Copyright © 2021. Harvard University Press. All rights reserved.
100 200 300 400 500 600 Y
Figure 2.1 Supply of goods.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 53
mainstream theory by individuals (or households) optimizing among alterna-
tive combinations of labor and leisure. Te vertical axis measures the real
wage rate, W/P, and the horizontal axis measures the level of employment, L.
Te LS schedule slopes upward or “backward” according to which of two con-
tradictory efects dominates, the substitution efect of higher wages making
leisure more costly or the income efect of higher wages making leisure more
afordable. When these two efects are exactly in balance, the LS schedule is
vertical, defning a single level of full employment. Otherwise, every point on
the LS schedule corresponds to a diferent level of “full” employment, since
this schedule is defned by the number of hours (or days or weeks) workers
wish to work at various levels of (real) wages. Figure 2.2 depicts a situation in
which the substitution efect dominates.
Figure 2.3(a) transposes the GS schedule from the space of Y × P/W to the
space of L × W/P.
Te GS schedule becomes a demand-for-labor schedule because, in the
short run, the capital stock is fxed and output varies with the level of employ-
ment: proft maximization can be equally well described in terms of increas-
ing output to the point that the real marginal cost is equal to the real price of
goods (P/W = L/Y) or adding labor to the point that the marginal produc-
tivity of labor falls to the real wage (W/P = Y/L). Te intersection of the
demand and supply schedules determines both the equilibrium real wage and
the equilibrium level of employment.
In Figure 2.3(a) the determination of output is described in terms of an
equilibrium level of employment that both provides full employment (the
optimal amount of work in the sense of balancing workers’ desire for the
W
P
1
Copyright © 2021. Harvard University Press. All rights reserved.
Supply of labor
3/4
1/2
1/4
1,000 2,000 3,000 4,000 5,000 6,000 7,000 L
Figure 2.2 Supply of labor.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
54 Raising Keynes
W P
P W
1 Supply of goods
Demand for labor
5
3/4
Supply of labor 4 E
Supply of
1/2 3 labor
2
1/4 E
1
2,000 4,000 6,000 L 100 200 300 400 500 600 Y
(a) In the space of employment × (b) In the space of output ×
real wage real price
Figure 2.3 Equilibrium in terms of employment and output, I.
goods that their wages can buy and the leisure they enjoy) and maximizes
profts for producers.
Te same result is described in Figure 2.3(b) in terms of the level of output.
Just as Figure 2.3(a) transposes the GS schedule into a labor-demand sched-
ule, Figure 2.3(b) transposes the supply of labor into a supply of output, the
amount of output that workers optimally balancing leisure and labor would
produce. Equilibrium exists at the real price for which the two levels of sup-
ply, the supply of output that is optimal for workers (labeled “Supply of labor”
even though it is output that is supplied) and the supply of output that maxi-
mizes profts for producers (labeled “Supply of goods”), coincide.
Tree salient results emerge from this simple construction, the results that
distinguished the mainstream from what Keynes would propose in Te Gen-
eral Teory. First of all, there is no such thing as involuntary unemployment
Copyright © 2021. Harvard University Press. All rights reserved.
in equilibrium. Second, there is no role for aggregate demand. As can be seen
by comparing Figures 2.3(a) and 2.3(b), the demand for labor refects only the
proft-maximizing supply of goods; it is derived from the supply conditions of
producers, not from demand for consumption or investment. Finally, the
mainstream results refect the classical dichotomy, according to which only
real variables matter to the determination of equilibrium output and income.
Let us look at these results in order. Te mainstream equilibrium, whether
characterized in terms of employment or output, is at a point at which every
worker can work as much as she wishes at the going real wage. She may
choose to work more or less—this is a voluntary decision—but nobody is out
of work because she can’t fnd a job. Only if the real wage is too high, which is
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 55
to say the real price is too low, say W/P = 1 (and P/W = 1), can we speak of
involuntary unemployment. In this case, the supply of labor would exceed the
demand, the proft-maximizing supply of goods would fall short of the supply
of goods that workers would choose. If employers were to prevail in such a
situation, producing only as much output as maximized profts, there would
not be a job for every willing worker. Tis is shown in Figure 2.4, where the
amount of unemployment at F in panel (a) is approximately 4,000 (= 8,000 −
4,000) and the amount of lost output at F in panel (b) is 200 (= 650 − 450).
Not only is the cause of unemployment—too high real wages—clear from the
diagram, so is the cure: eliminate the obstacles that prevent the real wage
from adjusting proft maximization and worker optimization to each other.
Trade unions were always a prime suspect in the search for rigidities that
prevented the adjustment mandated by mainstream theory. But the logic did
not compel u nion bashing. As chapter 1 noted, any systemic rigidity that gets
in the way of price and wage adjustment, any element of monopoly, be it on
the side of capital or the side of labor, could be the culprit.
Te second implication of the theory is that there is no role for aggre-
gate demand in determining the level of output and employment. Tere is
a demand schedule in the lef-hand panels of Figures 2.3 and 2.4, but it de-
rives from producer optimization, like the GS schedule in the right-hand
panels, not the “fnal” demand that comes from the desires of consumers,
businessmen, governments, and foreigners for consumption, investment,
public spending, and exports.
Te third lesson of the mainstream story—really an assumption rather than
W P
P W
Supply of goods
1 F
Demand for labor
5
3/4
Copyright © 2021. Harvard University Press. All rights reserved.
4 E
Supply of labor
1/2 Supply of 3
labor
2
1/4 E
1 F
2,000 4,000 6,000 L 100 200 300 400 500 600 Y
(a) In the space of employment × (b) In the space of output ×
real wage real price
Figure 2.4 Equilibrium in terms of employment and output, II.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
56 Raising Keynes
a theorem—is that equilibrium depends on real variables only. Except for a
short period in which adjustment takes place, a higher or lower nominal price
level leaves the volume of employment and output unafected.
Jacob Viner and the Depression: Diagnosis and Cure in
Search of an Analytic Framework
Chapter 1 noted that the absence of a theory hinging on aggregate demand
did not prevent arguments for defcit spending or other policies. How did
defcit spending ft into a purely supply-side framework? Te answer is “not
well.” In fact, a purely supply-side framework deprived these arguments of
a coherent theoretical basis. Jacob Viner’s policy interventions in the early
1930s are a case in point.
Viner, a stalwart of the University of Chicago economics faculty and a
teacher of both Paul Samuelson and Milton Friedman, became a leading critic
of Keynes’s General Teory. Indeed, he was the one participant in a sympo-
sium published by the Quarterly Journal of Economics in the fall of 1936 to
whom Keynes replied at length and the one whose intervention Keynes ac-
knowledged to be “the most important” (Keynes 1937a, p. 210). Most percep-
tive in his analysis of individual trees, Viner never could see Keynes’s forest.
Viner’s own espousal of defcit spending, even the intellectual apparatus be-
hind his prescription, had much in common with Te General Teory—ex-
cept for lacking a theoretical framework in which defcits made sense.
Like the majority of the profession, Viner understood the Depression in
terms of an imbalance between costs and prices. Without a framework that
allowed a separate and distinct role for aggregate demand, Viner could think
only in terms of supply-side remedies. So, as long as the United States was on
the gold standard, Viner, lecturing at the University of Minnesota in February
1933, could envisage a solution only in terms of what he called “balanced de-
fation” (1933a), which is to say, getting costs to fall to the same degree as
Copyright © 2021. Harvard University Press. All rights reserved.
prices. Hardly a month had passed, however, before Franklin Roosevelt, in
one of his frst acts as president, had efectively taken the United States of the
gold standard, which allowed a new remedy, infation, to accomplish the
supply-side goal of restoring the price:cost balance. Some months later, in a
paper presented at the Institute of Public Afairs at the University of Georgia
(1933b), Viner expanded a remedy ofered as a hypothetical aside in February
(1933a, pp. 25–27) into the centerpiece of his policy prescription: the goal
remained to fnd the right balance between prices and costs, but the solution
now lay in expansionary monetary policy.
Figure 2.4 can be deployed to clarify Viner’s basic argument. Imagine what
happens when the economy is out of equilibrium, as in Figure 2.5, and there
are no obstacles to adjustment of the wage rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 57
W P
P W Supply of goods
1 F
Demand for labor
5
3/4
4 E
1/2 A Supply of 3 Supply of labor
labor
2 A
1/4 E
1 F
2,000 4,000 6,000 L 100 200 300 400 500 600 Y
(a) In the space of employment × (b) In the space of output ×
real wage real price
Figure 2.5 Equilibrium in terms of employment and output, III.
Suppose the economy is displaced from the initial equilibrium E, pictured
in the right-hand panel as <450, 4>, to a new position F = <450, 1>. For pres-
ent purposes it doesn’t matter what causes the displacement; we can assume
either a monetary or a real shock that causes the price level to fall while leav-
ing the money wage unchanged. In either case, the disequilibrium refects
Viner’s imbalance between prices and costs: we may suppose that prices have
fallen by a factor of four while wages remain unchanged. At F marginal cost
exceeds price, and the marginal product of labor falls short of the real wage.
What happens now?
One story goes like this: producers respond to the shortfall of (real) price
relative to (real) marginal cost by reducing output. It no longer pays to pro-
duce at the level 450 because reducing output will lower costs more than it
does returns. At the same time unemployed workers, competing for jobs,
Copyright © 2021. Harvard University Press. All rights reserved.
drive money wages down and thus the real price up. Together, these two argu-
ments imply a trajectory that initially moves the economy from F to the
northwest, as the arrow indicates. (Te equivalent trajectory in the lef-hand
panel is to the southwest, the change in employment being governed by a
comparison of the marginal product of labor with the real wage and the
change in the real wage being governed by the level of unemployment.)
We can identify two phases in the trajectory of output and real price. First,
a depression phase in which output (and employment) fall even as the real
price begins to recover; second, a period of recovery of output once the real
price has risen to the point that it exceeds marginal cost. Te economy bot-
toms out at A, which marks the transition from the frst to the second phase.
Recovery is complete when the original equilibrium E—characterized by both
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
58 Raising Keynes
roft maximization and full employment—is reached. Observe that equilib-
p
rium money prices and wages are four times lower at the end of the story than
at the beginning. Prices were assumed to fall to one-quarter of their original
level as the economy moved from E to F, and the return to E is accomplished
by reducing the wage rate by the same percentage amount. Tis is the essence
of “balanced defation.”2
Viner has suf ficient faith in the market to trust that equilibrium will be re-
stored sooner or later, but the Depression has taught him that “later” can be
too long a time:
I still believe that even if nothing is done costs will eventually probably fall
suf ficiently to restore a p
roft margin. But I am becoming more and more
convinced that there is serious risk that they won’t do so quickly enough to
forestall wholesale economic collapse, if reliance is placed wholly on the
self-acting processes of recovery. (1933a, p. 10)
Speeding up a process that a competitive economy would follow anyway
thus spares people pain and suffering. In February 1933, Viner looked to
Washington to take action to bring about balanced defation—for instance,
enforcing antitrust legislation to force down the prices of raw materials such
as steel, which Viner sees as being maintained by “quasi-monopolistic organi
zation” (1933a, p. 14). He also suggests that even though the pressures of the
market have corrected President Hoover’s misguided attempts to maintain
money wages at pre-Depression levels, some wages—particularly the wages of
railway workers and government employees—must come down in order to
reduce freight charges and taxes.3
Once the dollar was no longer linked to gold, a new possibility for jump-
starting the recovery process exists. Since the key to moving the economy is
the ratio of real prices to real marginal costs, instead of relying on the “natu-
ral” forces of the market to reduce wages, prices can be artifcially raised to
accomplish the rebalancing necessary for industry to reverse course:
Copyright © 2021. Harvard University Press. All rights reserved.
Now that we are of [the gold standard], we have a freedom which, if prop-
erly exercised, gives us ground for hope that with the aid of monetary mea
sures we may be able to end the depression within a reasonable length of
time.
Te theory behind infation as a remedy for the depression consists sim-
ply in the proposition that if you can get prices up, and if costs do not go up
simultaneously or do not go up in the same proportion, p roft margins will
reappear for business and business men will then be willing to use such
spendable funds as they still own, and will be willing to ask and be success-
ful in persuading the bankers to create new spendable funds, to be used in
buying raw materials, hiring labor, and so on. Tis process of putting pur-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 59
chasing power into circulation will operate to increase the total volume of
money income, prices will rise, but in less proportion than business expen-
ditures, and the result will be an increased physical output, which will mean
an increased employment of the productive factors. (1933b, p. 123)
Among the measures Viner advocates is defcit spending on the part of the
federal government. In a passage that eerily anticipates Keynes, Viner writes,
If the government were to employ men to dig ditches and fll them up again,
there would be nothing to show aferwards. But, nevertheless, even these
expenditures would be an indirect contribution to business recovery. Teir
major importance would not be in the public works or the unemployment
relief which immediately resulted, but in the possibility of hope that a sub-
stantial expenditure would act as a priming of the business pump, would
encourage business men by increased sales, make them more optimistic,
lead them to increase the number of their employees, and so on. (1933b,
p. 130)
Tere is a step missing in the argument. How exactly is the purchasing
power created by public works going to stimulate private-sector production
and sales? Te answer is easy with aggregate demand in the picture as a force
driving output, mysterious at best without aggregate demand. To assume that
the infow of money from expanding government payrolls directly increases
private-sector sales and employment undercuts the premise of Viner’s argu-
ment for balanced defation or infation—namely, that the underlying cause of
the Depression is the relationship of prices to costs. Viner can’t have it both
ways—frst, arguing that the problem lies in an unfavorable price:cost ratio
and second, arguing that “substantial [government] expenditure” will prime
the business pump, leading to an expansion of sales and employment—at
least not without an intermediate step in which the infow of government
money frst raises prices without raising costs, producers then responding to
a more favorable price:cost structure by increasing output. Such an interme-
Copyright © 2021. Harvard University Press. All rights reserved.
diate step would, however, tacitly admit aggregate demand into the picture
through the back door.
Viner is trapped by a purely supply-side perspective. Lacking any way of
integrating defcit spending into a supply-side story, Viner’s argument for
defcit spending failed to gain traction. Tree years later, Keynes makes virtu-
ally the same argument—and it makes perfect sense within the framework of
Te General Teory:
If the Treasury were to fll old bottles with banknotes, bury them at suitable
depths in disused coalmines which are then flled up to the surface with
town rubbish, and leave it to private enterprise on well-tried principles of
laissez-faire to dig the notes up again . . . , there need be no unemployment
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
60 Raising Keynes
and, with the help of the repercussions, the real income of the community,
and its capital wealth also, would probably become a good deal greater than
it actually is. It would, indeed, be more sensible to build houses and the like;
but if there are political and practical dif
ficulties in the way of this, the above
would be better than nothing. (Te General Teory, p. 129)
Making Aggregate Demand Disappear
Robert Malthus, nineteenth-century population theorist and interlocutor of
David Ricardo, was rare among economists before Keynes in maintaining the
importance of aggregate demand. Te economics profession has ever since
worked very hard to divest itself of Malthus’s insight. Tere are in fact two
ways of making demand go away, but one of the two, which depends on as-
suming full employment (see chapter 4, esp. Figure 4.9), was a reaction to
Keynes. Te pre-Keynesian way is to argue that the case of output as a whole
difers from the market for a single good because supply creates its own de-
mand: aggregate demand is thus identically equal to what is supplied.
Te kernel of truth in the argument lies in an elementary fact of national
income and product accounting: income and output are two ways of looking
at the same thing. We can see this most easily in the simple case in which no
material inputs or services are purchased from upstream suppliers. In this
simplest of cases, production involves the sweat of the worker’s brow (and the
entrepreneur’s) in combination with the services of machines and the gifs of
nature embodied in land and other natural resources. When these “factors of
production” produce a dollar’s worth of output, a dollar’s worth of income is
also generated—wages to workers, rents to landlords, interest to bondholders
and banks, and proft to the entrepreneur. (Proft can be looked at as the bal-
ancing item that ensures the sum of income generated to the owners of land,
labor, and capital will just equal the value of the product.)
So output—supply—is equal to income. How does this ensure that the de-
Copyright © 2021. Harvard University Press. All rights reserved.
mand for output is equal to supply, which is to say that the expenditure on
output is equal to the income generated by its production? Te equality be-
tween the output of apples and the income generated in the production of
these apples does not ensure that the demand for apples necessarily equals the
supply; orchard workers or owners may choose to spend their incomes on
other goods. But looking at output as a whole, this imbalance does not arise:
if workers choose all of a sudden to spend less on apples, then they will spend
more on something else, maybe pears. Or bicycles. Or laptop computers. In
total, demand must equal income, and therefore output—and therefore can
be ignored.
Tere is a dif ficulty with this argument. What if agents wish to save some of
their income, for a rainy day, for retirement, or to make a major purchase
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 61
(like a house) at some point in the future? Isn’t saving a diversion from spend-
ing? Keynes seized upon what he perceived to be the weak link in the rela-
tionship between income and expenditure:
An act of individual saving means—so to speak—a decision not to have din-
ner to-day. But it does not necessitate a decision to have dinner or to buy a
pair of boots a week hence or a year hence or to consume any specifed thing
at any specifed date. Tus it depresses the business of preparing to-day’s
dinner without stimulating the business of making ready for some future act
of consumption. (Te General Teory, p. 210)
Tis sounds plausible, but mainstream economics had long since taken ac-
count of Keynes’s objection and had posited a mechanism for transforming
saving—abstaining from consumption—into investment—additions to the
stock of goods (machines, buildings, infrastructure) used to produce goods
and services in the future. Indeed, Adam Smith may have been the original
source of the argument that a general defciency of spending, or to see matters
from the other side, a general glut of production, was impossible in a market
system. As Smith put it,
Whatever a person saves from his revenue he adds to his capital, and ei-
ther employs it himself in maintaining an additional number of productive
hands, or enables some other person to do so, by lending it to him for an
interest, that is, for a share of the profts. As the capital of an individual can
be increased only by what he saves from his annual revenue or his annual
gains, so the capital of a society, which is the same with that of all the indi-
viduals who compose it, can be increased only in the same manner. . . . What
is annually saved is as regularly consumed as what is annually spent, and
nearly in the same time too. . . . Tat portion which [an individual] annually
saves . . . is consumed by labourers, manufacturers, and artifcers, who re-
produce with a p roft the value of their annual consumption. (1937 [1776],
Copyright © 2021. Harvard University Press. All rights reserved.
p. 321)
Keynes associated the view that supply creates its own demand not with
Smith but with David Ricardo, the great English economist of the frst half of
the nineteenth century, and with Jean-Baptiste Say, the nineteenth-century
French economist, whose eponymous Law embodied the conventional wis-
dom.4 A clear statement of this view, due to John Stuart Mill, is cited by
Keynes in Te General Teory (p. 18):
[If we could] suddenly double the productive powers of the country, we
should double the supply of commodities in every market; but we should, by
the same stroke, double the purchasing power. Everybody would bring a
double demand as well as a supply: everybody would be able to buy twice as
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
62 Raising Keynes
much, because everyone would have twice as much to ofer in exchange.
(Mill 1909 [1848], bk. 3, chap. 14, para. 2)5
Tis view was dominant as Keynes’s General Teory was taking shape. In
1933, a journeyman economist, Myron Watkins, summed up conventional
economic wisdom in the prestigious Quarterly Journal of Economics:
Te whole joint product of industry in any period is the same as the aggre-
gate income of the community during that period; it cannot be more and it
cannot be less. Te aggregate income of the community represents the total
available purchasing power of the community, nothing more and nothing
less; . . . an addition to the community’s stock of capital assets, through sav-
ings from whatever type of current income derived and in whatever volume
efected, constitutes a demand for a corresponding part of current produc-
tion. It follows that the total available purchasing power of the capitalistic
community must be exactly equal to the joint product of industry, however
swifly the latter may be increased and however inequitably it may be distrib-
uted . . .
[T]he erroneous assumption that production and consumption must
somehow be kept “in balance,” . . . rests, in turn, upon the naïve belief that
income which is not “consumed,” but “saved,” does not constitute a demand
for the current output of industry. More puerile nonsense than this would be
hard to imagine, and were it not for the frequency and volubility with which
such ideas are put forward, even occasionally—alas!—by economists with a
respectable reputation, . . . the space of a professional journal would not
need to be encumbered with their refutation. (1933, pp. 523–524)
Say’s Law is summarized in Figure 2.6. Output and expenditure are defned
in real terms and are independent of prices and wages, hence independent of
the real price. Te result is that, defning the level of aggregate demand as the
level of output for which expenditure and output are equal, the entire shaded
Copyright © 2021. Harvard University Press. All rights reserved.
area of Figure 2.7 represents the AD “schedule,” combinations of real price
and output for which expenditure and output are equal. According to Figure
2.7, output and expenditure are equal at every level of output, for every level of
the real price. Since this equality holds everywhere, it necessarily holds when
production is at full-employment levels of output or at proft-maximizing
levels of output. Tat is, aggregate demand overlaps with the LS schedule and
the GS schedule. Tis is shown in Figure 2.8. Te result is paradoxical: since
every point in Figure 2.7 is a point on the AD schedule, the AD schedule is
superfuous and can be ignored when it c omes to characterizing equilibrium.
How do Smith, Ricardo, Say, and Mill, not to mention Dr. Watkins, know
this? What assures that for every individual who decides to go without dinner
today there will be another individual who will step forward to use the re-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 63
Expenditure (E)
500
400
300 The expenditure function
according to Say’s Law
200
100
45°
100 200 300 400 500 Income = Output (Y)
Figure 2.6 Say’s Law: expenditure = income at every level of output.
P
W
Aggregate demand
Copyright © 2021. Harvard University Press. All rights reserved.
Y
Figure 2.7 Say’s Law: output = expenditure.
sources thus freed to fashion kitchen utensils for tomorrow’s dinner, so that
expenditure is always equal to income? How do they know that If Millie saves
one dollar of her income and thus breaks the chain of income and spending,
the link is restored by Mollie, who borrows the dollar to add to the capital
stock, in short, that the saving done by all the Millies is spent by all the Mol-
lies?
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
64 Raising Keynes
P P
W W
Aggregate demand
Supply of goods
Supply of labor
Aggregate demand
Y Y
(a) Aggregate demand = Goods supply (b) Aggregate demand = Labor supply
Figure 2.8 Aggregate demand, goods supply, and labor supply.
Te key, or so most “respectable” economists thought, was that the interest
rate guaranteed equilibrium between the demand for investment and the sup-
ply of saving—not that the past tense is appropriate: in the words of Robert
Frank and Ben Bernanke,6
Desired saving is equated with desired investment through adjustments in
the real interest rate, which functions as the “price” of saving. Te move-
ments of the real interest rate clear the market for saving in much the same
way that the price of apples clears the market for apples. (2007, p. 639)
Frank and Bernanke are hardly alone. Larry Summers (2016) explains why
low interest rates are evidence for a lack of investment demand in exactly the
same manner:7
Just as the price of wheat adjusts to balance the supply of and demand for
wheat, it is natural to suppose that interest rates—the price of money—ad-
just to balance the supply of savings and the demand for investment in an
economy. (p. 3)
Copyright © 2021. Harvard University Press. All rights reserved.
Tis may be the natural supposition, at least for a mainstream economist, but
not for Keynes.
Equilibrium vs. Equilibrating: Prices
and the Price Mechanism
Te problem is confusion between an equilibrium price and an equilibrating
price mechanism. To rely on the saving-investment equilibrium requires more
than the supposition that investment demand and saving supply are equal.
Tis argument also requires that if either the investment-demand schedule or
the saving-supply schedule shifs, then (1) the shif has no impact on the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 65
other schedule, and (2) the interest rate adjusts the levels of investment de-
mand and supply saving to a new equilibrium. Te basic diagram is given
below in Figure 2.9. Suppose we start out at a point with the rate of interest
equal to 0 and the level of investment and saving equal to I0. Tere is nothing
in Figure 2.9 with which Keynes would have any reason to take issue. For
Keynes, as well as for the lineage of Smith, the demand for investment de-
pends on the rate of interest, and while Keynes did not emphasize the infu
ence of the interest rate on the supply of saving, there is nothing in Te Gen-
eral Teory that rules out a role for the interest rate as one of the determinants
of saving. Indeed, the only diagram in the book is one that relates both saving
and investment to the rate of interest, introduced on page 180 precisely to
make the present point!8
Te diference between Keynes and Smith is not about the interest rate
characterizing an equilibrium between saving supply and investment de-
mand. It is about the interest rate as a mechanism for bringing about equilib-
rium. Suppose there is a shock to the economy that moves the investment-
demand schedule downward, as in Figure 2.10. (Imagine it is September 2008,
and Lehman Brothers is going under; entrepreneurs worry about future eco-
nomic prospects, and those who remain optimistic can’t get credit for their
projects.) How does the mainstream, from Smith to Knut Wicksell to Ber-
nanke and Summers, see the consequences?
Interest rate ( )
Supply of saving
Copyright © 2021. Harvard University Press. All rights reserved.
0
Investment demand
I0 Investment, Saving
Figure 2.9 Investment demand and saving supply, I.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
66 Raising Keynes
Interest rate ( )
Supply of saving
0
1
Investment demand
I1 I0 Investment, Saving
Figure 2.10 Investment demand and saving supply, II.
In the mainstream view, the existing interest rate, 0, will no longer clear
the market, but no matter, except for those troublesome rigidities that might
rear their heads in the short run. Although the composition of output will
change—there will be more consumption and less saving and investment—
there is no overall, macro, impact on the economy. Because the supply of sav-
ing at 0 exceeds the demand for investment, the interest rate will fall. At 1
where the new investment-demand schedule, the dashed line, intersects the
saving-supply schedule, saving will again equal investment at the new (lower)
level I1. Te rate of interest adjusts the amount of saving in the economy as
Copyright © 2021. Harvard University Press. All rights reserved.
well as the quantity of investment so that they are once again in equilibrium,
the price mechanism working as microeconomics teaches that it operates in
the apple market. In the end, there is less investment (and saving) but more
consumption. Aggregate demand is unafected.
We shall explore Keynes’s own model presently. For now it is enough to ob-
serve that the Keynesian mechanism that brings investment demand and sav-
ing supply into line is the level of production itself. A shif in the investment-
demand schedule leads to a shif in the saving-supply schedule: the two
schedules are not independent. And (to a frst approximation) there is no
impact on the rate of interest. Keynes’s argument is that the interest rate
mechanism equilibrates the so-called money market (shorthand for inter-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 67
twined markets in which the stocks of fnancial assets are traded). Tis mecha-
nism is thus not available to equilibrate fows of investment and saving. Yes, in
equilibrium, for Keynes as well as for the mainstream, the quantity of invest-
ment demanded and the quantity of saving forthcoming are equal, so Say’s
Law holds in the sense that expenditure is equal to income, but this is not as
an identity that is true for any level of output. Te picture is in Figure 2.11.
Tere are two points at which saving and investment are equal: the old equi-
librium between desired saving and investment at I0 and the new equilibrium
at I2, the intersection of the displaced saving-supply schedule, the red-dashed
line, and the new investment-demand schedule. A key diference from the
mainstream is that the new intersection of investment and saving schedules
does not correspond to a higher level of consumption; the level of consump-
tion falls, along with the level of output and income.
From the point of view of Te General Teory, relying on the rate of interest
to equalize saving and investment is not like relying on the price of apples to
equalize demand and supply in the apple market, the analogy proposed by
Bernanke and Frank, but like relying on the price of pears to do the job. If
there is a shif in the apple-supply schedule, then within the space of apple
demand and supply versus apple price, we rely on a movement along the
apple-demand schedule to restore the balance between demand and supply.
Interest rate ( )
Supply of saving
0
Copyright © 2021. Harvard University Press. All rights reserved.
Investment demand
I2 I0 Investment, Saving
Figure 2.11 Investment demand and saving supply, III.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
68 Raising Keynes
But if we focus on the price of pears, adjustment comes about mainly through
a shif in the apple-demand schedule rather than through the price mecha-
nism. Similarly, the adjustment of saving to a shif in the investment-demand
schedule comes about through a shif in the saving schedule rather than
movement along the original saving schedule.
Formally, it is a matter of adjustment speeds: for the mainstream the inter-
est rate responds quickly to a diference in the demand for investment and the
supply of saving, so there is no reason to suppose that output need respond
and that the saving schedule need shif when the investment schedule shifs,
as it does in Figure 2.11. For Keynes, however, the interest rate, busy equili-
brating demands and supplies in markets for fnancial assets, moves at a gla-
cial rate in response to discrepancies between desired investment and saving;
output does the heavy lifing in bringing investment and saving into line. For
Keynes,
Te rate of interest is not the “price” which brings into equilibrium the de-
mand for resources to invest with the readiness to abstain from present
consumption. It is the “price” which equilibrates the desire to hold wealth in
the form of cash with the available quantity of cash. (Te General Teory,
p. 167)9
Tis is as much an empirical as a theoretical proposition. As I argue more
fully in chapter 11, one can imagine a world in which the interest rate adjusts
desired saving and investment as the mainstream argues. Tis is a world
where there is no market for fnancial assets apart from current fows, re
fected in so-called “on-the-run” bonds, because bonds are always held to ma
turity. (Viner [1936] suggested this possibility in his critique of Te General
Teory in the Quarterly Journal of Economics symposium that I alluded to
earlier in this chapter.) In this case, the interest rate would have nothing else
to do but adjust desired investment and saving, at least if we assume that
bonds are issued only to fnance investment. Say’s Law would hold, and ag-
Copyright © 2021. Harvard University Press. All rights reserved.
gregate demand would be irrelevant for the real economy. No Keynes, and no
Raising Keynes.
But this is not the world that we inhabit, and it has not been since the dawn
of capitalism. Te relative magnitudes of asset-market turnover and saving
and investment make the mainstream view untenable: the average daily vol-
ume of domestic bonds in the United States was over $750 billion in 2017
(Securities Industry and Financial Markets Association 2018), whereas the
gross annual fow of investment (and saving) was just over $3 trillion. Assum-
ing 250 trading days in the year, the ratio of annual volume of bond trading to
investment is approximately 60.10
In the passage quoted above, Keynes in efect assumes the limiting case, a
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 69
zero speed of adjustment of the interest rate to fow discrepancies between
desired saving and investment. A reasonable simplifying assumption, I would
argue, in view of the relative magnitudes of the stocks and the fows.11
In 2020, it would be hard to fnd a defender of the doctrine that supply cre-
ates its own demand in exactly the sense of Mill.12 I expect that Bernanke and
Summers would prefer Knut Wicksell as their patron saint, situating their ar-
gument in a world in which a “natural” rate of interest emerges from the in-
terplay of the demand for investment and the supply of saving (Wicksell
1958). But Wicksell’s logic is essentially no diferent from Say’s and Mill’s.
In the fnal analysis, for all the attention Say’s Law received in Te Gen-
eral Teory and in the post–General Teory literature, it turns out to be a bit
of a red herring in the sense that it is neither necessary nor suf ficient for a
full-employment equilibrium. It is not necessary because Modigliani’s model
shows how full employment is possible in the framework of Te General Te-
ory, without Say’s Law; not suf ficient because of the indeterminacy of equilib-
rium with which it leaves us, as we shall see in just a moment.
First, note that Say’s Law is a silent partner in the adjustment story repre-
sented in Figure 2.5. Nowhere did the story rely on Say’s Law for determining
the level of output! Instead, the story, at least the story I told on Viner’s behalf,
has the adjustment of output depend on the ratio of price to marginal cost,
and the adjustment of money wages depend on the level of unemployment.
Because there is nothing in the story about the adjustment of nominal prices,
the adjustment of money wages is also the adjustment of real wages and real
prices.
Tis story gets the economy to a full-employment, proft-maximizing equi-
librium without specifcally invoking Say’s Law. But Say is nonetheless having
his say, by eliminating aggregate demand from any role in moving the econ-
omy. To see this, suppose we try a diferent story, one in which business
women adjust prices rather than output. Operating in the framework of Fig
ure 2.5(b), imagine that when the economy is to the lef of the GS schedule,
Copyright © 2021. Harvard University Press. All rights reserved.
businesses reduce prices to stimulate sales. Conversely, when to the right of
the GS schedule, businesses raise prices. Tis would in efect make the actions
of business symmetric with the actions of labor, who respond to unemploy-
ment by reducing money wages.
Te problem with this story is that we are lef with no story for output ad-
justment; indeed, Say’s Law precludes any story, since wherever output hap-
pens to be, it will remain. Indeed, Say’s Law suggests that recovery could stall
out somewhere before full employment is reached. Suppose the economy
happens to fnd itself at B in Figure 2.12. Being to the lef of both the GS and
LS schedules, both prices and wages are falling. If it so happens that prices
and wages are falling at the same percentage rate, the real price will not move.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
70 Raising Keynes
P
W
Supply of goods
4 E
Supply of labor
3 B
A
2
1 F
100 200 300 400 500 600 Y
Figure 2.12 Incomplete adjustment: at B, prices and wages fall at the
same rate.
At the same time, Say’s Law guarantees that output won’t move! So here, un-
less we bring in aggregate demand as a separate determinant of output, we
have no mechanism whatsoever for moving the economy. Say’s Law efec-
tively narrows the adjustment process down to a single algorithm, the one I
have attributed to Viner.
Copyright © 2021. Harvard University Press. All rights reserved.
Reading Keynes through Mainstream Lenses: The Eclipse
If the 1950s and 1960s were the time in which Te General Teory was re-
duced from a general to a special theory, the 1970s were the time in which the
frontal attack on Keynes emerged from the shadows. Attacks on Keynes were
nothing new—witness the inroads of Milton Friedman’s “permanent income
hypothesis” (1957) and Franco Modigliani’s “life cycle hypothesis” (Modi
gliani and Brumberg 1954; Ando and Modigliani 1963) into the naïve version
of Keynes’s consumption function that made income the unique determinant
of consumption, about which more in chapter 9—but the success of the at-
tacks of the 1970s was unprecedented.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 71
Te ground was prepared by the overreliance of 1960s Keynesians on a
simple version of Te General Teory in which demand alone determined
(real) output and income, and the role of supply considerations was limited to
determining the price level. Tis view found empirical support in the work of
A. W. Phillips (1958), whose eponymous curve documented an inverse rela-
tionship between the level of unemployment and the rate of infation. An
oversimplifed reading of Te General Teory transformed Phillip’s empirical
regularity into a causal relationship that eliminated the supply side from the
picture.
Te idea of a trade-of between unemployment and price stability worked
tolerably well for more than three decades afer the publication of Te General
Teory, but only because this was a period in which aggregate demand was
relatively volatile and supply relatively stable. It was once again Milton Fried-
man who, in his presidential address to the American Economic Association
in 1968, planted doubts about the plausibility of a stable Phillips curve in the
presence of persistent price infation. But the mortal blow to the Phillips
curve as a causal relationship was dealt by events, not by theory. Te 1970s
saw supply-side convulsions, mainly oil-price shocks, that led to “stagfa-
tion”—higher prices coupled with higher unemployment—the opposite of the
Phillips-curve relation. It became clear that the economy could not be mod-
eled in terms of demand varying against the backdrop of fxed supply.
By itself, this need not have led to an eclipse of Keynes’s theory. To be
sure, supply had to be given its due, and the shif to the aggregate-demand/
aggregate-supply framework dates from this period, even though its frst use
goes back at least to the late 1940s.13 Tere were promising new developments
dating from Robert Clower’s important essay on the relationship between the
framework of Te General Teory and Walrasian general equilibrium (1984
[1963]). Axel Leijonhufvud’s Keynesian Economics and the Economics of
Keynes (1968) argued for a return to the source, to Te General Teory, which
had been misunderstood and distorted by a generation of disciples and crit-
Copyright © 2021. Harvard University Press. All rights reserved.
ics. Tis call was not heeded, but the 1970s saw a fowering of innovative ap-
proaches, summarized in Mauro Boianovsky and Roger Backhouse (2003).
But in the end these attempts did not save the baby from being thrown out
with the bathwater. Te neoclassical revival carried the day, to the point that
both Keynesians and anti-Keynesians noted the absence of young Keynesian
economists from the academy afer the 1970s (Blinder 1988, p. 278; Lucas
2013).
Te new data with which the old theory had to cope were the fnal blow,
but the counterrevolution had been long in the making. Te doubts sown by
Modigliani’s 1944 article were multiplied by a new insistence on anchor-
ing economic behavior in the rational, calculating, self-interested individual,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
72 Raising Keynes
homo economicus. Money-wage rigidity could not easily be reconciled with
competitive markets, in which everybody acts in terms of calculation of his or
her own advantage.14
Robert Lucas coupled competitive markets with the “rational expectations”
argument that economic actors all share a correct idea of how the economy
works and do not make systematic mistakes in predicting the future—they
might be wrong individually, but their errors would cancel each other out,
so that the average prediction would hone in on the true values of future out-
put, employment, interest rates, and so forth. In this case, everybody would
agree on the equilibrium, and nothing would prevent the economy from
moving at once to a new rational-expectations equilibrium if something upset
the old one.
Te crucial point is that the equilibrium exists not only in the mind of Lu-
cas and God (interchangeable for many economists who came of age in the
1980s) but also in every agent’s mind as well. No need at all, therefore, for a
dynamic theory of adjustment, the third pillar of the Keynesian revolution.
For Lucas and his followers, there was no question about which model of the
economy was the correct one: in this crucial respect there was no advance
from Pigou’s day, in which perfect competition ruled the roost. Te only way
for aggregate demand to play a role in this world would be if a large fraction
of economic agents were to act “irrationally,” that is, if most of the people
were fooled most of the time.15
Te rational-expectations equilibrium perspective was ramped up a notch
in the “real business cycle” theory put forth by Finn Kydland and Edward
Prescott (1982). Kydland–Prescott is the classical dichotomy in overdrive: the
shocks that move equilibrium are technological in nature, and aggregate de-
mand has no infuence, not even in the short run.
Te formalism of real b usiness cycle theory was new, but the ideas were
not. As I observed in the previous chapter, “New Classical” economics, the
economics of Friedman, Lucas, Kydland, and Prescott, had come full circle,
Copyright © 2021. Harvard University Press. All rights reserved.
back to the economics in which unemployment was the result of frictions, ri-
gidities, and other imperfections.
Te New Classicals and the New Keynesians remain divided to this day—
but only on the short run, where the New Keynesians appeal to one form or
another of infexibility to argue somewhat defensively, at least until the 2008
crisis, for the efficacy of monetary and (to a lesser extent) fscal policy. As far
as theory is concerned, a well-established consensus conceding the long run
to an anti-Keynesian view of the world is the common face of mainstream
economics. At least in terms of consistency, it must be admitted that the New
Classicals would appear to have the better of the argument. I have never heard
a convincing story why frictions, rigidities, and other imperfections would
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
2 • What Were They Thinking? 73
characterize the short period but can be counted upon to disappear in the
long run. One simply has to have faith that (unnamed) forces must be at work
to undermine all imperfections. In the long run, the arc of the universe bends
not only toward justice, as Martin Luther King supposed, but also toward
perfect competition. It is just that these forces work slowly enough that the
short period is blemished.
Tis book takes issue both with the idea that rigidities are responsible for a
main result of Te General Teory—the possibility of an unemployment equi-
librium—and with the related idea that the writ of Keynes does not run when
we move from the short period to the long. We turn now to Keynes’s frst-pass
model.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. II .
Keynes Defeated
Static Models and the Critics
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 3 .
the determination of output
and employment
First and Second Passes at Equilibrium
Given the propensity to consume and the rate of new investment,
there will be only one level of employment consistent with equilib-
rium. . . . Tere is no reason in general for expecting it to be equal to
full employment. Te efective demand associated with full employ-
ment is a special case, only realised when the propensity to consume
and the inducement to invest stand in a particular relationship to
one another. Tis particular relationship, which corresponds to the
assumptions of the classical theory, is in a sense an optimum relation-
ship. But it can only exist when, by accident or design, current invest-
ment provides an amount of demand just equal to the excess of the
aggregate supply price of the output resulting from full employment
over what the community will choose to spend on consumption when
it is fully employed.
—john maynard keynes
Keynes’s frst goal, as we have seen, was to establish that even in the absence of
frictions, a capitalist economy might not settle down at a point where every
willing worker had a job, and the productive apparatus of the economy was
Copyright © 2021. Harvard University Press. All rights reserved.
fully utilized. Te style of Te General Teory, if not the argument, is of a by-
gone era and was uncongenial even in 1936 to the twenty-somethings who
would become the backbone of the Keynesian Revolution. Te big diference
between Keynes, between the rhetorical style of his generation, and the style
of subsequent generations is that there are no models in today’s sense of the
term. Indeed, in the twenty-frst century, economics students fnd Te Gen-
eral Teory unrecognizable as economic argument. What passes for a model
is no more than descriptions of how variables relate to one another, a list of
ingredients without the recipe for baking them into a cake. Our task here is to
bake the cake, or rather to bake a couple of cakes based on alternative ways of
putting the ingredients together.
77
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
78 Raising Keynes
Stripped to its essence, the theory developed in Te General Teory hinges
on fve relationships: (1) the relationship between the rate of interest and in-
vestment demand; (2) the relationship between the level of output and de-
sired saving; (3) the relationship between the level of output and the transac-
tions demand for money; (4) the relationship between the asset demand for
money (what Keynes called the speculative demand) and the rate of interest;1
(5) fnally, the relationship between goods supply and the real price. (Keynes
uses the term “wage units” for real price.)
Te frst and the third of these relationships could not have upset the most
orthodox reader of Te General Teory; the frst is the standard stuf of re-
ceived doctrine, in the form of Irving Fisher’s theory of investment, and the
third is the so-called quantity theory of money, which remains a part of the
monetary side of Te General Teory, but not the whole. Neither is the sec-
ond relationship necessarily controversial, though it would be challenged by
Milton Friedman (1957) and Franco Modigliani (Modigliani and Brumberg
1954; Ando and Modigliani 1963). In 1936 it would have been argued that the
dependence of saving on income is irrelevant because, as was pointed out in
chapter 2, the interest rate does the heavy lifing in equilibrating investment
demand and the supply of saving. Similarly, the ffh relationship is unre-
markable: the basis of supply decisions in a competitive economy is p roft
maximization, and the supply schedule is the schedule of the real marginal
cost of output.
Tis leaves the fourth relationship, what Keynes calls the liquidity-
preference theory of interest. It is in fact the interrelationship between liquid-
ity preference and the theory of investment demand that is the principal in-
novation of Te General Teory. Te fnancial and real sectors of the economy
are linked by the two-sided nature of interest, one side its role in the fnancial
sector via liquidity preference and the other its role in the real sector via its
impact on investment demand. Together with the other relationships, the
dual role of interest leads to a central role for aggregate demand in determin-
Copyright © 2021. Harvard University Press. All rights reserved.
ing the level of output and employment.
As those who have followed the history of Te General Teory from its
birth know, the translation of the key relationships into models is not as
straightforward as it might appear if we start looking back from 2020, or from
even a few years afer publication. In the frst place, besides what I have iden
tifed as key relationships, there are less important relationships. For example,
relatively early on, the post–General Teory literature started taking output to
be a single, homogeneous good, in keeping with the macro orientation of
Keynes’s project. But the frst reviews (for example, James Meade [1937]) paid
considerable attention to the relationship between the output of investment
goods and the output of consumption goods, as well as to their relative prices.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 79
It was by no means obvious which were the core relationships and which were
peripheral to Keynes’s argument.
Ten there is what is lef out. For now we shall stay squarely in the short
run, a period in which we do not need to concern ourselves with the growth
of either the capital stock or the labor force. So there is nothing about these
relationships in the list. Tis simplifes our task. Furthermore, instead of relat-
ing wages and labor supply, as in the mainstream theory sketched in the pre-
vious chapter, Keynes simply posits a fxed money wage. Despite his insis-
tence that this is an assumption made merely to simplify the exposition, the
basis of Modigliani’s (1944) contrary assertion was that money-wage rigidity
is essential to Keynes’s argument. We shall turn to Modigliani’s argument in
chapter 4.
Tere remains the problem of moving from theory to model. Te dif ficulty
is that there are more variables than there are relationships. Tis means that
some of the relevant variables have to be specifed in advance, predetermined.
Transactions demand and liquidity preference together specify a relationship
between the amount of money required to grease the wheels of commerce
and satisfy the asset demands of wealth holders on the one hand and output
and interest rates on the other. But this leaves us in the dark as to whether the
relationships specify a demand for money based on levels of output and inter-
est, or levels of output and interest based on a supply of money.
Since Keynes’s monetary theory is an outgrowth of the quantity theory, al-
beit with signifcant modifcation, we might hope that the earlier theory
would provide some guidance. Alas, this is not to be. Despite Milton Fried-
man’s assertion that the quantity theory is and always has been a theory
of money demand, there is considerable ambiguity about whether the M in
MV = PY represents demand or supply.
In point of fact, Keynes presents ingredients for two models. A frst-pass
model, which we shall investigate now, assumes a given rate of interest, de-
rives the corresponding level of output, and then calculates money demand
Copyright © 2021. Harvard University Press. All rights reserved.
from the needs of businessmen and wealth holders. A second-pass model,
which we shall develop later in this chapter, reverses the causality, taking the
money supply as given and deriving from it both output and interest rate.2
Baking the First-Pass Cake: The Real Side
We consider each of the fve relationships—investment demand, supply of
saving, transactions demand for money, asset demand, and goods supply—in
turn. Te relationship between the rate of interest and investment demand,
I = I( ), Keynes’s “marginal ef
ficiency of capital,” is shown in Figure 3.1. Te
higher the rate of interest—seen here as a cost of investing in plant, equip-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
80 Raising Keynes
ment, structures—the lower is the level of investment. At a rate of interest
equal to 0, the corresponding level of investment is 100.
Te easiest way to think about the “marginal ef ficiency of capital” schedule
is to imagine an array of potential investment projects, ranked by their re-
spective rates of return.3 For each value of the rate of interest, , the corre-
sponding amount of investment demand is the total capital cost of all projects
with rates of return at least equal to . Te marginal ef ficiency of capital is the
rate of return on the marginal project.
Te rate of interest is the hurdle rate that projects must meet or surpass to
be worth undertaking. If is the rate at which businessmen discount invest-
ment returns, projects will normally have positive present values if their rates
of return exceed . For a rationally calculating, proft-maximizing, business-
man, the rate of interest is the opportunity cost of capital. He makes a com-
parison of the return on capital expenditure and the return on a bond yield-
ing the going rate of interest and opts for the subset of projects with higher
returns than are available in the bond market. In Keynes’s words:
Tere is always an alternative to the ownership of real capital-assets, namely
the ownership of money and debts; so that the prospective yield with
which the producers of new investment have to be content cannot fall be-
low the standard set by the current rate of interest. (Te General Teory,
pp. 212–213)
Or, equivalently (for our purposes at least), a businesswoman can be assumed
to be comparing the returns of capital expenditure over and above operating
Interest rate ( )
Copyright © 2021. Harvard University Press. All rights reserved.
Marginal efficiency of capital
0
100 200 300 400 Investment
demand
Figure 3.1 Investment demand as a function of the rate of interest.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 81
costs with the interest and principal payments she has to make on loans she
incurs for the sake of investment.
Te second relationship, that between saving and output, is more novel—
not, I have observed, in asserting that the level of desired saving depends on
the level of output, but in asserting the importance of this relationship. Equal-
ity of desired saving and investment means that the circular fow from output
and income on the one hand to expenditure on the other is uninterrupted:
any leakage from the circuit running from output and income to consump-
tion expenditure is made up by expenditure on new capital goods, that is, by
investment. But what guarantees the equality of desired investment and de-
sired saving, and therefore of income and expenditure, is not the interest rate.
Instead, it is the level of income itself. Keynes, as we saw in the last chapter,
rejects the mainstream theory, whereby desired saving and desired investment
are quickly if not immediately equated by the interest rate, arguing instead
that it is the dependence of saving on income that ensures the equality of sav-
ing and investment. Say’s Law becomes an equilibrium condition for expendi-
ture and income that is satisfed by a unique level of output rather than a rela-
tionship between expenditure and income that holds for any level of output.
We start from Figure 3.2, which depicts the relationship between real in-
come and real expenditure. Assume that out of every dollar of income $0.75
is spent on consumption goods. In Keynes’s terminology, the marginal pro-
Expenditure (E) E Y
500 C D ID
A
400 C D = 0.75Y
B
Copyright © 2021. Harvard University Press. All rights reserved.
300
200
100 ID
45°
100 200 300 400 500 Income (Y)
Figure 3.2 Consumption and investment determine aggregate demand.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
82 Raising Keynes
pensity to consume is 0.75. Tis relationship is represented by the schedule
CD = 0.75Y, the consumption function in Keynesian jargon. Te gap be-
tween income and consumption expenditure is given by the distance between
the 45-degree line—on which expenditure and income are equal—and the
consumption function. Investment constitutes additional expenditure, here
assumed to be 100, independent of the level of income. Te schedule labeled
CD + ID, the expenditure function, represents for each level of income the
amount that would be spent. Aggregate demand is given by the point on this
schedule where there is no leakage from the circular fow, or more accurately,
where leakages from consumption expenditure are just matched by injections
of investment demand. Graphically, this occurs where the expenditure func-
tion crosses the 45-degree line, at A = <400, 400>.
Aggregate demand is thus a stationary, or equilibrium, level of expenditure
and income, the level of income at which Say’s Law holds. At any other point,
where expenditure and income are not equal, the dynamics of the system will
force a change in both. Suppose that economy were at B = <300, 325>, where
300 is being produced and distributed as income, but 325 is being spent on
consumption and investment together. What happens as a result of this im-
balance? Te frst result will be that goods will fy of of the shelves faster than
they are being replaced. What happens next? Te logic of Keynes’s previous
book, A Treatise on Money, was that prices will rise.
But a group of young faculty members at the University of Cambridge—the
“circus”—who had gathered to digest and analyze the Treatise suggested that
there was more to the story—as did Richard Kahn (1931) in his pioneering
article on the multiplier, the indirect repercussions on employment of govern-
ment spending.4 A key passage reads:
At normal times, when productive resources are fully employed, the supply
of consumption goods in the short period is highly inelastic. . . . But at times
of intense depression when nearly all industries have at their disposal a large
surplus of unused plant and labour, the supply curve is likely to be very elas-
Copyright © 2021. Harvard University Press. All rights reserved.
tic. Te amount of secondary employment is then large and the rise of prices
is small. (p. 182)
In the Treatise Keynes had glimpsed the possibility that an imbalance be-
tween expenditure and income would lead to a change in employment and
output. His parable of a “banana economy” argued that, in the absence of a
corresponding expansion in investment expenditure, a thrif campaign would
lead to a downward spiral in employment and output that would end only
when the balance between desired saving and investment was restored or the
economy bottomed out at zero:
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 83
Tere will be no position of equilibrium until either (a) all production
ceases and the entire population starves to death; or (b) the thrif campaign
is called of or peters out as a result of the growing poverty; or (c) investment
is stimulated by some means or another so that its cost no longer lags be-
hind the rate of saving. (1930, vol. 1, p. 178)
Missing from Keynes’s parable is the analysis of the process by which a fall (or
a rise) in output and income is self-limiting, in the sense that the change in
output would itself correct the imbalance between investment and saving.5
As we shall see, there are at least two diferent ways of thinking about how
an excess of expenditure relative to output drives changes in output. One is
“Marshallian,” in which output responds to a more favorable relationship be-
tween prices and costs (Viner’s thinking, as reconstructed in chapter 2). A
second, “Walrasian,” way of thinking is that when inventories are drawn down
to satisfy higher expenditure, the frst response is to replenish inventories.
We shall examine both these stories in some detail in chapter 5, in prepara-
tion for the dynamic models laid out in chapters 6 and 7. For now, suf fi ce it to
say that the Marshallian story is a better ft with the story Keynes tells in his
banana parable, a story continued by Joan Robinson (1933a, 1933b). Keynes
sees the collapse of the banana economy beginning with a situation in which
“the selling price of bananas [has] fallen and not their cost of production”
(1930, vol. 1, p. 177). Robinson’s starting point for an increase in output is a
situation in which “prices are in excess of costs” (1933b, p. 24). In both these
stories the imbalance of expenditure and income changes prices, and the im-
balance between prices and costs changes output.
Nonetheless, when expenditures and incomes are out of line, the assump-
tion that the impact on output is direct and immediate makes for an easier
story of adjustment.6 In Figure 3.2, let us suppose for concreteness that when
E equals 325, orders increase by the full amount of the shortfall in goods, 25.
Tis leads to greater production (325 in place of 300), greater incomes (325),
Copyright © 2021. Harvard University Press. All rights reserved.
and more spending on consumer goods (243.75 = 0.75 × 325 in place of 225 =
0.75 × 300). But this is only the beginning. Tere is still a shortfall, since
spending, including spending on investment as well as on consumption, is
now 343.75 (= 243.75 + 100), whereas production is only 325. So the process
repeats with businesses increasing orders by 18.75 to make up for the short-
fall. And the new, higher, level of output and income leads to even more
spending.
Tere are in efect an infnite number of rounds of additional expenditure
and income, but each adds progressively less to the total because at each
round a proportion, in our case one-quarter, of the newly generated income
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
84 Raising Keynes
escapes the circular fow, ending up as saving. If we ignore the time dimen-
sion, and confate all these rounds into a moment of time, or if we imagine
that the original injection of spending is repeated over and over again, we will
fnd that the process converges to the equilibrium A = <400, 400>, where ex-
penditure and income are equal, and there is no pressure of sales to expand
output further.
It is important to be clear what an equilibrium like A represents—and what
it does not represent. It is a point of rest with respect to the balance of expen-
diture and income; that is, it represents the level of aggregate demand—the
level of income at which Say’s Law holds—corresponding to the given as-
sumptions about consumption, saving, and investment. It is not an equilib-
rium for the economy, or for the “goods market,” as a generation of Keynes’s
followers (myself included once upon a time) were wont to label it, at least not
without assuming very special circumstances. For we have, so far, said noth-
ing about supply.
Equivalently, aggregate demand is the level of income and output at which
desired saving SD and desired investment ID are equal, as in Figure 3.3. Tis
follows directly from the defnitions of expenditure and income. In our model
expenditure is equal to desired consumption plus desired investment: E =
CD + ID. Since by defnition SD = Y − CD, Y = E implies ID = SD and vice versa.
Tis equilibrium condition is diferent from the defnitional identity I ≡ S,
I D , SD
500
400
Copyright © 2021. Harvard University Press. All rights reserved.
300
200
SD = 0.25Y
100 ID
100 200 300 400 500 Income (Y)
Figure 3.3 Investment and saving determine aggregate demand.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 85
which follows from three accounting facts: output is the sum of consumption
and investment, income is the sum of consumption and saving, and output is
equal to income. Algebraically, we have Y = C + I and Y = C + S, so I = S what-
ever the level of Y.
Except at equilibrium, actual investment (I), which includes changes in in-
ventories whether desired or not, will difer from desired investment (ID),
which includes only desired changes in inventories. Tis diference is crucial,
at least for the Walrasian story. In a Walrasian world, moving from B to A
hinges on unintended reductions in inventories triggering the expansion of
output; the opposite, an unintended accumulation of inventories as expendi-
ture falls short of income, moves the economy downward.
Figure 3.3 facilitates two important insights into Keynes’s conception of ag-
gregate demand: the central role of investment demand and the concept of
the multiplier. Consider Figure 3.4, which reproduces Figure 3.3 with the ad-
dition of a second investment schedule, represented by the dashed line at 125.
Te construction of Figure 3.4 tells us that an increase in investment demand
of 25 increases aggregate demand by 100, from 400 to 500. Te original im-
pulse is multiplied by 4.
How does this work? Te basic idea is the same as the idea behind the ad-
justment process in Figure 3.2, from the disequilibrium B to the equilibrium
A. Te original increase of expenditure of 25—let us say, to build a factory—is
I D , SD
500
400
Copyright © 2021. Harvard University Press. All rights reserved.
300
200
S D = 0.25Y
125
100 ID
100 200 300 400 500 Income (Y)
Figure 3.4 Investment “determines” aggregate demand.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
86 Raising Keynes
in turn partially spent on consumption by the workers, managers, and busi-
ness owners, who receive wages, salaries, and p rofts from the new construc-
tion activity. Tis leads to additional production to provide the newly de-
manded consumption goods and services. Tis new activity in turn creates
new income and new demand. Since there are no supply constraints (yet),
the process is limited only by “leakages” from the chain of spending, in the
present simplifed case, leakages into saving. Formally, we have the change in
Y (Y) equal to
25 + (0.75 × 25) + (0.75 × 0.75 × 25) + ,
which, as a geometric sum, adds to 100. In general, if the spending by income
recipients out of each additional dollar of income is a uniform amount, which
we shall denote MPC, for marginal propensity to consume, we have
1
∆Y = 1 + MPC + MPC 2 + MPC 3 + = .
1 − MPC
Here the original impulse, the additional investment, is assumed to be 1
rather than 25.
If—and this is a big if—the propensity to consume is given, then we can
interpret investment as driving aggregate demand and determining saving.
Since aggregate demand is the point at which ID = SD, the level of ID must also
be the level of SD. Tus, for a given level of investment, the level of output and
income must be just enough to generate this level of desired saving—and no
more. Change ID and SD changes by the same amount because the level of in-
come adjusts to produce the required change in desired saving.
Putting together the investment demand function and the saving function
in Figure 3.5, we have the real side of the frst-pass model, that is, a relation-
ship between the interest rate and the level of output.
ID I D , SD
Copyright © 2021. Harvard University Press. All rights reserved.
400
300
200
Marginal efficiency of capital SD
100 ID
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
Figure 3.5 Aggregate demand for given interest rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 87
Baking the First-Pass Cake: The Financial Side
Te other side of Keynes’s frst-pass model is the relationship between the in-
terest rate, the level of output, and the demand for money. For Keynes the
demand for money comes from both the transactions associated with any
given level of real output and the desires of agents to hold assets in liquid
form. Transactions are assumed to require a proportionate amount of money,
M1, the constant of proportionality depending on both the price level P and a
variety of institutional factors summarized in the parameter , which is the
inverse of the output-velocity of money:7
M1 = PY.
As far as transactions are concerned, Keynes did not stray from the orthodox
view refected in the quantity theory of money.
With regard to the second source of the demand for money, Te General
Teory breaks new ground: Keynes creates his novel liquidity-preference the-
ory of interest from the assumption that there is a relationship between the
interest rate and the amount of money, M2, agents wish to hold in their
portfolios of assets:
M2 = ( )A.
( ) is agents’ desired ratio of liquid assets, M2, to total nominal wealth, A.
Te picture is in Figure 3.6.
Interest rate ( )
Copyright © 2021. Harvard University Press. All rights reserved.
( )
min
0 1 Liquid wealth as
fraction of total
wealth (M2/A)
Figure 3.6 Liquidity preference as a function of the interest rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
88 Raising Keynes
Te liquidity-preference schedule ( ) has two salient properties. First,
desired cash is inversely related to the rate of interest. For Keynes, interest is
the reward for giving up immediate access to wealth, so that the higher the
interest rate, the less desirable, relatively, is liquidity. Second, the proportion
of wealth agents wish to hold in liquid form goes to 100 percent at a positive
rate of interest, min, rather than at = 0. Tis second condition is the form
the so-called liquidity trap takes in a model in which the only fnancial assets
are bonds and cash. We take up both aspects of the liquidity-preference
schedule, the downward slope and the positive minimum interest rate, in
chapters 11 and 12.
Following Keynes, at least for now, we assume that wealth consists of two
parts, bonds refecting the nominal value of the capital stock, PK, and the
amount of money in wealth portfolios, M2. We assume bonds are consols (af-
ter their origin as “consolidated annuities” in eighteenth-century Britain),
commitments to pay a fxed yearly nominal sum, R, forever. For this commit-
ment to be invariant in real terms with respect to the price of goods and real
capital, we assume that the number of bonds B varies proportionately with P.
Tat is, we assume the rentier claim on real income, RB, is independent of the
price level, which is the appropriate normalization in the comparative-statics
framework of this chapter and the next. For simplicity, assume there is one
bond for each dollar’s worth of capital, so that B = PK.
Tere is another price to reckon with, the price of bonds PB. In general, the
bond price is a decreasing function of the rate of interest, but with the as-
sumption that bonds are consols, the relationship between PB and is a par-
ticularly simple one, namely, PB = R/, so that the relationship is the rectangu
lar hyperbola in Figure 3.7.
With assets limited to bonds and cash we have
A = PB ( )B + M 2 = PB ( )PK + M 2 ,
where M 2 is the quantity of money available for wealth portfolios. For equilib-
Copyright © 2021. Harvard University Press. All rights reserved.
rium the demand for money must be equal to the supply, which is to say,
M 2 = β( ρ) PB ( ρ)B + M 2 = β( ρ) PB ( ρ)PK + M 2 ,
β( ρ) β( ρ)
M2 = M2 = P ( ρ)B = P ( ρ)PK .
1 − β( ρ) B 1 − β( ρ) B
We put asset and transactions demands together in Figure 3.8, in which the
second quadrant combines Figures 3.6 and 3.7 to produce the liquidity-
preference schedule. Te frst quadrant pictures transactions demand. Given
the level of aggregate demand Y = 400 and interest rate 0, we have M1 = 300
and M2 = 200. In consequence M = 500.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
R
3
PB ( )
•
Figure 3.7 Te relationship of
bond price and interest rate.
PB
M2 M2 M1 M 1
500
Liquidity preference
400 Transactions demand
( ) ( )
PB ( ) B PB ( )PK 300
1 ( ) 1 ( ) PY Figure 3.8 Money demand
for given , Y, and P.
200
The Determination of Output and Employment
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
100
89
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
90 Raising Keynes
Baking the First-Pass Cake: Combining the
Real and the Financial Sides of the Model
We now can put the real and fnancial sides of aggregate demand together in
the form of a single diagram with four quadrants: fip Figure 3.8 on the hori-
zontal axis and marry it to Figure 3.5 to obtain Figure 3.9. Figure 3.9 maps a
given interest rate, in this case 0, to corresponding demands for real output
and money, Y = 400 and M = 500.
Figure 3.9 holds the price level as well as the wage level constant. What
happens if we vary the price level, while holding the wage constant in money
terms? If the price level increases from P0 to P1 as in Figure 3.10, the dashed
lines represent the new transactions-demand and liquidity-preference sched-
ules. At the higher price level, the same interest rate and aggregate demand
generate a higher money demand, M1 increasing from 300 to more than 350,
and M2 increasing from 200 to 300.
In the frst-pass model, price changes do not afect the real side of the
picture—desired consumption, saving, and investment. For this reason, if we
relate aggregate demand to the price level, as in Figure 3.11, we fnd that the
AD schedule is a vertical line. Here we start from a given interest rate (and
ignore various complications that will be introduced in later chapters), so
output is determined solely by aggregate demand and is independent of the
price level.
But we are not free to vary the price level arbitrarily; in the frst-pass model
the price level is determined as part of the equilibrium confguration by a GS
schedule. Indeed, in the frst-pass model this is the only reason that goods
supply matters. By virtue of the assumption of perfect competition, the supply
curve, as we have seen, is simply the economy’s marginal-cost curve. In Fig
ure 3.12, which adds a GS schedule to the AD schedule of Figure 3.9, the
equilibrium price level is 3.5. Observe that by fxing the nominal wage rate
(W = W0), the nominal price level P fxes the real price.
Copyright © 2021. Harvard University Press. All rights reserved.
Tere is some irony in this representation of the economy in the light of the
transformation of Keynes by friends and enemies from critical theorist to
theorist of one more version of sand in the wheels. In the mainstream view,
the AD schedule has a negative slope and the aggregate-supply schedule is
vertical—except in the short run, when sand gets in the wheels. In Figure 3.12
the AD schedule is vertical, and the GS schedule has a positive slope. In con-
trast with the mainstream view, in which supply rules the roost and demand
matters only to determine the price level, supply here matters only to deter-
mine prices, the nominal price level and the real wage. Output is determined
solely on the demand side!
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
ID ID , SD
400
300
II I
200
Marginal efficiency SD
of capital
100 ID
Interest rate ( )
Income (Y)
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
100
200
Liquidity 300
preference
400 Transactions demand
500
III M2 M2 M1 M 1 IV
Figure 3.9 Aggregate demand and money demand determined by the interest rate.
ID I D , SD
400
300
II I
200
SD
Marginal efficiency of capital
100 ID
Interest rate ( ) Income (Y)
Copyright © 2021. Harvard University Press. All rights reserved.
Interest rate ( ) P = P0 100 200 300 400 500 Income (Y)
0
P = P1 100
200
Liquidity preference 300 P = P0
400 Transactions demand
500 P = P1
III IV
M2 M2 M1 M 1
Figure 3.10 Aggregate demand and money demand at two diferent price levels.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
92 Raising Keynes
2 Aggregate demand
100 200 300 400 500 Y
W = W0
Figure 3.11 Aggregate demand at a given interest rate.
Supply of goods
5
2 Aggregate demand
Copyright © 2021. Harvard University Press. All rights reserved.
100 200 300 400 500 Y
W = W0
Figure 3.12 Aggregate demand and goods supply at a given interest
rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 93
Unemployment in the First-Pass Model
Te equilibrium in Figure 3.12 determines a level of employment—the num-
ber of workers required to produce 400 units of output—but it says nothing
about the level of unemployment. In fact, Te General Teory begins with a
somewhat convoluted defnition of involuntary unemployment, which we
shall examine in the frst appendix to this chapter. Suf fi ce it to say here that
Keynes rejects the core idea of the mainstream LS schedule, which is that the
supply of labor, and hence full employment, depends on the real wage.
Whatever the merits of Keynes’s strictures on the mainstream depiction of
the supply of labor, his critique does not ft very well with his avowed purpose
of meeting mainstream economics on its own turf as far as the structure of
markets is concerned. For that reason I shall out-Keynes Keynes in assuming
that there is no money illusion and no concern with relative wages so that, as
the mainstream argues, the supply of labor depends on the real wage.
It is not clear how much importance Keynes attached to his own character-
ization of unemployment despite the prominence he gives to laying it out. In
any case, the cutting edge of his characterization, namely, that unemployment
can exist without any pressure on money wages, is a provisional assumption,
scafolding rather than part of the building, which allows him to lay out the
core ideas of Te General Teory. He clearly states his intention to abandon
this assumption when the building is complete and the scafolding no longer
necessary. In his own words:
We shall assume that the money-wage and other factor costs are constant
per unit of labor employed. But this simplifcation, with which we shall dis-
pense later, is introduced solely to facilitate the exposition. Te essential
character of the argument is precisely the same whether or not money-
wages, etc., are liable to change. (Te General Teory, p. 27)
Tis is not quite accurate. In one sense the argument needs serious recast-
Copyright © 2021. Harvard University Press. All rights reserved.
ing, a major purpose of this book! Reconstructing (or constructing) a general
theory that refects Keynes’s intuition but is intelligible to present-day stu-
dents of economics is better served by sticking to the standard defnition of
unemployment, competitive labor supply and all. It may bear repeating that
this is not to deny the existence of rigidities and other imperfections in the
labor market, but to make it clear that Keynes’s insights about the role of ag-
gregate demand do not depend on imperfections. In this framework, unem-
ployment will exist if the equilibrium of aggregate demand and goods supply
in Figure 3.12 lies to the lef of the supply of labor.
Figure 3.13 adds the supply of labor (in terms of the goods the labor will
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
94 Raising Keynes
produce) at the price 3.5, one point on the LS schedule. At P = 3.5 and W =
W0 (in other words, at the real price 3.5/W0), workers are ready, willing, and
able to produce 500. Te output gap between the AD = GS equilibrium and
full employment is 100 (= 500 − 400).
In the frst-pass model there is no mechanism that automatically leads to
full employment. Indeed, even if the economy happens to fnd itself at full
employment, it will not stay there. Suppose the economy is initially at F =
<450, 4> in Figure 3.14. At this point there is both full employment and p
roft
maximization, but the economy will move away from this point to the AD-GS
equilibrium, E = <400, 3.5> as long as aggregate demand remains at 400. Ex-
penditure falls short of income and output when Y = 450, and one way or
another this sets in motion a decline in output.
Te assumption of a given money wage W = W0 precludes the reaction that
the mainstream relies upon to sustain its version of equilibrium, namely, a fall
in money wages whenever unemployment is present. It is true that at the AD-
GS equilibrium the marginal disutility of labor for the unemployed is less
than the going real wage W0/3.5, but nothing follows from this inequality.
With aggregate demand driving the economy, full-employment equilibrium is
possible only if aggregate demand can be expanded to the level at which the
GS schedule and the LS schedule intersect. Figure 3.14 shows the new AD
schedule as a dashed line. Te real wage at F is lower than the real wage at E,
but this change is the consequence of the movement from E to F rather than a
cause—a thermometer rather than a thermostat.
Supply of goods
5
4
Copyright © 2021. Harvard University Press. All rights reserved.
Supply of labor
3
Aggregate demand
2
100 200 300 400 500 Y
W = W0
Figure 3.13 Aggregate demand, goods supply, and labor supply.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 95
P
Supply of goods
4 F
3 E Supply of labor
1
Aggregate demand
450
100 200 300 400 500 600 Y
W = W0
Figure 3.14 Aggregate demand, goods supply, and labor supply:
moving to full employment.
Figure 3.15 indicates the changes in the interest rate, investment, and
money required for the full-employment equilibrium at F in Figure 3.14: the
interest rate has to be 1 in order to coax out investment demand equal to
112.5, which is the amount of investment needed to generate a demand for
output equal to 450. At the higher level of output, the real marginal cost is
also higher, so the price level has to be higher to ensure that producers are
maximizing profts. Both because prices are higher and because output is
higher, transactions require a greater amount of money, and the lower rate of
interest and the higher price level also drive up the asset demand for money.
Money demand totals almost 900, with M1 just shy of 400 and M2 near 500.
Copyright © 2021. Harvard University Press. All rights reserved.
Monetary Policy and Its Limits
We can, in the spirit of comparative statics, imagine a planet on which the
central bank sets the interest rate at 1 and fxes the money supply to ac
commodate money demand at this interest rate. Here the possibility of full
employment depends on whether the central bank has suf ficient fexibility
with regard to and on whether investment demand is suf ficiently responsive
to —in Keynes’s terms, upon liquidity preference and the marginal ef ficiency
of capital. Because of the liquidity trap the central bank cannot drive the in-
terest rate below min, which could prevent the necessary expansion of in
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
96
ID ID , SD
400
300
II I
200
SD
Marginal efficiency of capital
112.5
100 ID
Interest rate ( ) Income (Y)
Interest rate ( ) P = 3.5 450 Income (Y)
0 1 100 200 300 400 500
Raising Keynes
P= 4 100
200
Liquidity preference 300 P = 3.5
P = 4 Transactions
400 demand
500
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
M2 M2 M1 M 1
III IV
Figure 3.15 Interest, investment, aggregate demand, and the price level: moving to full employment.
3 • The Determination of Output and Employment 97
vestment demand—so could the inelasticity of the marginal-ef ficiency sched-
ule: even at a zero rate of interest, there could be insuf ficient investment
demand.
Figure 3.16 shows both possibilities. As the liquidity-preference schedule is
drawn, min is too high for investment demand compatible with full employ-
ment. With = min investment demand is only 60, which, given a multiplier
of 4, generates aggregate demand of 240.
Even if there were no liquidity trap and the interest rate could be driven to
zero, the demand for investment may be too low to generate the aggregate
demand corresponding to a full-employment level of output and income: in
Figure 3.16 investment demand is only 80 at = 0, and aggregate demand is
320.
Keynes himself thought a foor to the hurdle rate of more theoretical than
practical signifcance. “Whilst this limiting case might become practically im
portant in future, I know of no example of it hitherto” (Te General Teory,
p. 207). Both in Te General Teory and in later writings, Keynes put more
emphasis on the inelasticity of the marginal ef ficiency of capital schedule. He
didn’t go quite as far as stagnationist Keynesians (like Alvin Hansen) who
thought that capitalism was on the verge of exhausting its possibilities for
proftable private investment even at very low rates of interest, thinking that it
would take decades rather than years for the stagnationist vision to be real-
ized (Te General Teory, chap. 24, sec. 2).8
Te policy takeaway is that eliminating involuntary unemployment might
require the government to go beyond monetary policy in order to compen-
sate for the limits of private investment demand. Although Keynes had long
advocated defcits—we have seen that he was hardly alone in this—in Te
General Teory he is vague about the means, saying only,
I expect to see the State . . . taking an ever greater responsibility for directly
organizing investment; since it seems likely that the fuctuations in the mar-
ket estimation of the marginal ef ficiency of diferent types of capital . . . will
Copyright © 2021. Harvard University Press. All rights reserved.
be too great to be ofset by any practicable changes in the rate of interest.
(p. 164)
Indeed, although the term “fscal policy” occurs several times in Te Gen-
eral Teory, there is no systematic discussion of the implications of the overall
argument for taxing and government spending.9 Te theoretical justifcation
for using fscal policy to compensate for shortages in private demand was
conceived in the womb of Te General Teory, but—see chapter 14—it was
Abba Lerner rather than Keynes who spelled out the radical implications of
Te General Teory for fscal policy.
Besides the doubt it casts on the efficacy of monetary policy, the frst-pass
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
98
ID ID , SD
400
300
II I
200
SD
100
Marginal efficiency of capital 80 ID
60
Interest rate ( ) Income (Y)
min 240 320
Interest rate ( ) 100 200 300 400 500 Income (Y)
Raising Keynes
100
200
Liquidity preference 300 Transactions demand
400
500
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
M2 M2 M1 M 1
III IV
Figure 3.16 Aggregate demand limited by liquidity preference and the marginal efficiency of capital.
3 • The Determination of Output and Employment 99
model lays out a very diferent approach from the mainstream story about the
determination of output and employment, in which supply-side factors are no
longer in the driver’s seat. Keynes’s frst-pass model is one in which the star-
ring role goes to aggregate demand, to the point that supply matters only in
determining the price level. Full employment may be achievable but it re-
quires the visible hand of government to manage demand; the self-correcting
market mechanism is permanently AWOL.
The Second-Pass Model: Capitalism Left to Its Own Devices
Keynes’s critical intention went beyond providing an apparatus for showing
the possibilities and limitations of monetary policy. He intended no less than
the overthrow of basic mainstream doctrine that a capitalist economy lef to
itself would reach an equilibrium in which each and every market—including
the labor market—would have a willing buyer for every willing seller, an equi-
librium in which involuntary unemployment was impossible. Te frst-pass
model doesn’t accomplish this, for it takes as given two parameters—the in-
terest rate and the money wage—whose fexibility the mainstream considered
essential in leading the economy to full employment. Keynes obliges with a
second-pass model, in which the money supply rather than the interest rate is
given. For now, the money wage is still assumed to be given; for Keynes this
remains a “simplifcation . . . introduced solely to facilitate the exposition”
(Te General Teory, p. 28).
Te change involved in relaxing the assumption of a fxed interest rate is
enough to complicate the determination of output and employment: although
Keynes claims that the demand side of the model continues to determine the
levels of output and employment in the second-pass model (Te General
Teory, p. 247), we shall see that this result no longer holds. Te consequence
of reversing the direction of causality in the liquidity-preference schedule is
that aggregate demand varies with the price level, and equilibrium output and
Copyright © 2021. Harvard University Press. All rights reserved.
employment are jointly determined by aggregate demand and goods supply.
Te easiest way to enter into the logic of the second-pass model is to view
the equilibrium represented in Figures 3.9 and 3.12 in reverse. Instead of fx-
ing the interest rate and deducing the implications for money demand, we fx
the money supply and examine the consequences for the interest rate, along
with output, prices, and wages.10
Go back to Figure 3.9 and the arrows that run from the interest rate 0 to
the components of the overall money demand M1 = 300 and M2 = 200. Te
frst question in constructing the second-pass model is how to go in the op-
posite direction, how to determine the interest rate, and therewith investment
demand and output, starting from an exogenously given money supply.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
100 Raising Keynes
Te frst problem is that money is money. How do we know the market will
break up a given total, like 500, into the appropriate components, money for
transactions (M1 = 300) and money as an asset (M2 = 200)? More generally,
for an arbitrary total money supply, how do we know what the appropriate
division is between money to facilitate transactions and money to hold as an
asset?
We shall deploy Hicks’s (1937) IS-LM model to answer these questions. Te
key insight behind IS-LM is that fnancial and real variables provide separate
and distinct relationships between the interest rate and the level of income
(output), both of which have to be satisfed to establish the level of aggregate
demand.
Look frst at the fnancial relationships that connect the transactions de-
mand to income and the asset-market equilibrium to the interest rate. As-
sume the total money supply is given as 500. Figure 3.17 reproduces the rele-
vant parts of Figure 3.9 with one modifcation: although we continue to
measure asset money in quadrant three from “north” to “south,” in quadrant
four we measure transactions money from bottom to top, south to north. Te
point M1 = M1 = 0 corresponds to the point M2 = M 2 = 500 in quadrant three.
In efect, we deduct money required for transactions from the overall money
supply to fnd the money available for holding as an asset. Tis allows us to
trace out the relationship between the interest rate and the level of aggregate
demand consistent with fnancial equilibrium, that is, consistent with the to-
tal demand for money (asset + transactions) being equal to total supply.
Divide the total money supply, M = 500, into M1 = 300 and M 2 = 200. Now
ask what the implications of this division are for the interest rate and the level
of income, that is, what levels of the interest rate and the level of income are
compatible with this division and thus with each other? Going up the vertical
axis from M1 = 0 to M1 = 300, we follow the horizontal arrow to its intersec-
tion with the transactions-demand schedule: M1 = 300 supports an income of
400. At the same time, M1 = 300 leaves 200 to satisfy the demand for money as
Copyright © 2021. Harvard University Press. All rights reserved.
an asset. We move to the lef from M 2 = 200 to the point of intersection with
the liquidity-preference schedule, namely = 0. Translating this through
quadrant two, we can fnally relate the rate of interest and the level of aggre-
gate demand to one another in quadrant one. Specifcally, the division of the
money supply into M1 = 300 and M 2 = 200 corresponds to the point <400, 0>
in the space of Y × . (We actually knew this already since this composition of
money demand emerged in the frst-pass model from the assumption = 0,
which led to an aggregate demand of 400.)
Now keep the money supply constant but change the composition of de-
mand. For example, set M1 = 200 and M2 = 300. Tis generates a second point
in quadrant one, as shown in Figure 3.18.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Interest rate ( ) Interest rate ( )
0
Interest rate ( ) 45° Income (Y)
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
100 400
200 300
Liquidity preference
300 200
400 100 Transactions demand
500 0
M2 M2 M1 M 1
Figure 3.17 Financial equilibrium: constructing the LM schedule, I.
Interest rate ( ) Interest rate ( )
0
Interest rate ( ) 45° Income (Y)
Copyright © 2021. Harvard University Press. All rights reserved.
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
100 400
200 300
Liquidity preference
300 200
400 100 Transactions demand
500 0
M2 M2 M1 M 1
Figure 3.18 Financial equilibrium: constructing the LM schedule, II.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
102 Raising Keynes
If we vary the composition of the money supply continuously, always hold-
ing the total constant at 500, we generate a continuous schedule in quadrant
one, each point on the schedule refecting the combination of and Y corre-
sponding to a particular composition of the money supply. Tis is the LM
schedule, shown in Figure 3.19. Te LM schedule traces out the requirement
of fnancial equilibrium in this very simple model: namely, that money de-
mand, M1 + M2, equals money supply, M,11 which is to say
β( ρ) β( ρ)
M = αPY + PB ( ρ)B = αPY + P ( ρ)PK .
1 − β( ρ) 1 − β( ρ) B
Let us look next at the real side of the model, the relationship between the
interest rate and aggregate demand, as mediated by the marginal ef ficiency of
capital. In Figure 3.20, quadrant three shows the marginal efficiency of capital
schedule; in particular, the interest rate 0 corresponds to an investment de-
mand of 100, which in turn corresponds, in quadrant four, to an aggregate
demand of 400. Te correspondence is refected in quadrant one, which re-
lates the interest rate to aggregate demand. Raising the interest rate produces
a new, lower, level of investment demand and income, as Figure 3.21 shows.
Just as we varied the composition of money demand to obtain the LM
Interest rate ( ) Interest rate ( )
0 LM
Interest rate ( ) 45° Income (Y)
Copyright © 2021. Harvard University Press. All rights reserved.
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
100 400
200 300
Liquidity preference
300 200
400 100 Transactions demand
500 0
M2 M2 M1 M 1
Figure 3.19 Financial equilibrium: constructing the LM schedule, III.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Interest rate ( ) Interest rate ( )
0
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
Marginal efficiency of capital 100 ID
SD
200
300
ID ID , SD
Figure 3.20 Equilibrium of expenditure and income: constructing the IS schedule, I.
Interest rate ( ) Interest rate ( )
0
Interest rate ( ) Income (Y)
Copyright © 2021. Harvard University Press. All rights reserved.
45°
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
Marginal efficiency of capital 100 ID
SD
200
300
ID ID , SD
Figure 3.21 Equilibrium of expenditure and income: constructing the IS schedule, II.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
104 Raising Keynes
schedule, we can vary the interest rate to obtain the IS schedule, which is the
set of combinations of interest rate and real output for which investment de-
mand and the supply of saving are equal:
I( ) = sY.
Figure 3.22 traces this out graphically.
Now superimpose the frst quadrant of Figure 3.19 on the frst quadrant of
Figure 3.22. We obtain the IS-LM diagram, pictured in Figure 3.23, which has
served as the alpha and omega of Keynesian economics for generations of
undergraduate economics majors. Te intersection of the two schedules, the
point <400, 0> defnes aggregate demand for the given price level. To be sure,
we don’t learn much from the diagram; we already learned earlier in this
chapter that the combination of Y = 400 and = 0 is compatible with a
money supply of 500 divided as M1 = 300 and M2 = 200. Indeed, with the in-
terest rate fxed exogenously, as it is when the central bank is assumed to be in
control, the IS-LM analysis is superfuous. Now it becomes important pre-
cisely because we start not from the interest rate but from the money supply.
And this makes a diference when we come to the next question: what hap-
pens when we vary the price level?12
Interest rate ( ) Interest rate ( )
0
IS
Interest rate ( ) Income (Y)
Copyright © 2021. Harvard University Press. All rights reserved.
45°
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
Marginal efficiency of capital 100 ID
SD
200
300
ID ID , SD
Figure 3.22 Equilibrium of expenditure and income: constructing the IS schedule, III.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 105
Interest rate ( )
LM
0
IS
100 200 300 400 500 Income (Y)
Figure 3.23 Expenditure-income and fnancial equilibria determine aggregate
demand.
Since the price level has no impact on real investment demand, it makes no
diference to aggregate demand what the price level is when the interest rate is
fxed exogenously—the frst-pass model—changes in the price level simply
translate into changes in the amount of money required to support the level
of income corresponding to the given interest rate. But in the second-pass
model, the interest rate is free to take on diferent values, and increasing the
price level has the same efect as decreasing the money supply. Or rather, may
be equivalent, depending on what is assumed about the wage level.
We start, with Keynes, assuming that the nominal wage is fxed, W = W0.
Ten the liquidity-preference and transactions-demand schedules in Figures
3.10, 3.11, and 3.12 shif with an increase in P from P0 to P1 to produce a new
LM schedule, the dashed curve in quadrant one of Figure 3.24.
For any given level of income, more money is necessary to carry out trans-
actions. Hence less is available to hold as wealth. For agents to be content with
holding the reduced amount of money in their portfolios, interest rates must
Copyright © 2021. Harvard University Press. All rights reserved.
be higher; this change is represented by a move along the liquidity-preference
schedule. Additionally, at least in a comparative-statics framework, a higher
price level implies a higher nominal stock of bonds (each with a par value of
one dollar), which is refected in a shif of the liquidity-preference schedule
from the solid to the dashed curve.
Observe that the interest rate lives a double life: it is a real variable in rela-
tion to investment demand and a nominal variable relative to the asset de-
mand for money. In its frst incarnation it is independent of the price level.
Saving too is determined in real terms, in the present model by real income
alone. Hence the IS schedule is unafected by a change in the price level.
Te result is that we can superimpose the IS schedule from Figure 3.22 to
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
106 Raising Keynes
Interest rate ( ) Interest rate ( )
LM
P = P1
P = P0
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 100 200 300 400 500 Income (Y)
100 400
P = P1
Liquidity preference 200 300
P = P0
300 200
P = P0
P = P1 400 100 Transactions demand
500 0
M2 M2 M1 M 1
Figure 3.24 Financial equilibrium: constructing the LM schedule for diferent price levels.
obtain a new aggregate-demand equilibrium. In Figure 3.25 the interest rate
1 and the level of income 360 correspond to the price level P = P1 = 4.8, while
the interest rate 0 and Y = 400 correspond to the price level P0 = 3.5.
Evidently, the relationship between the price level and aggregate demand
when the interest rate is held constant is diferent from the relationship be-
tween price and demand when the money supply is held constant. In the frst
case, the price level, varying with money demand, has no impact on aggre-
gate demand, at least not until we introduce considerations that go beyond
Copyright © 2021. Harvard University Press. All rights reserved.
Keynes’s frst-and second-pass frameworks, hence the vertical AD schedule
in the frst-pass model.
By contrast, in the second case, with the money supply fxed, there is an
inverse relationship between the price level and aggregate demand: the lower
are prices, the less is the amount of money required for transactions, hence the
more is available to satisfy the demand for money as an asset, hence, in turn,
the lower the interest rate and the higher is investment demand. Tis relation-
ship, frst noted by Keynes in chapter 19 of Te General Teory (p. 266), would
come to be known simply as the “Keynes efect.” It is represented by a shif
along the liquidity-preference schedule as the amount of money in asset port-
folios varies.
In Figure 3.24 there is also a “bond efect,” the shif in the liquidity-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 107
preference schedule. Tis shif refects a changing quantity of bonds as the
value of the capital stock varies with the price level.
Figure 3.26 assumes the price level varies continuously and maps the IS-
LM equilibria in Y × space of Figure 3.25 to Y × P space. In Figure 3.26 the
point <400, 3.5> corresponds to <400, 0> in Figure 3.25, and <360, 4.8> cor-
responds to <360, 1>.
Interest rate ( )
P = 4.8
1 P = 3.5
0
LM IS
360
100 200 300 400 500 Income (Y)
Figure 3.25 Expenditure-income and fnancial equilibria determine aggregate
demand: as the price level rises, the equilibrium level of income—the level of
aggregate demand—falls.
5
4.8
4
Copyright © 2021. Harvard University Press. All rights reserved.
3.5
3
Aggregate demand
2
360
100 200 300 400 500 Income (Y)
W = W0
Figure 3.26 Aggregate demand schedule determined by varying P while
holding M constant.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
108 Raising Keynes
Equilibrium requires that the economy be not only on its AD schedule but
also on its GS schedule. Figure 3.27 shows the AD-GS equilibrium at E. Un-
employment exists as long as the labor supply at the equilibrium real price
would produce more output than is produced at E. Here, the full-employment
level of output is 500, and the output at the AD-GS equilibrium is 400. In
Keynes’s story, there is no reason why a capitalist economy will provide a job
for every willing worker even if we swap the frst-pass model for the second-
pass model.
Te lesson of both the frst-and the second-pass models is that eliminating
unemployment requires an appropriate interest rate to coax out the requisite
level of investment demand and hence aggregate demand. Tis, according to
Keynes, is beyond the capacity of a capitalist economy lef to its own devices,
that is, a capitalist economy without the hand of a central bank at the mone-
tary tiller. And even the most steady hand may not be enough.
What happens if we drop the “scafolding” of a fxed money wage? Te fol-
lowing chapter explores Franco Modigliani’s (1944) argument that fexible
money wages can do the job of adjusting investment demand to the require-
ments of full employment. And, for good measure, we explore the fallback
argument, due originally to Gottfried Haberler (1939, 1941), that in the event
investment demand proves unresponsive, fexible wages can adjust consump-
tion demand to the needs of full employment. Together these arguments con-
stitute a powerful critique of Keynes’s theoretical claims—or so mainstream
economics has believed for more than half a century.
Aggregate demand
P
Supply of goods
4 E
Copyright © 2021. Harvard University Press. All rights reserved.
Supply of labor
3
100 200 300 400 500 600 Y
W = W0
Figure 3.27 Aggregate demand–Goods supply equilibrium: the supply of
labor is a nonbinding constraint.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 109
appendix 1: keynes’s defin it ion(s) of unemployment
As for the creation of Eve in Genesis, there are two accounts of the meaning of
involuntary unemployment in Te General Teory:
Men are involuntarily unemployed if, in the event of a small rise in the price
of wage-goods relatively to the money-wage, both the aggregate supply of
labour willing to work for the current money-wage and the aggregate de-
mand for it at that wage would be greater than the existing volume of em-
ployment. (p. 15)
An alternative, though equivalent, criterion is that . . . aggregate employ-
ment is inelastic in response to an increase in the efective demand for its
output. (p. 26)
Te frst defnition says that if the real wage goes down because of an in-
crease in the price level relative to the money wage, both the supply of labor
and the demand for labor would be higher than the amount of employment at
the original real wage. Te second, if we insert the adverb “completely” before
inelastic, is tautological: full employment is reached when there is no possibil-
ity for expanding output and employment.
Te frst defnition is problematic for several reasons, one being that it is
based on a counterfactual. Apparently, we can’t tell if there is involuntary un-
employment without asking what would happen if the real wage were to
change. Tis is odd enough. But for Keynes’s readers it must have been even
more odd to distinguish between a change in the real wage that takes place
because the price level changes and a change that takes place because the
money wage changes.
One reason for distinguishing between the efect of price changes and the
efect of wage changes is money illusion: workers focus on their money in
comes rather than on their real incomes. Te problem with money illusion
Copyright © 2021. Harvard University Press. All rights reserved.
from a mainstream point of view is that a canon of orthodoxy is the ability of
economic agents to see through the “veil” of money and determine their ac-
tions on the basis of real rather than monetary values. According to the clas-
sical dichotomy, it shouldn’t matter whether the real wage changes because of
a change in the price level or because of a change in the wage rate. One of the
early charges leveled at the argument of Te General Teory was precisely that
it hinges on money illusion.
In point of fact, Keynes does not rely on money illusion, but rather on the
importance of relative wages. Any single group of workers will resist wage
cuts because this puts them at a disadvantage relative to their fellow workers.
But the same workers will accept price increases that reduce real wages be-
cause all are hit equally, at least up to the point that the “reduction proceeds
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
110 Raising Keynes
so far as to threaten a reduction of the real wage below the marginal disutility
of the existing volume of employment” (Te General Teory, p. 14). Tat is,
employed workers customarily enjoy a rent because the marginal disutility of
labor is usually less than, rather than equal to, the going real wage.
Tis leads to the paradoxical result that there usually exists some unem-
ployment in the sense that, at the going real wage, workers would prefer to
work more. Moreover, Keynes’s reasoning leads to the conclusion that the
only reason workers do not accept pay cuts is because “reductions of money-
wages . . . are seldom or never of an all-around character” (Te General Te-
ory, p. 14).
Te focus on relative wages and the existence of wage rents need not lead to
the conclusion that workers are concerned with money wages rather than real
wages, or that they respond diferently to wage cuts and price increases. We
can defne labor supply as the amount workers wish to work at the going real
wage (or real price, as in the models of this chapter), provided we stipulate
that changes in the money wage, like changes in the money price, afect all
workers equally, which is to say that wage changes are not piecemeal. At any
real wage, unemployment exists if the actual level of employment and output
is less than the labor supply at that real wage. When unemployment exists, the
marginal disutility of labor is less than the real wage.
Tis defnition is compatible with Keynes’s second defnition of full em-
ployment in terms of a completely inelastic response of output and employ-
ment to aggregate demand. Te only additional assumption we have to make
is to rule out compulsory overtime and the like so that the LS schedule be
comes an absolute limit on the amount of employment.
Keynes may believe that he needs to focus on relative wages to explain
money-wage rigidity, but this is true only if money-wage rigidity is essential
to his argument. Tis Keynes denies, with the result that his defnition of in-
voluntary unemployment in terms of an asymmetry between the efect of a
wage cut and the efect of a price rise adds nothing but confusion. It is no
Copyright © 2021. Harvard University Press. All rights reserved.
wonder there remained doubts about Keynes’s theory even among those who
accepted the policy implications.
appendix 2: do interest rates adjust
saving and investment?
We can now revisit the question of the role of the interest rate in equilibrating
the demand for investment and the supply of saving and compare the main-
stream position examined in the previous chapter with Keynes’s own posi-
tion. Te mainstream position, before Keynes and aferward, has been that
desired investment and desired saving constitute the two blades of a Marshal-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 111
lian scissors exactly analogous to the textbook model of a competitive market.
And the rate of interest is the price that brings investment demand (Keynes's
marginal efficiency of capital) and saving into equilibrium.
We start from the equilibrium <I0, 0> in Figure 3.28 and suppose that the
equilibrium is disturbed by a collapse of investor confdence. If the apple
analogy suggested by Frank and Bernanke (2007, p. 639)—“the movements of
the real interest rate clear the market for saving in much the same way that the
price of apples clears the market for apples”—were valid, we would expect that
the excess supply of saving at the original equilibrium interest rate, 0, would
drive the rate of interest down as savers competed with each other to place
their funds now that fewer borrowers are in the market. Tis process would
continue until the interest rate coaxed out suf ficient investment to match the
desired saving. In Figure 3.28 the new equilibrium interest rate is 1. (For
simplicity, saving is depicted as completely inelastic with respect to the rate of
interest, a simplifcation that does not afect the logic of adjustment.)
Te General Teory ofers a diferent analysis. To a frst approximation, the
impact of a decrease in investment demand will fall on saving itself: the
saving schedule will move inward to accommodate the new, lower, invest-
ment demand at the original rate of interest, 0. Te picture is given in Figure
3.29.
Interest rate ( )
Supply of saving
0
Copyright © 2021. Harvard University Press. All rights reserved.
Investment demand
1
I0 Investment, Saving
Figure 3.28 Investment demand and saving supply: equilibrium restored by a
change in the interest rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
112 Raising Keynes
Interest rate ( )
Supply of saving
0
Investment demand
I1 I0 Investment, Saving
Figure 3.29 Investment demand and saving supply: equilibrium restored by a shif
in the saving schedule.
In Keynes’s story, the downward shif in the investment schedule consti-
tutes a reduction in aggregate demand, and the level of output falls in re-
sponse. Te fall continues until desired saving, responding to the level of
output and income, falls to the new level of investment demand. End of
story—at least in the frst-pass model.
Tings are more complicated in the second-pass model. With a fxed
money supply, there is a secondary efect because the price level falls and
more money spills over into the demand for fnancial assets. We can trace the
Copyright © 2021. Harvard University Press. All rights reserved.
efects of the downward shif in the investment-demand schedule in the third
quadrant of Figure 3.30. Tis shif translates into a fall in the level of output
that equilibrates desired saving and investment in the fourth quadrant, and a
downward shif in the IS schedule in the frst quadrant. As investment (and
saving) fall from I0 to I1, income falls from Y0 to Y1.
Tis, however, is still an approximation. For not only does the IS schedule
shif but there is a shif along the LM schedule (and, before we are done, a
shif in the LM schedule itself, as the price level responds to the change in the
level of economic activity).
Te shif along the LM schedule takes place because at a lower level of eco-
nomic activity, less money is needed for transactions and more money is thus
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Interest rate ( ) Interest rate ( )
3 •
0
IS
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 0 Y1 Y0 Income (Y)
l1
l0
Marginal efficiency of capital
SD
The Determination of Output and Employment
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
ID ID , SD
113
Figure 3.30 A decrease in investment demand shifs the IS schedule.
114 Raising Keynes
available to satisfy asset demands. But with more money chasing the same
amount of fnancial assets, the prices of bonds will rise and the interest rate
will fall. Tis efect is shown in Figure 3.31.
In this fgure the money released to the asset market by a reduction in in-
come and output from Y0 to Y1 is willingly held by agents only if the rate of
interest falls from 0 to 1. Tis fall in the interest rate stimulates movement
along the investment-demand schedule and thus leads to higher output and
income. For fnancial markets to be in equilibrium and expenditure to be
equal to income, both the LM and the IS schedules must be satisfed, as in
Figure 3.32 at the point <Y2, 2>.
We now have to modify Figure 3.29 to take account of these changes in the
fnancial sector. In place of the equilibrium <I1, 0>, we have the equilibrium
<I2, 2> in Figure 3.33.
But we are still not done. Te process by which the economy moves from
one equilibrium to another will be explored in detail in chapters 5, 6, and 7;
for now, suf fi ce it to say that we have dealt with only half of the process, the
demand side. Indeed, not even: we have only examined the relationship of
two points, each on a separate AD schedule but each assuming the same price
Interest rate ( ) Interest rate ( )
LM
0
1
Interest rate ( ) Income (Y)
45°
Copyright © 2021. Harvard University Press. All rights reserved.
Interest rate ( ) 0 1 Y1 Y0 Income (Y)
Liquidity preference
Transactions demand
M2 M2 M1 M 1
Figure 3.31 With a fxed money supply, a fall in income decreases the equilibrium interest rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Interest rate ( ) Interest rate ( )
LM
0
1 2
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 0 21 Y1 Y2 Y0 Income (Y)
Liquidity preference
Transactions demand
M2 M2 M1 M 1
Figure 3.32 A lower interest rate stimulates investment demand, a partial ofset to the initial
decline.
Interest rate ( )
Supply of saving
0
Copyright © 2021. Harvard University Press. All rights reserved.
2
Investment demand
1
I1 I2 I0 Investment, Saving
Figure 3.33 Shif in investment demand and induced shifs in supply of saving.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
116 Raising Keynes
level. If we repeat the exercise while allowing the price level to vary, the entire
AD schedule shifs in line with the assumed shif in the investment-demand
schedule. Te dashed line in Figure 3.34 represents the new AD schedule.
But what determines the price level? To answer this question we need to
bring in the supply side. By superimposing a GS schedule on Figure 3.34, we
can see that there is a further efect on the equilibrium level of output and
income via the efect of a lower price level on the real money supply and
hence on the interest rate and investment demand. As the level of output falls,
the price level falls with marginal cost, and at the lower price level the trans-
actions demand for money falls; the interest rate must fall to accommodate
the greater volume of asset money relative to the stock of bonds. Figure 3.35
shows the new AD-GS equilibrium (Y3) in the space of Y × P, and Figure 3.36
the new equilibrium (I3) in I × space. Te rate of interest 3 in Figure 3.36
corresponds to the price level P3 associated with the new equilibrium at Y3 in
Figure 3.35.
Keynes vs. the Mainstream on Interest Rates,
Saving, and Investment: A Summing Up
Te key diference between the mainstream story and Keynes’s story is the
diference between a movement along a given saving schedule in the main-
stream story and a shif in this schedule in Te General Teory story. Te
P
Copyright © 2021. Harvard University Press. All rights reserved.
P0
Aggregate demand
Y2 Y0 Y
W = W0
Figure 3.34 A fall in investment demand shifs aggregate demand
downward.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • The Determination of Output and Employment 117
P Supply of goods
P0
P3
Aggregate demand
Y2 Y3 Y0 Y
W = W0
Figure 3.35 Taking into account the supply-side consequences of a fall
in investment demand, I: in the space of income × prices.
Interest rate ( )
Supply of saving
0
Copyright © 2021. Harvard University Press. All rights reserved.
2 Investment demand
3
I1 I2 I3 I0 Investment, Saving
Figure 3.36 Taking into account the supply-side consequences of a fall in
investment demand, II: in the space of investment and saving × interest.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
118 Raising Keynes
mainstream relies on interest-rate adjustments, whereas, to a frst approxima-
tion, Keynes relies on output adjustment.
If we go beyond the frst approximation, the interest rate also adjusts in
Keynes’s story, albeit with two important diferences from the mainstream.
Te frst is that a shifing IS schedule and a stationary LM schedule—not a
shifing investment demand and a stationary saving supply—drive interest-
rate changes. Te second is that even afer all adjustments to a change in in-
vestment demand are made, IS-LM equilibrium moves in the same direction
as the original change; consequently aggregate demand moves in the same
direction.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Te frst-pass model is characterized by two equations on the real side and
two equations on the fnancial side. Te real-side equations are (1) the equal-
ity of investment demand and desired saving
I( ) = sY,
which defnes the AD schedule, and (2) proft maximization, which defnes
the GS schedule
P
Y = GS
W
by the frst-order condition of equality between price and marginal cost
P
= ( FL ) .
−1
W
FL ≡ ∂Y/∂L is the marginal productivity of labor, derived from the production
function Y = F(K, L). Diferentiating the price = marginal cost condition with
respect to P/W gives
dY
= − ( FLL ) ( FL ) ,
−1 3
GS′ ≡
P
d
W
which is positive since FLL ≡ ∂2Y/∂L2 < 0. Te models of this chapter also take
Copyright © 2021. Harvard University Press. All rights reserved.
the money wage as given: W = W0.
On the money side the two equations are the demand (1) for money for
transactions
M1 = PY
and (2) for the demand for cash in wealth portfolios
β( ρ) β( ρ)
M2 = PB ( ρ)BN = P ( ρ)PK .
1 − β( ρ) 1 − β( ρ) B
Causality runs from the interest rate and the money wage, = 0 and W = W0,
to the equilibrium level of output, Y0, determined by aggregate demand, and
119
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
120 Raising Keynes
to the equilibrium price level, P0, determined jointly by aggregate demand
and goods supply. Te price level and the level of output together deter-
mine the transactions demand for money, M1, and the price level together
with the rate of interest determine the asset demand for cash. Total demand is
M ≡ M1 + M2.
Unemployment is measured by the gap between (1) the level of output that
would be produced if all workers available at the equilibrium real wage W0/P0
were employed and (2) the level of output at the equilibrium of aggregate de-
mand and goods supply: the output gap is
P
LS − Y0 ,
W
where LS(P/W) is the level of output that workers would like to produce at the
real price P/W. Tat is,
P P
LS = F K , L ,
W
W
with L(P/W) the amount of labor workers wish to supply at the price P/W, as-
sumed to depend on the utility of goods and the disutility of work.
The Second-Pass Model
Tis model builds on the frst-pass model but reverses the causality between
money and interest. Te interest rate is determined endogenously by equating
the total demand for money M to an exogenous supply M. We have
I( ) = sY,
P
Y = GS ,
W
Copyright © 2021. Harvard University Press. All rights reserved.
β( ρ)
M ≡ αPY + P ( ρ)PK = M .
1 − β( ρ) B
Once again, W = W0, and equilibrium unemployment is given by the output
gap LS(P0/W0) − Y0. As observed in the text, this system of equations, unlike
the frst-pass system, is not decomposable, and the level of output is not de-
termined by aggregate demand alone. Te interest rate in the aggregate-
demand equation now depends, via the money equation, on the unknowns Y
and P.
Te IS-LM construction in the text varies P parametrically to derive aggre-
gate demand from the intersection in Y × space of the IS schedule, defned
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
3 • Mathematical Appendix 121
by I = S, and the LM schedule, defned by M ≡ M. Te negative slope of the IS
schedule is confrmed by diferentiating with respect to Y:
d s
= ,
dY I′
with I′ ≡ dI/d. Te derivative d/dY is negative because I′ is negative. Simi-
larly, diferentiating the LM schedule gives
dρ α
=− ,
dY β′PB βPB′
2
+ K
(1 − β) 1− β
which is positive since ′ ≡ d/d and PB′ ≡ dPb/d are both negative. Diferen-
tiating the IS and LM schedules with respect to P shows that the AD schedule
slopes downward in Y × P space. We have
d dY
I′ =s
dP dP
and
dY d ρ β′PB βPB′ β
αY + αP + 2
+ PK + PB K = 0.
dP dP (1 − β) 1− β 1− β
Substituting (s/I′)(dY/dP) for d/dP gives
dY M
=− .
dP 2 s β′PB βPB′
P α + + K
I ′ (1 − β) 2
1 − β
Since both the numerator and the denominator are positive—I′, ′, and PB′ are
all negative—we have dY/dP < 0.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 4 .
equilibrium with a given
money supply
Critical Perspectives on the Second-Pass Model
It is usually considered as one of the most important achievements
of the Keynesian theory that it explains the consistency of economic
equilibrium with the presence of involuntary unemployment. It is,
however, not suf ficiently recognized that, except in a limiting case to
be considered later, this result is due entirely to the assumption of
“rigid wages.”
—franco modigliani
When prices and wages fall, more and more money is released from
transaction duties. Idle funds grow in terms of money and still faster
(without any limit even if the quantity of money remains unchanged
or decreases) in real terms. . . . When money hoards (inactive depos-
its, Mr KEYNES’ M2) have reached a certain level people will stop
saving (that is, in this case, they will cease to add to their hoards).
Tat amounts to saying that the rate of saving is not only an increas-
ing function of the level of (real) income, but also a diminishing
function of the wealth the individual holds.
—gottfried haberler
Shortfalls in overall demand would cure themselves if only wages and
Copyright © 2021. Harvard University Press. All rights reserved.
prices fell rapidly in the face of unemployment.
—paul krugman
Has Keynes made good on the claim that, with a given money supply, a com-
petitive market economy may be at equilibrium but not, in the absence of ac-
tivist monetary or fscal policy, able to provide a job for every willing worker?
Te equilibrium level of output in Figure 3.27 is 400, and the unemployed
workers would add 100; so the economy, according to Keynes, is at equilib-
rium but not at full employment.
However, anyone steeped in mainstream microeconomic theory would
122
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 123
have dif ficulty accepting the aggregate demand-goods supply intersection at E
as a true equilibrium. Figure 3.27 assumes a given money wage; what if the
money wage were lower? Before we examine this question in the framework
of the second-pass model, it serves as a useful benchmark to put the question
in the framework of the frst-pass model.
In this case, the supply side determines only the price level; the level of the
money-wage rate makes no diference at all for the level of output. Suppose
we compare, as in Figure 4.1, two equilibria, one (E) for the original wage W0,
the second (E′) for a wage half as great, 0.5W0. All that changes is the price
level. With the wage 0.5W0 instead of W0, the equilibrium price level is 1.75
instead of 3.5.
In order to increase employment and output, it is necessary to stimulate
aggregate demand. Moving the AD schedule from Y = 400 to Y = 450, as in
Figure 4.2, leads to full employment, whether the wage is W0 or 0.5W0. Once
again, all that changes is the equilibrium price level, which is 4 if the wage is
W0 and 2 if the wage is 0.5W0.
Te second-pass model behaves quite diferently. With the addition of the
LS schedule, the model, like its frst-pass counterpart, is overdetermined, but
with this key diference: the AD schedule is now downward sloping, as in Fig
ure 4.3.
On the assumption that the wage rate is fxed, at most two of the three rela-
5
Supply of goods
4
E
Copyright © 2021. Harvard University Press. All rights reserved.
2 E
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
W = W 0 (solid GS and LS) W = 0.5W 0 (dashed GS and LS)
Figure 4.1 Aggregate demand, goods supply, and labor supply in the frst-
pass model, I.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
124 Raising Keynes
Supply of goods
4
2
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
W = W 0 (solid GS and LS) W = 0.5W 0 (dashed GS and LS)
Figure 4.2 Aggregate demand, goods supply, and labor supply in the frst-
pass model, II.
5 Supply of goods
Aggregate demand
4
E
3
Copyright © 2021. Harvard University Press. All rights reserved.
2
Supply of labor
1
100 200 300 400 500 600 Y
W = W0
Figure 4.3 An overdetermined system, I.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 125
tionships pictured in Figure 4.3 can be satisfed for any combination of Y
and P, barring a serendipitous coincidence. Keynes’s case for privileging the
AD = GS equilibrium over the other two possibilities is that the LS schedule
does not exist; he rejects the argument that a labor-market equilibrium re-
quires that the wage rate be equal to the marginal disutility of labor (see ap-
pendix 1 to chapter 3).
If however we stipulate that changes in money wages are general rather
than piecemeal, Keynes’s reason for suppressing the LS schedule loses its
force. And if we are really trying to situate the argument in an economy with-
out frictions and imperfections, we have no good reason to ignore the exis-
tence of the LS schedule.
Enter Franco Modigliani. Modigliani allows the money wage to vary (in
God’s mind) while holding the money supply constant. What appears to be
a minor change in the model dramatically changes the main result: in Modi-
gliani’s version of the second-pass model, barring a liquidity trap or a com-
pletely inelastic investment demand in the neighborhood of a zero interest
rate, equilibrium requires full employment. Since the only thing Modigliani
does to Keynes’s second-pass model is to substitute an LS schedule for a fxed
money wage, it is an easy leap to the conclusion that the reason why Keynes
obtains an unemployment equilibrium is money-wage rigidity.
Observe, frst of all, that, although Modigliani c omes to the same conclu-
sion as the composite “classical” economist whom Keynes identifed with
Pigou, he comes to this conclusion by a diferent, “Keynesian,” route. Modi
gliani does not assume the classical dichotomy that allows the economy to be
analyzed with money playing no more of a role than a medium of exchange
and a unit of account. Modigliani’s is a money economy, like Keynes’s, com-
plete with investment demand depending on an interest rate that clears the
market for fnancial assets. Like Keynes, he locates the determination of
the hurdle rate of interest on the fnancial side of the model, not in the
investment-saving nexus.
Copyright © 2021. Harvard University Press. All rights reserved.
It is through the dependence of the interest rate on the supply of money
available for fnancial markets that Modigliani is able to restore the pre-
Keynesian invisible hand. Te key to Modigliani is that a lower money wage is
associated with lower levels of prices, a smaller transactions demand, more
money in wealth portfolios and, therefore, a lower interest rate and a higher
level of investment—the Keynes efect.
Modigliani accepts the mainstream view that the level of employment is
determined by the intersection of GS and LS schedules, both of which depend
on the real price, P/W. In other words, GS and LS schedules jointly determine
both the equilibrium level of output (employment) and the real price. But the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
126 Raising Keynes
real price is a ratio, P/W; so once the money-wage rate is freed from its moor-
ings, the price level is also free to vary.
What then determines the price level (and, consequently, the wage level)?
Modigliani assumes that the nominal price of output settles at the level at
which the interest rate coaxes out just the right amount of investment, namely,
the investment that generates aggregate demand equal to the output deter-
mined by the intersection of GS and LS schedules.
Te logic is straightforward. First, Y and P/W are determined by the inter-
section of GS and LS schedules so that full employment obtains and the real
price is determined. But P and W are separately up for grabs. To determine P
we turn to the investment-saving nexus, I( ) = sY, and the money demand–
supply relationship, the equation M1 + M2 = M. With Y now a known quan-
tity, the relationship I( ) = sY determines I and . Once we know ,
the nominal price level falls out from the equations relating transactions de-
mand to real output, M1 = PY, and fnancial–market demand to the rate of
interest,
β( ρ)
M2 = P ( ρ)PK .
1 − β( ρ) B
We have money supply equal to money demand,
β( ρ)
M = αPY + PB ( ρ)PK .
1 − β( ρ)
Since the money supply is given, we can derive P from this equation as
M
P = .
β( ρ)
αY + PB ( ρ)K
1 − β( ρ)
Now that we have P, we can determine W from the equilibrium real price
Copyright © 2021. Harvard University Press. All rights reserved.
P/W.
Graphically, shifs in the GS and the LS schedules resolve the overdetermi-
nation in Figure 4.3. In Figure 4.4, the quantity of goods supplied for P = 1 on
the dashed schedule, namely Y = 400, is the same quantity that is supplied for
P = 3.5 on the solid schedule because both of these prices correspond to the
same real price. If we assume W0 = 1, then in both cases P/W = 3.5. Te same
is true for labor supply: Y = 500 in both cases.
Alternatively, we can picture the GS and LS schedules in relation to the
price level assuming the real-price level, P/W = 3.5, is fxed—as in Figure 4.5.
Evidently, there are two diferent concepts of the “price level” at work,
which suggests an inherent ambiguity. Up to now we have represented the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 127
5 Supply of goods
Aggregate demand
4
E
3
2
Supply of labor
1 F
100 200 300 400 500 600 Y
W = W 0 1 (solid GS and LS) W = 1 / 3.5 (dashed GS and LS)
Figure 4.4 An overdetermined system, II.
Supply of goods
4
Supply of labor
3
Copyright © 2021. Harvard University Press. All rights reserved.
100 200 300 400 500 600 Y
P
= 3.5
W
Figure 4.5 Goods supply and labor supply with the real price at 3.5.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
128 Raising Keynes
price level as P varying relative to a fxed money wage, W0, or in the case of
Figures 4.1, 4.2, and 4.4, two diferent money wages, W0 and W0/3.5. By con-
trast, in Figure 4.5, the nominal price level varies while the real price is fxed,
which is to say that the wage rate is assumed to vary in the same proportion as
all other prices.
Te price level is a composite, an index of the money values of the goods
and services that the economy produces. Varying the price level evidently
means varying all prices proportionately. Te ambiguity stems from how we
conceptualize labor, whether as diferent in kind from other goods and ser
vices, or as simply one more commodity. Treating labor diferently, the price
level measures prices independently of wages; treating labor the same as other
commodities, with its nominal price—the money wage—varying in strict
proportion to other prices, the price level measures what we will call the abso-
lute price level, the price level on the assumption that the real price (and the
real wage) is fxed.
Because the AD schedule is independent of the wage level, it is easy to
move between the two diagrams, Figures 4.4 and 4.5.1 Figure 4.6 superim-
poses the AD schedule on Figure 4.5; the points E and F represent the same
combinations of prices, wages, and outputs that are represented by E and F in
Figure 4.4.
Supply of goods
5
Aggregate demand
4
Supply of labor
Copyright © 2021. Harvard University Press. All rights reserved.
E
3
1 F
100 200 300 400 500 600 Y
P
= 3.5
W
Figure 4.6 Equilibrium holding the real price at 3.5.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 129
But what has been accomplished? Simply specifying a wage rate difering
from the original one, or specifying a particular real price, leaves us with the
same problem of overdetermination as before. In both Figure 4.4 and Figure
4.6, we still have three relationships to be satisfed by two variables.
Since in Modigliani’s model the real price is a variable, and the full-
employment, proft-maximizing level of output is associated with a specifc
real price, we can choose both an absolute price level P, on which aggregate
demand depends, and a real price level P/W, on which full employment and
proft maximization depend. All we have to do is to vary the money-wage
rate. In the present case, at a real price P/W = 4, we have goods and labor sup-
plies in sync with one another—meaning we have both full employment and
proft maximization—at an output of 450. To achieve both, the nominal price
level has to be set at the level that generates an aggregate demand of 450. Fig
ure 4.7 represents the required nominal price level as 2, which is to say that
the wage level is 0.5. Te ambiguity of Figure 4.3 is resolved because the
schedules converge on a single equilibrium; G is a point of serendipity at
which the economy is simultaneously on its AD, GS, and LS schedules.
Te same result can be pictured in the context of a variable absolute price
level. Figure 4.8 assumes P/W = 4, which makes the GS and LS schedules co-
incide. Tis is of course the vertical aggregate-supply schedule reproduced in
5 Supply of goods
Aggregate demand
4
E
3
Copyright © 2021. Harvard University Press. All rights reserved.
2 G
Supply of labor
1 F
100 200 300 400 500 600 Y
W = W0 1 (solid GS and LS)
W = 1 / 3.5 (dashed GS and LS)
W = 0.5 (dotted GS and LS)
Figure 4.7 Aggregate-demand, goods-supply, and labor-supply schedules
for three diferent wage rates.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
130 Raising Keynes
virtually every elementary text as representing long-run supply conditions. It
would no doubt help students if they were told that this supply schedule
comes from going back to the old, pre-Keynesian, view of output being deter-
mined by proftability (the GS schedule) and workers’ goods-leisure prefer-
ences (the LS schedule)! As it is, most students—and perhaps their teachers
too—are in a fog about what assumptions must be made to make the supply
schedule vertical, and how it is that labor and goods supplies, conceptually
distinct, come together in a single schedule. Te key is to assume full employ-
ment and proft maximization, so that instead of two separate supply sched-
ules, we have a single “aggregate” supply, which is at one and the same time
the supply of goods refecting proft maximization and the supply of labor
required to produce those goods. Superimposing aggregate demand on Fig
ure 4.8, the serendipity of a full-employment equilibrium is refected in the
point G in Figure 4.9.
Observe that we have come full circle. With a given money supply, but the
price level and the wage level variable, there exist prices and wages compatible
with full employment. Moreover, supply conditions uniquely determine the
level of output. Aggregate demand has no bearing on equilibrium output, en-
tering the picture only to determine nominal prices and wages. Keynes seems
to have unleashed a tempest in a teapot, at least insofar as the question is
Aggregate supply
P
Supply of goods
5
4 Supply of labor
Copyright © 2021. Harvard University Press. All rights reserved.
450
100 200 300 400 500 600 Y
P
= 4
W
Figure 4.8 Goods supply and labor supply, P/W = 4.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 131
the existence of full-employment equilibrium in a competitive economy un-
marred by imperfections and frictions—except for one, or rather two, special
cases.
In chapter 3 it was observed that investment demand can be insuf ficient to
reach a full-employment equilibrium either because the interest rate cannot
be brought to a suf ficiently low level regardless of how much money is avail-
able to satisfy the asset demand for money—the liquidity trap—or because
even at a zero hurdle rate of interest an inelastic investment-demand sched-
ule provides too little investment demand. Figure 4.10, which reproduces a
slightly modifed Figure 3.16, shows both cases; in the liquidity-trap case, in-
vestment demand is limited to 60 and aggregate demand to 240, and, in case
of an inelastic investment schedule, maximum investment demand is 80 and
the corresponding aggregate demand is 320.
Te consequences for aggregate demand are shown in Figure 4.11. In both
cases even a price level near zero would not lead to suf ficient aggregate de-
mand to employ the available work force. So, even afer Modigliani, Te Gen-
eral Teory retains some relevance as an argument about the limits of the
self-regulating mechanism of the market as well as the limits of monetary
policy. But this can provide little comfort for Keynes, for the special cases are
limited to a foor on the hurdle rate of interest, which he specifcally dis-
Supply of
P labor
5 Supply of goods
4
Copyright © 2021. Harvard University Press. All rights reserved.
2 G
1 Aggregate demand
450
100 200 300 400 500 600 Y
P
= 4
W
Figure 4.9 Full-employment equilibrium with P/W = 4.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
ID ID , SD
400
300
II I
200
SD
100 80
Marginal efficiency of capital ID
60
Interest rate ( ) Income (Y)
240 320
Interest rate ( ) min 100 200 300 400 500 Income (Y)
100 400
200 300
Liquidity preference 300 200
400 100
Transactions demand, limiting case as P 0
500 0
III M2 M2 M1 M 1 IV
Figure 4.10 Aggregate demand limited by liquidity preference and the marginal efficiency of
capital.
AD limited by marginal
efficiency of capital
P schedule
Supply of goods
5
Copyright © 2021. Harvard University Press. All rights reserved.
4 Supply of labor
3 AD limited by
liquidity
2 preference
schedule
240 320 450
100 200 300 400 500 600 Y
P
= 4
W
Figure 4.11 Limits to full-employment equilibrium.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 133
avowed as a theoretical possibility with no historical traction, and stagnation
of investment demand, which he thought to be a distant prospect rather than
an immediate threat.
What are we to make of Modigliani’s critique? In one sense there is nothing
new in Modigliani’s claim that there is a money-wage rate that is consistent
with full employment, at least apart from the exceptional circumstances pic-
tured in Figure 4.11. Keynes, had he lived long enough and found the time to
answer Modigliani, might have claimed that Modigliani exaggerated the dif-
ferences between them. As Keynes puts it immediately afer introducing the
Keynes efect,
We can, therefore, theoretically at least, produce precisely the same efects
on the rate of interest by reducing wages, whilst leaving the quantity of
money unchanged, that we can produce by increasing the quantity of money
whilst leaving the level of wages unchanged. (Te General Teory, p. 266)
Keynes goes on to note that the same limitations that apply to monetary pol-
icy—a foor to the hurdle rate or an inelastic investment-demand schedule—
might prevent a lower wage from having the desired efect on employment:
It follows that wage reductions, as a method of securing full employment,
are also subject to the same limitations as the method of increasing the
quantity of money. Te same reasons as those mentioned above, which limit
the efficacy of increases in the quantity of money as a means of increasing
investment to the optimum fgure, apply mutatis mutandis to wage reduc-
tions.
Te key point is that the money wage operates on aggregate demand in the
same way as the money supply—in reverse. To be sure, the efects are dia-
grammed diferently in the space of price level and real output. In the case of
an increase in the money supply, aggregate demand moves outward, as in Fig
ure 4.12. By contrast, reducing the money wage is depicted as a move along
Copyright © 2021. Harvard University Press. All rights reserved.
the AD schedule as the GS (and LS) schedules move downward with the
money wage, as in Figure 4.13. But the efect on output is the same, and for
the same reason: a larger nominal money supply increases the real supply of
money M/P, and so does a lower nominal wage via its efect on the price level.
Since the LM schedule is defned by the equality of money demand and
money supply, the equation for the real money supply, M/P, is
M β( ρ)
= αY + P ( ρ)K .
P 1 − β( ρ) B
Raising M moves the LM schedule in the same direction as reducing P. In
short, both wages and money supply operate on the rate of interest, and the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
134 Raising Keynes
Supply of goods
5
Aggregate demand
4
E F
3
2
Supply of labor
1
100 200 300 400 500 600 Y
W = W0
Figure 4.12 Aggregate demand, goods supply, and labor supply:
increasing the money supply.
Aggregate
5 demand Supply of goods
4
E
3
F
2
Supply of labor
Copyright © 2021. Harvard University Press. All rights reserved.
100 200 300 400 500 600 Y
W = W 0 (solid GS and LS) W = W1 (dashed GS and LS)
Figure 4.13 Aggregate demand, goods supply, and labor supply: reducing
the real wage.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 135
rate of interest operates on aggregate demand. We are a long way from the
mainstream emphasis on a direct line of causality running from the real wage
to employment and output, the mechanism of chapter 2.
It may be satisfying from the point of view of intellectual history that
Keynes saw that the nominal wage and the money supply operate in the same
way, through the interest rate.2 But there is nonetheless a crucial theoretical
diference between the two: wage-rate adjustment is supposed to operate au-
tomatically, whereas changing the money supply requires the visible hand of a
central bank. Given this diference, it is not surprising that Modigliani’s argu-
ment had an important efect on the evolving discussion of what were and
what were not the essential assumptions of Te General Teory. Modigliani’s
1944 paper inficted a grievous wound to the claim that Te General Teory
disproved the automatic tendency of a competitive market economy to pro-
vide a job for every willing worker. Te wound continues to cause pain, as the
Krugman epigraph to this chapter indicates.
Is Modigliani’s Critique Compatible with
Endogenous Money?
Modigliani claims that a competitive market economy has an equilibrium at a
full-employment level of output. In making this claim Modigliani assumes,
along with Keynes, a fxed money supply. What is the status of this assump-
tion? Is it merely part of the scafolding? Or is this assumption essential to the
conclusion that a full-employment equilibrium exists if only we allow wages
and prices to vary? Specifcally, what happens to aggregate demand if we as-
sume fractional-reserve banking, so that the money supply can vary endoge-
nously even in the absence of a central bank?
Since the IS schedule is determined by real variables, the answer lies in the
behavior of the LM schedule. As a benchmark, look again at the LM schedule
when there are no banks and money is simply a predetermined quantity of
Copyright © 2021. Harvard University Press. All rights reserved.
Venetian gold ducats. In the simplest version, there are two classes of eco-
nomic agents: households and restaurateurs. Households account for all the
net wealth in the economy and deploy their money in two ways. Tey lend
money to restaurateurs (perhaps themselves, wearing diferent hats) for work-
ing capital to carry out the transactions necessary in the restaurant business—
buying the ingredients for tonight’s dinner. Te rest of their money forms the
liquid part of their wealth. Te other part of their wealth is a stock of bonds
issued by restaurants as the fnancial counterpart of the stock of physical cap
ital—structures, furnishings, equipment.
Assuming P = 0.75 and = 1 (−1 is the transactions velocity of money),
every ducat’s worth of the fow of output requires a stock of 0.75 ducats. If
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
136 Raising Keynes
Y = 400, 300 ducats are required to grease the wheels of commerce. If M =
500, and M1 = 300, there are 200 ducats available to satisfy the asset demand
for money. For fnancial markets to be in equilibrium, the interest rate must
be equal to 0, as we observed in constructing the LM schedule from transac-
tions-and asset-demand schedules. Te relevant graphs are reproduced in
Figure 4.14. Liquidity preference is given by the solid curve in the third quad-
rant.
We assume the real capital stock is 1,333, so that with P = 0.75, the nominal
value of bonds is 1,000. Assume that the bond coupon is equal to the interest
rate, R = 0, so that PB( 0) = 1.
If we posit a diferent output level, or a diferent price level, the economy
will need more or less gold to carry out its transactions, and there will be cor-
respondingly fewer or more ducats to satisfy asset demand. Te consequence
is that the interest rate that equilibrates fnancial markets will be higher or
lower. If, for example, Y = 266.67, with = 1 and P = 0.75, the economy will
need only 200 ducats for commerce. For households to be willing to hold the
remaining 300 ducats as part of their asset portfolios, the interest rate has to
be lower, 1 in the diagram. With R fxed, the bond price will be higher. If we
assume that 1 = 0/1.2, the bond price increases to PB( 1) = 1.2.
If the price of output (and capital) is 1.0 instead of 0.75, one ducat is needed
to fnance one unit of output, and 400 ducats are required to sustain an output
Interest rate ( ) Interest rate ( )
P=1
P = 0.75
LM
Copyright © 2021. Harvard University Press. All rights reserved.
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 2 0 1 100 200 300 400 500 Income (Y)
100 400 P=1
Table 4.3
200 300
Liquidity preference Table 4.1 P = 0.75
300 200
Table 4.2
P=1
400 100 Transactions demand
P = 0.75 500 0
M2 M2 M1 M 1
Figure 4.14 Financial equilibrium: constructing the LM schedule, I.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 137
of 400. In this economy only 100 ducats are lef over to satisfy asset demand,
necessitating a higher interest rate, 2. Te transactions-demand schedule in
Figure 14.4 shifs from the solid to the dashed line.
With a higher price level, there is not only a shif along the original
liquidity-preference schedule because more ducats are required to grease the
wheels of trade, there is also a shif of the liquidity-preference schedule be-
cause at a higher price level the nominal value of fxed capital is higher and,
correspondingly, given R, there must be a larger stock of bonds. But since R is
fxed, a higher interest rate means a lower bond price—say, PB( 2) = 0.675.
Te market value of bonds is now 900. As a result of the shifs of the
transactions-demand and liquidity-preference schedules, the LM schedule
also shifs; the dashed line replaces the solid line.
Tables 4.1, 4.2, and 4.3 summarize the three snapshots of the economy. For
accounting purposes, restaurants are treated as entities for which assets and
liabilities balance—so that their net worth is zero.
Table 4.1 Y = 400; P = 0.75; = 1; = 0; PB = 1
Households Restaurants
Loans to
Nonliquid Assets Nonliquid Assets Working Finance
Liquid Assets (Bonds—Market (Loans to Restaurants Capital (Gold Fixed Working Quantity of
(Gold Ducats) Value) for Working Capital) Ducats) Capital Capital Bonds
200 1,000 300 300 1,000 −300 −1,000
Table 4.2 Y = 266.7; P = 0.75; = 1; = 1 = 0/1.2; PB = 1.2
Households Restaurants
Loans to
Nonliquid Assets Nonliquid Assets Working Finance
Copyright © 2021. Harvard University Press. All rights reserved.
Liquid Assets (Bonds—Market (Loans to Restaurants Capital (Gold Fixed Working Quantity of
(Gold Ducats) Value) for Working Capital) Ducats) Capital Capital Bonds
300 1,200 200 200 1,000 −200 −1,000
Table 4.3 Y = 400; P = 1; = 1; = 2 = 0/0.675; PB = 0.675
Households Restaurants
Loans to
Nonliquid Assets Nonliquid Assets Working Finance
Liquid Assets (Bonds—Market (Loans to Restaurants Capital (Gold Fixed Working Quantity of
(Gold Ducats) Value) for Working Capital) Ducats) Capital Capital Bonds
100 900 400 400 1,333 −400 −1,333
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
138 Raising Keynes
All this is straightforward—except for transactions money. Bonds are the
counterpart of physical capital, structures, furnishings, machinery, and so
forth. Indeed, the separation of wealth management from restaurant opera-
tion was introduced precisely to create a role for bonds as fnancial assets.
Tis separation logically obliges household balance sheets to take account of
the cash lent to restaurants for working capital, that is, for transactions pur-
poses. Tables 4.1, 4.2, and 4.3 classify these loans as nonliquid fnancial assets
of the lenders, which is to say assets akin to bonds, except perhaps for the
length of time for which working capital is borrowed and lent. At the same
time we assume that these loans have no impact on wealth management. Tey
exist in a kind of limbo.3
Commercial loans apart, there is no conceptual dif ficulty in assuming an
exogenous money supply of 500 that gets distributed between transactions
demand and asset demand with the interest rate adjusting to clear fnancial
markets according to the relative supplies of liquid and nonliquid assets. But
what happens if the Venetian money supply does not consist only of gold duc-
ats? Can we, in the context of a fractional-reserve banking system, sustain the
assumption of an exogenous money supply when part of it is money created
by the banks? Suppose that instead of 500 gold ducats there are only 200 coins
in existence. Suppose further that households deposit these coins in banks, all
of which operate under laws (or customs) that require them to hold reserves
of at least 0.4 gold ducats for each ducat of deposits.
When fully loaned up, the banks can maintain deposits equal in value to
500 ducats against the reserves of 200 gold ducats. In this context, to argue that
the money supply is fxed at 500 is to argue that banks will always be fully
loaned up—and that they have no way around reserve requirements.
Tere are now three types of players to contend with: in addition to house-
holds and restaurateurs, there are bankers, who act as intermediaries between
the original two groups. Households and restaurateurs may be the same peo-
ple wearing diferent hats, but this assumption undermines one of the pur-
Copyright © 2021. Harvard University Press. All rights reserved.
poses of a banking system, which is to form a bridge between households
with assets to deploy and businesses who need access to these assets to carry
on their businesses. If we are going to assume the existence of banks, it now
makes more sense to think of wealth holders and restaurateurs as two difer-
ent groups of people.
What happens to the three snapshots in Tables 4.1, 4.2, and 4.3 under the
new dispensation? If P = 0.75 and Y = 400, then the total transactions demand
will again be 300 ducats. Te 200 ducats that represent the deposits of house-
holds satisfy their demand for liquid assets. In this case, fractional-reserve
banking changes very little. PB = 1 and = 0 in Table 4.4, just as in Table 4.1.
And we can tie up the loose end of commercial lending: the anomalous sta
tus of working capital disappears when banks are assumed to be intermediar-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 139
ies between households and business. Commercial loans now refect money
created by the banking system, so-called inside money, that is an asset of the
banks and a liability of the restaurants. Its net contribution to wealth is zero.
Now construct Table 4.5 with Y = 266.67 and P = 0.75, as in Table 4.2.
Transactions demand is 200 ducats, 100 ducats less than in Table 4.4 (and
Table 4.1). At = 0 total money demand, M1 + M2, is now 400 rather than
500. What happens to the money supply now depends on the banks. One pos-
sibility is that bankers actively seek outlets for their unused lending capacity
of 100 ducats, which is to say that banks enter the market for bonds. To in-
duce wealth holders to part with their bonds, the bond price must rise. Asset-
market equilibrium is still achieved by balancing liquidity and yield, and the
equilibrating mechanism is still the interest rate, refected in the price of
bonds. But the equilibrium rate of interest must be lower in Table 4.5 than
in Table 4.2 if wealth holders’ liquidity preferences are unchanged. For, hav-
ing sold bonds worth 100 ducats to the banks, they now have a smaller num-
ber of bonds, but, as in Table 4.2, 300 ducats in cash. Te value of their
bonds must be less than 1,200 ducats. Why? Because if wealth holders possess
1,200 ducats worth of bonds, then the total value of all bonds, including the
banks’ share, would have to be 1,300, and so the price of bonds would be 1.3
(= 1,300/1,000). But with PB = 1.3, the interest rate would be lower than 1,
which is the interest rate associated with PB = 1.2 in Table 4.2. Tis contradic-
tion—the composition of wealth is the same in the two cases—means that the
value of wealth holders’ bonds cannot be 1,200 in both cases.
If the market value of bonds in wealth holders’ hands is less than 1,200, say
1,150, then the ratio of their liquid wealth (300) to their bond wealth (1,150)
is higher in Table 4.5 than in Table 4.2. Compared with Table 4.2, the interest
rate is lower, = 3 = 0/1.25, and the price of bonds is higher, PB = 1.25.
Banks now own bonds with a market value of 100 ducats.
Te novelty lies in the entry of banks into the bond market, which moves
the liquidity-preference schedule (see Figure 4.15, where the new schedule is
Copyright © 2021. Harvard University Press. All rights reserved.
shown with dots). A real-time story—the entry of banks into the bond market
increases the overall demand for bonds and leads to a higher bond price and
a lower interest rate—is tempting. But in a comparative-statics framework, all
we can say is that an asset-market equilibrium with fewer bonds in the hands
of wealth holders requires a higher bond price and a lower yield than an equi-
librium in which there are more bonds in the possession of wealth holders.
Assume now Y = 400 and P = 1, so that one ducat of transactions demand
accompanies each unit of output. Tere is now no way to meet the demand
generated by Y = 400 unless households are willing to accept lower levels of
liquid assets in their portfolios. In efect, households have to be willing to buy
100 ducats of commercial loans from the banks.
Here we face the same question that was elided in the previous fnancial
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
140 Raising Keynes
model, without banks: how do we classify the 100 ducats of commercial loans
owned by households? Again, we have two alternatives: either we classify
these loans as illiquid fnancial assets or we consider these loans to be like
cash. Tere are real-world analogies for both classifcations. Assume that
in both cases the banks securitize these loans, in one case as a CDO (for col-
lateralized debt obligation), in the other as the assets of a money-market
fund. In the frst scenario, CDOs average out the idiosyncratic risk of indi-
vidual loans but leave households with important systemic risks. Tis sce-
nario would make the CDO more like a bond than like cash. In the alternative
scenario, we suppose that the residual systemic risks are small enough that
the loans can be packaged in a liquid form—a money-market fund—that
households treat like cash. In this case we can imagine the money-market
fund selling shares to households and using the proceeds to buy commercial
paper issued by restaurateurs.
Either way, we assume that the issuers of notes against working capital, the
restaurants, maintain their bank deposits—their obligation to pay money for
the ingredients in tonight’s dinner is not changed. What then is the money
supply? Under narrow defnitions that limit money to cash and demand de-
posits, the money supply remains 500, 400 ducats of business deposits and
100 ducats of deposits belonging to households. But if the assets sold by banks
to households end up in money-market funds, a broader defnition of money
would also include the 100 ducats households are now assumed to lend indi-
rectly to restaurants as working capital. Despite the reserve requirement of 40
percent, 200 gold ducats get multiplied into 600 ducats!
We can summarize the three snapshots of the fnancial markets in the fol-
lowing tables. It is now bank intervention in asset markets that moves the
economy along the liquidity-preference schedules in quadrant three of Figure
4.14; households respond to bankers’ initiatives—at least if we step out of the
comparative-statics mold temporarily to tell a story of change in real time.
In this story the negative relationship between the proportion of bonds in
Copyright © 2021. Harvard University Press. All rights reserved.
households’ portfolios and the interest rate is the complement of banks’ pur-
suit of profts.
Te common feature of these snapshots is that in all cases banks are fully
loaned up. Te LM schedules corresponding to Tables 4.4, 4.5, and 4.6 are
shown in Figure 4.15. Te new feature is the dotted liquidity-preference sched-
ule, which corresponds to a goods-price level of 0.75 and the sale of bonds by
households to banks (Table 4.5). Table 4.7 is represented in Figure 4.16, with
a defnition of the money supply that includes the money-market funds.
How do we know that banks are always fully loaned up? Te answer is that
we must assume this to be the case, that idle reserves are anathema. With-
out this assumption the equilibrium condition M = M1 + M2 tells us noth-
ing about whether money demand accommodates itself to money supply, as
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 4.4 Y = 400; P = 0.75; = 1; = 0; PB = 1
Households Banks Restaurants
Nonliquid Assets Nonliquid Assets
Bank (Bonds—Market Commercial (Bonds—Market Commercial Bank Bank Fixed Commercial Quantity of
Deposits Value) Loans Value) Loans Reserves Deposits Deposits Capital Loans Bonds
4
200 1,000 0 0 300 200 −500 300 1,000 −300 −1,000
•
Table 4.5 Y = 266.67; P = 0.75; = 1; = 3 = 0/1.25; PB = 1.25
Households Banks Restaurants
Nonliquid Assets Nonliquid Assets
Bank (Bonds—Market Commercial (Bonds—Market Commercial Bank Bank Fixed Commercial Quantity of
Deposits Value) Loans Value) Loans Reserves Deposits Deposits Capital Loans Bonds
300 1,150 0 100 200 200 −500 200 1,000 −200 −1,000
Table 4.6 Y = 400; P = 1; = 1; = 2 = 0/0.675; PB = 0.675
Households Banks Restaurants
Nonliquid Assets Nonliquid Assets
Bank (Bonds—Market Commercial (Bonds—Market Commercial Bank Bank Fixed Commercial Quantity of
Deposits Value) Loans Value) Loans Reserves Deposits Deposits Capital Loans Bonds
100 900 100 0 300 200 −500 400 1,333 −400 −1,333
Equilibrium with a Given Money Supply
Table 4.7 Y = 400; P = 1; = 1; = 4 = 0/0.8; PB = 0.8
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Households Banks Restaurants
Nonliquid Assets Nonliquid Assets
141
Bank (Bonds—Market Commercial (Bonds—Market Commercial Bank Bank Fixed Commercial Quantity of
Deposits Value) Loans Value) Loans Reserves Deposits Deposits Capital Loans Bonds
100 1,067 100 0 300 200 −500 400 1,333 −400 −1,333
Copyright © 2021. Harvard University Press. All rights reserved.
142
Interest rate ( ) Interest rate ( )
P=1
P = 0.75
LM
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 2 0 3 100 200 300 400 500 Income (Y)
100 400 P=1
Table 4.6
Raising Keynes
200 300
Liquidity preference Table 4.4 P = 0.75
300 200
Table 4.5
P=1
400 100 Transactions demand
P = 0.75 500 0
M2 M2 M1 M 1
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Figure 4.15 Financial equilibrium: constructing the LM schedule, II.
4 • Equilibrium with a Given Money Supply 143
we have been assuming, or money supply responds passively to changes in
money demand.
Instead of assuming that banks are always at the upper limit of deposit:
reserve ratios, suppose that banks focus exclusively on providing short-term
working capital, never venturing into the bond market. Ten the money sup-
ply rather than its composition will change with the volume of loan demand.
When commercial-loan demand is lower, banks simply curtail their opera-
tions and hold excess reserves rather than replace these loans with purchases
of bonds. We then have Table 4.8 in place of Table 4.5. In Table 4.8 there is no
change in the public’s holdings of liquid and illiquid assets relative to the
original position in Table 4.4, no change in bond prices and interest rates. In
this case the aggregate money supply, M, as well as its composition, fuctuates
with the level of output and prices; when transactions demand is lower, the
reserve ratio will be higher and the money supply lower. Now 0 balances the
advantages of bonds and bank deposits for wealth holders independently of
transactions demand: the result is the horizontal LM schedule in Figure 4.17.
With a higher price level, say P = 1, and with the stock of bonds higher in
the same proportion, we have the results in Table 4.9. In order for households
to absorb the larger stock of bonds, a lower bond price—a higher interest
rate—is required. Since M2 is unchanged, the market value of bonds has to be
Interest rate ( ) Interest rate ( )
P=1
LM
Copyright © 2021. Harvard University Press. All rights reserved.
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 4 100 200 300 400 500 Income (Y)
100 500
200 400 P=1
Liquidity preference Table 4.7
300 300
P=1
400 200 Transactions demand
500 100
600 0
M2 M2 M1 M 1
Figure 4.16 Financial equilibrium: constructing the LM schedule with money-market funds.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 4.8 Y = 266.67; P = 0.75; = 1; = 0; PB = 1
Households Banks Restaurants
144
Nonliquid Assets Nonliquid Assets
Bank (Bonds—Market Commercial (Bonds—Market Commercial Bank Bank Fixed Commercial Quantity of
Deposits Value) Loans Value) Loans Reserves Deposits Deposits Capital Loans Bonds
200 1,000 0 0 200 200 −400 200 1,000 −200 −1,000
Interest rate ( ) Interest rate ( )
Figure 4.17 Financial
equilibrium: constructing
the LM schedule with
LM endogenous transactions
money, I.
P = 0.75
Raising Keynes
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
100 100
200 200
Tables 4.4 and 4.8 Table 4.8
Liquidity preference P = 0.75
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
300 300
Table 4.4
P = 0.75
400 400 Transactions demand
500 500
M2 M2 M1 M 1
4 • Equilibrium with a Given Money Supply 145
lower than in Table 4.8, while the higher interest rate translates into a lower
price of bonds. Observe that the higher price level shifs the LM schedule
upward in Figure 4.18.
So even though fractional-reserve banking fattens out the LM schedule
when banks respond passively to the transactions demand for money, the key
diference between the second-pass and the frst-pass models holds: aggregate
demand continues to be a downward sloping function of the price level. Ac-
cordingly, there will be a full-employment equilibrium if the money-wage rate
is up for grabs—at least barring the two exceptional cases that cancel the
negative relationship between the price level and aggregate demand.
Te causality is diferent. Tere is no Keynes efect; the smaller transactions
requirements of a lower price level do not release money that is incorporated
into asset portfolios. Tat avenue is closed of by the frewall between the two
kinds of money, implicit in the assumption that the supply of money is the
sum of a given amount of portfolio money and an amount of transactions
money determined by output and the price level. It is, rather, the assumption
that a change in the price level is accompanied by a change in the same direc-
tion of the stock of bonds that is responsible for the shif in the LM schedule.
Te result, a downward-sloping AD schedule, is the same because both the
bond efect and the Keynes efect change the proportion of cash and bonds in
the same direction, the bond efect by a positive relationship between the
price level and the stock of bonds, the Keynes efect by a negative relationship
between the price level and the stock of portfolio money.
Chapter 7 shows that there is no bond efect when changes in prices take
place in real time. Te consequences of assuming the money supply adjusts to
transactions demand turn out to be diferent in the context of the disequilib-
rium dynamics explored in the next three chapters. And if we go the next step
and assume that transactions demand and asset demand are complementary
rather than competitive uses of money, the complications are even more seri-
ous. As we shall see in chapter 13, in the absence of a central bank, the interest
Copyright © 2021. Harvard University Press. All rights reserved.
rate will in general be indeterminate!
The Real-Balance Efect
In a comparative-statics framework, Modigliani’s argument leaves one part of
Keynes’s edifce standing—some would say the most important part—in al-
lowing for the exceptions of a liquidity trap (the foor on the hurdle rate) or
an inelastic investment-demand schedule. An attack from another quarter, in
the form of a critique based not on the efect of the price level on the interest
rate and investment demand, but on the efect of the price level on consump-
tion demand, did not leave Modigliani’s exceptions standing.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 4.9 Y = 266.67; P = 1; = 1; = 5 = 0/0.6; PB = 0.6
Households Banks Restaurants
146
Nonliquid Assets Nonliquid Assets
Bank (Bonds—Market Commercial (Bonds—Market Commercial Bank Bank Fixed Commercial Quantity of
Deposits Value) Loans Value) Loans Reserves Deposits Deposits Capital Loans Bonds
200 800 0 0 266.67 200 −466.67 266.67 1,333 −266.67 −1,333
Interest rate ( ) Interest rate ( )
Figure 4.18 Financial
equilibrium: constructing
the LM schedule with
P=1
LM endogenous transactions
P = 0.75 money, II.
Raising Keynes
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 5 0 100 200 300 400 500 Income (Y)
100 100
Table 4.8
200 200
Liquidity preference Tables 4.8 and 4.9 P = 0.75
Table 4.9
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
300 300
P=1 Transactions demand
400 400
P=1
500 500
P = 0.75
M2 M2 M1 M 1
4 • Equilibrium with a Given Money Supply 147
Keynes always recognized the simplifcations involved in making con-
sumption depend solely on income, but what he regarded as a modest qualif
cation became in the hands of his critics a weapon to knock down the entire
enterprise of Te General Teory. Keynes’s consumption function was based
on the view that
Te expenditure on consumption in [real] terms . . . depends in the main, on
the volume of output and employment. (p. 96)
Tis justifes
summing up the other factors in the portmanteau function “propensity to
consume.” (p. 96)
Te takeaway, which serves as the basis for the simple consumption function
deployed in the frst-and second-pass models, is that
Men are disposed, as a rule and on the average, to increase their consump-
tion as their income increases, but not by as much as the increase in their
income. (p. 96)
(Keynes stresses that the marginal propensity to consume is likely to decline
with income, a complication we shall ignore until chapter 9.)
Keynes acknowledges that “other factors are capable of varying [the pro-
pensity to consume],” and he warns us parenthetically that “this must not be
forgotten” (p. 96). In particular, he has already pointed to the impact of wealth
on consumption, particularly surprise changes in wealth:
Te consumption of the wealth-owning class may be extremely susceptible
to unforeseen changes in the money-value of its wealth. Tis should be clas
sifed amongst the major factors capable of causing short-period changes in
the propensity to consume. (pp. 92–93)
And he reiterates that one reason for the propensity to consume to vary is
Copyright © 2021. Harvard University Press. All rights reserved.
that
A decline in income due to a decline in the level of employment, if it goes
far, may even cause consumption to exceed income . . . by some individuals
and institutions using up the fnancial reserves which they have accumu-
lated in better times. (p. 98)
As we shall see in chapter 9, the mainstream, under the tutelage of Milton
Friedman (1957) and Modigliani (in a series of papers, beginning with Modi-
gliani and Brumberg [1954]), turned Keynes’s views on consumption inside
out. What was for Keynes a qualifcation of the idea that the main determi-
nant of consumption is income became in the “permanent income hypothe-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
148 Raising Keynes
sis” and the “life-cycle hypothesis” a radically diferent view of consumption
in which wealth was the dog and income the tail.
For present purposes we need not go so far. Te real-balance efect attacked
Keynes not on the general grounds that consumption depends on total wealth
rather than on income, but on the specifcs of the supposed impact of the
price level on consumption through its efect on the cash component of
wealth. Te opening salvo was fred by Gottfried Haberler, a leading business-
cycle theorist whose 1936 summa of pre-Keynesian views, Prosperity and De-
pression, was rendered obsolete by Te General Teory even as the ink was
drying on the pages of the two books. A second edition of Prosperity and De-
pression, published in 1939, added a long chapter on Te General Teory and
the ensuing discussion, in which Haberler hinted at what was to become the
real-balance efect. But it was in the third edition, published in 1941, that
Haberler feshed this argument out. A generation later the real-balance efect
became, as I observed in chapter 1, the killer argument for an increasingly vo-
cal anti-Keynesian mainstream and the source of increasing bewilderment to
the shrinking band of brothers (and at least one sister) who had actually read
and digested Te General Teory.
What is the real-balance efect? To answer this question, we must back up
and examine how wealth impacts consumption and saving. Te simplest way
is to redraw the consumption and saving functions of the last chapter, replac-
ing the zero intercepts by nonzero intercepts—positive in the case of con-
sumption, negative for saving—as in the right-hand panels of Figures 4.19
and 4.20.
Wealth is assumed here to change both the amount of consumption and
saving that take place in the absence of any current income and the marginal
propensities to consume and save, that is to say, both the intercepts and the
slopes of the consumption and saving functions. What matters here is the in-
tercept: with a positive intercept, even people with no income can—tempo-
rarily at least—consume, which is to say that they can save a negative amount.
Copyright © 2021. Harvard University Press. All rights reserved.
And the more wealth, the more consumption is possible.
Te key to the real-balance efect is that some fraction of wealth is held in
the form of money, and the purchasing power of this money varies inversely
with the price level. Imagine that the price level is so low that a single thin
dime will buy a life of luxury, and you happen to fnd an old dollar bill stashed
in a dresser in the attic. Surely your consumption level will be much higher
than if you fnd that dollar bill with prices what they are today.
Graphically, the intercept in the right-hand panels of Figures 4.19 and 4.20
varies with the price level; the level of consumption at a zero level of income
increases and the level of saving decreases as the price level falls. Regardless of
what happens to marginal propensities, any amount of aggregate demand can
be generated by varying the price level. Haberler puts it like this:
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Consumption (C D ) Consumption (C D )
CD Y CD Y
500 500
400 C D 0.75Y 400 C D 1 00 0.5Y
300 300
4
Figure 4.19 Revising the
•
200 200 consumption function to refect
the infuence of wealth.
100 100
45° 45°
100 200 300 400 500 Income (Y) 100 200 300 400 500 Income (Y)
(a) Consumption depends (b) Consumption depends
only on income on income and wealth
Saving ( S D ) Saving ( S D )
500 500
400 400 Figure 4.20 Revising the
saving function to refect
300 300 the infuence of wealth.
200 200
Equilibrium with a Given Money Supply
S D = 0.25Y S D 1 00 0.5Y
100 100
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
100 200 300 400 500 Income (Y) 100 200 300 400 500 Income (Y)
149
−100
(a) Saving depends only on income (b) Saving depends on income and wealth
150 Raising Keynes
Pari passu with the fall in prices, existing money hoards (M2) rise in real
value and, sooner or later, the point will be reached where even the most
cautious individuals will fnd an irresistible temptation to stop hoarding and
to dishoard. (1939, p. 403)
In the 1941 edition, Haberler expands on this theme (the relevant passage
quoted as an epigraph to this chapter):
When prices and wages fall, more and more money is released from transac-
tion duties. Idle funds grow in terms of money and still faster (without any
limit even if the quantity of money remains unchanged or decreases) in real
terms. It was argued [in the passage cited above] that sooner or later when
money hoards (inactive deposits, Mr KEYNES’ M2) have reached a certain
level people will stop saving (that is, in this case, they will cease to add to
their hoards). Tat amounts to saying that the rate of saving is not only an
increasing function of the level of (real) income, but also a diminishing
function of the wealth the individual holds. (pp. 408–409)
Pigou (1943, 1947) independently developed the same idea, and Don Patin
kin (1948) extended the argument to include government debt.
In the context of comparative statics, the real-balance efect means that the
lower the price level, the greater aggregate demand—with no upper limit on
the level of demand. Unlike the AD schedules in Figure 4.11, the AD schedule
in Figure 4.21(a) will always intersect the aggregate-supply schedule provided
the absolute price level can be brought arbitrarily close to zero. Te same re-
sult is shown in Figure 4.21(b), with the price level varying with the money
wage at diferent levels. Because the level of wealth, and hence the level of
consumption demand, becomes infnite as the price level goes to zero, there is
no need to qualify the argument that the economy is self-regulating. Neither a
liquidity trap nor inelastic investment demand matters in this context.
We can see this by examining what happens to the various components of
Copyright © 2021. Harvard University Press. All rights reserved.
real wealth as the price level falls. Consider frst the case in which there are no
banks and money is specie. Nominal wealth is
PB ( )B + M 2 + M1 + −B + PK − M1 + Working Capital.
(Households) (Firms)
Since B = PK and M1 = Working Capital, the assets (PK + working capital) of
frms and their liabilities (B + M1) are exactly equal regardless of the price
level. So net fnancial wealth is equal to the assets of households,
PB( )PK + M2 + M1.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 151
Aggregate Aggregate
P demand P demand
Supply of
5 Supply of 5 goods
Supply of goods
4 labor 4
3 3
2 2
Supply of
1 1 labor
0.6 0.6
450 450
100 200 300 400 500 600 Y 100 200 300 400 500 600 Y
(a) Real price fixed at full-employment (b) Alternative levels of the
equilibrium level money wage
P W = W 0 1 (solid GS and LS)
= 4
W W = 0.5 (dotted GS and LS)
W = 0.1 5 (dashed GS and LS)
Figure 4.21 No limits to full-employment equilibrium: aggregate demand with a real-
balance efect.
Real wealth, which along with real income is now assumed to determine
real consumption, is obtained by dividing fnancial wealth by the price level.
We have
M M
PB ( )K + 2 + 1 .
P P
As the price level goes to zero, this expression becomes infnite.
In a regime of fractional-reserve banking, the calculations are diferent. We
have
PB ( )B + M 2 + M1 + M 2 − M1 − M 2 + −B + PK − M1 + Working Capital.
(Households) (Banks) (Firms)
Copyright © 2021. Harvard University Press. All rights reserved.
Once again we can ignore frms, because their assets and liabilities match.
Bond liabilities (B) are equal to the value of physical capital (PK), and bor-
rowing from banks (M1) is equal to working capital.
We can ignore banks as well, because their liabilities and assets also match.
For the banks, commercial lending to restaurants for working capital (M1) is
an asset, while the deposits of these same restaurants are a liability (M1 in
red). Likewise, the deposits of wealth holders (M 2 in red) are a liability for the
banks balanced by their reserves (M 2 in black).
In other words, as was pointed out by Michał Kalecki early on in the debate
(1944, p. 132) that followed Pigou’s formulation of the real-balance efect, so-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
152 Raising Keynes
called inside money created by the banking system does not constitute wealth
from the vantage point of the economy as a whole: every ducat of M1 in black
is exactly ofset by a ducat of M1 in red. We are lef with household wealth,
which now includes only asset money. Te result is the same as without
banks; real wealth
PB ( )B + M 2 M
= PB ( )K + 2
P P
becomes infnite as P goes to zero—and so does aggregate demand.
Taking Stock
Is the Haberler–Pigou–Patinkin real-balance argument the stake through the
heart of Te General Teory? Yes and no. On its own terms, yes. Already dis-
posed to interpreting Keynes’s economics in terms of rigidities and frictions,
mainstream economists did not have to accept the practical relevance of the
real-balance efect. All they had to do was to agree that, in a world without
imperfections, cash holdings would become more and more important for
consumption spending at lower and lower levels of prices—and eventually
provide enough demand to fully employ the labor force regardless of how low
investment demand might sink. Milton Friedman was refecting a consensus
in 1970 when he wrote that because of the real-balance efect there was, as a
matter of theory, no fatal faw in the price system:
[According to Keynes] as a purely theoretical matter, there need not exist,
even if all prices are fexible, a long-run equilibrium position characterized
by “full employment” of resources . . .
[Tis] proposition can be treated summarily because it has been demon-
strated to be false. Keynes’s error consisted in neglecting the role of wealth in
the consumption function or, stated diferently, in neglecting the existence
Copyright © 2021. Harvard University Press. All rights reserved.
of a desired stock of wealth as a goal motivating savings.5 All sorts of fric-
tions and rigidities may interfere with the attainment of a hypothetical long-
run equilibrium position at full employment; dynamic changes in technol-
ogy, resources, and social and economic institutions may continually change
the characteristics of that equilibrium position; but there is no fundamental
“faw in the price system” that makes unemployment the natural outcome of
a fully operative market mechanism6 (pp. 206–207).
5
Keynes, of course, verbally recognized this point, but it was not incor-
porated in his formal model of the economy. Its key role was pointed out
frst by Haberler (1941, pp. 242, 389, 403, 491–503) and subsequently by
Pigou (1947), Tobin (1947), Patinkin (1951), and Johnson (1961).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Equilibrium with a Given Money Supply 153
6
Tis proposition played a large role in gaining for Keynes the adher-
ence of many noneconomists, particularly the large band of reformers,
social critics, and radicals who were persuaded that there was something
fundamentally wrong with the capitalist “system.” Tere is a long history
of attempts, some highly sophisticated, to demonstrate that there is a
“faw in the price system” (the title of one such attempt [Martin 1924]),
attempts going back at least to Malthus. . . . But, prior to Keynes, these
attempts had been made primarily by persons outside of the mainstream
of the economics profession, and professional economists had little trou-
ble in demonstrating their theoretical faws and inadequacies. Keynes’s
attempt was therefore greeted with enthusiasm. It came from a profes-
sional economist of the very highest repute, regarded, and properly so, by
his fellow economists as one of the great economists of all time. Te ana-
lytical system was sophisticated and complex, yet, once mastered, ap-
peared highly mechanical and capable of yielding far-reaching and im
portant conclusions with a sminimum of input; and these conclusions
were, besides, highly congenial to the opponents of the market system.
Needless to say, the demonstration that this proposition of Keynes’s is
false, and even the acceptance of this demonstration by economists who
regard themselves as disciples of the Keynes of Te General Teory, has
not prevented the noneconomist opponents of the market system from
continuing to believe that Keynes proved the proposition, and continuing
to cite his authority for it.
Te main problem is that, like Modigliani, Haberler et al. relied on a
comparative-statics argument. None of these authors directly addresses the
issue of what happens when prices fall, as distinct from what happens when
prices are at a lower rather than a higher level and always have been.
Keynes fnally addresses the efects of real-time changes in the nominal
wage in chapter 19 of Te General Teory. Te problem is that he never for-
malized this third pass at a critique of self-regulation in the form of a model.
Copyright © 2021. Harvard University Press. All rights reserved.
Without the framework of a model, it was easy to ignore this chapter alto-
gether; it became possible for mainstream economists to disregard Keynes’s
clear statements about the provisional nature of the assumption of a given
money wage and to focus on this assumption as the distinguishing feature of
Te General Teory. Chapters 6 and 7 of this book provide the missing model,
afer what may appear to be a digression but is not: in chapter 5 we explore
the dynamics of adjustment. We start with the simple demand-supply models
of the nineteenth-century pioneers of neoclassical economics, Alfred Mar-
shall and Léon Walras.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Modigliani transforms Keynes’s second-pass model by dropping the assump-
tion W = W0 and replacing it with the assumption that in equilibrium workers
are on their supply schedules. L = L(P/W) coupled with Y = F(K, L) implies
Y = F(K, L(P/W)) ≡ LS(P/W). Tis adds a fourth equation along with the ad-
ditional unknown W. Te complete system is
I( ) = sY, (4.1)
P
Y = GS , (4.2)
W
β( ρ)
M ≡ P αY + P ( ρ)PK = M , (4.3)
1 − β( ρ) B
P
Y = F K , L . (4.4)
W
As outlined in the text, the solution is particularly easy since the system can
be solved sequentially. Te goods-supply and labor-supply equations, (4.2)
and (4.4), allow us to solve for P/W and Y. Once Y is known, equation (4.1)
determines , and, fnally, equation (4.3) determines P and hence W.
A solution is not guaranteed. Liquidity preference may prevent the value of
that emerges from the second step from satisfying the money equation.
Even if there is no lower limit on in equation (4.3), inelastic investment de-
mand may preclude equality of investment and saving at the full-employment,
Copyright © 2021. Harvard University Press. All rights reserved.
proft-maximizing equilibrium of the subsystem formed by equations (4.2)
and (4.4).
Te real-balance efect remedies both the problem of a liquidity trap and
the problem of inelastic investment demand. Replace the fxed-parameter
saving function S = sY in equation (4.1) by
M + PB ( )PK
S = S ( income, wealth ) = S Y , 2
P
and suppose that wealth has a negative impact on saving, so that
154
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
4 • Mathematical Appendix 155
M + PB ( )PK
S 0, 2
P < 0.
Suppose also that the derivative of saving with respect to wealth is negative
and bounded away from zero. Ten as P → 0, real wealth, and therefore con-
sumption demand, increase without bound. So aggregate demand increases
without bound as P → 0 regardless of what happens to investment demand.
Te saving function S(Y, (M2 + PB( )PK)/P) refects the assumption that
wealth consists of portfolio cash and bonds, M 2 + PB ( )PK . Cash for trans-
actions is assumed to be inside money created by the banking system, so that
as P → 0, M1 → 0. Te efect of real cash balances on aggregate demand
does not depend on whether we assume banks are always fully loaned up;
since we are assuming that M1 = PY and M2 = M 2, we have M1 + M2 = M and
M2 → M as P → 0.
But the choice of assumption about banking does afect the shape of the
LM schedule. If we assume that transactions money is created by the banking
system to meet the demand PY, and the amount of portfolio cash is fxed
exogenously, then we have two demand = supply equations in place of the
single money equation, (4.3):
M1 ≡ PY = M1, (4.3a)
β( ρ)
M2 ≡ P ( ρ)PK = M 2. (4.3b)
1 − β( ρ) B
M1 is now an unknown, determined by P and Y, whereas M2 is determined by
M 2. Equation (4.3b) represents the LM schedule in Figure 4.17. It is horizon-
tal because the level of that emerges from equation (4.3b), given P, is inde
pendent of Y. It is still the case, that, as P falls, the level of Y determined by the
intersection of the equations (4.1) and (4.3b) rises, which is to say that the AD
schedule is downward sloping in Y × P space. Te demonstration mimics the
Copyright © 2021. Harvard University Press. All rights reserved.
derivation of dY/dP in the mathematical appendix to chapter 3 and is omitted
here.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. III .
Keynes Vindicated
A Theory of Real-Time Changes
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 5 .
the price mechanism
Gospels According to Marshall and Walras
When . . . the amount produced . . . is such that the demand price is
greater than the supply price, then sellers receive more than is suf fi
cient to make it worth their while to bring goods to market to that
amount; and there is at work an active force tending to increase the
amount brought forward for sale. On the other hand, when the
amount produced is such that the demand price is less than the supply
price, sellers receive less than is suf
ficient to make it worth their while
to bring goods to market on that scale; so that those who were just on
the margin of doubt as to whether to go on producing are decided
not to do so, and there is an active force at work tending to diminish
the amount brought forward for sale. When the demand price is
equal to the supply price, the amount produced has no tendency
either to be increased or to be diminished; it is in equilibrium.
—alfred marshall
Let us take, for example, trading in 3 per cent French Rentes [consols]
on the Paris Stock Exchange and confne our attention to these opera-
tions alone . . .
Te three per cents, as they are called, are quoted at 60 francs. At
this price, brokers who have received some orders to sell at 60 francs
and other orders at less than 60 francs, will ofer a certain quantity of
3 per cent Rentes . . . Per contra, the brokers who have received some
Copyright © 2021. Harvard University Press. All rights reserved.
orders to buy at 60 francs and others at more than 60 francs will
demand a certain quantity of 3 per cent Rentes, when 60 francs is
quoted . . .
[Let us assume that] demand is greater than the ofer . . .
Te brokers with orders to buy can no longer fnd brokers with
orders to sell. Tis is a clear indication that the quantity of three per
cents demanded at 60 francs is greater than the quantity ofered at
that price. Teoretically, trading should come to a halt. Brokers who
have orders to buy at 60 francs 05 centimes or who have orders to buy
at higher prices make bids at 60 francs 05 centimes. Tey raise the
market price . . . In consequence of a two-sided movement, the difer-
ence between efective demand and efective ofer is reduced . . . A
new stationary state is thus found at a higher price.
—léon walras
159
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
160 Raising Keynes
Keynes’s critical endeavor in Te General Teory was to establish that a per-
fectly competitive market system would not normally provide full employ-
ment if lef to its own devices. Whether or not Keynes succeeded depends not
only on whom you ask but also on how you pose the question.
Keynes intends his second-pass model to provide the background for an-
swering a question about dynamics. Tis question begins with supposing an
economy to be at full employment. Te economy then suffers a shock that
drives the unemployment rate into double digits—not unlike what happened
afer the stock market crash of 1929 or the fnancial crisis of 2008. Keynes’s
question is whether the market mechanism lef to itself—with output, prices,
and money wages responding to the new conditions—will return the econ-
omy to full employment, or instead whether the economy will settle at an
unemployment equilibrium.
Keynes’s answer is that the economy is not guaranteed to return to full em-
ployment even if we give it all the time in the world and assume away frictions
and imperfections that interfere with the working of competitive markets. In
short, Keynes is arguing for the existence of an unemployment equilibrium,
using the second-pass model as scafolding on which to tell a story about
changing money wages.
But Modigliani, Haberler, et al. are asking a diferent question: if both the
price level and the wage level can be chosen and the money supply is exoge-
nously fxed, are there levels of money wages and prices and output and em-
ployment that satisfy AD, GS, and LS schedules? Tat is, do there exist levels
of output and employment for which expenditure equals income, producers
maximize profts, and a job is available for every willing worker? Here the
question is whether or not there exists a full-employment equilibrium. Te
answer to this question is also yes.
Te problem is that the two ways of asking about full employment—
whether a full-employment equilibrium exists, and whether once disturbed, a
competitive market economy will return to full employment—are not two
Copyright © 2021. Harvard University Press. All rights reserved.
versions of the same question. By distinguishing the two questions, I mean to
highlight the diference noted in chapter 1 between an equilibrium in the
mind of God and an equilibrium in a world in which people are situated in
and encumbered by history. In the one case, the equilibrium exists indepen
dently of how and indeed whether an economy might reach it; in the sec-
ond case, existence and adjustment are inexorably linked. Te frst question,
whether or not a full-employment equilibrium exists, is the question that
economists have imputed to Keynes, at least since Modigliani’s 1944 article.
Te second question, this book argues, is Keynes’s own question.
To answer Keynes’s question requires detailed attention to the price mecha-
nism—how adjustment takes place when the economy is not in equilibrium—
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 161
instead of following the usual practice (my own past practice included) of
treating dynamics as an aferthought. Tere are two reasons why the usual
practice won’t do. Te more general reason, one which goes far beyond the
purposes of this book, is that models which ignore the adjustment process
make unreasonable demands on agents in terms of the knowledge they have
about the economy.
A standard trope of mainstream economics, going back to Adam Smith,
asserts the contrary, not only that markets are self-regulating but that this
self-regulation makes very minimal demands on agents’ knowledge. Tey
need to know very little about the economy: indeed nothing but the vector of
equilibrium prices and their own consumption preferences and production
possibilities. Choices of optimal plans in terms of the self-interest of each pro-
mote the well-being of all, at least in the limited sense of Pareto ef ficiency.
But this in itself tells us little about knowledge requirements when the data
change, when, say, investment demand or saving desires change. What do
agents do then? To move immediately to a new equilibrium requires frst of all
that every agent knows what the equilibrium looks like. Each must possess a
comprehensive understanding of how the economy works in order to calcu-
late, à la Modigliani or Haberler, what the new confguration of equilibrium
prices will be, a much harder task than simply fguring out one’s own reaction
to this price vector. If investment demand falls or the desire to save increases,
then for a change in the interest rate to stimulate enough investment or for a
change in household wealth to stimulate suf ficient consumption to get to a
new full-employment equilibrium, the price level must fall. Tis much follows
from the qualitative structure of the argument. But—here is the rub—to de-
termine the new equilibrium confguration of prices, agents must have com-
plete knowledge of the production possibilities and preferences of all agents,
for these are the data that will determine the price vector at the new equilib-
rium. And, in addition, each agent must have confdence that other agents are
making the same calculations and are prepared to act on them.
Copyright © 2021. Harvard University Press. All rights reserved.
Tis part of the problem is not specifc to Keynes and his critics. Te world
imagined by Kenneth Arrow and Gerard Debreu (1954; see also Arrow 1963–
1964; Debreu 1959) sharpens our understanding as well as our wits, but it di-
verts attention from some of the more signifcant issues we face in unravel-
ing the mysteries of markets. By collapsing time into a form of product
diferentiation, the Arrow–Debreu framework makes it dif ficult to think about
what happens outside of equilibrium. In a very real sense, the Arrow–Debreu
framework sets too easy a task for economists at the same time it makes life so
much easier for economic agents.
Even if we assume that the equilibrium price vector is known to agents, the
demands made by the Arrow–Debreu framework on agents’ knowledge are
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
162 Raising Keynes
hardly negligible: agents must know their own production possibilities and
consumption preferences from the get-go. A more realistic assumption is that
when the clock starts running, each consumer knows whether or not her
marginal rate of substitution of apples for nuts exceeds the ratio of the price of
nuts to the price of apples, but not the utility-maximizing allocation of her
food budget. Each producer may know whether or not the marginal produc-
tivity of labor exceeds the wage, but not the precise point at which the two are
equal. In short, producers and consumers may have suf ficient knowledge to
proceed in the direction of equilibrium but not enough knowledge to go there
directly; agents know which way is up but little more. As consumers adjust in
the direction of increasing utility, and producers adjust in the direction of
more proft, they will presumably learn more and more about their utility and
production functions. A world of experiential knowledge is necessarily a
world of learning by doing, in which, faced with new data, agents adjust
slowly because of the very nature of their knowledge. (See below, chapters 9
and 10, and Marglin 2008, chapters 7–9, for more extensive discussion of the
assumption of knowledge and rationality in economics.)
To a mainstream economist, this looks like a rigidity, a friction imposed on
the competitive model, à la Guillermo Calvo (1983), that arbitrarily restricts
the ability of agents to adjust. Tis is a misunderstanding. Calvo-type lim-
its on the frequency of adjustment are arbitrary, introduced into the New
Keynesian version of dynamic, stochastic, general equilibrium precisely as a
form of sand in the wheels. My conception of agency, by contrast, involves
no constraints other than those of being human and therefore operating
with limited knowledge and limited cognitive power. Te local, experien-
tial, nature of knowledge makes it impossible to go to the utility-and
proft-maximizing equilibrium all at once. It is a stretch to cast an inherent
limitation of human beings—the boundedness of rationality—as a friction.
Tere is a second reason, one specifc to the introduction of aggregate de-
mand into the story, why adjustment is important: in my interpretation of
Copyright © 2021. Harvard University Press. All rights reserved.
Keynes, the basic model of the economy is overdetermined. In contrast with
just-determined models, we can’t even defne equilibrium apart from the ad-
justment process. Even if we wished to sidestep the whole question of dynam-
ics by invoking assumptions that get the economy to equilibrium without the
trouble and bother of disequilibrium, we d on’t have this option—whatever we
might choose to assume about the knowledge of economic agents. Figure 5.1
pictures the simplest model. As we shall see, once we take all three sched-
ules—AD, GS, and LS—into account, where the economy comes to rest can-
not be disentangled from the process it follows outside of equilibrium.
Despite its importance, adjustment processes get short shrif from econo-
mists. Most think that the problem has been adequately covered if a brief
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 163
P
W
4 Supply of goods
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
Figure 5.1 Aggregate demand, goods supply, and labor supply.
discussion of adjustment is tacked onto a demonstration of the existence of
equilibrium. Neglect begins early on. In elementary texts, how the economy
gets to equilibrium is usually covered (or covered up) in a few paragraphs.
Greg Mankiw’s Principles of Economics explains, “Te price of any good ad-
justs to bring the quantity supplied and quantity demanded into balance”
(2018, p. 78). It is tempting but wrong to put this down to Mankiw’s conserva-
tive bias; a leading “liberal” text, Economics, written by William Baumol, Alan
Blinder, and John Solow, says the same thing: “In a free market, the forces of
supply and demand generally push the price toward its equilibrium level, the
price at which quantity supplied and quantity demanded are equal” (2020,
p. 66). Graduate texts do better, but not much better: as I observed in chapter
2, one leading text devotes no more than ten pages out of a thousand to the
question of stability of equilibrium (Mas-Colell, Whinston, and Green 1995),
Copyright © 2021. Harvard University Press. All rights reserved.
the term of art for whether the economy gets to equilibrium, or more accu-
rately, whether the economy will fnd its way back to equilibrium if it is dis-
lodged by a shock to demand or supply.
On second thought perhaps we shouldn’t be surprised. As Mas-Colell and
his co-authors note (1995, p. 620) there is relatively little to say about what
happens outside of equilibrium—and much of what can be said is quite dam-
aging to mainstream theory. (See, for example, the illuminating but dead-end
investigation by Frank Fisher [1983].) Even when the road to equilibrium lies
in virtual space, with neither production nor exchange taking place to com
plicate matters, very special assumptions are required to guarantee that once
disturbed, equilibrium will be restored. Such a guarantee is provided by as-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
164 Raising Keynes
suming that all goods are substitutes for one another in the sense that if the
price of, say, apples rises, more pears will be demanded. (Tis may be a rea-
sonable assumption about apples and pears, but not necessarily about apples
and piecrusts.)
More problematic is the assumption that time stands still while the econ-
omy is groping its way to equilibrium (Walras’s word for his timeless-groping
process is tâtonnement): in order to maintain the fction of fxed demand and
supply curves, the fction that none of the data change, no transactions take
place while trial prices are adjusted. Te assumption of virtual adjustment is
not indispensable, but it certainly makes the problem of disequilibrium more
tractable.
Tere is another problem, noted by Tjalling Koopmans (1957) and Ken-
neth Arrow (1959) six decades ago: even in a world where all goods are sub-
stitutes and time conveniently stands still outside of equilibrium, we have
trouble telling a story consistent with the basic assumptions of perfect compe-
tition once we give up the fction of an “auctioneer” who stands outside
the market. Tis auctioneer was implicitly assumed (by Léon Walras [1954
(1874)], about whom more below) to maintain the idea that agents always
take prices as given and adjust quantities accordingly. But as Koopmans and
Arrow pointed out, without the auctioneer the assumption that prices change
in response to disequilibrium shortages and surpluses requires that one agent
or another is not taking prices as given. She thus acts more like a monopolist
than a perfect competitor.
Two Stories about Disequilibrium Adjustment
Tere are many stories we can tell about what happens outside of equilibrium,
but two stories are enough to illustrate the multiplicity of options. One is a
story generally attributed to Léon Walras, the nineteenth-century pioneer of
general equilibrium theory. For Walras, demand and supply are determined
Copyright © 2021. Harvard University Press. All rights reserved.
by price, and equilibrium is a matter of fnding prices for which the quantities
demanded and supplied are equal. Te Walrasian story also gives an account
of how demand and supply adjust out of equilibrium, and how in turn prices
adjust to shortages and surpluses.
Te alternative is a Marshallian story, afer Walras’s British contemporary
(and Keynes’s teacher), Alfred Marshall, whose Principles of Economics went
through eight editions between 1890 and 1920 and dominated Eng lish-
speaking economics for long afer. Marshall, unlike Walras, took quantity as
the independent variable and price as the dependent. (Marshall’s preemi-
nence perhaps explains why contemporary economists, following Walras in
making price the independent variable, nevertheless use the horizontal axis
for quantity and the vertical axis for price, against the grain of mathematical
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 165
convention.) In Marshall’s story, and Keynes’s for that matter, the “demand
price” associated with a given quantity is the price at which that quantity will
be demanded; the “supply price” is the price at which the quantity will be
forthcoming from producers. By the same token, the disequilibrium adjust-
ment process is diferent chez Marshall: quantities adjust according to the
diference between demand and supply prices, instead of prices adjusting ac-
cording to the diference between quantities demanded and supplied.
Both Walras and Marshall discuss disequilibrium adjustment, and both tell
real-time stories, but it is fair to say that Walras’s is more problematic. Walras
explains his dynamics in terms of the order books that stockbrokers maintain
on their clients’ behalf. Suppose there are two groups of brokers, one of which
has clients who in the aggregate wish to buy one thousand 3 percent rentes at
61 francs, two thousand at 60, three thousand at 59, the other with clients
who have placed orders to sell one thousand rentes at 59, two thousand at 60,
and three thousand at 61.
If the market opens at 61, ofers to sell will outweigh ofers to buy in the
ratio 3:1. Similarly, if the market opens at 59, ofers to buy will swamp the
market. In the frst case, there will be an excess supply of two thousand rentes,
and in the second case, a shortage of two thousand rentes. In both cases, ac-
cording to Walras, the price will respond, falling in the frst case and rising in
the second. Te price will come to rest only when demand and supply are in
balance, in the present case at 60. Te picture is given in Figure 5.2.
Although Walras suggests the process takes place in real time, on the trad-
ing foor (1954 [1874], lesson 5, sec. 42, pp. 84–86), his story makes more
sense as a virtual, premarket, process than as real-time, market, dynamics. In
real time, what happens when the price is, say, 61, and supply exceeds de-
mand? How many rentes are sold and at what price? What happens to the
subsequent demand and supply of rentes at 60 if trades take place at an open-
ing price of 61 (or 59)? If either demand or supply changes, does 60 remain
the equilibrium price?
Copyright © 2021. Harvard University Press. All rights reserved.
It is not that these questions cannot be answered but that the story quickly
gets very complicated. By assuming the adjustment process takes place before
markets open, before trade takes place, it becomes possible to stipulate that
nothing at all happens out of equilibrium. If adjustment takes place outside of
time, we need not even ask how quantities adjust, because the only quantities
that matter are the equilibrium quantities.
Nor need we be too fussy about how prices change. We can imagine a
Walrasian auctioneer who frst chooses a price at random (“by chance,” or
perhaps “haphazardly” better conveys the meaning of au hasard in this con-
text, but “at random” is the usual translation), then adjusts the price according
to the diference between demand and supply, continuing the adjustment of
price until the equilibrium is reached.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
166 Raising Keynes
Price
(francs per rente)
61
Supply
60
Demand
59
1,000 2,000 3,000 Quantity of
3 percent rentes
Figure 5.2 Demand and supply à la Walras.
Even as a virtual adjustment, there is no guarantee that the process will
converge to equilibrium. If the auctioneer is guided only by the size of excess
demand or supply, she could as well take the traders away from equilibrium as
toward equilibrium. Starting from, say, a price of 60.25 francs per rente, she
might overshoot the mark, responding to the oversupply by reducing the
price to 59.50 francs, then overshooting again by raising the price to 60.75
francs. Only if the adjustment of price takes place in suf ficiently small steps
and the auctioneer is guided by the diference between the quantity demanded
and the quantity supplied at each trial price, can we be sure that, as Figure 5.2
Copyright © 2021. Harvard University Press. All rights reserved.
is drawn, the process will converge to the equilibrium at <Q = 2,000, P = 60>.
Marshall, for all the inevitable abstraction of a simple model, has a more
convincing real-time story. He asks us (1920 [1890], p. 348) to imagine a fsh
market in an era with no refrigeration (refrigeration introduces the complex
ity of storage and inventory). Te supply for the day is determined by the
day’s catch, which is the result of both systematic factors (the size of the fsh-
ing feet, the quality of the fshing grounds, and so forth) and random factors
(weather, etc.). To keep matters at their simplest, however, we abstract from
the random factors and assume only an initial arbitrariness in the frst day’s
catch, assumed to be 1,500 pounds in Figure 5.3. Te equilibrium price on
day 1 is the price that clears the market, namely $5.50 per pound.
Note the diference between the schedule labeled “Supply” and the sched-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 167
ule “Supply on day 1.” Te upward-sloping supply schedule tells us the price
required to coax out a given supply on a consistent basis, for example, a price
of $4.50 to elicit a supply of 1,500 pounds per day, or a price of $5.00 to get
fshermen to bring 2,000 pounds to market. On the assumption that fshing is
perfectly competitive, the supply price is the marginal cost of a pound of fsh.
Te actual supply on day 1, by contrast, is accidental, arbitrary; a result, we
might suppose, of the fshermen’s initial ignorance of market conditions.
But on day 2 the fshermen have more information, namely, the market
price on day 1. How many fshermen set forth tomorrow depends on the rela-
tionship between today’s temporary equilibrium price—which in Marshall’s
story is always the demand price—and the supply price associated with to-
day’s catch—the marginal cost of fsh. If the demand price exceeds the supply
price, as it does in Figure 5.3 ($5.50 vs. $4.50), then fshermen can expect a
windfall proft for going out to sea (that is, a return over and above the mini-
mum required to coax them out of their homes), and the quantity of fsh
brought to market will increase. Depending on how strongly fshermen react
to the diference between demand and supply prices, the day-2 catch may fall
short of the equilibrium <Q = 2,000, P = 5> or it may, like the Walrasian auc-
tioneer’s price, overshoot the mark. But if we imagine a market that adjusts
by small steps, and if the adjustment is always proportional to the gap be-
tween demand and supply prices, and if demand and supply schedules do not
Price
($ per lb)
Supply on day 1 Supply on day 2
6.00 Supply
Copyright © 2021. Harvard University Press. All rights reserved.
5.50
5.00
4.50 Demand
4.00
1,000 1,500 2,000 2,500 3,000 Pounds of fish
Figure 5.3 Demand and supply à la Marshall.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
168 Raising Keynes
change, then, as Figure 5.3 is drawn, the process will converge to the equilib-
rium at <2,000, 5>.
Observe that Marshall’s story does not address the Koopmans–Arrow para-
dox that price-taking agents have to change prices in order to get to equilib-
rium. Rather, Marshall fnesses the contradiction by assuming that the econ-
omy is always in a temporary equilibrium at which the day’s catch equals
demand; he never tells us how the market arrives at the temporary equilib-
rium price. In his own words,
It may so happen that the stock to be sold is practically fxed. Tis, for in-
stance, is the case with a fsh market, in which the value of fsh for the day is
governed almost exclusively by the stock on the slabs in relation to the de-
mand: and if a person chooses to take the stock for granted, and say that the
price is governed by demand, his brevity may perhaps be excused so long as
he does not claim strict accuracy. (1920, p. 348)
We can tell a Walrasian fsh story too. Instead of an arbitrary initial quan-
tity, imagine an arbitrary initial price, say $5.50. Assuming this price will hold
for the day, fshermen set forth. How much fsh do they catch? If we abstract
from random factors like the weather, then according to Figure 5.4, the supply
will be 2,500 pounds. But at a price of $5.50 per pound, consumers are willing
to purchase only 1,500 pounds. With no refrigeration, we must assume the
rest gets tossed back into the sea, or perhaps onto a compost heap. Or we can
assume that the lefover 1,000 pounds gets sold at knockdown prices, presum-
ably somewhere between $5.50 and $4.50 per pound.
As in the Marshallian story, there is new knowledge that will likely afect
the price on day 2. Te question is how. We can continue to suppose that the
fshermen as a group, led by the unlucky ones who couldn’t fnd buyers at
$5.50, lower the price. (Perhaps the elders get together over beer in the eve-
ning—where are the anti-trust police?—and decide on the next day’s price.)
However they resolve the Koopmans–Arrow paradox, we assume that on day
Copyright © 2021. Harvard University Press. All rights reserved.
2 all the fshermen converge on a common price, say $5.25. Te supply of fsh
falls to 2,250 pounds and the demand increases to 1,750 pounds, assuming
once again that underlying demand and supply schedules do not change from
one day to the next. Still, fsh are in excess supply, and we can assume that the
price will fall further. Te end, once again, is equilibrium at <2,000, 5> if the
price response is proportional to the gap between supply and demand and the
steps are small enough.
Te two stories are, in a sense, mirror images, the Walrasian story hinging
on prices responding to the diference between quantities demanded and sup-
plied at given prices, the Marshallian story hinging on quantities responding
to the diference between demand and supply prices. In the Walrasian case,
the quantity supplied responds to price, in the Marshallian, price responds to
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 169
Price
($ per lb)
6.00 Supply
5.50 Price on day 1
Price on day 2
5.00
4.50 Demand
4.00
1,000 1,500 2,000 2,500 3,000 Pounds of fish
Figure 5.4 A Walrasian fsh market.
the quantity supplied. Te Walrasian story traces out a sequence of daily equi-
libria on the supply schedule; the Marshallian story, a sequence on the de-
mand schedule.
Observe that in the Marshallian case the signals agents receive from one
day to the next are signals about proftability. Te temporary equilibrium tells
producers whether the price of fsh exceeds or falls short of marginal cost,
hence whether it pays to expand or contract production. In the Walrasian
case the signal is very diferent, refecting how fast fsh are selling relative to
the catch. Te balance between quantity demanded and quantity supplied
tells producers whether to reduce or to increase price.
Copyright © 2021. Harvard University Press. All rights reserved.
One way of marrying the two stories is to assume that, afer experiencing a
Marshallian day-1 equilibrium, producers respond on day 2 à la Walras, tak-
ing the market-clearing price on day 1, $5.25, as given and delivering the as-
sociated quantity of fsh, 2,500 pounds. Tis becomes the supply on day 2, and
the new Marshallian temporary equilibrium requires a dramatic fall in price,
to $4.25. Te new price coaxes out a much-reduced supply on day 3, and so it
goes. Since the successive steps can no longer simply be assumed to be small,
there is no guarantee that the process converges to the demand-supply equi-
librium. Convergence depends on the relative elasticity of the demand and
supply schedules.1
We can also tell a slightly diferent Walrasian story that gives the same pat-
tern of price and quantity as the Marshallian-adjustment process. Suppose
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
170 Raising Keynes
that fshermen set the price each day and deliver the quantity of fsh de-
manded at the day’s price, D(p). Tey continue to adjust the price according
to the diference between the day’s production and the quantity that would
maximize profts at the day’s price, that is, according to the diference D(p) −
S(p). When S(p) exceeds D(p), as in Figure 5.5, fshermen reduce prices in
order to sell more fsh the next day.
Qualitatively, the path this modifed Walrasian story charts is observation-
ally equivalent to the original Marshallian story, the daily equilibria now trac-
ing out a path on the demand schedule rather than on the supply schedule.
But the signaling is very diferent.
Te Walrasian story pictured in Figure 5.5 may be less convincing than the
Marshallian one, but this is as much due to our following Marshall in his
choice of example than because the (modifed) Walrasian process is inher-
ently less plausible than Marshall’s: we may fnd it hard to imagine fshermen
providing fsh to order. But consider, say, a competitive handloom-weaving
industry, with weavers making a homogeneous product. We can certainly
imagine producers setting a quantity for the day with the price adjusting to
clear the market, but it would make as much, if not more, sense to imagine
producers setting a price for the day and producing however much buyers
order at that price, then adjusting tomorrow’s price on the basis of the dis-
tance they end up from their supply schedules. If demand is slack and many
Price
($ per lb)
6.00 Supply
5.50 Price on day 1
Copyright © 2021. Harvard University Press. All rights reserved.
Price on day 2
5.00
4.50 Demand
4.00
1,000 1,500 2,000 2,500 3,000 Pounds of fish
Figure 5.5 A modifed Walrasian fsh market.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 171
looms are running for less time than the weavers wish, it is plausible that they
will respond by reducing prices; by the same token, if the looms are humming
and weavers are obliged to work into the night to fll customers’ orders, in the
next period they well might raise prices to take advantage of the favorable
market, recognizing that this action will discourage demand.
Evidently, there are a variety of stories we can tell about adjustment, some
of them more appealing than others (all of them of course just-so stories if for
no other reason than that they assume small-scale production is the rule).
Because all the stories limit producers to a single signal, they share a common
problem. Tey all limit the trajectory of output and price to a dynamic in
which the adjustment process keeps production on one of the two sched-
ules—the demand or supply schedule—which defne the system. For this
reason neither Walras nor Marshall gives us any way of handling disequilib-
rium when the economy is away from both demand and supply schedules.
To see the problem, we ask a simple question, one we shall again ask when
we apply the argument to Keynes’s General Teory: what happens when an
equilibrium confguration like <2,000, 5> ceases to be an equilibrium because
one of the schedules moves? Suppose, as in Figure 5.6, a supply shock shifs
the supply schedule upward and to the lef (imagine an increase in the price of
the coal on which the fshermen depend to run their boats’ engines).
If, out of habit, fshermen continue to provide 2,000 pounds of fsh, Mar-
Price
($ per lb)
6.00 Supply
5.50
Copyright © 2021. Harvard University Press. All rights reserved.
5.00
4.50 Demand
4.00
1,000 1,500 2,000 2,500 3,000 Pounds of fish
Figure 5.6 Disequilibrium caused by a supply shock.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
172 Raising Keynes
shall’s story can be invoked to explain how the economy might move from the
old equilibrium to the new one at <1,500, 5.50>: on day 1 of the new dispen-
sation, the market clears at the old equilibrium price and quantity—remem-
ber that the day’s supply is perfectly inelastic, equal to 2,000 whatever the
price—but this price is now below marginal cost, which is well over $6.00. So
on day 2 fshermen cut back production, and the economy moves toward the
new equilibrium.
Walras is, as it were, at sea, at least if we stick to the original Walrasian dy-
namic. Tere is no provision in this story for producing anywhere but on the
supply schedule; we simply don’t have a story to go with <2,000, 5> as a dis-
equilibrium phenomenon. To be sure, we would face a similar problem with
the Marshallian story if, instead of a supply shock, the economy were to suffer
a demand shock; the Marshallian story always assumes the economy is on the
demand schedule.
Hybrid Stories
We can solve the problems associated with starting from a point away from
the demand and supply schedules by constructing a hybrid “Marshallian”
process in which quantity adjusts according to Marshall’s story and price ac-
cording to Walras’s.2 Te advantage of the hybrid is that producers can be as-
sumed to process all the information at their disposal, one signal refecting
(marginal) proftability and the other about how fast goods are piling up
or moving of the shelves. Te diference between price and marginal cost
drives quantity adjustment, and the diference between quantity demanded
and quantity supplied drives price adjustment. Instead of forcing the system
onto the demand or supply schedule even when it is not in equilibrium, we
can now say what happens when the starting point <Q0, P0> is not on either
schedule.
As is the case with pure Marshall and pure Walras, the demand and supply
Copyright © 2021. Harvard University Press. All rights reserved.
schedules together defne equilibrium: the demand schedule represents com-
plete adjustment of fsh production to demand, with the result that price does
not change when the fsh economy is on the demand schedule, and the supply
schedule represents adjustment of marginal cost and price, so that quantity
does not change when the economy is on the supply schedule. Te labels P = 0
and Q = 0 on the two schedules in Figure 5.7 identify the schedules as, respec-
tively, stationary loci of prices and quantities. P is the rate of change of price
with respect to time, and Q is the rate of change of quantity.3
If the starting point is <Q0, P0>, output expands in response to the positive
diference between price P0 and marginal cost S−1(Q0). At <Q0, P0> price falls
because the quantity demanded, D(P0), is less than the quantity produced, Q0.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 173
Price
($ per lb)
6.00 P 0 Supply
Q 0 , P0
5.50 D (P0 ) , P0
Q 0 , S 1 (Q 0 )
5.00
4.50 Demand
4.00 Q 0
1,000 1,500 2,000 2,500 3,000 Pounds of fish
Figure 5.7 Disequilibrium adjustment as a hybrid Marshallian process.
As I indicated when I frst told Walras’s story, the unlucky suppliers who do
not fnd ready buyers at the price P0 scramble to unload their goods, driving
the price down today, with tomorrow’s price set afer work, over beer in the
pub. In neither case do we assume that the price falls to the level that clears
the day’s market, as in the pure-Marshallian story; a portion of the catch sim-
ply goes unsold. Te result is to move the economy in the direction of the ar-
row, with price falling and quantity initially increasing. Once price falls to
marginal cost, the trajectory changes course, with quantity falling.4
If we wish to add a dose of realism to the proceedings, we might take note
of the likelihood that some producers will be luckier than others: the lucky
Copyright © 2021. Harvard University Press. All rights reserved.
ones, who are able to sell the day’s catch at the price P0, are receiving an en-
tirely positive message from the market, in that they can both sell their catch
and expand their operations p roftably. Of course the unlucky ones, who
are lef with fsh on their hands, are receiving a negative signal, but hope
springs eternal: perhaps they will have better luck in fnding customers to-
morrow; and if they do, it will have paid them to expand their operations. All
we need assume is that the positive, proftability, signal dominates with regard
to quantity adjustment, while the negative, sales, signal dominates with re-
gard to price, price-cutting by the unlucky fshermen perhaps forcing the
hands of all the rest. Once again, convergence to equilibrium—stability—is
guaranteed by assuming that adjustment takes place in small steps, continu-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
174 Raising Keynes
ously in the limit, and that prices respond positively to D(P0) − Q0, while
quantities respond positively to P0 − S−1(Q0).5
We can also imagine an alternative hybrid, a “Walrasian” process, in which
price and quantity adjustment are inverted, making price adjustment depend
on the distance the actual quantity is from the quantity producers would like
to provide, while the distance between actual production and the quantity
demanded drives changes in output. Te supply schedule becomes a station
ary locus of price, and the demand schedule a stationary locus of quantity. In
Figure 5.8 price adjustment is governed by Q0 − S(P0) and quantity adjust-
ment by Q0 − D(P0).6
Te economy now moves southwest rather than, as in Figure 5.7, southeast.
Since the actual provision of fsh on day 1 is below the level fshermen would
choose to supply at the price P0, they will lower the price of fsh on day 2 in
order to expand the market. At the same time, output falls in response to the
discrepancy between actual production, Q0, and the quantity consumers will
purchase at the going price when they equate marginal rates of substitution
with relative prices.
A key diference between the two adjustment processes is that, in the frst,
Marshallian, process, discrepancies between production and demand drive
price changes, whereas in the second, Walrasian, process, the same discrepan-
cies drive output changes. One signal, two diferent responses. Tis is not to
Price
($ per lb)
6.00 Q 0 Supply
Q 0 , P0
5.50 D (P0 ) , P0 S (P0 ) , P0
Copyright © 2021. Harvard University Press. All rights reserved.
5.00
4.50 Demand
4.00 P 0
1,000 1,500 2,000 2,500 3,000 Pounds of fish
Figure 5.8 A hybrid Walrasian process.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 175
say that quantities are unchanging in the Marshallian case or prices are un-
changing in the Walrasian case. P roft maximization also enters the picture,
driving quantity changes in the hybrid version of Marshall’s story while driv-
ing price changes in the hybrid version of Walras’s.
Why don’t price and quantity go immediately to equilibrium? What ac-
counts for the gradualness of the adjustment process? Here is where I depart
from the equilibrium discipline that Lucas (1977) successfully urged upon the
economics profession. My assumption, as I indicated in the beginning of this
chapter, is that agents possess only local knowledge; at a point like <Q0, P0>
producers know the direction in which to change price and quantity, but
not by how much. Tey know, for instance, that price exceeds marginal cost,
P0 − S−1(Q0) > 0, or that the current level of production is less than the amount
they would optimally provide at the going price, S(P0) − Q0 < 0, but they do
not know the equilibrium values of Q and P. Tough they cannot go immedi-
ately to the equilibrium, they can move on a trajectory that will eventually
bring the economy to equilibrium.
Alternative names for Marshallian and Walrasian processes are provided by
John Hicks (1974). He distinguishes between “fexprice” and “fxprice” mod-
els, the frst being more or less the Marshallian process, the second the Walra-
sian process. Hicks’s terminology is misleading: in both cases prices are fexi
ble, the diference lying in what drives prices and what drives quantities. In
both fex-and fxprice adjustment, agents receive information of two kinds.
One kind is information about how fast goods are moving of producers’
shelves, based on the diference between actual output and how much con-
sumers are willing to buy at the current price, Q0 − D(P0). A second kind is
how proftable business is, based (in the fexprice model) on the gap between
the actual price and marginal cost or (in the fxprice model) on the gap be-
tween actual output and proft-maximizing output at prevailing prices. Te
diference between fex-and fxprice models lies in how the information is
processed.
Copyright © 2021. Harvard University Press. All rights reserved.
Information processing out of equilibrium is diferent from information
processing in a competitive equilibrium. At equilibrium all that perfectly
competitive producers need to know are price and (marginal) cost. Te
Koopmans–Arrow paradox is that perfect competition, if it means price-
taking behavior, can exist only in equilibrium, where sales information be
comes redundant. Out of equilibrium this information is not redundant: how
fast goods are moving of the shelves is essential knowledge. Te situation is
even more complicated in overdetermined models, like the ones we shall en-
counter in the next chapter: the relevance of both kinds of information, sig-
nals about proftability and signals about market conditions, does not disap-
pear in equilibrium.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
176 Raising Keynes
Macrodynamics à la Viner
All this is preamble. Our real interest here is not in fsh or cloth, but in the
macroeconomy, a highly simplifed version of the economy characterized by a
single good that can be used interchangeably for consumption and invest-
ment. But the preamble is essential, because macro-adjustment stories simply
take elements from the basic Marshallian and Walrasian micro stories. For
example, the story told by Jacob Viner, or more accurately, the story I imputed
to him in chapter 2, is one of Marshallian adjustment, in which proftability
drives output, and the GS schedule is a locus of stationary output. In the
right-hand panel of Figure 2.5, reproduced below as Figure 5.9, the trajectory
from F to E is one in which output is driven by proftability and the real price
is driven by labor-market conditions. At F the vertical diference between real
marginal cost, 4, and the real price, 1, drives output, causing the economy to
contract. At the same time, in the labor market, money wages are driven by
the diference between the supply of and the demand for labor. In response to
unemployment, the money wage and hence the real wage fall, which is to say
P
W 0
W
Supply of goods
4 E
Supply of labor
3 B
Copyright © 2021. Harvard University Press. All rights reserved.
2 A
1 F
Y 0
100 200 300 400 500 600 Y
Figure 5.9 A Marshallian trajectory from disequilibrium to equilibrium
with the price of output fxed.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 177
the real price rises. Once the real price has risen enough for the economy to
reach A, where price and marginal cost are equal, contraction gives way to
expansion even as the real wage continues to fall and the real price continues
to rise in response to unemployment. (As in fsh or cloth markets, whether or
not the trajectory converges directly to equilibrium depends on the values of
adjustment parameters.) Money prices play no role in this story. Te real
price of output rises because the money wage falls, not because the nominal
price of output rises.
Te polar opposite of a fxed money price is a fxed money wage. Adjust-
ment of the real price must now depend on nominal price adjustment. We
suppose prices adjust in the manner of Walras, according to the gap between
actual output and the proft-maximizing level of output at the going real price.
And we continue to assume that output adjustment is Marshallian, respond-
ing to the gap between price and marginal cost. Ten the GS schedule is at
one and the same time the locus of stationary output and the locus of station
ary real price. Every point on the GS schedule in Figure 5.10 is an equilib-
rium. Te particular point to which output and the real price will gravitate (A
in the fgure) depends on the speeds of adjustment. Reaching full employ-
P
W
Supply of goods
4 E
Supply of labor
Copyright © 2021. Harvard University Press. All rights reserved.
2 A
1 F
Y 0 P 0
100 200 300 400 500 600 Y
Figure 5.10 A hybrid trajectory from disequilibrium can stall out.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
178 Raising Keynes
ment in this case would happen only if the price level adjusted infnitely more
rapidly than output.7
If we try to tell a pure-Walrasian story, in which the GS schedule is a locus
only of stationary prices, the economy will move only if we assume that out-
put and employment respond to diferences between the marginal disutility of
labor and the current real wage. Tis is admittedly a bit odd, for it presup-
poses that workers rather than capitalists determine the level of economic
activity. But economics is full of odd assumptions; even Nobel Laureates have
been known to express indiference to the realism of their assumptions. In
this case, the trajectory would follow a very diferent path from that in Fig
ures 5.9 and 5.10, with employment, output, and the nominal price rising un-
til full employment is reached at B, as in Figure 5.11.
As the economy continues from B until E in Figure 5.11, output exceeds
capacity and gradually falls as the continuing rise in the price level fnally
brings the economy back to equilibrium at E.
Observe that Say’s Law plays no role in either the Marshallian or the Walra-
sian story. Say’s Law provides no theory of how output changes; it is if any-
thing a theory of why output does not change. Demand is the dog which isn’t
barking.
P
W
Y 0 Supply of goods
4 E
Copyright © 2021. Harvard University Press. All rights reserved.
B
2 Supply of labor
1
F
P 0
100 200 300 400 500 600 Y
Figure 5.11 Walrasian adjustment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 179
Macrodynamics in The General Theory
Te story in which output is driven by the relationship of actual output to the
quantity demanded is one that has been told over the years to make Keynes-
ian dynamics plausible. One textbook even makes it an essential part of the
Keynesian story. In Robert Frank and Ben Bernanke’s telling, an adjustment
process based on meeting the demand determined by (temporarily) fxed
prices—think handloom weavers producing to order—uniquely characterizes
disequilibrium in Keynes’s view of the world. Tey call this a “key assumption
of the basic Keynesian model” (2007, p. 722).
Frank and Bernanke are mistaken. It is not this particular version of how
aggregate demand matters that is essential to Te General Teory, but the fact
that aggregate demand matters at all. Aggregate demand may fnd a clearer,
more direct, expression in a Walrasian, fxprice, version of adjustment, in
which the GS schedule is a locus of stationary prices. But, as will be shown in
chapter 6, Te General Teory can be laid out in a Marshallian, fexprice,
framework, in which goods supply is a locus of stationary output; moreover,
despite the emphasis on the Walrasian story in the exegesis of Te General
Teory, Keynes’s intellectual debts likely disposed him in toward a Marshal-
lian understanding of adjustment. Te struggle to escape the old, of which
Keynes was excruciatingly aware, perhaps made a Walrasian framework more
inviting but not necessarily easy to embrace.8
What happens if Keynes’s economy is not at equilibrium? Suppose the in-
terest rate and consequently investment demand are fxed, and we begin at a
position to the lef of the equilibrium, say A = <300, 3.5> in Figure 5.12,
which reproduces Figure 3.13 (without the point on the LS schedule).
In deriving the AD schedule from agents’ consumption and investment
propensities, chapter 3 argued the fxprice case that the excess of expenditure
over income will deplete inventories and lead businesses to increase orders.
With a propensity to consume equal to 0.75 and investment equal to 100,
Copyright © 2021. Harvard University Press. All rights reserved.
producers respond positively to the diference between output (300) and ex-
penditure, which at Y = 300 is 325 = (0.75 × 300) + 100. Initially the
expenditure-income gap leads to an increase in demand of 25; but this is only
the beginning of the process: production responds to expenditure, which
leads to an equal increase in income and a further increase in spending. Te
process of demand expansion continues until income and expenditure are in
balance, namely at Y = 400.
If we begin at a level of output to the right of the AD schedule, we simply
apply the logic of expansion in reverse. At B = <500, 3.5>, spending is only
475 = (0.75 × 500) + 100, and goods pile up on merchants’ shelves. Tis leads
to a cutback in orders and a reduction in output, and the downward spiral
continues until expenditure and income are in balance, once again at Y = 400.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
180 Raising Keynes
P
Supply of goods
5
4
A B
3
C
2
1
Aggregate demand
100 200 300 400 500 600 Y
W = W0
Figure 5.12 Aggregate demand and goods supply at a given interest
rate.
In efect, we constructed the AD schedule out of a story in which the bal-
ance between the quantity produced and the quantity demanded drives out-
put adjustment. However, as a story of adjustment of the economy as a whole,
this argument is at best incomplete: businesses are assumed to increase or
decrease production without any attention to whether the change in output is
proftable. What has happened to the logic of goods supply based on p roft
maximization, an assumption implicit not only in Jacob Viner’s mainstream
approach but also in Keynes’s explicit adoption of the “classical” postulate that
the real wage is equal to the marginal productivity of labor (Te General Te-
ory, pp. 17–18)?
In the present case, it might be argued that there is no diference between
Copyright © 2021. Harvard University Press. All rights reserved.
assuming that inventory changes drive output and assuming that proftability
drives output. Given the price P = 3.5 (a real price of 3.5/W0), it is proftable to
expand production when expenditure exceeds income, as at A in Figure 5.12.
And it is proftable to contract output when expenditure falls short of income,
as at B. Proftability and inventory changes point in the same direction.
However, at a point like C, the two signals lead to diferent actions. Te
excess of marginal cost over price signals producers to contract output, while
the shortfall of output relative to aggregate demand signals them to expand
production.
Tere is another problem. Neither of these signals necessarily says anything
about how the price level changes. Businesses may receive two signals—one
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 181
about whether goods are fying of the shelves or accumulating in the back-
room, the other about whether or not it is proftable to expand or contract
output—but so far they take action in only one dimension, output. Tis leaves
open the question of how prices change, and without an answer to this ques-
tion we cannot characterize the path from a disequilibrium starting point.
It should be clear by now that the question is not whether producers, even
competitive producers, change prices, but how and when they do so. What
information guides price changes; what information guides output changes?
Output frst. In the present case, if we wish to stick with the fxprice story
that inventory changes drive output, output increases when Y − AD is nega-
tive, and output falls when Y − AD is positive. If demand drives output, we
must look to the supply side for a signal to drive prices. Adapting the Walra-
sian process depicted in Figure 5.8, we assume that prices change according to
the diference between the actual level of production and the level that would
maximize profts at the present level of prices, which is to say according to the
diference Y − GS(P/W). At A in Figure 5.12, Y − GS(P/W) is negative, so pro-
ducers cut prices to increase sales; at B and C, Y − GS(P/W) is positive, so
producers raise prices to reduce the volume of output. Figure 5.13 illustrates
the adjustment process initiated by a downward shif in aggregate demand,
which makes the original equilibrium E untenable. Initially, only output
changes because producers are still on their supply schedules at E even afer
Y 0
P
Supply of goods
5
E
4
Copyright © 2021. Harvard University Press. All rights reserved.
3
F
2
P 0
1
Aggregate demand
100 200 300 400 500 600 Y
W = W0
Figure 5.13 Walrasian adjustment in the frst-pass model.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
182 Raising Keynes
the downward shif in demand. But as time goes on, the gap between actual
output and proft-maximizing output, Y − GS(4), leads producers to decrease
prices in the hope of increasing sales. As shown, the adjustment process con-
verges to a new equilibrium, at which both output and prices are lower.
Of course, we are not confned to a Walrasian story. Marshallian fexprice
adjustment marries the idea that output changes are driven by the relation-
ship between price and marginal cost with the idea that imbalances between
expenditure and income drive prices. Tis requires a diferent story about
what happens when inventories fall below, or rise above, planned levels: in-
stead of triggering orders for additional goods or triggering a cutback in or-
ders, the inventory signal triggers price changes, prices rising in response to
goods fying of the shelves, prices falling in response to goods piling up. In
Figure 5.14 the starting point E is the same as in Figure 5.13, but the trajec-
tory to the equilibrium at F is diferent. Initially, prices fall because a fall in
expenditure leaves output piling up on producers’ shelves; producers main-
tain the initial level of output because price and marginal cost are equal. But
in time, as prices fall below marginal costs, producers begin to scale back
output.
Observe that the economy’s trajectory from E to F need not be the direct
path in Figure 5.14. As was observed in note 6, it is possible for the economy
P 0
P
Supply of goods
5
E
4
Copyright © 2021. Harvard University Press. All rights reserved.
3
F
2
Y 0
1
Aggregate demand
100 200 300 400 500 600 Y
W = W0
Figure 5.14 Marshallian adjustment in the frst-pass model.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 183
to follow a path of damped oscillations—ever smaller cycles around the equi-
librium point. In the present case the approach to equilibrium will follow
such a path if prices adjust much more rapidly than output. Te picture is in
Figure 5.15.
Finally, we should note the possible asymmetry between how the economy
adjusts when desired expenditure (ED) exceeds output (Y) and what happens
when expenditure falls short. In the frst case, goods are fying of the shelves,
but once inventories are depleted, actual expenditure (E) cannot exceed Y. In
the second case (ED < Y), E can be less than Y. Te reason is that including
unwanted inventory accumulation in income as well as in output is prob
lematic (despite the practice of the National Income and Product Accounts).
Sole proprietors such as fshermen would presumably not include un-
wanted inventories in calculating their incomes. It doesn’t matter in the
partial-equilibrium case because the income of fshermen can be assumed to
have a negligible impact on the demand for fsh. But if all output and income
were generated by sole proprietors, the valuation of unwanted inventories
would matter a lot because of the circular relationship between output, in-
come, and expenditure. In this case we would need to take account of a pos-
sible gap between income and output.
P 0
P
Supply of goods
5
E
4
3
Copyright © 2021. Harvard University Press. All rights reserved.
F
2
Y 0
1
Aggregate demand
100 200 300 400 500 600 Y
W = W0
Figure 5.15 Marshallian adjustment in the frst-pass model: damped
cycles.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
184 Raising Keynes
To do this, write E = min (ED, Y) and reformulate the relationship between
output and expenditure as
E = cE + I with E = ED if ED < Y,
E = cE + I with E = Y if ED > Y,
where c = 1 − s is the marginal propensity to consume. So
I
E = if ED < Y,
1−c
whereas
E = cY + I if ED > Y.
Instead of Figure 3.2, reproduced below as Figure 5.16, we have the expendi-
ture function in Figure 5.17.
However, Keynes’s focus, and ours, is on a capitalist economy rather than
one composed of sole proprietors. Te diference is that under capitalism
goods are produced by means of wage labor. In this case it is more reasonable
to assume that output is equal to income, and expenditure is equal to a frac-
tion of income, c in the present case, regardless of whether inventory accumu-
lation is desired or inadvertent. If goods are produced, wages must be paid
whether or not the goods are sold. Te only part of the link between output
Expenditure (E D ) ED Y
500 C D ID
A
400
C D = 0.75Y
B
Copyright © 2021. Harvard University Press. All rights reserved.
300
200
100 ID
45°
100 200 300 400 500 Income (Y)
Figure 5.16 Consumption and investment determine aggregate demand.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • The Price Mechanism 185
Expenditure (E D ) ED Y
500
A
400 C D ID
B
300 C D = 0.75Y
200
100 ID
45°
100 200 300 400 500 Income (Y)
Figure 5.17 Expenditure function in a sole-proprietor economy.
and income that might be broken by excessive inventories is the proft link. It
would be a very unwise producer who would count the bird in the bush
alongside the bird in hand. So the slope of the expenditure function in Figure
5.17 might be less steep to the right of E = Y, but it would not be perfectly fat.
If we focus on qualitative results, we can continue to use the assumption built
into Figure 5.16, namely, that income and output are identical regardless of
whether ED < Y or ED > Y.
A Summing Up
Copyright © 2021. Harvard University Press. All rights reserved.
How do economies adjust when not at equilibrium? What knowledge do
agents mobilize? And how do they act on their knowledge? Tis chapter pro-
vides a framework for answering these questions by retelling stories that Al-
fred Marshall and Léon Walras ofered their readers a century and a half ago.
Te stories may not be as diferent as I make out here (see note 3), but the
point of emphasizing the diferences is to make clear that there is more than
one way to process the information that disequilibrium generates. Indeed, the
starting point for understanding disequilibrium adjustment is that the infor-
mation available to agents is much richer than in the simple tale of competi-
tive markets at equilibrium, in which price-taking behavior is the defning
characteristic—a good thing too, since it is also the case that the agents need
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
186 Raising Keynes
much more information in order to adjust behavior in ways that might propel
the economy toward equilibrium.
Out of equilibrium, agents not only have information on the proftability of
production, as they do at equilibrium, but they also can observe, and act on,
information about how fast goods are moving of their shelves. Tey need this
additional information because they have to make two decisions rather than
one. Not only must agents make choices about how much to produce, as they
do on the basis of equilibrium prices, they also must make decisions about
prices. Price-taking behavior is all well and good as a characterization of equi-
librium, but it doesn’t ft very well with stories about disequilibrium. Unless
the problem is fnessed, as Marshall did by assuming the economy is always in
(temporary) equilibrium even as the equilibrium shifs, somebody has to de-
cide whether and by how much to change prices.
Dynamics do not matter so much in standard demand-supply models—
where demand curves slope downward and supply schedules edge upward—
or, for that matter, in Keynes’s frst-pass model. But this is not the case when
we graf an LS schedule onto the frst-(and second-) pass models of Te Gen-
eral Teory. Te model becomes overdetermined—too many relationships for
the number of variables—and the language of static equilibrium cannot even
account for equilibrium. Te frst casualty is the identifcation of equilibrium
with market clearing. In its more general sense of a balance of forces moving
quantities and price, equilibrium survives, but only as the outcome of one
adjustment process or another; diferent adjustment processes produce difer-
ent equilibria. Nor does adjustment always lead to equilibrium. We have no
choice but to examine dynamics directly.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
In continuous time the hybrid Marshallian adjustment process is
P = 1 (D(P ) − Q),
Q = 2( P − S −1 (Q) ).
Te Jacobian is
1 D ′ − 1
J = ,
2 − 2 S′−1
with D′ ≡ (dD/dP) and S′−1 ≡ (dS/dP)−1 = dS−1/dQ. Te trace
tr J = 1D′ − 2 S′ −1
is negative, and the determinant
det J = − 1D′2S′−1 + 12
is positive, so the equilibrium is stable. Whether the approach to equilibrium
follows a path like the one in Figure 5.7 (that is, the equilibrium is a node) or
like the one described in note 5 (where the equilibrium is a focus) depends on
the sign of the discriminant,
( D ′ + S′ )
2
−1
1 2 − 4 1 2.
If demand is inelastic and supply elastic (D′ small and S′ large), 1D ′ − 2 S′−1
Copyright © 2021. Harvard University Press. All rights reserved.
will be small in absolute value. In consequence, the discriminant will be nega-
tive, and the equilibrium is a focus.
Te mathematics is similar for alternative adjustment assumptions, and so
is omitted here.
Macrodynamics à la Viner
Te text proposed alternative adjustment processes for the pre-Keynesian
model. Te frst assumes that the price level is fxed and adjustment of the real
price takes place via changes in the nominal wage rate. Te wage rate moves
inversely with unemployment, measured here by the output gap
187
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
188 Raising Keynes
P
F K, L − Y.
W
Output follows the logic of p
roft maximization and increases when the real
price exceeds real marginal cost, decreasing when P/W < FL−1 with Y = F(K, L)
and FL ≡ dY/dL. Te system of adjustment is given by
•
P
W P
= −W = 1 F K , L − Y ,
P W W
W
P
Y = 2 − FL−1 .
W
We have the Jacobian
1FL L′ − 1
J = ,
2 2 FL−3 FLL
where L′ ≡ dL(P/W)/d(P/W). Te trace is negative and the determinant posi-
tive (FLL ≡ ∂2F/∂L2 < 0). So the equilibrium is stable.
If we take the wage as fxed, and assume Walrasian price adjustment—the
price level adjusts according to the distance between output and the GS
schedule—the two adjustment equations become
•
P
W
= P = − GS P − Y ,
P P
1
W
W
Copyright © 2021. Harvard University Press. All rights reserved.
P
Y = 2 − FL−1 .
W
Now the Jacobian is
( F ) −1 F 3 1
J = 1 LL L
,
2 2 FL−3 FLL
since
GS′ = − ( FLL ) FL 3.
−1
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
5 • Mathematical Appendix 189
Te trace is negative, but det J = 0. Tis means that each and every point on
the GS schedule is an equilibrium. Te adjustment process stalls out once Y
approaches GS(P/W).
Finally, the adjustment depicted in Figure 5.11 is characterized by
•
P
W
= P = − GS P − Y ,
P P
1
W
W
P P
Y = 2 LS − ,
W W
where LS(P/W) = F(K, L(P/W)). Te Jacobian is
( F ) −1 F 3 1
J = 1 LL L
,
2 (LS′ − 1) 0
so both the trace and the determinant are negative.
Macrodynamics in the First-Pass Model
Flexprice adjustment gives the system
P
= θ1[E − Y ] = θ1[I ( ρ) + cY − Y ] = θ1[I ( ρ) − sY ],
P
P
Y = 2 − FL −1 .
W
Prices are driven by the balance between expenditure and income, which is
to say the balance between investment demand I( ) and desired saving sY,
Copyright © 2021. Harvard University Press. All rights reserved.
and output is driven by the balance between the real price P/W and the real
marginal cost FL−1. Te Jacobian is
0 − 1 s
J = −1 .
2W 2 FL−3 FLL
Te trace is negative and the determinant is positive, guaranteeing a stable
equilibrium. Once again the equilibrium can be either a node or a focus, de-
pending on the sign of the discriminant
( − F )
2
−3
2 L FLL − 4 1s 2W −1.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
190 Raising Keynes
With relatively slow price adjustment this discriminant is positive, and the
equilibrium will be a node, as pictured in Figure 5.14. If price adjustment is
very rapid compared to output adjustment, the discriminant will be negative,
and the equilibrium a focus, as in Figure 5.15.
Fixprice adjustment is governed by
P P
= −1 GS − Y ,
P W
Y = θ 2[I ( ρ) − sY ].
We have
( F ) −1 F 3 1
J = 1 LL L
.
0 − 2 s
Once again the trace is negative and the determinant is positive.
With fxprice adjustment the discriminant
( (F )
2
)
−1
1 LL FL 3 + 2 s
is positive, and the approach to equilibrium is as pictured in Figure 5.13.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 6 .
the gener al theory without rigid
prices and wages
My so-called “fundamental equations” [in Keynes’s earlier book, A
Treatise on Money] were an instantaneous picture taken on the
assumption of a given output. Tey attempted to show how, assum-
ing the given output, forces could develop which involved a p roft-
disequilibrium, and thus required a change in the level of output.
But the dynamic development, as distinct from the instantaneous
picture, was lef incomplete and extremely confused. Tis book, on
the other hand has evolved into what is primarily a study of the forces
which determine changes in the scale of output and employment as a
whole.
—john maynard keynes
Te real signifcance of the Keynesian contribution can be realized
only within the framework of dynamic economics. Whether or not an
underemployment equilibrium exists; whether or not full employ-
ment equilibrium always will be generated in a static system—all this
is irrelevant. Te fundamental issue raised by Keynesian economics is
the stability of the dynamic system: its ability to return automatically
to a full-employment equilibrium within a reasonable time (say, a
year) if it is subjected to the customary shocks and disturbances of a
peacetime economy. In other words, what Keynesian economics
claims is that the economic system may be in a position of underem-
ployment disequilibrium (in the sense that wages, prices, and the
amount of unemployment are continuously changing over time) for
Copyright © 2021. Harvard University Press. All rights reserved.
long, or even indefnite, periods of time.
—don patinkin
Very likely Keynes chose the wrong battleground. Equilibrium analy-
sis and comparative statics were the tools to which he naturally
turned to express his ideas, but they were probably not the best tools
Portions of this chapter were previously published in “Te General Teory afer 80 Years:
Time for the Methodological Revolution,” in Keynes, Geld, und Finanzen, ed. H. Hagemann, J.
Kromphardt, and M. Marterbauer (Marburg: Metropolis, 2018), pp. 27–55; and in “Raising
Keynes: A General Teory for the 21st Century,” EconomiA 19 (January–April 2018): 1–11.
191
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
192 Raising Keynes
for his purpose . . . Keynes’s comparative statics were an awkward
analytical language unequal to the shrewd observations and intuitions
he was trying to embody.
—james tobin
Keynesian macroeconomics neither asserts nor requires nominal
wage and/or price rigidity. It does assert and require that markets not
be instantaneously and continuously cleared by prices. Tat is a much
less restrictive assumption, and much less controversial. It leaves
plenty of room for fexibility in any commonsense meaning of the
word.
—james tobin
As Tobin says, it all depends how you understand the word “rigid.” When
Franco Modigliani reduced the diference between Keynes and the main-
stream to the assumption of rigid money wages (1944), at least rigid meant
what the dictionary says it means: “infexible,” “fxed.”1 And it is in this sense
that the models I develop to represent Te General Teory do not assume ri-
gidities; rather, prices and quantities respond fexibly to economic pressures.
But there is another, more contemporary sense of nominal rigidities, in
which rigid is not opposed to fexible. Rather, an economy with rigidities is
one which does not move immediately to equilibrium. In this sense, the ab-
sence of rigidities means there is no need to bother with adjustment at all. Te
very need for adjustment “proves” the existence of rigidities. Tis is merely
part of the “discipline imposed by equilibrium theory,” to use Robert Lucas’s
phrase (1977, p. 12).
Tere is nothing but respect for language to prevent us from labeling grad-
ual adjustment of prices and wages a case of nominal rigidities. But whatever
we might think in a general way of the discipline of equilibrium and the pros-
Copyright © 2021. Harvard University Press. All rights reserved.
pect of dispensing with disequilibrium altogether, it is not available in the
present context. Here we can’t even defne equilibrium apart from the dy-
namic processes that drive the economy when it is not in equilibrium.
Where Do We Start? The Choice of Models
I have observed that Keynes lists the ingredients only for two models, a frst-
pass model, in which the interest rate is fxed, and a second-pass model, in
which the money supply is fxed. He conspicuously fails to provide a corre-
sponding list to accompany chapter 19 of Te General Teory, where he fnally
drops the assumption of a fxed money wage. Without even a list of ingredi-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 193
ents, there can be no recipe, no model. Tis chapter attempts to provide not
only the ingredients but also the instructions for cooking, in short, the miss-
ing third-pass model.
Which model, the frst-or the second-pass model, do we build on? It is
tempting to build on the second-pass model for reasons internal and external
to Te General Teory. Te internal reason is that this is the model sketched in
chapter 18, immediately before the discussion of changing wages. Moreover, a
model with a fxed money supply may seem more suitable to a discussion of
the ability of the invisible hand to guide the economy to full employment. Te
external reason is that the assumption of a given money supply became inte-
gral to the presentation and understanding of Te General Teory almost im-
mediately, with the publication of Hicks’s IS-LM exegesis (1937).
But the second-pass model is better understood as an attempt at a bare-
bones refutation of the classical dichotomy than as a way of thinking about
the economy Keynes was trying to analyze. Once it is accepted that aggregate
demand matters, it is not clear that it makes sense to hold the money supply
constant while varying the price level—except in a world in which money is
limited to specie. Once the money supply is defned to include bank deposits,
fractional-reserve banking makes the money supply endogenous, because
money created by the banks (as distinct from the monetary base) is deter-
mined by the behavior of the banking system. As was observed in chapter 4,
with the money supply endogenous, varying the price level can change not
only the demand for money but also the supply. If the price level falls dra-
matically, it’s hard to imagine banks not curtailing the supply of loans for
transactions purposes (and hence the supply of money) in line with the much
reduced value of collateral, not to speak of reduced demand. Te question
then becomes this: What do the banks do with the (potential) money; that is,
what happens to the pyramid built on the monetary base? Chapter 4 pointed
out that it takes a strong assumption about bank behavior to ensure that
banks will replace commercial loans with purchases of other assets (bonds in
Copyright © 2021. Harvard University Press. All rights reserved.
our models). And that assumption is even stronger in the dynamic context in
which changes in prices and output are at issue, not diferent levels of prices
and outputs. We return to this point in chapter 7.
Treating the money supply as fxed may be a legacy of the quantity theory
situated in the gold standard or some other form of commodity money. Here
the regnant beliefs about the metallic origins of money on the part of Keynes’s
contemporaries may have swayed Keynes to assume a simple monetary sys-
tem based on a fxed amount of currency or specie.2 Absent such an intention,
the exercise of assuming away the central bank and other elements of a mod-
ern economy would likely have appeared pointless to someone as deeply in-
volved in contemporary fnance and fnancial institutions as was Keynes.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
194 Raising Keynes
If the existence of a central bank is assumed, as in the summary in chapter
18 of Te General Teory (see p. 247), a natural way to characterize the money
supply is that the central bank controls interest rates and varies the supply of
bank reserves (or overdrafs) as the interest rate requires. But even if we leave
out the central bank, it still makes sense to start from a given interest rate, as
in the frst-pass model, for this modeling strategy allows us to treat the inter-
est rate separately from the price level. Like Keynes, we can vary the interest
rate afer the basic contours of the model are established. Unlike Keynes we
shall do this with the help of a formal model, not on the basis of informal
discussion of the complications introduced by varying wages and prices.
The First-Pass Model Augmented by
a Labor-Supply Schedule
Tere is one important, indeed fundamental, diference between the model
laid out in this chapter and Keynes’s frst-pass model: how labor supply fg
ures in the picture. For purposes of the present argument, the labor market is
assumed to be competitive in the textbook sense of the term, with the supply
of labor—full employment—defned by how much employment workers de-
sire at the going wage. Te money wage adjusts in response to the diference
between actual employment and the supply of labor. Tis is not the whole
story of wage dynamics—see chapter 18—but wages do appear to respond to
the existence of massive involuntary unemployment. During the Great De-
pression, hourly wages in U.S. manufacturing declined by more than 20 per-
cent from their 1929 level to their level in 1932. Agricultural wages declined
by even more, almost 30 percent.3 And, as we shall see in chapter 19, the rate
of unemployment appears to exercise a continuing infuence on the rate of
infation.
Te consequence of introducing labor supply is to overdetermine the
model. Figure 6.1 presents aggregate demand, goods supply, and labor supply
Copyright © 2021. Harvard University Press. All rights reserved.
in a model in which the interest rate is fxed and the AD schedule is a vertical
line. Two AD schedules, corresponding to two diferent interest rates, are
shown. For the AD schedule represented by the solid line, there is an equilib-
rium <Y = 450, P = 4>, at which all three schedules are satisfed. Otherwise, as
for the dashed AD schedule at Y = 250, there is no equilibrium at which the
economy is on its AD schedule as well as on its GS and LS schedules.
If the existence of involuntary unemployment puts downward pressure
on wages, we have to redefne the vertical axis: it will no longer do to assume
W = W0. We have a choice. We could, as in the construction of equilibrium
in the contemporary AD-AS framework of mainstream texts, assimilate the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 195
5
Supply of goods
4
2
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
W = W0
Figure 6.1 Aggregate demand, goods supply, and labor supply.
wage as just one more price, as in Figures 4.8 and 4.9. But this would presup-
pose the answer to the central question of what happens when the model is
overdetermined; we would simply be assuming that workers are fully em-
ployed at the same time businesses are maximizing p rofts.
Instead of fxing the relative price of goods and labor at the full-employment,
proft-maximizing level, we make the real price, P/W, a state variable of the
model. As in Keynes’s model, the price of goods is determined endogenously,
and, as in the standard AD-AS model, the money-wage rate is also deter-
mined endogenously—but, unlike the AD-AS model, not by assuming full
employment.
Serendipity, refected in the solid AD schedule, apart, a conventional equi-
Copyright © 2021. Harvard University Press. All rights reserved.
librium does not exist in an overdetermined model. As I have already noted,
we cannot even talk meaningfully about equilibrium without specifying the
process that the economy follows outside of equilibrium.
Te emphasis on dynamics, on the real-time trajectory of the economy, has
many other implications, starting with the measure of economic activity
on the horizontal axis of Figure 6.1. It becomes problematic to take the LS
and GS schedules as given once and for all so that the full-employment,
proft-maximizing level of output is constant over time. Moreover, if we as-
sume that the process of adjustment takes place in real time, and that invest-
ment generates new productive capacity, the relationship between employ-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
196 Raising Keynes
ment and output necessarily changes over time. If investment adds, say 2
percent to the capital stock each year, then afer fve years the capital stock
will have grown by more than 10 percent, afer ten years by more than 20
percent. Similarly with labor, if the labor force grows every year, then the full-
employment level of output will increase accordingly.
In the case of the Great Depression, investment dried up so the capital
stock did not grow very much. But a growing labor force, not to mention
technological progress, made for a dramatic change in the relationship be-
tween unemployment and output: both the U.S. labor force and labor produc-
tivity grew by about 1 percent per year on average between 1929 and 1936.4
Te result was that although output returned to its pre-Depression level by
the time Te General Teory was published, unemployment was at 17 percent,
fourteen percentage points higher in 1936 than in 1929.
Keynes’s assumption that investment depends only on the current rate of
interest is also more problematic in the dynamic context than in a static con-
text. If we are interested in comparative statics, all the efects of the path that
the economy followed to get to its present state can be understood as back-
ground to the relationship of investment to the current rate of interest. But
path dependence plays a central role in the context of models that unfold over
time: the investment-demand schedule is what it is today because of what
happened yesterday. To assume away the infuence of the past on investment
as simply “background” means, among other things, that the investment
schedule reproduces itself from one moment to the next, which is to say that
what appears proftable to undertake today is unafected by the projects taken
up yesterday or the day before. Te economy, as it were, never uses up oppor-
tunities, for investment opportunities always come back today even if they
were exploited yesterday.5
Tere is another issue that takes on new meaning in the context of real-
time adjustment: namely, the dependence of investment demand on P/W. Te
higher the real price of goods, the lower is the real wage and the higher is the
Copyright © 2021. Harvard University Press. All rights reserved.
rate of proft, and p rofts and wages infuence diferent kinds of investment
very diferently. One might expect that investment undertaken to augment
capacity would react favorably to increases in P/W and consequent increases
in profts, whereas investment to cut operating costs by reducing labor inputs
would react unfavorably to the same increase in P/W. For investment demand
to be invariant with respect to P/W, as in Figure 6.1, it would take the unlikely
coincidence that changes in capacity-increasing investment and labor-saving
investment just balance each other out.
A similar logic leads to questioning the assumption that the propensity to
save is independent of the distribution of income, refected in the price level
in Figure 6.1. Moreover, the propensity to save out of current income depends
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 197
not only on today’s income but on yesterday’s saving and income, not to men-
tion tomorrow’s.
Tis said, we have plenty on our plate without taking into account just yet
either the endogeneity of labor and capital or the infuence of P/W on invest-
ment and saving. We will return to these issues in chapters 9, 10, 17, and 18,
but for the present we continue to assume the AD schedule is vertical and that
the capital stock and the labor force are fxed.
How Disequilibrium Unfolds
Te basic model of the economy is given in Figure 6.2, which reproduces Fig
ure 6.1 with a single change: the vertical axis no longer measures the nominal
price with a fxed money wage. Instead, since the money wage is now a vari-
able, the vertical axis measures the real price, P/W.
When the story opens, whether by design of the central bank, the cunning
of the invisible hand, or dumb luck, aggregate demand serendipitously aligns
with the intersection of goods and labor supply to produce the full-
employment equilibrium <Y = 450, P/W = 4>.6 Ten the unthinkable hap-
pens—the end of a housing bubble, the discovery that collateralized debt ob-
ligations are toxic assets, a fnancial meltdown. What was thought to be risk,
subject to the laws of probability, turns out to be radically uncertain. Financ
ing becomes problematic, investment prospects appear less bright, house-
holds begin to worry about the future. Tere is a drying up of capital expendi-
P
W
Supply of goods
Copyright © 2021. Harvard University Press. All rights reserved.
2
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
Figure 6.2 Aggregate demand, goods supply, and labor supply in the space of
output × real price.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
198 Raising Keynes
ture, or a retrenchment in consumer spending, perhaps both. As a result of
this shock, the AD schedule moves to the lef, to become the dashed line at
Y = 250. What happens now?
Despite the upheaval, inertia may keep the economy for the moment at the
original equilibrium <Y = 450, P/W = 4>, as my young colleagues supposed it
would that fateful day in September 2008, when Lehman Brothers crashed
(see chapter 1). But for how long? Te Keynesian argument is that with the
crisis of confdence comes a sharp pullback of expenditure. If output contin-
ues at its old pace, goods begin to pile up in shops and warehouses.
Te next page of the story depends on the dynamics of the economy: what
does a shortfall in expenditure relative to output communicate to sellers of
goods and services? As we saw in the previous chapter, there are at least two
possible answers, a Marshallian (fexprice) one and a Walrasian (fxprice) one.
In a Marshallian world, the world in which Keynes was raised and in which
he remained most comfortable, we would expect prices to fall in response to
the shortfall of expenditure. Te AD schedule becomes a locus of stationary
prices, P = 0. Although this is not the assumption conventionally made in tell-
ing Keynes’s story, we shall go along with it for the moment in a spirit of being
more loyal to Keynes’s teacher, Marshall, than was Keynes himself.
We can fnd a justifcation for the fexprice hypothesis in Keynes’s work, but
tellingly not in Te General Teory, rather in the Treatise. In his banana para-
ble (see chapter 3), Keynes argues,
A thrif campaign leads to an overproduction of bananas . . .
Since bananas will not keep, their price must fall. Tus as before, the pub-
lic will consume the whole crop of bananas, but at a reduced price-level. . . .
Since wages are still unchanged, only the selling-price of bananas will have
fallen and not their cost of production; so that the entrepreneurs will suffer
an abnormal loss. (1930, vol. 1, p. 177)
Copyright © 2021. Harvard University Press. All rights reserved.
If prices fall and nothing else happens, the economy moves vertically down-
ward and is no longer on either the LS or the GS schedule. Suppose the
economy comes to a momentary rest at <Y = 450, P/W = 3>, as in Figure 6.3.
Tere is now involuntary unemployment because (by assumption) more
work is desired at the new, higher, real wage; workers fnd themselves to the
lef of their supply schedule. (Recall that in the absence of money illusion, the
fall in the price level has the same impact on the supply of labor as a corre-
sponding rise in the nominal wage.) At the same time, producers are supply-
ing more than the quantity of goods that maximizes p rofts since they are to
the right of their supply schedule. Equivalently, since the supply curve refects
the marginal cost of output, producers comparing goods prices with marginal
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 199
P 0
P
W
Supply of goods
4
2
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
Figure 6.3 Flexprice adjustment to a shortfall of expenditure, I.
costs realize that, at the margin, costs are greater than revenues. Keynes con-
tinues the parable:
Te continuance of this will cause entrepreneurs to seek to protect them-
selves by throwing their employees out of work or reducing their wages. But
even this will not improve their position, since the spending power of the
public will be reduced by just as much as the aggregate costs of production.
(1930, vol. 1, p. 177)
What happens next depends on the combined force of the three processes at
work in the economy. Because we are to the right of the AD schedule, prices
continue to fall. Te efect of disequilibrium in the labor market is to put
Copyright © 2021. Harvard University Press. All rights reserved.
downward pressure on wages as workers compete over jobs, the LS schedule
being a stationary locus of money wages, W = 0. If wages are falling less rap-
idly than prices, the real price continues to fall, which is to say (P/W)• < 0.
Output also falls, because real marginal cost exceeds the real price. For Mar-
shall the GS schedule is a locus of stationary output, Y = 0, as was the fsh-
supply schedule in chapter 5.
If wages are falling more slowly than prices, the economy moves to the
southwest, as indicated in Figure 6.4. Both money prices and money wages
are declining, the frst because of the shortfall of expenditure relative to
income, the second because of the shortfall of jobs relative to the supply of
labor.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
200 Raising Keynes
P 0
P
W Y 0
5
Supply of goods
4
0
W
2
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
Figure 6.4 Flexprice adjustment to a shortfall of expenditure, II.
In Figure 6.4 the second step is pictured as a decline in P/W, but depending
on the relative speeds of adjustment of wages and prices, it is possible that
P/W will rise or, indeed, that P/W will not change at all. How do we chart the
evolution of prices and output?
Equilibrium in an Overdetermined System:
The Stationary Real-Price Locus
Consider Figure 6.5. On any line that we might draw between the AD sched-
ule and the LS schedule within the shaded area, there is at least one point at
which the relative rates at which prices and wages decline are equal, so that
the ratio P/W remains constant over time. How do we know this? Because at
Copyright © 2021. Harvard University Press. All rights reserved.
point A the price level is stationary while the wage is falling, whereas at point
B the price level is falling while the wage is constant. Tus if we go from A to
B, the percentage rate at which the price level changes goes from zero to a
negative number, and the rate at which the wage changes goes from a negative
number to zero. At some point on any path between A and B, prices and
wages are falling at the same percentage rate. In Figure 6.5 this point is labeled
C. (Tink of two cars travelling on the same road, one increasing its speed,
the other one slowing down. If the frst car is accelerating from 0 to 50, and
the second is decelerating from 50 to 0, there is some point at which the two
vehicles are necessarily traveling at the same speed.)
What is true of the line segment labeled AB is true of any other line seg-
ment connecting the AD and LS schedules in the shaded area, for instance the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 201
P 0
P
W Y 0
5 B
Supply of goods
4 C
3 A
0
W
2
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
Figure 6.5 Flexprice adjustment to a shortfall of expenditure, III.
P 0
P
W Y 0
5 B
Supply of goods
4 C
3 A E
F
0
W
2
D
Supply of labor
1
Aggregate demand
Copyright © 2021. Harvard University Press. All rights reserved.
100 200 300 400 500 600 Y
Figure 6.6 Flexprice adjustment to a shortfall of expenditure, IV.
line segment DE in Figure 6.6. Here the point where the cars are going at the
same speed is labeled F. Now connect the dots: C and F and all the other
points at which the percentage rates of change of wages and prices are equal
trace out a schedule, or locus, of stationary real prices, as in Figure 6.7.
Te intersection of the stationary real-price schedule, labeled (P/W)• = 0,
and the GS schedule is the equilibrium. It is not, of course, a market-clearing
equilibrium in which aggregate demand equals goods supply and labor sup-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
202 Raising Keynes
ply, as at the serendipitous equilibrium <Y = 450, P/W = 4> in Figure 6.2. But
it is an equilibrium in the sense that there are no forces internal to the model
to propel either P/W or Y away from the point E in Figure 6.8. Te downward
pressure on prices just balances the downward pressure on wages. Both are fall-
ing at the same percentage rate, so the ratio of the two does not change.
Several things are worth mentioning here. First, the equilibrium, as adver-
tised, is an equilibrium with involuntary unemployment: E lies to the lef of
P 0
P
W Y 0
5 B
Supply of goods
4 C
3 A E
F
0
W
2
D
Supply of labor
P 0
1
W
Aggregate demand Stationary real price
100 200 300 400 500 600 Y
Figure 6.7 Flexprice adjustment to a shortfall of expenditure, V.
P 0
P
W Y 0
Copyright © 2021. Harvard University Press. All rights reserved.
5
Supply of goods
4
E
3
0
W
2
Supply of labor
P 0
1
W
Aggregate demand Stationary real price
100 200 300 400 500 600 Y
Figure 6.8 Equilibrium with fexprice dynamics.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 203
the LS schedule. Second, the model is one of fexible wages (and prices), at
least in the sense of fexibility before the idea took root that a nonrigid econ-
omy would go to equilibrium immediately, instead of gradually adjusting to
imbalances. Tird, we need not suppress the standard assumption that unem-
ployment puts downward pressure on money wages; here unemployment not
only coexists with fexible money wages but also drives money wages and
partially drives real wages. In Figure 6.8, labor supply is a full and equal part-
ner in determining how the economy moves when not in equilibrium, and
the equilibrium is sensitive to any displacement of the economy relative to the
LS schedule in the same way that it is sensitive to displacement away from the
AD and GS schedules. Fourth, the equilibrium is stable—see the mathemati-
cal appendix—so that afer a displacement of the AD schedule, as in Figure
6.2, the economy fnds its way to the new equilibrium at E in Figure 6.8. Start-
ing from any point in Figure 6.8, the length and the direction of the vertical
arrow are determined by the distance from (P/W)• = 0 and the speeds of ad-
justment of prices and wages; likewise, the horizontal arrow by the distance
from Y = 0 and the speed of adjustment of output. Together, the two arrows
indicate the trajectory of P/W and Y.7 Finally, the exact position of the sta
tionary real-price schedule depends on the relative speeds of adjustment of
prices and wages. Te AD schedule is itself one limiting case, in which prices
react infnitely more rapidly than wages; at the other extreme, the LS schedule
is a limiting case in which wages react infnitely more rapidly. Te two cases
are laid out in the two panels of Figure 6.9. Panel (a) in efect captures
P
P 0
P P 0
W Aggregate demand W Aggregate demand
Y 0 Y 0
Supply of goods Supply of goods
P 0
Copyright © 2021. Harvard University Press. All rights reserved.
5 W 5
Stationary
4 4
real price Supply of labor E Supply of labor
0 W 0
3 W 3
E P 0
2 2 W
Stationary real price
1 1
100 200 300 400 500 600 Y 100 200 300 400 500 600 Y
(a) Rapid price adjustment (b) Rapid wage adjustment
Figure 6.9 Limiting cases of fexprice dynamics.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
204 Raising Keynes
Keynes’s frst-pass model, in which money wages adjust so slowly that we are
back in the fxed-wage model and the LS schedule is redundant, real output
being determined by aggregate demand, and the real-price level being deter-
mined by goods supply. Panel (b) is the mainstream case, in which the AD
schedule is irrelevant, equilibrium being determined by the conditions of
goods supply and labor supply.
Observe that unless prices adjust infnitely more rapidly than wages, equi-
librium will be characterized by the indefnite accumulation of unwanted in-
ventories.8 At E in Figure 6.8, expenditure falls short of output and income;
inevitably, goods pile up on producers’ and merchants’ shelves, and price re-
ductions do nothing to alleviate the problem. Te imbalance between expen-
diture and income persists because the real price remains constant, percent-
age reductions in money wages just matching (percentage) reductions in
money prices. Attempts by producers to cut real prices are frustrated by
workers’ attempts to cut real wages.
Tere are two possible escape routes. One is that producers take account of
ever-increasing inventories by raising the speed of adjustment of prices. If this
action is not countered by workers, the efect will be to move the stationary-
price locus ever closer to the AD schedule, a process for which the limiting
case is Figure 6.9(a). Te result is to eliminate the excess inventories by restor-
ing the frst-pass equilibrium. A second possibility is that producers will re-
duce output in response to inventory accumulation. Tis second possibility
brings us to an alternative adjustment process, which I have identifed with
Walras rather than Marshall.
Alternative Dynamics, Alternative Equilibrium
Producers get two signals from the market. Te frst comes from the AD
schedule, which represents an equilibrium between output and expenditure.
Inventories and order books play a central role in the story: as order books
Copyright © 2021. Harvard University Press. All rights reserved.
shorten and inventories pile up, Marshallian producers respond by cutting
prices, while, conversely, increasing backlogs and inventory depletion lead
them to increase prices. Te second signal comes from the GS schedule,
which represents an equilibrium between marginal cost and price. If price
exceeds marginal cost, Marshallian producers fnd it proftable to expand pro-
duction; if price falls short of marginal cost, it is proftable to contract output.
In the alternative—fxprice or Walrasian—story, the reaction to market
signals is reversed. An imbalance between expenditure and output triggers
production adjustments; an imbalance between actual output and desired
supply triggers price adjustments. Hence the immediate impact of shifing the
AD schedule in Figure 6.10 is the opposite to the Marshallian case: the AD
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 205
Y 0
P
W P 0
5 Supply of goods
0
W
2
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
Figure 6.10 Fixprice adjustment to a shortfall of expenditure, I.
schedule is now a locus of stationary output, while the GS schedule becomes
a locus of stationary (money) prices.
I have not been able fnd any reference to fxprice dynamics in Te General
Teory—nor for that matter anywhere else in the corpus of Keynes’s work. But
this was how my generation learned Keynesian economics. My own introduc-
tion, Paul Samuelson’s Economics (1955) put the argument this way:
Suppose income were higher than the intersection point [of the saving and
investment schedules that determine aggregate demand]. Ten . . . business-
men would be unable to sell enough to justify their current level of output.
So they would contract their production and lay workers of. (p. 116)
As I have previously noted, Robert Frank and Ben Bernanke’s introductory
text goes even further:
Copyright © 2021. Harvard University Press. All rights reserved.
Te basic Keynesian model is built on a key assumption . . . that frms do not
continuously change their prices as supply and demand conditions change.
Rather, over short periods frms tend to keep their prices fxed and meet the
demand that is forthcoming at those prices. (2007, p. 722)
Nothing is said about prices and wages in either text, but in a fexible world,
price and wage changes will follow. If we suppose once again that a negative
demand shock moves the AD schedule to the lef from serendipity, the short-
fall of expenditure leads to a contraction of production, moving the economy
in the direction of the arrow in Figure 6.10. At this point, workers, as before,
are to the lef of their supply schedule, with the result that money wages fall.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
206 Raising Keynes
Tere is a new element: producers, who also fnd themselves to the lef of
and above their (goods) supply schedule, reduce prices in order to increase
sales. Both prices and wages fall, but since we don’t know anything about the
relative speeds of adjustment, once again we can’t say which is falling more
rapidly. Hence we can’t say whether the ratio of prices to wages, P/W, is rising
or falling. At least not without more information.
As in the Marshallian case, the additional information is refected in the
locus of stationary real prices. Te logic should now be familiar: somewhere
between the GS schedule, which is now associated with stationary money
prices, and the LS schedule, which continues to be a locus of stationary
money wages, lies a schedule along which prices and wages are falling at the
same percentage rate, so that the ratio of prices to wages remains constant.
Te construction of this schedule can be seen in Figure 6.11.
Te points C and F are points at which money prices and money wages
fall at the same relative rate. Te stationary real-price locus, again labeled
(P/W)• = 0, connects all such points. With this information, we can map out
the trajectory the economy will follow, as well as the equilibrium to which this
trajectory leads. Figure 6.12 shows the resulting stationary real-price schedule
and the accompanying equilibrium at E. Te vertical and horizontal arrows
once again indicate the trajectory of the economy starting from diferent
points on the diagram.
Initially, in contrast with fexprice adjustment, the real price does not
change; the beginning point is on both the GS and the LS schedules, so nei-
Y 0
P
W P 0
5 Supply of goods
Copyright © 2021. Harvard University Press. All rights reserved.
4 C
Stationary real price
F
3 P 0
W
0
W
2
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
Figure 6.11 Fixprice adjustment to a shortfall of expenditure, II.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 207
ther P nor W changes. But once the economy begins to contract, it fnds itself
above the stationary real-price locus, and the real price falls. If wages are ad-
justing more rapidly than prices, the stationary real-price locus slopes down-
ward, as in Figure 6.13; in this case, the real price climbs along the trajec-
tory to E.
Once again, the adjustment process has two limiting cases, one in which
Y 0
P
W P 0
5 Supply of goods
4
Stationary real price
E
3 P 0
W
0
W
2
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
Figure 6.12 Fixprice adjustment to a shortfall of expenditure, III.
Y 0
P
W P 0
E
5 Stationary real price Supply of goods
P 0
Copyright © 2021. Harvard University Press. All rights reserved.
W
4
0
W
2
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
Figure 6.13 Fixprice adjustment to a shortfall of expenditure, IV.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
208 Raising Keynes
price adjustment is infnitely more rapid, the other in which wages adjust in-
fnitely more rapidly. Te two cases are shown in the two panels of Figure
6.14. In both cases the equilibrium level of output is completely determined
by the vertical AD schedule. Diferent speeds of adjustment afect only the
equilibrium real price.
By comparing limiting cases, we gain a new perspective on the role of as-
sumptions about the relative speed of adjustment of money wages and prices.
With the relative speed of wage adjustment going to zero, as depicted in Fig
ure 6.15, we are once again back to Keynes’s frst-pass model—a fxed money
wage is the limiting case of rapid price adjustment. In the limiting case in
which wages do not adjust at all, it turns out not to matter for equilibrium
whether the rest of the model is determined by fexprice or fxprice dynamics;
we have the same equilibrium in both cases. Te LS schedule becomes irrele-
vant and we can ignore dynamics, at least in the sense that the equilibrium
does not itself depend on the dynamic specifcation.
Te two limiting cases of rapid wage adjustment are also revealing. In the
frst panel of Figure 6.16, the mainstream model comes into its own. With
prices responding relatively slowly to aggregate demand, money-wage adjust-
ment becomes, in the limit, real-wage adjustment, which obviates Keynes’s
stricture (see note 9) against confounding the determination of money wages
and real wages. Te equilibrium in both panels is a full-employment equilib-
rium.
P
Y 0
P Y 0
W W Aggregate demand
Aggregate demand
E
P 0 P 0
Supply of goods Supply of goods
5 5
Copyright © 2021. Harvard University Press. All rights reserved.
4 4
P 0 Supply of labor
W W 0 Supply of labor
3 Stationary 3 0
E P 0 W
real price
W
2 2
Stationary real price
1 1
100 200 300 400 500 600 Y 100 200 300 400 500 600 Y
(a) Rapid price adjustment (b) Rapid wage adjustment
Figure 6.14 Limiting cases of fxprice dynamics.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 209
P P Y 0
P 0
W W Aggregate demand
Aggregate demand
Y 0 P 0
Supply of goods Supply of goods
P 0
5 W 5
Stationary
4 real price 4
Supply of labor P 0
0
W W
3 3 Stationary
E
real price E
2 2
Supply of labor
1 1 W 0
100 200 300 400 500 600 Y 100 200 300 400 500 600 Y
(a) Rapid price adjustment with (b) Rapid price adjustment with
flexprice dynamics fixprice dynamics
Figure 6.15 Limiting cases of rapid price adjustment.
P P 0 P Y 0
W Aggregate demand W Aggregate demand
Y 0 P 0
5 5
Copyright © 2021. Harvard University Press. All rights reserved.
4 4
E
3 0 3
P 0 W P 0 0
W
W W
2 Stationary real price 2 Stationary real price
1 1
100 200 300 400 500 600 Y 100 200 300 400 500 600 Y
(a) Rapid wage adjustment (b) Rapid wage adjustment
with flexprice dynamics with fixprice dynamics
Figure 6.16 Limiting cases of rapid wage adjustment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
210 Raising Keynes
In Figure 6.16(a), the Marshallian case, aggregate demand plays no role in
determining the real variables of the economy—output and the real price—its
infuence being limited to money prices. In Figure 6.16(b), aggregate demand
and labor supply jointly determine equilibrium. In this case, aggregate de-
mand matters for the real side of the economy; it is the goods-supply equation
that becomes inefectual except for determining the price level. Because wage
adjustment is (by assumption) so rapid, producers’ attempts to afect real
prices by changing money prices are completely subverted by wage adjust-
ments. Real-wage adjustment efectively resists any attempt to dislodge em-
ployment from the LS schedule.
Table 6.1 summarizes the four limiting cases, two “Keynesian” cases of
rapid price adjustment, and two cases of rapid wage adjustment, one refect
ing Modigliani’s version of the second-pass model and one a hybrid of Keynes
and the mainstream, in which “real wage resistance” (Joan Robinson’s term
[1962]) makes the LS schedule a full partner with aggregate demand in deter-
mining the equilibrium confguration.
Te mainstream may see confrmation in Table 6.1: both the (frst-pass)
Keynesian formulation (ignoring the LS schedule) and the mainstream for-
mulation (ignoring the AD schedule) turn out to be limiting cases of a more
general model. But there is really not much comfort for orthodoxy. Between
the limiting cases is a vast middle ground, in which equilibrium is deter-
mined jointly by the adjustment of prices, wages, and output. And this middle
ground belongs to Keynes. Every equilibrium, apart from the two limiting
cases of rapid wage adjustment, is an equilibrium with unemployment, the
main critical point of Te General Teory.
Te limiting cases suggest that Keynes’s rejection of the classical dichotomy
takes place at two levels. Besides the complex argument in terms of monetary
infuences on the rate of interest, there is the simpler and more direct argu-
ment that the classical dichotomy is belied by the very fact that wage bargains
are struck in money terms, with real wages emerging only as the price level is
Copyright © 2021. Harvard University Press. All rights reserved.
determined along with the level of output. Only in the limiting case in which
money wages adjust infnitely faster than prices is revising the money-wage
Table 6.1 Limiting-Case Determinants of Equilibrium
Flexprice Fixprice
Rapid Price Adjustment Keynesian Keynesian
AD and GS AD and GS
Rapid Wage Adjustment Mainstream Real-Wage Resistance
GS and LS AD and LS
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 211
bargain tantamount to revising the real-wage bargain. In this limiting case,
the classical dichotomy holds its own; in all other cases, wage bargains are
nominal variables with real consequences.
Diferences between Flexprice and Fixprice Models
In both models producers attempt to maximize profts, but—unlike Marshal-
lian producers—Walrasian producers are frustrated by the market. Instead of
expanding output in response to diferences between price and marginal cost,
producers respond to supply conditions by changing prices. Tey attempt to
sell more goods by reducing prices, expecting larger quantities of goods to be
demanded at lower prices. But they do not succeed for the same reason that,
as Keynes argues, workers cannot fx real wages:9 attempts to lower the real
price of goods are frustrated because slack labor markets counter price reduc-
tions with wage reductions.
Another diference between the two models is that in the fxprice model
equilibrium is conceptually closer to the vision captured in Keynes’s frst-pass
model, output being determined solely by the AD schedule. Tis is true re-
gardless of the speeds of adjustment of wages and prices.
It is also the case that the fxprice equilibrium does not carry with it the
indefnite commitment to unwanted inventories. Te equilibrium now lies on
the AD schedule, along which output and expenditure are equal; the only in-
ventories accumulated in equilibrium are desired inventories, those necessary
to facilitate an orderly process of manufacture and distribution.
Finally, a diference between fxprice and fexprice models with empirical
bite is the cyclical behavior of the real price. In fexprice models the real price
moves procyclically (and the real wage contracyclically) if fuctuations are
caused by demand shocks; this is because the stationary real-price locus
moves procyclically against a stationary GS schedule. By contrast, in fxprice
models the vertical AD schedule moves relative to a stationary real-price lo-
Copyright © 2021. Harvard University Press. All rights reserved.
cus that slopes upward or downward depending on the relative speeds of ad-
justment of nominal prices and wages. If wages adjust more rapidly, then the
stationary real-price locus slopes downward, and the real price moves contra-
cyclically and the real wage procyclically.
Tese results undo an implication of the frst-pass model that was from the
get-go a troubling issue for Keynes and his followers—namely, the implica-
tion that demand changes cause money prices to adjust in the same direction
as output, so that, with a fxed money wage, real wages and output move in
opposite directions. A young American, John Dunlop (1938), and a young
Canadian, Lorie Tarshis (1939), launched their academic careers by casting
empirical doubt on the proposition of a negative relationship between real
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
212 Raising Keynes
wages and output. (Subsequent research has generally led to the conclusion
that fuctuations in real wages are mildly procyclical.) In reply, Keynes (1939)
pointed out that the negative relationship between wages and economic activ-
ity was a consequence not of Te General Teory but of Marshallian assump-
tions to which the economics profession had generally assented. Te as
sumption of diminishing marginal productivity is built into the neoclassical
production function, and therefore so is increasing marginal cost. Te as-
sumption of price-taking producers, for whom p roft maximization implies
equality between marginal cost and price, and the assumption that equilib-
rium is characterized by proft maximization are part and parcel of perfect
competition.
One way out of the problem is to replace the assumption of perfect compe-
tition by an assumption of monopolistic competition or oligopoly. Tis chap-
ter shows another way: drop restrictions imposed by assuming a fxed money
wage, even though this complicates the idea of equilibrium. Te argument
developed in this chapter does not guarantee procyclical behavior of real
wages, but it changes the question from the slope of the GS schedule into one
of how adjustment takes place.
Conclusions
Despite the considerable ground this chapter has covered, the conclusions are
relatively simple and straightforward. In the frst place, one makes enormous
progress in understanding Keynes’s General Teory by jettisoning the static
framework in which the argument has been normally framed, at least since
Hicks’s classic statement (1937). Building on Hicks, Modigliani (1944) for-
malized Keynesian economics in a way that reduces Keynes’s contribution to
nothing more than the replacement of the standard mainstream assumption
of fexible money wages with the assumption of rigid money wages. Keynes’s
exposition gave ammunition to this point of view, and even if Keynes’s com-
Copyright © 2021. Harvard University Press. All rights reserved.
mitment to rigid money wages was, as he claimed, no more than an exposi-
tional strategy, it became problematic in the light of the subsequent turn that
both his followers and his critics took.10 I would certainly agree with the crit-
ics that, to say the least, the argument of chapter 19 in Te General Teory, in
which the assumption of a given money wage is fnally relaxed, is hardly a
model of clarity. Lacking even a list of ingredients for a model comparable to
Keynes’s frst-and second-pass models, the chapter was all the more easily
ignored.
By recasting the argument in dynamic terms and redefning equilibrium in
terms of equal rates of price and wage changes, we can incorporate the two
elements that determine neoclassical equilibrium, an LS schedule and a GS
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 213
schedule, and the two elements that (taking the rate of interest as given) de-
termine equilibrium in Keynes’s own model, the GS schedule and the AD
schedule. Te problem then is that the combined model is overdetermined:
there are three relationships, the AD schedule, the GS schedule, and the LS
schedule, but only two state variables, P/W and Y.
Robert Clower believed that the existence of these three schedules in a
single model implied a contradiction with Walrasian theory:
Either Walras’ law is incompatible with Keynesian economics, or Keynes
had nothing fundamentally new to add to orthodox economic theory. (1984
[1963], p. 41)
Walras’ law is an equilibrium condition that holds only when producers
and workers are (or assume they are) on their respective supply schedules. In
the models of this chapter, even at equilibrium workers are not on their sup-
ply schedules, and producers may or may not be, except in some limiting
cases. Walras’s law is not compatible with equilibrium, but adjustment along
Walrasian lines is perfectly compatible with the theory. Keynes had plenty “to
add to orthodox economic theory”; he simply lacked the tools to articulate his
message.
Te present framework allows us to get beyond the pervasive view that
nominal rigidities are the essence of Keynes’s theory. No more than Keynes is
it my intention to deny nominal rigidities; the point is rather that eliminating
these rigidities would not eliminate the central causal role of aggregate de-
mand, which is the core theoretical construction of Keynes’s General Te-
ory. Rigidities stemming from monopolistic competition, trade u nions, menu
costs, to mention only a few of the usual suspects, exist, but these do not get
at the heart of the problem.
Specifying the dynamic- adjustment pro cess involves two assumptions.
First, an assumption about relative adjustment speeds in the labor and goods
markets, and, second, an assumption about how producers respond to the
Copyright © 2021. Harvard University Press. All rights reserved.
dual signals they receive whenever they are not on the stationary loci of prices
and output.
Te crucial role of relative adjustment speeds of wages and prices becomes
clear if we look again at the two limiting cases of the fexprice and fxprice
models. In both models, when prices adjust rapidly and wages slowly, the
equilibrium approaches the equilibrium in Keynes’s frst-pass model.
At the other extreme, when wages adjust infnitely more rapidly than prices,
the dynamic model approaches either the mainstream model, in which ag-
gregate demand doesn’t matter, or a hybrid model of real-wage resistance, in
which goods supply d oesn’t matter. In both these cases, full employment
reigns at equilibrium.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
214 Raising Keynes
In between these two extremes, equilibrium is characterized by unemploy-
ment, and relative speeds of adjustment determine how much. Te in-between
models formalize Keynes’s vision about capitalism lef to its own devices:
there is no endogenous mechanism by which the economy sooner or later
gets to full employment.
Te second assumption, how producers process the dual signals they re-
ceive from goods markets when they are not on the AD and GS schedules, is
more complex. In contrast to behavior at a demand = supply equilibrium,
producers cannot in general be price takers who decide how much to bring to
market at a price dictated by the market. Tey have two decisions to make,
not one, and they receive two signals, not one. Te frst signal is the proftabil
ity of expanding or contracting output, measured either by the vertical dis-
tance from the supply curve, which refects the diference between price and
marginal cost, or by the horizontal distance, which refects the amount by
which producers would have to increase (or decrease) production in order to
maximize profts. Te second signal is whether it is easy or dif ficult to fnd
buyers for a given level of output, whether products are accumulating on sell-
ers’ shelves or warehouses or fying out the door faster than they can be re-
stocked.
By the same token, producers have two decisions to make, what to do about
prices and what to do about output. Te two dynamic systems, fexprice and
fxprice, difer in the assumptions about how the signals are processed: does
an accumulation of unwanted inventories trigger a reduction in prices (the
fexprice reaction) or a reduction in output (the fxprice reaction)? Does a
lack of proftability at the margin trigger a reduction in output (the fexprice
reaction) or an increase in price (the fxprice reaction)?
Once dynamics become the focus of the model—we ought to begin with
dynamics, with process, whether we are operating in a just-determined or an
overdetermined system, but we are compelled to do so in the second case—it
becomes essential to investigate the institutional basis of dynamic systems
Copyright © 2021. Harvard University Press. All rights reserved.
instead of requiring merely that the dynamics exhibit a surface plausibility.
Te question of whether producers react to disequilibrium by changing price
or quantity, or rather how do changes in price and quantity interact, is hardly
a new one, but it takes on a new importance once we commit ourselves to a
dynamic view.
appendix: a brief hist ory of stat iona ry
real-p rice equilibria
Te possibility of an equilibrium at which prices and wages are falling at the
same rate was noted by perceptive readers of Keynes, but they didn’t know
what to make of it other than to reject the idea on empirical grounds. Haber-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 215
ler seems to have been the frst, at least the frst to take notice in print of this
possibility:
A logical possibility would, of course, be that all money expressions (prices,
wages, money values) fall continuously, while the real magnitudes including
employment remain the same. Tat would be the implication of the assump-
tion that the Keynesian relations remain unchanged in real terms in the face
of such a situation. But this case is surely too unrealistic to be seriously con-
templated.
Tis observation c omes as a footnote to Haberler’s discussion of the role of
wage and price rigidity in Keynes and his predecessors:
According to a widely held view, which can be described as a sort of simpli
fed, popular Keynesianism, the possibility of underemployment equilib-
rium has been denied by the “classical” school and demonstrated by Keynes.
Te matter is, however, not so simple as that. Tis becomes quite clear if we
refect upon the intricate and crucial question concerning the role of wage
(and price) rigidity in the Keynesian system. Keynes assumes that (money)
wages are rigid downward. If this assumption, which is certainly not unreal-
istic, is rigidly adhered to, most of his conclusions follow: Underemploy-
ment equilibrium is then possible; an increase in the propensity to consume
will then reduce unemployment and a decrease in the propensity to con-
sume will produce unemployment (except if, as many classical writers as-
sumed, the demand for idle funds, the liquidity preference proper, is entirely
inelastic with respect to the rate of interest). But all this is entirely in accord
with pre-Keynesian theory, although these conclusions certainly had not
been generally realized and suf ficiently emphasized before the appearance of
Te General Teory.
If fexible wages—“thoroughgoing competition between wage earners” (in
Pigou’s words)—are assumed, the situation is radically changed. Obviously,
Copyright © 2021. Harvard University Press. All rights reserved.
underemployment equilibrium with fexible wages is impossible—wages
and prices must then fall continuously, which can hardly occur without fur-
ther consequences and cannot well be described as an equilibrium position.
(1946, pp. 190–191)
Lawrence Klein made a similar observation in 1947:
Within the framework of Keynesian economics wage fexibility does not
correct unemployment and leads merely to hyper-defation if carried to its
logical conclusion. But in the real world one observes neither hyper-defation
nor full employment. Te explanation is that wages are sticky; they are not
fexible . . .
When imperfections and rigid money wages are introduced into our
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
216 Raising Keynes
model of the Keynesian system, the results of under-employment equilib-
rium follow quite easily, providing a more realistic picture of how the eco-
nomic system looks. But it is not true, as many have said, that the Keynesian
equations in conjunction with a perfect, frictionless system will always yield
a full-employment solution. Unemployment is extremely likely even under
perfect competition. (p. 90)
Patinkin characterized Keynes’s theory of unemployment as a process of
slow disequilibrium adjustment, rather than as a stationary-price equilib-
rium:
Keynesian economics is the economics of unemployment disequilibrium. It
argues that as a result of interest-inelasticity, on the one hand, and distribu-
tion and expectation efects, on the other, the dynamic process of [a cap
italist economy lef to its own devices]—even when aided by monetary pol-
icy—is unlikely to converge either smoothly or rapidly to the full-employment
equilibrium position. Indeed, if these infuences are suf ficiently strong, they
may even render this process unstable. In such a case the return to full em-
ployment would have to await the fortunate advent of some exogenous force
that would expand aggregate demand suf ficiently. (1965, pp. 337–338)
Te diference between Patinkin and Haberler is at least partly semantic:
Patinkin, like Haberler, refuses to treat a situation in which prices and wages
are falling as a true equilibrium. Te crucial passage is
All, then, that Keynes means by the statement that the system may settle
down to a position of “unemployment equilibrium” is that the automatic
workings of the system will not restore the system to a position of full-
employment equilibrium. He does not mean “equilibrium” in the usual sense
of a term that nothing tends to change in the system. All that is strictly in
equilibrium is the level—or, possibly, only the fact—of unemployment; but
there is no equilibrium of the money wage rate. (1965, note K:3, pp. 643–
Copyright © 2021. Harvard University Press. All rights reserved.
644)
In an unpublished paper written in 1958, Clower echoes both Haberler and
Patinkin:
Perhaps the most curious aspect of the matter is the fact that if w and p [the
money wage and the money price of goods] just happen to fall at the same
rate over time then, starting from an initial position of Keynesian equilib-
rium (with excess supply in the labor market), the economy will remain “in
equilibrium” indefnitely although prices and wages are constantly falling
over time! Under these circumstances, it is perhaps natural to speak of the
diference N − N [labor supply less labor demand] as “involuntary unem-
s d
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • the general theory without rigid prices and wages 217
ployment”; but it [is] a curious of language to refer to the situation as a
whole as one of equilibrium. (p. 13; quoted in Plassard 2018, p. 279–280)
In this paper Clower explores dynamics akin to the disequilibrium dynamics
that characterize equilibrium in my models.11
As to whether or not “this case is surely too unrealistic to be seriously con-
templated,” as Haberler believed, see chapter 8. Klein, by the way, is inaccu-
rate in assuming that the “logical conclusion” of an equilibrium with falling
prices and wages is “hyper-defation,” if hyper-defation means an increasing
rate of defation. Te models in this chapter lead rather to a constant rate of
defation in a depression equilibrium. What Klein presumably meant was that
the equilibrium entails prices and wages that approach zero as time goes to
infnity. Te point is that—contrary to the mainstream view—there are no
endogenous forces in either the fxprice or the fexprice model that provide an
antidote. In this sense, prices and wages falling forever is indeed a prediction
of the model—chapter 8 examines how well it characterizes what actually
happened during the Great Depression
Although I was ignorant of these earlier models, in the process of research
and writing I came to know not only of Clower’s 1958 model but also of three
other attempts to model the dynamics of adjustment and the resulting equi-
librium. Te frst is by Robert Solow and Joseph Stiglitz (1968). Tough the
details difer, the Solow–Stiglitz model is very similar to my own in that the
equilibrium is characterized by a stationary real price as well as stationary
output, along with nominal prices and wages changing at the same rate.12
James Tobin’s 1975 article, from which I have included an excerpt in the
epigraph to this chapter, is in some ways more like my model and in some
ways less. In particular, Tobin explores alternative disequilibrium processes.
A major diference is that Tobin pays scant attention to the labor market. And
he characterizes equilibrium by a constant nominal price, so that the possibil-
ity of a self-sustaining depression (or boom) is tied to an unstable equilib-
Copyright © 2021. Harvard University Press. All rights reserved.
rium. In this sense, the Tobin model is more like the dynamic version of
Modigliani’s 1944 model that I develop in the mathematical appendix to
chapter 7—but with the Keynes efect replaced by the combined efects of real
balance and Fisher efects.
Finally, in the 1980s Rodolphe Dos Santos Ferreira and Philippe Michel
worked out a dynamic model that resolves the overdetermination in Figure
6.2 by introducing trade u nions. In this model, u
nions intermediate between
workers and employers, so that there are in efect two LS schedules, the (rep-
resentative) worker’s and the composite union’s. Te union LS schedule takes
account of the level of employment and the corresponding real wage. Te re-
sulting unemployment equilibrium corresponds to the u nion labor supply,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
218 Raising Keynes
and in that sense refects voluntary unemployment. Involuntary unemploy-
ment is either a disequilibrium phenomenon or it is the result of a union fo-
cus on relative wages that leads to abdication of overall wage setting to frms,
which in this model vary wages in line with changes in unemployment rather
than in response to its level. (Like Tobin’s, the model of Dos Santos Ferreira
and Michel is a version of Modigliani’s variable wage model. At an equilib-
rium without involuntary unemployment, aggregate demand plays only the
role of determining nominal price and wage levels.) Tis work circulated in
draf form from at least 1987 but was published only in 2013. (See also Dos
Santos Ferreira 2014.)
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Te third-pass model adds just one equation, representing the money-wage
response to unemployment, to the dynamic version of the frst-pass model
considered at the end of chapter 5. Te fexprice version of the complete
model is
P
= θ1[I ( ρ) − sY ], (6.1)
P
P
Y = 2 − FL−1 , (6.2)
W
W P
= 3 Y − LS , (6.3)
W W
where LS(P/W) = F(K, L(P/W)). However, equilibrium is characterized not by
stationary P and W, but by stationary P/W. From equations (6.1) and (6.3) we
have
•
P
W
≡ P − W
P
P = θ1[I ( ρ) − sY ] − θ 3 Y − LS . (6.4)
P W W
W
Te fexprice system is made up of equations (6.4) and (6.2). Te equilib-
rium is stable since tr J < 0 and det J > 0, where
Copyright © 2021. Harvard University Press. All rights reserved.
3 LS′ −1s − 3 3 LS′ − 1 s − 3
J = = .
2 2 FL−3 FLL 2 − 2 (GS′)−1
Te discriminant is
2
3 LS′ + 2 (GS′)−1 − 4 ( 1s + 3 ) 2.
As note 7 in the text suggested, the path to equilibrium may be direct or con-
voluted—the equilibrium may be a node or a focus—depending on the sign
of the discriminant, positive for a node and negative for a focus. Te discrim-
inant above is negative if 1 is large relative to 2 and 3; or if the elasticity of
219
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
220 Raising Keynes
labor supply is high, so that LS′ is large; or if the elasticity of goods supply is
low, so that GS′ is small. In these cases price adjustment will cause the econ-
omy to overshoot the equilibrium, but by progressively less over time.
Te position of the stationary real-price locus depends on the relative
speeds of adjustment of money prices and money wages. Te higher the ratio
1/3, the closer it is to the AD schedule. As this ratio increases, the response
to any departure of expenditure from income gets larger and larger, so the sta
tionary real-price locus is tied up ever more strongly with the AD schedule.
Te same logic in reverse means that as 1/3 falls, the stationary real-price
locus gets closer to the LS schedule.
In the limit as 1/3 → ∞, the stationary real-price locus fuses with the AD
schedule, and the LS schedule drops out of the picture. We are back to the
frst-pass model. Conversely, as 1/3 → 0, equilibrium is determined by the
LS and GS schedules; aggregate demand drops out of the picture.
Fixprice adjustment reverses the processing of the signals of demand pres-
sure and proftability. We have
P P
= −1 GS − Y , (6.5)
P W
Y = θ 2[I ( ρ) − sY ], (6.6)
W P
= 3 Y − LS . (6.7)
W W
Te stationary real-price locus is now
•
P
W
≡ P − W
P P
P = −1 GS − Y − 3 Y − LS . (6.8)
P W W W
W
Copyright © 2021. Harvard University Press. All rights reserved.
Te Jacobian of the system composed of equations (6.6) and (6.8) is
−1GS′ + 3 LS′ 1 − 3
J = ,
0 − 2 s
for which once again the trace is negative and the determinant positive. Te
discriminant is
( −1GS′ + 3 LS′ + 2 s ) ,
2
which is necessarily positive; the equilibrium is invariably a node.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
6 • Mathematical Appendix 221
As in the fexprice case, the distance of the stationary real-price locus from
the stationary nominal-price and nominal-wage schedules depends on rela-
tive speeds of adjustment of prices and wages. As 1/3 → ∞, the LS schedule
drops out of the picture, and equilibrium is determined by the AD and GS
schedules; as 1/3 → 0, the GS schedule drops out, and equilibrium is deter-
mined by aggregate demand and labor supply.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 7 .
dynamics vs. statics
Can the Economy Get from the Here of Unemployment
to the There of Full Employment?
Te real source of trouble is the confusion between comparisons of
equilibrium positions and the history of a process.
—joan robinson
One of the ways in which history and historical time matter for
macroeconomics is that they can afect the adjustment process very
intimately. Financial and other commitments from the past, and
expectations and apprehensions about the future, can easily interfere
with the economy’s ability to adapt to changes in aggregate demand
and supply.
—robert solow
We fnd ourselves confronted with this paradox: in order for the
comparative-statics analysis to yield fruitful results, we must frst
develop a theory of dynamics.
—paul samuelson
I can imagine a reader asking what has been accomplished by the long explo-
Copyright © 2021. Harvard University Press. All rights reserved.
ration of adjustment mechanisms in the previous two chapters. Yes, she might
say, it has been shown that the adjustment process determines the equilib-
rium, but how important are the insights that emerge from this new way of
looking at the economy? And, indeed, is it so new? Te focus is still on equi-
librium, to which the adjustment process eventually converges. Moreover, the
analysis removes one rigidity—the money wage—only to replace it with an-
other—a fxed interest rate.
Tere would be some truth in the criticism if we were at the end of the
story rather than the middle. It might be helpful to recall Paul Samuelson’s
report of the musings of his teacher Edwin Bidwell Wilson on the relativity of
equilibrium:
222
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 223
You leave your car in the MIT parking lot overnight. Te rubber tire is a
membrane which separates the inside of the tire from the atmosphere, and
because of this stif wall there’s an equilibrium diference in pressure. Wilson
would say, “Come back a thousand years later, and that tire will be fat.” Tat
was not strict equilibrium. It’s just a very slowly adjusting disequilibrium.
(Colander and Landreth 1996, p. 163)
Te same might be said about an equilibrium at which both money wages
and money prices continue to fall (or rise). Is such an equilibrium simply “a
slowly adjusting disequilibrium,” and what can be said about the continuing
disequilibrium process? In particular, what happens to aggregate demand
when the price level changes? In this chapter we examine four possible ave-
nues by which price changes in real time may have a diferent impact from
that of the virtual changes associated with comparative statics: the Keynes ef-
fect, the efect of fractional-reserve banking, the real-balance efect, and the
Fisher efect.
The Keynes Efect in a World of Commodity Money
For starters, what happens to the interest rate and investment demand in a
world of commodity money? We have seen in chapter 3 that a lower price
level frees up money from its role in facilitating commerce and production
and makes correspondingly more money available to satisfy the liquidity de-
mands of wealth holders. Tese wealth holders willingly accommodate the
extra cash only if the interest rate falls.
In chapter 6 I fnessed this complication by assuming the interest rate is
given, but is this a plausible assumption alongside an equilibrium at which
the nominal price level is continually falling? In the laundry list of reasons
Keynes provides in chapter 19 of Te General Teory why (and why not) fall-
ing money wages might solve the problem of unemployment, he includes the
Copyright © 2021. Harvard University Press. All rights reserved.
efect of an accompanying fall in the price level. Te resulting downward pres-
sure on the rate of interest—the Keynes efect—is key to Modigliani’s argu-
ment laid out in chapter 4.
So instead of a fxed interest rate, make the same assumptions that Modi
gliani made in constructing his version of the second-pass model, namely, a
fxed money supply and the adjustment of the interest rate to the price level.
But here we apply these assumptions in the context of money prices and
wages that change in real time rather than varying on diferent planets.
How do we refect this assumption graphically? Te move from the frst-
pass to the second-pass model in chapter 3 involved a replacement of a verti-
cal AD schedule with a downward-sloping one, but this modifcation won’t
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
224 Raising Keynes
work here. Te reason is that the price variable is now the real price, not the
nominal price, and it is the nominal-price level that infuences the interest
rate and investment demand via the Keynes efect. As the nominal price
changes, the vertical AD schedule moves in Y × P/W space, rather than, as in
the static second-pass model, a change in the nominal price causing a move-
ment along the downward-sloping AD schedule.
Output thus chases a moving target as it is pulled from an original position
of serendipity by a shock to aggregate demand. Afer its initial fall, the vertical
AD schedule reverses course and moves to the right (lower prices, lower in-
terest rate, more investment demand, more aggregate demand), in the direc-
tion of the full-employment equilibrium. In Figure 7.1, which assumes fx-
price adjustment, the frst impact of the shifing AD schedule is to slow down
the contraction of the economy. As the price level continues to fall and the
AD schedule continues to shif to the right, contraction turns into expansion;
at point A the economy’s trajectory is momentarily tangent to the moving AD
schedule. Te trajectory is vertical at this point, because expenditure is equal
to income and there is no reason for the economy either to contract or to ex-
pand. (Similarly, the economy’s trajectory is horizontal where it cuts the sta
tionary price locus.)
Te question is whether the economy converges to full employment, that is,
whether the dynamic adjustment path approaches the equilibrium E that
characterizes Modigliani’s version of the second-pass model. Tis is the only
possible equilibrium point in this model, because stationary output requires a
Y 0
P 0
Aggregate demand Supply of goods
P
W
Stationary real price
5
P 0
E
Copyright © 2021. Harvard University Press. All rights reserved.
W
4
A
3
W 0
2
Supply of labor
1
100 200 300 400 500 600 Y
Figure 7.1 Fixprice adjustment with a shifing aggregate-demand schedule.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 225
stationary interest rate, which in turn requires a stationary nominal price. So
the only feasible equilibrium on the stationary real-price locus is the original
starting point, E.
Convergence depends on how the adjustment process is specifed. In the
mathematical appendix to this chapter, the fexprice model is shown to con-
verge to full employment however rapidly one variable or another responds to
the pressure of disequilibrium. With fxprice models, convergence depends
on parameter values. Te economy does not return to full employment if
(paradoxically) money wages adjust very quickly to unemployment compared
with how rapidly money prices adjust to being of the GS schedule and if the
high speed of wage adjustment is complemented by a low elasticity of labor
supply. Under these conditions, the economy oscillates in an explosive orbit
around the full-employment equilibrium, a possibility Keynes intuited but
could not establish with the tools at his disposal: “If competition between
unemployed workers always led to a very great reduction of the money-wage,
there would be a violent instability in the price-level” (Te General Teory,
p. 253).1
Te picture is in Figure 7.2. Te explosive orbit is the trajectory of output
and real price, the greyhound following the mechanical rabbit of aggregate
demand. Te invisible AD schedule is pulled forward by a falling price level
and pushed back by a rising one, in an ever-widening orbit.
Alternatively, the economy might stall out because of a foor to the hurdle
P 0
W
P Stationary real price
W P 0
Supply of goods
5 P 0
W
Copyright © 2021. Harvard University Press. All rights reserved.
4
P 0
3
W
1 W 0
Supply of labor
100 200 300 400 500 600 Y
Figure 7.2 Fixprice adjustment with a shifing aggregate-demand schedule, unstable
equilibrium.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
226 Raising Keynes
Y 0
P 0
Aggregate demand
Supply of goods
P
W Stationary real price
5 P 0
W
4 E
W 0
2
Supply of labor
1
100 200 300 400 500 600 Y
Figure 7.3 Fixprice adjustment with liquidity trap or inelastic investment demand.
rate of interest or because of inelastic investment demand—just as in Modiglia
ni’s version of the second-pass model. In these cases the rightward march of
aggregate demand stops before it reaches the full-employment level of output,
as shown in Figure 7.3 by the dashed AD schedule, with the equilibrium at E.
Tese dynamic results contradict Modigliani’s static conclusions about the
Keynes efect. Once serendipity is disturbed, the economy may, but it may
not, return to full employment. Absent Modigliani’s exceptional cases—a li-
quidity trap or an inelastic investment-demand schedule—it all depends on
the adjustment process, on which information drives which changes (fx-
price vs. fexprice). In the fexprice case, the equilibrium is always stable, but
with fxprice adjustment, stability depends on how rapidly diferent variables
(wages, prices, output) adjust.
Copyright © 2021. Harvard University Press. All rights reserved.
Fractional-Reserve Banking in Real Time
I concluded in chapter 4 that in a static constant fractional-reserve banking
did not afect the plausibility of Modigliani’s critique of Keynes. Tough the
details of the argument change—a bond efect does the work of the Keynes
efect—in a comparative-statics context, wart-free capitalism remains self-
regulating even if the money supply is endogenous.
Te analysis and conclusions are diferent if we examine real-time changes
rather than comparative statics. In the spirit of chapter 4, I shall assume here
that there is no central bank and that deposits of gold ducats by households
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 227
constitute the reserves of banks. In the real-time context, endogenizing trans-
actions money, M1, while assuming an exogenous supply of asset money, M2,
in efect converts the second-pass model back into the frst-pass model. As I
shall demonstrate, the liquidity-preference schedule is invariant with respect
to P, and therefore so is the interest rate. And so is the LM schedule. Te re-
sult is that the AD schedule is insensitive to the price level, just as in the frst-
pass model, and, indeed, as in the third-pass model developed in the previous
chapter.
It turns out that there is a strong assumption about the very essence of
money built into Modigliani’s critique, albeit an assumption to which Keynes’s
argumentation gives some support, namely, that money is a commodity, the
supply of which is fxed by nature, technology, and human ingenuity and ef-
fort.2 Once we depart from this assumption, Modigliani’s results need not
hold. Te money supply may be a nonbinding constraint.
To see why, we return to the relationship between transactions money and
portfolio money in a world of commodity money with no banking system at
all and therefore no possibility of endogenous creation of money. Te discus-
sion of chapter 4 carries over from the comparative-statics framework to the
real-time framework with one qualifcation, which turns out to matter when
we drop the assumption that money consists of gold, silver, or cowrie shells.
Te dynamic picture of the fnancial sector is in Figure 7.4, which reproduces
Figure 4.14, except that the liquidity-preference schedule does not shif when
the price level shifs.
Why not? Te diference between real-time changes and comparative-
statics lies in what it makes sense to assume about the stock of bonds in the
two cases. In both cases we assume the coupon (R) is fxed. In a comparative-
statics framework, we assume that the quantity of bonds varies with the
nominal value of capital, that is, with the price level; in other words, B = PK.
So, when the price level changes, the liquidity-preference schedule shifs. Tis
shif takes place because we are considering separate planets that have each in
Copyright © 2021. Harvard University Press. All rights reserved.
its own way been the same since time immemorial. By assumption, planets
have the same physical capital stock, so the appropriate assumption is that the
planet with the higher price level has a larger number of bonds outstanding,
the fnancial counterpart of the higher nominal value of its capital stock.
In a real-time context, there is no mechanism to change the quantity of
bonds as the price level changes. Whatever happens to the price level, the
stock of bonds remains B = P0K, where P0 is the price level when the story
begins. Bond prices will change as the interest rate changes, but this is not the
same thing as the quantity of bonds changing.
Te real-time context transforms Tables 4.1, 4.2, and 4.3 of chapter 4 into
the corresponding Tables 7.1, 7.2, and 7.3. Te change is in the last cell of
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
228
Interest rate ( ) Interest rate ( )
P=1
LM P = 0.75
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 2 0 1 100 200 300 400 500 Income (Y)
100 400 P=1
Table 7.3
Raising Keynes
200 300
Table 7.1 P = 0.75
Liquidity preference
300 200
P=1 Table 7.2
P = 0.75 400 100 Transactions demand
500 0
M2 M2 M1 M 1
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Figure 7.4 Te LM schedule with commodity money.
7 • Dynamics vs. Statics 229
Table 7.1 Y = 400; P = 0.75; = 1; = 0; PB = 1
Households Restaurants
Loans to
Nonliquid Assets Nonliquid Assets Working Finance
Liquid Assets (Bonds—Market (Loans to Restaurants Capital (Gold Fixed Working Quantity of
(Gold Ducats) Value) for Working Capital) Ducats) Capital Capital Bonds
200 1,000 300 300 1,000 −300 −1,000
Table 7.2 Y = 266.7; P = 0.75; = 1; = 1 = 0/1.2; PB = 1.2
Households Restaurants
Loans to
Nonliquid Assets Nonliquid Assets Working Finance
Liquid Assets (Bonds—Market (Loans to Restaurants Capital (Gold Fixed Working Quantity of
(Gold Ducats) Value) for Working Capital) Ducats) Capital Capital Bonds
300 1,200 200 200 1,000 −200 −1,000
Table 7.3 Y = 400; P = 1; = 1; = 2 = 0/0.8; PB = 0.8
Households Restaurants
Loans to
Nonliquid Assets Nonliquid Assets Working Finance
Liquid Assets (Bonds—Market (Loans to Restaurants Capital (Gold Fixed Working Quantity of
(Gold Ducats) Value) for Working Capital) Ducats) Capital Capital Bonds
100 800 400 400 1,333 −400 −1,000
able 7.3: there are now only 1,000 bonds (compared with 1,333 in Table 4.3)
T
despite the increase in the nominal value of fxed capital. As a result, the inter-
est rate 2 that equilibrates fnancial markets is lower than the corresponding
interest rate in Figure 4.14.
Copyright © 2021. Harvard University Press. All rights reserved.
Te consequence is that the only change in the demand for money is along
the given liquidity-preference schedule, driven by changes in the bond price.
For the moment this does not change the qualitative behavior of the LM
schedule, namely, that it slopes upward and shifs upward with the general
price level. Te LM schedule continues to respond to the price level because a
shif in the transactions-demand schedule is enough to move the LM sched-
ule. A fxed money supply is still assumed, as in chapter 4, to be divided be-
tween transactions money and portfolio money, with more of the one mean-
ing less of the other. Tus the intersection of a given IS schedule with the
shifing LM schedule continues to produce a downward-sloping AD schedule
in Y × P space, and Modigliani’s conclusion, if not his argument, may sur-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
230 Raising Keynes
vive—“may survive” rather than “survives” because the AD schedule is not
stationary in Y × P/W space, and, as we have seen, may or may not return to a
full-employment equilibrium.
In the comparative-statics context, it turned out not to matter very much
whether or not banks are fully loaned up or hold excess reserves. Either way,
the crucial element of the second-pass model holds: the LM schedule shifs
with the price level, and consequently, the AD schedule slopes downward in
Y × P space.
In real time, the position of the LM schedule is sensitive to whether or not
banks are fully loaned up. Te reason is that when banks have excess reserves
the LM schedule is invariant with respect to the price level. As a reference
point, recall the comparative-statics result, for which the relevant diagram is
Figure 7.5 (Figure 4.18), in which the LM schedule is fat for a given price
level but moves with the price level. In Figure 7.5 the LM schedule moves be-
cause the quantity of bonds shifs with the price level, so that the liquidity-
preference schedule shifs, even though there is by assumption no link be-
tween M1 and M 2. Since M 2 is unchanging, the rate of interest has to adjust to
the higher quantity of bonds. With a higher rate of interest, the market value
of bonds must be lower for agents willingly to hold the same amount of cash
as at the lower price level. As observed in chapter 4, the bond efect and the
Keynes efect both decrease the stock of bonds relative to portfolio cash at
lower price levels, but by a diferent mechanism: the bond efect is the result
of fewer bonds, the Keynes efect the result of more cash.
In real time, with the quantity of bonds fxed, the liquidity-preference
schedule does not move, so the price of bonds and the interest rate remain
unchanged when M 2 is fxed. So Table 7.4 is the same as Table 4.8, but Table
7.5 difers from Table 4.9. As a result, the LM schedule does not change with
the price level, as in Figure 7.6.
Now that the LM schedule is stationary, the AD schedule Y × P space is no
longer downward sloping. Instead, it is vertical, just as it is in the frst-pass
Copyright © 2021. Harvard University Press. All rights reserved.
model. Because interest and investment do not vary when prices change, ag-
gregate demand is invariant with respect to the price level. Tere is neither a
Keynes efect nor the substitute of a varying quantity of bonds. Unlike the
picture presented in Figures 7.1, 7.2, and 7.3, the AD schedule in Y × P/W
space remains stationary as the price level falls. Te economy is constrained
by liquidity preference even in the absence of a liquidity trap.
Fractional-reserve banking creates the possibility that money creation is
endogenous. But there remains the possibility that banks are driven by mar-
ket forces to utilize their reserves to the maximum possible extent, so that the
theoretical possibility is a dead-end. Reserves would determine the money
supply (via the money multiplier, 2.5 in chapter 4), to which money demand
would adapt, just as commodity money regulates demand in a world without
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 231
Interest rate ( ) Interest rate ( )
P=1
LM
P = 0.75
Interest rate ( ) Income (Y)
45°
Interest rate ( ) 5 0 100 200 300 400 500 Income (Y)
100 100
200 200
Liquidity preference P = 0.75
300 300
P=1
400 400 Transactions demand
P=1
500 500
P = 0.75
M2 M2 M1 M 1
Figure 7.5 Financial equilibrium: constructing the LM schedule with endogenous transactions
money, I.
Interest rate ( ) Interest rate ( )
P=1
LM
P = 0.75
Copyright © 2021. Harvard University Press. All rights reserved.
Interest rate ( ) 45° Income (Y)
Interest rate ( ) 0 100 200 300 400 500 Income (Y)
100 100
200 200
Liquidity preference P = 0.75
300 300
P=1
P = 0.75 400 400 Transactions demand
P=1
500 500
M2 M2 M1 M 1
Figure 7.6 Financial equilibrium: constructing the LM schedule with endogenous transactions
money, II.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
232
Table 7.4 Y = 266.67; P = 0.75; = 1; = 0; PB = 1
Households Banks Restaurants
Nonliquid Assets Nonliquid Assets
Bank (Bonds—Market Commercial (Bonds—Market Commercial Bank Bank Fixed Commercial Quantity
Deposits Value) Loans Value) Loans Reserves Deposits Deposits Capital Loans of Bonds
200 1,000 0 0 200 200 −400 200 1,000 −200 −1,000
Raising Keynes
Table 7.5 Y = 266.67; P = 1; = 1; = 0; PB = 1
Households Banks Restaurants
Nonliquid Assets Nonliquid Assets
Bank (Bonds—Market Commercial (Bonds—Market Commercial Bank Bank Fixed Commercial Quantity
Deposits Value) Loans Value) Loans Reserves Deposits Deposits Capital Loans of Bonds
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
200 1,000 0 0 266.67 200 −466.67 266.67 1,333 −266.67 −1,000
7 • Dynamics vs. Statics 233
banks. Chapter 4 did not try to resolve the tension between endogenous and
exogenous money, but that irresolution should be reconsidered in a world of
real-time changes.
In the comparative-statics context it was argued on the side of exogene-
ity that if banks lacked commercial-lending opportunities they would deploy
reserves to purchase bonds, remaining fully loaned up one way or another.
Tis assumption is more dif ficult to defend when change takes place in real
time. In a comparative-statics context, bond prices, like everything else, never
change even though the price of a bond is diferent on Planet A from what it
is on Planet B. But in the context of real-time changes, the fear of losses on
bonds can outweigh the lure of interest payments and lead bankers to hold
idle reserves instead of bonds.
Te danger of fuctuations in asset prices, not to mention the possibility of
outright default, is the reasoning behind the “real-bills” doctrine, a variant
of endogenous-money theory that its proponents thought would make the
money supply respond to “legitimate” trade and insulate the fnancial system
from aiding and abetting “illegitimate” speculation.3 Tis is not the place to
examine the history of monetary theory in general or the real-bills doctrine in
particular.4 For present purposes, the main problem with real bills as the basis
of a theory of endogenous money is the implicit assumption of bank passivity.
Even if banks have no other outlet for lending, they need not passively re-
spond to demand for commercial loans. In the case of a downturn, with
prices and output falling, expectations of a quick recovery may make the sup-
ply of transactions money less volatile than the demand, M1 = PY, whereas if
pessimism leads to expectations of continuing defation, the supply may fall
faster than the demand. Te next chapter examines how expectations played
out in the Great Depression.
Te argument that fractional-reserve banking is compatible with a fxed
money supply assumes not only that banks are fully loaned up but that re-
serves themselves are fxed. In chapter 4 gold ducats formed the base of the
Copyright © 2021. Harvard University Press. All rights reserved.
money supply; once deposited into the banking system by households, they
remain there. In analyzing real-time changes this may be another inappropri-
ate assumption. As prices fall, businesses and individuals who were struggling
to meet their fnancial obligations even when prices were steady may go un-
der. When they can’t repay their loans, their creditors, including the banks
that have provided working capital, also may go bust.5 In the absence of de-
posit insurance (which was not introduced in the United States until 1934 and
has never applied to large deposits), wealth holders may react to bank failures
by withdrawing their deposits, changing the composition of the monetary
base in favor of currency.6 So the result of a decline in the price level may be a
decline in reserves as well as a growth in excess reserves. Te money supply
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
234 Raising Keynes
may be endogenous simply because the public shifs from bank money to
specie.
The Real-Balance Efect: Haberler, Pigou,
and Patinkin—and Fisher
As laid out in chapter 4, the basic idea of the real-balance efect, also known
as the “Pigou efect,” starts from the assumption that the money supply, or at
least a fraction of it, forms part of private wealth, so a nominally fxed money
supply increases in purchasing power, in real value, as the price level falls. In
principle, at least in the static context, there are no limits to the amount by
which aggregate demand can expand, so aggregate demand can be whatever
is required for full employment. In the context of chapter 4, critics like Milton
Friedman could make this idea the killer argument against the theoretical
claims of Te General Teory. Even friends of Te General Teory accepted the
real-balance efect as disproving Keynes, at least at the theoretical level. Abba
Lerner, someone who grasped Te General Teory early and may have under-
stood its implications for fscal policy better than Keynes (see chapter 14),
wrote
Tere is a limit to hyperdefations in that there is a certain amount of money
in the economy which does not disappear in the course of the hyperdefa-
tion and which becomes more and more valuable as prices fall. A great deal
of money does disappear: much credit money, whether issued by banks or
by others, may be destroyed, and therefore you could get wages and prices
continuing to fall along with the terrible disruption of the economy. But
there is a core of money which does not disappear—gold and the hard
money provided by government. As the hyperdefation proceeds, the value
of the stock of money becomes so great and the people who own this money
are so rich that this will overcome their tendency to hoard the increasingly
valuable money. Tey will increase their demand for goods and services and
Copyright © 2021. Harvard University Press. All rights reserved.
this will put an end to the hyperdefation and to the depression. (Lerner
1951, p. 205)
Paul Samuelson (1964, p. 333), while cautioning against any practical ap-
plication of the real-balance efect, agreed that at a theoretical level Keynes
lost the argument: “[Te real-balance efect] did serve to save face and honor
for the believers in the harmony of equilibrium.”
At the same time, prominent Keynesians—James Tobin comes to mind—
marveled at the pivotal role critics attached to the real-balance efect: “To an
astonishing degree, the theoretical fraternity has taken the real-balance efect
to be a conclusive refutation of Keynes” (1993, p. 59; see also Tobin 1980).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 235
Tis raises an interesting question: how can grown-ups, all of them intelli-
gent, all of them wise, all of them learned in the Torah of mainstream eco-
nomics, all of them claiming to be scientifc, disagree so fundamentally about
the signifcance of the real-balance efect? Te answer lies squarely in the dif-
ference between comparative statics and dynamics.
One preliminary: all would eventually agree with Lerner’s stipulation that it
is not the total money supply, but only outside money, the monetary base of
bank reserves and currency, that is relevant for the real-balance efect. Tat is,
inside money created endogenously by banks under a system of fractional-
reserve banking is not included in the cash balances that are the basis of the
real-balance efect. Te reason is that inside money, money created by the
banks in the process of making loans, is a wash, with one dollar of debt cre-
ated for every dollar of credit. But this refnement afects only the quantitative
importance of the real-balance efect, not its logic and therefore not the logi-
cal efect of letting the price level go to zero.
Tis logic is impeccable in the comparative-statics context. Te progenitors
of the real-balance efect, starting with Gottfried Haberler, cast their argu-
ment in dynamic language, in terms of what happens when the price level
changes, but the symmetry assumed between creditors and debtors depends
critically, if tacitly, on a comparative-statics context. In the timeless world of
comparative statics, the burden of debt and the benefts enjoyed by creditors
are the same now as yesterday, the same now as tomorrow.
Arthur Pigou, the real one (as distinct from the straw man Keynes created
to represent the mainstream position on the supposed self-regulating prop-
erty of capitalism), independently put forward the real-balance efect in a
multipronged attack on Alvin Hansen’s book Fiscal Policy and Business Cycles
(1941). (Te point of attack was Hansen’s stagnationist view that a drying up
of investment opportunities would lead to an excess of desired saving over
investment demand at a full-employment level of output and that there would
be no way of mopping up the excess saving other than for the government to
Copyright © 2021. Harvard University Press. All rights reserved.
match private surpluses with public defcits.) Pigou’s argument (1943, esp.
para. 10–11, pp. 349–350) is essentially Harberler’s, and again, though the
language is dynamic, the model itself, put forward in paragraph 11, is purely
static. Pigou’s model leads to the conclusion that there exists a price level at
which aggregate demand will be whatever full employment requires—without
the exceptions of an interest-rate foor and inelastic investment demand.
But there is nothing in Pigou’s model about what happens along the way to
get to this serendipitous state. Michał Kalecki, the economist credited by
many with independently discovering the basic ideas of Te General Teory,
immediately pointed this out in a comment on Pigou’s paper (1944). Afer
observing that a fall in the price level would not increase the net value of real
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
236 Raising Keynes
balances insofar as it acts on inside money, Kalecki goes on to make the cru-
cial distinction between the static and dynamic contexts:
Te adjustment required would increase catastrophically the real value of
debts, and would consequently lead to wholesale bankruptcy and a “conf
dence crisis.” Te “adjustment” would probably never be carried to the end.
(p. 132)
Te point is that in contrast with the static comparison of debts and credits
at diferent price levels, the dynamics involve revaluation of debt. Debts taken
on at one price level must be repaid with dollars earned at a lower price level.
If a farmer borrows $10,000 to acquire land or equipment with wheat at $1.25
per bushel, and must repay the loan with wheat valued at $0.50, he must in
efect pay back 2.5 times the number of bushels borrowed, 20,000 against
8,000. On an aggregate accounting, this is balanced by the creditor’s windfall,
but the real consequences are not symmetric unless the magnitude of the
change is small. When the price level falls dramatically, the farmer may be
driven to bankruptcy, while the creditor adds marginally to his consumption.
As Kalecki concludes, “If the workers persisted in their game of unrestricted
competition [driving down wages and prices], the Government would intro-
duce a wage stop under the pressure of employers” (1944, p. 132).7 In a world
in which workers possess very little, it would necessarily be employers who
are threatened with bankruptcy!
Keynes had in fact made a similar observation in his chapter 19 laundry list
of objections to falling wages as a cure for unemployment:
Te depressing infuence on entrepreneurs of their greater burden of debt
may partly ofset any cheerful reactions from the reduction of wages. Indeed
if the fall of wages and prices goes far, the embarrassment of those entrepre-
neurs who are heavily indebted may soon reach the point of insolvency,—
with severely adverse efects on investment . . .
Te method of increasing the quantity of money in terms of wage-units
Copyright © 2021. Harvard University Press. All rights reserved.
by decreasing the wage-unit increases proportionately the burden of debt;
whereas the method of producing the same result by increasing the quantity
of money whilst leaving the wage unit unchanged has the opposite efect.
Having regard to the excessive burden of many types of debt, it can only be
an inexperienced person who would prefer the former. (Te General Teory,
pp. 264, 268–269)
But, as is ofen the case with laundry, it is easy to lose one item or another on
the list: Keynes’s observation seems to have gone unnoticed by the various
participants in the debate over real balances. And Keynes himself seems to
have been unaware of Irving Fisher’s pioneering work on the asymmetric im-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 237
pact of rising real liabilities and rising real assets. Fisher (1933) took this
asymmetry to be the cause of depressions in general and the Great Depression
in particular. It is worth quoting Fisher at length:
Assuming . . . , that, at some point of time, a state of over-indebtedness ex-
ists, this will tend to lead to liquidation, through the alarm either of debtors
or creditors or both. Ten we may deduce the following chain of con
sequences in nine links: (1) Debt liquidation leads to distress selling and to
(2) Contraction of deposit currency, as bank loans are paid of, and to a slow-
ing down of velocity of circulation. Tis contraction of deposits and of their
velocity, precipitated by distress selling, causes (3) A fall in the level of prices,
in other words, a swelling of the dollar. Assuming, as above stated, that this
fall of prices is not interfered with by refation or otherwise, there must be
(4) A still greater fall in the net worths of business, precipitating bankruptcies
and (5) A like fall in profts, which in a “capitalistic,” that is, a private-proft
society, leads the concerns which are running at a loss to make (6) A reduc-
tion in output, in trade and in employment of labor. Tese losses, bankrupt-
cies, and unemployment, lead to (7) Pessimism and loss of confdence, which
in turn lead to (8) Hoarding and slowing down still more the velocity of circu-
lation.
Te above eight changes cause (9) Complicated disturbances in the rates of
interest, in particular, a fall in the nominal, or money, rates and a rise in the
real, or commodity, rates of interest.
Evidently debt and defation go far toward explaining a great mass of phe-
nomena in a very simple logical way. (pp. 341–342; emphasis in original)
Fisher tried to ft his trenchant insight into the only framework he knew, the
quantity theory, but it d idn’t ft very well, and his analysis remained a schol-
arly footnote in the historiography of the analyses provoked by the Great De-
pression.
If we transpose the comparative-statics argument of chapter 4 to the con-
Copyright © 2021. Harvard University Press. All rights reserved.
text of real-time changes, it is easy to see how Fisher’s observations impact the
real-balance efect. Recall the expression for total wealth in a world of com-
modity money,
PB ( )B + M 2 + M1 + −B + PK − M1 + Working Capital.
(Households) (Firms)
In the comparative-statics context, frms’ assets and liabilities always match,
and wealth reduces to household assets. Real wealth accordingly is
M 2 M1
PB ( )K + + ,
P P
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
238 Raising Keynes
which, as we saw in chapter 4, becomes infnite as the price level approaches
zero. Even though the real value of bonds increases only insofar as the price of
bonds (PB( )) increases (because the issue of bonds is assumed to vary with
the price level), the real value of both M1 and M2 becomes infnite regardless
of what is assumed about interest rates and bond prices.
In the real-time context by contrast, B is fxed once and for all when bonds
are frst issued. Households are still getting ever richer in real terms with ev
ery fall in the price level. But the situation is very diferent for frms. PK, the
dollar value of physical capital, declines with the price level, while the dollar
value of frms’ liabilities (B) remains fxed.8 Accordingly, the real value of
frms’ net assets
B
− +K
P
moves further and further into negative territory as the price level declines,
and goes to −∞ as P goes to zero, even as the real value of household wealth
PB ( )B M 2 M1
+ +
P P P
increases without bound. Provided PB does not fall to zero, changes in P will
raise the level of real household wealth by more than the decline in the net
worth of frms because of the cash component of household wealth. If the
change in P is small, so that we can assume a uniform propensity to consume
for creditors and debtors, the overall real-balance efect will be positive. Over-
all wealth is
PB ( )B M 2 M1 B
+ + − + K,
P P P P
which increases as P falls.
If the fall in P continues too long, however, the calculus changes. Firms will
Copyright © 2021. Harvard University Press. All rights reserved.
go into bankruptcy, which not only will diminish the bond wealth of house-
holds but will cause disruptions to production and therefore to incomes. Te
real-balance efect may cause the economy to spin even more out of control
once debtors start going broke.
Fractional-reserve banking, and with it the distinction between outside and
inside money, doesn’t change the distinction between comparative statics and
changes in real time. In the model of chapter 4, M 2 is the outside money that
constitutes the reserves of the banking system, and M1 is created by the banks
as the counterpart of commercial loans to fnance working capital. Now nom-
inal wealth is
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 239
PB ( )B + M 2 + M 1 + M 2 − M1 − M 2 + −B + PK − M1 + Working Caapital.
(Households) (Banks) (Firms)
In the comparative-statics context, assets ofset liabilities on the books of
banks as well as on the books of frms, so that the qualitative efects of price-
level changes do not difer from the simpler case of commodity money. Inside
money, M1, disappears from the calculation of real wealth.
In the real-time context, the results are diferent. In our simple model,
banks are not the problem as long as they stick to commercial lending. Banks
guided by the real-bills doctrine will remain solvent if they shrink their
nominal lending, M1 = PY, in line with declining price level; presumably
working-capital requirements will vary inversely with prices. In reality, banks
get into trouble if prices fall so rapidly that the value of collateral falls below
the amount of the debt. Te stock market crash of 1929 is a case in point, but
stock-market debt was the antithesis of real bills.
Te problem, as in the previous case of commodity money, is the part of
frms’ balance sheets that deals with fxed capital. A frm’s fnancial position
will suffer from a falling price level because the debt issued against its physical
capital is assumed to be fxed in nominal terms. As before, the net position of
frms in real terms is
B
− + K.
P
Te net position of households difers from the commodity-money case only
in the absence of M1 in the formula for household real wealth,
PB ( )B M 2
+ .
P P
Apart from the implicit assumption that debtors value the bonds they owe
in terms of their original value (B/P), whereas bondholders mark their bonds
Copyright © 2021. Harvard University Press. All rights reserved.
to market (PB( )B/P), the efects of real-time changes in the price level on the
real value of debt and credit are symmetrical if we assume uniform propensi-
ties to consume out of wealth and income. In this case, the overall impact of
the real-balance efect on aggregate demand remains positive since house-
holds also have a real cash balance, (M 2/P).
But this simplifcation obscures the same real-time issues that arise in the
absence of fractional-reserve banking. Once changes in the price level be-
come large, there is every reason not to assume that debtors and creditors
have the same propensities to consume and save. Even if a uniform propen-
sity to consume is a defensible assumption in a comparative-statics frame-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
240 Raising Keynes
work, in real time agents can be expected to respond diferently to prosperity
than to privation. If for no other reason, liquidity constraints are most likely
to occur in times of defation and a rising real burden of debt. If the propen-
sity to consume is lower for creditors than for debtors, the positive multiplier
for windfalls enjoyed by creditors may be overwhelmed by the negative mul-
tipliers attaching to rising real debt.
Te above assumes that the debtors manage to survive economically. As
prices continue to fall, survival becomes less and less likely. As noted, there is
no limit to the real burden of debt as the price level goes to zero—except
bankruptcy. At this point, the rising burden of debt is no longer ofset by the
increasing wealth of coupon clippers. And the ephemeral quality of the credi-
tor’s wealth is revealed the moment the debtor’s IOU becomes worthless. Te
real-balance efect, far from being the savior of the system, turns out to be an
agent of its destruction.
Te problem does not stop with the destruction of the bond wealth of both
lender and borrower. Te creditor’s cash hoard continues to increase in real
value as P falls, but not by enough to counter the real efects of bankruptcy—
not only the disruption of production at the site of the bankruptcy, but the
whole chain of collateral damage, the efects on suppliers, customers, employ-
ees, as well as the banks, which discover that real bills do not insulate them
from default.
Te contagion caused by the ballooning of real debt has been understood
by macroeconomists taking their cue from Fisher (1933) (see, for example,
Bernanke 1983, 1995). Te link to the real-balance efect, particularly how the
growth of real debt vitiates Friedman’s claim that the real-balance efect un-
dermines the argument of Te General Teory, seems to have eluded the post-
Friedman generation.
Don Patinkin attempted to rescue the real-balance efect by shifing the
focus from money to the qualitative diferences between private and public
debt. Patinkin argues that in a real-time context the real-balance efect is not
Copyright © 2021. Harvard University Press. All rights reserved.
only, or primarily, about cash balances, but about bonds, specifcally about
government bonds, which unlike private bonds do not, at least not in Patin
kin’s telling, pit creditor and debtor against each other:
What constitutes the “cash balance” whose increase in real value provides
the stimulatory efect of the Pigou analysis . . . [is] the net obligation of the
government to the private sector of the economy. Tat is, it consists of the
sum of interest-and non-interest-bearing government debt held outside the
treasury and central bank. Tus, by excluding demand deposits and includ-
ing government interest-bearing debt, it difers completely from what is usu-
ally regarded as the stock of money. (1948, pp. 550–551)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 241
Patinkin’s argument implicitly recognizes that for a falling price level to
stimulate real consumption via an efect on wealth, there must be an asset that
satisfes two conditions. First, the real value of the asset must vary in the op-
posite direction of the price level. Second, the asset must have no counterpart
as a liability. Te frst condition means that most physical assets would not
qualify: plant and equipment, houses, infrastructure—the money values of all
such assets can be expected to vary more or less in line with the overall price
level, so that their real values would, at least to a frst approximation, be in-
variant with respect to prices.9 Neither does working capital qualify. For simi-
lar reasons, equities would hardly be a candidate for inducing real-balance
efects; a stock market boom might indeed stimulate consumption demand,
and a crash might even provoke a crisis, but such movements are not conse-
quences of changes in the general price level.
Tis leaves us with debt fxed in nominal terms. For present purposes we
can rule out debt on working capital by assuming that it is suf ficiently short
term that a falling price level does not impair the capacity of frms to repay
their loans, and that as loans are rolled over, the fall in demand for loans is
matched by a fall in the supply of short-term credit.
Tere remain bonds, which satisfy the frst condition of varying in real
value inversely with the goods-price level. Te problem with bonds is that
they run up against the second condition, namely, that there be no liability
corresponding to the asset. On the contrary, for every bond owner there is a
debtor, and it has been observed that increases in the real value of bonds can
have disastrous efects on borrowers, efects not likely to be ofset by the addi-
tion to lenders’ real wealth.
Te one exception, the only asset that meets both of Patinkin’s conditions,
is a government bond, to which, indeed, he assimilates (outside) money, as a
form of noninterest-bearing government debt. Tis seems more promising as
a basis for the real-balance efect in real time, but the implicit assumptions in
the argument cannot go unremarked.
Copyright © 2021. Harvard University Press. All rights reserved.
Te distinguishing feature of assets that lead to a real-balance efect is now
bound up with the behavior of diferent classes of agents: governments, unlike
private borrowers, are supposed to maintain the status quo ante in terms of
taxing and spending even in the light of debilitating price changes. In Patin
kin’s words, “If we assume that government activity is not afected by the
movements of the absolute price level, then the net efect of a price decline
must always be stimulatory” (1948, p. 550).
To be sure, sovereign governments do not normally declare bankruptcy,
but short of bankruptcy there are budgetary obstacles to maintaining the
nominal value of government bonds. Again, the diference between the static
and dynamic analysis is critical. In a static analysis, it is difficult to see any
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
242 Raising Keynes
rationale for Patinkin’s supposed insensitivity of the nominal stock of govern-
ment debt to the level of prices. In a static model, real bonded indebtedness
should in principle be independent of the price level, which is to say that
nominal indebtedness should vary with the price level. If on Planet A the
price level is and always has been 1 and the public debt is 500, on what
grounds would the public debt in Planet B—identical in every respect to A
except that the price level is and always has been 2—be anything but 1,000?
In the dynamic analysis, by contrast, it is more appropriate to assume that
the quantity of government bonds is fxed once and for all when the debt is
issued, just as we must assume for private debt. Patinkin’s point is that unlike
private debt, the resulting increase in the real public debt as P falls need not
lead to adverse consequences for the economy. Tis is presumably what is
meant by “[assuming] that government activity is not afected by the move-
ments of the absolute price level.” But if spending and taxation continue at the
same levels in real terms, how does the government hold to its commitment
to pay the interest and principal stipulated in nominal terms—which afer all
is the only basis on which the nominal value of government bonds can be
maintained?
Tere is, to be sure, more than one way the government can uphold its
nominal commitments, but not by business as usual, not by assuming that
government activity is unafected; for unless the government takes positive
steps to counter the impact of a declining price level, tax revenues must de-
cline along with the value of nominal output and income and, eventually, the
value of property.
One thing the government can do is to raise taxes. But raising taxes will
impose a drag on aggregate demand that will ofset the stimulus of the rising
real value of government bonds.10 Or, more realistically, the government can
allow the real-balance efect to work—by borrowing more in order to meet its
interest obligations!
But think what this means: Patinkin’s version of the real-balance efect now
Copyright © 2021. Harvard University Press. All rights reserved.
hinges on a commitment to defcit fnance, and the proposition that there is
no limit to the amount by which aggregate demand can be expanded via the
real-balance efect turns out to be a proposition about the inability of unem-
ployment to persist in the presence of suf ficient fscal stimulus. In this form,
the real-balance efect may be alive and well. But it can hardly be what Milton
Friedman (1970) had in mind when he dismissed Keynes’s claim to have es-
tablished the possibility of unemployment in a perfectly competitive econ-
omy:
[Te proposition that] as a purely theoretical matter, there need not exist,
even if all prices are fexible, a long-run equilibrium position characterized
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 243
by “full employment” of resources . . . can be treated summarily because it
has been demonstrated to be false. (pp. 206–207; see chapter 4 for the com-
plete quotation)
Perhaps we are too quick to accept the assumption that government bonds
do not pit the interests of debtors against the interests of creditors. Te early
years of the Roosevelt Administration provide considerable food for thought
in this respect, particularly the “gold-clause” cases, Perry v. United States (294
U.S. 330, 1935) and Norman v. Baltimore and Ohio Railroad Co. (294 U.S. 240,
1935), which come within a whisker of a “natural experiment” with regard to
the plausibility of Patinkin’s behavioral assumptions.
Tese cases revolved around the constitutional legality of various con
gressional and presidential actions nullifying a standard clause in debt con-
tracts—bonds, mortgages, etc.—that called for repayment in dollars refecting
the gold value of the dollar at the time the debt was contracted. A gold clause
became part of bond boilerplate in the United States in the afermath of the
Civil War, when gold coins and “greenbacks” circulated side by side, both le-
gal tender, even though greenbacks exchanged for gold at a discount.
Prior to 1933, the value of gold was $20.67 per troy ounce, but soon afer
his inauguration, President Roosevelt efectively broke the link to gold by for-
bidding U.S. citizens to own gold, and Congress by joint resolution abrogated
the gold clauses in both private and public debts. Tese actions were further
codifed by various acts of Congress, culminating in the Gold Reserve Act of
1934. One purpose of all this legislation was to reduce the gold content of the
dollar in an attempt to put a halt to the spiraling defation. During 1933 the
president was authorized to change the gold content of the dollar within cer-
tain limits, until under the authority of the Gold Reserve Act he fxed the
price of gold to $35.00 per ounce, an increase of 69 percent over its pre-1933
level.
Creditors challenged the validity of the abrogation of gold clauses in both
private (Norman) and public (Perry) obligations. Te cases worked their way
Copyright © 2021. Harvard University Press. All rights reserved.
up through the courts and fnally reached the Supreme Court in 1935. Te
Court held by a 5–4 majority that the government was acting legally if not
morally in nullifying both private and public gold clauses. Norman did not
touch Patinkin’s version of the real-balance efect directly, because it involved
a transfer of wealth between private creditors and private debtors. But it pow-
erfully raised the question of the asymmetry between debtors and credi-
tors. As Roosevelt was planning to inform the nation had Norman gone the
other way,
If the letter of the law is so declared and enforced, it would automatically
throw practically all the railroads of the United States into bankruptcy. . . .
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
244 Raising Keynes
Te principle laid down today in the railroad case applies to every other
corporation which has gold bonds outstanding, driving many another huge
enterprise into receivership! It must be applied likewise to the obligations of
towns, cities, counties, and states; and these units of government, now work-
ing bravely to meet and reduce their debts, would be forced into the position
of defaulters. (quoted in Magliocca 2011, p. 33)
Even if we put to one side the asymmetrical efects of price changes on pri-
vate debtors and creditors, there remains Perry, which involved the essence of
the real-balance efect as interpreted by Patinkin, namely, the creation of pri-
vate wealth to the detriment of the government’s own balance sheet.11 In the
event, the Attorney General’s argument to the Supreme Court made no men-
tion of the benefts that might be enjoyed by allowing coupon clippers to
prevail. Rather, he stressed the harm to the economy and the government.
Roosevelt himself was prepared to reject the Court’s decision. Here are his
intended words:
Te actual enforcement of the gold clause against the Government of the
United States will not bankrupt the Government. It will increase our na-
tional debt by approximately nine billions of dollars. It means that this ad-
ditional sum must eventually be raised by additional taxation. In our present
major efort to get out of the depression, to put people to work, to restore
industry and agriculture, the literal enforcement of this opinion would not
only retard our eforts, but would put the Government and 125,000,000
people into an infnitely more serious economic plight than we have yet ex-
perienced . . .
To stand idly by and to permit the decision of the Supreme Court to be
carried through to its logical and inescapable conclusion would so imperil
the economic and political security of this nation that the legislative and
executive of ficers of the Government must look beyond the narrow letter of
contractual obligations. (quoted in Magliocca 2011, pp. 34–35)
Copyright © 2021. Harvard University Press. All rights reserved.
Tis particular “freside chat” was never given because the decisions in Nor-
man and Perry went in the government’s favor, but it reveals that Roosevelt
was willing to suspend the constitutional order because of his perception of
the catastrophic consequences of enforcing the law, a position not taken by
any American president since Lincoln suspended habeas corpus during the
Civil War. It apparently never occurred to any of the actors, certainly not to
Roosevelt, that the increase in private wealth might have a salutary efect on
the economy—or that (pace Professor Patinkin) the government’s operations
could not but be afected by an increase in the real value of its debt!
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 245
The Fisher Efect
Up to now we have been analyzing the efect on aggregate demand of price-
level changes via two routes, the efect on investment demand of the interest
rate, and the efect on consumption demand of wealth. It should now be clear
that moving from a static to a dynamic context transforms anti-Keynesian
arguments into much more tame, not to say lame, parodies of their static
selves. Indeed, in a dynamic context, a lower price level is likely to depress
demand rather than to increase it.
In this section we face a diferent challenge: the Fisher efect has no coun-
terpart in the static context, for it is the efect of a changing price level on the
interest rate—not a comparison of interest rates appropriate to two diferent
price levels. Te Fisher efect (I. Fisher 1896, 1930 [1907]) is the diference
between the so-called real rate of interest and the nominal rate of interest, a
diference that arises only if the price level is changing over time. Here “real”
means what it has meant throughout this study, value in terms of goods and
services, as distinct from value in terms of money: as its name suggests, the
real rate of interest captures the time premium attaching to present goods over
future goods, in contrast with the nominal rate of interest, which refects the
time premium of present money over future money.
In principle, as Fisher observed way back in 1896, there is a diferent inter-
est rate for each and every good: the interest rate for wheat will be the same as
the interest rate for oil if and only if the relative prices of wheat and oil do not
change. In general
interest rate in interest rate in rate of change of price of
= −
terms of wheat terms of oil wheat in terms of oil
WHEAT = OIL − W/O ,
where WHEAT is the wheat-specifc rate of interest, what Keynes in his enig-
matic chapter 17 calls the own rate of interest; OIL is the own rate of interest
Copyright © 2021. Harvard University Press. All rights reserved.
for oil; and W/O is the percentage rate of change of the price of wheat in terms
of oil, which is to say the diference between the rate of change of the dollar
price of wheat and the rate of change of the dollar price of oil. In other words,
the interest rate on two diferent goods will difer by the rate at which their
relative price changes.
To understand the logic of this proposition, think of goods as potential
stores of value. For oil and wheat to coexist as stores of value requires that
agents be indiferent between holding wealth in the form of promissory notes
in wheat and holding wealth in the form of promissory notes in oil. Ignoring
storage costs and the like, and assuming price changes are known in advance,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
246 Raising Keynes
for the two forms of wealth-holding to be equivalent in agents’ eyes, the wheat
rate of interest—the premium on present wheat relative to future wheat—can
difer from the oil rate of interest only by an amount equal to the rate at which
the price of oil changes relative to the price of wheat.
We can also think of the equation WHEAT = OIL − W/O as an equilibrium
condition that must hold if investment takes place both by sacrifcing some
present oil for future oil and by sacrifcing present wheat for future wheat. Te
right-hand side of the equation above is the return on an investment in oil,
and the lef-hand side is the return on an investment in wheat, both expressed
in terms of wheat. Te two investments must yield equal returns if the alloca-
tion of overall investment is in equilibrium, with both oil and wheat receiving
positive doses of investment.
For example, suppose the oil rate of interest is 6 percent and the wheat rate
of interest is 9 percent. If the relative price of the two goods is unchanging, all
agents would prefer to hold wealth in the form of promises to pay wheat and
to invest in wheat projects than to lend or invest in oil. For even if their ulti-
mate goal is to consume oil, they can do better by swapping present oil for
present wheat—let’s assume that the prices start out at $1.00 per bushel of
wheat and $1.00 per barrel of oil—and lending present wheat against future
wheat. In this way, agents can reap 109 bushels of wheat next year for every
100 barrels of oil they hold at present. If relative prices are unchanging, they
can then reswap the 109 bushels of wheat for 109 barrels of oil. By contrast, if
they directly lend present oil against future oil, there will be only 106 barrels
of future oil to show for every 100 barrels.
However, if (and only if) the price of wheat relative to oil falls by 3 per-
cent—let’s say oil remains at $1.00 while wheat falls to $0.97—the 109 bushels
of wheat will only buy 106 barrels of oil next year, with the result that lending
wheat versus lending oil is a wash, and agents will be indiferent between the
two forms of wealth holding. A similar story can be told for investment that
adds to the capacity to produce wheat or oil.12
Copyright © 2021. Harvard University Press. All rights reserved.
Te third-pass model developed in chapter 6 provides a special case of the
general proposition encapsulated in the equation WHEAT = OIL − W/O—a
special case in which bonds replace oil and “goods” replace wheat. If the
nominal rate on bonds is NOMINAL and the percentage rate at which the price
level is changing is , the real rate on bonds, REAL, is given by the formula
real rate = nominal rate − rate of change of price level
REAL = NOMINAL − .
For example, if the nominal interest rate is 5 percent per year but prices are
climbing at the rate of 4 percent, then the real interest rate is 1 percent. If
$100.00 is lent at interest, it will return $105.00 at the end of the year, but a
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 247
basket of goods that cost $100.00 at the beginning of the period will cost
$104.00 at the end. Te return in terms of goods will be the diference be-
tween the nominal return and infation, namely $1.00. If the rate of infation
is 10 percent and the nominal interest rate is 5 percent, the real rate is −5 per
cent; a loan at 5 percent still returns the borrower $105.00, but that $105.00
will buy goods that were worth only $95.00 at the beginning of the period,
5 percent less than the original $100.00 could buy.
As in the generic example of wheat and oil, we can interpret the equation
relating real and nominal rates of interest either as a defnition of the return
on bonds in terms of goods or as an equilibrium condition. As an equilibrium
condition, the equation says that the return on real investment (the lef-hand
side of the equation) must equal the return on bonds (the right-hand side).
How should a changing price level modify our analysis of the relationship
between the real and the fnancial sides of the economy? Te answer to this
question depends on whether the real rate determines the nominal rate or
vice versa and thus goes to a fundamental diference between Keynes and the
mainstream.
In the mainstream story based on the classical dichotomy, the real interest
rate is determined in the real economy; it is the price that equilibrates the real
supply of saving and the real demand for investment. In the equation REAL =
NOMINAL − , causality runs from REAL and to NOMINAL, which is to say that
the real rate is the master, the nominal rate the slave. So, for the mainstream,
the Fisher efect causes no revision of the theory of macroeconomic adjust-
ment; there is no need to take account of price changes except in describing
the nominal side of the classical dichotomy, namely, how the real interest rate
is refected in fnancial markets.13
Tis story breaks down completely when prices are falling: negative nomi-
nal rates may be required for the real rate to be the master. If, for example,
REAL = 2 percent and = −5 percent, then the nominal rate on bonds would
have to be −3 percent. A zero-coupon bond14 with a redemption value of
Copyright © 2021. Harvard University Press. All rights reserved.
$100.00 next year would have to sell at a premium today, at $103.00, the de-
cline in value over the year refecting the negative nominal interest on the
bond.15
In stark contrast with the mainstream, the starting point for Keynes is the
nominal rate of interest, determined in markets for fnancial assets: causality
runs from NOMINAL and to REAL. Te framework of Te General Teory thus
requires a story about how the real rate of interest—not the nominal rate—is
afected by changes in prices.
Tere are two ways of telling this story. Te frst allows us to continue to
work with nominal interest rates and refect price changes in terms of invest-
ment returns and investment demand. Consider an investment project which,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
248 Raising Keynes
for a sacrifce of $10.00 worth of goods today—say oil—gives back (net of
operating costs) $1.50 worth of goods per year for the next ten years, 1.5 bar-
rels at $1.00 per barrel. Paying back the cost of the project would require a
level payment of $1.30 for principal and interest over the ten-year period to
amortize the loan. Net of loan amortization, the project would return 0.2 bar-
rels of oil per year, worth $0.20 per year, as in Table 7.6.
But if prices are changing over time, the numbers tell a diferent story. Sup-
pose the price of oil is falling by 10 percent per year. Ten the money value of
the oil and the resulting p roft would follow the trajectory in Table 7.7. Tis
table suggests a simple approach to incorporating price level changes in the
analysis: adjust the investment-demand schedule to refect infation or defa-
tion in calculating project returns. In the present example, an attractive proj
ect under a regime of constant prices becomes a white elephant under a defa-
tionary regime.
Instead of adjusting the investment-demand schedule, we can adjust the
interest rate. We do all the calculations in real terms, but with this important
diference: instead of assuming the real rate is unafected by price changes, we
adjust the hurdle rate used to determine which investment projects pass mus-
ter and which do not. Te hurdle rate is now, as in the mainstream analysis, a
real rate of interest, but in the framework of Te General Teory this rate is
determined jointly by fnancial markets and the rate of infation or defation:
the equation REAL = NOMINAL − is read from right to lef, with the real (oil)
rate of interest the diference between a nominal hurdle rate and the rate of
change of the price of oil.
Assume, as above, a nominal hurdle rate of 5 percent. If the price of oil falls
by 10 percent each year, the hurdle rate calculated from the equation REAL =
NOMINAL − is 15 percent. At this rate of interest, it takes about 2 barrels of oil
to amortize the original 10 barrels of cost over ten years, a loss of 0.5 barrels
per year given that production is 1.5 barrels.
Tere are two important points that emerge from Table 7.7. Te frst is the
Copyright © 2021. Harvard University Press. All rights reserved.
role of expectations. If, as in the oil example, the expected project life is ten
years, then the relevant interest rate is the rate expected over the next ten
years. Since investment by defnition deals with costs incurred today for
benefts in the future, the Fisher efect puts expectations front and center.16
Te second point is the implications of the Fisher efect for liquidity prefer-
ence. Absent the Fisher efect, we have thus far followed Keynes in dividing
fnancial assets into two kinds: money (cash or cash equivalents such as de-
mand deposits), for which the rate of interest is zero, and long term bonds, for
which the interest rate is . Once prices are assumed to change, money as well
as bonds can deliver nonzero returns. If, for example, the price level is falling
by 10 percent per year, money yields a 10 percent real return even though its
nominal return is zero.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 7.6 Proft and Loan Amortization at 5 percent Interest
Year
1 2 3 4 5 6 7 8 9 10
Nominal Project Returns $1.50 $1.50 $1.50 $1.50 $1.50 $1.50 $1.50 $1.50 $1.50 $1.50
P&I −$1.30 −$1.30 −$1.30 −$1.30 −$1.30 −$1.30 −$1.30 −$1.30 −$1.30 −$1.30
7
Net Cash Flow $0.20 $0.20 $0.20 $0.20 $0.20 $0.20 $0.20 $0.20 $0.20 $0.20
•
Addendum: Allocation of Amortization between Interest and Principal; Principal Balance at Year End
Interest $0.50 $0.46 $0.42 $0.37 $0.33 $0.28 $0.23 $0.18 $0.12 $0.06
Principal Reduction $0.80 $0.83 $0.88 $0.92 $0.97 $1.01 $1.07 $1.12 $1.17 $1.23
Principal Balance (End of Year) $9.20 $8.37 $7.49 $6.57 $5.61 $4.59 $3.53 $2.41 $1.23 $0.00
Dynamics vs. Statics
Table 7.7 Proft and Loan Amortization at 5 percent Interest and 10 percent Defation
Year
1 2 3 4 5 6 7 8 9 10
Nominal Project Returns $1.35 $1.22 $1.09 $0.98 $0.89 $0.80 $0.72 $0.65 $0.58 $0.52
P&I −$1.30 −$1.30 −$1.30 −$1.30 −$1.30 −$1.30 −$1.30 −$1.30 −$1.30 −$1.30
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Net Cash Flow $0.05 −$0.08 −$0.20 −$0.31 −$0.41 −$0.50 −$0.58 −$0.65 −$0.71 −$0.77
249
250 Raising Keynes
Paradoxically, the variable governing asset demands remains the nominal
rate of interest on bonds even though what matters to the agent is the difer-
ence between the real return on bonds and the real return on money. Te
paradox disappears once it is recognized that price-level changes afect bonds
and money in exactly the same way because both money and bonds are de-
nominated in nominal terms. Consequently, the diference between real rates
is the same as the diference between nominal rates. Since the nominal rate on
money is zero, the diference between real rates on bonds and money reduces
to the nominal rate on bonds.
Algebraically we have the general formula for the fraction of wealth held in
liquid form as a function of the diference between real rates, ( RB − RM ), in
place of the original formulation, ( ), where superscripts distinguish bonds
and money, and subscripts distinguish real and nominal rates. Substituting
from the general equation relating real and nominal rates gives
ρ RB = ρ NB − φ
and
ρ RM = ρ NM − φ = − φ.
So ( RB − RM ) reduces to ( NB ), which is the original formula for the de-
mand for money as an asset.
As the oil example illustrates, the Fisher efect drives a wedge between the
interest rate relevant for investment decisions and the rate relevant for fnan
cial market equilibrium. Te frst is the real rate, the second the nominal rate.
Tat is, if returns are measured in real terms, investment demand is driven by
the real rate,
I D = I ( ρ RB ) = I ( ρ NB − φ ),
while equality between the asset demand for money and its available supply
Copyright © 2021. Harvard University Press. All rights reserved.
depends on the nominal rate,
β ( ρ NB )
M2 = M2 = PB ( ρ)B .
1 − β ( ρ NB )
Te wedge between the two rates is the rate of change of the price level.
In terms of the basic four-quadrant representation of aggregate demand,
the Fisher efect is refected in a linear displacement of the vertical axis sepa-
rating quadrants two and three. In Figure 7.7 the axis for quadrant three mea
sures the nominal interest rate and for quadrant two the real interest rate. If
the nominal rate is equal to 0 and prices are falling at the rate , the corre-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
ID ID , SD
400
300
II I
7
200
•
SD
Marginal efficiency of capital
100 ID
Real interest rate ( ) 1 Income (Y)
Nominal interest rate ( )
0
100 200 300 400 500 Income (Y)
100 100
200 200
Dynamics vs. Statics
Liquidity preference
300 300
400 400 Transactions demand
500 500
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
III M2 M2 M1 M 1 IV
Figure 7.7 Aggregate demand determined by the real interest rate.
251
252 Raising Keynes
Y 0
P 0
Aggregate demand Supply of goods
P
W E
5
P 0
W
4 Stationary real price
W 0
2
Supply of labor
1
100 200 300 400 500 600 Y
Figure 7.8 Fixprice adjustment: Fisher efect reduces aggregate demand.
sponding real rate is 1. (We assume given endowments B and M 2, so that the
liquidity-preference schedule is stationary despite the assumption of a falling
P.) In Figure 7.7 defation displaces the investment-demand schedule down-
ward from the solid line to the dashed line, and aggregate demand, instead of
being 400, where a hurdle rate of 0 would put it, is equal to 266.7.
Te consequence of a real interest rate exceeding the nominal rate is that
the economy is once again chasing a moving target of aggregate demand. Te
new complication is that the falling price level will have a knock-on negative
impact on aggregate demand because, in the framework of Te General Te-
ory, a falling price level, < 0, increases the real interest rate. Figure 7.8 illus-
trates the process assuming the nominal interest rate is fxed. Te AD sched-
ule moves to the lef as output and prices fall. It is not necessarily the case that
Copyright © 2021. Harvard University Press. All rights reserved.
the process converges to a new equilibrium, but, as the mathematical appen-
dix shows, the Fisher efect delays but does not derail the adjustment pro
cess—provided the propensity to save is sufficiently large and provided wages
adjust more rapidly than prices.
By Way of Summary
Moving from comparative statics to real-time changes reverses the verdict on
whether a capitalist economy can be self-regulating. In chapter 4 the verdict
was against Keynes. Whatever the particular assumption we a dopted about
money and banks, the arguments of Modigliani and Haberler carried the day.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Dynamics vs. Statics 253
One way or another, wage and price fexibility assured the existence of a full-
employment equilibrium.
But in a world of real-time changes, such as Keynes invites us to consider
in chapter 19 of Te General Teory, we get very diferent results. Even in a
world of commodity money, there is no guarantee that Modigliani’s full-
employment equilibrium is stable: in contrast with the Eagles’ “Hotel Califor-
nia,” once you depart from full employment, you may be barred from ever
returning.
If we accept that the framework of modern capitalism includes a fractional-
reserve banking system, then the question of whether capitalism can be self-
regulating becomes more complex. Te key issue is whether banks remain
fully loaned up when prices are falling. If banks do remain fully loaned up,
the money supply is always a constant multiple of the reserve base,17 and if
the reserve ratio is fxed exogenously, so is the money supply. A world of
fractional-reserve banking is then no diferent from a world of commodity
money because banks simply multiply an exogenously given quantity of out-
side money.
If, however, the money supply is endogenous because the creation of money
by banks varies directly with the price level, and there is no shifing between
transactions money and asset money, then there is no Keynes efect. More-
over, unlike the case of comparative statics, there is no change in the quantity
of bonds to accomplish the same result as the Keynes efect does, namely, to
increase aggregate demand as the price level falls. Consequentially, there is
nothing in the interest-investment nexus to propel the economy back to full
employment afer it has suffered a negative demand shock.
Te killer argument against Keynes was supposed by Milton Friedman to
be the real-balance efect because this argument, unlike Modigliani’s, operates
on the propensity to consume and is therefore not subject to the qualifcations
Modigliani was compelled to make for a liquidity trap and inelastic invest-
ment demand. Unlike the comparative-statics context, where one can fnesse
Copyright © 2021. Harvard University Press. All rights reserved.
frms and banks and focus on the wealth of households, this is no longer pos-
sible when change takes place in real time. In real time, frm liabilities are
fxed in nominal terms, while the value of frm assets varies with the price
level. Price-level reductions, which are supposed in the comparative-statics
context to drive household wealth and consumption demand ever upward,
instead drive frms into bankruptcy and undermine the wealth of their credi-
tors. Long before the bonanza of infnite wealth and unlimited aggregate de-
mand materializes, the economy dissolves into catastrophe. Far from solving
the problem of aggregate demand, the real-balance efect exacerbates it!
Finally, the Fisher efect piles on the bad news. A continuously falling price
level increases real rates of interest and thus reduces investment demand. Tis
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
254 Raising Keynes
reduces aggregate demand further and shifs the equilibrium to a lower level
of output and employment.
All these efects, with the exception of the Keynes efect, have to do with
the impact of real-time changes on the values of assets and liabilities. Consid-
ering how important the issue was to his purpose in writing Te General
Teory, it is curious how little attention Keynes pays to the interaction of the
price level, debt, and aggregate demand. He appears content to argue that
price fexibility is more of a problem than a solution. Tis argument is correct
as far as it goes, but incomplete at best. In any case, there is no systematic
treatment of the kind accorded to consumption, investment, and interest, and
the anecdotal comments ofered in his chapter 19 can hardly serve as a substi-
tute.
In fact, as was noted in connection with the real-balance efect, it is not to
Keynes but to Irving Fisher (1933) that one must turn for connecting aggre-
gate demand to the nexus of prices, assets, and liabilities—except that Fisher
had no place for aggregate demand in his conceptual framework, unquestion-
ably a serious shortcoming; the solution, I have suggested, is to marry Fisher
to Keynes.
In the context of real-time changes, none of the dynamic considerations
discussed in this chapter is totally unambiguous, but, except for the Keynes
efect, each is likely to depress aggregate demand as the price level falls. So the
Keynes efect would not only have to survive the move from comparative stat-
ics to real time it would also have to work so well that it canceled out the oth-
ers! It would take a tremendous act of faith (credere quia absurdum) to argue
that wage and price fexibility vitiate the message of Te General Teory: with-
out this faith, it is not plausible that a market system lef to itself will eliminate
involuntary unemployment.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Te frst issue is the dynamic behavior of the Modigliani second-pass model.
Te focus is on the fxprice version of the model since this is the adjustment
process that yields ambiguous results. (Te mathematics is similar for the
fexprice model.) We have the following system of equations:1
•
P
W
≡ P − W
P P
P = −1 GS − Y − 3 Y − LS , (7.1)
P W W W
W
Y = θ 2[I ( ρ) − sY ], (7.2)
P P
= −1 GS − Y , (7.3)
P W
β( ρ)
M ≡ M1 + M 2 = αPY + PB ( ρ)P0 K = M . (7.4)
1 − β( ρ)
For the sake of mathematical tractability I will assume that the adjustment
of the interest rate to the price level in equation (7.4) is instantaneous. Equa-
tion (7.4) difers from its counterpart in chapter 4 because in a world of real-
time change the quantity of bonds B is not assumed to change with the price
level. Instead it is fxed once and for all at B = P0K, the value of the capital
stock at the initial price level P0. Diferentiating equation (7.4) with respect to
Copyright © 2021. Harvard University Press. All rights reserved.
the price level gives
∂ρ − αY
ρ′ ≡ = > 0. (7.5)
∂P β′PB βPB′
+ P K
(1 − β)
2
1− β 0
Here the impact of assuming B to be fxed is negligible because the fungibility
of M1 and M2 means that varies directly with the price of goods.
We have three state variables, (P/W), Y, and P, and three diferential equa-
tions, equations (7.1), (7.2), and (7.3). Te conditions for a stable equilibrium
255
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
256 Raising Keynes
are correspondingly more complicated than in the two-equation systems that
we have studied up to now. In general, if the Jacobian matrix is
j11 j12 j13
J = j21 j22 j23 ,
j j32 j33
31
the stability conditions are
1 = − tr J = −(j11 + j22 + j33) > 0,
2 = −(j13 j31 + j12 j21 + j23 j32 − j11 j22 − j11 j33 − j22 j33) > 0,
3 = −det J > 0,
12 − 3 > 0.
Diferentiating equations (7.1), (7.2), and (7.3) with respect to the state
variables gives
−θ1GS′ + θ 3 LS′ θ1 − θ 3 0
J = 0 −θ 2 s θ 2 I ′ ρ′ ,
−θ1GS′ θ1 0
so
α1 = θ1GS′ − θ 3 LS′ + θ 2 s,
α2 = −θ1θ 2 I ′ ρ′ + ( θ1GS′ − θ 3 LS′ ) θ 2 s,
and afer simplifcation
α3 = −θ1θ 2 θ 3 I ′ ρ′(GS′ − LS′),
Copyright © 2021. Harvard University Press. All rights reserved.
α1 α2 − α3 = ( θ1GS′ − θ 3 LS′ + θ 2 s ) ( θ1GS′ − θ 3 LS′ ) θ 2 s − θ1θ 2 I ′ ρ′ θ 2 s + GS′ ( θ1 − θ 3 ) .
Te signs of the various terms guarantee that the expressions 1, 2, and 3
are positive regardless of the magnitudes of the variables and parameters. Te
last expression, 12 − 3, is more problematic. If 1 > 3, then this expression
also is positive and the system is stable, converging to the full-employment
equilibrium E in Figure 7.1. But if 1 < 3, and if 1GS′ is not too large and
3LS′ is sufficiently small, then 12 − 3 will be negative, and the system will
be unstable, as in Figure 7.2. To see this, suppose 0 < 1GS′ ≤ , LS′ ≈ 0. Ten
( α1α2 − α3 ) ≤ θ1θ 2 ( GS′s [ ε + θ 2 s ] − I ′ ρ′ θ 2 s − GS′ ( θ 3 − θ1 ) ),
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Mathematical Appendix 257
which is negative if 3 − 1 is suf
ficiently large. Tese are the conditions stated
verbally in the text (pp. 223–226).
Fractional-Reserve Banking
Real time makes an important diference for the impact of fractional-reserve
banking on the ability of a capitalist economy to recover from a shock to out-
put. With the quantity of bonds fxed once and for all, the price level has no
impact on the rate of interest. In place of equation (7.4), we have
M1 = PY, (7.4a)
β( ρ)
M2 = PB ( ρ)P0 K = M 2 , (7.4b)
1 − β( ρ)
with M1 determined by P and Y, and M 2 determining . Te LM schedule be
comes horizontal, as in chapter 4. Te novelty is that the LM schedule is now
invariant with respect to the price of goods because the bond issue is no
longer assumed to vary with the nominal value of capital. Assuming that the
LM schedule is characterized by equation (7.4b) breaks the link between the
interest rate and goods prices. Tere is now neither a bond efect nor a Keynes
efect. Aggregate demand is still characterized by I( ) = sY, but in place of the
inequality in (7.5) we have ′ ≡ ∂/∂P = 0. Te AD schedule in Y × P/W space
is once again vertical.
The Real-Balance Efect
Te starting point for the real-balance efect is the argument, perfectly sensi-
ble in itself, that consumption and saving depend on wealth as well as on in-
come. In place of the consumption function C = cY, we have in its simplest
form C = cY + aA, where A here represents real wealth and a is the marginal
Copyright © 2021. Harvard University Press. All rights reserved.
propensity to consume from wealth. Since S = Y − C, in place of S = sY, where
s = 1 − c, we have S = sY − aA. At lower and lower price levels, the cash com-
ponent of wealth is worth more and more in real terms. Tis inverse relation-
ship between A and P was the end of the real-balance story In chapter 4. Even
if a liquidity trap drives ′ to zero, or a lack of investment opportunities drives
I′ to zero, the AD schedule slopes downward so long as a is positive.
In chapter 4, bank lending and debt did not matter. In a comparative-statics
framework, money creation and debt are always in balance, and nothing ever
changes. It did not make any diference whether real balances included trans-
actions money (as when households were assumed to lend directly to frms),
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
258 Raising Keynes
or did not include transactions money (as when banks were assumed to f
nance trade by creating transactions money).
Here too bank lending is not central to the argument. Tere are two essen-
tial assumptions for the stunning reversal of the real-balance efect. Te frst is
that the propensities to consume of debtors and creditors difer. Te second is
that as the real burden of debt increases, there comes a point where bank-
ruptcy becomes the best of a bad lot of alternatives, if not the only way out of
debt. Te frst idea can be refected by simple modifcations to the basic third-
pass model.
To refect diferent propensities to consume, assume two distinct kinds of
agents: Type 1 are wealth holders, and Type 2 are business people whose
physical capital is exactly equal to their debts when the story opens and
P = P0. Write desired expenditure as
C = C1 + C2,
P ( )P0 K M 2
C1 = c1Y1 + a1 A1 = c1Y1 + a1 B + ,
P P
P ( )P0 K
C2 = c2Y2 + a2 A2 = c2Y2 + a2 K − B .
P
To simplify the calculations, I assume that debtors (Type 2) as well as credi-
tors (Type 1) mark their bonds to market. Desired saving is
S = S1 + S2 = s1Y1 − a1 A1 + s2Y2 − a2 A2
P ( )P0 K M 2 PB ( )P0 K
= s1Y1 − a1 B + + s2Y2 − a2 K − .
P P P
Suppose Y1 = Y, and Y2 = (1 − )Y. Ten aggregate demand is defned by
equality between desired expenditure and income:
Copyright © 2021. Harvard University Press. All rights reserved.
P ( ρ)P0 K M 2
I ( ρ) = s1 γY − a1 B +
P P
(7.6)
P ( ρ)P0 K
+ s2 (1 − γ)Y − a2 K − B .
P
In the absence of a Keynes efect, that is, with
β( ρ)
M2 = PB ( ρ)P0 K = M 2
1 − β( ρ)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
7 • Mathematical Appendix 259
and
′ = 0,
the efect of changes in the price level on aggregate demand is
PB ( ρ)P0 K M
− ( a1 − a2 ) − a1 22
∂Y P 2
P .
=
∂P s1 γ + s2 (1 − γ)
If a1 = a2, then a fall in the price level increases aggregate demand. But if
a2 > a1 then the negative impact on the balance sheets of business can easily
outweigh the positive efect on wealth portfolios.
Te efects of bankruptcy are harder to model. If there were no fallout be-
yond the direct efects on the fnancial assets and liabilities of the debtor and
creditor, bankruptcy might indeed have provided the salutary cleansing that
Depression-era economists and politicians alike claimed for it.2 Te balance
sheet of the creditor takes a hit, but this is ofset by the improvement of the
debtor’s balance sheet. If it is true that the debtor’s propensity to consume is
higher than the creditor’s, then bankruptcy would thus have a salutary efect
on aggregate demand. Writing the market value of bonds as
PB B = PB ( )P0 K ,
we have from equation (7.6)
∂Y
=
( a1 − a2 ) .
∂ ( PB B ) P [ s1 + s2 (1 − ) ]
As the value of fnancial assets falls, aggregate demand increases.
Te problem is of course the fallout beyond the contracting parties. As
noted in the text, the contagion to suppliers, customers, employees, banks
Copyright © 2021. Harvard University Press. All rights reserved.
quickly erodes whatever relief bankruptcy provides.
The Fisher Efect
Again we focus on fxprice adjustment. Te complete system is
•
P
W
≡ P − W
P P
= −1 GS − Y − 3 Y − LS , (7.1)
P
P W W W
W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
260 Raising Keynes
P P
Y = θ 2 I ρ − − sY = θ 2 I ρ + θ1 GS − Y − sY , (7.7)
P W
is the real interest rate. Te Jacobian becomes
where − P/P
−1GS′ + 3 LS′ 1 − 3
J = .
1 2 I ′GS′ − 2 ( I ′1 + s )
Te trace and determinant conditions necessary and suf ficient for a stable
equilibrium are satisfed if s > − I′1 and 3 > 1. Te frst inequality is suf fi
cient (but not necessary) for tr J < 0. Te two inequalities together are suf fi
cient (but again not necessary) for det J > 0. To see how the two inequalities
together ensure that the determinant condition is satisfed, calculate det J and
simplify the expression to obtain
det J = 123I′(GS′ − LS′) + 2s(1GS′ − 3LS′).
If s > − I′1, then
det J > 123I′(GS′ − LS′) − 2I′1(1GS′ − 3LS′) = 12I′GS′(3 − 1),
so det J > 0 if 3 > 1.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 8 .
a dose of reality
The Evidence of the Great Depression
Values have shrunken to fantastic levels; taxes have risen; our ability
to pay has fallen; government of all kinds is faced by serious curtail-
ment of income; the means of exchange are frozen in the currents of
trade; the withered leaves of industrial enterprise lie on every side;
farmers fnd no markets for their produce; the savings of many years
in thousands of families are gone.
More important, a host of unemployed citizens face the grim
problem of existence, and an equally great number toil with little
return. Only a foolish optimist can deny the dark realities of the
moment.
—franklin roosevelt, first inaugural address,
march 4, 1933
Many persons lef their jobs for the more proftable one of selling
apples.
—herbert hoover
Roosevelt puts some fesh on the bare bones of our models. Table 8.1 shows
the changes in real U.S. gross domestic product by sectors from 1929 to 1933.
One thing stands out from the economy-wide data: the massive decline in real
Copyright © 2021. Harvard University Press. All rights reserved.
GDP, led, as Table 8.2 shows, by private investment. By the time the econ-
omy reached bottom, output had fallen by almost one-fourth, and private
investment by four-ffhs. Investment in 1932 and 1933 was a shadow of its
1929 self.
Te employment numbers in Table 8.3 bear out the GDP data. Te major
exception was the farm sector, where real output was higher in 1933 than in
1929 and employment was only marginally lower. Indeed, other data, pre-
sented in Table 8.4, suggest that farm employment in 1932 was higher than in
1929.
Whether or not total agricultural employment held its own, as Table 8.4
indicates, or decreased by 5 percent, as in Table 8.3, is not terribly important
compared with the striking diference between the performance of agricul-
261
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
262
Table 8.1(a) Real GDP by Sector ($ billions at 1929 prices)
Gross Domestic Nonfarm Households Nonproft General State and
Product Business Business Farm and Institutions Households Institutions Government Federal Local
1929 103.60 89.40 80.50 8.90 9.00 7.40 1.50 5.20 1.10 4.10
1930 94.68 80.40 72.07 8.36 8.81 7.17 1.54 5.44 1.18 4.27
1931 88.54 74.42 64.95 10.12 8.51 6.85 1.56 5.54 1.13 4.41
1932 76.97 63.08 54.33 9.87 8.18 6.52 1.56 5.47 1.08 4.39
1933 75.98 61.76 53.32 9.40 8.10 6.47 1.53 5.77 1.33 4.42
Source: Bureau of Economic Analysis, National Income and Product Accounts.
Raising Keynes
Table 8.1(b) Real GDP by Sector (1929 = 100)
Gross Domestic Nonfarm Households Nonproft General State and
Product Business Business Farm and Institutions Households Institutions Government Federal Local
1929 100 100 100 100 100 100 100 100 100 100
1930 91 90 90 94 98 97 103 105 107 104
1931 85 83 81 114 95 93 104 107 103 108
1932 74 71 67 111 91 88 104 105 98 107
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1933 73 69 66 106 90 87 102 111 121 108
Source: Bureau of Economic Analysis, National Income and Product Accounts.
8 • A Dose of Reality 263
Table 8.2 Real Gross Domestic Product (billions of chained dollars of 1937, scaled to
1929 price level)
1929 1930 1931 1932 1933
Gross Domestic Product 104.6 95.7 89.5 78.0 77.1
Personal Consumption Expenditures 77.4 73.2 70.9 64.5 63.2
Goods 43.8 40.4 38.9 34.4 34.1
Durable Goods 9.8 8.2 7.0 5.4 5.3
Nondurable Goods 33.9 32.1 31.7 28.9 28.8
Services 33.6 33.0 32.1 30.2 29.0
Gross Private Domestic Investment 17.2 11.7 7.6 2.8 3.8
Fixed Investment 15.6 12.2 8.6 5.3 4.9
Nonresidential 11.6 9.7 6.5 4.1 3.8
Structures 5.5 4.7 3.0 1.9 1.4
Equipment 5.5 4.4 2.9 1.8 1.8
Intellectual Property Products 0.6 0.6 0.6 0.6 0.6
Residential 4.1 2.5 2.1 1.2 1.0
Change in Private Inventories 1.5 −0.6 −1.5 −3.4 −1.8
Net Exports of Goods and Services 0.4 0.2 0.0 0.0 −0.1
Exports 5.9 4.9 4.1 3.2 3.2
Goods 5.3 4.3 3.6 2.8 2.8
Services 0.6 0.5 0.5 0.4 0.4
Imports 5.6 4.8 4.2 3.6 3.7
Goods 4.5 3.9 3.3 2.8 3.0
Services 1.1 1.1 0.9 0.9 0.7
Government Consumption Expenditures 9.6 10.6 11.0 10.7 10.3
and Gross Investment
Federal 1.9 2.1 2.2 2.3 2.8
National defense 1.0 1.1 1.1 1.1 1.0
Nondefense 0.8 0.9 1.0 1.1 1.6
State and Local 7.7 8.4 8.8 8.4 7.6
GPDI as Percentage of GDP 16.44 12.21 8.51 3.53 4.94
Source: Bureau of Economic Analysis, National Income and Product Accounts.
Copyright © 2021. Harvard University Press. All rights reserved.
ture and nonagricultural business. A clue to this diference is provided by
breaking down farm work between labor supplied by the farm household and
that supplied by hired workers. It is clear that the reduction in total farm em-
ployment was limited to nonfamily workers; there were 5.5 percent more
family workers on American farms in 1933 than in 1929.
Nominal output fell more sharply than real output because prices were de-
clining as dramatically as was output. Tables 8.5 and 8.6 show nominal output
and implicit price levels over the four years. In the nonfarm business sector,
prices initially held their own. Even as nonfarm business output fell by more
than 10 percent between 1929 and 1930, prices declined by barely 2 percent.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 8.3 Nonfarm and Farm Employment and Unemployment Rates
264
Federal Unemployed, Not Counting
Civilian Emergency Federal Emergency Workers as
Labor Force Employed Unemployed Workers Employed
Percentage of Percentage of
Percentage of Civilian Private Percentage of Civilian Private
Civilian Labor Nonfarm Labor Civilian Labor Nonfarm Labor
Total Farm Government Number Force Force Number Force Force
Table Ba470 Table Ba471 Table Ba472 Table Ba473 Table Ba474 Table Ba475 Table Ba476 Table Ba477
(thousands) (thousands) (thousands) (thousands) (thousands) (percent) (percent) (thousands) (percent) (percent)
1929 47,757 46,374 10,541 3,065 1,383 2.90 4.05 0 2.90 4.05
1930 48,523 44,183 10,340 3,148 4,340 8.94 12.39 20 8.99 12.44
1931 49,325 41,604 10,240 3,563 7,721 15.65 21.74 299 16.26 22.58
1932 50,098 38,630 10,120 3,817 11,468 22.89 31.71 592 24.07 33.35
1933 50,882 40,247 10,092 5,361 10,635 20.90 30.02 2195 25.22 36.21
Source: Historical Statistics of the United States, Millennial Edition Online.
Raising Keynes
Table 8.4 Farm Labor: Family Workers and Hired Workers (thousands)
Total Farm Employment Family Workers Hired Workers
Table K 174 Table K 175 Table K176
1929 12,763 9,360 3,403
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1930 12,497 9,307 3,190
1931 12,745 9,642 3,103
1932 12,816 9,922 2,894
1933 12,739 9,874 2,865
Source: Bureau of the Census 1975.
Copyright © 2021. Harvard University Press. All rights reserved.
Table 8.5 Nominal GDP by Sector ($ billions)
Gross Domestic Nonfarm Households Nonproft General State and
Product Business Business Farm and Institutions Households Institutions Government Federal Local
1929 103.60 89.40 80.50 8.90 9.00 7.40 1.50 5.20 1.10 4.10
1930 91.20 77.30 70.40 7.00 8.50 6.90 1.60 5.40 1.10 4.20
1931 76.50 63.40 57.70 5.70 7.60 6.10 1.50 5.50 1.10 4.30
1932 58.70 47.00 43.00 3.90 6.60 5.20 1.40 5.20 1.10 4.10
8
1933 56.40 45.10 41.00 4.10 5.80 4.50 1.30 5.50 1.40 4.10
•
Source: Bureau of Economic Analysis, National Income and Product Accounts.
A Dose of Reality
Table 8.6 Sectoral Price Levels, Implicit Defators (1929 = 100)
Gross Domestic Nonfarm Households Nonproft General State and
Product Business Business Farm and Institutions Households Institutions Government Federal Local
1929 100 100 100 100 100 100 100 100 100 100
1930 96 96 98 84 96 96 104 99 93 98
1931 86 85 89 56 89 89 96 99 97 97
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1932 76 75 79 40 81 80 90 95 102 93
1933 74 73 77 44 72 70 85 95 105 93
265
Sources: Tables 8.2 and 8.5, above.
266 Raising Keynes
Only when output continued its freefall did prices begin to fall at comparable
rates.
Clearly there was something diferent about agriculture: prices fell dra-
matically from the beginning of the Depression, and continued to fall cata-
strophically until bottoming out in 1932. Table 8.7 summarizes the diferent
behavior of the farm and nonfarm economies in terms of sectoral Fisher ef-
fects—rates of change of prices—for both nonfarm businesses and the farm
sector. As Table 8.7 shows, afer 1930, when price cutting became the norm,
the Fisher efect became more signifcant in the nonfarm-business sector,
even though defation never became as calamitous as it was for farmers. Be-
tween 1930 and 1932 nonfarm businesses had some 10 percent per year
tacked onto nominal interest rates because they were repaying in appreciated
dollars. On the farm, debtors implicitly faced real interest rates as high as 33
percent per year even if they had been able to borrow at a zero nominal inter-
est rate!1 Te numbers in the example I gave in chapter 7 to fesh out the real-
balance efect—a fall in the price of wheat from $1.25 per bushel to $0.50—
were illustrative, not fanciful. In fact, wheat averaged $1.18 per bushel in the
decade 1920 to 1929 and plummeted to $0.38 in 1931. Wheat was hardly ex-
ceptional: the price of cotton averaged $0.21 per pound in the 1920s, falling
below $0.06 in 1931 (Historical Statistics of the United States, Millennial Edi-
tion Online, tables Da 719 and Da 757). At the end of this chapter, I return to
the question of what distinguished agriculture from the rest of the economy.
Wages followed the same downward trajectory as prices. In fact, as Table
8.8 shows, money wages in manufacturing—whether measured by annual
wages for full-time workers or by hourly wages—fell almost in lockstep with
prices. As a result, real wages (reported in the last two columns of Table 8.8)
remained virtually constant over the period 1929 to 1933, falling slightly for
full-time employees, rising slightly on an hourly basis for manufacturing
workers.
Te toll of the Fisher and real-balance efects show up in mortgage delin-
Copyright © 2021. Harvard University Press. All rights reserved.
quencies and business failures, shown in Tables 8.9 and 8.10. Te two tables
tell a similar story. Te number of nonfarm mortgage foreclosures almost
doubled between 1929 and 1932, and the number of corporate business fail-
Table 8.7 Percentage Change in the Price Level
Gross Domestic Product Nonfarm Business Farm
1930 −3.67 −2.32 −16.26
1931 −10.31 −9.06 −32.74
1932 −11.73 −10.91 −29.87
1933 −2.67 −2.85 10.42
Source: Table 8.6, above.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 8.8 Money and Real Wages
8
Avg. Real Annual
•
Avg. Annual Earnings, Avg. Hourly Earnings, Earnings, Avg. Real Hourly
Avg. Annual Earnings, Avg. Hourly Earnings, Full-Time Employees All Manufacturing Consumer Price Index Full-Time Earnings,
Full-Time Employees All Manufacturing (1929 = 100) (1929 = 100) (1929 =100) Employees All Manufacturing
Table D 722 Table D 802 Table D 727
1929 $1,405 $0.56 100.00 100.00 100.00 100.00 100.00
1930 $1,368 $0.55 97.37 98.21 97.37 100.00 100.87
1931 $1,275 $0.51 90.75 91.07 88.65 102.37 102.73
A Dose of Reality
1932 $1,120 $0.44 79.72 78.57 79.64 100.10 98.66
1933 $1,048 $0.44 74.59 78.57 75.37 98.97 104.25
Source: Bureau of the Census 1975.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
267
268 Raising Keynes
ures increased by almost 50 percent over the same period. Te liabilities of
these failed businesses totaled over 1.5 percent of GDP in 1932, double the
dollar amount of 1929, when prices were one-third higher. At the same time,
instead of the steady growth in the number of active businesses characteristic
of the 1920s, by 1933, voluntary attrition, as distinct from forced liquidation
due to bankruptcy, led to the closure of 10 percent of the number of corporate
businesses active in 1929. (I have not been able to fnd data on farm-mortgage
foreclosures or noncorporate business failures.)
Finally, what happened to banks, to the money supply, to interest rates? Te
most obvious sign of distress was the precipitous rise in the number of bank
Table 8.9 Nonfarm Mortgage Foreclosures
Per 1,000 Mortgaged
Number Structures
1926 68,100 3.6
1927 91,000 4.8
1928 116,000 6.1
1929 134,900 7.1
1930 150,000 7.9
1931 193,800 10.2
1932 248,700 13.1
1933 252,400 13.3
Source: Historical Statistics of the United States, Millennial
Edition Online.
Table 8.10 Business Failures
Business Failures
Active Firms
(thousands) Number Per 10,000 Firms
1920 1,821 8,881 49
Copyright © 2021. Harvard University Press. All rights reserved.
1921 1,927 19,652 102
1922 1,983 23,676 119
1923 1,996 18,718 94
1924 2,047 20,615 101
1925 2,113 21,214 100
1926 2,158 21,773 101
1927 2,184 23,146 106
1928 2,187 23,842 109
1929 2,203 22,909 104
1930 2,160 26,355 122
1931 2,127 28,285 133
1932 2,066 31,822 154
1933 1,986 19,859 100
Source: Historical Statistics of the United States, Millennial Edition Online.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 269
failures. Again, outright failure is only part of the story: there were 40 percent
fewer banks in 1933 than in 1929, weaker banks closing “voluntarily” or
merging with more solid banks (Board of Governors of the Federal Reserve
System 1943, table 1). Table 8.11 gives annual data for the relevant period,
data which has to be understood against a baseline of failures of approxi-
mately fve hundred per year over the decade following the end of World War
I. Tese bank failures culminated in the Bank Holiday that set Roosevelt’s
New Deal in motion. (Roosevelt closed the banks by executive order thirty-
six hours afer assuming of fi ce; those deemed to be solvent reopened a week
later.) In 1933 alone, the losses to depositors amounted to 1 percent of GDP.
Bank failures were both cause and efect of the decline in the money supply,
frst gradual and then precipitous, over the period leading up to the crisis of
1933. Table 8.12 shows the course of GDP and summary monetary data from
the end of the boom to the bottom of the Depression. Te narrow measure of
the money stock as currency plus demand deposits (M1, not to be confused
with Keynes’s M1) declined, but not as rapidly as nominal GDP, with the result
Table 8.11 Bank Failures
Deposits of Suspended Banks Losses to Depositors
Number of Suspensions ($ millions) ($ millions)
1929 659 231 77
1930 1,352 869 237
1931 2,294 1,691 391
1932 1,456 725 168
1933 4,004 3,601 540
Source: Board of Governors of the Federal Reserve System 1943, sec. 7, table 66.
Table 8.12 Nominal GDP, Money Supply, and Implicit Velocity, All Data for June of Each
Year except GDP ($ billions)
Copyright © 2021. Harvard University Press. All rights reserved.
Monetary Base =
Reserve Balances Currency +
Nominal GDP Demand Currency in with Federal Member-Bank
(calendar year) Deposits1 Circulation Reserve Banks Reserve Balances
Table 9 Table 9 Table 105
1929 103.6 22.5 3.6 2.3 6.0
1930 91.2 21.7 3.4 2.4 5.8
1931 76.5 19.8 3.7 2.4 6.0
1932 58.7 15.6 4.6 2.1 6.6
1933 56.4 14.4 4.8 2.2 7.1
Sources: Bureau of Economic Analysis, National Income and Product Accounts; Board of Governors
of the Federal Reserve System 1943.
1. Demand deposits are adjusted to eliminate double counting of checks in process of collection.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
270 Raising Keynes
that the income velocity of money, the ratio of GDP to M1, also declined,
from almost four in 1929 to just under three in 1933.
Te money supply shrunk for two reasons. Te frst is the shif from depos-
its to currency by people spooked by the increasing threat of losing their
money to bank failure as the Depression deepened. Tis lef the banks with
lower reserves and a correspondingly reduced capacity to lend. But the re-
duced capacity to lend was likely less important than the reduced willingness
to lend; excess reserves grew from 1931 on.
Tis brings us to the second reason, the decline in deposits associated with
the decline in bank lending. Te growth in idle reserves led to a sharp rise in
the ratio of total reserves to total deposits and a correspondingly sharp fall in
the money multiplier. Te decline in the money multiplier becomes clearer
from the more detailed data in Table 8.13 that exist for banks in the Federal
Reserve System. Tese member banks, which accounted for close to 75 per-
cent of total demand deposits at the outset of the Depression and almost 85
percent by the time the economy bottomed out, showed a steady rise in the
ratio of reserves to deposits. Between 1929 and 1931, demand deposits de-
clined slightly while total reserves increased. In June 1932 the picture was
very diferent: as against the previous June, reserves declined by $340 million,
even as required reserves were falling by $450 million. Excess reserves stood
at $234 million in June 1932 and grew by more than $100 million over
the next year. By June 1933 excess reserves were 20 percent of required re-
serves, enough to support some $3 billion of additional demand deposits at
an average required reserve ratio of approximately 12 percent, more or less
the amount by which demand deposits had declined since 1929.2
Te asset side of the banks’ balance sheets flls in the story. Tables 8.14 and
8.15 show both the decline and the change in the composition of bank assets,
Table 8.14 in nominal terms, Table 8.15 in percentages of 1929 levels. “CIAC
Loans” (commercial, industrial, agricultural, and consumer loans, excluding
Copyright © 2021. Harvard University Press. All rights reserved.
Table 8.13 Deposits and Reserves of Member Banks of Federal Reserve System ($ millions)
Reserves at Reserve Ratio
Adjusted Demand Time Federal Reserve Required Excess (total reserves/
Deposits Deposits Banks Vault Cash Reserves Reserves total deposits)
June Table 18 Table 18 Table 105 Table 18 Table 105 Table 105
1929 16,324 13,168 2,314 433 2,275 42 0.078
1930 16,043 13,476 2,392 484 2,338 54 0.081
1931 15,208 13,247 2,404 519 2,275 129 0.084
1932 12,433 10,560 2,062 478 1,827 234 0.090
1933 12,089 8,890 2,160 405 1,797 363 0.103
Source: Board of Governors of the Federal Reserve System 1943.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 8.14 Nominal Output of Business Sector ($ billions) and Member-Bank Assets ($ millions)
Nominal GDP,
Business Sector Total Loans Investments
Loans + Open Market Loans on Real Estate Loans to CIAC U.S. Government
As of June 30 Calendar Year Investments Total Paper Securities Loans Banks Loans1 Total Obligations Other
1929 89.40 35,711 25,658 447 9,759 3,164 670 11,618 10,052 4,155 5,898
1930 77.30 35,656 25,214 748 10,425 3,155 535 10,349 10,442 4,061 6,380
1931 63.40 33,923 21,816 885 8,334 3,218 457 8,922 12,106 5,343 6,763
1932 47.00 28,001 16,587 658 5,570 2,894 573 6,892 11,414 5,628 5,786
1933 45.10 24,786 12,858 594 4,704 2,372 330 4,857 11,928 6,887 5,041
8•
Sources: Bureau of Economic Analysis, National Income and Product Accounts, for data on output of business sector; Board of Governors of the Federal Reserve
System 1943, table 18, for Federal Reserve member-bank data.
1. CIAC loans are commercial, industrial, agricultural, and consumer loans not collateralized by securities.
Table 8.15 Indices of Nominal Output of Business Sector and Member-Bank Assets (1929 = 100)
A Dose of Reality
Nominal GDP,
Business Sector Total Loans Investments
Loans + Open Market Loans on Real Estate Loans to CIAC U.S. Government
As of June 30 Calendar Year Investments Total Paper Securities Loans Banks Loans1 Total Obligations Other
1929 100 100 100 100 100 100 100 100 100 100 100
1930 86 100 98 167 107 100 80 89 104 98 108
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
1931 71 95 85 198 85 102 68 77 120 129 115
1932 53 78 65 147 57 91 86 59 114 135 98
1933 50 69 50 133 48 75 49 42 119 166 85
271
Sources: Bureau of Economic Analysis, National Income and Product Accounts, for data on output of business sector; Board of Governors of the Federal Reserve
System 1943, table 18, for Federal Reserve member-bank data.
1. CIAC loans are commercial, industrial, agricultural, and consumer loans not collateralized by securities.
272 Raising Keynes
loans against securities as collateral), are a rough empirical counterpart to
Keynes’s theoretical M1. In 1929 CIAC loans constituted almost half of all
loans. By 1933 CIAC loans had fallen by almost 60 percent, even more than
the nominal value of output fell. Until 1932 the decline in the volume of CIAC
loans was less rapid (in percentage terms) than the fall in output, but the de-
cline in lending accelerated, so that between 1932 and 1933 CIAC loans fell by
30 percent while nominal output fell by “only” 5 percent; in other words,
CIAC loans fell six times more rapidly than output between those two years.
A plausible explanation is that banks tried to accommodate their customers
as long as they could but then gave up.
Te second largest category, “Loans on Securities,” represents all lending
against securities as collateral and thus includes both loans to stock-market
brokers and dealers and CIAC lending against securities. Te proportions
changed dramatically during the Depression: with the collapse of the stock
market, lending to brokers and dealers fell from 30 percent of all loans on se-
curities in June 1929 to 20 percent of a much-reduced total in June 1933.
As was observed in chapter 4, for the money supply to remain constant in
the face of a sharply contracting demand for transactions loans, banks would
have to replace these loans with other fnancial assets purchased on the open
market. Banks did expand their ownership of U.S. Treasury obligations, by a
large percentage; however, the absolute increase, some $2.7 billion, only re-
placed a modest 20 percent of the decline of $12.8 billion in total loans and
40 percent of the decline in CIAC loans.
Te breakdown of the U.S. public debt in Table 8.16, the breakdown of
bank holdings of Treasury obligations in Table 8.17, and the path of interest
rates together shed some light on why banks would prefer to maintain idle
excess reserves rather than purchase Treasuries. Te problem was not a short-
age of Treasury obligations. From just over 15 percent in 1929, the ratio of the
public debt to GDP rose to almost 40 percent by mid-1933. To be sure, this
increase was largely because of the fall in GDP, but the outstanding debt rose
Copyright © 2021. Harvard University Press. All rights reserved.
by $5.6 billion, by one-third of its level in 1929. And banks held substantial
portions of each class of debt, as Table 8.17 indicates.
But there were reasons for banks to avoid too much exposure to Treasury
debt of all maturities. At the long end of the spectrum, bonds would suffer
losses in capital values if and when interest rates recovered (as we shall see in
chapter 11, this is one reason why wealth holders prefer cash and short-term
Treasuries), especially in light of the fears of the fnancial community that
continuing Federal defcits and mounting debt signaled runaway infation
in the offing. At the short end, Treasury obligations provided smaller and
smaller nominal returns; by 1932 the return on short-term Treasuries had
fallen so low that it hardly paid to own these securities. Te middle ofered
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 273
both drawbacks, the possibility of a decline in capital values and very m
odest
nominal returns.
Indeed, despite Keynes’s of-quoted observation that the liquidity trap was
only a hypothetical possibility, Table 8.18 illustrates the phenomenon of a
positive lower bound on the hurdle rate of interest for investment—here
the yield on Baa-rated corporate bonds—when short rates are driven close
enough to zero to be essentially equivalent to cash.3
Table 8.16 Federal Debt and Its Composition ($ millions)
Long-Term Bonds Treasury Bills and
Total Debt (maturity greater Treasury Notes Certifcates of Indebtedness
As of June 30 Outstanding than 5 years) (maturity 1–5 years) (maturity up to 1 year)
1929 16,931 12,125 2,254 1,640
1930 16,185 12,111 1,626 1,420
1931 16,801 13,531 452 2,246
1932 19,487 14,250 1,261 3,341
1933 22,539 14,223 4,548 3,063
Source: Board of Governors of the Federal Reserve System 1943, table 146.
Table 8.17 Federal Reserve Member-Banks’ Holdings of U.S. Treasuries ($ millions)
Treasury Bills and
Bonds (maturity Treasury Notes Certifcates of Indebtedness
As of June 30 Total greater than 5 years) (maturity 1–5 years) (maturity up to 1 year)
1929 4,155 3,005 704 446
1930 4,061 3,340 463 259
1931 5,343 4,039 403 901
1932 5,628 4,163 503 962
1933 6,887 3,725 2,049 1,113
Source: Board of Governors of the Federal Reserve System 1943, table 20.
Copyright © 2021. Harvard University Press. All rights reserved.
Table 8.18 Yields on Bonds and Short-Term Securities
Short-Term U.S. Long-Term U.S.
Government Securities Government Bonds (due Baa-Rated
June–July Average (maturity 3–6 months) or callable afer 12 years) Corporate Bonds
Table 122 Table 122 Table 128
1929 4.68 3.67 5.32
1930 1.86 3.25 5.78
1931 0.48 3.14 7.22
1932 0.28 3.67 11.16
1933 0.13 3.21 6.85
Source: Board of Governors of the Federal Reserve System 1943.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
274 Raising Keynes
Tese are of course nominal returns. Real returns, taking account of price
level changes, were much higher on all classes of fxed-income securities, sur-
passing 20 percent on moderately risky Baa bonds in 1932. Nominal yields on
Baa-rated corporate bonds actually rose through mid-1932, and they only
began to return to anything like the pre-Depression normal afer the frst ten-
tative beginning of recovery in mid-1933.
What Do the Data Tell Us?
Te decline in nonfarm business output between 1929 and 1933 was the result
of a massive failure of aggregate demand. Te stock market crash of October
1929 is associated in the popular mind with the ensuing decline of output and
employment, but there were clearly other forces at work, and economists have
advanced various views as to the fundamental cause of the decline in demand.
Te fall in investment demand was most precipitous, but this does not auto-
matically make it the causa causans. Te real efects of the fall in demand
were compounded by the accompanying fall in nominal prices and wages.
As Tables 8.1 and 8.6 show, the fall in prices and wages followed the decline
in output, with nonfarm prices and wages holding fairly steady between 1929
and 1930 despite a signifcant fall in output. Afer 1930 the fall in prices and
wages matched the fall in output, and prices and wages continued to fall into
1933 even as output was stabilizing. Output leading prices and wages down
suggests that the nonfarm economy followed a fxprice-adjustment model.
Te fxprice story is made more plausible by the fact that inventories fell dur-
ing the Depression, the opposite of what would have happened in a fexprice
regime. In the fxprice model, a fall in aggregate demand leads to cutbacks in
production, and the ensuing unemployment drives wages down; prices fall as
business seeks to take advantage of lower costs. As we shall see below, these
internal dynamics of the nonfarm business sector were compounded by the
interplay between nonfarm business and the farm sector.
Copyright © 2021. Harvard University Press. All rights reserved.
Nonfarm business thus followed the trajectory indicated in Figure 8.1, from
a full-employment equilibrium at E to the unemployment equilibrium at F.
It is not clear whether the real economy of output and employment had
reached this equilibrium when Roosevelt took of fi ce. Fortunately, the ques-
tion was never put to the test; New Deal policy initiatives, particularly the
break with gold early on, changed the dynamic of price and wage adjustment,
and the economy began to recover from the Depression.
Te monetary side of the Depression is explained by the joint efects of the
fall in prices and output. Transactions demand, Keynes’s M1, declined with
the nominal value of output, PY, and the money supply contracted in re-
sponse. Te fall in prices reduced the value of collateral and pushed many
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 275
Y 0 Y 0
P
W P 0
5 Supply of goods
P 0
W E
4
Stationary real price F
3
2 W 0
Supply of labor
1
Aggregate demand
100 200 300 400 500 600 Y
Figure 8.1 From full employment to a depression equilibrium.
businesses to the point of bankruptcy—and beyond. Business failures took a
toll on banks, and both business and bank failures rose sharply as the Depres-
sion wore on. Banks became increasingly unwilling to lend, except to the
federal government, which took a further toll on investment and aggregate
demand.
Business default had two knock-on efects: frst, it threatened the solvency
of the banks themselves; second, even when the banks survived, they became
more cautious as defaults and default risk increased, hence, though at the out-
set bank deposits remained relatively immune to the fall in the value of col-
lateral, as banks were either unwilling to recognize publicly the threat to sol-
vency or were trying to accommodate their customers—perhaps some of
both. But as time went on and the economy failed to recover, bank deposits
Copyright © 2021. Harvard University Press. All rights reserved.
fell by more than strict proportionality with PY would have dictated. Table
8.14 bears this out: until 1932, CIAC loans fell less rapidly than output, but
between 1932 and 1933 much more rapidly. A contributing factor to the
decline in the money supply was the shif from bank deposits to cash, a
refection of the increasing systemic risk to which the banking system was
subject.
Businesses that managed to stay current on their debt with respect to work-
ing capital still suffered from the increase in the real value of debt tied to
equipment and structures. At the very least, this meant increased dif ficulties
in servicing their debts, and for some businesses it was just too much. In
chapter 7 this problem was formulated in terms of an increase in the real debt
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
276 Raising Keynes
on fxed capital, B/P. Te (real) net worth of the representative frm was
stated as
B M
− + Real Fixed Capital − 1 + Real Working Capital.
P P
Firms that managed to keep the last two terms in line with each other would
still have had to deal with a growing wedge between the real debt and the real
value of fxed capital.
Judging by the failure of interest rates to fall, the decline in M1 did nothing
for M 2. On the contrary: the decline in business confdence increased liquid-
ity preference, and the result was higher nominal interest rates, except for
government obligations. Te increase in the hurdle rate further dampened
investment demand. Even more important than the rise in nominal interest
rates was the Fisher efect, which implied a double-digit gap between real and
nominal interest rates.
With one important amendment this interpretation of the data conforms
to the vision if not the letter of Te General Teory. Te amendment concerns
the role of debt and defation. Both were very much in the public eye, topics
of lively political debate, but Keynes did not pay a lot of attention to the price
level and did not incorporate changing prices into the models of Te General
Teory.
If marginal for Keynes, debt and defation were, as I emphasized in the
previous chapter, central to Irving Fisher’s theory (1933). In the end, how-
ever, Fisher supplements Keynes rather than providing an alternative. Fisher
had no concept of aggregate demand and no argument for why equilibrium
should be characterized by persistent defation, nor did he provide a frame-
work for how debt and default, caused by the persistent defation, in turn fed
back into the economy through their negative impact on aggregate demand.
Data ofen do not speak with great clarity, and the data of the Great De-
pression are a prime example. Even if we accept that the ultimate cause of
Copyright © 2021. Harvard University Press. All rights reserved.
the Depression remains a mystery, there are two alternatives to the dynamics
of how it unfolded that are encapsulated in Figure 8.1. One deploys the static
second-pass model; the other argues that all would have been well if in
competence at the Federal Reserve had not transformed the relatively mild,
one might say routine, downturn of 1930 into the catastrophe of 1931 and
1932.
Keynes’s own version of the second-pass model, to be sure, assumes a given
money wage, leaving only the price level (and the real wage) to be determined
by aggregate demand. Tis model clearly does not correspond to the reality of
the Depression, characterized by falling prices and wages.
Modigliani’s version ofers more hope in this regard since it allows for a
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 277
variable level of wages as well as prices. Indeed, for Modigliani a falling price
level is the mechanism that ultimately restores full-employment equilibrium
in the face of falling demand. In Modigliani’s model a lower price level means
that less money is needed for transactions; for the redundant money to be
absorbed into asset portfolios, interest rates, including the hurdle rate, must
fall. Te stimulus to investment in turn boosts aggregate demand and restores
full employment.
Modigliani got the frst part right: falling wages and prices did reduce the
transactions demand for money. But none of the rest followed. First, lower
prices reduced not only transactions demand but also the supply of transac-
tions money. As demand fell, banks simply created less money, and as Table
8.13 shows, excess reserves grew, as did the ratio of reserves to deposits. Not a
dime moved into asset portfolios. Te overall money supply, measured by the
sum of demand deposits and currency in Table 8.12, contracted. Second, the
hurdle rate of interest, proxied by the Baa bond rate in Table 8.18, rose due to
a spike in liquidity preference. Tird, the Fisher efect made a bad situation
worse; as Table 8.7 shows, price declines added 9 percentage points to the real
interest rate in 1931 and more than 10 percentage points in 1932.
Milton Friedman and Anna Schwartz (1963a) blame the ineptness of the
Fed leadership. In summing up the lessons of their investigations, Friedman
and Schwartz write,
At all times throughout the 1929–33 contraction, alternative policies were
available to the [Federal Reserve] System by which it could have kept the
stock of money from falling, and indeed could have increased it at almost
any desired rate. Tose policies did not involve radical innovations. Tey
involved measures of a kind the System had taken in earlier years, of a kind
explicitly contemplated by the founders of the System to meet precisely the
kind of banking crisis that developed in late 1930 and persisted thereafer.
Tey involved measures that were actually proposed and very likely would
have been a dopted under a slightly diferent bureaucratic structure or distri-
Copyright © 2021. Harvard University Press. All rights reserved.
bution of power, or even if the men in power had had somewhat diferent
personalities. (p. 693)
For Friedman and Schwartz, there is no question that earlier, pre-Fed, mone-
tary arrangements would have prevented the disaster:
If the pre-1914 banking system rather than the Federal Reserve System had
been in existence in 1929, the money stock almost certainly would not have
undergone a decline comparable to the one that occurred. (p. 693)
Tey also argued that the monetary contraction was not simply independent
of declines in output and prices, it was the cause of these declines:
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
278 Raising Keynes
Te contraction is additional strong evidence for the economic indepen
dence of monetary changes from the contemporary course of income and
prices, even during the early phase of the contraction, from 1929 to 1931,
when the decline in the stock of money was not the result of explicit restric-
tive measures taken by the System. (p. 694)
Tere are many holes in the Friedman–Schwartz story—some of which
are explored in the appendix to this chapter. Te most glaring one is the
growth of excess reserves as the Depression deepened. For money demand to
accommodate to money supply—the key assumption underlying Friedman–
Schwartz—banks have to be fully loaned up or, equivalently, maintain a con-
stant margin of excess reserves. Te dramatic build-up of excess reserves in
1932, continuing into 1933—see Table 8.13, and Table 8.20 in the appendix—
tells a diferent story. In 1933, Jacob Viner summed up the situation of the
banks thus:
In the past three years the test of a successful banker has been the rate of
speed with which he could go out of the banking business and into the
safety-deposit business. Tose bankers have survived who have succeeded in
the largest degree and at the most rapid rate in converting loans into cash.
Tat has been good banking from the point of view of the individual banker,
or of his individual depositors; but from the social point of view it has been
disastrous. (1933b, p. 130)
Federal Reserve policy may or may not have been inept; as Barry Eichen-
green argues (2015), it is not clear that the Fed could have done much more
than it did until Roosevelt cut the link to gold.4 So even if Friedman is right
that the Fed didn’t do much to stop the rot, this failure hardly makes the Fed
the cause of the Depression. Te efectiveness of monetary policy in a period
of slack demand has rightly been likened to pushing on a string.5 Te reason
for this inefectiveness is summarized in Figure 8.1, which refects the theory
Copyright © 2021. Harvard University Press. All rights reserved.
developed in the third-pass model that I laid out in chapters 6 and 7. In this
theory the central bank has little to no leverage on the hurdle rate of interest
because of its limited control of the public’s stock of short-term and long-
term fnancial assets, and, given the rules and norms of twentieth-century
banking, even less control over the public’s liquidity-preference function.6 Te
Fisher efect compounds the misery.
Why Was Agriculture Diferent?
An important qualifcation to the Keynesian account of the Great Depression
is the anomalous behavior of the farm sector.7 As was noted at the beginning
of this chapter, agricultural production actually increased during the Depres-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 279
sion. But this did not mean farmers escaped unscathed. Relative prices turned
against agriculture, as Tables 8.6 and 8.7 indicate: compared with nonagricul-
tural goods, the price level of farm products in 1932 was half what it was in
1929.
Equally, perhaps more, important, than the terms of trade, the fall in the
nominal price level meant that debt quickly became unmanageable. As we
have seen, real-balance and Fisher efects gave the nonfarm business sector a
bad cold—farmers got pneumonia.
Te relationship between farm prices and the output of the nonfarm busi-
ness sector during both the Depression and the subsequent recovery is strik-
ing. Figure 8.2 plots nominal farm prices, and Figure 8.3 plots farm prices
110
Index of farm prices (1929 = 100)
100
1929
90
1930
80 1937
1936
70
1935
1938
60
1934 1931
50
1933
40 1932
30
60 65 70 75 80 85 90 95 100
Real nonfarm business output ($ billions)
Figure 8.2 Farm prices vs. nonfarm output.
Ratio of farm prices to prices of nonfarm
110
1929
100
business output (1929 = 100)
Copyright © 2021. Harvard University Press. All rights reserved.
1936 1937
90
1935 1930
80
1938
70
1934
1931
60
1933
50
1932
40
60 65 70 75 80 85 90 95 100
Real nonfarm business output ($ billions)
Figure 8.3 Relative prices vs. nonfarm output.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
280 Raising Keynes
relative to nonfarm prices. Te correlation is clearly strong in both cases,
stronger for relative prices.
Tis much is well understood, but there remains the question of how to
explain the correlation. How was it that real output and employment in agri-
culture actually rose during the Depression, and why were agricultural prices
battered by the decline in nonfarm output?
Te mainstream answer to these questions follows from its emphasis on
frictions and rigidities: the resilience of agricultural output and the collapse of
agricultural prices were due to the competitive structure of this sector. Tere
is more than a germ of truth in this explanation. Unlike industry and trans-
portation and large parts of wholesale and retail trade, which had become
increasingly infected with the imperfections of monopoly and oligopoly, agri-
culture remained close to the economist’s model of the price-taking producer.
Viner was one mainstream economist who saw the diference between
farm and nonfarm sectors in these terms. In his lecture at the University of
Minnesota in February 1933, he observed,
Agriculture has in this depression been the only major portion of the na-
tional economy which has continued to operate and to produce to the limit
of its capacity. Tere could be no business depressions if all sections of the
national industry were as readily to adjust their demands to the level at
which the market would absorb their full output. (1933a, p. 11)
Another was John Kenneth Galbraith though he was later to become a tren-
chant critic of mainstream economics. Prodigious author, senior of ficial in the
Of fi ce of Price Administration during World War II, adviser and ambassador
under John F. Kennedy, Galbraith was trained as an agricultural economist
and very likely his pre-Keynesian perspective on the causes of the Great De-
pression owed something to this training. In Galbraith’s 1936 view (1981,
p. 64), the problem was structural:8
Copyright © 2021. Harvard University Press. All rights reserved.
Free competition had given way to oligopoly and monopolistic competition,
and because of the latter too many resources were being wasted on advertis-
ing and salesmanship. Te shortfall in production from these defects caused
the depression. . . . Te remedy was more competition.
Te mainstream view has changed little since the 1930s. Giovanni Federico
(2005, p. 972) concluded his investigation into the causal role of agriculture in
the Great Depression thus:
Te . . . most plausible hypothesis [why agricultural prices fell by so much
more than non-farm prices] focuses on the asymmetry between agriculture
and the rest of the economy. Agriculture was a competitive sector and, in the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 281
peculiar conditions of the depression, its elasticity of supply was bound to be
low. In fact the opportunity cost of farmers’ labor was extremely low as it
depended on the expected wages in other sectors. Real wages were rising,
but the high rate of unemployment made the prospect of emigration to cities
hardly attractive. Tus farmers were likely to go on producing as long as
revenues covered their variable costs plus a minimum income for the pur-
chase of indispensable consumer goods. In this situation, demand shocks
caused prices to fall. In contrast, in the rest of the economy demand shocks
caused a fall in output because prices were sticky. A long interpretative tradi-
tion, dating back to Keynes, attributes this feature to wage stickiness, deter-
mined by the power of trade unions, by the existence of welfare benefts or
by the existence of long-term labor contracts. Another interpretation, frst
suggested by Steindl [no reference, presumably, Steindl 1952] and recently
revived by Madsen [2004a, 2004b] blames limited competition in manufac-
turing.
Federico has it partly right: the supply elasticity of farm goods was indeed
low, but the contrast with sticky prices elsewhere is misleading. He, along
with Viner and Galbraith, were of course correct in observing that oligopoly
and monopoly and all sorts of rigidities and frictions put more of a burden on
output and employment adjustment in the event of a shock to the economy.
But the argument of Keynes—the argument of this book—is that even in the
case of perfect competition, output and employment will suffer along with
prices. Tis didn’t happen in agriculture.
Farm output, if not fnancial solvency, weathered the storm of the Depres-
sion9 not because of the proximity of agriculture to the norms of competition,
but because of its distance. Te departure from the competitive model was
not in the goods market, where farmers were indeed price takers, as the
manual of perfect competition dictates, but in the labor market.
Consider how the competitive frm determines output and employment. In
Copyright © 2021. Harvard University Press. All rights reserved.
equilibrium it is a price taker both in goods markets and in labor markets.
Te competitive frm maximizes profts by producing, where real marginal
cost (MPL)−1 is equal to the given real price (P/W) at Q0 in Figure 8.4(a), and,
equivalently, where the marginal productivity of labor is equal to the given
real wage, at L0 in Figure 8.4(b). Although farmers were price takers in prod-
uct markets, family farms do not meet labor demands primarily by hiring
workers—at least they didn’t until the meaning of “family farm” changed fun-
damentally afer World War II.
Tis is not to say that hired labor played no role. According to the Depart-
ment of Agriculture data in Table 8.4, hired labor constituted some 25 percent
of the U.S. farm labor force as the United States entered the Depression. Te
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
282 Raising Keynes
W
P
P
W 1 MPL
MPL
W
P 0
P
W 0
Q0 Q L0 L
(a) Real marginal cost = real price at Q 0 (b) Real marginal productivity of
labor = real wage at L0
Figure 8.4 Proft maximization under perfect competition.
decline in hired labor between 1929 and 1933 was in fact smaller in percent-
age terms than the decline of nonfarm employment (15 percent vs. 25 per-
cent), presumably because of the complementarity between hired labor and
family labor. At peak times, especially at harvest, many family farms required
extra help beyond what the family could mobilize on its own.10
Te diference between hired labor and family labor is that it makes no
sense to apply the rule of p
roft maximization—add labor only so long as the
marginal productivity exceeds the real wage—to family members. Family
members have to be fed, clothed, and sheltered whether they work or not. Te
cost of employing a family member is a fxed cost, not a wage bill that varies
in proportion to the number of hours or days worked.11
Farm labor, both family labor and hired labor, had been a declining per-
Copyright © 2021. Harvard University Press. All rights reserved.
centage of the total labor force since the early days of the Republic: as the
World War I ditty went, “How can you keep ’em down on the farm afer
they’ve seen Paree?” But, as the economy weakened in 1930 and collapsed in
1931 and 1932, rising urban unemployment caused a hiatus in the steady fow
from the farm. On this point I have no disagreement with Federico or others
who oppose the fexible farm economy to the rigid nonfarm economy. If there
was no work to be had in the city, many a young man simply went home.12 As
Robert Frost wrote in his poem “Te Death of the Hired Man,” “Home is the
place where, when you have to go there, they have to take you in.” And if there
were more hands to feed, they might as well work.
Studying the very diferent context of subsistence peasant agriculture, the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 283
Russian economist Alexander Chayanov (1966 [1925]) suggested a model
that takes account of the distinctive features of family farms. Chayanov pro-
posed replacing proft maximization by the criterion of maximum output. In
the context of unlimited supplies of labor, which according to Chayanov char-
acterized the relatively undeveloped economy of czarist Russia, this meant
applying additional hours of labor until its marginal product fell to zero. Te
more general implication is that labor supply determines the marginal prod-
uct of labor, as in Figure 8.5, rather than, as in Figure 8.4, the wage rate deter-
mining the marginal product of labor from the demand side.
At the aggregate level, the intersection between the MPL schedule and the
LS schedule in Figure 8.5 looks just like the picture of how the real wage is de-
termined in the mainstream model: if we simply add up each farm’s marginal
productivity and LS schedules, we get an aggregate marginal-productivity
schedule and an aggregate LS schedule that mirror the labor-demand and LS
schedules of chapter 2. Te real wage is determined by the marginal produc-
tivity of labor at full employment.
However, the two models are very diferent. In the Chayanov model, the
marginal-productivity schedule is not a demand schedule in the sense of a
schedule of how much labor will be employed at diferent wage rates. It can’t
be, for there is no wage rate in Chayanov’s conception of the family farm. Te
relevance of productivity is limited to the case where the marginal productiv-
ity of labor goes to zero before utilizing the available labor. Moreover, in the
mainstream model, the LS schedule refects the disutility of labor, and unem-
Bushels of wheat MPL
per hour of work
Supply of labor
Copyright © 2021. Harvard University Press. All rights reserved.
L0 L
Figure 8.5 An alternative model of the family farm.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
284 Raising Keynes
ployment leads to workers bidding down real wages as they compete for jobs.
In Chayanov’s model, by contrast, since there are no wages there can be no
weighing of the disutility of labor with a real wage. Instead, labor supply di-
rectly drives family employment.
Tis diference between family farms and a capitalist enterprise makes it
possible for a failure of demand in the capitalist sector to bring down farm
prices while farm output hardly changes—and sow the seeds of bankruptcy
and disaster. Consider an economy that includes a sector of family farms co-
existing with a capitalist industrial sector.13 Te capitalist sector produces a
dual-purpose good used for both consumption and investment, whereas the
agricultural sector produces a single-purpose consumption good, food. We
denote industrial output by YI, agricultural output by YA, agricultural income
by YA. As we shall see, it is necessary to distinguish farm output and income
because in this model falling prices mean rising agricultural surpluses that
remain unsold. As noted at the end of chapter 5, an economy of sole propri-
etors—family farms ft the bill—drives a wedge between income and output.
Demand for industrial goods for consumption purposes, CI , is given in
nominal terms by
PI C I = II PI YI + AI PAYA .
Here, PI and PA are the prices of the two goods, II is the propensity to con-
sume industrial goods by the producers of industrial goods, and AI is the
propensity to consume industrial goods out of income from producing food.
In real terms we have
PA
C I = II YI + AI YA.
PI
Demand for industrial goods for investment is the sum of the demands for
additional capital for the capitalist sector and for additional capital for the
Copyright © 2021. Harvard University Press. All rights reserved.
farm sector:
I = II( ) + IA( ).
Total demand for industrial goods is
YI = CI + I,
so that
PA
( 1 − II ) YI = AI YA + I ,
PI
which is to say that saving out of income from producing industrial goods
must be equal to the consumption of industrial goods by farmers plus invest-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 285
ment. Tis equation defnes an industrial-demand schedule that plays the
same role here that the AD schedule plays in the one-sector economy.14
Analogously, the demand schedule in the agricultural sector is given by
equality between agricultural income and desired food consumption,
PAYA = PAC A = IA PI YI + AA PAYA ,
so that
−1
P IA
YA = A YI .
P
I 1 − AA
Figure 8.6 reproduces the structure of Figure 8.1 to show a serendipitous
full-employment equilibrium in the industrial sector at E and an unemploy-
ment equilibrium at F. Te new element in Figure 8.6 is that the price on the
vertical axis is the real price of industrial goods, and output on the horizontal
axis is the output of industrial goods. Figure 8.7 shows the corresponding
equilibria in the agricultural sector.
Initially, the agricultural economy is at E in Figure 8.7 because the indus-
trial economy is at E in Figure 8.6. At this point, PI = 0, so the stationary
relative-price locus, (PA/PI)• = 0, is identical to the stationary locus of the price
of agricultural goods, PA /PA = 0. Assuming fexprice adjustment in agricul-
ture, (PA/PI)• = 0 is the food-demand schedule.
Why fexprice adjustment? Te choice may appear arbitrary, especially hav-
ing argued that fxprice equilibrium makes more sense in the macro model
pictured in Figure 8.1. Tere is really no choice. Te logic of the family-farm
Y I 0 Y I 0
PI
W P I 0
Copyright © 2021. Harvard University Press. All rights reserved.
5 Supply of goods
PI
W 0 E
4
Stationary real price F
3 of industrial goods
2 0
W
Supply of labor
1
Demand for industrial goods
100 200 300 400 500 600 Y
Figure 8.6 Te capitalist (industrial) sector: fxprice adjustment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
286 Raising Keynes
Stationary relative prices
PA PA Y A 0
0
PI PI YA YAS
Supply of labor
PA
PI 0 E
PA Demand for food
G F
PA
PI 1 0
PA
1
P IA
Y A A YI
P
I 1 AA
YA
Figure 8.7 Te family-farm sector.
model dictates fexprice adjustment because output is determined by the ex-
ogenously given supply of labor. Output cannot respond to a shortfall in ex-
penditure, which explains why agricultural surpluses mounted during the
Depression while inventories in the nonfarm business sector shrunk.15
Once the industrial economy moves away from E, the intersection of the
demand and supply curves in Figure 8.7 no longer determines equilibrium
relative prices. When industry is at F in Figure 8.6, the price ratio PA/PI can
remain stationary only if PA is falling at the same rate as PI. But the price of
farm output falls if and only if farm output is to the right of the demand
schedule. So the new stationary-(PA/PI) locus, the red-dashed schedule in Fig
ure 8.7, lies to the right of the black-dashed food-demand schedule, which is
the stationary-PA locus when the industrial economy is at F in Figure 8.6.
Copyright © 2021. Harvard University Press. All rights reserved.
Te reason for the distinction between farm income and output should
now be clear. In Figure 8.7 the point F is the new equilibrium in the sense of
stationary output and relative prices. At F, agricultural prices must be falling
at the same rate as industrial prices at F in Figure 8.6, which implies an agri-
cultural surplus that fnds no market at the equilibrium (PA/PI)1. Farm income
is limited by what can be sold, represented by the point G. Mounting sur-
pluses, empirically problematic as a description of the macro economy, as
Table 8.2 shows, were part of the catastrophe that farmers experienced be-
tween 1929 and 1933.
Figure 8.7 determines equilibrium PA/PI for given YI, but in the demand
equation in Figure 8.6, YI depends on PA/PI. We have to take the two demand
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 287
equations together to determine both the level of industrial demand and
the corresponding relative price of food. Rearranging the industrial-demand
equation gives
AI PA 1
YI = YA + I,
1 − II PI 1 − II
and rearranging the agricultural-demand equation gives
PA IA
YA = Y.
PI 1 − AA I
Substituting from the agricultural equation into the industrial equation we
have
AI IA 1
YI = YI + I.
1 − II 1 − AA 1 − II
Collecting terms, we have an equation for industrial output as the product of
the multiplier and the level of investment,
1 − AA
YI = I.
( 1 − II ) ( 1 − AA ) − AI IA
We substitute back into the agricultural equation to obtain the value of farm
income determined by the level of investment:
PA IA
YA = I.
PI ( 1 − II ) ( 1 − AA ) − AI IA
Evidently, greater investment leads not only to greater industrial output but
also to higher relative prices for farm goods. Te corollary is that if invest-
ment dries up, as happened afer 1929, not only will industrial output and
Copyright © 2021. Harvard University Press. All rights reserved.
prices fall; farm prices will also fall and fall harder than industrial prices.16
Observe that on present assumptions the level of farm output plays no role
in the determination of industrial output. Changes in the level of farm output
are cancelled out by ofsetting changes in the ratio of agricultural to industrial
prices, PA/PI: as we have seen, the equilibrium level of industrial output de-
pends only on the level of investment and the multiplier.
Tis result depends on the assumption that the propensities to consume are
fxed in both sectors. Truth to tell, the only defense of this assumption is trac-
tability. In all likelihood, as relative prices vary, agents will vary the composi-
tion of their consumption baskets. It is reasonable to assume that industrial
workers will purchase more from agriculture and less from industry as the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
288 Raising Keynes
relative price of food falls. Perhaps their overall consumption as a fraction of
income will change too (Grigoli, Hermana, and Schmidt-Hebbel 2016).
Farmers will also react to relative price changes, but not necessarily in the
same way. As the prices of farm products fall relative to the prices of indus-
trial goods, farmers are worse of and the income efects may induce them to
purchase less from industry, despite substitution efects working in the oppo-
site direction. Again, their saving may or may not increase.
Diferent assumptions about how agents in the two sectors react to price
changes will lead to diferent conclusions about the efect of changes in de-
mand and supply. Of particular interest is causality running from agriculture
to industry. Te mathematical appendix to this chapter shows that the level of
industrial output is sensitive to the level of agricultural output if we drop the
assumption of fxed propensities to consume. Te model in the mathematical
appendix assumes instead that the fraction of income allocated to consump-
tion of industrial goods—by capitalists, industrial workers, and farmers—is
positively related to the relative price of food. For mathematical tractability I
also assume that propensities to save do not change (in other words, that the
fraction of income allocated to food goes down by exactly the same amount
as the increase in the fraction devoted to industrial goods). A surprising re-
sult is that as YAS increases, YI falls; the adverse efect on agricultural prices
more than ofsets the extra agricultural production.
Tis suggests that the increase in agricultural production in the 1920s, and
the consequent deterioration of the terms of trade, may have contributed to
the fall in industrial output from 1929 on,17 as Jakob Madsen (2001) suggests.
My own view is that causality running from industry to agriculture is more
important, the fall in investment (and/or the propensity to consume) afer
1929 producing the paradoxical result of more output in the agricultural sec-
tor. Indeed, the nadir of 1932–33 may have been an equilibrium: farm and
nonfarm output, the real price of industrial goods, and the relative price of
agricultural goods might have persisted indefnitely, with nominal prices (and
Copyright © 2021. Harvard University Press. All rights reserved.
wages) falling to ever-lower levels. Except that Franklin Roosevelt, inaugu-
rated as president on March 4, 1933, took several actions that halted the
downward spiral.
Tere has been relatively little analysis of how agricultural recovery infu
enced the return to prosperity. One exception is Peter Temin and Barrie Wig-
more (1990), who consider devaluation of the dollar (see chapter 7 above) the
lynchpin of the new economic regime, not only for its efects on commodity
prices but, more generally, for its impact as a signal that change was the new
order of the day. (Te Temin–Wigmore argument was formalized by Gauti
Eggertsson [2008].) Another exception is Joshua Hausman, Paul Rhode, and
Johannes Wieland (2019), who focus more directly on the farm recovery as
the engine of growth and recovery in the wider economy.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 289
It is certainly the case that the Roosevelt administration’s agricultural poli-
cies assumed a negative relationship between YAS and YI and attempted to har-
ness this relationship to increase both agricultural and industrial incomes.
Restricting agricultural production in order to raise prices and incomes was
the clear intent of the Agricultural Adjustment Act, and the U.S. Department
of Agriculture under Henry Wallace carried out the new policy with zeal.18
Looking through the rearview mirror in 1936, Chester Davis, the head of the
Agricultural Adjustment Administration, observed:
We know that the country accepted this identifcation of the farm problem
with the national problem. Rarely has an act of such moment encountered
less opposition at its birth. Nonfarmers, as well as farmers, expected to bene
ft from it. Nonfarmers who thought the matter through believed that fairer
prices to farmers would be bread cast upon the waters. Tey expected that
the cost of the adjustment programs to them in the form of higher prices of
food and fber would be counterbalanced by increased industrial employ-
ment and trade. (1936, p. 229)
Te logic is the logic of immiserizing growth (see note 17)—the negative rela-
tionship between YAS and YI—but in reverse. If the decrease in YAS is large
enough, it could in principle propel the economy back to full employment, as
in Figures 8.8 and 8.9. Te increase in PA/PI more than makes up for the fall in
agricultural production, driving demand upward in both sectors.
Stationary relative prices
PA
0
PI
Y A 0
PA
PI YA YAS
PA Supply of labor
Copyright © 2021. Harvard University Press. All rights reserved.
H
PI 2
Demand for food
PA G F PA
0
PI 1 PA
1
P IA
Y A A YI
P
I 1 AA
YA
Figure 8.8 Te family-farm sector adjusts to a fall in supply.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
290 Raising Keynes
Y I 0 Y I 0
PI
W P I 0
5 Supply of goods
PI
W 0 H
4
Stationary real price F
3 of industrial goods
2 0
W
Supply of labor
1
Demand for industrial goods
100 200 300 400 500 600 Y
Figure 8.9 Te industrial sector responds to a decrease in the supply of food.
Tere is one problem with this formulation of the relationship between
agriculture and industry. In my story, agricultural prices are determined en-
dogenously, which is tantamount to assuming a closed economy. Tis as-
sumption, as I noted in note 13, describes the U.S. industrial economy of the
interwar period accurately enough, and a large part of the agricultural econ-
omy was geared to the domestic market—think dairy farms, cattle ranches,
and pig farms; orange groves, market gardens, and apple orchards. My calcu-
lations indicate that about 70 percent of agriculture was efectively insulated
from foreign trade. But this still leaves a part of agriculture for which prices
were determined in international markets. Half the cotton crop was exported
in 1929 and over three-quarters in 1932.
As Temin and Wigmore (1990), Eggertsson (2008), and Hausmann, Rhode,
Copyright © 2021. Harvard University Press. All rights reserved.
and Wieland (2019) all note, the exchange rate was a critical determinant of
dollar prices of exports. Te United States efectively went of the gold stan-
dard in April 1933, and this had the immediate corollary of devaluing the
dollar relative to other currencies. Export prices in terms of dollars rose ac-
cordingly, and agricultural exports were a large share of the total in this pe-
riod; cotton alone accounted for 17 percent of total exports in 1932. Between
April 1933 and February 1934, when a new gold parity was fxed at $35.00 per
ounce, the price of cotton practically doubled in dollars but went up by only
15 percent in terms of sterling.19 Te price of wheat followed the same upward
trajectory, increasing by more than 75 percent between April 1933 and Febru-
ary 1934 in dollars, while remaining fat in London. For the most part, goods
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 291
that did not enter into international trade increased much less in value, if they
increased at all. Milk was up almost 20 percent, but cattle were only 10 per-
cent more expensive in February 1934 than in April of the preceding year;
hog prices were unchanged. Corn was an exception: its domestic price rose
more than the price of wheat even though little was exported—not even
$3 million worth, out of a crop valued at more than $1.2 billion in 1933 (Bu-
reau of the Census 1935).20 Despite infationary fears and scare tactics to in-
duce consumers to buy before prices rose, the U.S. consumer price index rose
less than 6 percent. (Food prices rose by just over 20 percent.)
A diferent model of price and output determination is required to account
for export crops. We can still use the two-sector schema, but causality is dif-
ferent once agricultural prices are determined outside the model. In this case,
the international price, along with the domestic demand schedule, deter-
mines the quantity consumed domestically, and the diference between total
production and domestic consumption goes to export markets. In Figure
8.10, the price is initially fxed at (PA/PI)0, and domestic consumption is deter-
mined by the equilibrium at E. Domestic consumption is YA0, and exports plus
additions to stocks are equal to YAS − YA0. Causality in this case runs in one
direction, from the exogenously fxed export price to farmers’ incomes to in-
dustrial demand. Te only feedback from industry to agriculture is in deter-
mining how much of the crop is exported—what’s lef over afer domestic
demand is met.
In this model, devaluation increases the relative price of farm products to
PA
Y A 0
PI
PA F
Supply of labor
PI 1
PA
E
Copyright © 2021. Harvard University Press. All rights reserved.
PI 0
Domestic demand
YA1 YA0 YAS YA
Figure 8.10 Te export-crop sector.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
292 Raising Keynes
(PA/PI)1. Te new equilibrium of domestic demand is at F. Industrial produc-
tion increases in response to the additional demand coming from farmers, as
in the closed-economy model. Exports also increase as a by-product of re-
duced domestic demand for agricultural products.
How well did the policies of devaluation and production restrictions suc-
ceed? Te problem of assessing the role of crop restrictions, and more gen
erally, the role of agricultural policy (counting devaluation as agricultural
policy), is not only that other policies were in play. To compound the prob
lem, Mother Nature collaborated with government in restricting output. Te
clouds of sand that covered the agricultural heartland and transformed it into
the dust bowl had a silver lining.
Food for Thought
In the Great Depression, much more havoc was wrought by wage and price
fexibility than by rigidity. Keynes came to this conclusion on theoretical
grounds in chapter 19 of Te General Teory, which is concerned with the
pros and (mostly) cons of fexible prices and wages. He concludes with this
observation:
To suppose that a fexible wage policy is a right and proper adjunct of a sys-
tem which on the whole is one of laissez-faire, is the opposite of the truth. It
is only in a highly authoritarian society . . . that a fexible wage-policy could
function with success. One can imagine it in operation in Italy, Germany or
Russia, but not in France, the United States or Great Britain. (p. 269)
Te mortgage-debt crisis brought on by the disastrous fall in prices, espe-
cially agricultural prices, points up the truth of the paradox that laissez-faire
requires an authoritarian government. In fact, the democratic process in the
United States responded quite sensibly to the intolerable position of mortgage
debtors: the increasing debt burden led the Minnesota legislature to enact a
Copyright © 2021. Harvard University Press. All rights reserved.
statute allowing loan repayments to be postponed on the fnding by a local
judge of fnancial hardship—a perfectly reasonable reaction to hard times.
Te problem was that the warp of American democracy on which the Min-
nesota legislature tried to weave debt relief was the Constitution of the United
States, and on every traditional canon of constitutional interpretation—lan-
guage, intent of the framers, and the spirit of the text—the Minnesota law was
plainly unconstitutional. Te founding fathers had deliberately constrained
democracy in order to safeguard property rights. And so it was argued before
the Supreme Court in the fall of 1933.
With respect to language, the relevant constitutional provision is clear:
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 293
it prevented states from enacting any law “abridging the obligation of con-
tract.” Moreover, this was the clear intent of the framers: the contract clause
was a reaction to the agitation in many states—including the famous rebel-
lion led by Daniel Shays in my backyard (Western Massachusetts)—to pro-
vide relief to debtors in the post-Revolutionary depression. Finally, the spirit
of the provision was clearly directed against populist actions on the part of
states, like the action of the Minnesota legislature to postpone debt repay-
ment.
Nonetheless, in Home Building & Loan Association v. Blaisdell (290 U.S.
398, 1934), a very conservative, very originalist, U.S. Supreme Court, by
a narrow vote of 5 to 4, upheld the Minnesota statute on the convoluted
grounds that although the Depression did not create special state power, it
might be the occasion for the exercise of such power.
Te power at issue was the police power of the states. Police power? On
second thought, quite right. Te Minnesota legislature had good reason to
fear riots and worse if hard-pressed debtors were not granted some relief.
In this regard authoritarian regimes clearly have a comparative advantage:
they are not constrained in the use of repressive force in the way that demo-
cratic regimes are. In 1934 Germany, the cousins of the angry farmers who
came together to disrupt foreclosure sales in Minnesota and throughout the
Midwest would have found themselves in concentration camps if they had
dared such a protest and lived to tell the tale. In the Soviet U
nion, peasants
were starved to death for resisting foreclosure of another sort.
appendix: milton friedman and anna schwartz
on what made the depression great
Friedman and Schwartz (1963a and 1963b) argue that the Fed could have
prevented the downturn of 1930 from turning into the Great Depression. Te
best evidence that they can muster is their analysis of open-market operations
Copyright © 2021. Harvard University Press. All rights reserved.
in 1932, when the Fed for a brief period actively purchased Treasury securi-
ties. In their view, this led to an expansion of the money supply, which in turn
led to a recovery in industrial production. Te rest of the story is that the
Fed’s failure to continue the program of buying Treasury securities aborted
the recovery, thus reinforcing the argument that the money supply deter-
mines, rather than refects, economic activity. By extension, all would have
been well if the Fed had intervened earlier and consistently as it did briefy in
1932.
Friedman and Schwartz rely heavily on the timing of changes in the money
supply relative to the uptick in industrial production:
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
294 Raising Keynes
Te tapering of of the decline in the stock of money and the beginning of
the purchase program were shortly followed by an equally notable change in
the general economic indicators . . . Wholesale prices started rising in July,
production in August. (1963a, p. 324)
Tey admit
Tere is, of course, no way of knowing that the economic improvement re
fected the monetary improvement. But it is entirely clear that the reverse
was not the case. Aside from the precedence in time of the monetary im-
provement, the program of large-scale open market purchases was a delib-
erative action undertaken by the Reserve System. And it was the major fac-
tor accounting for the monetary improvement. (p. 324)
Having feigned agnosticism, Friedman and Schwartz show their hand even
more clearly in the next paragraph:
Te timing relations, previous experience, and general considerations all
make it highly plausible that the economic improvement refected the infu
ence of the monetary improvement, rather than the only other alternative—
that it occurred shortly thereafer entirely by coincidence. We have observed
that, in the past, an increase in the rate of monetary growth—in the present
case, from rapid decline to mild decline and then mild rise—has invariably
preceded a trough in general business. Afer three years of economic con-
traction, there must have been many forces in the economy making for re-
vival, and it is reasonable that they could more readily come to fruition in a
favorable monetary setting than in the midst of continued fnancial uncer-
tainty. (p. 324)
Tis is thin gruel. “Past experience” is of very limited relevance, as are “gen-
eral considerations.” As Friedman and Schwartz were well aware, the Great
Depression was unprecedented in the economic history of the United States.
Copyright © 2021. Harvard University Press. All rights reserved.
We are lef with the fact that the episode of open-market purchases preceded
the uptick in industrial production.
Tere is more than one problem with this timing argument. Federal Re-
serve numbers do indicate a sizeable increase in industrial production be-
tween July and October 1932 (Board of Governors of the Federal Reserve
System [n.d.], pp. S-148–S-149), 19 percent without seasonal adjustment and
13 percent on a seasonally adjusted basis. Impressive at frst sight but less im-
pressive considering that in the frst four months of 1931 there was a 10 per-
cent increase in industrial production (3 percent with seasonal adjustment),
an increase that appears to be unrelated to Federal Reserve policy. Wholesale
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 295
prices too saw an upward spike, almost 9 percent between May and October
1932. As was the case for output, there was a milder spike in wholesale prices
in the spring of 1931, prices rising by 2 percent between May and July (Fed-
eral Reserve Bank of St. Louis).
Te possibility that there was nothing special about the uptick in the sum-
mer and fall of 1932 is reinforced by the revised series of the industrial-
production index put together by Jefrey Miron and Christina Romer (1990).
Teir series shows much more volatility in the monthly fgures, with month-
to-month increases of 10 percent or more in every year between 1929 and
1933. Te only sustained increases (four months) were the periods January to
April and September to December 1931. None in 1932!
Even without the Miron–Romer revisions, the fact that output and prices
turned upward briefy on an earlier occasion makes pure coincidence a plau-
sible explanation of the 1932 correlation between monetary expansion—or
rather a slowdown in the rate of contraction—and the increase in real output
and prices. Miron and Romer make us wonder if there is anything to explain.
But there is a more basic question. If we stipulate, as the lawyers say, that
there were coincidental increases in the money supply on the one hand and
prices and output on the other, what, we may reasonably ask, was the mecha-
nism by which the progression “from rapid decline to mild decline and then
mild rise” in the money supply turned the economy around?
Friedman and Schwartz’s Monetary History presents no mechanism relat-
ing money to output and prices, but a mechanism is described in a paper of
theirs (1963b) that can be read as the theoretical companion of the book. Te
basic idea outlined in the paper will be familiar: open-market purchases by
the Federal Reserve lead to lower interest rates, and lower interest rates stimu-
late investment. Te frst step is an increase in the reserves of the banks:
Although the initial sellers of the securities purchased by the central bank
were willing sellers, this does not mean that they want to hold the proceeds
Copyright © 2021. Harvard University Press. All rights reserved.
in money indefnitely. Te [central] bank ofered them a good price, so they
sold; they added to their money balances as a temporary step in rearranging
their portfolios. If the seller was a commercial bank, it now has larger re-
serves than it has regarded before as suf ficient and will seek to expand its
investments and its loans at a greater rate than before. If the seller was not a
commercial bank, he is not likely even temporarily to want to hold the pro-
ceeds in currency but will deposit them in a commercial bank, thereby, in
our fractional reserve system, adding to the bank’s reserves relative to its
deposits. In either case, therefore, in our system, commercial banks become
more liquid. (Friedman and Schwartz 1963b, p. 60)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
296 Raising Keynes
Te commercial banks will consequently increase the supply of money:
Both the nonbank seller and commercial banks will therefore seek to read-
just their portfolios, the only diference being that the commercial banks
will in the process create more money, thereby transmitting the increase in
high-powered money to the total money stock. (pp. 60–61)
More money means, for a given level of prices and output, more demand for
fnancial assets:
It seems plausible that both nonbank and bank holders of redundant bal-
ances will turn frst to securities comparable to those they have sold, say,
fxed-interest coupon, low-risk obligations. But as they seek to purchase
these they will tend to bid up the prices of those issues [and thus reduce
yields]. Hence they, and also other holders not involved in the initial central
bank open-market transactions, will look farther afeld. (p. 61)
And what do they fnd as their gaze widens?
Te banks . . . their loans; the nonbank holders . . . other categories of securi-
ties—higher-risk fxed coupon obligations, equities, real property, and so
forth. (p. 61)
Te end result is greater investment:
As the process continues, the initial impacts are difused in several respects:
frst, the range of assets afected widens; second, potential creators of assets
now more in demand are induced to react to the better terms on which they
can be sold, including business enterprises wishing to engage in capital ex-
pansion, house builders or prospective homeowners, consumers who are
potential purchasers of durable consumer goods—and so on and on. (p. 61)
Tere is nothing theoretically problematic in the Friedman–Schwartz trans-
mission theory. Teir story actually fts Keynes’s own second-pass model, as
Copyright © 2021. Harvard University Press. All rights reserved.
well as Modigliani’s version of this model. If the central bank is reducing the
money supply just when the sensible policy is expansion, we have a recipe for
disaster, not recovery. Here Friedman–Schwartz’s monetarism does not con-
tradict Keynes. On the comparative-statics interpretation provided by the
second-pass model, Te General Teory is entirely consistent with the idea
that a combination of misguided policy and sheer incompetence caused the
Great Depression.
Te third-pass, dynamic, model tells a very diferent story. Te data for
deposits at all banks over the period June 1931 to December 1932 do indicate
that the decline that began in 1930 bottomed out, at least temporarily, in July
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • A Dose of Reality 297
1932. But, as Table 8.19 shows, the composition of bank assets does not follow
the Friedman–Schwartz story line. Holdings of U.S. government securities
rose between June 1931 and June 1932. More to the point, the end purpose of
the operation was foiled: loans continued south throughout this period, de-
clining by 6.5 percent between June and December 1932. Te more detailed
data available for the member banks of the Federal Reserve System confrm
this pattern. Table 8.20, like Table 8.19, shows that adjusted deposits (deposits
net of interbank deposits) hit a foor in June 1932. On the liability side, the
table makes it clear that the increase in investments afer that date was due to
a growing stake in U.S. Treasuries, a stake that continued to grow well afer
the Fed’s open-market purchase program ended. No doubt the liquidity of the
banking system improved, as indicated by the growth in excess reserves,
which rose by $300 million between June and December 1932. Perhaps this
was due, as Friedman and Schwartz suggest, to nonbank sellers of Treasuries
depositing the proceeds of their sales into their bank accounts. In any case,
Table 8.19 Assets and Liabilities, All Banks ($ millions)
U.S. Government
End of Month Deposits Adjusted Deposits Total Investments Obligations Loans
Jun 1931 56,902 51,769 19,982 6,682 35,285
Dec 1931 49,509 45,925 18,651 31,395
Jun 1932 45,411 42,093 18,422 6,895 27,888
Dec 1932 45,886 41,752 19,060 26,109
Source: Board of Governors of the Federal Reserve System 1943, table 2.
Table 8.20 Assets and Liabilities, Member Banks of the Federal Reserve System ($ millions)
Reserves at
U.S. Federal
Adjusted Total Government CIAC Reserve Required Excess
Loans1
Copyright © 2021. Harvard University Press. All rights reserved.
Deposits Investments Obligations Banks Reserves Reserves
End of Month Table 18 Table 18 Table 18 Table 18 Table 105 Table 105 Table 105
Jun 1931 31,338 12,106 5,343 8,922 2,404 2,275 129
Sep 1931 29,312 12,199 5,564 8,722 2,333 2,212 120
Dec 1931 27,391 11,314 5,319 8,126 2,069 2,010 60
Mar 1932 1,899 1,840 59
Jun 1932 24,717 11,414 5,628 6,892 2,062 1,827 234
Sep 1932 24,860 12,121 6,366 6,398 2,181 1,836 345
Dec 1932 25,020 12,265 6,540 5,970 2,435 1,909 526
Source: Board of Governors of the Federal Reserve System 1943.
1. CIAC loans are commercial, industrial, agricultural, and consumer loans not collateralized by securities.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
298 Raising Keynes
the increased liquidity of the banking system did not lead to an easing of
credit. CIAC loans continued to fall at virtually the same pace in the six
months afer June 1932, as in the six months before.
Te open-market purchase program added U.S. Treasuries to the Federal
Reserve’s portfolio, almost $1 billion between March and June 1932 (Board of
Governors of the Federal Reserve System [1943, table 91, p. 343]). But these
additions did not in themselves contribute to the improved liquidity of the
banks. Reserves hardly changed between December 1931 and June 1932. It
was afer the purchase program ended that bank reserves began to climb.
Te remaining channel by which monetary easing might have contributed
to recovery was via the interest rate. Here, as has already been noted, the evi-
dence suggests a classic liquidity trap, at least in the sense of a foor to the
hurdle rate when the T-bill rate hits the zero lower bound. Table 8.21 gives
interest rates on a variety of instruments, both loans and investments. Te
main impact of the Fed’s open-market purchases, not surprisingly, was on
short-term Treasuries, the yield on which fell sharply in June 1932, as the
bond-buying program topped out, and continued to fall through the summer.
Yields on longer-term government bonds and on short-term commercial pa-
per fell throughout 1932, but it is not clear how much the decline was the re-
sult of Fed intervention and how much had to do with the easing of fears
about whether the United States would maintain its link to gold in the wake of
Britain’s decision to go of the gold standard in September 1931. Tere was
much less of an impact on commercial loans and the moderately risky Baa-
grade corporate bonds. Rates on both of these were much higher in June 1932
than they had been a year earlier, and even afer these rates fell over the next
six months, they remained higher than they had been in June 1931.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 8.21 Selected Interest Rates (percent per year)
Yield on Short-Term
8
U.S. Government Yield on Long-Term
•
Prime Commercial Securities (3–6 month U.S. Government
Two-Month Averages Paper (4–6 months Rates on Commercial Treasury Notes and Bonds (due or callable Yield on Baa-Rated Yield on Aaa-Rated
(e.g., June and July) New York) Loans (avg. 19 cities) Certifcates) afer 12 years) Corporate Bonds Corporate Bonds
Table 120 Table 125 Table 122 Table 128 Table 128 Table 128
Jun 1931 2.00 4.17 0.48 3.14 7.22 4.36
Sep 1931 2.57 4.19 1.08 3.44 8.56 4.77
Dec 1931 3.88 4.77 1.82 4.10 9.78 5.26
Mar 1932 3.57 4.88 1.68 3.80 9.65 5.08
A Dose of Reality
Jun 1932 2.63 4.77 0.28 3.67 11.16 5.34
Sep 1932 2.07 4.58 0.02 3.43 7.74 4.67
Dec 1932 1.44 4.44 0.43 3.29 8.22 4.52
Source: Board of Governors of the Federal Reserve System 1943.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
299
. mathematical appendix .
In the industrial sector of a two-sector model, desired expenditure is the sum
of (1) consumption demand on the part of capitalists and workers in the in-
dustrial sector; (2) consumption demand on the part of farm families; (3) in-
vestment demand from both sectors. Te demand schedule for industrial
goods is defned by equality between income and expenditure:
PA
YI = γ II YI + γ AI YA + I ( ρ).
PI
Equivalently, the demand schedule is defned by the two-sector analog of S = I:
PA
( 1 − γ II ) YI = γ AI YA + I ( ρ).
PI
Assuming fxprice adjustment, the dynamics of the industrial sector are sum-
marized by
•
PI
W
≡ PI − W
PI PI
PI = −1 GS − YI − 3 YI − LS , (8.1)
PI W W W
W
P
Y I = θ 2 γ AI A Y A + γ IIYI + I ( ρ) − YI . (8.2)
PI
Copyright © 2021. Harvard University Press. All rights reserved.
For agriculture we have expenditure equal to income when the consump-
tion of agricultural goods by nonfarmers is equal to the saving of farm fami-
lies:
−1
P
IA A YI = ( 1 − AA ) YA . (8.3)
PI
But, unlike the industrial sector, fexprice adjustment is assumed to character-
ize agriculture. Te adjustment equations are
YA = − 4 ( YA − YAS ), (8.4)
300
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • Mathematical Appendix 301
P A PA
−1
= θ 5 γ IA YI + γ AAY A − YA . (8.5)
PA PI
Combining equation (8.5) with the price-adjustment equation
PI P
= −1 GS I − YI
PI W
gives the adjustment equation for relative prices:
•
PA
−1
PI ≡ P A − P I = θ γ PA Y + γ Y − Y
PA 5 IA I AA A A
PA PI PI (8.6)
PI
P
+ θ1 GS I − YI .
W
Te complete system is made up of equations (8.1), (8.2), (8.3), and (8.4)
plus (8.6).
At equilibrium, the variables are all stationary, and denoting
= (1 − II)(1 − AA) − IAAI ;
we have, afer substitution,
1 − γ AA
YI I, (8.7)
∆
P P
1GS I − 3 LS I = ( 1 − 3 ) YI , (8.8)
W W
−1
Copyright © 2021. Harvard University Press. All rights reserved.
γ P
YA = IA A I , (8.9)
∆ PI
YA = YAS , (8.10)
P
YA + 1 GS I − YI = YA . (8.11)
5 W
Equation (8.11) explains the relationship between the food-demand sched-
ule and the locus of stationary relative prices in Figure 8.7.
Te stability conditions for a four-equation system are
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
302 Raising Keynes
1 > 0,
2 > 0,
3 > 0,
4 > 0,
12 − 3 > 0,
3(12 − 3) − (1)24 > 0,
where the ’s are the coef
ficients of the characteristic equation
4 + 1 3 + 2 2 + 3 + 4 = 0
formed by the determinant of the Jacobian matrix
j11 j12 j13 j14
j j22 j23 j24
21 =
j31 j32 j33 j34
j41 j42 j43 j44
−θ1GS′ + θ 3 LS ′ θ1 − θ 3 0 0
PA
0 −θ 2 ( 1 − γ II ) YI θ 2 γ AI θ 2 γ AI YA
PI
0 0 −θ 4 0 .
−1 −2
P PA
θ1GS′ θ 5 γ IA A − θ1 −θ 5 ( 1 − γ AA ) −θ 5 γ IA YI
PI PI
Taking account of the zeros in det J, the coef
ficients of the characteristic
equation are
1 = −tr J = −( j11 + j22 + j33 + j44),
2 = j11 j22 + j11 j33 + j11 j44 + j22 j33 + j22 j44 + j33 j44 − j24 j42,
Copyright © 2021. Harvard University Press. All rights reserved.
3 = −( j11 j22 j33 + j11 j33 j44 + j11 j22 j44 + j22 j33 j44 − j11 j24 j42 − j33 j24 j42 + j41 j12 j24),
4 = det J = j11 j22 j33 j44 − j11 j24 j42 j33 + j41 j12 j33 j24.
Fortunately, most of the terms disappear because of the zeros in the Jacobian,
but the calculations are still formidable.
For this reason, we work with a simpler version of the model that takes the
adjustment of farm output to the supply constraint to be instantaneous. Tis
gives us the system formed by equations (8.1), (8.2), (8.3), (8.6), and (8.10).
For this system the Jacobian is
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
8 • Mathematical Appendix 303
−θ1GS′ + θ 3 LS′ θ1 − θ 3 0
j11 j12 j13
0 −θ 2 ( 1 − γ II ) YI θ 2 γ AIYA
J = j21 j22 j23 = −1 −2 .
P PA
j
31 j32 j33 θ1GS′ θ 5 γ IA A − θ1 −θ 5 γ IA
YI
PI P
I
Te stability conditions for this system are
1 = − tr J = −( j11 + j22 + j33) > 0,
2 = −( j23 j32 − j11 j22 − j11 j33 − j22 j33) > 0,
3 = −( j11 j22 j33 − j11 j23 j32 + j31 j12 j23) > 0,
1 2 − 3 = − ( j11
2 2
j22 + j11 2
j33 + j22 2
j11 + j22 2
j33 + j33 2
j22 + j33 j11 + 2 j11 j22 j33 )
+ j23 ( j22 j32 + j33 j32 + j31 j12 ) > 0
once the zeros have been eliminated.
Te signs of j12 and j32 are ambiguous, which in turn make the signs of the
terms in boldface in the stability conditions ambiguous. If j32 < 0, which is
to say,
−1
P
θ 5 γ IA A < θ1,
PI
then 2 > 0. If, in addition, either GS′ ≈ 0 or 3 > 1, then j31 j12 j23 is either very
small (if GS′ ≈ 0) or negative (if 3 > 1), so 3 > 0. Finally, the same condition,
small GS′ or positive 3 − 1, ensures 12 − 3 > 0.
It should be emphasized that these conditions on j32, j12, and j31 are suf fi
cient rather than necessary. Te system will be stable even if j32 > 0, j12 > 0, and
GS′ is large, provided the centripetal forces represented by the (negative) di-
agonal elements of the Jacobian are stronger than the centrifugal forces repre-
Copyright © 2021. Harvard University Press. All rights reserved.
sented by positive of-diagonal elements.
Finally, we have the comparative statics results associated with changes in the
supply of agricultural output. Denoting dij/d(PA/PI) by ij′, and d/d(PA/PI)
by ′, we diferentiate equations (8.7)–(8.11) to obtain at the end of the day
(actually many days),
P
γ AA ′∆ + ∆′ ( 1 − γ AA ) I A
dYI PI
= ,
dYAS PA θ1 θ 3 ( GS′ − LS′ )
∆ YA − ( γ IA ′ + ∆′γ IA ) I − ( γ AA ′ − ∆′ ( 1 − γ AA ) )
2
I
PI θ 5 θ1GS′ − θ 3 LS′
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
304 Raising Keynes
P
d A YA
PI =
dYAS
PA
∆2
PI
− .
P θ θ ( GS′ − LS′ )
∆ YA − ( γ IA ′ + ∆′γ IA ) I − ( γ AA ′ − ∆′ ( 1 − γ AA ) ) A 1 3
2
I
PI θ 5 θ1GS′ − θ 3 LS′
Assuming II′ = −IA′ > 0 and AI′ = −AA′ > 0, we have
γ AA ′∆ + ∆′ ( 1 − γ AA ) = − γ II ′ ( 1 − γ AA ) ( 1 − γ AA − γ AI ) − γ AI ′ γ IA < 0
and
γ IA ′ + ∆′γ IA = γ IA ′ ( 1 − γ AA ) + γ AA ′ γ IA [1 − γ II − γ IA ] < 0,
so that both dYI/dYAS and d([PA/PI]YA)/dYAS are negative. Te economics is that
as long as farmers and industrial workers and capitalists (1) increase their
consumption of industrial goods when the relative price of agricultural goods
rises and (2) decrease their consumption of agricultural goods by the same
amount, a decrease in agricultural production (YAS ) will lead to an increase in
industrial output (YI) and an increase in the real income of farmers ([PA/PI]YA,
measured in terms of industrial goods). Both farmers and nonfarmers in-
crease their demand for industrial goods as the terms of trade turn in favor of
agriculture, but their reasons are diferent. It is plausible that income efects
drive farmers, whereas substitution efects drive nonfarmers.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. IV .
Building Blocks
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 9 .
consumption and saving
Aggregate income . . . is, as a rule, the principal variable upon which
the consumption constituent of the aggregate demand function will
depend . . .
Te fundamental psychological law, upon which we are entitled to
depend with great confdence both a priori from our knowledge of
human nature and from the detailed facts of experience, is that men
are disposed, as a rule and on the average, to increase their consump-
tion as their income increases, but not by as much as the increase in
their income.
—john maynard keynes
For Keynes, the “fundamental psychological law” that people spend some but
not all of any increases in income was the beginning rather than the end of
wisdom on consumption. Tis “law,” however, soon morphed into highly sim
plifed models in which income was the sole determinant of consumption,
models which were bound to be refuted by experience.
Tis transformation took place even though Keynes qualifed his funda-
mental psychological law in many ways that in fact anticipated the arguments
of both friendly critics like Franco Modigliani and hostile ones like Milton
Friedman. On the one hand, he emphasized that the propensity to consume
was likely to be lower in the short run than over a longer period:
[Te tendency for consumption to change less than income] is especially the
Copyright © 2021. Harvard University Press. All rights reserved.
case where we have short periods in view, as in the case of the so-called
cyclical fuctuations of employment during which habits, as distinct from
more permanent psychological propensities, are not given time enough to
adapt themselves to changed objective circumstances. For a man’s habitual
standard of life usually has the frst claim on his income, and he is apt to save
the diference which discovers itself between his actual income and the
expense of his habitual standard; or, if he does adjust his expenditure to
changes in his income, he will over short periods do so imperfectly. Tus a
rising income will ofen be accompanied by increased saving, and a falling
income by decreased saving, on a greater scale at frst than subsequently.
(Te General Teory, p. 97)
307
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
308 Raising Keynes
In the longer run individuals would be subject to conficting infuences.
Habit would play a smaller role as people adapt their standard of life to their
incomes. So we should expect that the marginal propensity to consume will
rise as we consider longer time periods. But Keynes ofers an argument in the
other direction:
A higher absolute level of income will tend, as a rule, to widen the gap be-
tween income and consumption. For the satisfaction of the immediate pri-
mary needs of a man and his family is usually a stronger motive than the
motives towards accumulation, which only acquire efective sway when a
margin of comfort has been attained. Tese reasons will lead, as a rule, to a
greater proportion of income being saved as real income increases. (Te
General Teory, p. 97)
Milton Friedman in particular, as we saw in chapter 4, chided Keynes for
ignoring the role of wealth in determining consumption. On the contrary:
Keynes is very clear that wealth forms a bufer that insulates consumption
from variations in income, especially in the short period:
A decline in income due to a decline in the level of employment, if it goes
far, may even cause consumption to exceed income . . . by some individuals
and institutions using up the fnancial reserves which they have accumu-
lated in better times. (Te General Teory, p. 98)
But changes in wealth may reinforce the efects of income changes even as
levels of wealth have the opposite efect:
Te consumption of the wealth-owning class may be extremely susceptible
to unforeseen changes in the money-value of its wealth. Tis should be clas
sifed amongst the major factors capable of causing short-period changes in
the propensity to consume . . .
If a man is enjoying a windfall increment in the value of his capital, it is
natural that his motives towards current spending should be strengthened,
Copyright © 2021. Harvard University Press. All rights reserved.
even though in terms of income his capital is worth no more than before;
and weakened if he is suffering capital losses. (pp. 92–93, 94)
Te upshot of these qualifying observations is to change the consumption
function from the simple linear form deployed so far, illustrated in Figure 9.1.
Instead, we have two consumption functions, one for the short run and an-
other for the long run, as in Figure 9.2.
In part stimulated by the accounting framework of Keynes’s theory, econo-
mists began to compile and analyze data on consumption, saving, and in-
come. In the United States, Simon Kuznets pioneered the collection of aggre-
gate data, producing a comprehensive series of national income and product
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 309
accounts that allowed historical comparisons on both an annual basis from
1919 to 1938 (Kuznets 1941a) and on a decadal basis from the late nineteenth
century (1941b [quoted in Ezekiel 1942a, 1942b]). Mordecai Ezekiel (1942a,
1942b) analyzed the data collected by Kuznets and others over the period
1919 to 1940 and concluded that, as Keynes had suggested, over short periods
the ratio of consumption to income tended to move inversely with income,
higher at lower levels of income, lower at higher levels.
Consumption (C D )
C D cY
45°
Income (Y)
Figure 9.1 Te relation between income and consumption, I.
Consumption (C D )
CD Y
Copyright © 2021. Harvard University Press. All rights reserved.
Long-period consumption
function
Short-period consumption
function
45°
Income (Y)
Figure 9.2 Te relation between income and consumption, II.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
310 Raising Keynes
But according to Ezekiel, the data only supported Keynes in part. For one
thing, while the marginal propensity to consume was roughly constant, the
short-period consumption function on the way up in the 1920s and in the
post-1933 recovery lay below the corresponding function on the way down
between 1929 and 1933; people appear to have maintained their standards of
living in the face of an eroding economy, while rebuilding fnancially be-
fore spending on consumption as the economy recovered from a downturn
(1942a, fg. 5, p. 32).
More important, over longer periods, Kuznets’s data (1941b) did not sup-
port the idea of a secular decline in the propensity to consume, as the long-
period consumption function in Figure 9.2 implies. Rather, over the period
1879 to 1928, decadal average rates of gross capital-formation in the United
States remained remarkably constant, at about 20 percent of gross national
product. Only the decade 1929 to 1938 witnessed a signifcant change: the
capital-formation ratio fell to 15 percent (Ezekiel 1942b, table IV, p. 301). On
this basis—without taking Ezekiel’s dynamic modifcation into account—the
relationship between short-and long-period consumption functions becomes
that of Figure 9.3, with the long-run consumption function once again linear,
and going through the origin if extrapolated backward.
Te issues raised by the existence of two consumption functions were both
theoretical and practical. Indeed, Te General Teory challenged orthodox
economics before it had much impact on the economy. It is ironic that it was
in managing the economy during World War II, and even more so in the
Consumption (C D )
CD Y
Copyright © 2021. Harvard University Press. All rights reserved.
Long-period consumption
function
Short-period consumption
function
45°
Income (Y)
Figure 9.3 Te relation between income and consumption, III.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 311
planning of the postwar peacetime economy, that Keynesian economics came
into its own—ironic because of the origins of Te General Teory in a time of
deep depression.
Samuelson on the Prospects for Full-Employment
after World War II
Paul Samuelson’s take on postwar opportunities and challenges is emblematic.
Writing in 1943 he refected the widespread concern that the economy would
lapse into unemployment once the government was no longer spending vast
sums on military procurement. In other words, would the private investment
needed to match anticipated saving at full employment be forthcoming? If
not, what could be done to prevent the calamity of renewed depression?
Samuelson was quick to dismiss the possibility of a redo of the collapse
of 1929–1933. “Regardless of plans and intentions,” he wrote, “any party in
power would be forced by the mere sweep of catastrophic political events to
provide suf ficient demand to prevent this from happening” (1943, p. 28).
Rather, Samuelson’s concern was stagnation:
Te real danger lies in the possibility that we shall lag ever farther behind
our true productive potential—that we shall be content with a half loaf in-
stead of insisting upon the whole loaf which can be ours. Te thing to fear is
an ever-widening gap between our attained levels of output and employ-
ment and our true productive potential. (p. 28)
Samuelson, as was to become the norm, collapsed the LS schedule to a sin-
gle level of output compatible with “full employment,” a fgure he estimated
on the basis of the wartime economy. Using a highly simplifed Keynesian
framework, he calculated the amount of consumption demand that the econ-
omy would generate at full employment and, correspondingly, the full-
employment amount of saving. Absent any other sources of spending, full-
Copyright © 2021. Harvard University Press. All rights reserved.
employment saving tells us how much investment demand is required for the
economy to achieve its potential.
But which consumption function is relevant, the short-run or the long-
run? In answering this question, Samuelson frst makes the plausible argu-
ment that a long-run schedule like the one in Figures 9.2 and 9.3 is simply a
locus of actually observed points on successive short-run consumption func-
tions:1
an enlarged scale of wants . . . causing an upward shif in the [short-run]
consumption function at about the same rate as improvements in our pro-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
312 Raising Keynes
duction potential [and] yielding a stable relation between percentage con-
sumed out of national incomes corresponding to a given fraction of full-
employment income. (1943, p. 33)
Given the time frame, the long-run schedule may be of little relevance. It is
the short-run consumption function when the war ends that matters because
it is the immediate postwar period with the inevitable dislocations caused by
demobilization and its afermath on which Samuelson focuses.
Figure 9.4 adds an assumption of full-employment output and income
(Y = YFE) to the short-period consumption function of Figures 9.2 and 9.3.
Te expenditure required to ofset full-employment saving is given by the
vertical distance between the 45-degree line and the short-period consump-
tion function, namely, YFE − CD FE.
Samuelson poses the problem of achieving full employment in terms of a
list of potential ofsets to full-employment saving, in which he includes not
only investment but also initiatives that the government might take to in-
crease aggregate demand directly—government spending on goods and ser
vices—and indirect measures to redistribute income from people more likely
to save to people more likely to consume. Tat is, Samuelson’s ofsets include
measures to fll the gap YFE − CD FE both by mobilizing additional sources of
spending and by increasing CD FE. Samuelson concluded not only that invest-
ment was likely to prove inadequate but that all the measures together were
likely to fall short.2
For our purposes, the most important of these ofsets, the one around
which debate would coalesce, is the possibility that deferred demand, the con-
Expenditure (E D )
ED Y
YFE
Copyright © 2021. Harvard University Press. All rights reserved.
Full-employment saving
Short-period consumption
CD FE function
45°
YFE Income (Y)
Figure 9.4 Full-employment saving.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 313
sequence of the shortages of consumer goods because of the war efort, would
provide additional consumption demand beyond the normal demand repre-
sented by the consumption function. Samuelson is far from sanguine about
the quantitative boost that this will provide to ordinary consumption de-
mand. In the event, the optimists (including Keynes, despite his reservations
about the long-run prospects for self-sustaining demand) proved to have
better crystal balls than the pessimists such as Samuelson. But this matters
less than the importance that both sides attached to the role of wealth in con-
sumption; as Samuelson notes, “Te real backlog of deferred demand . . . will
be accompanied by the fnancial means to make it efective” (1943, p. 46).
In 1943 the debate could still be framed in terms of consumption expen
diture that ordinarily depends on income but may be infuenced by other
factors, including wealth and backlogs of unmet demands. Little more than
a decade later this framework itself would be challenged. Instead of being
an additional factor complementary to income, as Keynes originally argued,
wealth was to become the central determinant of consumption.
Wealth, Consumption, and Saving: The Life-Cycle and
Permanent-Income Hypotheses
Just as Keynes had charged his fellow economists with making a category
mistake about interest-rate determination, the mainstream now returned the
favor in arguing about how consumption and saving are determined. In efect,
the mainstream charged Keynes with failing to understand that rational eco-
nomic agents would allocate stocks of assets to fows of consumption, rather
than basing current consumption fows on current income fows. Te details,
as we shall see, difer, but both Franco Modigliani’s “life-cycle hypothesis”
(Modigliani and Brumberg 1954) and Milton Friedman’s “permanent-income
hypothesis” (1957), attempt to correct Keynes’s supposed mistake by assum-
ing that agents determine current consumption and saving as part of a long-
Copyright © 2021. Harvard University Press. All rights reserved.
term consumption plan. Te operative constraint is not current income but
long-term wealth.
Tis theoretical argument received a boost from the studies of family-
consumption data that were undertaken in the 1930s and the early postwar
period. (Robert Ferber [1962] summarizes this literature.) Studies based on
cross-sectional data (consumption and income data collected for diferent
families at a single period of time) gave results qualitatively similar to the re-
sults gleaned from aggregate time-series data over short periods of time,
namely, a positive increase in consumption as household income rose, but
an increase that was m odest in comparison with propensities to consume
implicit in the long-run Kuznets data. If we scale the family-consumption
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
314 Raising Keynes
data diferently from the aggregate data, we can superimpose the family-
consumption function on a long-term “Kuznets” consumption function along
which the marginal propensity to consume is constant. Te qualitative rela-
tion between the two consumption functions in Figure 9.5 is the same as in
Figure 9.2.
Working from the same general premise, Modigliani and Friedman ofer
diferent, but mutually compatible, reasons why the long-run propensity to
consume refected in the aggregate data is greater than the propensity to con-
sume refected in the family data. Both the life-cycle and the permanent-
income hypotheses start from the standard microeconomic theory of choice.
In the simplest version, there are two commodities, say (with Greg Mankiw,
Principles of Economics [2018]) pizza and cola, and the agent is a student with
a snacks budget of, say, $1,000—provided by kind parents—which she can al-
locate as she pleases between the two goods. Figure 9.6 superimposes a bud-
get constraint on a set of indiference curves.
As generations of students of introductory economics have learned, the
consumer’s optimum is where the budget constraint intersects with the high-
est attainable indiference curve, in the present case the indiference curve
marked I2. With smooth indiference curves, this optimum is characterized by
equality between the slope of the indiference curve I2 and the slope of the
budget constraint. Tese slopes refect the marginal rate of substitution of
cola for pizza (the ratio of the marginal utility of pizza to the marginal utility
of cola) and the ratio of the price of pizza to the price of cola. Optimization
Expenditure (E D )
2,000 125,000 ED Y
1,600 100,000 Long-period consumption
function according to Kuznets
Copyright © 2021. Harvard University Press. All rights reserved.
1,200 75,000
Family consumption function
800 50,000
400 25,000
45°
25,000 50,000 75,000 100,000 125,000 Family income ($)
400 800 1,200 1,600 2,000 Aggregate personal
income ($ billions)
Figure 9.5 Aggregate and family consumption functions.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 315
requires that the frst, a subjective parameter that depends on the individual’s
utility function, be brought into line with the second, an objective parameter
given by the market.
Te beauty of the theory of consumer choice is its universality. We can go
from pizza and cola to intertemporal consumption planning in a fash, simply
by relabeling the axes and reinterpreting the budget constraint and the indif-
ference curves. Figure 9.7 represents the intertemporal problem as the choice
between consumption in two periods, present and future. Te indiference
curves are now the isoquants of a utility function refecting the satisfac-
tion associated with alternative consumption patterns over time, and the
budget constraint now refects the long-term resource—wealth—available to
the agent. As before, the optimum is the point of tangency between the high-
est attainable indiference curve and the budget constraint, which refects the
adjustment of subjective marginal preferences, achieved by varying the con-
sumption pattern, to the objectively given market prices.
Te interpretation of preferences and prices changes. Te marginal rate of
substitution, given by the slope of the indiference curve, remains the relative
marginal utility of the two goods, but now these goods are present and future
consumption; the price ratio, given by the slope of the budget constraint, is
the relative price of present and future consumption. Te position of the bud-
Quantity of cola
Copyright © 2021. Harvard University Press. All rights reserved.
Consumer’s optimum
I3
I2 Indifference curves
I1
Budget constraint
Quantity of pizza
Figure 9.6 Te rational consumer.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
316 Raising Keynes
Future consumption
Consumer’s optimum
I3
I2 Indifference curves
I1
Long-term resources
Present consumption
Figure 9.7 Optimal intertemporal choice.
get constraint is given by the agent’s wealth. Its slope, the trade-of between
the two goods, is given by the interest rate, since the price of present con-
sumption relative to future consumption is one plus the market rate of inter-
est between the two periods.
An important diference between Modigliani and Friedman is the time
frame each has in mind when applying the future versus present model. For
Modigliani, the future is the lifetime of an individual agent, who is assumed
to maximize his or her utility with no thought for partner or progeny. Con-
sider the simplest case, in which the rate of interest is assumed to be zero and
Copyright © 2021. Harvard University Press. All rights reserved.
indiference curves are assumed to be symmetric around the 45-degree line
representing equal consumption, which is tantamount to assuming that the
marginal utility of consumption in each period is the same when the level of
consumption is unchanging over time. Suppose agents are supported by their
parents until age twenty and continue their studies until age twenty-fve,
when they enter the workforce. (To keep matters simple, these agents receive
full scholarships for tuition, books, and all costs associated with their studies,
but they have to cover their living costs.) Once in the paid labor force, agents’
earnings rise gradually from a beginning salary of $50,000 to a maximum of
$121,500, when their careers end in retirement at age sixty. Agents live an-
other twenty years. Nobody receives a pension; rather, everybody must save
during his or her working life to provide twenty years’ income in retirement.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 317
Income, Consumption, Saving
($ thousands)
125
100 Labor income
75 Saving
Optimal
50 consumption
25 Dissaving Dissaving
20 40 60 80 Age
Figure 9.8 Income, consumption, and saving according to the life-cycle hypothesis.
Under the assumption that the marginal utility of income is constant over
time when the level of consumption does not change, the optimum consump-
tion plan is to consume the same amount in each period. Our assumptions
about income, schooling, and retirement lead to an optimum consumption
level of $50,000, as depicted in Figure 9.8. Saving is initially negative since as
students aged between twenty and twenty-fve agents must incur debt to sup-
port the optimum consumption plan. Positive saving—frst to pay of student
loans, then to provide for retirement—takes place when an agent enters the
work force.
Modigliani’s life-cycle agents do not provide for their survivors. Indeed,
they have no motivation for saving other than to repay student debt and, later,
to provide for their own retirement. If they plan properly, they die as paupers.
We can make the model more realistic by supposing that agents pay inter-
est on student loans and earn a return on the assets they accumulate once
Copyright © 2021. Harvard University Press. All rights reserved.
these loans are paid of. Another dose of realism would be to replace the as-
sumption of an earnings trajectory known with certainty by a set of probabil-
ities with respect to earnings and interruption of earnings due to lay-ofs, ill
health, and other contingencies of work and life. We could also make life span
a probabilistic variable rather than something known with certainty.
But the main lesson of the life-cycle hypothesis is clear even in a simple
model without these complications: lifetime earnings, in this case $3 million,
are the determinant of consumption at each point of time. Te role of current
income is that it contributes to one’s lifetime earnings, not that it constrains
today’s consumption possibilities. Imagine, for example, that the agent is a
sports star, for whom lifetime income is the same but the trajectory is the op-
posite of what is pictured in Figure 9.8.3 Instead of a gradually rising income,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
318 Raising Keynes
this agent’s peak income takes place the moment she enters the labor force,
and declines gradually until retirement, as in Figure 9.9. If she has the same
utility function as the agent depicted in Figure 9.8, she will choose the same
consumption profle!
If we now plot either agent’s consumption function against income, and
superimpose this consumption function on an aggregate consumption func-
tion, we have the result depicted in Figure 9.10. Evidently, this is an extreme
version of Figure 9.5, for here the family consumption function is fat, a result
due, among other things, to the rather extreme assumption of a zero interest
rate.4 According to Modigliani’s life-cycle hypothesis, the consumption func-
tion based on household budgets refects the age distribution of the popula-
tion, with the young and the elderly typically consuming more than they earn,
and the middle-aged less.
In the life-cycle model, life-time saving is zero. How then is it possible for
aggregate saving to be positive, as the long-run consumption function sug-
gests? Tere are two possible sources of saving in this model, in the frst place,
population growth. If there are more thirty-somethings who are earning and
saving than are necessary to balance the retired folk who are spending down
their previous savings, net saving will be positive. (Te dissaving of young
adults is a complication, but one that doesn’t change the basic picture.) Aggre-
gate saving will be a weighted sum of individual saving distributed along the
individual consumption function, with greater weight on savers than on dis-
savers simply because there are more savers when the population is growing.
Te second path to positive saving in the life-cycle model is technological
progress. Whatever we assume about how wages are determined, it is reason-
able to assume that technological progress raises lifetime earnings. In this
Income, Consumption, Saving
($ thousands)
Copyright © 2021. Harvard University Press. All rights reserved.
125
100 Labor income
75 Saving
Optimal
50 consumption
25 Dissaving Dissaving
20 40 60 80 Age
Figure 9.9 A sport star’s optimum—according to the life-cycle hypothesis.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 319
case, each cohort has a diferent consumption function, as in Figure 9.11, and
each cohort saves accordingly.
Te class of 1955 enters the paid work force in 1955 (at age twenty-fve),
and the class of 1975 does the same thing in 1975. Assume both cohorts are
Expenditure (E D )
2,000 125,000 ED Y
1,600 100,000
1,200 75,000 Long-period consumption
function according to Kuznets
800 50,000 Family consumption function
Age 70 Age 25 Age 60
Age 20
400 25,000
45°
25,000 50,000 75,000 100,000 125,000 Family income ($)
400 800 1,200 1,600 2,000 Aggregate personal income
($ billions)
Figure 9.10 Aggregate and family consumption functions according to the life-cycle
hypothesis, I.
Expenditure (E D )
ED Y
2,000 125,000
Long-period consumption
function according to Kuznets
1,600 100,000
Age 70
Copyright © 2021. Harvard University Press. All rights reserved.
Age 20 Age 25 Age 60
1,200 75,000 Class of ’75
Family consumption functions
800 50,000 Class of ’55
Age 70 Age 25 Age 60
Age 20
400 25,000
45°
25,000 50,000 75,000 100,000 125,000 Family income ($)
400 800 1,200 1,600 2,000 Aggregate personal income
($ billions)
Figure 9.11 Aggregate and family consumption functions according to the life-cycle
hypothesis, II.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
320 Raising Keynes
the same size. In the year 2000, the positive saving of the class of 1975, now
ffy years old, is based on a higher level of life-time income and will swamp
the dissaving of the retirees of the class of 1955, now age seventy. Without
technological progress (or population growth) and higher life-time earnings,
earnings at age ffy would, as Figure 9.8 indicates, yield just enough saving to
provide for about one year’s retirement.
Milton Friedman’s permanent-income hypothesis emphasizes a diferent
aspect of lifetime optimization, namely, the diference in the efects on con-
sumption of changes in income that are expected to be permanent from
the efects of transitory changes, changes that are expected to be temporary.
Friedman (1957) is deliberately vague about the time frame that distinguishes
“permanent” from “transitory,” but we can illustrate his argument within the
life-cycle framework.
What happens to the consumption of the agent whose earnings profle is
given in Figures 9.8 or 9.9 if she is laid of for a year at age forty-two?5 Con-
sumption will take a hit, but only a modest one. If she anticipates the loss, her
lost income will be spread out over her entire adult life, so that 1/60 of the
loss, approximately $1,500 ($85,714/60 = $1,429) will be felt each year. In
Friedman’s world, marginal propensities to consume when changes in income
are transitory are determined by the ratio of the length of the transitory time
period to the length of one’s consuming life. In the present case, the one-year
marginal propensity to consume is 1/60 when the loss is anticipated. Even If
the loss is unanticipated, its efects will still be spread out over time. Te opti-
mal response to an unanticipated shortfall (or for that matter windfall) is to
spread the efects out over one’s remaining life span, in this case 37.5 years.
Accordingly, the marginal propensity to consume will be 1/37.5.
In Modigliani’s telling, the long-period propensity, the propensity to con-
sume over one’s whole lifetime, will be one. Friedman, however, is not com-
mitted to a lifetime horizon, and, accordingly, the permanent-income hy-
pothesis leaves room for a long-run propensity to consume that is less than
Copyright © 2021. Harvard University Press. All rights reserved.
one; we can imagine saving taking place to provide for one’s heirs, or to build
a monument to oneself, or because of the inability to predict when one will
die—if an agent is risk averse, she will err on the side of over-providing for
her retirement.
In the permanent-income hypothesis, the long-run marginal propensity to
consume is relevant for evaluating the impact of short-period changes in in-
come if an unanticipated change in current income leads the agent to change
her beliefs about the trajectory her income will follow in the future. If being
out of work today leads her to believe she will suffer a permanent decline in
income, the ratio of the change in current consumption to the change in cur-
rent income—the short-run propensity to consume—will increase. If an agent
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 321
receives an unanticipated increase in income that carries with it the expecta-
tion of permanence, as distinct from a windfall in the form of a one-time bo-
nus, optimal consumption planning will lead the agent to spend according to
her new evaluation of her long-term resources: her increased spending will be
in line with her long-period propensity to consume, not with the short-period
propensity associated with transitory income changes.
To see the rationale of distinguishing between permanent and transitory
income, assume that fve individuals are twenty-fve years old and all initially
expect to have the income trajectory depicted in Figure 9.8. Anna receives a
jolt in the form of a one-year postponement of her starting date and takes a
temporary job at half her expected starting salary, $25,000 instead of $50,000.
Betsey has to settle for a lower-paying job in her chosen feld, with a starting
salary of $40,000, and her long-term income prospects fall proportionately.
Carrie is luckier: she receives a one-time hiring bonus of $25,000. Delia is
luckier still, starting her career at $75,000, which leads her to project higher
incomes over her entire working life. Only Ellie is unafected, remaining on
the initial trajectory while all her friends are jostled by the winds of fortune.
Te position of each woman, in terms of her relationship between consump-
tion and current income, is indicated by her initial in Figure 9.12. Anna
and Carrie are on a consumption function refecting transitory departures
from the average (Ellie). Te consumption function along which Betsey and
Delia fnd themselves, by contrast, refects changes in permanent income. Te
transitory-income consumption function evidently corresponds to the indi-
vidual consumption function in the life-cycle analysis; the permanent-income
consumption function corresponds to the aggregate function.
Expenditure (E D )
ED Y
125,000
Permanent-income
Copyright © 2021. Harvard University Press. All rights reserved.
100,000 consumption function
75,000
D Transitory-income
50,000
E consumption function
A
C
25,000
B
45°
25,000 50,000 75,000 100,000 125,000 Family income ($)
Figure 9.12 Responses to changes in permanent and transitory income.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
322 Raising Keynes
As far as the aggregate relationship between consumption and income is
concerned, individual good and bad luck cancels out, at least so it is assumed.
If there are 160 million workers, the equilibrium E will also refect the rela-
tionship between aggregate consumption and aggregate income, measured
along the lower scale. For aggregate consumption and income to shif upward
over time, we once again have to appeal either to population growth or to
technological progress (or both). In this case, the permanent-income con-
sumption function at the individual level coincides with the aggregate-
consumption function, with the proviso that the two functions in Figure 9.13
are scaled diferently.
Te similarities of the two versions of optimal consumption trajectories are
more important than the diferences. Te fundamental assumption is that
consumption today (and every day) is based on optimization of long-term
utility subject to a long-term resource constraint. Tis means that current
income plays little to no role in determining current consumption, unless
changes in current income are taken as harbingers of future changes. Tat is
the key takeaway.
In any case, there is no fundamental incompatibility between the life-cycle
and the permanent-income hypotheses. One can believe that the most impor
tant reason for saving is to allow for more or less constant consumption even
though income varies sharply over one’s lifetime, as the life-cycle hypothesis
stresses, saving taking place to provide for a comfortable retirement (or to pay
Expenditure (E D ) Permanent-income
ED Y
consumption function
2,000 125,000
Long-period consumption
1,600 100,000 function according to Kuznets
1,200 75,000
Copyright © 2021. Harvard University Press. All rights reserved.
Transitory-income
800 50,000
consumption function
400 25,000
45°
25,000 50,000 75,000 100,000 125,000 Family income ($)
400 800 1,200 1,600 2,000 Aggregate personal income
($ billions)
Figure 9.13 Aggregate and family consumption functions according to the permanent-income
hypothesis.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 323
of student debt). One can at the same time believe that saving is a shock ab-
sorber against unfavorable but transitory hits to income and a rainy-day fund
when income gets a temporary boost, which is what the permanent-income
hypothesis stresses.
What Accounts for the Triumph of Life-Cycle and
Permanent-Income Thinking?
Te two hypotheses, permanent income and life cycle, quickly carried the day
as far as the economics profession was concerned. And while both have re-
ceived considerable refnement over the years,6 the original idea that wealth,
not income, drives consumption has not been substantially modifed; it re-
mains the basis of mainstream thinking about the consumption function.
One might imagine that the life-cycle and permanent-income hypotheses
supplanted the simple Keynesian consumption function because these theo-
ries were better at explaining the data. Tis is only partly true. Te common
core of these two theories is that they support the phenomenon of consump-
tion smoothing. But other theories that lead to consumption smoothing have
fallen by the wayside or never gained traction in the frst place. Keynes him-
self ofered reasons why consumption would change less rapidly in the im-
mediate afermath of an income change and only gradually catch up: in the
excerpt from Te General Teory quoted on the frst page of this chapter,
Keynes says, “A man’s habitual standard of life usually has the frst claim on
his income.”7 Paul Samuelson echoes Keynes’s observation in arguing that
some time is required [for consumption] to become adjusted to increased
levels of income so that in the short run consumption increases less with
increased income than it does in the long run, saving taking up the slack.
Moreover, when income drops, consumption is maintained at the expense of
savings. According to this . . . point of view, the short-run marginal propen-
sity to consume is less than the long-run marginal propensity to consume.
Copyright © 2021. Harvard University Press. All rights reserved.
(1943, p. 34)
Some years ago I tried to formalize this idea of delayed adjustment in what
I called the disequilibrium hypothesis (Marglin 1984, chaps. 17–18; I guess
I’ve always been attracted to disequilibrium). My formalization ofered a the-
oretical argument for the relatively low propensities to consume implicit in
family-budget studies—higher-income agents are more likely to be those who
are in the process of adjusting to a change for the better in their economic sta
tus, and lower-income agents are more likely to be in the process of adjusting
downward.
Tis theory accounted econometrically for the aggregate data at least as
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
324 Raising Keynes
well as the life-cycle/permanent-income approach did. No surprise here be-
cause the very diferent theoretical bases of the two approaches are muted
once the theories are formulated in a manner suitable for econometric com-
parisons; approaching the data from the point of view of disequilibrium
adjustment and approaching the data on the basis of lifetime earnings or
permanent income both lead to estimating equations that are consistent with
observed parameter values even while difering radically on how the param-
eters are understood (see chapter 15). Neither was it surprising that no weak-
ening of support for Modigliani’s and Friedman’s formulations followed from
my eforts or the eforts of others working the same street before me, Tillman
Brown (1952) and Hendrik Houthakker and Lester Taylor (1966), to mention
only two studies.
Te same fate befell theories that started life with more promise by virtue of
appearing before there was a life-cycle or permanent-income hypothesis with
which to compete. James Duesenberry’s relative-income hypothesis (1949) is
a case in point. Duesenberry explained the apparent contradiction between
cross-sectional (family-budget) data and long time-series (aggregate) data by
the role of relative income in determining the proportion of income that
agents consume. Te pressures to spend on lower-income households are
greater not only because of absolute needs that exist independently of what
anybody else might be consuming but also because richer individuals deter-
mine a standard of life that the relatively poor a dopt for the sake of personal
dignity. Richer folk, with fewer or no Joneses to keep up with, consume a rela-
tively smaller part of their income.
Duesenberry’s ideas took hold when he frst published them but were soon
eclipsed by Modigliani’s and Friedman’s. Duesenberry’s theory has resurfaced
from time to time, for example, in the work of Robert Frank (1999) and Juliet
Schor (1998), but it has not challenged the hold of life-cycle and permanent-
income reasoning on economists’ thinking. As with the disequilibrium hy-
pothesis, I think it’s fair to say that the relative failure of relative income is not
Copyright © 2021. Harvard University Press. All rights reserved.
due to the econometrics.
What then does account for the success of the life-cycle and permanent-
income hypotheses? I would argue that the main reason is the ft between
these theories and the basic premises of mainstream economics, in particular,
the key role of utility maximization subject to a resource constraint as the
basis for all action. Te attractiveness of these two theories is because of the
aid and comfort they ofer to the project of assimilating Keynes to main-
stream economics and stripping Te General Teory of its most subversive
aspects.
From this perspective, empirical ft, while hardly irrelevant, is not decisive.
Indeed, the theoretical apparatus of mainstream economics, in particular the
pride of place accorded to the rational, utility-maximizing, agent, is not due
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 325
to the role of this framework in predicting or accounting for behavior. In the
context of so-called positive or descriptive economics, rationality is not so
much an incorrect assumption as an irrelevant one. Te relevant information
about the demand for apples is contained in the demand schedule itself, not
in the utility maximization that is supposed to underlie the demand schedule.
Utility maximization adds very little empirically, especially once we move
from an individual agent to total demand.
Te apparatus exists rather to justify a particular, might I say peculiar, pen-
chant of mainstream economics: the separation of ef ficiency from distribu-
tion and the emphasis on ef ficiency—distributional issues being regarded as
political, philosophical, and thus distinguished from the “hard science” of
economic ef ficiency. More specifcally, the point is to justify capitalism as a
self-regulating economy that maximizes ef ficiency when lef to itself—the
very argument that Keynes set out to demolish in Te General Teory.
I mentioned earlier that the universality of the framework of constrained
maximization is part of its appeal. In fact, this universality is an illusion. Te
model is plausible for pizza and cola because the budget constraint is well-
defned—$1,000 per year, thanks to Mom and Dad—and the utility function
can be assumed to exist on the basis of early trial and error before the history
of rational choice begins.
Neither of these assumptions carries over to the context of long-term con-
sumption planning. Consider the budget constraint, which in the life-cycle
version of the story is an agent’s lifetime earnings. Even if the agent starts her
working life on a well-defned career ladder, how sturdy is the ladder? How
high does it go? What if she gets stuck? Or falls of? Layofs, illness, any num-
ber of factors can interfere with the steady progress assumed in Figure 9.8, or
the steady decline in Figure 9.9.
Even if a lifetime budget constraint can be assumed, what about the indif-
ference curves? George Akerlof has argued (2007) that the utility function
consumers are supposed to maximize leaves out an important consideration,
Copyright © 2021. Harvard University Press. All rights reserved.
namely, norms of behavior. People save because it is the right thing to do, not
because they have maximized a utility function whose arguments are con-
sumption at diferent points of time. And people consume for the same rea-
son. Adam Smith noted long ago that consumption follows social norms:
By necessaries I understand, not only the commodities which were indis-
pensably necessary for the support of life, but whatever the custom of the
country renders it indecent for creditable people, even of the lowest order, to
be without. A linen shirt, for example, is, strictly speaking, not a necessary
of life. Te Greeks and the Romans lived, I suppose, very comfortably,
though they had no linen. But in the present times, through the greater part
of Europe, a creditable day-labourer would be ashamed to appear in public
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
326 Raising Keynes
without a linen shirt, the want of which would be supposed to denote that
disgraceful degree of poverty, which, it is presumed, no body can well fall
into without extreme bad conduct. (1937 [1776], pp. 821–822)
I agree with Akerlof (and Smith) on the importance of norms but doubt
that it is illuminating to incorporate norms into the utility function. Tis
smacks of adding epicycles onto the Ptolemaic system. Yes, one can account
for the phenomena by means of epicycles, but the fundamental thinking in
terms of utility maximization is unafected. As I argue in Te Dismal Science
(2008, pp. 71–73), an obligation to feed the hungry is fundamentally diferent
from the warm glow you get from the look on a child’s face afer you have
given her an ice-cream cone. Similarly, an obligation to save is diferent from
the maximization of the utility of lifetime consumption.
Additionally, incorporating norms into the utility function reinforces the
view that the utility function is a primitive of the argument, a given. Tis begs
an important question. In the pizza-cola context, it is at least plausible to as-
sume a prehistory of trial and error through which agents learn the contours
of their utility functions. But what prehistory is available in the context of a
lifetime consumption plan? Even believers in reincarnation do not generally
believe that we can recall much from our past lives. Norms arguably fll the
vacuum.
If norms are internalized cultural presuppositions, there is an obvious link
to the social conditioning under which these norms are formed—and an im
portant link from the budget constraint to the utility function. People who
learn from experience that foresight pays of are more likely to believe in the
virtue of the economist’s formulation of consumption choice and to inculcate
the virtue of constrained optimization in their ofspring. People who learn
from experience—the experience of parents, friends, neighbors, as well as
their own—that they have little control over their own lives may react by try-
ing obsessively to plan the areas of their lives where they do feel in control.
More likely, in my view, they will react, as the origin myth of a weaving caste
Copyright © 2021. Harvard University Press. All rights reserved.
in the Indian state of Orissa suggests,
Niti araji, niti khāibi (I will earn every day, I will eat every day)
Jalāgud i bate, bela cāhinbi (I will see the passage of time through the window)
(quoted in Marglin 1990, p. 272)
Diferent Strokes for Diferent Folks
Some of the problems articulated here can in theory be remedied, I have sug-
gested, by recasting the framework in terms of probabilities. For instance,
instead of lifetime earnings of $3 million, as in Figures 9.8 and 9.9, the indi-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 327
vidual might think in terms of a one-third probability of earning $2 million, a
one-third probability of earning $3 million, and a one-third probability of
earning $4 million.
But to think that recasting choice in probabilistic terms can rescue main-
stream consumption theory is an illusion. Can people think in probabilis-
tic terms when the uncertainties are enormous and the information sparse?
More to the point, do people think in these terms? Te salient diference be-
tween risk and uncertainty, brought to the fore by Frank Knight (1921), is
precisely the diference between situations where the agent has enough infor-
mation to think in probabilistic terms, Knight’s risk, and situations more akin
to asking the chances of surviving Niagara Falls in a barrel, uncertainty.8
No doubt some people do think in probabilities. People whose career paths
are reasonably certain, for whom the threats of unemployment and illness are
other people’s problems (until these threats materialize in their own lives),
people who can face the world with the confdence that comes from actually
being in control of their own lives (or at least have the illusion of being in
control). Duesenberry once quipped that the life-cycle hypothesis is exactly
what one would expect from a middle-aged college professor, thus showing
that some people’s quips are as profound as other people’s theories.
In point of fact, it is not only college professors who have a reasonably good
idea of their future earnings and may have enough experience to weigh alter-
native lifetime consumption plans against one another—good enough that
the constrained maximization framework of Figure 9.7 makes sense as a way
of characterizing choice. Other professional w omen (and men), even those
lacking formal tenure of the kind enjoyed by academics (or at least some of
us), might also qualify.
But people lacking employment security and ofen lacking good health and
the means to pay for prolonged illness might easily fnd rational consumption
planning to be a bad joke, especially because their whole lives teach them that
they have very limited control over how much they will earn over their life-
Copyright © 2021. Harvard University Press. All rights reserved.
times and, indeed, how much they will earn even over shorter periods that
make the distinction between permanent and transitory income meaningful.
And then there are the very rich, the 1 percent upon whom the Occupy
movement and the 2016 Democratic primary campaign of Bernie Sanders
focused public attention. It is really a stretch to imagine the very rich plan-
ning consumption and saving in terms of trade-ofs between the future and
the present.9 Once again, a Duesenberry story is apposite. Te future Aga
Khan once came up to the lectern afer a class in which Duesenberry had
presented the standard mainstream theory of consumer choice. “Sir,” the
young prince is reported to have asked, “that was very interesting. But how
does it work without the budget constraint?”
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
328 Raising Keynes
A reasonable conclusion is diferent strokes for diferent folks: it is a mis-
take to look for a single explanation of consumption and saving. In the mid-
dle, maybe 10 to 25 percent of the population, is a professional class for which
the standard mainstream choice model and its life- cycle or permanent-
income corollary are a realistic basis on which to understand consumption
and saving decisions. Below them on the socioeconomic scale are the vast
majority of the population for whom the standard model makes no sense and
for whom a simple disequilibrium-adjustment story seems to me to capture
the process by which consumption is adjusted to income. For this, the work-
ing class, the simple Keynesian short-period consumption function, coupled
with a long-term function à la Kuznets, is pretty much all we need, especially
since working-class households possess relatively little wealth. At the top of
the pyramid is a small class—the proverbial 1 percent. If we are honest, we
must admit we know practically nothing about what motivates their con-
sumption and saving decisions.
In the United States, I would suggest a cut-of between the working class
and the professional middle class somewhere around $100,000 of family in-
come. According to Emmanuel Saez (2014, slide 7), this cut-of was at ap-
proximately the 90th percentile of the income distribution in 2012; for that
year, the Bureau of the Census (2019b, table A-2, p. 26) puts the $100,000
cut-of at approximately the 75th percentile. Saez calculates the income share
of the segment of the population between the 90th and 99th percentiles at just
under 30 percent, and the share of the top 1 percent at just over 20 percent,
with an income foor of $400,000. If we use Saez’s fgure for the top-income
share, the Bureau of the Census (2014, table 2, p. 9) data makes the segment
between the 75th and 99th percentiles approximately 40 percent of overall
income.10 So the portion of income to which life-cycle/permanent-income
logic might apply would be somewhere in the neighborhood of 30 to 40 per-
cent. Te rest of the population, those with incomes below $100,000 and
those with incomes in excess of $400,000, obey diferent logics. Some 40 to 50
Copyright © 2021. Harvard University Press. All rights reserved.
percent of income accrues to people below the $100,000 threshold, for whom
the precariousness of their economic position imposes a diferent, Keynesian,
rationality. Some 20+ percent of income accrues to the top 1 percent, whose
afuence also mocks the rationality of the middle class.
Te mainstream recognizes that one size does not ft all and that a large
part of the population does not act according to any reasonable facsimile of
life-cycle or permanent-income models. John Campbell and Greg Mankiw
(1989) estimate the proportion of consumption based on wealth constraints
and the proportion of consumption based on “rules of thumb,” in which in-
come rather than wealth is the driving force. Tey fnd the split to be roughly
50–50.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 329
Te standard explanation is that these households are “liquidity con-
strained,” meaning that they are unable to access their wealth and convert it
into consumption. Such households are perforce limited in their spending by
current income and respond to income changes à la Keynes. Twenty-some
things, for example, might wish to draw on future labor income to smooth
consumption in the manner of Figures 9.8 and 9.9. If indentured servitude
were allowed, future labor income might be the basis for such loans; however,
lacking acceptable collateral, these households may not be able to borrow in
accordance with their optimal consumption plans.11 In short, institutional
constraints against slavery—a market imperfection if you will—may make it
impossible to follow “optimal” consumption plans.
For the mainstream, it is acceptable to argue that agents would like to fol-
low optimal consumption plans but are frustrated by government interven-
tion that makes slavery illegal or tapping into equity dif ficult. Te important
point is that agents are trying to be good economic men and w omen. It is less
acceptable to argue that agents aren’t even trying to be rational, as rational is
defned in the economic canon.
Tere is some hope: behavioral economics has made it possible to think the
unthinkable and still be an economist in good standing. But the pace of
change is glacial. Behavioral economics may be a big plus in describing eco-
nomic reality, but it undermines the normative wing of economics, the wing
that goes beyond trying to tell it like it is to make arguments that what is is the
best of all possible worlds—or would be were it not for frictions and imper-
fections that can be addressed only by making the world more like the com-
petitive model.
To be sure, what is at stake is not so much the specifcs of the consumption
function, but the larger issue of how to do macroeconomics. First of all, it
doesn’t matter very much empirically if we base the consumption function on
the inability of consumers to access capital markets rather than on a totally
diferent mindset from that implied by the utility-maximizing framework.
Copyright © 2021. Harvard University Press. All rights reserved.
Both “imperfections” are undoubtedly at play and reinforce each other rather
than acting at cross purposes.
Even to the extent that Modigliani and Friedman got it right, the conse-
quence is merely to reduce the short-term marginal propensity to consume,
and thus the multiplier, relative to what might be expected in a Keynesian
framework. But this d oesn’t matter all that much, frst because there are more
important reasons why multipliers that make sense in the simple Keynesian
framework are likely to exaggerate actual multipliers once we factor in salient
real-world features of the economy (see chapter 15). Second, the size of the
multiplier indicates only the extent to which an initial shock to the economy
or a government induced countershock is amplifed by the internal workings
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
330 Raising Keynes
of the economy. Te logic of Te General Teory does not depend on the size
of the multiplier even if the size of the multiplier is critical in evaluating the
efects of monetary and fscal policy.
The Cambridge Saving Theory
Whereas class diferences with respect to consumption propensities appeared
to Keynes (and to Samuelson) the natural way to explain the cross-sectional
evidence that the ratio of consumption to income declines with income, a key
result of both life-cycle and permanent-income reasoning is that no such in-
ference should be drawn from these cross-sectional studies. Class may deter-
mine the absolute amount of saving, but not the proportion of income saved.
Redistribution of income from the rich to the poor (or vice versa) will not
have any impact on aggregate demand! For Modigliani, higher savers are or-
dinary people nearer the peak of their earning power. Tey are not necessarily
richer than anybody else in terms of their lifetime command over resources.
For Friedman, higher savers may simply be people enjoying temporary wind-
falls, for example, farmers who reap bountiful crops while agricultural prices
are high. Te lean years will follow, just as surely as they did for Pharaoh in
the land of Egypt.
An older view of consumption and saving assumed that working people
simply lacked the economic capacity to save and that the rich were responsible
for whatever saving the community was able to muster. In the nineteenth cen-
tury, economists were not reluctant to associate the distribution of income-
with class: capitalists saved a substantial share of their profts, while workers,
with meager wages, had all they could do to keep body and soul together. In
the book that transformed him from middle-level civil servant to world-
renowned public intellectual, Te Economic Consequences of the Peace, Keynes
himself made a class-based argument about saving, attributing the remarkable
growth of capital and income between 1870 and World War I to a tacit social
Copyright © 2021. Harvard University Press. All rights reserved.
bargain that allowed capitalists to call a substantial part of the national prod-
uct their own—on condition that they plowed it back into capital formation.
Te class-based view of saving was revived by Cambridge economists—
Nicholas Kaldor (1966), Luigi Pasinetti (1962), and Joan Robinson (1962,
1966)—following the lead of Michał Kalecki (1971 [1933]), whom, as noted
on p. 235, some also credit with anticipating the essentials of Te General
Teory. In the simplest form of the “Cambridge saving theory,” saving is pro-
portional to total proft, S = s, s representing capitalists’ propensity to
save and total profts. Saving is now proportional to the rate of proft r since
= rK, and in the short run K is given:
S = srK.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 331
Te AD schedule is now defned by
I( ) = srK.
We have
−1
P
rK = Y − L,
W
and the equilibrium in r × I space that defnes the AD schedule in the second
quadrant of Figure 9.14 is unique. But in the space of I × Y, for any given real
price (P/W) there are in general two points at which saving and investment
are equal. Observe that all the equilibria in the frst quadrant correspond to
the same rate of proft and the same rate of saving and investment: Y1 and Y4
are points where desired saving and investment are equal at the price level
(P/W)2, Y2 and Y3 with (P/W) = (P/W)1. Multiple equilibria arise because at a
given price level profts and therefore saving rise as long as the marginal pro-
ductivity of labor exceeds the real wage, and then fall, while (in this simple
case) investment demand is constant.
Figure 9.15 shows the implications for aggregate demand. For relatively
high real prices, like (P/W)2 and (P/W)1, the two intersections of the desired-
saving schedule and the desired-investment schedule in Figure 9.15(a) trans-
late into two points on the AD schedule in Figure 9.15(b). At a low price, like
(P/W)0, there is no level of output that generates p roft suf ficient to balance
the desired level of investment.
Observe that the GS schedule goes through the minimum point of the AD
schedule. Tis is because, for any given real price, maximum saving corre-
sponds to the point of p roft maximization, which in turn defnes the GS
schedule. At the minimum of the AD schedule, there is only one level of out-
put—namely, the proft-maximizing level—that generates enough p roft and
saving to cover the given amount of desired investment.
Copyright © 2021. Harvard University Press. All rights reserved.
Class-based saving requires us to revisit the analysis of equilibrium. With
fexprice adjustment, the story changes very little. In Figure 9.16, the logic of
the equilibrium at E is the same as the logic laid out in chapter 6. Te arrows
indicate disequilibrium trajectories of real price and output.
Te story changes somewhat more in a fxprice regime. Tere are now, as in
Figure 9.17, two equilibria. But only the equilibrium at E is stable. If we start
to the right of E′, investment demand exceeds saving, and the economy is
pushed further to the right. If we start to the lef of E′, saving exceeds invest-
ment, and the economy moves to the lef.
Tere is also the possibility of an infationary equilibrium, in which pres-
sure in the labor market drives wages up, and producers fnd themselves in
the happy position of too much demand, to which they respond directly (the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
332
ID , SD ID , SD
1
P
SD s Y L
W 2
SD s rK 1
P
SD s Y L
W 1
ID I( ) ID I( )
Raising Keynes
1
P
SD s Y L
W 0
r Y1 Y2 Y3 Y4 Y
Figure 9.14 Aggregate demand determined by propensities to save and invest.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 333
(a) Investment and saving holding real-price constant
ID , SD
1
P
S D S Y L
W 2
1
P
S D S Y L
W 1
I( )
1
P
S D S Y L
W 0
Y1 Y2 Y3 Y4 Y
P
W
Supply of goods
P Aggregate
W 2 S D ID demand
P
W 1
P
W 0
ID S D
Copyright © 2021. Harvard University Press. All rights reserved.
Y1 Y2 Y3 Y4 Y
(b) Aggregate demand
Figure 9.15 Aggregate demand.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
334 Raising Keynes
fexprice case) or indirectly (under a fxprice regime) by raising prices. Te
fexprice equilibrium in Figure 9.18 is stable regardless of adjustment speeds,
whereas the fxprice equilibrium E in Figure 9.19 is stable only if prices and
wages adjust faster than output. As in Figure 9.17, E′ in Figure 9.19 is unsta-
ble. North-and southeast of E′, the arrows take the economy to the east, fur-
ther and further from E′. North-and southwest of E′, the economy moves
toward E. (See the mathematical appendix to this chapter for proofs.)12
P
W
Supply of goods
Stationary real price Y 0
P 0 Aggregate demand
W P 0
E
W 0
Supply of labor
Figure 9.16 Equilibrium with saving determined by the Cambridge
saving theory: fexprice adjustment.
P
W
Supply of goods
P 0 Stationary
real price
E
P 0
Copyright © 2021. Harvard University Press. All rights reserved.
W
E
Aggregate demand
Y 0
W 0
Supply of labor
Figure 9.17 Equilibrium with saving determined by the Cambridge
saving theory: fxprice adjustment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 335
In more general forms of the Cambridge saving theory, both capitalists and
workers save, but their propensities to save are diferent, capitalists saving the
fraction s, and workers saving the fraction sw, with s > sw. Afer some years
of oblivion, this more general model may once again become relevant as a
result of Tomas Piketty’s (and his less well-known colleagues’) investigations
into inequality and its causes (Piketty 2014): Piketty’s argument that the long-
run tendency of capitalism is to produce greater and greater inequality is
P Supply of goods
W Y 0
Aggregate demand
E P 0
Stationary real price
P 0
W
W 0
Supply of labor
Figure 9.18 Infationary equilibrium with saving determined by
the Cambridge saving theory: fexprice adjustment.
P
W Supply of goods
P 0
E
Copyright © 2021. Harvard University Press. All rights reserved.
E Aggregate demand
Stationary real price Y 0
P 0
W
W 0
Supply of labor
Figure 9.19 Infationary equilibrium with saving determined by
the Cambridge saving theory: fxprice adjustment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
336 Raising Keynes
made in the context of a two-class model in which there is saving both from
capital income and labor income, that is, from p
rofts and wages.13
I address this question in the mathematical appendix to chapter 18. Te
empirical richness of Piketty’s Capital in the Twenty-First Century is well
worth the price of admission. But Piketty’s claims about the relationship be-
tween growth and distribution should be the beginning rather than the end of
the discussion about inequality.
Conclusions
Wealth matters in the determination of consumption. Nobody disputes that.
But to replace income by wealth as the most important determinant of con-
sumption is, to say the least, an overreaction to the oversimplifcation of
Keynes’s argument in Te General Teory. Even the mainstream is on board
with this conclusion, although its path to a realistic assessment is made tor-
turous by its insistence on liquidity constraints as the primary reason for in-
come to play the primary role in the determination of consumption à la
Keynes. Tis is a self-inficted torture, one made necessary by the main-
stream’s allegiance to the utility-maximizing framework of the rational con-
sumer, an allegiance in turn justifed by the commitment to the idea that
markets are ef ficient.
Te path of wisdom is to acknowledge that one size does not ft all. A
wealth-centered theory combining insights of the permanent-income and
life-cycle hypotheses may be a good basis for understanding the consumption
behavior of a minority that we can characterize as middle-class profession-
als—doctors, lawyers, academics, and the middle range of managers and
technocrats in both the private and the public sectors.
Te vast majority, what was once called the working class, is better charac-
terized by a much simpler model, in which liquidity constraints play a role,
but a supporting role. Tis class is driven by the uncertainty that surrounds
Copyright © 2021. Harvard University Press. All rights reserved.
both the intertemporal budget constraint and the intertemporal utility func-
tion, which makes it highly irrational to allow life-cycle or permanent-income
considerations to dominate decision making. Instead, the majority operates
according to simple rules of thumb that make income rather than wealth the
most important variable driving consumption.
And then there are the rich and the super rich, the 1 percent. Te fact is
that we know almost nothing about what drives the consumption and saving
decisions of this element of society. Tis gap in our knowledge would matter
little were it not for the disproportionate amount of saving done by the 1 per-
cent.
Te one conclusion of the permanent-income and life-cycle hypotheses
that can be most frmly rejected is the proposition that the long-run saving
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Consumption and Saving 337
propensity is constant over the entire population, that rich and poor save the
same proportions of their permanent or lifetime incomes. Te contrary view
is the basis of the Cambridge saving theory, which revives the nineteenth-
century idea that saving comes disproportionately from the income of the
rich, in a capitalist world from profts. For navigating the complexities of sav-
ing in a capitalist economy, the two-class model is a better vehicle than the
mainstream vehicle of the representative consumer. Te meaning of the work-
ing class in the context of two-class models may be diferent from its classical
meaning in the works of Smith, Ricardo, and Marx, but the basic structure of
the models does not change. Tanks to Tomas Piketty and his collaborators
these models may come back into favor.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Te new element in the story is the Cambridge saving theory. We now have
fexprice adjustment characterized by the equations
•
P
W
=θ
−1
P P
1 I ( ρ) − s
Y − L − θ 3 Y − LS , (9.1)
P π
W W
W
P
Y = 2 − FL −1 . (9.2)
W
Te Jacobian is
P
−1
−2 W
−θ s P L + θ LS′ −θ1s π 1 − −θ
J = 1 π W 3
FL
3
,
P
θ2 −θ 2 (GS′)−1
W
P
−2
1 π L + θ 3 LS′
− θ s −θ 3
= W ,
P
θ2 −θ 2 (GS′)−1
Copyright © 2021. Harvard University Press. All rights reserved.
W
since, by virtue of equation (9.2), 1 − (P/W)−1/FL vanishes in the neighbor-
hood of equilibrium. We thus have tr J < 0 and det J > 0, so the equilibrium in
both Figure 9.16 and Figure 9.18 is stable.
Fixprice dynamics are a bit more complicated. We have
•
P
W
= − GS P − Y − Y − LS P , (9.3)
P 1 3 W
W
W
338
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
9 • Mathematical Appendix 339
P
−1
Y = θ 2 I ( ρ) − s π Y − L , (9.4)
W
and the Jacobian is
−θ1GS′ + θ 3 LS ′ θ1 − θ 3
−1
J = −2
P
W .
−θ 2 s π P L −θ 2 s π 1 −
W FL
To the lef of the intersection of the AD and GS schedules,
−1
P
W
1− > 0,
FL
so tr J is unambiguously negative. Provided 1 > 3, as in Figure 9.17, det J > 0,
so that the equilibrium E is stable.
To the right of the intersection of the AD and GS schedules, the sign of tr J
depends on the speed of adjustment of prices and wages relative to the speed
of adjustment of output. If 1 and 3 are large compared to 2, then tr J will be
negative despite the fact that 1 − (P/W)−1/FL is negative. With 1 > 3, det J is
positive if and only if the slope of the stationary-price locus (relative to the
Y-axis) exceeds the slope of the AD schedule, which is to say,
P
−1
1 − W
1 − 3 FL
> − . (9.5)
1GS′ − 3 LS′ P
−2
W L
Copyright © 2021. Harvard University Press. All rights reserved.
Tis explains why in Figure 9.19 E is stable, and why E′ is unstable in both
Figure 9.17 and Figure 9.19.
Observe that the condition det J > 0 precludes a stable infationary equilib-
rium if wages adjust faster than prices. If 3 > 1, then the inequality (9.5) can’t
possibly hold, since the lef-hand side is negative while the right-hand side is
positive.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 10 .
investment
Te level of output and employment as a whole depends on the
amount of investment. I put it in this way, not because this is the only
factor on which aggregate output depends, but because it is usual in a
complex system to regard as the causa causans that factor which is
most prone to sudden and wide fuctuation.
—john maynard keynes
In Te General Teory, and in the macroeconomics that Te General Teory
spawned, investment means the commitment of abstract purchasing power to
specifc physical forms—plant, equipment, houses, infrastructure. It is that
commitment that conceptually distinguishes investment from saving. In con-
trast with investment, saving constitutes withholding a commitment, namely,
the commitment of resources to consumption.1
Tough investment and saving are conceptually distinct, actual saving and
actual investment are always equal, at least in a simple economy with no gov-
ernment expenditures and revenues and no foreign trade. Te resources with-
held from consumption are always equal to the commitment of resources to
investment, provided we include the accumulation of inventories of fnal
goods and goods in process as part of investment whether or not this accu-
mulation is intended or is instead the result of a miscalculation on the part of
producers. Tis is why saving and investment in national-income accounts
are always equal, except for statistical errors that arise because of the diferent
Copyright © 2021. Harvard University Press. All rights reserved.
ways that realized investment and saving are measured.
Intended investment and saving are another matter. Tere is no reason why
investment desires and saving desires should always coincide. Tis is espe-
cially the case in fractional-reserve banking systems: when banks can create
claims on resources, investment does not depend on prior decisions about
saving. Nonetheless, a condition of equilibrium between income and expen-
diture (again, absent government and foreign trade) is that desired saving and
desired investment equal each other. Tat is, the amount of resources agents
Portions of this chapter were previously published in Te Dismal Science: How Tinking Like
an Economist Undermines Community (Cambridge, Mass.: Harvard University Press, 2008).
340
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 341
collectively wish to withhold from consumption and the amount they wish to
commit to specifc capital goods must be equal. Say’s Law holds that this
equality is true regardless of the level of income and output. In Te General
Teory this equality is the condition that defnes the AD schedule.
Tis chapter is concerned with how the amount of resources people wish to
commit to specifc capital goods, how investment demand, is determined. Te
frst question perhaps is why the demand for investment, the fow of new cap
ital goods, attracts our attention (or Keynes’s) at all. Why not focus on capital
as a stock?
If capital were Lego sets and could be dismantled and put together in new
forms at low cost, the action would be in a market for stocks of capital, not in
a market for fows of investment. Each frm could adjust its capital stock as
easily as it can adjust, or at least is assumed to be able to adjust, its labor force.
Te demand for capital would be determined by its marginal-productivity
schedule, and the intersection with today’s supply schedule would determine
the current marginal productivity and the interest rate—in a world of Legos,
the focus would be on short-run rates, because specifc capital goods would
exist only so long as these particular forms of capital are proftable for their
owners.
Tis would still leave open the question of how the pace of capital accumu-
lation is determined. In Lego world, today’s investment demand would be in-
fnitely elastic at an interest rate equal to today’s level of the marginal pro
ductivity of capital. Logically, the rate of capital-stock growth must then be
determined by the supply of saving.
Tis is precisely the model that Paul Samuelson proposed in the various
editions of his elementary text, Economics, from the 1950s to the 1980s (I have
consulted the second, third, and the eleventh—the last he authored by him-
self). In the eleventh edition he writes that net investment ceases when the
interest rate is “low enough to choke of all desire to save, . . . low enough to
make the community’s average propensity to consume equal to 100 percent of
Copyright © 2021. Harvard University Press. All rights reserved.
income” (1980, p. 562).
As we shall see in chapter 17, this is also the model implicit in the canonical
Solow growth model (1956). Observe that in the absence of any role for in-
vestment demand, there is no question of expenditure difering from income.
We are back in the Garden of Eden with Jean-Baptiste Say.
Implicit in Te General Teory view of investment demand as the critical
element in determining the level of aggregate output is what I believe to be
the correct intuition: that for the most part, once resources are committed to
specifc capital goods, these resources are frozen in place, a conception of the
relationship between capital and investment captured by the ideas of vintage
capital, irreversible investment, and putty-clay. In this view, investment is
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
342 Raising Keynes
necessary for innovation as well as expansion. Indeed, any new product or
new method, as well as additions to output, require a new commitment of
resources to specifc capital goods rather than simply a rearrangement of ex-
isting pieces of a Lego set.
Tis does not preclude markets in secondhand capital goods, nor does it
preclude markets for fnancial assets—stocks, bonds, and the like—that cor-
respond, at least dimensionally, to the agglomerations of capital goods. But,
for the most part—we will consider exceptions down the road—the market in
secondhand capital goods falls far short of what would be needed to make the
world conform to the model of Lego capital. And the market for fnancial as-
sets resembles a market for capital goods only dimensionally in that it is a
market for stocks (like capital) rather than a market for fows (like invest-
ment). Indeed, when Keynes rejected the idea that interest rates are deter-
mined by demand and supply in a market for investment and saving, he care-
fully avoided any suggestion that interest rates are determined in capital
markets understood as markets for physical capital goods. Rather
Te rate of interest is not the “price” which brings into equilibrium the de-
mand for resources to invest with the readiness to abstain from present
consumption. It is the “price” which equilibrates the desire to hold wealth in
the form of cash with the available quantity of cash. (Te General Teory,
p. 167; emphasis added)
Investment, in this view, is the way agents compensate for the missing market,
the one in which the price of capital—the interest rate—is theoretically deter-
mined by the demand for and supply of capital goods. Tis market is not
temporarily absent without leave because of frictions or imperfections; it is
permanently missing because the very nature of capital goods makes it ex-
tremely dif ficult to transform clay back into putty.
Tis said, in many ways the theory of investment is the least original ele-
ment of Te General Teory. Keynes builds up a demand schedule for invest-
Copyright © 2021. Harvard University Press. All rights reserved.
ment, or more precisely, for the resources that investment requires, by assum-
ing there exists an array of discrete projects ofering higher or lower returns.
In this, he follows Marshall and, even more closely, Irving Fisher. In the sim-
plest case, projects ofer a perpetual and constant annual return, net of oper-
ating costs, repairs, and maintenance. It simplifes the argument further, with-
out any loss of generality, to assume that all projects entail the same original
commitment, say, one dollar. In this case, projects can be ranked by their an-
nual return, and the schedule in Figure 10.1 refects, on the horizontal axis,
the amount of investment yielding at least as much as the interest rate speci
fed on the vertical axis. For example, the diagram says there are 100 one-
dollar projects yielding at least 0. At a level of investment of 100, 0 is
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 343
Interest rate ( )
Investment demand (marginal efficiency of capital)
0
100 200 300 400 Investment
Figure 10.1 Investment demand.
Keynes’s marginal ef ficiency of capital, though it would be more accurate to
refer to 0 as the marginal ef ficiency of investment.
Te MEC schedule becomes an investment-demand schedule if we treat
the vertical axis as the independent variable measuring a hurdle rate of inter-
est. If the alternative to investing in the real economy of plant, equipment,
infrastructure, houses is assumed to be holding a fnancial asset, the hurdle
rate of interest is the return available in asset markets on securities with a pat-
tern of returns comparable to the array of investment projects (and with the
same degree of uncertainty). In the present case, the hurdle rate is the coupon
yield on private-sector consols, bonds ofering a perpetual fxed payment with
no redemption date, with a default risk comparable to the probability of fail-
ure of potential investment projects. In the twenty-frst century, preferred
stock is the closest real-world analog of such obligations.
Copyright © 2021. Harvard University Press. All rights reserved.
Suppose that the consol yield is . Anyone with access to the array of proj
ects whose returns are represented by the schedule in Figure 10.1 then has a
choice between placing his one dollar in a consol yielding $ per year or un-
dertaking a physical investment costing one dollar. How much investment
will these agents undertake? If they are optimizing, they will choose projects
for which the annual return exceeds the opportunity cost, which is to say
projects for which the quasi-rent, q, the return net of variable costs for labor
and other inputs associated with the project, is greater than (or, at the margin,
equal to) .
When = 0, all projects represented in the part of the MEC schedule lying
above 0 pass the test, so investment demand is 100. Te MEC schedule, I( ),
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
344 Raising Keynes
is thus the bridge between the nominal economy of fnancial assets and the
real economy of production, consumption, and investment.
Durability and the Elasticity of Investment Demand
What determines the slope of I( )? Part of the answer, which is important for
how the efects of monetary policy are distributed among various sectors of
the economy, is investment durability. In this respect, investment in plant and
equipment is no diferent from consumer durables. A sensible consumer will
pay much less attention to the rate of interest on a car loan than she will to the
rate of interest on a home mortgage.
Similarly for business: the rate of interest looms larger the more durable is
a project. When a company is deciding whether or not to replace its employ-
ees’ laptops, which might last for three years, the interest rate will not play
much of a role in the decision. A decision about a new of fi ce building or new
railway track, expected to last for thirty, forty, or more years, is exceedingly
sensitive to the rate of interest.
We can formalize this intuition by assuming that instead of lasting forever,
projects are “one-hoss shays” that operate for a fnite amount of time and then
collapse into worthless heaps of junk.2 Assume as well that these investments,
like perpetuities, throw of constant cash fows (net of all operating and main-
tenance costs). Instead of being equal to the hurdle rate, as is the case for
perpetuities, this net cash fow now has to exceed the hurdle rate to allow for
amortization of the cost of the investment. In place of the requirement for a
perpetuity, q − > 0, the project has to yield a discounted present value
q
Q = q ∑ 1 (1 + ρ)− τ =
m
1 − (1 + ρ)−m
ρ
in excess of one dollar, the capital cost of the project. (Te rate of return of a
project is the value of for which Q = 1.) Te condition for an investment to
Copyright © 2021. Harvard University Press. All rights reserved.
be worthwhile, Q > 1, is equivalent to
q− > 0.
1 − (1 + )−m
Tis formula generalizes the condition
q−>0
for projects that return q in perpetuity.
Since [1 − (1 + )−m] increases as m increases, the required excess of the
quasi-rent over the hurdle rate varies inversely with the durability of the in-
vestment. Table 10.1 compares the cash fow required to cover interest and
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 345
principal when these charges are blended into a level payment for projects of
diferent durabilities that require the same initial outlay of one dollar. For a
three-year project, the required cash fow per dollar of investment changes
modestly, from $0.33 to $0.40, as the hurdle rate varies from 0 to 10 percent,
that is, by less than 25 percent. For longer-lived projects, the percentage
change is much greater, more than 60 percent for the ten-year project, and
300 percent for the thirty-year project. (Te absolute change in the quasi-rent
is similar in the three cases, close to $0.07.) Te impact on investment de-
mand is pictured in Figure 10.2, which shows investment-demand schedules
corresponding to the project durabilities in Table 10.1, with a perpetual re-
turn as a benchmark. (Figure 10.2 normalizes the amount of investment in
projects of diferent durability so that for each class there are 1,000 units of
investment when the hurdle rate is 0.01.)
Te importance of durability in determining the interest elasticity of in-
Table 10.1 Required Quasi-Rent in Relation to Project Durability
and Hurdle Rate
Durability
3 10 30
0.0 0.333 0.1 0.033
Hurdle Rate 0.05 0.367 0.13 0.065
0.1 0.402 0.163 0.106
0.05
0.04
0.03
Hurdle rate
Copyright © 2021. Harvard University Press. All rights reserved.
0.02
0.01
0.0
0 200 400 600 800 1,000 1,200
Investment demand
3 10 30 Perpetuity
Figure 10.2 Investment demand for projects of varying durability (years).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
346 Raising Keynes
vestment demand provides an important clue to how monetary policy oper-
ates, at least how it works through the interest-rate channel. Because changing
the hurdle rate of interest disproportionately afects potential investment with
a long pay-out period, monetary policy has historically focused on stimulat-
ing or contracting residential construction—housing being among the long
est lived in the array of potential investment projects. Te overreliance on
monetary policy to bring the economy out of the recession of 2001 con
tributed to the housing boom of the frst decade of this century. And this
boom was the proximate cause of the fnancial crisis of 2008 and the ensuing
recession.
Make vs. Buy and Tobin’s Q
Tere is another way to look at investment decisions. According to William
Brainard and James Tobin (1968, p. 104), “Investment is stimulated when cap
ital is valued more highly in the market than it costs to produce it, and dis-
couraged when its valuation is less than its replacement cost.” At one level, the
ratio of the value of new capital to its cost of production, which has been en-
shrined in the literature as Tobin’s Q, contains no new information. If we
continue to think in terms of projects with a capital cost of one dollar each,
with Q the discounted present value of the quasi-rents, then the investment is
proftable if Q > 1, and is unproftable if Q < 1. But this is precisely the same
information on which the investment-demand schedules in Figures 10.1 and
10.2 are based.
If this were all there were to Tobin’s Q, no one would fnd very much value
added in the concept. But Brainard and Tobin intend something more. In-
stead of a comparison of returns to a prospective project and its costs, they
have in mind a market valuation of capital goods versus the cost of producing
capital goods. Tis is an idea that Keynes had toyed with in his response to a
symposium organized by the Quarterly Journal of Economics shortly afer the
Copyright © 2021. Harvard University Press. All rights reserved.
publication of Te General Teory:
Capital-assets are capable, in general, of being newly produced. Te scale on
which they are produced depends, of course, on the relation between their
costs of production and the prices which they are expected to realize in the
market. Tus if the level of the rate of interest taken in conjunction with
opinions about their prospective yield raise the prices of capital-assets, the
volume of current investment (meaning by this the value of the output of
newly produced capital-assets) will be increased; while if, on the other hand,
these infuences reduce the prices of capital-assets, the volume of current
investment will be diminished. (1937a, pp. 217–218)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 347
Lego capital ofers an extreme version, in which the existence of a market
for capital goods vitiates the idea of a downward-sloping investment demand.
Tis idea evidently had some appeal for Tobin: his classic paper (1969, sec. 6)
models a long run in which Q = 1 holds as an equilibrium condition, which
makes sense for Lego capital. By contrast, in the short run, there are impedi-
ments to equating the cost of capital goods with their market values, and this
discrepancy drives investment.
Te original Brainard–Tobin model is more expansive. Instead of applying
the theory at the level of individual capital goods, Brainard and Tobin apply Q
to an ensemble of capital goods that constitutes an entire company:
One of the basic theoretical propositions motivating the model is that the
market valuation of equities, relative to the replacement cost of the physical
assets they represent, is the major determinant of new investment. (Brainard
and Tobin 1968, pp. 103–104)
Again, Keynes had had the same idea:
Te daily revaluations of the Stock Exchange, though they are primarily
made to facilitate transfers of old investments between one individual and
another, inevitably exert a decisive infuence on the rate of current invest-
ment. For there is no sense in building up a new enterprise at a cost greater
than that at which a similar existing enterprise can be purchased; whilst
there is an inducement to spend on a new project what may seem an ex-
travagant sum, if it can be foated of on the Stock Exchange at an immediate
proft. Tus certain classes of investment are governed by the average expec-
tation of those who deal on the Stock Exchange as revealed in the price of
shares, rather than by the genuine expectations of the professional entrepre-
neur. (Te General Teory, p. 151)
In this formulation, Q no longer represents the investor’s calculation of
the present value of a stream of future returns. It is now a market valuation of
Copyright © 2021. Harvard University Press. All rights reserved.
this stream, as refected in the price of the shares of the company that owns
a particular ensemble of capital goods. If you are thinking of entering an
industry, one option is to buy up all the shares of XYZ, an existing company
in the feld. Te other is to launch a whole new enterprise. If we normalize
the cost of building new at one dollar, and the cost of the shares of XYZ as
Q, we are formally in the same place: Q > 1 is the signal to invest, that is, to
build new; Q < 1 is the signal not to invest, but to buy an existing company
instead.
Tis decision is diferent from the decision that has engaged our attention
until now. It is no longer a choice between undertaking a project or placing
the same purchasing power in the market for fnancial assets. Te choice now
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
348 Raising Keynes
is to make or buy physical assets, with “make” implying a decision to under-
take investment and “buy” representing a decision to take over an existing
company.
Tere are certainly examples of this logic at work. When Microsof fnally
looked at the search-engine business, it had the choice of acquiring an exist-
ing business or starting from scratch. In the event, neither option proved
proftable because the Google train had long since lef the station. When
Amazon decided to enter the food-distribution business, it could have started
from scratch to build a new supermarket chain. It decided instead to acquire
Whole Foods.
In this world, the driving force of investment is the stock market. As Brain-
ard and Tobin put it, and as subsequent commentators have understood the
meaning of Q, “Investment is encouraged when the market yield on equity . . .
is low relative to the real returns to physical investment (1968, p. 104).”
A stock market boom stimulates investment because investment becomes
cheaper than acquisition. By the same token, a slump makes investment less
attractive because companies now become dirt cheap.
But is buying as realistic as the Brainard–Tobin version of Q-theory as-
sumes? Te question is how representative is Microsof’s choice about how to
get into the search-engine business, or Amazon’s choice about how to learn
the ropes of the grocery business. Te econometric evidence is mixed, re-
searchers fnding a strong infuence of the stock market on investment or
none, according to how they specify the model. My own view is that a coher-
ent business strategy normally is not, and cannot be, pursued by acquisition;
investment is the principal means by which new processes, methods, and
products, as well as additional capacity to produce existing goods, are intro-
duced. Investment is the way an actual economy compensates for the absence
of a market in capital goods. Te absence of this market, it bears repeating, is
not the result of imperfections and frictions and thus cannot be brought into
existence by the magic transition from the short run to the long run.
Copyright © 2021. Harvard University Press. All rights reserved.
Tere are important exceptions. Warren Bufett’s Berkshire Hathaway is
one, perhaps the exception par excellence. Bufett apart, housing is an entire
sector where making and buying are real options. New and existing homes are
not perfect substitutes for one another but are close enough to make Q-theory
relevant. And in fact, as a ratio of the value of residential housing to its re-
placement cost, Tobin’s Q, correlates closely with residential construction, at
least since the mid-1990s. Figure 10.3 shows the data from the early 1950s to
2019. Q-theory works here because it is applied to individual capital assets:
when home prices are high relative to construction costs, new home con-
struction is very proftable and the industry booms. When housing prices
come down, residential construction c omes down too.
We now have a second possible interest-rate channel through which mon-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 349
0.07 2.50
0.06 2.25
0.05 2.00
0.04 1.75
0.03 1.50
0.02 1.25
1960 1970 1980 1990 2000 2010
Tobin’s Q: value of real estate/replacement cost (right axis)
Residential fixed investment/potential GDP (left axis)
Figure 10.3 Tobin’s Q and residential construction (1952–2019). Sources: BEA,
Board of Governors, CBO. myf.red/g/pYZn
etary policy might drive the demand for residential construction. Prices of
existing homes are sensitive to interest rates because of their durability, and
new construction is sensitive to the price of existing homes relative to new
construction. Te idea, in other words, is that interest rates drive Q.3
Capital Widening vs. Capital Deepening
In chapters 3 to 9, and up to now in this chapter, investment demand depends
only on the rate of interest. But I noted in chapter 3 that the efect on invest-
Copyright © 2021. Harvard University Press. All rights reserved.
ment demand of the real price P/W would be addressed in due course. “Due
course” is now.
Te responsiveness of investment demand, and hence aggregate demand,
to the real-price level—and hence to the distribution of income between
profts and wages—depends on what kind of investment one has in mind, and
the predominance of diferent kinds of investment will vary with the cycle.
Te frst type is investment that adds to productive capacity, capital widening
in an older literature.4 For this type of investment, chapter 5 gives us a way of
thinking about the decision that goes beyond a simple comparison of the an-
ticipated return with the hurdle rate. Tere we explored how producers are
guided by two diferent signals from the market, one about how quickly
goods are moving of the shelves, the other about the proftability of selling
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
350 Raising Keynes
more goods. Te same two signals can be assumed to be at work in invest-
ment decisions. Capacity utilization provides information about how likely it
is that additional goods will fnd a market. Te more production presses on
capacity, the more likely it is that new capacity can be proftably deployed. Te
real price provides additional information, information about how proftable
additional production will be if it can be sold. Tis, as we shall see in chapters
18 and 19, is the type of investment on which growth models, including my
own (Marglin 1984, chap. 20; Marglin and Bhaduri 1990; Bhaduri and Mar-
glin 1990), typically have focused.
Tere is another kind of investment, investment that substitutes capital for
labor (or other inputs), capital deepening. Capital deepening difers from cap
ital widening in two ways. First, it is in principle less susceptible to the state of
the market, to whether goods are moving quickly or slowly. Te point of cap
ital deepening is to cut costs, not to bring more goods to market; so its proft
ability does not depend on a growing demand. Second, since the name of the
game is cutting costs, the higher the real wage (the lower the real price), the
more proftable is capital substitution.
To capture the diference between these two basic types of investment, con-
sider investment proftability in a highly simplifed world in which capital is
infnitely durable. Te quasi-rent is the diference between (1) the gross re-
turn (equal to the sum of the marginal productivity of capital and the output
attributable to any associated change in employment) and (2) the wages as-
sociated with the change in employment. Te net return per unit of invest-
ment (K = 1) is the quasi-rent less the annual interest cost:5
−1
∆L P ∆L
q − ρ = MPK + MPL − − ρ.
∆K W ∆K
In the mainstream conception of equilibrium, the above expression van-
ishes: the marginal productivity of capital (MPK) is equal to the interest rate
Copyright © 2021. Harvard University Press. All rights reserved.
( ), and the marginal productivity of labor (MPL) is equal to the real wage
(W/P). In this book, there is no mechanism that brings capital productivity
and the interest rate into equality once we leave the world of Lego capital.
Moreover, the marginal productivity of labor and the real wage are equal only
at a fexprice equilibrium. Otherwise, the change in labor input (L) shows
up in both the return and the cost of the investment, which is why the real
price has opposite efects on capital widening and capital deepening.
For capital widening the generic formula for q − above becomes
∆L P ∆L
−1
q − ρ = Ω(Y ) MPK + MPL − − ρ,
∆K W ∆K
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 351
where (Y) is a parameter that varies from 0 to 1 as Y increases, the value of
refecting the confdence that business has in its ability to sell additional
output. If we assume that for capital widening the incremental labor:capital
ratio is the current average labor:capital ratio,
∆L L
= ,
∆K K
we can rewrite the net return per unit of investment as
L P L
−1
q − ρ = Ω(Y ) MPK + MPL − − ρ .
K W K
With constant returns to scale
L Y
MPK + MPL =
K K
so
Y P 1 L
q (Y ) (Y )r ,
K W K
which is to say that the anticipated quasi-rent becomes the existing proft rate,
r, multiplied by the probability of selling the additional goods, (Y). Te net
annual return is
q − r = (Y)r − .
For capital deepening, we assume that output is unchanged, which is to say,
MPKK + MPLL = 0,
so that
∆L MPK
Copyright © 2021. Harvard University Press. All rights reserved.
=− .
∆K MPL
In this case the net annual return is
−1 −1
P ∆L P MPK
q − ρ = − − ρ= − ρ.
W ∆K W MPL
Te quasi-rent q is the annual saving in wages, −(W/P)L, per unit of invest-
ment, K. Since the amount of labor displaced by a unit of additional capital,
−L/K, is equal to the inverse ratio of the respective marginal productivi-
ties, MPK /MPL, the quasi-rent increases with Y and decreases with (P/W).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
352 Raising Keynes
In a world of perfect certainty, or in a world of Lego capital, where capital
can be reshaped as new information becomes available, an excess of the quasi-
rent over and above the hurdle rate of interest, q − > 0, would call forth in-
vestment until the ratio MPK/MPL falls to the point at which q − = 0. But in
an uncertain world in which investment is irreversible, the most that we can
assume is that investment demand is an increasing function of the annual re-
turn, (q − ). In this case the demand for capital-widening investment is
P Y P −1 L
I , ρ = ψ (q − ρ) = ψ Ω ( Y ) − − ρ ,
W K W K
and the demand for capital deepening is
P P −1 MP
K
I , ρ = ψ(q − ρ) = ψ − ρ .
W W MPL
In consequence, the AD schedule is no longer vertical in the space of real
price and output even if we hold constant. Its slope depends on the mix of
the two kinds of investment as well as on the elasticity of substitution in pro-
duction.
We will examine investment demand in more detail in chapter 18. For now,
the important takeaway is that the equilibrium depends on whether invest-
ment tends to widen capital or deepen capital, which will depend importantly
on where the economy is situated in the business cycle. Te proftability of
capital widening is infuenced by investors’ perceptions with respect to the
market for additional output, captured here by the magnitude of the parame-
ter (Y). Tis parameter will be smaller to the point of vanishing when there
is considerable slack in the economy; it will be close to one during a boom. So
capital widening may be expected to be an important, if not necessarily the
predominant, form of investment in prosperous times. During a recession,
Copyright © 2021. Harvard University Press. All rights reserved.
when (Y) is a very small number, capital deepening may be the only game
in town.
For each type of investment, we can derive the AD schedule by a construc-
tion analogous to the construction in Figure 9.15. Figure 10.4 constructs the
AD schedule for capital widening. Tere is a critical diference between Fig
ure 10.4 and Figure 9.15: for the AD schedule corresponding to the Cam-
bridge saving theory, points above the schedule represent an excess of saving
over investment and hence an excess of income over expenditure; points be-
low refect an excess of investment over saving. Here the situation is reversed.
Above the AD schedule, desired investment exceeds saving, whereas below,
saving exceeds investment. Consequently, the dynamics are very diferent in
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 353
(a) Investment and saving holding real-price constant
ID , SD
SD sY
1
Y P L
ID (Y )
K W 2 K
1
Y P L
ID (Y )
K W K
1
1
Y P L
ID (Y )
K W 0 K
Y1 Y 2 Y3 Y4 Y
P Supply of goods
W
Aggregate demand
P
W 2
ID SD
P
W 1
P SD ID
W 0
Copyright © 2021. Harvard University Press. All rights reserved.
Y1 Y 2 Y3 Y4 Y
(b) Aggregate demand varying real price
Figure 10.4 Aggregate demand with capital-widening investment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
354 Raising Keynes
the two cases. Figures 10.5 and 10.6 depict fexprice equilibria, and Figures
10.7 and 10.8 depict fxprice equilibria. Te stability of these equilibria de-
pends on the relative speeds of adjustment of prices, wages, and output. In the
fexprice case, stability is guaranteed if wages adjust much more rapidly than
prices—the opposite of Keynes’s assumption of a fxed money wage. Stable
P
W
Stationary real price Supply of goods
Y 0
P 0
W
Aggregate demand
P 0
E
W 0
Supply of labor
Figure 10.5 Defationary equilibrium with capital-widening
investment: fexprice adjustment.
Stationary real price
P
P 0
W
W
Supply of goods
Y 0
E Aggregate demand
Copyright © 2021. Harvard University Press. All rights reserved.
P 0
W 0
Supply of labor
Figure 10.6 Infationary equilibrium with capital-widening
investment: fexprice adjustment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 355
Stationary real price
P 0
W
P
W
Supply of goods
P 0
E Aggregate demand
Y 0
W 0
Supply of labor
Figure 10.7 Defationary equilibrium with capital-widening investment:
fxprice adjustment.
Aggregate
demand
P P
Supply of
W W goods
Aggregate Stationary
demand real price
E Supply of goods E
E
Stationary
real price
E
Copyright © 2021. Harvard University Press. All rights reserved.
Supply of labor Supply of labor
Y Y
E is stable E is stable
E is unstable E is unstable
(a) Equilibrium real price falls (b) Equilibrium real price rises
with output with output
Figure 10.8 Equilibria with capital-widening investment: fxprice adjustment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
356 Raising Keynes
equilibrium in the fxprice case requires (1) that the slope of the AD schedule
exceed the slope of the stationary-price locus (as it does at E in both Figure
10.7 and Figure 10.8 but not at E′), and (2) if the equilibrium entails defation,
that prices and wages adjust more rapidly than output. Details are in the
mathematical appendix to this chapter.
An AD schedule corresponding to capital deepening is constructed along
the same lines as in Figure 10.4. In Figure 10.9, investment demand and the
(a) Investment and saving holding real-price constant
ID , SD P 1 MPK
ID
W 1 MP
L
P 1 MPK S D sY
ID
W 0 MP
L
P 1 MPK
ID
W 2 MP
L
P
W Aggregate demand
P
W 2
SD ID ID SD
Copyright © 2021. Harvard University Press. All rights reserved.
P
W 1
P
W 0
Y0 Y1 Y2 Y
(b) Aggregate demand varying real price
Figure 10.9 Aggregate demand with capital-deepening investment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 357
supply of saving are shown in panel (a), and the corresponding AD schedule
is shown in panel (b).
Te frst question is why the investment-demand schedule slopes upward
in Figure 10.9(a). We have assumed from the get-go that the marginal prod-
uct of labor declines as more workers are employed with any given capital
stock. And as the marginal product of labor declines, the marginal productiv-
ity of capital increases, thus giving a double boost to investment proftability.
But this is at best a partial answer: as the AD schedule in Figure 10.9(b) is
drawn, investment proftability responds so strongly to the level of output that
the slope of the investment-demand schedule increases with Y. Te reason is
an implicit assumption that technical possibilities for substituting capital for
labor are limited, which formally is an assumption that the elasticity of sub
stitution in production is less than one. Tis means that, as output expands,
the ratio of MPK to MPL increases at a faster (percentage) rate than does the
labor:capital ratio L/K. On this assumption, the marginal product of labor
falls so rapidly (and the marginal product of capital increases so rapidly) that
investment demand and hence aggregate demand respond more than propor-
tionately to changes in output.6
Flexprice equilibria are depicted in Figures 10.10 and 10.11. Te stability of
the equilibria in these fgures depends on the assumption that the GS sched-
ule is steeper than the stationary-price locus.
Te fxprice case is depicted in Figure 10.12. Tere are two equilibria: E′ is
unstable, and E is stable. As in Figure 9.19, stability depends on the relative
slopes of the two schedules that determine the equilibria; in Figure 10.12,
however, the stationary-price locus must be steeper than the AD schedule
P
Supply of goods Stationary real price
W
Y 0
P 0
W
Copyright © 2021. Harvard University Press. All rights reserved.
E
Aggregate demand Supply of labor
P 0 0
W
Figure 10.10 Defationary equilibrium with capital-deepening
investment: fexprice adjustment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
358 Raising Keynes
P
W
Supply of goods
Y 0
Stationary real price
P 0
W
Aggregate demand Supply of labor
P 0 0
W
Figure 10.11 Infationary equilibrium with capital-deepening
investment: fexprice adjustment.
P Y 0
E Aggregate demand
W
Supply of goods
Y 0
Stationary real price
P 0
W
E
Supply of labor
W 0
Copyright © 2021. Harvard University Press. All rights reserved.
Figure 10.12 Multiple equilibria with capital-deepening investment:
fxprice adjustment.
for a fxprice equilibrium to be stable. In addition prices must adjust faster
than wages. Once again the details are in the mathematical appendix to this
chapter.
Observe that the upward slope of the AD schedule does not contradict the
assumption that capital-deepening investment demand is a decreasing func-
tion of the real price. Tis assumption is refected in Figure 10.9(a), in which
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 359
P Y 0
E Aggregate
W demand
F
Supply of goods
Y 0
Stationary real price
P 0
W
E
Supply of labor
W 0
Figure 10.13 A shif in the aggregate-demand schedule due to a
decrease in investment demand or an increase in desired saving.
investment-demand schedules associated with lower real prices refect more
demand at each level of output. Te AD schedule in Figure 10.9(b) slopes
upward only because this greater demand is absorbed by the available saving
at lower levels of output. Even though the AD schedule slopes upward, shocks
to aggregate demand shif equilibria as our intuition would suggest: a change
in investment demand will lead to a change in output and employment in the
same direction; a change in desired saving will lead to output and employ-
ment changes in the opposite direction. If an increase in the hurdle rate of
interest displaces the investment-demand schedule ID = (q − p) downward,
or if the saving propensity increases, so that SD = sY is displaced upward, then
the AD schedule moves downward, as in Figure 10.13. Te (stable) equilib-
Copyright © 2021. Harvard University Press. All rights reserved.
rium at E shifs downward to F, where output and the real price are both
lower.
The Psychological Determinants of Investment:
Algorithm vs. Experience
I have followed Keynes (and Irving Fisher) in measuring an investment’s at-
tractiveness by the diference between its quasi-rent and the annual cost of
interest and amortization, which, with our simplifying assumptions is equiva-
lent to measuring investment worth by the diference between the discounted
present value of its future returns and its initial cost. Alas, the formalism of
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
360 Raising Keynes
the theory obscures its real content, or, perhaps I should say, lack of content.
Te problem is that the returns of any project, lying in the future, are not ob-
jectively given but are a subjective construction on the part of the investor.
Tese returns depend on conjunctions of events about which the investor
may have only vague notions and hunches, hardly the stuf of meaningful
formal calculation. For Keynes, returns are vessels more or less flled accord-
ing to the optimism or the pessimism of the investor herself. For Keynes, the
“animal spirits” of investors play a crucial role in investment demand, and
when investors’ courage fails, so does investment demand. Indeed, one of Te
General Teory’s most celebrated passages is a rif on animal spirits:
Tere is the instability due to the characteristic of human nature that a
large proportion of our positive activities depend on spontaneous optimism
rather than on a mathematical expectation, whether moral or hedonistic or
economic. Most, probably, of our decisions to do something positive, the full
consequences of which will be drawn out over many days to come, can only
be taken as a result of animal spirits—of a spontaneous urge to action rather
than inaction, and not as the outcome of a weighted average of quantitative
benefts multiplied by quantitative probabilities. Enterprise only pretends to
itself to be mainly actuated by the statements in its own prospectus, however
candid and sincere. Only a little more than an expedition to the South Pole,
is it based on an exact calculation of benefts to come. Tus if the animal
spirits are dimmed and the spontaneous optimism falters, leaving us to de-
pend on nothing but a mathematical expectation, enterprise will fade and
die. (pp. 161–162)
When animal spirits fail, all investors have to fall back on is the wisdom of the
crowd. As Chuck Prince, erstwhile CEO of Citigroup, put it just before the
economy began to unravel in 2007,
When the music stops, in terms of liquidity, things will be complicated. But
Copyright © 2021. Harvard University Press. All rights reserved.
as long as the music is playing, you’ve got to get up and dance. We’re still
dancing. (quoted in Nakamoto and Wighton 2007)
Keynes anticipated this reliance on the wisdom (or folly) of the crowd. In his
Quarterly Journal of Economics response to critics, he expanded on the role of
animal spirits and what happens when animal spirits fail:
We have, as a rule, only the vaguest idea of any but the most direct conse-
quences of our acts . . .
By “uncertain” knowledge, let me explain I do not mean merely to distin-
guish what is known for certain from what is only probable . . . Te sense in
which I am using the term is that in which the prospect of a European war is
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 361
uncertain [Keynes was writing this in 1937], or the price of copper and the
rate of interest twenty years hence, or the obsolescence of a new invention,
or the position of private wealth-owners in the social system in 1970. About
these matters there is no scientifc basis on which to form any calculable
probability whatever. We simply do not know. Nevertheless, the necessity for
action and for decision compels us as practical men to do our best to over-
look this awkward fact and to behave exactly as we should if we had behind
us a good Benthamite calculation of a series of prospective advantages and
disadvantages, each multiplied by its appropriate probability, waiting to be
summed.
How do we manage in such circumstances to behave in a manner which
saves our faces as rational, economic men? We have devised for the purpose
a variety of techniques, of which much the most important are the three fol-
lowing:
(1) We assume that the present is a much more serviceable guide to
the future than a candid examination of past experience would show it to
have been hitherto. In other words we largely ignore the prospect of future
changes about the actual character of which we know nothing.
(2) We assume that the existing state of opinion as expressed in prices and
the character of existing output is based on a correct summing up of future
prospects, so that we can accept it as such unless and until something new
and relevant comes into the picture.
(3) Knowing that our own individual judgment is worthless, we endeavor
to fall back on the judgment of the rest of the world which is perhaps better
informed. Tat is, we endeavor to conform with the behavior of the majority
or the average. Te psychology of a society of individuals each of whom is
endeavoring to copy the others leads to what we may strictly term a conven-
tional judgment. (1937a, pp. 213–214)
Not only do agents lack certainty about the returns of prospective investment
Copyright © 2021. Harvard University Press. All rights reserved.
projects, they cannot even attach objective probability assessments to the
range of possible outcomes. Te returns that govern proft calculations are
embedded not in the investments themselves but in the heads, and as Keynes
would have it, in the hearts of the investors. Proftability, like beauty, lies in
the eye of the beholder.
Frank Knight (1921) distinguished uncertainty and risk on the grounds of
whether meaningful probabilities can be calculated; for Knight, meaningful
probabilities were associated with objective frequencies. As we saw in chapter
9, assimilating uncertainty to risk became part of the project of restoring the
hegemony of mainstream microeconomics afer Te General Teory. Tis
project was partly about coherence, to reconcile Keynes’s macroeconomic ar-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
362 Raising Keynes
guments with the principles of microeconomics for the sake of a unifed the-
ory—or, failing reconciliation, to toss out Keynes’s arguments as bathwater—
even if this necessitated the demise of the baby as well.
Te mainstream was also, perhaps more so, interested in restoring the le-
gitimacy of the claim that markets work well for people, a central component
of which was utility maximization—the “good Benthamite calculation” to
which Keynes adverts in his 1937 discussion of uncertainty.7 Te problem for
the mainstream was how people could possibly maximize utility when they
can’t calculate probabilities.
Te frst step in the assimilation process was to blur the Knightian distinc-
tion by transforming Knight’s binary opposition into a continuum and ob-
serving that, at the risk end of the spectrum, we can never really know an
empirical (as distinct from logical) probability distribution; we have only
more or less relevant information on particular samples. By the same token, at
the uncertainty end, we are never totally without information about the likeli-
hood of alternative outcomes. Moreover, the essential (for Knight) institu-
tional diference between risky and uncertain events—the dif ficulties of in-
surance in the case of uncertainty—is also blurred: markets exist in a wide
range of situations that are closer to uncertainty than to risk; for instance, fu-
tures markets in commodities and the ad-hoc insurance contracts that have
long been the specialty of Lloyd’s of London.
Te blurring in practice of the distinction between risk and uncertainty
ignores the obvious in Knight’s theory—that these categories are ideal types.
(In this, Knight was likely infuenced by the great German sociologist Max
Weber.) Te existence of mixed cases and fuzzy lines becomes the pretext for
abolishing the distinction altogether. If one is disposed in that direction, it is
an easy intellectual step from the fuzziness at the edges of the distinction to
the idea that all probabilities are personal or subjective in nature.
Frank Ramsey (1980 [1926]) and Bruno de Finetti (1980 [1937]) provided
axiomatic foundations for subjective probabilities, which captured the atten-
Copyright © 2021. Harvard University Press. All rights reserved.
tion of both economists and statisticians in the wake of the seminal work on
game theory by John von Neumann and Oskar Morgenstern (1944). Tese
ideas took shape in the formal model of decision making elaborated by Jim-
mie Savage (1954), according to which optimal decisions are characterized by
the maximization of expected utility, with the probability distribution given
by the subjective evaluations of the decision makers in a way that parallels the
utility function as the subjective valuation of diferent outcomes.
Savage had not even published his theory when Maurice Allais (1953) of-
fered an example of a choice situation in which experimental subjects were
inclined to violate the Savage axioms. (Savage himself was, on his own admis-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 363
sion [1954], tempted, but upon refection recanted intuition in favor of the
axioms.) A few years later Daniel Ellsberg (1961) crafed an example to sepa-
rate risk from uncertainty and found once again that Savage, along with many
other distinguished economists and decision theorists, was disposed to reject
Savage, whose axioms have no room for the “ambiguity”—Ellsberg’s word—of
pure uncertainty. Indeed, Keynes had anticipated Ellsberg, distinguishing am-
biguity (which he called the “weight of an argument”) from probability un-
derstood as degree of belief. However, Keynes went only half-way down the
street of subjective probability theory: A Treatise on Probability (published in
1921 as a revised version of his fellowship dissertation of 1908) interpreted
probability as degree of belief rather than relative frequency, but Keynes held
to the view that degree of belief could be given an objective, interpersonal
meaning as belief rationally justifed by the evidence.8
Subsequent criticism, initiated by Daniel Kahneman and Amos Tversky
(1979), has been even more damning, since it questions the very ability of
people to make the kind of calculations required by probabilistic choice with
any degree of consistency. For example, Tversky and Kahneman (1981)
showed that decision makers respond to the way in which the problem is
posed, a 90 percent chance of surviving a medical procedure eliciting a difer-
ent response than a 10 percent chance of succumbing. Te line of investiga-
tion begun by Kahneman and Tversky has gathered a substantial following
within economics, its value recognized in the award of a Nobel Prize to Kah
neman in 2002.9
Nonetheless, subjective probability has a frm hold on the economist’s
imagination. Tis, it should be understood, is not thanks to the predictive or
normative power of the approach. Te hidden agenda is to take human be-
havior, strip it of all its vagaries, conficts, and contradictions, and then com-
press it within the confnes of the maximization paradigm. In the positivist
euphoria of the mid-twentieth century, the appeal of subjective probability
was doubtless enhanced by its consistency, at least in Savage’s version, with
Copyright © 2021. Harvard University Press. All rights reserved.
the axioms of revealed preference: in principle, subjective probabilities, like
subjective utilities, could be recovered from choices and actions. But for this
project to succeed, subjective probabilities have to be stable, rooted in a fun-
damental psychology that is resilient to changing facts on the ground. If prob-
abilities change with the phases of the moon, it will be neither possible to re-
cover these probabilities from agents’ choices nor to characterize the choices
people make as optimal in any but the most ephemeral manner. In the ab-
sence of stable subjective probabilities, the attempt to recover subjective prob-
abilities from observed behavior would not contribute anything to the argu-
ment that markets work well for people. If one gives up on this argument—as
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
364 Raising Keynes
well we should—it is hard to come up with any reason for maintaining the
purity of economic man as a maximizing agent.10
I expressed my own view in the previous chapter: under conditions of un-
certainty, consumption decisions are unlikely to rely very heavily on an ap-
paratus of calculation and maximization. Tis is true in spades for investment
decisions. Moreover, to the extent calculation governs investment, subjective
probabilities are not likely to be stable. And without objective data on which
to peg probabilities, individuals necessarily fall back on intuition, conven-
tional behavior, authority—in short, on experience—to form subjective prob-
abilities. Experience constitutes a diferent system of knowledge from the
knowledge that drives maximizing behavior, what I call algorithm. It not only
has a diferent epistemology, it also has diferent rules of transmission and in-
novation, not to mention a diferent implicit politics.
I have elaborated the contrast and complementarity of algorithm and expe-
rience in a variety of publications, most recently in my book on how econom-
ics has undermined community (2008). Suf fi ce it to say that here the point of
a systems-of-knowledge approach is not to deny a role for calculation, but to
argue that a framework of calculation based on subjective probabilities should
be understood as one cause of the vagaries of investment demand. And, thus,
subjective probability actually creates a space for the elements that Keynes
thought essential to understanding investment behavior, from the intuition of
animal spirits to the conventional judgments of the crowd.11
Te subjective element in investment calculation explains why investment
demand, and hence aggregate demand, is likely to be volatile, more volatile
than if it were grounded in objective data subject to rational calculation. Con-
sider the expression for the annual return anticipated on capital widening:
Y P −1 L
q − ρ = Ω(Y ) − − ρ.
K W K
Copyright © 2021. Harvard University Press. All rights reserved.
In this expression the parameter (Y), which measures the perceived proba-
bility of making use of new capacity, does not refect a rational calculation of
the state of the market, but is rather a refection of hopes and fears. Tis pa-
rameter will respond to nonrational waves of optimism and pessimism that at
one point in time buoy up investment demand and at another drive invest-
ment demand into the ground. Nor is the expression within brackets, the re-
turn on investment conditional on the new production fnding a market,
engraved in stone. As a projection into the future, Y/K − (P/W)−1L/K is a
property of the investor’s mind, not of the investment project.
In any case, the prospective annual return is not dispositive of the amount
of investment demand. Twixt cup and lip is the function relating the pro-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Investment 365
spective return to some amount of new physical capital: is not fxed by any
objective data but rather embodies the animal spirits that Keynes invoked to
explain the vagaries of investment demand.
I will give Keynes (almost) the last word:
We have seen above that the marginal ef ficiency of capital [schedule] de-
pends, not only on the existing abundance or scarcity of capital-goods and
the current cost of production of capital goods, but also on current expecta-
tions as to the future yield of capital-goods. In the case of durable assets it is,
therefore, natural and reasonable that expectations of the future should
play a dominant part in determining the scale on which new investment is
deemed advisable. But, as we have seen, the basis for such expectations is
very precarious. Being based on shifing and unreliable evidence, they are
subject to sudden and violent changes. . . .
Ofen the predominant, explanation of the crisis is . . . a sudden collapse
in the marginal ef ficiency of capital [schedule].
Te later stages of the boom are characterised by optimistic expectations
as to the future yield of capital-goods suf ficiently strong to ofset their grow-
ing abundance and their rising costs of production and, probably, a rise in
the rate of interest also. It is of the nature of organised investment markets,
under the infuence of purchasers largely ignorant of what they are buying
and of speculators who are more concerned with forecasting the next shif of
market sentiment than with a reasonable estimate of the future yield of cap
ital-assets, that, when disillusion falls upon an over-optimistic and over-
bought market, it should fall with sudden and even catastrophic force. (Te
General Teory, pp. 315–316)
Te volatility of investment is central to understanding both the limits of
self-regulation and the limits of monetary policy. In 1930, the frst year of the
Great Depression, fxed investment fell by 22 percent, from 15 percent of
GDP to 13 percent of a much reduced level of output. In 2009, the frst year of
Copyright © 2021. Harvard University Press. All rights reserved.
the Great Recession, fxed investment fell by 17 percent, from 16 to 14 percent
of GDP.
Te diference in what happened next lies in how the federal government
responded to the decline in investment demand. Chapter 15 argues that the
Obama stimulus, whatever its faults, made up for the shortfall in private de-
mand, at least enough to stave of a repeat of the Great Depression. In 1932,
fxed investment was 31 percent of what it had been in 1929. By 2011, fxed
investment had recovered to 90 percent of its 2008 level.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Capital Widening: Flexprice Dynamics
We take up capital widening and capital deepening in turn. Flexprice dynam-
ics for the capital-widening case are
•
P
W
= θ ψ Ω(Y ) Y − P L K −1 − ρ − sY
−1
P 1
W
(10.1)
W
P
− θ 3 Y − LS ,
W
P
Y = 2 − FL −1 . (10.2)
W
At equilibrium the Jacobian is
P
−2
L Y P −1 L
θ1 ψ′Ω
+ θ 3 LS′ θ1 ψ′ Ω′ − − s − θ3
J = W K K W K .
θ2 −θ 2 (GS′)−1
Copyright © 2021. Harvard University Press. All rights reserved.
Without further specifcation, the signs of both the trace and the determinant
are ambiguous. A suf ficient condition for a stable equilibrium, as in Figures
10.5 and 10.6, is that wages adjust much more rapidly than prices. If 3/1 is
sufficiently large, then both 1′(P/W)−2L/K + 3LS′ and 1′{′[Y/K −
(P/W)−1L/K] − s} − 3 will be negative and therefore det J will be positive.
Te trace condition, tr J < 0, is also met if 1′(P/W)−2L/K + 3LS′ is neg
ative.
366
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Mathematical Appendix 367
Capital Widening: Fixprice Dynamics
Te fxprice case is characterized by
•
P
W
= − GS P − Y − Y − LS P , (10.3)
1 3 W
P W
W
P
−1
Y = θ 2 ψ Ω(Y ) Y − L K −1 − ρ − sY , (10.4)
W
−θ1GS′ + θ 3 LS′ θ1 − θ 3
P
−1
J = W .
K −1 + ψ′Ω′ Y − P L
−2 −1
θ ψ′Ω P L
θ 2 ψ′Ω 1 −
− s
2 W K FL
K W K
Te condition det J > 0 requires
P
−1
W −1
ψ′ Ω 1 − K −1 + Ω′ Y − P L − s
FL
θ1 − θ 3 K W K
<− (10.5)
θ1GS′ − θ 3 LS′ −2
P L
ψ′Ω
W K
stated in the text as the requirement that the slope of the stationary-price lo-
cus be algebraically smaller than the slope of the AD schedule. When the
equilibrium takes place on the upward-sloping part of the AD schedule, like E
in Figure 10.8, the term
Copyright © 2021. Harvard University Press. All rights reserved.
P
−1
W
K −1 + Ω′ Y − P L − s
−1
ψ′ Ω 1 − (10.6)
FL K W K
is necessarily negative, so the trace condition, tr J < 0, is satisfed.
For a defationary equilibrium to be stable, 2 must be small relative to 1
and 3 since 10.6 is positive. Observe that a fxprice defationary equilibrium
is not possible if 1 > 3 since this would violate the condition det J > 0.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
368 Raising Keynes
Why Does a Stable Capital-Widening Equilibrium Look like a
Cambridge Saving Equilibrium but Behave Diferently?
Qualitatively, the pictures for the Cambridge saving theory, Figures 9.16 to
9.19, are similar to the pictures for capital-widening investment, Figures 10.5
to 10.8. But the stability conditions are very diferent; under a fxprice regime
the determinant condition on the slopes of the stationary-price locus and the
AD schedule are opposite to one another.
Moreover a change in saving or investment afects the AD schedule in op-
posite directions. To see why, look at the diference between displacing the
saving or investment schedules in Figure 9.15 and displacing these schedules
in Figure 10.4: the same displacement shifs the intersection between desired
saving and desired investment in opposite directions. But because of the dif-
ference in the stability conditions, the resulting displacement, from E to F, is
in the direction of greater output in both cases, as Figure 10.14 shows.
We illustrate the general result by calculating the derivatives dY/ds for the
case of capital widening in Figure 10.14(a) and dY/ds in Figure 10.14(b). Te
AD schedule in Figure 10.14(a) is given by
P
−1
ψ Ω(Y ) Y − L K −1 − ρ − sY = 0.
W
P P
W W
Aggregate Aggregate
Supply of goods demand E F Supply of goods demand
E Stationary Stationary
real price real price
F
Copyright © 2021. Harvard University Press. All rights reserved.
Supply of labor Supply of labor
Y Y
(a) Capital-widening investment (b) Cambridge saving theory
Figure 10.14 Displacement of stable equilibrium by an increase in investment or a decrease
in saving.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Mathematical Appendix 369
So, holding P/W constant,
dY Y
= .
ds P
−1
W
K −1 + Ω′ Y − P L − s
−1
ψ′ Ω 1 −
FL K W K
Tat is, if s decreases, then Y will decrease if the AD schedule is downward
sloping, so that the denominator on the right-hand side is positive, and Y will
increase if the AD schedule slopes upward.
By contrast, the Cambridge saving theory generates an AD schedule from
the equation
P
−1
I ( ρ ) − s π Y − L = 0.
W
We have
1
P
Y L
dY W
1
,
ds P
W
1
FL
so that dY/ds is negative when the denominator is positive, which is to say,
when the AD schedule lies to the lef of the GS schedule and is downward
sloping. When the AD schedule slopes upward, the denominator is negative,
and dY/ds is positive.
Te displacement of equilibrium is calculated by diferentiating the entire
system of equations. In the case of capital widening with a uniform propen-
sity to save, the system is
Copyright © 2021. Harvard University Press. All rights reserved.
P P
−1 GS − Y − 3 Y − LS = 0, (10.7)
W W
P
−1
ψ Ω(Y ) Y − L K −1 − ρ − sY = 0. (10.8)
W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
370 Raising Keynes
From 10.7, we have
P
d
W = 1 − 3 dY
, (10.9)
ds 1GS′ − 3 LS′ ds
and from 10.8 we have
P
−1
W −1
′
ψ Ω 1 − K −1 + ψ′Ω′ Y − P L − s dY
FL ds
K W K (10.10)
P
d
L W
−2
P
+ ψ′Ω − Y = 0.
W K ds
Substituting from equation (10.9) into equation (10.10) and solving for dY/ds
we obtain
Y
P
−1
W
K −1 + ψ′Ω′ Y − P L − s
−1
ψ′Ω 1 −
FL
dY K W K
= −2
.
ds P L
ψΩ
W K θ1 − θ 3
1+
P
−1
θ1GS′ − θ 3 LS′
W
K −1 + ψ′Ω′ Y − P L − s
−1
ψ′Ω 1 −
FL K W K
Copyright © 2021. Harvard University Press. All rights reserved.
At the equilibrium E in Figure 10.14(a), on the upward-sloping part of the
AD schedule, the expression
P
−1
W −1
ψ′Ω 1 − K −1 + ψ′Ω′ Y − P L − s
FL
K W K
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Mathematical Appendix 371
is negative, and by virtue of inequality (10.5) we have
−2
P L
ψ′Ω
W K θ1 − θ 3
1+ > 0.
P
−1
θ1GS′ − θ 3 LS′
W
K −1 + ψ′Ω′ Y − P L − s
−1
ψ′Ω 1 −
FL K W K
So dY/ds < 0. Tat is, taking both the stationary-price locus and the AD
schedule into account, the impact of a decrease in the saving propensity is to
raise the equilibrium level of output. It follows from equation (10.9) that the
impact on the real price is also positive.
With the AD schedule based on the Cambridge saving equation (which
replaces equation [10.8]), the efect of displacing equilibrium is
P
−1
Y − L
W
P
−1
1 − W
dY FL
=− .
−2
ds π P
W L θ1 − θ 3
1 + −1
P θ 1GS ′ − θ 3 LS′
1 − W
FL
At the stable equilibrium E in Figure 10.14(b),
Copyright © 2021. Harvard University Press. All rights reserved.
P
−1
1 − W > 0,
FL
so both the numerator and the denominator are positive. Tus in this case too
dY/ds < 0; from equation (10.9) we also have d(P/W)/ds < 0.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
372 Raising Keynes
Capital Deepening
Te frst order of business is the shape of the investment-demand schedule in
Figure 10.9(a). We have
P −1 F
I = ψ K
− ρ .
W FL
We start from the production function Y = F(K, L), which we assume refects
constant returns to scale and diminishing marginal productivities. We defne
l ≡ L/K so that dividing through by K, we can work with f(l) = F(1, L/K) = Y/K.
Diferentiating f(l)K = F(K, L) with respect to L and K, we have
f l = FL and f − f l l = FK.
We now rewrite investment demand as
P −1 f − f l l
I = ψ − ρ
W fl
and obtain
∂I f f l l P −1
= −′ , (10.11)
( fl ) K W
3
∂Y
which is positive. (f l l < 0 by virtue of diminishing marginal productivities.)
But whether ∂I/∂Y is itself increasing with Y or decreasing depends on the
elasticity of substitution in production, . By defnition, is the percentage-
rate change in the labor:capital ratio corresponding to a one percent change
in the ratio of factor marginal productivities, (f − f l l ) / f l ; that is,
Copyright © 2021. Harvard University Press. All rights reserved.
dl ( f − f l l) / f l
= .
d ( f − f l l ) / f l l
Expanding the derivative gives
fl ( f − fl l ) FL FK
=− = .
f l fl l YFLK
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Mathematical Appendix 373
Production functions with a constant elasticity of substitution have the ge-
neric form
1
F (K , L) = A ( λ1K ζ + λ2 Lζ ) ζ ,
1
(
f (l ) = A λ1 + λ2l ζ ) ζ,
with 1 = 1 − 2, and = 1/(1 − ). → 0 as → −∞; → ∞ as → 1; and
< 1 ⇔ < 0.
We can now determine the shape of the investment-demand schedule by
rewriting equation (10.11) as
∂I I + ψ′ρ
=
∂Y σf l lK
and diferentiating to obtain
∂2I
∂I
( )
f l − f l l l + f l (I + ψ′ρ) ( f l K )
1 ∂Y l
−1
=
∂Y 2 ( fl l )
2
σK
(10.12)
(I + ψ′ρ) 1− σ
= fl − fll l .
( fl l )
2
σK 2
fl σ
Te sign of ∂2I/∂Y2 depends on the elasticity of substitution. If < 1, then
∂2I/∂Y2 > 0, which is to say that the slope of the investment-demand schedule
increases as Y increases, as is assumed in Figure 10.9(a).
Consequently, the AD schedule in Figure 10.9(b) is upward sloping. To see
this, start from
P −1 f − f l l
ψ − ρ − sY = 0. (10.13)
Copyright © 2021. Harvard University Press. All rights reserved.
W fl
Diferentiating equation (10.13) gives the slope of the AD schedule:
P I + ψ′ρ P
d −s
W = σf l lK W 1 sY fl l
= − .
dY −2
P f − fl l f l lK σ I + ψ′ρ f
W fl
Since sY/(I + ′) and f l l / f are both less than one, the entire expression ex-
ceeds one if < 1.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
374 Raising Keynes
Flexprice-adjustment equations are
•
P
W
= θ ψ P f − f l l − ρ − sY − θ Y − LS P ,
−1
(10.14)
P 1
3 W
W fl
W
P −1
Y = 2 − ( f l ) . (10.2)
W
Te Jacobian is
P
−2
f − fl l I + ψ′ρ 1 sY fl l
−θ1 ψ′ + θ 3 LS′ θ1 − − θ3
J =
W fl f l lK σ I + ψ′ρ f .
P
θ2 −θ 2 (GS′)−1
W
Te trace is negative and the determinant is positive provided
I + ψ′ρ 1 sY fl l
θ1 − − θ3
f l lK σ I + ψ′ρ f (GS′)−1
−2
< .
P f − fl l P
θ1 ψ′ − θ 3 LS′ W
W fl
Tat is, the stationary real-price locus must be less steep than the GS sched-
ule, as it is at the equilibrium in Figure 10.10.
Te adjustment equations for a fxprice regime are
•
P
W
= − GS P − Y − Y − LS P , (10.3)
P 1 3 W
W
W
Copyright © 2021. Harvard University Press. All rights reserved.
P −1 f − f l
Y = θ 2 ψ l
− ρ − sY . (10.15)
W fl
Te Jacobian is
−θ1GS′ + θ 3 LS′ θ1 − θ 3
−2
J = P f − fl l I + ψ′ρ 1 sY fl l .
− θ ψ′
2 W θ2 −
fl f l lK σ I + ψ′ρ f
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
10 • Mathematical Appendix 375
Te trace is negative provided that, compared to output, wages and prices
ficiently rapidly (1 and 3 both large), so that
adjust suf
I + ψ′ρ 1 sY fl l
θ2 − < θ1GS′ − θ 3 LS′.
f l lK σ I + ψ′ρ f
Te other condition for stability, det J > 0, will be met provided that (1) prices
adjust more rapidly than wages (1 − 3 > 0) and (2) the stationary price locus
is steeper than the AD schedule,
I + ψ′ρ 1 sY fl l
−
f l lK σ I + ψ′ρ f θ1 − θ 3
< .
−2
P f − fl l θ1GS′ − θ 3 LS′
ψ′
W fl
In Figure 10.12, E, but not E′, satisfes both (1) and (2).
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 11 .
the theory of interest, i
Liquidity Preference in a World of Money and Bonds
Tere is, I am convinced, a fatal faw in that part of the orthodox
reasoning which deals with the theory of what determines the level of
efective demand and the volume of aggregate employment; the faw
being largely due to the failure of the classical doctrine to develop a
satisfactory theory of the rate of interest.
—john maynard keynes
Money, it is well known, serves two principal purposes. By acting as a
money of account it facilitates exchanges without its being necessary
that it should ever itself come into the picture as a substantive object.
In this respect it is a convenience which is devoid of signifcance or
real infuence. In the second place, it is a store of wealth. So we are
told, without a smile on the face. But in the world of the classical
economy, what an insane use to which to put it! For it is a recognized
characteristic of money as a store of wealth that it is barren; whereas
practically every other form of storing wealth yields some interest or
proft. Why should anyone outside a lunatic asylum wish to use
money as a store of wealth?
Because, partly on reasonable and partly on instinctive grounds,
our desire to hold Money as a store of wealth is a barometer of the
degree of our distrust of our own calculations and conventions
concerning the future. Even tho this feeling about Money is itself
Copyright © 2021. Harvard University Press. All rights reserved.
conventional or instinctive, it operates, so to speak, at a deeper level
of our motivation. It takes charge at the moments when the higher,
more precarious conventions have weakened. Te possession of actual
money lulls our disquietude; and the premium which we require to
make us part with money is the measure of the degree of our disqui-
etude.
Te signifcance of this characteristic of money has usually been
overlooked; and in so far as it has been noticed, the essential nature
of the phenomenon has been misdescribed. For what has attracted
attention has been the quantity of money which has been hoarded;
and importance has been attached to this because it has been
supposed to have a direct proportionate efect on the price-level
through afecting the velocity of circulation. But the quantity of
376
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 377
hoards can only be altered either if the total quantity of money is
changed or if the quantity of current money-income (I speak broadly)
is changed; whereas fuctuations in the degree of confdence are
capable of having quite a diferent efect, namely, in modifying not
the amount that is actually hoarded, but the amount of the premium
which has to be ofered to induce people not to hoard. And changes
in the propensity to hoard, or in the state of liquidity-preference as I
have called it, primarily afect, not prices, but the rate of interest; any
efect on prices being produced by repercussion as an ultimate conse-
quence of a change in the rate of interest.
—keynes
Te initial novelty [of Te General Teory] lies in my maintaining that
it is not the rate of interest, but the level of incomes which ensures
equality between saving and investment. Te arguments which lead
up to this initial conclusion are independent of my subsequent theory
of the rate of interest, and in fact I reached it before I had reached the
latter theory. But the result of it was to leave the rate of interest in the
air. If the rate of interest is not determined by saving and investment
in the same way in which price is determined by supply and demand,
how is it determined?
Te resulting theory, whether right or wrong, is exceedingly
simple—namely, that the rate of interest on a loan of given quality
and maturity has to be established at the level which, in the opinion
of those who have the opportunity of choice—i.e. of wealth-holders—
equalizes the attractions of holding idle cash and of holding the loan.
—keynes
No part of Te General Teory has caused readers more grief than Keynes’s
theory of interest. It is chock full of ambiguities and in the end—I will argue
in the next chapter—is at best incomplete: it is not a theory of the rate of in-
terest but a theory of interest spreads. Moreover, as the last epigraph to this
Copyright © 2021. Harvard University Press. All rights reserved.
chapter indicates, it is an aferthought. Having rejected the mainstream view
that the rate of interest equilibrates desired saving and desired investment,
Keynes had to ofer an alternative.
Or did he? Could Keynes not have simply posited a central bank in control
of the interest rate and done away with the need for a novel theory of interest?
Yes and no. Yes, to the extent that his goal was to provide an operating manual
for an economy in which the need for a governmental hand on the tiller is
already recognized, as in the frst-pass model elaborated in chapter 3. No, to
the extent that the point was to demonstrate the need to steer the economy
rather than to rely on putative self-regulating forces to maintain full employ-
ment. Te theory of interest Keynes ofers really c omes into its own in the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
378 Raising Keynes
second-pass model, which was intended to serve the second of these two
overlapping purposes—and of course in the third-pass model, which I intend
to serve both purposes.
Keynes’s interest theory departs dramatically from the mainstream: it re-
jects the idea that the interest rate is determined by the forces of productivity
and thrif, the forces that govern the fows of desired saving and investment:
Te rate of interest is not the “price” which brings into equilibrium the de-
mand for resources to invest with the readiness to abstain from present
consumption. It is the “price” which equilibrates the desire to hold wealth in
the form of cash with the available quantity of cash. (Te General Teory,
p. 167)
Keynes likely believed that, once stated, this dictum about where and how
interest rates are determined—by demands and supplies in asset markets, not
by investment and saving—is obvious. Probably the sheer size of asset mar-
kets, being markets in stocks, compared with the fows of investment and sav-
ing, was supposed to convince the reader. For me the stock versus fow argu-
ment is indeed persuasive, and even more persuasive is the volume in trading
of fnancial assets relative to the volume of investment and saving. As chapter
2 noted, in 2017 U.S. saving and investment were approximately $3.4 trillion.
By contrast, the daily volume of trading in bond markets was over $750 bil-
lion. Assuming 250 trading days, the annual trading volume works out to
$187.5 trillion.
In any case, Te General Teory ofers no justifcation for the shif from in-
vestment and saving to asset markets. Te result was a controversy, “liquidity
preference versus loanable funds,” that has never been fully settled. One rea-
son is that loanable funds is a slippery concept. What are loanable funds, ac-
cumulated saving, that is, wealth, or current saving, this year’s additions to
wealth? If loanable funds are wealth, then there is no conceptual confict with
Keynes’s foundational assumption that the interest rate clears the market for
Copyright © 2021. Harvard University Press. All rights reserved.
fnancial assets. If loanable funds correspond to current saving, then there is
no daylight between the equilibrium in the loanable funds market and the
equilibrium determined by desired saving and desired investment. Frictions,
rigidities, and other imperfections apart, Say’s Law is alive and well, and with
it so is the rejection of a role for aggregate demand.1
Te controversy over whether fnancial-asset markets or markets for saving
and investment are the site for interest-rate determination can be understood
as an argument about relative speeds of adjustment of the level of output and
the rate of interest when saving and investment do not balance. If Y is the
speed at which output adjusts and is the speed at which the interest rate
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 379
adjusts, the mainstream view is that we live in a world of the limiting case
where Y / → 0. Te interest rate does the heavy lifing of adjusting ID and
SD. Keynes rather assumes the opposite limiting case, Y / → ∞. In Te Gen-
eral Teory interest rates, kept busy adjusting asset demands and asset sup-
plies, react with glacial slowness to imbalances between desired investment
and desired saving. In consequence the heavy lifing is lef to the adjustment
of output. In the limiting case output adjusts so much faster to imbalances in
saving and investment that we can ignore the interest rate as a factor in ad-
justing investment and saving. And focus entirely on asset markets.2
Te saving-investment view of interest-rate determination, that is, the lim-
iting case Y / → 0, is certainly a logical possibility. One way for this to hap-
pen is that there be complete segmentation between the fows of saving and
investment and the stocks of fnancial assets. In this situation it is not a ques-
tion of whether loanable funds or liquidity preference determines the interest
rate; instead there are two interest rates, with loanable funds determining one
in balancing fows of desired investment and saving, and liquidity preference
determining the other in balancing demands and supplies for the stock of f
nancial assets.
An extreme case of segmentation would be that there is no market for f
nancial assets at all. Tis is perhaps the world that Adam Smith had in mind
when he argued his version of Say’s Law:
Whatever a person saves from his revenue he adds to his capital, and either
employs it himself in maintaining an additional number of productive
hands, or enables some other person to do so, by lending it to him for an
interest, that is, for a share of the p
rofts. (1937 [1776], p. 321)
In principle, there need be no market for the bond that the entrepreneur is-
sues to the saver. Te saver may have no choice—and possibly no desire—but
to hold the bond until it matures.
Tis would leave the rate of interest unemployed in asset markets, and so
Copyright © 2021. Harvard University Press. All rights reserved.
free to balance investment and saving. Te grain of truth in this view was
noted by Jacob Viner in his critique of Te General Teory.
Tere are, however, in every country large numbers of investors who have
been taught to buy gilt-edge securities on the basis of their yield to maturity
and to disregard the fuctuations in their day-to-day market values. (1936,
p. 153)
Viner’s observation is as valid in the twenty-frst century as it was in the
1930s. In a world of cash and bonds, a fnancial planner might counsel her
younger clients to hold portfolios that mature over the expected span of their
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
380 Raising Keynes
golden years, with nary a side glance to cash.3 Tese agents have no need for
the liquidity that cash ofers and no need for a bond market once they have
made their initial commitments.
Te disproportionate volume of trading in the most recently issued T-
bonds, so-called “on-the-run” bonds, likely refects the importance of Viner-
type agents (or in 2020 the pension funds that act on behalf of these agents).
Trading in newly issued, on-the-run, bonds with maturities from two to ten
years dwarfed the volume of trading in of-the-run securities in Michael
Fleming’s data (2013), accounting for over 60 percent of all the trading in
Treasury securities. Presumably, this disparity refects a signifcant presence
of agents who buy bonds when issued and hold them to maturity.
But the grain of truth in the diference in on-the-run and of-the-run trad-
ing does not really diminish the importance of the market in fnancial assets
as the site for determining interest rates. Even the 40 percent of this volume
that represents trading in of-the-run long-term bonds and short-term bills
dwarfs saving and investment by a factor of more than twenty.4
If segmentation is a logical possibility, there is considerable evidence to the
contrary. In the eighteenth century, at the dawn of fnancial capitalism, gov-
ernment bonds and shares in the companies that had monopolies over their
countries’ overseas trade—the Dutch and British East India Companies are
the most famous—were frequently used as collateral for loans to fnance new
undertakings (Carlos and Neal 2011, passim). Tis would suggest a certain
degree of integration of the market for existing assets with the market for new
investment. Given this integration, it seems implausible that even in the 1700s
capital markets were so compartmentalized that one interest rate (or, more
realistically, one set of interest rates) was determined by the demand for in-
vestment and the supply of saving, and another was determined in bond and
share markets.
Even if there were merit in the segmentation hypothesis as history, by the
time Keynes wrote Te General Teory, capital markets were highly inte-
Copyright © 2021. Harvard University Press. All rights reserved.
grated. To imagine capital markets so segmented that interest rates had no
other role to play than to adjust the demand for investment and the supply
of saving is out of touch with reality, Keynes’s reality of the frst decades of
the twentieth century, not to mention the reality of the frst decades of the
twenty-frst century.
But does it matter whether we focus on the investment-saving nexus or on
the market for fnancial assets? Early on, Frank Knight dismissed the liquidity
preference versus loanable funds controversy as a tempest in a teapot:
It is self-evident that at any time (and at the margin) the rate of interest
equates both the desirability of holding cash with the desirability of holding
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 381
nonmonetary wealth and the desirability of consuming with that of lending.
(1937, pp. 112–113)
John Hicks agreed with Knight: Value and Capital argues that the choice be-
tween liquidity preference, with the interest rate being determined in markets
for fnancial assets, and loanable funds, with the interest rate determined by
the demand for investment and the supply of saving, is simply a matter of
choosing a numéraire (1946 [1939], chap. 12).
A reviewer of a preliminary draf of this current book went even further.
Since fows of investment and saving are simply changes over time in stocks of
assets, the fows must be equal if asset markets are continuously in equilib-
rium. If DA = SA continuously, where DA and SA are the demand for and supply
of fnancial assets, then in a closed economy without government, ID = S A and
SD = D A. An implication of continuous equilibrium in asset markets is equal-
ity of demand for investment and the supply of saving.
Knight, Hicks, and my anonymous referee are undoubtedly correct. Te
simplifed model of Te General Teory view of adjustment to inequality be-
tween ID and SD presented in chapter 2 and the more complete model in the
appendix to chapter 3 both end up with ID and SD equal at the equilibrium
rate of interest.
But this misses the point. As Lawrence Klein pointed out long ago:
Nothing has been proved by this argument. It does not tell whether the rate
of interest is the mechanism which allocates funds into idle hoards as op-
posed to earning assets or which brings the supply and demand for [invest-
ment] into equilibrium. (1947, pp. 118–119)
I should be clear as to what is at stake. From one point of view, a lot. Te im-
plication of Keynes’s dictum that interest rates are determined in asset mar-
kets is that the level of output rather than the rate of interest adjusts the de-
mand for investment with the supply of saving. It is no exaggeration to say
Copyright © 2021. Harvard University Press. All rights reserved.
that this is the basis of his entire theory.
From a narrower perspective, not very much changes: that the rate of inter-
est is determined in the market for stocks of fnancial assets hardly eliminates
the possibility—nor does Keynes deny—that the rate of interest might infu
ence fows of both investment and saving. Indeed, Keynes’s theory of invest-
ment is taken over lock, stock, and barrel from Irving Fisher, and the rate of
interest is the determinant of investment once the array of potential invest-
ment projects is given. (In one sense, Keynes, in emphasizing the psychologi-
cal dimension of estimates of the returns of a given array of projects—animal
spirits—departs signifcantly from Fisher.) And, while the infuence of the
rate of interest on saving is not emphasized, the only diagram in Te General
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
382 Raising Keynes
Teory (p. 180) illustrates joint dependence of saving on the rate of interest
and income.
Causality running from the rate of interest to investment and saving is not
the sticking point between Keynes and the mainstream. Te sticking point is
rather the idea that causality runs in the other direction also, from saving and
investment decisions to the rate of interest, with the rate of interest the adjust-
ment mechanism for bringing saving and investment decisions into line with
each other. For Keynes the arrow running from saving and investment deci-
sions to the rate of interest is broken, so the rate of interest cannot harmonize
investment and saving desires. As he puts it in the last epigraph to this chap-
ter, the rate of interest becomes superfuous in bringing about equilibrium
between investment and saving once the equilibrating function of the level of
output is understood.
Liquidity Preference: The Basics
In addition to shifing the focus of interest-rate determination from fows of
investment and saving to stocks of fnancial assets, Keynes ofers a specifc
theory of asset-market equilibrium, his theory of liquidity preference. Chap-
ter 3 summarized Keynes’s theory of interest in a function ( ) that relates
the desire to hold cash as a fraction of total wealth to the rate of interest on
bonds. Total nominal assets are A, the demand for money in wealth portfolios
is M2, and the interest rate on bonds is . Te demand for money as a fraction
of total wealth, M2/A = ( ), is pictured in Figure 11.1. Te demand for
money in wealth portfolios is simply M2 = ( )A. Chapter 3 translated the
relationship between desired cash holdings and total wealth into a relation-
Copyright © 2021. Harvard University Press. All rights reserved.
M2
( )
A
min
1 M2
A
Figure 11.1 Liquidity preference as a function of the interest rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 383
ship between desired cash and the value of bonds. If there is no government
debt and no other fnancial asset (like equity), and we assume that physical
capital is monetized in the form of bonds, then the stock of bonds, B (each
with a par value of one dollar), corresponds to the nominal value of physical
capital, PK. If each bond provides an annual payment of R, R is the interest
rate on these bonds at par, but the interest rate consistent with equilibrium in
the market for fnancial assets, , is determined by the interaction of the value
of bonds, PBB; the stock of money available for wealth portfolios, M 2; and the
liquidity-preference function ( ). We have
A = PB B + M 2 = PB PK + M 2.
For consols—bonds for which R is fxed in perpetuity—the relationship
between the price of the bonds and the interest rate is particularly simple: the
price of a consol is the ratio of R to , PB( ) = R/, as in Figure 11.2.5 If we put
Figures 11.1 and 11.2 together, we can determine the interest rate that clears
the market for fnancial assets. Te demand for money as an asset, M2, is
given by
M 2 = β( ρ) PB ( ρ)B + M 2 ,
and asset-market equilibrium requires M2 = M 2 so that
β( ρ)
M2 = M2 = PB ( ρ)B .
1 − β( ρ)
Copyright © 2021. Harvard University Press. All rights reserved.
R
PB ( )
PB
Figure 11.2 Bond price and interest rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
384 Raising Keynes
Asset-market equilibrium is pictured in Figure 11.3, with the liquidity-
preference schedule the locus of intersections of the money-demand and
money-supply schedules.
Te budget constraint requires that the sum of the demand for bonds (PBB,
as distinct from PBB) and the demand for money equals the sum of the en-
dowments:
M 2 + PB ( )B = M 2 + PB ( )B.
So equality of the demand for and the supply of money implies equality of the
demand for and the supply of bonds.
What happens when the price level of goods, P, changes? In the models
deployed in this book, the price of physical capital changes in lockstep since
capital is just a stock of congealed goods. In chapters 3 and 4, where the focus
was on comparative statics, the appropriate assumption was that the quantity
of bonds varies with the goods-price level. If two planets difer only in their
respective goods-price levels, then the planet with the higher price level will
have a proportionately larger volume of bonds outstanding.
In the dynamic context of chapters 5 to 8, where the focus is on real-time
changes, there is no way for the number of bonds to vary as the goods-price
level varies. Te appropriate assumption is rather that the quantity of bonds is
M2 1 00 M2 325
M2 ( ) PB ( ) B 325
Copyright © 2021. Harvard University Press. All rights reserved.
Liquidity preference
( )
M2 ( ) PB ( ) B 100 M2 M2 PB ( ) B
min 1 ( )
100 200 300 400 500 M2 , M2
Figure 11.3 Equilibrium between portfolio cash and the interest rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 385
fxed by the capital stock and the goods-price level at t = 0 and that it is invari-
ant with respect to whatever goods-price level happens to obtain at any later
time t.
Tree questions present themselves. First, why is positive rather than
equal to zero? In other words, why do people hold any cash at all in their asset
portfolios given that bonds pay interest and cash does not? (Why d on’t the
schedules in Figures 11.1 and 11.3 coincide with the vertical axis?) Second,
why is the amount of cash that agents wish to hold inversely related to the
interest rate? (Why do the schedules in Figures 11.1 and 11.3 slope down-
ward?) Finally, why does liquidity preference become absolute at a positive
rate of interest min, so that the demand for cash becomes infnite at a positive
rate of interest? (Why does the money-demand schedule reach its maximum
of one in Figure 11.1 before approaching the vertical axis, and why does the
rate of interest in Figure 11.3 approach a positive lower bound?)
Transactions, Precaution, Retirement, and Speculation
Te fact that money is a medium of exchange explains why agents hold cash
in anticipation of transactions. If you plan a trip to Mexico or India, you will
need pesos or rupees. In the United States, we need dollars. We can’t pay our
bills with bonds or other assets. But money’s role as the medium of exchange
does not explain liquidity preference. Te question here is a diferent one:
why do agents hold cash in their asset portfolios, geared not to present trans-
actions but to future ones.
One possible answer is the need for reserves against a rainy day; fnancial
planners routinely advise clients to hold a precautionary cash reserve. But this
is not an answer to the question posed in the epigraph to this chapter: why
does any sane person hold cash? Why not bonds or other assets? Again, why is
positive?
Te premise of the precautionary motive is that the time when cash is
Copyright © 2021. Harvard University Press. All rights reserved.
needed might be an extremely bad time to sell assets. As a Wells Fargo fnan
cial-planning website once put it,
[since] the unexpected can always be expected, having a cash reserve ready
is essential, [and] a good rule of thumb . . . is to have six to 18 months of
cash to cover your living expenses and any other unexpected costs.
Wells Fargo advised its clients accordingly:
In the Wells Fargo asset allocation strategy, across the board we hold a 3
percent allocation to cash . . .
Cash is cash . . . If it’s described as “like cash” then it’s not cash, and that’s
how people got themselves in trouble before 2008. (Wells Fargo 2013)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
386 Raising Keynes
( )
M2
0.0 0.03 1
A
Figure 11.4 Precautionray demand for cash as a function of the
interest rate.
As far as the precautionary motive is concerned, the Wells Fargo view is
well-nigh universal.6 I have never come across a fnancial planner who fac-
tored the rate of interest into his calculations of the appropriate size of a pre-
cautionary cash reserve, so that, if we represent Wells Fargo’s 3 percent rule in
Figure 11.4, we should see a vertical line, not a schedule along which the
proportion of wealth held in liquid form varies inversely with the interest rate
and reaches unity at a positive level of the interest rate.
Te very lack of interest sensitivity makes the precautionary demand for
cash more like the transactions demand. And indeed, Keynes merges the two,
arguing,
In normal circumstances the amount of money required to satisfy the
transactions-motive and the precautionary-motive is mainly a resultant of
Copyright © 2021. Harvard University Press. All rights reserved.
the general activity of the economic system and of the level of money-
income. (Te General Teory, p. 191)
His formal statement of the demand for money echoes this assimilation of pre
cautionary and transactions demands and their separation from what he calls
the speculative demand on the basis of what drives the demand for money:
Let the amount of cash held to satisfy the transactions-and precautionary-
motives be M1, and the amount held to satisfy the speculative-motive be M2
. . . Tus
M = M1 + M2 = L1(Y) + L2(r),
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 387
where L1 is the liquidity function corresponding to an income Y, which de-
termines M1, and L2 is the liquidity function of the rate of interest r, which
determines M2. (Te General Teory, pp. 199–200)
Money held for precautionary reasons is technically part of wealth, as is
transactions money to the extent it is not ofset by debt. But money held for
these reasons is fundamentally diferent from money held for speculative pur-
poses. You and I hold small amounts of liquid assets to buy groceries and gas,
and larger amounts to prepare for a rainy day on which our jobs disappear,
illness strikes, or some other calamity occurs. And—a motive Keynes does
not analyze—we hold bonds (and other fnancial assets) to provide for our
retirement. In a cash versus bonds model, neither precaution nor retirement
will lead us to be sensitive to the interest rate in designing the composition of
our portfolios. Rainy-day funds will be entirely in cash, and retirement funds
entirely in bonds, provided the bond maturities can be matched to an indi-
vidual’s expected time of retirement and life expectancy.
By contrast, Keynes’s speculators derive a beneft in having liquid assets
available to exploit targets of opportunity if and when they arise, like the op-
portunity to buy fnancial or physical assets on the cheap, or to stave of
creditors if a business venture goes sour. Keynes’s speculators are always cal-
culating whether the interest potential on bonds is suf ficient to induce them
to limit access to their capital. Teir answer—and therefore the composition
of their portfolios—will depend on the return to illiquidity.
In this view, speculators drive the market since other agents will always
choose either cash or bonds. Portfolios designed in the light of the precaution-
ary motive will contain only cash. Te retirement motive (or planning for a
child’s college education) produces Viner-type agents whose focus is on the
wealth available at some well-defned period in the future; in the present
model, with assets limited to cash and bonds, these agents will hold only
bonds.7
Copyright © 2021. Harvard University Press. All rights reserved.
Two Notions of Liquidity
Tere are two distinct reasons why cash is an attractive asset, not only when it
comes to a reserve for emergencies but also when it comes to being prepared
to deal with fnancial targets of opportunity. First, noncash assets may difer
widely in their liquidity in the sense of the spread between bid and ask prices,
that is prices that potential buyers are prepared to ofer and prices that sellers
are prepared to accept. (Call this “bid-ask” liquidity.) If you have to sell a boat,
a second home, a rental property, or a factory to meet an unexpected need for
cash, a relatively thin market may lead to a wide divergence between the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
388 Raising Keynes
prices ofered and the prices you would expect if you w eren’t in a rush to sell.
You know that if you wait, a buyer willing to pay the asking price, or very
near, will very likely come forward, but you also know that you can’t wait:
you need the money now. Tis is why, as the Wells Fargo website (2013) put it,
a cash reserve “prevents you from making short-term decisions with long-
term assets.”
But this reasoning d oesn’t apply to, say, U.S. T-bonds or shares of IBM
stock. In the sense of the spread between bid and ask, these securities are ex-
tremely liquid. In September 2014, the bid-ask spread on 30-year T-bonds
falling due ten years hence (in 2024) was $0.70 on a bond with a price tag of
about $1,450, approximately one-twentieth of 1 percent of the price of the
bond, and this spread was considerably higher than the spreads on recently
issued, on-the-run, 10-year bonds. Moreover, these are retail spreads for rela-
tively small purchases and sales.
Tere is, however, a second meaning of liquidity, namely, the possibility of
a capital loss on the asset, and this less-intuitive meaning is the one that
Keynes has in mind when discussing his theory of interest.8 (Call this price-
fuctuation liquidity.) If interest rates go up, the price of a bond with a fxed
payment R will decline, and vice versa—even if we assume that it has zero
default risk. So even though the market value at maturity is perfectly certain,
a T-bond can fuctuate widely in value in the interim before it matures. It is
perfectly liquid in the frst (bid-ask) sense of liquidity but illiquid in terms of
preservation of capital value. For this reason, wealth holders may be willing to
forego the returns it ofers, opting instead to hold cash.
Tere is an important diference between these two notions of liquidity,
apart from the diference in relation to ordinary language. Bid-ask liquidity
does not require that the value of an asset (measured by the average of bid and
ask prices) vary over time for the bid-ask spread to be a factor, even a decisive
factor, in portfolio management. An agent doesn’t have to speculate about
whether or not she will take a hit if she must quickly sell illiquid assets for
Copyright © 2021. Harvard University Press. All rights reserved.
which the bid-ask spread is great. She does not face a risk of capital loss but
rather a virtual certainty.
Price fuctuations are another matter altogether. If we assume that gains
and losses in capital value are equally likely, an agent who has to sell in a
hurry is as likely to realize a windfall as to suffer a loss. Nonetheless, on stan-
dard models of utility maximization, in which wealth ofers diminishing mar-
ginal utility, a gain of one dollar will count for less than the loss of one dollar.
Tat is, agents will be risk averse, and their risk aversion may be suf ficient to
justify placing a portion of one’s wealth in cash even in the absence of bid-ask
liquidity concerns.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 389
If average bond prices are not expected to be stationary, but rather to fall
over time, wealth holders have another reason for holding cash. Assuming no
default risk, it pays to forego bond interest today if tomorrow the bond can be
bought at a discount to today’s price, provided the expected reduction in price
is greater than the foregone interest.
Default risk adds another dimension. Indeed, default risk is a special case
of price fuctuations, in which the bond price falls dramatically, to zero in the
limit.
To be sure, Keynes’s frst reason for liquidity preference is formally compat-
ible with both bid-ask liquidity and price-fuctuation liquidity:
A need for liquid cash may . . . arise before [a bond falls due . . . Conse-
quently,] there is a risk of loss being incurred in purchasing a long-term debt
and subsequently turning it into cash, as compared with holding cash. (Te
General Teory, p. 169)
Te preference for cash comes from “the risk of loss” associated with long
term debt, and loss can logically come about either from bid-ask illiquidity or
price-fuctuation illiquidity. But in deference to the realities of bond markets,
particularly the market for Treasuries, we can put bid-ask liquidity to one side
in order to focus on price fuctuations. Keynes implicitly assumes—and we
will too—that all fnancial assets are bought and sold in dense markets with
only small spreads between bids and ofers.9
Te question then is how price fuctuations explain why the liquidity-
preference schedules in Figures 11.1 and 11.2 slope downward, which is to
say, why agents typically desire to hold a larger portion of their portfolios in
cash the lower the interest rate.
Insurance and Risk Aversion
Copyright © 2021. Harvard University Press. All rights reserved.
In Te General Teory Keynes deploys an analogy between interest payments
and insurance to explain why people might wish to hold increasing amounts
of cash as interest rates fall:
Every fall in [the interest rate] reduces the current earnings from illiquidity
[that is, from holding wealth in the form of long term bonds], which are
available as a sort of insurance . . . to ofset the risk of loss on capital account.
(p. 202)
Te analogy suggests that it is risk aversion that makes price fuctuations
problematic, for there is an obvious link between insurance and risk aversion.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
390 Raising Keynes
But, as we shall see, the apparatus we develop to explain cash holdings in
terms of risk aversion also helps to understand other motives for holding cash.
Understanding the idea of interest payments as insurance requires us to
distinguish three measures of the return on bonds. Te frst is the coupon
yield, coup ≡ R/PB, measured by the ratio of the annualized periodic payment,
the “coupon,” R, to the price of the bond, PB.10 Te second is the yield to matu
rity, mat , which is the solution to the following equation,
T
R ∑ ( 1 + mat ) + ( 1 + mat )
−t −T
= PB,
1
that is, the yield to maturity is the discount rate that makes the present value
of returns on a bond, including the return of principal (assumed here to be
one) at maturity in year T, equal to the bond price today. In continuous time,
the equation becomes
T
∫ 0 Re
− matt − matT
dt + e = PB.
Finally, the holding yield, hold, the return on holding the bond for one time
period, say one year, is the sum of the coupon yield and the percentage
change in the price of the bond,
R ∆PB
ρhold ≡ + .
PB PB
In continuous time, the holding yield is the instantaneous return, which for
convenience we continue to measure at an annual rate,
R P
hold ≡ + B.
PB PB
Evidently, the coupon yield, coup , and the yield to maturity, mat , are re-
Copyright © 2021. Harvard University Press. All rights reserved.
lated. Indeed, consols have the convenient property that the coupon yield is
exactly equal to the yield to maturity.11 Since for all bonds the percentage
change in the price of a bond is equal to the percentage change in its coupon
yield, for consols the percentage change in the price of the bond is also the
percentage change in the yield to maturity.
Why do we focus on the return from holding a bond for one year (or for
an instant in continuous time)? Why one year (or one second) and not ten
years? A short-term focus is not a self-evident given. Indeed, a more sophisti-
cated model of speculation might take into account more explicitly the uncer-
tainty surrounding the time when speculators might need to access their
portfolios.12
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 391
But in order to get on with the story, we leave this point to one side. I shall
simply follow Keynes and much of the subsequent literature, which implicitly
assumes that speculators, who do not know when they might need to access
their wealth, focus on the short term.
Suppose now that P B/PB (or PB/PB) is a random variable with mean zero,
which is to say that capital gains and capital losses are equally likely. Tis as-
sumption makes sense if we are dealing with consols and if there is no expec-
tation that interest rates will change. Suppose further that the probability dis-
tribution of P B/PB is independent of PB, and therefore of R/PB (since R is fxed
once and for all at the time bonds are issued). Since we are modeling portfolio
choices, what matters are expectations; these expectations may or may not be
based on frequency distributions of past performance and may be idiosyn-
cratic or widely shared.
We can picture the resulting distribution of capital gains and losses in the
frst two quadrants of Figure 11.5. Te third quadrant shows the breakeven
coupon yield, the combinations of price changes, P B/PB, and coupon yield,
R/PB, that makes the holding yield, hold , just equal to zero. For any level of the
coupon yield, this schedule tells us the maximum rate at which the bond price
can fall without making the holding yield negative.
Tis maximum rate determines the “uninsured” portion of the probability
distribution, that is, the part of the probability distribution for which the loss
Probability distribution
of expected P B/PB
P B
Copyright © 2021. Harvard University Press. All rights reserved.
PB −0.04 −0.02 0.02 0.04
Breakeven coupon yield 0.02
0.04
R
PB
Figure 11.5 Coupon yield and “insurance” against holding yield loss, I.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
392 Raising Keynes
Probability distribution
of expected P B/PB
P B uninsured
PB −0.04 −0.02 0.02 0.04
Breakeven coupon yield 0.02
0.04
R
PB
Figure 11.6 Coupon yield and “insurance” against holding yield loss, II.
of capital value exceeds the interest payment. For instance, with R/PB = 0.04,
the uninsured portion of the probability distribution is represented in Figure
11.6. For R/PB = 0.02 the uninsured portion is as in Figure 11.7. Te key point
is that the lower the coupon yield, the larger is the uninsured zone. As long as
bond-price changes stay to the right of this zone, holding bonds is superior to
holding cash.
None of this matters to agents who are neutral with respect to risk. Given
the assumed symmetry of the probability distribution around zero, gains are
equally likely as losses, and no insurance is needed. But for risk-averse agents,
Copyright © 2021. Harvard University Press. All rights reserved.
the insurance bufer is important because it ofsets the greater pain caused by
loss in comparison with the pleasure of gain. Since higher coupon yields ofer
more protection against price fuctuations, it is plausible that wealth holders
will wish to hold a smaller portion of their wealth in the form of cash when
yields are higher.
Tis is as far as Keynes goes; the observation that higher coupon yields of-
fer more protection against price fuctuations, like many of the trenchant ob-
servations in Te General Teory, is not incorporated into a formal model re-
lating liquidity preference to risk aversion. In fact it took twenty years for the
model to appear in print, in a deservedly famous article by James Tobin
(1958). Tobin’s formalization allows us to derive a liquidity-preference func-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 393
Probability distribution
of expected P B/PB
P B uninsured
PB −0.04 −0.02 0.02 0.04
Breakeven coupon yield 0.02
0.04
R
PB
Figure 11.7 Coupon yield and “insurance” against holding yield loss, III.
tion ( ) that answers the question of why this function slopes downward. I
present a pre-Tobin version of Tobin’s model in the mathematical appendix to
chapter 12.
Liquidity Preference Without Risk Aversion:
Reversion of Bond Yields to Normal
Te risk-aversion story is lacking in one crucial respect: there is no guarantee
of a liquidity trap. Te ratio of money demand to money supply must, as has
Copyright © 2021. Harvard University Press. All rights reserved.
been observed, go to one as the bond price becomes infnite and the yield
goes to zero. But a liquidity trap requires that the ratio of money demand to
money supply goes to one before PB becomes infnite and R/PB goes to zero.
Tere is nothing in the risk-aversion story that assures this result.
Keynes has another argument for liquidity preference that does imply a li-
quidity trap: namely, that agents do not necessarily expect bond prices and
yields to be stationary over time. Instead, Keynes argues that there are re
ference prices and, correspondingly, reference yields, against which current
prices and yields are judged to be high or low, with prices and yields expected
to revert to these reference levels over time.
Tat is, unlike the case of pure risk aversion, expectations about bond-price
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
394 Raising Keynes
fuctuations are not centered on zero. Rather the mean of the distribution of
P B/PB may itself vary; in particular, it may vary with the coupon yield. If the
current yield R/PB is high relative to the reference yield, agents will expect it
to come down, which is to say the mean of (anticipated) price changes is
greater than zero, E(P B) > 0.13 And vice versa: if the coupon yield is low,
the same agent may expect the yield to rise and the price of the bond to fall. If
E(P B) < 0, it may pay to defer purchasing the bond until it can be had more
cheaply. “May pay,” rather than “will pay,” because the other side of the ledger
is the loss in interest foregone while waiting for bond prices to fall.
Keynes’s terminology in Te General Teory for the reference rate is a “safe”
level, though his defnition of “safe” is somewhat circular; he refers to “what is
considered a safe level of [the coupon yield], having regard to those calcula-
tions of probability which are being relied on” (p. 201). Te important point is
that, as the current rate fuctuates,
If the general view as to what is a safe level of [the coupon yield] is un-
changed, every fall in [the coupon yield] reduces the market rate relatively to
the “safe” rate and therefore increases the risk of illiquidity [that is, the risk
of holding bonds]. (p. 202)
Te safe rate soon came to be called a “normal rate,” normal in terms of
agents’ historical experience. As Keynes put it:
Any level of interest which is accepted with suf ficient conviction as likely to
be durable will be durable; subject, of course, in a changing society to fuc
tuations for all kinds of reasons round the expected normal. (p. 203)
Figure 11.8 presents two distributions of P B/PB, each corresponding to a
diferent coupon yield. R/PB = 0.04 is assumed to be the normal rate at which
the probability distribution of P B/PB, depicted by the solid curve, is equi-
poised between gains and losses, and so has its expected value at the origin.
At R/PB = 0.03, with the price of consols greater by one-third, price declines
Copyright © 2021. Harvard University Press. All rights reserved.
are perceived to be more likely than increases, and the probability distribu-
tion, represented by the dashed curve, moves to the lef. Te consequences for
the amount of insurance provided by the coupon yield are evident in Figures
11.9 and 11.10.
As the breakeven coupon yield falls and the uninsured portion of the prob-
ability distribution increases in size, even risk-neutral agents will fnd it ad-
vantageous to put all their assets into cash at some point before the coupon
yield falls to zero. Tis is in contrast to risk-averse agents facing a symmetric
probability distribution (as in Figures 11.4, 11.5, 11.7, and 11.9), who will al-
ways hold some portion of their portfolios as bonds regardless of how high
bond prices are.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 395
Probability distribution Probability distribution
of expected P B/PB of expected P B/PB
corresponding corresponding
to R/PB 0.03 to R/PB 0.04
P B
PB −0.04 −0.02 0.02 0.04
Breakeven coupon yield 0.02
0.04
R
PB
Figure 11.8 Coupon yield and “insurance” against holding yield loss with
reversion to normal.
Probability distribution
of expected P B/PB
corresponding
to R/PB 0.04
P B uninsured
PB
Copyright © 2021. Harvard University Press. All rights reserved.
−0.04 −0.02 0.02 0.04
Breakeven coupon yield 0.02
0.04
R
PB
Figure 11.9 Coupon yield and “insurance” against holding yield loss with
reversion to normal (R/PB = 0.04).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
396 Raising Keynes
Probability distribution
of expected P B/PB
corresponding
to R/PB 0.03
uninsured
P B
PB −0.04 −0.02 0.02 0.04
Breakeven coupon yield 0.02
0.04
R
PB
Figure 11.10 Coupon yield and “insurance” against holding yield loss with
reversion to normal (R/PB = 0.03).
In a world of risk-neutral agents, each has such a tipping point. In Keynes’s
words,
Unless reasons are believed to exist why future experience will be very dif-
ferent from past experience, a long-term [coupon yield] of (say) 2 per cent.
leaves more to fear than to hope, and ofers, at the same time, a [coupon
yield] which is only sufficient to ofset a very small measure of fear. (Te
General Teory, p. 202)
Copyright © 2021. Harvard University Press. All rights reserved.
In Figure 11.11, the tipping point is 1.5 percent, where the expected value of
the change in price is just equal to the insurance provided by the coupon
yield. Here a risk-neutral agent will be indiferent between bonds and cash.
If all risk-neutral agents have the same expectations, there is a unique
equilibrium interest rate, that is, a unique interest rate that balances the de-
mand for cash with the supply. With positive endowments of money and
bonds, the only portfolio that is optimal for the agents and for which asset
demands and supplies are equal is the knife-edge case shown in Figure 11.12;
Figure 11.12(a) shows the demand for cash as a proportion of assets, and
Figure 11.12(b) pictures the equilibrium interest rate, which is to say the
bond yield at which the demand for money and its supply are equal. In a
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 397
Probability distribution
of expected P B/PB
corresponding
to R/PB 0.015
uninsured
P B
PB −0.04 −0.02 0.02 0.04
Breakeven coupon yield 0.02
0.04
R
PB
Figure 11.11 Coupon yield and “insurance” against holding yield loss with
reversion to normal (R/PB = 0.015).
0.05 0.05
0.04 0.04
0.03 M2
( ) 0.03 M2 (M2 PB B ) M2
0.02 A 0.02
0.01 0.01
M2
1 M2 , M2
A
Copyright © 2021. Harvard University Press. All rights reserved.
(a) Cash demand as proportion (b) Asset-market equilibrium
of wealth
Figure 11.12 Normal reversion and risk-neutral representative agents.
representative-agent, normal-reversion model, there is a liquidity trap—the
horizontal segment of Figure 11.12(b)—but not a downward-sloping money-
demand schedule.14
Observe that the very nature of a normal rate places a time limit on rever-
sion to normal. Te reason that speculators wish to hold cash is that the inter-
est rate on bonds is (perceived to be) on an upward trajectory. But if this tra-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
398 Raising Keynes
jectory is expected to return the bond yield to normal rapidly, then the very
basis of the argument—the positive gap between the normal rate and the ac
tual rate—will soon be eliminated. If expectations of a relatively rapid in-
crease in yield (and corresponding decrease in price) are disappointed, if the
bond yield moves upward at a snail’s pace or not at all, the gap between the
actual yield and the normal yield remains. But, as Keynes says, any rate that
persists long enough becomes normal, and a downward revision of the nor-
mal yield would also eliminate the gap between normal and actual.
Dennis Robertson saw reliance on reversion to normal as a faw in the
theory of liquidity preference:
Te rate of interest is what it is because it is expected to become other than
it is; if it is not expected to become other than it is, there is nothing lef to tell
us why it is what it is—the organ which secretes it has been amputated, and
yet it somehow still exists—a grin without a cat. (1937, p. 433)
And so it is—if liquidity preference rests entirely on reversion to normal.
But reversion to normal is not an alternative to risk aversion; it is, rather, a
complementary argument, one that delivers the missing piece—the liquidity
trap—in the risk-aversion story.
Bulls and Bears
We do not need to rely on a combination of motives in order to end up
with the liquidity-preference schedule of Figure 11.1: we can transform the
fat segment of the money-demand schedule in Figure 11.12 into (almost) a
continuously downward-sloping schedule and thus eliminate the knife-edge
problem—without invoking risk aversion. But to do so we need another of
Keynes’s ideas, namely, that expectations are not uniform:
Diferent people will estimate the prospects [for bond prices] diferently and
Copyright © 2021. Harvard University Press. All rights reserved.
anyone who difers from the predominant opinion as expressed in market
quotations may have a good reason for keeping liquid resources in order to
proft, if he is right, from [lower bond prices later on] . . . Te individual,
who believes that future rates of interest will be above the rates assumed by
the market, has a reason for keeping actual liquid cash, whilst the individual
who difers from the market in the other direction will have a motive for
borrowing money for short periods in order to purchase debts of longer
terms. Te market price will be fxed at the point at which the sales of the
“bears” and the purchases of the “bulls” are balanced. (Te General Teory,
pp. 169–170)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 399
Te same agent can be a bull or a bear, depending on the bond yield. In
Figure 11.12, which assumes that the mean of the distribution of bond-price
changes is −1.5 percent, the agent is a bull at yields above 1.5 percent, desiring
to hold all her wealth in the form of bonds. (Indeed, Keynes goes further; if
she is “borrowing money . . . to purchase debts of longer terms,” she is actually
holding a negative portion of her wealth in cash!) She becomes a bear when
coupon yields fall below 1.5 percent, cashing out her entire bond position to
await lower bond prices.
At the same time, diferent individuals will evaluate the future diferently
and may have diferent probability distributions over P B/PB. For one thing,
they need not agree on the normal rate. But even if they share a view on what
is the normal rate of interest, they need not agree on how fast today’s rate will
revert to normal.
Suppose there are two agents endowed with equal resources, (M 2 + PBB)/2,
but heterogeneous beliefs; agent 1 has a tipping point between cash and bonds
at 3.5 percent, agent 2 at 1.5 percent. Teir individual money-demands, de-
noted M21 and M22, are represented in Figure 11.13(a) and the aggregate
schedule in Figure 11.13(b). Te general formula for money demand in the
aggregate is
M 2 + PB B
M 2 = m ( PB ) .
2
For interest rates not equal to the tipping points, m(PB) represents the number
of agents who desire to hold only cash; m(PB) = 0 for > 0.035, and m(PB) = 1
when 0.035 > > 0.015. For = 0.035, m(PB) = 1 ≤ 1, and for = 0.015,
M2 PB B
M2 m (PB ) M2
2
m (PB ) 0
Copyright © 2021. Harvard University Press. All rights reserved.
0.05 M21 0.05
1 ( ) 0 m (PB ) 1
0.04 A 0.04
0.03 M22 0.03 m (PB ) 1
2 ( ) m (PB ) 1 2
0.02 A 0.02
0.01 0.01
M2i
1 PB0 B PB1 B M2 , M2
A
2 2
(a) Cash demand as proportion (b) Asset-market equilibrium
of wealth
Figure 11.13 Normal reversion with heterogeneous beliefs.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
400 Raising Keynes
m(PB) = 1 + 2 < 2. At these tipping points, think of m(PB) as including frac-
tional agents, the fraction equal to the proportion of the agent’s wealth de-
voted to cash.
At equilibrium money demand and supply are equal:
M 2 + PB B
M 2 = m ( PB ) = M 2.
2
We have
=
M 2 =
M 2 0, > 0.035
=
M 2 =
M 2 1PB B, = 0.035
=
M 2 =
M 2 PB B , 0.035 > > 0.015
1 + 2
M2 = M2 = PB B . = 0.015
1 − 2
Observe that the equilibrium money supply corresponding to interest rates
between 0.035 and 0.015 is equal to the value of the total endowment of
bonds. Tis follows from the assumptions (1) that the two agents have equal
endowments and (2) that in the interval (0.015, 0.035) one agent wants to
hold only cash and the other only bonds. For this to be the case, the value of
the bond endowment must equal the value of the cash endowment. Te
liquidity-preference schedule slopes downward in this interval because the
value of bonds increases as the interest rate falls: the total value of bonds for
= 0.035 is PB0 B = R/0.035; the value of bonds increases to PB1B = R/0.015 at
= 0.015.
Te more diversity of beliefs, the smaller is each step in the aggregate
money- demand schedule; as diversity increases, we obtain an aggregate
schedule that more and more resembles the smooth schedule in Figure 11.1.
Each step corresponds to a diferent division of bulls and bears, for, assuming
risk neutrality, agents will generally allocate all or nothing of their portfolios
Copyright © 2021. Harvard University Press. All rights reserved.
to bonds according to whether their individual expected values of the hold-
ing yield on bonds are positive or negative; at each tipping point, at least
one agent will be indiferent among alternative mixes of bonds and cash. A
liquidity trap, the foor to the coupon yield, is encountered when there is so
much cash in the hands of speculators and the price of bonds has reached
such a high level that almost everybody has become a bear. Te last holdout is
willing to maintain a position in bonds, but she too would sell if the bond
price got any higher, that is, if the yield got any lower. Since this would leave
no one willing to hold bonds, her tipping point provides a foor to the bond
yield.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 401
Default Risk
Tere is one more argument that Keynes ofers for liquidity preference, an
argument that receives all too little attention relative to its intrinsic impor-
tance, especially at times of fnancial crisis. Up to now we have implicitly as-
sumed that the default risk on bonds is the same as the default risk on money.
Te simplest assumption along these lines is that there is no default risk:
bonds are issued by the government in its own currency, a currency over
which the government has total control, and money consists of specie and
bank deposits fully insured by the government.
Alternatively, we could simply ignore specie and assume (as Keynes does in
Te General Teory, p. 168) that bonds and bank deposits both carry a posi-
tive default risk, which just happens to be identical for the two assets. But if
we drop this unrealistic assumption, and assume instead that bonds carry a
greater default risk than bank deposits, we have yet another reason for liquid-
ity preference. If money is free of default risk and the perceived default risk on
bonds is 2 percent per year, and if this risk is independent of the coupon yield,
then risk-neutral agents will display liquidity-preference schedules with the
shape pictured in Figure 11.11—even though the reasoning is diferent.15 In
the present case, the expected holding yield on bonds will be 2 percent less
than the coupon yield, and thus will be positive when the coupon yield ex-
ceeds 2 percent and will be negative when the coupon yield falls short of 2
percent. At 2 percent, risk-neutral agents will be indiferent between holding
bonds and holding cash.
Once again, the assumption of heterogeneous beliefs transforms the aggre-
gate liquidity-preference schedule. If diferent agents have diferent beliefs
about default risk, the demand for money becomes a downward-sloping step
function, as in Figure 11.13(b), and with enough heterogeneity the money-
demand function morphs into the smooth money-demand function of Fig
ure 11.3.
Copyright © 2021. Harvard University Press. All rights reserved.
Conclusions
In its simplest version liquidity preference is the desire to hold cash rather
than a less liquid form of wealth, bonds. It would hardly seem necessary
to devote considerable space to analyzing this desire, except that for Keynes
it is the basis of a new theory of interest. For Keynes, interest is not, as in
mainstream economics, a reward for deferring consumption, and the rate
of interest does not balance productivity and impatience. Interest is rather
a reward for overcoming a preference for liquid assets. Liquidity preference
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
402 Raising Keynes
is thus a theory of portfolio management, which with given endowments
of fnancial assets gives us a theory of asset-market equilibrium. Te inter-
est rate, or more realistically complex of rates that emerges from equilib-
rium in markets for stocks of fnancial assets (bonds and money in the sim-
plest version of the model), is the dog that wags the tail of investment and
saving.
Liquidity preference cannot be reduced to a single cause. Keynes rather
suggests three diferent motives—risk aversion, reversion to normal, and de-
fault risk—for wealth holders to hold a portion, or even all, of their wealth in
cash despite the prospect of receiving periodic payouts of interest on bonds.
A risk-averse agent foregoes the returns on bonds because bond-price fuc
tuations can cause the agent to lose money if she is obliged to sell the bond
before it matures. Te possibility of making money, even if it balances the
possible fnancial loss, does not ofset the pain of loss. Risk aversion is another
way of saying that equal probabilities of gains and losses do not cancel each
other out because of diminishing marginal utility of wealth.
An altogether diferent reason for liquidity preference is embodied in an
agent who believes the current bond yield is below normal and will gradually
revert to normal. Tis agent may calculate that the potential loss in the value
of the bond outweighs the periodic interest payment associated with the
bond. She may decide to hold cash even though she is risk neutral.
Default risk is the fnal reason why agents might prefer barren cash to
interest-paying securities. If the default risk exceeds the nominal return, hold-
ing bonds is a money-losing strategy, and agents will choose to hold cash in-
stead.
Tis chapter explored how the various motives behind portfolio manage-
ment animate the three characteristics of the liquidity-preference schedule:
frst, that the asset-demand for money is a smoothly decreasing function of
the bond yield, which is to say that a small change in yield triggers a small
change in money demand in the opposite direction; second, that the demand
Copyright © 2021. Harvard University Press. All rights reserved.
for money increases without bound as the bond yield falls; third, that there is
a positive lower bound to the bond yield, a positive minimum below which
the bond yield cannot fall.
It turns out—see the mathematical appendix to chapter 12—that none
of the motives for holding cash—risk aversion, reversion to normal, or de-
fault risk—can by itself provide a justifcation for all three properties of the
liquidity-preference schedule. Risk aversion does not necessarily give rise to a
positive lower bound to the bond yield, a liquidity trap. Reversion to normal
and default risk provide the basis for a positive lower bound to the bond yield,
but when beliefs about the future course of interest rates are homogeneous,
the liquidity-preference schedule consistent with either reversion to normal
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 403
or default risk is hardly the smoothly declining schedule that liquidity-
preference theory assumes. Rather, it is a fat line refecting a tipping point at
which a risk-neutral agent moves from a portfolio consisting of all bonds to a
portfolio consisting only of cash. At the tipping point, she is indiferent be-
tween holding money and holding bonds.
Te smooth downward-sloping equilibrium between money demand and
money supply, with a positive lower bound to the bond yield, requires that
agents difer in their beliefs about the normal bond yield, or in their per
ceptions of default risk. With heterogeneous agents, there will be bulls and
bears, agents who believe current bond prices undervalue bonds and agents
who believe that bonds are overvalued. Te frst group will put all their wealth
into bonds, the second all their wealth into cash. When opinions difer widely,
only a small fraction of wealth holders will be indiferent between bonds
and cash, and small changes in the bond yield will lead to small changes in
the proportions of bulls and bears. Even with (almost) all agents in one
camp or the other, the equilibrium bond price will be an increasing func-
tion of the money supply, and this equilibrium price will have an upward
limit.
Risk aversion, normal reversion, and default risk are not mutually exclusive
reasons for liquidity preference. We can imagine a population consisting of
both risk-averse and risk-neutral agents, of agents who believe that bond
yields will revert to a normal level and agents who believe that bond yields
follow a random walk so that the current yield is always “normal.” Or we can
simplify the argument by assuming identical risk-averse agents who also be-
lieve in reversion to normal but with heterogeneous beliefs about the level of
the normal rate or the speed with which current rates will revert to normal.
We can also imagine that bonds carry a default risk, so that even without re-
version to normal, there will be a lower bound to the bond yield. With any of
these assumptions, liquidity preference provides a logically consistent theory
of interest, with the equilibrium bond yield a decreasing function of the
Copyright © 2021. Harvard University Press. All rights reserved.
money supply that approaches a fnite foor as the money supply increases
without bound.
Tis chapter would thus appear to vindicate Keynes vis-à-vis critics who
challenged his assertion that liquidity preference determines the level of inter-
est rates. We have not only an argument why asset-market equilibrium rather
than equilibrium between the demand for investment and the supply of sav-
ing determines the rate of interest but also an explanation of the properties of
the liquidity-preference schedule that characterizes the equilibrium.
Of course, liquidity preference is not the sole determinant of the interest
rate that obtains when the economy is in equilibrium. For the equilibrium, we
need to specify the other components of the model with which liquidity pref-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
404 Raising Keynes
erence interacts. In the third-pass model developed in chapters 6 and 7, in-
vestment demand, the supply of saving, the transactions demand for money,
the GS schedule, and the LS schedule complete the story.
A lesson of that model was that the dei ex machinis of the second-pass
model, the Keynes and real-balance efects, may well exacerbate the problem
of unemployment rather than curing it. Price and wage fexibility turn out to
be part of the problem of depression, not the solution.
Te question remains whether, like a fxed money wage, a world of bonds
and cash is a simplifying assumption that can be amended once the structure
of the argument is understood or a critical assumption on which Keynes’s
edifce depends. Tis question is explored in the next chapter.
appendix: bond coupons as insurance
against price declines
Keynes implicitly makes a stronger assumption than is refected in the section
on Insurance and Risk Aversion, above. He argues that the insurance pro-
vided by the coupon yield (“running yield” in his terminology) varies in pro-
portion to the square of the coupon yield and that this geometric relationship
is the most important reason why any attempt to reduce the (long-term) in-
terest rate is frustrated by liquidity preference. Te full statement is,
Every fall in [the running yield] reduces the current earnings from illiquid-
ity, which are available as a sort of insurance premium to ofset the risk of
loss on capital account, by an amount equal to the diference between the
squares of the old rate of interest and the new. For example, if the rate of in-
terest on a long-term debt is 4 per cent., it is preferable to sacrifce liquidity
unless on a balance of probabilities it is feared that the long-term rate of in-
terest may rise faster than by 4 per cent. of itself per annum, i.e. by an
amount greater 0.16 percent. per annum. If however the rate of interest is
Copyright © 2021. Harvard University Press. All rights reserved.
already as low as 2 per cent., the running yield will only ofset a rise in it of
as little 0.04 per cent. per annum. Tis, indeed, is perhaps the chief obstacle
to a fall in the rate of interest to a very low level. Unless reasons are believed
to exist why future experience will be very diferent from past experience, a
long-term rate of interest of (say) 2 per cent. leaves more to fear than to
hope, and ofers, at the same time, a running yield which is only suf ficient to
ofset a very small measure of fear. (Te General Teory, p. 202)
Keynes does not make his reasoning explicit. But a rational reconstruction
suggests that he is assuming that the distribution of the expected change in
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
11 • The Theory of Interest, I 405
the coupon yield, coup, rather than the expected percentage change in the
coup/coup, is invariant with respect to the coupon yield. Since
coupon yield,
coup
PB
=−
coup PB
the breakeven coupon yield is now given by the formula
2
R
coup
= .
PB
Tus, the boundary of the insured portion of the probability distribution
moves in proportion to the square of coup (rather than in proportion to coup)
as in Figure 11.14, which replaces Figures 11.6 and 11.7.
But two considerations have to be addressed before it can be agreed that
this quadratic feature of the breakeven coupon yield can be considered to be
“the chief obstacle to a fall in the rate of interest to a very low level.” First, the
boundary of the insured portion of the probability distribution is one ele-
ment, but not the only one, in determining the size of the insured por-
Probability distribution
of expected c
2
R
coup
PB
12
12
16
6
04
04
8
08
01
00
00
00
00
00
00
00
.0
.0
.
0.
0.
.
0.
0.
−0
−0
−0
−0
Copyright © 2021. Harvard University Press. All rights reserved.
0.02
Breakeven coupon yield
0.04
R
coup
PB
Figure 11.14 Coupon yield and “insurance” against holding yield loss assuming
coup is invariant with respect to coup.
distribution of expected
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
406 Raising Keynes
tion; diferent probability distributions will give diferent relationships of the
size of the insured portion to its boundary. Second, if my reconstruction
of Keynes’s reasoning is accurate, the argument makes a strong assumption
about the probability distribution itself. Since coup = −coup P B/PB, to assume
the distribution of coup is independent of coup is to assume that the variance
of the distribution of P B/PB varies inversely with coup, which is in turn to
say that P B/PB varies directly with PB. In other words, higher bond prices im-
ply higher price volatility. Denoting variance by Var, we have Var( coup) =
Var(−coupP B/PB) = 2coupVar(P B/PB). Tus if Var( c ) remains constant as the
coupon yield varies, then Var(P B/PB) must fall with the coupon yield. Since
the coupon is fxed, the bond price varies inversely with the coupon yield, so
Var(P B/PB) varies directly with the price of the bond.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Tis appendix demonstrates the key feature of the construction of Figure
11.3, namely, that, where the liquidity-preference and money-demand sched-
ules intersect—which is to say where (asset) money demand and supply are
equal—the liquidity-preference schedule
β( ρ)
M2 = M2 = PB ( ρ)B (11.1)
1 − β( ρ)
is fatter than the money-demand schedule
M 2 = β( ρ) PB ( ρ)B + M 2 . (11.2)
Mathematically, since both dM 2/d and dM2/d are negative, this amounts to
dM 2 dM 2
< (11.3)
d d
wherever M2 = M 2.
Diferentiating (11.1) we have
dM 2 β′ β
= PB+
2 B
PB′ B , (11.4)
dρ (1 − β) 1− β
and diferentiating (11.2) gives
Copyright © 2021. Harvard University Press. All rights reserved.
dM 2
= β′ ( PB B + M 2 ) + βPB′ B, (11.5)
dρ
where ′ = d/d, and PB′ = dPB/d. Since the liquidity-preference schedule is
defned by the locus of intersections between demand for money and supply,
we can substitute the right-hand side of equation (11.1) for the lef in equa-
tion (11.5). Collecting terms, we obtain
dM 2 β′
= P B + βPB′ B,
dρ 1− β B
407
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
408 Raising Keynes
thus
dM 2 1 dM 2
= ,
dρ 1 − β dρ
which guarantees that the inequality (11.3) holds.
A liquidity trap exists at = min because lim (dM 2/d) = lim (dM2/d)
→ min → min
= −∞.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 12 .
the theory of interest, ii
Liquidity Preference as a Theory of Spreads
To the extent that “money” includes deposit accounts bearing interest,
the theory [of liquidity preference] becomes not a theory of the rate
of interest but of the gap between diferent rates of interest, viz., the
yield on Government securities and the interest on bank deposits.
—dennis robertson, personal correspondence to
keynes, february 3, 1935
It would be a mistake, which would be as damaging to further analy-
sis of liquidity preference as it would be to classical doctrines, if it
were thought that uncertainty and liquidity diferentials are the sine
qua non for the existence of a [positive] rate of interest. Such a view
can be compared with a theory of land rent based upon diferences in
the quality of diferent kinds of land. I believe that the analogy is not
a superfcial one.
—paul samuelson
I observed at the outset of the previous chapter that nothing caused the read-
ers of Te General Teory more grief than liquidity preference. It is thus not
surprising that, from the very moment of publication of Te General Teory,
this part of Keynes’s overall argument was subjected to intense criticism. Ulti-
Copyright © 2021. Harvard University Press. All rights reserved.
mately, liquidity preference provides only a partial explanation of the phe-
nomenon of interest and the level of interest rates. For good reasons (see the
epigraph to this chapter) and bad (see the quotation in the previous chapter),
Dennis Robertson had strong reservations about liquidity preference, and this
chapter at least partly vindicates his intuition: Keynes’s theory tells us why
bonds of diferent quality and maturity commonly ofer diferent yields but
not whether the overall level of yields is high or low.
We can see why by asking a simple question: if illiquidity were the sole de-
terminant of interest, what would be the limiting value of the rate of inter-
est as the term to maturity of a bond without default risk gets shorter and
shorter? Keynes’s answer has to be zero. But this is decisively disproven by the
409
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
410 Raising Keynes
data on the rate charged for overnight loans between banks (in the United
States the so-called Federal Funds rate, the name deriving from what is ac
tually borrowed and lent: namely, funds on deposit with the Federal Reserve
banks). Figure 12.1 shows the Fed Funds rate for the period 1954 to 2019.
Only rarely during that time was the Fed Funds rate been below 2.5 percent—
the eight years that followed the fall of Lehman Brothers are an obvious ex-
ception to this rule.
Te basic problem is that the theory presented in the previous chapter is
too simple: liquidity preference explains interest in a world with two assets,
money and bonds, because the spread determines the interest rate on the
bond. Given that the yield on money is zero, the spread between the yield on
bonds and the yield on money is the yield on bonds, so in this special case,
determining the spread is tantamount to determining the level of the (long-
term) interest rate.
As a matter of principle, Keynes would no doubt agree with the need to
generalize the argument to include short-term interest-bearing assets. But in
Te General Teory he ofers only a fudge:
We can draw the line between “money” and “debts” at whatever point is
most convenient for handling a particular problem. For example, we can
treat as money any command over general purchasing power which the
25.0
20.0
15.0
Percent
Copyright © 2021. Harvard University Press. All rights reserved.
10.0
5.0
0.0
1960 1970 1980 1990 2000 2010
Effective Federal Funds rate
Figure 12.1 Fed Funds rate (1954–2019). Source: Board of Governors of the
Federal Reserve System (U.S.). myf.red/g/pXAD
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • The Theory of Interest, II 411
owner has not parted with for a period in excess of three months, and as
debt what cannot be recovered for a longer period than this; or we can sub-
stitute for “three months” one month or three days or three hours or any
other period . . . It is ofen convenient in practice to include in money time-
deposits with banks and, occasionally, even such instruments as (e.g.) trea
sury bills. (p. 167n)
Treating “the line between ‘money’ and ‘debts’” as a matter of convenience ac
tually highlights the limitations of liquidity-preference theory. Allow money
to include interest-bearing assets that dominate cash (by yielding a greater
return without sacrifcing liquidity) and it becomes clear that liquidity prefer-
ence does not speak to the question of why interest exists or ofer an expla
nation of the overall level of interest rates. Liquidity preference becomes a
theory of interest-rate diferentials or spreads.
But do interest-bearing bills dominate noninterest-bearing demand depos-
its and banknotes? Te central issue is the relationship between money as a
medium of exchange and money as a store of value. It is the second with
which liquidity preference is concerned. For transactions, agents need cash,
bank deposits, or other forms of legal tender, but this does not necessitate
holding cash or deposits as a store of value. Te last chapter noted that if you
travel to Mexico, you will need pesos, and if you travel to India, you will need
rupees, but this does not mean that you will necessarily hold pesos or rupees
as part of your asset portfolio or that your holdings of foreign currencies will
be sensitive to the exchange rate. Just so with wealth portfolios: your portfolio
need not contain cash—dollar bills or noninterest-bearing demand depos-
its—even though when it comes time to paying your bills you have to pay
with cash. For clarity, I shall (pace Keynes) limit the use of the term “money”
to the medium of exchange, for which legal tender is of the essence. In wealth
portfolios, money in this sense is replaced by short-term interest-bearing as-
sets, like T-bills.
Copyright © 2021. Harvard University Press. All rights reserved.
Tis separation of two textbook functions of money suggests that we can
separate transactions demand for money from portfolio demand for short-
term fnancial assets and thus ignore the costs of converting one into the
other. Te operative assumption is that decisions about the assets that go into
wealth portfolios are separate from decisions about the assets held for trans-
actions purposes. Tis means one of two things: either the amount of money
agents hold for transactions purposes is interest insensitive or agents who
engage in transactions do not hold asset portfolios.
Chapter 13 will return to the issue of whether or not transactions demand,
and hence cash, is relevant to the determination of interest rates. For the mo-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
412 Raising Keynes
ment we simply bracket this question and assume that interest rates are deter-
mined by liquidity preference and the public’s holdings of a spectrum of inter-
est-bearing fnancial assets.
Liquidity Preference Without Money
Tere is of course more than one short-term interest-bearing asset, and many
of these assets are perceived as normally difering little in default risk. Over
the period 1954 to 2019 the Fed Funds rate moved pretty much in tandem
with rates on 3-month T-bills and 3-month commercial paper, as Figure 12.2
indicates. For brief periods, T-bills have sold at a premium (which is to say
they yielded less than the Fed Funds rate and less than commercial paper),
but for the most part the market has judged these three securities to be good
substitutes for one another.1
What a diference a panic makes. Figure 12.3 shows the rates on these as-
sets over the year following the collapse of Lehman Brothers. Fed Funds, T-
bills, and commercial paper were no l onger perceived to be substitutes for one
another. In the immediate wake of Lehman’s collapse, the Fed Funds rate was
more than two percentage points above the 3-month T-bill rate, and then the
market in commercial paper threatened to freeze up completely (see chapter
20.0
17.5
15.0
12.5
Percent
10.0
7.5
Copyright © 2021. Harvard University Press. All rights reserved.
5.0
2.5
0.0
1960 1970 1980 1990 2000 2010 2020
Effective Federal Funds rate
3-month Treasury bill: secondary market rate
3-month commercial paper rate (discontinued in August 1997)
3-month Aa financial commercial paper rate (from August 1997)
Figure 12.2 Short-term rates (1954–2019). Source: Board of Governors of the
Federal Reserve System (U.S.). myf.red/g/pY1W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • The Theory of Interest, II 413
13, note 9). Liquidity preference with a vengeance! Only as the economy bot-
tomed out in early 2009, and it became clear that the Great Recession would
not repeat the descent into economic hell of the Great Depression, did fnan
cial paper once again become a plausible substitute for Fed Funds or T-bills.
Afer June 2009, we see the old relationships among these three kinds of
short-term assets.
If in normal times liquidity preference does not create much of a spread
between high-grade short-term commercial paper and short-dated T-bills,
the spreads between short-term and long-term rates, as well as the spreads
between government and private long-term paper, are a diferent matter. We
turn now to analyzing the diference it makes when the liquid alternative to
long-term bonds is an interest-bearing short-term asset rather than cash.
Fortunately, we already have in hand an apparatus for modeling these
spreads; the logic is the logic of the relationship between bonds and cash in
the two-asset model studied in chapter 11. Figure 12.4 charts the relationship
between 3-month and 10-year Treasuries. Tis fgure has three notable fea-
tures. First, the yield on T-bills is generally below the yield on 10-year bonds.
Second, the spread is inversely related to the level of yields. Finally, there are
occasions—early 2007 is a case in point—in which the spread is inverted, so
that short-term bills yield more than long-term bonds. Both risk aversion
5.0
4.0
3.0
Percent
2.0
Copyright © 2021. Harvard University Press. All rights reserved.
1.0
0.0
−1.0
Apr 2008 Jul 2008 Oct 2008 Jan 2009 Apr 2009 Jul 2009 Oct 2009
Effective Federal Funds rate
3-month Treasury bill: secondary market rate
3-month Aa financial commercial paper rate
Figure 12.3 Short-term rates (2008–2009). Source: Board of Governors of the
Federal Reserve System (U.S.). myf.red/g/pY22
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
414 Raising Keynes
17.5
15.0
12.5
10.0
Percent
7.5
5.0
2.5
0.0
1970 1980 1990 2000 2010
3-month Treasury bill: secondary market rate
10-year Treasury constant maturity rate
Figure 12.4 3-month Treasury bill and 10-year Treasury note (1962–2019).
Source: Board of Governors of the Federal Reserve System (U.S.). myf.red/g/pY2a
and the expectation of reversion to normal can explain a yield premium on
longer-dated securities, but reversion to normal cannot by itself account for
the tendency of bond yields to exceed bill yields. And as risk aversion was
characterized in the previous chapter, it cannot account at all for the periods
in which short-term bills yield more than long-term bonds.
Both risk aversion and reversion to normal become more complicated
when cash is replaced by T-bills or other interest-bearing liquid assets. Con-
sider risk aversion. Cash, paying no interest, has no interest-rate risk. By con-
trast, as an alternative to long bonds, bills are not immune to the possibility,
Copyright © 2021. Harvard University Press. All rights reserved.
indeed, the likelihood, that the interest rate will fuctuate over time. (Not
within the lifetime of a single bill but over a succession of bills held as an al-
ternative to bonds.) Tis is clear from Figures 12.1 to 12.4. Here the agent’s
time horizon becomes crucial. For investors with short time horizons, these
fuctuations are of no concern; the nominal return on short-term bills, bar-
ring default, is certain. But if you have a long time horizon, if your portfolio is
arranged to put your three-year-old through college ffteen years from now or
to provide for your retirement twenty-fve years hence, then the liquid asset
may actually be riskier than the long bond, rather than the other way around,
as it is when cash is the alternative to bonds.2
If long-horizon agents dominate, and they are motivated by risk aversion
alone, we would expect bills to yield more than bonds to compensate for the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • The Theory of Interest, II 415
extra risk. Tis might be temporarily upset by the belief that the bill rate is
below normal so that bond prices are expected to fall, but the relationship
between the bond yield and the bill rate would in this case follow a pattern
opposite to the pattern in Figure 12.4. Te actual pattern strongly suggests,
even if it does not compel, the view that participants in asset markets gener-
ally have a short time horizon, so that risk aversion focuses on variations in
wealth over a short period. Tere may indeed be agents with much longer
horizons, but only on the relatively rare occasions when short rates exceed
long yields are these agents obliged to balance risk and return. Otherwise,
they get a free ride: they would be willing to give up returns in order to hold
long bonds, but the market only occasionally compels them to do so.3
A short-period focus for risk averse agents provides an explanation of the
relative rarity of inversions of the customary pattern of a yield premium on
bonds. Even if people were to expect interest rates to go up or down with
equal frequency, normal reversion would have to overcome risk aversion in
order to wipe out the yield premium on bonds. Tis was a main point of John
Hicks’s exposition of liquidity preference in Value and Capital: “Te short rate
can only lie above the long rate if the short rate is regarded as abnormally
high” (1946, p. 152).
Risk aversion and normal reversion work in combination to account for all
three characteristics of Figure 12.4—provided we assume, frst, that risk aver-
sion is focused on the near term and, second, that over time the mean of the
distribution of anticipated price changes, E(PB)/PB, averages to zero. Te yield
on bills is generally below that on bonds because of risk aversion. Te spread
widens at low levels of interest rates because reversion to normal reinforces
risk aversion and, even in the absence of default risk, the yield premium en-
joyed by bondholders rises because of greater volatility in bond prices. Te
exceptional occasions when the term structure is inverted, like 2007, can be
explained as times at which reversion to normal and risk aversion are work-
ing at cross purposes. Agents are willing to commit to long-term bonds dur-
Copyright © 2021. Harvard University Press. All rights reserved.
ing these periods because they believe on balance that long-term yields will
fall; they are motivated to buy while bonds are perceived to be cheap. In this
case, expectations of reversion to normal dominate the price-fuctuation risk
of holding long-term bonds, and, unusually, agents have to be compensated in
the form of higher returns, not for holding illiquid bonds, but for holding
short-term paper. Here is where long-horizon agents become players in asset
markets rather than passive free riders who enjoy lower risk and higher re-
turns by specializing in long-dated bonds.
In the mathematical appendix to this chapter we derive the relationship
between short- and long-term yields on the basis of risk aversion alone, as-
sumed from here on in to focus on the variability of wealth in the short term.
Ten we examine the spread assuming risk-neutral agents who believe that
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
416 Raising Keynes
present rates will revert to normal. Finally, we combine the two hypotheses,
assuming that risk aversion and normal reversion operate at the same time.
Here we summarize the results.
The Interest Rate Spread with Risk-Averse Agents
Te simplest way to introduce interest-bearing securities into the picture is to
have bills replace money in the agent’s endowment. As in chapter 11, her en-
dowment is M 2 + PBB but M 2 now consists of a stock of Treasury (or com-
mercial) bills rather than a sum of money.4 Te diference is that the short-
term asset now ofers a riskless return of s ≥ 0, rather than strictly zero. Te
equilibrium spread between the bill rate s and the bond yield coup = R/PB,
pictured in Figure 12.5, has two noteworthy properties. A positive relation-
ship between the bond yield and the bill rate replaces the negative relation-
ship between the bond yield and the quantity of portfolio money. Tis change
in the meaning of liquidity preference changes the meaning of a liquidity trap.
Now a liquidity trap is a positive lower limit to the bond yield at the zero
lower bound (zlb) of the bill rate.
Tis is analogous to the defnition of the liquidity trap in a bonds versus
cash model, but it is not the same thing. In chapter 11 we defned the liquidity
trap in terms of the behavior of the bond yield as the cash endowment in-
creases without bound, and we found that risk aversion by itself was no guar-
antee of a liquidity trap. Here the endowment of bills (and bonds) is assumed
to be fxed;5 what varies is the bill rate, and the liquidity trap is now defned as
a positive lower bound to the bond yield as the bill rate approaches the zlb.6
s s
Copyright © 2021. Harvard University Press. All rights reserved.
Liquidity Liquidity
preference preference
45° 45°
coup coup
(a) Risk aversion without (b) Risk aversion with
a liquidity trap a liquidity trap
Figure 12.5 Liquidity preference without money, I: risk-averse agents.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • The Theory of Interest, II 417
Now, if risk aversion is the only game in town, a liquidity trap may or may
not exist. As the liquidity preference schedule is drawn in Figure 12.5(a), the
liquidity-preference schedule goes through the origin; assuming spreads are
driven solely by risk aversion, and assuming the short-term rate is at the zlb,
the bond price may have to become infnitely high, so that the bond yield falls
to zero, before wealth holders will put all their eggs in the bill basket. It is also
possible for the liquidity-preference schedule to exhibit a liquidity trap, that
is, a positive intercept on the coup-axis, as in Figure 12.5(b). (See the mathe-
matical appendix to this chapter for details.) In this case, desired bond hold-
ings go to zero when the bond price passes a fnite threshold.
Te most important takeaway from substituting bills for cash as the liquid
store of value is that we can infer from the assumption of asset-market equi-
librium only the spread between the yields on the assets that comprise the
market. Robertson’s observation to this efect—see the frst epigraph to this
chapter—was correct in 1935, and Samuelson (quoted in the second epi-
graph) was right in 1947.7
In a way this should not be surprising. Tat liquidity preference determines
only spreads is the counterpart of a more general limitation of market equilib-
rium, namely, that with n goods only n − 1 prices emerge, which is to say that
only relative prices are determinate. In asset markets there are not enough
degrees of freedom to determine the separate yields.
Nor is this result problematic in a world with central banks. For most of the
last century, monetary policy has consisted of choosing the bill rate with an
eye to fxing the bond yield. In other words, the central bank has taken on the
task of anchoring the spread at the short end, leaving it to asset markets to
determine bond yields and associated hurdle rates of return for new capital
expenditure.
Reversion to Normal
Copyright © 2021. Harvard University Press. All rights reserved.
But I get ahead of my story. Risk aversion is only one of the arguments for li-
quidity preference. What happens to normal reversion when we substitute
interest-bearing short-term assets for cash? Little changes qualitatively, but
the existence of bills as an interest-bearing alternative to cash makes it neces-
sary to fesh out the normal-reversion argument.
In a model with cash and bonds, the short rate is the return on cash; it is
fxed at zero and so cannot revert to anything else. Reversion in the cash-
bond case necessarily refers only to the bond yield. In the present model, by
contrast, reversion is fundamentally a property of the short-term bill rate, and
we can derive the trajectory of the breakeven bond yield, the yield that makes
an agent indiferent between bonds and bills, from the expected trajectory of
short rates.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
418 Raising Keynes
Te starting point is that, in the absence of risk aversion, the willingness of
agents to hold both bills and bonds requires the anticipated holding yield on
bonds ( hold) and the yield on bills ( s) to be equal. In continuous time this
indiference condition is
R E ( PB )
hold ≡ + = s,
PB PB
where E(P B)/PB is the mean of the distribution of anticipated bond-price
changes. If the above equality holds for the future as well as the present, we
can infer a “demand price,” E(PB,t), for the bond at any time t from the solu-
tion to the integral equation:
τ
E ( PB ,t ) = Re ∫t
∞ − ρ s ( x )dx
∫t d τ,
where
τ
− ∫ ρ s (x ) dx
t
is the discount factor at time t for a future time , namely, the value at time t
of one dollar available at time when the discount rate for each point in time
between t and is given by the value of s. At t = 0, we have
τ
E ( PB , 0 ) = Re ∫0
∞ − ρ s ( x )dx
∫0 d τ.
Whether or not it makes fnancial sense for the agent to hold bonds de-
pends on how the actual price today compares with her demand price, that is,
the price based on her view of expected reversion to normal. If the actual
price exceeds the demand price, then she is better of putting her fnancial
resources into short-term bills. If the actual price is lower than the demand
Copyright © 2021. Harvard University Press. All rights reserved.
price, it makes sense to buy bonds. If the two prices are exactly equal, she can
expect capital losses to just ofset the coupon and would logically be indifer-
ent between bills and bonds.
In terms of yields, the demand price defnes a coupon yield R/E(PB,0) at
which the agent will be indiferent between holding bonds and holding bills.
She will put her whole wealth into bonds if the actual coupon yield exceeds
R/E(PB,0) and prefer bills if R/E(PB,0) is above the actual.
In this model, reversion to normal of the bill rate drives the demand price
of bonds and the corresponding yield. So how do we characterize the ex-
pected path of short-term rates? Te simplest story is that the short rate is
expected to make up the distance between the current rate s and the normal
rate *s at a speed proportional to the distance:
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • The Theory of Interest, II 419
(
s = −θ ρ s − ρ∗s .
ρ )
In this formulation, the subjective elements of an agent’s belief system reduce
to two parameters, the normal bill rate (*s ) and the speed of adjustment ().
If the agent projects this process into the future, the expected future rate at
time is given by a weighted average of the current rate and the normal rate,
with the weight on the present declining as we move forward in time:
ρ s (τ) = ( 1 − e − θ( τ −t ) ) ρ∗s + e − θ( τ −t ) ρ s (t ).
Substituting into the equation for the demand price, we obtain
∞ ρ (0 ) − ρ∗s − θτ
− ρ∗s τ + s (e −1)
E ( PB , 0 ) = ∫ Re θ
d τ,
0
and the critical value of the current coupon yield—the value below which the
agent will hold only bills, and above which only bonds—becomes
−1
∞ − ρ∗s τ + ρs (0) − ρs ( e − θτ −1 )
∗
R
E ( ρcoup, 0 )
E ( PB , 0 ) ∫0
= = e θ
dτ .
Figure 12.6 charts the relationship between s and E( coup, 0) on the as-
sumptions *s = 0.04 and = 0.1. If the current short-term rate is zero, Figure
12.6 says that the critical value of the long-term bond yield is 0.029. (Te
mathematical appendix provides a numerical solution to the equation that
defnes the schedule in Figure 12.6.) If coup > 0.029, the agent will commit her
portfolio entirely to bonds; if coup < 0.029, entirely to bills. Evidently, if all
agents are alike, the only long-term yield consistent with agents’ holding both
bills and bonds is coup = 0.029. At this coupon yield, all agents believe that
capital losses will just cancel out interest earnings and are indiferent between
Copyright © 2021. Harvard University Press. All rights reserved.
alternative portfolio mixes of bills and bonds.
If agents have diferent beliefs about how rapidly s will revert to normal
(or diferent beliefs about what constitutes normal, or both), then only a sub-
set of agents need be equally comfortable with alternative portfolio mixes.
Everybody else will specialize either in bonds or in bills. Figure 12.7 assumes
fve types of agents difering only in their assumptions about the speed of re-
version to normal. Evidently, there is a liquidity trap, defned now as the exis-
tence of a positive foor to the (equilibrium) bond yield when the bill rate is at
the zlb.
Te relationship between bills and bonds is pictured in Figure 12.8, which
resembles Figure 12.5 but with two diferences. First, a liquidity trap is an es-
sential part of the story. Even in the case in which the present (t = 0) value of
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
420 Raising Keynes
0.08
0.07
Expected value of coupon yield
0.06
Hold only bills
Short-term bill yield
0.05
0.04
0.03
Hold only bonds
0.02
0.01
0.0
0.0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Long-term bond coupon yield
Figure 12.6 Term structure: expected short- and long-term rates with reversion to
normal.
0.08
0.07
0.06
Short-term bill yield
0.05
0.04
Copyright © 2021. Harvard University Press. All rights reserved.
0.03
0.02
0.01
0.0
0.0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Long-term bond coupon yield
0.05 0.08 0.1 0.25 1.0
Figure 12.7 Term structure: relationship between short- and long-term rates with
diferent adjustment speeds.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • The Theory of Interest, II 421
s is zero, the bond yield will be positive because it is an average of short-term
rates over the life of the bond, and these rates are expected to be positive as
the short rate reverts to normal.
Second, when the bill rate is above normal, the relationship between the
bond yield and the bill rate is inverted. Unlike the risk-aversion case, in which
the yield premium exists regardless of the level of interest rates, in the nor-
mal-reversion case short-term bill rates exceed the bond yield when the cur-
rent bill rate is above the normal rate. Although short-horizon risk aversion
cannot account for the inversion of short and long rates pictured in Figure
12.4, normal reversion can.8
Reversion to normal is a plausible reason for interest-rate spreads. But re-
version to normal is no more adequate by itself than is risk aversion. For one
thing, if reversion to normal were the sole force at work, we should expect
that, over time, short-term rates would be distributed more or less symmetri-
cally around the normal rate, so that the mean of the diference between the
current short-term rate and the normal rate would be zero. Tis would imply
that inversions of the yield premium would be common, not the relatively
rare events they are in Figure 12.4. Moreover, a major implication of the ab-
sence of risk aversion is that individual portfolios are specialized to bonds or
bills except when agents are on the margin and willing to hold both securities
in combination. When all is said and done, the idea of portfolios consisting of
only one kind of security is only marginally more palatable than the idea of
s
Liquidity preference
Copyright © 2021. Harvard University Press. All rights reserved.
s
s coup
Figure 12.8 Liquidity preference without money, II: reversion
to normal.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
422 Raising Keynes
identical risk-neutral agents, which leads to demand oscillating wildly be-
tween short- and long-dated securities in response to small changes in the
spread. Tere may be wide diversity of opinion, but with risk-neutral agents
there is practically no diversifcation!
Combining Risk Aversion with Reversion to Normal
in a Theory of Interest-Rate Spreads
Risk aversion and normal reversion are not mutually exclusive theories.
Rather, as in chapter 11, the two theories are complementary. Risk aversion
answers a question to which normal reversion provides no answer, namely,
why do agents diversify their holdings? And normal reversion answers a
question to which risk aversion provides no answer, namely, why does the
term structure sometimes exhibit an inversion of the usual positive spread
between long and short coupon yields?
Figure 12.9 combines these results, showing how normal reversion dis-
places the liquidity-preference schedule in Figure 12.5. Observe that at A,
corresponding to a bill rate equal to the normal rate, the two schedules inter-
sect. At A an agent who believes that current rates always revert to normal has
no diference of opinion with an agent who does not believe at all in reversion
to normal; where s = *s , both agents share the belief that E(P B)/PB = 0. When
risk aversion and normal reversion are combined, the short-term rate must be
s
Liquidity preference combining
risk aversion and normal reversion
B
0s
Liquidity preference
with pure risk aversion
Copyright © 2021. Harvard University Press. All rights reserved.
s A
s coup
Figure 12.9 Liquidity preference without money, III: combining
risk aversion and normal reversion.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • The Theory of Interest, II 423
more than marginally greater than the normal rate for an inversion of the
yield premium to take place. In Figure 12.9 only when the short rate exceeds
0s is the bill rate higher than the bond yield.
Te two extremes of pure risk aversion and pure normal reversion reduce
to extreme assumptions about portfolio optimization. Pure risk aversion,
without normal reversion, can be expressed as the assumption E(P B)/PB = 0;
pure normal reversion reduces to an assumption about the utility function,
namely, −U2/U1 = 0. In this formulation, explored more fully in the mathe-
matical appendix to this chapter, U1 is the marginal utility of (expected)
wealth, and U2 refects the marginal utility of holding wealth in liquid form.
Normal reversion without risk aversion means that the marginal utility of li-
quidity is zero.
Te empirical appendix to this chapter assesses the relative importance of
risk aversion and normal reversion in determining the yield premium over
time. Te conclusion is that most of the time normal reversion is of relatively
minor importance—but not all the time. At critical junctures—including the
period since the fnancial crisis developed in the fall of 2008—normal rever-
sion not only matters, it has been the dominant force driving interest-rate
spreads, at least for spreads between Treasury securities.
Default Risk
Up to now we have considered two assets, one of which has price risk. But
neither entails any default risk. T-bills and T-bonds are the canonical exam-
ples, though (as was observed at the beginning of the chapter) high-grade
commercial paper as well as Fed Funds usually are interchangeable with
short-term T-bills. Since our focus is ultimately on how the hurdle rate for
private investment decisions is determined, we need to extend the story to
take account of the possibility, always present in private undertakings, that
the borrower may default.9 Te hurdle rate relevant for private investment is
Copyright © 2021. Harvard University Press. All rights reserved.
not the yield on Treasuries, but the yield on bonds issued by corporations
with a risk of default comparable to the risk that a particular investment in
physical assets will go sour.
How do we conceptualize the relationship between yields on Treasuries
and yields on corporate bonds? Te logic of this comparison is the same as
the logic for comparing short- and long-term government obligations: agents
are assumed to compare the expected holding yields on the two types of
bonds, taking account of the impact of default on the expected change in
bond price. Risk-averse agents presumably require a premium refecting the
greater price volatility associated with corporate bonds—once default risk is
factored in.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
424 Raising Keynes
coup
Liquidity
preference
CORP
Figure 12.10 Liquidity preference without money, IV: optimizing
between corporate and Treasury bonds.
Figure 12.10 summarizes the results of adding default risk to liquidity pref-
erence. Te spread between Treasury and corporate bonds is assumed to be
decreasing with the Treasury yield. Tis is intended to refect the increase in
both perceived and actual default risk at times of slack aggregate demand,
when the Federal Reserve typically reduces Treasury yields but corporate de-
fault risk, and hence the spread CORP − coup , increases.
Tis is particularly salient at times of fnancial panic. From 1990 to 2007,
the diference between the yield on corporate bonds and the yield between
Treasuries of comparable maturity suggests an implicit default risk on the
lowest investment-grade corporate bonds (Moody’s Baa rating) of the order
of 1.5 percent per year.10 By contrast, in the year following the collapse of
Copyright © 2021. Harvard University Press. All rights reserved.
Lehman Brothers, the implicit default risk rose on average to 4 percent, peak-
ing just above 5.5 percent in December 2008.
Figure 12.11 shows how the relationship between short and long Treasuries
is modifed by the addition of default risk. It is still theoretically possible to
have an inverted term structure—possible for the corporate bond to have a
lower yield to maturity than a short-term T-bill—though it takes a higher
short-term rate to ofset the higher price risk when default is part of the pic-
ture. In fact, even though the term structure of Treasuries exhibited inversion
six times between 1962 and 2019, the short-term bill rose above the corporate
bond yield only three times, and very briefy at that (in 1973, 1980, and 1981),
as Figure 12.12 shows.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • The Theory of Interest, II 425
s
Liquidity preference combining
risk aversion and normal reversion
(government bonds)
Liquidity preference with
risk aversion, normal
reversion, and default risk
(corporate bonds)
coup , CORP
Figure 12.11 Liquidity preference without money, V: adding in default
risk.
17.5
15.0
12.5
10.0
Percent
7.5
Copyright © 2021. Harvard University Press. All rights reserved.
5.0
2.5
0.0
1970 1980 1990 2000 2010
3-month Treasury bill: secondary market rate
10-tear Treasury constant maturity rate
Moody’s seasoned Baa corporate bond yield
Figure 12.12 3-month Treasury-bill, 10-year Treasury note, and Moody’s BBB
corporate bond index (1962–2019). Sources: Board of Governors, Moody’s.
myf.red/g/pY2y
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
426 Raising Keynes
Liquidity Preference and Monetary Policy
Innovations in the Great Recession
If we couple the construction of the liquidity-preference schedule in Figure
12.11 with the investment-demand and saving schedules, we can determine
the hurdle rate of interest, the level of aggregate demand, and the demand for
transactions money, as in Figure 12.13. Liquidity preference may provide only
a theory of spreads, but with a central bank’s hand on the steering wheel, a
theory of spreads is all that is needed to determine aggregate demand: the
central bank fxes the short-term bill rate—s = 0.05 in Figure 12.13—and
bond markets take care of the rest.
Te Federal Reserve’s immediate response to the fnancial crisis and the
ensuing recession was to reduce the bill rate to its lower bound of zero. Tis
had only a m odest efect on the hurdle rate because of both increased risk
aversion—like investment demand, liquidity preference is based on expecta-
tions and highly volatile—and the expectation that interest rates would revert
to normal: the spread between the T-bill rate and the corporate bond yield—
see Figure 12.12—was more than 500 basis points throughout 2009. Given the
endowments of bonds and bills in the hands of the public, the Fed could do
little to change the spread.
Tese endowments are not immutable in theory and were not in practice.
Te Fed followed up its policy of setting the bill rate at zero by “quantitative
easing,” essentially swapping bills for the public’s bonds, and by this means
shifing the liquidity-preference schedule inward. (In actuality the swap was
central-bank reserves for bonds; see chapter 13, note 10.) In the mathematical
appendix to this chapter, it is shown that for changes in the endowments that
are wealth preserving, that is, for which
M 2 + PB B = M 2 + PB B = const.,
the relationship between the equilibrium bond price and the quantity of
Copyright © 2021. Harvard University Press. All rights reserved.
bonds in the hands of the public is negative when certain assumptions about
risk preferences hold. Under these assumptions an increase in M and a cor-
responding decrease in B shif the liquidity-schedule inward, as in Figure
12.14.
A second novelty in Fed policy as recovery slowed was the commitment to
maintain the short-term rate virtually at zero. Reversion to normal has been
framed as a process in which the current value of the short-term interest rate
is fxed by the central bank, and its expected evolution follows a path of
gradual adjustment to normal. Under this assumption, the central bank may
set s as low as it chooses, even at zero (as was efectively the case from the fall
of 2008 to the end of 2015), but expectations embodied in determine the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
ID , SD
12•
Investment demand in relation
to corporate bond yield
SD
ID
0.125 0.10 0.075 0.05 0.025
corp Income (Y)
100 200 300 400 500
0.025 100
0.05 200
Liquidity preference combining
risk aversion, reversion to 0.075 300
The Theory of Interest, II
normal, and default risk Transactions demand
0.10 400 for money
0.125 500
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
s M1 M 1
427
Figure 12.13 Aggregate demand and the transactions demand for money.
Copyright © 2021. Harvard University Press. All rights reserved.
428
ID , SD
Investment demand in relation
to corporate bond yield
SD
ID
corp 0.125 0.10 0.075 0.05 0.025
Income (Y)
100 200 300 400 500
0.025 100
Raising Keynes
0.05 200
Liquidity preference combining Transactions demand
risk aversion, reversion to 0.075 300 for money
normal, and default risk
0.10 400
0.125 500
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
s M1 M 1
Figure 12.14 Aggregate demand and the transactions demand for money with quantitative easing.
12 • The Theory of Interest, II 429
bond yield. Tis is what makes a liquidity trap, defned as the level of coup (or
CORP) associated with s = 0, possible.
However, is not etched in stone; is a variable under (partial) control of
the central bank. A commitment to maintain s below the normal rate in ef-
fect reduces , and the longer the duration of its commitment, the lower the
current bond yield. Te efect is, like the efect of quantitative easing depicted
in Figure 12.14, to shif the liquidity-preference schedule inward, with the dif-
ference that the shif is more pronounced the further the economy is from the
normal rate.
Te limit to the central bank’s control over is the credibility of its commit-
ment. In the limit, a credible commitment to s = 0 for the entire term of a
T-bond drives to zero over this whole period. Te result is that the price of
the T-bond exceeds its par value, and the gradual fall in the price of the bond
exactly ofsets the coupon; the yield to maturity is zero. Observe that in con-
trast with a commitment to maintain s = 0 for a defnite amount of time, a
commitment that expires when a trigger is pulled (for example, the unem-
ployment rate reaching 5 percent) leaves the path of uncertain because of
the uncertainty as to when the gun will be fred.
Conclusions
Tis chapter addressed the major shortcoming of liquidity preference as a
theory of interest: the alternative to holding bonds is not to hold cash or bank
deposits, but short-term bills that normally ofer an interest payment to their
owners. Te cash-bonds model, unlike the rigid money-wage model, turns
out not to be a part of the scafolding that can be jettisoned once the building
is in place!
When all is said and done, Keynes’s critics were right to question liquidity
preference as a theory of interest. Liquidity preference is instead a theory of
interest rate diferences. In a world of money and bonds, liquidity preference
Copyright © 2021. Harvard University Press. All rights reserved.
provides a coherent and complete theory of the rate of interest, but only be-
cause a theory of diferences between yields on the short-term asset (money)
and the long-term asset (bonds) is necessarily a theory of the rate of interest
on the long-term asset. Te nominal return on money is by defnition zero, so
the diference between the two rates is simply the nominal return on bonds.
Tis result does not generalize to a more realistic world in which wealth
holders choose among an array of assets of varying terms, yields, and default
risks, an array in which, as a store of wealth, money is dominated by short-
term bills. In this world, the writ of liquidity preference runs no further than
the spreads between these various yields. In a world in which a central bank
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
430 Raising Keynes
steers the economy by imposing a short-term interest rate, this limitation is
not, in principle, a problem, because all that is needed is a theory of spreads.
As in the money-bonds world of the previous chapter, liquidity preference
is a big tent in which risk aversion and reversion to normal as well as default
risk all infuence the structure of interest rates. And as in the simpler cash-
bonds model, none of these motives for holding liquid assets is suf ficient by
itself to account for observed patterns of interest-rate structures. Short-hori-
zon risk aversion leads to the prediction that interest-rate diferentials do not
depend on the existing short rate, and that bond yields will always be above
short-term rates. As we shall see in the empirical appendix to this chapter, the
data reveal some dependence of the spread on the level of the bill rate, some
thing we can see in Figures 12.4 and 12.12.
Reversion to normal accounts for the widening of the diference between
long and short yields at low interest rates, but by itself would predict that in-
versions of the term structure would be as frequent as the usual term struc-
ture, in which yields rise with bond term. Neither aversion to the risk of
bond-price fuctuations nor reversion of interest rates to normal accounts for
the persistent gap between yields on government securities and yields on cor-
porate bonds of comparable maturity. Here, liquidity-preference theory has to
appeal to another kind of risk, namely, the risk of default.
Te various motives for holding more liquid assets are not mutually exclu-
sive. We can imagine agents who embody both risk aversion and a belief in
reversion to normal. Or we can imagine that some agents are risk averse with-
out believing rates will revert to normal, and others are risk neutral while be-
lieving in reversion to normal. Either way, we will get a liquidity-preference
relationship between short rates and long rates, and more to the point, be-
tween riskless short rates and the hurdle rate that governs investment deci-
sions. Even if the short-term rate is equal to its zlb, the hurdle rate will be
positive. And higher short-term rates generally correspond to higher long-
term yields. Together these two results are the essence of liquidity preference.
Copyright © 2021. Harvard University Press. All rights reserved.
It should come as no surprise that the liquidity-preference schedule, like
the investment-demand schedule, is likely to be extremely volatile. Te reason
is the same here as in chapter 10: the central role of expectations and the fra-
gility of the knowledge on which expectations are based.
We should expect risk aversion to vary over the cycle because both actual
default rates and, more important, the fear of default rise in bad economic
times. But risk aversion can, as we shall see when we examine the data, in-
crease even with respect to T-bonds, a result that takes some explaining.
Te impact of normal reversion also changes over the cycle. Most of the
time, the data suggest, normal reversion is noticeable for its absence. Most of
the time, agents have little clue as to which way interest rates will move. But
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • The Theory of Interest, II 431
sometimes normal reversion is the most important, if not the only, driver of
interest-rate spreads. From the beginning of the fnancial crisis of 2008 until
the end of 2015, interest rates had nowhere to go but up. In this period, the
question was the speed with which interest rates would move, not the direc-
tion in which they would move. During that period, there was increasing
speculation about whether the normal rate itself has changed. In 2020 we are
still waiting for the answer.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Tobin’s 1958 model quickly became the canonical model for analyzing liquid-
ity preference in terms of risk aversion. In the intervening decades, there have
been many modifcations, but the basic ideas remain the same. Tobin assumes
frst that the distribution of price changes, PB/PB (in a discrete time model),
is independent of the price of bonds, as in Figure 11.4, and, second, that
agents maximize a utility function based on the mean and the standard devia-
tion of the one-period performance of portfolios that include bonds and cash.
Tobin rationalized the emphasis on the mean and standard deviation in terms
of quadratic expected utility à la von Neumann and Morgenstern (1944). John
Campbell and Luis Viceira (2002, chap. 2) review alternative assumptions
about the utility function and the distribution of returns that justify a focus
on the short-term mean and variance.
In this section I follow Tobin and his heirs in focusing on risk aversion, but
the main points of the argument can be made more transparently with
fewer restrictive assumptions—in particular, without the assumption of von
Neumann–Morgenstern utility and corresponding restrictions on the distri-
bution of bond-price changes. Te essential idea, laid out in chapter 11, is that
money provides utility over and above its contribution to wealth because of
the fexibility it provides to take on targets of opportunity in an uncertain
world.
Here we suppose that agents choose a combination of money and bonds
to maximize an (ordinal) utility function U(E(A), M), in which A represents
wealth, E(A) represents expected value of wealth at the end of the year (the
Copyright © 2021. Harvard University Press. All rights reserved.
mean of the probability distribution of year-end portfolio values), and M (no
subscript is necessary for present purposes) represents the money in the port-
folio at the beginning of the year. Utility increases with E(A), but for any
given level of expected wealth, the greater the fraction of the portfolio in
money, the greater is the level of utility. Utility increases with M because more
money means less risk of fuctuations in wealth; the initial price of bonds PB is
assumed to be given, but the terminal price PB + PB is a random variable.
Expected wealth is
R E ( ∆PB )
M + 1 + + PB B.
PB PB
432
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 433
Te utility function is
R E ( ∆PB )
U (E( A), M ) = U M + 1 + +
P B , M (12.1)
P P
B
B B
where B and M are the agent’s holdings of bonds and money.
Indiference curves associated with U(E(A), M) are pictured in Figure
12.15. Te positive relationship of utility to the portfolio’s expected value and
to the portion of the portfolio committed to money is refected in the signs of
the derivatives U1 ≡ ∂U/∂E(A) > 0 and U2 ≡ ∂U/∂M > 0.
To construct a liquidity-preference schedule along the lines of Figures 11.1
and 11.2, we need to impose restrictions on the indiference map: (1) indifer-
ence curves become steeper as we move upward, that is, holding the amount
of portfolio money constant; (2) indiference curves fatten out if the agent
increases her holdings of money while maintaining a given level of expected
wealth; and, fnally, (3) indiference curves become less steep when E(A) and
M increase at the same rate. Assumption (1) means that agents become more
risk averse as their expected wealth increases relative to their money holdings.
Assumption (2) implies that agents become less risk averse as their money
holdings increase relative to their expected wealth. Together these two as-
sumptions imply that the indiference curves are convex to the origin, so each
one becomes less steep from northwest to southeast (in the direction of the
E(A)
(3)
(1)
(2)
I3
Copyright © 2021. Harvard University Press. All rights reserved.
I2
I1
Figure 12.15 Liquidity preference as aversion to risk.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
434 Raising Keynes
curved arrow).1 Assumption (3) guarantees that the impact of additional
wealth on the risk-return trade-of is smaller than the impact of additional
liquidity, measured by an equal increase in M. Tese three assumptions are
refected in the changes in the slope of the indiference curves along the direc-
tional arrows (1)–(3) in Figure 12.15.
We obtain formal expressions for Assumptions (1)–(3) from the behavior
of the slope as we move in one direction or another. Write the slope as
dy U
= h( y , x ) = h(E( A), M ) = − 2 .
dx U1
Assumption (1), that indiference curves become steeper as we move verti-
cally in Figure 12.15, means that the derivative of h with respect to expected
wealth is negative, which is to say
U 21U 1 − U 11U 2
h1 = − < 0. (12.2)
U 12
An increase in the slope −U2/U1 holding E(A) constant, Assumption (2),
implies
U 22U 1 − U 12U 2
h2 = − > 0. (12.3)
U 12
Convexity means that every indiference curve fattens out as we move along
it in a southwest direction. Tis is to say
dy U U −U U U 2 U 22U 1 − U 12U 2
h1 + h2 = − 21 1 2 11 2 − − > 0. (12.4)
dx U1 U1 U 12
Assumption (1) implies that the frst term in equation (12.4) is positive, and
Assumption (2) implies that the second term is positive. Assumption (3) re-
quires that h1 be small in absolute value relative to h2. Starting from any point
Copyright © 2021. Harvard University Press. All rights reserved.
in Figure 12.15, the slopes of successive indiference curves increase along the
45-degree line of equality between (E(A)) and M:
U 21U 1 − U 11U 2 U 22U 1 − U 12U 2
h1 + h2 = − − > 0. (12.5)
U 12 U 12
Optimization is constrained by wealth:
M + PB B = A0 = M + PB B , (12.6)
where M and B are, respectively, the agent’s endowment of cash and bonds.
Te agent is assumed to maximize expression (12.1) subject to the portfolio
composition possibilities refected in equation (12.6).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 435
Te return on cash is by assumption zero, so that if the portfolio consists
entirely of cash, expected wealth at the end of the year is the initial wealth
A0 = M + PBB. To the extent the portfolio includes bonds, expected wealth
increases by the expected holding yield, multiplied by the amount of the port-
folio in bonds,
R E ( ∆PB )
+ PB B.
PB PB
In the case of pure risk aversion, E(PB)/PB = 0, so the relationship between
the expected value of the portfolio and the holdings of bonds and money is
given by
R R R
E( A) = M + 1 + PB B = A0 + PB B = A0 + ( A0 − M ),
PB PB PB
and the problem is to maximize
R
U (E( A), M ) = U A0 +
( A0 − M ) , M . (12.7)
PB
In the text, cash is replaced by short-term bills as the alternative to bonds.
Tis does not change the optimization problem, except that now we regard M
as the desired holdings of bills, and M as the endowment of bills. Equation
(12.6) continues to represent the budget constraint, but expected wealth and
the utility function change because bills are assumed to yield s. Expected
wealth is now
R R
E( A) = ( 1 + s ) M + 1 + PB B = ( 1 + s ) A0 + − s ( A0 − M ),
PB P
B
and utility is
Copyright © 2021. Harvard University Press. All rights reserved.
R
U (E( A), M ) = U ( 1 + s ) A0 + − s ( A0 − M ) , M .
PB
Figure 12.16 shows the menu of alternative portfolios of bills and bonds and
the optimal portfolio at E.
In Figure 12.16 the optimal balance between return and risk is struck at
E, with the demand for bills given by M = M*, and the demand for bonds
PBB* = M + PBB − M*. Te frst-order condition for an interior maximum is
R U
− s − 2 = 0,
PB U1
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
436 Raising Keynes
E(A)
I3
I2
I1
(1 s ) A 0
Portfolio composition possibilities
(1 s ) (M PB B ) R
E ( A ) (1 s ) A 0 s ( A 0 M)
PB
M M A 0 M PB B M
Figure 12.16 Balancing coupon yield, capital loss, and risk.
which can be rewritten as
R U
− s = 2 .
PB U1
U2/U1 is a risk premium, the amount by which the expected yield on bonds
must exceed the riskless return available on short-dated assets, “must” in
the sense of the dictates of constrained maximization of the wealth holder’s
utility.
A change in the yield R/PB changes the optimal balance of risk and return
Copyright © 2021. Harvard University Press. All rights reserved.
in two ways. If PB rises, which is to say that the yield decreases, the value of
the agent’s endowment increases with the price of the bond to the extent that
her endowment takes the form of bonds—a “wealth efect.” At the same time,
the reward for holding bonds decreases, a “substitution efect.” Te two efects
are shown in Figure 12.17, the wealth efect by the shif in the position of the
intercept of the schedule of portfolio possibilities, the substitution efect by
the shif in the slope of this schedule.
At the two limits B = 0 and M = 0, portfolio possibilities shif as in Figure
12.18. F is the new equilibrium when the agent’s endowment consists only of
bills and F′ is the new equilibrium when only bonds. Observe that in the sec-
ond case the wealth efect is necessarily negative.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 437
Portfolio composition possibilities
R
E ( A ) (1 s ) A10 1 s (A10 M )
E(A) P
B
F I3
I2
(1 s ) A 20 (1 s )(M PB2 B ) I1
(1 s ) A10 (1 s )(M PB1 B ) R
E ( A ) (1 s )A 20 2 s (A 20 M )
PB
M A 10 M PB1B M A 20 M PB2B M
Figure 12.17 Te efect of a change in the bond price and interest rate.
Even though the wealth efect can work against the substitution efect, the
increase in the price of bonds from PB1 to PB2 generally leads to an increase in
the portfolio demand for bills and a corresponding decrease in the demand
for bonds. In addition to the restrictions on the indiference-curve slopes em-
bodied in Assumptions (1)–(3), a suf ficient condition for this result is that the
return on the optimal holding of bonds be bounded by the expected value of
the bond endowment,
R ∗ R
Copyright © 2021. Harvard University Press. All rights reserved.
B < 1 + B.
PB PB
Otherwise, a (negative) wealth efect may swamp the substitution efect, and,
consequently, the demand for money might decrease, as is the case in Figure
12.18 at the limit A0 = M, where the new equilibrium at F involves a lower
demand for bills.
To see this, substitute from the portfolio constraint into the utility function
to transform the optimization problem from a choice of B to a choice of M:
R
U ( E( A), M ) = U M + PB B + s M +
P
( M + PB B − M ) , M . (12.8)
B
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
438 Raising Keynes
Portfolio composition possibilities
R
E(A) (1 S ) A02 2 S (A02 M )
pB
E(A)
E F
(1 S ) A02 (1 S )(PB2 B )
I3
F
I2
I1
(1 S ) A01 (1 S )(M PB1B ) R
E(A) (1 S ) A02 2 S (A02 M )
(1 S ) A02 (1 S )(M ) p
B
M
M A01 M PB1B M A02 M M A02 PB2 B
Figure 12.18 Te efect of a change in the bond price and interest rate: two limiting
cases.
Te frst-order condition for an interior solution remains the same, but now
we write the frst order condition as
R U
H ( M , PB , M , B , s ) ≡ − s − 2 = 0. (12.9)
PB U1
With PB given along with M, B, and s, we can solve equation (12.9) for the
Copyright © 2021. Harvard University Press. All rights reserved.
optimal amount of bills and write the solution as
M = M ∗ ( PB , M , B , s ).
Taking the total derivative of the function
(
H M ∗ ( PB , M , B , s ) , PB , M , B , s = 0 )
with respect to PB tells us how the demand for bills (M = M*(PB, M, B, s)) is
related to the price of bonds. We have
∂M ∗
H1 + H 2 = 0,
∂PB
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 439
so that
∂M ∗ H
= − 2.
∂PB H1
By virtue of equation (12.8), diferentiating equation (12.9) with respect to
M gives
U U −U U R U U −U U
H1 = 21 1 2 11 2 − s − 22 1 2 12 2 ,
U1 PB U1
R R
( U 21U1 − U11U 2 ) 1 + P B − B
PB
R B
H2 = − − .
( PB )
2
U 12
H1 is positive because of Assumptions (1) and (2). Assumption (1) also en-
sures H2 is negative if
R R ∗
1 + B − B ≥ 0.
PB PB
Te takeaway is that, unless the demand for bonds is large compared to the
endowment, the wealth efects of bond-price changes dominate, and the de-
mand for money or bills increases as the bond price rises and the bond yield
falls.
Equilibrium in a Representative-Agent Model
Demand is of course just half the story of asset-market equilibrium; the other
half is supply. In general, demand-supply balance in any market is an aggrega-
tive property and thus requires us to sum demands and supplies over all
Copyright © 2021. Harvard University Press. All rights reserved.
agents.
A representative-agent model (a model in which all other agents are clones
of the representative) simplifes our task: the representative agent is a micro-
cosm of the world, and the overall supply-demand equilibrium must be repli-
cated at the level of the individual agent. With identical agents, not only must
the agent’s demand for bills and bonds add up to her total wealth, but each
separate demand must be equal to the respective endowment, just as for the
population as a whole. If endowments of bills and bonds are respectively M
and B, the agent’s choice has to respect both the overall constraint
M + PB B = A0 = M + PB B
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
440 Raising Keynes
and also the individual constraints M = M and B = B. (Evidently if M = M, the
budget constraint guarantees B = B, so we don’t need to consider the second
condition separately.)
With a given endowment M, it is precisely the function of the interest rate
to bring equilibrium about, which is to say that, given the short-term interest
rate s, the equilibrium bond price PB and interest rate R/PB are endogenously
determined by the endowments to equilibrate the demand for and supply of
both bills and bonds.
Te tangencies refected in Figures 12.16 and 12.17 continue to play a role
but are necessary rather than suf ficient to characterize equilibrium. Te tan-
gency between an indiference curve and the portfolio constraint is now
subject to the conditions M = M and PB = PB(M, B, s). Tat is, the endow-
ments M and B determine the bond price PB(M, B, s) that just makes the
demand for bills equal to the supply. At the individual level, we have the
demand function
M = M ∗ ( PB , M , B , s )
and equality of demand and supply
(
M = M ∗ PB ( M , B , s ) , M , B , s = M . )
With the agent representing the cosmos, market equilibrium
∑ M = ∑ M ∗ ( PB , M , B, s ) = ∑ M
reduces to the same relationship at the level of the representative agent
M = M ∗ ( PB , M , B , s ) = M .
Does an equilibrium exist? A simple fxed-point theorem shows that a suf
fcient condition for existence of equilibrium in this model is that the econ-
Copyright © 2021. Harvard University Press. All rights reserved.
omy not be in a liquidity trap; that is, the condition s > 0 must hold. Con-
struct the demand function by starting with a price PB1 such that R/PB − s = 0.
Tat is, PB1 = R/s. At this price, it does not pay to hold bonds, and with B = 0
M + PB B = M = M + PB B > M .
Now consider the lower end of the price range. As PB → 0, we have
lim M + lim ( PB B ∗ ) = M + lim PB B = M ,
PB → 0 PB → 0 PB → 0
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 441
with the result
lim M < M if lim ( PB B ∗ ) > 0,
PB → 0 PB → 0
lim M = M if lim ( PB B ∗ ) = 0.
PB → 0 PB → 0
So we have
(
M = M ∗ PB1 , M , B , s > M , )
lim M = lim M ∗ ( PB , M , B , s ) ≤ M ,
PB → 0 PB → 0
and, since M is a continuous function of PB, there must exist a price, PB = PB*,
for which M = M. Te picture in Figure 12.19 assumes
lim M = lim M ∗ ( PB , M , B , s ) < M .
PB → 0 PB → 0
Observe that this argument fails at the zlb. When s = 0, there is no f-
nite bond price at which R/PB = 0, so there may be no fnite bond price PB1 at
which the demand for bonds vanishes and M exceeds M. Pure risk aversion is
consistent with the absence of a liquidity trap since it may be the case that for
s = 0 the equilibrium bond yield may also be zero, as in Figure 12.5(a).
M, M
M (PB1 , M, B, s )
M M (PB , M, B, s )
E
M
Copyright © 2021. Harvard University Press. All rights reserved.
lim M M
PB 0
PB PB1 PB
Figure 12.19 Equilibrium in a representative agent model.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
442 Raising Keynes
The Impact of Bill Supply on the Price of Bonds
What happens when the endowment of bills increases? Te bill supply and
the demand for bills both increase, and the result of these two counteracting
forces is ambiguous. In Figure 12.20 the equilibrium bond price rises, and the
bond yield falls, but, as we shall see, this result is guaranteed only if s is small.
Te math is as follows. If M* = M, then PB is no longer a free parameter.
Instead we have PB = PB(M, B, s) such that
( ( )
H M ∗ PB ( M , B , s ) , M , B , s , PB ( M , B , s ) , M , B , s = 0. )
If we diferentiate this equation with respect to M, we obtain
∂M ∗ ∂PB ∂M ∗ ∂PB
H1 + + H 2 + H 3 = 0.
∂PB ∂M ∂M ∂M
Since
(
M = M ∗ PB ( M , B , s ) M , B , s = M , )
we have
∂M ∗ ∂PB ∂M ∗
+ = 1,
∂PB ∂M ∂M
M, M
M M (PB , M2 , B, s )
M M (PB , M1 , B, s )
F
M2
E M1
Copyright © 2021. Harvard University Press. All rights reserved.
lim M M
PB 0
PB
Figure 12.20 An increase in the supply of bills raises the equilibrium bond
price.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 443
so
∂PB H + H3
=− 1
∂M H2
with
U U −U U R
H 3 = − 21 1 2 11 2 1 + .
U1 PB
Te sum H1 + H3 is
U U −U U U U −U U
H1 + H 3 = − 21 1 2 11 2 ( 1 + s ) − 22 1 2 12 2
U1 U1
U U −U U U 22U 1 − U 12U 2 U 21U 1 − U 11U 2
= − 21 1 2 11 2 − − s . (12.10)
U1 U 12 U 12
Te sign of the right-hand side of equation (12.10) is ambiguous since, by
Assumption (3), the sum of the frst two terms is positive, while Assumption
(1) makes the last term negative. Te ambiguity vanishes if s is small since
the last term goes to zero with s. In this case we have
∂PB +
sgn = − = +,
∂M −
which is to say that the equilibrium bond yield is a decreasing function of the
endowment of bills in the neighborhood of the zlb.
Equilibrium Spreads
For a given endowment of bills, we calculate the relationship between the bond
price and the bill rate by taking the total derivative of H(M*(PB(M, B, s),
Copyright © 2021. Harvard University Press. All rights reserved.
M, B, s), PB(M, B, s), M, B, s) with respect to s and equating it to zero.
Tis gives
∂PB
H2 + H5 = 0
∂ s
since the stock of bills is constant and therefore
∂M ∗ ∂PB ∂M ∗
+ = 0.
∂PB ∂ s ∂ s
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
444 Raising Keynes
Te derivative H5 is
U 21U 1 − U 11U 2
H 5 = −1 − M,
U 12
which is negative. So
∂PB H −
sgn = sgn − 5 = − = −,
∂ s H2 −
and in turn the equilibrium bond yield is a positive function of the bill rate.
Quantitative Easing
In this chapter (see Figure 12.14) quantitative easing is represented as an ex-
change of the public’s bonds for the monetary authority’s bills. We have M*(PB,
M, B, s) = M and B = B, so
M ∗ ( PB , M , B , s ) + PB ( M , B , s ) B = M + PB ( M , B , s ) B . (12.11)
When the central bank buys bonds from the public, the corresponding change
in the public’s stock of bills is given by total diferentiation of equation (12.11).
Since M* = M and B = B, all terms but the derivatives of M* and M cancel out
on the two sides, and we are lef with
∂M ∗ dPB ∂M ∗ ∂M ∂M ∗ dM
+ + = .
∂PB dB ∂M ∂B ∂B dB
Since ∂M/∂B = −PB and ∂M */∂B = PB∂M */∂M, we can write dPB/dB as
dM
dPB
= dB∗ .
Copyright © 2021. Harvard University Press. All rights reserved.
dB ∂M
∂PB
Because an increase in the availability of bonds must decrease the availability
of bills, the numerator is negative. So the sign of the fraction depends on the
denominator. Provided s is small, ∂M*/∂PB > 0, and we have
dPB −
sgn = = −.
dB +
In the neighborhood of the zlb, as the stock of bonds in the hands of the pub-
lic falls, the equilibrium price of bonds rises and bond yields fall.
Te dynamics of bond-price adjustment lead to a more general result,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 445
which does not rely on s being small. Assume that PB adjusts proportionately
to the excess demand for bonds
PB
= B ( B − B )
PB
and thus negatively with the excess demand for bills
(
PB = −θ B M ∗ ( PB , M , B , ρ s ) − M . )
PB must increase when the supply of bonds in the hands of the public falls and
the quantity of bills rises. We have
dPB ∂M ∗ ∂M ∂M ∗ ∂M H3 H
= − B + − = − B − ( −PB ) − 4 − ( −PB )
dB ∂M ∂B ∂B ∂B H1 H1
and
R
( U 21U1 − U11U 2 ) 1 + P PB
H4 = − B
.
U 12
Combining this equation with equation (12.10) eliminates PBH3 − H4, giving
the result
dPB
= − B PB.
dB
Te picture is in Figure 12.21. Suppose asset markets are originally in
equilibrium, with B = B and M = M. At this point PB = 0. Now let B and
P B
Copyright © 2021. Harvard University Press. All rights reserved.
P B B (M (PB , M, B, S ) M )
M M M M M M B
B B B B B B
Figure 12.21 Disequilibrium price change (P )as a function of
the quantity of bonds in the public’s hands (B ).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
446 Raising Keynes
M remain fxed while B falls and M increases in accordance with the con-
straint
M + PB B = const.
A fall in B drives up the price of bonds: to the lef of the equilibrium PB > 0.
Te increase in PB leads to a new, higher, level of M* and a lower level of B,
and eventually to a new equilibrium price PB* that exceeds the original price,
and this independently of the size of s.
Bringing in Normal Reversion
When the current short-term rate is below normal, normal reversion simply
reinforces risk aversion. Te equilibrium condition is now
R E ( ∆PB ) U 2
− ρs = − + .
PB PB U1
Te yield premium R/PB − s exceeds the risk premium U2/U1 by the absolute
value of the expected capital loss, −E(PB)/PB. Compared with the case of
pure risk aversion, in which E(PB)/PB = 0, the portfolio constraint shifs in-
ward, as in Figure 12.22.
Portfolio composition possibilities
R
E ( A ) (1 s ) A 0 s ( A 0 M )
P
B
E(A)
I3
Copyright © 2021. Harvard University Press. All rights reserved.
I2
I1
(1 s ) A 0 (1 s ) (M PB B )
R E ( PB )
E ( A ) (1 s ) A 0 s ( A 0 M )
P
B PB
M A 0 M PB B M
Figure 12.22 Normal reversion changes portfolio composition possibilities.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 447
Te more interesting case is when the current short rate exceeds the nor-
mal rate. In this case normal reversion and risk aversion are working against
one another, since −E(PB)/PB < 0 and −U2/U1 > 0. If the expected capital gain
is great enough, the combined force of normal reversion and risk aversion can
be negative
R E ( ∆PB ) U 2
− ρs = − − < 0.
PB PB U1
Tat is, the yield premium R/PB − s can become negative at the optimum F in
Figure 12.23.
free riding by agents with long habitats An agent looking to the
risk and return of her portfolio at retirement or when her last child is about to
enter university will derive positive utility from bonds and will be averse to
holding bills. Her indiference curves will have the shape in Figure 12.24, so
provided the schedule of portfolio composition possibilities has a negative
slope, the optimum will typically be a corner solution like E. Having a long
“preferred habitat,” she gets a free ride provided the yield premium R/PB − s
is positive and the expected price change is non-negative. For an interior so-
lution, in which the agent holds bills as well as bonds, the bill rate must ex-
Portfolio composition possibilities
F R E ( PB )
E(A) E ( A) (1 s ) A 0 s ( A 0 M )
PB PB
E
I3
Copyright © 2021. Harvard University Press. All rights reserved.
I2
I1
(1 s ) A 0 (1 s ) (M PB B ) R
E ( A ) (1 s ) A 0 s ( A 0 M )
PB
M A 0 M PB B M
Figure 12.23 A negative yield premium.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
448 Raising Keynes
I3
E(A)
E
I2
I1
(1 s ) A 0 (1 s ) (M PB B )
M A 0 M PB B M
Figure 12.24 Equilibrium with a long preferred habitat.
ceed the bond yield by more than the expected increase in the bond price. Te
frst-order condition remains
R E ( ∆PB ) U 2
− ρs + + = 0,
P PB U1
B
but U2 is now negative.
what if agents are risk neutral? In the absence of risk aversion,
the indiference curves become fat, refecting the assumption that no disutil-
Copyright © 2021. Harvard University Press. All rights reserved.
ity (or utility) attaches to owning bonds, which is to say that bills afect utility
only because they are part of wealth, not because of the added fexibility they
provide.
With horizontal indiference curves, agents are driven to put all their eggs
in the bond or money basket according to the sign of
R E ( PB )
+ − s ,
PB PB
the instantaneous holding yield on bonds minus the bill rate. Tree possibili-
ties are shown in the three panels of Figure 12.25.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Portfolio composition possibilities
R E ( PB )
E ( A ) (1 s ) A 0 s ( A 0 M)
P
B PB
12
E(A) E(A) E(A)
•
I4 I4 I4
I3 I3 I3
Indifference curves
(1 s ) A 0 Indifference curves
I2 I2 I2
(1 s ) (M PB B ) Indifference curves
I1 I1 I1
M M
Mathematical Appendix
M A 0 M PB B M A 0 M PB B M A 0 M PB B M
R E ( PB ) R E ( PB ) R E ( PB )
(a) s 0 (b) s 0 (c) s 0
PB PB PB PB PB PB
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Figure 12.25 Optimization with normal reversion and risk neutrality.
449
450 Raising Keynes
Te main body of this chapter laid out the mathematics for calculating an
agent’s demand price for a bond on the assumption of continuous equality
between the holding yield on a bond and the bill rate, depicted in Figure
12.25(b), coupled with the assumption of a simple linear reversion to normal
of s,
(
s = −θ ρ s − ρ∗s .
ρ ) (12.12)
If the actual bond price is lower than demand price, then the agent will devote
her portfolio entirely to bonds, as in Figure 12.25(a); if higher, to bills, as in
Figure 12.25(c). Or, as expressed in Figure 12.6, if the actual coupon yield
coup exceeds E( coup), where the expected coupon yield at t = 0 is
−1
∞ − ρ∗s τ + ρs (t ) − ρs ( e − θτ −1 )
∗
R
(
E ρcoup,0 ) = = ∫e
E ( PB , 0 ) 0
θ
dτ ,
(12.13)
then the optimal strategy is to commit 100 percent to bonds, whereas if coup <
E( coup), then the optimizing agent holds only bills.
If the evolution of the bill rate is governed by equation (12.12), we can cal-
culate equation (12.13) by a formula derived in an old paper (Marglin 1970):
−1
ρ (0) − ρ∗ −n ∞ ρ∗
−1
R 1 1 +
∞
= ∗
E ( PB , 0 ) ρ s ∑ θ ∏ θ .
s s s
+ j
n =1 j =1
the relationship between expected bond-p rice changes and
the bill rate We continue to assume normal reversion and risk-neutral
agents. Diferentiating the equation
∞ τ
E ( PB ,t ) = ∫ Re ∫ t
− ρ s ( x )dx
dτ
Copyright © 2021. Harvard University Press. All rights reserved.
with respect to t, we have
E ( PB ,t ) = −R + s E ( PB ,t )
or
E ( PB ,t ) R
=− + s,
E ( PB ,t ) E ( PB ,t )
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 451
so
E ( PB , t )
∂
E ( PB , t ) R ∂E ( PB , t )
= + 1.
E ( PB , t )
2
∂ s (t ) ∂ s (t )
Since
∞ τ ∞ ρ (t ) − ρ∗s ( − θτ − θt )
− ∫ ρ s ( x )dx − ρ∗s ( τ − t ) + s e −e
E ( PB ,t ) = ∫ Re t
d τ = ∫ Re θ
dτ,
t t
we have
E ( PB ,t )
∂
E ( PB ,t ) R
∞ 1
( e − θτ − e − θt )
∂ρ s (t )
=
E ( PB ,t )
2 ∫eθ d τ + 1,
t
and, setting t = 0,
E ( PB , 0 )
∂
E ( PB , 0 ) R
∞
e(
1 − θτ
−1 )
∂ρ s (0)
=
E ( PB , 0 )
2 ∫e θ
d τ + 1.
0
Since the right-hand side exceeds one, the expected percentage change in
the bond price is in the same direction as a change in the bill rate, but greater
than the absolute change in the bill rate.
Equilibrium with Normal Reversion and Risk Neutrality
Suppose there are n agents, each having the same endowment of bills and the
Copyright © 2021. Harvard University Press. All rights reserved.
same endowment of bonds. Each agent has her own forecast of the normal
rate and the speed of reversion to that rate, so each has her own tipping point
between bonds and bills. Te total demand for bills is the sum of the individ-
ual demands
m ( PB , s ) ( M + PB B ),
where m(PB, s)is the number of agents desiring to hold only bills, and M + PBB
is the value of the individual endowment.
When we include individuals on the margin of indiference, m can take on
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
452 Raising Keynes
fractional values. Suppose we order individuals by the bond price that tips
them from bonds to bills. If m(PB0, s ) maps to the interval between individual
20 and individual 21, this means that, when PB = PB0, twenty agents desire to
hold only bills and the twenty-frst agent, who is indiferent between bills and
bonds, is willing to hold any fraction of her wealth, from zero to one, in bills.
Both elements of the demand for money, m(PB, s ) and (M + PBB), are in-
creasing in PB. Figure 12.26 shows the relationship between the price of bonds
and the number of agents, including fractional agents, wishing to hold only
bills, and Figure 12.27 shows the demand for bills as well as the aggregate sup-
ply, nM. At the equilibrium E, one agent holds both bonds and bills.
Te math is straightforward. Equilibrium is characterized by the equality
m ( PB , s ) ( M + PB B ) = nM
or
mPB B = (n − m)M . (12.14)
As before, let
PB = PB ( M , B , s ).
Diferentiate equation (12.14) with respect to M. Te result is
∂m dPB dP
∂PB dM
( M + PB B ) + m 1 + B B = n
dM
m (PB , s )
Copyright © 2021. Harvard University Press. All rights reserved.
m (PB , s )
PB
Figure 12.26 Number of agents desiring to hold bills in relation to
bond price.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 453
or
dPB n−m
= .
dM ∂m
∂PB
( M + PB B ) + mB
Te expression on the right is positive unless everybody is 100 percent into
bills, so just as in the case of pure risk aversion, the bond yield falls as the
endowment of bills rises. Tis establishes the downward-sloping liquidity-
preference function in the space of coup × M.
For given endowments of bonds and bills, the equilibrium bond price de-
creases (or at least does not increase) as the bill rate increases. Diferentiating
equation (12.14) with respect to s for fxed M and B, we obtain
∂m
dPB ∂ s
( M + PB B )
=− .
d s ∂m
∂PB
( M + PB B ) + mB
Since ∂m/∂s ≥ 0 and ∂m/∂PB ≥ 0, we have dPB/ds ≤ 0, so that the bond
price cannot increase with the bill rate, and the bond yield is a nondecreasing
function of the bill rate. Figure 12.28 shows the mappings m(PB, 1s ) and
m(PB, 2s ) with 2s > 1s . Figure 12.29 shows the two corresponding demand
schedules.
What about the liquidity trap? Te chapter 11 version of the liquidity trap
m (PB , s ) (M PB B ) , nM
m (PB , s ) (M PB B )
Copyright © 2021. Harvard University Press. All rights reserved.
nM
E
PB
Figure 12.27 Demand and supply in bills market.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
454 Raising Keynes
m (PB , s )
m (PB , 2s )
m (PB , 1s )
PB
Figure 12.28 Number of agents desiring to hold bills as the bill rate
varies.
m (PB , s ) (M PB B ) , nM m (PB , 1s )(M PB B )
m (PB , 2s )(M PB B )
F
nM
E
Copyright © 2021. Harvard University Press. All rights reserved.
PB
Figure 12.29 Demand and supply in bills market as the bill rate varies.
is a positive lower bound to the coupon yield coup and hence an upper bound
to PB as M increases without bound. Write the tipping-point price of the nth
agent (the last agent to switch from bonds to bills as PB increases) as PBn. Since
for PB > PBn we have
m ( PB , s ) ( M + PB B ) = n ( M + PB B ) > nM ,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Mathematical Appendix 455
so that the demand for bills exceeds the supply. At the upper limit PBn we write
m as n − 1 + n, where n is the proportion of bills in the wealth portfolio of
the nth agent. Te demand for bills is
(n − 1 + (
n ) M + PBn B . )
Equation (12.14) becomes
(n − 1 + n ) PBn B = ( 1 − n ) M .
Demand and supply are maintained in balance as M increases because n in-
creases in tandem. n → 1 in the limit as M → ∞.
In this chapter, a liquidity trap exists if the equilibrium bond price remains
fnite when the bill rate is at its lower bound, s = 0. For a given endowment of
bills and bonds, the proof of the existence of a liquidity trap in this sense par-
allels the proof of the existence of equilibrium in the risk-aversion case repre-
sented in Figure 12.19. For PB > PBn, the demand for bills exceeds supply. At
PB = 0 there is no demand for bills. Since the mapping m(PB, 0)(M + PBB) is
continuous, it must intersect the supply schedule somewhere on the interval
[0, PBn].
The Spread between Treasury Yields and Corporate Yields
Liquidity preference is supposed to lead to a theory of how the hurdle rate is
determined, but so far I have limited my attention to T-bills and T-bonds. To
fnish the story, imagine a separate market in which Treasury and corporate
bonds are traded.2 Portfolio possibilities are given by
PCORP BCORP + PB B = PCORP BCORP + PB B ,
where PCORP and BCORP are prices and quantities of corporate bonds, and
B CORP is the endowment of corporate bonds. As before, PB and B are the price
Copyright © 2021. Harvard University Press. All rights reserved.
and quantity of Treasuries, and B is the agent’s endowment of Treasuries. Ex-
pected wealth is
R E ( ∆PCORP ) R E ( ∆PB )
E( A) = 1 + CORP + PCORP BCORP + 1 + + PB B
PCORP PCORP PB PB
R E ( ∆PB )
= PCORP BCORP + PB B + + PB B
PB PB
R E ( ∆PCORP )
+ CORP + ( PCORP BCORP + PB B − PB B ) .
PCORP PCORP
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
456 Raising Keynes
With the utility function now
U ( E( A), PB B ) = U ( E( A), PCORP BCORP + PB B − PCORP BCORP ),
the frst-
order condition characterizing agents who hold both types of
bonds is
R E ( ∆PB ) RCORP E ( ∆PCORP ) U 2
+ − + = .
PB PB PCORP PCORP U1
As before, U1 refects the marginal utility of expected wealth, but U2 now
measures the comparative safety of holding Treasuries relative to holding cor-
porates. Assuming that the two types of bonds are of comparable maturities
eliminates diferences in price risk that are independent of default risk, so that
the diference
E ( ∆PB ) E ( ∆PCORP )
−
PB PCORP
measures the default risk on the corporate bond. Now the optimization con-
dition is that the yield premium on corporate bonds
RCORP R
CORP − coup = −
PCORP PB
is equal to the sum of the default risk and the illiquidity premium,
E ( ∆PB ) E ( ∆PCORP ) U2
− + .
PB PCORP U1
In the absence of risk aversion, U2 = 0, so the yield premium is exactly equal
to the default risk. With risk aversion, the yield premium must be higher than
the default risk to compensate for the aversion to loss.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. empirical appendix .
What Do the Data Say?
Tis appendix examines the evidence for liquidity preference as a theory of
how interest-rate spreads are determined by expectations of the course of fu-
ture short-term interest rates. More specifcally, the focus is on sorting out the
relative importance of reversion to normal, risk aversion, and default risk in
determining the spread between the hurdle rate of interest and the return on
safe, short-term assets.
Near the end of the last century, John Campbell ofered an authoritative as-
sessment of the empirical state of play of interest-rate determination, based
on his own joint work with Robert Shiller as well as on his comprehensive
knowledge of a vast literature. Campbell argued that the data contradict a key
provision of pure normal reversion (the “pure expectations hypothesis” in
Campbell’s terminology).1 Te hybrid hypothesis that both normal reversion
and risk aversion are at work (Campbell’s “expectations hypothesis”) doesn’t
fare any better. Campbell regresses changes in bond yields on spreads be-
tween yields on bonds of various maturities and the short rate and fnds
In these regressions, the spread is scaled so that if the expectations hypoth-
esis holds, the slope coef ficient should be one. In fact, all but one of the slope
coef ficients are negative; all are signifcantly less than one, and some are sig
nifcantly less than zero. When the long-short yield spread is high, the long
Copyright © 2021. Harvard University Press. All rights reserved.
yield tends to fall, amplifying the yield diferential between long and short
bonds, rather than rising to ofset the yield spread as required by the expec-
tations hypothesis. (1995, pp. 138–139)
Nobody to my knowledge has contradicted Campbell’s assessment, and a
more recent summary of the term-structure literature (Gürkaynak and Wright
[2012]) affi rms Campbell’s position. Campbell is right that the data contradict
the expectations hypothesis, in both its pure and hybrid forms, at least as the
post–World War II literature has framed the argument. Te problem is that
Campbell and others neither ofer nor test a theory of the expectations hy-
pothesis, pure or hybrid, but rather a joint test of the expectations hypothesis
457
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
458 Raising Keynes
and a very specifc—and highly implausible—theory of how expectations are
formed.
Beginning in the 1970s, the expectations hypothesis was married to ratio-
nal expectations, an idea that would have been summarily rejected by Keynes.2
In its new life the hypothesis being tested became a hypothesis not only that
present spreads depend on future short rates but also that agents correctly
predict the future (up to a random error). Framed this way, looking for con
frmation of the expectations hypothesis is a fool’s errand. It is hard to see how
any agent—except perhaps Nostradamus—could correctly forecast future
rates ten years into the future, as agents endowed with rational expectations
must be assumed to do when they calculate expected yields on 10-year bonds.
As in the main body of this chapter, I assume that short and long rates re-
late to one another both through risk aversion and through expectations
formed on the hypothesis that the short rate will revert to a normal rate taken
for the purposes of this exercise to be fxed exogenously. Tese expectations
are not “rational.” Neither are they irrational. Tey simply are.
From Consols to Long Bonds, From Coupon
Yields to Yields to Maturit y
Before we turn to empirical tests of various aspects of liquidity-preference
theory, the theory has to be modifed to take account of an important difer-
ence between the models we have laid out in chapters 11 and 12 and the real
world, namely, the virtual absence of consols from bond markets. In the
United States, consols have never been a regular part of Treasury debt, though
apparently some of the debt issued in connection with fnancing the Panama
Canal, long since retired, took this form.3 In the United Kingdom, the original
home of the consol, this particular debt instrument has gone the way of the
dodo (see chapter 11, note 5).
Tere is instead a spectrum of Treasury obligations of various fnite ma-
Copyright © 2021. Harvard University Press. All rights reserved.
turities, running, in the United States, from one day to thirty years. Tose
with a term of more than a year are issued as “coupon” bonds, paying interest
twice a year. Short (one year or less) instruments are bills issued at a discount
to their value at maturity. Returns to obligations maturing at diferent times
are conventionally summarized in a yield curve, representing the relationship
between the yield to maturity and the bond’s term. Figure 12.30 represents the
yield curve for Treasury obligations in March 2014, when the short-term rate
was virtually zero and the 30-year bond, maturing in March 2044, ofered a
yield to maturity of 3.6 percent.
By defnition, the yield to maturity is the interest rate that just makes the
present value of the bond’s lifetime returns equal to the current price of the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 459
0.04
0.035
Yield to maturity 0.03
0.025
0.02
0.015
0.01
0.005
0.0
0 5 10 15 20 25 30
Bond term in years
Figure 12.30 Yield curve (March 2014).
bond, P. In continuous time, assuming the bond is redeemed for $1.00 at time
t + m,
t +m
∫t
−ρ ( m , t ) ⋅( τ − t ) −ρ (m, t )⋅ m
P (m, t ) = Re mat d τ + e mat
=−
R
ρmat (m, t )
e mat (
− ρ (m, t )⋅m − ρ (m, t )⋅ m
− 1 + e mat , )
where R is the (annualized) return, t is the calendar date at which the bond
price is evaluated, and m is the term to maturity. For notational convenience,
the subscript in the expression for the bond price is omitted in this appendix.
Te bond price is denoted by P or by P(m, t), showing the functional depen
dence of the bond price on both the time to maturity (m) and the calendar
date (t) when the extra detail is helpful. (Te price of goods, which was previ-
ously denoted by P, does not enter the analysis here.)
For a consol, we can defne the yield to maturity as mat(∞) = lim mat(m, t);
m→∞
the above equation reduces to
Copyright © 2021. Harvard University Press. All rights reserved.
R
P (m, t ) =
mat (∞)
so that
R
mat (∞) = coup ≡ .
P
Tat is, the limiting yield to maturity is equal to the coupon yield R/P.4 By
contrast, in the case of fnite maturities, the two yields will generally difer.
Only when the bond trades at par, that is, at its redemption value, P = 1, will
the two conceptually distinct yields coincide in value.
Whatever the term to maturity, agents continue to anticipate the holding
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
460 Raising Keynes
yield as the sum of the coupon yield and the expected (percentage) change in
the bond’s price,
R E ( P )
hold (m, t ) ≡ + .
P (m, t ) P (m, t )
And market equilibrium, as characterized by equality between the expected
holding yield on long bonds and the sum of the short rate plus a risk pre-
mium, U2/U1, continues to hold for fnite-maturity bonds, except that U2/U1
now depends on the term to maturity, since the longer the life of the bond,
the greater the expected variability of its price and the greater the sensitivity
of utility to the value of bonds in the portfolio. Denoting the risk premium
U2/U1 by (m), we have the equilibrium condition for agents to hold both
bonds and bills:
R E ( P )
+ = ρ s (t ) + α(m).
P P
Te spread between the coupon yield and the short-term rate continues to
be equal to the diference between the risk premium and the change in price:
R E ( P )
− ρ s (t ) = α(m) − .
P P
When we were dealing with consols, this result made it easy to deal with the
two polar cases where only risk aversion or only normal reversion is at play.
Te frst assumption—no risk aversion—implies = 0, whereas the second
assumption—no reversion to normal—implies E(P)/P = 0. Te frst of these
two assumption carries over to a world of fnite bond maturities: the absence
of risk aversion implies (m) = 0 whatever the term of the bond. But the ab-
sence of reversion to normal does not imply E(P)/P = 0 when m is fnite.
Unlike consols, every fnite maturity bond is characterized by a terminal
Copyright © 2021. Harvard University Press. All rights reserved.
condition, namely, a condition that the price must approach the redemption
value of the bond as we approach the redemption date.5 Although this termi-
nal condition is independent of whether risk aversion or normal reversion
drives bond prices, its efect plays out very diferently in the two cases.
In the case of pure risk aversion (no reversion to normal), expected bond
price and yield will have the general shape of Figure 12.31. Tis fgure as-
sumes that the short-term rate is zero when the bond is issued (March 2014),
and will continue to be equal to zero as far out as the eye can see (or rather the
mind can imagine). On this assumption it represents the trajectory of the
price and yields (both the coupon yield and the yield to maturity) of a 30-year
bond with a par value of $1.00 and a coupon of $0.036.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 461
0.04 1.12
0.035 1.10
Bond yields and bill rates
0.03 1.08
Bond price
0.025
1.06
0.02
1.04
0.015
0.01 1.02
0.005 1.00
0.0 0.98
Mar 2014 Mar 2024 Mar 2034 Mar 2044
Yield to maturity Coupon yield
Bill rate Bond price (right axis)
Figure 12.31 Bond yields, bond prices, and bill rates: risk aversion without
reversion to normal.
How do we account for the shapes of the various schedules in Figure 12.31?
Observe frst that the yield-to-maturity schedule is the mirror image of the
yield curve in Figure 12.30: the long bond starts life as a 30-year bond but
over time morphs into bonds of successively shorter maturities. At every
point in the bond’s life, the condition of market equilibrium is that the hold-
ing yield (R/P) + (E(P)/P) on the long bond equal the short rate s plus a risk
premium (m), where m is the time remaining until the bond matures.
In 2043 the 30-year bond issued in 2014 will be equivalent to a 1-year bill
in price risk, and, therefore, equivalent in its return to a short-term bill issued
in 2043 (apart from tax treatment and bid-ask liquidity considerations). Tis
is to say that in 2043 the 30-year bond issued in 2014 requires a very small
premium to ofset its price volatility. With the 1-year bill rate expected to re-
main at the near-zero level obtaining in 2014, and with virtually no risk pre-
mium on the 2014 vintage bond ( ≈ 0), in 2043–44 the holding yield on this
Copyright © 2021. Harvard University Press. All rights reserved.
bond is now expected to be near zero. But this can only happen if the price is
expected to fall by about as much as the coupon, namely, by about $0.036.
Tis tells us that the price in 2043 must be in the vicinity of $1.036 since it is
(by assumption) certain that the bond will be redeemed for $1.00 in 2044.
What comes down must frst have gone up. Long bonds, by assumption,
start life at par, so the price of the 30-year bond issued in 2014 must rise early
in its life to be able to come down at the end of its life. If, as expected, the
bond price begins to rise upon issuance, both the yield to maturity and the
coupon yield initially fall since the two yields start out life together.
Evidently, it is no longer the case that E(P)/P = 0. Rather, the expected
value of bond-price changes with the remaining maturity of the bond. When
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
462 Raising Keynes
risk aversion rules, price changes are expected to be positive at the beginning
of the bond’s life and negative at the end.
Tis contrasts sharply with the expected price behavior of the same 30-year
bond in a world of reversion to normal, without risk aversion. Figure 12.32
shows the expected course of the yield to maturity, the coupon yield, and the
price of the same 30-year bond issued at par with a coupon rate of 3.6 per-
cent, along with the expected course of the short-term bill rate, under the as-
sumption that the bill rate will revert to a normal rate of 6 percent at the rate
= 0.25, afer an initial three-year period at zero and two more years of
slower adjustment. Te diferences between Figures 12.31 and 12.32 evidently
hinge on the diferent assumptions about the trajectory of the bill rate as well
as on the relationship between the holding yield and the short rate. Because
the short-term rate is expected to rise, the yield to maturity on the long bond
must also rise as we approach the redemption date. Te bond price must ini-
tially fall since market equilibrium continues to require
R E ( P )
+ = ρ s + α(m).
P P
But in the absence of risk aversion, = 0 for all m, not just at the short end
of the term structure, so when the bond is issued at par, the expected bond
price must fall in order to equalize returns on bills and bonds. As the price
falls, the coupon yield rises because the coupon yield is the ratio of a fxed
coupon to a varying price. Once the short-term rate catches up to the coupon
yield, the downward price trajectory is reversed, and the price once again
reaches par when the bond is redeemed.6
Observe that the yield curve, a static picture at one point in time, itself tells
0.07 1.05
Bond yields and bill rates
0.06 1.00
Copyright © 2021. Harvard University Press. All rights reserved.
0.05 0.95
Bond price
0.04 0.90
0.03 0.85
0.02 0.80
0.01 0.75
0.0 0.70
Mar 2014 Mar 2024 Mar 2034 Mar 2044
Yield to maturity Coupon yield
Bill rate Bond price (right axis)
Figure 12.32 Bond yields, bond prices, and bill rates: reversion to normal without
risk aversion.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 463
us nothing, or rather very little, about the roles of risk aversion and normal
reversion in the determination of the spread between long and short rates.
Te only diference between the two theories that might reveal itself in the
yield curve is the incompatibility between pure risk aversion and a downward-
sloping (inverted) yield curve. As has been observed, Hicks argued that a
downward-sloping yield curve presupposes that “the short rate is regarded as
abnormally high” (1946 [1939], p. 152).7
In any case, an upward-sloping yield curve is consistent both with risk
aversion and with reversion to normal. To demonstrate this consistency, Fig-
ure 12.33 shows the actual yield curve for March 2014, alongside a hypotheti-
cal yield curve constructed on the basis of reversion of a zero-coupon bond
(see below) to normal, with = 0.8 and *s = 0.0385. Evidently, the hypotheti-
cal yield curve does a good job of approximating the actual curve. If, however,
we project the curves fve years into the future, from 2014 to 2019, and as-
sume that the actual 2014 curve is based solely on risk aversion, the resulting
curves behave very diferently. Pure risk aversion implies that the yield curve
does not change over time (unless wealth holders become more or less risk
averse); pure normal reversion implies that the yield curve fattens out. So,
under the hypothesis of pure risk aversion, we would expect a 30-year bond
issued in 2019 to have the same yield to maturity as one issued in 2014. Under
normal reversion, the 30-year bond issued in 2019 is expected to yield the
average of short-term rates forecast in 2014 for fve to thirty-fve years hence.
Figure 12.34 pictures 2014 predictions for the two yield curves in 2019, both
of them now hypothetical.
0.04
0.035
0.03
Yield to maturity
0.025
Copyright © 2021. Harvard University Press. All rights reserved.
0.02
0.015
0.01
0.005
0.0
0 5 10 15 20 25 30
Bond term
Hypothetical yield curve, 0.8, s 0.0385
Actual yield curve, March 2014
Figure 12.33 Yield to maturity (2014).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
464 Raising Keynes
0.045
0.04
0.035
Yield to maturity
0.03
0.025
0.02
0.015
0.01
0.005
0.0
0 5 10 15 20 25 30
Bond term
Hypothetical yield curve, 0.8, s 0.0385
Actual yield curve, March 2014
Figure 12.34 Yield to maturity (2019).
Sorting Out Risk Aversion and Normal Reversion
in a World of Zero-Coupon Bonds
Coupon bonds are the real-world norm, but in exploring the implications of
risk aversion and normal reversion, it is useful to assume that all bonds pro-
vide returns not by periodic payments of interest but by virtue of a diference
between the lower price paid at the time of purchase and the redemption
value when the bond matures (assumed to be one dollar). Such bonds, so-
called zero-coupon bonds, are unavailable even in theory in a world of con-
sols because nobody would hold a bond that ofers no periodic payment and
will never be redeemed. But in the real world of fnite maturities, zero-coupon
Copyright © 2021. Harvard University Press. All rights reserved.
bonds have become the focus of both theoretical and empirical research be-
cause it is simpler to analyze a bond with only one payment than a bond with
periodic payments and a fnal payment of a diferent amount. Real-world, m-
year coupon bonds can be understood as composite securities put together
from m zero-coupon bonds, each corresponding to a single payment of inter-
est (assumed to take place once per year), with the last payment including the
repayment of principal.8
One obvious diference between zero-coupon and ordinary bonds is that
ordinary bonds generally begin their lives at par (and are referred to as “par
bonds”) and, as Figures 12.31 and 12.32 show, can be expected to fuctuate
diferently according to whether risk aversion or normal reversion is operat-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 465
ing on bond prices. By contrast, the price of a zero-coupon bond increases
over time regardless of whether risk aversion or normal reversion is the driv-
ing force. Or, rather, bond prices will be expected to increase. Actual bond
prices may fall, but nobody will hold a zero-coupon bond that is expected to
fall in price as long as holding money is costless. By contrast, yields to matu
rity, which difer from changes in bond prices when yields change over time,
can be expected to rise or fall, depending on whether risk aversion or rever-
sion to normal is calling the tune.
As is the case for par bonds, the trajectories of two zero-coupon bonds of the
same maturity and the same initial yield to maturity—one refecting normal
reversion and the other risk aversion—must start out together and end at the
same point. Te diference between the price trajectories in Figures 12.31 and
12.32 translates into the diference in Figure 12.35. Both bonds start from a
value of just over $0.30 when frst issued, corresponding to a yield to maturity
of 0.038. But the slope of the schedule depicting normal reversion is fat at the
outset, and the schedule depicting risk aversion is fat when the bond matures.
To see why, start from the relationship between the price of a zero-coupon
bond and the yield to maturity. By defnition, the yield to maturity is the in-
1.20
1.00
0.80
Bond price
0.60
Copyright © 2021. Harvard University Press. All rights reserved.
0.40
0.20
0.0
Mar 2014 Mar 2024 Mar 2034 Mar 2044
Price with normal reversion Price with risk aversion
Figure 12.35 Anticipated trajectories of zero-coupon 30-year bond price in limiting
cases.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
466 Raising Keynes
terest rate that makes the return on purchasing the bond today equal to the
value of the bond at maturity:
mat (m, t )⋅m
P (m, t )e = 1.
Equivalently, the price today is equal to the present value of the redemption
price, discounted at the yield to maturity,
− mat (m, t )⋅m
P (m, t ) = e .
Te holding yield on the bond is just the (percentage) rate of price apprecia-
tion,9
P d ˆ mat
mat m,
P dt
where d ˆ mat /dt is the total derivative of mat, taking account of the interde
pendence between the remaining time to maturity m and the calendar time t,
m = T − t, with T the redemption date:
d ˆ mat ˆ
m mat m mat m .
dt t m
Te holding-yield condition is that the expected price increase be equal to
the sum of the short rate and the risk premium
P d ˆ mat
mat m ˆ s (t ) (m).
P dt
Te frst-order diferential equation
d ˆ mat ˆ (t ) (m)
mat s
dt m m
Copyright © 2021. Harvard University Press. All rights reserved.
has the solution
mat (m, t ) m 1 ˆ s (t )dt (m)dt m 1c
m 1 ˆ s (t )dt (m)dm m 1c.
To solve for the constant term, c, we take limits
ˆ s (t )( m) (m)m c
lim mat (m, t ) mat (0, T ) lim .
m 0 m 0 m m
Since mat(0,T) = lim ˆ s (t) = ˆ s(T), we have c = lim (m)m = 0. Hence
m o m→o
mat (m, t )m ˆ s (t )dt (m)dm.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 467
In words: the holding-yield condition implies that discounting at the yield
to maturity is equivalent to discounting at the average expected short-term
rate plus the corresponding risk premium. Te relationship between bond
price, yield to maturity, and short-term rates is
t m
t m m m t ˆ s ( ) d
ˆ s (t )d ( )d
,
mat ( m,t )m
P (m, t ) e e t 0
e t
where (m + t − ) is the risk premium at time , (m + t − ) refecting the
time remaining until the bond matures, and ˆ s () representing the estimate
of the short-term interest rate at . Te holding-yield condition can be de-
rived from this equation by taking total derivatives on both sides, but there is
new information in the limits of integration. Take logarithms on both sides
t m
ln P (m, t ) mat m (m t ) ˆ s ( ) d
t
and then take partial derivatives with respect to both calendar time t and ma
turity m. We obtain two new conditions that characterize the instantaneous
relationship between the yield to maturity, forecasts of short-term rates, and
the risk premium.
First we diferentiate the discount factor matm with respect to t, holding m
constant. Te resulting partial derivative ˆ mat/∂t is the estimate of the rate of
change over time of the yield to maturity on an m-period bond, holding ma
turity constant, that is, the rate of change of the constant-maturity yield:
ˆ mat t m
m (0) ˆ s (t m) (m) s (t) t d .
t t
Te last term can be seen to be equal to (m) − (0) by making the substitu-
tion = − t and noting that
Copyright © 2021. Harvard University Press. All rights reserved.
t +m ∂α m dα
∫t ∂t
d τ = −∫ −
0 dω
d ω = α(m) − α(0).
Tus the equation reduces to the consistent forecasting condition
ˆ mat
m ˆ s (t m) s (t ),
t
which says that the anticipated change in the constant-maturity discount fac-
tor, matm, must equal the diference between the expected gain at the end of
the bond’s life, ˆ s(t + m), and the loss at the beginning, s(t).10
A second condition is obtained by diferentiating matm with respect to ma
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
468 Raising Keynes
turity, holding time constant. Te derivative ∂mat/∂m is the rate of change of
the yield to maturity along the yield curve. (Since the yield curve is known to
the agent, this is not an estimate; hence no hat.) Tis gives the forward-rate
condition
mat (m, t ) mat m ˆ s (t m) (m).
m
Te lef-hand side is the forward rate implicit in the yield curve. Te for-
ward rate is the yield that an agent can obtain in the future if she enters into a
swap in which she makes ofsetting sales and purchases of bonds that mature
at time t + m and time t + m + . If time is divided into discrete periods, we
can imagine an agent making a short sale of one bond maturing m − 1 periods
hence and with the proceeds buying bonds maturing m periods hence. At the
present time she need make no cash outlay; the cash outlay takes place m − 1
periods from now, when she must redeem the bond she sold short today. A
market equilibrium in which agents hold both long-and short-term securities
requires the forward rate to be equal to the expected short rate plus the risk
premium.11
Why does the forward rate equal the sum of the yield to maturity, mat(m, t),
and the maturity-weighted change along the yield curve, (∂mat/∂m)m? An
example might help. Suppose time is divided into discrete periods of one year
each. Assume that for bonds with terms of zero to four years, the yield to ma
turity is 1 percent; that is, mat(4, t) = 0.01, and that the yield to maturity on
5-year bonds is 2 percent, mat(5, t) = 0.02. Te agent of the previous para-
graph—let’s call her Naomi—sells a 4-year bond short and buys 5-year bonds.
With continuous compounding, the bond she sells, with a redemption value
of $1.00, nets her $0.961. Tis allows her to buy 1.062 5-year bonds, since the
5-years are worth $0.905 each.
Te cost of the 5-year bond is ofset for four years by the short sale of the
4-year bond. During this period Naomi has no money at risk and earns noth-
ing. At the end of the fourth year, however, Naomi has to return the bond she
Copyright © 2021. Harvard University Press. All rights reserved.
borrowed to initiate the process, and to do so she has to lay out $1.00 to make
the lender of the bond whole. At the end of year fve, she receives $1.062 when
it comes time to redeem the 5-year bonds. Te net return of $0.062 can be
broken down into a 1 percent yield (the yield on 4-year bonds), plus an addi-
tional 1 percent per year over fve years, or 6 percent in all,
∆ρmat
ρmat (4, t ) + × 5 = 0.01 + (0.01 × 5),
∆m m = 5
which is precisely the formula on the lef-hand side of the forward-rate condi-
tion.
Equilibrium requires that the forward rate equal the expected short rate at
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 469
time t + m plus a risk premium (m). Why is a risk premium necessary to
equilibrate expectations about future short rates and opportunities for future
gain available with certainty today (assuming no default)? If the short rate
expected in year fve is say, 2 percent, how could Naomi’s ofsetting transac-
tions yield 6 percent? Afer all, she lays out $1.00 at the end of year four and
reaps $1.06 at the end of year fve, a short-term return on a short-term invest-
ment. Te answer is that, even though no cash is required of Naomi until year
fve, she is committed to the transaction from the get-go and thus runs the
price risks that would accompany premature (literally) unwinding of her po-
sitions. Tis commitment, rather than the actual laying out of cash, is the
reason for a risk premium in the frst place.
Te forward rate provides only upper bounds for ˆ s(t + m) and (m), be-
cause it tells us only what the sum ˆ s(t + m) + (m) must be when Naomi is
indiferent between committing to the (partially) ofsetting purchase and sale
today and holding short-term bonds in the future.12 A given value of the for-
ward rate is compatible with any combination of the expected short-term rate
and the risk premium that sums to this value. Tat is, any given forward rate
is compatible with a high value of ˆ s(t + m) and a low value of (m), or vice
versa. Of course, the total absence of risk aversion implies that the forward
rate on long bonds must be exactly equal to the expected future short rate for
the forward market to be in equilibrium. At the other extreme, the absence of
expectations of reversion to normal means that the expected short rate is
equal to today’s short rate. In the example, a forward rate of 6 percent for year
fve is compatible with an expected short rate of 6 percent coupled with a risk
premium of zero, or a risk premium of 0.06 coupled with an expected short
rate of zero—or with in-between values that sum to 0.06.
If we combine consistent forecasting with the forward-rate condition, we
obtain the holding-yield condition
P mat ˆ
mat (m, t ) m mat m s (t ) (m).
Copyright © 2021. Harvard University Press. All rights reserved.
P m t
Rearranging terms, we obtain a relationship between the expected change in
the yield to maturity (taking account of the change in term to maturity) and
the spread between long-bond yield and short-term rate:
ˆ mat
m mat m mat (m, t ) s (t ) (m).
t m
Analysis of this equation allows us to understand the very diferent ways
that expectations have been understood to interact with interest rates. As I
have argued in this chapter, Keynes’s liquidity-preference theory is really a
theory of how interest-rate spreads are determined by expectations about the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
470 Raising Keynes
future course of interest rates and risk aversion. So if causality is read as going
from right to lef, we have
ˆ mat
mat (m, t ) s (t ) m mat m (m),
t m
with the yield premium the dependent variable and the terms on the right-
hand side the independent variables. Te key to teasing out the separate im-
pact of risk aversion from the separate impact of normal reversion is the be-
havior of ˆ mat /∂t and ∂mat/∂m, particularly how these terms relate to the
current short-term rate of interest. If we combine the forecasting-consistency
condition
ˆ mat
m ˆ s (t m) s (t )
t
and the forecasting equation
ˆ s ( ) 1 e (t ) s e (t ) s (t ),
the solution to the diferential equation
mat ˆ
mat (m, t ) m mat m s (t ) (m)
m t
becomes
m 1 − e − θm ∗ 1 − e − θm
ρmat (m, t ) = m −1 ∫ α(τ)d τ + 1 − ρ s + ρ s (t ) + m −1c .
0
θm θ m
We solve for the constant term c by invoking the boundary conditions,
mat(0, t) = s(t) and (0) = 0.
Te solution is c = 0, so the spread is given by
Copyright © 2021. Harvard University Press. All rights reserved.
m 1 − e − θm ∗ 1 − e − θm
ρmat (m, t ) − ρ s (t ) = m −1 ∫ α(τ)d τ + 1 − ρ s + − 1 ρ s (t ).
0
θm θm
Te two limiting cases of no risk aversion and no normal reversion are char-
acterized by
1 − e − θm ∗ 1 − e − θm
ρmat (m, t ) − ρ s (t ) = 1 − ρ s + − 1 ρ s (t )
θm θm
no risk aversion: () = 0, ∈ [0,m],
m
ρmat (m, t ) − ρ s (t ) = m −1 ∫ α(τ)d τ no normal reversion: = 0.
0
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 471
0.05
0.04
Yield to maturity
0.03
0.02
0.01
0.0
0 5 10 15 20 25 30
Bond term
Figure 12.36 Yield curve (March 2007).
One consequence of assuming away normal reversion is to limit the yield
curve to the general shape of Figure 12.30, with a positive yield premium, one
that increases with the term to maturity. If we add the assumption of an upper
limit to (m) as m → ∞, we obtain an asymptotic limit to the yield to matu
rity. Te absence of normal reversion thus rules out inverted yield curves,
yield curves along which the spread between long and short bonds falls with
the maturity of the bond. Assuming agents are focused on short-term wealth,
the fact that from time to time we actually observe inverted yield curves, as
for example in early 2007 in Figure 12.36, implies that risk aversion is not the
whole story.
Testing Whether Both Risk Aversion and
Normal Reversion Matter
Which is it, risk aversion or normal reversion—or both? Testable diferences
Copyright © 2021. Harvard University Press. All rights reserved.
between risk aversion and normal reversion follow from the implications of
the two hypotheses for regressions of the yield premium on the short-term
rateals
mat(m,t) − s(t) = a0 + a1 s(t) + .
In the absence of risk aversion, this regression should give
1 − e − θm ∗
a0 = 1 − ρ s > 0,
θm
1 − e − m
−1 < a1 = − 1 < 0.
m
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
472 Raising Keynes
And in the absence of normal reversion the resulting coef
ficients should be
m
a0 = m −1 ∫ α(τ)d τ > 0,
0
a1 = 0.
In between the two limiting cases, we would expect
m 1 − e − θm ∗
a0 = m −1 ∫ α(τ)d τ + 1 − ρ s > 0,
0
θm
1 − e − m
−1 < a1 = − 1 < 0,
m
with both the constant term (a0) and the coef ficient on the short rate (a1)
larger in absolute value the greater is m.
Regressing the spread mat − s (or the bond yield mat) against the short-
term rate is problematic if the error term is correlated with s. Te standard
remedy is to change the regression technique, to abandon ordinary least
squares (OLS) in favor of two-stage least squares (TSLS). In the frst stage
of the process one or more new variables are used as instruments to replace
actual observations of the independent variable ( s) with synthetic observa-
tions that are stripped of the ofending correlation with . For this process to
provide statistically consistent estimates of the regression coef ficients, the
instrument(s) should meet two criteria, relevance and exogeneity. An instru-
ment is relevant if it accounts for the variation in the independent variable; it
is exogenous if it is uncorrelated with the error term.
Te instrument deployed in the regressions that follow is the rate of infa
tion. Tis instrument is obviously relevant, but its exogeneity is less clear
since it depends on what is assumed about the relationship between infation
and long-bond yields. For now I assume that infation directly afects only the
Copyright © 2021. Harvard University Press. All rights reserved.
current short rate and that the direct efect feeds through to long-bond yields
via expectations about future short rates, which are still assumed to be driven
by the equation
ˆ s ( ) 1 e (t ) s e (t ) s (t ).
Te key point is that the normal rate of interest is assumed to be unafected
by the current infation rate. On this assumption, the rate of infation will not
be correlated with the error term, so that the exogeneity condition is satis
fed.13
Infation (INF) is measured by the urban Consumer Price Index (CPI).
Tese data, like interest-rate data, are reported on a monthly basis, but I use
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 473
year-on-year changes in the index to eliminate noise in the monthly data.
Separate regressions over periods of rising infation and periods of falling in
fation (before and afer 1980) make it clear that the coef ficient on infation in
the frst-stage regression is markedly higher in the period afer 1980. Tis
makes sense if the short rate actually depends on a distributed lag of past and
present infation, since lagged infation will be lower than current infation in
the frst period and higher in the second. Without introducing an explicit
lag function, this efect can be approximated by adding a dummy variable
(DUMINF), which takes the value zero until June 1980 and is equal to the in
fation rate thereafer.
Te results of the frst-stage regression of the short rate and the second-
stage regression of the diference between the long rate and the short rate are
reported below. TB3MS is the monthly return on 3-month T-bills in the sec-
ondary market rate, which I use as the measure of the short-term bill rate, s.
YLDPREM is the diference between the yield to maturity of a zero-coupon
10-year T-note and the short rate, which represents mat − s.
TB3MS = .000949 + .697 INF + .405 DUMINF
(.0002) (.033) (.050)
R2 = .57
YLDPREM = .00284 − .313 TB3MS
(.0009) (.020)
R2 = .38
Robust Standard Errors in Parentheses
Observations: 497 F(2, 494) = 223.7
Variable Obs Mean Std Dev Min Max
YLDPREM 497 0.0015 0.0012 −0.0028 0.0036
TB3MS 497 0.0044 0.0027 0.0000083 0.0136
INF 497 0.0036 0.0025 −0.0017 0.0123
Copyright © 2021. Harvard University Press. All rights reserved.
Interest and infation are per month, so that 0.001 corresponds to 12 percent per year.
A couple of pictures will help us understand these results. Figure 12.37 is a
scatter of the yield premium versus the short rate, along with the ftted values
of the two-stage regression (and the OLS values for comparison). Te next
picture, Figure 12.38, shows the ftted equation with a plot of the yield pre-
mium against the estimate of the current short-term rate calculated in the
frst stage. Te vertical lines in Figure 12.38 demarcate an interval of one stan-
dard deviation on each side of the mean of the ftted short-term rate. Figure
12.38 makes clear that the strength of the regression derives from the outliers,
when estimated 3-month bill rates are more than one standard deviation away
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
474 Raising Keynes
0.004
0.002
Yield premium
0.0
−0.002
−0.004
0.0 0.004 0.008 0.012
3-month Treasury-bill rate (secondary market)
Actual values Fitted values (TSLS) Fitted values (OLS)
Figure 12.37 Yield premium on a 10-year Treasury note vs. actual 3-month Treasury bill rate.
0.004
0.002
Yield premium
0.0
Copyright © 2021. Harvard University Press. All rights reserved.
−0.002
−0.004
0.0 0.0023 0.004 0.0065 0.008 0.012
3-month Treasury-bill rate
Estimated values Fitted values (TSLS)
Figure 12.38 Yield premium on a 10-year Treasury note vs. estimated 3-month Treasury bill rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 475
from the mean, that is, less than 0.0023 and greater than 0.0065. For if we re-
peat the regression limiting the sample to observations within the one stan-
dard deviation interval, the second-stage coef ficient a1 is closer to zero and
has a much smaller t-value. Te coef ficient goes from −0.313 to −0.185, and
its t-value falls from 15.6 to 4.2.
TB3MS = .0006 + .795 INF + .401 DUMINF
(.0003) (.060) (.068)
R2 = .28
YLDPREM = .00237 − .185 TB3MS
(.0002) (.045)
R2 = .15
Robust Standard Errors in Parentheses
Observations: 399 F(2, 396) = 91.2
Tese results support the plausible notion that both normal reversion and
risk aversion play a role in determining the spread between the long-bond yield
and the short-term rate. Te negative coef ficient on TB3MS supports a role for
normal reversion, while the fact that this result is driven mostly by the outli-
ers suggests that most of the time normal reversion matters relatively little.14
We can exploit regression results for bond maturities other than ten years
to learn about the empirical relationship between risk aversion and the term
of the bond. Risk aversion predicts that (m) increases with m because a
given change in the yield to maturity will cause the bond price to fuctuate
more the longer is the bond term. Te result is that on theoretical grounds
m
m−1 ()d should increase with m, reaching its maximum ᾱ as m goes to ∞.
0
So on this account alone, we would expect a0 to be an increasing function of
m. But normal reversion makes the same prediction for the other part of a0,
[(1 − (1 − e−m)/m)]*s , at least in the range of m that is relevant for the value
of implicit in the estimate of a1 (namely, = 0.0067).15 Since we can’t sort out
Copyright © 2021. Harvard University Press. All rights reserved.
the response of a0 between its two components, there is not much to be
learned about (m) from the relationship of a0 to m.
On the other hand we can learn something from the ratio a0/a1. In the gen-
eral case we have
m 1 − e − θm ∗
m −1 ∫ α(τ)d τ + 1 − ρ m
a0 0
θm s θ ∫ α(τ)d τ
=− =− 0
− ρ∗s .
a1 1 − e − θm ( (
θm − 1 − e − θm )
)
1 −
θ m
m
Both ()d and m − (1 − e−m) increase with m, but the ratio can rise or
0
fall, depending on which term rises more rapidly. With ≈ 0.007, the elastic-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
476 Raising Keynes
ity of the denominator with respect to m is approximately two, so a0/a1 rises
or falls in absolute value according to whether the elasticity of the n umerator
m
m(m)/ ()d is greater or less than two. If (m) were a linear function of
0
m, then its elasticity would be just equal to two, so a0/a1 i ncreasing in absolute
value would indicate that (m) has an increasing slope in m × (m) space; a
decreasing absolute value suggests a decreasing slope.
In the event, Table 12.1 suggests that (m) increases slowly at low levels of
m, then almost but not quite linearly, more or less in line with the yield curve
in Figure 12.30. On the lef-hand side of the table, a0/a1 falls over intermediate
Table 12.1 Regressions of Yield Premium Against Short-Term Bond Rate
Sample Period
1971–2012 1985–2012
Maturity (years) a1 a0 a0/a1 a1 a0 a0/a1
1 −0.01052 0.00049 −0.04658 0.0018 0.04838 0.00023 0.00480
2 −0.07502 0.00097 −0.01296 0.0066 0.01068 0.00058 0.05399
3 −0.12564 0.00135 −0.01076 0.0076 −0.03432 0.00091 −0.02661
4 −0.16738 0.00167 −0.00998 0.0079 −0.07789 0.00123 −0.01573
5 −0.20247 0.00194 −0.00959 0.0078 −0.11854 0.00151 −0.01275
6 −0.23223 0.00218 −0.00937 0.0077 −0.15575 0.00177 −0.01138
7 −0.25755 0.00238 −0.00924 0.0075 −0.18931 0.00201 −0.01061
8 −0.27899 0.00256 −0.00916 0.0072 −0.21920 0.00222 −0.01012
9 −0.29726 0.00271 −0.00910 0.0070 −0.24552 0.00240 −0.00978
10 −0.31267 0.00284 −0.00907 0.0067 −0.26849 0.00256 −0.00955
11 −0.28837 0.00270 −0.00938
12 −0.30548 0.00282 −0.00925
13 −0.32012 0.00293 −0.00914
14 −0.33259 0.00301 −0.00906
15 −0.34317 0.00309 −0.00900
16 −0.35213 0.00315 −0.00894
Copyright © 2021. Harvard University Press. All rights reserved.
17 −0.35970 0.00320 −0.00889
18 −0.36609 0.00324 −0.00884
19 −0.37148 0.00327 −0.00880
20 −0.37604 0.00329 −0.00875
21 −0.38043 0.00331 −0.00870
22 −0.38377 0.00332 −0.00866
23 −0.38663 0.00333 −0.00861
24 −0.38909 0.00333 −0.00857
25 −0.39124 0.00333 −0.00852
26 −0.39311 0.00333 −0.00847
27 −0.39477 0.00332 −0.00842
28 −0.39626 0.00332 −0.00837
29 −0.39761 0.00331 −0.00831
30 −0.39884 0.00329 −0.00826
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 477
bond maturities of one to fve years and levels out when the regression is run
for maturities of six to ten years.
Te results of regressing the yield premium against the short rate over the
full range of maturities are on the right side of Table 12.1.16 Te ratio a0/a1
does not change very much from the value on the lef-hand side, at least
for maturities in the seven to ten year range; a0/a1 falls relatively sharply
when bond maturities range from three to six years, and then falls slowly.
By contrast, at the short end of the spectrum a1—contrary to the predictions
of both risk aversion and normal reversion—is signifcant at the 99 per-
cent level but has the wrong sign, which makes the ratio a0/a1 also have the
wrong sign.
Is the Risk Premium Stationary?
So far it has been assumed that the risk premium depends only on the (re-
maining) term until the bond matures. Both theory and data suggest other-
wise. Refet Gürkaynak and Jonathan Wright (2012) survey a larger literature
on variable risk premia. Te theory developed in chapter 11 and this chapter
suggests specifc reasons why the risk premium ought to vary.
One is variation in the relative supplies of bonds and bills. Suppose the bill
rate is given, then the more bonds in the mix, the lower the price of bonds
and the higher the yield. Conversely, the more bills, the lower the bill price
and the higher the bill yield. Tus the yield premium ought to vary directly
with the proportion of bonds in the mix of Treasury obligations. A conve-
nient measure of the bond:bill ratio is the average maturity (AVGMAT) of
Treasury obligations.
A second reason why the risk premium varies, namely, changes in the rate
of unemployment, is less obvious. As a measure of business conditions, un-
employment might be a reason for default rates to vary, but default is not an
issue for Treasury obligations. However, as a proxy for the degree of uncer-
Copyright © 2021. Harvard University Press. All rights reserved.
tainty about the economic future, it is plausible for the unemployment rate to
infuence the degree of perceived liquidity risk through an efect on the vola-
tility of bond prices. Liquidity preference as aversion to risk suggests that the
more volatile are bond prices, the greater will be the risk premium.
We can test this relationship by asking whether or not the volatility of
bond-price changes is systematically related to the unemployment rate. Ac-
cording to Figure 12.39, there is a clear relationship.
Te vertical axis measures monthly price changes of zero-coupon bonds,
estimated by the formula
P ∂ ∂
= mat (m, t ) + mat m − mat m,
P ∂m ∂t
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
478 Raising Keynes
0.2
Monthly price change as fraction of current price
0.1
0.0
−0.1
Standard deviation
0.019 0.031 0.034 of price change
−0.2 168 162 167 Number of observations
3 5 5.5 7 9 11
Unemployment rate
Figure 12.39 Change in price of a 10-year Treasury note (1971–2012).
in which the price change P/P and the shif in the yield curve ∂ /∂t are ac
mat
tual rather than expected price changes. At low levels of unemployment, the
standard deviation of bond-price changes is two-thirds its value at higher
levels, so the unemployment rate makes sense as a proxy, if not a direct cause,
of bond-price volatility.
If we linearize a risk-premium function that depends on the unemploy-
ment rate (UNRATE, measured as a percentage of the civilian labor force),
the average maturity of Treasury obligations (AVGMAT, measured in months),
Copyright © 2021. Harvard University Press. All rights reserved.
and bond maturity ((m)), the risk premium becomes the sum of a constant
term (m) and linear terms refecting the impact of the unemployment rate
and the mix of bonds and bills:17
(m, UNRATE, AVGMAT) = (m) + b1UNRATE(t) + b2AVGMAT(t).
Te diferential equation relating the holding yield on the bond to the bill rate
becomes
d ˆ mat (t ) (m) b1UNRATE(t ) b2 AVGMAT(t )
mat s .
dt m m
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 479
And the solution becomes
mat (m, t ) m 1 ˆ s (t m)dm (m)dm
b1UNRATE(t ) b2 AVGMAT(t ) m 1c.
As before, examination of the limiting case as m → 0, gives c = 0. With
short rates forecast by the equation
ˆ s ( ) 1 e ( t ) s e ( t ) s (t ),
the yield premium is now related to the short-term bill rate by the equation
m 1 − e − θm ∗ 1 − e − θm
ρmat (m, t ) − ρ s (t ) = m −1 ∫ α(τ)d τ + 1 − ρ s + − 1 ρ s (t )
0
θm θm
+ b1UNRATE(t ) + b2 AVGMAT(t ).
Once again we use TSLS to estimate the equation for the yield premium,
which is now
mat(m,t) − s(t) = a0 + a1 s(t) + b1UNRATE(t) + b2AVGMAT(t) + .
Te results are
TB3MS = .000974 + .661 INF + .453 DUMINF − .00176 UNRATE − .0000137 AVGMAT
(.0002) (.044) (.071) (.00006) (.000009)
R2 = .58
YLDPREM = .00274 − .291 TB3MS + .000375 UNRATE + .0000228 AVGMAT
(.00009) (.019) (.00002) (.000002)
R2 = .60
Robust Standard Errors in Parentheses
F(4, 492) = 115.34
Variable Obs Mean Std Dev Min Max
Copyright © 2021. Harvard University Press. All rights reserved.
YLDPREM 497 0.0015 0.0012 −0.0028 0.0036
TB3MS 497 0.0044 0.0027 0.0000083 0.0136
INF 497 0.0036 0.0025 −0.0017 0.0123
UNRATE 497 6.43 1.58 3.8 10.8
AVGMAT 497 56.22 12.41 29.0 74.0
Te coef ficient on the short-term rate hardly changes. But in place of the con-
stant term
m 1 − e − θm ∗ −1 m
a0 = m −1 ∫ α(τ)d τ + 1 − ρ s = m ∫ 0 α(τ)d τ − a1 ρ s = 0.00274
∗
0
θm
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
480 Raising Keynes
we have
m 1 − e − θm ∗
m −1 ∫ α(τ)d τ + 1 − ρ s + b1UNRATE(t ) + b2 AVGMAT(t )
0
θm
m
= m −1 ∫ α(τ)d τ − a1 ρ∗s + b1UNRATE(t ) + b2 AVGMAT(t )
0
= .00274 + .000375 UNRATE + .0000228 AVGMAT,
which fuctuates considerably around its mean, as Figure 12.40 shows. With-
out an assumption about the normal rate, we cannot sort out the stationary
m
element of the risk premium m−1 ()d from the normal-rate term −a1 *s
0
because the constant 0.0274 is the sum of the two terms. If we resolve the
ambiguity by assuming that the normal rate is equal to the sample mean of
TB3MS, *s = 0.0044 (0.053 on an annual basis), then we have
m
m 1 ( )d 0.00146 ,
0
which translates into an annual risk premium on the 10-year bond of 0.0175.
Tis estimate is extremely close to the estimate of 0.0171 in John Campbell
and Luis Viceira (2002, p. 73), whose model is built on very diferent assump-
tions from mine.
In any case, the fuctuations over time of the expression
m 1 e m
m 1 ( ) d 1 s b1UNRATE(t ) b2 AVGMAT(t )
0
m
= .00274 + .000375 UNRATE + .0000228 AVGMAT
involve only the risk premium if we assume that the normal rate is unchang-
ing. Observe that over time AVGMAT acts overall as a counterweight to UN-
RATE; as Figure 12.41 shows, average maturity peaks about the same time
Copyright © 2021. Harvard University Press. All rights reserved.
that the unemployment reaches its foor. But over certain intervals of time,
UNRATE and AVGMAT reinforce each other in terms of their efects on the
yield premium. Tis is the case in the periods 1979–1983, 1993–1996, and
2003–2007. Tese are also, it turns out, periods in which the changes in the
risk premium reinforce the efects of changes in the short rate. In 2003–2007,
for example, changes in UNRATE and AVGMAT together account for almost
one-fourth of the total fall in the yield premium, and these changes partly
explain the failure of long yields to respond to the dramatic increase in short
rates over this period.18 (Te other part of the mystery is explained by expec-
tations that short rates were, in Hicks’s phrase, “abnormally high” and were
therefore expected to fall—as indeed they did.)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 481
0.005
0.00274 0.000375 UNRATE 0.0000228 AVGMAT
0.004
0.003
0.00274
0.002
0.001
0.0
1970 1975 1980 1985 1990 1995 2000 2005 2010 2015
m 1 − e − θm ∗
∫0
−1
Figure 12.40 m a(τ)d τ + 1 − ρS + b1UNRATE + b2 AVGMAT.
θm
12
70
UNRATE (percentage of labor force)
Treasury obligations in months)
AVGMAT (average maturity of
10
60
Copyright © 2021. Harvard University Press. All rights reserved.
8
50
6
40
4 30
1970 1975 1980 1985 1990 1995 2000 2005 2010 2015
1979 1983 1993 1996 2001 2003 2007
UNRATE AVGMAT
Figure 12.41 Unemployment and average maturity of Treasury obligations (1971–2012).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
482 Raising Keynes
0.015
0.004
0.01 0.0035
YLDPREM, TB3MS
Risk premium
0.003
0.005
0.0025
0.0
0.002
−0.005 0.0015
1970 1980 1990 2000 2010
1976 1979 1983 1987 1993 1995 2001 2003 2007 2013
YLDPREM TB3MS Risk premium
Figure 12.42 Short-term rate, yield premium, and risk premium.
Figure 12.42 graphs the short rate, the yield premium, and the risk pre-
mium, this last equal to the sum of the constant element, 0.00274, and the
variable elements, 0.000375 UNRATE + 0.0000228 AVGMAT. Te risk pre-
mium pulls the yield premium down between 2003 and 2007, reinforcing the
efect of the rise in the short rate. In 1983 to 1987, however, the fall in the risk
premium worked at cross-purposes with the decline in the short rate, so that
the yield premium hardly budged. Since 2010, the risk premium has fallen as
the economy has improved; and the yield premium has fallen even as the
short rate has hovered near the zlb.
Bill Clinton’s time as president was a period when changes in the risk pre-
mium both reinforced and counteracted the efect of the short rate on the
yield premium. Over the eight years of the Clinton administration, the yield
Copyright © 2021. Harvard University Press. All rights reserved.
premium (on a monthly basis) fell by more than 0.003, almost 0.04 on an an-
nual basis. Unemployment—proxying for bond-price variability—appears to
have been driving the fall in the risk premium of approximately 0.0015, or
0.018 on an annual basis. But half of the reduction in both the risk premium
and the yield premium took place in the frst two years of Clinton’s tenure,
when the reduction in average maturity reinforced the fall in the unemploy-
ment rate, rather than, as in the period from 1995 to 2001, when the two
variables moved in opposite directions. Table 12.2 summarizes the relevant
data.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 12.2 Short Rate, Yield Premium, Unemployment, and Average Maturity (selected months)
Levels Changes
12
Date TB3MS YLDPREM UNRATE AVGMAT TB3MS YLDPREM UNRATE AVGMAT
•
Jan 1976 0.00406 0.00228 7.9 29
Jan 1979 0.00779 −0.00051 5.9 39 0.00373 −0.00279 −2.0 10
Jan 1983 0.00655 0.00226 10.4 48 −0.00124 0.00277 4.5 9
Jan 1987 0.00453 0.00164 6.6 64 −0.00203 −0.00063 −3.8 16
Jan 1993 0.0025 0.00334 7.3 70 −0.00203 0.00171 0.7 6
Jan 1995 0.00476 0.00171 5.6 65 0.00226 −0.00163 −1.7 −5
Jan 1996 0.00417 0.00064 5.6 62 −0.00059 −0.00108 0.0 −3
Jan 2001 0.00429 0.00019 4.2 69 0.00013 −0.00045 −1.4 7
Empirical Appendix
Jan 2003 0.00098 0.00274 5.8 64 −0.00332 0.00255 1.6 −5
Jan 2007 0.00415 −0.00016 4.6 58 0.00318 −0.0029 −1.2 −6
Jan 2010 0.00005 0.00334 9.7 54 −0.0041 0.0035 5.1 −4
Dec 2012 0.00006 0.0014 7.9 54 0.00001 −0.00194 −1.8 0
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
483
484 Raising Keynes
How Does Infation Bear on Interest-Rate Forecasts?
Te results reported in this appendix depend not only on a theory of how the
risk premium is determined but also on a theory of how short-term rates are
forecast, particularly how infation is incorporated into projections of the fu-
ture course of interest rates. Te data permit at least a limited test of my as-
sumption about infation, namely, that infation works its way into long rates
by a progressive ramping up of the short-term rate according to the formula
ˆ s ( ) 1 e ( t ) s e ( t ) s (t ).
An alternative is that infation-induced changes in the short rate are im-
mediately incorporated into forecasts of the normal rate, which is to say that
these changes are treated as permanent. (In this alternative a 2 percent in-
crease in the current rate of infation translates into the expectation that infa
tion will be 2 percent higher indefnitely.) Te propagation of infation in ex-
pected short-term rates is diferent under this assumption, with expected
rates higher at each point in time by the amount of today’s infation. Denoting
infation at time t by INF(t), and real rates by the superscript R, we have real
normal and current rates given by
∗
Rs = ∗s − INF(t ),
ˆ Rs ( ) ˆ s ( ) INF(t ).
In real terms, the relationship between the expected short rate, the current
rate, and the normal rate is now
ˆ Rs ( ) 1 e ( t ) Rs e ( t ) Rs (t ),,
whereas the nominal relationship does not change:
ˆ s ( ) 1 e ( t ) Rs e ( t ) Rs (t ) INF(t ) 1 e ( t ) s e ( t ) s (t ).
Copyright © 2021. Harvard University Press. All rights reserved.
Since the real yield to maturity on the long bond is given by
R
mat (m, t ) = mat (m, t ) − INF(t ),
the real yield premium and the real short rate are now driven by the same
equation as in the regression reported earlier:
m
R
ρmat (m, t ) − ρ Rs (t ) = ρmat (m, t ) − ρ s (t ) = m −1 ∫ α(τ)d τ + b1 ( UNRATE(t ) )
0
1 − e − θm R∗
+ b2 ( AVGMAT(t ) ) + 1 − ρ s
θm
1− e − θm
+ − 1 ρ Rs (t ),
θm
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 485
except that now the nominal interest rates on the right-hand side are replaced
by real rates. On the lef-hand side, the spread is the same whether expressed
in real or nominal terms since infation now afects the short rate and the
long-bond yield equally.
We can rewrite the spread equation as
m
ρmat (m, t ) − ρ s (t ) = m −1 ∫ α(τ)d τ + b1UNRATE(t ) + b2 AVGMAT(t )
0
1 − e − θm R ∗ 1 − e − θm
+ 1 − ρ s + − 1 ρ s (t )
θm θm
1 − e − θm
− − 1 INF(t ) = ao + a1 ρ s (t ) + a2 INF(t )
θm
+ b1UNRATE(t ) + b2 AVGMAT(t ).
Tis leads to a straightforward test of the two hypotheses about how infation
afects the relationship between the short rate and the yield premium. On the
hypothesis that the normal rate is independent of the rate of infation and that
infation afects the adjustment to normal (H1), the test statistic is the coef fi
cient a2: the hypothesis is a2 = 0. On the alternative hypothesis, infation is
immediately incorporated into forecasts, so that the expected real short rate is
a weighted average of today’s real rate and the normal real rate (H2), the test
statistic is sum of the coefficients: according to H2, we expect a2 + a1 = 0.
Te results of running a two-stage least-squares regression augmented by
the addition of the variable INF are
TB3MS = .000974 + .661 INF + .453 DUMINF − .00176 UNRATE − .0000137 AVGMAT
(.0002) (.044) (.071) (.00006) (.000009)
R2 = .58
YLDPREM = .00269 − .219 TB3MS + .000374 UNRATE + .0000187 AVGMAT − .0712 INF
Copyright © 2021. Harvard University Press. All rights reserved.
(.00009) (.058) (.00002) (.000004) (.049)
2
R = .63
Robust Standard Errors in Parentheses
Observations: 497 F(4, 492) = 115.34
Te standard errors of the estimated coef ficient are consistent with H1 but not
with H2. Te data do not reject a2 = 0 since the coef ficient is .071 and the
standard error is .05. But the data do reject a2 + a1 = 0 since the sum of these
two coefficients is −.29 and the standard error of this sum is .02 (already esti-
mated in the previous regression).
To summarize: the data are consistent with a theory of interest-rate spreads
in which both normal reversion and risk aversion matter. Additionally, the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
486 Raising Keynes
data support the view that the risk premium is sensitive both to relative sup-
plies of bonds and bills and to economic conditions, the second of these two
infuences being proxied by the unemployment rate. It is important to bear in
mind that economic conditions enter the picture not because they afect de-
fault risk, as would be expected for private obligations, but because economic
conditions correlate with the volatility of bond prices. Finally, of two models
of how infation afects interest-rate forecasts, the data are consistent with a
model in which the normal rate is fxed in nominal terms but reject the alter-
native model in which reversion is determined by real rates. Tis supports,
but does not prove, the assumption that infation meets the exogeneity crite-
rion in the TSLS regressions above.
Te data also reinforce the commonsense view of the balance between nor-
mal reversion and risk aversion. Figure 12.38 suggests that when short-term
rates are abnormally high or abnormally low normal reversion matters a lot.
But it is the nature of abnormality that it be relatively rare.
Normal Reversion Matters a Lot,
But Only Some of the Time
Experience says that most of the time, even if individual agents have strong
views, there is not generally a strong consensus. But sometimes—like during
the Great Depression, or in the wake of the Great Recession—agents have
strong opinions and there is a strong consensus. In 2014 there was general
agreement that short rates were going up—they had nowhere to go but up—
but there was considerable disagreement about how rapidly interest rates
would move. Even the members of the Federal Open Market Committee, who
are presumably in a good position to know, and who had been uniform in
their view that short-term rates would rise over the next few years, diverged
widely with respect to the pace of the anticipated change. (Tey had been poll-
ing themselves at regular intervals since 2012 and continue to do so.) Figure
Copyright © 2021. Harvard University Press. All rights reserved.
12.43 is an example, emerging from the FOMC meeting in September 2014
(p. 3, fg. 2). Figure 12.43 shows both the uniformity of views with respect to
the direction of change and the divergence with respect to the pace of change.
Te so-called “dot plot,” as the note to the fgure explains, tells us where indi-
vidual members of the FOMC believed the Federal F unds rate would be at
year’s end from 2014 through 2017, as well as in an unspecifed “longer run,”
a period in which the normal rate might be expected to come into its own.19
Both the central tendency and the variation are interesting. Starting from
the (September) 2014 rate of (near) zero, the median of the forecast for (De-
cember) 2015 was an annual rate of 1.375 percent; for (December) 2017 it was
3.75 percent.20 But while everybody agreed that short-term rates would rise,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 487
5.0
4.5
4.0
3.5
3.0
2.5
Percent
2.0
1.5
1.0
0.5
0.0
2014 2015 2016 2017 Longer run
Figure 12.43 Appropriate pace of policy frming: midpoint of target range or target
level for the Federal Funds rate. Note: Each shaded circle indicates the value
(rounded to the nearest one-eighth percentage point) of an individual participant’s
judgment of the midpoint of the appropriate target range for the Federal Funds rate
or the appropriate target level for the Federal Funds rate at the end of the specifed
calendar year or over the longer run.
Copyright © 2021. Harvard University Press. All rights reserved.
there is a wide band around the median forecast. Te Federal Reserve of this
period was committed to keeping interest rates low until labor-market condi-
tions improved, but views difered on what constituted suf ficient improve-
ment and how quickly improvement was likely to take place. And it is evident
that the Fed was hardly unanimous with regard to its perceptions of the im-
plications of an improving labor market for price stability—or, for that matter,
its perceptions of the relative importance of the two elements of its dual man-
date. Given these diferences it is hardly surprising that the variability in indi-
vidual forecasts increases as the time horizon lengthens.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
488 Raising Keynes
Te most interesting thing about the chart is its very existence. I don’t have
in mind the transparency of the FOMC, though that in itself refects a histori-
cal sea change in the conduct of central banking, but rather that members had
views about the course of future interest rates. If you had polled the FOMC in
2004, you well might have got similar answers to the polls in 2014; at that
time, too short rates had nowhere to go but up. But in 1996 or 2006 I imagine
that the FOMC would not have had any view at all, certainly not a frm view,
as to where interest rates were going. Lesser mortals perhaps did, but they
were paid to have frm views; nobody shells out good money for a forecaster
to say “I don’t have a clue which way interest rates will move.” It is no wonder
that for most of the observations in Figure 12.38—399 out of 497—the level of
the short rate has relatively little predictive power with respect to the yield
premium but that when short rates are abnormally high or abnormally low (as
they were in 2015)—the other 98 observations—the short rate has consider-
able weight.
The Spread Between Treasury and Corporate Debt
To complement the analysis of the spread between returns on short-and
long-term Treasury debt, we turn now to the spread between long Treasuries
and long corporate debt. Te principle diference between the two kinds of
debt is default risk, totally absent in the analysis of Treasury debt and of vary-
ing importance in the analysis of corporate debt, intermittently the domi-
nant factor, but hardly the only factor: analogously to the efect of the com-
position of Treasury debt on the spread between long and short Treasury
yields, the relative amounts of Treasury and corporate debt ought to afect the
spread between Treasury and corporate bonds (as Arvind Krishnamurthy
and Annette Vissing-Jorgensen [2012] have argued).
Te risk premium on corporate bonds relative to short-term T-bills can be
written as
Copyright © 2021. Harvard University Press. All rights reserved.
C (m, UNRATE, AVGMAT, DEFTRAIL, DELDEF, CORPFRAC)
= C (m) + d1UNRATE(t ) + d2 AVGMAT(t ) + d3DEFTRAIL(t )
+ d 4 DELDEF(t ) + d5CORPFRAC.
Two new variables measuring default risk are DEFTRAIL, the trailing
12-month default rate for all U.S. corporate bond issues (calculated by
Moody’s Analytics), and DELDEF, the frst diference of this series.21 DELDEF
is arguably a better indicator of the current default rate than DEFTRAIL since
DELDEF, measuring the diference between the most recent observation and
the observation from one year further back, captures the momentum of de-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 489
fault. Te last variable, CORPFRAC, measures the volume of corporate bonds
outstanding relative to the total bond volume, including securities issued by
government agencies and government sponsored entities, as well as state and
municipal bonds, but excluding the holdings of the Federal Reserve.22
Te integral equation for the corporate-bond-yield premium over the T-
bill rate is
mCORP (m, t ) m 1 ˆ (t m )dm (m)dm d UNRATE(t ) d AVGMAT(t )
s C 1 2
d3DEFTRAIL(t ) d 4 DELDEF(t ) d5CORPFRAC(t ) m 1c,
where mCORP is the yield to maturity on corporate bonds that mature in m
years. Once again we invoke limiting values of the lef-and right-hand sides
to fx c, and once again we have c = 0.
Alas, direct estimation of the coef
ficients is problematic because of the lack
of a series of zero-coupon bonds comparable to the series that Houston
McCulloch originally developed for Treasuries in the 1970s (1971, 1975).23 In
principle one could construct a zero-coupon corporate bond following the
procedures that have been applied in the case of Treasuries, but in view of the
considerable resources that would be necessary, I am taking the easier route
of simply subtracting the equation for Treasuries
mat (m, t ) m 1 ˆ (t m)dm (m)dm b UNRATE(t )
s 1
1
b2 AVGMA T(t ) m c
from the equation for corporates. Tis gives
ρmCORP (m, t ) − ρmat (m, t ) = m −1 ∫ ( αC (m) − α(m) ) dm + ( d1 − b1 ) UNRATE(t )
+ ( d2 − b2 ) AVGMAT(t ) + d3DEFTRAIL(t )
+ d 4 DELDEF(t ) + d5CORPFRAC(t ).
Copyright © 2021. Harvard University Press. All rights reserved.
For this equation to generate unbiased estimates of the relevant parameters, it
is necessary to assume that the diference between yields to maturity on zero-
coupon and coupon bonds is the same, up to a random error, for Treasuries
and for corporates. Tis seems plausible, but it must be recognized that there
is no way to test this hypothesis short of constructing a time-series of zero-
coupon corporate bonds and estimating the yields.
Te data set now includes the yields on 10-year corporate bonds from the
Treasury “High Quality Market” Corporate Bond Yield Curve (U.S. Depart-
ment of the Treasury [n.d.]). Using quarterly data, the regression for the pe-
riod 1984 to 2012 is
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
490 Raising Keynes
mCORP(m,t) − mat(m,t) = −.0005974 + .0001114 UNRATE(t) − .0000119 AVGMAT(t)
(.001369) (.0000437) (.0000151)
+ .003932 DEFTRAIL(t) + .0359 DELDEF(t) + 0.004 CORPFRAC(t)
(.00380) (.01014) (0.001)
R2 = .54
Newey-West Standard Errors in Parentheses
Variable Obs Mean Std Dev Min Max
mCORP(m,t) − mat(m,t) 116 0.0018 0.0006 0.0011 0.0042
UNRATE 116 6.1483 1.5071 3.9000 9.9333
AVGMAT 116 63.138 6.775 46.0 73.0
DEFTRAIL 116 0.0222 0.0158 0.0057 0.0829
DELDEF 116 0.00008 0.0055 −0.0264 0.0231
CORPFRAC 116 0.3558 0.0505 0.2710 0.4396
Yields are per month, data are quarterly.
Te coef ficient of UNRATE, the estimate of the diference d1 − b1, 0.0001114,
taken together with the coef ficient of UNRATE in the Treasury yield-premium
regressions, 0.000375, implies that the efect of unemployment on corporate
bonds is almost 0.0005. Tat is, an increase of one percentage point in the
unemployment rate leads to an increase in the corporate-bond yield of 0.05
percentage points per month, or 0.6 percentage points on an annual basis.
Tis suggests that if UNRATE is indeed a proxy for price variability, corporate-
bond price variability should exhibit the same correlation with unemploy-
ment that shows up for Treasuries in Figure 12.39. But the data show no such
correlation, as Figure 12.44 shows. Except for the two outliers in the 6 to 8
percent unemployment interval, the pattern of corporate-bond price changes
does not show any sensitivity to the rate of unemployment. It may be that this
is due to the truncated sample of 10-year corporate bond yields. Te 497 ob-
Copyright © 2021. Harvard University Press. All rights reserved.
servations in the (monthly) Treasury regressions include the high unemploy-
ment years of the 1970s and 1980s. In contrast, the data for 10-year corpo-
rates begins in 1984. More likely, the current unemployment rate is a better
proxy for default risk than the trailing index of actual default over the previ-
ous twelve months: the coef ficient on the index itself, the variable DEFTRAIL,
is insignifcant in the presence of UNRATE, but is signifcant (t = 2.7) when
the unemployment variable is suppressed. By contrast, the coef ficient on the
change in the default index, DELDEF, is statistically signifcant whether or
not the unemployment rate is included in the regression.
Te interpretation of the unemployment rate as a proxy for default risk is
plausible, but further investigation suggests this is not all there is to the story.
If we replace the trailing index of default by the default risk over the subse-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
12 • Empirical Appendix 491
0.1
Monthly price change as fraction of current price
0.05
0.0
−0.05
Standard deviation
0.013 0.016 0.021 0.014 of price change
Number of
88 114 103 43 observations
−0.1
4 5 6 8 10
Unemployment rate
Figure 12.44 Change in price of 10-year corporate bonds (1984–2012).
quent year, DEF, and the change in the index by the change of the index over
the next year, labeled DELDEF+1, both variables are signifcant, and the re-
gression ft improves markedly:
mCORP(m,t) − mat(m,t) = .0006821 + .0001204 UNRATE(t) − .000012 AVGMAT(t)
(.0009217) (.0000312) (.00000953)
+ .0093 DEF(t) + .0439 DELDEF + 1(t) + 0.00375 CORPFRAC(t)
(.00284) (.0102) (0.000866)
Copyright © 2021. Harvard University Press. All rights reserved.
R2 = .71
Newey-West Standard Errors in Parentheses
Variable Obs Mean Std Dev Min Max
mCORP(m,t) − mat(m,t) 116 0.00085 0.00051 0.00026 0.00352
UNRATE 116 6.14828 1.50706 3.90 9.93
AVGMAT 116 63.1379 6.77507 46.0 73.0
DEF 116 0.02228 0.01573 0.00573 0.08293
DELDEF+1 116 0.00008 0.00549 −0.02637 0.02307
CORPFRAC 116 0.35576 0.05053 0.27097 0.43961
Yields are per month, data are quarterly.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
492 Raising Keynes
An obvious problem with this regression is its reliance on the ability of
agents to forecast default rates. As I indicated at the beginning of this appen-
dix, I am not enamored of rational expectations, but the problem here is less
acute than in forecasting interest rates over a much longer (ten-year) period.
It is at least arguable that information about the likelihood of future defaults is
available to bond-market participants, and in the absence of a time series of
forecasts, relying on actual future defaults may be defensible. But in this case
the unemployment variable becomes a supplementary indicator of default
sentiment rather than a replacement for the actual default rate. We must now
argue that the forward-looking default rate is simply grist for the mill of ex-
pectations, along with current unemployment, rather than arguing that agents
actually know future default rates.
What Do the Numbers Mean?
Te regressions in this appendix provide a quantitative estimate of the contri-
bution of various factors—the short rate, unemployment, average maturity,
default risk—to the desirability of T-bills relative to longer Treasury obliga-
tions and to corporate bonds. Of particular interest, given the focus of
Keynes’s General Teory, is the consequence for the liquidity trap, here mean-
ing the spread between the short rate and the corporate bond yield when the
short rate approaches the zlb.
In Tables 12.3 and 12.4 we compare predicted and actual spreads before the
fnancial crisis unleashed by the fall of Lehman Brothers in September 2008
and three observations afer the crisis began to unfold. Table 12.3 refects
trailing default risk and Table 12.4 future default risk.
Te relationship between short and long Treasuries is the same in both ta-
bles. Te T-bill rate falls from almost 4 percent per year in the fall of 2007 to
0.67 percent a year later, and then to almost zero (where it remained until the
Federal Reserve began to raise rates in December 2015). Between fall 2007
Copyright © 2021. Harvard University Press. All rights reserved.
and fall 2008, the spread between the 10-year Treasury and the 3-month bill
widens by just over 150 basis points. According to the decomposition in both
tables, this increase is driven in equal measure by unemployment, presumably
a proxy for price volatility, and the decline in the bill rate, presumably an indi-
cator of anticipated reversion to normal. Te increase in the proportion of
bills in the mix of Treasury debt (AVGMAT declines from 57.3 to 46 months)
mitigates the rise in the spread: according to the regression coef ficients the
spread would have widened by 180 basis points without the decline in the
average maturity of Treasury debt.
By contrast, default risk is the driver of the corporate-Treasury spread. As
the federal defcit rises, the decrease in the relative volume of outstanding
corporate debt sofens the blow, but not by much.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 12.3 Estimated Spreads with Varying Unemployment and Default Risk (trailing default rates)
Fall 2007 Fall 2008 Winter 2009 Fall 2009 Cumulative
Sample Mean Observations (Oct–Dec) (Oct–Dec) (Jan–Mar) (Oct–Dec) (Fall 2007– Fall 2009)
TB3MS = s(t) (percent, annual) 5.26 1971:6–2012:12 3.39 0.3 0.21 0.06
UNRATE (percent) 6.43 1971:6–2012:12 4.8 6.87 8.27 9.93
AVGMAT (months) 56.0 1971:6–2012:12 57.3 46.0 47.0 51.3
DEFTRAIL (index) 0.022 1984:1–2012:12 0.006 0.025 0.041 0.083
12
DELDEF (index) 0.00008 1984:1–2012:12 −0.0018 0.0068 0.0161 0.0046
•
CORPFRAC (fraction) 0.356 1984:1–2012:12 0.437 0.403 0.412 0.407
Actual (percent, annual)
YLDPREM = mat(m,t) − s(t) 1.76 1971:6–2012:12 1.06 3.62 3.15 3.78
mCORP(m,t) − mat(m,t) 1.02 1984:1–2012:12 1.44 4.22 3.72 1.49
mCORP(m,t) − s(t) 2.78 1984:1–2012:12 2.5 7.84 6.87 5.28
Estimates (percent, annual)
YLDPREM = mat(m,t) − s(t) 1.76 1971:6–2012:12 1.6 3.12 3.8 4.72
Empirical Appendix
mCORP(m,t) − mat(m,t) 1.02 1984:1–2012:12 1.15 1.89 2.58 2.42
mCORP(m,t) − s(t) 2.78 1984:1–2012:12 2.75 5.01 6.38 7.14
Decomposition of Estimated Changes
[mat(m,t) − s(t)] = −0.2909278 TB3MS + 0.0003748 UNRATE + 0.0000228 ∆AVGMAT
[mat(m,t) − s(t)] 1.52 0.68 0.91 3.11
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Contribution of TB3MS 0.9 0.02 0.05 0.97
Contribution of UNRATE 0.93 0.63 0.75 2.31
Contribution of ∆AVGMAT −0.31 0.03 0.12 −0.16
493
Copyright © 2021. Harvard University Press. All rights reserved.
Table 12.3 (continued)
Fall 2007 Fall 2008 Winter 2009 Fall 2009 Cumulative
Sample Mean Observations (Oct–Dec) (Oct–Dec) (Jan–Mar) (Oct–Dec) (Fall 2007– Fall 2009)
494
Decomposition of Estimated Changes
[mCORP(m,t) − mat(m,t)] = 0.0001114 UNRATE − 0.0000119 ∆AVGMAT
+ 0.0039318 DEFTRAIL + 0.0358872 ∆DELDEF
+ 0.0039981 ∆CORPFRAC
[mCORP(m,t) − mat(m,t)] 0.74 0.69 −0.16 1.27
Contribution of UNRATE 0.28 0.19 0.22 0.69
Contribution of ∆AVGMAT 0.16 −0.01 −0.06 0.09
Contribution of DEFTRAIL 0.09 0.08 0.2 0.36
Contribution of ∆DELDEF 0.37 0.4 −0.49 0.28
Contribution of ∆CORPFRAC −0.16 0.04 −0.02 −0.14
Table 12.4 Actual vs. Estimated Spreads (current and future default rates)
Raising Keynes
Fall 2007 Fall 2008 Winter 2009 Fall 2009 Cumulative
Sample Mean Observations (Oct–Dec) (Oct–Dec) (Jan– Mar) (Oct–Dec) (Fall 2007–Fall 2009)
TB3MS = s(t) (percent, annual) 5.26 1971:6–2012:12 3.39 0.3 0.21 0.06
UNRATE (percent) 6.43 1971:6–2012:12 4.8 6.87 8.27 9.93
AVGMAT (months) 56.0 1971:6–2012:12 57.3 46.0 47.0 51.3
DEF (index) 0.022 1984:1–2012:12 0.009 0.041 0.064 0.07
DELDEF+1 (index) 0.00008 1984:1–2012:12 0.0033 0.0161 0.0231 −0.0127
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
CORPFRAC (fraction) 0.356 1984:1–2012:12 0.437 0.403 0.412 0.407
Actual (percent, annual)
YLDPREM = mat(m,t) − s(t) 1.76 1971:6–2012:12 1.06 3.62 3.15 3.78
mCORP(m,t) − mat(m,t) 1.02 1984:1–2012:12 1.44 4.22 3.72 1.49
mCORP(m,t) − s(t) 2.78 1984:1–2012:12 2.5 7.84 6.87 5.28
Copyright © 2021. Harvard University Press. All rights reserved.
Estimates (percent, annual)
YLDPREM = mat(m,t) − s(t) 1.76 1971:6–2012:12 1.6 3.12 3.8 4.72
mCORP(m,t) − mat(m,t) 1.02 1984:1–2012:12 1.29 2.64 3.49 1.83
mCORP(m,t) − s(t) 2.78 1984:1–2012:12 2.89 5.76 7.29 6.54
12
Decomposition of Estimated Changes
•
[mat(m,t) − s(t)] = −0.2909278 TB3MS + 0.0003748 UNRATE + 0.0000228 ∆AVGMAT
[mat(m,t) − s(t)] 1.52 0.68 0.91 3.11
Contribution of TB3MS 0.9 0.02 0.05 0.97
Contribution of UNRATE 0.93 0.63 0.75 2.31
Contribution of ∆AVGMAT −0.31 0.03 0.12 −0.16
Decomposition of Estimated Changes
[mCORP(m,t) − mat(m,t)] = 0.0001204 UNRATE − 0.000012 ∆AVGMAT + 0.0093002 DEF
+ 0.0439496 ∆DELDEF+1 + 0.0037535 ∆CORPFRAC
Empirical Appendix
[mCORP(m,t) − mat(m,t)] 1.35 0.85 −1.67 0.54
Contribution of UNRATE 0.3 0.2 0.24 0.74
Contribution of ∆AVGMAT 0.16 −0.01 −0.06 0.09
Contribution of DEF 0.36 0.26 0.07 0.68
Contribution of ∆DELDEF+1 0.68 0.37 −1.89 −0.84
Contribution of ∆CORPFRAC −0.15 0.04 −0.02 −0.13
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
495
496 Raising Keynes
Te residuals in the regressions for the spread between corporate and Trea
sury are much larger in Table 12.3 than in Table 12.4. In fall 2008, the direness
of the economic situation had yet to be refected in the numbers: unemploy-
ment was less than 7 percent—it would rise to 10 percent before the end of
the next year—and default rates had only begun to rise. In Table 12.3 the esti-
mate of the yield premium on 10-year corporate bonds relative to Treasuries
of the same maturity for this period is less than 45 percent of the actual pre-
mium, 1.89 percent annually versus 4.22 percent. Tis model does better by
the following winter, but the estimate is still less than 70 percent of the actual.
In Table 12.4 by contrast, the estimate for fall 2008 is 2.64, percent and for the
winter of 2009 it is 3.49, respectively 63 percent and 94 percent of the actual.
Finally, the composition of the overall debt, both the relative size of corpo-
rate and Treasury debt and the average maturity of Treasury debt, is statisti-
cally signifcant in both of the equations for mCORP(m,t) − mat(m,t), but the
efects of CORPFRAC and AVGMAT are each relatively small; moreover, the
two efects worked in opposite directions. Te burgeoning Treasury debt be-
tween 2007 and 2009 reduced the fraction of corporate bonds by almost four
percentage points, which in turn had the expected negative impact on the
corporate-Treasury spread (the estimate is −.14 percentage points in Table
12.3 and −.13 percentage points in Table 12.4). Tis reduction was shaved by
more than half by the shortening of the AVGMAT of Treasury debt from 57
to 51 months.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 13 .
taking money seriously
Te wild duck has dived down to the bottom—as deep as she can
get—and bitten fast hold of the weed and tangle and all the rubbish
that is down there, and it would need an extraordinarily clever dog
to dive afer and fsh her up again.
—john maynard keynes
We are a long way from the simple models of chapter 3. In those models the
interest rate is either fxed directly by the central bank or is determined by the
money supply and two sources of demand, a demand for money to grease the
wheels of commerce and industry and a demand for money as a constituent
of asset portfolios. In the frst case, which I dubbed the frst-pass model, ag-
gregate demand alone determined the level of output, whereas in the second-
pass model, real aggregate demand depended on the nominal price of output,
and the equilibrium level of output depends not only on aggregate demand
but on supply conditions as well.
Te frst-pass model lays out the interconnections of the elements of
Keynes’s vision in the simplest possible form, and serves to demonstrate both
the possibilities and the limits of monetary policy as a tool in the service of
full employment.
Te second-pass model can be interpreted either as a slightly more con-
crete formalization of how monetary policy may or may not be suf ficient to
Copyright © 2021. Harvard University Press. All rights reserved.
direct the economy to a full-employment equilibrium or as a model of how a
capitalist economy lef to its own devices may or may not spontaneously
achieve full employment. In the frst interpretation, a central bank is assumed
to set the money supply. In the second, there is no central bank and we have
two choices. In the simplest model, the money supply is an exogenously given
supply of gold, silver or cowrie shells. In a more sophisticated fnancial world,
commodity money is but one form that money takes. Te overall money sup-
ply is determined by a fractional-reserve banking system, which is free to
create money up to the maximum ratio of deposits to reserves permitted by
law or custom.
Even without the complications introduced with respect to propensities to
consume and save (chapter 9), investment demand (chapter 10), and the the-
497
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
498 Raising Keynes
ory of interest (chapters 11–12), the second-pass model foundered once we
lef the safe harbor in which money is a commodity. Once we assume that
banks create money to fulfll the transactions demand, we no longer can
count on a simple equation in which the transactions demand and the asset
demand add up to a given supply. Crucially, it is no longer the case that mon-
etary equilibrium requires more of one money demand to mean less money
available to supply the other. A frewall now exists between the two kinds of
money.
None of this turned out to matter very much in a comparative-statics
framework. Even if transactions demand is passively satisfed by the banking
system, so that transactions money and asset money are independent of one
another, the volume of bonds (by assumption the only other fnancial asset)
will vary with the price level of output. Te result in chapter 4 was that the
equilibrium interest rate, determined by the demand for and supply of f
nancial assets, varied directly with the price level. Tus the essential quality
of Keynes’s second-pass model, the downward-sloping AD schedule, was pre-
served even though its raison d’être is now a bond efect, which makes the
stock of bonds decrease with the price level, rather than the Keynes efect,
which makes the stock of portfolio money increase as the price level de-
creases.
In a world of real-time changes, it makes no sense to assume that the vol-
ume of bonds will vary with the nominal value of capital. Rather, the appro-
priate assumption is that the volume of bonds is fxed once and for all. Tis
seemingly small change makes a big diference. Now not only is the supply of
asset money independent of the money demand (and supply) for transac-
tions, but also the quantity of bonds is independent of the general price level.
Te consequence, as we saw in chapter 7, was that demands for fnancial as-
sets become independent of the price of goods, and so does the interest rate.
Te AD schedule becomes vertical in the space of output versus price; the
second-pass model collapses into the frst-pass model. Tere is neither a
Copyright © 2021. Harvard University Press. All rights reserved.
Keynes efect nor a bond efect, one or the other a necessary condition for a
downward-sloping AD schedule.
Te purpose of this chapter is to revisit these results in the light of the the-
ory of interest developed in the previous chapter. We begin with an issue that
turns out to be central to the story but got virtually no attention in Te Gen-
eral Teory.
The Missing Theory of Money
Te title is Te General Teory of Employment, Interest and Money. As we have
seen, there is not one, but two theories of employment, one for a given inter-
est rate, one for a given money supply.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 499
What about the other two elements of the title, interest and money? Te
theory of interest is incomplete, and the theory of money is at best implicit.
Tere is a theory of the demand for money, and this theory is central
to Keynes’s argument. Reprised below, liquidity preference is Keynes’s an-
swer to the so-called quantity theory of money. Te quantity theory (which,
vampire like, is still with us) is not a theory of money at all; if truth in label-
ing prevailed, it would be called the quantity-of-money theory of prices.
But Keynes is as remiss as the quantity theorists. Despite lots of talk about
properties of money, including the impenetrable chapter 17 on own rates of
interest, Te General Teory ofers no theory of how money comes to be or
what it is.
Te absence of a theory of money turns out to be an important reason why
Keynes doesn’t deliver on his promise to replace the conventional theory of
interest, in which productivity and thrif rule the roost, by his own theory of
liquidity preference. And this failure in turn supports the argument, which
prevails to this day, that unemployment is the result of frictions and imperfec-
tions, problems that can be fxed by making the world more like the textbook
heaven of deregulation and unfettered self-interest—or by just waiting for the
long run, which magically dissolves all rigidities.
Te theory of money is best approached through the theory of interest. Li-
quidity preference starts from the argument that the mainstream theory is
built on a category mistake, namely, that the rate of interest is determined in a
market for new capital goods in which demand comes from projects for add-
ing to the capital stock (plant, equipment, housing, infrastructure), and sup-
ply comes from the willingness of the populace to abstain from consumption
and thus provide resources for investment. For Keynes the locus of interest-
rate determination is not to be found in the fows of investment and saving,
the tail of asset markets, but on the dog itself, in markets for fnancial assets
that correspond, dimensionally at least, to the stock of capital goods. Te in-
terest rate (singular for ease of exposition but plural in reality) that emerges
Copyright © 2021. Harvard University Press. All rights reserved.
from buying and selling in these markets becomes the “hurdle rate” for in-
vestment, separating the sheep from the goats among potential projects. Be-
sides determining the demand for investment, the interest rate prevailing in
the market for fnancial assets may also infuence the supply of saving. But
this is very diferent from arguing that interest rates are determined by invest-
ment demand and saving supply.
What does this have to with a theory of money?
As presented in Te General Teory, there are two uses of money. One is
to grease the wheels of commerce. Tis leads to a transactions demand for
money, along the lines of the quantity theory that goes back to David Hume.
Te other use for money is to satisfy appetites for a store of value that can be
reliably counted upon to meet future contingencies. Tis leads to a separate
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
500 Raising Keynes
and distinct demand for money in wealth portfolios, which Keynes calls the
speculative demand.
Why speculative? Keynes starts from a theory of wealth management in
which portfolios are driven by two conficting factors, return and risk. Re-
turns matter for obvious reasons. Risk, or rather uncertainty, matters because
agents might need cash to meet unpredictable contingencies of life.1 Illness
and job loss come to mind as the contingencies that most concern most peo-
ple, but these are not the contingencies that Keynes would have had in mind
if the theory was to apply to wealth management in the real world.2 For the
simple reason that few people, especially in his day, accumulate signifcant
amounts of wealth, whereas for the people who do, questions like how to pay
the doctor’s bill or how to get along for a while without a salary are not up-
permost in their minds as they think about how to deploy wealth. More im
portant would be business losses, or for that matter business opportunities,
situations where ready cash is necessary either to stave of serious reverses or
to take advantage of extraordinary possibilities.
In making choices among the various forms in which to hold wealth,
agents trade of the potential gains of holding assets that are expected to pay
dividends, interest, or rents against the potential losses to which these instru-
ments are subject. Losses are possible even if the asset is a long-term U.S. T-
bond, which is perfectly safe in the sense of having no risk of default: bond
prices go up and down as interest rates fall and rise. Given the potential for
loss, economic actors might prefer to hold some or all of their wealth in the
form of cash and forego the potential return on bonds and other assets, in
order to ensure against loss if they need cash before the bonds mature. Te
possibility of default adds another degree of uncertainty, another reason for
holding cash.
Keynes thus sees agents as necessarily speculating in two senses. First, they
inevitably are speculating on the likelihood of the various contingencies that
might force them to need cash; second, they must speculate on the prices they
Copyright © 2021. Harvard University Press. All rights reserved.
will receive for their assets if they have to cash out. In a simplifed world in
which a long-term T-bond stands in for the entire range of fnancial assets
other than ready money, the second speculation is in part a gamble about the
future course of interest rates and bond prices. Risk-neutral agents will not
hold bonds if, averaging the probability of a capital gain and the probability of
a capital loss, an expected loss in the value of a bond exceeds the coupon
yield. But even if the probabilities of gain and loss cancel out, risk-averse
agents will, in Keynes’s theory, require compensation for holding long-term
bonds; to a risk-averse agent, the damage caused by a loss outweighs the
beneft of an equal gain. In return for foregoing liquidity, she will require a
higher yield on the T-bond than she would if there were no possibility of
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 501
changes in bond yields and bond prices. Risk aversion and the expectation
that interest rates will revert to normal reinforce each other in creating a de-
mand for liquidity when bond prices are expected to fall, but they work in
opposite directions when bond prices are expected to rise. In any case, the
higher the return on a bond, the more attractive it is. In a two-asset world,
cash and long-term Treasuries, the speculative demand for money is an in-
verse function of the rate of interest on long bonds.
Diferent considerations govern the demand for money as a means of pay-
ment in business (and personal) transactions. Te transactions demand is
determined by the level of production and the price level. Higher levels of
nominal economic activity mean either more transactions or more money per
transaction (or both).3
Te endowment of bonds and the nominal level of output determine the
total demand for money. Te other side of the coin, so to speak, is the supply
of money. For any given supply of money, the rate of interest and the level of
economic activity balance the overall demand for money and its supply. Tis
is the essence of Keynes’s theory of interest; its emphasis on balancing return
against the risks associated with changing bond prices and yields explains
why it is called the liquidity-preference theory of interest.
For Keynes, unlike the mainstream, money is not a veil, but an active com-
ponent in determining how much is produced, consumed, and invested. Te
supply of money has a direct efect on the real side of the economy through its
efect on interest rates and investment demand.
Tree important conclusions follow. First, in a capitalist economy in which
the supply of money is really and truly beyond human control, in an economy
where nobody is at the monetary helm, there is no endogenous stabilizing
mechanism that guarantees full employment. It is possible for the economy to
settle down at a level of employment that leaves many willing workers without
jobs. Second, control of the money supply is not only necessary but is usually
suf ficient to steer the economy to full employment. Tird, there are excep-
Copyright © 2021. Harvard University Press. All rights reserved.
tions to the usual case, a liquidity trap or an unresponsiveness of investment
demand, that call for something more than activist monetary policy in order
to keep the economy of the shoals of unemployment. (Keynes was vague
about the “something more.” Te General Teory provides a glimpse of the
promised land of fscal policy as the additional tool, but a combination of
Victorian virtue and his sense of the politically possible made Keynes shrink
from its implications. Some years later, possibly under the tutelage of one
of his acolytes, Abba Lerner, he overcame his inhibitions (see chapter 14,
note 3).
Tere is an alternative to increasing the nominal money supply, which in
theory can have the same positive efect on investment demand: reducing
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
502 Raising Keynes
nominal wages. Usually associated with Franco Modigliani’s (1944) formal-
ization of Te General Teory, the argument was anticipated by Keynes:
If the quantity of money is virtually fxed, it is evident that its quantity in
terms of wage-units [that is, in real terms] can be indefnitely increased by a
suf ficient reduction in money wages. (Te General Teory, p. 266)
At a lower level of wages, prices will also be lower, and less money will be re-
quired to facilitate the same level of production. Because the overall supply of
money is fxed, the money no longer needed for transactions must spill over
into the market for fnancial assets, that is, into the market for long-term
Treasuries. But agents are willingly holding the cash in their portfolios be-
cause the interest rate is what it is. Tey won’t hold more unless the interest
rate falls. So to restore equilibrium between overall demand for money and
overall supply, the interest rate must be lower. Tere is thus a positive rela-
tionship between the overall price level and the rate of interest. Tis coupled
with the inverse relationship between the interest rate and investment de-
mand implies that lower wages go along with greater aggregate demand, the
Keynes efect.
For Keynes, the Keynes efect showed the limitations of a policy of wage
reductions:
It follows that wage reductions, as a method of securing full employment,
are also subject to the same limitations as the method of increasing the
quantity of money. (Te General Teory, p. 266)
But wage reductions, like an increase in the quantity of money, “may exert an
inadequate infuence over the long-term rate of interest” (p. 266). Moreover,
wage fexibility is a cumbersome and politically dif
ficult way of achieving
what is more practically accomplished by deliberate monetary policy:
A method which it is comparatively easy to apply should be deemed pref
erable to a method which is probably so dif ficult as to be impracticable.
Copyright © 2021. Harvard University Press. All rights reserved.
(p. 268)
Modigliani, as we saw in chapter 4, drew a diferent conclusion: wage and
price fexibility, Modigliani contended, up-ends Keynes’s conclusion that even
a purely competitive capitalist economy, one free of the vices of frictions, im-
perfections, and infexibilities, will provide full employment only under a
serendipitous constellation of economic forces. Specifcally, Modigliani de-
ployed the Keynes efect to argue that the same positive efects on demand
that can be accomplished by increasing the money supply can be accom-
plished by the invisible hand, without any top-down intervention, if the wage
level, and accordingly the price level, are suf ficiently fexible.4
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 503
What Does Keynes’s Liquidity-Preference
Theory Assume about Money?
Despite the lack of an explicit theory of money, an implicit theory is hard at
work here—in two places. First, the supply of money is exogenous, either a
fxed amount of a commodity given by the available resources and technology
or a fxed amount dictated by the monetary authority. Second, the two uses of
money, speculation and transactions, are assumed to compete for the avail-
able supply of money. One more dollar, euro, or yen greasing the wheels of
commerce means one fewer available to satisfy the desire of wealth holders
for a hedge against bond-price fuctuations.
Tere is more than one way to justify these two assumptions, but in the
absence of an explicit theory of money, the justifcation must be inferred. In
arguing about the capacity of capitalism to provide full employment without
intervention from above, that is, in the absence of a central bank, the simplest
assumption is that money is a commodity: gold, silver, or cowrie shells.
commodity money With commodity money, the stock is fxed by the
costs of fnding, mining, and refning. And commodity money can be used
for only one purpose at a time. A Venetian ducat used to pay for the fsh
needed for tonight’s menu at the Poste Vecie cannot at the same time serve as
an umbrella for a rainy day for the restaurant’s owner or anybody else.
Commodity money fts the story that we learn from the texts, the one
about money emerging spontaneously to allow trade to take place without the
double coincidence of wants that makes barter ef ficient. Precious metals are
uniquely positioned to act as money because they are, well, precious, and thus
concentrate value, measured in terms of volume or in terms of weight. Pre-
cious metals are also durable and relatively easy to divide into smaller units,
as needed. Abraham’s hoard of silver could be easily divided to extract the
four hundred shekels he needed to purchase Sarah’s burial plot from Ephron
Copyright © 2021. Harvard University Press. All rights reserved.
the Hittite.
Commodity money thus fts neatly with the requirements of Keynes’s
liquidity-preference theory. It is both fxed in supply, and it can be used only
for one thing at a time.
If money is not based on gold or cowrie shells, it is harder to justify either
of these requirements. Te problem is that, as history, the spontaneous evolu-
tion of precious metals into money has long since been discredited as at best
incomplete and at worst a just-so story. Georg Friedrich Knapp published
Te State Teory of Money in 1905 (as Staatliche Teorie des Geldes). A. Mitch-
ell Innes published “What is Money?” in 1913 and “Te Credit Teory of
Money” in 1914. Teir titles tell us their respective emphases, each of which
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
504 Raising Keynes
captures part of the critique of the commodity theory. Christine Desan’s 2014
account of the evolution of English money from Anglo-Saxon times to the
early modern period makes the limitations of the commodity theory clear
and suggests that the questions raised by Knapp and Innes have been debated
for centuries.
Keynes himself, in the Treatise on Money, recognized that the commodity-
money story had long since passed its sell-by date. In the Treatise he invokes
the authority of Knapp, but he had also read at least the frst of Innes’s articles,
having reviewed it in the Economic Journal the year afer it appeared in print.
According to Keynes,
Te right [to determine what constitutes money] is claimed by all modern
States and has been so claimed for some four thousand years at least. It is
when this stage in the evolution of Money has been reached that Knapp’s
Chartalism—the doctrine that money is peculiarly a creation of the State—is
fully realised.
Tus the Age of Money had succeeded to the Age of Barter as soon as
men had a dopted a money-of-account. And the Age of Chartalist or State
Money was reached when the State claimed the right to declare what thing
should answer as money to the current money-of-account . . . To-day all
civilized money is, beyond the possibility of dispute, chartalist. (1930, vol. 1,
pp. 4–5)
Chartalist money is fat money. Te state is centrally involved. A declaration
like that of the U.S. government on every piece of our paper currency, “Tis
note is legal tender for all debts, public and private,” may not be necessary for
fat currency to be accepted as a means of payment or store of value, but it
certainly helps—as does the state’s ability to enforce its declaration.
With fat money, the usual story is that the supply is fxed by the monetary
authority, typically but not necessarily a central bank. Formally, money is still
exogenous—we can ask what happens to various elements of the economy,
Copyright © 2021. Harvard University Press. All rights reserved.
including the rate of interest, at diferent levels of the money supply—but exo-
geneity doesn’t mean the same thing anymore. A central bank is part of the
fnancial, economic, and political system, and the substantive meaning of as-
suming that the money supply is determined outside the model is very difer-
ent when the supply is fxed by economics and politics than when it is fxed by
nature and technology.
fractional-r eserve banking If the issue is the ability of a capitalist
ship to right itself afer a storm without the intervention of a helmsman, we
have to take account of fractional-reserve banking. Tis question is prior to
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 505
the role of the state, and thus prior to the introduction of a central bank,
which is quintessentially a state institution (even when, like the Bank of Eng
land, it was legally a private entity for most of its history).
What happens if money is created out of whole cloth, as it is under
fractional-reserve banking? (If you are devoted to the commodity theory, re-
phrase the question as What happens when commodity money is supplanted
by, or if even that is too strong, supplemented by, fractional-reserve banking?)
How does fractional-reserve banking enter into Keynes’s story about inter-
est and employment? In a word, if a central bank changes the meaning of ex-
ogenous in “exogenous money supply,” fractional-reserve banking delivers the
frst blow to the either/or conception of using money to satisfy transactions
demand and using money to satisfy the appetites of wealth holders.5
Fractional-reserve banking is a system for creating money in the form of
credit. It is like fat money because it is divorced from a commodity base even
when its issuers purport that it is as “good as gold” (or cowrie shells). It is
pseudo-fat money, because it does not require a state guarantee—though a
state guarantee is certainly useful, and may be necessary, in encouraging wide
acceptance. Fractional-reserve banking allows the supply of money to vary
endogenously with demand via the creation and destruction of bank deposits,
creation and destruction that are part and parcel of the expansion and con-
traction of bank credit. Key here is the acceptance of bank deposits (or bank
notes) as a form of money, a way of discharging fnancial obligations. Tis is
one reason a state guarantee is practically if not logically essential.
Fractional-reserve banking need not change the story that transactions
demand and speculative demand are in competition with one another. But to
ensure that the story holds, we have to make additional assumptions, strong
assumptions that cannot be warranted in the name of expositional simplicity.
To see the dif ficulties, try to retell the story of the Keynes efect in a world
with fractional-reserve banking. Output or the price level falls. Firms require
less working capital to fnance their on-going operations. Transactions de-
Copyright © 2021. Harvard University Press. All rights reserved.
mand falls, and, so, by the way, does the value of the collateral that frms must
put up to cover their loans, especially if this collateral takes the form of physi-
cal goods (for example, the inventory of automobiles that car dealers pledge
as collateral—or gondolas in Venice).
Now what happens? In the earlier story, “the money no longer needed for
transactions must spill over into the market for fnancial assets, that is, into
the market for long-term Treasuries.” And this drives down interest rates, and
so encourages investment. Something like this will happen in the present
story if banks replace the working-capital loans they make to car dealers, res-
taurants, and others or their holdings of short-term commercial paper by
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
506 Raising Keynes
adding to their portfolios of long-term T-bonds. Tat is, the reduction in
transactions demand may be compensated by an increase in speculative de-
mand on the part of banks.
In this case, the story goes through pretty much as before, though the
mechanism is diferent. Banks operate on money demand, keeping the overall
supply constant while varying the composition of demand. Te crucial as-
sumption here is that banks always attempt to deploy their assets so as to have
no excess reserves, that they always try to stay fully loaned-up—and succeed
in the attempt!6
Without these assumptions the either/or conception of transactions money
and speculative money breaks down. Te consequence is that the theoretical
equivalence of money-supply changes and changes in the price and wage lev-
els can no longer be invoked, à la Modigliani, to argue for the self-regulating
capacity of a capitalist economy. As Keynes wrote, somewhat cryptically, just
before introducing the Keynes efect,
If the quantity of money is itself a function of the wage-and price-level,
there is indeed, nothing to hope for in this direction. (Te General Teory,
p. 266)
We can still write the equilibrium equation
M1 + M 2 = M ,
where M1 represents transactions demand and M2 asset demand, but M is no
longer exogenous. Assuming banks operate under the same ideas of liquidity
preference that Keynes attributed to nonbank agents, there may be no con-
nection between transactions money and portfolio money. Instead of fungi-
bility between the two kinds of money, the appropriate metaphor is a frewall:
the supply of money is now determined by the transactions demand (for
money as a means of payment, M1) plus an exogenously given endowment of
Copyright © 2021. Harvard University Press. All rights reserved.
money (as a store of value, M 2) in the hands of wealth holders
M ≡ M1 + M 2.
Since M1 ≡ M1, the equilibrium equation reduces to
M2 = M2.
A change in M1 is no longer automatically ofset by an equal and opposite
change in M2.
In this story the supply of transactions money is endogenous, reacting to
demand in a kind of inverse Say’s Law, at least up to the point that banks
reach the limits of reserve ratios and are fully loaned up. (For a comprehen-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 507
sive account of the fnancial side of an endogenous-money economy, see
Marc Lavoie [2014, chap. 4.])
Te supply of money for speculative purposes, that is, money as an asset in
the portfolios of wealth-holding agents, requires a separate story. If Keynes’s
M 2 is taken as given, the mechanism that produces the Keynes efect may
completely break down even though liquidity preference may be alive and
well in relating the speculative demand for money to the interest rate. (What-
ever story we tell about M2 and M 2, it can hardly be argued that the logic of
endogenous money creation for transactions purposes applies to the supply of
money as a store of value.)
Tere is another complication introduced by fractional-reserve banking
even if banks are fully loaned up: recall chapter 4, where we assumed a reserve
ratio of 40 percent and bank reserves of 200 gold ducats representing the de-
posits of wealth holders. Banks initially made loans of 300 ducats based on
these reserves, and the interest rate settled at a level that made wealth holders
content with their division of wealth between bonds and bank deposits. Now
suppose the price level increases and with it transactions demand, to, say, 400
ducats. How does money shif from asset markets to the market for working
capital? How do banks make the additional loans?
Te answer is they can’t. Banks are fully loaned up, with loans and com-
mercial paper equal to 300 ducats in their portfolios, and deposits of 500—
300 corresponding to these loans, plus the 200 ducats of wealth- holder
deposits.
Breaking the Rules to Create More Money
Banks can’t make the additional loans—until someone thinks to change the
rules of the game. Imagine that you own a bank, Banco Ziro, which is faced
with the problem of being fully loaned up. (For the purposes of this exer-
cise, assume Banco Ziro is a monopoly, which makes the accounting easier.)
Copyright © 2021. Harvard University Press. All rights reserved.
You realize that you can solve the shortage of money by selling loans worth
100 ducats to the wealth holders whose deposits form your reserves. How
will that help? By getting 100 ducats of Banco Ziro’s books, as in Tables
4.6 and 4.7. Te trick is that this transaction leaves total deposits unchanged
at 500, but instead of the original split between wealth holders (200) and
restaurants (300), the split is now 100 and 400. Te borrowers, the Vene-
tian restaurateurs, still need to keep their money in the bank in order to set-
tle their accounts with suppliers. For the settling of accounts it d oesn’t
matter that their borrowing is now only partially mediated by the banking
system.
You do have one obstacle, namely, that wealth holders might hesitate to
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
508 Raising Keynes
take on these loans because they do not know the borrowers well enough to
separate the sketchy from the solid. (Tis is normal; your business as a banker,
afer all, is to know your customer.) Te solution proposed in chapter 4 was to
consolidate the loans into a collateralized debt obligation, a CDO, which will
eliminate much of the risk in each individual loan, not to mention—no coin-
cidence—that CDOs can fnd a ready market where loans cannot, precisely
because of the idiosyncratic risks, both known and unknown, of individual
loans. Or, equivalently for present purposes, assume that a money-market
fund accepts deposits and uses them to buy commercial paper issued by the
restaurants. Once again, bank reserves are lef untouched because businesses
need deposit accounts to pay their bills.
Either way, Banco Ziro continues to have 200 ducats in reserves, against
500 in deposits. Of these deposits, 400 ducats correspond to the commercial
loans and commercial paper held by the bank (300) and the money-market
fund (100). Te remaining 100 ducats are the liquid assets of wealth holders
on deposit with the bank.
Te question remains whether wealth holders treat their money-market as-
sets (or CDOs) as liquid, or instead assimilate these assets to the inherently
risky bonds in their wealth portfolios.
In chapter 4 I made a case both ways. On the one hand, these fnancial in-
struments, by construction, are short term in nature and therefore carry little
price risk. On the other hand, money-market funds and CDOs carry some
default risk even if this is minimized by risk pooling.
Te other hand, that wealth holders treat these assets as risky bonds, is less
disruptive to liquidity-preference theory. In this case, the theory actually con-
tinues pretty much unchanged, with higher interest rates the price that must
be paid to wealth holders to induce them to give up 100 ducats of liquidity.
But now we would need an additional asset class to refect the intermedi-
ate position of the new assets between safe T-bills and long-term corporate
bonds.
Copyright © 2021. Harvard University Press. All rights reserved.
Te frst possibility, that the restaurateurs’ debts are transformed into
money-market funds ofering liquidity similar to that of cash deposits, is also
compatible with liquidity preference, superfcially at least. It simply ofers a
free ride to the economy, allowing the banking system to mobilize the money
supply toward satisfying transactions demand (up to 500 ducats in the pres-
ent example) without compromising liquidity and thus without incurring
any cost in the form of higher interest rates as the quid pro quo of reducing
liquidity.
However, this “free ride” turns out to be in its own way subversive of the
liquidity-preference edifce. Te problem is that money-market funds nor-
mally pay interest, a (partial) pass-through of the interest paid by restaura-
teurs and other businesses. If these assets pay interest and have little to no
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 509
price risk, why would anybody ever store their wealth in noninterest-bearing
accounts?
One answer is default risk, but for purposes of this exercise we shall as-
sume away this complication. Absent default risk, no one would ever hold
noninterest-bearing deposits or cash as a store of value when equally liquid
interest-bearing options are available. Checking deposits and cash still have
their uses as means of payment, but that is not relevant to liquidity-preference
theory, which is a theory of interest based on the relative benefts and costs of
holding wealth in diferent forms.
A second possibility is that deposit accounts will begin to pay interest, in
order to compete with the money-market funds—as indeed they did in real-
ity, at least prior to the long reign of the zero lower bound (zlb) that began
when Lehman Brothers crashed.
Fractional-reserve banking thus raises two problems for a theory of money.
Te frst is whether, as is the case with commodity money, transactions and
speculation are fungible, one more dollar dedicated to either of the two pur-
poses meaning one dollar fewer for the other. If the money supply expands
and contracts with transactions demand, then transactions and portfolio de-
mands become separate, sealed, compartments, each refecting a diferent use
of money, the frst money as a medium of exchange, a means of payment, the
second money as a store of value.7
Te possibility of transcending the limits of fractional-reserve banking by
moving assets of the banks’ books, as in the Banco Ziro example, is the sec-
ond problem, namely, that transactions and portfolio demands are actually
complementary activities, the same dollar serving the day-to-day needs of
business at the same time it serves the liquidity needs of wealth holders—in
this case, not only does the equilibrium equation become M2 = M 2. In addi-
tion with money serving two purposes at once, the overall equilibrium is
characterized by an inequality
M ≡ M1 + M 2 > M ,
Copyright © 2021. Harvard University Press. All rights reserved.
where M represents the sum of the endowment of asset money and the
amount of transactions money when this endowment forms bank reserves
and banks are fully loaned up.
Why Would Agents Hold Money When There
Are Safe Short-Term Bills Available?
Te preceding section raises another money-theoretic question: what is the
status of short-term credit in the theory of liquidity preference? As we saw in
chapter 12, Keynes’s own answer to this question was to assimilate cash and
short-term safe assets: “We can draw the line between ‘money’ and ‘debts’ at
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
510 Raising Keynes
whatever point is most convenient for handling a particular problem” (Te
General Teory, p. 167n). In other words, money is one end of a spectrum of
short-term credit.
Tis view leads logically to the conclusion that as debt maturity shrinks to
zero, the interest rate on this debt will similarly shrink to zero—the nominal
rate of interest on cash. If interest “is the reward for parting with liquidity for
a specifed period” (Te General Teory, p. 167), then, as the period goes to
zero, the rate of interest must go to zero.
But experience says otherwise. As was observed in the last chapter, the Fed
Funds rate, the rate charged by banks to one another for overnight loans of
reserves at the Federal Reserve, is as close as we can get to the limiting case of
a zero-duration loan. Te fact is that it is rare for Fed Funds rates to be any-
where near zero, though someone whose memory goes back only to 2008
would not have that impression.
Tis would not matter if there were a good reason why money should be
part of an agent’s portfolio despite its having to compete with short-term f
nancial assets that pay interest. Beginning with John Hicks (1946 [1939]), the
“good reason” has been the cost of converting other stores of value into an
acceptable medium of exchange. If transactions costs are high enough, then
a portion of wealth should normally be allocated to money. At the mar-
gin agents will be indiferent not only between long bonds and short-term
interest-bearing bills but also between these bills and the medium of ex-
change. Terefore, as in the cash-bonds model, the spread would be anchored
by the zero return to money.
A problem with this line of argument, as Hicks recognized, is that transac-
tions costs are likely to be proportional to the number of transactions rather
than varying with the amount transacted. Size matters. In particular, holders
of large amounts of wealth engage in transactions that are generally so large
that the cost of transacting will not be an issue:
Relatively large transactions can usually be made with very little more trou-
Copyright © 2021. Harvard University Press. All rights reserved.
ble than small transactions, but the total interest ofered on a large sum is
much larger than on a small sum; thus large capitalists will be tempted to
buy bills much more easily than small capitalists. (1946 [1939], p. 165)
For “large capitalists,” interest-bearing bills will always dominate cash, and
will be preferred as long as the interest rate is positive. Hicks argued that if
large capitalists called the tune, the demand for bills would drive their yield to
zero, so that the distinction between bills and money would disappear:
If . . . all traders reckon . . . a particular bill as perfectly safe, then there is no
reason why that bill should stand at a discount. (p. 165)
In other words, short-term bills would yield nothing to their owners.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 511
To avoid this conclusion, Hicks assumes that “small capitalists,” for whom
transactions costs are signifcant, also trade of liquidity against returns in
markets for fnancial assets. Tere are not enough large capitalists to bid the
price of bills up to par with money (and the interest rate down to zero), so the
yield on bills normally remains positive. Hicks can therefore end where he
started, with the argument that
Te imperfect “moneyness” of those bills which are not money is due to
their lack of general acceptability; it is this lack of general acceptability
which causes the trouble of investing in them, and that causes them to stand
at a discount. (pp. 165–166)
Hicks elides the key issue: he simply asserts that in the absence of transac-
tions costs, wealth holders would have such a large appetite for bills that
yields would go to zero. Later theorists have attempted to provide a basis for
substitution to take place between money and short-term assets. Te locus
classicus is the argument of William Baumol (1952) and James Tobin (1956).
Baumol and Tobin develop a model in which agents’ holdings of money for
meeting transactions optimally vary with the rate of interest on short-term
liquid assets.
Te data do suggest that the interest rate on safe bills infuences overall
money demand.8 But this does not establish that transactions demand afects
the demand for portfolio assets because the interest sensitivity of transactions
demand is a necessary rather than a suf ficient condition for integrating trans-
actions and speculative demands into a unifed portfolio theory. It must also
be the case that agents’ holdings of money for transactions impact their
wealth and hence their demands for various forms of wealth. Only if both
these conditions—the interest sensitivity of transactions demand and substi-
tutability between transactions money and speculative money—hold, would
it make sense to consider the demand for transactions money along with the
demand for interest-bearing assets in a unifed theory à la Baumol and Tobin.
Copyright © 2021. Harvard University Press. All rights reserved.
Tis theory evidently is more plausible for some agents than for others.
Inconvenience and cost may be a factor in determining the money holdings
of a majority of households without it mattering very much in the aggregate.
One reason is that the majority of households account for little wealth, and
even less fnancial wealth. Whatever the elasticity of their transactions de-
mand with respect to the rate of interest on short-term bills, it wouldn’t mat-
ter for portfolio demands because these households are not players in markets
for fnancial assets. Even were we to suppose that households with m odest
means manage their wealth portfolios in terms of the risks and returns of
various assets—a dubious assumption on behavioral grounds—the impera-
tives of convenience and cost would lead to a very inelastic demand for the
medium of exchange unless the cost of transactions was extremely low or the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
512 Raising Keynes
interest rate extremely high. In plain English, the typical household will under
normal conditions not vary its cash holdings in response to interest-rate
variation.
Tis analysis does not apply to the important players in asset markets. For
wealthy households, for nonprofts managing million-and even billion-dollar
endowments, for insurance companies, pension funds, banks, and other f
nancial institutions, the reasoning is exactly opposite to the reasoning for
households who live paycheck to paycheck. For wealthy households, as well as
for nonprofts and fnancial businesses, far from being the essential determi-
nant of the separation of transactions and portfolio holdings, the costs and
inconvenience of moving assets around are likely to be extremely small rela-
tive to the size of their portfolios. In the Baumol–Tobin model, the ratio of
money holdings to income or wealth goes to zero as the denominator in-
creases. Tese agents are the equivalent of Hicks’s “large capitalists” for whom
money is not a substitute for interest-bearing bills.
Tis leaves nonfnancial business. Businesses, especially businesses that
depend on bank credit or on the issuance of commercial paper, have both
motive and means for taking the interest rate into account in the manage-
ment of current operations. Te interest rate on short-term T-bills, or other
instruments that approximate Hicks’s safe bills, is refected directly in the
cost of borrowing for working capital. It is true that the cost impact is rela-
tively modest compared with the impact of the interest rate on longer-term
investment projects, but it can’t be ignored. Even relatively small businesses
deal in larger amounts of money than the average household, so the incon
venience and cost of transactions may not weigh as heavily as the interest
cost of holding money balances. But this may matter less than the fact that
businesses have ways of operationalizing a response to a change (say, an in-
crease) in the interest rate, namely, pressuring fnancial managers to speed up
action on accounts receivable and to slow down making good on accounts
payable. Tese tactics are not generally available to households, and the difer-
Copyright © 2021. Harvard University Press. All rights reserved.
ence between households and businesses in this and other respects supports
the hypothesis that the behavior of nonfnancial business accords with the
observed sensitivity of transactions demand to the interest rate on liquid
assets.
Of all agent classes, nonfnancial business thus ofers the best ft with the
Baumol–Tobin argument that the demand for transactions money is sensitive
to interest rates. I would go further and hypothesize that the (very m odest)
interest elasticity that shows up in the data (see note 7) is probably due to this
class of agents. But to the extent that these agents are operating on borrowed
funds, changes in transactions demand have no impact on wealth. No matter
how much their transactions demand may respond to interest rates, there is
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 513
no good basis for arguing the converse, that interest rates respond to transac-
tions demand.
Tis compartmentalization implicitly assumes that nonfnancial businesses
limit themselves to producing goods and services, distributing the fruits of
enterprise, and investing in projects designed to enhance production or pro-
duce more cheaply in the future. If businesses are also important as holders of
fnancial assets for the purpose of being able to strike at fnancial targets of
opportunity as they arise—Apple with its mountains of ofshore fnancial as-
sets comes to mind, though its motive seems to be tax avoidance more than
anything else—then the demand for transactions money for current opera-
tions would be thoroughly enmeshed with the demand for safe assets in their
wealth portfolios. In this case, the interest sensitivity of transactions demand
would in theory once again anchor the spread between various interest rates,
and we would once again have a theory of the level of interest rates, including
the hurdle rate, for investment demand. But it is dif ficult to imagine these
frms, like the large capitalists of Hicksian theory, paying much mind to the
costs of going back and forth between safe, liquid, stores of value and the me-
dium of exchange.
In any case, combining wealth holding with business pursuits in a single
portfolio is exceptional enough, I believe, that it makes sense to separate the
two functions of money, medium of exchange and store of value. For wealth
holders who are strictly portfolio managers, safe short-term bills will drive
out cash and noninterest bearing checking deposits as a liquid store of value,
relegating these forms of money to means of payment. Once short-term credit
substitutes for checking deposits and cash as the liquid portfolio asset, we are
frmly in the world we tentatively entered in chapter 4, where we considered
the possibility of of-balance-sheet short-term assets answering to transac-
tions demand and at the same time serving the liquidity demand of wealth
holders. In the perspective of Te General Teory, we can now answer the
question that frames this section: Is credit money? For purposes of the theory
Copyright © 2021. Harvard University Press. All rights reserved.
of asset management, the answer hinges on whether credit instruments can
be considered liquid assets. In my revision of liquidity-preference theory,
short-term credit plays the role that Keynes assigns to money, an instrument
that has little price risk and little perceived default risk.
And so it is in reality: T-bills exemplify this dual role, providing a safe asset
(no default risk) and at the same time a liquid one (no price fuctuations), and
so, at least most of the time, do money-market funds, repos, and other even
more exotic fnancial instruments.9 On the one hand, these instruments fund
day-to-day business transactions and government operations; on the other
hand, the same instruments provide liquidity that satisfes the portfolio de-
mand of wealth holders.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
514 Raising Keynes
What is Left of Liquidity Preference?
Once we drop the assumption of commodity money and replace it with the
more realistic assumption of fractional-reserve banking and related forms of
credit, we have to contend with both the endogeneity of money and the com-
plementarity of speculative and transactions demands. What then is lef of li-
quidity preference as a theory of interest? Surprisingly, everything that was
there before.
Chapter 12 showed that liquidity preference could never be a theory of in-
terest, even when the most favorable assumptions of commodity money were
assumed to hold. Liquidity preference can never be more than a theory of
interest-rate diferentials, spreads between interest rates on obligations of dif-
ferent maturities and diferent default risks.
An exogenously given money supply, whether commodity money or fat
money, appears to transform a theory of spreads into a theory of levels only
because of the addition of a crucial assumption, namely, that wealth portfolios
include a zero-interest asset—cash or checking deposits. Tis assumption al-
lows fnancial-market equilibrium to determine the levels of various bond
rates, including the hurdle rate for investment, because it anchors spreads
with an asset yielding a zero rate of interest.
In the world we actually inhabit, central banks substitute for commodity
money in the sense that their control over short-term rates provides the an-
chor that cash does in Keynes’s version of liquidity-preference theory. Indeed,
if central banks did not exist, we would have to invent them just for the sake
of economic theory. Alan Greenspan, Ben Bernanke, Janet Yellen, and, as of
2020, Jerome Powell are the helmsmen (and helmswoman!) who have an-
chored the spectrum of interest rates in this century.
Te role of the modern central bank, up to the fnancial crisis of 2008,
was precisely to fx the short-term interest rate and leave the rest to fnancial
markets, where demands and supplies for a variety of assets determine the
Copyright © 2021. Harvard University Press. All rights reserved.
spectrum of interest rates. In theory, the actions of the central bank and the
myriad agents acting in fnancial markets jointly produce a hurdle rate for
investment and an accompanying level of aggregate demand consistent with
full employment. In this world, liquidity preference is restored to a prominent
role even if this role is not quite what Keynes thought it was.10
Te present model implies a very diferent monetary regime from the re-
gime assumed in the simple bonds-money model of chapter 11. Now the di-
vorce between portfolios and transactions means that injections of money by
the central bank do not afect asset markets directly: agents’ portfolios are as-
sumed to hold bills and bonds, and they have no reason to hold money. In a
regime of reserve scarcity, the central bank’s choice of a particular level of the
short-term rate afects the amount of transactions money by varying the re-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 515
serves held by banks. If the banking system is geared to providing work-
ing capital for commerce and industry, we can tell a plausible story of a
fractional-reserve system in which banks create deposits that serve as the
medium of exchange for transactions, up to the limit of the required reserve
ratio. In return the banks acquire assets in the form of commercial loans or
commercial paper. Te transactions demand for money becomes the entire
demand for money.
Tis is a key departure from Te General Teory. Fractional-reserve bank-
ing erects a frewall between portfolios and transactions, and consequently M1
and M 2 become apples and oranges even if the two assets are both denomi-
nated in dollars (or pounds, euros, and yen) and are identical in appearance;
they are no longer the interchangeable parts of a homogeneous money supply.
Te existence of safe and liquid interest-bearing fnancial assets completes the
divorce between money as a means of payment and money as a store of
wealth. M1 and M1 are demands for and supplies of money—cash or checking
deposits—whereas M2 and M 2 are demands for and supplies of a short-term
interest-bearing asset. M1 and M1 are money as the asset that businesses (and
households) need to settle debts incurred in carrying on the economic activi-
ties of producing and consuming. M2 and M 2 are money as a store of value in
agents’ wealth portfolios.
Until 2008, when the Fed started to pay interest on reserves, it relied on
reserve scarcity to exercise its control over the short-term rate of interest.
From 1994, when the Fed began to announce a target for the Fed Funds rate,
it was able to vary the short-term rate without much actual change in the
overall reserve position of the banks, relying on the efect of its announce-
ments to accomplish the desired result (B. Friedman 1999; B. Friedman and
Kuttner 2011).
It is not surprising that the Fed could change rates simply by announcing a
new rate: like any monopolist, the Fed can set whatever price it chooses for its
product (bank reserves) and therefore for close substitutes—provided it is
Copyright © 2021. Harvard University Press. All rights reserved.
prepared to supply the quantity demanded at that price. More surprising is
that once the policy of announcing Fed Funds targets was put in place changes
in the rate induced very little change in demand for reserves and so very little
need for open-market operations to change the amounts of reserves.11
Since the 2008 meltdown the Federal Reserve has paid interest on reserves.
Accordingly, for banks there is little reason to prefer comparable assets like
T-bills or high-grade commercial paper. Te new regime of paying interest
has produced an unprecedented accumulation of excess reserves, just under
$1.8 trillion in August 2018, down from a peak of almost $2.7 trillion in 2014.
As in the pre-2008 period, the Fed simply announces a target rate for Fed
Funds transactions, but now no change in reserves is needed to implement a
change in the target rate.12 Te counterpart to the reserves on the liability side
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
516 Raising Keynes
of the Fed’s balance sheet is its holdings of Treasuries and mortgage-backed
securities. Its holdings of all securities reached a peak of $4.5 trillion in early
2014, of which $2.5 trillion were T-bonds and $1.7 trillion were mortgages.
By August 2018, the Fed’s assets had come down much less than bank re-
serves: the Fed still owned $2.3 trillion worth of Treasuries and $1.7 trillion of
mortgages, with total assets equal to $4.3 billion.
Real and Nominal Rates: Is Central Bank
Freedom Limited by Necessity?
So far in this story the only limit to a central bank’s power over the short-term
rate is the zlb. Tis is actually a bit anachronistic. As long as cash, with its zero
nominal holding yield, is an option, a zlb for bills makes perfect theoretical
sense, but as long as there are costs to holding cash (storage, insurance, and
the like), negative short rates become theoretically—and more important,
practically—possible. Te continuing recession in Europe led the European
Central Bank to experiment with negative short-term rates in the form of
charging banks interest on reserves. Indeed, for a large part of 2016, negative
rates were not confned to the short end of the spectrum: ten-year govern-
ment bond yields in Germany and Switzerland (and in Japan as well) also
were negative.13
Tis does not mean a central bank can set short-term rates as far into nega-
tive territory as it might like, or that there are no limits on how far bond yields
can fall. Te costs of holding cash, even if not zero, limit the discretion of
central banks as well as the possibilities for the market to drive bond yields
down. Nothing short of a cashless economy or Silvio Gesell’s stamped money
(Te General Teory, chap. 23) would remove the limit posed by the zero
nominal yield of cash. Carrying costs shif the lower bound to the short rate
into negative territory, but no further than the charge represented by these
carrying costs.
Copyright © 2021. Harvard University Press. All rights reserved.
More interesting is the question of the freedom of the central bank to con-
trol the real rate of interest, which difers from the nominal rate by the
amount of infation. Te mainstream view is that at least in the long run,
when frictions and imperfections are overcome, the real rate of interest (what
Knut Wicksell [1936 (1898)] calls the “natural rate”) is determined by forces
of productivity and thrif. In the view put forward by Irving Fisher in 1896
and by Wicksell as well, a central bank can temporarily set the short-term rate
of interest at a level incompatible with the natural rate, but economic forces
will eventually make the central bank adjust to the natural rate. If, for in-
stance, the central bank sets the short-term rate at such a low level that the
bond yield—Wicksell’s “money rate of interest”—is below the natural rate,
economic activity will be stimulated. But with the economy normally at full
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 517
employment, there would be no outlet for the stimulus other than to raise
prices. Higher prices, however, would require more transactions money, and
price infation would eventually bump up against a fxed money supply
(Wicksell 1936 [1898], pp. 116–117). Only by bringing the short rate, and
thus the money rate of interest, into line with the natural rate can the infa
tionary pressure be relieved and transactions demand brought into line with
the supply of money.
Twenty-frst-century mainstream economics refects the Fisher–Wicksell
view that the real rate is determined by productivity and thrif and is in the
long run independent of central bank policy. Te diference from the older
view is that the central bank is free to choose a short-term rate and the cor-
responding long rate. But the central bank’s choice of interest rate is purely
nominal, with only a transitory efect on the natural rate, that is, on the real
rate. In consequence, the central bank is choosing a rate of infation or defa-
tion, which is the diference between the nominal rate chosen by the central
bank (Wicksell’s money rate) and the real rate (the natural rate) determined
by desired investment and saving at full employment (the natural rate). In the
equation linking real and nominal rates,
NOMINAL − REAL =
where represents the rate of infation, causality is read from lef to right,
from the real rate and the nominal rate to the rate of infation.
Liquidity preference tells a diferent story. With the short-term rate fxed,
the yield on a long-term bond is determined by a combination of risk aver-
sion, reversion to normal, and default risk. With only bonds and bills avail-
able, portfolio choices are purely nominal, but nominal rates determine real
rates. A rearrangement of the equation linking nominal and real rates, also
read from lef to right,
NOMINAL − = REAL
Copyright © 2021. Harvard University Press. All rights reserved.
implies a very diferent causality. Chapters 6 and 7 told the story and chapter
8 provided the numbers for the Great Depression: in this argument equilib-
rium rates of price (and wage) defation are endogenously determined by ag-
gregate demand, goods supply, and labor supply, and along with the nominal
rate of interest, this equilibrium also determines the real rate of interest.
Money, Interest, and Output in the Revised Model
We can now pull all the strands of preceding chapters together. In the simplest
case of investment demand depending only on the interest rate, and saving
depending only on the level of output, aggregate demand is determined as in
Figure 12.13, reproduced as Figure 13.1. Te central bank sets the short-term
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
ID , SD
518
Investment demand in relation to
corporate bond yield
SD
ID
0.125 0.10 0.075 0.05 0.025
corp Income (Y)
100 200 300 400 500
0.025 100
Raising Keynes
0.05 200 Transactions demand
Liquidity preference combining for money
risk aversion, reversion to 0.075 300
normal, and default risk
0.10 400 M1 0 ( s ) PY
0.125 500
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
s M1 M 1
Figure 13.1 Aggregate demand and the transactions demand for money.
13 • Taking Money Seriously 519
rate, and fnancial markets determine the hurdle rate. Aggregate demand cor-
responds to the level of output at which the supply of saving equals the resul-
tant investment demand. More refned assumptions about investment and
saving complicate the AD schedule, but the logic does not change. Te AD
schedule is of course just one of the elements of the revised model. In addi-
tion goods supply, labor supply, and the adjustment mechanism all enter into
the determination of equilibrium.
Money exists in this model, but in a strict sense only as a medium of ex-
change. Paradoxically, we are back to a world in which the quantity equation
holds, transactions demand being given by M1 = PY, with the inverse of
the income velocity of money. Te only amendment to the equation as pic-
tured in the fourth quadrant of Figure 12.13 is that the parameter is no
longer a constant. If businesses weigh the costs and benefts of holding cash à
la Baumol and Tobin, then the income velocity of transactions money de-
pends on the bill rate
α = α0 ( ρ s ) ,
ε
where 0 is a constant and is the elasticity of transactions demand with re-
spect to the bill rate.14
Te quantity equation holds, but not the quantity theory. Causality does not
run from M1 to P or Y or s, but from PY and s to M1. As J. Laurence Laugh-
lin (1911) observed in debate with Irving Fisher on the operation of the quan-
tity equation (see note 6), equilibrium between transactions demand and
transactions supply is achieved because of the ability of the banking system to
create money endogenously to satisfy transactions demand.
On the assumption of reserve scarcity, open-market operations target the
supply of bank reserves to satisfy bank demands for reserves, which are a
function of the public’s demand for transactions money. Te supply of re-
serves and the corresponding supply of transactions money must be consis-
tent with the short-term rate, 0.05 in Figure 13.1.
Copyright © 2021. Harvard University Press. All rights reserved.
What Happens If There Is No Central Bank? Money, Interest,
and Output When Capitalism Is Left to Its Own Devices
Te crucial issue is what we assume about money. Te models of chapters 3 to
10 presupposed a world in which, one way or another, the rate of interest
was determined by the demand for money and its supply. In all these mod-
els, money is money: there is no diference between money as a medium of
exchange and money as a store of value. Chapter 11 developed Keynes’s
liquidity-preference theory along these lines and changed nothing essential in
the models of previous chapters. But once money as a store of value becomes
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
520 Raising Keynes
conceptually diferent from money as a medium of exchange, as in chapter 12,
the stage is set for a very diferent assessment of how a capitalist economy
functions when lef to its own devices. Without a central bank to fx the inter-
est rate on money as a store of value, that is, on short-term bills, this interest
rate is indeterminate.
Te result is that bond yields, including the hurdle rate for private invest-
ment, are also indeterminate, and so are the resulting levels of investment
demand and aggregate demand. In efect we have an IS schedule but no LM
schedule. So we can’t pin down a point on the IS schedule to associate with
each potential level of the price of goods.
Te rate of interest turns out to be a conventional phenomenon rather than
a refection of market equilibrium. Many years ago, while trying to under-
stand the psychology of Indian peasants encountering new agricultural tech-
nologies, I asked about local rates of interest charged by moneylenders. Ac-
cording to my informants, the annual interest rate, 18.75 percent, did not
change from year to year or season to season or village to village. It may be
that year afer year in village afer village 18.75 percent was the Goldilocks
rate that just equalized the demand for the moneylender’s services and their
supply. But this is not how the villagers explained the uniformity. Rather, their
explanation was that in British times, with the rupee divided into 16 annas
and the anna into 4 paise, the conventional interest charge was one paisa of
interest per month for one rupee of debt. One paisa per rupee works out to
1/64, or 12/64 on an annual basis (without compounding). Now 12/64 =
0.1875, a rate that compares favorably with the interest rates that U.S. house-
holds pay on their credit-card debt in 2020. Apparently, one paisa in the rupee
was simply a convenient heuristic, a conventional rather than a market an-
swer to balancing demands and supplies.
In any case, under a regime of capitalism lef to itself, operating in a re-
gime of fractional-reserve banking, there is no “natural rate of interest” that
emerges from the free play of market forces and regulates the accumulation of
Copyright © 2021. Harvard University Press. All rights reserved.
capital. Asset markets are one equation short, and there is no plausible way of
anchoring the spectrum of interest rates.
But there are at least two implausible ways. As chapter 10 noted, if capital
were Lego sets and could be dismantled and put together in new forms at zero
cost, we could imagine the hurdle rate being determined in a capital market.
A frm could adjust its capital stocks just as it is assumed to adjust employ-
ment. Te demand for capital (not investment!) would be determined by
frms’ marginal-productivity schedules; the intersection of the economy-wide
demand schedule with the given supply-of-capital schedule would determine
the current marginal productivity and the interest rate—in a world of Legos
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 521
there would be no distinction between long and short run because specifc
capital goods would exist only so long as they are proftable for their owners.
Tis leaves open the question of how the pace of capital accumulation
would be determined. If, as in this model, there is no investment demand per
se, the rate of growth of the capital stock must be determined by the supply of
saving. Tis is indeed the model that Paul Samuelson proposes in various edi-
tions of his elementary text (1948, 1955, 1980), and it is the model implicit in
the canonical Solow growth model (1956) that we will consider in chapter 17.
In Samuelson’s or Solow’s model, there is no investment demand and there-
fore no question of a distinct role for aggregate demand.
Te alternative, equally implausible, is to assume that capital is not fungible
but that saving and investment are insulated from markets for fnancial assets.
Or even simpler, that there is no market for fnancial assets apart from the
market for new issues that accompany the undertaking of new investment
projects. We would be in a world where the dog’s tail exists, but not the dog.
Savers commit to capital formation in the same way that entrepreneurs do, or
at least are supposed to do: once in, they are in for the duration.
Tis model is diametrically opposed to the model with fungible capital
goods. In this case there is an investment-demand schedule that is regulated
by the hurdle rate. Te anchor comes from the other end of the spectrum,
short-term interest rates adjusting (via liquidity preference) to the hurdle rate
determined by the demand for investment and the supply of saving. In this
case, unlike the case of Lego capital, an aggregate demand exists for output,
but the price mechanism—the adjustment of the hurdle rate to the demand
and supply of investment and saving—ensures that aggregate demand will not
difer from aggregate supply.
In both cases we are thrust back into Jean-Baptiste Say’s Garden of Eden,
where aggregate demand is always equal to aggregate supply. Keynes’s liquid-
ity preference plays no role in the determination of aggregate demand, out-
put, and employment.
Copyright © 2021. Harvard University Press. All rights reserved.
In the real world, or rather in a world as real as can be imagined with cap
italism lef to its own devices, capital goods are not fungible and the hurdle
rate of interest is the tail wagged by the market for fnancial assets. In this
world we are lef without any determinate theory of interest. Consequently,
the level of aggregate demand and levels of output and employment are also
indeterminate. Liquidity preference does not substitute for the neoclassical
theory of interest even as it destroys this theory!
Keynes did not think his theory lef the level of interest rates indetermi-
nate,15 but in a sense, the theoretical critique becomes even more devastating
when the indeterminacy is recognized: capitalism lef to itself is not a world of
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
522 Raising Keynes
equilibrium with unemployment, but a world of drif. All we are lef with is
historical accident to determine the level of interest rates and the accompany-
ing levels of capitalist economic activity.
Conclusions
Tere is no theory of money in Te General Teory. Or, rather, there is no co-
herent theory of money. Given the existence of a central bank, Keynes un-
doubtedly thought of the money supply as fxed by the monetary author-
ity, and the hurdle rate of interest responding, perhaps insuf ficiently, to the
money supply. But his mechanism for transmission of monetary policy in-
volves a substitution of money between diferent uses, which makes sense
only if money can be used for only one thing at a time, transactions or as a
store of wealth. And this in turn requires a homogeneity of the asset used as a
medium of exchange and the asset used as a store of value.
Te homogeneity and substitutability between store of value and medium
of exchange make sense in a world of commodity money in which there are
no interest-bearing short-term assets. It does not make sense in a world of
fractional-reserve banking, in which banks create the bulk of transactions
money in response to the public’s demand for the medium of exchange, and
in which there are alternative safe assets—safe in the sense both of being free
of default risk and of being stable in nominal value. In this world, liquidity
preference is a theory of spreads, which requires a monetary authority to an-
chor these spreads, as central banks have done by fxing the short-term rate.
For all the added complexity caused by recasting liquidity preference as a
theory of interest-rate spreads, the basic message of Te General Teory sur-
vives pretty well. At least if the question is one of how the economy works
when the short-term interest rate is under the control of a central bank.
Te results obtained in chapters 9 and 10 continue to hold. In particular,
these chapters introduce the possibility of multiple equilibria and the possi-
Copyright © 2021. Harvard University Press. All rights reserved.
bility of instability. In some cases the instability depends not only on the ad-
justment process, but on parameter values, specifcally, on the relative speeds
of adjustment of prices, wages, and output.
Liquidity preference reverses the conventional view of the relationship be-
tween nominal and real interest rates. Te conventional view is that the “natu-
ral” real rate is determined by the forces of productivity and thrif—the de-
mand for investment and the supply of saving—and the nominal rate or the
rate of infation adjusts to the real rate. Keynes’s view, which the modifcations
introduced in chapter 12 do not change, is that nominal rates are determined
jointly by the central bank and liquidity preference. Together with investment
and saving propensities and supply conditions, the nominal hurdle rate of in-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Taking Money Seriously 523
terest determines the levels of output and employment and the rate of infa
tion, and as part of this process the real rate of interest.
Te incoherence of Keynes’s monetary theory is in part due to the double
purpose of Te General Teory. A model in which the central bank steers the
economy is conceptually clearer than a model of capitalism as a self-regulating
system: we face the inherent ambiguity of a counterfactual in specifying what
we mean by capitalism lef to its own devices once we try to take account of
money and banking. Te sticking point is how money is defned. Te simplest
assumption is that money is a commodity, but with this defnition we exclude
fractional-reserve banking and the creation of money by the banking system.
A central bank is clearly out of bounds, but fractional-reserve banking seems
well within the realm of possible capitalisms lef to their own devices—as
does the existence of safe interest-bearing assets that dominate money in
wealth portfolios.
Te two assumptions, commodity money or fractional-reserve banking
with safe interest-bearing assets, lead to very diferent results. With commod-
ity money, the reasoning of Te General Teory leads to a determinate level
(or schedule) of aggregate demand. With the other elements of the model,
goods supply and labor supply, we have a determinate equilibrium once we
specify the adjustment mechanism. Te problem is that the equilibrium may
fall short of full employment (or, in theory at least, may lead to overfull
employment). In the revised model, with commodity money replaced by
fractional-reserve banking and a wedge between the medium of exchange
and the store of value, the problem is more profound. With a central bank,
aggregate demand responds to the short-term rate, and the overdetermina-
tion introduced by adding an AD schedule to the GS and LS schedules can be
resolved by focusing on the adjustment process. But without a central bank,
there is an equilibrium hurdle rate for each value of the bill rate, but no way to
anchor the spread, and therefore no determinate equilibrium. Without a cen-
tral bank, the AD schedule is itself indeterminate, so specifying an adjust-
Copyright © 2021. Harvard University Press. All rights reserved.
ment process does not determine an equilibrium. Te implication is even
more disturbing than Keynes imagined: not only might capitalism lef to its
own devices fail to deliver jobs for all willing workers, the model is incapable
of determining how much employment will be generated by the economy
given the underlying data.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Baumol–Tobin
Te Baumol (1952)–Tobin (1956) formula for the transactions demand for
money is
1
fPY 2
M1 = , (13.1)
2 s
where
M1 = the average holding of money for transactions purposes,
f = bank fee per conversion of bills to money (and vice versa), assumed to
be independent of transaction size,
s = the bill rate, assumed to be riskless,
P = price level,
Y = real income.
Equation (13.1) is the frst-order condition for an interior solution to the
problem of minimizing the direct and indirect costs of holding cash instead of
bills. Te agent is assumed to have the option of holding some of her income
in the form of interest-paying bills instead of noninterest-bearing cash. Te
downside of bills is that the bank charges a fee every time the agent makes a
change in the composition of her portfolio. Te cost of bank transactions is
the direct cost, fm, where
Copyright © 2021. Harvard University Press. All rights reserved.
m = number of conversions between bills and money
plus the interest foregone for every dollar held as cash instead of bills, namely,
sM1. Total transactions costs are the sum of the two components
TC = fm + sM1, m = 0, 2, 3, ... (13.2)
Te initial transaction is an exchange of money, the assumed means of pay-
ment for the agent’s income PY, for bills. Hence, if there are any transactions
at all, there must be at least two.
To make the math more manageable, we assume that exchanges of bills for
524
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
13 • Mathematical Appendix 525
money are of equal size and that the interval between them is constant and
the same as the interval between the initial exchange of money for bills and
the frst conversion of bills into money. Te money itself is assumed to be paid
out at a constant rate between bank transactions. For the purpose of this exer-
cise, assume saving is equal to zero, so that the entire income is spent over the
year. On these assumptions, if PY, the agent’s annual income, is paid in a lump
sum at the beginning of the period, then her average holding of money is
PY
M1 = if m ≥ 2, (13.3)
2m
PY
M1 = if m = 0. (13.4)
2
Substituting from equation (13.3) into (13.2) gives
s PY
TC = fm + if m ≥ 2, (13.5)
2m
s PY
TC = if m = 0. (13.6)
2
If we treat m as a continuous variable, we can diferentiate equation (13.5)
with respect to m to obtain the frst-order condition for an interior solution
(m ≥ 2) to the agent’s cost-minimization problem
1
PY 2
m= s . (13.7)
2f
Again substituting from equation (13.3), equation (13.7) gives the equation
for the average money holding
1
fPY 2
Copyright © 2021. Harvard University Press. All rights reserved.
M1 = . (13.1)
2 s
Te transactions fee f must be small relative to potential interest sPY for the
interior solution to dominate the corner solution (m = 0). In this case, divid-
ing both sides of equation (13.1) by PY gives
1
M1 f 2
= . (13.8)
PY 2 s PY
So, as PY → ∞, M1/PY → 0 and m → ∞. At the other end of the spectrum—
think of agents who are paid relatively small sums weekly or even monthly—
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
526 Raising Keynes
f is large relative to potential interest sPY. In this case, the corner solution
(m = 0) will dominate.
Te important point is that, in both extreme cases, the lesson we draw is
opposite to the usual lesson of the Baumol–Tobin framework. Optimal money
holdings will not be sensitive to small changes in the bill rate. At m = 0 the
interest rate would have to be suf ficiently large to displace the optimum into
the interior of the solution space before it can have any efect on money hold-
ings. For an interior solution we have
M
∂ 1
PY s = − 1 s . (13.9)
∂ s M1 2 PY
PY
So, at the other end of the spectrum, as Y → ∞, the elasticity of demand for
cash goes to zero.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. V .
Fiscal Policy in Theory and Practice
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 14 .
functional finance and the
stabilization of aggregate demand
It is obvious that in the long run the budget must balance, but what is
the logical period?
—jacob viner
As soon as it is recognized as a duty of the government—perhaps
even the primary duty of the government—to ensure the mainte-
nance of full employment, and that any so-called principle of “sound
fnance” that might interfere with this task can have no possible
justifcation, the instruments by which full employment can be
maintained stand out clear and unmistakable.
—abba lerner
Te General Teory argues that a capitalist economy lef to its own devices
may come up short in the jobs department and that there are limits to the
ability of monetary policy to steer the economy to full employment. Not only
might there be a liquidity trap that prevents the hurdle rate of interest from
falling to a level low enough to coax out investment demand suf ficient for full
employment; but even if the hurdle rate could be driven to zero, the result-
ing investment demand might be insuf ficient. Tere is a third reason that
emerged from the study of investment demand in chapter 10. Since it is long-
Copyright © 2021. Harvard University Press. All rights reserved.
term investment that responds most to the interest rate, residential construc-
tion becomes the swing element on which monetary policy operates. As long
as the swings are not enormous, the overweighting of one element of invest-
ment demand might not matter. But the reliance on monetary policy to bring
the economy out of the 2001 recession had unintended consequences both for
the price of the existing stock of housing and for new construction. Te U.S.
and world economies reaped the whirlwind when the housing boom turned
to bust in 2007 and 2008.
All three of these arguments provide a theoretical justifcation for fscal
policy. In the presence of a liquidity trap or inelastic investment demand, full
employment requires the active management of taxation and spending. Even
529
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
530 Raising Keynes
when the economy is not in extremis, the possibility that monetary policy will
skew the composition of investment makes fscal policy an appropriate com-
plement for monetary policy at times when monetary policy alone would be
sufficient to move the economy to full employment.
Keynes, as I have noted, has surprisingly little to say on the subject of fscal
policy. By my count the term appears seven times in Te General Teory; all
seven are in the context of how taxation might afect the propensity to con-
sume.1
Te most clear reference in Te General Teory to fscal policy is indirect:
If the Treasury were to fll old bottles with banknotes, bury them at suitable
depths in disused coalmines which are then flled up to the surface with
town rubbish, and leave it to private enterprise on well-tried principles of
laissez-faire to dig the notes up again (the right to do so being obtained, of
course, by tendering for leases of the note-bearing territory), there need be
no more unemployment and, with the help of the repercussions, the real
income of the community, and its capital wealth also, would probably be-
come a good deal greater than it actually is. It would, indeed, be more sen-
sible to build houses and the like; but if there are political and practical dif
fi
culties in the way of this, the above would be better than nothing. (p. 126)
Keynes’s point here is to emphasize the multiplier efects of government
spending, and to argue that if there are no good projects on the table, the mul-
tiplier alone justifes unproductive spending; in the second and subsequent
rounds of spending, the newly employed workers are producing useful goods.
Keynes’s activist fscal policy did not begin with Te General Teory. I ob-
served in chapter 1 that he had a long record of supporting government
spending as a means of stimulating the economy, going back to the U.K. elec-
tion campaign just before the Depression. Nor was Keynes alone in seeing
government spending this way. Chapter 2 pointed out that Jacob Viner had
used strikingly similar language in 1933, three years before the publication of
Copyright © 2021. Harvard University Press. All rights reserved.
Te General Teory.2 Te diference between Keynes, on the one hand, and
Viner and other proponents of activist fscal policy, on the other, was that
Keynes alone ofered a framework in which the cause of unemployment—a
failure of aggregate demand—and the limits of monetary policy—a liquidity
trap and/or inelastic investment demand—are integrated, a framework in
which activist fscal policy then makes logical sense. In the absence of such
a framework, the trained economist, who had mastered the art of seeing
through common sense, was no more disposed to fscal activism than any
other solid citizen.
It is not surprising that Keynes became identifed in the public mind with
the idea of compensatory public fnance, the idea that the government should
lean against the wind if there is a shortfall or an excess of private expenditure
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
14 • functional finance and the stabilization of ad 531
relative to the supply of goods at full employment. And Keynes became iden
tifed also with an upsetting corollary: defcit fnance and unbalanced bud-
gets. Although it turns out that the logic of activist fscal policy does not
require defcit fnance, there is a good reason for running defcits under con-
ditions of slack private demand. Defcits provide more bang for the buck than
a balanced-budget approach to compensating for a lack of private demand.
Te fact is that Keynes, like Viner, does not appear to have challenged the
idea that the government should balance its budget over time, though, again
like Viner, he questioned the need to do so in every short period. But there is
an important diference between Keynes and Viner. Viner never rejected the
conventional logic that a government’s budget, like that of a household or a
business, must balance, at least in the long run. For Keynes the issue was not
logic but practicality in the light of very strong public resistance to the idea
of government debt growing without bound. It was not Keynes but one of
Keynes’s disciples, Abba Lerner, who laid out the logic of compensatory f
nance in its most provocative form, “functional fnance,” which Lerner delib-
erately contrasted with “sound fnance” (1941, 1943, 1944).3
How Does Fiscal Policy Get the Economy
to Full Employment?
Functional fnance emerged from the discussion and debate about what the
government could do to avoid a slump once World War II ended. Recall Fig
ure 9.4, which summarizes Paul Samuelson’s (1943) understanding of the
problem. Figure 14.1 is adapted from Figure 9.4 with the consumption func-
tion simplifed by assuming that the propensity to consume is constant. In
this fgure the LS schedule is collapsed into a vertical line representing a
unique full-employment level of output, here represented by YFE = 480. Te
vertical distance between the 45-degree line and the consumption function
represents the additional spending necessary for aggregate demand to equal
Copyright © 2021. Harvard University Press. All rights reserved.
the full-employment level of output. In a closed private-enterprise economy,
that is, an economy without government or foreign trade, the spending gap
consists of the saving that would be undertaken at full employment.
Samuelson’s (and Keynes’s) doubts and fears for the post–World War II
economy revolved around whether private investment demand would be suf
fcient to ofset the saving gap, 120 in this simplifed example. One possible
source of salvation, we saw in chapter 9, was the “wealth efect.” Accumulated
wealth and unfulflled consumer demand, both the result of wartime re
strictions on the production of consumer goods, might combine to shif the
consumption function upward, at least temporarily, and narrow the spend-
ing gap.
But the pessimists doubted that this would be suf ficient. If the stimulus to
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
532 Raising Keynes
Expenditure (E D ) ED Y
600 Consumption function
500
E D FE 480
Full-employment spending gap =
400 Full-employment saving
C D FE 360
300
200
100
45°
YFE 480
100 200 300 400 500 600 Income (Y)
Figure 14.1 Full-employment spending gap, I.
consumption from the wealth efect (CA) is 20 and investment demand (ID) is
40, this still leaves a gap, as in Figure 14.2. Te expenditure function, which is
given by adding the additional consumption and investment to the original
consumption function, now intersects the 45-degree line at Y = 240, refecting
the multiplier of 4 (= 1/(1 − MPC) = 1/(1 − 0.75)). Tis level of aggregate de-
mand is only half what is required for full employment. With a multiplier of 4,
the spending gap is now 60.
A Keynesian solution is for the government to substitute peacetime ex
penditure for military expenditure. If G, government expenditure on the pur-
chase of goods and services, is 60, the expenditure function will be displaced
upward, as in Figure 14.3. When G = 60 is added to CA + ID = 60, the addi-
Copyright © 2021. Harvard University Press. All rights reserved.
tional spending closes the original gap, with aggregate demand equal to 480.
Te impact of the additional spending is magnifed by the multiplier. New
government spending equal to 60 has an immediate impact since it represents
an excess of total expenditure over total output and income. Tis excess in-
creases the price level (with fexprice adjustment) or depletes inventories
(fxprice adjustment), or some of both. Te result is to increase production
and thus income: under a fxprice regime producers respond directly to the
new expenditure by replenishing inventories; under a fexprice regime the
increase in prices leads to a positive gap between price and marginal cost,
which in turn leads to greater production.
Expansion leads to a further increase in consumption demand, since three-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
14 • functional finance and the stabilization of ad 533
Expenditure function
C D ID 0.75Y + 60
Expenditure (E D )
ED Y
600
500
ED FE 480
Full-employment spending gap
400
Consumption function
C D 0.75Y + 20
300
Aggregate demand
240
200
100 CA ID 60
CA Additional
45° consumption = 20
240 YFE 480
100 200 300 400 500 600 Income (Y)
Figure 14.2 Full-employment spending gap, II.
Expenditure function
C D ID 0.75Y +1 20
Expenditure (E D )
ED Y
600
500 Aggregate demand
E D FE 480
Consumption function
400 C D 0.75Y + 20
Copyright © 2021. Harvard University Press. All rights reserved.
300
240
200
CA ID G 1 20
100
CA Additional
45° consumption = 20
240 YFE 480
100 200 300 400 500 600 Income (Y)
Figure 14.3 Full-employment aggregate demand with spending gap flled by government
spending.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
534 Raising Keynes
P
W
5
Supply of goods
4
3 Supply of labor
1
Aggregate demand
240 480
100 200 300 400 500 600 Y
Figure 14.4 Government expenditure shifs aggregate demand.
quarters of each dollar of income is spent on consumption. If the economy is
on its AD schedule,4 the process continues until “leakage” into saving, twenty-
fve cents on the dollar, puts an end to further expansion of output. Te mul-
tiplier, the ratio of the change in output to the original stimulus, is 4, so that
government purchases of goods and services equal to 60 leads to an increase
in aggregate demand of 240. Whether the adjustment regime is fxprice or f-
exprice, the new level of aggregate demand produces a full-employment equi-
librium. Te picture is in Figure 14.4, in which the AD schedule without
government spending is shown by the solid line, and the AD schedule with
G = 60 is shown by the dashed line.
From the perspective of traditional canons of government fnance, there is
an obvious problem: the government runs a defcit as long as private demand
falls short of full-employment output. To match spending by taxation will re-
Copyright © 2021. Harvard University Press. All rights reserved.
duce consumption demand and undermine the positive efect on aggregate
demand from G.
To see this, suppose that private consumption is determined by disposable
income, the amount lef over afer individuals meet their tax obligations. If
taxes are a constant fraction of income, t, consumption is now determined by
(1 − t)Y. Let t = 0.2. Ten the consumption function is given by
CD = 0.75(1 − 0.2)Y + 20 = 0.6Y + 20
and aggregate expenditure is
ED = CD + ID + G = 0.6Y + 120 = 300.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
14 • functional finance and the stabilization of ad 535
Expenditure function
C D ID 0.6Y +1 20
Expenditure (E D )
ED Y
600
500
E D FE 480
Full-employment spending gap
400
Consumption function
Aggregate demand
300 C D 0.6Y + 20
200
CA ID G 1 20
100
CA Additional
45° consumption = 20
240 YFE 480
100 200 300 400 500 600 Income (Y)
Figure 14.5 Balancing the budget.
Te picture is in Figure 14.5. Taxation has the efect of reducing the marginal
propensity to spend. Taxes, like saving, represent a leakage that reduces the
multiplier, in this case from 4 to 2.5, with the result that the additional spend-
ing of 120 (as compared with Figure 14.1) leads to aggregate demand equal to
300 rather than to 480. In particular, the 60 of government spending leads to
an additional 150 of output, rather than the 240 associated with G = 60 in Fig
ure 14.3.
Tree points are noteworthy. First, the government budget is balanced
since tY = G = 60. Second, the new level of aggregate demand is well below the
full-employment level of 480. At full employment there is once again a spend-
ing gap, now equal to 72. Tird, government spending with a balanced budget
Copyright © 2021. Harvard University Press. All rights reserved.
still has a positive efect on output compared with no government spending
(and no taxes). Instead of aggregate demand equal to 240, as in Figure 14.2, in
Figure 14.5 aggregate demand is 300.
Alternative Fiscal Policies for Full Employment
Indeed, under the simplifying assumptions of the present model, the so-
called “balanced-budget multiplier,” the addition to aggregate demand result-
ing from an increase in government purchases of goods and services just bal-
anced by an increase in taxation, is one. Why? Te answer lies partly in the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
536 Raising Keynes
logic of the multiplier, and partly in the special assumptions of the present
model. Te logic of the multiplier is that taxation reduces spending by the
same amount as the increase in spending by those who sell goods and services
to the government.
So why isn’t the multiplier zero? Te reason is that the initial round of gov-
ernment spending involves the purchase of goods and services, the creation of
new output, and this has no counterpart in the multiplier chain correspond-
ing to taxation. It would be diferent if instead of purchasing goods and ser
vices government expenditure took the form of transfer payments, payments
such as unemployment insurance, for which no output shows up on the other
side of the ledger. In that case, the multiplier would indeed be zero.
Tis is the general logic, but the specifcs of the model, particularly the as-
sumption of a uniform marginal propensity to consume, play a role as well.
As will be seen in chapter 15, if the marginal propensity to consume difers
across the economy, the impact of taxation will depend in large part on the
marginal propensity to consume of the taxpayers whose taxes are raised to
balance government spending. And the impact of spending will depend on
the propensities to consume of suppliers of goods and services to the govern-
ment. Te upshot is that the balanced-budget multiplier will be greater than
one if the propensities to consume on the spending side are greater than the
propensities to consume of taxpayers, less than one in the reverse scenario.5
For present purposes, we ignore these complications and assume that the
balanced-budget multiplier is precisely one.
Te theory of fscal policy developed by Lerner regards sound fnance in
the form of balanced budgets to be a hindrance to efective economic man-
agement, one to be done away with under the new dispensation of functional
fnance. Budget balance does indeed present a constraint on fscal policy, but
the constraint is not so simple: governments are not required to run a defcit
to achieve the goal of full employment.
Continue to assume that the possibilities for expanding investment de-
Copyright © 2021. Harvard University Press. All rights reserved.
mand have been exhausted. I = 40 is the maximum that will be forthcoming
when the central bank has done its all to lower rates. In this case, fscal policy
is the only remaining tool.
Tere is more than one fscal policy that will lead to aggregate demand of 480.
One possibility, as we have seen, is for the government to spend 60 while col-
lecting no taxes. Another is to make use of the balanced-budget multiplier to
arrive at an amount of spending that is both balanced by taxation and suf fi
cient to raise demand to a level of 480.
To fnd the requisite rate of taxation and the corresponding level of govern-
ment expenditure for a balanced-budget solution to the problem of inade-
quate private demand, we calculate the aggregate expenditure
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
14 • functional finance and the stabilization of ad 537
ED = CD + ID + G = 0.75(1 − t)Y + 60 + G
with ED = Y = 480 and tY = G. Solving the equations together gives G = 240,
t = 0.5, and CD = 200. Te picture is in Figure 14.6.
Te possibilities given by Figures 14.3 and 14.6 are only two of an infnite
number of combinations of G and t that would give rise to a full-employment
level of aggregate demand. Even though each is associated with a diferent
level of government expenditure and taxation (hence a diferent level of the
defcit) and a diferent value of the multiplier, all of these fscal policies shif
the level of aggregate expenditures from 240 to 480.
Tis raises a question about the necessity of rejecting sound fnance in
order to achieve the full-employment objective of functional fnance. Why
do we have to look beyond the balanced-budget confguration of taxes and
spending represented in Figure 14.6? If there is a balanced-budget solution to
the problem of unemployment, in what sense is sound fnance a constraint on
efective economic management?
Te problem is that a balanced budget commits us to a particular size of
the government budget. In the above numerical example, a balanced-budget
path to full employment involves government expenditure (and taxes) of 300,
while the defcit path involves expenditure of 60 (and taxes equal to 0). Te
defcit unleashes the full power of the multiplier, whereas a balanced budget
prevents the multiplier from kicking in.6
Te size of government is problematic for two reasons. One is the need to
Expenditure (E D )
ED Y
600
500 Aggregate demand Expenditure function
E D FE 480
C D ID 0.375Y + 300
400
Copyright © 2021. Harvard University Press. All rights reserved.
CA ID G 300
300
200 Consumption function
C D 0.375Y + 20
100
CA Additional
45° consumption = 20
YFE 480
100 200 300 400 500 600 Income (Y)
Figure 14.6 Balancing the budget at full employment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
538 Raising Keynes
balance the benefts of public and private expenditure. Tis problem will be
addressed in chapter 16. Te second problem is that a large government bud-
get is difficult to manage once the context shifs from a one-time calculation
of a full-employment saving gap to the calculation and recalculation of this
gap in the context of fuctuating private demand. Even if the government had
all the relevant information about economic conditions, it would be like turn-
ing a battleship around in a small harbor to continually adjust taxes and
spending to the state of private demand. As in the analysis of equilibrium in
the absence of government, real-time changes in budgets are diferent from
comparative statics. In comparative-static comparisons of the positions of a
battleship in the harbor, the battleship doesn’t move, but moving is precisely
the problem in the analysis of real-time changes.
Lack of information about the battleship complicates the problem of turn-
ing it around. Milton Friedman (1961) identifed two issues. Te frst is a
recognition lag, a delay in gathering and interpreting statistics that indicate
the state of private demand. Te second is an implementation lag, a delay in
taking the steps to formulate and enact a plan to counter a shortfall or an ex-
cess of private demand relative to what full employment requires.
Both recognition and implementation lags present a problem for monetary
policy as well as for fscal policy, but implementation is orders of magnitude
more dif ficult for fscal policy. With monetary policy, the implementation
problem is to form a consensus among twelve more or less like-minded pro-
fessionals who make up the Federal Open Market Committee (the policy
committee of the Federal Reserve). For fscal policy, the problem is to form a
consensus of two house of Congress and the executive branch, each with its
own political agenda, indeed each with multiple conficting agendas. Herding
cats seems easy by comparison.7
In any event, post–World War II capitalism has relied more on changes in
the defcit than on changes in the size of the government budget to regulate
aggregate demand. Instead of changing expenditures and taxes in the same
Copyright © 2021. Harvard University Press. All rights reserved.
direction, as required to maintain a balanced budget, the Keynes–Lerner pre-
scription for stabilizing aggregate demand envisages smaller swings in both
expenditures and taxes—in opposite directions: reducing taxes and raising
government expenditures when private demand slackens, and moving in the
opposite direction when private demand strengthens.
Tis prescription has been followed to a great extent even by the fercest
ideological opponents of everything Keynes stood for. Even before the publi-
cation of Te General Teory—see the data on the government debt during
the Great Depression in Table 8.16—capitalist economies had incorporated
one part of the Keynesian prescription for stabilizing aggregate demand into
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
14 • functional finance and the stabilization of ad 539
the very architecture of public fnance, under the rubric of automatic stabili-
zation.
Automatic Stabilization
Automatic stabilization has two components. First, expenditures tend to re-
main relatively stable over the cycle. Second, tax revenues tend to vary more
or less proportionately with output as long as tax rates do not change. Te
consequence is that the defcit rises in slack times and falls in good times—
even if nobody has a hand on the tiller. Te consequences are clear in Figure
14.7. Te only clues to the identity of the party or the president in power are
the dates at the bottom of the graph. Te defcit rose together with unemploy-
ment in every recession since 1970, and the defcit and unemployment fell
together in every recovery.
It’s not quite true that in practice nobody’s hand was at the tiller—even
though the helmsmen sometimes had very odd ideas on how to steer the
economy. Discretionary fscal policy, as distinct from automatic stabilization,
is relatively rare for the same reasons that relying on the size of the budget is
cumbersome. And in fact the U.S. government has used this option only in
extremis. But automatic stabilizers were supplemented by Ronald Reagan’s
10.0
7.5
5.0
2.5
0.0
Copyright © 2021. Harvard University Press. All rights reserved.
−2.5
−5.0
1950 1960 1970 1980 1990 2000 2010
Deficit as percentage of GDP
Unemployment as percentage of civilian labor force
Figure 14.7 Unemployment and the federal defcit (1948–2018). Sources: BLS,
OMB. myf.red/g/pY6r
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
540 Raising Keynes
supply-side tax cuts in the early 1980s. And George W. Bush’s supply-side tax
cuts in the early years of this century led to the largest increase in the defcit
since the Great Depression, at least until the Great Recession. More tradi-
tional, demand-oriented, tax cuts were put in place by Bush in 2008, at the
end of his second term, as the downturn was getting underway. Both tax cuts
and purchases of goods and services (principally by means of additional
transfers to state governments) were a major piece of the Obama stimulus
enacted in 2009.
Even without these extraordinary interventions, the very structure of taxa-
tion and spending should have meant that the ups and downs of private de-
mand would not have had the same impact as in the pre–Great Depression
era of small government. But did they do so in practice? If we look at the
overall rate of unemployment, fuctuations do not seem to have moderated.
Figure 14.8 suggests that post-Depression fuctuations have been as violent as
the fuctuations between 1890 and 1930.
Appearances, however, are deceiving, because two contradictory forces are
at work. On the one hand, the importance of agriculture in terms of both
GDP and employment prior to the Great Depression dampened fuctuations
in the unemployment rate. On the other hand, the role of government, as we
have seen, has been increasing in importance since the Depression. Tis has
to some extent ofset the decline in the role of agriculture.
Te reason why the decline in agriculture matters for the measurement of
fuctuations is that the agricultural labor force is part of the denominator of
the unemployment rate, but in the days when family farms dominated, agri-
culture—for reasons explored in chapter 8—did not function according to the
25.0
Percent of the civilian labor force
20.0
Copyright © 2021. Harvard University Press. All rights reserved.
15.0
10.0
5.0
0.0
1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
Figure 14.8 Unemployment. Sources: 1890–1990, Historical Statistics of the
United States; post-1990: myf.red/g/pY6r
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
14 • functional finance and the stabilization of ad 541
norms of capitalism. Indeed, if we defne capitalism in terms of the predomi-
nance of wage labor, agriculture was not really part of the capitalist economy
except as a reservoir from which to draw labor. For the same reasons, agricul-
tural unemployment tended to be insignifcant in bad times as well as in good
times. So it makes sense to subtract agricultural employment from both the
denominator and the numerator in calculating capitalist-sector employment
and unemployment rates.
In the postwar period, government has, as we have seen, grown markedly.
Like agriculture, government doesn’t play by capitalist rules. Governments,
especially state and local governments, do engage in layofs, but unemploy-
ment is much rarer in this sector than in the private sector. If we are trying to
understand diferences over time in fuctuations of the unemployment rate
under capitalism, it makes sense to subtract government employment from
both the numerator and the denominator in calculating the unemployment
rate.
If we eliminate both agriculture and government from the calculations and
compute unemployment as a percentage of the nonfarm civilian labor force,
we get unemployment rates that are consistently higher than the of ficial se-
ries. More to the point, the unemployment rate fuctuates much less in the
postwar period than in the pre-Depression period. Figure 14.9 plots the two
series. If we focus on the capitalist-sector unemployment rate, as refected in
35.0
30.0
25.0
20.0
Percent
15.0
Copyright © 2021. Harvard University Press. All rights reserved.
10.0
5.0
0.0
1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
Unemployment as percentage of civilian private nonfarm labor force
Unemployment as percentage of civilian labor force
Figure 14.9 Two Measures of Unemployment. Sources: 1890–1990, Historical
Statistics of the United States; post-1990: myf.red/g/pY6r
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
542 Raising Keynes
the unemployment rate as a percentage of the civilian nonfarm labor force, it
is clear that there has indeed been a decrease in the volatility of unemploy-
ment.8
Summary: Functional Finance as a Tool for Stabilization
In principle, budget defcits are not necessary for managing aggregate de-
mand with an eye to full employment. If society is indiferent to the size of
government and if we abstract from the lags that prevent timely changes in
the government’s budget, aggregate expenditure can be increased or reduced
by changing spending and taxes in the same direction, with the government
budget always in balance.
But neither of these assumptions holds in practice. Recognition and imple-
mentation lags are endemic, and the size of government is a perennial issue.
In practice, compensatory, countercyclical, variations in the defcit are a cor-
nerstone of fscal policy. Moreover, these changes in fscal stance are built into
the fscal system via the inertia of tax rates and spending levels. Automatic
stabilization is responsible for much of the observed correlation between
the defcit and the unemployment rate, a correlation that transcends personal
and partisan preferences. In point of fact, automatic stabilization has played a
consistent if unsung role in mitigating the fuctuations in private demand.
To be sure, this success does not show up in the overall unemployment
statistics, which suggest that the economy has been no less volatile since
World War II than it was before the Great Depression. But unemployment
relative to the nonfarm-private-sector labor force (as close as we can come to
a measure of unemployment in the capitalist sector) tells a diferent story,
namely that capitalism has become noticeably less volatile since the post–
Great Depression regime of automatic stabilization was put in place.
In extremis, however, notably during the recession of the 1980s and the
recession that followed on the fnancial crisis of 2008, automatic stabilization
Copyright © 2021. Harvard University Press. All rights reserved.
would have been insuf ficient. Ronald Reagan spoke the language of supply-
side economics but took appropriate Keynes–Lerner measures to stimulate
demand when the private sector faltered. When recession loomed again in
2001, George W. Bush undertook discretionary measures to stimulate de-
mand, while covering his fscal butt with Reagan’s supply-side language. And
he again deployed discretionary measures (this time abandoning the supply-
side veneer) in 2008, in the face of the gathering storm that became the Great
Recession.
It remained for Barack Obama to react with what, up to 2020, was the
greatest peacetime fscal stimulus in the history of the Republic, the American
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
14 • functional finance and the stabilization of ad 543
Recovery and Reinvestment Act. Te next chapter explores whether or not
the Obama stimulus was successful.
Tere remains another question: what about the debt? Te theoretical pos-
sibility of balancing the budget at full employment may not be practical, but
nothing precludes balance over the cycle, with surpluses in the good years
canceling out the defcits in lean times, a là Joseph in Pharaoh’s Egypt. By
varying the mix between the size of the government and the size of the defcit
(and surplus), it is possible to keep the economy on an even keel of full em-
ployment—without a rising debt. If deploying fscal policy to stabilize the
economy at full employment does not explain the long-term rise in the na-
tional debt, what does? We shall return to this question in chapter 16 afer we
examine the Obama stimulus in chapter 15.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 15 .
did the obama stimulus work?
You know, you d on’t need to be an economist to know that jobs are
the engine of our economy. And without jobs, people can’t earn. And
when people can’t earn, they can’t spend. And if they d on’t spend, it
means more jobs get lost. It’s a vicious cycle. And that’s the vicious
cycle we’re in today. And it’s one of the reasons that this bill is so
desperately needed. We have to reverse that cycle.
But everyone knows that jobs are more than just about a job.
Tey’re about dignity. Tey’re about respect. Tey’re about being able
to get up in the morning, look your child in the eye and say, “Every
thing is going to be okay; everything is going to be okay.”
—vice president joe biden, remarks at the signing of
the american recovery and reinvestment act,
denver, colorado, february 17, 2009
And what we see with [President Obama’s] plan is a lot of spending
that I just don’t think will work . . . We can’t borrow and spend our
way back to prosperity . . . Providing $300 billion of this package to
states—$166 billion in direct aid to the states, another $140 billion
in education funding—this is not going to do anything, anything to
stimulate our economy.
—house minority leader john boehner, on
“meet the press,” january 25, 2009
In late 2008 and early 2009 the U.S. economy was hemorrhaging jobs at a rate
Copyright © 2021. Harvard University Press. All rights reserved.
unseen since the Great Depression. Indeed, the chart of unemployment from
2006 through 2009 in Figure 15.1(b) looks uncannily like a chart of unem-
ployment over the years 1927 to 1930 in Figure 15.1(a). Fortunately, the sub-
sequent history turned out very diferently, as Figure 15.2 shows. Instead of
plunging into depression, the economy gradually recovered afer 2009. From
early 2009 through early 2012, the ARRA (the American Recovery and Re
investment Act) injected over $750 billion of spending and tax cuts in order
to stimulate the U.S. economy, most of the money being spent between mid-
Portions of this chapter were previously published in Stephen A. Marglin and Peter Spiegler,
“Unpacking the Multiplier: Making Sense of Recent Assessments of Fiscal Stimulus Policy,” So-
cial Research 80 (Fall 2013): 819–854.
544
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 545
10 10
9 9
8 8
7 7
6 6
Percent
Percent
5 5
4 4
3 3
2 2
1 1
0 0
1927 1928 1929 1930 2006 2007 2008 2009
(a) Then (b) Now
Figure 15.1(a) Unemployment in the Great Figure 15.1(b) Unemployment in the Great
Depression (1930). Recession (2009).
25 25
20 20
15 15
Percent
Percent
10 10
5 5
0 0
31
11
33
34
32
13
28
08
09
12
10
07
26
06
30
29
27
19
20
19
19
19
20
19
20
20
20
20
20
19
20
19
19
19
(a) Then (b) Now
Figure 15.2(a) Unemployment in the Great Figure 15.2(b) Unemployment in the Great
Depression (1930–1934). Recession (2009–2013).
Copyright © 2021. Harvard University Press. All rights reserved.
2009 and mid-2011. Its efects are still being debated. Did the ARRA stave of
a slump that would have rivaled the Great Depression? Or was it a colossal
waste? One node of the continuing dysfunction of U.S. politics is the polariza-
tion over this question.
In the case of the ARRA, there was in fact little additional spending by the
federal government on direct purchases of goods and services; most of the
stimulus took the form of transfer payments to individuals and tax reduc-
tions, both of which increased the disposable income of households and busi-
nesses. Tere were also sizeable transfers to state governments. Efectiveness
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
546 Raising Keynes
thus boils down to two questions: frst, to what extent did the recipients of
transfers and tax breaks actually spend the money they received, rather than
using it to shore up their fnances, and, second, to what extent was their
spending multiplied by a further chain of increased output, income, and
spending?
I take up the second question, the size of the multiplier, frst. Aferward, I
address the question of whether there was anything to multiply by examining
in some detail the most controversial part of the ARRA, the portion that went
to the states (which I will here call the S-ARRA), that is, the transfers over and
above the large amounts of fnancial aid routinely provided by the federal
government over the last forty years of so-called fscal federalism. Te S-
ARRA is of particular interest for two reasons. First, it comprises a substantial
portion of the overall stimulus—roughly $250 million if we include supple-
mental Medicaid grants. Second, the fact that there are only ffy states allows
us to explore the relevant behavioral and counterfactual conditions of the re-
cipients of S-ARRA in relatively fne-grained detail.
The American Recovery and Reinvestment Act:
A Brief Sketch
Te of ficial website of the stimulus program, www.recovery.gov, divided the
stimulus into three roughly equal parts, as in Table 15.1.1 Each of these cate-
gories includes a multitude of programs, of which the largest four in each
category are listed. Te largest single “Tax Beneft” program was an across-
the-board tax cut enjoyed by over 116 million taxpayers, Making Work Pay,
which provided a two-year $400 credit for an individual and an $800 credit
for a couple with two working spouses, phasing out only at relatively high
levels of $75,000 for an individual taxpayer and $150,000 for a couple fling
joint returns. Te total beneft amounted to $104.4 billion. At the other end
of the spectrum were adjustments to the Alternative Minimum Tax Exemp-
Copyright © 2021. Harvard University Press. All rights reserved.
tion totaling over $65 billion, which benefted only thirteen million taxpayers.
Tax Benefts also included some $25.5 billion of tax breaks for students and
$33 billion of tax breaks for businesses, including an adjustment to deprecia-
tion allowances totaling almost $24 billion.
“Entitlements,” the smallest amount of benefts of the three parts in the ta-
ble, was dominated by three programs: increases in the amount of Medicaid/
Medicare covered by the federal government ($88.8 billion, all but $2 billion
of which was Medicaid grants to states), extension of unemployment insur-
ance ($61 billion), and family services ($38.4 billion, of which the lion’s share
was food stamps). One-time payments of $250 to social-security recipients in
2009 added another $13.2 billion of entitlements.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 547
Table 15.1 Allocation of ARRA Funds through 2011
Total Benefts
($ billions)
Tax Benefts 299.8
Making Work Pay 104.4
Increased Alternative Minimum Tax Exemption 65.0
American Opportunity Tax Credit 25.5
Special Allowance for Certain Properties Acquired in 2009 23.9
Entitlements 220.2
Medicaid/Medicare 88.8
Unemployment Insurance Programs 61.0
Food Stamps 31.1
Social Security One-Time Payments 13.2
Contracts, Loans, and Grants 225.8
Education 89.6
Transportation 33.3
Infrastructure 25.6
Energy/Environment 23.2
Source: American Recovery and Reinvestment Act website (www.recovery.gov), accessed
February 25, 2012.
“Contracts, Loans, and Grants” consisted chiefy of grants to states and lo-
cal governments (the vast majority to states) for a variety of purposes. Educa-
tion accounted for $89.6 billion, and transportation for another $33.3 billion.
Infrastructure ($25.6 billion) included grants from the Environmental Protec-
tion Agency to the states as well as supplements to the budgets of a variety of
federal agencies, from the General Services Administration, responsible for
the upkeep of federal buildings, to the Army Corps of Engineers, responsible
for food control on navigable rivers. Energy/Environment was also a hodge-
podge, including $1.5 billion awarded to Savannah River Nuclear Solutions to
clean up the Savannah River Site, where production for the military’s nuclear
Copyright © 2021. Harvard University Press. All rights reserved.
arsenal took place during the Cold War, and a half-billion dollar loan guaran-
tee for Solyndra, the solar-panel manufacturer that ARRA opponents love to
hate—it went bust in 2011 despite the support of the federal government.
Unlike politicians, leading economists generally believed that the Obama
stimulus succeeded in its stated goal of job creation. Te University of Chica-
go’s Booth School of Business periodically surveys a panel of prominent econ-
omists on policy questions. In early 2012, an overwhelming majority, more
than 90 percent, concurred with the view that the ARRA had added jobs
(ChicagoBooth 2012).2
John Taylor and John Cogan were at the forefront of the minority who be-
lieved that the ARRA was inefective (Taylor 2011; Cogan and Taylor 2012).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
548 Raising Keynes
In their view, the S-ARRA, like the rest of the ARRA, failed because state
governments, like other agents, smooth expenditures over time à la Milton
Friedman and Franco Modigliani (see chapter 9). As a result, any temporary
fscal stimulus will be largely, if not completely, saved rather than spent. In the
Taylor–Cogan view, the stimulus did nothing more than provide debt relief—
by substituting the debt of the United States for the debt of households, busi-
nesses, and state governments.
Chapter 9 argued that the premises and conclusions of the hypothesis of
permanent-income/life-cycle expenditure smoothing are at best dubious for
a large percentage of households. Tese are households that are liquidity-
constrained or (for good reason) unable or unwilling to follow the main-
stream logic of utility maximization that underpins the expenditure-
smoothing argument.
Expenditure smoothing, rational or otherwise, is even less plausible for
states. Te second half of this chapter supports this contention with three
separate investigations of the impact of S-ARRA. My conclusion is that states,
like most households, are unable and/or unwilling to smooth expenditures in
the face of fuctuating revenues.
To begin we focus on a more general question about the stimulus: how
much bang was there for each buck? In the language of economics, how large
was the multiplier?
The Multiplier in Theory
Te multiplier is part and parcel of Keynes’s General Teory, but it actually
antedates Te General Teory by several years. In a pamphlet co-authored
with H. D. Henderson to support the Liberal Party’s election campaign in
1929 (summarized in Keynes 1931a), Keynes argued that ripple efects from
government spending would enhance the impact of the original outlay on the
economy. In this connection he assigned his student, Richard Kahn, the task
Copyright © 2021. Harvard University Press. All rights reserved.
of quantifying these ripple efects. How much additional spending and in-
come could be expected from an initial expenditure of one pound? What was
the multiplier?
Te basic idea, as we saw in chapter 3, is that a new purchase not only calls
forth an immediate addition to production but also is an immediate increase
in income for the producer, and therefore leads to a subsequent increase in
his purchases. Tese purchases in turn represent new income for other pro-
ducers, and new spending on their part. In principle, the chain continues
indefnitely.
Ten why isn’t the multiplier infnite? Kahn’s insight (1931) was that though
the number of rounds might be infnite, each round of spending would be
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 549
smaller because some of the income would “leak” into saving and imports,
not to mention taxes. So from ₤1 of government spending, the workers, con-
tractors, and other direct recipients of income might spend only half a pound,
50p (or 10 shillings as Kahn would have said), on domestically produced
goods and services, creating only 50p of additional income. If in turn the re-
cipients of this 50p also spend only half, the next round of spending will pro-
duce only 25p of new output and income.
In this analysis, the crucial determinant of the ripple efects is the propor-
tion of new income that the typical individual spends on domestically pro-
duced goods and services—in Keynes’ vocabulary, her marginal propensity to
consume (MPC). Te ultimate efect on output (∆Y) of an initial impulse—
say an increase in government spending (∆G)—is
∆Y = ∆G + MPC ∆G + MPC 2 ∆G + MPC 3 ∆G +
1
(
= ∆G 1 + MPC + MPC 2 + MPC 3 + = ) 1 − MPC
∆G,
and the multiplier is the ratio of the overall increase in output and income to
the original injection of spending
∆Y 1
= .
∆G 1 − MPC
A reduction in taxes or an increase in transfer payments (social security,
food stamps, unemployment insurance) also has a multiplier efect, but the
tax multiplier is generally less than the spending multiplier because, as was
observed in chapter 14, the frst round does not involve the production of any
goods or services. If taxes are cut by ∆T, and the recipients of the tax cut
spend the same fraction of income as the average for the economy as a whole,
then the direct spending is MPC × ∆T. Tis creates an equal amount of in-
come and new spending of MPC × MPC × ∆T. Taking all rounds of spending
together we have
Copyright © 2021. Harvard University Press. All rights reserved.
∆Y = MPC ∆T + MPC 2 ∆T + MPC 3 ∆T +
MPC
(
= MPC ∆T 1 + MPC + MPC 2 + MPC 3 + = ) 1 − MPC
∆T
so that
∆Y MPC
= .
∆T 1 − MPC
Simple models, with saving as the only leakage from successive rounds of
spending, create the impression of very large multipliers. Saving equal to one-
tenth of income would lead to a spending multiplier of ten and a tax multi-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
550 Raising Keynes
plier of nine. Even if households were to save one-quarter of their incomes,
the two multipliers would be four and three.
Observe that in both cases, the diference between the spending multi-
plier and the tax multiplier is one. Under the simplifying assumption of a
uniform MPC, this result holds whatever the MPC might be. Tis is the so-
called “balanced-budget multiplier” theorem, of which chapter 14 made ex-
tensive use.
Te world is obviously a more complicated place than the stripped-down
expositional models of this book, or for that matter the models of Te General
Teory. Not only are there leakages into saving, economic agents typically pay
taxes, and the multiplier chain is also broken when income is spent on im-
ports. (Imports add to domestic product only to the extent of processing and
distribution expenditures within the importing country’s national boundar-
ies.) For these reasons empirical estimates of the multiplier are typically much
lower than the multipliers suggested by a focus on saving alone. At two places
in Te General Teory (pp. 121–122 and pp. 127–128), Keynes estimates the
multiplier as lying between 2 and 3 for a moderately open economy.
An example might help here. Imagine the chain of money income gener-
ated by a hypothetical injection of €1.00 of new spending that leads to the
production and sale of €1.00 of new output. Assume that one-third of the sale
proceeds end up as the return to capital (gross proft), and that, in the short
run at least, none of capital’s share is spent. (Investment plans have already
been made, and extra proft ends up as retained earnings or is paid as taxes.)
Tis leaves €0.67 to be distributed as wages. Assume that one-quarter of the
wage share is paid as taxes, and three-quarters is disposable income. Tus of
the original expenditure of €1.00, €0.50 ends up as the disposable income of
households. Now assume that one-tenth of this disposable income is saved,
and nine-tenths is consumed, that is, €0.45 spent on consumption. Of this,
assume that three-quarters goes to domestic producers, and one-quarter goes
for imports. So of the original €1.00 outlay, only 3/4 × €0.45, or €0.3375, is
Copyright © 2021. Harvard University Press. All rights reserved.
actually recycled into the economy. In this case the multiplier is
1 1
= = 1.51,
1 − MPC 1 − 0.3375
a far cry from the multiplier values implicit in the expositional models even
though saving out of disposable income is one-tenth. Te ratio of personal
saving to disposable income turns out to exert relatively little infuence on the
multiplier compared with the division of income between labor and capital,
tax rates, and the openness of the economy to imports.
Tere are two assumptions that require special scrutiny in evaluating ex-
penditure and tax multipliers. Te frst deals with how much economic activ-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 551
ity is displaced by government spending, the second with diferences be-
tween the spending propensity of direct and indirect benefciaries of stimulus.
Mainstream theory, as we shall see in chapter 17, argues that crowding out
takes place because government borrowing competes with investment de-
mand for the available supply of saving, driving up interest rates and reducing
the availability of credit. In Keynes’s logic, there is no impact of government
spending on investment via the interest rate or the credit channel; the pres-
sure on resources is refected in higher prices. In each case, under both main-
stream and Keynesian reasoning, output would rise less than 1/(1 − MPC).
We can refect crowding out in the multiplier formula by adjusting the
multiplicand. If CO represents the fraction of new spending that displaces
existing production and m = 1 − CO, initial spending of €1.00 now gen
erates only €m of new output. Assuming subsequent rounds of spending
are subject to the same degree of crowding out, the second round generates
€m × (m × MPC), the third round €m × (m × MPC) × (m × MPC), and so on.
Te multiplier sum becomes
∆Y
∆G
(
= m 1 + mMPC + m2 MPC 2 + m3 MPC 3 + , )
and the multiplier is
m
Spending Multiplier with Crowding Out = .
1 − mMPC
Observe that m appears both in the numerator and in the denominator if, as
we assume, crowding out is assumed to afect every round of spending equally.
Te same logic applies to taxation and transfers. Recipients of tax breaks or
enhanced entitlements may spend the increase in disposable income, but
their spending may simply crowd out other spending. Te tax multiplier be
comes
mMPC
Copyright © 2021. Harvard University Press. All rights reserved.
Tax Multiplier with Crowding Out = .
1 − mMPC
Te balanced-budget multiplier, the diference between spending and tax
multipliers, now becomes less than one, approaching this upper limit only as
crowding out goes to zero.
Te logic of the tax multiplier breaks down if the spending of direct benef-
ciaries of the stimulus does not mirror the average spending pattern in the
economy. Although we can justify using an economy-wide average for the
second, third, and subsequent rounds of spending—there is no way of tracing
out the expenditure of each income recipient in the chain—we can surely do
better in measuring the impact on spending in the frst round. Te direct ben-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
552 Raising Keynes
efciaries of a program to (say) rebuild the transportation infrastructure may
include a large number of workers whose spending habits are markedly difer-
ent from those of the general population. On the tax side, diferent kinds of
tax cuts will produce very diferent responses from the benefciaries. We now
have the spending multiplier as
1 + m(v − MPC )
Spending Multiplier with Crowding Out = m ,
1 − mMPC
where v is the frst-round MPC, the fraction spent by the direct benefciaries
of government spending. Te same logic applies to taxes and transfers. If v is
the fraction spent by the direct benefciaries of tax reductions, transfers, or
grants, the tax multiplier becomes
mv
Specific Tax Multiplier with Crowding Out = .
1 − mMPC
We can think of v as a valve controlling the fow of the initial stimulus into
the economy. As might be expected, specifc spending and tax multipliers are
higher than their generic counterparts if v > MPC, and lower if v < MPC. At
the lower limit, with v = 0, the valve is shut completely. Te multiplier is zero,
so spending increases income by m, and tax cuts do not increase income at all
regardless of the size of the MPC.
Taylor and Cogan (Taylor 2011; Cogan and Taylor 2012) argued that v = 0
because agents follow the dictates of the mainstream theory of consumer
choice and rationally smooth expenditures.3 Tey reason that benefciaries of
tax rebates and transfers, recognizing the temporary nature of federal lar-
gesse, would treat the increase in disposable income as a one-time addition to
their assets, to be doled out in little bits over the long-term future.
As chapter 9 observed, a rational individual will optimally insulate spend-
ing from fuctuations in income by lending in times of plenty and borrowing
in times of dearth. Despite an array of refnements that have been added to
Copyright © 2021. Harvard University Press. All rights reserved.
the original work of Milton Friedman on the permanent-income hypothesis
and by Franco Modigliani and associates on the life-cycle hypothesis, the es-
sential kernel of both theories remains intact: rational agents smooth expen-
ditures when income fuctuates.
A new wrinkle is that the same logic that applies to households also applies
to state and local governments.4 Cogan and Taylor note Edward Gramlich’s
pioneering work on the efects of federal grants on state budgets. Gramlich
ends up skeptical of the efficacy of trying to stimulate the economy through
grants to states, arguing, as do Cogan and Taylor a generation later, that
grants end up fortifying state balance sheets (Gramlich 1978, 1979).
Before Gramlich, the terrain of how government spending is determined
had been lef mostly to students of politics. As early as the 1960s, Aaron Wil-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 553
davsky argued the position that would later inform Gramlich’s work: last
year’s expenditures are the primary determinant of this year’s expenditures.
An important diference between Wildavsky, on the one hand, and Gramlich,
Taylor, and other economists who invoke expenditure smoothing, on the
other, is that Wildavsky claimed no rational basis—on the contrary—for the
workings of the budgetary process. (See Wildavsky 1964; Davis, Dempster,
and Wildavsky 1966, 1974.) Nor did he apply his arguments to the operation
of state and local government. His focus was rather on the process that deter-
mined agency budgets within the federal government. A crucial diference
between the federal government and state and local governments is that no
balanced-budget constraint operates at the federal level.
It was observed in chapter 9 that expenditure smoothing can take place for
reasons that have nothing to do with the supposed rationality of basing ex-
penditure on long-term resources rather than on current income. Tis idea,
labeled the disequilibrium hypothesis in chapter 9, is like Wildavsky’s theory
in positing that agents are creatures of habit and inertia, which implies that
adjustment of spending to income takes place only gradually. Applied to indi-
viduals, the disequilibrium hypothesis in its pure form is
∆C = (Y − C−1),
where is the speed of adjustment. We can combine the idea of disequilib-
rium with the more conventional Keynesian argument of consumption as a
constant fraction of income, C = Y, where is the marginal (and average)
propensity to consume. In terms of changes,
g
∆C = ∆Y = Y
1+ g
with
∆Y = gY−1.
Copyright © 2021. Harvard University Press. All rights reserved.
Adding the two expressions for ∆C, we have
g g
∆C = θ ( Y − C −1 ) + ζ Y = θ + ζ Y − θC −1.
1+ g 1+ g
As was also noted in chapter 9, this formulation can be made equivalent
observationally to the permanent-income version of consumption smooth-
ing. For a follower of Friedman,
C = Y P
and
∆C = ∆Y P,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
554 Raising Keynes
where Y P represents permanent income, and the propensity to consume out
of permanent income. Permanent income is not directly observable, and we
have some latitude in choosing the particular formula that we impute to
agents for revising their notions of permanent income. For the sake of obser-
vational equivalence, assume a recursive formula that gives weight both to the
current gap between actual income and last year’s notion of permanent in-
come, Y − Y−1P , and to the trend rate of growth of permanent income, :
(
∆Y P = η Y − Y−P1 + φY−P1. )
Together with the two equations above, this adjustment formula gives
∆C = Y + ( − )C−1.
In this case the permanent-income hypothesis and the disequilibrium hy-
pothesis lead to the same equation, one which says that the change in con-
sumption will depend on two variables, current income and lagged consump-
tion. And both formulations have two parameters, and , which can be
estimated from the data.5
While the two formulations are econometrically identical, the interpreta-
tion of the parameters is very diferent, unsurprisingly in view of the very
diferent premises of the two hypotheses. In the case of disequilibrium, the
observed coef ficient of lagged consumption is an estimate of the speed of ad-
justment to a gap between current income and last year’s consumption. In
the case of permanent income, the same coef ficient − is an estimate of the
diference between the growth rate of permanent income and the rapidity
with which the revision of permanent income takes place in response to a gap
between actual income and lagged permanent income. By the same token, the
interpretation of the coef ficient of current income difers: in the one case, it is
the sum of the adjustment-coef ficient and the product (g)/(1 + g); in the
other case, it is the product .
Whichever interpretation one might prefer, there are good reasons to be
Copyright © 2021. Harvard University Press. All rights reserved.
skeptical about expenditure smoothing as a general assumption. First, many
agents are simply unable to engage in expenditure smoothing, at least when
incomes are falling. Tey have little or no saving and equally little access to
credit markets. Chapter 9 noted the large literature on what are called in the
jargon liquidity-constrained households. An equivalent liquidity constraint
exists for the states to the extent that they are in fact bound by law or custom
to cover current expenditures without borrowing.
Rational expenditure smoothing is especially suspicious. Te economist’s
notion of rationality makes stringent demands on individuals with respect
both to their intertemporal utility functions and to their intertemporal bud-
get constraints. As observed in chapter 9, most people simply do not know
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 555
enough about their future needs and wants, much less about their future in
comes, for the standard theory of consumer choice to make sense. Instead,
people fall back on habit, rules of thumb, and other perhaps less elegant but
more realistic ways of coping. Moreover, real-world rationality may suggest a
higher premium on solidarity and sharing than the economist’s paradigm of
individual choice allows. A poor person embedded in community may feel
that sharing a tax rebate with her less fortunate neighbors, particularly the
neighbor faced with eviction if the rent goes unpaid or a blackout if paying
the electricity bill is put of, is a higher priority than replenishing her own
bank account. She knows that someday it will be her turn to rely on the com-
munity (Stack 1975, quoted in Marglin 2008, p. 23). For a large segment of the
population it seems more plausible to attribute consumption smoothing to
disequilibrium adjustment than to a carefully thought out plan to balance
long-term expenditure and long-term income.
It remains to be seen how much of the consumption-smoothing reasoning,
of either permanent-income or disequilibrium inspiration, carries over to the
states. Tis question goes beyond whether expenditure smoothing is a plau-
sible behavioral assumption for private agents. Tere is also the question of
the extent to which the constraints state governments face difer from those
faced by households. One obvious diference is that as a practical matter we
can take permanent income to be beyond the control of the household (even
though in theory permanent income is not given—it is, rather, the outcome of
a maximization process that trades of goods against leisure). But for states,
permanent income is clearly endogenous. States, unlike households, have
the power to levy taxes, so their maximization problem is more complicated
than one of optimizing expenditures relative to an exogenously given income
stream.
Multipliers for the Obama Stimulus
Copyright © 2021. Harvard University Press. All rights reserved.
Teory and hypothetical examples, however, can take us only so far in esti-
mating multipliers. Ideally, we would be able to settle disagreements over the
appropriate value—actually values, see below—by appealing to data. Unfortu-
nately, there are signifcant challenges to doing so. Te greatest challenge is to
isolate the efect of the stimulus from other macroeconomic activity—a stan-
dard dif ficulty of econometric analysis, but one that is made particularly acute
in the case of estimating stimulus multipliers due to the paucity of relevant
data. Not only are fscal stimulus programs relatively rare, they are also idio-
syncratic, responses to diferent kinds of economic stress. Tis turns a rela-
tively small number of historical stimulus programs into an even smaller pool
of diferent programs deployed under diferent circumstances (see note 4).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
556 Raising Keynes
Concretely, this means that when we are prospectively assessing the wisdom
of a particular stimulus program we cannot draw on a large sample of similar
past episodes as a guide to its likely impact.
Notwithstanding, economists have deployed a variety of techniques to esti-
mate multipliers. Valerie Ramey’s (2011) review of the literature reports esti-
mates from eighteen studies. She concludes:
Te range of plausible estimates for the multiplier in the case of a temporary
increase in government spending that is defcit fnanced is probably 0.8 to
1.5 . . . If the increase is undertaken during a severe recession, the estimates
are likely to be at the upper bound of this range.
To be sure, Ramey qualifes her conclusion by noting that
there is signifcant uncertainty involved in these estimates. Reasonable peo-
ple could argue that the multiplier is 0.5 or 2.0 without being contradicted
by the data. (pp. 680–681)
Te range of these estimates should not surprise us, especially if we take
into account the problems attendant to the various estimation procedures.
Peter Spiegler (Marglin and Spiegler 2013) has grouped these techniques
under three rubrics: structural vector auto-regression (SVAR), dynamic sto-
chastic general equilibrium (DSGE), and large-scale macroeconomic models
(LSM).
Vector auto-regressions have been an important tool since the 1980s, be-
ginning with a paper by Nobel Laureate Christopher Sims (1980) that ofered
this technique as an alternative to what Sims saw as a hodgepodge of ad hoc
and sometimes inconsistent restrictions imposed on separate equations of a
complex system. Vector auto-regressions were proposed as a way of letting
the data speak for themselves by imposing only a minimal structure on how
the variables interact. By its very nature, this virtue comes at the cost of ambi-
guity about causal relationships. SVAR models address this ambiguity by
Copyright © 2021. Harvard University Press. All rights reserved.
making assumptions about which variables are exogenous and therefore driv-
ers of the system.
Tere remain problems. First, as a practical technique, SVAR requires us to
limit the number of variables and is therefore not well suited to analyzing the
impact of stimulus on diferent categories of benefciaries. Second, it requires
us to treat distinct episodes as sharing a common framework and therefore
to suppress the idiosyncratic nature of stimulus; as Jonathan Parker (2011,
p. 709) observed, “By assumption, the impulse response to an unexpected or
exogenous increase in government spending is constrained to be the same in
dependent of the state of the business cycle.”
DSGE models go to the opposite extreme; they impose so much structure
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 557
that multiplier estimates from these models make sense only to the extent that
you buy into the model. DSGE models derive from the New Keynesian prem
ise that a perfectly competitive economy will lead to a (Pareto) ef ficient allo-
cation of resources. Recessions, unemployment, the space for stimulus are
accounted for by grafing one or more imperfections on the textbook com-
petitive model.
A relevant piece of the structure assumed in DSGE models is the reliance
on a representative agent’s consumption decisions based on intertemporal
optimization à la Friedman and Modigliani. If this is the appropriate charac-
terization, then the estimate of the multiplier resulting from an injection of
stimulus refects permanent-income/life-cycle consumption smoothing. Oth-
erwise, the impact of stimulus is simply an unexplained correlation; it is rea-
soning in a circle to argue from parameter estimates to model validation. Tis
is of course just one example of a general problem that Sims and others have
noted: in Sims’s words (2007, p. 153), “Making forecasts, policy projections,
and (especially) welfare evaluations of policies with these models as if their
behavioral interpretation were exactly correct is a mistake.”
LSM models are the third way of estimating multipliers. Tese models have
the virtue of nuance that both SVAR and DSGE models lack. Tat is, one can
in principle estimate a variety of multipliers from the thousands of equations
that these models typically mobilize to describe the economy. Both the federal
government and private economists (for example, Alan Blinder and Marc
Zandi [2010]) deployed LSM models to estimate the efects of the Obama
stimulus. Tese models however do not escape the need for assumptions
about the structure of the economy, and in their case, extremely detailed as-
sumptions. Te multiplier estimates that emerge are only as good as the as-
sumptions are accurate.
A major problem with evaluating the usefulness of LSM models is that
the same models were used ex post to evaluate the impact of the stimulus,
thus the importance of independent analysis of the kind provided by Gabriel
Copyright © 2021. Harvard University Press. All rights reserved.
Chodorow-Reich et al. (2012), Daniel Wilson (2012), and others—including
the present chapter.
One takeaway is that the multiplier is highly variable: diferent strokes
not only for diferent folks, but for diferent types of stimulus and at difer-
ent points in the economic cycle. A stimulus program targeted toward low-
income people will give very diferent results from one directed toward upper-
income groups, and the same target groups will react diferently when the
economy is in recession than in more prosperous times. A generic value of
“the multiplier” that covers all situations at all times may be a legitimate sim
plifcation when introducing the idea for the frst time but cannot be the stuf
of policymaking.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
558 Raising Keynes
Tis point is well understood by practitioners. In evaluating the Obama
stimulus, the Congressional Budget Of fi ce, for example, used a variety of mul-
tipliers, and indeed a range of values rather than a single point estimate for
each. Te ranges varied from 0 to 0.4 for certain corporate tax breaks and 0.1
to 0.6 for tax cuts for high-income people, to 0.3 to 1.5 for tax cuts for middle-
income folks and 0.4 to 2.1 for transfer payments such as food stamps and
unemployment compensation. For payments to states to supplement educa-
tion and Medicaid budgets, the CBO multiplier ranged from 0.4 to 2.1 (Con
gressional Budget Of fi ce 2012, table 2, pp. 6–7).6
Te diferences between the multiplier estimates for diferent elements of
the stimulus does raise an obvious question: if the point was to add demand
to a weakening private sector, and thereby to restore prerecession levels
of employment and output, why would stimulus money take the form of
tax breaks directed to corporations and high-income individuals? A much
greater bang for the buck was available from tax cuts for middle-income
people and transfers to the poor and unemployed, not to mention transfers to
the states.
Te likely answer is that tax breaks for the rich and tax breaks for corpora-
tions were never intended to stimulate. Perhaps as much as one-fourth of the
total was the political price of stimulus, not stimulus itself. Tax breaks for the
wealthy represent the price of getting action from a Congress dominated by
special interests.7
Consider again the three-part division of the ARRA into Tax Benefts; En
titlements; and Contracts, Loans, and Grants in Table 15.1. A rough-and-
ready division of Tax Benefts according to the specifcs of the various pro-
grams makes it possible to identify eight programs that appear to beneft
better-of segments of the population, listed in Table 15.2, along with the
number of benefciaries and the aggregate benefts of each. Table 15.2 also
includes the two programs that beneft business, the last two items, which
amounted to $43 billion. Te total “stimulus” for the better-of segments of
Copyright © 2021. Harvard University Press. All rights reserved.
the economy totaled over $175 billion!
We suppose that the rest of the money included under Tax Benefts in Table
15.1, more than $120 billion, went to taxpayers who would not have operated
by Friedman–Modigliani logic to the extent that those higher up the income
distribution did. Tese programs included Making Work Pay, the $100+ bil-
lion program that was the centerpiece of the ARRA tax breaks for individuals;
the Earned Income Credit ($5.1 billion); and the exclusion of a portion of
unemployment benefts from taxable income ($6.3 billion).
It is undoubtedly an exaggeration to assume that not a penny of the $175+
billion benefting the better-of was spent in the frst round (v = 0), and to as-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 559
Table 15.2 Tax Beneft Programs Chiefy Benefting Better-Of Taxpayers
Number of Taxpayers
Benefting Benefts
(millions) ($ billions)
Child Tax Credit 15.6 18.4
American Opportunity Tax Credit 10.2 25.5
First-Time Homebuyer Tax Credit 1.0 10.4
Extension of Alternative Minimum Tax Relief for 13.0 4.4
Nonrefundable Personal Credits
Increased Alternative Minimum Tax Exemption 13.0 64.7
Residential Energy Credit 4.3 11.0
Business Incentives 34.5
Manufacturing and Economic Recovery 9.2
Total 178.1
Source: American Recovery and Reinvestment Act website (www.recovery.gov); data now stored at
data.nber.org/data/ARRA.
sume that the entire addition to disposable income was spent (v = 1) when tax
benefts fowed to the less well-of. But I can defend this generalization as
much closer to the mark than assuming a uniform frst-round MPC.
If we also assume that entitlements fow to the less well-of, and so are en-
tirely spent, we are a long way toward assessing the impact of the ARRA on
GDP. With the assumption that there is no crowding out, so that m = 1, the
multiplier for tax benefts and individual entitlements depends only on the
division between rich and poor, plus what is assumed about the average MPC
in the economy. Here we assume MPC = 1/3, so that the tax/transfer multi-
plier appropriate to tax benefts and entitlements is
v
= 1.5v,
1 − MPC
Copyright © 2021. Harvard University Press. All rights reserved.
in line with the upper end of Ramey’s plausible range and in line with the
CBO estimates as well.
Table 15.3 summarizes a provisional reckoning. Observe that Table 15.3
shifs the $86.9 billion of Medicaid enhancements from the Entitlements row
in Table 15.1 to the Contracts, Loans, and Grants row. Te reason is that these
grants were in fact grants to the states, not direct grants to individuals. As
such they were fungible, and it is more reasonable to lump them together with
other grants than to treat them like, say, food stamps or unemployment insur-
ance, which were grants to individuals even though they fowed through the
states. Although the states were bound by the terms of Medicaid grants to
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
560 Raising Keynes
Table 15.3 ARRA Spending and Impact, by Spending Category ($ billions)
(1) (2) (3) (4) (5) (6)
Direct
Reallocating First-Round Spending of Total
Reported Medicaid to MPC Recipients Impact
Spending Grants =v = (3) × (4) Multiplier = (4) × (5)
Tax Benefts, upper- 178.1 178.1 0.0 0.0 0.0 0.0
income recipients
and business
Tax Benefts, lower- 121.7 121.7 1.0 121.7 1.5 182.6
income groups
Entitlements 220.2 133.3 1.0 133.3 1.5 200.0
Contracts, Loans, and 225.8 312.7 ? ? ? ?
Grants
Total 745.8 745.8 ?
Source: Column (1) from Table 15.2; remaining data, author’s calculations.
maintain the programs in place before the fnancial crisis and ensuing Great
Recession, the states did not have to come up with their own funds to meet
the claims of a much enlarged program.8
We turn now to substituting numbers for the question marks in the bottom
row of Table 15.3: how much of the money spent by the federal government
under the heading of contracts, loans, and grants actually led to the purchase
of additional goods and services?
What Did the States Do With the Money?
Indeed, the focus is a little narrower: I concentrate on what happened to the
money the ffy states received. Tis limited focus makes sense because the
data for the states is more tractable than for other entities (local governments
and regional authorities, private businesses and nonprofts); moreover, ac-
Copyright © 2021. Harvard University Press. All rights reserved.
cording to my reckoning for the fscal-year 2010, FY-2010, the year on which
the cross-section analysis focuses, states received some 80 percent of the
money under this category.
Tere are two issues. Te frst is the value of the MPC for all the spending
afer the frst round. We shall continue to assume MPC = 1/3, and hence a
multiplier of 1.5 for state outlays—even though the estimates implied by the
cross-sectional analysis later in this chapter are somewhat higher. Te second
issue, on which the argument here focuses, is the frst-round MPC, the v in
the multiplier formulas developed earlier in this chapter. I ofer three argu-
ments why it is reasonable to suppose that the states spent all the money they
received under S-ARRA programs, that is, v = 1.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 561
Te frst argument examines the time-series evidence that John Cogan and
John Taylor (2012) put forward in support of the hypothesis that state govern-
ments base their spending not on current revenues but on long-term re-
sources, with the implication that for state governments v = 0. Far from ofer-
ing independent support, the time-series evidence argues for expenditure
smoothing only if one is already committed to that view.
A second line of inquiry examines the diferential impact of the S-ARRA
money on spending across the states. In conducting this exercise we are in ef-
fect starting from an agnostic position on whether the states could have found
other funds in order to continue to spend as usual, and so might simply have
substituted S-ARRA funds for drawing down their bank accounts or borrow-
ing. Tis cross-sectional analysis leads to the conclusion that the S-ARRA
money that went to the states was indeed spent.
Te third approach is to see what the boots on the ground, state budget of
fcers, have to say about the hypothesis that, absent the S-ARRA, states could
have sustained prerecession levels of spending. Tese interviews lead to a
clear conclusion: with very few exceptions, state budget of ficers say that their
states could not have avoided spending cuts in the absence of the S-ARRA.
What Do Time-Series Regressions Say?
Cogan and Taylor (2012) use aggregate time series to investigate the impact of
grants to states, complementing Taylor’s earlier article (2011) that estimates
the impact of the portions of the ARRA that involved tax reduction and trans-
fers to individuals. All of this work fnds that the stimulus had no impact on
spending.
Cogan and Taylor’s analysis is straightforward. Te states had two options
with respect to how they responded to S-ARRA grants. A state could have
added to its expenditures or shored up its balance sheet. Te funds, it is as-
sumed, were fungible. Teir point is that enhanced Medicaid reimburse-
Copyright © 2021. Harvard University Press. All rights reserved.
ment or education grants would not—and did not in fact—increase Medicaid
spending or school spending if a state was planning to allocate its own money
to these ends. Te extra Medicaid or education dollars would then be allo-
cated to other expenditures or to building up depleted assets or to reducing
borrowing. (One possibility Cogan and Taylor don’t seem to consider is the
possibility of reducing taxes, or more realistically, raising taxes less than a
state otherwise would have.9 As we shall see, their intuition is sound here:
though this is a possibility a priori, there is no support in the data for this op-
tion.)
Cogan and Taylor estimate the impact of the S-ARRA by regressing aggre-
gate purchases of goods and services (G) and transfers (E) by states on non-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
562 Raising Keynes
ARRA revenues (R) and ARRA grants (A), along with lagged values of the
dependent variables:
G = a0 + a1G−1 + a2R + a3A = 3.66 + 0.876G−1 + 0.113R − 0.097A,
E = b0 + b1E−1 + b2R + b3A = −6.1 + 0.743E−1 + 0.056R + 0.163A.
From these two equations, Cogan and Taylor estimate aggregate saving—the
addition to fnancial assets or the reduction in liabilities—on the part of the
states (L) as the diference between revenues (R + A) and outlays (G + E):
L = −(a0 + b0) − a1G−1 − b1E−1 + (1 − a2 − b2)R + (1 − a3 − b3) A,
L = 2.442 − 0.876G−1 − 0.743E−1 + 0.831R + 0.933A.
I had questions about how Cogan and Taylor handled the data for these
regressions, and my answers to these questions led to certain modifcations,
the most important of which was to collapse the distinction between transfer
payments and purchases of goods and services.10 Afer all these modif
cations,
replacing G and E by the sum O = G + E gives the result
O = a0 + a1O−1 + a2R + a3A = − 0.001 + 0.788O−1 + 0.233R + 0.049A.
(0.010) (0.071) (0.109) (0.168)
Tis is to say that one dollar of S-ARRA funds only increased state-government
outlays by a nickel, virtually identical to the Cogan–Taylor result that all but
seven cents went to shore up state balance sheets. Newey-West standard
errors (in parentheses below the coef ficients) indicate that the coef
ficients
on lagged outlays and revenues are signifcant at usual confdence intervals
and that the coef ficient on S-ARRA spending is statistically indistinguishable
from zero.
Te heavy lifing in all the estimating equations is done by the lagged de
pendent variables, a result certainly consistent with expenditure smoothing.
But two qualifcations are in order. First, as has been observed, expenditure
Copyright © 2021. Harvard University Press. All rights reserved.
smoothing may or may not mean that spending decisions are driven by long-
term considerations in the manner of the permanent-income and life-cycle
hypotheses. Te coef ficient on lagged outlays is consistent not only with long-
term budgeting that smooths over revenue fuctuations, but also with the idea
that lags occur because old habits die hard, and habit may be reinforced by
institutional rigidities in the adjustment of spending to revenue.
To see this, look at the above equation rerun without the S-ARRA term and
with the change in outlays as the dependent variable. We obtain
∆O = a0 + a1O−1 + a2R = − 0.001 − 0.204O−1 + 0.224R.
(0.009) (0.048) (0.083)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 563
As noted in the section titled Te Multiplier in Teory, the disequilibrium
interpretation of the coef
ficients is
ˆ 0.204, ˆ 0.294 ,
and the permanent-income interpretation is
ˆ 0.276 .
ˆ 0.81,
Tese estimates assume that the rate of growth of nominal “permanent reve-
nue” is equal to the trend rate of growth of nominal revenues over the sample
period,
= g = 0.072.
Te constant term apart, the interpretation of the coef ficients under the
disequilibrium hypothesis is that outlays change by a bit over 20 percent of
the gap between revenues and lagged outlays, with an additional immediate
adjustment of almost 30 percent. Under the permanent-income hypothesis,
the interpretation is that permanent income (revenues), Y P, increases 7 per-
cent every year and is further adjusted by just over one-fourth of the difer-
ence (Y − Y−1P )/Y−1P , that is, 0.276 times the percentage gap between actual in-
come and the lagged value of permanent income; the change in outlays is just
over 80 percent of the change in permanent income.
A second, more important, qualifcation is the possibility that the coef fi
cient of lagged outlay plays an outsize role in the regression because of serial
correlation. Serial correlation in the data can generate the observed result that
lagged outlay matters a lot even in the case where lagged outlay is actually ir-
relevant to current outlay. Suppose that in fact—a messenger from on high
told me so—it is the other variables (ordinary revenues and S-ARRA reve-
nues) that are driving expenditures. Nonetheless, a lagged dependent variable
will still show up with a positive coef ficient and bias the estimates of the true
Copyright © 2021. Harvard University Press. All rights reserved.
drivers downward, provided that in the correct specifcation (the one that the
messenger vouched for) the independent variables and the error term are se-
rially correlated (Achen 2001).
I analyze the data for serial correlation in the statistical appendix to this
chapter. Te results are consistent with the hypothesis that only revenues mat-
ter for expenditures, and the regression coef ficients that emerge from the
specifcation including lagged outlays are spurious. If we exclude lagged out-
lays, we obtain the following result:
O = a0 + a1R + a2 A = − 0.008 + 1.094R + 1.630A.
(0.012) (0.095) (0.213)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
564 Raising Keynes
In this specifcation, the S-ARRA has a huge impact on spending. For every
dollar of ARRA spending there is an additional $1.63 of outlay.
Te upshot is that a high t-value for the coef ficient of lagged expenditure
does not provide independent support for expenditure smoothing. But we
have to be careful in what we take away from this. Tat the high t-value on
lagged outlays could be due to serial correlation does not prove that it is due to
serial correlation. We cannot infer that there is no causal relationship running
from expenditure smoothing (via a signifcant coef ficient on lagged expendi-
ture) to present expenditure, that is, we cannot infer that last year’s expendi-
ture does not drive this year’s. Neither is there an implication that the alterna-
tive of basing present expenditure only on present revenues is the correct
approach. All we can fairly say is that the impressive statistics that character-
ize the Cogan–Taylor regressions, either in their original form or in my modi
fed form, add nothing to the argument for expenditure smoothing. Te an-
swer to the question of what efect the S-ARRA had on state expenditures is
built into the regression assumptions.
What Can We Learn from Cross-Sectional Regressions?
In view of the indeterminacy of time-series regressions, it makes sense to see
what is added by other approaches. In this section I deploy cross-sectional
evidence to test the hypothesis that states spent the bulk of S-ARRA monies
they received against the hypothesis that these monies had little or no efect
on spending. Tis exercise provisionally commits us to the stipulation that the
states had considerable latitude in this regard, that they could have main-
tained their actual expenditures if no ARRA monies had been forthcoming
and, accordingly, that the S-ARRA money might simply have shored up their
balance sheets.11
My conclusion is that even if states could have continued to spend without
the S-ARRA, these grants had a considerable impact on spending. Tis result
Copyright © 2021. Harvard University Press. All rights reserved.
supports the results of previous cross-sectional tests, including the pioneering
analyses by Chodorow-Reich et al. (2012) and Wilson (2012).
Before turning to the analysis, I need to say a few words about the data. In
contrast with the time-series analysis of Cogan and Taylor (2012), the data
here are restricted to state governments, leaving out the portion of S-ARRA
grants channeled directly to local governments and other agencies at one re-
move from state governments.
Tere are several reasons for focusing on state governments. One is that the
data are of better quality for the states than for the consolidated accounts of
state and local governments; state government data are assembled from a sur-
vey of state governments, whereas local government data is collected through
a sampling procedure and are therefore subject to sampling error. Another
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 565
reason for focusing on the states has already been noted: the bulk of the
ARRA monies in the S-ARRA—including both grants to government entities
and contracts and loans to nongovernment entities—in fact went to the states.
Substantial amounts of S-ARRA grants were in turn transferred by the states
to localities as well as to higher educational institutions and other nonprofts,
but for reasons exemplifed by Medicaid grants, I regard these transfers as es-
sentially equivalent to purchases of goods and services.12
For all the information that was available on the of ficial website of the
stimulus program, recovery.gov website, no breakdown of ARRA grants was
provided between states, localities, universities and other nonprofts, and pri-
vate businesses. For the portion of grants covered by the recipient-reporting
requirement (Section 1512 of the ARRA), I separated state grants out by using
a set of keywords like “department,” “education,” “executive of fi ce,” “human
services.” For the programs not subject to Section 1512 reporting, the largest
of which was Medicaid, I used the fgures of the relevant federal depart
ments. Because the quarterly listing of recipient reports lumped together dis-
bursements through September 30, 2009, I also relied on reports of the De-
partment of Education and the Department of Transportation to separate
grants received by the states during FY-2009 from grants received during FY-
2010.
Finally, a disclaimer. Since estimating the model, I have found errors in
processing the raw S-ARRA data. Tese errors are small and do not afect the
results, but it is possible there are errors that have gone undetected.13
Te model is designed both to account for the impact of the S-ARRA on
state expenditures, revenues, and borrowing, and to quantify the efects of
these intermediate variables on output. Te dif ficulties are obvious: all these
relationships involve reciprocal cause and efect. For example, if the S-ARRA
led to a decrease in taxes, this should in principle have stimulated the econ-
omy just as an increase in outlays does. At the same time, higher output will
generate higher tax revenues. Output and revenue are thus simultaneously
Copyright © 2021. Harvard University Press. All rights reserved.
determined. Te same story repeats itself with expenditure and output. As
others have noted (Chodorow-Reich et al. 2012; Wilson 2012; Leduc and Wil-
son 2017; Feyrer and Sacerdote 2011; Conley and Dupor 2013), S-ARRA
money may itself be infuenced by the variables it is supposed, directly or in-
directly, to explain. Tis leads to estimating parameter values by three-stage
least squares (3SLS).
Te endogenous variables are the change in outlays, that is, changes in state
purchases of goods and services directly or indirectly (through transfers that
are used for the purchase of goods and services), ∆O; non-ARRA revenues, R;
S-ARRA grants, A; and gross state product, Y. All variables are measured per
capita on the basis of state populations in April 2010.
Te general idea of these regressions is that if the S-ARRA had an impact,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
566 Raising Keynes
it should show up in greater expenditures by states receiving more ARRA
money and (or) by a reduction in taxation—both relative to baseline levels in
the absence of S-ARRA funds. Hence there are two separate equations in
which the change in expenditures and the change in non-ARRA revenues are
the dependent variables. To estimate the impact of state-government spend-
ing requires another equation in which the dependent variable is the change
in the state’s gross product.
Te basic cross-sectional model is
∆O = a0 + a1R + a2O−1 + a3A + a′1Z1 + + a′nZn,
∆Y = b0 + b1∆O + b′1Z1 + + b′nZn,
∆R = c0 + c1∆Y + c3A + b′1Z1 + + c′nZn,
A = d0 + d′1Z1 + + d′nZn,
where Z1, , Zn is a set of exogenous variables that include the instruments
for estimating S-ARRA grants and control variables relevant for determining
the changes in outlays, production, and revenues.
Without the controls, variation across states of changes in spending in rela-
tionship to variations in non-ARRA revenues, prior levels of expenditure, and
injections of S-ARRA money appear to confrm the time-series result that the
S-ARRA had no statistically signifcant impact. With the change in outlays as
the dependent variable, the time-series result is
O = a0 + a1O−1 + a2R + a3A = − 0.001 − 0.212O−1 + 0.233R + 0.049A,
(0.010) (0.071) (0.109) (0.168)
whereas 3SLS estimation of cross-sectional results for forty-fve states—I omit
Alaska as an outlier as well as the four states with fscal years ending on dates
other than June 30—gives
O = a0 + a1O−1 + a2R + a3A = 137 − 0.311O−1 + 0.270R + 0.917A.
Copyright © 2021. Harvard University Press. All rights reserved.
(154) (0.077) (0.082) (0.647)
Te time-series and cross-sectional estimates are not inconsistent because the
standard error on the coef ficient of A is so much larger in the second equa-
tion.
Te picture changes when we add fnancial and political controls: the im-
pact of S-ARRA funds on state spending becomes signifcant both statistically
and economically. And the coef ficient on S-ARRA funds is 2.8 for the forty-
fve state regressions, which is to say that one dollar transferred to a state led
to $2.80 of additional spending. When the states are separated by population,
not only do the smaller states receive more S-ARRA funds per capita, but the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 567
estimates of the impact on state spending are markedly diferent for the
smaller and larger states: the coef
ficient on S-ARRA funds for small states is
more than 4.5, and the estimate for large states falls below one. Te results are:
forty-fve states
∆O = a0 + a1R + a2O−1 + a3A + a′1Z1 + + a′nZn =
−323 − 0.065O−1 + 0.015R + 2.801A + a′1Z1 + + a′nZn
(136) (0.087) (0.083) (0.661)
twenty-two small states
∆O = a0 + a1R + a2O−1 + a3A + a′1Z1 + + a′nZn =
−613 − 0.153O−1 + 0.032R + 4.58A + a′1Z1 + + a′nZn
(180) (0.152) (0.125) (0.826)
twenty-three large states
∆O = a0 + a1R + a2O−1 + a3A + a′1Z1 + + a′nZn =
−233 − 0.141O−1 + 0.181R + 0.922A + a′1Z1 + + a′nZn.
(165) (0.106) (0.115) (0.484)
Notably, the impact of S- ARRA funds on state spending depends on
whether the governor is a Democrat or a Republican, Republican states in-
creasing spending by approximately ffy cents less per dollar of S-ARRA
funds received.
Te multiplier estimates—as refected by the coef ficient on O in the equa-
tion in which Y is the dependent variable—are on the high side compared
with other investigations. My estimates of the multiplier are 2.6 for the forty-
fve states, 2.8 for the twenty-two small states, and 4.7 for the twenty-three
large states. To some extent the diferences can be explained by the larger
amount of spillovers from one state to another in smaller states.
In marked contrast with the time-series results, these fndings are robust to
how the relationship between spending and revenues and lagged expendi-
Copyright © 2021. Harvard University Press. All rights reserved.
tures is specifed. In particular, the results are relatively little afected by
whether or not we attach importance to expenditure smoothing.
What Did We Learn from State Budget Of
ficers?
Te econometrics adds to the consensus that the S-ARRA stimulated the
economy. But I would be the last to claim that this analysis is conclusive. Not
only because, as I have noted, there were errors in data processing, but be-
cause even regression results from a pristine data set are no better than the
models used to interpret the data.
For this reason the quantitative analysis was supplemented by a direct ex-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
568 Raising Keynes
amination of the plausibility of the counterfactual assumption on which the
analysis was based, namely, that the states could have maintained expendi-
tures despite the reduction in revenues and the reduction in Rainy-Day fund
balances. Having risen by some 6 percent per year from FY-2003 to FY-2008,
non-ARRA revenues fell by 3 percent in FY-2009 and barely recovered in FY-
2010. Since upward pressure on expenditures intensifed, Rainy-Day funds
were bound to suffer. Here the aggregates are misleading since two states,
Alaska and Texas, accounted for more than one-quarter of Rainy-Day assets
at the end of FY-2008 and more than half of these assets when FY-2009 came
to a close. As Table 15.4 shows, if we leave out these two states, Rainy-Day
funds were practically exhausted by the end of FY-2010 and only began to
recover in FY-2011, well afer the ARRA spigot had been turned on.
To fnd out what the states did with the S-ARRA money, Peter Spiegler and
I sent a questionnaire to all ffy state budget of ficers. Te questions were
framed to provide a basis for conversation without being so restrictive that
they would prevent us from learning things about state-budgeting practice
that we had not anticipated (Marglin and Spiegler 2013, pp. 842–844). Budget
officers from twenty-nine states responded, and as near as we could tell these
states did not difer systematically from the nonresponders.
Needless to say, this approach is unorthodox. Economists generally resist
asking agents for information about why they do what they do or what they
would have done if the circumstances had been diferent.14 Ofen, there is
good reason for this reluctance: there are too many agents, it is hard to get a
representative sample, and agents may have trouble reconstructing the cir-
cumstances of their decisions well enough to answer, especially when the
questions involve a counterfactual. Fortunately, none of these reasons apply to
the case at hand. Tere are only ffy states, and state budget of ficers are a
highly professional group of men and w omen. It thus seemed reasonable to
ask these of ficers what they would have done had there been no ARRA funds
to ofset lost revenues and increased demands for expenditure that were the
Copyright © 2021. Harvard University Press. All rights reserved.
twin results of the Great Recession.
Te responses to our questionnaire indicated that, aside from “fossil-fuel”
Table 15.4 Rainy-Day Fund Balances by Fiscal Year ($ millions)
2008 2009 2010 2011
Rainy-Day Fund Balances, All States 32,943 29,006 21,034 24,651
Alaska 5,601 8,898 10,364 12,981
Texas 4,355 6,276 7,693 5,012
Rainy-Day Fund Balances, All Other States 22,987 13,832 2,977 6,658
Number of States with Zero Balance 5 10 15 17
Source: National Association of State Budget Of
ficers, Te Fiscal Survey of States, various years.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 569
states, which generally rely on severance taxes on the extraction of natural
resources rather than on income and sales taxes, internal funds15 were insuf fi
cient to support operating expenditure at the level actually observed. More-
over, maintaining expenditure by shifing it to the capital budget was so lim-
ited that it would not have added signifcantly to the spending resources of
the states. On the capital-expenditure side, the responses indicated that states
were very limited in their ability to increase capital borrowing during the
Great Recession, as their capital borrowing limit is tied to projected reve-
nues.16
In short, it is implausible that states could have and would have increased
net borrowing to fund spending at the levels observed with the S-ARRA. Te
collective view of the state budget of ficers was that the great majority of the
states would have cut spending signifcantly without the S-ARRA. Te evi-
dence indicates that states not only would have done so but that in almost all
cases they could not have done otherwise.
Conclusions
Did the stimulus work? Te short answer is yes. If we fll in the blanks in Table
15.3, we can arrive at an estimate for the ARRA as a whole. Table 15.5 takes a
conservative stance and estimates the frst-round MPC of the states as one, a
fgure roughly two standard deviations below the point estimates of A in the
cross-sectional regressions, and continues to assume a multiplier of 1.5, also a
conservative fgure relative to the cross-sectional estimates. On this basis we
arrive at an estimate of $469 billion for the category of Contracts, Loans, and
Grants.17
Table 15.5 ARRA Spending and Impact, by Spending Category ($ billions)
(1) (2) (3) (4) (5) (6)
Direct
Copyright © 2021. Harvard University Press. All rights reserved.
Reallocating First-Round Spending of Total
Reported Medicaid to MPC Recipients Impact
Spending Grants =v = (3) × (4) Multiplier = (4) × (5)
Tax Benefts, upper- 178.1 178.1 0.0 0.0 0.0 0.0
income recipients
and business
Tax Benefts, lower- 121.7 121.7 1.0 121.7 1.5 182.6
income groups
Entitlements 220.2 133.3 1.0 133.3 1.5 200.0
Contracts, Loans, 225.8 312.7 1.0 312.7 1.5 469.0
and Grants
Total 745.8 745.8 851.7
Source: Column (1) from Table 15.2; remaining data, author’s calculations.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
570 Raising Keynes
Te overall impact of the ARRA up to December 2011 works out to $852
billion, a little more than 2 percent of GDP over the two and a half years be-
tween the summer of 2009 and the end of 2011. Tis is in line with the fnal
estimate of the Council of Economic Advisers (2014). If the Obama adminis-
tration can be faulted on its stimulus program, it is for failing to appreciate the
gravity of the situation it inherited in January 2009, for lacking the courage or
foresight to ask for more stimulus over a longer time period, for failing to ar-
gue forcefully enough that more of the stimulus should be directed to lower-
income benefciaries who would have been more likely to spend than to save.
Taken together, the available evidence suggests that state-government ex-
penditure was signifcantly increased by the S-ARRA. Cogan and Taylor pre
sent time-series evidence that expenditure smoothing was responsible for the
lack of an increase in spending beyond what non-ARRA revenues would have
permitted. But this evidence requires a prior commitment to the idea that
expenditure smoothing drives spending. Because of serial correlation in the
data, roughly the same coef ficients on lagged expenditure would show up
even if no expenditure smoothing took place. So a high statistical signifcance
for the coef ficient of lagged expenditure and a correspondingly low signif
cance for the coef ficient of S-ARRA grants is not independent evidence for
expenditure smoothing.
Tis chapter reinforces the consensus view with respect to the stimulus as a
whole: both econometric analysis of cross-sectional state data and interviews
with state budget of ficers suggest that the S-ARRA allowed the states to main-
tain spending programs that would have been drastically cut if the stimulus
had not been enacted. A portion of the ARRA monies might have gone to
shore up state balance sheets—as indeed was the intention of the ARRA legis-
lation—but far less than Cogan and Taylor contend. My estimate, based on a
conservative interpretation of the cross-sectional data, is that for FY-2010 the
states on average spent all the S-ARRA money they received—and then some.
Unsurprisingly, states with Democratic governors increased their spending
Copyright © 2021. Harvard University Press. All rights reserved.
more than states with Republicans in the statehouse.
Taylor rightly argues that much of the post-hoc vindication of the ARRA
could have been—and in fact was—written before one dime of ARRA monies
had been spent. But this is a case of the pot calling the kettle black: whatever
the truth of the Cogan–Taylor hypothesis, their methodology guaranteed that
the data on state and local governments would confrm expenditure smooth-
ing. In contrast, the cross-sectional regressions presented here did not pre
determine the outcome one way or another; the results could have been un
favorable to the hypothesis that the S-ARRA increased state-government
spending.
Given the limits to what can be learned from any regression analysis, ad-
ditional light on the question of causality was provided by state budget of fi
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Did the Obama Stimulus Work? 571
cers—the very individuals who would have been in the thick of cause and ef-
fect. Here too every attempt was made to frame questions in a way that would
have permitted answers on both sides. Tere was a uniformity of responses—
but not unanimity—with respect to how the S-ARRA actually afected expen-
ditures, even when the respondents obviously difered in their evaluation of
the ARRA as a policy. So, while it is true that the framework of analysis afects
the results, it is not the case that all frameworks are created equal.
Te fnal lesson is skepticism about the conventional distinction between
positive and normative economics. Mainstream economists believe that de
scription can be separated from values, the frst representing science, the sec-
ond ideology. But just as there are no facts without theory, there is no separate
realm for description that does not embody values. Ideology ought not to be,
as it is glossed in the Cambridge Dictionary of Philosophy (Audi 1999, p. 406),
“a disparaging term used to describe someone else’s political views which one
regards as unsound.” Acting on ideology is not a failing or disease of the
Other, against which Taylor (or I) can claim immunity. As Joseph Schumpeter
insisted (1949, p. 349), we all operate on the basis of assumptions that cannot
be proved or disproved, and ideology is the coin of the vast realm of what is
beyond our powers to confrm or deny. Tis does not mean there is nothing
to discuss, nothing to learn. To the contrary. We may seek to transcend ideol-
ogy, but we will never do so until we admit that it is the necessary starting
point of any serious discussion about policy.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. empirical appendix .
Regressions and Teir Discontents
Tis appendix takes up three issues: frst, the modifcations of the Cogan–
Taylor (2012) regressions and the reasons for these modifcations; second,
why lagged dependent variables may introduce spurious correlation; and, f
nally, the cross-sectional regressions summarized in the main body of this
chapter.
Modifying Cogan–Taylor
A relatively inconsequential change is that I translated Cogan and Taylor’s
quarterly data to fscal years, using the period July 1 to June 30 as the fscal
year, since this is the fscal year for the overwhelming majority of the states.1
Te reason is that the fscal year is the unit of time over which budget deci-
sions are normally made (even though in an emergency like the one that fol-
lowed the abrupt downturn of the economy in the fall of 2008, many states
were obliged to make midcourse adjustments).
More important, I excluded the imputed value of capital services in the
National Income and Product Accounts. NIPA calculations of state govern-
ment purchases include these imputations, which lead to overstating pur-
chases G and understating transfers E. Tis modifcation has a big impact
Copyright © 2021. Harvard University Press. All rights reserved.
on individual regression coef ficients but does not change the qualitative re-
sults that Cogan and Taylor obtain. Te key result is that the coef ficients of
S-ARRA grants in the two estimated equations imply a larger negative impact
of the S-ARRA on G, which is not ofset by a larger positive impact on E; the
overall impact of the S-ARRA on the states’ net addition to their fnancial as-
sets (L) remains positive. Indeed, my time-series estimates suggest that one
dollar of S-ARRA money leads to an increase of net saving of $1.36—whereas
Cogan and Taylor’s estimate is that just over $0.90 of every dollar of S-ARRA
grant money is saved. My results are summarized in the frst three columns of
Table 15.6.
A further modifcation of the regression procedure is more consequential:
column (4) consolidates transfer payments and purchases of goods and ser
572
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Empirical Appendix 573
Table 15.6 Regressions of Nominal Purchases and Transfers on Lagged Dependent
Variables, Revenues, and ARRA
(1) (2) (3) (4)
Dependent Variable
G E L O
Constant 8.85 −1.595 −7.255 5.169
1.944 1.622 2.435
G−1 0.78 −0.78
0.032
E−1 0.854 −0.854
0.06
O−1 0.783
0.04
R 0.198 0.054 0.748 0.269
0.025 0.017 0.041
A −0.586 0.226 1.359 −0.314
0.057 0.065 0.108
Adj R2 0.9996 0.999 0.9997
Frequency/ FY FY FY FY
Time Period 1969–2011 1969–2011 1969–2011 1969–2011
N 43 43 43 43
Sources: Bureau of Economic Analysis; Department of Health and Human Services.
Note: Newey-West standard errors appear below coefficients.
vices into a single variable, “outlays,” denoted by O and equal to E + G. At the
state level it makes little sense to distinguish between the two because trans-
fers by states difer in important ways from transfers from the federal govern-
ment. Most direct federal transfers to individuals come with few or no strings
attached—think social security—and it is reasonable to consider such trans-
Copyright © 2021. Harvard University Press. All rights reserved.
fers simply as putting more money in the pockets of recipients. Food stamps,
for example, though nominally tied to the purchase of groceries, are in fact
fungible since they free up cash for other purposes. Estimates suggest that
only about $0.26 to $0.35 of each one dollar of food stamps generates addi-
tional spending on food (Mark Nord and Mark Prell [2011], p. 17). Te re-
maining $0.65 to $0.74 simply replaces spending that the recipient would
have made out of her own pocket, freeing up spending on other goods and
services.
Unlike food stamps, social security, and other federal programs,2 the bulk
of transfer payments made by states and localities are not really payments to
the nominal recipients—except by national income accounting convention.
Medicaid, the largest single transfer program, appears in the national income
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
574 Raising Keynes
accounts as a transfer payment to individuals, but the individual never sees
any cash. Te payments are actually made to vendors of medical goods and
services—for visits to doctors, surgical procedures, prescription drugs—and
are purchases of goods and services every bit as much as direct purchases by
state governments. For the purpose of analyzing the impact of the S-ARRA, it
makes little diference whether states purchase goods and services directly or
indirectly.
Following this logic, the relevant dependent variable is total current out-
lays, the sum of current purchases of goods and services and transfer pay-
ments. Te result, in column (4) of Table 15.6, with total outlays, O = G + E, as
the dependent variable, is that the coef ficient of S-ARRA grants on outlays
becomes negative, lying between the two separate coef ficients for purchases
and “transfers.” More important, this coef ficient is insignifcantly diferent
from zero.
Tere remains a problem: the data are dominated by an upward trend in all
the variables. Tese trends hide the complementary behavior of purchases
and transfers, purchases of goods and services being correlated with fuctua
tions in revenues, and transfers being correlated with the upward trend in
revenues. For this reason, there is a stronger correlation between revenues
and total outlays than between revenues and either of the two components of
outlays.
Normalizing the variables by dividing by potential GDP removes some of
the problems of using the trend-dominated data in its raw form. In columns
(1) and (2) of Table 15.7, the negative efect of the S-ARRA on states’ pur-
chases of goods and services remains signifcant, both statistically and eco-
nomically, and is not ofset by the positive stimulus to transfers. But the re-
gression in which the dependent variable is total outlay (column [4]), suggests
a negligible impact on spending overall—the coef ficient on A is not only
small, suggesting that less than a nickel of each S-ARRA dollar got spent, it
is statistically indistinguishable from zero. Tis regression, like the regres-
Copyright © 2021. Harvard University Press. All rights reserved.
sions reported by Cogan and Taylor, suggests that the lion’s share of S-ARRA
grants, the unspent $0.95, went toward increasing fnancial assets or reducing
fnancial liabilities.
In short, all the modifcations that I introduced into the Cogan–Taylor
analysis do nothing to change their qualitative conclusions.
Spurious Correlation with Lagged Dependent Variables
Serial correlation in the data can generate the key result in the Cogan–Taylor
regressions, namely, that lagged outlay matters a lot, and current revenues and
S-ARRA funds matter hardly at all—even if lagged outlay is actually irrelevant
to current outlay. Suppose that in the true specifcation revenues, including
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Empirical Appendix 575
Table 15.7 Regressions of Normalized Purchases and Transfers on Lagged Dependent
Variables, Revenues, and ARRA
(1) (2) (3) (4)
Dependent Variable
G E L O
Constant 0.007 0.004 −0.011 −0.001
0.010 0.004 0.010
G−1 0.813 −0.813
0.084
E−1 0.980 −0.980
0.037
O−1 0.788
0.071
R 0.098 −0.022 0.924 0.233
0.078 0.035 0.109
A −0.465 0.189 1.277 0.049
0.081 0.113 0.168
Adj R2 0.8101 0.982 0.9441
Frequency/ FY FY FY FY
Time Period 1969–2011 1969–2011 1969–2011 1969–2011
N 43 43 43 43
Sources: Bureau of Economic Analysis; Department of Health and Human Services.
Note: Newey-West standard errors appear below coefficients.
S-ARRA funds, drive outlays, and lagged outlays are irrelevant. Lagged out-
lays still show up with a positive coefficient and bias the estimates of the true
drivers downward if the independent variables, revenues and S-ARRA funds,
Copyright © 2021. Harvard University Press. All rights reserved.
and the error term are serially correlated (Achen 2001).
To see the efect of serial correlation in a simpler version of the present
case, ignore the efect of the S-ARRA and suppose the correct specifcation is
Ot = a0 + a1Rt + t.
In this regression both R and exhibit high serial correlation, with respective
coefficients of 1.003 and 0.731. Tat is, if successive values of the independent
variable and the error term are related by a frst-order process, we have the
results
Rt = 1Rt−1 + t = 1.003 Rt−1 + t,
t = 2t−1 + t = 0.731 t−1 + t.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
576 Raising Keynes
If we substitute into the equation for Ot, we have
t = Ot − a0 − a1Rt = 2t−1 + t = 2(Ot−1 − a0 − a1Rt−1) + t.
Since
Rt−1 = ( 1)−1Rt − ( 1)−1t,
the equation for Ot becomes
Ot = (1 − 2)a0 + ( 1 − 2)( 1)−1a1Rt + 2Ot−1 + t + 2( 1)−1a1t.
Te lagged dependent variable Ot−1 sneaks in because of the serial correla-
tion of the error term in the original equation and—when 1 > 2 and 1 is
close to unity—at the same time reduces the coef ficient on the true explana-
tory variable Rt. In the case at hand, since 1 = 1.003 and 2 = 0.731, the esti-
mated coef ficient of R in this last equation is ( 1 − 2)( 1)−1a1, which is ap-
proximately one-quarter of its true value a1. Table 15.8 presents the relevant
data. Te estimates of the constant term and the coef ficient of revenues in
column (2) are â0 = −0.006 and â1 = 1.083, with the result that the equation we
derived for O becomes
Oˆ t 0.001 0.294 Rt 0.731Oˆ t 1 t 0.789 t .
Observe that these numbers are close to what is reported in column (1) as
the results of estimating the same equation by ordinary least squares,
Ot = b0 + b1Rt + b2Ot−1 + t,
for which bˆ0 = −0.001, bˆ1 = 0.224, and bˆ2 = 0.796. Unsurprisingly, these coef
fi
cients difer little from the coef ficients reported in column (4) of Table 15.7,
which includes A as an explanatory variable.
Te formulas for the limiting values of the direct estimates of the coef fi
cients in column (1) are (Achen 2001, pp. 5–6)
Copyright © 2021. Harvard University Press. All rights reserved.
1 R2
plim bˆ1 1 1 2 a1,
1 12 R 2
1 R2
plim bˆ2 2 ,
1 12 R 2
where R2 is from the equation reported in column (2) of Table 15.8, and is
equal to 0.7275.3 Without knowing the true value of a1 we cannot estimate
the limiting value of bˆ1; we can only estimate this limiting value conditional
on the estimate â1. On this basis we have plim bˆ1 = 0.276. In addition, on the
assumption that the true coef ficient of lagged expenditure is zero, we have
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Empirical Appendix 577
Table 15.8 Regressions of Total Spending with and without Lagged Dependent Variable
(1) (2) (3) (4)
Dependent Variable
O O R
Constant −0.001 −0.006
0.009 0.012
O−1 0.796
0.048
R 0.224 1.083
0.083 0.090
R−1 1.003
0.005
−1 0.731
0.084
Adj R2 0.944 0.721 0.999 0.498
Frequency/ FY FY FY FY
Time Period 1969–2011 1969–2011 1969–2011 1969–2011
N 43 43 43 43
Sources: Bureau of Economic Analysis; Department of Health and Human Services.
Note: Newey-West standard errors appear below coefficients.
plim bˆ2 = 0.743, which once again is close to the direct OLS (ordinary least
squares) estimate of 0.796.
Estimating the Impact of the S-A RRA by
Cross-Sectional Regressions
For now—I will modify the model as we proceed—the model is
Copyright © 2021. Harvard University Press. All rights reserved.
∆O = a0 + a1R + a2O−1 + a3A + a′1Z1 + + a′nZn + ,
∆Y = b0 + b1∆O + b′1Z1 + + b′nZn + ,
∆R = c0 + c1∆Y + c3A + b′1Z1 + + c′nZn + ,
A = d0 + d′1Z1 + + d′nZn + ,
where Z1, , Zn are the exogenous variables, including both instrumental
variables and other control variables.4 Table 15.9 summarizes the data.
Te frst question is the extent to which cross-sectional evidence duplicates
the Cogan–Taylor time-series results. I start with an OLS estimation of the
outlay equation with all the Zs omitted. Te frst column in Table 15.10 pre
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
578 Raising Keynes
Table 15.9 Summary Statistics
Obs Mean Std Dev Min Max
Variable
∆O 46 139 203 −269 921
R 46 5,475 2,036 3,490 16,023
∆R 46 30 226 −1160 427
O−1 46 5,596 1,720 3,559 14,124
R − O−1 46 −121 517 −907 1,899
A 46 308 82 158 599
∆Y 46 239 1,315 −5,249 2,150
Control Variables
Med2008 46 910 260 460 1,483
Roads2008 46 108 182 2 816
Mine%2010 46 3 6 0 36
NF−1 46 1,986 9,451 −3,641 60,584
∆NF−1 46 −323 681 −3,735 1,719
Rainy−1 46 371 1,838 0 12,528
∆Rainy−1 46 68 703 −584 4,642
RepGov × A 46 142 177 0 599
DumRepGov 46 0.435 0.501 0 1
R−1 46 5,445 2,096 3,425 16,174
Control Variables
Med2008 = State payments to vendors for Medicaid, FY-2008.
Roads2008 = Federal aid to state and local governments for highway trust fund, FY-2008.
Mine%2010 = Percentage of gross state product due to mining and resource extraction, FY-2010.
NF−1 = Net fnancial assets, end FY-2009.
∆NF−1 = Change in net fnancial assets, end FY-2009 minus end FY-2008.
Rainy−1 = Rainy-Day fund, end FY-2009.
∆Rainy−1 = Change in Rainy-Day fund, end FY-2009 minus end FY-2008.
RepGov × A = Product of DumRepGov and A.
DumRepGov = Dummy variable equal to 1 if the state house was occupied by a Republican at the
beginning of FY-2010, equal to 0 if occupied by a Democrat.
Except for Mine%2010 and DumRepGov, all variables are expressed per capita of population (as
Copyright © 2021. Harvard University Press. All rights reserved.
of April 2010).
sents the results of this regression. Te OLS results suggest that S-ARRA
funds had a large impact on spending: the coef ficient is close to one, and the
t-statistic exceeds two. Nothing much changes if we take account of the endo-
geneity of S-ARRA money by instrumenting for A (column [2]). Te coef fi
cients change very little, and the standard errors increase only marginally. At
ficient of A no longer passes mus-
a strict 5 percent signifcance level, the coef
ter, but p = 0.058 is very close.
Te frst-stage equation, deploying instruments similar in spirit to those
devised by Wilson (2012) and Chodorow-Reich et al. (2012), yields an F-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 15.10 Te Efect of S-ARRA on the Change in State Government Outlays between FY-2009 and FY-2010
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
OLS 2SLS(1) 2SLS(2) 3SLS 3SLS 3SLS 3SLS 3SLS 3SLS OLS
A and R
A Estimated Estimated in with with with with with with
15
in First Stage First Stage controls controls controls controls controls controls
•
Constant 334.100 318.402 516.692 454.976 2.854 −304.127 −283.059 −318.612 −350.366 −13.923
125.100 187.838 290.886 210.723 145.421 128.075 123.300 137.660 132.371 124.856
R 0.190 0.186 0.296 0.325 0.119 0.044 0.175
0.059 0.093 0.124 0.103 0.055 0.076 0.076
∆R 0.015
0.191
O−1 −0.274 −0.272 −0.379 −0.404 −0.216 −0.067 −0.018 −0.194
Empirical Appendix
0.067 0.101 0.139 0.104 0.061 0.083 0.027 0.083
R − O−1 0.051
0.074
A 0.961 1.049 0.404 0.535 2.365 2.365 1.913 2.323 2.138 1.265
0.411 0.555 0.695 0.833 0.657 0.588 0.432 0.711 0.467 0.394
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
579
Copyright © 2021. Harvard University Press. All rights reserved.
580
Table 15.10 (continued)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
OLS 2SLS(1) 2SLS(2) 3SLS 3SLS 3SLS 3SLS 3SLS 3SLS OLS
A and R
A Estimated Estimated in with with with with with with
in First Stage First Stage controls controls controls controls controls controls
Control Variables
NF−1 0.009 0.005 0.011 0.007 0.005
0.008 0.007 0.010 0.007 0.008
∆NF−1 0.171 0.189 0.199 0.213 0.112
0.068 0.057 0.053 0.043 0.068
Rainy−1 −0.215 −0.209 −0.208 −0.209 −0.168
0.082 0.077 0.083 0.080 0.084
Raising Keynes
∆Rainy−1 0.420 0.436 0.408 0.447 0.279
0.174 0.167 0.191 0.172 0.184
RepGov × A −0.231 −0.437 −0.441 −0.449 −0.471 −0.300
0.149 0.135 0.133 0.137 0.132 0.143
R2 0.373 0.372 0.3126 0.280 0.294 0.523 0.561 0.512 0.513 0.552
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
N 46 46 46 46 46 46 46 46 46 46
Note: Standard errors appear below coef
ficients.
15 • Empirical Appendix 581
value of 28. Like them, I looked to the three areas that constituted the bulk
of S-ARRA aid to the states. I was unable to fnd a good instrument among
the obvious candidates in education spending, but FY-2008 payments by
the states to vendors of Medicaid services (Med2008) and federal aid for
roads in FY-2008 (Roads2008) proved serviceable for health care and highway
spending.
To capture the particularities of states that rely directly or indirectly on
extraction of natural resources for tax revenues, I added a variable repre
senting the percentage of gross state product due to resource extraction
(Mine%2010). Tis variable was one of several with which I experimented to
capture the special features of these states. As Figure 15.3 shows, there is a
great diference among the states in this dimension, and I thought this instru-
ment would capture the drop in the prices and production levels of oil, natu-
ral gas, and coal as well as the diferential impact of these declines on the
economies of the several states.
Something else must be at work. Despite steep declines in both production
and prices in the second half of calendar 2008 (the frst half of FY-2009),
prices fell relatively modestly fscal year to fscal year, 12 percent for gas and
Copyright © 2021. Harvard University Press. All rights reserved.
10.03–37.60
5.12–10.03
1.99–5.12
1.43–1.99
0.75–1.43
0.20–0.75
No data
Figure 15.3 Percentage of state gross product from mining and resource
extraction (2010).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
582 Raising Keynes
8 percent for oil, while production was almost fat, as Figures 15.4(a) and
15.4(b) indicate.
Although endogenizing S-ARRA by itself has little efect, the coef ficients
change much more if we also treat non-ARRA revenue as endogenous (col-
umn [3] in Table 15.10). In particular, the coef ficient of A falls to 0.4. In this
case, the results are qualitatively similar to the time-series results, with similar
coef ficients on revenues and lagged expenditures. Table 15.11 compares col-
umn (4) of Table 15.7 with column (4) of Table 15.10. (Te frst regression is
rewritten so that the dependent variable is ∆O rather than O.) Te coef ficient
on S-ARRA funds is greater in the cross-section than in the time series, but in
both cases the standard error on the coef ficient of A is suf ficiently high that
we cannot reject the Cogan–Taylor view that the true value of the coef ficient
is zero. Observe that in the cross-sectional case, unlike the time-series case,
serial correlation cannot be behind the high t value of the coef ficient of O−1.
Introducing control variables makes a big diference to the coef ficient on
S-ARRA funds. In column (5) of Table 15.10, the coef ficient increases mark-
edly, to 2.37, and it becomes statistically signifcant (z = 3.6). Te imprecision
of the estimate in column (4) appears to be at least partly the result of lump-
ing states with Republicans in the state house with states with Democratic
governors. For when we separate the two by means of a dummy variable
(equal to 1 for Republican states and equal to 0 for Democratic states), we
fnd a negative impact of Republican control on the amount of S-ARRA rev-
enue that is translated into state government outlays and a larger and more
precise estimate of the coef ficient of A. Te negative coef ficient of RepGov × A
Price indices (2002 = 100) Production indices (2012 = 100)
600 140
500 120
100
400
Copyright © 2021. Harvard University Press. All rights reserved.
80
300
60
200
40
100 20
0 0
08
08
09
09
09
09
10
10
10
10
20
20
20
20
20
20
20
20
20
20
ul
ul
an
ul
an
ul
an
ul
an
ul
1J
1J
1J
1J
1J
1J
1J
1J
1J
1J
Gas Oil Coal Gas Oil Coal
Mining Industrial production
(a) Prices (b) Production
Figure 15.4 Energy extraction: prices and production (FY-2009 and FY-2010).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Empirical Appendix 583
Table 15.11 Comparison of Time-Series and Cross-
Sectional Regressions with Expenditure Smoothing
Time Series Cross-Section
Dependent Variable
∆O ∆O
Constant −0.001 454.976
0.010 210.723
O−1 −0.212 −0.404
0.071 0.104
R 0.233 0.325
0.109 0.103
A 0.049 0.535
0.168 0.833
R2 0.2140 0.280
Frequency/ FY
Time Period 1969–2011
Coverage States with FY
ending June 30
N 43 46
suggests that Republicans can be taken at their word: they will cut expendi-
tures when they get the chance.
Introducing two kinds of fnancial controls, net fnancial assets and Rainy-
Day funds, increases the precision of the estimate of the infuence of political
party and (except for column [8]) reduces the standard error of the estimate
of the impact of S-ARRA funds on outlays. Tis combination of a large value
of the coef ficient and a high level of statistical signifcance is robust to the
precise specifcation of the model, as columns (5) to (9) show.5
Apart from the Rainy-Day fund, all the coef ficients have the expected sign.
Copyright © 2021. Harvard University Press. All rights reserved.
It turns out that the anomaly is due to Alaska, which is an extreme outlier in
many respects. On a per-capita basis, its 2008 Rainy-Day fund was eighteen
times the size of the Rainy-Day fund of the next highest state (Wyoming, also
a natural-resource rich state). Its net fnancial assets per capita were more
than three times the assets of the next state (also Wyoming), and its revenues
and expenditures more than one and a half times the next state’s (again Wyo-
ming). If we omit Alaska, the summary data are given in Table 15.12.
Te fnancial controls now make more sense, as Table 15.13 shows. Except
for the last regression, levels of neither net fnancial assets nor the Rainy-
Day fund are signifcant, while the changes in both variables have a positive
sign. Since ∆NF−1 includes ∆Rainy−1, the coef ficient of the change in Rainy-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
584 Raising Keynes
Table 15.12 Summary Statistics (omitting Alaska)
Obs Mean Std Dev Min Max
Variable
∆O 45 149 196 −248 921
R 45 5,241 1,287 3,490 9,658
∆R 45 34 227 −1160 427
O−1 45 5,407 1,156 3,559 8,933
R − O−1 45 −166 422 −907 1634
A 45 302 70 158 521
∆Y 45 316 1,220 −5,249 2,150
Control Variables
Med2008 45 898 250 460 1,483
Roads2008 45 92 149 2 691
Mine%2010 45 3 6 0 36
NF−1 45 683 3,404 −3,641 18,552
∆NF−1 45 −247 451 −1,744 1,719
Rainy−1 45 101 143 0 706
∆Rainy−1 45 −33 138 −584 315
RepGov × A 45 132 165 0 521
DumRepGov 45 0.422 0.499 0 1
R−1 45 5,207 1,348 3,425 10,818
Day funds is approximately three times higher than the coef ficient on the
change in other net fnancial assets, that is, net fnancial assets excluding
Rainy- Day funds. (Rewrite the control variable ∆NF−1 and ∆Rainy−1 as
[∆NF−1 − ∆Rainy−1] and ∆Rainy−1, and recalculate the two coef ficients as they
appear in the table, call them a′∆NF−1 and a′∆Rainy−1, as the coef ficients a′1 and
a′2 of [∆NF−1 − ∆Rainy−1] and ∆Rainy−1. We have a′1 = a′∆NF−1 and a′2 =
a′∆NF−1 + a′∆Rainy−1.) Roughly speaking, for every dollar that a state’s Rainy-Day
fund declines between FY-2008 and FY-2009, its outlays decline by $0.50,
Copyright © 2021. Harvard University Press. All rights reserved.
while for every dollar that non-Rainy-Day net assets decline, outlays decline
by $0.15. Tis is not surprising unless, like Cogan and Taylor, you believe in
the complete fungibility of state funds.6
Observe that the fnancial coef ficients do not strongly support Friedman–
Modigliani expenditure smoothing even if—given the large standard errors
for the level variables—it is not rejected by the data. An agent who optimizes
according to permanent-income logic bases outlays on total wealth, which is
to say, on net fnancial assets (if we suppose the states to take revenues as
given). Te higher the level of net assets, the greater should be the level of
outlays. Rainy−1, by this logic, should be completely fungible with other assets,
so its coefficient should be zero. By contrast, changes in net fnancial assets or
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Table 15.13 Te Efect of S-ARRA on the Change in State Government Outlays Between FY-2009 and FY-2010 (omitting Alaska)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
OLS 2SLS(1) 2SLS(2) 3SLS 3SLS 3SLS 3SLS 3SLS 3SLS OLS
A and R
A Estimated Estimated in with with with with with with
15
in First Stage First Stage controls controls controls controls controls controls
•
Constant 83.745 105.195 65.698 136.960 −95.181 −323.106 −279.347 −306.450 −326.057 −12.947
122.833 157.736 182.005 153.691 132.463 135.740 130.859 141.565 137.643 126.792
R 0.263 0.266 0.255 0.270 0.187 0.015 0.173
0.053 0.072 0.116 0.082 0.052 0.083 0.080
∆R 0.058
0.197
O−1 −0.274 −0.300 −0.292 −0.311 −0.245 −0.065 −0.047 −0.193
Empirical Appendix
0.067 0.072 0.108 0.077 0.054 0.087 0.030 0.085
R − O−1 0.038
0.083
A 0.990 0.890 1.077 0.917 2.006 2.801 1.855 2.655 2.004 1.277
0.350 0.628 0.753 0.647 0.544 0.661 0.451 0.722 0.485 0.413
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
585
Copyright © 2021. Harvard University Press. All rights reserved.
586
Table 15.13 (continued)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
OLS 2SLS(1) 2SLS(2) 3SLS 3SLS 3SLS 3SLS 3SLS 3SLS OLS
A and R
A Estimated Estimated in with with with with with with
in First Stage First Stage controls controls controls controls controls controls
Control Variables
NF−1 0.008 0.006 0.010 0.005 0.005
0.009 0.008 0.010 0.008 0.009
∆NF−1 0.156 0.202 0.162 0.221 0.111
0.071 0.061 0.058 0.045 0.070
Rainy−1 0.061 −0.106 0.075 −0.058 −0.135
0.248 0.231 0.236 0.221 0.273
Raising Keynes
∆Rainy−1 0.364 0.371 0.333 0.380 0.275
0.179 0.177 0.194 0.182 0.190
RepGov × A −0.165 −0.463 −0.429 −0.451 −0.462 −0.303
0.130 0.143 0.139 0.138 0.137 0.147
R2 0.511 0.508 0.510 0.508 0.458 0.406 0.522 0.443 0.487 0.595
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
N 45 45 45 45 45 45 45 45 45 45
Note: Standard errors appear below coef
ficients.
15 • Empirical Appendix 587
in Rainy-Day funds should have no impact on outlays. Since the year-on-year
change between FY-2008 and FY-2009 is already included in the level at the
end of FY-2009, both ∆NF−1 and ∆Rainy−1 should have zero coef ficients.
For our purposes, the fnancial variables are not ends in themselves, but
means to accurate measurement of the efects of the S-ARRA, as refected in
the coef ficient of A and the coef ficient of RepGov × A. Evidently, omitting
Alaska does not change the results qualitatively. Te coef ficient of A is insig
nifcant until we add control variables. Once we add controls, the efect per
dollar of S-ARRA funds in states with Democratic governors lies between
$1.85 and $2.80 (omitting Alaska); in Republican states, the amount is about
$0.45 less. Within these limits, the estimate depends on how revenues and
lagged expenditures are supposed to afect the change in expenditures. Irre-
spective of the precise specifcation, the coef ficient of A remains statistically
signifcant, with z approximately equal to four in all cases. Of course there is a
wide band on these estimates, and using a conservative lower bound fxed at
two standard deviations from the point estimate would move the estimates
closer to one.
I turn now to the other parts of the overall model, the impact of the S-
ARRA on output and income, and its impact on tax revenues. S-ARRA reve-
nues are supposed to afect output through the demand efect of additional
state government outlays, whereas the impact on tax revenues has two poten-
tial channels, the frst a direct one running from S-ARRA grants to changes in
tax policy, the second via taxation of output at existing rates. Table 15.14
presents the complete model, with results for all forty-fve states in columns
(1), (2), and (3). Column (1) reproduces the results of column (6) in Table
15.13 and so needs no further comment. Te coef ficient of ∆O in the output
equation (column [2]) is the multiplier. Its value of 2.6 is somewhat higher
than the consensus range, but not implausibly so. Daniel Shoag, whose work
I shall examine shortly, estimates the state-outlay multiplier to be greater
than three in slack times (2010, table 12). In principle, the coef ficient of ∆O
Copyright © 2021. Harvard University Press. All rights reserved.
should underestimate the multiplier since it does not take account of spill-
overs of demand across state lines. We shall take up the question of spillovers
momentarily.
Te revenue equation in Table 15.14 (column [3]) is interesting for two
reasons. First, the coef ficient of S-ARRA funds, though it has the expected
negative sign, is insignifcant, a result that is robust to alternative specifca
tions. Te implication is that there is no clear impact of S-ARRA funds on
taxes. Tis runs counter to my intuition, which is that, without S-ARRA fund-
ing, the states would have been obliged to enact more and greater tax in-
creases than the $24 billion actually enacted (see note 1, above). Te second
point of interest is the coef ficient of ∆Y. If we interpret causality as run-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
588
Table 15.14 Modeling the Efect of the S-ARRA on Outlays, Gross State Product, and Revenues
(1) (2) (3) (4) (5) (6) (7) (8) (9)
3SLS 3SLS 3SLS 3SLS 3SLS 3SLS 3SLS 3SLS 3SLS
with with with with with with
with with with controls; controls; controls; controls; controls; controls;
controls controls controls small states small states small states large states large states large states
Dependent Variable
∆O ∆Y ∆R ∆O ∆Y ∆R ∆O ∆Y ∆R
Constant −323.106 266.687 119.465 −612.635 211.849 39.070 −232.762 −179.840 258.492
135.740 169.546 118.818 179.564 217.091 174.796 165.260 225.269 167.903
R 0.015 0.032 0.181
0.083 0.125 0.115
∆R
Raising Keynes
O−1 −0.065 −0.153 −0.141
0.087 0.152 0.106
R − O−1
A 2.801 −0.394 4.580 −0.072 0.922 −0.750
0.661 0.393 0.826 0.543 0.484 0.624
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
∆O 2.616 2.826 4.675
0.756 0.711 1.622
Copyright © 2021. Harvard University Press. All rights reserved.
Mine%2010 −125.652 −7.351 −153.223 −11.025 12.376 −5.585
19.868 5.223 20.639 7.330 41.619 7.397
∆Y 0.170 0.138 0.036
0.031 0.041 0.041
15
Control Variables
•
NF−1 0.008 0.012 0.040
0.009 0.010 0.019
∆NF−1 0.156 0.053 0.016
0.071 0.124 0.081
Rainy−1 0.061 0.396 0.317
0.248 0.307 0.461
∆Rainy−1 0.364 0.356 0.057
0.179 0.203 0.486
Empirical Appendix
RepGov × A −0.463 −0.469 −0.412
0.143 0.189 0.143
R2 0.406 0.574 0.388 0.510 0.779 0.680 0.456 0.053 −0.183
N 45 45 45 22 22 22 23 23 23
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Note: Standard errors appear below coef
ficients.
589
590 Raising Keynes
ning from output and income to tax revenues, the coef ficient of ∆Y suggests
an implicit tax rate of 0.17, much higher than the actual average tax rate of
about 0.1.
Te coef ficients of Mine%2010 in both the output equation and the revenue
equation are implausibly high, suggesting that a one percentage point in-
crease in the contribution of mining to a state’s output resulted in a loss of
$126 per capita, not to mention a $7 loss in per-capita tax revenue (though
the coef ficient of Mine%2010 in the revenue equation is statistically insignif
cant).7 Tis reinforces the suspicion that Mine%2010 is capturing something
other than mining.
In order to test the hypothesis that demand spillovers are important, I ran
the regression model reported in columns (1), (2), and (3) separately on the
twenty-two states with the smallest populations (excluding Alaska) and the
twenty-three largest states. Figure 15.5 shows the two sets. Te small states, it
will be noted, have quite diverse economies, and this variation appears to be
driving the results of these regressions. Te results are reported in Table
15.14, columns (4), (5), and (6) for the twenty-two small states, and columns
(7), (8), and (9) for the large states. My expectation was that the multiplier
would be larger for the larger states because spillovers would be smaller. Te
point estimates support this conjecture, but the multiplier for the large states
exceeds four, and is thus implausibly large. (For small states the multiplier is
Copyright © 2021. Harvard University Press. All rights reserved.
3,800,000–37,000,000
560,000–3,800,000
No data
Figure 15.5 Small and large states.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Empirical Appendix 591
less than three, still large but not implausibly so.) By the same token, the im-
pact of one dollar of S-ARRA money on state-government outlays in the
small states is estimated to be over four dollars, once again implausibly large.
Interestingly, the efect of Republican governors seems to be the same in large
and small states: the coef ficients of RepGov × A difer by only 0.05 in the two
samples. Finally, the diferential impact of Mine%2010 can be explained by
underlying economic diferences: states whose economies rely heavily on re-
source extraction are concentrated within the states with low populations, so
that Mine%2010 is signifcant in the small-state sample but not in the large-
state sample.
Tables 15.15 and 15.16 present summary statistics for the two sets of states.
Perhaps the most interesting information that emerges from a comparison of
the two tables is the diference in the average size of S-ARRA grants: $333 per
head in small states as against $272 in large states, a diference of approxi-
mately one standard deviation of the small-state distribution—not surprising,
perhaps, when Wyoming, with a population of 500,000, and California, with a
population north of 35 million, each has two senators.8
Te other noteworthy diference between the two sets of states is the much
greater variation in the small-state data. Tis is presumably why the small-
state regressions produce much higher R2 values.
Table 15.15 Summary Statistics, Small States (population < 3.8 million)
Obs Mean Std Dev Min Max
Variable
∆O 22 181 248 −248 921
R 22 5,819 1,485 3,498 9,658
∆R 22 12 303 −1160 427
O−1 22 5,835 1,318 3,687 8,933
R − O−1 22 −16 527 −907 1,634
A 22 333 70 220 521
Copyright © 2021. Harvard University Press. All rights reserved.
∆Y 22 231 1,608 −5,249 2,150
Control Variables
Med2008 22 930 280 460 1,483
Roads2008 22 172 183 34 691
Mine%2010 22 3 8 0 36
NF−1 22 1,726 4,552 −3,641 18,552
∆NF−1 22 −241 610 −1,744 1,719
Rainy−1 22 162 179 0 706
∆Rainy−1 22 −51 188 −584 315
RepGov × A 22 167 181 0 521
DumRepGov 22 0.50 0.51 0 1
R−1 22 5,807 1,580 3,425 10,818
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
592 Raising Keynes
Table 15.16 Summary Statistics, Large States (population > 3.8 million)
Obs Mean Std Dev Min Max
Variable
∆O 23 118 125 −181 380
R 23 4,688 746 3,490 6,278
∆R 23 56 119 −198 274
O−1 23 4,997 808 3,559 6,377
R − O−1 23 −308 218 −776 −6
A 23 272 57 158 401
∆Y 23 397 704 −1,128 1,419
Control Variables
Med2008 23 867 219 510 1,461
Roads2008 23 16 8 2 33
Mine%2010 23 2 3 0 13
NF−1 23 −314 1,127 −3,453 1,563
∆NF−1 23 −252 228 −642 243
Rainy−1 23 41 52 0 188
∆Rainy−1 23 −17 60 −133 141
RepGov × A 23 99 144 0 401
DumRepGov 23 0.35 0.49 0 1
R−1 23 4,632 739 3,473 6,187
One fnal exercise is worth reporting. In two papers, Daniel Shoag (2010,
2013) has proposed a clever way around the endogeneity problem in esti-
mates of the multiplier associated with government spending: he uses “wind-
fall” returns to state pension funds as an instrument for estimating purchases
of goods and services by the state. Shoag argues that because state employees
normally receive pensions based on some combination of years of service,
average earnings, and peak pay—in 2008, 97 percent of state pension-fund
assets were in so-called defned-beneft plans (Shoag 2010, p. 5)—the states
rather than the pensioners are the residual claimants on pension-fund assets.
Copyright © 2021. Harvard University Press. All rights reserved.
As residual claimants, their current obligation to pay into their retirement
fund is inversely related to the fund’s performance, so that direct purchases of
goods and services (education, highway maintenance) and indirect purchases
(Medicaid “transfers”) can be expected to vary positively with pension-fund
performance.9 Some portion of the diference in performance is explained by
diferences in the degree of risk a state is willing for its pension fund to take
on. But controlling for asset allocation, idiosyncratic pension-fund perfor-
mance (what Shoag calls windfalls) should be a reliable instrument for esti-
mating the state-spending multiplier since it should be a good predictor of
nonpension spending while being independent of difering economic condi-
tions across the states.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
15 • Empirical Appendix 593
Shoag fnds support for this hypothesis in the data. In his earlier paper
(Shoag 2010), he estimates that a dollar improvement in pension-fund perfor-
mance is associated with $0.43 of increased government spending (2010, table
9) and that each dollar of additional spending increases personal income
within the state by $2.12 (2010, table 12), and indeed by more than $3.00 in
times of slack. Tis last estimate is consistent with the estimates in the cross-
sectional regressions reported here. Shoag’s later paper, which is limited to the
postcrisis period, fnds lower estimates of pension windfalls on government
spending, $0.29, and of government spending on income, $1.43 (Shoag 2013,
table 1).
I was unable to fnd an impact of pension-fund performance on state-
government spending. When the controls included the change between end
FY-2008 and end FY-2009 in per-capita pension-fund assets, expressed either
as the per-capita change or the change as a percentage of FY-2008 assets, the
coef ficient of this variable had the wrong (negative) sign and was statistically
insignifcant. Tis was true independently of the particular specifcation of
the relationship between non-ARRA revenues and lagged expenditures, on
the one hand, and the change in expenditures, on the other. Te FY-2009 level
of pension-fund assets, either in combination with year-on-year changes or
by itself, had the expected positive sign but was statistically insignifcant.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 16 .
functional fin ance and the
composition of aggregate demand
A family which takes its mauve and cerise, air-conditioned, power-
steered and power-braked automobile out for a tour passes through
cities that are badly paved, made hideous by litter, blighted buildings,
billboards, and posts for wires that should long since have been put
underground . . . Tey picnic on exquisitely packaged food from a
portable icebox by a polluted stream and go on to spend the night at
a park which is a menace to public health and morals. Just before
dozing of on an air mattress, beneath a nylon tent, amid the stench of
decaying refuse, they may refect vaguely on the curious unevenness
of their blessings.
—john kenneth galbraith
Equality of the marginal social benefts from the diferent directions
of permissible spending is a necessary condition for the optimum use
of the resources that can be employed . . .
What many people fnd most disconcerting . . . is the complete
disregard for, and even the absence of any reference to, the principle
of balancing the budget.
—abba lerner
How large a role should the government play in the economy? Or, rather,
what should be the composition of aggregate demand—how much private
Copyright © 2021. Harvard University Press. All rights reserved.
consumption, how much investment, how much devoted to the government
purchase of goods and services? What is the right level of transfer payments?
And how much of the government budget should be fnanced out of taxes,
how much from debt?
We saw in chapter 14 that even if there are limits to how much investment
is forthcoming, full employment can, in principle, be achieved without violat-
ing a commitment to sound fnance, that is, while honoring the principle of a
balanced budget. It is possible to do this not only over the cycle, in which
surpluses in fat years balance defcits in lean ones, but, even if only in theory,
at every moment of time.
594
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 595
We also saw that there are good reasons for rejecting this theoretical pre-
scription in practice, including opposition to the enlargement of government
that would likely take place if the commitments to full employment and
sound fnance were both honored at all times. Yet, it is a fact that the size of
government has grown dramatically over the past century everywhere in the
capitalist world. In the United States, the combined state and federal budget
doubled in a single decade, total government expenditures rising from less
than 10 percent of GDP just before the Great Depression to over 15 percent
on the eve of World War II. Relative to the economy the combined govern-
ment budget doubled again in the intervening years, reaching 33 percent of
GDP in 2018. And the weight of the federal government in the mix doubled,
from one-third in 1929 to two-thirds in 2018.
Growth in the size of government relative to GDP was a staple of the fve
decades from the 1930s through the 1970s, albeit the growth was in transfer
payments rather than in the purchase of goods and services. Systemic defcit
has characterized U.S. fscal policy ever since. For a brief period, during Bill
Clinton’s presidency, the federal budget was in surplus, and the public debt
actually came down, but otherwise the years between Ronald Reagan’s tenure
in the White House and 2018 saw a steady rise in the debt. Federal govern-
ment debt held by the public climbed from 25 percent of GDP to 75 percent.
We now have both large government and a large debt. Willy-nilly, we have
embraced big government at the same time we have embraced continuing
defcits.1
Determining the Optimal Size of Government
Te size of government and continuing defcits are central to the contro-
versy Abba Lerner raised in opposing functional fnance to sound fnance.
Te size of government purchases of goods and services and the volume of
transfer payments at which Lerner’s optimality condition “equality of the
Copyright © 2021. Harvard University Press. All rights reserved.
marginal social benefts from the diferent directions of permissible spend-
ing”—will only by chance be consistent with a balanced budget. And the size
of the defcit will depend as well on the mix of private consumption and pri-
vate investment.
Te constraint posed by sound fnance becomes clear if we pose the size of
government as a choice problem in which the government acts as society’s
agent in selecting one out of all combinations of private consumption (C),
private investment (I), and government purchases of goods and services (G),
which I shall refer to for short as “collective goods.”2 In this exercise we ignore
transfer payments (or, equivalently, consider transfers as negative taxes) even
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
596 Raising Keynes
though to do so obscures the reason why government’s presence in the econ-
omy has grown so large. Society is assumed to have a well-defned social-
welfare function
C I G
W , , ,
YFE YFE YFE
where YFE represents full-employment output, as in chapter 14 taken to be a
unique level of output corresponding to a vertical LS schedule.
I do not mean to suggest that this abstraction represents the actual political
process by which government budgets are determined in the United States or
anywhere else. Not only does it ignore the problems of passing from individ-
ual utility to collective welfare, about which there is a vast “social-choice” lit-
erature;3 even if these problems could be wished away, the most I can claim is
that my model allows us to impute some measure of rationality to the “re-
vealed preferences” of the U.S. government.
To represent social welfare in two-dimensional space, we shall for the mo-
ment assume that I is fxed. With I/YFE = I0/YFE, the choice problem reduces to
C G I0
Max = W , ,
YFE YFE YFE
subject to
C G I
+ =1− 0 .
YFE YFE YFE
Graphically, we can represent the choice problem in Figure 16.1. Te prob
lem resembles the standard problem of a consumer maximizing her utility
subject to a budget constraint, except that here the consumer is “society,”
and the “budget constraint” is given by full-employment output afer pro
viding for a stipulated amount of investment. Te optimum confguration is
Copyright © 2021. Harvard University Press. All rights reserved.
<C*/YFE , G*/YFE >.
We can derive the tax rate t* associated with the optimum from the con-
sumption function. Ignore consumption out of wealth and assume a linear
relationship between C and Y:
C = c(1 − t)Y,
with c the marginal propensity to consume out of disposable income (1 − t)Y.
We then have
C∗
= c(1 − t ),
YFE
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 597
G
YFE
Isoquants of equal social welfare
G∗
Full-employment constraint
YFE I
C G
+ = 1− 0
YF E YF E YF E
C∗ C
YFE YFE
Figure 16.1 Maximization of social welfare subject to an investment constraint.
and if we solve this equation for the optimal level of the tax rate, we obtain
C∗
t∗ = 1 − .
cYFE
Te key point for the theory of functional fnance is that only by chance
will t* be consistent with sound fnance. Tat is, only by chance will t* satisfy
the equation t* = G*/YFE.
Tis is easily seen to be the case by superimposing the balanced-budget
Copyright © 2021. Harvard University Press. All rights reserved.
constraint, G/YFE = t, where = Y/YFE, on Figure 16.1, as in Figure 16.2. Te
horizontal intercept is given by setting G/YFE = 0, and therefore t = 0. Absent
consumption out of wealth, output at the intercept is equal to the product of
the multiplier and investment, so that
Y 1 I0
≡ = .
YFE 1 − c YFE
We also have C = cY, which gives
C c I0
c = = .
YFE 1 − c YFE
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
598 Raising Keynes
G
YFE
G
= t
YFE
Isoquants of equal social welfare
B
A Full-employment constraint
C G I
+ = 1− 0
YF E YF E YF E
C c I0 C
=
YF E 1 − c YF E YFE
Figure 16.2 Maximization of social welfare subject to two constraints: full
employment and sound fnance.
As drawn in Figure 16.2, the sound-fnance constraint is a vertical line.
Tis follows from the logic of the balanced-budget multiplier: purchases of
goods and services by the government add to output, but, when balanced by
taxation, do not add to private consumption. (Tat is why the balanced-
budget multiplier is one.)
Lerner’s contention that sound fnance should not be a constraint on fscal
policy—see the second epigraph to this chapter—is illustrated by the difer-
Copyright © 2021. Harvard University Press. All rights reserved.
ence between the constrained optimum without the sound-fnance constraint,
point A, and the optimum with this constraint, point B: the sound-fnance
constraint requires society to settle for a lower level of welfare. Figure 16.3
represents the impact of sound fnance by demarcating the subset of choices
that are available—the shaded area of the diagram—when society commits to
a balanced-budget constraint.
Balancing the budget is not all there is to the doctrine of “sound fnance.”
Low taxes have been part of the sound-fnance rhetoric since the barons of
medieval England insisted that the king “should live on his own.” And the
low-tax party has become ever more vocal. Milton Friedman, for one, always
insisted that achieving a balanced budget was less important than reducing
spending and taxes:
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 599
G
YFE
G
t
YFE
Surplus Deficit
Full-employment constraint
C G I
1 0
YF E YF E YF E
C c I0 C
YF E 1 c YF E YFE
Figure 16.3 Sound fnance restricts society’s choices to the shaded area.
I have written repeatedly that while I would prefer that the budget be bal-
anced, I would rather have government spend $500 billion and run a defcit
of $100 billion than have it spend $800 billion with a balanced budget.
(1983)
Te usual justifcation is a supply-side argument about distortions introduced
into supply schedules by high taxes, particularly disincentives to work and to
invest.4 Te analytics of “starving the beast” as sound fnance are in appendix
Copyright © 2021. Harvard University Press. All rights reserved.
1, below.
We can also approach the choice problem by assuming that C is fxed and
the full-employment trade-of is between private investment and collective
goods, in other words, that the social-choice problem is to maximize
C G I
W 0 , ,
YFE YFE YFE
subject to
I G C
+ =1− 0 .
YFE YFE YFE
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
600 Raising Keynes
Fixing C = C0 in turn fxes the tax rate at full employment, since
C0 = c(1 − t)YFE
defnes a tax rate t1 that holds everywhere on the full-employment constraint
in Figure 16.4. Te policy tool controlling the mix of I and G for a given level
of C is not the tax rate but the hurdle rate of interest h.5 A higher level of h
means less investment and thus more room for government expenditures.
Observe that the full-employment constraint may not extend all the way
down to the I-axis; there may be a nonzero minimum level of G required for
full employment (for C/YFE = C0/YFE) if either a liquidity trap or an inelastic
investment demand is a limiting factor.
Although Figure 16.4, like Figures 16.2 and 16.3, refects the balanced-
budget constraint as a possible impediment to optimal choice, the constraint
is now preventing choice to the lef of the balanced-budget constraint. Te
shaded area once again refects the choices remaining afer the imposition of
a balanced-budget constraint.
Finally, we can pose the choice problem as one of specifying a fxed level of
government expenditure, G/YFE = (G/YFE)0, and optimizing among alternative
feasible combinations of C and I, as in Figure 16.5. Te novelty is that the
G
YFE
G
t
YFE
h
Deficit Surplus
Copyright © 2021. Harvard University Press. All rights reserved.
Full-employment constraint
I G C
1 0
YF E YF E YF E
h min
I 1 c C0 I
1 YFE
YF E c YF E
Figure 16.4 Choosing G and I with given C.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 601
balanced-budget constraint is no longer vertical. Taking into account the
change in the tax rate as Y increases, a change in investment drives output up
by a multiplier of 1/(1 − c), and each dollar of output leads to an increase of c
dollars of consumption. Te result is, as is shown in the mathematical appen-
dix to this chapter,
dC c
= .
dI 1−c
Te relationship between the policy tools and the objectives is clear from Fig
ure 16.5: the interest rate determines I, and the tax rate determines C.
We have in efect reformulated the determination of the optimal mix of ag-
gregate demand as a problem of choosing two instruments, <h, t>. Formally,
the problem is
C G I
Max = W (t ), , ( h )
h , t Y
FE YFE YFE
subject to
C G I
+ + = 1.
YFE YFE YFE
I
YFE
h min
Su
Copyright © 2021. Harvard University Press. All rights reserved.
r
pl
G
us
t
YFE
D
ef
ici
t
Full-employment constraint
I C G
1 0
YF E YF E YF E
h
t t C
YFE
Figure 16.5 Choosing C and I with given G.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
602 Raising Keynes
Evidently, functional fnance is not about fscal policy alone, or, rather, fscal
policy cannot be separated from monetary policy in the functional-fnance
framework.
In this framework we can understand a persistent defcit and a large gov-
ernment presence as a preference for private consumption (C) and collec-
tive goods (G), with relatively little weight on private investment (I). Opting
for this mix of aggregate demand implies low taxes (to stimulate consump-
tion) and a large government budget (to provide collective goods). Invest-
ment suffers as long as the economy’s full-employment budget constraint is
also satisfed.
Are Defcits Sustainable?
Functional f nance requires us to address a long- standing debate about
whether or not defcits are sustainable. Once we get beyond clichés,6 two ar-
guments for limiting defcits stand out. Te frst is that defcits lead to unsus-
tainable debt.
Tat continuing defcits lead to growing debt is incontrovertible. Tat the
debt is unsustainable is much more complex. Te ratio of the nominal debt,
D, to nominal GDP, and thus the sustainability of the debt, depends on four
factors:
d = the ratio of the primary defcit to GDP (the primary defcit leaves out
interest on the debt),
= the nominal rate of interest on the debt (conceptually distinct from
the hurdle rate h; in practice, an average of rates on debts of diferent
maturities),
g = the rate of growth of nominal GDP,
t = taxes as a fraction of private income (GDP + D).
Copyright © 2021. Harvard University Press. All rights reserved.
Imagine that defcits begin today, and that the primary defcit continues to be
a constant proportion of GDP. Assume the rate of interest and the rate of
growth are also constant over time. Te mathematical appendix to this chap-
ter shows that the ratio D/GDP converges to a fnite limit provided the rate of
growth of GDP is greater than the interest on the debt net of taxes. With con-
tinuous compounding of interest and growth, this limit is given by the follow-
ing formula:
D 1
lim =d .
GDP g − (1 − t )
Te assumption that g exceeds (1 − t) is realistic for the United States over
most of its history (see appendix 2, below). (Te formula has to be modifed if
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 603
the interest on the debt is not included in personal income and taxed as such;
see appendix 3.) If (1 − t) > g, the debt ratio D/GDP becomes infnite over
time.
Do Defcits Crowd Out Investment?
A second argument against functional fnance is that long-term defcits crowd
out investment. At full employment, there are evidently trade-ofs between
government expenditure, investment, and consumption, but this does not
mean that defcits necessarily crowd out investment. With
D
C = c(1 − t ) Y + ,
P
where P is the price level, the aggregate-demand equation
Y=I+G+C
becomes
D
Y = I + G + c(1 − t ) Y +
P
or
D D D
I + G+ − t Y + = (1 − c)(1 − t ) Y + ,
P P P
which in words is
Investment + Defcit = Private Saving.
In the presence of government expenditure, taxation, and public debt, aggre-
gate demand is no longer given by the equality of desired investment and sav-
ing, as in the models of chapters 3 to 10. Te demand for private saving is now
Copyright © 2021. Harvard University Press. All rights reserved.
the sum of private investment demand and the government defcit, Invest-
ment + Defcit, whereas the supply of private saving is the saving propensity,
1 − c, multiplied by the fraction of personal income lef afer the taxman
has taken his bite, 1 − t, multiplied in turn by the level of personal income,
Y + (D/P).
If you take the right-hand side of the equation, Private Saving, as fxed,
then a rise in Defcit on the lef-hand side means that Investment must de-
cline by an equal amount. Crowding out of investment is one for one.7 In
the mainstream story, however, consumption as well as investment will be
crowded out if saving responds positively to changes in the hurdle rate of
interest, now labeled h. Recall the picture from chapter 2, revisited in appen-
dix 2 to chapter 3, and reproduced below as Figure 16.6. Introducing a defcit,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
604 Raising Keynes
Interest rate ( h )
Supply of saving
h0
Investment demand
I0 Investment, Saving
Figure 16.6 Investment demand and saving supply.
as in Figure 16.7, causes private investment to fall from I0 to I1, less than the
amount of the defcit.
In a Keynesian approach to defcits, the economy’s response will depend on
the dynamics of adjustment, instead of the interest rate automatically rising to
curtail investment demand to the extent necessary to accommodate the gov-
ernment defcit. To begin with, we will focus on fxprice adjustment and ex-
amine two diferent cases, the frst with saving determined by the distribution
of income between wages and profts, as in the Cambridge saving theory, and
investment demand determined solely by the rate of interest; the second case
Copyright © 2021. Harvard University Press. All rights reserved.
assumes a uniform propensity to save with capital-widening investment. We
then look at fexprice adjustment.
In the frst case, the starting point is the serendipitous equilibrium at E in
Figure 16.8. Te mathematical appendix shows that this equilibrium is stable
provided prices adjust more rapidly than wages. (We ignore the other equilib-
rium involving super-full employment because it is unstable regardless of ad-
justment speeds.) Assume that the government budget is in balance at E,
government purchases of goods and services just equal to tax revenues.
Suppose now an increase in government purchases of goods and services
or a decrease in the tax rate on p rofts moves the AD schedule upward, shif-
ing the equilibrium from E to F. (For F to be stable requires not only that
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 605
Interest rate ( h )
Deficit Supply of saving
1h
h0 Investment + government deficit
Investment demand
I1 I0 Demand for and
supply of saving
D D
I1 G t Y
P P
Figure 16.7 Te mainstream view of crowding out.
P W 0
W Stationary real price
Supply of labor
Supply of goods P 0
P 0 W
Aggregate demand
Y 0
F
E
Copyright © 2021. Harvard University Press. All rights reserved.
YFE Y
Figure 16.8 Fixprice response to an increase in G or a decrease in t with Cambridge
saving.
prices adjust more rapidly than wages, as at E, but also that output adjusts
more slowly than prices and wages.) Te additional spending means an excess
of expenditure over income, which has the immediate efect of making goods
disappear from the shelves faster than they are being replenished. Under a
fxprice regime, output and employment increase because of the additional
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
606 Raising Keynes
demand.8 Tis expansion of output decreases proftability and private sav-
ing and so adds to the defcit. But there is an ofsetting, positive, efect on
profts due to the increase in the equilibrium real price. Te accompanying
redistribution of income from wages to p rofts more than ofsets the nega-
tive impact of the increase in output. It is an increase in the rate of p roft
rather than an increase in the rate of interest, and it is a reduction in con-
sumption rather than a reduction in private investment, that accommodates
the defcit.
Instead of assuming that the distribution of income accommodates the
defcit, suppose the propensity to save is uniform across the population, and
that investment responds positively to the rate of p roft, as described in chap-
ter 10 for capital widening. Also assume that wages respond so much more
quickly than prices that the stationary real-price locus is practically vertical.
Te picture is in Figure 16.9.
Once again, producers respond to the increase in expenditure by increasing
output. As output expands beyond the GS schedule, producers raise prices,
but only marginally. Workers, however, are more than able to demand higher
wages. So the real price falls. Tis combination of increasing output and fall-
ing real price continues until expenditure and income are in balance at H; the
real price and the proft rate continue to fall, and this in turn makes invest-
ment unattractive, with the result that output begins to contract. Te adjust-
ment process continues; depending on the relative speeds of adjustment of
wages and output, either the economy goes directly to F, as in Figure 16.9,
Stationary real price
P 0
P
W
W
Supply of goods
P 0
Supply of labor
Copyright © 2021. Harvard University Press. All rights reserved.
0
W Aggregate demand
Y 0
F H
YFE Y
Figure 16.9 Fixprice response to an increase in G or decrease in t
with capital-widening and rapid wage adjustment, I.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 607
or the economy approaches the new equilibrium with cycles that become
smaller and smaller. Te details are in the mathematical appendix.
Te fexprice version of this case has a similar outcome but a diferent tra-
jectory: the initial equilibrium in Figure 16.10 persists even as aggregate de-
mand increases. In a fexprice regime, the response of producers to additional
demand is to raise prices, but in the present case they do so very slowly, and
workers in turn are able to nullify producers’ attempts to raise real prices.
Consequently, output does not change, and the new equilibrium at F coin-
cides with the original equilibrium at E.
Tere is infation and crowding out in all three defcit scenarios—at full
employment increasing one component of aggregate expenditure must crowd
something out—but there is nothing automatic about what gets crowded out.
Investment demand is unafected in Figure 16.8. (It may actually increase: if
the nominal interest rate doesn’t change, the real interest rate is reduced by
the amount of infation.) By contrast, investment falls in Figure 16.9, since the
proft rate is lower at F than at E. Consumption rather than investment bears
the brunt of the infation. In the fexprice scenario in Figure 16.10, it is not
clear what gets crowded out. Figures 16.9 and 16.10 both refect the “forced
saving” that Keynes relied on in the Treatise on Money to adjust expenditure
to income, but Figure 16.10 does not answer the question of who is forced to
do the saving.
Shrinking the government may also lead to a defcit that does not crowd
out investment. Instead of assuming a change in government spending or a
Stationary real price
P 0 Supply of goods
P
W Y 0
W
Supply of labor
Copyright © 2021. Harvard University Press. All rights reserved.
0 E
W Aggregate demand
F P 0
YFE Y
Figure 16.10 Flexprice response to an increase in G or decrease in t
with capital-widening and rapid wage adjustment, I.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
608 Raising Keynes
change in taxes, consider a decrease in government spending accompanied by
a decrease in the tax rate, with the reduction in taxes tailored to generate just
enough additional consumption spending to keep the economy on the full-
employment frontier. In this case, the reduction in taxes must exceed the re-
duction in government purchases of goods and services. Paradoxically—the
logic is the logic of the balanced-budget multiplier—this will increase the
defcit but will not crowd out any private investment because private saving
will increase by the same amount as the defcit. In the equation
D D D
I +G+ − t Y + = (1 − c)(1 − t ) Y + ,
P P P
the right-hand side is no longer fxed, and it is the accommodation of Private
Saving to Defcit that prevents the crowding out of Investment. Diferentiat-
ing this equation with respect to t tells what must happen to the tax rate, as
well as to levels of consumption, the defcit, and private saving, for the econ-
omy to stay on the full-employment frontier when the level of government
spending changes. We have
dt 1
= ,
dG D
c Y +
P
dC
= −1,
dG
dDeficit 1−c
=− ,
dG c
dPrivate Saving 1−c
=− .
dG c
Te defcit increases as G falls, but this is entirely due to the decrease in taxes
Copyright © 2021. Harvard University Press. All rights reserved.
necessary to stimulate consumption. If taxes were not cut, the assumed de-
crease in government spending would depress the economy.
Rhetoric vs. Reason in Debating Fiscal Policy
In contrast with what happens at full employment, crowding out does not
take place at all when the economy is operating inside the full-employment
constraint. In this case, increasing the defcit may if anything crowd in invest-
ment demand. Moving from inside the choice set to the full-employment
frontier will likely enhance the probability that new capacity will be utilized,
and this will make capital widening more attractive. Te mainstream logic re
fected in Figures 16.6 and 16.7 is not only faulty but would be inapplicable in
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 609
a stagnant economy even if it were correct for an economy operating on all
cylinders.9
Rhetoric has more than made up for the gaps in the logic of the main-
stream position. Te blue- ribbon commission empanelled by the White
House in the wake of the post-2008 debt explosion set the tone:
Federal debt this high is unsustainable. It will drive up interest rates for
all borrowers—businesses and individuals—and curtail economic growth
by crowding out private investment. By making it more expensive for entre-
preneurs and businesses to raise capital, innovate, and create jobs, rising
debt could reduce per-capita GDP, each American’s share of the nation’s
economy, by as much as 15 percent by 2035. (National Commission on Fis-
cal Responsibility and Reform 2010, p. 12)
Te deleterious efects on investment and thus on growth became a central
theme of the opposition to stimulus and defcits. Paul Ryan, then Chair of the
House Budget Committee, began his alternative budget plan with a frighten-
ing picture of the future if defcits were to continue:
Te only solutions to a debt crisis would be truly painful: massive tax in-
creases, sudden and disruptive cuts to vital programs, runaway infation, or
all three. Tis would create a huge hole in the economy that would be exac-
erbated by panic. (House of Representatives 2011, p. 21)
If the apocalypse d
oesn’t materialize, defcits would kill the economy slowly:
Even if high debt did not cause a crisis, however, the nation would still be
in for a long and grinding period of economic decline if it stayed on its cur-
rent path. A recent study completed by [Carmen] Reinhart and economist
Ken Rogof of Harvard confrms this common-sense conclusion. Te study
found conclusive empirical evidence that total debt exceeding 90 percent of
the economy has a signifcant negative efect on economic growth.
Te study looked specifcally at the United States, focusing on growth and
Copyright © 2021. Harvard University Press. All rights reserved.
infation relative to past periods when this nation has experienced high debt
levels. Te study found that not only is average economic growth dramati-
cally lower when gross U.S. debt exceeds 90 percent of the economy, but in
fation also becomes a problem.
Essentially, the study confrmed that massive debts of the kind the nation
is on track to accumulate are associated with “stagfation”—a toxic mix of
economic stagnation and rising infation. (Ibid., p. 21)
If we look at the Reinhart–Rogof study (2010), the “conclusive empirical evi-
dence” is actually less than compelling. It is in fact not evidence at all. Te key
Reinhart–Rogof chart, reproduced below as Figure 16.11, encapsulates the
case that “economic growth [is] dramatically lower when gross US debt ex-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
610 Raising Keynes
5 7
GDP growth (bars, left axis)
Inflation
Debt/GDP (line, right axis)
below 30% Debt/GDP
4 30% to 60% Debt/GDP 6
60% to 90%
3 5
Growth rate
2 4
Inflation
1 3
Average Median
0 2
Average Median Average Median Average Median
−1 1
−2 Debt/GDP 0
above 90%
Figure 16.11 U.S. growth and infation at various debt levels. Source: Reinhart and
Rogof 2010, p. 10.
ceeds 90 percent of the economy.” Te bars show mean and median growth
rates for various ranges of the debt:GDP ratio. Tis ratio appears to have little
efect on growth when it ranges between 0 and 0.9, especially in the range 0.3
to 0.9, where the debt ratio has spent most of the time since World War II. But
once the debt:GDP ratio goes beyond 0.9, growth turns south, sharply.
Reinhart and Rogof ’s data set consists of the entire historical record of U.S.
Copyright © 2021. Harvard University Press. All rights reserved.
debt and output, 216 observations in all. As Reinhart and Rogof note
Te number of observations for the four debt groups are: 129 for debt/GDP
below 30%; 59 for debt/GDP 30 to 60%; 23 observations for debt/GDP 60 to
90%; and 5 for debt/GDP above 90%, for a total of 216 observations. (2010,
p. 10)
Look at Figure 16.12, which includes both the debt held by the public (the
usual measure of the government debt) and the total debt including the por-
tion held by the Social Security Trust Fund and the Federal Reserve. Total
debt is Reinhart and Rogof ’s preferred measure (though social-security debt
is certainly a case of owing ourselves). Te total-debt ratio is indeed in excess
of 90 percent for fve years (six by my count), but these constitute fve or six
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 611
120
110
100
90
80
70
60
50
40
30
20
1940 1950 1960 1970 1980 1990 2000 2010
Gross debt Publicly held debt
Figure 16.12 Debt as percentage of GDP (1939–2018). Sources: BEA, CEA, St.
Louis Fed. myf.red/g/pYBJ
observations only because time is divided into years. If we divided time into
days, we would have almost two thousand observations.
However many observations we reckon there are, there is only one episode,
the last years of World War II and its immediate afermath—1944 to 1949—
bounded by the vertical lines in Figure 16.12. Real growth was low, even
negative, afer the war. Te economy was demobilizing in 1945 and 1946, and
there was a brief recession in 1949. Along with low growth, infation was an
inevitable result of fnancing the war with debt and utilizing price controls
and rationing to balance demands and supplies of many goods. Afer years of
Copyright © 2021. Harvard University Press. All rights reserved.
privation during depression and war, people were fush with the wealth accu-
mulated during the war, wealth on which there was little to spend. Respond-
ing to public pressure to remove price controls immediately afer the end of
hostilities but before conversion to peacetime production d idn’t help matters.
Given these specifcs, it is hard to see that this episode has any lessons for the
consequences of peacetime defcits and debt, especially during periods of
slack capacity utilization.10
Is There a Moral Argument Against Defcits?
Tere has been no shortage of argument about defcits that, Micawber-like,
ignore the diference between governments and private entities. In two books,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
612 Raising Keynes
Ben Friedman has presented a more challenging argument for reducing gov-
ernment defcit and debt, an argument that I applaud but ultimately reject.
My applause is for Friedman’s implicit endorsement of the basic premise of
functional fnance, namely, that the mix of aggregate demand is a matter of
social choice that requires serious political deliberation:
Citizens in a free society, and especially one as rich as ours, should have the
right to choose how much of their income to devote to strictly private pur-
suits and how much of it to spend on public goods like defense or highways
or parks or law enforcement. (1988, p. 24)
Friedman argues that there are moral reasons for choosing a high-investment
aggregate demand, which means that either collective goods or private con-
sumption has to give way. Tis has implications for the defcit.
Te argument has two parts, made in two books separated by a decade and
a half. Day of Reckoning, published in 1988 at the tail end of the Reagan ad-
ministration, emphasizes what Friedman regards as a tacit but nonetheless
morally binding intergenerational pact, one that Reagan broke by running
defcits and piling up debt that shifed the composition of aggregate demand
from investment to consumption:
Te radical course upon which United States economic policy was launched
in the 1980s violated the basic moral principle that had bound each genera-
tion of Americans to the next since the founding of the republic: that men
and women should work and eat, earn and spend, both privately and col-
lectively, so that their children and their children’s children would inherit a
better world. Since 1980 we have broken with that tradition by pursuing a
policy that amounts to living not just in, but for, the present. We are living
well by running up our debt and selling of our assets. (p. 4)
By now it is clear that this sense of economic well-being [enjoyed in the
1980s] was an illusion, an illusion based on borrowed time and borrowed
Copyright © 2021. Harvard University Press. All rights reserved.
money. Jobs are plentiful and p rofts are high because we are spending am-
ply, but more than ever before what we are spending for is consumption. . . .
Our afer-tax incomes are rising because we are continuing to receive the
usual variety of services and benefts from our government, but we are not
paying the taxes to cover the cost. . . .
Te trouble with an economic policy that artifcially boosts consumption
at the expense of investment, dissipates assets, and runs up debt is simply
that each of these outcomes violates the essential trust that has always linked
each generation to those that follow. We have enjoyed what appears to be a
higher and more stable standard of living by selling our and our children’s
economic birthright. (p. 5)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 613
Our unprecedented splurge of consumption fnanced by borrowing has bro-
ken faith with the future. (p. 6)
Continued government spending not matched by the taxes to restrain pri-
vate consumption is eroding the material basis for American society as we
know it. What would America be like—what would Americans be like—
without the fact, and consequently the idea, of progress? How long will it
take before the rigidity, complacency, and mediocrity characteristic of eco-
nomically stagnant societies set in? Which of our open and democratic in-
stitutions will survive the more fractious disputes over shares of a national
product that is not increasing? (p. 10)
A policy that intentionally provides too little tax revenues to pay for the cost
of government, as determined by the voters, while accumulating ever greater
debt both at home and abroad amounts to willful bankruptcy: the deliberate
curtailment of our economic growth for fear of how the next generation of
Americans will use it. (p. 24)
We have “broken with . . . tradition,” lived “an illusion based on borrowed
time and borrowed money,” followed “an economic policy that artifcially
boosts consumption,” and by an “unprecedented splurge of consumption f
nanced by borrowing . . . broken faith with the future.” And it will lead, in the
end, to the “rigidity, complacency, and mediocrity characteristic of economi-
cally stagnant societies,” not to mention “willful bankruptcy.”11
Friedman’s Moral Consequences of Economic Growth, published in 2005 at
the beginning of George W. Bush’s second term, expands on a theme he only
hints at in Day of Reckoning:
Periods of economic expansion in America and elsewhere, during which
most citizens had reason to be optimistic, have also witnessed greater open-
ness, tolerance, and democracy. (p. 9)
Copyright © 2021. Harvard University Press. All rights reserved.
Tese positive externalities of growth mean that we must go beyond the cal-
culus of private benefts and costs:
To the extent that economic growth brings not only higher private incomes
but also greater openness, tolerance, and democracy—benefts that we value
but that the market does not price—and to the extent that these unpriced
benefts outweigh any unpriced harm that might ensue, market forces alone
will systematically provide too little growth. (p. 15)
It helps to separate three major components of Friedman’s argument. First,
we owe future generations the same growth in material well-being that past
generations (Friedman’s and mine in particular) have enjoyed. Second, be-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
614 Raising Keynes
yond the economic calculus, positive externalities for the social and political
fabric of the nation mean that growth, and hence investment, are even more
important than would be the case if we were to deliberate solely on the basis
of our commitment to our progeny. Tird, defcits undermine this moral
commitment to the future.
Let me frst address the last issue, the role of defcits. I agree with Friedman
on two points. If we take full employment as a hard constraint, more of one
thing must mean less of something else. More consumption—given the level
of government spending—means less investment. In the model of this chap-
ter, higher defcits brought about by lower tax rates and accompanied by
higher interest rates are the only way of efectuating the substitution of con-
sumption for investment while maintaining both full employment and the
level of government spending. By the same token, if we hold consumption
and the tax rate constant, more collective goods also mean less investment
and a higher defcit.
But this does not imply the converse, namely, that defcits necessarily mean
less investment. As we have seen, a higher defcit will also result from trading
of government spending for private consumption when the tax rate (and pri-
vate saving) adjusts to keep the economy at full employment. In this case, the
trade-of involves no change in the level of investment. It is simply not true
that “a policy that intentionally provides too little tax revenues to pay for the
cost of government . . . while accumulating ever greater debt . . . amounts to
. . . the deliberate curtailment of our economic growth.”
My basic disagreement with Friedman is on causality, not on the facts. Te
1980s did see a rise in consumption as a share of GDP, and a decline in invest-
ment, while government spending relative to GDP hardly budged. But this
was not the inexorable logic of a defcit putting pressure on saving, and the
rate of interest rising in response to this pressure. It was rather the result of
two policy decisions, the decision to reduce taxes and the decision to main-
tain high real interest rates even afer infation had been tamed. With lower
Copyright © 2021. Harvard University Press. All rights reserved.
interest rates and a reduction in government spending—this was perhaps
Reagan’s plan—the government would still have incurred a defcit but a
smaller one, with no impact on investment and growth.
Te point is not to endorse either what actually happened in the 1980s or
the alternative I am imputing to Reagan. From the point of view of functional
fnance, the confation of the defcit with rates of investment and growth is the
principal issue.
Te moral imperative is less clear than Friedman suggests. I am sympa-
thetic to the idea that the choice between consumption and investment is not
simply a matter of taste, like the choice between vanilla and strawberry ice
cream. And I accept the criticism that the framework laid out in Figures 16.1
to 16.5 elides moral issues.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 615
And Friedman is of course correct that many of us are descended from im-
migrants who came to this country in part, if not entirely, for the opportunity
to provide their children with better economic opportunities than existed in
the old country. Our ancestors worked long hours at dif ficult and unpleasant
jobs so that their ofspring could have the opportunities that we have enjoyed.
We no doubt have a moral debt to our own forefathers and foremothers, just
as we owe a diferent debt to those who were brought here in chains and la-
bored without any realistic hope that their children would do better.
But it is not clear to me that this adds up to a moral imperative to maintain
a high rate of investment and growth. It is not at all clear what our debt to our
ancestors implies about our debt to our own children, just as it is unclear what
our debt to people who endured slavery and its afermath implies about our
debt to their children. I fully endorse Ta-Nehisi Coates’s call for reparations to
the descendants of slaves, defned as “the full acceptance of our collective bi-
ography and its consequences” (2014), but this is a matter of distributive jus-
tice rather than a reason for higher growth. On the count of obligation to our
progeny, whether descendants of slaves, descendants of slaveholders, or nei-
ther of the above, the best I think Friedman can hope for is a Scotch verdict—
unproven.
Does the verdict change when the externalities of growth are brought into
the picture? I do not disagree with Friedman’s assessment of American his
tory, especially post–World War II history. Te growth regime of the immedi-
ate post-war period combined a high rate of growth and a wide distribution
of the fruits of growth, and saw the civil rights movement, second-wave femi-
nism, and the beginnings of liberation from all kinds of gender stereotypes.
Although, as Friedman acknowledges, none of these movements can be ex-
plained solely by economics, I would not disagree that economic prosperity
played a role.
In the late 1970s and early 1980s this “Golden Age” came to a screeching
halt. Ever since, the U.S. rate of growth has been much reduced and its bene
Copyright © 2021. Harvard University Press. All rights reserved.
fts much more skewed toward the very rich. Te slowdown of growth and
the skewed distribution of its benefts fed the Bernie Sanders insurgency in
the Democratic primaries of 2016 and was an element in the coalition that
elected Donald Trump. Indeed the accession of Donald Trump to the White
House is proof enough of Friedman’s basic thesis:
Rising intolerance and incivility and the eroding generosity and openness
that have marked important aspects of American society in the recent past
[Friedman wrote this in 2005!] have been, in signifcant part, a consequence
of the stagnation of American middle-class living standards (p. 9)
Friedman is again right on the facts, and I agree with much of his analysis
of how growth allows greater tolerance, openness, and civility (2005, chap. 4).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
616 Raising Keynes
If we focus on comparing ourselves with our parents, then a growing econ-
omy can make us all winners. And when people are winning, there is much
more to go around. Let growth grind to a halt, or just slow down, and we be-
come more anxious about losing what we have. And diference becomes more
threatening because it unmoors us from what we perceive, through rose-
colored glasses, as a halcyon past. For Friedman these externalities clinch the
argument for growth, and not incidentally the argument for capitalism as the
engine of growth.
Capitalism has indeed proven itself the only economic system that has ac-
tually delivered sustained growth, and, ironically if not unsurprisingly, the
failure to deliver ever-more prosperity for ordinary people has led to a call for
an even hefier dose of the growth medicine. But perhaps this is the wrong
takeaway. Maybe it is time to question the premise of an ever-higher standard
of living as the moral basis for our economy, society, and polity. Is there no
point at which we can say enough is enough?
When Will We Have Enough?
Te short answer is that the more than half a billion residents of the United
States, the European U nion, Japan, and one or two other countries already
have enough, and would have a lot lef over if our wealth were distributed
reasonably evenly. Given the standard of living already reached in these parts
of the world, the further pursuit of consumption has real costs that outweigh
the benefts that Friedman foregrounds.
Te market system is unequaled as a device for ramping up production,
consumption, and material wealth. Even without introducing Friedman’s ex-
ternalities into the reckoning, the good side of this is that enough economic
growth provides the material basis for a dignifed and meaningful life. With a
large enough economic pie, we can meet our absolute needs—which is what
Keynes (1931b) called the needs we have independent of what others con-
Copyright © 2021. Harvard University Press. All rights reserved.
sume—with very modest efort and time, freeing us from economic necessity
and opening up possibilities for self-exploration and self-realization that have
historically been available only to elites.12
But at the same time that capitalism is delivering the goods (literally as well
as metaphorically), it is producing people. More specifcally, markets produce
precisely the attitudes, values, and beliefs that are necessary for success in the
market—self-interest, a focus on the material as the measure of human value,
limitless wants, an ideology of knowledge that denigrates experience in favor
of algorithm (see chapter 10): we must become homine economici in order to
succeed economically.
In short, along with “hard work, diligence, patience, discipline, and a sense
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 617
of obligation” (B. Friedman 2005, p. 17), capitalism has relied on every man
being for himself and leaving the hindmost to the devil. Community and
spirituality, which require very diferent ways of relating than those cultivated
by the market, are prime victims of economic growth. Market ways of think-
ing, being, and doing crowd out the ways of the community and the ways of
the spirit.
Tere is another question: whether or not striving for more and more is
desirable, is it feasible?
Tere are two threats to feasibility. A lack of sources: where will the world
fnd the resources needed for the global economy to expand as it did in the
twentieth century? And a problem of sinks: where do we throw the garbage,
the waste products of growth? Te most urgent problem is carbon emissions.
As part of its preparation for the Paris climate-change summit at the end
of 2015, the International Energy Agency put forward a plan to augment
the existing intentions of the various countries so that emissions would stay
within the (relatively) safe boundary of 450 ppm CO2 (International Energy
Agency 2015). Te problem is that the IEA plan relied heavily on untested
technologies such as carbon capture and storage. If these technologies for de-
carbonization pan out, the world might escape the confict between contin-
ued growth and a safe environment, at least as far as the climate-change
boundary is concerned. If GDP growth can be decoupled from increased en-
ergy use or energy use decoupled from CO2 emissions, we could hope to
achieve emissions targets even with twentieth-century rates of economic
growth.
However, the practical reality in 2020 is that decarbonization on the requi-
site scale would require global rates of improvement in energy systems several
times faster than any historical experience. Te major players—the United
States, China, India, Japan, and the European U nion—would have to go well
beyond their declared intentions in 2015. China and the United States would
have to reduce carbon intensity the most. For China almost 30 percent more
Copyright © 2021. Harvard University Press. All rights reserved.
than the plan its leaders presented to the Paris climate-change summit; more
than 20 percent for the United States. And—most important—the optimists
have to be right about the potential of technological fxes.
Tis was clear before the ink dried on the Paris Agreement. Hardly a year
later, the United States had pulled out of the agreement altogether, and soon
thereafer the Intergovernmental Panel on Climate Change had submitted a
report calling global warming to be held to 1.5° Celsius (2018). Bigger tech-
nological fxes would be necessary.
Even if technology does rescue us on the climate-change front, the growth
regime of the past remains problematic. Continued growth on a global scale
at historical rates may sooner or later run into shortages of raw materials,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
618 Raising Keynes
or, equivalently, sharply increasing costs of raw-material extraction. Indeed,
technological progress on energy efficiency and decarbonization may bring us
more quickly to other ecosystem constraints.
Unlike climate change and other issues surrounding the detritus of growth,
the problem of resource limitations has long been recognized; Tomas Mal-
thus raised the frst warning at the end of the eighteenth century—which
brings up an enduring problem that reprises confict about how to deal with
climate change: the limits of technological fxes.
On the one side is the plain fact that the Earth is fnite and so, therefore,
are the resources available for expanding production and consumption. On
the other side is the plain fact of technological ingenuity, which has con-
founded predictions of resource exhaustion since Malthus’s day. Malthus took
it as self-evident that under the most favorable circumstances, agricultural
output could grow linearly, whereas population increases geometrically:
Population, when unchecked, increases in a geometrical ratio. Subsistence
increases only in an arithmetical ratio. A slight acquaintance with numbers
will shew the immensity of the frst power in comparison of the second.
(1798, p. 4)
In the event, agricultural production has so outstripped population growth
that the more than seven and a half billion of us in 2020 are better fed than
the one billion souls alive in Malthus’s own time, with a much smaller per-
centage of the world’s labor and output. In much of the world, govern
ment policy is concerned with containing food surpluses, not in coaxing out
enough to feed people.
For technological optimists, the lesson is obvious, and so are its applica-
tions: for one, the impending exhaustion of petroleum reserves is always im-
pending, never realized. Table 16.1, produced by James Gwartney and his
collaborators (n.d.) based on data from the website of the U.S. Energy Infor-
mation Administration, illustrates how wild have been the miscalculations of
Copyright © 2021. Harvard University Press. All rights reserved.
those who ignore the prospects for technological improvement.
In 2013, the world had almost thirty times the reserves of a century ago,
Table 16.1 Estimated World Oil Reserves, Usage, and Years to Depletion at Current
Usage Rates
Reserves Usage
Year of Estimate (billion barrels) (billion barrels per year) Years to Depletion
1920 60 6 10
1970 531 14.5 32
2000 1,000 26 38
2013 1,650 32.8 50
Source: From Gwartney, Stroup, Sobel, and Macpherson (n.d., p. 4).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 619
having exhausted all the projected reserves of 1970 in the interim. Similar ta-
bles could probably be constructed for other major resources.
Tere is some irony in the fact that to respect the carbon boundary, it is
likely that much of the existing petroleum reserves, not to mention reserves
as yet undiscovered, will have to be lef in the ground. But there can be little
doubt that since Malthus’s day the technological optimists have generally had
the better of the argument.
Should the pessimists therefore concede? Can we stop worrying about re-
source limits? My own view is that, as Chou En-Lai is supposed to have said
about whether or not the French Revolution was a good thing, “It’s too early
to tell.” Te pessimists argue that depleting resources in the hope that technol-
ogy will save the day is a form of Russian roulette; just because the bullet is
not in the chamber the frst few times you pull the trigger, you have no guar-
antee that you are going to escape unscathed the next time. Indeed, if you play
long enough, you are sure to lose.
We lack suf ficient evidence to reject either the pessimistic or the optimistic
argument. Te important point is that we are in the grip of massive uncertain-
ties—uncertainty about whether extraction technologies can evolve quickly
enough to counter the increasing dif ficulties of access to resources, uncer-
tainty about the true extent of reserves, uncertainty about the technologies
for transforming resources into products. In fact, massive uncertainty is a
common theme whether we are considering sources or we are considering
sinks.
Te question is what to do when you d on’t know and your probability as-
sessments are shrouded in darkness. In one sense the problem is akin to the
problem of determining whether or not to add capacity to manufacture a
product for which the demand is highly uncertain. In this context Keynes
observed (see chapter 10) that we must rely on the innate optimism of entre-
preneurs, their willingness to take a chance when meaningful forecasts are
nonexistent. But even the most optimistic among us might shy away from
Copyright © 2021. Harvard University Press. All rights reserved.
betting the planet on a technological fx or fxes that are at best in the plan-
ning stage and at worst a hope based on the assumption that technology al-
ways comes to the rescue.
Some have concluded that uncertainty is a good reason to do nothing, to
wait until we know more, and then take action. Tis is a counsel of folly.
Rather, prudence requires us to assume the worst with regard to the possibili-
ties for sustainable growth. In countries such as the United States, this eco-
logical argument is bolstered by the lack of any compelling need for more;
quite apart from the question of feasibility, it is dif ficult to make an economic
case for growth.
Te logic of our planetary situation suggests that a new global polity may
emerge in the coming decades—good or bad, beautiful or ugly. In one sce-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
620 Raising Keynes
nario, we will descend into a latter-day version of Hobbes’s war of all against
all, powerful nations fghting for access to the limited sources of materials and
energy for growth and perhaps even for access to the limited sinks into which
to throw out the garbage. In another scenario, we will go forward in apprecia-
tion of what unites us, building solidarity and equality, justice and compas-
sion, quality of human life and ecological fourishing.
You would think this would be an easy choice.
Against these considerations, Ben Friedman’s positive externalities consti-
tute a weak case for growth. On the one hand, the moral obligation to provide
a rising standard of living for future generations pales against the obligation
to leave our children a livable planet. On the other hand, the positive exter-
nalities of growth, what Friedman ofers, in the tradition of Bernard Mande
ville’s Fable of the Bees (1988 [1714]), is a consequentialist argument for an
economy based on devil take the hindmost: the growth that such an economy
can generate allows us to avoid confronting basic problems, for starters rac-
ism, sexism, homophobia. Mandeville, as his subtitle, Private Vices, Publick
Benefts, indicates, discovered a moral paradox, not a moral argument.13
Ernest Gellner put it very well. In Nations and Nationalism, contemporary
capitalism (what he calls industrial society to include the Soviet version of the
modern economy) is presented as “the only society ever to live by and rely on
sustained and perpetual growth” (1983, p. 22). For Gellner, growth is a means
of “buying of social aggression with material enhancement” (p. 22). His char-
acterization, “universal Danegeld,” is apt. Te important point is that the costs
of universal Danegeld in deforming our humanity and threatening our sus-
tainability are simply too high.
For all my disagreement with the particulars of Friedman’s argument, I can
still applaud his call for a national conversation on the appropriate mix of ag-
gregate demand. Tree decades on from when he wrote Day of Reckoning, it is
high time for the reckoning.14 And that reckoning will have to address the
logic of debt sustainability in a low-or no-growth world. In the past, growth
Copyright © 2021. Harvard University Press. All rights reserved.
has held the debt:GDP ratio in check—see appendix 2—but in its absence the
relationship between g and (1 − t) will not favor debt sustainability. We will
have to fnd other means, for example, a progressively larger share of the debt
ending up in the hands of central banks, or substituting collective consump-
tion and higher taxes for private consumption, which may be desirable in its
own right while at the same time keeping the debt in check.
What Happens When We D
on’t Owe the Debt to Ourselves?
Tere remains one more issue: the efect of defcits when we owe the resulting
debt to foreigners. Te data show clearly the increase in foreign holdings of
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 621
U.S. debt, already an issue for Ben Friedman in the 1980s—and certainly an
issue for Keynes as he was contemplating the future of the British economy as
World War II was winding down. As Figure 16.13 shows, as a percentage of
the publicly held debt, foreign ownership rose from about 5 percent in 1970
to 20 percent at the end of the decade, declined in the 1980s as total debt was
rising, only to resume its rise in the 1990s, reaching a maximum of almost
50 percent in 2014. (Te foreign share of the debt has since come down
modestly.)
Foreign ownership is a game changer. For Friedman the consequences cut
to the heart of the role of the United States as a global power:
World power and infuence have historically accrued to creditor countries. It
is not coincidental that America emerged as a world power simultaneously
with our transition from a debtor nation dependent on foreign capital for
our initial industrialization, to a creditor supplying investment capital to the
rest of the world. But we are now a debtor again, and our future role in world
afairs is in question. People simply do not regard their workers, their ten-
ants, and their debtors in the same light as their employers, their landlords,
and their creditors. Over time the respect, and even deference, that America
had earned as world banker will gradually shif to the new creditor countries
50
40
30
20
Copyright © 2021. Harvard University Press. All rights reserved.
10
0
1975 1980 1985 1990 1995 2000 2005 2010 2015
Federal debt held by foreign and international investors
as a percentage of gross federal debt held by the public
Figure 16.13 Foreign held debt as percentage of publicly held debt (1970–2018).
Sources: CEA, Treasury. myf.red/g/pYA9
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
622 Raising Keynes
that are able to supply resources where we cannot, and America’s infuence
over nations and events will ebb. (1988, p. 13)
Contemporary defcit hawks have echoed Friedman’s concerns. Again, Paul
Ryan (before the leopard changed his spots in 2017):
In the end, the debate about rising U.S. debt is not just about dollars and
cents, but also about America’s status as a world power and its freedom to
act in its own best interests. If the nation stays on its current path, interest
payments on the national debt will begin to exceed yearly defense spending
just 11 years from now. In just 16 years, yearly interest expenses will be
double national defense spending.
If it stays on its current fscal path, the United States will be unable to af-
ford its role as an economic and military superpower. Other nations with
very diferent interests will rush in to fll that role. Last year in Foreign Af-
fairs magazine, fnancial historian Niall Ferguson surveyed some of the great
empire declines throughout history and observed that “most imperial falls
are associated with fscal crises. All the . . . cases were marked by sharp im-
balances between revenues and expenditures, as well as dif ficulties with f
nancing public debt. Alarm bells should be ringing loudly . . . [for] the
United States.” (House of Representatives 2011, p. 21)
Friedman and pre-2017 Ryan may be right about the connection between
foreign debt and empire, but foreigners may be forgiven for having doubts
about the benefcence of U.S. dominance on the world stage.
Even some Americans have some doubts about our role as the economic
and military superpower—I count myself among the doubters. Tis is not to
advocate a retreat to isolationism but to argue for a more genuine multilater-
alism than we have contemplated since the end of World War II. Ecological
constraints make a multilateral course of action all the more imperative.
In any case, whereas the problem of foreign ownership of the debt is real,
Copyright © 2021. Harvard University Press. All rights reserved.
its link to the fscal defcit is tenuous. Tere is no question that it is in the
power of government to decrease the trade defcit by decreasing the budget
defcit—if we do not regard full employment as a binding constraint. Decreas-
ing government purchases of goods and services, decreasing transfer pay-
ments (social security, food stamps, unemployment insurance, and the like),
or increasing taxes will reduce disposable incomes, consumption spending,
production, and hence imports. Unless other countries counter this move
with policies designed to discourage imports from our country, our balance
of payments will improve. But at what cost?
Te more interesting, and one would hope, policy-relevant question is what
happens when fscal and monetary policies work in tandem to keep the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 623
economy on the full-employment frontier. In Friedman’s telling, the link be-
tween the fscal defcit and the trade defcit is a corollary of the impact of the
defcit on interest rates: as interest rates rise, not only is investment crowded
out but foreign funds are attracted to the United States, which drives up the
exchange rate and increases imports (1988, pp. 10–11). Te second part of the
logic makes sense in a Keynesian perspective as well as in the mainstream
view, but the frst part is, as we have seen, problematic. One must embrace the
mainstream view of crowding out to accept that an increase in the budget
defcit automatically leads to an increase in the rate of interest.
In the Keynes–Lerner perspective of this chapter, the key point is that if the
economy stays on the full-employment frontier, the composition of aggre-
gate demand afects the trade defcit only to the extent that diferent compo-
nents of demand entail diferent ratios of imports to domestic content. If
consumption is more import intensive than government purchases of goods
and services, then a substitution of private consumption for collective goods
will increase both the budget defcit and the trade defcit. But if consumption
is less import intensive, the budget defcit will increase while the trade defcit
decreases.
Even if he has the causality wrong, Friedman still has an important point
with regard to foreign ownership of U.S. assets: the share of foreign ownership
of U.S. debt is the consequence of persistent trade defcits. Tis puts a very
diferent light on the issue of foreign ownership. Evidently, if the United States
runs current-account defcits, as it has done almost every year since 1970, its
trading partners will end up owning more and more U.S. assets. Given this
inescapable logic of accounting, the question becomes not whether foreigners
will own U.S. assets but which assets.
I think most Americans, myself included, would agree with Friedman’s
preference for foreigners to own U.S. government bonds rather than the news
papers we read, the companies for which we work, or the houses in which we
live (1988, pp. 12–13). But this need not lead us to share Friedman’s view that
Copyright © 2021. Harvard University Press. All rights reserved.
original sin lies in the budget defcit. Te rising share of debt owned by for-
eigners does not change the game for the better, but it is the least of evils in
terms of how foreign-owned assets are deployed. By all means let us attack the
current-account defcit, but until the attack is successful, we should be ap-
plauding a rising share of foreign ownership of the debt, not deploring it.
If it is public policy to contain the trade defcit, whether this be to preserve
empire or to maintain U.S. ownership of U.S. assets, the efect is to add an-
other consideration to the problem of choosing an optimal mix of aggregate
demand. To the extent that the exchange rate responds to demands for U.S.
fnancial assets and these demands are sensitive to the spread between U.S.
and foreign interest rates, interest rates do double duty. Not only do interest
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
624 Raising Keynes
rates regulate investment demand, they regulate the exchange rate, and thus
the balance of payments.15
If we were to take a zero trade defcit as an absolute constraint, the hurdle
rate would have to be set at a level that attracted just enough demand for the
stock of U.S. fnancial assets for the exchange rate to balance the fow of ex-
ports and imports. In this scenario, the interest rate is no longer available
as an instrument for determining investment demand. Respecting the zero
trade-defcit constraint determines investment demand as a by-product.
Imagine, in line with Friedman’s argument, a policy of trade-defcit re
duction that is implemented by decreasing interest rates and the value of the
dollar. Te efect would be to raise investment demand and shif the full-
employment constraint in C × G space inward at the same time as the sound-
fnance constraint is shifed outward. Te picture is in Figure 16.14. As the
two constraints shif, the optimum shifs from A to B.
Te result is to reduce the budget defcit, but this is not the same thing as
saying that the budget defcit causes the trade defcit. In this example, causal-
ity runs the other way, from a deliberate policy initiative to reduce the trade
defcit to a reduction in the budget defcit. An increase in the budget defcit
that results from lower government expenditure accompanied by lower taxes
G
D D
YFE G t Y
P P
Isoquants of equal YF E YF E
social welfare
Copyright © 2021. Harvard University Press. All rights reserved.
Full-employment constraint
A
C G I
1 2
YF E YF E YF E B
Full-employment constraint
C G I
1 1
YF E YF E YF E
C c I2 C
YF E 1 c YF E YFE
Figure 16.14 A balance-of-payments constraint shifs the full-employment and
sound-fnance constraints.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 625
(rather than higher interest rates) would have no such efect. Indeed, a take-
away from Figure 16.14 is that the logic of choice is unafected. Te new opti-
mum at B still represents a mix of output that requires a budget defcit, and
the trade-of between diferent components of the mix is unafected.16
Summary
Functional fnance has two components. Chapter 14 considered the prob
lem of regulating aggregate demand countercyclically, exploring reasons why
compensatory fscal policy takes the form of adjusting the budget defcit
rather than the size of the government sector. Tis chapter looks at the long-
run issue of managing the composition of aggregate demand, in which the
size of the government sector is central.
Te obstacle posed by the constraint of a balanced budget is similar in the
two cases. In the long run as over the business cycle, requiring a balanced
budget removes a degree of freedom. In the long run, only if the stars align
themselves can the economy achieve (1) an optimal combination of collec-
tive goods, private consumption, and private investment; (2) full employ-
ment of the available resources; and (3) a balanced budget. Lerner was right
to argue that once the analogy between government debt and private debt
is exposed as false, it is easy to see which of these three ought to give. We
are lef with an optimization problem of choosing instruments, the hurdle
rate of interest h and the tax rate t, to maximize society’s social-welfare func-
tion W(C/YFE, G/YFE, I/YFE), subject to the full-employment constraint C/YFE +
G/YFE + I/YFE = 1.
Private consumption contributes to our immediate well-being. Private in-
vestment is valued for its contribution to the future availability of goods and
services, for its contribution to growth. Government expenditure on collec-
tive goods is desired for its contribution both to investment and to consump-
tion, through the provision of goods and services more efectively supplied
Copyright © 2021. Harvard University Press. All rights reserved.
collectively than privately, ofen (but not necessarily) because these goods and
services are imbued with an essential characteristic of so-called public goods,
namely, that your use is not rival to mine. All three—G, I, and C—provide
social benefts; all three have opportunity costs. Te trick is, as Lerner states
in his epigraph to this chapter, to equalize the “marginal social benefts from
the diferent directions of permissible spending.”
In this framework, taxes exist not for achieving Mr. Micawber’s defnition
of happiness but for regulating the level of private consumption. With the
hurdle rate of interest regulating the level of investment, the level of govern-
ment purchases of goods and services emerges as the consequence of the full-
employment constraint.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
626 Raising Keynes
Functional fnance faces many challenges. First is the issue of debt sustain-
ability. Te bogeyman of insolvency turns out to be just that, a bogeyman, as
long as we grant the assumption that agents consider government bonds to be
part of their private wealth. Te conditions for sustainability are complicated,
involving the primary defcit, the rate of interest on government debt, the tax
rate, and the growth rate of output. As appendix 2 shows, plausible numbers
suggest a debt to GDP ratio that does not threaten the economy.
A second challenge is the mainstream contention that defcits crowd out
private investment and thereby reduce growth. Tere is a certain truth to this
charge, but not the truth based on the mainstream model, in which defcits
raise the hurdle rate of interest for private investment. Tis model is based on
the category mistake underlying Say’s Law, namely, that the rate of interest is
determined, à la Knut Wicksell, Ben Bernanke, and Larry Summers, by the
demand for and supply of saving; whatever the limitations of the alternative
of liquidity-preference theory, Keynes was on target in rejecting the Wicksell-
ian model. At full employment, government expenditure has to crowd out
something, but whether it is private investment or private consumption that
gives way depends on how the instruments of taxation and monetary policy
are deployed as well as on how the economy adjusts to shocks.
A third issue is the moral argument against defcits. Te version ofered by
Ben Friedman does not appeal to false analogies between private debt and
public debt, but rather to the efects of defcits on the mix of private invest-
ment and private consumption: given the level of government expenditure,
the lower the tax rate, the higher is the defcit and the higher is the level of
consumption relative to investment. For Friedman, high defcits threaten the
basis of political and social stability in the United States, because the Ameri-
can Dream revolves around an increasing material standard of living, for
which high rates of investment have been an essential ingredient.
Friedman raises an important question, one that Micawber-like noise has
prevented us from discussing: to what extent is the American society and pol-
Copyright © 2021. Harvard University Press. All rights reserved.
ity dependent on growth? Put another way, can we ever say enough is enough,
and shif the focus of the American Dream to something other than ever-
increasing material prosperity? Tis question becomes especially salient in
the light of ecological constraints that humankind is wrestling with for the
frst time in our history. My own conclusion is that the moral imperative is for
the rich nations of the world to slow down growth if not bring it to a complete
stop. Depending on the mix of consumption and government spending, this
may imply greater defcits as far out as the eye can see.
Finally, what happens when, as became more and more the case afer 1980,
we no longer owe the debt to ourselves? Even Abba Lerner, for all his commit-
ment to functional fnance, recognized that foreign debt is diferent from do-
mestic debt (1944, p. 305). Te impact of foreign ownership is to enlarge the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 627
scope of the social optimization problem to include trade-defcit reduction
either as a goal or as a constraint. Optimization becomes more complex, but
not diferent in kind.
And though what scares people—foreigners exercising infuence if not con-
trol over domestic economic policy—turns out to be real enough, the solution
of reducing the budget defcit is chimerical. It is the balance of payments that
is the culprit, not the budget defcit. Indeed, as long as the United States con-
tinues to run trade defcits, foreigners will have a larger and larger claim on its
assets. Holding government bonds is likely the most benign way they can de-
ploy these claims!
Especially as we enter an era in which—whether or not by design—growth
may be much less robust than in the past, we must hope that functional f
nance is an idea whose time has come. Both stabilizing the economy over the
cycle and choosing an optimal mix of output are no more likely to be consis-
tent with sound fnance than in the past. And our margin for error may be
much lower.
appendix 1: sound fin ance as starving the beast
If we add a maximum-tax constraint to Figure 16.3, that is, a constraint
t ≤ t0 < t*, we further restrict the choice set. For any t we have
I0 G
+
YFE YFE
=
1 − c(1 − t )
and
I G
c(1 − t ) 0 +
C Y
FE YFE ,
=
YFE 1 − c(1 − t )
Copyright © 2021. Harvard University Press. All rights reserved.
with the result that the slope of any tax-rate constraint as Y varies is constant
dG 1 − c(1 − t )
= .
dC c(1 − t )
Te horizontal intercept is inversely related to the tax rate
I
c(1 − t ) 0
C YFE
= .
YFE 1 − c(1 − t )
Te picture is in Figure 16.15. A tax-rate constraint, like a balanced-budget
constraint, cuts down the feasible choice set, limiting the available options to
the shaded area.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
628 Raising Keynes
G
YFE
G
t
YFE
Surplus Deficit
t t0 c (1 t 0 )(G I0 )
C
YF E 1 c (1 t 0 ) YF E
t < t0
Full-employment constraint
C G I
1 0
YF E YF E YF E
C c I0 C
YF E 1 c YF E YFE
Figure 16.15 A maximum tax rate restricts society’s choices to the shaded area.
Te problem of achieving an optimal government budget is compounded
by the simultaneous imposition of a balanced budget and a maximum tax
rate. Under these circumstances, there may be no way even to reach full em-
ployment, much less a way to reach the unconstrained optimum. In the pres-
ent case, the choice set that satisfes the balanced-budget constraint and the
tax-rate constraint is limited to the shaded area in Figure 16.16; in this setting,
full employment is a distant dream.
Copyright © 2021. Harvard University Press. All rights reserved.
appendix 2: the empirics of debt sustainability
Te postwar period begins with a very high debt to GDP ratio, as Figure 16.12
shows. Te ratio c omes down gradually under Republicans and Democrats
alike and reaches a low in the 1970s. Under Presidents Reagan and George
H. W. Bush, the debt climbs again, but the 1980s trend is reversed under
Bill Clinton. Te presidency of George W. Bush saw a m odest rise in the
debt:GDP ratio until the black days of 2008, when the debt begins to rise
sharply. Tis upward trajectory continued until 2011, with Barack Obama
overseeing the sharpest rate of increase since the Great Depression and World
War II. Te debt:GDP ratio leveled out during Obama’s second term.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 629
G
YFE
G
t
YFE
t
Surplus Deficit
t t0 c (1 t 0 )(G I0 )
Full-employment constraint C
C G I YF E 1 c (1 t 0 ) YF E
1 0
YF E YF E YF E t < t0
C c I0 C
YF E 1 c YF E YFE
Figure 16.16 Te choice set determined by a balanced budget and a maximum tax rate.
Infation coupled with low interest rates whittled away at the debt in the
second half of the 1940s—the high growth rates of nominal GDP in that
period largely refected rising prices rather than more output. Real output
growth and low primary defcits (even some surpluses!) were the reasons the
debt:GDP ratio declined from the 1950s to the 1970s. Higher interest rates
and slower growth reversed the trend in the 1980s, but as interest rates came
down and the primary surplus increased in the 1990s, the debt:GDP ratio
once again fell. In the early years of the new century, higher infation made up
Copyright © 2021. Harvard University Press. All rights reserved.
for some of the fall in the primary surplus, so the debt:GDP ratio remained
relatively stable. In 2008 it all came apart. Despite the lowest rates of interest
since the Great Depression, defcit, low growth, stable prices all combined to
drive the debt:GDP ratio to levels not seen since the 1940s. Since then the
only thing that changed is that the defcit declined markedly and the rate of
growth of the debt:GDP ratio moderated. Te ebb and fow of the various ele-
ments that determine the size of the debt relative to GDP are indicated in
Figure 16.17.
Without claiming any powers of prediction, we can use the values of the
relevant variables in Table 16.2 to see what might be in store for the U.S.
economy with regard to the public debt. Table 16.2 provides values for the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
630 Raising Keynes
20
15
10
Percent
−5
−10
1950 1960 1970 1980 1990 2000 2010
Interest rate Primary deficit relative to GDP
GDP growth rate Tax rate
Figure 16.17 Determinants of debt sustainability (1947–2018). Sources: BEA,
CEA, OMB. myf.red/g/pYzk
relevant variables and the limiting values for the debt:GDP and the debt-
service:GDP ratios, if the data for any one year were reproduced from that
year on, using the formulas
D 1
lim =d
PY g − (1 − t )
and
D 1
lim = d .
PY g − (1 − t )
(Te formulas do not hold for 2009, when the economy contracted with the
Copyright © 2021. Harvard University Press. All rights reserved.
result that (1 − t) exceeded g.)
Tese fgures do not—pace Professor Rogof—suggest that debt unsustain-
ability is a reason for Americans to stay awake at night—at least not as long as
the numbers in Table 16.2 hold up. To be sure, the numbers for the frst years
of the Obama administration would have been reasonable cause for alarm if
there were grounds to believe that the Great Recession and the American Re-
covery and Reinvestment Act would continue indefnitely: a debt to GDP ra-
tio of eight or nine, as in the next-to-last column of Table 16.2, and debt
service in excess of 25 percent of GDP (in the last column) is likely unsustain-
able. But even though growth remained anemic throughout the Obama years,
the numbers in Table 16.2 for the period following 2011 are quite reassuring
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 631
Table 16.2 Limiting Values of Debt and Debt Service
Ratio of Tax Growth Rate Limiting Ratio of Limiting Ratio
Revenues to Ratio of Interest Primary of Nominal Publicly Held Federal of Debt Service
GDP Payments to Debt Defcit GDP Debt to GDP to GDP
t d g D/PY D/PY
2009 0.078 0.035 0.078 −0.018 ∞ ∞
2010 0.085 0.032 0.075 0.038 9.039 0.290
2011 0.095 0.032 0.062 0.037 8.418 0.273
2012 0.097 0.029 0.047 0.042 2.959 0.086
2013 0.104 0.027 0.017 0.036 1.352 0.036
2014 0.108 0.027 0.012 0.044 0.608 0.016
2015 0.111 0.026 0.009 0.040 0.499 0.013
2016 0.108 0.025 0.014 0.027 3.204 0.081
2017 0.103 0.025 0.003 0.043 0.147 0.004
2018 0.095 0.026 0.026 0.054 0.829 0.022
Data source: Bureau of Economic Analysis (various dates). Te primary defcit is measured by the diference
between the total defcit and total interest payments.
(except perhaps for 2016, when economic growth ground to a halt). Even the
2018 spike in the defcit caused by the 2017 tax cut did not signal a huge prob
lem with the debt, because growth picked up dramatically in 2018.
By comparison, in the Reagan era debt unsustainability was a serious issue:
from 1982 until 1986 interest rates exceeded the growth rate of GDP, a rever-
sal of the normal relationship that previously had happened only in times of
recession and then only briefy. Only in the mid-1990s did the combination of
steady growth, lower interest rates, a smaller defcit—leaving out debt service,
the budget was in surplus from 1995 to the end of the century—reverse the
growth of the debt. No wonder Ben Friedman (and many others) were con-
cerned about the rising debt. But the problem was high interest rates, not the
defcit.
Copyright © 2021. Harvard University Press. All rights reserved.
appendix 3: are government bonds private wealth?
and what difference does it make for the
sustainability of the debt?
If the government debt is not equal to zero, the condition for equality between
real government expenditures and tax revenues—the balanced-budget con-
straint—is
D D
G+ t Y +
P = P
,
YFE YFE
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
632 Raising Keynes
G
YFE
G
t
YFE
C c (1 t A ) (G I0 )
YFE 1 c (1 t A ) YF E
Full-employment constraint
C G I
1 0
YF E YF E YF E
C c I0 C
YF E 1 c YF E YFE
Figure 16.18 Serendipitous maximization of social welfare subject to full
employment and sound fnance.
where P is the price of goods. Te question remains as to how private agents
view their interest income (D/P) and their bond holdings D/P. In this ap-
pendix I address this question by exploring what happens when a serendipi-
tous equilibrium, point A in Figure 16.18, is disturbed.
At A, the tax rate t = t *A, which characterizes the optimum A = <C A* , G A* >
given by the solution to the equation
C A∗ (
c(1 − t ) G A∗ + I 0 )
Copyright © 2021. Harvard University Press. All rights reserved.
= = c(1 − t ).
YFE [1 − c(1 − t ) ] YFE
Suppose there is a shock to confdence that reduces maximum investment
demand I1 below the initial level I0. With I1/YFE < I0/YFE, the economy can
continue to provide full employment only if consumption and government
purchases of goods and services take up the slack.
Te sound-fnance constraint and the full-employment constraint adjust as
in Figure 16.19. Te dashed lines represent the new normal of an economy
relying on increased level of government expenditure and private consump-
tion to avoid long-term demand stagnation: as the full-employment con-
straint moves outward, the amount of income the economy can generate
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 633
G
YFE
G
t
YFE
B
A A Full-employment constraint
C G I
A 1 1
YF E YF E YF E
C c (1 t A ) (G
I1)
YFE 1 c (1 t A ) YF E
C c I1 C
YF E 1 c YF E YFE
Figure 16.19 Consequences of a decline in investment demand.
without government expenditures and without taxation—the intercept of the
sound-fnance constraint—shifs inward, as does the constraint refecting the
tax rate at A. All this happens with the public debt still at zero.
What happens to the defcit and the debt depends on what fscal policy the
government follows (on the assumption that monetary policy has been de-
ployed to its limits). If government adheres to the existing fscal policy, which
is to say the combination of G A* and the tax rate t *A, the equilibrium will shif
from A to A′. Tis implies a reduction in C/YFE, but automatic stabilization
Copyright © 2021. Harvard University Press. All rights reserved.
limits the economic damage: A′ is associated with a higher level of income
and output—a higher level of government spending as well as a higher level of
consumption—than A″, which maintains the tax rate t *A while continuing to
enforce the sound-fnance constraint.17 At A′, in contrast with A″, defcits will
become the order of the day, and the debt will rise.
A′ is superior to A″, but A′ also shows the limits of automatic stabilization.
Te new optimum is at B, which is to say that in order to counter the shock to
investment demand, government spending needs to be raised and/or the tax
rate needs to be reduced.
What happens to the debt at B = <C B* , G B* >? Lerner’s answer is that the
debt itself will generate income to bondholders and taxes to the government’s
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
634 Raising Keynes
cofers, and in time this will solve the debt problem. If bonds form part of
private wealth,18 and private income in turn is the sum of output Y and
the real value of the interest income on the public debt (D/P), then con-
sumption is
D
C = c(1 − t ) Y + ,
P
and output is given by
D
G + I + c(1 − t )
Y =C+I +G = P,
1 − c(1 − t )
so that for a given tax rate and given investment we have
D
c(1 − t ) G + I1 +
C P
= .
YFE [1 − c(1 − t )]YFE
Te picture is in Figure 16.20. Te sound-fnance schedule continues to be
vertical because the balanced-budget multiplier still holds. It is also indepen
dent of D/P. Along the sound-fnance constraint
D D
G+ Y + P
P = t ,
YFE YFE
and consequently
I1 D
Y = − ,
(1 − c)(1 − t ) P
c
C = I.
1−c 1
Copyright © 2021. Harvard University Press. All rights reserved.
Tat is, changing the debt (like changing government purchases of goods and
services) does not change consumption on the sound-fnance constraint be-
cause the tax rate changes to ofset any change in the debt.
Holding the tax rate constant at t B* , the consumption schedule moves to the
right with the level of debt because, as the debt increases, private income and
consumption go up; given the full-employment constraint, this means lower
government spending. As the t = t B* schedule moves to the right, t B* becomes
progressively less optimal since the optimal consumption schedule must go
through B. But for any given level of debt, achieving the optimal point B
means a steeper consumption schedule—the solid line in Figure 16.20—
therefore a tax rate t B** higher than t B* . Since G B* doesn’t change, a higher tax
rate means that the primary defcit declines over time.19
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 635
G
YFE D
c (1 tB ) G I1
C P
YFE 1 c (1 tB ) YF E
D
c (1 t B ) G I1
C P
YFE 1 c (1 t B ) YF E
D D
G t Y
P P
YF E YF E
B
Full-employment constraint
C G I
1 1
YF E YF E YF E
C c I1 C
YF E 1 c YF E YFE
Figure 16.20 Te evolution of the tax rate to maintain the optimum mix of C and G.
Lerner went further than asserting that the primary defcit would decline
over time. In a 1943 article laying out the essence of functional fnance, he
argued that the process would end with a balanced budget:
As the national debt increases it acts as a self-equilibrating force, gradually
diminishing the further need for its growth and fnally reaching an equilib-
rium level where its tendency to grow comes completely to an end. Te
greater the national debt the greater is the quantity of private wealth. Te
Copyright © 2021. Harvard University Press. All rights reserved.
reason for this is simply that for every dollar of debt owed by the govern-
ment there is a private creditor who owns the government obligations (pos-
sibly through a corporation in which he has shares), and who regards these
obligations as part of his private fortune. Te greater the private fortunes the
less is the incentive to add to them by saving out of current income. As cur-
rent saving is thus discouraged by the great accumulation of past savings,
spending out of current income increases (since spending is the only alter-
native to saving income). Tis increase in private spending makes it less
necessary for the government to undertake defcit fnancing to keep total
spending at the level which provides full employment. When the govern-
ment debt has become so great that private spending is enough to provide
the total spending needed for full employment, there is no need for any def
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
636 Raising Keynes
cit fnancing by the government, the budget is balanced and the national
debt automatically stops growing. (p. 49)
Te idea of a self-correcting defcit requires that the propensity to consume
rises with wealth because, in Lerner’s words, there is “less . . . incentive to add
to [wealth] by saving out of current income.” Tis assumption runs counter to
the usual hypothesis: as people become richer, the propensity to consume
falls. So a rising propensity to consume is—to say the least—a strong assump-
tion. Perhaps for this reason Lerner abandoned the whole line of argument in
later expositions of functional fnance.20 It is a pity, since even without the
controversial assumption of a rising propensity to consume (maybe especially
without this assumption), the argument might have helped to cement the case
for functional fnance: in a growing economy, government debt may be part
of the solution to the problem of inadequate aggregate demand.
Truth to tell, even if Lerner had assumed a constant propensity to con-
sume, this line of thinking would have had rough going once the rational-
expectations revolution had become the new orthodoxy. A corollary of ratio-
nal consumer behavior, as it came to be understood in the post–World War II
period (see chapter 9) was the polar opposite of the Lernerian argument that
government debt is private wealth: so-called Ricardian equivalence, an argu-
ment identifed with Robert Barro (1974, 1989), holds that the rational agent
will understand that debt service (and principal payments for bonds with f-
nite maturities) will require higher taxes in the future. Tis taxation will ofset
the nominal returns from holding debt. Te upshot is that today’s propensity
to consume will adjust so that the efect of defcit-fnanced expenditure on
demand is exactly the same as the efect of balanced-budget expenditure.
Te logic of Ricardian equivalence is most clearly seen in the event that the
government borrows to ransom King Richard from his Austrian captors. (In
Ricardo’s case it was to fnance the Napoleonic Wars.) Let the ransom be £1
million and assume for simplicity that the Austrians are content to keep the
Copyright © 2021. Harvard University Press. All rights reserved.
£1 million in the bank—a simpler example for present purposes than a war
because of the lack of repercussions of the additional spending, which allows
us to focus on alternative fnancing arrangements. Barro’s argument is that
levying a tax of £1 million today has the same efect on aggregate demand as
foating a consol in this amount and levying taxes every year to pay interest
on the bonds. Te bonds would carry a coupon of 5 percent, and would thus
require a perpetual levy of £50,000 per year to pay the coupon.
According to Ricardian equivalence, agents would recognize the negative
impact of debt service and understand that their real wealth today had been
reduced by £1 million, so the efect of the two options would be identical. Tis
position is easy to understand if the annual tax for debt service were applied
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Functional Finance and the Composition of AD 637
only to the holders of the new bonds, but then of course nobody would buy
the bonds: the tax would reduce the net return to zero, which would make the
bonds worthless! But in Barro’s world the efect on private wealth is the same
if the additional taxes are spread over all forms of income. Even workers with
no holdings of bonds or any other fnancial assets would take a hit to their
wealth as defned by the life-cycle or permanent-income hypotheses—for
these agents wealth is the present value of their future earnings net of taxes—
and would act accordingly.
Whatever the merit of Barro’s argument in the context of agents possessed
of perfect foresight who optimize over an infnite future, there are many prob
lems in applying his ideas to a world in which agents are fnite in so many
respects. One is the fundamental issue of knowledge and belief: do agents
recognize the putative equivalence of taxation and borrowing? Ricardo him-
self had doubts on this score, which led at least one commentator to sug-
gest that Ricardian equivalence should actually be called Ricardian nonequiv-
alence:
But the people who pay the taxes never so estimate them, and therefore do
not manage their private afairs accordingly. We are too apt to think, that the
war [with Napoleon] is burdensome only in proportion to what we are at
the moment called to pay for it in taxes, without refecting on the probable
duration of such taxes. It would be difficult to convince a man possessed of
20,000 pounds, or any other sum, that a perpetual payment of 50 pounds
per annum was equally burdensome with a single tax of 1000 pounds. (Ri-
cardo 1810; cited in O’Driscoll 1976, p. 208)
If agents treat the income from government bonds as an addition to the in-
come generated by the production of goods and services, there can of course
be no equivalence between defcit fnancing and taxation.
Observe that the assumptions about information and information process
ing go beyond those that underlie the life-cycle and permanent-income hy-
Copyright © 2021. Harvard University Press. All rights reserved.
potheses, namely, that consumption is based on wealth rather than on in-
come. We can assume that agents base their consumption decisions on wealth
without committing to the hyper-rationality of Ricardian equivalence; indeed,
the idea of consumption smoothing, the essence of the shif from income to
wealth as the driver of consumption, is perfectly consistent with the argument
of functional fnance. Debt service, D, represents the permanent-income
proxy for bond wealth in the consumption function deployed in Figures 16.19
and 16.20.
Te key diference between my version of Lerner and Barro’s version of
Ricardo is in the treatment of the propensity to consume out of disposable
income and the treatment of the tax rate. For Barro, the propensity to con-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
638 Raising Keynes
sume is a choice variable rather than a behavioral parameter, and the tax
rate does not afect the optimal consumption pattern. For Lerner, and in my
models, the propensity to consume constrains tax-policy choices, and opti
mization dictates both government expenditures and tax revenues, with no
presumption that the two will match—indeed, quite the contrary. Tis is
transparently simple in the choice models in which investment is assumed to
be fxed: the policy choice of an optimal level of consumption C* together
with the given propensity to consume c dictates the tax rate from the equation
C = c(1 − t)(Y + (D/P)), and the optimal level of government expenditure G*
determines the defcit or surplus from G* + (D/P) − t(Y + (D/P)).
For all its abstract appeal—Ricardo was but the frst of a long line of emi-
nent economists who appreciated the intellectual attractiveness, indeed,
beauty, of equivalence between taxation and debt fnance—this idea is in the
end a snare of abstract rationality against which, as I note in the Acknowledg-
ments, Keynes warned us in the preface to Te General Teory.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Sound Finance with G/Y FE Fixed
To derive the slope of the sound-fnance constraint, we calculate dC/dI, taking
into account the relationship between the tax rate and the level of output. We
have aggregate demand
Y=C+I+G
and the sound-fnance constraint
G = tY.
We also have the consumption function
C = c(1 − t)Y.
Because G is fxed, t has to change inversely with Y,
dt
0=t+ Y,
dY
so
dt t
=− .
dY Y
Te slope of the sound-fnance constraint is
dC dC dY
Copyright © 2021. Harvard University Press. All rights reserved.
= .
dI dY dI
Substituting for C and G in the aggregate-demand equation gives
I
Y =
(1 − t )(1 − c)
and
dt dY
(1 − c) I
dY 1 dY dI ,
= −
dI (1 − t )(1 − c) [(1 − t )(1 − c)]2
639
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
640 Raising Keynes
so that substituting −t/Y for dt/dY and solving for dY/dI we obtain
dY 1
= .
dI 1−c
From the consumption function we have
dC
= −cY ,
dt
dC dC dt t
= c(1 − t ) − = c(1 − t ) − cY − = c .
dY dt dY Y
Te result is
dC dC dY c
= = .
dI dY dI 1−c
The Debt:GDP Ratio Over Time
In continuous time the equation for the rate of change of the debt is the difer-
ence between total expenditures, the sum of debt service and spending on
goods and services, D + PG, and current receipts, t(PY + D)
D = D + PG − t (PY + D) = (1 − t ) D + dPY .
Te parameters d = (PG − tPY)/PY, , t, and g are assumed to be constant over
time. Te rate of change of the debt ratio = D/PY is
•
D D (PY )• D D P Y D
γ = = − = − − = [ (1 − t ) ρ − g ] γ + d.
PY D PY PY D P Y PY
For any time , the solution to this diferential equation is
Copyright © 2021. Harvard University Press. All rights reserved.
D
PY
(
(1− t ) ρ − g τ
= 1 − e ) d
g − (1 − τ) ρ
(1− t ) ρ − g τ D
+ e PY ,
0
where (D/PY)0 is the initial debt ratio. It is clear that the ratio D/PY becomes
infnite if (1 − t) > g, and converges to d/[g − (1 − t)] if (1 − t) < g. Observe
that the limiting debt ratio is independent of the initial debt level.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Mathematical Appendix 641
Crowding Out
Te model underlying Figure 16.8, adapted from chapter 9, is
•
P
W
= − GS P − Y − Y − LS P ,
P 1 3 W
W
W
P
−1
Y = θ 2 I ( ρh ) + G − ( s π (1 − t ) + t ) Y − L ,
W
with the defcit assumed to be equal to zero. Te Jacobian is
−θ1GS′ + θ 3 LS′ θ1 − θ 3
−1
J = −2
P
W .
−θ 2 ( s π (1 − t ) + t ) P L −θ 2 ( s π (1 − t ) + t ) 1 −
W FL
At E in Figure 16.8, 1 − (P/W)−1/FL = 0, so tr J is unambiguously negative.
Provided 1 > 3, as in Figure 16.8, we have det J > 0, so that this equilibrium
is stable.
To the right of the intersection of the AD and GS schedules, the sign of tr J
depends on the speed of adjustment of prices and wages relative to the speed
of adjustment of output. If 1 and 3 are large compared to 2, then tr J will be
negative despite the fact that 1 − (P/W)−1/FL is negative. With 1 > 3, det J is
positive if and only if the slope of the stationary-price locus (relative to the
Y-axis) exceeds the slope of the AD schedule, which is to say,
Copyright © 2021. Harvard University Press. All rights reserved.
P
−1
1 − W
1 − 3 FL
> − .
′
1GS − 3 LS ′ P
−2
W L
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
642 Raising Keynes
Crowding Out with a Uniform Saving Propensity and
Capital-Widening Investment
From chapter 10, the fxprice case is characterized by
•
P
W
= − GS P − Y − Y − LS P ,
P 1 3 W
W
W
P
−1
Y = θ 2 ψ Ω(Y ) Y − L K −1 − ρh + G − (s(1 − t ) + t )Y
Y
W ,
−θ1GS′ + θ 3 LS′ θ1 − θ 3
P
−1
J = −2 ψ′Ω 1 − W K −1 .
P L
θ 2 ψ′Ω θ2
FL
W K
−1
+ ψ′Ω′ Y − P L − (s(1 − t ) + t )
K W K
Te combination of a vertical LS schedule and the limiting assumption 1 =
≈ 0 means that the Jacobian takes the form
− ε GS′ ε − θ3
P
−1
J = −2 ψ′Ω 1 − W K −1 ,
P L
θ 2 ψ′Ω
θ2
FL
W K −1
+ ψ′Ω′ Y − P L
Copyright © 2021. Harvard University Press. All rights reserved.
− (s(1 − t ) + t )
K W K
so that both the trace condition and the determinant condition are satisfed at
E and F.
Whether the result is the trajectory in Figure 16.9, in which F is a node, or
the confguration in Figure 16.10, in which F is a focus, depends on the sign
of the discriminant. Ignoring terms in
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
16 • Mathematical Appendix 643
2
P
−1
W Y P L −1
′ −1
∆ = θ2
ψ Ω1 − F K + ψ′Ω′ K − W K − (s(1 − t ) + t )
L
−2
P
L
− 4 θ 3 θ 2 ψ′Ω .
W K
If > 0, which will be the case if 3 is small, then F will be a node, as in Figure
16.9. If < 0, which is to say 3 is large, then F will be a focus, as in Figure
16.21.
Flexprice adjustment is governed by
•
P
W
= θ ψ Ω(Y ) Y − P L K −1 − ρ + G − (s(1 − t ) + t ) Y
−1
W [ ]
1
P
W
P
− θ 3 Y − LS ,
W
P
Y = 2 − FL −1 .
W
Stationary real price
P P 0
W W
Supply of goods
P 0
Supply of labor
0 Aggregate
W demand
Y 0
Copyright © 2021. Harvard University Press. All rights reserved.
F H
YFE Y
Figure 16.21 Fixprice response to an increase in G or decrease
in t with capital-widening and rapid wage adjustment, II.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
644 Raising Keynes
With LS′ = 0 and ≈ 0, the Jacobian is
P
−2
L Y P −1 L
εψ′Ω εψ′Ω′ − − s(1 − t ) + t − θ 3
J = W K
K W K
,
−1
θ2 −θ 2 ( GS′ )
for which the trace is negative and the determinant positive.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. VI .
Keynes in the Long Run
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 17 .
first steps into the long run
Harrod, Domar, Solow, and Robinson
When did the right stock of capital come into existence, and what
mechanism, supposing that it did, keeps accumulation going at the
right rate? Te argument of the General Teory, which shows that
there is no such mechanism in a private-enterprise economy, cannot
be true at each moment of time and yet untrue “in the long run.”
—joan robinson
Keynesian growth theory, like the son of upper-caste Hindus, is twice born, a
fact that may account for the wrong turn taken as a teenager: in the hands of
the mainstream, Keynesian growth theory abandoned the most distinctive
feature of Keynes’s revolution, the insistence on a role for aggregate demand
in determining real economic outcomes. Joan Robinson’s attempt to save
Keynes from the mainstream misfred for the same reasons that Hicks’s mis-
fred—the absolute neglect of the supply side.1 Furthermore, the entire exer-
cise is characterized by the total absence of prices, rather odd for economic
analysis of any kind, but perhaps the natural outcome of the tacit acceptance
of a false premise, namely that Te General Teory was all about a world with
rigid prices and wages.
Chapter 6 touched on the dynamic questions of what happens when labor
Copyright © 2021. Harvard University Press. All rights reserved.
supply and the capital stock are growing over time. Despite investment and
saving, despite population growth, the basic model of chapter 6 assumes that
the capital stock and the labor force remain constant, so that equilibrium can
be characterized by a stationary level of output and real price. Tis model is
pictured in Figure 17.1.
Harrod: The Tension between Growth in Aggregate Demand
and Growth in Labor Supply
Te original attempt to take account of growth in the capital stock and labor
force was Roy Harrod’s “An Essay in Dynamic Teory.” Tis essay was based,
as Harrod tells us, on three propositions:
647
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
648 Raising Keynes
P
W
5
Supply of goods
4
2
Supply of labor
1 Aggregate demand
100 200 300 400 500 600 Y
Figure 17.1 Aggregate demand, goods supply, and labor supply.
(1) that the level of a community’s income is the most important determi-
nant of its supply of saving; (2) that the rate of increase of its income is an
important determinant of its demand for saving, and (3) that demand is
equal to supply. (1939, p. 14)
Tese three propositions are summarized in the growth equation, g = s/v,
which perpetuates Harrod’s name—along with the name of Evsey Domar, the
parent in the theory’s other birth. For Harrod, g is the rate of growth of output
(income); v is the incremental capital:output ratio, the stock of additional
goods (plant, equipment, inventories of goods in process) for each additional
unit added to the fow of output; s is the ratio of aggregate saving to aggregate
output.
Like many important economic formulas—I = S or MV = PY—the Harrod–
Copyright © 2021. Harvard University Press. All rights reserved.
Domar equation is an identity that acquires economic meaning only when we
specify how the separate elements of the formula are determined. Te tautol-
ogy reveals itself immediately if we write out the various terms. Taking time
to be made up of discrete periods (say, a year), we have g = Y/Y, v = K/Y,
and s = S/Y = K/Y. So whatever the values of the individual terms, the equa-
tion g = s/v must hold.
Te tautology becomes a distinctive theory of growth when we distin-
guish the determinants of desired saving and desired investment. Following
Keynes, Harrod’s frst proposition, at its simplest, is that desired saving is
given by SD = sY, with s a fxed parameter. Harrod’s second proposition is a
variation on Keynes: for Harrod it is not the interest rate but expected income
growth that determines, along with v, the demand for additional capital. De-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 649
noting the expected change in income by ∆YE, we can write investment de-
mand as ID = v−1∆YE. (Here, as in most of the growth literature, the focus is on
capital widening; capital deepening is ignored.) Harrod’s third proposition
requires ID = SD.
If we treat v as determined by technology, there is an immediate problem:
there are two unknowns in the equation ID = SD, the level of income and
the expected change. Harrod solves this problem by focusing on expectations
that are in fact realized (rational expectations avant la lettre)—in Harrods’s
words expectations “which, if executed, will leave entrepreneurs in a state
of mind in which they are prepared to carry on a similar advance” (1948,
p. 82). If expectations are realized, we have ∆YE = ∆Y ≡ Y − Y−1, and, assum-
ing Y−1 is given, investment demand, like saving, depends on current income:
ID = vY = v(Y − Y−1).
Te equality of desired saving and desired investment then determines
Y and also converts the tautology g = s/v into a relationship in which causal-
ity runs from right to lef. Te key is that at an income level where ID = SD
both investment intentions and saving intentions are realized. So the actual
saving:income ratio ∆K/Y is equal to the desired ratio s. Harrod reserves the
term warranted rate of growth, denoted gw, for the particular value of g that
corresponds to the values of s and v given respectively by consumer psychol-
ogy and production technology.
Harrod notes two problems. First, the level of Y determined by equality of
desired saving and investment is not a stable equilibrium. In Harrod’s model
there are no prices, so the only available adjustment mechanism is fxprice
adjustment. If the economy fnds itself at a point where ID and SD are unequal,
fxprice adjustment leads the economy away from rather than toward the
point where the equation gw = s/v. To understand this strange result, consider
frst the simple textbook model of aggregate-demand determination laid out
in chapter 3, reproduced below as Figure 17.2. Were output and income to
exceed 400, desired saving would exceed desired investment, which is to say
Copyright © 2021. Harvard University Press. All rights reserved.
that aggregate expenditure would fall short of output and income. Invento-
ries would pile up on store shelves, and, with fxprice adjustment, producers
would get a strong signal to curtail output. Were output and income to fall
short of 400, producers would get the opposite message: expenditure would
exceed income and output, inventories would disappear and order backlogs
would increase. Producers would increase output. (With fexprice adjustment,
the process is more complicated, since in the frst instance unwanted invento-
ries trigger price changes rather than output changes, but since there are no
prices in Harrod’s world, this is a nonissue.) Tus, in Figure 17.2, as far as the
demand side is concerned, 400 represents a stable equilibrium of desired sav-
ing and investment.
Te picture is quite diferent in Harrod’s case. Desired investment is no
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
650 Raising Keynes
ID , SD
500
400
300
200
SD sY
100 ID
100 200 300 400 500 Income (Y)
Figure 17.2 Investment and saving determine aggregate demand, I.
longer fxed in amount, but rather varies with the change in output this year
relative to the previous year: v, by assumption, refects a relationship between
the amount of new capital that businesses must have to accommodate an ex-
pected change in output. With ID = vY = v(Y − Y−1), we have Figure 17.3.
Here the equilibrium is unstable. If output exceeds 400, expenditure will
exceed output and income, moving the economy further away from 400; if
output falls short of 400, expenditure will fall short of output, and output will
contract further.
It may appear that the instability is an accident of the drafsman’s pen in
making the investment schedule steeper than the saving schedule. To be sure,
Copyright © 2021. Harvard University Press. All rights reserved.
instability is the consequence of the fact that the investment schedule cuts the
saving schedule from below rather than from above. But this is not the drafs-
man’s fault. It is, rather, inherent in the logic of Harrod’s ID schedule. Since last
year’s output, Y−1, is positive, the intercept of the investment schedule on the
vertical axis must be positive, that is, ID = 0 where Y = Y−1 (unrealistically as-
sumed to be at a level of 350, implying a warranted growth rate Y/Y−1 =
50/350, or approximately 14 percent). Te investment schedule cannot cut the
saving schedule from above, as would be necessary to ensure stability, without
assuming that the previous period’s income was negative!
Harrod’s second problem is that the warranted rate of growth speaks to the
demand side alone: gw is the rate of growth that keeps the economy on the AD
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 651
ID , SD
500
ID v (Y Y 1 )
400
300
200
SD sY
100
100 200 300 400 500 Income (Y)
Figure 17.3 Investment and saving determine aggregate demand, II.
schedule. Supply enters the theory in what Harrod calls the natural rate of
growth, gn, the rate of growth determined by labor-force growth and technical
change. If N denotes the size of the labor force and we ignore technological
progress, we have gn = N/N. Observe the conceptual diference: gn has, in
principle, nothing to do with gw, which is the rate of growth for which desired
investment and desired saving are equal. In efect, Harrod identifes the sup-
ply side with labor supply, assumed to be a constant fraction of an exoge-
nously given population.
Copyright © 2021. Harvard University Press. All rights reserved.
Tis may seem insuf ficient and even odd, but it is odd only with the beneft
of twenty-twenty hindsight. For some years afer the publication of Te Gen-
eral Teory, the focus was on aggregate demand, and supply was brought in
only in the form of additional constraints on output determination, such as
an upper limit refecting full employment. As noted earlier, price adjustment
was ignored altogether.
Terminological confusion added to the problem. Te terminology em-
ployed by Harrod (and others) distinguished the demand for investment from
the supply of saving, a terminology that facilitated the belief that saving deci-
sions formed the supply side of the economy rather than being an aspect of
demand determination.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
652 Raising Keynes
Be all this as it may, the distinction between warranted and natural rates of
growth served Harrod’s purpose of repositioning Te General Teory in terms
of a changing economy rather than in terms of a short-term static equilib-
rium. Since the natural rate gn is determined independently of the warranted
rate gw, we have to take account of the interaction between the two rates.
If we were to take the warranted rate as the center of gravity of economic
activity, as Keynes meant us to understand the corresponding notion of equi-
librium in his own model, then gw < gn would be a signal of economic stagna-
tion. Saving and investment desires would be satisfed by a growth rate at
which output is expanding less rapidly than the labor force, and if employ-
ment grows at the same rate as output, the imbalance between output growth
and labor-force growth must mean growing unemployment. By the same to-
ken, gw > gn would signal economic exhilaration. Te economy would be
straining at the bit, trying to grow faster than its labor resources allow.
But these are not the conclusions Harrod draws, and for good reason: the
logic of taking the warranted rate as a center of gravity is built on the assump-
tion that gw represents a stable equilibrium of desired saving and desired in-
vestment. If, as in Figure 17.3, the equilibrium is unstable, then the implica-
tions of gw < gn, and gw > gn, are virtually the opposite of those of the previous
paragraph. Consider: if gw < gn, then booms are at least possible. If the econ-
omy fnds itself to the right of the level of output corresponding to gw in Fig
ure 17.3, then expansion will continue until the economy runs up against the
labor constraint embodied in gn. At this point, real expansion must give way
to infation, since the economy can no longer respond to an excess of expen-
diture over income by increasing output. (Needless to say, infation is dif ficult
to model without prices.) Of course, if the economy fnds itself to the lef of
the point corresponding to gw, there is nothing to stop it from running into
the ground.
If, however, gw > gn, then sustained booms are not possible. A boom can
take place only if the actual growth rate exceeds the natural rate, g > gn. But
Copyright © 2021. Harvard University Press. All rights reserved.
g > gw > gn will lead to an explosive increase in g that must end sooner rather
than later in infation without real growth, and gw > g > gn means continuing
decay. Te natural rate gn is a barrier to expansion if g > gw but no barrier to
decline when g < gw.
Clearly, the instability of the saving-investment equilibrium is central to
Harrod’s argument about the interaction of gw and gn. One way around the
instability of gw is to assume v to be a variable rather than a constant, to as-
sume in efect that v is psychologically rather than technologically deter-
mined. In this interpretation, v refects the decisions of capitalists about how
much capital it is appropriate to add in the light of expectations about output,
rather than positing a technologically required amount of capital per addi-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 653
tional unit of output. Technology may still be an important factor in the de-
termination of v, but that is hardly equivalent to making v into an aspect of
the production function.2 If v were to decrease with Y—not an implausi-
ble assumption—then we might have the picture represented in Figure 17.4,
where the unstable equilibrium at 400 coexists with a stable equilibrium at
800. (Don’t worry about the precise magnitudes; the stable equilibrium im-
plies an enormous growth rate, even less plausible than the 14 percent growth
rate implied by the equilibrium of Figure 17.3.)
Allowing v to vary with Y restores the intuitive association of gw < gn
with stagnation, and gw > gn with exhilaration—the same result could be ob-
tained by making the propensity to save, s, vary positively with the level of
income—only if the level of output dictated by the natural rate of growth were
to lie to the right of the frst, unstable, saving-investment equilibrium. If the
full-employment level of output, corresponding to gn, were to lie to the lef of
Y = 400 in Figure 17.5, then even though gw > gn, the economy would fnd it-
self on a downward spiral; unable to sustain output in excess of the natural
rate, the economy would end up in the centrifugal part of the diagram with
nothing to cushion the fall. A qualitatively similar result obtains if we follow
an alternative approach to making gw stable by allowing growth expectations
to be revised by experience.
Whether or not the warranted-rate equilibrium is unstable, the very oppo-
sition between warranted and natural growth rates highlights a problem that
a simple comparative statics of aggregate demand obscures—and thus pro-
vides another reason for questioning the ability of a market system to provide
full employment. Even if the economy were to manage to reach full employ-
ment at any given moment, there is no guarantee that the economy would
evolve in precisely the way required to resolve the confict between an equilib-
rium that satisfes the requirements of investors and savers and an equi
librium that satisfes the requirements of providing a job for every willing
worker in the (growing) labor force.
Copyright © 2021. Harvard University Press. All rights reserved.
We can summarize the results that emerge from Harrod’s model in the
form of the 2 × 2 matrix in Table 17.1, even though Harrod himself focused
on only one column, the lef-hand column, of the matrix. One way or another,
the result, apart from a fortuitous equality between warranted and natural
growth rates, is that the economy either strains at the bit, persistently to the
lef of its AD schedule, or ends up to the right of its AD schedule (or exactly
on the schedule), but in either case further and further from the LS schedule.
An economy that is perpetually straining at the bit, held back by inadequate
labor-force growth, must eventually give itself over to fexprice adjustment,
even though Harrod d oesn’t go there: since Harrod’s economy cannot adjust
forever to an excess of expenditure over income by increasing output, the re-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
654
ID , SD
500
400
300
SD sY
200 I D v (Y Y 1 )
Raising Keynes
100
100 200 300 400 500 600 700 800 900 1,000 Income (Y)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Figure 17.4 Investment and saving determine aggregate demand, III.
17 • First Steps into the Long Run 655
ID , SD
500
400
300
200
SD sY
(I I )D
100 ID
Y
100 200 300 400 500 Income (Y)
Figure 17.5 Warranted growth in the Domar model.
Table 17.1 Growth Determined by Warranted and
Natural Rates
gw unstable gw stable
g w < gn g → gn g → gw
infation unemployment
or
Y→0
unemployment
g w > gn Y→0 g → gn
unemployment infation1
Copyright © 2021. Harvard University Press. All rights reserved.
1. Assuming gn lies to the right of the unstable
equilibrium in Figure 17.3.
sult must be perpetual infation, as the price level is the only means of adjust-
ing excess demand. Te other possibility is long-run unemployment, as the
growth in output and capital, and hence of the demand for labor, lags behind
the growth in the labor force. Harrod’s economy lef to itself ofers no good
outcome.
Observe that there is no role for goods supply based on proftability as a
separate element of the model. Harrod implicitly assumes that actual output
(and employment) grows at the warranted rate unless impeded by a lack of
hands, that is, by a failure of the labor force to grow suf ficiently rapidly to ac-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
656 Raising Keynes
commodate warranted-rate growth. Tere can’t be a GS schedule based on
proft maximization in the absence of a role for prices and wages.
Domar: The Tension between Growth in Aggregate Demand
and Growth in Goods Supply
Goods supply is the heart of the alternative interpretation of the Harrod–
Domar equation provided by Evsey Domar (1946, 1947). But it is not goods
supply responding to p roft maximization. For Domar, as for Harrod, there
are no prices in the model and no p roft maximization. Domar conceives of
goods supply in terms of additions to productive capacity, and in this respect
he is on the same page as Harrod. But capacity is no longer linked directly to
labor-force growth.
Although Domar does not employ the terminology of warranted and natu-
ral rates, we can see both the similarities and the diferences between his
model and Harrod’s by using the warranted/natural framework. In Domar’s
model, the relationship between the warranted rate, the propensity to save,
and the capital:output ratio is, as in Harrod’s model, gw = s/v. But now gw =
ID/ID, and s = SD/Y. Tat is, the warranted rate is the rate of growth of
desired investment, now defned as a rate of growth of investment demand
that is assumed to be given exogenously: think animal spirits and a given
hurdle rate of interest. And the saving parameter is the marginal propensity to
save. (Te diference in the defnition of saving disappears if we assume for
simplicity that saving is a constant fraction of income, so that the marginal
and average propensities to save are equal.)
Te capital:output ratio, v, is formally the same ratio as it is in Harrod’s ap-
proach, K/Y, but it has a diferent meaning. Here it is neither technologi-
cally nor psychologically determined; it is, rather, the multiplier process that
fxes the value of Y: income must change to generate an amount of addi-
tional saving that matches the additional investment I. Te warranted-rate
Copyright © 2021. Harvard University Press. All rights reserved.
equation is now read from lef to right, from the exogenously given growth
rate of investment demand to the change in output, Y. As for Harrod, the
key assumption is that desired investment and saving are equal.
Te picture is in Figure 17.5. Assuming that the economy begins and ends
on the AD schedule, the change in desired saving, SD, must be equal to the
exogenous change in investment, ID. From the expression for s, we can cal-
culate the required change in income, Y; s is simply the inverse of the multi-
plier. Tis fxes v, since SD = ID .
Te natural rate of growth is also conceptually diferent. For Domar the
natural rate of growth is the rate of growth of desired saving that results from
full utilization of the capital stock. Te capital:output ratio fgures once again,
but it is defned diferently, this time as a technologically determined param-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 657
eter relating additional capital, K, to the additional output that can be gener-
ated if the new capital is fully deployed.3 On standard mainstream assump-
tions we have
Y = MPKK + MPLL,
where MPK and MPL are the marginal products of capital and labor. If we set
L = N, which is to say that the entire addition to the labor force is em-
ployed, we can solve for Y, which determines v = K/Y. Te natural rate
follows as
(sY )
S D (sY ) s Y .
gn
SD sY v K
Y
Besides the change in the meaning of v, the natural-rate equation dif-
fers from its warranted-rate counterpart because gn is determined by v—the
equation is read from right to lef—opposite to the direction of causality in
Domar’s warranted-rate equation.
Domar’s equilibrium, like Harrod’s, is defned by equality of the warranted
and natural rates of growth: for them both, this requires not only continuous
equality of desired saving and investment but also continuing equality be-
tween the expansion of aggregate demand and the expansion of productive
capacity. Te diferences lie in what drives aggregate demand and in what
drives supply. Harrod endogenizes investment demand by making it depend
on a capital:output ratio that is based on some combination of technology and
the psychology of the businessmen and -women who undertake the invest-
ment. For Domar, investment demand is a black box.
With regard to supply, Harrod emphasizes labor-force growth, whereas for
Domar the labor force is simply a background variable that afects the growth
of capacity. If goods were produced by robots without involving any human
Copyright © 2021. Harvard University Press. All rights reserved.
labor, Domar’s model would survive unscathed—we would simply set L =
N = 0 in the equation for determining the expansion of productive capacity.
Te parameter v in the natural-rate equation would then be equal to the in-
verse of the marginal productivity of capital.
One equation, two very diferent interpretations. Nevertheless, there is an
important common theme: even if the conditions for a serendipitous full-
employment equilibrium were to be achieved at one point in time, there is no
guarantee that the economy would grow in a way that maintained full em-
ployment. Both formulations point to the precariousness of macroeconomic
equilibrium.
In the late 1940s, when Harrod and Domar launched Keynes into the long
run, concern for maintaining full employment over time was a central ques-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
658 Raising Keynes
tion. But their mutual concern for the precariousness of sustained full em-
ployment faded away pretty quickly. For this there is an empirical reason: the
robust performance of capitalist economies over the three decades following
the end of World War II. At the level of theory, we have Robert Solow to
thank—or to blame, as the spirit moves us.
Solow: Resolving the Tension between Growth in the
Capital Stock and Growth in Labor Supply
Solow’s contribution (1956) was in many ways a breath of fresh air. Confusion
about the meaning of the Harrod–Domar equation was rampant. Te very
fact that the same equation was deployed without any self-consciousness to
summarize two diferent theories is indicative.4 And Solow’s attempt at clarif
cation is both elegant and convincing. Indeed, his paper marks a watershed in
economics, from the discursive literary argumentation of Keynes and Harrod
to the razor-sharp logic of calculus and allied mathematical tools, a major
step in the mathematization of economics that has characterized the years
since.
Solow addresses himself to the opposition between warranted and natural
rates of growth, summarizing the problem in these terms:
Te characteristic and powerful conclusion of the Harrod–Domar line of
thought is that even for the long run the economic system is at best balanced
on a knife-edge of equilibrium growth. Were the magnitudes of the key pa-
rameters—the savings ratio, the capital-output ratio, the rate of increase of
the labor force—to slip ever so slightly from dead center, the consequence
would be either growing unemployment or prolonged infation. In Harrod’s
terms the critical question of balance boils down to a comparison between
the natural rate of growth which depends, in the absence of technological
change, on the increase of the labor force, and the warranted rate of growth
Copyright © 2021. Harvard University Press. All rights reserved.
which depends on the saving and investing habits of households and frms.
(1956, p. 65)
Tese are of course the results summarized in Table 17.1. But Solow sees the
problem not as a problem of insufficient or excessive aggregate demand, as
Harrod (and Domar too in his own way) do, but as the result of a very specifc
assumption about technology, an assumption that is inappropriate once we
are no longer in the short run. Nor does Solow recognize that the instability
of the aggregate-demand equilibrium in Harrod’s model is only one reason
for the precariousness of full employment. According to Solow
Fundamental opposition of warranted and natural rates turns out in the end
to fow from the crucial assumption that production takes place under con-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 659
ditions of fxed proportions. Tere is no possibility of substituting labor for
capital in production. If this assumption is abandoned, the knife-edge no-
tion of unstable balance seems to go with it. (1956, p. 65)
For Solow, the “knife-edge” problem stems from the assumption that the
demand for labor is constrained by the size of the capital stock: on the one
hand, growing unemployment characterizes the economy when the growth of
output and hence the growth of demand for labor fall short of the natural rate
of growth, and, on the other hand, excess demand for labor and infation
characterize the economy when output and hence the demand for labor grow
faster than the supply of labor. Te weak link in this chain, Solow argues, is
the link between output and the demand for labor. If it is possible to produce
a given output with various combinations of labor and capital, there need be
no rigid link between output and employment. In the limiting case of con-
tinuous substitution between capital and labor, we can characterize the rela-
tionship between output and inputs in terms of the so-called neoclassical pro-
duction function, according to which substitution possibilities are refected in
the relative marginal productivities of the factors of production, which are
assumed to vary smoothly with changes in factor proportions.
Tis gives Solow an extra degree of freedom to model the interaction of
goods supply, labor supply, and aggregate demand. He could have used this
degree of freedom along the lines of chapter 6, with explicit assumptions
about the dynamics of goods supply, labor supply, and aggregate demand. In-
stead, following in the footsteps of Modigliani (1944), he simply assumes full
employment and shows that adjustment of the capital:labor ratio will resolve
any discrepancy between gw and gn. In efect, Solow eliminates the AD sched-
ule in Figure 17.1, so that the problem becomes one of ensuring that the equi-
librium of goods supply and labor supply evolves in a way that maintains a
stationary capital:labor ratio. But like the models of Harrod and Domar, there
are no prices in Solow’s model; goods supply does not depend on proft maxi-
Copyright © 2021. Harvard University Press. All rights reserved.
mization, and labor supply does not depend on utility maximization.
Along the way, Solow substitutes his own defnition of the warranted rate of
growth. Solow’s warranted rate is the rate of growth of the capital stock, deter-
mined solely by saving propensities and the output:capital ratio. Te impor
tant change here is not that capital replaces output, but that investment de-
mand no longer plays any role. Solow keeps the defnition of the natural rate
as the rate of growth of the labor force, but with the assumptions of continu-
ous substitution and full employment, the natural rate is now equal to the rate
of growth of employment as well.
Solow takes it as his task to show that the automatic workings of the econ-
omy will bring the warranted and natural rates into line with one another—
no booms or busts, no infation or unemployment. To see how this happens,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
660 Raising Keynes
consider Figure 17.6, which reproduces Solow’s Figure 1 (1956, p. 70). Tis
fgure relates warranted and natural rates to the capital:labor ratio.
We start with an aggregate production function Y = F(K, L). Assume con-
stant returns to scale, so that we can divide both sides of the equation by L
to obtain a relationship between output per worker and capital per worker
Y/L = F(K/L, 1). Defning k ≡ K/L, and f(k) ≡ F(K/L, 1), we can write this
equation more compactly as Y/L = f(k). In Solow’s formulation, gw = K/K
and, since K = sY, the growth of the capital stock per worker is given by the
equation gwk = sf(k). In Solow’s notation the natural rate of growth is n, so the
growth of the capital stock per worker required for the capital:labor ratio to
maintain stationary is gnk.
Te argument is that if gw > gn, the growth in the amount of capital at each
worker’s disposal (gwk) must exceed the growth in the amount of capital
needed to maintain the existing capital:labor ratio (gnk). So k must be increas-
ing over time. By the same token, if the inequality is reversed, and we start
with gw < gn, k must be decreasing. Te capital:labor ratio comes to rest only
when gw = gn. In the picture, the gap between gwk and gnk becomes smaller as
k moves from its initial position, k0 or k1; at the point k*, k is neither increas-
ing nor decreasing, which makes k* an equilibrium to which the adjustment
process converges.5
QED. Which was to be demonstrated.
Or was it?
One problem is that we have changed the meaning of the production func-
tion in applying it to the long run. In the short run, changes in the cap
Required growth to maintain existing g k
n
level of capital per worker
Actual growth
Capital stock growth per worker
g w k sf(k)
Copyright © 2021. Harvard University Press. All rights reserved.
k0 k k1
Capital stock per worker
Figure 17.6 Warranted and natural rates in the Solow model.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 661
ital:labor ratio have been assumed to be the result of changing the level of
utilization of an existing capital stock. Tat is, the capital:labor ratio rises if
there are fewer workers manning the machines, or if the workforce is on short
time. Te capital:labor ratio falls if more workers are employed, or if workers
put in longer hours.
In the long run, the production function represents diferent technologies
rather than diferent intensities of using an existing capital stock; a shif in the
capital:labor ratio now represents a change in the technology by which goods
and services are produced. Te long-run production function raises funda-
mental issues: can we represent the distinct confgurations of capital goods
associated with diferent technologies in terms of an aggregate capital, K;
what meaning attaches to this aggregate when the technology changes? Te
questions multiply: Can we order distinct technologies in terms of an overall
quantity of capital? What is now the meaning of the marginal product of cap
ital, and for that matter, the marginal product of labor? Tese are questions
that have occupied some of the keenest minds of twentieth-century econom-
ics; even afer the dust had settled, the mid-century “capital controversies”
engendered no fewer than four entries in Te New Palgrave published in 1987
(Eatwell, Milgate, and Newman). Nor is the subject over and done with; see
David Rezza Baqaee and Emmanuel Farhi (2019) for more recent discussion.
For all its problems, I embrace the aggregate production function and its
problematic deployment of an aggregate capital stock. For one thing, like
Keynes, I am committed to engaging mainstream economics on its own turf,
of which the aggregate production function is an important constituent. Per-
haps more important is the plain fact that I do not know how to argue about
the long run without an aggregate production function and its aggregate
arguments.
Even within this framework, Solow is problematic. He does indeed dispose
of one of Harrod’s problems, but not the other. And not Domar’s. Figure 17.6
speaks to the convergence of warranted and natural rates of growth under the
Copyright © 2021. Harvard University Press. All rights reserved.
assumption that saving at full employment determines investment and capital
stock growth but ignores the Keynesian problems of adjustment of aggregate
demand that were at the heart of Harrod’s and Domar’s inquiries—shades of
J. B. Say.
Solow is aware both of what he accomplished and of what he had not ac-
complished. Concluding his seminal paper with a section titled “Qualifca
tions,” he begins with this observation:
Everything above is the neoclassical side of the coin. Most especially it is full
employment economics—in the dual aspect of equilibrium condition and
frictionless, competitive, causal system. All the dif ficulties and rigidities
which go into modern Keynesian income analysis have been shunted aside.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
662 Raising Keynes
It is not my contention that these problems don’t exist, nor that they are of
no signifcance in the long run. (1956, p. 91)
But Solow did nothing to address these problems, and the legacy of the paper
was to push the Keynesian issue of aggregate demand to the sidelines. Te
promise of a Keynesian theory of the long run that seemed well within reach
to Harrod and Domar was never fulflled.
A comparison of Harrod, Domar, and Solow brings to mind the Indian tale
of the blind men, each of whom characterizes an elephant according to the
part of the body he happens to touch: as a wall, a pillar, or a rope (to men-
tion only three reports), corresponding to the elephant’s side, leg, or tail. A
nineteenth-century English retelling of the tale concludes
And so these men of Indostan
Disputed loud and long,
Each in his own opinion
Exceeding stif and strong,
Tough each was partly in the right,
And all were in the wrong!6
Embedding Harrod, Domar, and Solow in a Common Model
Te partial nature of each view emerges clearly if we embed Harrod, Domar,
and Solow in the common framework of the model elaborated in chapter 6.
Te starting point is Figure 17.7, which appends stationary real-price sched-
ules consistent with fexprice and fxprice dynamics to Figure 17.1. Te long-
run problem turns out to be the same as the short-run problem: the system is
Stationary real price
P 0 Stationary real price
W
Copyright © 2021. Harvard University Press. All rights reserved.
P P
P 0
W Supply of goods W Supply of goods
W E
5 Y 0 5 P 0
4 4
3 E 3 W 0
W 0
2 2 Supply of labor
Supply of
1 P 0 1 Y 0
Aggregate demand labor Aggregate demand
100 200 300 400 500 600 Y 100 200 300 400 500 600 Y
(a) Flexprice dynamics (b) Fixprice dynamics
Figure 17.7 Equilibrium in the short run.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 663
overdetermined, so even though Harrod, Domar, and Solow each fnds a sat-
isfactory solution to the problem he himself has posed, none of them solves
the problem that the others have posed.
À la Solow, we replace the short-run measure of economic activity, Y, with
a measure more suitable to an evolving economy. Our measure is the inverse
of Solow’s k, namely, the labor:capital ratio l ≡ L/K. We now have Y/K = f(l) ≡
F(1, L/K).
Long-run equilibrium is defned as a state of afairs where both the real
price and the labor:capital ratio, and thus the output:capital ratio, are station
ary. Unlike the level of output, which evidently must change over time as cap
ital accumulates and employment increases, the labor:capital ratio and the
output:capital ratio can remain stationary even as the capital stock and labor
force evolve—at least as long as we ignore technological change.7
Figure 17.8 transposes Figure 17.1 to the new space of real price and
labor:capital ratio. Observe that the GS schedule now refects a rising mar-
ginal cost. (Te linear marginal cost curve that has been deployed up to now
has nothing to recommend it but the drafsman’s convenience.)
Te point of changing the state variable from Y to l is to avoid having to
deal with moving AD, GS, and LS schedules. We do not succeed completely
P
W
Supply of goods
Copyright © 2021. Harvard University Press. All rights reserved.
Supply of labor
Aggregate demand
Figure 17.8 Aggregate demand, goods supply, and labor supply
in the space of l × P/W.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
664 Raising Keynes
until (in the next chapter) we modify the assumption that labor-force growth
is exogenously given.
Consider frst aggregate demand. In the context of an economy in which
output grows over time, investment and saving must grow too. But if we as-
sume that investment demand and saving supply per unit of capital are sta
tionary over time, then aggregate demand per unit of capital will also be sta
tionary. In the simplest version we assume that the hurdle rate of interest h is
the sole driver of investment per unit of capital, which we (for now at least)
denote ID = ( h). Similarly, the simplest saving schedule assumes that a con-
stant proportion of all income is saved, so that we can write SD = sY/K. Ag-
gregate demand is defned by ID = SD and is thus stationary provided that the
investment-demand schedule, ( h), and the propensity to save, s, do not
change over time.
Observe that this defnition of investment elides two problems at once.
First, the instability of Harrod’s equilibrium disappears once investment de-
mand no longer depends on income, for it will be recalled that instability
arose from the greater sensitivity of investment demand to output as com-
pared with the sensitivity of saving. Tis does not, of course, mean that we are
done with Harrod, for instability was only one of his issues. Even with this
issue resolved, there remains the argument that the determinants of the war-
ranted rate and the determinants of the natural rate are distinct and sepa-
rate, and there is no mechanism for bringing the two rates into line with one
another.
Second, Keynes’s own problem of declining investment opportunities is
also elided in this construction. Keynes saw the long run as providing an en-
vironment less and less favorable for investment, for which the only solu-
tion was a progressive diminution in the rate of interest, a diminution that
would eventually lead to the “euthanasia of the rentier” (Te General Teory,
pp. 375–376). Keynes’s view was developed by Alvin Hansen into a vision of
the future in which capitalism is beset by a chronic failure of investment de-
Copyright © 2021. Harvard University Press. All rights reserved.
mand—what Hansen labeled secular stagnation. In turn, investment pessi-
mism became the basis for Abba Lerner’s theory of functional fnance, which
assigned to fscal policy the primary role of compensating for the putative
lack of investment demand, whatever the consequences for the government
defcit and the national debt (Hansen 1938, 1941; Lerner 1944, chap. 24).8
Like aggregate demand, goods supply is invariant over time in the space
l × P/W provided the production function does not change. (Remember, we
are abstracting from systematic technological change, so the only thing that
can change the production function is an isolated shock, something we con-
sider in the next chapter.) Te GS schedule, defned by equality between the
real price and real marginal cost, requires P/W = f ′(l)−1, which, like the equa-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 665
tion expressing the equality of desired saving and investment, is invariant
over time.
Tis leaves the LS schedule. Te natural assumption in the context of an
evolving economy is that an underlying labor-supply function is fxed in the
sense that, for a given real-price level, the ratio of desired employment (LS) to
the labor force (N) does not vary over time as a function of P/W. We write the
functional relationship between LS/N and P/W as LS/N = (P/W). Denoting
the ratio of labor supply to capital by lS ≡ LS/K, we have lS = (P/W) × N/K.
Since the ratio N/K enters into the determination of lS, the LS schedule is not
stationary over time unless the labor force and the capital stock are growing at
the same rate.
If, as is characteristic of depressions, the rate of growth of the capital stock
slows markedly so that N grows proportionately faster than K, the result will
be that lS grows for any given (P/W). Tis is to say that the LS schedule will
move outward over the course of time; Figure 17.9 shows the evolution of
a fexprice equilibrium, in which the original LS schedule (the solid line)
morphs into the dashed one. Successive short-period equilibria are pictured
on the assumption of fexprice adjustment. Te later equilibrium F, refects a
higher labor:capital ratio, the result of the relative decline in the capital stock.
P
W
Supply of goods
Copyright © 2021. Harvard University Press. All rights reserved.
F
Supply of labor
E
Aggregate demand Stationary
real price
Figure 17.9 Evolution of fexprice equilibrium when labor force
grows more rapidly than capital.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
666 Raising Keynes
Note that moving from E to F refects an economy that is getting poorer, not
one in which more people are fnding jobs.9
Observe that the process ends in disaster if it is allowed to continue un-
checked: the labor:capital ratio increases without bound. Output per unit cap
ital approaches a maximum if the limiting elasticity of substitution between
capital and labor is less than one, and it increases without bound if this elas-
ticity is greater than one. In both cases the increase in Y/K takes place because
there is so little capital to go around; output per worker falls to zero, and the
real price climbs higher and higher—the real wage goes to zero.
Te alternative possibility is that the rate of growth of capital exceeds the
rate of labor-force growth. Tis possibility is depicted in Figure 17.10, again
under the assumption of fexprice adjustment. Here the results are topsy-
turvy. As time passes and the original equilibrium E moves to F, labor be
comes scarce, with employment eventually exceeding the labor supply! At F,
money wages rise as a result of labor-market pressure, and prices rise because
of an excess of expenditure over income. As long as investment per unit of
capital exceeds population growth, the process continues, with not only a
continually falling real-price level but also an ever-increasing rate of infation.
Under a fexprice regime, there is a built-in potential check to an infa
tionary spiral but not to a defationary spiral, the details of which are lef to
P
W
Supply of goods
Copyright © 2021. Harvard University Press. All rights reserved.
E
F Supply of labor
Aggregate demand Stationary
real price
Figure 17.10 Evolution of fexprice equilibrium when labor
force grows less rapidly than capital.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 667
the appendix to this chapter. If the warranted rate of growth exceeds the natu-
ral rate, extending the model of chapter 6 to the long run provides a mecha-
nism whereby a “quasi-warranted” rate of growth (quasi because investment
intentions are partially frustrated by a lack of saving) falls until it is brought
into line with the natural rate of growth. However, it is a one-way bridge. If
the problem is a lack of demand, then the adjustment process does nothing to
raise the rate of accumulation of productive capital. To be sure, the rate of
growth of the total capital stock increases, but the increase takes the form of
unwanted inventories that do nothing to resolve the tension between war-
ranted and natural rates.
If we assume fxprice adjustment, the model of chapter 6 provides no way
of reconciling warranted and natural rates. Figure 17.11 shows what happens
when the warranted rate falls short of the natural rate and Figure 17.12 what
happens when the warranted rate is higher than the natural rate.
With fxprice adjustment, equilibrium is always on the AD schedule, so
expenditure and income are in balance, hence so are desired investment
and desired saving. Movement of the LS schedule simply shifs the equilib-
rium along the (stationary) AD schedule, resulting in an ever-higher real
price and an ever-higher rate of defation when the labor force grows more
rapidly than the capital stock (Figure 17.11), and a lower and lower real
P
W
Supply of goods
E
Stationary real price
Copyright © 2021. Harvard University Press. All rights reserved.
Supply of labor
Aggregate demand
Figure 17.11 Evolution of fxprice equilibrium when labor force
grows more rapidly than capital.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
668 Raising Keynes
P Supply of labor
W
Supply of goods
F
Stationary real price
Aggregate demand
Figure 17.12 Evolution of fxprice equilibrium when labor force
grows less rapidly than capital.
price and ever-greater infation when the labor force grows less rapidly (Fig
ure 17.12).
Taking all these results together, the model of chapter 6 does not hold out
much hope for the economy’s ability on its own to correct discrepancies be-
tween warranted and natural rates of growth. Te consequence is to reinforce
the pessimistic conclusions that Harrod and Domar drew from their respec-
tive analyses.
Bringing monetary policy into the picture does little to change this conclu-
sion. Even if we assume away obstacles to successful deployment of monetary
Copyright © 2021. Harvard University Press. All rights reserved.
policy in the short period, the long run presents new challenges. Consider the
serendipitous equilibrium in Figure 17.13. For l to equal l0, a particular value
of ( h), and thus of h, is required. Barring a liquidity trap or highly inelastic
investment demand, the central bank can put the right level of the hurdle rate
into efect. But the corresponding warranted rate of growth is totally indepen
dent of the natural rate of growth. (Tis, of course, was Harrod’s original
point!) To bring the warranted rate into line with the natural rate may require
more or less investment. Suppose ( h) > gn, so that it is necessary to increase
h as a way of scaling back aggregate demand to l = l1. Te new equilibrium,
whether defned by a fexprice or a fxprice process, involves unemployment
and continuing defation. Te price of dynamic equilibrium of warranted and
natural rates is continuing economic slack. If, by contrast, ( h) < gn, then the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 669
P
W
Supply of goods
Supply of labor
Aggregate demand
l1 l0 l
Figure 17.13 Serendipitous equilibrium not so serendipitous.
remedy is stimulus, monetary ease that reduces h and moves the AD sched-
ule to the right. In this case, the price of dynamic equilibrium is continuing
labor-market tightness (indeed, continuing overfull employment) accompa-
nied by permanent infation.
Toward a Keynesian Theory of the Long Run
Evidently, the two rates of growth, warranted and natural, can be aligned by
appropriate intervention, but at a price: either chronic unemployment or en-
Copyright © 2021. Harvard University Press. All rights reserved.
demic infation. Tat at least is where Harrod’s theory (and Domar’s) leads us
once we introduce prices and wages into their models. Te historical record is
somewhat more reassuring. Although the world has suffered bouts of both
unemployment and infation, neither seems as pervasive in post–World War
II capitalism as pre-Solow apocalyptic visions suggested. So where did Harrod
and Domar go wrong?
Te error was to assume that the labor supply is exogenously given, a num-
ber of bodies independent of the needs of the economy. Like Solow’s mis-
placed emphasis on factor substitution, exogenous labor supply was unhesi-
tatingly (and unthinkingly) incorporated into the canon. Growth theory has
seen many innovations since Harrod’s time, but the assumption that the rate
of growth must conform to the constraint of a labor force given by population
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
670 Raising Keynes
growth has rarely been questioned. Harrod’s 1939 assumption has cast a long
shadow.
In this regard, Joan Robinson (1966 [1956], 1962) is exceptional. Robinson,
Keynes’s disciple and standard-bearer in Cambridge (as well as the theorist,
along with Edward Chamberlin, of monopolistic competition), continued
until her death to argue for a role for aggregate demand in analyzing the long
run. But she and her followers were fghting a rearguard battle against an eco-
nomics increasingly dominated by the mainstream view that the only possi-
bility for aggregate demand to infuence real outcomes lay in a short run
characterized by frictions and rigidities.
Robinson’s analysis contained three innovations, of which an endogenous
labor force was one, but one easily lost sight of. First, in determining the rate
of investment, she shifed the emphasis from the rate of interest to the return
on capital. Instead of the equation ID = ( h), in which animal spirits com-
bine with the hurdle rate of interest to drive investment demand, for Robin-
son investment demand is determined by the rate of proft, in the present
notation, f(l) − (P/W)−1l. In this spirit, chapter 10 developed an investment-
demand function for capital widening, in which the diference between the
rate of proft and the hurdle rate of interest drives investment demand:
P P
−1
ID = ψ l , , ρh = ψ ( Ω(l )r − ρh ) = ψ Ω(l ) f (l ) − l − ρh .
W W
Now, the rate of proft and the hurdle rate of interest both play a role in the
determination of investment demand, but capital-stock adjustment does not
bring the two variables into line with one another. As chapter 10 noted, things
are diferent in a world of Lego capital, where every producer’s capital can be
adjusted in to bring about equality between the marginal productivity of cap
ital and its rental cost. Once Lego capital is abandoned in favor of a more re-
alistic putty-clay production function, the adjustment of the capital stock
Copyright © 2021. Harvard University Press. All rights reserved.
does not necessarily equalize marginal productivity and the rate of proft, or
the expected proftability (l)r and the interest rate, even in the long run.
A second innovation, one Robinson shared with other Cambridge econo-
mists working the Keynesian street, was to introduce class diferences into
consumption and saving.10 Instead of the assumption of uniform propensi-
ties to consume and save for all types of income and all types of agents, the
so-called Cambridge saving theory, explored in chapter 9, posits diferent
consumption and saving propensities for p rofts and wages. In its simplest
version, no wages are saved while a portion s of p rofts are saved. In conse-
quence, the saving schedule provides a second relationship between growth
and proft rates, namely SD = sr.
Te problems of aggregating the value of individual capital goods (see
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 671
above, pp. 660–661) led Robinson to believe that a production function that
assumes continuous substitution between inputs, the so-called neoclassical
production function, is the root of all evil in mainstream economics. What-
ever the merit of her antipathy, an antipathy which evidently I do not share,
fxed-factor proportions make the depiction of equilibrium much easier. With
l fxed, Robinson’s version of Harrod’s warranted rate of growth is a rate of
capital-stock growth desired by savers (SD) and investors (ID), both dependent
on the rate of proft. In Figure 17.14 (as in Figure 17.4) there are two rates of
proft at which investment and saving desires coincide. Te higher warranted
rate, g w0 , is stable, and g 1w is unstable.
Te reasoning is the same as for Harrod’s own model, with the rate of p roft
rather than the level of income driving adjustment outside of equilibrium.
Te investment function in Figure 17.14 assumes that successive increments
of proftability call forth less and less additional investment. Eventually, incre-
ments in proftability call forth more saving than investment. Robinson does
not justify this assumption, which plays the same role in her model that the
lower responsiveness of investment to income, compared with the respon-
siveness of saving, plays in the version of Harrod’s model in Figure 17.4. For-
mally, a suf ficient condition for stability (but not a necessary one, as we shall
see in the next chapter) is ′ < s, where ′ represents the response of in-
r
SD Desired saving per unit of capital
rw0
Copyright © 2021. Harvard University Press. All rights reserved.
ID Investment per unit of capital
rw1
g w1 g w0 ID , SD
Figure 17.14 Investment and saving à la Robinson.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
672 Raising Keynes
vestment demand to a change in the rate of p roft when the labor:capital ratio
is fxed.
In Robinson’s formulation, the equilibrium real price and thus the distribu-
tion of income are determined by the rate of proft at the equilibrium levels of
ID and SD. Substituting this rate of proft into the equation r = f(l0) − (P/W)−1l0
allows us to solve for the real wage. Te infuence of the supply side is com-
pletely masked by assuming fxed proportions between capital and labor; f(l)
and l become parameters rather than variables.
Robinson’s formulation of the warranted rate of growth can be understood
in terms of the AD schedule in Figure 17.15, which we will formally derive in
the next chapter. Take a vertical slice of Figure 17.15 at l = l0. Varying P/W we
can derive the relationship in Figure 17.14 between investment and saving
and the rate of proft from ID = ((l)r − h) and SD = sr. Te key is the equa-
tion r = f(l0) − (P/W)−1l0.
Tis is aggregate demand with a vengeance! Tere is no supply-side equa-
tion in Figure 17.14.
Like Harrod, Domar, and Solow, Robinson in efect sees a portion of the
elephant. Perhaps she sees more than her predecessors because of her third
innovation: how she addresses the tension between the natural rate and the
warranted rate. A variety of possible resolutions are cataloged in Robinson
(1962), each with a colorful name (some samples: “golden age,” when the war-
ranted rate and the natural rate just happen to coincide; “bastard golden age,”
“creeping platinum age,” and so forth). For my purposes, the interesting pos-
sibility is that the natural rate is not so natural, that labor-force growth is not
given exogenously:
Capitalist industry does not employ the whole work force in any country.
Domestic service, paid or unpaid, jobbing work and small-scale trade, and,
P
Copyright © 2021. Harvard University Press. All rights reserved.
W
s
s Aggregate
demand
rW0 , g w0 •
•r 1 1
W , gw
l0 l
Figure 17.15 Generalizing investment and saving à la Robinson.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
17 • First Steps into the Long Run 673
in most countries, agriculture, hold a reservoir of labour which flls up when
regular employment is not expanding as fast as the population. (1962, p. 15)
Te next chapter develops this idea more fully.
appendix: inventory accumulation as a brake on output
Tere is a peculiarity in the behavior of the economy under a fexprice re-
gime. Take the simple case where the AD schedule is vertical. To the right of
the AD schedule, it does not make sense to assume that all investment adds to
the productive capital stock. Neither does it make sense to assume that the
capital stock can continue to grow at a rate in excess of labor-force growth
when the economy is to the lef of the AD schedule.
To see this peculiarity more clearly, consider the inequality between expen-
diture and income at a defationary fexprice equilibrium. At E in Figure 17.16
expenditure falls short of income and output. Consequently, as was noted in
chapter 6, there is unwanted inventory accumulation. In the short period this
need not complicate matters unduly, since it can be argued that producers
factor in the possibility that they will be unable to sell all they produce. All
that may be required to induce enough producers to take the chance associ-
ated with overproduction is a premium of price over marginal cost.
Tis assumption allows us to address the short-run problem of unwanted
inventory accumulation by positing that goods supply involves a displace-
ment of the GS schedule and hence a displacement of equilibrium from E to
F. To the right of the AD schedule, where expenditure falls short of output,
P 0
P
W Y 0
Supply of goods
Copyright © 2021. Harvard University Press. All rights reserved.
5
F
4
E
3
0
W
2
Supply of labor
P 0
1
W
Aggregate demand Stationary real price
100 200 300 400 500 600 Y
Figure 17.16 Taking account of unwanted inventories.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
674 Raising Keynes
the premium is equal to the vertical distance between the solid GS schedule
and the dashed marginal-cost schedule.
But in the longer run, the wedge between expenditure and output may un-
dermine the very notion of equilibrium with unintended inventory accumu-
lation. Implicitly—the time has come to make the assumption explicit—the
model assumes that the accumulation of productive capital is determined by
the minimum of desired investment and desired saving, which allows for in-
ventory accumulation in situations where expenditure falls short of output.
Tis is to say that total investment is always equal to desired saving, with any
positive diference between desired saving and desired investment showing
up as undesired investment in inventory. To the right of the AD schedule in
Figure 17.16, the growth of unwanted inventory is equal to the gap sY − ( h)
K, that is, to the gap between total saving and the addition to productive cap
ital.
With fexprice adjustment this gap is a permanent feature of a defationary
equilibrium.11 If the gap between desired saving and investment per unit of
capital, sY/K − ( h), relative to productive investment per unit of productive
capital, ( h), exceeds the ratio of unwanted inventories to the productive
capital stock, then inventories will grow faster than productive capital. Te
presence of so much inventory must eventually defeat even the most optimis-
tic producer calculating his chances of selling what he produces currently. If
the gap is more modest, so that the ratio [sY/K − ( h)]/( h) is less than the
ratio of unwanted inventories to the productive capital stock, then the inven-
tory ratio will gradually diminish, and an equilibrium like F in Figure 17.16,
at which businesses weigh profts against the likelihood of fnding a market,
may be sustainable.
Tings are diferent when the economy operates to the lef of the AD
schedule. Here the problem is that expenditure can continue to exceed output
only as long as businesses are willing and able to run down inventories. In a
fexprice regime, prospective purchasers will eventually have to be turned
Copyright © 2021. Harvard University Press. All rights reserved.
away, and either consumption or investment will suffer. If we assume it is in-
vestment plans that are frustrated by infation so that actual investment is
constrained by saving, then the rate of growth of productive capital must slow
down as inventories are depleted. Te reversal of the inequality between de-
sired investment and saving, ( h) > sY/K, implies a very diferent outcome
from the previous case. Since the rate of accumulation of productive capital
is falling, capital-stock growth will eventually come into equality with the ex-
ogenously given rate of growth of the labor force as the LS schedule moves to
the right.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 18 .
keynes in the long run
A Theory of Wages, Prices, and Employment
Tere are two propositions in the classical system which can be tenta-
tively discarded. One is the population doctrine, the proposition that
the supply of labour is infnitely elastic at a certain real wage, that
wage being determined by what the labouring classes of the country
regard as their minimum standard of living with suf ficient frmness to
infuence their conduct in reproduction. Tis doctrine may still have
relevance to large poverty-stricken areas of the world of to-day. It is
one of the doctrines that may perhaps be regarded as valid in relation
to certain circumstances although not universally valid. I am inter-
ested now particularly in the economies of the United States, Great
Britain, Western Europe and other advanced countries. In this
context we may regard the size of the population not, as in the old
classical system, as a dependent but as an independent variable. To
put the matter otherwise, changes in it may be regarded as exogenous
changes.
—roy harrod
A long-run Keynesian theory is generally considered to be an oxymoron. Te
process of consolidating the Keynesian revolution made a role for aggregate
demand contingent on one form or another of market imperfection or fric-
Copyright © 2021. Harvard University Press. All rights reserved.
tion, and the resulting sand in the wheels was supposed to operate in the
short term, not over the long run. By the late 1960s, when the neoclassical
counterrevolution had begun in earnest, the Keynesians had already aban-
doned the long run to the neoclassicals. As chapter 17 argued, Robert Solow’s
1956 essay, “A Contribution to the Teory of Growth,” was widely understood
to have demonstrated the irrelevance of aggregate demand, even though the
“demonstration,” as Solow himself recognized (p. 91), is simply an assump-
Portions of this chapter were previously published in “Generalizing Lewis: Unlimited Sup-
plies of Labor in the Advanced Capitalist World,” Research in the History of Economic Tought
and Methodology 37A (2019): 165–171; and in “Wages, Prices, and Employment in a Keynesian
Long Run,” Review of Keynesian Economics 5 (Autumn 2017): 360–425.
675
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
676 Raising Keynes
tion. In the standard “New Keynesian” model, aggregate demand can have a
transitory efect on the economy, but the efect dissipates as the economy re-
turns to long-run equilibrium—an example is given at the end of the mathe-
matical appendix to this chapter.
It can’t be otherwise if we maintain the assumption of a fxed labor force,
fxed not once and for all, but fxed by population growth. In this world there
is a natural rate of growth, and we have two choices. One is to follow Solow
and deny a role for aggregate demand. Te other is to follow Joan Robinson in
letting aggregate demand play a central role, with the result that—serendipity
apart—the theory leads either to a “limping golden age” or even a “leaden
age,” characterized by perpetual growth of redundant labor, or to a “restrained
golden age” or a “platinum age,” characterized by rampant infation (1962,
pp. 51–59). Solow’s route is unattractive for obvious reasons, and Robinson
provides a way forward only if we generalize her insight that “capitalist indus-
try does not employ the whole work force in any country” (1962, p. 15). In the
models that follow, the tension between the rate of growth warranted by de-
mand conditions and the rate of growth consistent with the supply side of the
economy is resolved by reformulating the supply side, in particular the as-
sumption that the labor supply is exogenously given.
Unlimited Supplies of Labor: The Road Not Taken
A key assumption of the models in this chapter is that capitalism operates in a
world of unlimited supplies of labor. Tis assumption connects my model
both to W. Arthur Lewis’s seminal 1954 paper, and, in an odd way, to Harrod’s
pioneering attempt to apply Keynes’s conceptual framework to a growing
economy. Had they survived into the twenty-frst century, it might not have
surprised Harrod or Lewis that China has been able to sustain a remarkable
rate of growth by virtue, among other things, of its huge agricultural sector
from which to draw labor for industry. But neither thought the Lewis model
Copyright © 2021. Harvard University Press. All rights reserved.
of unlimited labor applicable to advanced capitalism. Lewis confned his ar-
gument to the poor “developing” world, and Harrod, as the epigraph to this
chapter notes, discarded this hypothesis as irrelevant to the rich countries.
For the rich countries, Harrod argued, it is more appropriate to take the labor
supply as exogenously given. Harrod’s choice has become the standard as-
sumption about labor supply in growth models. In this respect, the present
models constitute a road not taken.1
Te idea of unlimited supplies of labor is counterintuitive because we nor-
mally focus on the supply of labor to an entire economy and society closed to
immigration. In such a setting it is natural to think of labor supply in terms of
population, and to think of population as given by the net reproduction rate.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 677
Te facts tell a diferent story. With the exception of Japan, labor supplies in
the rich countries have been augmented by immigration for the better part of
a century, and in the United States from the frst permanent European settle-
ment.
Moreover, there is no analytic reason to focus on the entire economy. In-
deed, it is tendentious to argue that any single model can explain all economic
behavior in all settings; and in practice, growth models implicitly assume that
production is guided by proft maximization and that labor is a commodity,
wage labor. In short, the focus of the present inquiry, like the focus of virtually
all growth economics, is the behavior of a capitalist economy. What is novel in
my approach is to take seriously the fact that the capitalist sector is always
embedded in a larger economic formation, which includes sectors that follow
a logic diferent from the logic of capitalism.
In particular, we can distinguish a “family-enterprise” sector, in which pro-
duction is oriented toward the market but most if not all of the labor is sup-
plied by family members who are not paid wages. Te family farm—see chap-
ter 8—is the most important case in point, and this is the sector that Lewis
and other theorists of unlimited supplies of labor had in mind. But it is not
the only source of labor supply for the capitalist sector. We can also distin-
guish a “household” sector, in which not only is wage labor minimal but pro-
duction is for the immediate satisfaction of wants and needs, unmediated by
the market. Food is prepared, clothes washed, children driven to football or
piano practice—mostly by their mothers—with nary a dollar, euro, pound, or
yen changing hands.
Both these sectors have historically been important constituents of the in-
ternal “reserve army.” At the beginning of the last century, the agricultural
sector, mostly family farms, contained some 40 percent of the labor force in
the United States. By midcentury, 40 percent had fallen to 10 percent, and by
1970 to less than 5 percent. In the twenty-frst century, agriculture’s share of
the labor force has never been above 2 percent (Federal Reserve Bank of St.
Copyright © 2021. Harvard University Press. All rights reserved.
Louis). Most of the rich capitalist world lagged behind the United States but
eventually caught up. As late as 1960 some 30 percent of the labor force in
both Japan and Italy and close to 25 percent of the French labor force were
engaged in agriculture. By the beginning of this century, these countries more
or less caught up with the United States; in none of them was agricultural
employment more than 5 percent of the labor force (Bureau of Labor Statis-
tics 2005, p. 30).
Te household sector similarly provided a steady stream of recruits to the
capitalist sector. Women didn’t necessarily leave the kitchen altogether; most
added paid labor to their domestic duties. Female participation rates rose in
the United States from 35 percent in 1950 to 60 percent, declining slightly fol-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
678 Raising Keynes
lowing the beginning of the recession in 2008. In the teens of this century,
women’s and men’s participation rates have difered by hardly ten percentage
points.
Tese internal recruits have complemented the external reserve army, the
immigrant population that has served as a reliable if politically contentious
source of labor, especially for those jobs that native populations have been
reluctant to fll at going rates of pay. Te time pattern of immigration into the
United States is especially revealing. Afer an abrupt fall in the wake of World
War I, a decline that lasted for almost half a century, immigration picked up
around 1970, just as domestic agriculture dried up as a source of labor—see
Figure 18.1.
Figure 18.2 shows the impact of the reserve army on the paid labor force in
the United States. Whereas the native-born population grew at an annual rate
of 1.3 percent over the twentieth century, paid private employment outside
agriculture was able to grow half again as fast, at a rate of more than 2 percent.
Te important point is that the reserve army is not a fxed body of men and
women, the battalion of workers made redundant by progress in technique
(or in our own day by outsourcing of whole sectors of the economy). Rather,
the reserve army is constituted and reconstituted in terms of the needs of the
capitalist economy: as one source of labor, domestic agriculture, dried up,
other spigots were opened. In practical terms, labor supply is unlimited, not
only in the poor countries, but in the rich countries as well.2
70
60
50
Percentage
40
Copyright © 2021. Harvard University Press. All rights reserved.
30
20
10
0
1900 1920 1940 1960 1980 2000
Agricultural employment as percentage of U.S. civilian labor force
Foreign born as percentage of U.S. population
Female participation as percentage of female U.S. population
Figure 18.1 Sources of the Reserve Army in the twentieth century. Source:
Statistical Abstracts of the United States.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 679
What then determines the wage? Te classical economists—Smith, Ricardo,
and Marx—were on the right track in emphasizing subsistence as the deter-
minant of wages, though they have been misunderstood by later generations
who took subsistence to mean a minimal standard of living that would ensure
the reproduction of the labor force. (Tis presumably was Harrod’s under
standing of the term and for him a reason to reject the whole idea of an un-
limited labor supply.) Neither Smith nor Ricardo nor Marx conceived of sub-
sistence solely in biological terms. For all three, there were historical, cultural,
and institutional dimensions that entered the determination of real wages.
Marx, not surprisingly, emphasized class power, but this emphasis did not
preclude a recognition that class power is situated in a matrix of culture and
history. Workers are better able to press their claims to the extent that public
opinion is on their side. Subsistence, in short, is a social norm:3
[Te worker’s] natural wants, such as food, clothing, fuel, and housing, vary
according to the climatic and other physical conditions of his country. On
the other hand, the number and extent of his so-called necessary wants, as
also the m
odes of satisfying them, are themselves the product of historical
development, and depend therefore to a great extent on the degree of civili-
sation of a country, more particularly on the conditions under which, and
consequently on the habits and degree of comfort in which, the class of
free labourers has been formed. In contradistinction therefore to the case
of other commodities, there enters into the determination of the value of
labour-power a historical and moral element. (Marx 1959 [1867], p. 171)
13
12
Copyright © 2021. Harvard University Press. All rights reserved.
Log scale
11
10
9
1900 1920 1940 1960 1980 2000
Log (native-born population) Log (nonfarm employment)
Figure 18.2 Growth of native-born population and nonfarm employment.
Source: Gibson and Jung 2006.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
680 Raising Keynes
To summarize: the long-run theory of labor supply proposed here reverses
the mainstream relationship between the supply of labor and the real wage.
Instead of an exogenous labor supply and an endogenous real wage, as the
mainstream would have it, I here posit an endogenous labor supply. Te equi-
librium real wage is endogenously determined, but an ingredient of this equi-
librium is an exogenous real wage that forms the basis of workers’ monetary
demands. To avoid confusion, I shall label this wage not a subsistence wage
but a conventional wage (as did Lewis 1954, pp. 150, 172), conventional refer-
ring to both the customary elements and the contractual elements that enter
into wage determination. In a dynamic setting the conventional wage is per-
haps better understood as a target wage, the wage that the working class aims
for in order to achieve norms of consumption conventionally conceived and
understood if not unanimously agreed upon.
None of this is to say that productivity is irrelevant to wage determination.
Productivity determines the size of the pie that the contending parties strug-
gle over. Nineteenth-century British output had to be of a certain size before
workers could obtain a share large enough to permit them to buy refned white
bread rather than the coarse grains to which the lower classes had accommo-
dated themselves in earlier times; twentieth-century American production
had to be of a certain size before workers could successfully struggle for a
share large enough to include an automobile or a mortgage payment as part of
the wage packet. But a productive technology did not guarantee that either
the nineteenth-century British pie or the twentieth-century American pie
would be sliced in a way that would allow workers to achieve their aims. For
this to happen, certain community standards, social norms, were essential.
Tese community standards underlie, for example, the very idea of a legal
minimum wage—and the erosion of community underlies the erosion of the
minimum wage in the United States since the 1970s. Nor is it just the minimum
that is at issue. During this same period, there has been an ever-widening gap
between workers in the middle of the wage distribution and the richest 1 per-
Copyright © 2021. Harvard University Press. All rights reserved.
cent of the income distribution, which includes both top managers and large
stockholders. Tere is no l onger even the pretense that we are all in it together,
that the United States is a community when it c omes to matters economic.
Not only is productivity relevant to wage determination, so are rates of
price increases or decreases. Workers, as Keynes insists, may bargain in
money terms, but they are bargaining about real wages, so that infation or
defation plays an important role in determining money bargains.
Unemployment also matters. Te ability of workers to achieve their ambi-
tions in the sphere of wages, or to hold onto a level of real wages already
achieved, depends on the state of the labor market. In the short period, in
which wage claims are made (and contested), unemployment is a fact of life
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 681
and may be the dominant fact, even if the labor force is endogenous in the
long run.
Making the unemployment rate one among other determinants of changes
in wages (rather than the sole determinant) resolves another problem. Te
idea that wages respond to unemployment, and unemployment alone, pro-
vides a tolerably good account of wage behavior during the period 1929 to
1933, but not since. Until the Depression bottoms out in 1933, changes in
money wages are closely correlated with the unemployment rate, as Table 18.1
shows. But in 1933 wages stopped falling even though unemployment re-
mained high. If we extend the table into the recovery, we fnd that wages pick
up afer 1933 even though unemployment never fell below 10 percent until
the country entered World War II. Table 18.2 provides the extended data.
Table 18.1 Relationship between Unemployment Rate and Changes in Money
Wages
Unemployed, Not Counting Federal Year-on-Year Change in Hourly
Emergency Workers as Employed Manufacturing Wage
(percent) (percent)
1930 9.0 −1.8
1931 16.3 −7.3
1932 24.1 −13.7
1933 25.2 0.0
Source: Historical Statistics of the United States, Millennial Edition Online.
Table 18.2 Relationship between Unemployment Rate and Changes in Money
Wages (1930–1941)
Unemployed, Not Counting Federal Year-on-Year Change in Hourly
Emergency Workers as Employed Manufacturing Wage
(percent) (percent)
1930 9.0 −1.8
Copyright © 2021. Harvard University Press. All rights reserved.
1931 16.3 −7.3
1932 24.1 −13.7
1933 25.2 0.0
1934 22.0 20.5
1935 20.29 1.9
1936 17.03 1.9
1937 14.32 12.7
1938 19.05 0.0
1939 17.17 1.6
1940 14.59 4.8
1941 9.94 10.6
Source: Historical Statistics of the United States, Millennial Edition Online.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
682 Raising Keynes
Clearly a simple model of wages responding solely to the unemployment level
will not do even if the long run is as limited as the transition from decline to
recovery. Te regression equation below, covering the period 1930 to 2011,
adds labor’s share of output, productivity growth, and the change in energy
prices as explanatory variables, along with dummy variables meant to refect
changes in the conventional wage at three critical times in its evolution.4 Un-
employment still plays a role, but it has to share the stage with these other
variables.
∆W WL ∆Prod ∆PE
= 43.48 − 76.53 + 0.4708 Prod + 0.1196 P − 0.3058 UNRATE
W−1 PY −1 −1 ( E )−1
(21.10) (35.25) (0.1439) (0.0221) (0.1091)
+ 5.062 DUM1936 + 2.138 DUM1970 − 4.773 DUM1994
(2.301) (0.9402) (1.0337)
R2 = 0.63
Newey-West Standard Errors in Parentheses
Observations: 82 (1930–2011) F(7, 74) = 14.1
Variable Description Mean Std Dev Min Max
∆W
Percentage Change in Nominal Wages 4.740 4.063 −13.210 17.539
W−1
WL
PY Labor Share of Output, Lagged One Year 0.560 0.018 0.514 0.596
−1
∆Prod
Percentage Change in Output per Employee Hour 2.215 2.391 −6.641 12.552
Prod
−1
∆PE
Percentage Change in Price of Energy 4.641 11.852 −26.045 55.164
( E )−1
P
Copyright © 2021. Harvard University Press. All rights reserved.
UNRATE Unemployment Rate 7.241 4.849 1.200 24.900
DUM1936 = 0 prior to 1936, = 1 in 1936 and later years
DUM1970 = 0 prior to 1970, = 1 in 1970 and later years
DUM1994 = 0 prior to 1994, = 1 in 1994 and later years
Te conventional-wage share emerges from the model.5 Money-wage dy-
namics, without the control variables, are given by
∆W WL ∗ WL
= − PY ,
W−1 PY −1
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 683
where (WL/PY)* is the conventional wage, and is the speed of adjustment.
Translating this formula into a regression equation we have
∆W WL a WL
= a0 + a1 ≡ −a1 0 − + controll variables.
W−1 PY −1 −a1 PY −1
Tus the speed of adjustment, , is equal to −a1, and the conventional wage
share, (WL/PY)*, is equal to the ratio a0/(−a1). In the above regression, this
ratio gives an initial value of the conventional wage share equal to 0.57 (se =
0.033, computed by the delta method). (WL/PY)* increases to 0.63 in 1936,
and increases again to 0.66 in 1970. In 1994 the conventional-wage share falls
to 0.6, close to its value in the early 1930s. We will revisit the determination of
wage changes afer the theoretical investigation in the remainder of this chap-
ter and the empirical analysis in the next chapter.
The Checkered History of the Phillips Curve
One reason for moving beyond the short run is that the policy question that
animated Keynes is not limited to the here and now, but is an issue for the
long run as well. Can demand management improve the performance of the
economy? Do changes in aggregate demand have long-run efects on employ-
ment and other variables? And, to go beyond Keynes, what is the impact of
wages, or rather wage aspirations refected in the conventional wage, on the
level of economic activity?
Solow answered these questions by assuming aggregate demand away. Mil-
ton Friedman (1968) and Edmund Phelps (1968) took a diferent tack, argu-
ing that aggregate demand would in the long run afect only the price level, as
in Figure 4.9. Te positive relationship between employment and infation
enshrined in the Phillips curve, understood by Keynesians to refect the op-
eration of aggregate demand, was dismissed by the counterrevolutionaries.
Copyright © 2021. Harvard University Press. All rights reserved.
Te Phillips curve was now understood to be the result of misperceptions that
would necessarily disappear as agents developed more sophistication about
the economy. Te implication was that demand cannot matter in the long
run. Indeed, Friedman and Phelps predicted that periods of high infation
would not be accompanied by higher economic activity. In the long run there
is no Phillips curve, no trade-of between economic activity and price stabil-
ity. In this perspective, raising the conventional wage may lead to higher
money wages, but this is necessarily an exercise in futility: higher money
wages can mean only higher infation, with no impact on employment and
output. Te classical dichotomy with a vengeance!
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
684 Raising Keynes
And the data do appear to bear out the prediction of no relationship be-
tween employment and infation, at least not a simple one. Figure 18.3 plots
the data over more than half a century. Analyzing these data economists
have found, if anything, a negative relationship between real economic activ-
ity and infation. Cross-sectional studies by Stanley Fischer (1993), Robert
Barro (1996), and others have found a signifcant negative correlation be-
tween growth and infation. But since these results are dominated by high in
fation rates, their relevance is perhaps limited to the relatively poor countries
where high infation has been more common. More striking therefore are the
fndings of Moshin Khan and Abdelhak Senhadji (2001), who separate poor
and rich countries and fnd that for the rich countries the threshold above
which infation is associated with lower GDP growth is only 1 to 3 percent per
year. Other researchers, to be sure, have taken issue with these results, but
nobody to my knowledge has found the strong, positive relationship between
infation and growth that the Phillips curve suggests.
None of this should surprise us. Once it has been determined that demand
does not matter in the long run, it makes sense to treat all observations sym-
metrically and look for supply-side efects.
Appearances notwithstanding, macroeconomists learned from the impact
of rising oil prices in the 1970s to sort out supply shocks from demand shocks
and thus to make sense of the data in terms of a Phillips curve along which
movements refect demand shocks but which is itself moved by supply shocks
(see, among others, Robert Gordon [1984, 2013]). Sorting out demand and
supply shocks is central to my approach as well. But it will not do to lump
all supply shocks together since wage and price shocks may have diferent
impacts. Tis possibility is taken account of in this chapter: wage, price,
14.0
12.0
Inflation (change in CPI)
Copyright © 2021. Harvard University Press. All rights reserved.
10.0
8.0
6.0
4.0
2.0
0.0
−2.0
90 91 92 93 94 95 96 97
Employment as percentage of civilian labor force
Figure 18.3 Employment vs. infation (1956–2011).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 685
and employment dynamics derive from the interaction of aggregate demand,
proft maximization, and a conventional wage.
Are High Wages Good for Capitalism?
Te relationship between wages, prices, and employment, specifcally whether
higher wages promote higher employment, has long been controversial. Te
question did not begin or end with Keynes, but Te General Teory provided
a framework in which it became possible to analyze the impact of the wage
rate on aggregate demand. Tere is some irony here: in the main, Te General
Teory argues that the real wage rate is determined endogenously, a ther-
mometer of economic conditions rather than a thermostat.6
Despite Keynes’s own relative neglect of the issue, the potential for stimu-
lating aggregate demand by redistributing profts and other forms of prop-
erty income to wage earners became identifed with Te General Teory.
With good reason: even though Keynes paid little attention to this issue,
the importance of Te General Teory framework to the macroeconomics of
redistribution made Keynes a fgure lionized by the social-democratic Lef
and despised by the Right. For the Lef, it was a no-brainer. As long as the
economy is operating at less than full employment, income redistribution
amounts to a free lunch for the economy as a whole: shifing income from
rich people with low propensities to consume to poor people with higher con-
sumption propensities would boost aggregate demand. For the Right, redistri-
bution was anathema, a threat not only to the free market but to social order
and morality.
Among economists, the idea that redistribution could stimulate the econ-
omy came under attack from two sides. Te permanent-income and life-cycle
hypotheses (see chapter 9) challenged the idea that the propensity to consume
is a function of income. And with the rise of supply-side economics, aggre-
gate demand was relegated to being at best a short-run phenomenon of fric-
Copyright © 2021. Harvard University Press. All rights reserved.
tions and market imperfections, a fate for which Keynes’s followers, especially
in the United States, had helped to pave the way.
Tere was a minority view: the lef Keynesians, who took inspiration from
Joan Robinson and Michał Kalecki, continued to argue that aggregate de-
mand was a problem that ran much deeper than frictions and imperfections,
and Robinson, as I observed in chapter 17, devoted much efort to building
long-run models in which outcomes depend on aggregate demand (1966
[1956], 1962). Te distributional issue remained central in these models.
Bob Rowthorn (1982) formalized the argument that redistributing income
from rich to poor would increase aggregate demand because the propensity to
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
686 Raising Keynes
consume declines with income. A novel aspect of Rowthorn’s argument was
that higher consumption demand would be self-reinforcing through its ef-
fects on investment demand. Higher consumption demand implies higher
employment, higher capacity utilization implies a higher rate of p roft, and a
higher rate of proft implies higher investment demand.
Amit Bhaduri and I challenged Rowthorn’s logic, in particular, the impact
of the wage rate on investment demand (Marglin and Bhaduri 1990; Bhaduri
and Marglin 1990). We argued that the two factors that together determine
the proft rate, namely, the share of profts in GDP and the rate of capacity
utilization, should be treated separately because each has a distinct impact
on investment demand. Te p roft share refects the prospects for making
money from new capacity, whereas the rate of capacity utilization refects the
likelihood of selling additional goods. When the proft share has a large posi-
tive efect on investment, the negative impact of higher wages on the p roft
share may outweigh the positive impact on capacity utilization via consump-
tion demand. Our conclusion was that growth can be stimulated by higher
wages under some circumstances, and by higher profts—lower wages—un-
der others. Growth can be “wage led” or “proft led.”
Te intention was to change the debate about the impact of the wage rate
on growth from one that focused entirely on consumption to one that in-
cluded investment as well. Judging from the voluminous literature that has
appeared since we published our papers, we succeeded. (See, for example, the
symposium in the Review of Keynesian Economics, 2016 and 2017.)
Nonetheless, I have come to view our work as defcient, particularly in our
focus on comparative statics and our corresponding neglect of dynamics. Te
version of our argument that appeared in the Cambridge Journal of Economics
ends with the observation
Te recognition that quantities (capacity utilization) and prices (the real
wage) may adjust simultaneously in a more general dynamic model raises a
deeper conceptual issue regarding the interpretation of the IS-curve itself. It
Copyright © 2021. Harvard University Press. All rights reserved.
can be treated either as the locus of stationary capacity utilization . . . as has
been implicitly assumed in our . . . analysis, or as the locus of stationary
price level . . . Ultimately it boils down to one of the most important unset-
tled questions of modern macroeconomics: does excess demand for com-
modities lead primarily to quantity or to price adjustment? We cannot pre-
tend to have an answer; but dynamic analysis cannot be undertaken without
addressing this important, and as yet unsettled, question. (Bhaduri and
Marglin 1990, p. 390)
I make no claim to be able to answer defnitively the question of how prices
and quantities adjust out of equilibrium. But the data on employment and in
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 687
fation shed light on dynamics, especially if we separate wage and price ad-
justment rather than combining the two in the level of the mark-up.
Tere is another problem with the analysis Bhaduri and I ofered in 1990:
we lumped all investment together, implicitly assuming that investment takes
place solely to expand capacity, an assumption that goes back to Roy Harrod
(1939, 1948). Output-enhancing investment, capital widening in an older lit-
erature, is important, but it is not the only kind of investment. In addition to
enhancing output, investment takes place in order to cut costs by substituting
capital for labor, energy, or other inputs—capital deepening for short.
As chapter 10 observed, cost-cutting investment has a diferent logic from
output-enhancing investment, in particular, a very diferent relationship to
wages and profts. For output-enhancing investment, the lower the real wage,
the better: lower real wages mean, ceteris paribus, higher p rofts. By contrast,
cost-cutting investment, specifcally labor-saving investment, ought to re-
spond positively to higher wages: the higher the real wage, the greater the
savings in labor costs from a given investment—hence the more p rofts.
Te result is to make the distinction between wage-led and p roft-led
growth somewhat problematic. A lower real wage makes one kind of invest-
ment, capital widening, more proftable. But a lower real wage makes the
other kind of investment, capital deepening, less proftable. In the one case,
proft-led growth benefts from lower real wages, in the other, from higher
real wages. Te diference is that in the frst case higher proftability as a
stimulus to investment conficts with higher real wages as a stimulus to con-
sumption, whereas in the second case the stimulus to investment and con-
sumption complement one another.
Te diferent impact of capital-widening and capital-deepening investment
is particularly important because the salience of the two types of investment
is likely to difer markedly over the cycle. Over the decade 2008 to 2017, for
example, there was considerable excess capacity and, consequently, relatively
little demand for output-enhancing investment. A recession thus enhances
Copyright © 2021. Harvard University Press. All rights reserved.
the case for wage-led growth; wage-led growth becomes more attractive since
high wages not only stimulate consumption demand directly but also stimu-
late the investment demand for capital deepening. If capacity utilization is at
90 percent, there may be little interest in output-enhancing investment, but
cutting costs on 90 percent of maximum output is only 10 percent less worth-
while than cutting costs on 100 percent.
When bust is replaced by boom, capital widening becomes more important
whether or not it takes center stage. In boom situations, the positive relation-
ship between low wages and investment proftability bolsters the argument
for proft-led growth.
Te countercyclical response of overall investment to the real-wage rate has
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
688 Raising Keynes
an important impact on the recurring efort to determine whether particular
economies are wage or proft led. Te search may reveal the economy to be a
chimera. Any given economy may respond positively to high real wages in
slack times and negatively in boom times; the average response over the cycle
may be of limited interest.
A Long Run Model in the Spirit of The General Theory
Te basic long-run model, depicted in Figure 18.4, takes the real price level
P/W and the labor:capital ratio l ≡ L/K as describing the state of the economy.
(At any moment of time the capital stock is given, so in the short run, l is a
measure of employment.) Formally we have
Y = F(L, K),
L
f (l ) ≡ F 1, .
K
In the simplest case, investment demand per unit of capital is a function of
the hurdle rate of interest alone, ID = ( h); desired saving is a constant frac-
tion, s, of income, and per unit of capital is SD = sf(l).
P
W
Supply of goods
Copyright © 2021. Harvard University Press. All rights reserved.
P Supply of labor
W
Aggregate demand
l0 l
Figure 18.4 Aggregate demand, goods supply, and unlimited labor supply.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 689
Figure 18.4 difers from the models of chapter 6 in how labor supply is rep-
resented. Te same forces are at play in determining wage dynamics in the long
run as in the short run, but in the short run the focus was entirely on unem-
ployment; we ignored the conventional wage. In the long-run context, the
emphasis is on the conventional wage: to keep matters simple we shall assume
that the supply of labor is literally unlimited at the conventional real wage. Tis
ignores the complication that, though the labor force is endogenous, it doesn’t
adjust instantaneously; the unemployment rate is an important variable in de-
termining the creation and mobilization of one or another reserve army. (Te
mathematical appendix to this chapter lays out a model that incorporates
both unemployment and the conventional wage as determinants of wages.)
Like the short-run models of previous chapters, the model in Figure 18.4 is
overdetermined.7 Any two of the three schedules are suf ficient to determine
employment and the real price. All three together make the usual concept of
equilibrium insuf ficient. We can make sense of Figure 18.4 only if we shif the
focus to the dynamics of adjustment.
With fexprice dynamics, the picture is in Figure 18.5. Te equilibrium is
at E, where both employment per unit of capital and the real price are station
ary. E is an infationary equilibrium, with prices and wages rising at the same
P
W Stationary real price l 0
Supply of
P 0
goods
W
Copyright © 2021. Harvard University Press. All rights reserved.
P Supply of labor
W 0
W
Aggregate demand
P 0
l0 l
Figure 18.5 Equilibrium with fexprice dynamics.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
690 Raising Keynes
percentage rate. Aggregate expenditure exceeds aggregate income, so prices
rise. And because the equilibrium real wage is below the conventional wage,
P/W > (P/W)*, there is constant infationary pressure on wages.
As in the short run, the alternative to fexprice adjustment is what Hicks
(1974) called fxprice adjustment. In the long run as well as in the short
run, this terminology is misleading since prices are not fxed; rather, the
direct impact of an excess or a shortfall of expenditure relative to income
is on output and employment. Prices (and wages) are afected indirectly.
Prices are driven by the relationship between actual employment and the
proft-maximizing rate of employment, that is, by the horizontal distance be-
tween today’s level of employment and the GS schedule. Money wages con-
tinue to be driven by the gap between the actual real price and the conven-
tional real price.
Tis process defnes the equilibrium in Figure 18.6. Since the GS schedule
is now a locus of stationary prices and the LS schedule is the locus of station
ary money wages, the stationary real-price locus, (P/W)• = 0, lies between
them.
As Figure 18.6 is drawn, the equilibrium, like the equilibrium in Figure
18.5, is characterized by infation: producers raise prices because they are los-
ing money at the margin, while workers, having to make do with less than the
P
W P 0
Supply of goods
P 0
W
Stationary real price
Copyright © 2021. Harvard University Press. All rights reserved.
E
P Supply of labor
W 0
W
Aggregate demand
l 0
l0 l
Figure 18.6 Fixprice dynamics.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 691
conventional wage, put pressure on wages. At E, the pressure on prices and
pressure on wages just balance, so that the real price remains stationary.8
Some Comparative Statics: Phillips and Anti-Phillips Curves
What happens when there is a shock to the system so that one or another
schedule is displaced? Does the shif in the equilibrium produce a Phillips
outcome, in which employment and infation move in the same direction, or
an anti-Phillips outcome, in which employment and infation move in oppo-
site directions?
Look frst at what happens when aggregate demand changes, as in Figures
18.7(a) and 18.7(b). In both cases the increase in aggregate demand leads to
an increase in economic activity accompanied by higher infation, that is, a
Phillips outcome.
Supply shocks are diferent, because demand and supply are not indepen
dent. As David Colander observed in 1995, a change in production condi-
tions will not only shif the GS schedule but will normally move the AD
schedule as well. A change in output is at the same time a change in income.
With fexprice dynamics, the results are ambiguous. An upward shif in the
GS schedule produces a higher price level and higher infation, but it may in-
crease or decrease the level of employment depending on the relative speeds
at which prices and wages adjust: a Phillips outcome, with higher employ-
ment and higher infation, is a possible result of an upward shif in the supply
schedule, but so is an anti-Phillips efect, coupling lower employment with
higher infation. In the case of fxprice dynamics, we always get a Phillips out-
come since a lefward shif of the GS schedule means more workers are
needed to produce the same output.
Te simplest way of seeing this is to expand the production function to in-
clude a third variable, say, energy, denoted E. We write
Y = F(K, L, E)
Copyright © 2021. Harvard University Press. All rights reserved.
and assume constant returns to scale, so that
Y
= F=
(1, l , e) f (l , e),
K
where e = E/K. We further assume that energy is purchased on a competitive
market at the real price, = PE/P, where PE is the nominal price of energy. Te
efects of a change in the price of energy depend on whether energy is pro-
duced domestically or imported. Te simplest story is that energy is en-
tirely imported and paid for by transferring goods abroad. h(l, ) is then the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
692 Raising Keynes
(a) Flexprice dynamics
Stationary real price
P
P 0
W
W l 0
Supply of goods
F
E
P Supply of labor
0
W W
Aggregate demand
P 0
l0 l
(b) Fixprice dynamics
P 0
P Supply of goods
W
Stationary
real price
P 0
W
Copyright © 2021. Harvard University Press. All rights reserved.
F
E Supply of labor
P
W 0
W
Aggregate demand
l 0
l0 l
Figure 18.7 An aggregate-demand shock.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 693
output:capital ratio net of the cost of energy. Tis assumption allows us to
write
h(l , ) = max [ f (l , e) − e ] = f ( l , e(l , ) ) − e(l , ),
e
where e(l, ) is the amount of energy per unit of capital that, for given l and ,
maximizes the expression f(l, e) − e.
What happens when the real price of energy, , changes? Figure 18.8 shows
how an increase in afects both goods supply and aggregate demand. Pre-
cisely how equilibrium shifs depends both on the adjustment process and on
the relative speeds of adjustment.
With fexprice adjustment, there are contradictory forces at work, and the
contradictions may be resolved in diferent ways. Two possibilities are shown
in Figures 18.9 and 18.10. In Figure 18.9, the labor:capital ratio at F, the
new equilibrium, is greater than at E, whereas in Figure 18.10, the new l is
smaller than the old. In both cases the new real price is higher, which is to
say that an increase in the price of energy leads to greater infation, but
the infation is now “cost-push” with its origins on the supply side, as dis-
tinct from the “demand-pull” infation caused by an increase in aggregate
demand.
P
W
Supply of goods
Copyright © 2021. Harvard University Press. All rights reserved.
P Supply of labor
W
Aggregate demand
l0 l
Figure 18.8 An increase in energy price.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Stationary real price
P
W P 0
W
l 0
Supply of goods
P Supply of labor
W 0
W
Aggregate demand
P 0
l0 l
Figure 18.9 Increase in energy price with fexprice dynamics, I:
sluggish wage adjustment.
P
W
l 0
Supply of goods
Stationary real price
P 0
W
Copyright © 2021. Harvard University Press. All rights reserved.
P E Supply of labor
W 0
W
Aggregate demand
P 0
l0 l
Figure 18.10 Increase in energy price with fexprice dynamics, II:
rapid wage adjustment.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 695
Why is the efect on output diferent in the two diagrams? Te impact of the
shif in the GS schedule is unambiguous: when the price of energy rises, the
real marginal cost of output also rises (labor and energy are complements),
and producers respond to the incentive to reduce output and employment.
Te ambiguity is the result of contradictory efects on demand. Te AD
schedule moves outward since the higher energy price reduces real income
and thus requires a greater income to generate suf ficient saving to match the
given level of investment demand. Tat is, output per unit of capital net of
energy costs, h(l, ), falls, and therefore so does saving. Because investment
demand is assumed to be given, aggregate expenditure rises relative to in-
come, and this causes the price level to rise. Tis makes production more
proftable at the margin, which induces an increase in output and employ-
ment. However—this is the contradictory efect—the rise in prices reduces
real wages. Tis leads to wage pressure that counters the positive efect of
higher prices on output and employment. How these conficting tendencies
are resolved depends on the relative adjustment speeds of money prices and
money wages.
Consider the limiting cases of rapid price adjustment and sluggish price
adjustment (rapid wage adjustment). In the frst case, which is Keynes’s case
of money-wage rigidity, and therefore real-wage fexibility, there is nothing to
constrain the adjustment of the price level to an energy-price shock, and the
workers’ share of the burden is felt in lower real wages and higher infation.
Employment per unit of capital actually increases.
If wages adjust much more rapidly than prices, employment efects domi-
nate. In the limit, the real wage is rigid, and the workers’ share of the burden
associated with a supply shock is through a reduction in employment per unit
of capital. Once again, the rise in the price of energy leads to a permanently
higher rate of infation.
Tis analysis suggests a fssure in the working class with regard to a supply-
side shock; if asked to choose between taking a hit in the form of reduced
Copyright © 2021. Harvard University Press. All rights reserved.
wages or in the form of reduced employment, all workers do not have the
same interests. Tose with secure employment, or with good job prospects in
the face of job loss, might be expected to opt for rigid real wages. If these folks
have their way, the stationary real-price locus will coincide with the conven-
tional wage, and employment per unit of capital will bear the entire adjust-
ment burden. Tose at the margins with respect to employment opportunities
might opt for rigid money wages. If they have their way the stationary real-
price locus will coincide with the AD schedule, and the entire hit takes the
form of lower real wages, employment being unafected.
With fxprice dynamics, the picture is in Figure 18.11. An increase in the
price of energy increases employment while lowering real wages and increas-
ing the equilibrium level of infation. Employment increases because more
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
696 Raising Keynes
workers are required to produce the same output; with fxprice dynamics
equilibrium is always on the AD schedule. Real wages fall because producing
the new level of output at F is proftable only at a higher real price; infation
increases because workers increase the pressure on money wages as the real
wage falls. With fxprice adjustment, the results are unambiguous: employ-
ment and infation both rise in conjunction with a falling real wage. How
much infation rises and the real wage falls depends on the relative speeds at
which prices and wages adjust. Te more rapidly prices adjust, the greater the
hit to infation and the real wage.
Te results of varying the conventional wage are similar to the results of
a supply shock—without the ambiguity in the efect on employment that ex-
ists with fexprice adjustment. Tis is another case of cost-push infation; in-
deed, wage pressure was practically synonymous with cost-push before the oil
shocks of the 1970s.
As Figure 18.12(a) shows, in the fexprice case, an increase in the conven-
tional wage, which is to say a fall in (P/W)*, leads unambiguously to a reduc-
tion in employment per unit capital and an increase in infation (except in the
limiting case of rigid money wages); in the fxprice case, Figure 18.12(b),
there is also an increase in the equilibrium rate of infation, but employment
is unafected.
P
W
Stationary
real price
P 0
Supply of goods W
P 0 F
Copyright © 2021. Harvard University Press. All rights reserved.
E Supply of labor
P
W 0
W
Aggregate demand
l 0
l0 l
Figure 18.11 An energy-price shock with fxprice dynamics.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 697
(a) Flexprice dynamics
Stationary real price
P l 0
P 0 Supply of goods
W
W
E
F
P Supply of labor
W 0
W
Aggregate demand
P 0
l0 l
(b) Fixprice dynamics
P P 0
W Supply of goods
Stationary real price
P 0
W
Copyright © 2021. Harvard University Press. All rights reserved.
F
P Supply of labor
W 0
W
Aggregate demand
l 0
l0 l
Figure 18.12 An increase in the conventional wage.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
698 Raising Keynes
Te fxprice result that employment is unafected by a shif in the conven-
tional wage is the result of a vertical AD schedule, the consequence of assum-
ing, frst, that saving is a constant fraction of income, and, second, that invest-
ment demand is exogenously given as a function of one variable, the hurdle
rate of interest. As we relax these simplifying assumptions in the next section,
we shall obtain diferent results.
Te assumption of a uniform saving propensity makes it impossible to ex-
amine the lef Keynesian view that higher wages are a boon to employment
since this proposition rests on the assumption that the fraction of wages spent
on consumption is higher than the fraction of p rofts. To see the distribu-
tional efect in action, we replace the assumption of a uniform propensity to
consume with the assumption—itself highly simplifed in the other direc-
tion—that all wages are consumed while only a fraction of p rofts are con-
sumed, what was called in chapter 9 the Cambridge saving theory.
To see the efect of diferent assumptions about investment, we distinguish
capital widening from capital deepening. Here, in contrast with chapter 10,
we examine these diferent assumptions about investment in conjunction
with the Cambridge saving theory. Te mathematics is in the appendix.
Demand and Supply Shocks with the Cambridge Saving
Theory and Capital-Widening Investment
As chapter 9 noted, a pre-Keynesian view of consumption and saving as-
sumed that working people simply lacked the economic capacity to save and
that the rich were responsible for whatever saving the community was able to
muster. In the last chapter, we saw how Joan Robinson deployed this assump-
tion along with an innovative characterization of investment demand as a
function of the rate of p
roft. Tis section formalizes Robinson’s argument—
in a way she would likely deplore (given her aversion to a production function
incorporating aggregate capital and continuous factor substitution)—in order
Copyright © 2021. Harvard University Press. All rights reserved.
to make clear the role of what is assumed about investment and saving in the
analysis of the long run.
Saving frst. Taking account of energy as well as capital and labor in the
production function, we have the rate of p roft as
−1 −1
Π P P
r = = f (l , e) − l − ξe = h(l , ξ) −
K W
W
and the simplest version of the Cambridge saving theory on a per-unit of cap
ital basis as
P
−1
SD = s πr = s π h(l , ξ) − l .
W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 699
Tis saving function is pictured in Figure 18.13, with the real price increasing
from (P/W)1 to (P/W)2.
With respect to investment, we examine frst the expansion of productive
capacity—capital widening—which, as I have noted, is the nigh-universal as-
sumption in the literature initiated by Roy Harrod in 1939. Capital widening
responds positively to the real price as well as to the level of output, while cap
ital deepening is relatively unresponsive to the level of economic activity and
responds negatively to the real price. Taking account of energy and writing
investment demand as a ratio to the stock of capital, capital-widening invest-
ment is
P P
−1
I D = ψ r , l , , ρh , ξ = ψ(Ω(l )r − ρh ) = ψ Ω(l ) h(l , ξ) − l − ρh .
W W
Te argument of the investment function is the anticipated annual return
from a unit of investment that lasts forever. Te gross return, what chapter 10
termed the quasi-rent, is the product of the subjective probability of fnding a
market for additional output, (l), and the prospective rate of p
roft assuming
the additional output can be sold, namely, h(l, ) − (P/W)−1l. Te net return is
the diference between the quasi-rent, the probability-weighted proft, and the
hurdle rate of interest. Figure 18.14 shows how capital-widening investment
demand responds to the capital:labor ratio at two diferent levels of the real
price.
Aggregate demand is defned by the equality of investment demand and
desired saving, graphically determined by the intersection of the schedules in
Figures 18.13 and 18.14. Figure 18.15(a) assumes that investment is more re-
1
P
S D s h (l , ) l
SD W 2
Copyright © 2021. Harvard University Press. All rights reserved.
1
P
S D s h (l , ) l
W 1
Figure 18.13 Cambridge saving at diferent real prices.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
700 Raising Keynes
1
P
ID ID (l ) h (l , ) l h
W 2
1
P
ID (l ) h (l , ) l h
W 1
Figure 18.14 Capital-widening investment at diferent real prices.
sponsive to the rate of proft than is saving, ′ − s > 0, at low rates of proft,
whereas at higher rates of proft, it is the other way around, ′ − s < 0. Note
that the assumption that investment is more responsive at low rates of proft is
necessary for the existence of the AD schedule; the intercept of the invest-
ment function in Figures 18.14 and 18.15(a) is negative, so if ′ − s < 0
everywhere, investment never becomes equal to the amount of desired saving.
Te assumption that ′ − s is negative at higher rates of proft is not re-
quired for existence of the AD schedule, but it is, as we shall see, necessary for
the lef-Keynesian intuition that higher wages stimulate employment.
Figure 18.15(b) refects the assumption that ′ − s switches sign: there
are two distinct segments of the AD schedule, formed by the projection of
the points of intersection of the investment and saving schedules in Figure
18.15(a) onto the space of l × P/W.
Copyright © 2021. Harvard University Press. All rights reserved.
Unsurprisingly, equilibrium is more complicated in this model. To stream-
line the argument we will focus on fxprice adjustment. Not only does this
simplify, fxprice dynamics will prove, in the next chapter, to be more relevant
empirically, just as in chapter 8.
Te shape of the AD schedule gives rise to multiple equilibria, as in Figure
18.16. Te equilibria at E′ and F′ are unstable. Southwest of F′ the trajectory is
downward, approaching the stationary-P/W locus as l falls to zero. Tis is
secular stagnation with a vengeance: investment demand falls evermore short
of saving with the capital stock per worker rising, not because of strong in-
vestment demand but because of ever-weaker demand for output and em-
ployment. Southwest of E′ and northeast of F′, the economy enters the orbit of
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 701
(a) Investment and saving holding real-price constant
1
P
ID (l ) h (l , ) l h
W 2
1
ID , S D P
S D s h (l , ) l 1
W P
2 ID (l ) h (l , ) l h
W
1
1
P
SD s h (l , ) l
W 1
P
W
s
s
P
W 2
Aggregate demand
ID S D
P
W 1
ID S D
ID S D
Copyright © 2021. Harvard University Press. All rights reserved.
l
(b) Aggregate demand varying real price
Figure 18.15 Aggregate demand with Cambridge saving and capital widening.
E; northeast of E′, the trajectory is toward F. Te equilibrium at E is stable
provided prices adjust more rapidly than quantities. Te equilibrium at F is
stable whatever the speeds of adjustment. Details are in the mathematical ap-
pendix.
Te response to a demand shock or a price shock turns out to be the same
for all stable equilibria. But the impact of a wage shock depends on whether
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
702 Raising Keynes
Stationary real price
Supply of goods
P P 0
P 0
W
W
F
s
E s
Aggregate
demand
E
l 0
Supply of labor
F
0
W
Figure 18.16 Equilibrium with Cambridge saving and capital widening.
or not investment is more responsive to proftability than is saving, that is,
on which segment of the AD schedule the equilibrium lies. We take these
changes one at a time.
Consider an increase in investment demand or a decrease in desired sav-
ing. Te AD schedule shifs as in Figure 18.17. Starting from E or F, the new
equilibrium is at E′ or F′; employment increases along with the real price level
and the rate of infation.
Now suppose the price of energy increases, as in Figure 18.18. Net produc-
tivity—output per unit of capital net of energy costs, h(l, )—decreases. Tis
Copyright © 2021. Harvard University Press. All rights reserved.
fall in productivity is refected in the shif in the AD schedule. By itself, this
shif would increase employment when saving is more responsive to proft
ability (starting from E) since more output and employment are required to
generate saving, and would decrease employment when investment is more
responsive (starting from F). But the increase in the price of energy also de-
creases proftability, which is refected in the shif in the GS schedule. Starting
from E (where ′ − s < 0), the increase in employment puts pressure on
prices, and the concomitant fall in proftability leads to a contraction of pro-
duction and employment. Starting from F (′ − s > 0), the fall in employ-
ment reduces the pressure on prices, and employment rises. Figure 18.18 as-
sumes this tension is resolved in favor of lower employment (at F′), but this
result is by no means guaranteed.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 703
Stationary real price
P 0
P
W
W Supply of goods
F F
P 0
s
E s
Aggregate
demand
l 0 E
Supply of labor
0
W
Figure 18.17 An increase in investment or a decrease in saving with Cambridge saving and
capital widening.
Stationary
real price
P Supply of goods
F P 0
W P 0
W
F
s
Copyright © 2021. Harvard University Press. All rights reserved.
s
Aggregate
demand E
l 0
E
Supply of labor
0
W
Figure 18.18 An increase in the price of energy with Cambridge saving and capital widening.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
704 Raising Keynes
Similarly, the efect on the price level and infation depends on the balance
of the two opposing forces. Indeed, prices and employment can move in the
opposite directions—stagfation—as in Figure 18.18. Or they can move in the
same direction. Once again the details are in the mathematical appendix.
Consider now a change in the conventional wage. As Figure 18.19 shows,
the efect of a higher conventional wage depends critically on the relative re-
sponsiveness of investment and saving to changes in proftability. In both
cases, proftability falls as the conventional wage rises and greater pressure is
put on wages. Starting from E, the efect of higher wages is refected in lower
profts and lower saving. Employment per unit of capital increases so that sav-
ing will continue to match the (relatively) fxed level of investment demand.
Infation also increases—overall, a Phillips outcome, in which infation and
employment move in the same direction.
If the starting point is at F, where investment is more sensitive to proftabil
ity than is saving, the higher conventional wage leads to less employment be-
cause of the negative impact on investment demand. Higher wages mean
lower profts in this case as well, but lower p rofts discourage investment de-
mand more than saving, and employment falls because of the reduction in
aggregate expenditure. In this case, infation also falls, once again a Phillips
outcome.
Bhaduri and I highlighted this possibility as a counterpoint to the lef-
Stationary real price
P
W P 0
F
Supply of goods W
P 0 F
s
Copyright © 2021. Harvard University Press. All rights reserved.
s
Aggregate demand E
l 0 E
Supply of labor
0
W
Figure 18.19 An increase in the conventional wage with Cambridge saving and capital widening.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 705
Keynesian contention that higher wages are always good for employment
(Marglin and Bhaduri 1990; Bhaduri and Marglin 1990). Our intention was
not to ofer a countervailing dogma, but to argue that the sensitivity of invest-
ment demand to proftability had to be taken into account to assess the im-
pact of wages on employment. Te contrast in Figure 18.19 between displac-
ing the equilibrium at E and displacing the equilibrium at F makes this clear.
But there is more to the story. Once we drop the assumption that invest-
ment invariably takes place solely for the purpose of widening capital, sensi-
tivity of investment demand to proftability may reinforce rather than under-
mine lef-Keynesian policy views.
Capital Deepening
Capital deepening is in crucial respects opposite to capital widening. Te pro-
spective return to a single unit of investment is
−1 −1
P MPK P h − hl l
W − h ≡ − h ,
MPL W hl
since the marginal product of labor is the derivative of h(l, ) with respect to l
MPL = hl
and with constant returns to scale
MPK = h − hll.
Investment demand becomes
P −1 h − h l
l
ID = ψ − ρh ,
W hl
with the function, as before, refecting the animal spirits of the business
Copyright © 2021. Harvard University Press. All rights reserved.
community.
Assuming that saving is a constant fraction of output (rather than of p roft),
the AD schedule can be derived in the same manner as we derived Figure
18.15. Te picture is in Figure 18.20.
Tere is a new wrinkle that sets the long-run case apart from the short-run
analysis of chapter 10. Unlike capacity-augmenting investment, which is as-
sumed to maintain the labor:capital ratio unchanged, cost-cutting investment
is specifcally intended to reduce the labor:capital ratio. As shown in the
mathematical appendix, the stationary-l locus must shif to take capital deep-
ening into account. In a fxprice regime, the stationary-l locus is a rightward
displacement of the AD schedule.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
706 Raising Keynes
(a) Investment and saving holding real-price constant
I D , SD P 1 h hl l
ID h
W 1 h
l S D sh
P 1 h hl l
ID h
W 2 h
l
( h ) Aggregate demand
P
W
P
W 2 S D ID
ID S D
P
W 1
l1 l2 l
(b) Aggregate demand varying real price
Copyright © 2021. Harvard University Press. All rights reserved.
Figure 18.20 Aggregate demand with capital-deepening investment and
uniform saving propensity.
Te fxprice equilibria are pictured in Figure 18.21. Te equilibrium E′ is
unstable and E is stable. Here the unstable equilibrium is more interesting
than in other cases. If the starting point is to the lef of E′, the economy moves
on a trajectory that follows the stationary-price schedule lefward. Tis trajec-
tory involves ever-less employment per unit of capital. Real wages may ini-
tially rise or fall depending on the starting point, but eventually the real wage
must rise. In a way this smacks of secular stagnation, in which a lack of ag-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 707
Supply of goods
P Stationary real price
W Aggregate demand
P 0
W
Stationary labor:
capital ratio
E l 0
Supply of labor
E
Figure 18.21 Equilibrium with capital-deepening investment and a
uniform saving propensity.
gregate demand means that substituting capital to cut costs eventually takes a
toll on employment. Te saving grace is that stagnation does not imply falling
living standards—at least not for the workers who retain their jobs.
Te equilibrium at E is a bit strange: the boss’s right hand does not know
what the lef hand is doing, and vice versa. Te right hand is substituting cap
ital for labor because it is cheaper to produce with more capital and less labor.
At the same time, the lef hand is hiring more workers because goods are fy
ing of the shelves. (To the right of the AD schedule, expenditure exceeds in-
come and output.) Te two actions just balance each other at E, with the re-
Copyright © 2021. Harvard University Press. All rights reserved.
sult that the labor:capital ratio does not change.9
More than one aspect of economic reality is hidden by consolidating all
output into a single sector in which goods serve equally well for consumption
and investment. Moreover, in the present model the demand and supply ef-
fects of capital deepening take place instantaneously (which is why the sta
tionary-l locus lies to the right of the AD schedule). In reality, the supply side
lags behind the demand efect, so capital deepening plays out diferently in
the short run and the long run.
In the short run, capital deepening, like capital widening, adds to de-
mand: additional output and employment are required to put new plant and
equipment in place. But unlike capital widening, which adds to productive
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
708 Raising Keynes
capacity as the new capital comes on line, capital deepening leaves capacity
unchanged while reducing employment. Te specter of technological unem-
ployment lurking behind capital deepening has haunted capitalism since the
days of the Luddites.10
In the past, for the most part, successive waves of capital deepening, sup-
ported by capital widening, have managed to maintain the demand for la-
bor overall even as particular workers in particular industries have suffered.
But the question has resurfaced, even with some mainstream economists,
whether the future of capital deepening is a threat at the macroeconomic
level as well as the micro level. (See Daron Acemoglu and Pascual Restrepo
[2018].)
Te comparative statics of parameter changes are similar in most respects
to what has already been encountered. Figures 18.22 through 18.25 show the
efect of a higher level of investment demand (or a lower level of desired sav-
ing), the efect of a higher energy price, and, fnally, the efect of a higher
conventional wage.
As has been the case right along, an increase in investment demand or a
decrease in desired saving, pictured in Figure 18.22, expands employment
Supply of goods Stationary real price
P
P 0
W
Aggregate demand W
F
l 0
E Stationary labor:
Copyright © 2021. Harvard University Press. All rights reserved.
capital ratio
Supply of labor
Figure 18.22 Capital-deepening investment: an increase in investment
demand or a decrease in desired saving.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 709
Supply of goods
Stationary real price
P
Aggregate
W P 0
demand
W
Stationary labor:
F
capital ratio
l 0
E
Supply of labor
Figure 18.23 Capital deepening: an increase in the price of energy.
Supply of goods
P
W Stationary real price
Aggregate demand P 0
W
Stationary labor:
Copyright © 2021. Harvard University Press. All rights reserved.
E capital ratio
F l 0
Supply of labor
Figure 18.24 Capital deepening: an increase in the conventional wage, I.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
710 Raising Keynes
Supply of goods
P
Stationary
W real price
Aggregate demand
P 0
W
Stationary labor:
capital ratio
l 0 E
F
Supply of labor
Figure 18.25 Capital deepening investment: an increase in the conventional
wage, II.
and output and raises infation. Te efect of a higher price of energy also par-
allels earlier results. Both the real price and the level of employment can go up
or down. And they can move in opposite directions, as in Figure 18.23, or in
the same direction.
A change in the conventional wage, pictured in Figures 18.24 and 18.25,
refects a key diference between capital deepening and capital widening.
When the purpose of investment is to substitute capital for labor, an increase
in the conventional wage is unambiguously associated with a higher level of
economic activity. Te reason is that a higher conventional wage drives the
equilibrium real wage up; a higher real wage means more investment demand
Copyright © 2021. Harvard University Press. All rights reserved.
because the higher the real wage, the more proftable it is to substitute capital
for labor. Tis is of course the opposite of capital widening: when the purpose
is to expand capacity, investment demand falls as the conventional wage rises
and the proft rate falls. Since the mix of capital deepening and capital widen-
ing is likely to vary countercyclically, this diference is an important reason
why the relationship between real wages and macroeconomic performance
will vary over the cycle.
Observe that a higher conventional wage is not only expansionary, it is also
infationary, that is, the new equilibrium F in Figure 18.24 is associated with a
higher level of infation than is the original equilibrium E. Displacement of
equilibrium on the falling portion of the stationary-l schedule, as in Figure
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Keynes in the Long Run 711
18.25, is more complicated, but the mathematical appendix shows that the
overall efect is to increase the rate of infation in this case as well.
Wages, Prices, and Employment: A Summary of Results
Tables 18.3 to 18.6 summarize the results of the previous sections.11 A striking
conclusion is that the positive relationship between employment and infation
embodied in the Phillips curve exists independently of how saving and in-
vestment are characterized.
A second result is the variability of the response to an exogenous price
shock. An anti-Phillips, stagfationary, result, in which infation and employ-
ment move in opposite directions, is one possibility, but Phillips-type move-
ments, in which infation and employment move in the same direction, are
also possible. More surprising is the possibility that an increase in the price of
energy can lead to higher employment.
Finally, results with respect to the conventional wage put lef-Keynesian
Table 18.4 Cambridge Saving with Capital
Table 18.3 Fixed Investment with Saving a Widening: Saving More Sensitive to
Fixed Fraction of Income Proftability
Infation Infation
Employment Employment
Investment↑ or Saving↓ ↑ Investment↑ or Saving↓ ↑
↑ ↑
Price of Energy↑ ↑ Price of Energy↑ ?
↑ ?
Conventional Wage↑ ↑ Conventional Wage↑ ↑
0 ↑
Copyright © 2021. Harvard University Press. All rights reserved.
Table 18.5 Cambridge Saving with Capital
Widening: Investment More Sensitive to Table 18.6 Capital Deepening with Saving a
Proftability Fixed Fraction of Income
Infation Infation
Employment Employment
Investment↑ or Saving↓ ↑ Investment↑ or Saving↓ ↑
↑ ↑
Price of Energy↑ ? Price of Energy↑ ?
? ?
Conventional Wage↑ ↓ Conventional Wage↑ ↑
↓ ↑
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
712 Raising Keynes
beliefs on the benefcial efects of high wages in perspective. If investment
demand is confned to capital widening and saving depends on profts, the
relationship between wages and employment depends on the relative respon-
siveness of investment and saving to proftability. A higher conventional wage
is associated with greater aggregate demand and higher employment if saving
is more responsive, but a lower conventional wage promotes employment
if investment is more responsive. By contrast, the efect of the wage level on
capital deepening is not at cross purposes with the efect on saving: a higher
wage level stimulates capital deepening at the same time as it stimulates con-
sumption demand. In both Tables 18.4 and 18.6 the efect is higher employ-
ment and higher infation.
It remains to be seen how a Keynesian long-run model stacks up against
the historical data for the U.S. economy. Tis is explored in the next chapter.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. mathematical appendix .
Te basic fexprice model is
•
P
W P P ∗
≡ P − W
= [ I − S] − − ,
1 3 (18.1)
P P W W W
W
P −1
l = 2 − ( hl ) , (18.2)
W
and the fxprice version of the model is
•
P
W P P ∗
= θ l − GS P , ξ − θ − , (18.3)
P 1 W 3
W W
W
l = 2 (I − S). (18.4)
Te novel element is the wage part of equations (18.1) and (18.3)
W P P ∗
= 3 − ,
W W W
which says that workers are able to put pressure on money wages when the
actual real wage is below the conventional wage, in other words, when P/W >
(P/W)*, and that employers have the upper hand when this inequality is re-
versed.
Copyright © 2021. Harvard University Press. All rights reserved.
I and S denote, respectively, investment demand and desired saving per
unit of capital. In the simplest case investment is a function of the hurdle rate
alone
I = ( h), (18.5)
and a uniform propensity to save gives the saving function
S = sh(l, ), (18.6)
713
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
714 Raising Keynes
where h(l, ) is the output:capital ratio, net of real energy costs, the assump-
tion being that all energy is purchased abroad on a competitive market at an
exogenously given nominal price PE:
h(l , ) = max[ f (l , e) − e] = f (l , e(l , )) − e(l , ). (18.7)
e
Tat is,
Y ≡ F(K, L, E),
Y
≡ F (1, l , e) ≡ f (l , e),
K
P
≡ E,
P
L
l ≡ ,
K
E
e ≡ ,
K
hl = f l ,
h = −e,
el = − f el f ee−1,
hl = f le e = f le f ee−1,
hl = −el = f le f ee−1,
( )
hl l = f l l + f l e el = f l l − f l2e f ee−1 = f ee−1 f l l f ee − f l2e .
In equation (18.7), e = e(l, ) is the amount of energy that, for given l and ,
maximizes the expression f (l, e) − e. We assume diminishing returns to each
Copyright © 2021. Harvard University Press. All rights reserved.
input, so that both f ll and fee are negative. We also assume that energy and la-
bor are complements, with the result that fel (= f le) is positive. hll must be nega-
tive; otherwise equality between the marginal productivity of labor and the
real wage
−1
P
hl = f l =
W
characterizes minimum rather than maximum p rofts. In other words, the as-
sumption of an interior solution to the p
roft-maximization problem implies
that diminishing returns to each input outweigh the complementarity of the
two inputs.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 715
With fexprice adjustment and investment and saving in the simple form of
(18.5) and (18.6), we have
J = W
(
1 I P − S P − 3
W
) − 1 Sl −
= 3
−1shl
, (18.8)
2 hll ( hl )
−2
−2
2 2 hll ( hl ) 2
for which the equilibrium is stable, since tr J < 0 and det J > 0.
For the comparative statics of varying aggregate demand, goods supply,
and the conventional wage, we calculate
P
∂W
∂x
∂l
∂x
as the solution to the equation system
P
∂W ∂AA
∂x
J ∂x = − ∂BB . (18.9)
∂l
∂x
∂x
Equation (18.9) is obtained by diferentiating the system
P P ∗
AA ≡ θ1 ψ ( ρh ) − sh − θ 3 − = 0, (18.10)
W W
P −1
BB ≡ 2 − ( hl ) = 0, (18.11)
W
with respect to the parameters x = h, , and (P/W)*. (We omit a calculation of
the efect of changing the parameter s since the math is symmetrical with the
Copyright © 2021. Harvard University Press. All rights reserved.
corresponding operation with respect to h.) Te solution to (18.9) is
P
∂W ∂AA
−1
∂x
∂x = − J ∂BB , (18.12)
∂l
∂x
∂x
−1
where J is the inverse of the Jacobian J.1 In the 2 × 2 case
j11 j12
J = ,
j21 j22
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
716 Raising Keynes
we have
j22 j12
J 1 1 ,
j21 j11
where = det J. In the specifc case of (18.6), we have
θ h ( h ) −2 θ1shl
J −1 = ∆ −1 2 ll l .
−θ 2 −θ 3
Diferentiating equation (18.10) gives
∂AA
∗
∂AA ∂AA ∂ P
∂ρ θ ψ′ −θ1shξ W θ 3
h = 1 , ∂ξ = , = ,
∂BB 0 ∂BB θ 2hl ξ ( hl ) −2 ∂BB 0
∂ρ
∂ξ P ∗
h
∂
W
with the results
P
∂W
−1 θ ψ′θ h ( h )
−2
−
∂ρh = −∆ 1 2 ll l = , (18.13)
−θ 2 θ1 ψ′ −
∂l
∂ρh
P
∂W θ θ s ( h ) −2 ( −h h + h h )
−1 1 2 l ξ ll l lξ
= + ,
∂ξ = −∆
( )
(18.14)
θ θ sh − θ 3 ( hl ) hl ξ ±
−2
∂l 2 1 ξ
∂ξ
P
Copyright © 2021. Harvard University Press. All rights reserved.
∂W
∗
P
−1 θ 3 θ 2hll ( hl )
∂ −2
+
W
= − ∆ = . (18.15)
− θ θ
+
∂l 3 2
∗
∂ P
W
Tese results confrm the graphical reasoning in Figures 18.7(a), 18.9,
18.10, and 18.12(a). Te ambiguity in the response of employment to an
energy-price shock is because the two terms in the expression 1sh − 3(hl)−2hl
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 717
have opposite signs. Which dominates depends on the relative magnitudes of
the adjustment speeds 1 and 3.
Te efect on infation of these shocks is more complicated. We have
P
= θ1 ψ ( ρh ) − sh(l , ξ) . (18.16)
P
Te complication is that for demand and energy-price shocks, the rate of
price infation, measured by the horizontal distance between the equilibrium
and the AD schedule (along which P/P = 0), moves not only because the equi-
librium changes but also because the AD schedule itself shifs. Te general
formula is
P P P P P
d ∂ ∂ ∂ ∂
P = P W + P ∂l + P . (18.17)
dx P ∂x ∂l ∂x ∂x
∂
W
From equation (18.16) we have
P P
∂ ∂
P = 0, P = − sh
1 l
P ∂l
∂
W
and
P P P
∂ ∂ ∂
P = θ ψ′, P = −θ sh , P = 0.
1 1 ξ ∗
∂ρh ∂ξ P
∂
W
Putting these results together with equations (18.13), (18.14), and (18.15),
we have
Copyright © 2021. Harvard University Press. All rights reserved.
P
d
P = −∆ −1θ θ ψ′θ sh + θ ψ′ = −∆ −1θ θ θ h h −2 ψ′ = − ,
2 1 1 l 1 1 2 3 ll ( l )
d ρh
P
d
dξ 2 (
1 ξ 3 ( l ) lξ 1 l 1 ξ )
P = ∆ −1θ θ sh − θ h −2 h θ sh − θ sh = −∆ −1θ θ θ s h −2 h h − h h = +,
1 2 3 ( l ) ( l lξ ξ ll )
P
d
P = −∆ −1θ θ θ sh = − .
∗ 1 2 3 l
P
d
W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
718 Raising Keynes
Again, these results confrm the efects of parameter changes analyzed graph-
ically in the text: the increase in aggregate demand in Figure 18.7(a), which
we may imagine to be the result of a decrease in the hurdle rate of interest;
the increase in the price of energy in Figures 18.9 and 18.10; and the increase
in the conventional wage in Figure 18.12(a). All three of these parameter
changes lead to an increase in infation in line with the above formulas.
For the fxprice system (18.3) and (18.4), with investment and saving deter-
mined by (18.5) and (18.6), the Jacobian is
−1GS P − 3 1
J = W ,
0 − 2 shl
and its inverse is
−θ 2 shl − θ1
J −1
=∆ −1
0 −θ1GS P − θ3 .
W
As in the fexprice case, tr J < 0, and det J > 0, so the equilibrium is stable.
Comparative statics of demand, energy- price, and conventional- wage
shocks are given by (18.12). In the fxprice case, we have
P P P ∗
AA = θ1 l − GS , ξ − θ 3 − = 0,
W W W
BB = θ 2 ψ ( ρh ) − sh = 0,
and the derivatives are
∂AA
∗
∂AA ∂AA ∂ P
∂ρ 0 ∂ξ −θ1GSξ W θ 3
Copyright © 2021. Harvard University Press. All rights reserved.
h = , = , = ,
∂BB θ 2 ψ′ ∂BB −θ 2 s πhξ ∂BB 0
∂ρ ∂ξ P ∗
h
∂
W
so the fxprice counterparts of (18.13), (18.14), and (18.15) are
P
∂ W
−θ 2 θ1 ψ′ −
∂ρh
(
= −∆ − θ GS + θ θ ψ′ = ,
)
−1
∂l 1 P 3 2 −
W
∂ρh
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 719
P
∂W
θ 2 shl θ1GSξ + θ1θ 2 shξ +
( )
= −∆ θ1GS P + θ 3 θ 2 s πhξ = ,
−1
∂ξ
∂l +
W
∂ξ
P
∂W
∗
P
∂ −1
−θ 2 shl θ 3 +
W = −∆ 0
= .
∂l 0
∗
∂ P
W
With fxprice adjustment, price infation is governed by
P P
= θ1 l − GS , ξ ,
P W
and therefore
P P
∂ ∂
P = − GS , P = ,
1 P 1
P ∂l
∂ W
W
P P P
∂ ∂ ∂
P = 0, P = −θ GS , P = 0.
1 ξ ∗
∂ρh ∂ξ P
∂
W
Consequently, infation results are
Copyright © 2021. Harvard University Press. All rights reserved.
P
d
P = ∆ −1θ θ θ ψ′ = −,
1 2 3
d ρh
P
d
P = −∆ −1θ θ θ sh + sh GS = + ,
1 2 3 ( ξ l ξ )
dξ
P
d
P = −∆ −1θ θ θ sh GS = −.
∗ 1 2 3 l P
P W
d
W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
720 Raising Keynes
Tese results parallel the graphical results in Figures 18.7(b), 18.11, and
18.12. Te novelties are that (1) the possibility of an energy-price shock lead-
ing to stagfation is eliminated—in the fxprice case both employment and
infation increase, and (2) the conventional wage has no bearing on the level
of employment—in the fxprice case the conventional wage has no impact on
investment or on saving and therefore no impact on aggregate demand.
Complicating Investment and Saving
At the price of mathematical complication, we can replace the simple invest-
ment and saving functions based on the hurdle rate of interest and a uniform
propensity to save by investment functions that take account of capital widen-
ing, capital deepening, and the Cambridge saving theory. In its most general
form, we can write investment as a function of P/W, l, and h, namely, I(P/W,
l, h), and saving as a function of P/W and l, S(P/W, l). Specifcally, we refect
capital widening in the investment equation
P
−1
I = ψ ( Ω(l )r − ρh ) = ψ Ω(l ) h(l , ξ) − l − ρh , (18.18)
W
and capital deepening in the equation
P −1
h − hl l
I = ψ − ρh . (18.19)
W hl
Te simple version of the class-based theory of saving incorporated in the
Cambridge saving theory is
P
−1
S = s πr = s π h(l , ξ) − l . (18.20)
W
As in the body of this chapter, we focus on the fxprice model summarized
in equations (18.3) and (18.4). Te stability conditions for various permuta-
Copyright © 2021. Harvard University Press. All rights reserved.
tions of this model, as well as the conditions for the fexprice variant, are
examined in my 2017 Review of Keynesian Economics paper in exhaustive,
one might say excruciating, detail. Here I can be more selective, focusing
on the diference between a relatively high responsiveness of investment
to proftability, ′ − s > 0, and a relatively high responsiveness of saving,
′ − s < 0, as well as examining the diference between capital widening
and capital deepening.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 721
For (18.18) and (18.20) we have
−θ1GS P − θ 3 θ1
W
P
−1
J = ψ′Ω′ h − l .
W
−2
θ ( ψ′Ω − s ) P l
π θ2 −1
2 W + ψ′Ω − s h − P
( π)
l W
Te sign of j22 as well as j21 depends on whether the expression ′ − s is
positive or negative. Te slope of the AD schedule is given by the equation
P
−1
P
−1
P θ2 ψ′Ω′ h − l + ( ψ′Ω − s π ) hl −
d W W
W = −
,
−2
dl AD P
θ 2 ( ψ′Ω − s π ) l
W
and knowing the sign of ′ − s determines the sign of the denominator of
the expression on the right-hand side. Whether a point is on the downward-
or upward-sloping portion of the AD schedule determines the sign of the ex-
pression as a whole, numerator and denominator. Given the minus sign in-
front of the ratio, the numerator must be of the same sign as the denominator
on the downward-sloping portion of the AD schedule and of the opposite
sign on the upward-sloping portion. Tat is, the inequality (d(P/W)/dl)AD < 0
holds on the downward- sloping portion, so ′ − s > 0 implies that
the numerator must be positive. On the upward-sloping portion we have
(d(P/W)/dl)AD > 0, so the implications of the sign of ′ − s for the numera-
tor are reversed.
Tis information is crucial to determining the stability of equilibrium.
Copyright © 2021. Harvard University Press. All rights reserved.
Consider frst the case where investment is less responsive than saving to
changes in proftability, ′ − s < 0. Te condition tr J < 0 is satisfed at E in
Figure 18.16 only if the adjustment speed of employment is slow relative to
the adjustment speeds of prices and wages since
P
−1
P
−1
θ 2 ψ′Ω′ h − l + ( ψ′Ω − s π ) hl −
W W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
722 Raising Keynes
is necessarily positive. Te determinant condition, det J > 0, is satisfed since
the stationary real-price locus is steeper than the AD schedule. We have
P
−1
P
−1
−θ1GS P − θ 3 θ 2 ψ′Ω′ h − l + ( ψ′Ω − s π ) hl −
W W W
P
−2
> θ1 θ 2 ( ψ′Ω − s π ) l
W
because
P
d
W
• =
θ1
dl = 0 θ1GS P + θ 3
P
W
W
P
−1
P
−1
θ 2 ψ′Ω′ h − l + ( ψ′Ω − s π ) hl −
W W
>− −2
P
θ 2 ( ψ′Ω − s π ) l
W
P
d
W
= .
dl AD
In contrast, at F the trace condition is automatically satisfed since, with
′ − s > 0, the inequality
P
−1
P
−1
θ 2 ψ′Ω′ h − l + ( ψ′Ω − s π ) hl − < 0
W W
Copyright © 2021. Harvard University Press. All rights reserved.
holds on the rising portion of the AD schedule. Te condition det J > 0 is also
satisfed because at F the AD schedule is steeper than the stationary real-price
locus. Tat is,
P dl
d > P .
W d P •
dl AD W = 0
W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 723
Te efects of demand and supply shocks are the total derivatives of the
equation system
P P P ∗
AA ≡ θ1 l − GS , ξ − θ 3 − = 0,
W W W
P P
−1 −1
BB ≡ θ 2 ψ Ω(l ) h(l , ξ) − l − ρh − s π h(l , ξ) − l θ 2 = 0,
W W
with respect to the exogenous parameters x = h, , and (P/W)*. We have
−θ1GS P − θ 3 θ1
W
P
−2 P
−1
−1
P
θ 2 ( ψ′Ω − s π ) W l θ 2 ψ′Ω′ h − l + ( ψ′Ω − s π ) hl −
W W
P
∂W ∂AA
∂x
∂x = −
∂l ∂BB
∂x
∂x
for which the solution is
P
−1
ψ′Ω′ h − l
P W ∂AA
∂W θ2 −1
− θ1
−1
+ ( ψ′Ω − s π ) hl − P
∂x .
∂x = −∆ ∂BB
∂l W
∂x
P
−2
∂x
−θ 2 ( ψ′Ω − s π ) l −θ1GS P − θ 3
Copyright © 2021. Harvard University Press. All rights reserved.
W W
For x = h, we have
∂AA
∂ρ 0
h = .
∂BB −θ 2 ψ′
∂ρ
h
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
724 Raising Keynes
Te results are
P
∂
W = θ1θ 2 ψ′ = − ,
∂ρh ∆
∂l
=
( θ GS
1 P
W
)
+ θ 3 θ 2 ψ′
= −.
∂ρh ∆
Combining the efects of interest-rate changes on real price and employ-
ment gives the impact on infation
P P P P
d ∂ ∂ ∂
P = P W + P ∂l = θ1θ 2 θ 3 ψ′ = −
d ρh P ∂ρh ∂l ∂ρh ∆
∂
W
since
P
∂
P = − GS ,
1 P
P
∂ W
W
P
∂
P = .
1
∂l
Monetary easing, refected here by a reduction in h, stimulates the econ-
omy on a permanent basis, at least to the point beyond which the hurdle rate
can no longer be reduced (a liquidity trap) or to the point that the responsive-
Copyright © 2021. Harvard University Press. All rights reserved.
ness of investment demand ′ goes to zero. Tese limits apart, monetary eas-
ing increases the labor:capital ratio as well as the infation rate: the Phillips
curve exists in the long run as well as in the short period.
Supply-side changes are more complicated. A change in the real price of
energy gives the right-hand side of equation (18.12) as
∂AA
∂ξ −θ1GSξ
= .
∂BB θ 2 ( ψ′Ω − s π ) hξ
∂ξ
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 725
In consequence,
P
−1
′ ′
ψ Ω h − l
θ1θ 2 W GS + θ θ ψ′Ω − s h
−1 1 2 ( π) ξ
P P ξ
∂ + ( ψ′Ω − s π ) hl −
W = W
= ±,
∂ξ ∆
−2
P
−θ1θ 2 ( ψ′Ω − s π ) lGSξ + θ1GS P + θ 3 θ 2 ( ψ′Ω − s π ) hξ
∂l W
= W = ±,
∂ξ ∆
P P P P P
d ∂ ∂ ∂ ∂
P = P W + P ∂l + P
dξ P ∂ξ ∂l ∂ξ ∂ξ
∂
W
−1
P
−1
P
GSξ ψ′Ω′ h − l + ( ψ′Ω − s π ) hl −
θ1 θ 2 θ 3 W W
(
π) ξ
+ ψ ′Ω − s h = ±.
=
∆
Since GS = −hl/hll < 0, the signs of all these derivatives are ambiguous; the
efects on demand and supply work in opposite directions. Suppose the efect
of a change in the price of energy on the supply side is negligible, GS ≈ 0. Te
demand-side efect on the real price and infation is via −h = e; the strength
of the efect depends on the energy-intensity of production, but in any case
both the real-price level and the rate of infation fall provided investment is
more responsive than saving to a change in proftability, ′ − s > 0. In this
Copyright © 2021. Harvard University Press. All rights reserved.
case, employment also falls because demand falls along with productivity and
proftability. If ′ − s < 0, then real price, infation, and employment re-
spond positively to an increase in the price of energy; now, the fall in produc-
tivity means that more workers are required to meet the relatively inelastic
demand.
At the other extreme—when e ≈ 0—supply-side efects dominate, and the
decrease in proftability drives prices and production up when ′ − s > 0,
and down when the inequality is reversed. When both supply and demand
are afected, the consequences for employment, the real price, and infation
cannot be determined on the basis of the qualitative structure of the model,
but we can say that the stronger is the energy intensity of production, e, the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
726 Raising Keynes
more likely is an increase in to increase employment, the real price, and in
fation.2 Except for the extreme cases GS ≈ 0 and e ≈ 0, there is no reason why
employment and prices will necessarily move in the same direction: in the
general case, the product
P
∂
W ∂l
∂ ∂
is itself sign indefnite. Figure 18.18 pictures the stagfationary case in which
prices and employment move in opposite directions.
Te efect of a change in the conventional wage is more straightforward.
We have
P
−1
P
−1
P −θ 3 θ 2 ψ′Ω′ h − l + ( ψ′Ω − s π ) hl −
∂
W = W W
∗
P ∆
∂
W
+ F in Fig. 18.16, ψ′Ω − s π > 0
= at ,
_ E in Fig. 18.16, ψ′Ω − s π < 0
−2
P
θ 3 θ 2 ( ψ′Ω − s π ) l
∂l W + F in Fig. 18.16, ψ′Ω − s π > 0
= = at ,
P
∗
∆ _ E in Fig. 18.16, ψ′Ω − s π < 0
∂
W
P
−1
P
−1
P ( π ) l
∂ θ θ θ
1 2 3 ψ′Ω′ h − l + ψ′Ω − s h − W GS P − AD P
P = W W W
∗
P ∆
∂
W
Copyright © 2021. Harvard University Press. All rights reserved.
± F in Fig. 18.16, ψ′Ω − s π > 0
= at .
_ E in Fig. 18.16, ψ′Ω − s π < 0
Where ADP/W is the inverse of (d(P/W)/dl)AD. Te sign of ∂(P/P)/∂(P/W) *
evi-
dently depends on the sign of GSP/W − ADP/W. We know this expression
is negative at E because the stationary-price locus is steeper than the AD
schedule:
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 727
P
−1
ψ′Ω′ h − l
W
θ2 −1
P
P + ( ψ′Ω − s ) h − d P
d π l W
W θ1 W
P • = >− −2
= ,
dl = 0 θ1GS P + θ 3 P dl AD
W
W θ 2 ( ψ′Ω − s π ) l
W
so
3
GS P − AD P < − .
W W
1
Te takeaway is that an increase in the conventional wage, a decrease in
(P/W)*, can increase the equilibrium price level and infation, and—as lef-
Keynesians have argued since the afernoon of the appearance of Te General
Teory—increase employment. Te critical assumption is that in Figure 18.19
we start from E: saving is class-based and responds more to proftability than
does investment.
Capital Deepening
In the long-run analysis of capital deepening, we cannot ignore the change in
the labor:capital ratio due to the investment itself. Unlike capacity-augmenting
investment, which is assumed to maintain the labor:capital ratio unchanged,
cost-cutting investment is specifcally intended to reduce the labor:capital ra-
tio. As noted in the body of this chapter, this supply-side efect is supposed to
take place instantaneously, at the same time as investment adds to aggregate
demand. To capture the impact of capital deepening on supply, we start from
the identity
L K L
l = −
Copyright © 2021. Harvard University Press. All rights reserved.
L K K
and the assumption that capital substitutes for labor
( h − hl l ) K = −hl L ,
so
h − hl l P h − hl l
−1
L h − hl l K K
=− =− ψ − ρh l −1.
L hl K L hl W hl
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
728 Raising Keynes
Assuming for this exercise that investment intentions are realized, which is
to say,
K P −1 h − h l
l
= ψ − ρh ,
K W hl
the supply-side efect is
h P h − hl l
−1
l = − ψ − ρh .
hl W hl
In the fxprice version of the model, with saving a constant fraction of total
income, the stationary-l locus combining demand and supply efects is
P −1 h − h l h P −1 h − h l
l = θ 2 ψ l
− ρh − sh(l , ξ) − ψ l
− ρh .
W hl
hl l W hl
How to go from aggregate demand to stationary l is shown in Figure 18.26. In
panel (a) investment demand
P −1 h − h l
l
ψ − ρh
W hl
is replaced by
h P h − hl l
−1
Z = θ2 −
ψ − ρh , (18.21)
hl l W hl
and desired saving sh(l, ) is replaced by 2sh(l, ). Te construction in Figure
18.26(b) follows from the fact that equation (18.21) has an interior maximum.
To see this, set ∂Z/∂l equal to zero. Tis gives the following equation:
( hl )
2 −1
Copyright © 2021. Harvard University Press. All rights reserved.
l − hhll l − hl h h P hh
− ψ= − θ 2 ψ′ − ll 2 . (18.22)
( hl l ) W ( hl )
2
hl l
Assuming a constant-elasticity production function,
f(l, e) = A(1 + 2l + 3e)1/ (18.23)
and combining equations (18.7) and (18.23), the elasticity of substitution σ is
given by the equation
1 fl ( f − fl l ) hl ( h − hl l )
σ = =− =− .
1−ζ f lf ll hlhll
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 729
(a) Investment, saving and stationary l holding
real-price constant
I, S P 1 h hl l
2 h
W 1 hl
2 sh
1
h P h hl l
2 h
h l
l W 1 hl
h
2 h l ( h ) l
l
P
W
Aggregate demand l 0
P
W 1
l
P P
(b) Aggregate demand and l 0 for
W W 1
Figure 18.26 Aggregate demand and stationary l.
We have therefore
hll (1 − ζ) λ1
Copyright © 2021. Harvard University Press. All rights reserved.
− = ,
( hl ) λ1hl l + λ2hl l ζ +1
2
h λ + λ lζ
= 1 ζ2 .
hl l λ2 l
So equation (18.22) becomes
ζ λ + λ lζ P −1
ψ = (1 − ζ) 1 ζ 2 − θ 2 ψ′ . (18.24)
l λ2l W
We assume < 1 and consequently < 0. So if is bounded away from
zero, the lef-hand side of equation (18.24) falls from +∞ to −∞ as l goes from
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
730 Raising Keynes
zero to ∞. Te intercept of the right-hand side at l = 0 is negative, and the RHS
goes to +∞ as l goes to ∞, so there must exist an interior solution to this equa-
tion. Te solution corresponds to a maximum of Z rather than a minimum
because the LHS is falling and the RHS is rising, or at least is falling less rap-
idly. Te shape of the stationary-l locus in Figures 18.21 to 18.25 follows from
varying the real-price in Figure 18.26.
With price changes governed by
•
P
W
= θ l − GS P , ξ − θ P − P
∗
P 1 W 3
W W
W
and changes in the labor:capital ratio by
h P h − hl l
−1
l = θ 2 −
ψ − ρh − θ 2 sh(l , ξ) ,
hl l W hl
the Jacobian is
j j12
J = 11
j21 j22
− θ1GS P − θ 3 θ1
W
( )
2
h l − hh l − h h
− l ll l
ψ
=
−2
( hl l )
2
− θ − h ψ′ h − hl l P
2 hl l hl W h P
−1
h h
ψ′ − − θ 2 shl
ll
+ θ2 −
hl l W ( hl ) 2
Copyright © 2021. Harvard University Press. All rights reserved.
−θ1GS P − θ 3 θ1
W
ζ λ
λ + λ lζ P
−1
= ψ + (1 − ζ) θ 2 − 1 ζ 2 ψ′ 1 ζ
−2
λ + λ lζ λ P
− θ 2 − 1 ζ 2 ψ′ 1 ζ
l λ2 l W λ2 l
λ2 l λ2 l W −θ 2 shl
− +
= .
− ±
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 731
Te negative sign of j21 is the result of the inequality 2 > (1 + 2l )/2l , which
is necessary for the existence of an equilibrium that accommodates desired
saving, desired investment, and the negative adjustment of l to investment,
that is, for the existence of the stationary-l schedule. Te sign of j22 is positive
on the upward-sloping branch of the stationary-l schedule, negative on the
downward-sloping branch.
Consequently, since stability requires both
ζ λ + λ lζ P −1 λ1
tr J = −θ1GS P − θ 3 + ψ + (1 − ζ) θ 2 − 1 ζ 2
ψ′ − θ 2 shl < 0
l W λ2l
ζ
W λ2l
and
ζ λ + λ lζ P −1 λ1
det J = −θ1GS P − θ 3
l
ψ + (1 − ζ) θ 2 − 1 ζ 2 ψ′ − θ 2 shl
W λ2l
ζ
W λ2l
−2
λ1 + λ2l ζ λ P
+ θ1 θ 2 − ′ 1
λ2l ζ ψ λ l ζ W > 0
2
on the upward-sloping branch of the stationary-l locus, prices and wages
must adjust faster than employment, and the stationary real-price locus must
be steeper than the stationary-l locus. Tus the equilibrium at E′ in Figure
18.21 is unstable, whereas the equilibrium at E is stable provided the trace
condition is satisfed. On the downward-sloping branch of the stationary-l
locus, both stability conditions are satisfed since j22 < 0.
Comparative statics lead to the now familiar results that a reduction in the
hurdle rate of interest promotes both employment and higher prices, whereas
an increase in the price of energy leads to ambiguous results. A higher con-
ventional wage—a lower (P/W)*—leads to more employment, the same result
that takes place when the impact on saving dominates the impact on capital-
widening investment, but this result is achieved by a very diferent mecha-
Copyright © 2021. Harvard University Press. All rights reserved.
nism. Te impact on infation is in the same direction as the impact on em-
ployment.
We consider the comparative-statics of displacing h, , and (P/W)* in or-
der. We have
P
AA
W j
1 22
j12 x
x j ,
l 21 j11 BB
x
x
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
732 Raising Keynes
where now
P P P ∗
AA = θ1 l − GS , ξ − θ 3 − ,
W W W
h P h − hl l
−1
BB = θ 2 −
ψ − ρh − θ 2 sh(l , ξ).
hl l W hl
Te results for monetary policy are
P
∂
W = −∆ −1θ θ − λ1 + λ2l
ζ
1
2 ψ′ = − ,
∂ρh λ2l ζ
∂l λ + λ lζ
= −∆ −1 θ1GS P + θ 3 θ 2 − 1 ζ 2 ψ′ = −,
∂ρh λ2l
W
P
d
P = −∆ −1θ θ θ − λ1 + λ2l
ζ
1 3 2 ψ′ = − .
d ρh λ2l ζ
Te results for an energy-price shock are
P
∂
W = −∆ −1 − j θ GS + j θ sh = ± ,
∂ξ
( 22 1 ξ 12 2 ξ )
∂l
= −∆ −1 ( − j21θ1GSξ + j11θ 2 shξ ) = ±,
∂ξ
P
d
P = −θ j θ GS + j θ sh = ± .
3 ( 22 1 ξ 12 2 ξ )
dξ
Copyright © 2021. Harvard University Press. All rights reserved.
Finally, a change in the conventional wage gives
P
∂
W = − ∆ −1θ ζ ψ + (1 − ζ) θ − λ1 + λ2l P −1 λ1
ζ
∗ 3 2 ψ′ − θ 2 shl
l W λ2l
ζ
P λ 2l ζ
∂
W
dl
+ if >0
P
d
= W l= 0 ,
dl
− if < 0
P
d
W l= 0
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 733
∂l −1 λ1 + λ2l ζ λ1 P
−2
∗
= − ∆ θ 3
θ 2 − ψ′ ζ = −,
P
∂
λ2l ζ λ2l W
W
P
d
P = ∆ −1θ θ GS j + j = −θ 1 + ∆ −1θ j
P
∗ 3 1
P 22 21
3 (
3 22 )
d W
W
θ3
= − θ 3 1 − = −.
dl
θ 3 + θ1 GS P −
W d P
W l= 0
Te efect of a change in the conventional wage on employment is unam-
biguously positive. Te change in infation is unambiguously negative, but the
algebra is more complicated. On the rising branch of the stationary-l locus in
Figure 18.24, we can use the fact that j22 is positive, which makes 1 + −13 j22
positive, to conclude that infation moves in the same direction as em
ployment. At an equilibrium on the downward-sloping portion of the sta
tionary-l locus, as in Figure 18.25, both j22 and j21 are negative, which ensures
GSP/W j22 + j21 < 0.
Extending the Cambridge Saving Theory:
Mr. Piketty, Meet Mr. Pasinetti
Luigi Pasinetti (1962) extended the Cambridge saving theory to allow for
workers’ savings and consequent ownership of a share of the capital stock.
Tomas Piketty’s monumental study of wealth dynamics (2014) made Pasi-
netti newly relevant since Pasinetti provides a framework for assessing Piket-
Copyright © 2021. Harvard University Press. All rights reserved.
ty’s claim that wealth distribution is driven by the relationship between the
rate of growth g and the rate of proft r—more specifcally, that the wealth
dynamics resulting from an excess of r over g are responsible for the growing
concentration of wealth. At one point, Piketty calls r > g “the fundamental
force for divergence” (p. 25); at another, “the central contradiction of cap
italism” (p. 571). Here is his argument in a nutshell:
When the rate of return on capital signifcantly exceeds the growth rate of
the economy . . . , then it logically follows that inherited wealth grows faster
than output and income. People with inherited wealth need save only a por-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
734 Raising Keynes
tion of their income from capital to see that capital grow more quickly than
the economy as a whole. (p. 26)
Piketty (2015, p. 49) makes it clear that the problem is not the inequality r > g
per se:
that r > g is certainly not a problem in itself. Indeed, the inequality r > g
holds true in the steady-state equilibrium of the most common economic
models, including representative-agent models where each individual owns
an equal share of the capital stock.
Rather, the problem is that r > g implies a growing concentration of wealth.
Piketty reaches this result with the help of two assumptions. First, r and g
are treated separately rather than as joint outcomes of the growth process.
Second, Piketty’s focus is on the impact of r − g on the fortunes of family dy-
nasties, particularly its role in concentrating wealth by multiplying random
shocks to accumulation, shocks that would otherwise dissipate over time
(Piketty and Saez 2015, sec. 15.5, esp. sec. 15.5.4, pp. 1351–1356).
By contrast, in a Pasinetti framework the focus is on the interaction be-
tween steady-state r and g endogenously determined as outcomes of a growth
model, and the emphasis is on what happens to the capitalist class as a whole
rather on what happens to particular capitalists. In this framework we reach a
conclusion opposite to Piketty’s: there is no logical reason why inherited
wealth should grow more rapidly than the capital stock or output, or faster
than newly acquired wealth, even if the p roft rate exceeds the growth rate.
To apply Pasinetti’s generalized Cambridge saving theory to Piketty’s argu-
ment requires us to redefne capitalists and workers because the traditional
working class plays no role in Piketty’s conception of wealth dynamics. Pre-
sumably, like workers in the original Cambridge theory, chez Piketty workers
save nothing and accumulate no wealth. As far as saving is concerned, their
place is taken by executives, managers, and technocrats, what Piketty calls a
Copyright © 2021. Harvard University Press. All rights reserved.
“patrimonial middle class” (2014, p. 373). Members of the PMC rely on work
as well on accumulated capital for their income. Te capitalists of the Cam-
bridge saving theory are for Piketty rentiers whose wealth is inherited and
who rely completely on p rofts for consumption and further accumulation. As
an empirical shorthand we can identify Piketty’s rentiers with the proverbial
“1 percent” and the PMC with the next 9 percent down, or perhaps more ac-
curately, we can identify the two groups with the top 0.1 percent and the next
9.9 percent.3
Pasinetti’s extension of the Cambridge saving theory adds two assump-
tions, frst, that “workers” (henceforth the PMC)4 save a portion of their in
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 735
comes, sw , albeit a smaller portion than do “capitalists” (henceforth rentiers),
that is, sw < s; second, that, since members of the PMC save, they acquire a
portion of the capital stock and therefore a portion of p rofts. Te saving
function for the rentier class is
K π K P
−1
Sπ = = π δ = s π h(l , ξ) − l δ = s πr δ,
K Kπ W
where = K/K, the fraction of the capital stock owned by rentiers. Te PMC
saves both from its share of p
rofts and from its labor income:
K W P
−1
P
−1
P
−1
Sw = = sw h(l , ξ) − l [1 − δ] + sw l = sw r (1 − δ) + sw l ,
K W W W
where 1 − = KW/K represents the PMC share of the capital stock.
Total saving is the sum of the saving of rentiers and the saving of the PMC:
K K + K W P
−1
= π = s π h(l , ξ) − l δ
K K W (18.25)
P
−1
P
−1
+ sw h(l , ξ) − l [1 − δ] + sw l.
W W
Te shares and 1 − are determined endogenously since the relative
amounts of saving of the two classes determines their relative shares. We have
K K − KK
K π K K π
δ = π π
= − = ( s π r − g ) δ. (18.26)
K2 Kπ K K
It is clear from equation (18.26) that an excess of sr over g will increase the
concentration of wealth in the hands of rentiers. But sr > g is a much more
stringent condition than r > g; the two conditions coincide only if s = 1.
Chapter 9 suggested that we have no theory worth the name of the saving
Copyright © 2021. Harvard University Press. All rights reserved.
behavior of the very rich. Te rich are diferent from the rest of us not only
because they have more money, but there is no reason to believe that all or
practically all of their income is saved.
Moreover, even if we amend Piketty’s central contradiction to refect the
assumption that s is less than one, it is by no means clear that the inequality
sr − g > 0 spells more and more concentration of wealth until rentiers end up
owning all but a vanishing share of the capital stock. Wealth dynamics are
endogenous to the model. It may be the case that sr − g > 0 on the transi-
tional path to equilibrium, while the equilibrium value for (the value of at
which sr = g and = 0) is less than one. Indeed, sr < g is compatible with
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
736 Raising Keynes
equilibrium only if the rentier class ends up with a vanishing share of the cap
ital stock, that is, with = 0.
Substituting from equations (18.18) and (18.25) we obtain equation (18.27),
and the complete dynamic fxprice system with capital widening is
•
P
W
= θ l − GS P , ξ − θ P − P ,
∗
1 W 3 (18.3)
P W W
W
ψ ( Ωr − ρh ) − s π r δ −
l = θ (I − S) = θ −1
2 2 P , (18.27)
sw r[1 − δ] − sw W l
P
−1
δ = [ (1 − δ)s π + δsw ] h(l , ξ) − l − sw h(l , ξ) δ. (18.28)
W
Te condition for a stationary value of requires special comment. For =
0, either = 0, or = (sr − swh)/(s − sw)r. Evidently, > 0 implies that the
growth rate of rentier capital, sr, exceeds the growth rate of PMC capital, sWh,
associated with the PMC owning all but a vanishing share of the capital stock.
A stationary > 0 also implies that the growth rate of rentier capital is equal
to the growth rate of the capital stock, in which case the growth rate in equi-
librium must be less than the p roft rate unless s = 1. Conversely, = 0 im-
plies swh > sr since the growth rate of rentier capital, sr, is necessary less
than the PMC saving rate, swh, when the patrimonial middle class owns the
entire capital stock.
Te conditions for stability are complicated, and to obtain defnite results
we have to make simplifying assumptions. Te Jacobian of equations (18.3),
(18.27), and (18.28) is
Copyright © 2021. Harvard University Press. All rights reserved.
−θ1GS P − θ 3
W
−2
θ 2 ψ′Ω − ( s π − sw ) δ P l
J = W
−2
− (1 − δ)s + δs P l δ
π w
W
θ1 0
P
−1
θ 2 ψ′Ω′r + ( ψ′Ω − ( s π − sw ) δ ) hl − − sw hl −θ 2 ( s π − sw ) r
W
.
P
−1
(1 − δ)s π + δsw hl − − sW hl δ − ( s π − sw ) r δ
W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 737
Writing
j11 j12 j13
J = j21 j22 j23 ,
j j32 j33
31
it is clear that only the frst of the four stability conditions
1 = −tr j = −(j11 + j22 + j33) > 0,
2 = −(j13 j31 + j12 j21 + j23 j32 − j11 j22 − j11 j33 − j22 j33) > 0,
3 = −det J > 0,
12 − 3 > 0
will be satisfed unless restrictions are imposed on the magnitudes of the jij’s.
For starters, −j13 j31 = 0; −j12 j21 < 0 if ′ − (s − sw) > 0; and −j23 j32 < 0 if at
equilibrium is positive. Unless the of-diagonal elements of J are small rela-
tive to the diagonal elements, there is no guarantee that the centripetal forces
making for stability will dominate the centrifugal forces.5
Assume that money wages adjust so slowly that we can take 3 ≈ 0, and that
prices adjust very rapidly, which is to say that 1 is very large. Also assume
= 1 so that ′ = 0. With these assumptions hl ≈ (P/W)−1 and we have
−θ1GS P θ1 0
W
−2
P
J = θ 2 ψ − ( s π − sw ) δ l
′ −θ 2 sw hl −θ 2 ( s π − sw ) r .
W
P
−2
− [ (1 − δ)s π + δsw ] l δ −sw hl δ − ( s π − sw ) r δ
W
Copyright © 2021. Harvard University Press. All rights reserved.
Now all four of the stability conditions are satisfed provided that both
marginal saving propensities with respect to the rate of p roft, the long-run
propensity to save (s) and the short-run propensity to save ([s − sw]), ex-
ceed the marginal propensity to invest (′)
α1 = −tr J = − ( j11 + j22 + j33 ) = θ1GS P + θ 2 sw hl + ( s π − sw ) r δ > 0,
W
α2 = −( j13 j31 + j12 j21 + j23 j32 − j11 j22 − j11 j33 − j22 j33 )
−2
P
= −θ1θ 2 ψ′ − ( s π − sw ) δ l + θ1θ 2 sw hl + θ1θ 2GS P ( s π − sw ) r δ > 0,
W W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
738 Raising Keynes
−2
P
α3 = −det J = −θ1θ 2 ( ψ′ − s π ) l ( s π − sw ) r δ > 0,
W
P
−2
−θ 2 ψ′ − ( s π − sw ) δ l
α1 α2 − α3 ≈ ( θ1 ) GS P
2
W > 0
W + θ 2 sw hl + θ 2GS P ( s π − sw ) r δ
W
for large 1.
Te picture is in Figure 18.27.
Te equilibrium E′ is necessarily unstable, since along the lower portion
of the aggregate-demand function, investment is more responsive to proft
ability than is saving, ′ − s > 0, which violates the determinant condition
3 = − det J > 0. Tis condition is satisfed at E, but stability also requires
′− [s − sw] < 0, namely, that the responsiveness of investment to the rate of
proft be less than the response of saving. Stability also requires that the price
level responds more rapidly to disequilibrium than does employment, which
will be the case if 1 is suf ficiently large. (Otherwise the economy will spiral
outward from E in an ever-wider orbit.)
In the limiting case of instantaneous price adjustment, we have hl = (P/W)−1,
and the proft rate is equal to the marginal productivity of capital:
r = h − hl l. (18.29)
P Supply of goods
W P 0
s
Stationary real price
Copyright © 2021. Harvard University Press. All rights reserved.
P 0
W s
Aggregate
demand E
l 0 0
W
E
Supply of labor
Figure 18.27 Equilibrium with Cambridge saving and capital
= 0 and > 0, = 0 and → ∞, = 1.
widening: 3 1
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 739
Te fxprice model composed of equations (18.3), (18.27), and (18.28) then
collapses into
P
−1
l = θ 2 ( I − S ) = θ 2 ψ(r − ρh ) − s πr δ − sw r[1 − δ] − sw l , (18.30)
W
δ = {[ (1 − δ)s π + δsw ] r − swh(l, ξ)} δ. (18.31)
Equations (18.30) and (18.31) essentially bring us back to Keynes’s frst-
pass model with the modifcation that both investment and saving depend on
the proft rate. Money wages are implicitly assumed to be fxed, and the GS
schedule is smuggled in via equation (18.29). Te novelty introduced by Pasi-
netti is that the saving function makes the distribution of wealth endogenous.
Te Jacobian is now
J =
{
−θ 2 ψ′ − ( s π − sw ) δ hll l + sw hl
} −θ 2 ( s π − sw ) r
.
− {[ (1 − δ)s π + δsw ] hll l + sw hl } δ − ( s π − sw ) r δ + [ (1 − δ)s π + δsw ] r − sw h
ficient condition for tr J < 0 is that the two diagonal terms be negative.
A suf
Te condition
{
−θ 2 ψ′ − ( s π − sw ) δ hll l + sw hl < 0 } (18.32)
has a clearer economic meaning if we make use of the relationship between
output and the rate of p
roft
∂h
∂h h
= ∂l = − l . (18.33)
∂r ∂r hll l
Copyright © 2021. Harvard University Press. All rights reserved.
∂l
Substituting into equation (18.32), we can rewrite the trace condition as
θ 2 hl ∂h
ψ′ − ( s π − sw ) δ − sw < 0. (18.34)
∂h ∂r
∂r
Te expression inside the braces is the diference between the responsiveness
of investment demand to a change in the p roft rate and the short-run respon-
siveness of saving (short-run because is fxed). For the inequality (18.34) to
hold, investment must be less responsive than saving.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
740 Raising Keynes
Te second diagonal element
− ( s π − sw ) r δ + [ (1 − δ)s π + δsw ] r − sw h (18.35)
is unconditionally negative. If > 0, equation (18.28) implies
[ (1 − δ)s π + δsw ] r − sw h = 0.
If = 0, expression (18.35) reduces to sr − swh, which, as noted earlier, is nec-
essarily the case when the PMC owns the entire capital stock.
Te condition det J > 0 is satisfed if investment is less responsive to profts
than is long-run saving. When > 0, det J > 0 if and only if ′ < s. In this case,
the growth rate of rentier capital is equal to the growth rate of the entire cap
ital stock, and the responsiveness of saving to a change in r is simply s. When
= 0, det J > 0 if and only if ′ < sw(∂h/∂r). PMC capital is the entire stock of
capital, and the responsiveness of capital-stock growth to proftability is given
by sw(∂h/∂r).
Pictures may be helpful here. Te vertical axis in Figures 18.28 and 18.29
measures the rate of proft. Te horizontal axis measures the rate of growth of
the capital stock, equal to rentier saving per unit of rentier capital, sr, when
> 0, and equal to PMC saving, sWh, when = 0. Figures 18.28 and 18.29 as-
sume = 0.
In Figures 18.28 and 18.29, the global saving and growth rate is denoted
r r g
1
1
swh
0
Copyright © 2021. Harvard University Press. All rights reserved.
s r
S (r ) max ( s r, sw h )
0
Figure 18.28 A two-class model of saving à la Pasinetti: < 1.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 741
S(r). We have S(r) = max(sr, swh). Why? Te argument has two strands, both
of which follow from equation (18.31): when > 0 and is stationary, equation
(18.31) guarantees both that the global saving rate is equal to sr and that
sr > swh; when = 0, the same equation leads to the global saving rate being
equal to swh, and swh > sr.
Te relationship between the sr and swh schedules depends on the elastic-
ity of substitution in production, . With production net of the cost of energy
given by h(l, ), defned by equation (18.7), the formula for can be expressed
as on p. 728:
hl ( h − hl l )
=− .
hlhll
Substituting into equation (18.33) gives
∂h
= ,
∂r h − hl l
h
which is to say that, given equation (18.29), the slope of swh (with respect to
the r-axis) is the ratio of the elasticity of substitution to the proft share. Te
opposite behavior of factor shares under the two regimes, < 1 and > 1,
explains the divergent shapes of swh in Figures 18.28 and 18.29. When < 1,
we have
r s r r g
0
sw h
Copyright © 2021. Harvard University Press. All rights reserved.
0
S (r ) max ( s r, sw h )
1
1
Figure 18.29 A two-class model of saving à la Pasinetti: > 1.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
742 Raising Keynes
h − hl l h − hl l
lim = 0 and lim = 1.
l →0 h l →∞ h
When > 1, these limits are reversed:
h − hl l h − hl l
lim = 1 and lim = 0.
l →0 h l →∞ h
Hence ∂h/∂r goes from ∞ to zero when < 1, and from zero to ∞ when > 1.
As a result, swh exceeds sr for small values of l and r in the frst case, and swh
exceeds sr for large values of l and r in the second.
Here’s the economics. As the proft share approaches zero or one, the wage
share approaches the opposite limit. So, as l and r increase, the decrease in the
wage share concentrates wealth further in the hands of rentiers when < 1.
But when > 1, the wage share increases with l and r, and this tips the balance
toward the PMC with regard to ownership of capital. In this case, the higher
is the rate of proft, the greater the equilibrium ( = 0) share of capital held by
the PMC.
Te construction of Figures 18.28 and 18.29 refects only saving, and the
saving function plus the built-in supply-side assumptions do not by them-
selves determine equilibrium. We also need the investment function, as in
Figure 18.30.
Two results emerge from the picture. First, as in Figure 18.27, there are two
equilibria. Te equilibrium at E′ is necessarily unstable since the determinant
condition, ′ < s, is not satisfed. Te equilibrium at E satisfes the determi-
nant condition
r S (r ) max ( s r, sw h )
r g
sw h E
Copyright © 2021. Harvard University Press. All rights reserved.
I (r h )
s r
E
Figure 18.30 Equilibrium in a two-class model: < 1.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 743
h
ψ′ < s π = ( s π − sw ) δ + sw
r
but may violate the inequality
{
−θ 2 ψ′ − ( s π − sw ) δ hll l + sw hl l < 0, } (18.32)
and, consequently, tr J may not be negative. With < 1, ∂h/∂r is falling as r
increases, so h/r > ∂h/∂r. Tus at E, it is possible to have det J > 0 and at the
same time tr J > 0.
Tis complication does not exist when = 0 because long-run and short-
run saving responses to the rate of proft are identical. For det J to be positive
when = 0 we must have
∂h
′ − sw < 0,
∂r
which is exactly the same as the trace condition (18.32).
It is clear from Figure 18.30 that the equilibrium rate of p roft may exceed
the rate of growth of the capital stock without the shares of the capital stock
changing, that is, with rentier capital growing at exactly the same rate as the
overall capital stock. In the present model, the equilibrium distribution of
wealth is endogenously determined by investment and saving, and can lie
anywhere in the closed interval [0, 1].
Change investment or saving, and the equilibrium will change, including
the equilibrium distribution of capital. For instance, if the rate of interest h
falls, investment demand will increase and so will the rate of growth, but the
equilibrium rentier share of capital will increase. From the Jacobian, we have
∂δ ψ′(σ − 1)sw h
=− ,
∂ρh ( ψ′ − s π ) ( s π − sw ) r 2
which for < 1 and ′ < s is negative. Tat is, if the interest rate falls, the
distribution of wealth shifs in favor of rentiers despite the higher rate of
growth. Te rate of p roft rises by more than the rate of growth.
Copyright © 2021. Harvard University Press. All rights reserved.
What happens when > 1? In Figure 18.31 the equilibrium at E is stable
since the trace and determinant conditions are both satisfed. Observe that
as in Figure 18.30, at E the rate of proft exceeds the equilibrium growth rate.
And this is true despite the fact that rentiers own a vanishing share of the
capital stock. With a lower interest rate, the investment function lies farther
to the right, and it could be the case that the growth rate exceeds the proft
rate at equilibrium. (Tis is not possible for the stable equilibrium in Figure
18.30.)
With a suf ficiently low investment demand, the stable equilibrium will lie
on the linear portion of the saving schedule, on which the rentier class owns a
positive share of the capital stock. In this case, an increase in investment de-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
744 Raising Keynes
s r
r
r g
S (r ) max ( s r, sw h )
E
swh
I (r h )
Figure 18.31 Equilibrium in a two-class model: > 1.
mand will, as in Figure 18.30, increase the rate of growth and the rate of
proft, but the rentier share of capital will decrease; with > 1, ∂/∂h is posi-
tive. Te economics is that in Figure 18.31 an increasing wage share more
than ofsets the efect of a rising rate of proft, whereas in Figure 18.30 the
proft share increases along with the rate of p roft at higher levels of invest-
ment demand.
An obvious question is whether it is more likely that the elasticity of substi-
tution is greater or less than one. In the second case, < 1, a higher equilib-
rium rate of p roft will be associated with a greater concentration of wealth in
the hands of rentiers, whereas in the frst case, > 1, the opposite is true. In
this case, if at equilibrium ∈ (0,1], a higher equilibrium rate of p roft is as-
sociated with a lower rentier share of the capital stock!
Copyright © 2021. Harvard University Press. All rights reserved.
Even so, Piketty believes that > 1, and justifes his belief on the grounds
that “there are many diferent uses of capital in the long run” (2014, p. 221).
Te fexibility of capital surely has a bearing on the overall elasticity of substi-
tution, but most empirical work suggests relatively low values of , well below
one. (For a selective survey of the vast research aimed at estimating , see
Robert Chirinko [2008]. Simon Koesler and Michael Schymura [2015] ofer a
more recent set of estimates by industry and geographical region.) In any
case, an elasticity of substitution greater than one has an implausible feature:
> 1 implies that it is possible to produce goods with only one factor of pro-
duction, that we can make an omelet with our labor alone, without eggs. or,
indeed, vice versa: with eggs that stir and then fry themselves. (Te latter ex-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 745
ample is perhaps less far-fetched: it is the central assumption of the utopia—
or dystopia—of complete automation.)
Te inequality r > g by itself tells us little about the dynamics of wealth dis-
tribution in a Pasinetti framework, but this ambiguity does not address the
impact of a change in the rate of growth. Piketty assumes that the dynamics of
g and r are independent of one another, and that, accordingly, an increase in
the rate of growth decreases the concentration of wealth. In a Pasinetti frame-
work, by contrast, the dynamics are intertwined.
Suppose, once again, a lower interest rate. Regardless of whether is less
than or greater than one, with > 0, higher investment shifs the equilibrium
to the right:
∂g ∂g ∂l s ψ′
= = π < 0.
∂ρh ∂l ∂ρh ψ′ − s π
Te initial impact on employment and wages may increase the PMC’s share of
capital (temporarily in case < 1), but if we assume that wages adjust imme-
diately, the rate of proft will rise by even more than the rate of growth—see
Figures 18.30 and 18.31—so that the equilibrium diference between r and g
also increases. With g = sr
∂(r − g ) ( 1 − s π ) ψ′
= < 0.
∂ρh ψ′ − s π
Whether increases or decreases depends on the elasticity of substitution:
< 1 implies an increase in the rentier share of wealth, and > 1 implies a de-
crease. Te takeaway is that the efect of higher growth on the concentration
of wealth is ambiguous.
I do not wish to leave the impression that my negative conclusions with
respect to Piketty’s “central contradiction of capitalism” are dispositive. Te
model I have elaborated to explore the equilibrium distribution of the capital
stock is much too simple to draw any defnitive conclusions from it. Te
Copyright © 2021. Harvard University Press. All rights reserved.
analysis should be read rather as an invitation to consider the interactions of r
and g and to focus on class dynamics rather than the dynamics of dynasties.
A More General Theory of the Long Run: Incorporating
Unemployment as a Determinant of Labor Supply
Te model elaborated in the body of this chapter can be made more plausible
by incorporating the dependence of both wage dynamics and labor-supply
dynamics on the rate of unemployment. Instead of wage dynamics that de-
pend solely on the relationship between the actual real wage and the conven-
tional wage
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
746 Raising Keynes
W P P ∗
= 3 − ,
W W W
assume that wage dynamics refect the wage equation on page 681 in includ-
ing unemployment as an independent variable:
W P P ∗
= θ3 − − θ 4 (1 − η), (18.36)
W W W
L
=
= employment as a fraction of the labor supply .
LS
To keep the model tractable, I do not incorporate the price of energy, despite
its signifcance in the wage equation of this chapter and the next.
Te assumption that the labor supply is unlimited is replaced by the more
realistic assumption that labor-supply growth depends on two factors: one a
“natural” growth, nLS, the other a function of the unemployment rate. Te
natural rate is not quite as natural as it is in the Harrod tradition; the param-
eter n is the sum of the population growth rate and a normal rate of immigra-
tion—from both domestic and foreign sources—into the capitalist economy
at full employment. Natural or not, this part of labor-force growth is assumed
to be exogenous. Te endogenous part of labor-supply growth is the variation
in the rate of immigration due to job-market conditions: the unemployment
rate (1 − ) is assumed to proxy for how easy or dif ficult it is to fnd a job. We
have
( LS )• = nLS − 5(LS − L)
and
(LS)•
Copyright © 2021. Harvard University Press. All rights reserved.
= n − θ 5 (1 − η).
LS
With fxprice adjustment, the rate of capital accumulation is equal to the
rate of investment as a ratio to the capital stock, I = K /K, so
l L
= −I
l L
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 747
and
l
•
η L ( LS ) ( LS )•
= − = +I− = θ 2 (I − S) + I − n + θ 5 (1 − η). (18.37)
η L LS l LS
Te complete model is
•
P
W
= θ l − GS P , ξ − θ P − P + θ (1 − η),
∗
P 1 W 3 4
W W
W
l = 2 (I − S),
and equation (18.37).
We assume a Pasinetti saving function, but to keep the mathematics trac-
table we let the capital shares of the rentiers and the PMC, and 1 − , adjust
instantaneously. Tis allows us to write the saving function as
S = max(sr, swh).
Additionally, assume that investment responds positively to the real price
P/W as a proxy for proftability and to as a proxy for the pressure on capac-
ity, but does not respond at all to changes in the labor:capital ratio l. With
subscripts denoting the respective derivatives of the investment and saving
functions, the Jacobian is
−θ1GS P − θ 3 θ1 −θ 4
j11 j12 j13 W
(
j23 = θ 2 I P − S P
)
J = j21 j22 − θ 2 Sl θ2 I η . (18.38)
j j32 j33 W W
( )
31 θ2 I P − S P + I P − θ 2 Sl (1 + θ2 ) I η − θ5
W W W
Copyright © 2021. Harvard University Press. All rights reserved.
Te analysis of stability is facilitated by subtracting the second row of the
Jacobian (18.38) from the third, which gives us
−θ1GS P − θ 3 θ1 −θ 4
j11 j12 j13 W
J = j21
j
j22
j32 j33
(
j23 = θ 2 I P − S P
W W
) − θ 2 Sl θ 2 I η .
31 IP 0 I η − θ5
W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
748 Raising Keynes
Tis has the efect of replacing the state variable and the adjustment equation
•
L ( LS )
= −
L LS
by the state variable K/LS = /l and the adjustment equation
•
l K LS ⋅
K K
LS = − = −
l l K LS LS
or
•
K
LS ⋅
= − l = K − LS .
K
l K LS
LS
Te stability conditions
1 = −tr J = −(j11 + j22 + j33) > 0,
2 = − (j13 j31 + j12 j21 + j23 j32 − j11 j22 − j11 j33 − j22 j33) > 0,
3 = − det J > 0,
12 − 3 > 0
are satisfed if we make plausible restrictions on the values of the coef
ficients:
(1) j33 = I η − θ 5 < 0 ,
growth of the capital stock responds more slowly than growth of the
labor supply to changes in employment;
j 2 Sl j 1
(2) − 22 = > − 12 = ,
( )
Copyright © 2021. Harvard University Press. All rights reserved.
j21 2 I P − S P j11 1GS P + 3
W W W
the slope of the AD schedule relative to the l-axis in l × P/W space is
steeper than the slope of the stationary P/W schedule;
j S j θ
(3) − 22 = l > − 12 = 1 ,
j23 Iλ j13 θ4
the slope of the AD schedule in l × K/LS space is steeper than the slope
of the stationary P/W schedule;
(4) 2 is small relative to other speeds of adjustment.
Condition (1) guarantees the trace condition, 1 > 0, since all the other terms
in the trace are negative. Conditions (1) and (2) guarantee 2 > 0, since, given
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 749
condition (1) the only expression of doubtful sign is j13 j31 + j12 j21, and condi-
tion (2) ensures this expression is negative. Conditions (1), (2), and (3) to-
gether guarantee the determinant condition, 3 > 0, since, given condition
(1), condition (2) ensures j12 j21 j33 − j11 j22 j33 < 0, and condition (3) guarantees
j31 j12 j23 − j13 j31 j22 < 0. Finally, condition (4) guarantees 12 − 3 > 0. Observe
that these conditions are suf ficient rather than necessary for stability.
Consequences of Changes in Demand and
Supply in the Long Run
Comparative-statics exercises confrm the main points of the simpler analysis
without the adjustment of the labor supply. Te equilibrium equations are
now
P P P ∗
AA ≡ θ1 l − GS , ξ − θ 3 − + θ 4 (1 − η) = 0,
W W W
BB ≡ I − S = 0,
CC ≡ I − n + 5(1 − ) = 0.
To fnd the efect of exogenous changes in demand or supply conditions, we
diferentiate the system of equations with respect to x, which can represent
any of the exogenous parameters, x = h, s, sw, (P/W)*, . For simplicity, the
analysis is limited to the demand side; specifcally, the efect of the hurdle rate
on employment. Again, to keep the math simple, we limit ourselves to the
case 0 < ≤ 1, that is, the case where rentiers own a nonvanishing share of the
capital stock. Te result is the equation system
P
∂ ∂AA
−θ1GS P − θ 3 −θ 4 W ∂x
Copyright © 2021. Harvard University Press. All rights reserved.
θ1
W ∂x
I −S ∂BB
− Sl I η ∂l = − . (18.39)
P P
∂x
W W ∂x
IP 0 I η − θ 5 ∂η ∂CC
W ∂x
∂x
Te generic inverse is
j22 j33 j23 j32 j13 j32 j12 j33 j12 j23 j13 j22
J 1 1
j23 j31 j21 j33 j11 j33 j13 j31 j13 j21 j11 j23 ,
j j j j j12 j31 j11 j32 j11 j22 j12 j21
21 32 22 31
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
750 Raising Keynes
where
∆ = det J = j11 ( j22 j33 − j23 j32 ) − j12 ( j21 j33 − j23 j31 ) + j13 ( j21 j32 − j222 j31 ).
ficient matrix in the equation system (18.39), we have
For the coef
∆ = [ −θ1GS′ − θ 3 ] [ −Sl ] I η − θ 5 − θ1 − I P θ 5 − S P I η − θ 5 − θ 4 Sl I P .
W W W
We will assume −(IP/W)5 > SP/W[I − 5], which is to say that IP/W is small rela-
tive to SP/W. Tis assumption guarantees < 0.
For x = h, equation (18.39) becomes
P
∂
−θ1GS P − θ 3 θ1 −θ 4 W
W ∂ρh 0
I −S − Sl I η ∂l = − I ,
P P ρh
W W ∂ρh I
IP 0 I η − θ5 ρh
W ∂η
∂ρh
and from the inverse J −1 we have
P
∂
W = ( θ 1 − θ 5 ) − θ 4 Sl − I
∂ρh ∆ ρh ( )
< 0 if and only if 1 5 − 4 Sl > 0,
θ1GS P + θ 3 θ 5 + θ 4 S P
∂l
∂ρh
= W
∆
W
− I ρh ( ) < 0,
Copyright © 2021. Harvard University Press. All rights reserved.
θ1S P + θ1GS P + θ 3 Sl
∂η
∂ρh
= W
∆
W
− I ρh < 0.( )
Te efect on infation of a change in the hurdle rate is given by
P P P
∂
P
∂ ∂ ∂
P = P W + P ∂l =
θ1θ 3 θ 5 + θ1θ 4 GS P + S P
W W
( ) ( −I ρh ) < 0.
∂ρh P ∂ρh ∂l ∂ρh ∆
∂
W
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 751
All of these derivatives are sign defnite with the exception of ∂(P/W)/∂h. A
fall in the hurdle rate stimulates employment and infation, but the real price
level may fall or rise because the relative efect of greater investment on em-
ployment and on the labor force depends on relative adjustment speeds.
Consider the simpler case of assuming SP/W = IP/W = 0. We have
∆ = −θ1GS P − θ 3 [ −Sl ] I η − θ 5 ,
W
θ1GS P + θ 3 θ 5
∂l
∂ρh
= W
−θ1GS P − θ 3 [ −Sl ] I η − θ 5
− I ρh = ( θ5
)
Sl I η − θ 5
I ρh , ( )
W
θ1GS P + θ 3 Sl
∂η 1
∂ρh
= W
−θ1GS P − θ 3 [ −Sl ] I η − θ 5
− I ρh = ( )
I η − θ 5
I ρh . ( )
W
In this simple case—shades of the frst-pass model, except that investment as
well as saving depends on income and employment—it is clear that whether
l or responds more vigorously to changes in the hurdle rate depends on
the relative magnitudes of 5 and Sl. Since l ≡ L/K and ≡ L/LS, if 5 > Sl ,
then the labor supply is responding more strongly than the capital stock. In
case 5 < Sl, is responding more strongly than l, which is to say that the cap
ital stock adjusts more rapidly than does the labor supply.
Observe that in this model, as in the other models elaborated in this chap-
ter, shocks are permanent and have permanent efects. Tis is a principal dif-
ference between a Keynesian long run and the standard, New Keynesian,
treatment of the long run, in which shocks are transitory however long a time
it might take for their efects to dissipate. Greg Mankiw’s simple New Keynes-
Copyright © 2021. Harvard University Press. All rights reserved.
ian model (2016, chap. 15) lends itself to a dynamic version suitable for com-
parison. Using Mankiw’s notation
Y = Natural level of output
Y = Output
r = Real rate of interest
= Natural rate of interest
= Demand shock
= Supply shock
= Rate of infation
* = Target rate of infation
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
752 Raising Keynes
the model is
Y = θ1 [ Y − Y − α(r − ρ) + ε ],
π = θ 2 [ Y − Y + ν ],
r = θ π [ π − π∗ ] + θY [ Y − Y ].
Te equations for output and infation are supposed to refect the workings of
the economy; the natural level of output is assumed to be fxed for simplicity.
Te adjustment of the interest rate is a central-bank reaction function à la
Taylor (1993).
Te Jacobian for this system is
− θ1 0 − θ1 α
J = θ2 0 0 .
θ θπ 0
Y
Te system is stable if and only if the product 1Y is greater than the prod-
uct 2.
Te efect of a demand shock is given by inverting the Jacobian matrix
to obtain
∂Y
∂ε 0 −θ1 αθ π 0 θ1
∂π = −∆ −1 0 −θ1 αθY −θ1 αθ 2 0 ,
∂ε
∂r θ θ
2 π − θ1 θ π 0 0
∂ε
∂Y
= 0,
∂
∂π
Copyright © 2021. Harvard University Press. All rights reserved.
= 0,
∂ε
∂r 1
= .
∂ε α
Te result is that a demand shock has no permanent efect either on output or
on infation. Te real rate of interest has to rise in order to curb demand and
make room for the demand shock.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
18 • Mathematical Appendix 753
By contrast a supply shock has a permanent efect:
∂Y
∂ν 0 −θ1 αθ π 0 0
∂π = −∆ 0
−1
−θ1 αθY −θ1 αθ 2 θ 2 ,
∂ν
∂r θ θ
2 π − θ1 θ π 0 0
∂ν
∂Y
= −1,
∂
∂π θ
=− Y ,
∂ν αθ π
∂r 1
=− .
∂ν α
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
. 19 .
inf lat ion and employment empirics in
the keynesian long run
Te central forecast to which [M. Friedman’s (1968) and Phelps’s
(1968)] reasoning led was a conditional one, to the efect that a high-
infation decade should not have less unemployment on average than
a low-infation decade. We got the high infation decade, and with it
as clear-cut an experimental discrimination as macroeconomics is
ever likely to see, and Friedman and Phelps were right. It really is as
simple as that.
—robert e. lucas, jr.
Te U.S. Phillips curve appears to be one of the most stable empirical
macroeconomic relationships of the postwar era, one that shows no
sign as of yet of being subject to Lucas’s econometric critique
—robert gordon
How do the models elaborated in the previous chapter stand up to the data?
In particular, to what extent does this framework explain the ephemeral as-
pect of the Phillips curve that has led to the denial of the existence of a long-
run trade-of between price stability and employment?
Were we living in a world without technological change, it would be rea-
sonable to apply the theory developed in chapter 18 to data relating the
labor:capital ratio to the real price of labor (or the real wage) and to translate
Copyright © 2021. Harvard University Press. All rights reserved.
demand and supply shocks into changes in the labor:capital ratio and changes
in the real price. But in a world of changing technology, the labor:capital ra-
tio will vary independently of demand and supply shocks and so will the real
price. Without controlling for technology, it is pointless to look for an associa-
tion between changes in the labor:capital ratio and changes in the real price.
Te regression equation in chapter 18 controlled for technology by substi-
tuting a conventional wage share for the conventional wage rate in explaining
wage changes over time. Here I follow a version of this strategy in replacing
Portions of this chapter were previously published in “Wages, Prices, and Employment in a
Keynesian Long Run,” Review of Keynesian Economics 5 (Autumn 2017): 360–425.
754
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
19 • inflation and employment empirics 755
the labor:capital ratio by the employment ratio in explaining the rate of infa
tion. Despite the endogeneity of the labor force, it is fxed at each moment of
time and adjusts to demand only over time, as in the model elaborated in the
mathematical appendix to the last chapter.
Te models of the text and the mathematical appendix to chapter 18 con-
verge on one point: positive demand shocks, represented in the model by in-
creases in investment demand or reduction in desired saving, lead to an in-
crease in employment and an increase in infation. In other words, according
to the theory, the Phillips curve should be refected in the long-term data.
Beyond this prediction, the models suggest that there are a variety of plau-
sible responses to supply shocks. Both a negative association between infa
tion and employment (anti-Phillips stagfation) and a positive association
(the Phillips relationship) are possible.
Te challenge is then to make sense of the haphazard collection of points in
Figure 18.3, which shows the relationship between the employment ratio and
infation in the United States over the period 1956 to 2011.1 Tat fgure does
not hold out much hope for a Phillips relationship until, as Robert Gordon
(1984, 2015) and others have argued, we separate demand and supply shocks.
We begin by looking at the data for the frst years of this period, when
Keynesians believed all too easily that the world, like their models, could be
understood in terms of the behavior of demand, with nary a side glance at
supply. Figure 19.1 presents the subset of data from Figure 18.3 for the period
1956 to 1969. Tese data, especially for the earlier part of this period, do not
look very promising. We do not need a regression line to see that employment
and infation do not move together until 1965.
However, we have not exploited one feature of the model laid out in chapter
6, namely, the diferent implications of the two adjustment processes for lags
6.0
Copyright © 2021. Harvard University Press. All rights reserved.
Inflation (change in CPI)
1969
5.0
4.0 1968
1957 1966
3.0
1958 1967
2.0
1963 1960 1965 1956
1961 1962 1964
1.0 1959
0.0
93 94 95 96 97
Employment as percentage of civilian labor force
Figure 19.1 Employment vs. infation (1956–1969).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
756 Raising Keynes
between changes in employment and changes in the rate of infation. Flex-
price adjustment implies that prices move frst, in response to positive dif
ference between desired expenditure and output; employment adjusts later,
when the rise in prices makes it proftable to expand production. We can
capture this in a simple way by assuming that employment responds with a
one year lag to infation, with changes in the rate of infation in turn triggered
by a change in aggregate demand. Figure 19.2 presents the data for the period
1956 to 1969 with this lag structure. No improvement in the ft would be an
overstatement.
However, if we assume fxprice adjustment and impose a lag structure that
goes in the opposite direction—frst comes the change in output and employ-
ment, then comes the change in the rate of infation—the results look more
promising. Figure 19.3 relates employment in year t to the rate of infation in
year t + 1. Te ft is now very much in line with the prediction of the model
6.0
Inflation (change in CPI)
1969
5.0
4.0 1968
1957
3.0 1967
1958 1966
2.0
1964 1965
1960 1962 1956
1963
1.0 1959
1961
0.0
93 94 95 96 97
Employment as percentage of civilian labor force
Figure 19.2 Employment(+1) vs. infation (1956–1969).
Copyright © 2021. Harvard University Press. All rights reserved.
6.0 1969
Inflation (change in CPI)
1968
5.0
4.0 1967
1956
3.0 1965 1966
1957
2.0 1959
1963 1964
1961 1962 1960
1.0 1958
0.0
93 94 95 96 97
Employment as percentage of civilian labor force
Figure 19.3 Employment vs. infation(+1) (1956–1969).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
19 • inflation and employment empirics 757
for demand-driven changes, but the Phillips curve is really a curve rather
than a linear relationship: higher levels of employment lead to more than pro-
portionate changes in the rate of infation. In fact, a log-linear specifcation
does much better, as in Figure 19.4, in which one percentage point of addi-
tional employment leads to the same percentage increase in the infation rate
regardless of the employment ratio.
You probably know what happens when we extend the range of the data
beyond the so-called Golden Age. Keeping to the lag structure suggested by
fxprice adjustment, we have Figure 19.5, in which the years 1970 to 1975
have been added. Te Phillips curve falls apart. Something other than aggre-
gate demand is clearly in play.
If we take a somewhat longer view, we can make better sense of the data.
Add the period through 1983, and the data sort themselves out diferently. In
Figure 19.6, there are two distinct Phillip curves covering, respectively, the
Log inflation (change in CPI)
2.0
1968 1969
1.5 1967
1956
1965 1966
1.0 1957
0.5 1959 1964
1963
1961 1962 1960
0.0 1958
−0.5
93 94 95 96 97
Employment as percentage of civilian labor force
Figure 19.4 Employment vs. log (infation[+1]) (1956–1969).
3.0
Copyright © 2021. Harvard University Press. All rights reserved.
Log inflation (change in CPI)
2.5
1973
1974
2.0
1975 1972 1969
1968
1.5 1970 1967
1971 1956
1.0 1965
1957 1966
0.5 1959 1964
1961 1963
1962 1960
0.0 1958
−0.5
90 91 92 93 94 95 96 97
Employment as percentage of civilian labor force
Figure 19.5 Employment vs. log (infation[+1]) (1956–1975).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
758 Raising Keynes
periods 1956 to 1969 and 1973 to 1983. Te Phillips curve is somewhat fatter
in the second period, but the more important distinction is that the intercepts
are very diferent.2 Te Phillips curve has shifed.
Figure 19.7 brings the story up to the present century. In the fgure there
are important changes in the intercept, and the Phillips-curve trade-of in-
creases moderately in the period 1997 to 2001. Te period 1994 to 1996 is a
transition period, like the period 1970 to 1972 in Figure 19.6, not belonging
to either the era before or the era afer.
Starting in the 1970s, changes took place that allowed any given employ-
ment ratio to be achieved at a progressively lower rate of infation. For in-
stance, in 1982, an employment ratio of just over 90 percent (an unemploy-
ment rate approaching 10 percent!) coexisted with an infation rate of 3.1
3.0
Log inflation (change in CPI)
2.5 1979
1980 1978 1973
1977 1974
2.0 1976
1981 1972 1969
1975 1968
1.5 1983 1970 1967
1982 1971 1956
1.0 1965
1957 1966
0.5 1959 1964
1963
1961 1962 1960
0.0 1958
−0.5
90 91 92 93 94 95 96 97
Employment as percentage of civilian labor force
Figure 19.6 Employment vs. log (infation[+1]) (1956–1983).
3.0
Copyright © 2021. Harvard University Press. All rights reserved.
Log inflation (change in CPI)
1979
2.5 1978
1980 1973
1974
2.0 1977
1976 1981
1975 1988 1989
1.5 1983 1987 1990
1984 1986 1994 1999
1982 1992 1991 1996
1.0 1993 2000
1995 1998
1985
0.5 1997 2001
0.0
−0.5
90 91 92 93 94 95 96 97
Employment as percentage of civilian labor force
Figure 19.7 Employment vs. log (infation[+1]) (1973–2001).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
19 • inflation and employment empirics 759
percent the next year (exp(1.13) = 3.1); approximately a decade later, in 1991,
the employment ratio was over 93 percent, with the same rate of infation; by
the end of the century, in 2000, the employment ratio was just shy of 96 per-
cent, with infation even less than 3 percent. Indeed, Figure 19.8 suggests that
by the late 1990s the relationship between employment and infation was back
to where it had been in the Golden Age of the 1950s and 1960s. To illustrate,
the data points for 1956 and 1999 are virtually indistinguishable; in 1956 as
well as in 1999, employment was just under 96 percent and infation in both
1957 and 2000 was 3.3 percent.
Figure 19.9 brings the story into the second decade of this century. Some
thing new has happened. Not only has the dispersion of the data increased,
but the trade-of between employment and infation has become much fat-
3.0
Log inflation (change in CPI)
2.5
2.0
1968 1969
1.5 1967
1999 1956
1965
1.0 1957 2000 1966
1998
1964
0.5 1959 2001
1963 1962 1997
1961 1960
0.0 1958
−0.5
90 91 92 93 94 95 96 97
Employment as percentage of civilian labor force
Figure 19.8 Employment vs. log (infation[+1]) (1956–1969 and 1997–2001).
3.0
Copyright © 2021. Harvard University Press. All rights reserved.
Log inflation (change in CPI)
2.5
2.0
1.5 2007
2005
2010 2004 1999
1.0 2006 2000
2003 2002 1998
2011
0.5 2009 2012 2001
2013
1997
0.0
−0.5
90 91 92 93 94 95 96 97
Employment as percentage of civilian labor force
Figure 19.9 Employment vs. log (infation[+1]) (1997–2013, excluding 2008).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
760 Raising Keynes
ter.3 If the data for the years 2002 to 2013 defne a new Phillips curve, we
could have been having much higher employment in this century without
touching of much more infation.
According to the theory laid out in chapter 18, the Phillips curve shifs over
time because of supply shocks, which I model as changes in the real price of
energy—the price of energy relative to the price of all goods other than en-
ergy and food. As Figure 19.10 shows, the real price of energy does indeed
correlate with the observed shifs in the Phillips curve. Te increase between
1973 and 1980 correlates with the upward shif in the Phillips curve in the
1970s, and the subsequent downward shif in the 1980s correlates with the
downward shif of the Phillips curve in the later period.
Since the scatter of points in Figures 19.1 to 19.9 refect both supply and
demand factors, the visual appearance of the Phillips Curve relationship be-
tween infation and employment in these diagrams is not an accurate repre-
sentation of the relationship between infation and aggregate demand. In the
1970s, when, as Figure 19.10 indicates, the real price of energy was on an up-
swing, aggregate demand would have had less efect on infation than the ob-
served data in Figures 19.6 and 19.7 suggest—regression lines summarizing
the relationship between the two variables on the basis of the observed data
would overstate the true relationship between aggregate demand and infa
tion. In the 1980s and early 1990s, when energy prices were falling sharply,
aggregate demand would have had more of an efect on infation than the data
in Figure 19.7 suggest. For the period 2000 to 2008, similar to the 1970s in
terms of the rising trajectory of energy prices, aggregate demand would have
had even less infuence on infation than the data of Figure 19.9 indicate.
How do I account for the shifs in the Phillips curve during periods in
which relative energy prices were fat, like the periods in Figure 19.10 shown
in red and gray, 1970 to 1972 and 1994 to 1996? Here, I would argue, the driv-
ing force is wage shocks rather than price shocks. In these exceptional peri-
Copyright © 2021. Harvard University Press. All rights reserved.
120
100
80
60
40
20
0
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
Figure 19.10 Energy price level as percentage of price level of all goods
except energy and food (1982 = 100).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
19 • inflation and employment empirics 761
ods, employment and infation appear to have been responding to changes in
the conventional wage. Te dummy variables DUM1970 and DUM1994 in
the wage equation of chapter 18 were intended to capture this phenomenon,
albeit in a somewhat stylized way by concentrating the efect into two years,
1970 and 1994. I interpret the positive coef ficient on DUM1970 as refecting
the ability of American workers to raise the conventional wage (as a share of
output) in the face of rising unemployment. In the second period, the nega-
tive coef ficient on DUM1994 suggests that their children were unable to simi-
larly defend themselves against the erosion of the conventional wage share
even as unemployment was falling.
Why this diference between the two generations? In the late 1960s workers
were riding high. Prosperity, strong unions, two terms of a Democratic Con-
gress and administration, an economy relatively well insulated against low-
wage competition from abroad—all combined to empower workers to press
wage claims vigorously. By contrast, in the 1990s, while the economy was
again prosperous, none of the other ingredients of working-class vigor was
present. Tere was indeed a Democrat in the White House, but the Demo-
cratic Party had changed dramatically in a generation. With neither political
party looking afer their interests, workers had little power to prevent jobs
from moving ofshore and even less power to defend their wages. U nions had
been in decline since the 1960s. Te result was a fall in the conventional wage
share, the culmination of trends and policies that had prevented real wages
from rising very much over the previous two decades. Te upside was a
downward shif in the Phillips curve.
Here the specifcs of the model come into play. In a fxprice regime in which
capital widening is the dominant form that investment took, upward pressure
on the conventional wage in the early 1970s is consistent with the observation
of increasing infation (and falling employment). And downward pressure on
the conventional wage would have produced the opposite result in the 1990s.
To tie demand and supply together, consider the following regression,
Copyright © 2021. Harvard University Press. All rights reserved.
loosely based on the model formulated in the section “A More General Te-
ory of the Long Run: Incorporating Unemployment as a Determinant of La-
bor Supply” in the mathematical appendix to chapter 18. Tis model incorpo-
rates the dependence of labor supply on the level of unemployment and the
dependence of wage changes on the unemployment rate. But the empirics are
still only loosely based on the theoretical model because, like the analysis of
wage changes in the previous chapter, the regression analysis here also takes
account of the reality of technical change and hence modifes the conven-
tional wage: instead of a wage rate, the conventional wage is modeled as a
wage share. Equilibrium is modeled as a relationship between nominal price
infation on the one hand and, on the other hand, wage infation, technical
change, and the change in the price of energy.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
762 Raising Keynes
Assuming a stationary proft rate and ignoring capital deepening, the re-
gression equation for price infation is
∆P ∆W ∆Prod ∆PE
= a0 + a1 + a2 + a3 .
P−1 W−1 Prod −1 ( PE )−1
Te variables are defned below. In equilibrium the coef
ficients should be:
a0 = 0.
a1 = 1.
a2 = reciprocal of wage share, with sign reversed, if Prod accurately mea
sures technical change. a2 will be smaller in absolute magnitude than
the wage share if the measure of technical change overstates the true
rate. Tis is likely the case here since Prod is actually a measure of out-
put per employee hour, which includes the efects of capital deepening.
a3 = the ratio of the share of energy in output to the labor share.
Ordinary least squares (OLS) regressions do not work in this situation since
the error term is almost certainly correlated with the change in wages. In-
ficients by means of two-stage least squares (TSLS).
stead, we estimate the coef
Te model in the mathematical appendix to chapter 18 suggests the conven-
tional wage share and the unemployment rate as instruments for estimating
the change in wages in the frst-stage equation. Additionally, three dummy
variables refect the changes in the conventional wage associated with the
emergence of the U.S. economy from the Great Depression (dated 1936), the
apex of working-class power (dated 1970), and the triumph of neoliberalism
(dated 1994). Te frst- and second-stage regression results are, respectively,
∆W WL ∆Prod ∆PE
= 43.48 − 76.53 + 0.4708 + 0.1196
W−1 PY −1 Prod −1 ( PE )−1
(19.00) (32.02) (0.1330) (0.0195)
Copyright © 2021. Harvard University Press. All rights reserved.
− 0.3058 UNRATE + 5.062 DUM1936 + 2.138 DUM1970 − 4.773 DUM1994
(0.11744) (2.005) (0.7530) (0.8795)
2
R = 0.63
∆P ∆W ∆Prod ∆PE
= 0.1913 + 0.7465 − 0.4043 + 0.0925
P−1 W−1 Prod −1 ( PE )−1
(0.3035) (0.0865) (0.1401) (0.0223)
2
R = 0.84
Robust Standard Errors in Parentheses
Observations: 82 (1930–2011) F(7, 74) = 13.82
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
19 • inflation and employment empirics 763
Variable Description Mean Std Dev Min Max
∆P
Percentage Change in Consumer Price Index 3.263 3.955 −10.301 14.389
P−1
∆W
Percentage Change in Nominal Wages 4.740 4.063 −13.210 17.539
W−1
WL
PY Labor Share of Output, Lagged One Year 0.560 0.018 0.514 0.596
−1
∆Prod
Percentage Change in Output per Employee Hour 2.215 2.391 −6.641 12.552
Prod
−1
∆PE
Percentage Change in Price of Energy 4.641 11.852 −26.045 55.164
( PE )−1
UNRATE Unemployment Rate 7.241 4.849 1.200 24.900
DUM1936 = 0 prior to 1936, = 1 in 1936 and later years
DUM1970 = 0 prior to 1970, = 1 in 1970 and later years
DUM1994 = 0 prior to 1994, = 1 in 1994 and later years
Te regression coef ficients all have the right sign, and, with the exception of
the coef ficient on wage changes in the second-stage equation, are of the mag-
nitudes predicted by the model. Te misbehavior of a1 suggests that prices fail
to adjust rapidly enough to achieve equilibrium in every year. Delay in adjust-
ment is also refected in Figures 18.3 to 18.9, in which price infation is as-
sumed to adjust with a one-year lag to the level of employment.
Observe that the frst-stage equation is the same equation that was de-
ployed in chapter 18 to explain wage changes, from which the conventional-
wage share and its evolution over time were calculated.4 At that point in the
argument I could not motivate this equation, not yet having developed the
long-run theory that was the subject matter of chapter 18 and its mathemati-
cal appendix.
Copyright © 2021. Harvard University Press. All rights reserved.
Based on that theory, the two-stage regression describes an equilibrium
between price infation and wage infation. Another interpretation presents
itself if we regard the two-stage regression as a structural model of the rela-
tionship between wage and price dynamics. In this interpretation the frst-
stage regression captures the dependence of wage changes on the distance
from the conventional wage share, unemployment, productivity growth, and
energy-price changes; the second stage refects the dependence of price infa
tion on wage infation, productivity growth, and energy-price changes.
Whichever interpretation we adopt, the important takeaway from all this
is, frst, that there is indeed a trade-of between employment and output on
the one hand and price stability on the other, not only in the short run
but also in the long run. And, second, the starting point for the trade-of
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
764 Raising Keynes
epends on supply conditions. Higher output due to greater demand must be
d
paid for with higher infation, but how much infation is consistent with any
given level of employment depends on the supply side, in my model refected
in the change in the price of energy and the level of the conventional wage.
Now we can see why Milton Friedman (1968) and Edmund Phelps (1968)
were correct in predicting that high infation would not see less unemploy-
ment, but Robert Lucas could still be stunningly wrong in theorizing that this
meant the demise of the Phillips curve. As Lucas observed—see the epigraph
to this chapter—the average rate of unemployment over the quarter century
preceding Lucas’s 1981 article did not depend very much on the level of infa
tion because, as the Phillips curves in Figure 19.6 show, the displacements due
to supply-side shocks change the average rate of infation more than the aver-
age rate of employment.
But there is no way that the analysis is as simple as Lucas claimed. Te lack
of correlation between the average rate of unemployment and the rate of infa
tion is totally consistent with the existence of a demand-induced trade of:
within each of the two periods covered in Figure 19.6, higher employment
goes along with higher infation, even though across periods there is less dif-
ference between the average rate of unemployment (4.9 percent vs. 7.3 per-
cent) than there is between the average rate of infation (2.1 percent vs. 7.9
percent). Experience confrmed the letter of Friedman and Phelps, but not the
interpretation that there is no Phillips curve in the long run.
Tere is an important corollary of more recent policy relevance. If I have
the logic of investment demand right, it is precisely in times when the econ-
omy is in a deep slump, like the one that began in 2008, that higher wages will
stimulate the economy, not only via higher consumption demand but also via
the inducement that high wages provides for capital deepening. Te presence
of a large amount of unused capacity means that cost-cutting investment, if
not the only game in town, is where most of the investment action is taking
place. Tis is not to argue that shifing the conventional wage is a superior
Copyright © 2021. Harvard University Press. All rights reserved.
alternative to interventions to move the AD schedule, but rather that raising
wages is another element in the toolbox of policy intervention—one which
becomes more attractive when the limits of fscal stimulus and monetary eas-
ing have been reached.
What’s So Bad about Infation?
In the theory developed in the previous chapter, higher levels of activity re-
sulting from higher demand are always accompanied by higher rates of infa
tion. In recession this is hardly a downside, indeed, as the experience of the
Great Depression illustrates all too vividly, the consequences of defation can
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
19 • inflation and employment empirics 765
be even more catastrophic than the failure to make use of existing capacity.
But what about the long run? What are the costs of infation if we leave out
the special case of an underperforming macroeconomy?
Tere is a disconnect between the man and the woman in the street and the
economist. Economists tend to regard infation as neutral between prices and
wages, at least in the long run, and on this basis one can argue that infation
does not hurt the “representative agent,” for whom all forms of income—
profts, wages, rents—go up or down at the same rate. Te man and the
woman in the street difer, regarding infation as a thief robbing them of their
hard-earned wages.
In the mid-1990s, Robert Shiller documented the public fear and loathing
of infation in three countries: the United States, which has dealt with moder-
ate infation over most of the post–World War II period; Germany, where
prices have been much more stable over this period, but second-and third-
hand memories of the disastrous hyperinfation of the early 1920s are sup-
posed to haunt public and policymakers alike; and Brazil, which for much of
the second half of the twentieth century faced chronic infation. Interviews
with ordinary people revealed more similarities than diferences among the
three countries, as well as between older and younger respondents, despite
the great diferences in their actual experience of infation. Among many sur-
prising results, perhaps the most surprising was the strong preference voiced
by the Americans and the Germans for relatively stable prices even at the cost
of high unemployment. In these two countries, both those born before 1940,
and presumably more sensitive to the evils of high unemployment as well as
high infation, and those born afer 1940 preferred a hypothetical combina-
tion of annual infation at a 2 percent rate coupled with 9 percent unemploy-
ment over a combination of an infation rate of 10 percent per month coupled
with 3 percent unemployment. Te fraction of older German respondents
who preferred the low infation combination was two-thirds, while younger
Germans opted fve to one for this package. In the United States there was
Copyright © 2021. Harvard University Press. All rights reserved.
relatively little diference between the generations, with approximately three-
quarters of both groups favoring low infation even at the cost of high unem-
ployment. Only the Brazilians, the one population with the more recent expe-
rience of very high infation, were more or less evenly split between the two
options (Shiller 1997, p. 27).
In Shiller’s survey, the public was consistently at odds with economists over
both the causes and the consequences of infation. Much of the diference evi-
dently lies in how the two groups perceive the efects of infation on real in
comes. Economists tend, I have noted, to conceptualize infation as neutral
between prices and wages, the second rising as rapidly as the frst, so that in
fation has no impact on the distribution of income (Shiller 1997, passim, esp.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
766 Raising Keynes
p. 29). Noneconomists believe that their own real incomes are likely to suffer
when infation is high (Shiller 1997, p. 29).
Who’s right? Greg Mankiw ofered a perceptive comment on Shiller’s essay
(1997): supply shocks necessarily reduce the real income of some segment of
the population, and infation is simply one way of absorbing the hit. Te real
culprit is the supply shock, but to the extent the association between supply
shocks and infation holds, it is no wonder that the public associates price in
fation with reductions in the real standard of living. In other words, it is rea-
sonable to blame the messenger (infation) for the message (the hit to real in-
come) when the message is so frequently delivered by the same messenger.
Mankiw might have added that adverse demand shocks may also reduce
real incomes; even positive demand shocks can have negative consequences
somewhere along the line, especially when these shocks play out in terms of
infation. Tis is true whether the demand shock is the result of an expansion
of the government’s claim on resources (the locus classicus being the German
hyperinfation of the 1920s, the result of the inability of the Weimar govern-
ment to levy the taxes that would have been required to make reparations
payments while maintaining price stability) or whether the demand shock is
the result of expansion of private investment demand or, for that matter, con-
sumption demand. Te older term of art for the process by which the econ-
omy balances its books in accommodating a demand shock is “forced saving,”
forced highlighting the contrast between ordinary, voluntary saving and the
involuntary nature of the reduction in real consumption that accompanies the
failure of (some) money incomes to keep pace with rising prices.
In A Treatise on Money (1930), Keynes made important use of the concept
of forced saving by workers whose wages fail to keep pace with prices; it is the
mechanism by which resources are diverted from consumption to investment
when investment demand runs ahead of profts. (By the time of Te General
Teory, having dropped the assumption of the Treatise that the normal state of
the economy was one of full employment, Keynes was able to argue that no
Copyright © 2021. Harvard University Press. All rights reserved.
forced saving is necessary to accommodate an expansion of investment.)
Keynes already hints at forced saving in his polemic against the Versailles
Treaty, Te Economic Consequences of the Peace. No doubt heavily infuenced
by his experience as a middle-level adviser to the British Treasury during
World War I, he defended capitalists against the charge of p rofteering:
“Profteers,” [quotation marks are Keynes’s] are, broadly speaking, the entre-
preneur class of capitalists, that is to say, the active and constructive element
in the whole capitalist society, who in a period of rapidly rising prices can-
not help but get rich quick whether they wish it or desire it or not. If prices
are continually rising, every trader who has purchased for stock or owns
property and plant inevitably makes p rofts. (1919, pp. 236–237)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
19 • inflation and employment empirics 767
Keynes expands on the process of forced saving in the context of wartime
government fnance in a lecture delivered at about the time he was putting the
fnishing touches on the Treatise. He explains how this mechanism worked to
fuel infation during World War I:
When a government orders munitions of war at a rate faster than that of
current savings as supplemented by taxation—the inevitable efect of such
measures must be to cause prices to rise faster than wages. Te government
secures purchasing power for itself at the expense of the consumer, who is
constantly fnding that the real value of his income is less than he had sup-
posed . . .
Te transference . . . puts some resources directly into the hands of the
government. But to a very great extent the gains accrue in the frst instance
not to the government, but to business men, who, owing to this rise of price,
are able to sell what they have produced at an unexpectedly high price which
yields them a proft in excess of what they had anticipated. Te selling price
of their goods is rising all the time faster than their cost of production. Tat
is to say, if this method of forced transference is a dopted, the business men
are made, in the frst instance, the collectors—the agents, so to speak—for
the government, to collect the purchasing power which has been thus forci-
bly diverted from the consumers . . . If, having allowed them to receive this
additional sum, you then proceed to withdraw it from them through taxes
and in fact treat them as having been agents for the government, then this
device is far and away the most ef ficient that exists for collecting purchasing
power from the consumers and transferring it into the hands of the govern-
ment. (Lecture to the Royal United Service Institution, February 13, 1929, in
Keynes 1981, pp. 785–786)
Economists today may be less convinced than Keynes that infation hurts
consumers, at least in the representative-agent model, but most are not ready
to let infation of the hook. Given the mainstream’s concern (one might say
Copyright © 2021. Harvard University Press. All rights reserved.
obsession) with ef ficiency, a focus on the putative inefficiencies brought about
by infation is hardly surprising. Two favorites are “shoeleather” and “menu”
costs, each a metaphor for a kind of wasteful activity imposed on economic
agents by infation. Shoeleather costs hark back to an era when depositors had
to visit the bank to withdraw money or shif funds from interest-bearing ac-
counts to demand deposits; they would use up more shoeleather in having to
make more trips to the bank, trips they must make to avoid holding currency
or noninterest-bearing checking accounts that are continually depreciating in
value. Menu costs recall the same bygone era, for these costs were incurred
because menus had to be reset in hard type as prices rose. Te faster prices are
rising, the more frequently menus have to be printed. Whatever the ineffi-
ciencies caused to previous generations by higher costs in terms of shoe-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
768 Raising Keynes
leather and menus, neither of these can be considered formidable costs in the
age of the computer, the internet, and the web.
Another inef ficiency to which economists point—this time with more rea-
son—is tax distortions, particularly in the treatment of capital gains. Tere is
no question that, absent special provisions in the tax code, infation causes
people with signifcant capital gains to suffer at the hands of the taxman. Ef fi
ciency (and fairness as well) would arguably be served if the real gain could
be separated from the nominal gain, so that the purely nominal part of cap
ital gains could be exempted from taxation. In fact, it is extremely dif ficult
to do so: I know of no country in which the tax code efectively separates
out the real component of taxable gains and forgives the purely nominal com-
ponent. In any case, this is a rather parochial interest in terms of the percent-
age of taxpayers seriously afected, and it is hard to make a popular case
against infation on these grounds, especially if, as in the U.S. Internal Reve-
nue Code, there is a lower tax rate on capital gains on property held more
than one year, as well as a very generous exemption for the sale of one’s prin-
cipal residence, the largest (and sometimes only) holding in most Americans’
asset portfolios.
Although economists for the most part hold that infation does not have
much of an impact on the distribution of income, they charge that infation
redistributes wealth in an arbitrary fashion. Since infation can never be en-
tirely anticipated, debtors gain at the expense of creditors. But to argue, as
two leading elementary texts do, that this redistribution is arbitrary is to
stretch the defnition of “arbitrary.”5 No doubt some creditors are people of
modest means. But the plain fact is that the holdings of fxed-income securi-
ties—which are the assets most afected by infation—are highly concentrated
among the upper tiers of wealth holders. In consequence, any redistribution
that results from the erosion of the real value of assets fxed in nominal terms
is hardly arbitrary. It is another question whether such redistribution is fair—
William Baumol, Alan Blinder, and John Solow may (or may not) be on the
Copyright © 2021. Harvard University Press. All rights reserved.
money when they write, “Te gainers do not earn their spoils, and the losers
do not deserve their fate” (2020, p. 477). But the redistribution is as systematic
as can be. Wall Street has good reason to fear and loath infation.
Keynes himself in his pre-Keynesian manifestation was partial to the view
espoused by today’s textbooks. Indeed, for the younger Keynes, infation re-
distributed not only arbitrarily but apocalyptically. In a famous passage from
Te Economic Consequences of the Peace he wrote:
Lenin is said to have declared that the best way to destroy the Capitalist Sys-
tem was to debauch the currency. By a continuing process of infation, gov-
ernments can confscate, secretly and unobserved, an important part of the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
19 • inflation and employment empirics 769
wealth of their citizens. By this method they not only confscate, but they
confscate arbitrarily; and, while the process impoverishes many, it actually
enriches some. Te sight of this arbitrary rearrangement of riches strikes not
only at security, but at confdence in the equity of the existing distribution of
wealth. Tose to whom the system brings windfalls, beyond their deserts
and even beyond their expectations or desires, become “profteers,” who are
the object of the hatred of the bourgeoisie, whom the infationism has im-
poverished, not less than of the proletariat. As the infation proceeds and the
real value of the currency fuctuates wildly from month to month, all perma-
nent relations between debtors and creditors, which form the ultimate foun-
dation of capitalism, become so utterly disordered as to be almost meaning-
less; and the process of wealth-getting degenerates into a gamble and a
lottery.
Lenin was certainly right. Tere is no subtler, no surer means of overturn-
ing the existing basis of society than to debauch the currency. Te process
engages all the hidden forces of economic law on the side of destruction,
and does it in a manner which not one man in a million is able to diagnose.
(1919, pp. 235–236)
In the end, none of the economic reasons—the distortions caused by
hyperinfation, the inefficiencies associated with more modest rates of infa
tion, the arbitrary redistribution of wealth—really explains the fear and the
loathing of the public. Hyperinfation does distort the economy in major
ways, but there is a missing step in the argument. I have never read a convinc-
ing defense of the argument that infation is like pregnancy: there can be no
such thing as a little bit, and initially modest rates of infation, if not nipped in
the bud, will escalate into hyperinfation. Te other inefficiencies, shoeleather
and menu costs, are anachronistic in the sense that they are based on obsolete
technologies. Te arbitrary redistribution of wealth turns out to be not so
arbitrary afer all, and this argument can easily explain why the very rich
Copyright © 2021. Harvard University Press. All rights reserved.
are sworn enemies of infation—but it doesn’t explain the attitudes of the rest
of us.
I think Keynes was in touch with emotions that play a huge role in the pub-
lic’s dislike of infation. Shiller reports a strong public identifcation with the
idea that infation undermines social cohesion and international prestige, and
in the end can, if unchecked, lead to “economic and political chaos” (1997,
pp. 37–46). But this is only part of the story. In a political system where
money speaks loud and clear, and threatens to drown out every other voice, it
is not credible that the very real economic interest of the creditor class in
keeping prices in check plays no role in shaping political attitudes. Main
Street may have its reasons, right or wrong, for disliking infation, but central
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
770 Raising Keynes
banks and fnance ministers are more attuned to Wall Street in emphasizing
price stability as the holy grail of economic policy.
Keynes for All Seasons
Tis chapter and the last develop and test a model that responds to Harrod’s
plea for a framework suitable for analyzing the economy over a period in
which investment adds to the capital stock, and thus adds to the supply of
goods while at the same time being a source of aggregate demand, indeed the
key source of private demand so long as consumption is assumed to respond
passively to income. As Harrod (1937) wrote in his review of Te General
Teory,
Te only criticism of Mr. Keynes which I venture to ofer is that his system is
still static . . . In the dynamic theory, as I envisage it, . . . our question will
then be, what rate of growth can continue to obtain, so long as the various
surrounding circumstances, including the propensity to save, remain the
same? (pp. 85–86)
My theory immediately departs from Harrod and the ensuing literature in
rejecting the idea of an exogenously determined labor force. Tis involves two
major conceptual changes. First, we have to replace that mythical notion, “the
economy,” with the portion of the economy for which the essential features of
wage labor and proft maximization are present. Tis is not to say that sectors
like the household, the family farm, and the government—sectors in which
one or the other distinguishing feature of capitalism is missing—are not wor-
thy of study. Rather, these noncapitalist structures require a very diferent
framework of analysis. One size does not ft all.
Second, we have to replace the idea that the labor force available to the cap
italist sector is given by population, and, indeed, that population is given.
Over the last several centuries, capitalist economies have grown much more
Copyright © 2021. Harvard University Press. All rights reserved.
rapidly than the larger economic structures in which they have been embed-
ded. To do so, they have had to draw labor from noncapitalist parts of the
economy. And when domestic sources of labor have not suf fi ced, permeable
national borders (whatever the of ficial stance of governments has been to-
ward immigration) have flled the gap. Japan is the most notable exception to
this dictum, relying only marginally on immigration, and at that only during
its short-lived experiment with empire in the frst half of the twentieth cen-
tury. South Korea and China, each for its own reasons, have followed the path
charted by Japan.
Like the short-run model developed earlier in this book, the long-run
model is overdetermined because it has three independent schedules—aggre-
gate demand, goods supply, and a conventional wage—but only two state
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
19 • inflation and employment empirics 771
variables, real price and the labor:capital ratio. As in the short run, equilib-
rium cannot be defned apart from the dynamic process by which the econ-
omy—strictly speaking, the capitalist sector—is assumed to adjust. Te equi-
librium associated with a fexprice process and the equilibrium associated
with a fxprice process difer from one another, but both share the property
that the equilibrium is characterized by a stationary labor:capital ratio and a
stationary real-price level.
Defation and stagnation are possible in the long run. But with rare excep-
tions (the Great Depression of course, and, more recently, Japan again comes
to mind), over several centuries the story of capitalism has been one of ex-
panding output and ever-higher prices.
Interpreted through the lens of this model, the historical data for the U.S.
economy from the 1950s to the 2010s suggest three important results. First,
infation responds with a lag to the level of economic activity, which implies
that fxprice adjustment fts actual experience better than does fexprice ad-
justment. With fxprice adjustment, the frst response of the economy to dis-
equilibrium caused by a demand shock is for output to adjust, with price ad-
justments following as producers struggle to get back on their GS schedules,
and money wages respond to the accompanying reduction in employment.
(With fexprice adjustment, the sequence is the other way around: the frst
response is for prices to change, with output and wages both responding to the
initial price change.) Second, aggregate demand shocks lead to a trade-of be-
tween the rate of infation and the level of demand, à la A. W. Phillips and his
eponymous curve. Tird, shocks to the GS schedule, emanating from changes
in the relative price of energy, and to a lesser extent shocks to the conven-
tional wage, shif the Phillips curve while changing the demand-side trade-of
between infation and employment relatively little. Only in this century does
the slope of the Phillips curve change markedly, while during the whole pe-
riod from 1956 to 2013 the Phillips curve shifs no fewer than fve times.
One takeaway is that infation is the price society pays for higher levels of
Copyright © 2021. Harvard University Press. All rights reserved.
economic activity. In contrast with the widespread view that infation is the
enemy of a thriving real economy, the theory presented here leads to the con-
clusion that aggregate demand matters in the long run as well as in the short
run, and that higher levels of aggregate demand produce higher levels of eco-
nomic activity along with higher prices.
Te implication is that monetary policy has real consequences in the
long run as well as in the short run. Tis is of course contrary to the near
unanimous view of the economics profession, which the late Fed Chair Paul
Volcker summarized in these words:
I know that it is fashionable to talk about a “dual mandate”—the claim that
the Fed’s policy should be directed toward the two objectives of price stabil-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
772 Raising Keynes
ity and full employment. Fashionable or not, I fnd that mandate . . . illusory
. . . It is illusory in the sense that it implies a trade-of between economic
growth and price stability, a concept that I thought had long ago been re-
futed not just by Nobel Prize winners but by experience. (2013, pp. 32–33)
Te existence of a long-run trade-of between price stability and employ-
ment weakens the po lit
i
cal argument for an in de
pendent central bank:
namely, that the central bank can and should be an apolitical institution be-
cause there is no reasonable alternative to a focus on price stability, so-called
infation targeting (see, especially, Paul Tucker [2018].) If I am correct in
claiming that there is a confict between employment and price stability, it be
comes relevant to ask who benefts from diferent policies.
Clearly, the benefts of higher levels of economic activity and the benefts of
more stable prices are not shared evenly. Main Street may fear infation, but
Main Street clearly benefts from high levels of economic activity. Wall Street
has good reason to be at best ambivalent about high levels of economic activ-
ity, but more than ample reason to loath infation. And Wall Street has a much
easier time of getting itself heard in the halls of government, particularly in
the boardrooms of central banks.
Since the Great Recession, Europe has been mired in a recovery that has
moved at glacial speed, while an entire generation of Greeks, Italians, and
Spaniards are being sacrifced on the altar of fscal rectitude, a pillar of which
is price stability. Keynes may have been right when he suggested, at the end of
Te General Teory,
Te ideas of economists and political philosophers, both when they are right
and when they are wrong, are more powerful than is commonly understood.
Indeed, the world is ruled by little else. Practical men, who believe them-
selves to be quite exempt from any intellectual infuences, are usually the
slaves of some defunct economist. (p. 383)
He may have been right as well to stress that “the power of vested interests
Copyright © 2021. Harvard University Press. All rights reserved.
is vastly exaggerated compared with the gradual encroachment of ideas.” But
the marketplace of ideas is a market far from the norm of perfect competition.
It welcomes those backed by purchasing power with open arms, while those
who challenge money power struggle to be heard. Other than the 1 percent,
all of us pay the price for this particular market failure.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
epilogue
Attack Them in Their Citadel
Tus, if the heretics . . . are to demolish the forces of nineteenth-
century orthodoxy, . . . they must attack them in their citadel . . .
Only if they are successfully attacked in the citadel can we reason-
ably ask them to look at the problem in a radically new way.
—john maynard keynes
Keynes wrote Te General Teory to save capitalism from itself. Not the rug-
ged individualism of every man for himself and devil take the hindmost, but
a more humane, twentieth-century, version in which the state would provide
the control and direction required for a prosperous economy while preserv-
ing individual liberty.1
Te frst step was to remove the rose-colored glasses through which eco-
nomic orthodoxy had long viewed every man for himself. Among the sup-
posed virtues of unfettered capitalism was the capacity to provide a job for
every willing worker. Tis was an illusion, and a dangerous one at that, for it
paralyzed governments from taking the actions necessary to correct the mar-
ket failure that allowed the economy to get stuck far from full employment.
Tis was the heavy lifing of Te General Teory. Once it was understood
that the economy was not self-regulating, Keynes believed it would become
Copyright © 2021. Harvard University Press. All rights reserved.
legitimate for the government to keep the economy humming and the labor
market at full employment. In the normal course of events, maintaining inter-
est rates at suf ficiently low levels would do the trick by making private invest-
ment suf ficiently proftable to employ workers who would otherwise be re-
dundant; in extremis, it would be necessary to use the government’s power of
the purse. Fiscal policy would complement monetary policy, and the need for
more extensive state planning and control could be avoided.
Te diference between the inaction of the U.S. Government in the early
Portions of this chapter were previously published in “History vs. Equilibrium One More
Time: How Keynes’s General Teory Foundered on the Rocks of Comparative Statics,” Review of
Social Economy 78, no. 1 (2020): 35–52.
773
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
774 Raising Keynes
1930s and its interventions in 2008 and 2009 are testimony to Keynes’s suc-
cess. Te weakness of these interventions and the continuing commitment to
austerity in Europe are testimony to Keynes’s failure.
Tis book explains what Keynes got right and where he went wrong, and
translates his vision into a theoretical basis for a macroeconomics for the
twenty-frst century. With Keynes, I set the argument in a framework of per-
fect competition. Tis may seem an unnecessary concession to the orthodoxy
that Keynes was determined to bring down. Keynes understood—as did his
contemporaries, at least the ones he took seriously—that the capitalist econ-
omy of his day, like the capitalist economy of ours, was riddled with imperfec-
tions. So why not assume these imperfections instead of assuming them
away? Te problem was and is the orthodox belief in a self-regulating econ-
omy. Tis belief is born of total immersion in the teachings of the classical
economists and their modern heirs, epitomized from the mid-twentieth cen-
tury on by the models of Kenneth Arrow and Gerard Debreu (Arrow and
Debreu 1954; Arrow 1963–1964; Debreu 1959). A corollary of their world of
perfect competition, in which there is indeed a job for every willing worker,
just as there is a customer for every apple, is that lapses from full employment
must be the result of imperfections, frictions, rigidities, warts on the body of
capitalism. Once, over lunch, I described my project in this book to a col-
league; his immediate response was “What’s your imperfection?”
I guess the answer is people. As it has developed since the World War II
era, orthodox economics has increasingly relied on an exaggerated degree of
rationality and abundance of knowledge, not to mention an exaggerated de-
gree of self-centered greed, to make its central point about the virtue of mar-
kets. In the hands of Robert Lucas and his New Classical colleagues, a conse-
quence was that, if a full-employment equilibrium is disturbed by a shock to
aggregate demand (or to supply, for that matter), economic agents possess the
knowledge and calculating ability to move immediately to a new full-employ-
ment equilibrium. In this view any need to search for the new equilibrium
Copyright © 2021. Harvard University Press. All rights reserved.
refects limited knowledge or limited cognitive ability, an imperfection or
friction to the smooth functioning of the economy. Alas, this is an imperfec-
tion or friction that can be avoided only if people are more than the human
beings we actually are, endowed with limited knowledge and limited cogni-
tive power.
Orthodoxy has always recognized the problems caused by warts on the
body of capitalism; indeed, orthodoxy has embraced the warts as impedi-
ments to the self-regulating abilities of the economy. At the same time, the
perfection of the textbook economy, wart free, remains a principle, indeed a
litmus test, of the canon. For the mainstream of the economics profession,
these two propositions are not in confict but rather serve as the basis for en-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Epilogue 775
lightened government policy: remove the warts, make the economy over in
the image of the textbook. Period.
Tis is not mere theoretical speculation. Te push to deregulate the econ-
omy, especially the fnancial sector, in the last decades of the twentieth cen-
tury was founded on the idea that a competitive economy is self regulating.
Alan Greenspan was shocked! shocked! shocked! when he discovered that
bankers did not put aside enough capital to cushion themselves against the
risks they took on in the run-up to the fnancial crisis of 2008.
Keynes assumed away imperfections to focus on the deeper problem of
how aggregate demand enters into the determination of economic outcomes.
Te Achilles heel was that Keynes lacked the tools to make a convincing argu-
ment. Tis book makes up for that defciency by recasting Te General Teory
in dynamic terms, that is, in terms of the adjustment process that takes place
when equilibrium is disturbed. Tis leads to radically diferent results from
the comparative-statics method of simply comparing equilibria.
I imply no criticism in suggesting that Keynes was biting of more than he
could chew. Te necessary mathematical techniques were not available even
to the relatively small group of economists that was mathematically literate, a
minority in which Keynes could count himself. But this was changing. Afer
World War II, mathematics was clearly becoming the preferred language of
economics, and in that generation there were plenty of economists with the
requisite skill set. As I note in the appendix to chapter 6, by the 1960s there
was a growing recognition of the inadequacy of comparative statics and the
consequent need for an analysis of dynamics. Many economists—the names
of Don Patinkin (1951 [1948], 1965 [1956]), Bob Clower (1958, 1960, 1984
[1963], and Axel Leijonhufvud (1968) come immediately to mind—recog-
nized the problem; the frst publication of a dynamic model appeared in 1968
(Solow and Stiglitz). Tere were at least two other attempts (Tobin 1975;
Dos Santos Ferreira and Michel [1987]; and Dos Santos Ferreira and Michel
[2013]), but it remains a mystery why Keynesians missed the opportunity to
Copyright © 2021. Harvard University Press. All rights reserved.
put Keynes on a dynamic footing. Most of the efort to update Keynes went in
a diferent direction, typifed by the models of Robert Barro and Herschel
Grossman (1971, 1976) and Edmond Malinvaud (1977, 1980). In any case, by
the 1970s the tide had turned against Keynes, and the clever young men (still
not many women in the economics profession in the 1970s) were focused on
demolishing his theory of economics and putting something new in its place.
Te 1970s saw the birth of the New Classical economics, an apt name for an
approach that combined new heights of technical sophistication with old
theories about the self-regulating nature of the economy.
Te Keynesian project had run out of steam, and in 1979 it was pronounced
dead by one of the leading architects of the New Classical school. In a talk to
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
776 Raising Keynes
the annual management conference held at the Graduate School of the Uni-
versity of Chicago, Robert Lucas proclaimed,
Keynesian economics is dead (maybe “disappeared” is a better term). I do
not exactly know when this happened but it is true today and it was not true
two years ago. Tis is a sociological not an economic observation, so evi-
dence for it is sociological. For example, one cannot fnd a good, under-40
economist who identifes himself, works as “Keynesian.” Indeed, people
even take ofense if referred to in this way. At research seminars people do
not take Keynesian theorizing seriously any more—audience starts to whis-
per, giggles to one another. Leading journals are not getting Keynesian pa-
pers submitted any more. (2013, pp. 500–501; quoted in De Vroey 2016,
pp. 210–211)
A prominent New Keynesian, Alan Blinder, echoed Lucas: “By about 1980, it
was hard to fnd an American academic macroeconomist under the age of 40
who professed to be a Keynesian” (1988, p. 278).
For the anti-Keynesians this was Gresham’s Law in reverse, good ideas
driving out bad ones. Te reality is more complicated: as I argue in chapter 1,
the attack on Keynes was three pronged: in addition to the theoretical attack
waged by Lucas and other New Classical economists, the stagfation of the
1970s was (mis)understood to be inconsistent with Keynesian theory. And
politics infuenced the receptivity to anti-Keynesian theories: particularly in
the United States, Keynes was anathema on the right, which saw Keynesian
policies as a stalking horse for an alien conspiracy intent on imposing social-
ism where God intended rugged individualism to thrive. Te 1970s saw the
collapse of the New Deal consensus in the United States and the eclipse of
social democracy in Europe: the 1980s belonged to Reagan and Tatcher.
Even where social democracy could claim a popular mandate, as in France, it
was a shadow of its earlier self.
Teoretical issues plagued the development of Keynes’s theory long before
Copyright © 2021. Harvard University Press. All rights reserved.
Lucas and friends appeared on the scene. Anticipating Lucas by a decade,
Milton Friedman could write in 1970 that in the absence of imperfections
there was nothing to Keynes’s argument that a capitalist economy could not
be self-regulating. “Tere is,” Friedman concluded, “no fundamental ‘faw in
the price system’ that makes unemployment the natural outcome of a fully
operative market mechanism” (p. 207).
Keynes Defeated: Static Models and the Critics
How could Friedman be so sure? To answer this question, we have to go back
to the static structure of Te General Teory and the ensuing critique. Keynes
elaborated Te General Teory in the context of a fxed money wage, an as-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Epilogue 777
sumption that he regarded as scafolding to simplify the exposition of a novel
theory. It would be time enough to remove the scafolding once the theoreti-
cal groundwork had been laid. And remove the scafolding he did: chapter 19
is titled “Changes in Money Wages.” But this chapter was bound to disappoint
a reader looking for a unifed theoretical argument. Instead, there are a series
of separate reasons why reducing money wages would only exacerbate eco-
nomic dif ficulties when aggregate demand is defcient. Te orthodox argu-
ment for reducing money wages is for Keynes fallacious because it confuses
money wages and real wages. Te money wage may fall in the presence of
unemployment (or by the fat of an authoritarian ruler—Keynes contrasted
the possibilities open to a Hitler, Mussolini, or Stalin with the possibilities
available to leaders of the capitalist democracies), but the real wage is deter-
mined endogenously by the equilibrium price level corresponding to any
given level of money wages. A fall in money prices will generally vitiate any
direct stimulus to production from decreasing money wages.
Within Keynes’s framework the only cogent reason for reducing money
wages is an indirect efect on investment demand. Even though a fall in
money prices defeats the hopes of those who mistakenly think that a reduc-
tion in money wages is tantamount to a reduction in real wages, it can have a
positive efect on output and employment through its efect on the demand
for money. Te Keynes efect, as it came to be called, relies on three assump-
tions, the frst and the third squarely orthodox, the one in the middle novel
and controversial: (1) a fall in money prices reduces the amount of money
required for transactions purposes; (2) this in turn makes more money avail-
able for wealth portfolios and causes interest rates to fall, a fall in interest rates
being a necessary condition for wealth holders to hold the extra cash will-
ingly; and (3) a fall in interest rates expands investment demand. Keynes
considered the logic of this argument to be impeccable but doubted its practi-
cal value. Much simpler, less costly, and much less politically fraught—espe-
cially in a democracy—to expand the real money supply (the amount of
Copyright © 2021. Harvard University Press. All rights reserved.
money divided by the price level) by direct means rather than through the
roundabout procedure of reducing money wages. Besides, the possibilities for
expanding investment demand might be limited by a foor to the long-term
interest rate that provides a hurdle for potential investment projects.
In 1944 Franco Modigliani published a formal model of the Keynes efect
but drew opposite conclusions from Keynes’s. For Modigliani, the lesson was
that a fxed money wage was not scafolding that could be removed once the
building was erected. It was an essential assumption.
Modigliani casts his argument in the conventional form of comparative
statics, in efect asking us to consider separate worlds that are identical in all
but one respect. Population, conditions of production, preferences, endow-
ments of resources (and money!) are all the same. Labor supply is determined
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
778 Raising Keynes
by utility maximization, by workers who optimize in terms of leisure and
goods. Te supply of goods is determined by proft maximization on the part
of capitalist producers. Aggregate demand is determined by consumption
propensities and the demand for investment.
Te only diference is the level of the money wage. Each planet has a difer-
ent money wage. So changing the money wage has a diferent meaning from
what it means in ordinary speech. It is not the case that the wage is $20.00 per
hour today, and $10.00 per hour tomorrow. Planets are unchanging, the same
today as yesterday and tomorrow. Change in this context involves traveling
from world to world and observing the diferences that result from diferent
levels of the money wage.
If we temporarily suppress the labor-supply decision, as Keynes did for
eighteen chapters, we will fnd that in worlds with lower money wages the
equilibrium of aggregate demand and goods supply will involve greater em-
ployment and output. Tis is the consequence of coupling (1) the expansion
of aggregate demand via the Keynes efect with (2) standard orthodox as-
sumptions of proft maximization and diminishing marginal productivity of
labor.
Now bring back the supply of labor. Because investment demand is low in
a high money-wage world, the high wage AD-GS equilibrium provides less
output and employment than the levels consistent with workers’ labor versus
leisure choices, which is to say less employment than is consistent with work-
ers being on their supply curves. Te opposite is true at low levels of money
wages. So, as the money wage falls on our interplanetary travels, somewhere
between very high money wages and very low money wages is the Goldilocks
economy at which all three conditions—aggregate demand, goods supply, and
labor supply—are satisfed. Tis is Modigliani’s full-employment equilibrium.
Modigliani recognizes that the orderly succession from one world to an-
other of greater output and employment associated with an AD-GS equilib-
rium may break down because of the second of the three assumptions behind
Copyright © 2021. Harvard University Press. All rights reserved.
the Keynes efect: more money available for wealth portfolios may not drive
interest rates down suf ficiently. If wealth holders cease to need an inducement
to adjust their portfolios to the infux of cash released from transactions du-
ties, there will be no efect on the interest rate and hence no efect on invest-
ment. So one of the main arguments of Te General Teory remains even afer
Modigliani’s criticism of the assumption of money-wage rigidity.
But it is another supposed efect of falling wages and prices, the real-bal-
ance efect, on which Friedman relied to reach his scathing conclusion that
Keynes had failed in his attack on the citadel. Tis is the story of the efect of
falling prices on consumption demand, making no assumption about invest-
ment demand. Te real-balance efect begins with the determinants of con-
sumption demand. In Te General Teory Keynes argues that consumption
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Epilogue 779
depends primarily on income, with other factors playing smaller roles. Vari-
ous critics—Gottfried Haberler (1939, 1941) was the frst, though the real-
balance efect is generally associated with another critic, Arthur Pigou (1943,
1947), and is ofen referred to as the Pigou efect—noted that wealth also af-
fected consumption. Specifcally, the cash component of wealth would be-
come more valuable the lower the price level; with a low enough price level, a
single rumpled dollar bill you fnd in your grandma’s attic will buy the entire
production of the nation! In these circumstances even if wages and prices
move down in lockstep, the real-balance efect will ensure the self-regulating
nature of the economy, whatever happens to investment demand.
Under the weight of the Keynes efect and the real-balance efect, Keynes-
ians retreated to the position that afer all it was the warts that were the prob
lem. Paul Samuelson claimed that he and his Keynesian colleagues had always
thought Keynes’s argument rested on a bed of imperfections (see chapter 1 for
the relevant quote; Colander and Landreth 1996, pp. 160–161). Be that as it
may, by 1970 there was little opposition among Keynesians to Friedman’s dic-
tum that Keynes was merely a better expositor of sand in the wheels. Leijon-
hufvud’s On Keynesian Economics and the Economics of Keynes, published in
1968, was important in emphasizing the distinction between Keynes and his
followers in this regard. (I overstate the case: fundamentalist Keynesians, par-
ticularly the old guard who had been present at the creation in Cambridge,
England, or arrived soon afer, continued to argue that there was more to the
message than messengers like Samuelson understood. Teir eforts, joined by
younger economists who would become “Post-Keynesians,” met with little
success in the mainstream Keynesian community.)
Tis retreat was a mistake; coupled with the new classical attack it has set
macroeconomics back forty years. Te basic problem was a misunderstand
ing about what the Keynes efect and the real-balance efect really mean.
Te misunderstanding lies in the diference between what Modigliani et al.
thought they had proved and what they had really proved.
Copyright © 2021. Harvard University Press. All rights reserved.
what they thought Suppose there is a shock to a full-employment
equilibrium and the economy moves, temporarily at least, to a Keynesian un-
employment equilibrium, that is, an equilibrium of AD and GS schedules at
which aggregate demand is too weak for full employment. Unemployment
will put pressure on money wages. If the money wage falls from $20.00 to
$10.00, then the price level will fall pari passu, and if prices fall, then invest-
ment (Modigliani) or consumption (Haberler et al.) will increase, and a full-
employment equilibrium will be attained.
what they really proved If we consider a variety of planets that are
identical except for the money wage, which varies on a continuum from zero
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
780 Raising Keynes
to ∞, and we fnd the equilibrium for each of these worlds, there is one Gold-
ilocks world in which full employment obtains, along with the economy being
on its AD and GS schedules. None of these worlds has a history, so we can’t
properly speak of change in this context, at least not of change as Keynes con-
ceived it, where one day the going wage is $20.00 per hour and the next it is
$10.00.
Tere is a way, but it is a torturous one, of transforming change à la Modi-
gliani and Haberler to change à la Keynes. Suppose all agents know that the
Keynesian unemployment equilibrium is untenable because of the pressure
on money wages to fall. And they know more: they each know the model that
generates an AD-GS equilibrium for every world once the money wage for
that world is known. So they know the money wage that corresponds to the
Goldilocks, full-employment, economy. Tey can now recalibrate Goldilocks,
taking account of the change in investment demand that triggered the move
away from full employment. And they can take account of the real-balance
efect, though that is somewhat more complicated because it requires agents
to factor in the change in the value of debts and credits as the price level
changes. If each agent accepts the “equilibrium discipline” that Robert Lucas
introduced to the economics profession in 1977, and each agent believes that
all the others do too, then the economy will move at once to the new Goldi-
locks confguration. Comparative statics morphs into dynamics, even if the
dynamics is concentrated in the split second of time it takes these agents to
calculate the new full-employment equilibrium. Big bang with a vengeance!
Te torture lies in how much knowledge agents must be assumed to have.
In fact, a great virtue of the decentralized market system is how little agents
need to know. At a perfectly competitive equilibrium, a homo economicus or a
mulier economica needs to know his or her own consumption preferences, his
or her own production possibilities, and market prices—nothing else. But
once equilibrium is displaced, those agents who needed to know so little at
the original equilibrium need to be virtually omniscient to immediately fnd
Copyright © 2021. Harvard University Press. All rights reserved.
their way to a new one.
Te equilibrium real wage presumably h asn’t changed, because it depends
only on goods supply and labor supply, and by assumption these have not
changed. But nominal price and wage levels will be diferent, and the only way
agents can know the new levels of nominal prices and wages is to know the
AD schedule. And behind aggregate demand lie investment demand and
liquidity preference, as well as consumption demand. In accord with the
Keynes efect and the real-balance efect, both investment demand and con-
sumption demand will change with the price level. Agents need to know the
model for the economy to fnd the equilibrium for the economy once it’s dis-
placed.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Epilogue 781
Tis is not all. Agents need not only to know but to act. It makes sense to
take actions based on the new Goldilocks only if each agent believes that oth-
ers are going through the same process and are all going to make the same
leap of faith.
In short, there is a possible theory of agent knowledge and behavior that
makes comparisons across space relevant for comparisons over time. But the
theory is on its face implausible. Whatever the merits of equilibrium disci-
pline, it is a most egregious example of making the economy over in the mold
of the textbook to assume that the agent as well as the theorist can (and
should!) subject herself to the discipline of thinking only in terms of equilib-
rium. As I indicated earlier in this chapter, it is an odd use of the word “im-
perfection” to characterize human beings as imperfect because we do not
possess the knowledge and cognitive power of gods.
Keynes Vindicated: A Theory of Real-Time Changes
Central to translating Keynes’s vision into a theory for the twenty-frst cen-
tury is a dynamic alternative to omniscient agents fnding their way back to
equilibrium in a split second of ultrarationalism. To be sure, this alternative
also makes considerable demands on agents once a full-employment equilib-
rium is displaced. Tey must now make decisions about prices and quantities,
whether to change them or to stand pat, two decisions instead of the one that
economic men and women make at a competitive equilibrium. Fortunately,
agents have additional information at their disposal: they can observe not
only whether they are making money but also whether goods are piling up
unsold or fying of their shelves. And they need to know only at the margin.
Boundedly rational consumers need to know only whether the marginal util-
ity per dollar spent on pizza is greater than the marginal utility of a dollar
spent on beer, and to adjust at the margin in accordance with this knowledge
(unlike the omniscient consumer who knows her entire indiference map and
Copyright © 2021. Harvard University Press. All rights reserved.
therefore the optimum consumption bundle). Boundedly rational produc-
ers need to know only whether or not the real price exceeds the real mar-
ginal cost of production, and adjust at the margin (unlike the omniscient
agent who knows her entire production possibilities and therefore knows the
proft-maximizing level of output).
If we think of equilibrium as a combination of a real price and a level of
output that do not change over time, the immediate problem is how to char-
acterize the equilibrium when there are AD, GS, and LS schedules that must
be satisfed. Te mainstream solved (and still solves) the problem by ignoring
aggregate demand, on the grounds that Say’s Law guarantees that the econ-
omy will be on its AD schedule at any combination of real price and output—
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
782 Raising Keynes
except when imperfections muddy the waters. Consequently, equilibrium can
be characterized in terms of goods supply and labor supply. Keynes provision-
ally jettisoned the LS schedule, in order to characterize equilibrium in terms
of aggregate demand and goods supply, but he recognized that ignoring the
LS schedule was not a satisfactory solution—hence, a whole chapter on what
happens when the wage rate changes, not planet to planet but in real time on
the particular planet we happen to inhabit.
We can take account of all three schedules if we are willing to drop the
identifcation of equilibrium with the equality of supply and demand. Instead,
a stationary real price is achieved by nominal prices and wages falling (or ris-
ing) at the same percentage rate. One way this can happen is that employers
reduce prices in order to promote sales, while workers reduce wages in order
to increase employment. If the percentage rates of price and wage reduction
are the same, the real price will not change. Both workers and capitalists are
frustrated: they are trying to change real wages and real prices, but their ac-
tions are ofset by the actions on the other side. Everybody may be frustrated,
but the economy is in equilibrium.
Te resulting equilibria are diferent from the conventional supply equals
demand equilibria in two important respects. First, equilibrium cannot be
determined without specifying the adjustment process outside of equilibrium,
and, second, there are consequently multiple equilibria. Te determining fac-
tor is how producers process the two signals that the market sends them, one
on proftability, the other on whether sales are running ahead of or behind
production. For example, producers can respond to lagging sales by reducing
prices or by cutting back on production. Similarly, producers can respond to
negative profts at the margin by cutting output or by raising prices to dis-
courage demand.
Te story gets more complicated: once we leave the simplest models, it has
to be demonstrated that a given process leads to equilibrium. Tis is not guar-
anteed, and indeed a dynamic version of Modigliani’s model may or may not
Copyright © 2021. Harvard University Press. All rights reserved.
lead to equilibrium.
Te diference between Keynes and the mainstream is the diference be-
tween night and day. An economy composed of omniscient agents who have
internalized Lucas’s equilibrium discipline will vindicate Modigliani, Haber-
ler, Friedman, and the entire anti-Keynesian counterrevolution. If disturbed
by a demand shock, the economy will fnd its way back to the original full-
employment equilibrium. Tis economy is self-regulating, and the best the
government can do to aid the process is—nothing.
An economy composed of agents with bounded knowledge and bounded
rationality will perform very diferently. Tis economy has the potential for
prolonged and self-fulflling depression, an economy characterized by equi-
librium unemployment and continuing defation. Wages and prices fall, but
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Epilogue 783
there is no Keynes efect: the lower price level does not drive down interest
rates and stimulate investment. Nor do falling prices stimulate consumption
via the real-balance efect.
Te reason is that the Keynes efect does not work in real time as it is sup-
posed to work in a comparative-statics framework. To be sure, a fall in prices
leads to a reduction in transactions demand, but this does not lead to greater
amounts of money pouring into wealth portfolios. “Money” turns out not to
be homogeneous, and reductions in the demand for money to transact busi-
ness are refected in reductions in bank lending, not in an infux of money
into wealth portfolios.
Instead of decreasing the rate of interest and increasing investment de-
mand, the efect of falling prices is the opposite. Te efective rate of interest
for investment decisions is the real rate of interest, the sum of the nominal
interest rate and the rate of defation. (Te impact of price changes on the ef-
fective rate of interest is called the Fisher efect, afer Irving Fisher, who iden
tifed the wedge between real and nominal interest rates due to changing
prices in 1896.) Te greater the equilibrium rate of defation, the less invest-
ment.
Te real-balance efect is also undone by the change in prices: as prices fall,
holders of cash or fxed-income securities prosper, as Haberler, Pigou, and
especially Don Patinkin (1948) emphasized. But continuing defation spells
increasing hardship and possibly disaster for debtors. When the real debt
reaches intolerable levels, the windfall to creditors is swamped by the wave of
bankruptcies that wipe out the debtors. Te real-balance efect is ashes in the
mouths of every agent who took out loans in the good times and has to pay
interest and principal fxed in nominal terms when her products bring in sig
nifcantly less cash. Neither the Fisher efect nor the calamitous efect of defa-
tion on debtors has any counterpart in the comparative statics of omniscient
agents. Both efects depend on real-time changes in the price level that have
no place in interplanetary comparisons.
Copyright © 2021. Harvard University Press. All rights reserved.
Te asymmetrical efects of falling prices on debtors and creditors was
mentioned only in passing by Keynes. But it is the heart of Irving Fisher’s
1933 contribution. Keynes couldn’t ft debt and defation into his static frame-
work, and Fisher had no framework of aggregate demand into which to ft his
“Debt-Defation Teory of Great Depressions.” A shame because the marriage
of Fisher and Keynes, under the canopy of a dynamic theory, goes a long way
to explaining the calamity of the Great Depression.
Agriculture is a case in point. In terms of employment and output, agricul-
ture suffered no depression: both the number of people working on U.S.
farms and the crops they produced were larger in 1933 than in 1929. But
farmers suffered nevertheless. Coupled with the predominance of the family
farm, the dynamics of adjustment led agricultural prices to fall much more
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
784 Raising Keynes
than in the rest of the economy. Farmers who had taken out loans in the
1920s, when wheat was at $1.25 per bushel and cotton at $0.20 per pound,
were obliged to pay back the interest and principal with $0.50 wheat and
$0.07 cotton. Te surprise is not that so many farmers went under but that so
many remained solvent.
Building Blocks
Recasting the argument in a dynamic form is a main element in rescuing
Keynes’s vision. But it is not enough by itself. Afer more than eighty years of
critical discussion, all the building blocks of Te General Teory are in need of
reevaluation. Most important is the most novel, Keynes’s theory of interest.
Keynes was surely right to argue that the rate of interest is not the “price” that
balances desired saving and the demand for investment. If it were, the entire
argument of Te General Teory would fail, because the price mechanism—
the interest rate—would always adjust to maintain equality between desired
saving and investment and thus between aggregate demand and the level of
output. In other words, Say’s Law.
Instead, Keynes places the action of interest rate determination where it
belongs: in markets for fnancial assets. Te interest rate, he says, is the price
of parting with liquidity, of giving up the advantages of holding cash, which
an agent must sacrifce when she puts her wealth into bonds or other assets
that fuctuate in value. Keynes’s theory of interest is a theory of liquidity pref-
erence.
So far, so good. Te problem with the theory is that it determines the
spreads between the yields of diferent assets but not the level of interest rates.
In Keynes’s two-asset model, the spread is anchored by the assumption that
the safe alternative to bonds is cash, which yields no interest. In this case the
spread between the yield on cash and the yield on bonds is the yield on bonds.
But this is a special case. In reality, the relevant liquid alternative to bonds
Copyright © 2021. Harvard University Press. All rights reserved.
is not cash but assets such as short-term T-bills, which unlike cash normally
pay interest. In this more general case, the spread between the liquid asset and
other assets does not determine the level of interest rates.
So what does? Nothing, at least nothing inside the model. When Keynes
asserted that a capitalist economy lef to its own devices was not self-regulat-
ing, he meant that a competitive capitalist economy would likely settle at an
equilibrium with less than full employment. Without a theory of interest, the
level of investment demand is indeterminate and so, consequently, are the
level of aggregate demand and the equilibrium.
Tis does not mean that the real world is in a state of perpetual drif, but
that the model is incomplete. One way of closing the model to bring it more
in line with reality is to introduce a central bank, whose function is pre-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Epilogue 785
cisely—or was before quantitative easing was added to the quiver of mone-
tary-policy arrows—to fx the rate of interest on safe short-term assets like
T-bills. Once this rate is determined, spreads translate into levels, and the in-
determinacy disappears.
A chief result of liquidity-preference theory survives its reformulation as a
theory of spreads, namely, the liquidity trap. But the meaning of the liquidity
trap is diferent. In the context of agents who choose between cash and bonds,
the liquidity trap refers to the behavior of the bond yield. In the context of
agents who choose between bills and bonds, the reference is to the zero lower
bound on the bill rate; the spread between bills and bonds at the zlb deter-
mines the efect on the bond yield. In both cases, however, the liquidity trap
(not Keynes’s terminology—see Ingo Barens [2018]) is a situation where a
foor to the bond yield prevents suf ficient investment demand for the econ-
omy to achieve a full-employment equilibrium; at this point the central bank
of Keynes’s day—a day that lasted until the a doption of quantitative easing in
the wake of the fnancial crisis of 2008 and the ensuing Great Recession—has
lost control over the economy.
Keynes’s failure to provide a theory of interest reveals a deeper problem
with Te General Teory. Te full title is Te General Teory of Employment,
Interest and Money, but, in fact, there is no theory of money. Most of the time
money is what Milton Friedman later called helicopter money, money created
by the central bank and deposited to the accounts of economic agents (Fried-
man 1969). At other points, though Keynes surely knew better, money is a
commodity—gold, silver, or cowrie shells. At only one point in Te General
Teory (in the chapter on wage changes) does Keynes allude to the possible
efects of the expansion and the contraction of the money supply by a frac-
tional-reserve banking system.
Te Keynes efect is a creature of this confusion. So is the identifcation of
money as a means of payment (cash in the form of dollars, euros, or pounds)
with money as a store of value (safe short-term assets such as T-bills) and the
Copyright © 2021. Harvard University Press. All rights reserved.
consequent argument that liquidity preference is a theory of the level of inter-
est rates rather than a theory of the spread between interest rates.
Fiscal Policy in Theory and Practice
For Keynes, Te General Teory was not an exercise in pure theory. It was in-
tended to disarm the opposition to government intervention to manage ag-
gregate demand, hence the need to make his readers understand, frst, that
the economy was not self-regulating, and, second, that there are limits to the
efectiveness of monetary policy as a means of steering the economy to full
employment.
Tis is the background for the essential policy innovation that emerges
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
786 Raising Keynes
from Te General Teory, the management of aggregate demand through the
adjustment of taxation and government spending to the vicissitudes of private
demand. Te importance of Te General Teory in this regard is not the de-
velopment of a theory of fscal policy. Tere is virtually nothing of the kind in
the book, and what there is eerily resembles the observations of an economist
who was to become a leading critic of book. Read the following passage from
Te General Teory
If the Treasury were to fll old bottles with banknotes, bury them at suitable
depths in disused coal mines which are then flled up to the surface with
town rubbish, and leave it to private enterprise on well-tried principles of
laissez-faire to dig the notes up again (the right to do so being obtained, of
course, by tendering for leases of the note-bearing territory), there need be
no more unemployment and, with the help of the repercussions, the real
income of the community, and its capital wealth also, would probably be-
come a good deal greater than it actually is. (p. 129)
and compare it with what Jacob Viner wrote three years before the publication
of Te General Teory
If the government were to employ men to dig ditches and fll them up again,
there would be nothing to show aferwards. But, nevertheless, even these
expenditures would be an indirect contribution to business recovery. Teir
major importance would not be in the public works or the unemployment
relief which immediately resulted, but in the possibility of hope that a sub-
stantial expenditure would act as a priming of the business pump, would
encourage business men by increased sales, make them more optimistic, lead
them to increase the number of their employees, and so on. (1933b, p. 130)
Te diference between the two is that Viner had no theoretical framework
in which his policy prescription made sense. (Tis is why Viner is forgotten
except by specialists in the history of economic thought while Keynes is re-
Copyright © 2021. Harvard University Press. All rights reserved.
membered not only by economists but by a much broader public.) In the
context of Te General Teory, the same advice makes eminent sense. Keynes’s
contribution to fscal policy lay in providing a theoretical framework for
the role of aggregate demand in the determination of equilibrium. Te Gen-
eral Teory undermines the Micawber-like devotion to balanced budgets that
characterized both the economics profession and “responsible” politicians.
It was lef to Keynes’s disciples (and critics) to develop a Keynesian theory
of fscal policy. A major role was played by Abba Lerner, whose 1944 book
developed a theory of “functional fnance,” which he opposed to the “sound
fnance” of Mr. Micawber. Te emphasis in functional fnance is that taxation
and spending should be in the service of stabilizing the economy at full em-
ployment, with the corollary that the defcit is irrelevant.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Epilogue 787
As a matter of theory, it is not the case that functional fnance need lead to
permanent defcits and a growing public debt. In principle, aggregate demand
can be managed with a balanced budget if the size of the government budget
is not a concern and we ignore the problems of ramping up and ramping
down public expenditure in a timely manner. Even if these conditions are not
met, as indeed they will not be met in practice, functional fnance does not
preclude balancing the budget over good times and bad. When the economy
is booming, functional fnance dictates a surplus to prevent aggregate de-
mand from running ahead of capacity.
Stabilizing the economy in accord with the theory of functional fnance has
been a dif ficult sell, even though public opinion proved much more receptive
to government defcits when the economy went south in 2008 than during the
Great Depression. Nonetheless, whether or not politicians or economists ac-
knowledge it, the U.S. government has been generally faithful to the dictates
of functional fnance. To a great extent this has been automatic, because taxa-
tion and transfer payments (think unemployment insurance) change more or
less in line with economic activity, while government expenditure is relatively
insensitive to the state of the economy.
But when automatic stabilization fails to do the trick, as in the recessions of
the early 1980s and the early years of this century and, of course, during the
Great Recession, the government has embraced stimulus in the form of a re-
duction of tax rates and an increase in spending. Ronald Reagan didn’t ac-
knowledge that his 1981 tax cuts were in the service of stimulating aggregate
demand, nor did George W. Bush in the recession of 2001; the rhetoric was all
about the supply side. In 2008 and 2009, both Bush and Barack Obama made
no bones about the need to stimulate demand.
Tere remains some question about the efectiveness of the Obama stimu-
lus. A part of the counterrevolution to Keynes was the argument that eco-
nomic agents do not base expenditures on current income but rather on long-
term resources. Te permanent-income (M. Friedman 1957) and life-cycle
Copyright © 2021. Harvard University Press. All rights reserved.
(Modigliani and Brumberg 1954; Ando and Modigliani 1963) hypotheses
formalized this idea with respect to individual consumers. According to per-
manent-income and life-cycle logic, rational agents engage in expenditure
smoothing and have very muted responses to any change in disposable in-
come perceived as temporary. In this view neither the adverse impact of a re-
cession nor the attempt to ofset this impact through discretionary fscal pol-
icy has any impact on consumer spending.
Tere is no doubt that some households determine expenditures according
to the dictates of permanent-income, life-cycle rationality. Chapter 9 argues
that these households are likely to be a small proportion of the total, namely,
middle-class professionals with stable job prospects and accordingly stable
income trajectories. For the rest—the working class and the 1 percent (or
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
788 Raising Keynes
maybe the 0.1 percent)—the assumptions underlying the so-called rational
agent have minimal bearing on actual behavior. Most workers have too little
idea of their future incomes and needs to undertake such planning, and at the
other end of the spectrum there is no need for such planning.
Te most controversial part of the Obama stimulus, however, was not the
tax cuts and transfers to individuals, but the transfers to state governments.
Did these transfers simply shore up state fnances with no efect on spending,
or did the states add to their overall expenditure? John Cogan and John Tay-
lor (2012) applied the permanent-income/life-cycle argument to the states
and hypothesized that the states would logically apply the corollary of expen-
diture smoothing. Tey fnd support for this hypothesis in time-series regres-
sions.
Te analysis in chapter 15 exposes faws in Cogan and Taylor’s time-series
approach and, on the basis of both cross-sectional regressions and interviews
with state budget of ficers, concludes that state governments actually spent the
money they received. Tis analysis supports the conclusions of other econo-
mists (for example, Chodorow-Reich et al. 2012; Wilson 2012) who have ana-
lyzed the impact of the portion of the Obama stimulus channeled through the
states.
A second aspect of functional fnance normally gets short shrif but is in
my judgment as important as the stabilization aspect: the appropriate level of
government spending. Stabilization can be achieved by diferent combina-
tions of the level of government spending with the size of the defcit or sur-
plus. From the point of view of stabilization, these diferent combinations
may be equivalent, but not from the point of view of social welfare. Once the
level of output is fxed, government expenditure competes with consumption
and investment for resources. Te relevance for functional fnance is that a
good resolution of this competition may involve a government surplus or a
government defcit. Te diference from stabilization policy is that the surplus
or the defcit is not cyclical in nature, but a permanent feature of the optimal
Copyright © 2021. Harvard University Press. All rights reserved.
confguration of government spending, private investment, and private con-
sumption. So the question of the public debt inevitably raises its ugly head.
My position is that a national discussion of our priorities is long overdue, but
“sound fnance” should not distort the argument.
Keynes in the Long Run
Even economists who accept the label of Keynesian fnd it something of an
oxymoron to place “Keynes” alongside “long run.” Limiting the relevance of
Keynes to (at best) a short period makes sense if we accept a faulty premise
and an illogical corollary. Te premise is that the problem is the warts rather
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Epilogue 789
than the body of capitalism, and the corollary is that these warts are limited to
the short run, magically disappearing with the passage of enough time. Ac-
cepting neither the premise nor the “corollary,” I argue that Keynes is as rele-
vant in the long run as he is in the short period. More specifcally, aggregate
demand matters in the long run as well as in the short run—not instead of
supply, I hasten to add, but in addition.
Te obvious objection to my view is that the unemployment rate shows no
particular trend over time. If aggregate demand infuences output and em-
ployment in the long run, what prevents growing unemployment when de-
mand falls short, or growing pressure on capacity as the demand for labor
squeezes the supply? It would be a remarkable coincidence if aggregate de-
mand matters in the long run but just happens to stay in balance with the
supply of labor so that the rate of unemployment displays no trend.
Implicit in this objection is the assumption that labor-force growth is given,
the consequence of population growth and the exogenous forces determining
participation rates. My assumption is rather that labor-force growth is endog-
enous, determined in the end by demand factors. Behind this assumption is a
view of capitalism as embedded in a larger economy and able to draw on the
“reserve armies” (to use Karl Marx’s terminology) of other sectors as the need
arises. Historically, the family farm and the kitchen have constituted princi-
pal reserve armies throughout the capitalist world. And when farmers and
women have not been available in suf ficient numbers, immigrants from the
capitalist periphery have made up the diference, at least most of the time and
in most places. (Japan is the most notable exception.) Te key point here, and
a diference at least in emphasis from Marx, is that the reserve army is not
given once and for all but constituted and reconstituted over time in the ser
vice of capitalism. Te resulting model may be seen as the ofspring of a mar-
riage between Keynes’s General Teory and W. Arthur Lewis’s 1954 “Eco-
nomic Development with Unlimited Supplies of Labour,” with the marriage
consummated not on the bed of the poor, densely populated, countries that
Copyright © 2021. Harvard University Press. All rights reserved.
Lewis had in mind, but on the capitalist bed that was Keynes’s focus.
Assuming the labor force is endogenous requires us to revisit the theory of
wage determination. In analyzing the short run, I uncritically adopted a sim-
ple view of wage dynamics depending only on the level of unemployment.
Tis simplifcation fts with the idea that the labor supply is exogenous, and a
long-run theory need not dispute this as far as the short run is concerned. But
the simplest long-run theory of wage determination consistent with an en-
dogenous labor force is to ignore altogether the infuence of unemployment
on wages. Instead, wage dynamics are based on an exogenously given conven-
tional wage. By conventional I mean what the classical economists, Adam
Smith and David Ricardo, as well as Marx, meant by subsistence: not a mini-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
790 Raising Keynes
mum that holds body and soul together, but a social norm that refects the
economic, political, and social conditions of a time and place. In the main-
stream literature, the same idea is captured by the concept of a target wage
born of the aspirations of the working class and their power to achieve these
aspirations, both existing in a matrix of culture and history.
Te emphasis on a conventional wage does not mean that the current level
of unemployment is irrelevant to wage dynamics, or, for that matter, to the
evolution of the labor supply. But the mathematics are complicated when the
level of unemployment is included, so the argument is placed in the mathe-
matical appendix to chapter 18 rather than in the main body of that chapter.
Te resulting model deploys the same dynamic structure as the short-run
model used to understand depression economics, with this important difer-
ence: in the long run the interaction of aggregate demand, goods supply, and
labor supply has historically led to infation rather than defation.
One virtue of the model is that it provides a unifed framework for analyz-
ing both demand shocks and supply shocks. Tis permits us to understand
the confusion around the Phillips curve: specifcally, does, or does not, a
trade-of between employment and infation exist in the long run? A long-run
Keynesian model allows us to sort out movements along the Phillips curve
from shifs in the Phillips curve. Te conclusion is that, contrary to both New
Keynesian and New Classical theories, the Phillips curve is, as Robert Gordon
put it in the title of a 2013 working paper, alive and well—or at least was until
the afermath of the Great Recession.
A consequence is that the scope of monetary and fscal policy is not limited
to a short run in which imperfections temporarily open up a space for gov-
ernment to infuence aggregate demand. In contrast with the New Keynesian
view, monetary and fscal policy can impact employment and output in the
long run as well. Governments may choose not to push employment and out-
put at the expense of price stability, but this is a political choice, not the con-
sequence of a world in which infation brings only pain without any gain.
Copyright © 2021. Harvard University Press. All rights reserved.
My model also sheds new light on a perennial question that is as old as—
indeed, older than—Te General Teory: are high wages good for capitalism,
more specifcally, good for employment? Keynes did not spend much time on
this question; in Te General Teory, the real wage is endogenous, determined
as part of the equilibrium confguration, so there is no way to even ask the
question. Tis theoretical issue apart, Keynes thought redistribution to work-
ers was indeed a means of increasing aggregate demand, since workers would
likely consume more of their incomes than capitalists and other segments of
the population. But he considered redistribution as a last resort; there were
better means, like reducing the rate of interest, for stimulating aggregate de-
mand.
Still the question has persisted, especially among lef Keynesians and their
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Epilogue 791
social-democratic allies in politics. My answer, developed originally with
Amit Bhaduri (Marglin and Bhaduri 1990; Bhaduri and Marglin 1990) and
expanded here, is “it depends.” In the model of chapter 18, the question is put
as whether a higher conventional wage is associated with greater employ-
ment.2 Against the positive impact on demand when workers have a higher
propensity to consume than other agents must be weighed the efects of a
higher conventional wage on investment demand. Tese efects are far from
one-sided. A higher real wage means a lower p roft share, but the impact of
the proft share depends on the kind of investment. Lower profts make cap
ital widening—investment that expands capacity—less attractive. But a higher
real wage makes capital deepening—investment that substitutes capital for
labor—more attractive. To complicate matters further, the balance of the two
kinds of investment likely depends where the economy is in the business cy-
cle. In boom times, with output pressing on capacity, capital widening will be
more on point than in slack times, when there is considerable excess capacity.
Capital deepening is less subject to swings in employment and output, so it
will be relatively more important in slack times.
Te implication is that in boom times, high real wages work in opposite
directions on consumption and investment, and it is not possible on theoreti-
cal grounds to say which is more important. By contrast, in slack times, in-
vestment and consumption tell the same positive story about the efect of
high wages. So, yes, “it depends,” but in the conditions that persisted for most
of the decade following the fnancial crisis of 2008, conditions that still persist
in much of southern Europe, the model speaks clearly: a higher conventional
wage would stimulate employment and output.
“It depends” also casts doubt on the idea that economies can be classifed as
wage led, where higher wages stimulate output and employment, or p roft led,
where lower wages (and higher profts) promote these goals. Timing is all: ev
ery capitalist economy is likely to be wage led in bad times and at least more
likely to be proft led in boom conditions.
Copyright © 2021. Harvard University Press. All rights reserved.
What Will It Take to Change Economics?
It takes new ideas, and this book has several worth pondering. First, the focus
on real-time adjustment as the basis of equilibrium. Second, liquidity prefer-
ence as a theory of interest-rate spreads. Tird, the updating of the concep-
tion of money and the critical diference between money as a medium of ex-
change and money as a store of wealth. Fourth, the formulation of a model of
the long run in which aggregate demand plays a critical role. Fifh, the recast-
ing of Lerner’s functional fnance to separate stabilization from the composi-
tion of aggregate demand.
Tere is some creative destruction at work too. Te focus on real-time ad-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
792 Raising Keynes
justment exposes the shortcomings of the comparative-statics exercises on
which rests the claim that unemployment is the consequence of frictions and
rigidities. Te mainstream theory of interest is seen to be based on a category
mistake: there is no natural rate of interest determined by the interplay of in-
vestment demand and desired saving; indeed, there is no determinate level of
interest rates unless we posit a central bank or its equivalent to anchor the
spreads determined by liquidity preference. Tere may be a long-run ten-
dency to full employment, or at least to an equilibrium rate of unemployment,
but this is because in the long run the labor force is itself endogenous, not
because there is an endogenous mechanism for adjusting the level of employ-
ment to an exogenously given labor force. Tere is a Phillips curve in the long
run as well as in the short run, and the trade-of between employment and
price stability requires political choices. Finally, while this book does not dis-
pute the idea that a negative efect of high wages on investment might counter
its positive efect on consumption, it adds some perspective by analyzing the
diference between capital widening and capital deepening. Whereas capital
widening creates a confict between the impact of higher wages on investment
and consumption, capital deepening reinforces the consumption argument
for higher wages. Te distinction between the two kinds of investment is es-
pecially important in thinking about wage policy in slack times, when capital
deepening is likely to predominate.
Ideas are necessary, but not suf ficient, especially if these ideas challenge the
sacred proposition of economics, the benefcent invisible hand that guides the
economy to an ef ficient allocation of resources unless it is undermined by
imperfections, frictions, rigidities. Keynes carried the day, or rather the gen-
eration of 1936, because the new ideas of Te General Teory made common
cause with a newly ascendant political movement, the New Deal in the United
States and social democracy in Europe. Even for those who understood little
of the theory, the conclusions, especially the justifcation for an activist fscal
policy, ft well with the basic underlying philosophy that an important role for
Copyright © 2021. Harvard University Press. All rights reserved.
government was to address market failures. Not with more and better (read
more competitive) markets, but with direct intervention; spending and tax-
ing in the interest of stabilizing aggregate demand is a leading example. In-
deed, outside the economics profession—and to some extent within it—
“Keynesian” came to be identifed with government intervention, including
many forms of intervention to which Keynes was, at least until late in his life,
at best cool, the welfare state a case in point.3
It is no coincidence that the dominance of Keynesian macroeconomics
ended at the same time that the social-democratic and New Deal coalitions
collapsed. Nor is the contemporaneous rise of New Classical economics and
the rise of Margaret Tatcher and Ronald Reagan pure coincidence. New
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Epilogue 793
movements in economics and politics, both on the lef and on the right, draw
strength from one another. It is probably not too strong to say that the one is
a precondition for the other.
Te lack of success of the New Lef in the 1960s supports this view. I was
one of many of my generation who hoped that fundamental change would
emerge from the civil rights movement, the anti- (Vietnam) war movement,
May 1968 in France, and the hot autumn in Italy the same year. Tese move-
ments did have lasting efects, but not the fundamental social change we
hoped for. Tis is not the place, nor am I the person, to ofer a full analysis of
the failures of the 1960s. Nonetheless I think it’s fair to say that one reason
was the lack of any coherent set of ideas about the structures to be put in
place; high on the list of what was missing was a new economics.
Te abundance of new political movements in the second decade of the
twenty-frst century also illustrates my point. It is too early to tell if any of
them, right or lef, is more than a passing reaction to various (and deep) so-
cial ills. But I predict that if any of these movements shows staying power, it
will be, at least in part, because it has forged an alliance with a new economics
that feeds and is fed by the new politics.
In the meantime, those of us who challenge the reigning orthodoxy
are planting seeds. Whether these seeds will germinate; if they germinate,
whether they will grow; if they grow, whether the fruit will be sweet—de-
pends on the soil, the weather, and the myriad other factors beyond our con-
trol. Tis book has planted a few seeds. I hope you will join me not only in
praying for a good crop but in working to bring in the harvest.
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes
References
Acknowledgments
Index
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
notes
Prologue
Epigraphs: Keynes 1973a, p. 406: “A note, probably written in 1932,” according to
Moggridge; Samuelson 1964.
Chapter 1
Epigraph: Keynes 1973a, p. 492.
1. Full disclosure: almost everybody in my department is younger than I am, but
these people were really young, so young that I d idn’t even know them by name.
2. To put matters in perspective, at the worst of the Great Recession, the unemploy-
ment rate in the United States briefy touched 10 percent. In 1936, when John
Maynard Keynes published Te General Teory, the U.S. unemployment rate
stood at 17 percent, almost double the 2009 rate. Tis represented a signifcant
improvement over 1932 and 1933.
At the lowest depths of the Great Depression, in 1933, of ficial statistics re
corded one out of four workers in the United States as out of work, but the of ficial
statistics underestimated the unemployment rate as a fraction of the wage-labor
force. In the 1930s agriculture was a signifcant sector of the economy in terms of
both employment and output, and because of the limited role of wage labor in
pre–World War II agriculture, this sector did not suffer from unemployment in
the same way as industry and commerce did. Until the 1950s American agricul-
Copyright © 2021. Harvard University Press. All rights reserved.
ture was dominated by the family farm, and family farms normally do not lay of
family members as a way of absorbing economic shocks. Te impact of the Great
Depression on agriculture was not on employment and output but on prices and
thus on farmers’ money incomes and, even more so, on the real burden of debt
that farmers carried. (More about this in chapter 8.) Leaving the agricultural sec-
tor out of the denominator, which makes sense if we are making comparisons be-
tween the Great Depression and the Great Recession, the 1933 unemployment
rate in the United States was more like one in three; this is the ratio of the number
of unemployed to the nonagricultural labor force, which was then only 75 percent
of the total labor force, not the 98.5 percent it is in the twenty-frst century.
Among other capitalist countries, Germany suffered the most, with unemploy-
ment rates at the U.S. level. Nor was the calamity solely economic: the Great
797
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
798 Notes to Pages 18–29
Depression must take a good share of the blame for the political calamity named
Adolf Hitler and indirectly for World War II.
3. Schumpeter (1954, p. 41) defnes the pre-analytic vision in these terms:
In order to be able to posit to ourselves any problems at all, we should frst
have to visualize a distinct set of coherent phenomena as a worth-while ob-
ject of our analytic eforts. In other words, analytic efort is of necessity pre-
ceded by a preanalytic cognitive act that supplies the raw material for the
analytic efort.
Schumpter meant the distinction between vision and analysis as a general one but
illustrates it with Keynes’s General Teory.
4. Arthur Cecil Pigou was professor of political economy at the University of Cam-
bridge from 1908 to 1943 and fellow of King’s College, Keynes’s university (and
college), at a time and place when professor was a title mostly used in the singular:
Alfred Marshall, teacher of both Pigou and Keynes, was the frst professor of po
litical economy, holding the chair from 1884 (when economics was still part of the
feld of moral sciences—it was established as a separate feld in 1903) to 1908.
Dennis Robertson, who had earlier been a close collaborator of Keynes but broke
with him over Te General Teory, succeeded Pigou as professor of political econ-
omy, serving in that capacity from 1944 to 1957. In the 1950s, the professorial
ranks began to expand exponentially, and the British academic hierarchy came to
resemble the American model, in which professors occupy the top foor of an aca-
demic apartment house rather than the pinnacle of a pyramid. (Tanks to Geof-
frey Harcourt, personal communication on February 17, 2014, for setting me
straight on the chronology of economics at Cambridge.)
5. Friedman adds a telling footnote afer the sentence that ends in the assertion that
Keynes’s error lay in “neglecting the existence of a desired stock of wealth as a goal
motivating savings”:
Keynes, of course, verbally recognized this point, but it was not incorporated
in his formal model of the economy. Its key role was pointed out frst by
Haberler (1941, pp. 242, 389, 403, 491–503) and subsequently by Pigou
(1947), Tobin (1947), Patinkin (1951), and Johnson (1961).
All these references are to the so-called real-balance efect, which is a very pecu-
Copyright © 2021. Harvard University Press. All rights reserved.
liar, to say the least, application of the idea that Friedman did so much to pro-
mote, namely that wealth rather than income drives consumption. See below in
this chapter and more extensively in chapters 4 and 7 for a discussion of the real-
balance efect and chapter 9 for a discussion of wealth as a driver of consumption.
6. A search in J-Stor turned up classical or neoclassical dichotomy as early as 1936,
but the term seems to have been frst used in the sense of a division between real
and nominal quantities by Don Patinkin in 1949.
7. ”So-called” precisely because the quantity theory of money was not a theory of
money at all, but rather a theory of the price level!
8. ”Our method of analyzing the economic behavior of the present under the infu
ence of changing ideas about the future is one which depends on the interaction
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 36–37 799
of supply and demand, and is in this way linked up with our fundamental theory
of value. We are thus led to a more general theory, which includes the classical
theory with which we are familiar, as a special case” (Te General Teory, p. vii).
9. Samuelson’s formulation, that full employment would “validate and bring back
into relevance the classical verities” (1955, p. vi), echoed Keynes:
If our central controls [over the level of investment] succeed in establishing
an aggregate volume of output corresponding to full employment . . . , the
classical theory c omes into its own again . . . then there is no objection to be
raised against the classical analysis of the manner in which private self-
interest will determine what in particular is produced, in what proportions
the factors of production will be combined to produce it, and how the value
of the fnal product will be distributed. (Te General Teory, pp. 378–379)
10. Some actually learned from that event. At a hearing of the House Committee on
Oversight and Government Reform on October 23, 2008, Alan Greenspan, for-
mer head of the Federal Reserve, had this to say:
Tose of us who have looked to the self-interest of lending institutions to
protect shareholders’ equity—myself especially—are in a state of shocked
disbelief . . .
A Nobel Prize was awarded for discovery of the pricing model that un-
derpins much of the advance in derivatives markets. Tis modern risk man-
agement paradigm held sway for decades. Te whole intellectual edifce,
however, collapsed in the summer of last year, because the data inputted into
the risk management models generally covered only the past two decades, a
period of euphoria . . .
I made a mistake in presuming that the self-interest of organizations, spe
cifcally banks and others, were such is [sic] that they were best capable of
protecting their own shareholders and their equity in the frms . . . (U.S.
Government Printing Of fi ce 2008)
Te chair of the committee, Representative Henry Waxman, had pressed Green
span on his persistent and consistent championing of deregulation over a long
career:
You were, perhaps, the leading proponent of deregulation of our fnancial
Copyright © 2021. Harvard University Press. All rights reserved.
markets. Certainly you were the most infuential voice for deregulation. You
have been a staunch advocate for letting markets regulate themselves.
Waxman then cut to the chase: “My question for you is simple, were you wrong?”
Afer Greenspan’s less than convincing defense of the hands-of Fed, which em-
phasized that, for the most part, deregulation (or rather self-regulation) worked
just fne, Waxman interrupted to ask: “Well, where did you make a mistake then?”
Greenspan’s response deserves to be quoted more fully. Afer acknowledging his
mistake—see above—Greenspan went on to say,
Te problem here is something which looked to be a very solid edifce, and,
indeed, a critical pillar to market competition and free markets, did break
down. And I think that, as I said, shocked me. I still do not fully understand
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
800 Notes to Pages 40–48
why it happened and, obviously, to the extent that I fgure out where it hap-
pened and why, I will change my views. If the facts change, I will change.
(Keynes is ofen cited as the source for the last sentence. Accused as he ofen was
of inconsistency, he was said to have replied on one occasion, “Sir, when I get new
facts, I may change my mind. What do you do?”)
Greenspan’s self-searching, such as it was, was echoed in the economics pro
fession. But self-examination and self-criticism in the mainstream economics
profession lasted about ten minutes. Afer a brief dalliance with letting a hun
dred fowers bloom, the profession closed ranks, and it was back to business as
usual.
11. Schumpeter’s sin apparently was twofold. First, as noted, he was pessimistic about
the prospects for capitalism, a thesis expounded in Capitalism, Socialism and De-
mocracy. Second, he took Karl Marx seriously despite his rejection of Marx’s fun-
damental doctrines. I have considerable personal experience with the taint that
comes from taking Marx seriously.
12. Te controversy over postwar economic prospects provides a good illustration
of Keynes’s view of models as mental constructions for organizing thought, as
against a mechanical view of models as mirrors, if not of nature, of economic
structures. Followers of Keynes who saw in the simple consumption function
of Te General Teory a formula for prediction were misled: a fall in income
would in their view mechanically produce a proportionate decrease in consump-
tion in line with the marginal propensity to consume. Others, like Keynes himself,
who did not take Te General Teory model literally in its predicating consumer
spending on income, were mentally freer to take into account the special circum-
stances of the war and its legacy of both unprecedented liquid wealth and unprec-
edented unmet desires for goods.
13. Strangely, despite its policy shif, the mission statement remains basically un-
changed. I consulted the ECB website again in the fall of 2015 and in the spring of
2020. Compared with 2009, the 2020 statement changed only the “hierarchy of
objectives”: it now reads:
To maintain price stability is the primary objective of the Eurosystem and of
the single monetary policy for which it is responsible. Tis is laid down in
Copyright © 2021. Harvard University Press. All rights reserved.
the Treaty on the Functioning of the European U nion, Article 127 (1).
“Without prejudice to the objective of price stability,” the Eurosystem
shall also “support the general economic policies in the Union with a view to
contributing to the achievement of the objectives of the Union.” Tese in-
clude inter alia “full employment” and “balanced economic growth.”
Te Treaty establishes a clear hierarchy of objectives for the Eurosystem.
It assigns overriding importance to price stability. Te Treaty makes clear
that ensuring price stability is the most important contribution that mone-
tary policy can make to achieve a favourable economic environment and a
high level of employment.
For the powers that be it is evidently less embarrassing to be consistently wrong
than to change their mind.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 51–58 801
Chapter 2
Epigraphs: Pigou 1933, p. 252; Viner 1933a, p. 9.
1. Keynes’s critique of Pigou in the appendix to chapter 19 of Te General Teory at
tributes an explicit demand-supply model to Te Teory of Unemployment, but on
my reading of Pigou it is more accurate to say that this demand-supply model is
implicit rather than explicit. Te Teory of Unemployment focuses one-sidedly on
the demand for labor that derives from p roft maximization by producers. Te
supply schedule is hardly mentioned, and indeed Pigou’s language ofen suggests
that workers pick a point on the demand schedule even in the absence of mo-
nopoly or other restrictions that replace competition by concerted action; there
are repeated references to the “real wage stipulated by workpeople.” Nonetheless,
Pigou’s general argument can be understood only in the context of an LS schedule
that would combine with the demand schedule to provide full employment were it
not for disturbances to demand and various institutional impediments (trade
unions, minimum wages, unemployment insurance, public sentiment about
“what constitutes a reasonable living wage,” p. 255) to workers’ expressing their
supply preferences. On page 252, in the passage quoted in the epigraph to this
chapter, he comes reasonably close to the picture Keynes paints of him.
Pigou is clearer in his 1927 discussion of what would later be termed struc-
tural unemployment (“structural” to distinguish it from “cyclical” unemploy-
ment):
Before the Great War there can be little doubt that wage-rates in Great Brit-
ain were adjusted in a broad way to the conditions of demand and supply . . .
It was nowhere suggested that the general body of wage-rates had been
forced up too high relatively to the openings for employment, in such wise
that, even had no industrial fuctuations taken place, a substantial number of
healthy persons seeking employment must have been always unable to fnd
it. In the post-war period, however, there is strong reason to believe that an
important change has taken place in this respect; that, partly through direct
State action, and partly through the added strength given to workpeople’s
organisations engaged in wage bargaining by the development of unemploy-
Copyright © 2021. Harvard University Press. All rights reserved.
ment insurance, wage-rates have, over a wide area, been set at a level which
is too high in the above sense; and that the very large percentage of unem-
ployment which has prevailed during the whole of the last six years is due in
considerable measure to this new factor in our economic life. (p. 353)
Pigou’s 1933 treatise is indeed a curious book with a curious history. Apart
from the question of whether or not it contains the demand-supply theory that
Keynes attributes to it, it has the dubious distinction of being praised in the high-
est possible terms in the reviews that appeared just afer publication, only to be-
come in Keynes’s new dispensation the epitome of all that was wrong with the old.
2. Figure 2.5 is not faithful to Viner in one respect because it identifes costs solely
with wages. So what was for Viner an imbalance between producers’ prices and
costs, such as transportation and utilities, is pictured here as an imbalance be-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
802 Notes to Pages 58–64
tween prices and wages. But the logic of Figure 2.5 is clearly Viner’s: the cause of
depression is that the prices businessmen receive for their outputs are well below
the costs of their inputs, and the cure is to restore the balance between costs and
prices.
3. Viner explicitly deplored President Hoover’s exhortations to business to maintain
money wages in the face of falling prices:
At the beginning of the depression, Hoover pledged industry not to cut
wages, and for a long time large-scale industry as a rule adhered to this
pledge. Tere is not time for an adequate exposure of the errors of economic
analysis from which this high-wage doctrine derives its plausibility. But its
basic fallacy, to which, unfortunately, many of my professional colleagues
have succumbed, can be made clear in a few words. All that is guaranteed by
wages higher than employers can aford to pay and still give employment to
the available supply of labor is unemployment. Te doctrine asserts that high
wages mean high purchasing power, but an unemployed laborer has no pur-
chasing power at all, however high may be the wage rate he would get if he
had a job. (1933a, pp. 12–13)
4. Whether or not J. B. Say is the author of Say’s Law has been the subject of much
argument. See William Baumol (1999) for a summary, and Alain Béraud and Guy
Numa (2018).
5. Like Hume (see chapter 1), Mill qualifed the bald statement quoted here to allow
for short-run deviations from a fxed relationship between money and prices. And
like Viner, Mill appears to view the problem caused by temporary deviations from
Say’s Law as one of unbalanced defation—which makes sense in terms of the
quantity theory that informed their thinking. Mill writes,
It is also evident that this temporary derangement of markets is an evil only
because it is temporary. Te fall being solely of money prices, if prices did
not rise again no dealer would lose, since the smaller price would be worth
as much to him as the larger price was before. (1909, bk. 3, chap. 14, para. 4)
6. Te same Bernanke who was the Chair of the Board of Governors of the Federal
Reserve System from 2006 until 2014, that is before, during, and afer the fnancial
crisis of 2008–9 and the consequent recession. If the Frank–Bernanke claim were
Copyright © 2021. Harvard University Press. All rights reserved.
true, Bernanke would have had a much easier time as Fed Chairman!
7. Tis is a common trope in mainstream texts. See Mankiw (2018, chap. 26, fg. 1
and accompanying text, pp. 549–557).
Both the Frank–Bernanke and the Mankiw texts distinguish between a “long
run,” in which the interest rate is determined by equilibrium between desired in-
vestment and desired saving, and a “short run,” in which the interest rate is deter-
mined by the demand for and supply of money (Frank and Bernanke 2007, p. 772;
Mankiw 2018, p. 742). Te two texts root the distinction in terms of the rigidity of
prices and wages in the short run versus fexibility in the long run. Presumably if
prices and wages are fexible in the short run—as I shall assume when developing
my preferred model in chapters 5 through 7—the long-run model applies in the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 65–69 803
short run as well. (By the same token, if prices and wages are assumed to be rigid
in the long run, the Frank–Bernanke and Mankiw short-run models would apply
to the long run.)
Mankiw’s text, the leading text in terms of market share, has a decidedly con-
servative bent. But the more liberal Baumol–Blinder–Solow (2020) text, while
avoiding the idea that the interest rate equilibrates desired investment and saving
in their discussion of macroeconomics, makes the same argument as Mankiw in
its presentation of the theory of distribution (pp. 403–404).
Te Voluntary National Content Standards in Economics, the codifcation of the
mainstream consensus of what economics should be taught to U.S. schoolchil-
dren, says high school graduates should know that “real interest rates rise and fall
to balance the amount saved with the amount borrowed [for investment]. Tis
afects the allocation of scarce resources between present and future uses” (Coun-
cil for Economic Education 2010, p. 30).
8. If the interest rate has no impact on saving, the basic logic of Figure 2.9 is un-
changed: “no impact” means only that the saving-supply schedule becomes verti-
cal. Similarly, were we to assume that saving is interest-dependent but investment
is given independently of the interest rate, the identical logic would still apply to
the determination of equilibrium.
9. Peter Bauer, who later became a well-known opponent of all kinds of government
intervention—he was made a life peer by Margaret Tatcher—published a review
in German in 1938 of the English language discussion of Te General Teory. Tis
review, apparently for the frst time in print, clearly indicated the diference be-
tween the role of stocks of fnancial assets in Keynes’s theory of interest and the
role of fows of investment and saving in the mainstream theory. Tibor Scitovsky
(1940) analyzed the diference between the theories of Keynes and the main-
stream in terms of stocks and fows. (I have not found an earlier English-language
discussion in these terms, but I cannot rule out that one exists.) In an idealized
world without government and foreign trade, the money market might be under-
stood as a market in claims on the capital stock, but in fact the money market in-
cludes a variety of fnancial assets, chiefy government debt, that have no counter-
part in physical capital.
Copyright © 2021. Harvard University Press. All rights reserved.
10. A disproportionate amount of trading is in on-the-run Treasuries. Even allowing
for this disproportion, it remains true that bond trading dwarfs current invest-
ment. See chapter 11 for more detail.
11. As far as I know, nobody has ever tested formally the extent to which interest rates
adjust to desired investment and saving. Te twentieth-century fnance literature
simply assumed that interest rates are determined in asset markets, without bring-
ing in current rates of investment and saving. In this century there have been at-
tempts to include macroeconomic variables in explaining the structure of interest
rates—see Refet Gürkaynak and Jonathan Wright (2012, sec. 4.1) for a summary.
Te contrast with mainstream macro is striking: in this literature, from intro-
ductory textbooks on up, interest rates are determined by the demand for invest-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
804 Notes to Pages 69–79
ment and the supply of saving. at least in the pristine setting of the long run. Te
textbook waters are frequently muddied by the substitution of the term “loanable
funds” for saving—see chapter 11, note 1.
12. But the same thing is accomplished through the assumption that goods supply
and labor supply can be collapsed into an aggregate-supply schedule, which in the
long run is vertical. See chapter 4.
13. An early, perhaps the frst, exposition of the aggregate-supply schedule as the pro
jection in Y × P space of the equilibrium of goods supply and labor supply is in
Jacob Marschak’s Income, Employment, and the Price Level, lectures given to grad-
uate students at the University of Chicago in 1948 and 1949. Marschak’s exposi-
tion is in the context of his elaboration of an aggregate-demand, aggregate-supply
model based on the paper written by his PhD student Modigliani (1944), to which
I have alluded earlier. Tis AD-AS model may be also be a frst. AD-AS did not
catch on for several decades, but once it did it became the gold standard for pre-
senting (and circumscribing) Keynes. Marschak’s lectures do not appear to have
had much infuence outside of Chicago. Income, Employment, and the Price Level
was checked out of the Harvard library exactly once between 1963 and 2010.
14. In the 1970s and 1980s mainstream macroeconomists formalized nominal wage
rigidity in terms of overlapping wage contracts (for example, Fischer 1977). Over-
lapping contracts prevent nominal wages from fully adjusting for a period defned
by the longest extant contract. Afer this short run of nominal rigidity, wages be-
come fully fexible over a longer period. Subsequently, John Taylor (1979, 1980)
developed overlapping contract models in which rigidities could persist over a
longer period.
15. Irrationally is in quotes because, as we shall see in chapter 9, it is an abuse of lan-
guage to identify rationality with calculation and stigmatize the absence of calcu-
lation as irrationality.
Chapter 3
Epigraph: Te General Teory, p. 28.
1. I depart from Keynes in what is included under transactions demand and what is
Copyright © 2021. Harvard University Press. All rights reserved.
included under asset demand. Te division between a transactions demand deter-
mined by the level of output and the price level and an asset demand determined
by the rate of interest leaves the precautionary motive for holding money—provi-
sion for a rainy day—somewhat in limbo. See chapter 11.
2. Tere is a theoretical equivalence between the frst-and the second-pass models
since in equilibrium there is a one-to-one correspondence between interest rate
and money supply. But there is a crucial diference once we leave behind the sim-
plifying assumption of a fxed money wage, one that Keynes does not appear to
notice. He moves seamlessly from the frst-pass model to the second-pass model
when he takes stock of the argument (Te General Teory, chapter 18) prior to
taking on the question of what happens if money wages change.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 80–97 805
3. We ignore the possibility of esoteric time-streams of investment returns and costs
that can produce multiple rates of return.
4. Te work of the circus and Kahn’s article were not independent sources for the re-
consideration of what happens when expenditure and income are out of balance.
Kahn was not only a member of the circus, but the intermediary through whom
Keynes was kept informed of the progress of the group’s discussions and through
whom Keynes suggested questions and issues for it to consider (Keynes 1973a,
pp. 338–339). Kahn’s multiplier article systematized an argument that Keynes and
his collaborator H. D. Henderson had made on behalf of the Liberal Party in the
1929 election campaign (Keynes 1931a).
5. Two articles by Joan Robinson, also a member of the circus, published in February
and October 1933 (1933a and 1933b) refect the shif from emphasizing price ef-
fects to emphasizing quantity efects. In February, Robinson is still in the mode of
the Treatise, with the primary impact of an imbalance between expenditure and
income falling on prices, though there is a recognition of the possibility of an im-
pact on output and employment. By October, the shif in emphasis is clear:
When prices are in excess of costs windfall profts are earned by entrepre-
neurs, and however much of these profts the entrepreneurs spend the to-
tal of p
rofts remains unchanged, since spending by one entrepreneur only
serves to increase the windfall profts of others. Tis argument is valid upon
the assumption that an increase in demand for consumption goods leads to
no increase in their supply. Now to assume that the supply of goods is per-
fectly inelastic is a natural simplifcation to make, at the frst step in the argu-
ment, if we are primarily interested in the price-level, but to make such an
assumption when we are primarily interested in the volume of output is to
assume away the whole point of the argument. (p. 24)
6. Tis indeed is the story by which my generation of economists was introduced to
the concept of aggregate demand. It is only one story—see chapters 5 and 6—but
for most of the economy it is likely the more relevant story empirically—see chap-
ters 8 and 19.
7. If prices matter for the transactions demand for money, does the wage rate also
matter? In principle, yes, but the infuence is conventionally ignored. Te simplest
Copyright © 2021. Harvard University Press. All rights reserved.
assumption is that wages are paid out of the proceeds of output sales, so that
money is needed only to fnance nonlabor inputs to the production process, not
to fnance wage payments.
8. Lawrence Klein (1947, esp. pp. 84–87) also emphasizes the inelasticity of invest-
ment demand as the primary reason for the existence of an unemployment equi-
librium. See chapter 17, especially note 7, for a more extensive discussion of stag-
nation.
9. Te discussion of fscal policy is with respect to the factors that may afect the
propensity to consume (Te General Teory, pp. 94–95). Te closest Keynes c omes
to endorsing defcit spending, to which his name is inseparably linked in the pub-
lic imagination, is in the concluding section of chapter 10 in Te General Teory,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
806 Notes to Pages 99–104
where Keynes lays out the relationship between the propensity to consume and
the multiplier. It is here one fnds the ironic endorsement of unproductive public
expenditure as better than no public expenditure at all, quoted at length in chap-
ter 2.
10. Fixing the total money supply may seem the obvious modeling strategy for inves-
tigating whether laissez-faire will produce full employment. In fact, it is obvious
only in the context of an economy where money consists exclusively of cash, an
economy without banks or at least without fractional-reserve banking, a world of
gold coins. As soon as we enter a world in which money is created endogenously
by a fractional-reserve banking system, the assumption of a given money supply is
hardly compelling. We return to the theoretical issue in chapters 4 and 7, and
again in chapter 13. Endogenous money plays out diferently in a General Teory
based in comparative statics, as against a General Teory based in real-time ad-
justment. Chapter 8 examines the empirics of money-supply determination dur-
ing the Great Depression.
11. Observe that if there is no asset demand for money, as the classical quantity the-
ory of money assumes, the LM schedule becomes a vertical line: in the equation
MV = PY, the quantity of money (M) uniquely determines the level of income (Y)
once the income velocity of money (V) and the price level (P) are fxed. Te LM
schedule approaches this classical limiting case as the elasticity of the asset de-
mand for money with respect to the interest rate approaches zero.
12. It should be noted that the IS-LM framework is being used of label. First, in
Hicks’s original version (1937), “output” is the nominal value of production in a
two-sector model consisting of a consumption-goods sector and an investment-
goods sector. Second, and more important, goods prices are not given exoge-
nously but are determined endogenously by proft maximization. Te intersection
of the IS and LM schedules in this construction thus combines demand-and
supply-side considerations and is an equilibrium for the economy as a whole.
Te problem with Hicks’s construction is the inherent ambiguity of the efect of
a wage change on equilibrium. An increase in the money wage increases the equi-
librium price level while decreasing the equilibrium level of real output. So nomi-
nal output can either rise or fall. In any case, what matters is the impact of wage
Copyright © 2021. Harvard University Press. All rights reserved.
changes on real output, about which Hicks’s formulation in terms of nominal
variables is unhelpful.
Subsequent expositors of Keynes modifed Hicks’s construction. Te two-sector
model was typically condensed into a single sector in which output is homoge-
neous, and the argument was formulated in real terms. Crucially, Hicks’s endoge-
nous determination of the price level was abandoned. Alvin Hansen, whose Guide
to Keynes was the essential pony for economics students in the days when they
were expected to read Te General Teory, is a key link in the evolving construc-
tion of IS-LM. Hansen’s IS-LM model has only one output, but Hansen does not
make it clear whether output is measured in nominal or real terms. He doesn’t
have to choose because he treats the price level as given exogenously once and for
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 128–135 807
all, in which case it simply d
oesn’t matter whether we are dealing with nominal or
real output. (Incidentally, Hansen is responsible for what became the standard
nomenclature; for Hicks the schedules were labeled LL and IS.)
With a given price level, the intersection of the IS and LM schedules generates
a single point on the AD schedule. In my own deployment of the IS-LM frame-
work, below, the price level is exogenous but varies parametrically. With output
defned in real terms, and with the simplifying assumptions that both the supply
of saving and the demand for investment are independent of the distribution of
income, the IS schedule is stationary with respect to the price level. In conse-
quence, a varying price level moves only the LM schedule, and the shifing inter-
section with the stationary IS schedule traces out the AD schedule.
I am indebted to Ingo Barens for sending me back to the drawing board with
respect to Hicks’s 1937 paper and for pointing me to his unpublished papers
(1997, 2001) that deal with the history of IS-LM post Hicks.
Chapter 4
Epigraphs: Modigliani 1944, p. 65; Haberler 1941, pp. 408–409; Krugman 2009, p. 182.
1. Tis is a strong assumption. Even with a uniform propensity to consume (an as-
sumption examined in chapter 9), to assume money demand is independent of
the wage rate ignores the possibility, indeed, the likelihood, that wages form part
of the working capital that is fnanced by the transactions component of money
demand—one more area in which we are obliged to suspend disbelief.
2. At one point in chapter 18 of Te General Teory, Keynes takes stock of the argu-
ment up to the point before he departs from static equilibrium to ask what hap-
pens when the wage rate changes in real time. At this point he explicitly recog-
nizes the possibility of a Modigliani-type outcome (as well as the possibility I
examine in chapter 7 that dynamic fuctuations do not lead to full employment):
If competition between unemployed workers always led to a very great re-
duction of the money-wage, there would be a violent instability in the price-
level. Moreover there might be no position of stable equilibrium except in
conditions consistent with full employment; since the wage-unit might have
Copyright © 2021. Harvard University Press. All rights reserved.
to fall without limit until it reached a point where the efect of the abundance
of money in terms of the wage-unit on the rate of interest was suf ficient to
restore a level of full employment. At no other point could there be a resting-
place. (p. 253)
Te argument is confused (see below, chapter 7, note 1), but in any case this is less
of a concession than a literal reading might suggest. Keynes is not arguing that a
laissez-faire economy will actually reach full employment, à la Pigou and Viner.
He rather treats the possibility that equilibrium exists only at full employment,
along with the failure of the economy to achieve full employment in practice, as
an argument why other factors generally intervene to prevent the necessary wage
fexibility:
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
808 Notes to Pages 138–172
Whether or not [the idea that forces exist to limit fuctuations in money
wages] is plausible a priori, experience shows that some such psychological
law must actually hold. (p. 253)
For otherwise, as the previous quotation suggests, the economy would indeed
come to equilibrium only at full employment. Here, as elsewhere in Te General
Teory, the argument is somewhat opaque because Keynes fails to make a clear
distinction between static equilibrium and dynamic disequilibrium paths.
We shall return to this point in chapter 7. Here we focus on the statics of Modi-
gliani’s argument, in particular, how he arrives at a conclusion fundamentally at
odds with Keynes.
3. Tis simply fnesses the problem. We could take explicit account of commercial
lending as a third fnancial asset or consider these loans to be like cash. At this
point, introducing another fnancial asset would complicate the analysis more
than is useful. Treating loans to restaurants like cash would require us to assume
that working-capital loans have little default risk and are so short term that they
hardly fuctuate in value as the interest rate on working capital varies. If both these
assumptions hold, commercial paper is as good as gold (so to speak). But this
raises the question of why anybody would ever hold cash as part of their wealth
portfolio.
Tis is not a question we are ready to take up, but it is an important one. Te
problem is that even in a world of commodity money, the equation M = M1 + M2
breaks down if the same coin can fulfll a transactions demand and serve as a liq-
uid asset for households. In this case, rather than competing with each other, asset
demand and transactions demand become complementary. Commercial loans
transfer the usufruct of money from households to restaurateurs, but restaura-
teurs’ obligations remain liquid assets in wealth portfolios. Tis point is developed
below (p. 140) and in chapter 13. For now, we simply ignore the anomaly, which is
why I say that these assets exist in a kind of limbo, counted as illiquid fnancial
assets but not infuencing wealth management.
An alternative way of dealing with commercial lending is to blur the line be-
tween wealth holding and business by assuming that households own and operate
the restaurants, and we can therefore consolidate the two balance sheets. Transac-
Copyright © 2021. Harvard University Press. All rights reserved.
tions demand becomes simply a claim on household cash, and no lending is in-
volved. But then households must be assumed to directly own the physical capital
associated with the restaurant business, and there is no need for bonds to exist, so
no balancing of fnancial assets.
Chapter 5
Epigraphs: Marshall 1920, p. 345; Walras 1954 [1874], pp. 84–85.
1. Nicholas Kaldor (1934, esp. pp. 133–135) gave the name “cobweb theorem” to this
process, attributing the idea of such a recursive process to Henry Schultz and Um-
berto Ricci.
2. Tis moment may be as good as any to fess up that the opposition between Walras
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 172–192 809
and Marshall is almost certainly overdone to the point of being misleading. Al-
though Walras lays out a “Walrasian” adjustment process early on in the Elements
when he is concerned with a pure exchange economy, once production enters into
the model, the adjustment process becomes the same as Marshall’s—unless I am
misreading Walras’s lesson 21 (1954, sec. 208–220, pp. 243–254).
3. To simplify the mathematics, we assume changes take place continuously rather
than once a day. Te notation P and Q refect this assumption. I will continue to
tell discrete-time stories since periodic change is easier to grasp intuitively than is
continuous change.
4. It may seem contradictory that a fall in price is accompanied by an increase
in production. It is contradictory if we think about equilibrium responses to a
change in demand, when prices and output move in the same direction. Te re-
sponse here is more akin to what happens when the supply schedule moves out-
ward. Te lure of p roft drives quantity decisions—price exceeds marginal cost—
while excess supply drives price down.
5. Whether the trajectory goes straight to equilibrium, as in Figures 5.7 and 5.8, or
follows a more convoluted path depends on parameter values—see the mathemat-
ical appendix to this chapter. If demand is inelastic and supply elastic, the econ-
omy initially overshoots the equilibrium, and goes to equilibrium via a path of
dampened oscillations around the equilibrium.
6. Alternatively, we can imagine consumers determining output adjustment by how
the marginal utility of fsh compares with its price. (Te vertical distance between
the starting point <Q0, P0> and the demand schedule measures the initial difer-
ence between price and marginal utility.) If, as in Figure 5.8, price initially exceeds
marginal utility, consumers will cut back on their fsh orders. Te trajectory is
qualitatively the same as in Figure 5.8, but the logic is diferent, with producers no
longer calling all the shots.
7. Evidently, the momentum of an increasing real price could be sustained beyond
A by assuming that money wages respond negatively to unemployment at the
same time as the nominal price level responds to proft maximization. Tis would
lead to the expansion of output and eventually back to the full-employment,
proft-maximizing equilibrium at E.
Copyright © 2021. Harvard University Press. All rights reserved.
8. Tere is some evidence that this was so for the circus, the group of younger faculty
members who served as a sounding board (and more) as Keynes was moving
from A Treatise on Money to Te General Teory (see the references in chapter 3 to
the group).
Chapter 6
Epigraphs: Te General Teory, p. vii; Patinkin 1948, pp. 562–563; Tobin 1975,
pp. 195–196; Tobin 1993, p. 46.
1. Well, maybe it didn’t. Modigliani defnes “rigid wages” as an “infnite elasticity of
the supply curve of labor when the level of employment is below ‘full’” (1944,
p. 65, note 23), whereas “fexible wages” is later defned as shorthand for “homo-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
810 Notes to Pages 193–203
geneity of zero degree of the supply-of-labor function” (1944, p. 70, note 35).
Since the supply of labor is, in Modigliani’s telling, a function of the real wage, he
presumably did not mean that the supply-of-labor function is homogeneous of
degree zero, which is to say, that a doubling of the real wage would leave labor
supply unafected. Rather, he presumably meant that the supply of labor written as
L = L(W, P) is homogeneous of degree zero in the arguments W and P, which is to
say that doubling W and P would leave the value of L unchanged. Tis is, to say
the least, an odd defnition of fexibility even if one accepts my clarifcation of
what Modigliani intends in his defnition of fexible wages.
2. Te counterargument to the dominant view that money originated as conve
nience to simplify barter trade—the metallic theory—is that money originated as
a way of facilitating taxation—the “chartalist,” or state theory of money. Te impli-
cation of the chartalist view is that money always has been fat money, receiving
its value from the ability of the state both to compel people to pay taxes and to fx
the unit in which taxes are paid. For educating me on the chartalist alternative to
the standard origins story, I am indebted to Christine Desan, especially to her
2014 book Making Money; see also L. Randall Wray (2014) and Pavlina Tcherneva
(2007). An all but forgotten early twentieth-century statement of the chartalist
view is Innes (1913, 1914). Te standard reference is Georg Friedrich Knapp
(1924 [1905]). Keynes fully accepted the chartalist view, paying homage to Knapp
in the opening pages of A Treatise on Money. By contrast, Te General Teory
doesn’t go into the consequences of chartalism for theorizing about the money
supply. See chapter 13 below.
3. Keynes’s take on this fact:
It is not very plausible to assert that unemployment in the United States in
1932 was due either to labour obstinately refusing to accept a reduction of
money-wages or its obstinately demanding a real wage beyond what the pro-
ductivity of the economic machine was capable of furnishing. (Te General
Teory, p. 9)
Tis observation supports the idea that the assumption of a constant money wage
was an expositional tactic rather than a basic assumption of the model. As we
have seen, Keynes’s protests to the contrary have not stopped the development of
Copyright © 2021. Harvard University Press. All rights reserved.
the idea that a fxed money wage and the irrelevance of the LS schedule are core
assumptions of Te General Teory.
4. Troughout this book, formal models do not take account of technical progress.
Alexander Field (2011) has an informative account of technological change dur-
ing the Great Depression.
5. Keynes emphasized the dependence of tomorrow’s investment-demand schedule
on today’s investment: “A rate of investment, higher (or lower) than prevailed
formerly, begins to react unfavourably (or favourably) on the marginal ef ficiency
of capital if it is continued for a period which, measured in years, is not very large”
(Te General Teory, p. 251).
6. Observe that P/W = 4 in Figure 6.2 implies that W0 = 1 in Figure 6.1.
7. Stability is no guarantee that the path to the new equilibrium is direct. Te equi-
librium may be stable but approachable only on a path of smaller and smaller os-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 204–225 811
cillations around the equilibrium. Again, see the mathematical appendix to this
chapter.
8. I owe this point to Peter Skott. Commenting on a workshop presentation of this
model, Skott rightly took me to task for ignoring the consequences of inventory
accumulation. See Chapters 8 and 17.
9. “Tere may exist no expedient by which labour as a whole can reduce its real wage
to a given fgure by making revised money bargains with the entrepreneurs” (Te
General Teory, p. 13).
10. A hedge is in order. Tere is some evidence in Te General Teory that Keynes,
likely because of the British experience during the Depression, saw money-wage
rigidity as a fact of life. But he nevertheless recognized the need to modify this
assumption if his theory was to live up to his claim that Te General Teory was
indeed more general than competing—mainstream—explanations of unemploy-
ment. And, as observed in note 3, he explicitly recognized that in the United
States, money wages fell during the Depression.
11. Clower published a much shorter revision of this paper in the Quarterly Journal of
Economics in 1960. Te published version not only omits the paragraph quoted
above, it also leaves out the dynamics that make Clower’s 1958 paper an intellec-
tual precursor of my own work. Tanks to Romain Plassard for his discussion
of Clower’s 1958 paper in his 2018 article “Clower’s Volte-Face Regarding the
‘Keynesian Revolution’” and for sharing the Clower paper with me.
12. However, the emphasis of the Solow–Stiglitz model was diferent; their purpose
was to shed light on an important controversy of the 1960s, namely, the role of
marginal productivity in determining the real wage, and the assumptions of their
model are designed to this end. In particular, the assumption of a uniform saving
propensity is replaced by the assumption of class-based saving, the Cambridge
saving theory (see chapter 9). In its attempt to infuence the marginal-productivity
debate, the Solow–Stiglitz paper fails utterly: it miscasts marginal-productivity
theory as simply the equality of marginal products with factor prices, in particu-
lar, the equality of the marginal productivity of labor with the real wage. Tis
equality is an immediate corollary of proft maximization and competitive mar-
kets and applies to static General Teory models of chapters 3 and 4 as well as to
Copyright © 2021. Harvard University Press. All rights reserved.
the neoclassical alternative discussed in chapter 2. Marginal-productivity theory
involves a stronger assumption, namely, that, in addition to proft maximization
and competition, the real wage is endogenously determined by an exogenously
given labor supply—in other words, the assumption that the real wage clears the
(competitive) labor market. Te diference between endogenous and exogenous
determination of the real wage is discussed in chapter 18.
Chapter 7
Epigraphs: Robinson 1980, p. 57; Solow 1979, p. 345; Samuelson 1941, p. 102.
1. Te passage concludes
Moreover there might be no position of stable equilibrium except in condi-
tions consistent with full employment; since the wage-unit might have to fall
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
812 Notes to Pages 227–236
without limit until it reached a point where the efect of the abundance of
money in terms of the wage-unit on the rate of interest was suf ficient to re-
store a level of full employment. At no other point could there be a resting-
place.
Keynes’s assertion is garbled. If there is “violent instability in the price level,” so
must there be violent oscillations in output and money wages. In this case there
will be no stable equilibrium consistent with full employment. Keynes got oscilla-
tions in the price level right, but he was incorrect in conjecturing the existence of
a stable equilibrium in the limiting case as W → 0. In Figure 7.2 the money wage
oscillates, like real output, along an ever-widening trajectory.
2. See chapter 13. In chapter 4, I noted Keynes’s rejection of commodity money in
the Treatise. Moreover, in chapter 18 of Te General Teory, the money supply is
assumed to be fxed by a central bank (p. 247) rather than by an exogenous supply
of gold, and in chapter 19 he notes the possibility that the money supply is endog-
enous, and that this would undo the Keynes efect: “If the quantity of money is
itself a function of the wage-and price-level, there is indeed nothing to hope in
this direction” (p. 266). But beyond this passing comment, the idea of endogenous
money is not developed.
Indeed, in the concluding section of Te General Teory, Keynes suggests that
the quantity of money might be limited by the availability of precious metals, for
example in this passage:
Now, if the wage-unit is somewhat stable and not liable to spontaneous
changes of signifcant magnitude (a condition which is almost always satis
fed), if the state of liquidity-preference is somewhat stable, taken as an aver-
age of its short-period fuctuations, and if banking conventions are also sta-
ble, the rate of interest will tend to be governed by the quantity of the
precious metals, measured in terms of the wage-unit, available to satisfy the
community’s desire for liquidity. (p. 326)
3. “Te real bills doctrine is a rule purporting to gear money to production via the
short-term commercial bill of exchange, thereby ensuring that output generates
its own means of purchase and money adapts passively to the legitimate needs of
trade.” (Humphrey 1992, p. 3)
Copyright © 2021. Harvard University Press. All rights reserved.
4. Te real-bills doctrine bears some responsibility for the passivity not only of the
commercial banking system but also of central banks during the Great Depres-
sion; see chapter 8, note 5.
5. Te word bankrupt derives from the Italian banca rotta, describing the ritual by
which bank failure was recognized and communicated: the banca (bench) where
bankers conducted their business in the market place would be broken to signify
a fnancial collapse.
6. Since the fnancial crisis of 2008, when the maximum was last raised, deposit in-
surance is limited to $250,000 per account in the United States.
7. Pigou revisited the real-balance efect some years later (1947). He came in the in-
terim to recognize the diference between outside and inside money, though he
never acknowledged Kalecki’s priority in making this distinction. (Perhaps it was
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 238–247 813
mere forgetfulness. According to William Baumol [2000, p. 1n], “Dennis Robert-
son repeatedly told me how on passing Pigou’s lair, the great man would regularly
emerge, demanding ‘Robertson—tell me, what is the Pigou efect?’”).
More importantly, Pigou came to understand the diference between the com-
parative statics of real balances and the dynamics. His reversal is stunning:
It is ridiculous to suppose that the public authorities would stand passive in
the case of catastrophic disturbances. If a situation arose in which money
income was being driven inexorably downwards in the way contemplated . . .
no government would allow money wage rates to rush downwards very far
. . . Tus the puzzles we have been considering . . . are academic exercises, of
some slight use perhaps for clarifying thought, but with very little chance of
ever being posed on the chequer board of actual life. (1947, p. 188)
8. Te dollar value of frm liabilities actually rises if bonds are marked to market
because PB( ) rises as the interest rate falls.
9. Not always. Te dramatic changes in the real value of the housing stock, both up
and down, are implicated in virtually every story about the causes of the Great
Recession, as was the stock market crash in the story of the Great Depression.
10. Tus Keynes: “Te efect of the lower price-level on the real burden of the Na-
tional Debt and hence on taxation is likely to prove very adverse to business con
fdence” (Te General Teory, p. 264).
11. Tis way of putting the issue refects the government’s position. Te plaintifs in
Perry could have argued that the issue was whether the Supreme Court would al-
low the destruction of private wealth to the beneft of the government.
Perry and Norman overlapped in many respects, but Perry raised the additional
question of the constitutional meaning of the delegation to Congress in Article
One, Section Eight, of the power to “borrow money on the credit of the United
States” and the meaning of “the validity of the public debt” in Section 4 of the
Fourteenth Amendment.
12. Observe that the formula WHEAT = OIL − W/O gives an instantaneous relationship
between own rates of interest on diferent goods. For fnite periods, we must con-
tinuously compound interest and price changes, so that the one-year return of
interest is e − 1 rather than , and the annual change in prices is e − 1 rather
Copyright © 2021. Harvard University Press. All rights reserved.
than . Te one-year formula relating the own rate of interest for wheat to the
wheat rate of interest for oil is
ρW ρO φW / O ρO − φW / O
e − 1 = (e /e ) −1 = e − 1.
13. I’m not sure what to make of Keynes’s own account of the relationship between
real and nominal interest rates. He argues:
Te expectation of a fall in the value of money stimulates investment, and
hence employment generally, because it raises the schedule of the marginal
efficiency of capital, i.e., the investment demand-schedule; and the expecta-
tion of a rise in the value of money is depressing, because it lowers the
schedule of the marginal ef ficiency of capital. (Te General Teory, pp. 141–
142)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
814 Notes to Pages 247–248
So far, so good, though he is now expressing the marginal-ef ficiency-of-capital
schedule in nominal terms rather than in real terms.
He goes on to make a distinction between anticipated and unanticipated price
changes, and his argument becomes harder to follow. If a rise or a fall in prices is
unanticipated, it will have no efect on expectations of returns, and so can be ig-
nored. (At least for the time being, but presumably today’s unanticipated price
change will have an impact on expectations tomorrow.) If, however, price changes
are anticipated,
Te prices of existing goods will be forthwith so adjusted that the advantages
of holding money and of holding goods are again equalized, and it will be
too late for holders of money to gain or to suffer a change in the rate of inter-
est which will ofset the prospective change during the period of the loan in
the value of the money lent. (p. 142)
Keynes appears to be saying here that there cannot be infation or defation that
is correctly anticipated by agents! I take up this question in chapter 13. On the
fundamental question of whether real rates drive nominal rates, or vice versa, I
am with Keynes and for vice versa.
14. A “zero-coupon” bond is a bond that pays no periodic interest, its yield being the
annualized diference between the purchase price and the redemption value at
maturity. T-bills are short-term, zero-coupon bonds. Te empirical appendix to
chapter 12 makes extensive use of zero-coupon bonds.
15. Negative bond yields have become a reality. Except for a short period during 2018,
Swiss government 10-year bonds have been in negative territory continuously
since 2015. Te 10-year German bund danced around zero through much of 2016
and went negative in 2019. Even private companies have been able to borrow long
term at negative rates.
Tere are obvious limits. Were nominal rates to spiral downward, the inconve
nience of holding cash would eventually be outweighed by the nominal yield of
zero on cash; cash would once again be king. To dethrone the king permanently,
cash would have to lose value over time, like Silvio Gesell’s stamped money. Ac-
cording to Keynes (Te General Teory, pp. 353–358), Gesell, as a means of stimu-
lating investment, proposed that a cost be imposed on holding money by a peri-
Copyright © 2021. Harvard University Press. All rights reserved.
odic requirement for stamps to be affixed to money to maintain its nominal value.
Te cost of a stamp would in efect impose a negative return on money held as a
store of value. Ignoring the inconvenience of cash and the risk of bonds, to com-
pete with stamped cash, bonds would need only to incur a smaller annual loss
than the cost of a stamp. If the stamp for a one-dollar bill cost three cents, any
negative bond yield closer to zero than −3 percent would be suf ficient to drive
people from cash to bonds.
16. Fisher himself was aware of the importance of expectations, and while a main
point of his argument in Appreciation and Interest was to argue for the validity of
the classical dichotomy in the long run, he recognized that in the short-run ex-
pectations play an independent role in determining real interest rates. He even
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 253–270 815
made expectations and their sluggish adjustment the centerpiece of his own the-
ory of fuctuations (Dimand and Betancourt 2012).
17. Not quite always. See Tables 4.6 and 4.7 in chapter 4.
Chapter 7 Mathematical Appendix
1. Evidently, we could replace equation (7.1) by the wage equation
W P
= 3 Y − LS ,
W W
since equation (7.1) is obtained by subtracting the above equation from equation
(7.3).
2. Joseph Schumpeter applied his ideas of creative destruction to fnd the silver lin-
ing in the Depression:
Now we have had combines and dry farming, more ef ficient methods of pro-
ducing electricity, rayon and motors and radios, and a thousand similar
things. Tis is really at the bottom of the recurrent troubles of capitalist soci-
ety. Tey are but temporary. Tey are the means to reconstruct each time the
economic system on a more ef ficient plan. But they infict losses while they
last, drive frms into the bankruptcy court, throw people out of employment,
before the ground is clear and the way paved for new achievement of the
kind which has created modern civilization and made the greatness of this
country. (1934, p. 13)
According to Herbert Hoover (1952, p. 30), his Treasury Secretary, Andrew Mel-
lon, advised:
Liquidate labor, liquidate stocks, liquidate the farmers, liquidate real estate
. . . It will purge the rottenness out of the system. High costs of living and
high living will come down. People will work harder, live a more moral life.
Values will be adjusted, and enterprising people will pick up the wrecks from
less competent people.
Both Laurence Ball (2018, chap. 12) and Barry Eichengreen (2015, p. 6) suggest
that the same spirit animated the government's decision to allow Lehman Broth-
Copyright © 2021. Harvard University Press. All rights reserved.
ers to fail in 2008.
Chapter 8
Epigraphs: Roosevelt 1933; Hoover 1952, p. 195.
1. Creditors received no ofsetting windfall unless they consumed wheat, cotton,
and other farm products disproportionately. Teir real returns were governed by
changes in the overall price level, of the order of 10 percent annually over the
critical period 1930 to 1932, not the 30+ percent premium that farmers paid.
2. In the frst phase of the Depression, the shif in the composition of deposits in fa-
vor of time deposits reduced reserve requirements since time deposits carried a
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
816 Notes to Pages 273–278
much lower reserve requirement (3 percent) than did demand deposits (from 7 to
13 percent). By contrast, from 1931 on, deposits—both demand deposits and time
deposits—and reserves contracted sharply, but deposits declined faster than re-
serves.
Reserve requirements on demand deposits also varied with bank location.
Banks in central reserve cities were required to maintain a reserve ratio of 13 per-
cent, banks in reserve cities 10 percent, and country banks 7 percent. All banks,
irrespective of location, were required to hold reserves equal to 3 percent of time
deposits (Goodfriend and Hargraves 1983, p. 11).
3. In his discussion of liquidity preference, Keynes says
Tere is the possibility . . . that, afer the rate of interest has fallen to a certain
level, liquidity-preference may become virtually absolute in the sense that
almost everyone prefers cash to holding a debt which yields so low a rate of
interest . . . Whilst this limiting case might become practically important in
future, I know of no example of it hitherto (Te General Teory, p. 207).
Richard Sutch (2018) has argued persuasively that Keynes’s meaning here of what
came to be called the liquidity trap is not the meaning now ascribed to this con-
cept, namely, the gap between the T-bill rate and the hurdle rate as the bill rate
approaches its zero lower bound. See chapter 12.
4. For historical reasons peculiar to the United States, the Federal Reserve was not
only a latecomer to the ranks of central banks, but only became a full-fedged
central bank in stages, and the process was not complete when the Depression
struck. (See Roger Lowenstein’s account of the creation, 2015.) In the period from
1929 to 1933, the Fed was much more fragmented in its governance than the post-
Depression Fed; much more power lay with the individual Federal Reserve Banks,
twelve of them. Te New York Fed was the most infuential but lacked the power
to force concerted action by the other banks. Tese institutional peculiarities
compounded the constraint posed by the gold standard, though for the United
States, as distinct from Great Britain, Germany, and Austria, the constraint was
essentially psychological.
Te psychology of central banking was shaped by the real-bills doctrine (see
chapter 7). As Eichengreen points out (2015, p. 2):
Copyright © 2021. Harvard University Press. All rights reserved.
Central bankers, for their part, were in thrall to the real bills doctrine, the
idea that they should provide only as much credit as was required for the le-
gitimate needs of business. Tey supplied more credit when business was
expanding and less when it slumped, accentuating booms and busts. Ne-
glecting their responsibility for fnancial stability, they failed to intervene as
lenders of last resort. Te result was cascading bank failures, starving busi-
ness of credit. Prices were allowed to collapse, rendering debts unmanage-
able.
5. For a more systematic view of the holes in Friedman and Schwartz’s argument, a
good starting point is Peter Temin’s 1976 Did Monetary Forces Cause the Great
Depression? Temin fnds little support for Friedman and Schwartz in the data. He
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 278–282 817
favors what he calls the spending hypothesis, basically the failure of aggregate de-
mand, not because the data overwhelmingly support this view but because it re-
mains the plausible alternative afer the shortcomings of the Friedman–Schwartz
analysis are exposed. Eichengreen’s 2015 account of how the Depression unfolded
provides no more comfort for Friedman and Schwartz than does Temin.
6. Tis theory is modifed in chapters 11, 12, and 13 in ways which give more agency
to central banks in determining hurdle rates of interest.
7. Remember that we are not talking about the miniscule share of the U.S. economy
of agriculture in 2020; agricultural employment came to almost 23 percent of total
employment in 1929, and over 25 percent in 1932; and output was more than 10
percent of GDP in 1929, and close to 15 percent in 1933 (at 1929 prices).
8. In the summer of 1936, Galbraith was recruited to explain economics to Henry
Dennison, one of a small number of businessmen who saw good in Roosevelt’s
New Deal. Dennison’s idea was to collaborate with like-minded industrialists to
write a book supporting FDR, and Galbraith was to provide the heavy economic
artillery. Te problem for Galbraith was that Dennison had his own ideas, which
veered toward the unorthodox, especially in his intuitive grasp, avant la lettre, of
the central idea of Te General Teory, namely, the crucial role of aggregate de-
mand. As Galbraith writes,
In the very same weeks that I was writing my brief for my views on competi-
tion, I was reading Te General Teory. As I did, I discovered that Keynes
was with Dennison and not with me . . . I was shaken. Tis was not the
primitive instinct of a businessman; this was the sophisticated case of a
greatly renowned economist. (1981, p. 65)
Galbraith quickly came to embrace Keynes and Te General Teory and later be-
came an important advocate for demand management. But his conversion makes
a main point of chapter 1 clearer, namely, that to justify active demand manage-
ment nothing less than an all-out attack on the received doctrine, one that went
well beyond the superfcialities of frictions and market structure, would do. Given
the hegemony of the belief in the fundamental resilience and benefcence of mar-
kets, it had to be shown that even competitive markets were defective when it
came to providing jobs.
Copyright © 2021. Harvard University Press. All rights reserved.
Keynes provided no explanation for the agricultural anomaly; indeed, he never
took note of it. Likely this was because of the relative unimportance of agriculture
to the British economy; Britain relied mostly on imports for food and fber. One
can only wonder whether the distinct behavior of a family-farm sector would have
made more of an impression on Keynes if he had lived in the United States (or
France), where the family farm persisted until mid-century.
9. Even though the Great Plains did not do so well when it came to the weather. Te
drought that began in 1933 gave rise to a new name for the region: Dust Bowl.
10. Te fgures in Table 8.1 likely overstate the role of hired help in the farm economy.
Te explanatory notes for this table state that to be counted as a family worker, the
family member had to put in ffteen hours or more per week, whereas one hour
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
818 Notes to Page 282
per week was suf ficient to be counted as hired labor (Bureau of the Census 1975,
p. 453).
Te role of hired labor varied markedly from one part of the country to an-
other. According to the reporting farmers on whom the Crop Reporting Board of
the Department of Agriculture relied, between one-half and two-thirds of farm
workers were hired hands in the states bordering on the Pacifc (sometimes
referred to in Board reports as the Far Western region, sometimes as the Pacifc
region); in other parts of the country, the fgure was below one-third. See, for
example, Crop Reporting Board, Bureau of Agricultural Economics, U.S. De
partment of Agriculture, “Farm Wage Rates and Related Data, April 1, 1933,
with Comparisons” (USDA, Washington, D.C., April 11, 1933 [mimeo]), http://
usda.mannlib.cornell.edu/MannUsda/viewDocumentInfo.do?documentID=1063
(accessed November 1, 2017).
11. As paraphrased by Lorie Tarshis, Keynes’s lectures to Cambridge undergraduates
during the years leading up to Te General Teory made the essential institutional
distinction:
If we could have a society in which the employer would feel responsible for
the employee, whether he was working or not, you wouldn’t have to worry
about unemployment. If the employer had to pay a worker’s wage or salary
whether there was work for him or not, he would be working. (Colander and
Landreth 1996, p. 60)
12. Some excerpts from the Crop Reporting Board’s summaries of farm-labor condi-
tions:
On April 1 [1930], all classes of [agricultural] wages . . . were at the lowest
level since 1923. Tis is a refection of the large supply of farm labor due to
the small volume of industrial employment at the present time. (“Farm La-
bor and Wages, April 1, 1930, with Comparisons,” April 9, 1930, p. 1)
Te supply of farm labor in per cent of normal was reported on January 1
[1932] 120.9%, and demand at 60.5%. Supply expressed as a per cent of de-
mand was 199.8, which is the highest ratio recorded since the beginning of
the record in 1918.
Reports received by the Department particularly in the North Central
Copyright © 2021. Harvard University Press. All rights reserved.
States contain numerous instances where farm laborers are working for
board and lodging alone. (“Farm Labor and Wages, January 1, 1932, with
Comparisons,” January 18, 1932, p. 1)
Reports from crop correspondents indicated a supply of farm labor 125.8 per
cent of normal on the frst of the month [April, 1933] as compared with a
supply 122.2 per cent of normal a year earlier. Te larger supply was related,
in turn, to the reduced level of employment in manufacturing industries.
(“Farm Wage Rates and Related Data, April 1, 1933, with Comparisons,”
April 11, 1933 [mimeo], p. 5)
http://usda.mannlib.cornell.edu/MannUsda/viewDocumentInfo.do?documentID
=1063 (accessed November 1, 2017).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 284–291 819
While the Crop Reporting Board focuses on hired labor, the same process
brought thousands of unemployed industrial workers back to the family farms
they had lef when factories were humming.
13. In principle, the economy is the world, because the essence of the model is the
endogenous determination of the relative prices of agricultural and industrial
goods. For this we need a closed economy. Te actual economy of the pre–World
War II United States was a mix: parts of the agricultural economy were very
open—see below—whereas the industrial economy was virtually closed.
14. Once we disaggregate, even to the limited extent of a two-sector model, there is a
demand schedule for industrial goods, and there is a food-demand schedule—see
below—but there is no such thing as an AD schedule.
15. U.S. stocks of wheat at the end of the harvest season more than doubled between
1929 and 1932, from 28 percent of 1929 production to 50 percent of the slightly
smaller production in 1932 (Bureau of the Census 1975, table K 509). Worldwide,
wheat stocks increased from an average 13 percent of output over the years 1922
to 1929 to 18 percent over the years 1930 to 1932 (Farnsworth 1940, p. 63, for
stocks; Federal Reserve Bank of St. Louis for production, https://fred.stlouisfed
.org/series/A010131WA393NNBR, accessed May 23, 2019). Te U.S. stock of cot-
ton quadrupled between 1929 and 1932, rising from 15 percent of production to
75 percent (Bureau of the Census 1975, table K 556). Nineteen thirty-two is the
appropriate reference year, because New Deal programs restricting output kicked
in during 1933.
16. Te same thing will happen if the propensity to save rises, which Peter Temin
(1976) and Frederic Mishkin (1978) emphasize in their accounts of the downturn.
Te emphasis on falling consumption as the driver of the 1930 downturn is sup-
ported by Robert Shiller’s account of the role of the stock market crash of 1929 on
consumer psychology (2017, esp. pp. 988–994).
17. Tis is a form of “immiserizing growth” (Jagdish Bhagwati [1958]); Frederic Pryor
(2007) provides an overview to a literature going back in its modern form to
Harry Johnson (1955). Te novelty of the present argument is twofold. Here
growth is immiserizing all around, not just for the growing economy. Moreover,
the institutional set-up is diferent.
Copyright © 2021. Harvard University Press. All rights reserved.
18. Te AAA was declared unconstitutional by the Supreme Court in 1936. It was re-
vived in 1937 with a work-around to overcome the constitutional objection.
19. Afer a period of fuctuating gold prices and exchange rates, the Gold Reserve Act,
signed into law by Roosevelt on January 30, 1934, authorized the president to fx a
new parity. Te price fxed by Roosevelt the next day, $35.00 per ounce, was 1.69
times the old parity. (It remained in efect until the United States broke the link to
gold altogether in 1971.) Te U.S. Treasury held a large share of the world’s gold
supply and was prepared to buy and sell gold to all comers—provided they were
not domiciled in the United States Americans were prohibited from owning golds
apart from ornamental and industrial uses.
20. Some if not all of the increase in the price of corn was weather related. Production
in 1933 was almost 20 percent below 1932 (Bureau of the Census 1935, table 613).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
820 Notes to Pages 311–323
Chapter 9
Epigraph: Keynes, Te General Teory, p. 96.
1. For reasons that remain mysterious, Samuelson refuses to call this a long-run
consumption function. A footnote ofers this explanation:
Some might choose to interpret [this locus] as a very long-run consumption
function, although I myself would not. Even if regarded as such, the fact that
it does not show an increased percentage of saving as income rises does not
in any way vitiate the application of the usual saving-investment analysis. It
is necessary to emphasize this because in some quarters Prof. Kuznets’s his-
torical fndings are taken as disproving the Hansen–Keynes long-run analy-
sis [namely, a declining ratio of consumption to income]. (1943, p. 36)
2. Keynes difered. Tough he believed that over a longer period of time the demand
for investment would fall below full-employment saving, he viewed the immedi-
ate postwar period as one in which investment demand would outstrip full-
employment saving (“Te Long-Term Problem of Full Employment” and reply
to Josiah Wedgewood, in Keynes 1980, pp. 320–325 and 350–351). To be sure,
Keynes’s focus was on the United Kingdom, and Samuelson’s on the United States,
but it is not clear that this is the source of the diference. Curiously, Samuelson
punts on the question of defcit spending as a means of bridging the investment-
saving gap, referring his readers to another article in the book (by Seymour Har-
ris) on the public debt, while at the same time suggesting that “careful thought
should be given to the alternative of the controlled issuance of noninterest-bearing
debt” (1943, p. 43).
3. Evidently, a poorly paid sports star. No doubt because of gender: this sports star is
assumed to be female.
4. If the rate of interest is positive instead of zero, consumption will optimally in-
crease with time, in line with the rate of interest. Moreover, the consumption
profle will depend on when income is earned as well as on total lifetime earnings.
5. Age forty-two is chosen for convenience because both the representative agent of
Figure 8 and the sports star of Figure 9 have identical incomes in that year. Both
are at the midpoints of their careers, earning approximately $86,000.
Copyright © 2021. Harvard University Press. All rights reserved.
6. See especially Robert Hall (1978), who brought the argument into the rational-
expectations framework.
7. For this reason Keynes was doubtful about the ability of temporary tax cuts to
stimulate consumption. In arguing with James Meade in “Te Maintenance of Full
Employment” (the heading of his memo) about the potential role of countercycli-
cal fscal policy in postwar Britain, Keynes expands the point made in Te General
Teory about the role of habit:
People have established standards of life. Nothing will upset them more than
to be subject to pressure constantly to vary them up and down. A remission
of taxation on which people could only rely for an indefnitely short period
might have very limited efects in stimulating their consumption. (Keynes to
Meade, April 25, 1943, in Keynes 1980, p. 319)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 327–336 821
For Keynes the unresponsiveness of consumption to temporary changes in dis-
posable income was a reason to modify the “fundamental psychological law” of
Te General Teory “that men are disposed, as a rule and on the average, to in-
crease their consumption as their income increases, but not by as much as the in-
crease in their income” (p. 96). For Modigliani and Friedman, this unresponsive-
ness was a reason to jettison Keynes’s framework.
8. In the interwar years of the twentieth century, probability was recast as subjective
in nature rather than as an objective distribution based on empirical frequen-
cies, precisely to overcome the risk-uncertainty distinction. Subjective probability
caught hold from midcentury onward as a means of defending the identifcation
of microeconomic general equilibrium with an ef ficient allocation of resources. It
is another thing altogether to argue that people actually make decisions this way.
9. One percent of the population would be easy to ignore if we were interested only
in the determination of aggregate consumption. However, the other side of the
coin is saving, and this particular 1 percent does the lion’s share of the personal
saving.
10. Saez’s data comes from Internal Revenue Service records, the Census Bureau’s
from household and family surveys. Te data are not strictly comparable, hence
the wide range in the estimates of income shares. Census Bureau data estimate the
2012 threshold of the top 5 percent at just over $190,000 for households, and at
$210,000 for families. Te 2012 share of the top 5 percent, according to Census
Bureau data is just over 20 percent, whether for families or for households, slightly
higher for households than for families (Bureau of the Census 2019a). Tis is the
income share that Saez attributes, on the basis of IRS data, to the top 1 percent of
the distribution.
11. Labor income apart, until the ref era, working-class households found it dif ficult
to tap into their main source of wealth, namely, the equity in their homes, even if
this was what their optimal consumption plan dictated. Exceptionally, during the
housing boom that preceded the crash of 2008—to which mortgage refnancing
was at the very least an important contributing factor—working-class as well as
more affl uent households were able to convert housing equity into cash and spend
it on current consumption.
Copyright © 2021. Harvard University Press. All rights reserved.
12. Assuming that changes in money wages are determined solely by an LS schedule
dictates that infation can occur only when there is excess demand for labor. Te
simplest assumption is that the overfull employment at the infationary equilibria
in Figures 9.18 and 9.19 is achieved through compulsory overtime. An alternative
is to redefne the LS schedule as a stationary money-wage locus along which there
is a positive rate of involuntary unemployment. Te issue of wages and employ-
ment is at the heart of the theory of the long run presented in chapters 17, 18, and
19. (Observe that the possibility of an infationary equilibrium does not require
the complications introduced by the Cambridge saving theory; it exists in the
simpler models of chapters 6 through 8, where saving depends only on income,
not its distribution.)
13. To be sure, working-class saving is no longer identifed with the traditional work-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
822 Notes to Pages 340–350
ing class. Tat class is lef out of the picture altogether, presumably on the grounds
that, though far wealthier than the working class of Adam Smith’s, David Ricar-
do’s, or Karl Marx’s times, it is still without the means to accumulate assets within
the capitalist sector, managing at best gradually to become homeowners. “Work-
ers” in a Piketty-type model are a “patrimonial middle class” (2014, p. 373), a class
of executives, managers, professionals, and bureaucrats reminiscent of John Ken-
neth Galbraith’s “technostructure” (1967). Capitalists, as in the original Cam-
bridge model, are a class of rentiers who rely solely on p rofts for their income.
Chapter 10
Epigraph: Keynes 1937a, p. 221.
1. Keynes well understood the diference, as witness his remark:
An act of individual saving means—so to speak—a decision not to have din-
ner to-day. But it does not necessitate a decision to have dinner or to buy a
pair of boots a week hence or a year hence or to consume any specifed thing
at any specifed date. Tus it depresses the business of preparing to-day’s
dinner without stimulating the business of making ready for some future act
of consumption. (Te General Teory, p. 210)
But he was wrong in thinking that this distinction by itself refuted Say’s Law. See
chapter 2.
2. “Have you heard of the wonderful one-hoss-shay,
Tat was built in such a logical way?
It ran a hundred years to a day, . . .”
(“Te Wonderful ‘One Hoss Shay,’” by Oliver Wendell Holmes, 1858)
3. Alas, the data do not appear to support this hypothesis. Q is not correlated with
interest rates, even afer introducing appropriate controls. Moreover, the data
suggest that, from the mid-1990s on, Tobin’s Q substitutes for monetary policy
rather than jointly determining residential construction. Te visual impression
that make versus buy became important only in the mid-1990s is borne out by
time-series regressions: nominal mortgage rates are highly correlated with resi-
dential construction until 1995, and Q is not. Afer 1990, this situation is reversed:
Copyright © 2021. Harvard University Press. All rights reserved.
mortgage rates no longer appear to matter, whereas Tobin’s Q seems to drive
homebuilding single handedly.
4. Te distinction between capital widening and capital deepening appears to have
originated with Ralph Hawtrey (1937, p. 36):
Te widening of the capital equipment means the extension of productive
capacity by the fotation of new enterprises, or the expansion of existing en-
terprises without any change in the amount of capital employed for each unit
of output. Te deepening means an increase in the amount of capital em-
ployed for each unit of output.
5. I should perhaps make it clear that I posit this simple characterization of invest-
ment for one purpose only: to drive home the diference between capital deepen-
ing and capital widening.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 357–364 823
6. Empirical estimates of the elasticity of substitution, conventionally denoted by ,
generally lie in a range well below one. A compelling theoretical argument for as-
suming < 1 is that the opposite assumption, > 1, leads to the odd result that
only a single input, capital or labor, is required for production.
7. Behavioral economics has repeatedly shown that the maximization postulate is
not only unnecessary but also counterproductive if all we wish to do is to describe
how the economy actually works. And for almost half a century microeconomics
has recognized that the assumption of utility maximization poses virtually no
constraints on the shapes of individual demand functions.
8. Keynes was later persuaded otherwise, apparently by Ramsey.
9. If Tversky had not died, he would, I presume, have shared Kahneman’s Nobel
Prize. Allais also received a Nobel. Ellsberg narrowly escaped prison—not for his
theory, but for leaking the Pentagon Papers.
To be fair, some pioneers of decision theory—the names of Howard Raifa and
Savage himself spring to mind—never regarded consistency with the Savage axi-
oms as innate. Generations of students have paid, and are still paying, good
money to the leading business schools of the world to learn how to apply proba-
bilities consistently. I do not wish even to hint that students d on’t get their mon-
ey’s worth, but I am very skeptical about how much their ability to deal with un-
certainty is enhanced.
For the purest of pure theorists, molded by Milton Friedman’s positivism
(1953), it does not matter whether individuals consciously calculate the subjective-
probability distributions required by the theory. Just as with utility maximization
uncomplicated by probabilities, “as if ” behavior will do just fne.
10. High mainstream theory makes use of subjective probabilities with regard to the
motives and choices underlying saving rather than investment. In fact, the most
sophisticated theory, due to Kenneth Arrow and Gerard Debreu (1954), denies
any role to the investor’s state of mind. Tis theory teams up the states of mind of
individuals as consumers with the idea of complete contingent markets. Decisions
on saving require subjective probability calculations about the likelihood of difer-
ent states of the world. Tese decisions are refected in the purchase of consump-
tion claims contingent on which potential state of the world is actually realized;
Copyright © 2021. Harvard University Press. All rights reserved.
saving thus does not require agents to hold physical capital to back their hunches
about the future. In this model, investment responds to objective prices attached
to goods in diferent states of the world, not to subjective utilities and proba
bilities. Te assumption that these states are objectively describable removes all
vagueness from future-project returns—the state of the world completely and
uniquely determines investment returns—and the prices of contingent claims as-
sociated with particular states of the world guide the frm’s investment decisions.
Although the stock market, as Peter Diamond showed in a seminal paper half a
century ago (1967), can in principle play the role of a universal contingent market,
the number of companies that would be required to span the economically rele-
vant states of the world is beyond the capacity of stock markets to list, fnancial
experts to analyze, or computers to store in memory.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
824 Notes to Pages 364–378
In practice, contingent markets are the exception rather than the rule, material-
izing only under unusual circumstances; otherwise opinions about the future are
backed not by the purchase of contingent commodities but by commitments to
particular forms of capital, by investment. In practice, frms’ investment decisions
are guided not by objective market prices but by prospective investment returns
and associated hunches about the possibilities of realizing these returns, seen
through a glass, darkly.
11. In this, Keynes had an unlikely ally. Well before Keynes extolled animal spirits in
Te General Teory, Joseph Schumpeter published Te Teory of Economic Devel-
opment, in which he wrote,
As military action must be taken in a given strategic position even if all the
data potentially procurable are not available, so also in economic life action
must be taken without working out all the details of what is to be done. Here
the success of everything depends upon intuition, the capacity of seeing
things in a way which aferwards proves to be true, even though it cannot be
established at the moment, and of grasping the essential fact, discarding the
unessential, even though one can give no account of the principles by which
this is done. Torough preparatory work, and special knowledge, breadth of
intellectual understanding, talent for logical analysis, may under certain cir-
cumstances be sources of failure. (1961 [1912], p. 85)
Chapter 11
Epigraphs: Keynes 1934, p. 850; 1937a, pp. 215–216; 1937b, p. 250.
1. Greg Mankiw characterizes the loanable-funds market as the fnancial market of a
(simplifed) economy to which “all savers go . . . to deposit their [current] saving,
and all borrowers go to take out their loans . . . [Current] saving is the source of the
supply of loanable funds” (2018, p. 549–550; italics in original). Loans in this
model are for adding to the real capital stock, that is, for investment: “investment
is the source of the demand for loanable funds” (2018, p. 550; italics in original). For
Mankiw, “saving” is the excess over and above an agent’s own investment, and
“investment” is the excess over and above an agent’s own saving, but it does no
Copyright © 2021. Harvard University Press. All rights reserved.
harm to assume that saving and investment are carried out by diferent agents, so
that the investment on the part of savers is zero, as is saving on the part of agents
who invest, that is, who add to the physical stock of capital.
Paul Krugman and Robin Wells (2018, pp. 706–715) generally agree with Man
kiw’s exposition of loanable funds. But when they discuss interest-rate equaliza-
tion across national fnancial markets (pp. 711–714), they veer over toward the
Keynesian notion of interest rates clearing markets for fnancial assets rather than
fows of saving and investment.
William Baumol, Alan Blinder, and John Solow, wisely in my judgment, avoid
the concept of loanable funds altogether in their text (2020), as do Robert Frank
and Ben Bernanke (2007). At least neither textbook contains an entry for loanable
funds in its index or glossary.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 379–387 825
2. What about combining the two views by bringing saving and investment into the
determination of interest rates in asset markets? Chapter 2 (note 11) observed that
the separation between the fnance literature, which focuses on asset markets
as the site of interest-rate determination, and the macro literature, for which
the saving-investment nexus is the focus, has been breached in this century
(Gürkaynak and Wright 2012, sec. 4.1). A case can be made for bringing macro-
economic variables into agents’ demands for fnancial assets, but this is a far cry
from investment and saving determining interest rates.
3. As we shall see in the next chapter, the calculations become more dif ficult when
there are a variety of fnancial assets available. Young agents might opt for T-bills
if bills pay a premium over the long-dated bonds that dovetail more closely with
their retirement plans.
4. An analogy with the determination of exchange rates may be helpful. Are ex-
change rates determined by the balance of trade or by the balancing of asset port-
folios? Once again, the statistics are orders of magnitude apart. Foreign trade,
measured by exports, imports, or the balance of trade pale in comparison with the
volume of transactions on foreign exchange markets. In 2010, U.S. exports and
imports were of the order of $2 trillion, and the trade defcit of the order of $500
billion, while transactions of U.S. dollars against other currencies on the spot
market alone averaged over $1 trillion per day for April of that year (data from
Bank for International Settlements 2010, table E-1).
5. Consols, short for consolidated annuities, were frst introduced in the mid-eigh
teenth century by the British government to consolidate the many outstanding
debt issues into a single uniform security. Consols remained an important part of
British public debt into the twentieth century. Te last outstanding consols were
redeemed in 2015 (HM Treasury and Te Rt Hon George Osborne 2015).
6. Wells Fargo has since updated its recommendations. Its current website advises
“investors to set aside the equivalent of at least 3 to 6 months of living expenses in
an emergency fund.” It also clarifes the meaning of cash:
By “cash,” we don’t mean just dollar bills or even the money in your checking
account. We use the term as shorthand for a spectrum of assets that have
typically been very stable in value and can usually be liquidated quickly
Copyright © 2021. Harvard University Press. All rights reserved.
when you need to cover an expense. Short-term Treasury bills and money
market funds are classic examples. (2019)
Te latest guidelines make no reference to the 3 percent rule, perhaps because the
bank’s investment-advising section is trying to expand its client base: for 3 percent
of wealth to cover six months of spending requires total wealth to be 16.7 times
annual expenditure, which is true only of high-wealth individuals (and misers).
7. Keynes actually fudges the distinction between speculation and precaution. He
includes holding cash against “unforeseen opportunities of advantageous pur-
chases,” in his characterization of the precautionary motive (Te General Teory,
p. 190), a reason my taxonomy attributes to speculation rather than to precaution,
because it makes the demand for money sensitive to the interest rate.
In the end, the focus on speculation is not necessary to Keynes’s argument. We
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
826 Notes to Pages 388–412
could assimilate the precautionary motive, and the retirement motive as well, to
speculation, and assume all three motives involve agents who are sensitive to a
trade-of between the returns available from tying up their funds in illiquid assets
and the potential for capital losses on these assets.
8. In the Treatise on Money, Keynes confates bid-ask liquidity with price-fuctuation
liquidity. One asset is more liquid than another if it is “more certainly realisable at
short notice without loss” (1930, vol. 2, p. 67). “Short notice” invokes bid-ask li-
quidity, “without loss” price fuctuations.
9. In short, Keynes would have had no reason to deny the relevance of bid-ask li-
quidity, but would regard it as a friction of the kind he was at pains to minimize.
10. U.S. T-bonds and notes pay a semi-annual coupon, but it is simpler to assume the
coupon is paid once a year.
11. Strictly speaking, the “yield to maturity” does not exist for a consol since consols
never mature. Te obvious substitute is the limiting value of the yield to maturity
as the maturity date increases without bound.
12. People who hold wealth against a rainy day or to prepare for retirement face a
similar issue of the choice of an appropriate horizon. For precautionary wealth
holders, the issue is the same as for speculators: they d on’t know when they will
need to tap into their wealth.
Retirement poses a diferent issue if we suppose that people have a reasonably
good idea of when they will retire, and at least probabilistic knowledge of the time
span of their retirement. But in a model with only cash and consols, agents cannot
tailor their portfolios to this knowledge. Consols, unlike bonds chosen with their
maturity dates corresponding to the projected retirement period, involve the same
price risks as they do for speculators. Like speculators, agents whose portfolios are
geared to retirement may be assumed to trade the potential gains of holding con-
sols against the variability of consol prices, but the holding yields would be mea
sured over decades rather than over a year or instantaneously.
13. Again we confne ourselves to consols. For bonds with fnite maturity dates, price
fuctuations will not be centered on zero even without expectations of reversion to
normal. See the empirical appendix to chapter 12.
14. A representative-agent model is one with identical agents—same preferences and
Copyright © 2021. Harvard University Press. All rights reserved.
same endowments. For some purposes, representative-agent models simplify the
analysis greatly without much loss of generality. But here not: as the following
section (“Bulls and Bears”) shows, agent diversity makes the normal-reversion
liquidity-preference schedule slope downward.
15. Two percent may seem a very modest annual risk, but over ten years it amounts to
nearly a 20 percent probability of default. (Te exact risk is 1 − (.98)10 = 0.18 if
default in any one year is independent of previous years.)
Chapter 12
Epigraphs: Robertson 1935, p. 499; Samuelson 1947, pp. 122–123.
1. Why aren’t short-term bank loans included as substitutes for the short-term paper
charted in Figure 12.2? Te answer lies in the form of illiquidity introduced in
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 414–416 827
chapter 11 only to be put to one side, namely, bid-ask illiquidity. Bank loans are
generally too idiosyncratic to command the dense markets necessary to eliminate
this kind of risk. Securitization—see chapter 4—is a way of overcoming at least
some of the idiosyncrasies of individual bank loans, but it is an imperfect way.
2. Jacob Viner (1936), in the critical passage on liquidity preference cited in chapter
11, likely had such agents in mind when he argued that holding long bonds
through thick and thin would be the preferred strategy in the face of fuctuating
yields. Franco Modigliani and Richard Sutch (1966) distinguished agents accord-
ing to “preferred habitats,” their way of characterizing the time horizons of difer-
ent agents. Joseph Stiglitz (1970) developed a theoretical framework for analyzing
portfolio management in the context of choosing an optimal consumption profle.
In this framework, risk aversion can work both ways, leading to a premium on
short-dated assets or on long-dated assets. John Campbell, working with various
collaborators, revisited the problem of time horizon; see Campbell and Luis Vi-
ceira (2002).
3. If your infation expectations exceed the expectations refected in current bond
yields, bills may be a better bet than bonds even if you have a long horizon.
4. Winston W. Chang, Daniel Hamberg, and Junichi Hirata (1983) hold the same
view as I do with regard to the short-term options available to agents in choosing
asset portfolios. Chang et al. demonstrate that with short-term riskless bills yield-
ing a positive return, optimization precludes holding money. Teir argument is
limited to deriving asset demands as functions of interest rates; they do not inves-
tigate the properties of asset-market equilibrium, specifcally the property that
equilibrium determines the spread between bond and bill rates, not their levels. I
am grateful to Korkut Alp Erturk for this reference.
Subsequently, Ben Friedman referred me to earlier work by Albert Ando and
Karl Shell (1975). Ando and Shell recognize that, from the point of view of a store
of value, short-term bills dominate cash, but like Hicks (see chapter 13) they as-
sume that at the margin this dominance is ofset by the convenience/cost advan-
tages of holding cash, given that as a medium of exchange cash dominates any
other asset.
5. Until we take up the important monetary-policy response to the Great Recession,
Copyright © 2021. Harvard University Press. All rights reserved.
namely, quantitative easing.
6. Richard Sutch (2018) is exceptional in distinguishing the two defnitions of the
liquidity trap. I am persuaded by Sutch that confusion between these two defni
tions is responsible for the puzzlement surrounding Keynes’s assertion that a li-
quidity trap—not Keynes’s terminology—is of more theoretical than practical in-
terest:
Whilst this limiting case might become practically important in future, I
know of no example of it hitherto. (Te General Teory, p. 207)
Since Keynes does not use the terminology of a liquidity trap, it is not precisely
clear what he means by “this limiting case.” But the context suggests, as Sutch ar-
gues, that he is referring to a foor to the hurdle rate of interest, not to the bill rate
itself approaching zero. Readers who think of a liquidity trap in terms of the be-
havior of bond yields when the bill rate is at the zlb would necessarily be puzzled:
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
828 Notes to Pages 417–455
the short-term (U.S.) T-bill rate had reached the zlb in 1932, and except for a brief
period in 1933 had stayed near zero while Te General Teory was in preparation
(and for many years thereafer).
7. Robertson’s letter included a number of criticisms of Te General Teory draf to
which Keynes did respond, but he overlooked this one. To the best of my knowl-
edge, Keynes never did counter Robertson on this point. Robertson went on to
present other criticisms of the liquidity-preference theory of interest (for example,
1937), but none was as cogent and on the mark. Never acknowledged by Keynes
and apparently forgotten by Robertson, Robertson’s 1935 criticism seems to have
surfaced only when Keynes’s Collected Works was published in 1980.
8. Agents’ time horizons are crucial here. It has been noted that in a world of long-
horizon risk aversion, the short-term bill is more risky than the long-term bond.
If long horizons dominated, we would expect an inverted yield curve to be the
rule rather than the exception. If we assume away risk aversion, agents’ time hori-
zons no longer matter. If portfolio choice is grounded only in normal reversion,
the issue becomes trading of the loss of principal against interest income, and the
time horizon is irrelevant: long-horizon agents will make the same calculations as
short-horizon agents.
9. It is not only private issuers of debt who may default but any issuer not in control
of the currency in which the debt is denominated. U.S. states and municipalities,
not to mention the otherwise sovereign countries that make up the Eurozone, are
all subject to default risk. Te contrast is with dollar bonds issued by the U.S.
Treasury, or for that matter yen bonds issued by the Japanese government, or ster-
ling bonds issued by the United Kingdom.
A disclaimer is in order. In the summer of 2011, in order to extract concessions
from President Obama, the Republican controlled House of Representatives de-
layed extension of the debt limit to the last minute, arousing fears of a default. Te
assumption that default is precluded if debt is issued in a currency controlled by
the issuer needs to be qualifed to exclude governments divided against them-
selves with at least one party playing with fre.
10. Tis diference varied between 1.2 and 2.8 percent, averaging just under 1.7 per-
cent. I am here measuring default risk by the diference between yields to maturity
Copyright © 2021. Harvard University Press. All rights reserved.
on 30-year U.S. T-bonds and Moody’s index of long-term Baa-rated corporate
bonds. Tis index is based on seasoned bonds with remaining maturities of at
least twenty years with maturities above twenty years (Federal Reserve Bank of St.
Louis 2020). When I checked with Moody’s Analytics in 2013 (personal commu-
nication, November 15, 2013), the average maturity of the index was twenty-eight
years.
Chapter 12 Mathematical Appendix
1. Tis is to say that agents have less risk aversion the more their portfolios are com-
mitted to money, with expected wealth decreasing by just enough to ofset the re-
duction in risk while maintaining the same level of utility.
2. Tis is a simplifcation of course. A more general formulation would be to treat all
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 457–460 829
bonds and bills as partial substitutes in a general-equilibrium framework. But the
relatively small size of the corporate bond market justifes a sequential framework
in which the Treasury yield is determined frst. In 2018, the average volume of
daily trading in Treasuries ($547.8 billion) was seventeen times the volume of
corporate bond trading ($31.5 billion) (Securities Industry and Financial Markets
Association 2020).
Chapter 12 Empirical Appendix
1. Campbell’s negative conclusions themselves were hardly novel. A quarter century
before Campbell, Ed Kane (1970) wrote
It is generally agreed that, ceteris paribus, the fertility of a feld is roughly
proportional to the quantity of manure that has been dumped upon it in the
recent past. By this standard, the term structure of interest rates has become
. . . an extraordinarily fertile feld indeed. (p. 361)
In 1989 Kenneth Froot ofered pretty much the same assessment, although his
language was considerably less colorful:
If the attractiveness of an economic hypothesis is measured by the number
of papers which statistically reject it, the expectations theory of the term
structure is a knockout. (p. 283)
Nonetheless, the rigor and comprehensiveness of Campbell’s analysis make his
summary the obvious starting point for assessing the data.
2. It was hardly a shotgun wedding. One of the very frst applications of rational ex-
pectations was to the relationship of long-and short-term rates (Sargent 1972).
Te ft between theorizing interest-rate forecasts and rational expectations must
have appeared irresistible, love at frst sight.
3. Some T-bonds issued in the nineteenth century were called consols, but these
bonds were callable (redeemable at the discretion of the Treasury) and so were not
really perpetual bonds. Some private railroad bonds issued in the nineteenth and
twentieth centuries had maturity dates so far in the future that they might as well
have been perpetuities. Te West Shore Railroad, whose track was leased for 475
years by the New York Central in the nineteenth century (with an option to renew
Copyright © 2021. Harvard University Press. All rights reserved.
for another 500 years), issued bonds that were coterminous with the expiration of
the initial lease in 2361. Te Canadian Pacifc Railway also issued perpetual de-
bentures in the nineteenth century. According to the Toronto Globe and Mail,
C$31 million were still outstanding in 2011 (Mittelstaedt 2012).
Today the closest fnancial asset to consols is the preferred stock of public com-
panies. Preferred stock is actually a conditional consol, for which the specifed
coupon payment can be omitted (or postponed) under certain circumstances.
4. In a world of consols, there is, strictly speaking, no yield to maturity because a
consol never matures; hence the defnition as a limit. In any case, we have no need
for this concept when dealing with consols: the limiting value of the yield to ma
turity is equal to the coupon yield. In a world of fnite-maturity bonds, the yield to
maturity plays a separate and distinct role.
5. Te analogous condition in the case of consols is a so-called transversality condi-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
830 Notes to Pages 462–472
tion, an externally imposed limiting value for the price of the consol. Without
such a transversality condition, there is nothing to prevent an infnitely long price
bubble from equating the holding yield on a consol to any given value of the
short-term interest rate, which is the optimality condition without a risk pre-
mium. For example, suppose the short-term interest rate is fxed at 5 percent. If a
consol with a $5 coupon were priced at $200, giving it a coupon yield of 2.5 per-
cent, it can ofer a holding yield of 5 percent if it increases in value to $205. In
subsequent years, the capital gain would have to be progressively greater than $5
for the holding yield to remain at 5 percent, but there is nothing to prevent this
scenario other than the assumption that bubbles eventually pop. A transversality
condition in efect rules out infnite bubbles.
6. It is of course possible that the expected price change is zero over time, but this re-
quires a fortuitous combination of the rates of change of s and (m). For E(P )/P
to equal 0, we must have
R
= ρ s + α(m) = const. ,
P
which is to say that s must increase over time at exactly the same rate that (m)
is decreasing.
7. I say “might reveal itself ” because I have been unable to prove Hicks’s conjecture
for coupon bonds, though it is clearly true for zero-coupon bonds, about which
more below.
8. By separating the interest payments from the repayment of interest, bond dealers
have created interest-only and principal-repayment (zero-coupon) bonds in de-
rivatives markets. Tese derivatives are called STRIPs (for Separate Trading of
Registered Interest and Principal of Securities).
9. “Hats” in general denote forecasts, as distinct from actual, unhatted, values.
10. Te consistent-forecasting condition obtained by diferentiating the defnite inte-
gral is diferent from the corresponding derivative of the indefnite integral. Te
upper limit of integration introduces the term ˆ s(t + m). A similar change charac-
terizes the forward-rate condition (below).
11. Te forward-rate condition contains no new information since it can be derived
Copyright © 2021. Harvard University Press. All rights reserved.
from the holding-yield condition and the forecasting-consistency condition. Te
forward rate is conceptually diferent from the holding yield even though both
refect short-period returns from long bonds. Te holding yield refects the return
to a commitment today with respect to the agent’s portfolio over the next years,
whereas the forward rate refects a commitment m years hence to hold a bond
maturing a further years in the future. If the forecast ˆ s(t + m) turns out to be
accurate, today’s forward rate for a point in time m years hence is equal to the
holding yield on an (m + )-year bond at that future time.
12. When the forward rate is equal to zero, the yield curve itself tells us something
about market expectations with respect to short-term rates. In this case equilib-
rium requires s(t + m) + (m) to be equal to zero. With both s(t + m) and (m)
constrained to be non-negative, this in turn implies s(t + m) = (m) = 0.
13. Later in this appendix, in the section titled “How Does Infation Bear on Interest-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 475–486 831
Rate Forecasts,” we examine in more detail the nexus between the rate of infation
and interest-rate forecasting.
Given the attraction of the Taylor rule (the original statement is John Taylor
[1993]), which emphasizes a trade-of between infation and unemployment in
setting the Fed funds rate, another obvious instrument for the bill rate is the un-
employment rate. It makes little diference to the results to include this variable as
an additional instrument. In fact I do so by default in later regressions, when un-
employment is introduced as an explanatory variable for the risk premium .
14. Tough her concerns are very diferent, Yacine Ait-Sahalia’s conclusions (1996)
about the actual behavior of the seven-day Eurodollar deposit rate during the pe-
riod 1973 to 1995 lead to similar results:
Te nonlinear mean-reverting drif pulls the rate back strongly into this
middle region whenever it wanders outside (for example, below 4 percent in
1993 and early 1994 or above 17 percent in 1981 to 1982). Tis makes the
process globally stationary . . . (p. 406)
Loosely speaking, markets may become more nervous outside the central
region, at both ends of the interest rate spectrum. Market participants may
expect the Federal Reserve to credibly return the short-term interest rate to
its middle range at some point, but are uncertain about the precise timing of
the intervention. Tis rather speculative interpretation would also be consis-
tent with the strong pull-back drif . . . at either end of the spectrum. (p. 407)
I am indebted to John Campbell for this reference.
15. Table 12.1 indicates the estimate of implicit in the value of a1 is relatively stable
with respect to the bond term, except for the shortest maturity of one year. When
the bond maturity is between two and ten years, varies between 0.0066 and
0.0079, indicating an expectation that the gap between the normal short rate and
the actual rate will close by 8 to 10 percent per year.
16. Bonds with maturities longer than ten years were issued in the 1970s, but the data
for the full spectrum of maturities up to thirty years is available over from the
1980s.
17. UNRATE is the deviation from the mean unemployment rate (6.43 percent) and
AVGMAT is the deviation from the mean average maturity (56.2 months) over
Copyright © 2021. Harvard University Press. All rights reserved.
the 497 observations.
18. Alan Greenspan, testifying before Congress in February 2005, referred to the fail-
ure of the long-bond yields to respond to the increases in short rates during 2004
as a conundrum:
In this environment, long-term interest rates have trended lower in recent
months even as the Federal Reserve has raised the level of the target federal
funds rate by 150 basis points. Tis development contrasts with most experi-
ence, which suggests that . . . increasing short-term interest rates are nor-
mally accompanied by a rise in longer-term yields. For the moment, the
broadly unanticipated behavior of world bond markets remains a conun-
drum. (Federal Reserve Board 2005)
19. Tere is also disagreement about what that normal rate might be. In Europe, with
the yield on the German 10-year T-bond near zero (April 2015), there was evi-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
832 Notes to Pages 486–500
dently a widespread belief that the new normal was signifcantly lower than his-
torical experience would suggest. In mid-2014 the economist-journalist Anatole
Kaletsky posted this question on his blog:
What accounts for the rock-bottom levels not only of the overnight interest
rates that central banks set directly, but also the long-term rates that depend
on the willingness of pension funds, insurers and private investors to tie up
their savings for 10 years or more in government bonds? (Kaletsky 2014)
In answering his question, Kaletsky ofered as one possibility that the normal rate
has fallen:
If investors were absolutely confdent that short-term rates set by the central
banks would remain near zero for many years ahead, then the seemingly
paltry returns—varying from 2.6 percent down to 0.6 percent—on 10-year
bonds issued by the U.S., European and Japanese governments would seem
generous. Rational investors would be happy to lock up their money for a
decade at these rates.
In other words, reversion to normal is alive and well, but the normal rate itself is
not what it used to be.
20. Estimates for 2017 were uniformly higher than what turned out to be the case.
Te actual Fed Funds rate at the end of 2017 was just below 1.5 percent.
21. Tese data were made available by Moody’s Analytics by special agreement, and
the results using these data are reproduced here by permission (Moody’s Agree-
ment No. 00043372.0).
22. CORPFRAC is measured by the ratio of two variables extracted from Flow-of-
Funds data “Financial Accounts of the United States -Z.1,” https://www.federal
reserve.gov/releases/z1/20180607/html/default.htm#levels (accessed August 17,
2018). Te numerator is Table L.213 Corporate and Foreign Bonds, line 1, Total
liabilities. Te denominator is the diference between total long-term debt and the
long-term debt held by the Federal Reserve:
+ Table L.210 Treasury Securities, line 4, Other Treasury notes, bonds, and TIPS
− Table L.109 Monetary Authority, line 1, Other Treasury securities
+ Table L.211 Agency-and GSE-Backed Securities, line 1, Total liabilities
− Table L.109 Monetary Authority, line 1, Agency-and GSE-backed securities
Copyright © 2021. Harvard University Press. All rights reserved.
+ Table L.212 Municipal Securities, line 2, State and local govt. liabilities
− Table L.212 Municipal Securities, line 3, Short term
+ Table L.213 Corporate and Foreign Bonds, line 1, Total liabilities
23. Refet Gürkaynak, Brian Sack, and Jonathan Wright re-estimated this series in
2006, and it is now updated on a continuous basis (https://www.quandl.com/data/
FED/SVENY, accessed September 10, 2014).
Chapter 13
Epigraph: Keynes, Te General Teory, p. 183—the wild duck was borrowed from
Henrik Ibsen’s play.
1. Predictable needs for cash—a new baby, children going to college, retirement—
are another matter.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 500–506 833
2. Tere is actually a third motive for holding money in Te General Teory: the
precautionary demand deals with the contingencies of ordinary life, like the ones
mentioned above. As I noted in chapter 11, Keynes exhibits some ambivalence
about where to place the precautionary demand relative to the transactions and
speculative demands. In the end, the precautionary motive is assimilated to the
transactions motive and plays no distinct role in the analysis of liquidity prefer-
ence.
3. Does the money for transactions belong to the agents engaged in production, or is
it borrowed from others, either in the commercial-paper market or in the form of
bank loans? With the appropriate simplifying assumptions, it d oesn’t matter. Te
key assumption is that transactions money is unavailable to satisfy the speculative
demand for money. Te simplest story is that wealth holders and businessmen
and -women are the same people, but they keep their asset money in one pocket
and their transactions money in the other. Money deployed for transactions does
not enter into wealth portfolios, and vice versa. Money shifs from one pocket to
the other according to the needs of trade.
4. Keynes suspected, but could hardly articulate, much less demonstrate, that the
virtues of price and wage fexibility might vanish completely if the framework of
analysis were a dynamic one. Focusing on real-time change, as the models of
chapters 6 and 7 do, rather than the static world of virtual change that was then
(and continues to be) the playground of economic theory, in turn up-ends Modi-
gliani.
5. Tere is a connection between central banking and fractional-reserve banking,
but it is not a logical one. On the one hand, a central bank does not presuppose a
fractional-reserve banking system—100 percent reserves are a logical possibility,
but in practice the set of 100 percent reserve banking systems is an empty one. On
the other hand, fractional-reserve banking does not logically require a central
bank, nor does it in practice. Indeed, there was fractional-reserve banking long
before central banks appeared in Europe or the United States. Neither, in princi-
ple, does fractional-reserve banking require the visible hand of the state.
6. One hundred years ago, economists were debating the impact on the quantity
theory of the transition from cash money to bank-deposit money, a transition that
Copyright © 2021. Harvard University Press. All rights reserved.
was still under way in the United States despite the huge steps taken in this direc-
tion during the nineteenth century. At its annual meeting in 1911, the American
Economics Association held a symposium on the causes of the changes in the
price level experienced in the previous decade and a half. Diferences of opinion
hinged on assumptions about bank behavior with respect to reserves.
Irving Fisher (1911, p. 38) emphasized the standard interpretation of the quan-
tity theory, in which causality runs from the quantity of money, exogenously
given, to the price level. J. Laurence Laughlin, a critic of the quantity theory, ar-
gued that causality actually ran in the opposite direction:
When the price is fxed, the credit medium by which the commodity is
passed from seller to buyer comes easily and naturally into existence . . . Tat
is, the quantity of the actual media of exchange thus brought into use is a
result and not a cause of the price-making process. (1911, pp. 29–30)
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
834 Notes to Pages 509–511
Implicitly, Laughlin was making the case for endogenous money, by assuming
that banks would vary reserve ratios according to the transactions demand for
money rather than maintaining constant (fully loaned-up) ratios by moving in
and out of speculative assets. Edwin Kemmerer, a third contributor, explicitly rec-
ognized the complication that fractional-reserve banking introduced into the
quantity theory—only to dismiss it. Gold formed the basis of bank reserves and
reserves were, he argued, exogenous. Proft maximization could be counted upon
to ensure that reserves were always fully utilized:
Banks do not make interest on money held in reserves, and accordingly take
measures to invest such surplus money, keeping these reserves as low as is
consistent with law and their ideas of safety. (1911, p. 56)
Consequently, though banks theoretically can vary reserve ratios, self interest
would prevent them from doing so.
Te Great Depression settled the question empirically. As was noted in chapter
8, excess reserves grew markedly in 1932 and 1933; in mid-1933 Jacob Viner saw
these excess reserves as the result of a self-interested prudence on the part of
bankers:
In the past three years the test of a successful banker has been the rate of
speed with which he could go out of the banking business and into the
safety-deposit business. Tose bankers have survived who have succeeded in
the largest degree and at the most rapid rate in converting loans into cash.
Tat has been good banking from the point of view of the individual banker,
or of his individual depositors; but from the social point of view it has been
disastrous. Which is preferable during a depression—a bank that continues
to fnance business and thus endangers its solvency, or a bank that acts on
the principle that during an acute depression good banking means no bank-
ing? Te latter have survived the crisis and now have the confdence of the
public. Tey should now be able to serve efectively in taking care of the
present needs of business if they are willing to return to the banking busi-
ness. (1933b, p. 130)
7. Fractional-reserve banking has the potential to segment the demand for money
into two noncompeting branches, but it doesn’t necessarily do so. Te intent of at
Copyright © 2021. Harvard University Press. All rights reserved.
least some of the framers of the Federal Reserve seems to have been to institution-
alize this separation: through the discount window, the Federal Reserve Banks
were to provide an “elastic currency” that would accommodate the needs of
trade—transactions demand—hermetically sealed of from the speculative de-
mands that operate in asset markets.
8. Te Baumol–Tobin theory is consistent with a large amount of data supporting
the idea that the demand for transactions money is sensitive to the interest rate,
even if the elasticity of transactions demand turns out to be very low. My own re-
gressions for the period 1959 to 2016, in which a narrow defnition of money sup-
ply is deployed as a proxy for transactions demand, give similar results to those
reported by Stephen Goldfeld and Daniel Sichel (1990), among others, for data
covering shorter periods of time. Goldfeld and Sichel regressed the real money
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Page 513 835
supply on real GDP, infation, and interest rates on time deposits and commercial
paper, whereas I use the nominal 3-month T-bill rate along with the frst two vari-
ables, but the diference in regressors has little impact on the empirical relation-
ship between transactions-money demand and the short-term rate. Observe that
for money supply to be a plausible proxy for money demand, it must be assumed
that supply adjusts to demand, in other words, that the money supply is created
endogenously.
Both my results and the Goldfeld–Sichel results point to considerable inconsis-
tency in the impact of GDP and infation over time on transactions demand. Most
striking is the fall in the estimated interest elasticity of money demand from ap-
proximately 20 percent in the decade of monetary turbulence lasting from the
mid-1970s to the mid-1980s, to approximately 1 percent in the decade following
the fnancial turmoil of 2008. I would attribute this fall to the decline in the cost
and inconvenience of moving between safe, interest-bearing, short-term assets
and cash once we entered the age of electronic banking (the rise of money-market
funds, daily sweeps from interest-bearing accounts to demand deposits, and the
like).
9. In normal times high-quality commercial paper and the money-market funds
that own this paper are also safe and liquid. A hallmark of fnancial crisis is that
high-quality commercial paper loses its luster. One leading money-market fund,
the Reserve Primary Fund, held close to $800 million of Lehman Brothers’ com-
mercial paper when the crash came. Immediately afer Lehman’s fall and the
write-down of this paper (approximately 1 percent of the fund’s assets), the Re-
serve Primary Fund “broke the buck”; it was forced to revalue its shares, normally
priced at $1.00, at $0.97. A run on money-market funds followed, customers
withdrawing some $350 billion over the next few days (Ball 2018, p. 44). Te con-
tagion was contained by a U.S. Treasury guarantee of money-market share prices
put in place at the end of September 2008 (U.S. Department of the Treasury 2008)
and the establishment of the Commercial Paper Funding Facility by the Federal
Reserve in early October. At its peak in January 2009, the CPFF held some $350
billion of commercial paper, 20 percent of the total outstanding of $1.75 trillion,
this total itself down from a peak of $2.2 trillion in August 2007 (Adrian, Kim-
Copyright © 2021. Harvard University Press. All rights reserved.
brough, and Marchioni 2011, charts 1, 6, and 7). Te facility was wound down
over 2009 and closed in early 2010.
Laurence Ball believes that the fall of Lehman Brothers could have been
avoided. In his view the action, or rather, inaction of the Fed in allowing Lehman
to go under resulted from the combination of Treasury Secretary Henry Paulson’s
desire to teach the markets a lesson and Fed Chair Bernanke’s deference to the let-
ter of the law and, more importantly, to Paulson. Coupled with the ignorance of
who held how much Lehman paper, the judgment of the Treasury and Fed princi-
pals that the damage of a Lehman bankruptcy would be contained was—as the
previous paragraph indicates—wildly of base.
Ball’s account contrasts with the of ficial story, or rather the story that offi
cials later told to justify their actions (Bernanke 2015a; Geithner 2014; Paulson
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
836 Notes to Pages 514–515
2010): Lehman was insolvent as well as illiquid, leaving the government no choice
when various attempts to arrange a takeover fell through. Ball goes through a de-
tailed accounting of Lehman assets and liabilities to bolster his claim, but in the
end he is unconvincing since so much of Lehman’s assets took the form of securi-
ties for which there was efectively no market, so no way to establish market val-
ues.
Tere is a somewhat sofer, but more convincing, version of Ball’s claim, namely,
that with nobody in a position to know the value of Lehman assets, it was impos-
sible to say whether or not Lehman was solvent when it faced its terminal liquidity
crisis. If this was the case, then it would have been possible to defend bailing
Lehman out, and Ball’s emphasis on the predilections of Paulson and Bernanke
makes sense as the real reason for allowing Lehman to fail.
It is dif
ficult to avoid the echo of Milton Friedman and Anna Schwartz (1963a),
who see the Fed as the villain in the 1930s (see chapter 8, appendix). But in both
cases, the Fed was an accessory rather than the perpetrator.
10. Of course, at times anchoring the spectrum, even at the zlb, fails to generate suf fi
cient investment demand to make for a healthy economy. Having reached the zlb
soon afer the fnancial crash of 2008, the Federal Reserve had no room for ma-
neuver with short-term rates. Te Fed (and belatedly the European Central Bank)
followed Keynes’s advice in Te General Teory:
Perhaps a complex ofer by the central bank to buy and sell at stated prices
gilt-edged bonds of all maturities, in place of the single bank rate for short-
term bills, is the most important practical improvement which can be made
in the technique of monetary management. (p. 206)
Keynes’s “complex ofer” became actualized as quantitative easing (QE), in the
event a very simple ofer to purchase T-bonds and mortgage-backed securities is-
sued by government-sponsored entities like Fannie Mae. Liquidity-preference
theory predicts that, by increasing the supply of short-term debt relative to the
supply of long-term debt in the hands of the public, the spread between the short
and the long end of the spectrum will narrow, thereby reducing the hurdle rate.
(See above, pp. 426–429.) QE seems to have done so, but it is questionable whether
this contributed very much to stimulating investment demand. Remember, be-
Copyright © 2021. Harvard University Press. All rights reserved.
sides the level of the hurdle rate, the elasticity of investment demand is critical.
Once the recovery from 2008 got underway, U.S. corporations sat on unprece-
dented piles of cash. Te efective limit to investment demand would appear to be
opportunity rather than the hurdle rate of interest.
11. Friedman and Kuttner explain the week-to-week inelasticity of demand by the
intricacies of the lags in calculating reserve requirements. But, as they recognize,
this explanation has no force for a longer period. Beyond a very short period, the
relatively low elasticity of demand for reserves is better explained by the low inter-
est elasticity of transactions demand (see note 8). If banks passively respond to
borrowers who meet their credit standards, and if loan demand changes little
when the interest rate changes, then reserves will not vary much in response to
interest-rate changes.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 515–531 837
12. In principle, the Fed announces a “corridor” rather than a single rate, with the rate
paid on reserves as the foor, and the rate charged at the discount window as the
ceiling. But given the large volume of excess reserves, the ceiling has been largely
irrelevant. With a foor and a ceiling, the only impediment to controlling the
Fed Funds rate is self-inficted, namely, the Fed’s acceptance of deposits from
government- sponsored enterprises like the mortgage giants Fannie Mae and
Freddie Mac. Te Fed does not pay interest on these deposits, which gives the
GSE’s an incentive to undercut the interest rate paid by the Fed.
13. At the end of 2018, Swiss 10-year government bond yields were still in negative
territory. Te yields in question are yields to maturity, not coupon yields. See
chapter 11 for an explanation of the diference, which is important for fnite-
maturity bonds but plays no role in a world of consols.
14. Tis supposes that transactions demand is given by the needs of business for
short-term fnance. It will not change things materially if banks also accommo-
date the needs of households for deposit accounts, or if banks also make longer
term loans for investment in equipment, structures, and other long-term projects.
15. Maybe he did: “Any level of interest which is accepted with suf ficient conviction as
likely to be durable will be durable.” (Te General Teory, p. 203)
Chapter 14
Epigraphs: Viner 1933b, p. 129; Lerner 1944, p. 302.
1. Keynes’s argument will surprise those who misunderstand Keynes as believing
that the interest rate received by savers has no efect on how much is saved, for it
is precisely this dependence that Keynes invokes in arguing that fscal policy—in
the form of taxation—will reduce saving and increase consumption. Keynes’s ar-
gument is that taxation of capital income will reduce the incentive to save because
it reduces the afer-tax return (Te General Teory, pp. 94–95, and chap. 24, pas-
sim).
2. Viner:
If the government were to employ men to dig ditches and fll them up again,
there would be nothing to show aferwards. But, nevertheless, even these
Copyright © 2021. Harvard University Press. All rights reserved.
expenditures would be an indirect contribution to business recovery. Teir
major importance would not be in the public works or the unemployment
relief which immediately resulted, but in the possibility of hope that a sub-
stantial expenditure would act as a priming of the business pump, would
encourage business men by increased sales, make them more optimistic,
lead them to increase the number of their employees, and so on. (1933b,
p. 130)
3. Lerner and others gave accounts of an interchange with Keynes that suggest that
not only did Keynes fnd protracted defcit spending problematic from a practical
point of view, but he also failed to grasp the logic of continual defcits. According
to Lerner’s own account, he was present at a lecture Keynes gave during one of his
wartime visits to Washington. Te subject was the ability of the U.S. economy to
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
838 Notes to Page 531
sustain full employment afer the end of hostilities and the end of the enormous
military expenditure involved in the war efort (see chapter 9). Here is how Lerner
recalled the exchange some years later:
At a lecture to the Federal Reserve in Washington in 1944, [Keynes] showed
concern that there might be “too much saving” afer the war. When I pointed
out that the government could always induce enough spending by incurring
defcits to increase incomes, he at frst objected that this would only cause
‘even more saving’ and then denounced as “humbug” my suggestion that the
defcits required to induce enough total spending could always be fnanced
by increasing the national debt. (I must add here that Evsey Domar, at my
side, whispered: “He ought to read the General Teory” and that a month
later Keynes withdrew his denunciation.) (Lerner 1978, p. 67)
Lerner’s interpretation of Keynes’s response suggests that Keynes was not en-
tirely successful in his “long struggle of escape . . . from habitual modes of thought
and expression.” (Te General Teory, preface, p. viii). In this view Keynes came
around to the idea of a long-run budget defcit only afer he read and digested
Te Economics of Control, in which Lerner lays out the theory of functional f
nance.
It is certainly true that Keynes wrote Lerner an efusive letter, worth quoting at
length:
At sea, Septr 1944
My dear Lerner,
Your book [Economics of Control] arrived in London whilst I was away at
Bretton Woods. But now again I am on the sea for yet another visit to the
USA, and the sea voyage has given me an opportunity to read it.
It is a grand book worthy of one’s hopes of you. A most powerful piece of
well organized analysis with high aesthetic qualities, though written more
perhaps than you see yourself for the cognoscenti in the temple and not for
those at the gate. Anyhow I prefer it for intellectual enjoyment to any recent
attempts in this vein . . .
In the second of the two books which you have placed within one cover, I
have marked with particular satisfaction and proft three pairs of chapters—
Copyright © 2021. Harvard University Press. All rights reserved.
chap 20 and 21, chap 24 and 25, and chap 28 and 29. Here is the kernel of
yourself. It is very original and grand stuf. I shall have to try when I get back
to hold a seminar for the heads of the Treasury on Functional Finance. It will
be very hard going—I think I shall ask them to let me hold a seminar of their
sons instead, agreeing beforehand that, if I can convince the boys they will
take it from me that it is so . . .
Yours ever
Keynes
(quoted in Colander and Landreth 1996, pp. 116–117)
But there is considerable evidence for the view that Keynes’s skepticism about the
feasibility of continued defcits was motivated not by a failure to understand the
theory, but by practical considerations, including the dif ficulty of persuading
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 534–538 839
people that a government’s budget and debt are fundamentally diferent from the
budget and debt of a business or of an individual. As David Colander (1984)
notes, Keynes had much earlier, certainly before the seminar encounter with Ler
ner—the actual date of which is disputed—written to James Meade while both
were occupied with war fnance and postwar planning at the UK Treasury:
I recently read an interesting article by Lerner on defcit budgeting, in which
he shows that, in fact, this does not mean an infnite increase in the national
debt, since in course of time the interest on the previous debt takes the place
of the new debt which would otherwise be required. (He, of course, is think-
ing of a chronic defciency of purchasing power rather than an intermittent
one.) His argument is impeccable. [Not quite: see chapter 16, pp. 635–636.]
But, heaven help anyone who tries to put it across to the plain man at this
stage of the evolution of our ideas. (April 25, 1943, in Keynes 1980, p. 320;
quoted in Collander 1984, pp. 1573–1574)
Robert Skidelsky (2000, p. 276) supports Colander’s view, quoting Keynes’s 1943
assessment that Lerner’s conception of functional fnance
runs directly contrary to men’s natural instincts . . . about what is sensible . . .
spoiling a splendid idea by pretending it can be crudely put into force, and
then refusing to look in the face of all practical dif ficulties.
4. As will be the case in general with fxprice adjustment. At a full-employment
equilibrium, the economy is on the AD schedule in a fexprice regime as well.
5. In principle, there will also be diferences in the propensities to consume in suc-
cessive rounds of spending, on the one hand, and retrenchment, on the other. But
in practice, the best the analyst can hope for is to be able to identify the frst-
round players, the taxpayers who are directly hit by the insistence on budget bal-
ance and the suppliers of goods and services to the government. She is not likely
to do better than an economy-wide average for successive multiplier rounds.
6. To be sure, in these simple examples the only leakages from spending are taxes
and saving. Tis simplifcation leads to an overestimation of the multiplier. See
chapter 15.
7. Formulating monetary policy is not as easy as it sounds. Even afer Lehman
Brothers went under, the FOMC was still debating whether infation or recession
Copyright © 2021. Harvard University Press. All rights reserved.
was the more serious danger. Janet Yellen, then the president of the San Francisco
Fed, and, even more so, Eric Rosengren of Boston, stand out for their recognition
of the looming danger of a recession. Favoring a signal in this direction, Rosen-
gren said,
Te failure of a major investment bank, the forced merger of another, the
largest thrif and insurer teetering, and the failure of Freddie and Fannie are
likely to have a signifcant impact on the real economy. Individuals and frms
will become risk averse, with reluctance to consume or to invest. Even if
frms were inclined to invest, credit spreads are rising, and the cost and avail-
ability of fnancing is becoming more dif ficult. Many securitization vehicles
are frozen. Te degree of fnancial distress has risen markedly. Deleveraging
is likely to occur with a vengeance as frms seek to survive this period of sig
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
840 Notes to Page 542
nifcant upheaval. Given that many borrowers will face higher interest rates
as a result of fnancial problems, we can help ofset this additional drag by
reducing the federal funds rate. I support [reducing] the fed funds rate 25
basis points. (Federal Open Market Committee 2008a, pp. 30–31)
In the event, the FOMC voted unanimously to do nothing (Rosengren was a non-
voting member) but vowed to watch the economy closely.
Te problem lay both in recognizing the new facts on the ground and in devis-
ing new instruments to deal with new problems, such as the freezing of the
commercial-paper market (see chapter 13, note 9). Financial conditions contin-
ued to deteriorate afer the September FOMC meeting, and three weeks later, in a
highly irregular between-meetings conference call, the FOMC unanimously en-
dorsed a joint action by the world’s leading central banks to cut interest rates, in-
cluding a 50 basis point (one-half percentage point) cut in the Federal Funds rate.
Chairman Bernanke summarized the state of play:
On the economic growth side, what is particularly worrisome to me is that,
before this latest upsurge in fnancial stress, we had already seen deceleration
in growth, including the declines, for example, in consumer spending. Ev
eryone I know who has looked at it—outside forecasters and the Greenbook
producers here at the Board [of Governors of the Federal Reserve System]—
believes that the fnancial stress we are seeing now is going to have a signif
cant additional efect on growth. Larry [Slifman, a Fed staf member] gave
some estimates of unemployment above 7 percent for a couple of years. So
even putting aside the extraordinary conditions in fnancial markets, I think
the macro outlook has shifed decisively toward output risks and away from
infation risks, and on that basis, I think that a policy move is justifed.
I should say that this comes as a surprise to me. I very much expected that
we could stay at 2 percent for a long time, and then when the economy be-
gan to recover, we could begin to normalize interest rates. But clearly things
have gone of in a direction that is quite worrisome. (Federal Open Market
Committee 2008b, pp. 13–14)
I suppose that for all the delay, this story actually illustrates the point made above,
namely, that it is easier to form a consensus on monetary policy than on fscal
Copyright © 2021. Harvard University Press. All rights reserved.
policy! No consensus was ever formed around President Obama’s stimulus pack-
age—which received nary a Republican vote in the House of Representatives. In
any case, the implementation lag was not due to the inherent dif ficulties of pass-
ing complex legislation. Te American Recovery and Reinvestment Act (ARRA)
was signed into law on February 17, 2009, within a month of Obama’s inaugura-
tion. Te problem was that the stimulus did not begin to take efect until six
months later. (See chapter 15 for an analysis of the Obama stimulus.)
8. Christina Romer (1986) criticized this conclusion as an artifact of diferent meth-
ods of computing unemployment. Te data in Figure 14.9 refect the re-analysis of
the data by David Weir (1992). See Susan Carter (2006, pp. 2-30 to 2-31) for an
overview of the data. Tanks to Gabe Gold-Hodgkin for pointing me to Romer’s
article.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 546–552 841
Chapter 15
Epigraphs: Biden 2009; Boehner 2009.
1. Tis website is no longer operative; the data are now stored at data.nber.org/data/
ARRA/.
2. However, barely half agreed that the game was worth the candle. Te doubters
were either uncertain or disagreed with the statement
Taking into account all of the ARRA’s economic consequences—including
the economic costs of raising taxes to pay for the spending, its efects on fu-
ture spending, and any other likely future efects—the benefts of the stimu-
lus will end up exceeding its costs. (ChicagoBooth)
Two and a half years later, almost two-thirds (of a slightly larger pool) concurred
with the conclusion that the benefts exceeded the costs, a signifcantly greater
percentage, but still short of what might be expected if economics were, as its
practitioners are wont to claim, a science like physics.
Te economics profession was hardly unanimous. An advertisement placed by
the conservative Cato Institute in the New York Times and the Wall Street Journal
on January 28, 2009, had as its central message “we the undersigned do not
believe that more government spending is a way to improve economic perfor-
mance.” Tis ad was signed by some two hundred economists, including several
Nobel Laureates (and future Laureates), http://object.cato.org/sites/cato.org/fles/
pubs/pdf/cato_stimulus.pdf (accessed February 2, 2016).
3. Cogan and Taylor were not the only dissenters from the consensus opinion of the
economics profession that the ARRA created jobs and income. Before the ink was
dry on the legislation authorizing the Obama stimulus, Robert Barro and John
Cochrane separately argued that the multiplier would be zero. Barro (2009) fo-
cused on crowding out: if m = 0, then the multiplier is zero regardless of the value
of the other parameters. Cochrane (2009) agreed with Barro on crowding out, but
also argued that v = 0 on the grounds that any rational agent who receives a tax
cut, transfer, or grant will take into account the debt that the federal government
incurs to fnance the stimulus, according to the “Ricardian Equivalence” theory
developed by Barro in the 1970s and 1980s (Barro 1989)—see chapter 16, appen-
Copyright © 2021. Harvard University Press. All rights reserved.
dix 3. According to this theory, any new spending by stimulus recipients will be
just cancelled out by spending reductions elsewhere in the economy.
Crowding out is surely relevant to stimulus programs launched in times of
high-capacity utilization. High-capacity utilization was the reality of World War
II, one of the periods of rapid military mobilization on which Barro focused in
order to avoid problems of data interpretation that arise when government expen-
diture and employment might be correlated but not because government expendi-
ture is the cause of the increase in employment. Episodes of sharp increases in
military spending avoid the econometric problem of spurious correlation, but
they have little bearing on the Great Recession.
Robert Gordon and Robert Krenn (2010) argue that the multiplier fell from
two to one as slack was eliminated in the military build-up that took place be-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
842 Notes to Pages 552–565
tween the fall of France in the spring of 1940 and the entry of the United States
into World War II at the end of 1941. In that period, unemployment fell in half,
from 15 percent to 7.5 percent.
By mid-2009, when the ARRA stimulus kicked in, the unemployment rate had
climbed to almost 10 percent. Tere was, accordingly, plenty of spare capacity and
an abundance of available labor. Crowding out was hardly an issue. (For a con-
trary view, see Timothy Conley and Bill Dupor [2013].)
4. Tis point was taken up by others and widely cited in the ensuing discussions of
the fscal multipliers and of the wisdom of fscal stimulus policy. See, for example,
Feyrer and Sacerdote (2011), Ramey (2011), Auerbach (2012), Jonas (2012), Con-
ley and Dupor (2013), and Leduc and Wilson (2017).
5. If the trend rate of growth of actual income, g, is stationary, then = g.
6. Similarly, the Blinder–Zandi (2010) multipliers are higher for programs aimed at
those lower on the income scale—for example, 1.24 for a payroll-tax holiday ver-
sus 0.37 for making dividend and capital-gains tax cuts permanent. Romer and
Bernstein (2009) used a disaggregated set of historical multipliers generated by
the Federal Reserve Board/U.S. model and that of a private frm to estimate the
impact of ARRA. Teir overall multiplier was a weighted average of multipliers
for the various parts of the stimulus program, reaching 1.57 afer two years.
7. On another, more generous, reading, these tax breaks were not intended to stimu-
late spending directly, but rather to help private agents get their balance sheets in
order afer the excesses of the Bush years. Stimulus, or at least a substantial part of
it, like the TARP (Troubled Asset Relief Program), was really about swapping
high-quality federal government debt for the tarnished (if not absolutely toxic)
debt of private individuals and businesses, as well as for the debt of state and local
governments. A case can be made that these agents were not in a position to
spend until their own fnancial houses were in order. Tis might qualify as indi-
rect stimulus under an elastic defnition of the term, but not stimulus as conven-
tionally defned.
8. I return to the question of how to treat Medicaid below, in note 10.
9. According to the National Association of State Budget Of ficers (NASBO), in 2009
the states collectively enacted $24 billion of tax increases for FY-2010 (NASBO
Copyright © 2021. Harvard University Press. All rights reserved.
2009, table 7, p. 18). Two states, California and New York, accounted for two-
thirds of the total. Te overall increase in expenditures in FY-2010 was $34 billion.
10. I explain why in the appendix to this chapter. Te argument for collapsing the
distinction between transfers and purchases of goods and services is illustrated by
Medicaid. In the national income accounts, Medicaid shows up as a transfer pay-
ment to individuals (like social security). In fact, the recipient never sees any cash:
Medicaid “transfers” are in fact payments to vendors of medical services.
11. Tis stipulation must be understood as provisional: the next section argues, on
the basis of reports of state budget of ficers, that most states could not have main-
tained their actual spending without the ARRA.
12. Although the supplement published by the Bureau of Economic Analysis on the
impact of the ARRA provides only aggregate data for state and local governments,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 565–577 843
National Income and Product Accounts (NIPA) data break down federal grants
between states and localities. Tese data show only a very modest increase in total
federal grants to localities over the period of the ARRA. It follows that ARRA
grants to the localities could not have been very large. Tis is confrmed by analy-
sis of the detailed ARRA data on the recovery.gov website. My calculation is that,
through the end of calendar 2011, if we add the total contracts, loans, and grants
reported in line with Section 1512 of the Recovery Act, plus Medicaid, 85 percent
went to state governments. Te rest went to private nonproft entities (like univer-
sities), to private businesses, as well as to local governments or their agencies.
13. Worse, I did not keep a careful record of all the transformations of the data, so
that the only way to verify that the errors are small would be to build up the
regression-data set from the original ARRA data.
14. Tere are a few notable exceptions. For example, H. D. Henderson’s (1938) and
Meade and Andrews’ (1938) use of interviews with businessmen to explore the
impact of the interest rate in the determination of investment, and Blinder et al.
(1998) and Bewley (1999), who engage with relevant economic actors to explore
the reasons behind the stickiness of prices and wages, respectively.
15. Internal funds are the sum of general budget-stabilization funds (including Rainy-
Day funds) and special funds.
16. Te fossil-fuel states are an exception to this, as they usually fund their capital
expenditures from the operating budget. Te question of whether they can engage
in countercyclical capital borrowing, then, is not pertinent. North Dakota, enjoy-
ing a boom due to drilling and extracting oil and natural gas from the Baaken
feld, was in a class by itself. Te unemployment rate never went above 4.2 percent
and was under 4 percent for most of the period in question. North Dakota’s re-
sponse to our questions on the efect of the recession on state operations was in
efect, “Recession, what recession?”
17. Tis assumes that the 20 percent of this category that did not go to state govern-
ments was also spent in its entirety.
Chapter 15 Empirical Appendix
Copyright © 2021. Harvard University Press. All rights reserved.
1. Four states, Alabama, Michigan, New York, and Texas, start their fscal years on
other dates: Alabama and Michigan begin their fscal year on October 1, as does
the federal government. New York’s fscal year begins on April 1, and Texas’s on
September 1.
2. Food stamps, though administered by the states, is a federal program and is not
part of any state’s budget.
3. Te value for R2 difers slightly from the value reported for R2 adjusted for degrees
of freedom, reported at the bottom of column 2.
4. A logical case can be made for using the diference between FY-2010 and FY-2009
S-ARRA receipts rather than the FY-2010 level to measure the injection of ARRA
money: FY-2009 revenues are already built into the level of FY-2009 state expen-
ditures, so that what matters for FY-2010 is the additional input of S-ARRA funds.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
844 Notes to Pages 583–595
Tough this argument is plausible, there is good reason for rejecting it and using
the level of FY-2010 ARRA money as the independent variable. Te level of S-
ARRA money received by a state in FY-2010 has virtually no correlation with the
level received in FY-2009. (Te correlation coef ficient is 0.07.) Tis suggests that
FY-2009 S-ARRA revenues did not serve as a meaningful baseline for FY-2010.
Tis conclusion is supported by the fact that the regression coef ficient on FY-2009
S-ARRA revenue is close to −1 in a frst-stage regression of S-ARRA revenue on
its instruments like the last one in Table 15.10—with the modifcation that the de
pendent variable is the diference between FY-2010 S-ARRA receipts and FY-2009
receipts. Te independence of FY-2010 S-ARRA receipts from FY-2009 S-ARRA
receipts makes intuitive sense because the relatively modest FY-2009 S-ARRA
transfers to the states were largely determined by Medicaid reimbursement; a
more complex allocation of S-ARRA funds among the states, evidently infuenced
by political bargaining during the legislative process leading up to the enactment
of the ARRA, did not really kick in until FY-2010.
5. Experiments with alternative combinations of fnancial controls, omitting one or
another of the measures of fscal solvency, are not reported in Table 15.10, but the
same pattern holds: coef ficients in excess of 1.75, and z-values in excess of 3.
6. In a debate at Harvard University in February 2012 between Taylor, a former Un-
dersecretary of the Treasury in the George W. Bush administration, and Larry
Summers, former Treasury Secretary (under President Bill Clinton) and Obama
adviser, Summers observed that constitutional or statutory balanced-budget re-
quirements prevented the states from borrowing and smoothing expenditure.
Taylor, in response, suggested that states could borrow on capital account to ad-
just their overall spending. In response to an email request for clarifcation, Taylor
repeated that “borrowing for infrastructure investment is one means of fexibility”
(Taylor, personal communication, February 29, 2012.)
7. By contrast, the ratio 7/126, which implies that tax revenues are 5.5 percent of
mining income, is not unreasonable.
8. James Feyrer and Bruce Sacerdote (2011) make political clout in the federal legis-
lative process, as measured by the average seniority of each state’s house delega-
tion, the instrument for explaining the distribution of ARRA moneys.
Copyright © 2021. Harvard University Press. All rights reserved.
9. In theory, tax revenues rather than expenditures could be afected, but Shoag
(2010, sec. 5.4) fnds no efect on taxes, a fnding similar to what my regressions
on S-ARRA spending suggest.
Chapter 16
Epigraphs: Galbraith 1998 [1958], pp. 187–188; Lerner 1944, pp. 316, 318.
1. As a percentage of GDP, overall government purchases of goods and services ac
tually declined signifcantly, from a peak of more than 30 percent of GDP during
World War II to a range of 15 to 20 percent in the postwar years. Te fgure was 15
percent in 2018. Te federal government’s purchases of goods and services were
about 5 percent of GDP in 2018, almost exactly where it was on the eve of the
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Page 595 845
military build-up in 1940. Te increasing size of government is rather the result of
the long-term increase in transfer payments (social security, grants to states, etc.).
As a share of federal government expenditure, purchases of goods and services
declined from just under 60 percent in 1940 to less than 25 percent in 2018. Over
the same period, transfer payments rose from just over 20 percent to over 60 per-
cent. (Te remainder goes to pay interest on the public debt, some 10 percent of
the federal budget in 1940 and 12 percent in 2018.)
Tese numbers are somewhat misleading. Transfers in the federal budget in-
clude payments to state and local governments under the rubric of fscal federal-
ism. Te states and localities are thus a conduit for the federal purchase of goods
and services. At an earlier time these transfers were not signifcant, but they be-
came increasingly important, rising from less than 8 percent of federal expendi-
tures in 1940 (3 percent during the height of the Cold War in the 1950s) to 13
percent in 2018. Transfers to states and their subdivisions show up in the National
Income and Product Accounts either as direct purchases of goods and services by
states and localities (as for highway construction or education) or as purchases by
households (as for Medicaid). As chapter 15 noted, the states administer Medicaid
and pay a share of its costs but receive transfer payments from Washington in sup-
port of their programs. In the NIPA tables, Medicaid is a transfer to households
and a purchase of medical services by these households. Of course the households
never see any money coming in or going out; it is purely an accounting conven-
tion to attribute Medicaid purchases to the private sector.
If we shif the 13 percent of its budget that the federal government transfers to
the states into the category of purchases of goods and services, this category in-
creases from about one-quarter of the budget to three-eighths in 2018. Transfers
are reduced to about 50 percent of the budget.
Accordingly, a better picture of the relative roles of transfers and purchases of
goods and services is obtained by consolidating government accounts. It is still
the case that by 2018 the balance of expenditure had shifed markedly, but the dif-
ference between the consolidated account and the federal government account in
isolation is quantitatively important. In 1952 the composition of overall govern-
ment expenditure was broadly similar to the composition of the federal budget:
Copyright © 2021. Harvard University Press. All rights reserved.
close to 75 percent went to the purchase of goods and services, with 15 percent
going to transfer payments and just under 10 percent to interest payments. In
2018 transfer payments and purchases of goods and services took about the same
share of the consolidated government budget, just under 45 percent each. Te re-
mainder is interest on government debt.
2. Te distinction between government purchases of goods and services and private
purchases is not the same distinction as between private and public goods. In the
private versus public distinction, private goods are those for which one more unit
for me means one less for you—a loaf of bread—and public goods are those we
consume simultaneously—the internet—in which there is no such trade-of be-
tween your consumption and mine (until a website crashes). Tere is no logical
necessity for consumption and investment provided collectively by the govern-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
846 Notes to Pages 596–611
ment to be identifed solely and strictly with public goods, but there is an obvious
affinity between public and collective, and the affinity is not purely semantic.
Governments typically provide such goods as education, transportation infra-
structure, and military, police, and fre protection, all of which have a large com-
ponent of the public even if they d on’t ft the defnition exactly.
Te national income accounts categorize government purchases of goods and
services, collective goods in my language, as consumption or investment. In ac
tuality, most goods provided by the government defy being compartmental-
ized in this way, combining as they do elements of both consumption and invest-
ment.
3. In the feld of economics, this literature dates from Kenneth Arrow (1963 [1951]).
4. Arthur Lafer (2004) and his eponymous curve is an extreme version. Lafer gives
both Ibn Khaldun and Keynes authorial priority. I’m not sure what Ibn Khaldun
had in mind, but Keynes clearly was not focused on supply. Early on in Te Means
to Prosperity (1933), Keynes writes,
Nor should the argument seem strange that taxation may be so high as to
defeat its object, and that, given suf ficient time to gather the fruits, a reduc-
tion of taxation will run a better chance, than an increase, of balancing the
Budget. (p. 7)
Lafer quotes this passage without, apparently, ever realizing that the whole pam-
phlet was about the efects of taxes on demand.
5. A crucial assumption here is that taxes are proportional to output. A tax levied on
profts would change the story; in particular, investment would presumably be-
come sensitive to the tax rate.
6. On the one hand, Mr. Micawber’s “Annual income twenty pounds, annual expen-
diture nineteen six, result happiness. Annual income twenty pounds, annual ex-
penditure twenty pound ought and six, result misery” (Charles Dickens, David
Copperfeld).
On the other hand, “We owe it to ourselves.” Even in a closed economy, in
which the entirety of the debt is held domestically, this simplifcation obscures the
concentrated ownership of the public debt, the consequence of which is to place a
disproportionate burden of debt service on the large majority who own a small to
Copyright © 2021. Harvard University Press. All rights reserved.
vanishing part of the debt. Sandy Hager (2016) explores the distribution of the
ownership of the debt in the United States and fnds that in 2015 the top 1 percent
of the distribution owned about half of the privately held domestic share of the
U.S. public debt.
7. See, for example, N. Gregory Mankiw (2018, pp. 72–74).
8. For the moment, we can assume compulsory overtime. (See chapter 9, note 12.) In
the long-run context of chapters 18, 19 and 20, the result is an expansion of the
labor force.
9. See the next chapter for a brief discussion of stagnation as a characterization of
the long-run prospects for capitalism, a characterization that was in vogue in the
wake of the Great Depression and, not surprisingly, revived in the wake of the
Great Recession.
10. Priority for recognizing the peculiarities of the 1944 to 1949 U.S. debt episode
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 613–620 847
goes to John Irons and Josh Bivens (2010, p. 2). I am grateful to Michael Ash for
pointing me to the Irons–Bivens paper and saving me the embarrassment of any
claim to originality.
Te Irons–Bivens argument should not be confated with the discovery by
Tomas Herndon, Michael Ash, and Robert Pollin of data errors that called other
Reinhart–Rogof conclusions into doubt (2014). Responding to Herndon et al.,
Rogof argued that these errors did not vitiate the conclusions of their 2010
paper(s) or his later work, with Carmen Reinhart and Vincent Reinhart (2012).
Rogof (2013) emphatically disavowed any political motivation or interest: “Is our
research politically motivated, as [Herndon et al] infer? No, we are centrists, our
academic research has always been completely apolitical.” Figure 16.11 does not
appear in the published version of Reinhart and Rogof ’s 2010 paper. Perhaps for
reasons of space.
11. Friedman is not indiscriminate in his condemnation of defcits. When the econ-
omy is operating below its productive capacity, defcits may be appropriate to
stimulate the economy. Te problem comes when the defcit continues even afer
the economy has reached full employment. (1988, pp. 142–143)
For this reason, it is a bit inconvenient for Friedman’s argument that most of
the Reagan defcits took place when the economy was actually quite far from full
employment, with an average rate of unemployment close to the average rate for
the eight years following the fnancial crisis of 2008: the economy was not operat-
ing on the frontier of the choice space in Figures 16.1 to 16.5 but well in the inte-
rior. By the end of Reagan’s second term, the unemployment rate was down to 5
percent, a rate that mainstream macroeconomists at that time considered to be
full employment, the “non-accelerating infation rate of unemployment” (NAIRU,
for short). In any case, 5 percent was a great improvement over the average rate of
unemployment during Reagan’s tenure in of fi ce. But by 1989 the defcit had also
come down sharply. Te total defcit was 3 percent of GDP, and the primary bud-
get was in surplus.
12. In this essay, “Economic Possibilities for Our Grandchildren,” originally published
in 1930 to cheer up his fellow Brits as the world was beginning to feel the shock of
the Great Depression, Keynes distinguished between absolute needs and relative
Copyright © 2021. Harvard University Press. All rights reserved.
needs, the needs we must fulfll to keep up with the Joneses. Te second class, he
recognized, knows no bounds, but he evidently believed that relative needs can be
ignored because they are important only while the economy is in the process of
accumulating suf ficient capital to satisfy absolute needs. Keynes’s implicit corol-
lary was that once this purpose has been fulflled, relative needs will wither away.
It remains to be seen whether the genie of emulation will go back in the bottle
once it is no longer functional to the project of creating the economic basis of a
good and meaningful life. Max Weber’s iron-cage metaphor certainly suggests
otherwise (1930, pp. 181–182): the dignity threshold might be an ever-receding
target. (Tanks to Will Rhatigan for bringing Weber to my attention in this re-
gard. See Marglin 2008 [chap. 11], for a more extensive treatment of Keynes’s ar-
gument.)
13. It was lef to Adam Smith (1982 [1759], pp. 304, 312–313) to embrace consequen-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
848 Notes to Pages 620–633
tialism fully as a moral argument. With some verbal sleight of hand, he was able
transform vice into virtue and resolve the paradox. (See Marglin 2008 [chap. 6].)
14. Friedman published Day of Reckoning in 1988. In the intervening three decades,
serious study has gone into providing the background for such a debate. Jon
Bakija, Lane Kenworthy et al. (2016) provide a good introduction.
15. Te argument is tricky. Te claim is that the exchange rate determines the balance
of payments, but not that the balance of payments determines the exchange rate.
Keynes’s liquidity-preference logic applies here. Te starting point for liquidity-
preference theory is that the interest rate is determined by stocks of various assets
rather than fows of saving and investment. But even though investment and sav-
ing do not determine the rate of interest, the rate of interest (along with income)
is a determinant of investment and saving.
Te same is true of exchange rates. Tese are determined by stocks of assets,
not in markets for exports and imports of goods and services (fows). But ex-
change rates in turn (along with other variables) determine exports and imports
and hence the balance of trade.
16. An alternative to imposing trade balance as a constraint is to incorporate the
trade defcit as an argument of the social-welfare function. In this case, the bene
fcial efects of low interest rates on the trade defcit would be weighed against any
putative welfare losses from driving investment beyond the socially desirable
level.
Tis trade-of could be avoided if it were possible to set the exchange rate inde
pendently of the interest rate. In this case, the addition of the new goal of control-
ling the balance of trade would be matched by the introduction of a new instru-
ment (the exchange rate). Te optimization problem would be more complex but
no diferent in kind from the two-instrument, two-goal problem of choosing
combinations of consumption, investment, and government spending subject to
the full-employment constraint.
It is debatable whether a country can determine its exchange rate even if it sub-
ordinates monetary policy to this goal. Tere are certainly examples of countries
trying to do so and failing. But if any one country is able to do so, it would pre-
sumably be the United State. Te dollar is the dominant reserve currency and, not
Copyright © 2021. Harvard University Press. All rights reserved.
coincidentally, U.S. resources for intervention in foreign-exchange markets are
enormous.
17. Te evolution from B to A″ captures the plight of the Eurozone under the impact
of the 2008 fnancial crisis and the ensuing recession. Te European Central Bank
moved relatively quickly to reduce short-term interest rates, but the zero lower
bound and inelastic investment demand prevented conventional monetary policy
from restoring full employment in an increasingly beleaguered southern Europe.
Te short-term interest rate was reduced below zero in 2014, and quantitative eas-
ing was instituted early in 2015, but these unconventional forms of monetary
policy were too little, too late.
Te functional-fnance solution would have been an aggressive fscal policy
aimed at both higher government spending and higher consumption spending; it
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 634–653 849
was a situation calling for a good measure of defcit fnance. But, no institutions
for delivering a Europe-wide defcit exist, and even if the institutions had been in
place, from 2010 on the ideology of sound fnance dominated German, and con-
sequently, European politics (Bibow 2018).
During the teens of this century, sound fnance, aka austerity, crippled the Eu-
ropean economy. For the Eurozone as a whole, unemployment in 2014 was more
than half again as high as it was in 2007, 11.6 percent against 7.2 percent. In
Greece and Spain, unemployment reached levels not seen in the capitalist world
since the Great Depression, with more than 25 percent of the labor force of ficially
unemployed. As is usually the case, the young were hit especially hard: in both
countries, unemployment rates among people under 25 topped 50 percent, and
the fgure doesn’t include those who fed the land of their birth.
18. How does the acquisition of government bonds increase private wealth? Tere
would be no change in private wealth were the exchange of cash for bonds simply
an alteration of the composition of existing portfolios, as quantitative easing was
modeled in chapter 12. Here private wealth portfolios grow, since agents are ex-
changing their current income for newly issued bonds.
19. Key to Lerner’s story is the idea that public debt is private wealth. Here Lerner
anticipates Don Patinkin’s version (1948) of the real-balance efect. As chapter 7
noted, Patinkin’s argument relies on defation to raise the real debt:GDP ratio to
whatever might be required for economic stimulus. Lerner argues for a falling
primary defcit and hence a falling debt:GDP ratio.
20. Tis is surmise, hence “perhaps.” Lerner’s Economics of Control, published in 1944,
contains no discussion at all of the efect of debt on private wealth. David Colan-
der, who I believe knows the evolution of Lerner’s views as well if not better than
any other living soul, could not enlighten me on why the argument was dropped.
Chapter 17
Epigraph: Robinson 1962, p. 14.
1. Tough in Hicks’s case the problem was not in his original argument but in how
the argument was understood, or rather misunderstood. See chapter 3, note 12.
Copyright © 2021. Harvard University Press. All rights reserved.
2. Harrod’s second proposition—that the rate of increase of income is an important
determinant of the demand for saving [investment]—is actually quite vague. It is
usually interpreted as providing a theory of investment demand based on a tech-
nological accelerator principle, but it is also consistent with a psychological basis
for the incremental capital:output ratio. In the psychological interpretation, the
key element is the optimism or the pessimism of businessmen and -women with
respect to the durability of any expected increase in output. When animal spirits
(to use Keynes’s phrase) are strong, business will react enthusiastically to a given
expectation of increased demand, but the same upsurge can trigger much less in-
vestment in a pessimistic climate in which the anticipated growth in output is not
expected to be sustained. In chapter 10 we captured this by introducing as a
subjective measure of the probability of selling additional output.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
850 Notes to Pages 657–664
Te usual interpretation of v as a technologically determined parameter is more
the product of subsequent theorizing than of a clearly stated view on Harrod’s
part. In a series of papers (see especially Besomi 2001), Daniele Besomi argues
that subsequent theorizing misunderstood Harrod’s concern in attributing to him
a focus on the conditions of steady growth; in Besomi’s view, Harrod’s concern
was rather with the trade cycle, and fuctuations in v were a critical piece of his
story.
3. In fact, Domar has a moment of ambiguity where he toys with the idea that v may
refect investor psychology:
In some problems [1/v] may be interpreted as the minimum annual output
per dollar invested which will make the investment worth undertaking. If
this output falls below [1/v], the investor suffers a loss or at least a disap-
pointment, and may be unwilling to replace the asset afer it has depreciated,
(1947, p. 44)
But this interpretation goes against the grain of Domar’s overall argument, which,
when dealing with the natural rate of growth, emphasizes technological determi-
nation of v as a supply-side parameter.
4. A twenty-frst-century reader might wonder why there was no apparent attempt
by Harrod and Domar to sort out their diferences, despite the fact that the two
authors were developing their respective theories at pretty much the same time.
Harrod’s original paper was published in 1939 and Domar’s in 1946, followed by a
somewhat lengthier exposition in 1947.
Likely, the main reason was that World War II diverted attention and made
scholarly communication across the Atlantic very dif ficult. Domar was evidently
unaware of Harrod’s contribution when he wrote his two papers. In the second of
the two, there is a curious footnote, added afer the paper was completed, in
which Domar acknowledges Harrod’s 1939 article as “[containing] a number of
ideas similar to those presented here” (1947, p. 42n). In the preface to Towards a
Dynamic Economics, Harrod, for his part, acknowledged being preoccupied by
other duties during the war so that an apology was due “to those writers overseas
whose recent contributions to the subject may appear to have been neglected in
these pages” (1948, p. v).
Copyright © 2021. Harvard University Press. All rights reserved.
5. Tis assumes that the elasticity of substitution in production does not exceed one,
which guarantees that the slope of the production function f(k), and hence sf(k), is
initially infnite and falls to zero as k increases. See the mathematical appendix to
chapter 18 for a discussion of the elasticity of substitution.
6. http://en.wikisource.org/wiki/The_poems_of_John_Godfrey_Saxe/The_Blind
_Men_and_the_Elephant
7. Te reason for ignoring technological change is hardly that it is unimportant.
Rather, the opposite: technological change is too important to be tacked on to a
book that has as its purpose the feshing out of a macro theory in the spirit of
Keynes.
8. For an informative history of secular stagnation, see Roger Backhouse and Mauro
Boianovsky (2016). Tis idea had a revival under the aegis of Larry Summers,
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 666–678 851
with Ben Bernanke, Robert Gordon, Paul Krugman, Ken Rogof, and other prom-
inent economists weighing in. (See Summers 2014, 2015; Bernanke 2015b; Gor-
don 2014, 2015, chaps. 17 and 18; Krugman 2014; Lo and Rogof 2015.)
As with the blind men and the elephant, the participants in this debate stress
their diferences, but there are basically two arguments: Krugman and Summers
ofer reasons why investment demand might be defcient, whereas Bernanke and
Rogof stress the reasons why saving might be excessive. Gordon’s contribution to
the debate is to change the subject: his focus is not on aggregate demand, the fo-
cus of Hansen’s original theory and of the other contributors to the revival, but
rather is about reasons why productivity growth can be expected to be lower in
the future than in the past. It is a supply-side rif on a demand-side debate.
My own take is that all the arguments are plausible—except for the argument
that the low level of real interest rates refects the role of the interest rate in equili-
brating the demand for investment with the supply of saving, relatively low rates
indicating low investment demand (Summers) or high savings (Bernanke). Both
Bernanke and Summers are stuck in a Wicksellian world (see chapter 2, note 7,
and the section titled “Real and Nominal Rates: Is Central Bank Freedom Limited
by Necessity?” in chapter 13).
But plausibility is not enough. Summers’s brief for stagnation (or one or an-
other of the variations on Summers’s theme) may turn out to be prophetic—or it
may not, as Hansen learned when the post–World War II boom trumped his 1938
reasons for anticipating stagnation. Twenty-frst-century stagnation may also turn
out to be one of those ideas that seems attractive at the time but whose time is yet
to come.
9. Tis is not just a theoretical possibility: over the four years of economic collapse
from 1929 to 1933, the U.S. labor force grew by 5 percent while the capital stock
actually shrunk! Tis would only have made matters worse for those lucky enough
to still be at work: equilibrium at F would imply a higher real-price level and thus
a lower real wage than equilibrium at E.
10. As noted above, the original source was likely Michał Kalecki, who in some ways
anticipated Keynes, and with respect to the theory of saving went back to the fu-
ture in adopting the nineteenth-century class-based view of saving. See Kalecki
Copyright © 2021. Harvard University Press. All rights reserved.
(1971 [1933]). See also Kalecki (1982 [1936]).
11. Unlike the fxprice model, where a gap between desired saving and desired invest-
ment is a disequilibrium phenomenon.
Chapter 18
Epigraph: Harrod 1948, pp. 19–20.
1. One exception is John Cornwall (1977, esp. chaps. 4 and 5). Another is Donald
Harris (1978, chap. 10, esp. pp. 273–276).
2. Figure 18.2 understates the historical contribution of the reserve army to the labor
force. Te correct comparison with the growth of the paid labor force is not the
growth of the native-born population; at any given time the native-born popula-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
852 Notes to Pages 679–685
tion includes children of immigrants, whom we might consider second-generation
members of the reserve army. More appropriate than the growth of the native-
born population as a standard of reference would be a hypothetical population
based on the initial population at time zero and the hypothetical fertility of this
population extrapolated over time.
Tis century may see the end of the labor-surplus economy. Even China, which
in the twentieth century was the poster child for labor surplus, is approaching the
time when the supply of workers from its hinterland will be exhausted (Fang Cai
and Meiyan Wang [2011]). In 2016, the percentage of the Chinese labor force
employed in agriculture was less than 30 percent (World Bank 2016). And, as
of 2012, China’s female participation rate was the highest in the world and close
to the male participation rate, so there is not much lef of that reserve army (Bu-
reau of Labor Statistics 2012). China is unlikely to be willing or able to tap the
reserves of foreign workers. According to World Bank data from 2016, India is
much further behind in terms both of the ratio of employment in agriculture (ap-
proximately half the labor force) and female participation rates (barely one-
quarter).
3. Arthur Pigou, who for Keynes epitomized the failings of the mainstream to un-
derstand the nature of unemployment under capitalism, endorsed the idea of a
“subsistence” wage and the key role of public opinion in determining its level:
“Public opinion in a modern civilised State builds up for itself a rough estimate of
what constitutes a reasonable living wage” (1933, p. 255). Tis, for Pigou, was one
reason why the real wage might persist at a level higher than the market-clearing
wage, resulting in persistent unemployment.
Te decline of unions is itself a consequence, at least in part, of the shif in pub-
lic opinion against the working class. Richard Freeman and James Medof (1984)
argued that workers themselves were more likely to organize in unions to the ex-
tent the public had a favorable image of u nions (cited in Surowiecki 2011).
Tere is a large literature on how norms enter into the determination of wages.
Solow (1990) provides a good introduction and references to articles published in
the 1980s, a (local) peak period of interest in the subject.
4. Olivier Blanchard and Larry Katz (1999) present a similar econometric analysis
Copyright © 2021. Harvard University Press. All rights reserved.
but with a very diferent interpretation of the variable that I interpret as a conven-
tional wage.
5. Te conventional wage is formulated as a share of product rather than as a wage
rate. In the theoretical models of this chapter, we can assume the conventional
wage is a fxed rate because there is no productivity growth to propel wage expec-
tations. But this will hardly do for empirical analysis.
6. In discussing the (im)possibility of curing unemployment by reducing money
wages, Keynes observes, “Te transfer from wage-earners to other factors is likely
to diminish the propensity to consume” (Te General Teory, p. 262). Tis recog-
nition of class-based diferences in the propensity to consume is presumably the
basis for his subsequent argument that an increase in overall consumption de-
mand can be engineered by a deliberate policy of redistribution. Keynes saw no
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 689–715 853
immediate need for such a policy but argued that it might be necessary in a gen-
eration or two in order to counteract the efect of investment satiation (Te Gen-
eral Teory, pp. 325–326, 374–377).
7. In the sections that follow, as in earlier chapters, the capitalist economy is as-
sumed to be perfectly competitive in the sense that no agent has market power.
Besides market power, real-world features such as frictions are notable for their
absence in the model. Te point, evidently, is not to provide a faithful description
of the world as it is, but to forestall the argument that the results would not hold
in the absence of one imperfection or another.
8. If the AD schedule lies to the lef of the intersection of the GS and LS schedules,
the stationary real-price schedule and AD schedule will intersect at a point where
prices and wages are falling at the same rate, as in chapter 6. Te distinction be-
tween infation-with-growth and defation-with-decline is not a theoretical dis-
tinction between the long run and the short run, but an empirical one; both are
possible long-run outcomes. In fact, the history of capitalism has been mostly a
history of rising output and infation. Te Great Depression and the Great Reces-
sion are exceptional in historical perspective.
9. What is at issue is the lack of coordination between diferent decisions, a phenom-
enon that we have encountered before, beginning with the frst foray into dynam-
ics in chapter 5. One way of understanding why the right and lef hands appear
at odds is that they might belong to diferent agents, for example, a capital-
expenditure manager and a production manager.
Te present formulation collapses two meanings of the production function.
Chapter 17 noted the diference between the short-run function, which refects
varying intensities of utilization of an existing capital stock, and the long-run
function, which refects diferent production techniques. Presumably, the produc-
tion manager is operating on the short- run function, whereas the cap ital-
expenditure manager is thinking in terms of the long-run function. Adding the
changes in the labor:capital ratio corresponding to the two distinct meanings of
the production function is, to say the least, suspicious.
10. Te original Luddites have got a bad rap for indiscriminately opposing labor-
saving machinery. Oppose they did, but with considerable discrimination. Under
Copyright © 2021. Harvard University Press. All rights reserved.
the leadership of the mythical General Lud, these nineteenth-century English ar-
tisans demolished machinery that was destroying their trades. Tey had no argu-
ment with machinery that was compatible with artisanal production. On the
Luddites as they really were, see E. P. Tompson (1963, pp. 521–602).
11. Flexprice results corresponding to Table 18.3 are omitted.
Chapter 18 Mathematical Appendix
1. Consider the generic case
u = f(u, v, x)
v = g(u, v, x).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
854 Notes to Pages 726–734
At equilibrium u = v = 0, so
∂u ∂v
fu + fv + fx 0
∂x ∂x
∂u ∂v = 0
gu + gv + gx
∂x ∂x
and
∂u ∂v
fu + fv
∂x ∂x = − f x .
∂u ∂v g
gu + gv x
∂x ∂x
With
f fv
J = u
gu g v
we have
∂u
∂x f
J = − x
∂v gx
∂x
and
∂u
∂x −1 f x
∂v = − J g .
x
∂x
2. My ROKE paper (Marglin 2017) is in error on this point. Te expression on page
391 for ∂(P/W)/∂ is the same as here, but I mistakenly assert that this expression
is unambiguously positive.
3. Piketty (2014, pp. 346–347) identifes the patrimonial middle class as the 40 per-
cent of the population between the top decile and the bottom half. Lumping four
Copyright © 2021. Harvard University Press. All rights reserved.
deciles together likely exaggerates the size of the group with signifcant wealth. If
the top quarter of this group, that is, the decile lying between the 80th and 90th
percentiles, possesses double the wealth of the average of the other three deciles
and if the group as a whole owns 40 percent of total wealth, as in France in 2015
(Piketty 2020, fg. 4.2), then the three deciles lying between the 50th and 80th
percentiles together hold only 24 percent of total wealth. Moreover, this wealth is
mostly equity in their residences and other real estate (Piketty 2020, fg. 11.17). I
haven’t seen corresponding numbers for the United States, but I would be sur-
prised if the qualitative pattern were diferent.
4. Observe that in the production function, L now represents executive, managerial,
and technological labor. As I have noted, the traditional working class plays no
role in the model; indeed, the simplest assumption is that this working class has
been automated out of existence. More realistically, we could assume that a work-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Notes to Pages 737–773 855
ing class exists along with the PMC and the rentier class and that the PMC appro-
priates a percentage of total wages and salaries. Te PMC’s share of labor income
plus its share of profts becomes the basis of its consumption and saving. Te re-
mainder of wages and salaries may be assumed to go to the rest of the working
population, but the working class plays no further role in the analysis: by assump-
tion, all its income is consumed. I do not think that greater realism in this regard
would add much to the economics, but it would certainly complicate the mathe-
matics.
5. For = 0, the Jacobian is
−θ1GS P − θ 3 θ1 0
W
P
−2 −1
P
J = θ 2 ψ′Ω l θ 2 ψ′Ω′r + ψ′Ω hl − − sw hl − ( s π − sw ) r .
W W
0 0 sw h(l , ξ) − s πr
Tis is essentially the model with a uniform propensity to save and capital widen-
ing, in which the condition for stability is that the stationary-price locus is fatter
than the AD schedule.
Chapter 19
Epigraphs: Lucas 1981, p. 560; Gordon 1984, p. 42.
1. Infation is measured by the consumer price index, which elides the relationship
between wage changes and price changes. In the models of the previous chapter
(and the mathematical appendix), this relationship is a simple one: these models
assume an equilibrium in which the rate of price infation and wage infation are
equal.
2. If we replace employment by capacity utilization (defned as the ratio of GDP to
potential GDP), the slope of the Phillips curve changes very little over the sample
until the frst decade of the twenty-frst century.
3. Te year 2008 is omitted because infation was negative and the log of the infa
tion rate therefore undefned. Tat year is literally of the chart. Evidently, includ-
Copyright © 2021. Harvard University Press. All rights reserved.
ing 2008 would increase the dispersion dramatically!
4. Standard errors are calculated diferently in the TSLS procedure.
5. According to Mankiw, “Unexpected infation redistributes wealth among the pop-
ulation in a way that has nothing to do with either merit or need” (2018, p. 645).
According to Baumol, Blinder, and Solow, “Infation redistributes income in an
arbitrary way” (2020, p. 477).
Epilogue
Epigraph: Keynes 1934, pp. 850–851.
1. Rod O’Donnell (1999) and, more recently, James Crotty (2019) argue that Keynes’s
enlightened capitalism is really “liberal socialism,” a term invoked by Keynes to
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
856 Notes to Pages 791–792
describe his economic philosophy in a 1939 interview with Kingsley Martin in the
New Statesman and Nation:
Te question is whether we are prepared to move out of the nineteenth-
century laissez-faire state into an era of liberal socialism, by which I mean a
system where we can act as an organised community for common pur-
poses and to promote economic and social justice, whilst respecting and
protecting the individual—his freedom of choice, his faith, his mind and its
expression, his enterprise and his property. (Keynes 1982, p. 500; quoted in
O’Donnell, p. 164)
2. Note the resort to the language of comparative statics. No apologies: comparative
statics is a useful tool when it does not force us to assume away essential difer-
ences between real-time change and ahistorical comparisons.
3. Maria Cristina Marcuzzo (2010) contrasts Keynes with Sir William Beveridge,
whose eponymous Report laid the foundations for the post–World War II British
welfare state. Crotty (2019, chap. 22) emphasizes Keynes’s support for the Bev-
eridge Report as evidence of his commitment to “liberal socialism.”
Crotty relies on a letter to Beveridge in which Keynes expresses “wild enthusi-
asm for your general scheme” (Keynes 1980, p. 204). However, most of his subse-
quent interventions—an entire chapter of the volume of Keynes’s collected writ-
ings on his activities as adviser to the Treasury (Keynes 1980, chap. 4) is given
over to Keynes’s interactions with Beveridge and others during the drafing of the
Report—concerned the fnancial details of unemployment insurance, children’s
allowances, and pensions. Robert Skidelsky, the magisterial biographer of Keynes,
concludes,
Te truth seems to be that he [Keynes] was not interested in social policy as
such, and never attended to it. Te sole question in his mind was whether
the Exchequer could “aford” Beveridge. (2000, p. 270)
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
references
Acemoglu, Daron, and Pascual Restrepo. 2018. “Artifcial Intelligence, Automation
and Work.” MIT Working Paper 18-01, January.
Achen, Christopher H. 2001. “Why Lagged Dependent Variables Can Suppress the
Explanatory Power of Other Independent Variables.” Unpublished working paper,
Department of Political Science and Institute for Social Research, University of
Michigan.
Adrian, Tobias, Karen Kimbrough, and Dina Marchioni. 2011. “Te Federal Reserve’s
Commercial Paper Funding Facility.” FRBNY Economic Policy Review (May): 25–
39. https://www.newyorkfed.org/medialibrary/media/research/epr/11v17n1/1105a
dri.pdf (accessed December 2, 2018).
Ait-Sahalia, Yacine. 1996. “Testing Continuous-Time Models of the Spot Interest
Rate.” Review of Financial Studies 9:385–426.
Akerlof, George. 2007. “Te Missing Motivation in Macroeconomics.” American Eco-
nomic Review 97:3–36.
Allais, Maurice. 1953. “Le Comportement de l’Homme Devant le Risque: Critique
des Postulats et Axioms de l’Écôle Américaine” (Human Behavior Toward Risk:
A Critique of the Postulates and Axioms of the American School). Econometrica
21:503–546.
Ando, Albert, and Franco Modigliani. 1963. “Te ‘Life Cycle’ Hypothesis of Saving:
Aggregate Implications and Tests.” American Economic Review 53:55–44.
Ando, Albert, and Karl Shell. 1975. “Demand for Money in a General Portfolio
Model in the Presence of an Asset Tat Dominates Money.” Appendix to Ando
Copyright © 2021. Harvard University Press. All rights reserved.
and Franco Modigliani, “Some Refections on Describing Structures of Financial
Sectors.” In Te Brookings Model: Perspective and Recent Developments, ed. Gary
Fromm and Lawrence Klein, pp. 560–563. Amsterdam: North Holland.
Arrow, Kenneth. 1959. “Toward a Teory of Price Adjustment.” In Te Allocation of
Economic Resources: Essays in Honor of Bernard Haley, ed. Moses Abramovitz.
Stanford: Stanford University Press.
———. 1963. Social Choice and Individual Values. 2nd ed. New York: Wiley. [First
published in 1951.]
———. 1963–1964. “Te Role of Securities in the Optimal Allocation of Risk-
Bearing.” Review of Economic Studies 31:91–96. [Originally published in French
in 1953.]
857
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
858 References
Arrow, Kenneth, and Gerard Debreu. 1954. “Existence of Equilibrium for a Competi-
tive Economy.” Econometrica 22:265–290.
Audi, Robert, ed. 1999. Te Cambridge Dictionary of Philosophy. 2nd ed. Cambridge:
Cambridge University Press.
Auerbach, Alan. 2012. “Te Fall and Rise of Keynesian Fiscal Policy.” Asian Economic
Policy Review 7:157–175.
Backhouse, Roger, and Mauro Boianovsky. 2016. “Secular Stagnation: Te History of
a Macroeconomic Heresy.” European Journal of the History of Economic Tought
23:946–970.
Bakija, Jon, Lane Kenworthy, Peter Lindert, and Jef Madrick. 2016. How Big Should
Government Be? Oakland: University of California Press.
Ball, Laurence. 2018. Te Fed and Lehman Brothers: Setting the Record Straight on a
Financial Disaster. Cambridge: University of Cambridge Press.
Bank for International Settlements (BIS). 2010. “Triennial Central Bank Survey of
Foreign Exchange and Derivatives Market Activity in 2010, Annex Tables.” http://
www.bis.org/publ/rpfxf10t.htm (accessed September 12, 2011).
Baqaee, David Rezza, and Emmanuel Farhi. 2019. “Te Microeconomic Foundations
of Aggregate Production Functions.” Journal of the European Economic Associa-
tion 17:1237–1392.
Barens, Ingo. 1997. “Destruction by Simplifcation: IS–LM afer 60 Years.” Presented
at the 1997 Conference of the History of Economics Society, Charleston, June.
———. 2001. “Patinkin on Keynesian Economics and Hicksian IS-LM: Te Neglected
Supply Side of the Commodity Market.” Presented at the Patinkin Conference,
Lausanne, September.
———. 2018. “Robertson’s ‘Liquidity Trap’ as an Answer to Keynes’s ‘Banana Parable,’
or: Did the General Teory Really Have to be Written?” Presented at the 22nd
Annual Conference of the European Society for the History of Economic
Tought, Madrid, June 7–9.
Barro, Robert. 1974. “Are Government Bonds Net Wealth?” Journal of Political Econ-
omy 82:1095–1117.
———. 1989. “Te Ricardian Approach to Budget Defcits.” Journal of Economic Per-
spectives 3:37–54.
Copyright © 2021. Harvard University Press. All rights reserved.
———. 1996. “Infation and Growth.” Review, Federal Reserve Bank of St. Louis
(May–June): 153–169.
———. 2009. “Voodoo Multipliers.” Economists’ Voice 6 (2): article 5.
Barro, Robert, and Herschel Grossman. 1971. “A General Disequilibrium Model of
Income and Employment.” American Economic Review 61:82–93.
———. 1976. Money, Employment and Infation. Cambridge: Cambridge University
Press.
Bauer, Peter. 1938. “Die Allgemeine Teorie von Keynes und ihre Kritiker” (Te Gen-
eral Teory of Keynes and Its Critics). Zeitschrif für Nationalökonomie 9:99–106.
Baumol, William. 1952. “Te Transactions Demand for Cash: An Inventory Teo-
retic Approach.” Quarterly Journal of Economics 66:545–556.
———. 1999. “Retrospectives: Say’s Law.” Journal of Economic Perspectives 13:195–
204.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 859
———. 2000. “What Marshall D idn’t Know: On the Twentieth Century’s Contribu-
tions to Economics.” Quarterly Journal of Economics 115:1–44.
Baumol, William, Alan Blinder, and John Solow. 2020. Economics: Principles and Pol-
icy. 14th ed. Boston: Cengage.
Béraud, Alain, and Guy Numa. 2018. “Beyond Say’s Law: Te Signifcance of J.-B.
Say’s Monetary Views.” Journal of the History of Economic Tought 40:217–241.
Bernanke, Ben. 1983. “Nonmonetary Efects of the Financial Crisis in the Propaga-
tion of the Great Depression.” American Economic Review 73:257–276.
———. 1995. “Te Macroeconomics of the Great Depression: A Comparative Ap-
proach.” Journal of Money, Credit, and Banking 27:1–28.
———. 2015a. Te Courage to Act: A Memoir of Crisis and Its Afermath. New York:
Norton.
———. 2015b. “Why Are Interest Rates so Low?” and “Why Are Interest Rates so
Low, Part 2: Secular Stagnation.” Ben Bernanke’s Blog, March 30 and 31. http://
www.brookings.edu/blogs/benbernanke/posts/2015/03/30whyinterestratesso
low?wpmm=1&wpisrc=nl_wonk; http://www.brookings.edu/blogs/benbernanke/
posts/2015/03/31whyinterestrateslowsecularstagnation.
Besomi, Daniele. 2001. “Harrod’s Dynamics and the Teory of Growth: Te Story of
a Mistaken Attribution.” Cambridge Journal of Economics 25:79–96.
Bewley, Truman. 1999. Why Wages D on’t Fall During a Recession. Cambridge, Mass.:
Harvard University Press.
Bhaduri, Amit, and Stephen Marglin. 1990. “Unemployment and the Real Wage: Te
Economic Basis for Contesting Political Ideologies.” Cambridge Journal of Eco-
nomics 14:375–393.
Bhagwati, Jagdish. 1958. “Immiserizing Growth: A Geometric Note.” Review of Eco-
nomic Studies 25:201–206.
Bibow, Jörg. 2018. “How Germany’s Anti-Keynesianism Has Brought Europe to Its
Knees.” International Review of Applied Economics 32:569–588.
Biden, Joe. 2009. “Remarks by the President and Vice President at Signing of the
American Recovery and Reinvestment Act, February 17, 2009.” Te White
House, President Barack Obama. https://www.whitehouse.gov/the-press-of fi ce/
remarks-president-and-vice-president-signing-american-recovery-and-reinvest
Copyright © 2021. Harvard University Press. All rights reserved.
ment-act (accessed February 2, 2016; site now https://obamawhitehouse.archives
.gov/the-press-of
fi ce/remarks-president-and-vice-president-signing-american-re
covery-and-reinvestment-act).
Blanchard, Olivier, and Larry Katz. 1999. “Wage Dynamics: Reconciling Teory and
Evidence.” American Economic Review 89:69–74.
Blinder, Alan. 1988. “Te Fall and Rise of Keynesian Economics.” Economic Record
68:278–294.
Blinder, Alan S., Elie R. D. Canetti, David E. Lebow, and Jeremy B. Rudd. 1998. Ask-
ing About Prices: A New Approach to Understanding Price Stickiness. New York:
Russell Sage Foundation.
Blinder, Alan, and Marc Zandi. 2010. “How the Great Recession Was Brought to an
End.” http://www.economy.com/mark-zandi/documents/end-of-great-recession
.pdf (accessed February 3, 2013).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
860 References
Board of Governors of the Federal Reserve System. n.d. “Industrial Production,
1957–59 Base.” Washington, D.C., pp. S-148–S-149. Retrieved from FRASER,
http://fraser.stlouisfed.org/docs/meltzer/bogind62.pdf (accessed June 11, 2013).
———. 1943. Banking and Monetary Statistics, 1914–1941, Part 1. Washington, D.C.
https://fraser.stlouisfed.org/title/38 (accessed May 24, 2020).
Boehner, John. 2009. Interview by David Gregory, Meet the Press. “‘Meet the Press’
Transcript for Jan. 25, 2009.” NBC News. http://www.nbcnews.com/id/28841300/
ns/meet_the_press/t/meet-press-transcript-jan/#.VrVB0vkrK00 (accessed Febru-
ary 5, 2016).
Boianovsky, Mauro, and Roger Backhouse. 2003. Transforming Modern Macroeco-
nomics: Exploring Disequilibrium Microfoundations, 1956–2003. New York: Cam-
bridge University Press.
Brainard, William, and James Tobin. 1968. “Pitfalls in Financial Model Building.”
American Economic Review 58:99–122.
Brown, Douglass V., Edward Chamberlin, Seymour E. Harris, Wassily W. Leontief,
Edward S. Mason, Joseph A. Schumpeter, and Overton H. Taylor. 1934. Te Eco-
nomics of the Recovery Program. New York: McGraw-Hill.
Brown, Tillman. 1952. “Habit Persistence and Lags in Consumer Behavior.” Econo-
metrica 20:355–371.
Buckley, William. 1951. God and Man at Yale. Chicago: Henry Regnery.
Bureau of the Census. 1935. Statistical Abstract of the United States, 1935. Washing-
ton, D.C.: U.S. Department of Commerce. https://www.census.gov/library/publi
cations/1935/compendia/statab/57ed.html (accessed October 7, 2018).
———. 1975. Historical Statistics of the United States, Colonial Times to 1970. Wash-
ington, D.C.: U.S. Department of Commerce.
——— (Carmen DeNavas-Walt and Bernadette D. Proctor). 2014. Income and Pov-
erty in the United States: 2013, P60-249. Washington, D.C.: U.S. Department of
Commerce. https://www2.census.gov/library/publications/2014/demographics/
p60-249.pdf (accessed October 28, 2019).
———. 2019a. Historical Income Tables: Income Inequality. Washington, D.C.: U.S.
Department of Commerce. https://www.census.gov/data/tables/time-series/demo/
income-poverty/historical-income-inequality.html (accessed October 28, 2019).
Copyright © 2021. Harvard University Press. All rights reserved.
——— (Jessica Semega, Melissa Kollar et al.). 2019b. Income and Poverty in the
United States: 2018, P60-266. Washington, D.C.: U.S. Department of Commerce.
https://www.census.gov/content/dam/Census/library/publications/2019/demo/
p60-266.pdf (accessed October 27, 2019).
Bureau of Economic Analysis. various dates. National Income and Product Accounts.
Washington, D.C.: U.S. Department of Commerce. https://apps.bea.gov/iTable/
iTable.cfm?reqid=19&step=2#reqid=19&step=2&isuri=1&1921=survey (accessed
various dates).
Bureau of Labor Statistics. 2005. Comparative Civilian Labor Force Statistics, 10
Countries, 1960–2004. Washington, D.C.: U.S. Department of Labor. http://www
.bls.gov/fs/fslforc.pdf (accessed February 11, 2011).
———. 2012. Charting International Labor Comparisons. Washington, D.C.: U.S. De-
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 861
partment of Labor. http://www.bls.gov/fs/chartbook/2012/chartbook2012.pdf
(accessed November 27, 2016).
Cai, Fang, and Meiyan Wang. 2011. “Chinese Wages and the Turning Point in the
Chinese Economy.” East Asia Forum, January 29. www.eastasiaforum.org (ac-
cessed March 13, 2016).
Calvo, Guillermo. 1983. “Staggered Prices in a Utility Maximizing Framework.” Jour-
nal of Monetary Economics 12:382–398.
Campbell, John. 1995. “Some Lessons from the Yield Curve.” Journal of Economic
Perspectives 9:129–152.
Campbell, John, and N. Gregory Mankiw. 1989. “Consumption, Income and Interest
Rates: Reinterpreting the Time Series Evidence.” In NBER Macroeconomics An-
nual 1989, vol. 4, ed. Olivier Jean Blanchard and Stanley Fischer, pp. 185–246.
Cambridge, Mass.: MIT Press.
Campbell, John, and Luis Viceira. 2002. Strategic Asset Allocation: Portfolio Choice for
Long-Term Investors. Oxford: Oxford University Press.
Carlos, Ann M., and Larry Neal. 2011. “Amsterdam and London as Financial Centers
in the Eighteenth Century.” Financial History Review 18:21–46.
Carter, Susan. 2006. “Labor Force.” Historical Statistics of the United States Millennial
Edition Online, pp. 2-3 to 2-35. Cambridge: Cambridge University Press. http://
hsus.cambridge.org.ezp-prod1.hul.harvard.edu/HSUSWeb/essay/showessaypdf.do
?id=Ba.ESS.02 (accessed December 24, 2017).
Chamberlin, Edward. 1962. Te Teory of Monopolistic Competition. 8th ed. Cam-
bridge, Mass.: Harvard University Press. [First published in 1933.]
Chang, Winston W., Daniel Hamberg, and Junichi Hirata. 1983. “Liquidity Prefer-
ence as Behavior toward Risk Is a Demand for Short-Term Securities—Not
Money.” American Economic Review 73:420–427.
Chayanov, Alexander. 1966. A. V. Chayanov on the Teory of Peasant Economy. Ed. D.
Torner, B. Kerblay, and R. Smith. Homewood, Ill.: Irwin. [Chayanov’s theory of
a peasant economy was frst published in 1925.]
ChicagoBooth. 2012. “Economic Stimulus.” IGM Forum, February 15. http://www
.igmchicago.org/igm-economic-experts-panel/poll-results?SurveyID=SV_cw5O9
LNJL1oz4Xi.
Copyright © 2021. Harvard University Press. All rights reserved.
Chirinko, Robert. 2008. “: Te Long and Short of It.” Journal of Macroeconomics
30:671–686.
Chodorow-Reich, Gabriel, Laura Feiveson, Zachary Liscow, and William Gui
Woolston. 2012. “Does State Fiscal Relief During Recessions Increase Employ-
ment? Evidence from the American Recovery and Reinvestment Act.” American
Economic Journal: Economic Policy 4:118–145.
Clower, Robert. 1958. “Keynes and the Classics: A Reinterpretation.” Robert Clower
Papers, Box 4. David M. Rubenstein Rare Book and Manuscript Library, Duke
University.
———. 1960. “Keynes and the Classics: A Dynamical Perspective.” Quarterly Journal
of Economics 74:318–323.
———. 1984. “Te Keynesian Counter-Revolution: A Teoretical Appraisal.” In
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
862 References
Money and Markets: Essays by Robert Clower, ed. Donald Walker, pp. 34–58.
Cambridge: Cambridge University Press. [First published in 1963 as “Die
Keynesianische Gegenrevolution: Eine Teoretische Kritik.” Schweizerische Zeit-
shrif für Volkswirtschaf und Statistik 99:8–31.]
Coates, Ta-Nehisi. 2014. “Te Case for Reparations.” Te Atlantic, June. www.the
atlantic.com/magazine/archive/2014/06/the-case-for-
reparations/361631/?wpisrc=nl_daily202&wpmm=1 (accessed February 11,
2019).
Cochrane, John. 2009. “Fiscal Stimulus, Fiscal Infation, or Fiscal Fallacies?” www
.faculty.chicagobooth.edu/john.cochrane/research/papers/fscal2.htm (accessed
June 24, 2012).
Cogan, John, and John Taylor. 2012. “What the Government Purchases Multiplier
Actually Multiplied in the 2009 Stimulus Package.” In Government Policies and
the Delayed Economic Recovery, ed. Lee. E. Ohanion, John B. Taylor, and Ian J.
Wright, pp. 85–114. Stanford, Calif.: Hoover Institution Press.
Colander, David. 1984. “Was Keynes a Keynesian or a Lernerian?” Journal of Eco-
nomic Literature 22:1572–1575.
———. 1995. “Te Stories We Tell: A Reconsideration of AS/AD Analysis.” Journal of
Economic Perspectives 9:169–188.
Colander, David, and Harry Landreth. 1996. Te Coming of Keynesianism to America.
Cheltenham, UK: Edward Elgar.
Congressional Budget Of fi ce. 2012. “Estimated Impact of the American Recovery
and Reinvestment Act on Employment and Economic Output from October
2011 Trough December 2012: A CBO Study.” CBO Publication No. 4435. Wash-
ington, D.C.: United States Government Printing Of fi ce. Retrieved from http://
www.cbo.gov/sites/default/fles/cbofles/attachments/02-22-ARRA.pdf.
Conley, Timothy G., and Bill Dupor. 2013. “Te American Recovery and Reinvest-
ment Act: Solely a Government Jobs Program?” Journal of Monetary Economics
60:535–549.
Cornwall, John. 1977. Modern Capitalism: Its Growth and Transformation. London:
Martin Robertson.
Council for Economic Education. 2010. Voluntary National Content Standards in
Copyright © 2021. Harvard University Press. All rights reserved.
Economics. 2nd ed. New York: CEE.
Council of Economic Advisers. 2014. Te Economic Impact of the American Recovery
and Reinvestment Act Five Years Later: Final Report to Congress. Washington,
D.C.
Crotty, James. 2019. Keynes Against Capitalism: His Economic Case for Liberal Social-
ism. New York: Routledge.
Davidson, Paul. 1991. “Is Probability Teory Relevant for Uncertainty? A Post
Keynesian Perspective.” Journal of Economic Perspectives 5:129–143.
Davis, Chester. 1936. “Te Agricultural Adjustment Act and National Recovery.”
Journal of Farm Economics 18:229–241.
Davis, Otto A., M. A. H. Dempster, and Aaron Wildavsky. 1966. “A Teory of the
Budgetary Process.” American Political Science Review 60:529–547.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 863
———. 1974. “Towards a Predictive Teory of Government Expenditure: US Domes-
tic Appropriations.” British Journal of Political Science 4:419–452.
de Finetti, Bruno. 1980. “Foresight: Its Logical Laws, Its Subjective Sources.” In Stud-
ies in Subjective Probability, 2nd ed., ed. Henry Kyburg, Jr. and Howard Smok-
ker, pp. 53–118. Huntington, N.Y.: Krieger. [Originally published in French in
1937.]
de Tocqueville, Alexis. 1969. Democracy in America. Trans. George Lawrence. Ed. J.
P. Meyer. Garden City, N.Y.: Doubleday. [Original French publication in two vol-
umes in 1835 and 1840.]
De Vroey, Michel. 2016. A History of Macroeconomics from Keynes to Lucas and Be-
yond. Cambridge: Cambridge University Press.
Debreu, Gerard. 1959. Teory of Value. New York: Wiley.
Department of Health and Human Services. various dates. Medicaid Financial Man-
agement Report. www.hhs.gov (accessed various dates).
Desan, Christine. 2014. Making Money: Coin, Currency, and the Coming of Cap
italism. Oxford: Oxford University Press.
Diamond, Peter. 1967. “Te Role of the Stock Market in a General Equilibrium
Model with Technological Uncertainty.” American Economic Review 57:759–776.
Dimand, Robert, and Rebeca Gomez Betancourt. 2012. “Irving Fisher’s Appreciation
and Interest (1896) and the Fisher Relation.” Journal of Economic Perspectives
26:185–196.
Domar, Evsey. 1946. “Capital Expansion, Rate of Growth, and Employment.” Econo-
metrica 14:137–147.
———. 1947. “Expansion and Employment.” American Economic Review 37:34–55.
Dos Santos Ferreira, Rodolphe. 2014. “Mr. Keynes, the Classics and the New Keynes-
ians: A Suggested Modelling.” European Journal of the History of Economic
Tought 21:801–838.
Dos Santos Ferreira, Rodolphe, and Philippe Michel. 1987. “Te Dynamics of
Keynesian Price and Wage Adjustments.” Bureau d’Économie Téorique et Ap-
pliquée, University of Strasbourg, Document 8702.
———. 2013. “Keynes’ Wage-Price Dynamics.” Metroeconomica 64:44–72.
Duesenberry, James. 1949. Income, Saving, and the Teory of Consumer Behavior.
Copyright © 2021. Harvard University Press. All rights reserved.
Cambridge, Mass.: Harvard University Press.
Dunlop, John. 1938. “Te Movement of Money and Real Wages.” Economic Journal
48:413–434.
Eatwell, John, Murray Milgate, and Peter Newman, eds. 1987. Te New Palgrave: A
Dictionary of Economics. London: Macmillan.
Eggertsson, Gauti. 2008. “Great Expectations and the End of the Depression.” Ameri-
can Economic Review 98:1476–1516.
Eichengreen, Barry. 2015. Hall of Mirrors: Te Great Depression, Te Great Recession,
And the Uses—And Misuses—of History. Oxford: Oxford University Press.
Ellsberg, Daniel. 1961. “Risk, Ambiguity, and the Savage Axioms.” Quarterly Journal
of Economics 75:643–669.
European Central Bank. n.d. “Objective of monetary policy.” http://www.ecb.europa
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
864 References
.eu/mopo/intro/objective/html/index.en.html (accessed December 26, 2009 and
April 26, 2020).
Ezekiel, Mordecai. 1942a. “Statistical Investigations of Saving, Consumption, and In-
vestment. Part I: Saving, Consumption and National Income.” American Eco-
nomic Review 32:22–49.
———. 1942b. “Statistical Investigations of Saving, Consumption, and Investment.
Part II: Investment, National Income and Te Saving-Investment Equilibrium.”
American Economic Review 32:272–307.
Farnsworth, Helen. 1940. “‘World’ Wheat Stocks, 1890–1914 and 1922–39.” Wheat
Studies of the [Stanford] Food Research Institute 16:39–66.
Federal Deposit Insurance Corporation (FDIC). n.d. A Brief History of Deposit Insur-
ance in the United States, Prepared for the International Conference on Deposit In-
surance, Washington, D.C., September 1998. https://www.fdic.gov/bank/histori-
cal/brief/brhist.pdf (accessed May 24, 2020).
Federal Open Market Committee. 2008a. “FOMC Meeting Transcript, September 16,
2008.” Federal Reserve Board. http://www.federalreserve.gov/monetarypolicy/
fles/FOMC20080916meeting.pdf (accessed February 24, 2014).
———. 2008b. “FOMC Conference Call Transcript, October 7, 2008.” Federal Re-
serve Board. http://www.federalreserve.gov/monetarypolicy/fles/FOMC2008
1007confcall.pdf (accessed August 18, 2015).
———. 2014. “Chair’s FOMC Press Conference Projections Materials, September 17,
2014.” Federal Reserve Board. https://www.federalreserve.gov/monetarypolicy/
fles/fomcprojtabl20140917.pdf (accessed March 27, 2019).
Federal Reserve Bank of St. Louis. 2020. “Moody’s Seasoned Baa Corporate Bond
Yield (DBAA).” https://fred.stlouisfed.org/series/DBAA (accessed May 12, 2020).
———. various dates. https://fred.stlouisfed.org/series (accessed various dates).
Federal Reserve Board. 2005. “Testimony of Chairman Alan Greenspan. Federal Re-
serve Board’s semiannual Monetary Policy Report to the Congress Before the Com-
mittee on Banking, Housing, and Urban Afairs, U.S. Senate, February 16, 2005.”
https://www.federalreserve.gov/boarddocs/hh/2005/february/testimony.htm
(accessed February 26, 2017).
Federico, Giovanni. 2005. “Not Guilty? Agriculture in the 1920s and the Great De-
Copyright © 2021. Harvard University Press. All rights reserved.
pression.” Journal of Economic History 65:949–976.
Ferber, Robert. 1962. “Research on Household Behavior.” American Economic Review
52:19–63.
Feyrer, James, and Bruce Sacerdote. 2011. “Did the Stimulus Stimulate? Real Time
Estimates of the Efects of the American Recovery and Reinvestment Act.” Na-
tional Bureau of Economic Research Working Paper #16759. http://www.nber.org/
papers/w16759.
Field, Alexander. 2011. A Great Leap Forward: 1930s Depression and U.S. Economic
Growth. New Haven: Yale University Press.
Fischer, Stanley. 1977. “Long-Term Contracts, Rational Expectations, and the Opti-
mal Money Supply Rule.” Journal of Political Economy 85:191–205.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 865
———. 1993. “Te Role of Macroeconomic Factors in Growth.” Journal of Monetary
Economics 32:485–512.
Fisher, Frank. 1983. Disequilibrium Foundations of Equilibrium Economics. Cam-
bridge: University of Cambridge Press.
Fisher, Irving. 1896. Appreciation and Interest. New York: Macmillan.
———. 1911. “Recent Changes in Price Levels and Teir Causes.” American Economic
Review 1:37–45.
———. 1930. Te Teory of Interest, as Determined by Impatience to Spend Income
and Opportunity to Invest It. New York: Macmillan. [Originally published in
1907.]
———. 1933. “Te Debt-Defation Teory of Great Depressions.” Econometrica
1:337–357. Reprinted in Review of the International Statistical Institute 1 (January
1934): 48–65.
Fleming, Michael. 2013. “Information on Dealer Activity in Specifc Treasury Issues
Now Available.” Posted on the Federal Reserve Bank of New York website, Lib-
erty Street Economics, August 26. http://libertystreeteconomics.newyorkfed.org/
2013/08/information-on-dealer-activity-in-specifc-treasury-issues-now-available
.html#.VAtXIvmwLYg (accessed August 26, 2013).
Fox, Justin. 2008a. “Te Comeback Keynes.” Time, October 23. content.time.com/
time/magazine/article/0,9171,1853302,00.html (accessed January 10, 2011).
———. 2008b. “Bob Lucas on the Comeback of Keynesianism.” Time, October 28.
http://curiouscapitalist.blogs.time.com/2008/10/28/bob-lucas-on-the-come
back-of-keynesianism/#ixzz1AfQ4lkta (accessed January 10, 2011).
Frank, Robert. 1999. Luxury Fever: Money and Happiness in an Era of Excess. New
York: Free Press.
Frank, Robert, and Ben Bernanke, with the assistance of Roger Kaufman. 2007. Prin-
ciples of Economics. 3rd ed. New York: McGraw-Hill.
Freeman, Richard, and James Medof. 1984. What Do U nions Do? New York: Basic
Books.
Friedman, Benjamin. 1988. Day of Reckoning: Te Consequences of American Eco-
nomic Policy. New York: Random House.
———. 1999. “Te Future of Monetary Policy: Te Central Bank as an Army with
Copyright © 2021. Harvard University Press. All rights reserved.
Only a Signal Corps?” International Finance 2:321–338.
———. 2005. Te Moral Consequences of Economic Growth. New York: Knopf.
Friedman, Benjamin, and Kenneth Kuttner. 2011. “Implementation of Monetary Pol-
icy: How Do Central Banks Set Interest Rates?” In Handbook of Monetary Eco-
nomics, vol. 3B, ed. Benjamin M. Friedman and Michael Woodford, pp. 1345–
1438. Te Netherlands: North-Holland.
Friedman, Milton. 1953. “Te Methodology of Positive Economics.” In M. Friedman,
Essays in Positive Economics, pp. 3–43. Chicago: University of Chicago Press.
———. 1956. “Te Quantity Teory of Money—A Restatement.” In Studies in the
Quantity Teory of Money, ed. M. Friedman. Chicago: University of Chicago
Press.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
866 References
———. 1957. A Teory of the Consumption Function. Princeton: Princeton University
Press.
———. 1961. “Te Lag in Efect of Monetary Policy.” Journal of Political Economy
69:447–466.
———. 1962. Capitalism and Freedom. Chicago: University of Chicago Press.
———. 1968. “Te Role of Monetary Policy.” American Economic Review 58:1–17.
———. 1969. “Te Optimum Quantity of Money.” In M. Friedman, Te Optimum
Quantity of Money and Other Essays. Chicago: Aldine.
———. 1970. “A Teoretical Framework for Monetary Analysis.” Journal of Political
Economy 78:193–238.
———. 1983. “Washington: Less Red Ink.” Te Atlantic, February. https://www
.theatlantic.com/magazine/archive/1983/02/washington-less-red-ink/305450/
(accessed January 15, 2018).
———. 1987. “Te Quantity Teory of Money.” In Te New Palgrave: A Dictionary of
Economics, vol. 4, ed. J. Eatwell, M. Milgate, and P. Newman, pp. 3–20. London
and Basingstoke: Macmillan.
Friedman, Milton, and Anna Schwartz. 1963a. A Monetary History of the United
States, 1867–1960. Princeton: Princeton University Press.
———. 1963b. “Money and Business Cycles.” Review of Economics and Statistics 45
(1), pt. 2, Supplement (February): 59–63.
Froot, Kenneth. 1989. “New Hope for the Expectations Hypothesis of the Term
Structure of Interest Rates.” Journal of Finance 44:283–305.
Galbraith, John Kenneth. 1967. Te New Industrial State. Boston: Houghton-
Mifin.
———. 1981. A Life in Our Times. Boston: Houghton-Miffl in.
———. 1998. Te Af fl uent Society. Boston: Houghton-Miffl in. [First published in
1958.]
Geithner, Timothy. 2014. Stress Test: Refections on Financial Crises. New York:
Crown.
Gellner, Ernest. 1983. Nations and Nationalism. Ithaca, N.Y.: Cornell University
Press.
Gibson, Campbell, and Kay Jung. 2006. “Historical Census Statistics on the Foreign-
Copyright © 2021. Harvard University Press. All rights reserved.
Born Population of the United States: 1850–2000.” Washington, D.C.: U.S. Bureau
of the Census. https://www.census.gov/population/www/documentation/twps
0081/twps0081.html (accessed May 25, 2020).
Goldfeld, Stephen, and Daniel Sichel. 1990. “Te Demand for Money.” In Handbook
of Monetary Economics, vol. 1, ed. Benjamin Friedman and Frank Hahn, pp. 300–
356. Amsterdam: Elsevier.
Goodfriend, Marvin, and Monica Hargraves. 1983. “A Historical Assessment of the
Rationales and Functions of Reserve Requirements.” Economic Review, Federal
Reserve Bank of Richmond (March–April): 3–21.
Gordon, Robert. 1984. “Supply Shocks and Monetary Policy Revisited.” American
Economic Review 74:38–43.
———. 2013. “Te Phillips Curve Is Alive and Well: Infation and the NAIRU During
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 867
the Slow Recovery.” NBER Working Paper 19390. http://www.nber.org/papers/
w19390 (accessed October 26, 2015).
———. 2014. “Te Turtle’s Progress: Secular Stagnation Meets the Headwinds.” In
Secular Stagnation: Facts, Causes, and Cures, ed. C. Teulings and R. Baldwin,
pp. 47–60. London: CEPR Press.
———. 2015. Te Rise and Fall of American Growth. Princeton: Princeton University
Press.
Gordon, Robert, and Robert Krenn. 2010. “Te End of the Great Depression 1939–
41: Policy Contributions and Fiscal Multipliers.” NBER Working Paper 16380.
Cambridge, Mass.: National Bureau of Economic Research.
Gramlich, Edward. 1978. “State and Local Budgets the Day afer It Rained: Why Is
the Surplus So High?” Brookings Papers on Economic Activity 1978 (1):191–216.
———. 1979. “Macro Policy Responses to Price Shocks.” Brookings Papers on Eco-
nomic Activity 1979 (1):125–166.
Grigoli, Francesco, Alexander Hermana, and Klaus Schmidt-Hebbel. 2016. “Te Im-
pact of Terms of Trade and Macroeconomic Regimes on Private Saving.” Econom-
ics Letters 145:172–175. http://dx.doi.org/10.1016/j.econlet.2016.06.020 (accessed
November 29, 2017).
Gürkaynak, Refet, Brian Sack, and Jonathan H. Wright. 2006. “Te U.S. Treasury
Yield Curve: 1961 to the Present.” Finance and Economics Discussion Series,
Federal Reserve Board, Washington, D.C.
Gürkaynak, Refet, and Jonathan Wright. 2012. “Macroeconomics and the Term
Structure.” Journal of Economic Literature 50:331–367.
Gwartney, James, Richard Stroup, Russell Sobel, and David Macpherson. n.d. “Are
We Running Out of Resources?” http://commonsenseeconomics.com/wp-con
tent/uploads/CSE_H_Reading-Gwartney_Stroup_Sobel_Macpherson-Are_We
_Running_Out_of_Resources.pdf (accessed March 14, 2016).
Haberler, Gottfried. 1939. Prosperity and Depression: A Teoretical Analysis of Cycli-
cal Movements. Rev. ed. Geneva: League of Nations.
———. 1941. Prosperity and Depression: A Teoretical Analysis of Cyclical Movements.
3rd ed. Geneva: League of Nations.
———. 1946. “Te Place of the General Teory of Employment, Interest, and Money in
Copyright © 2021. Harvard University Press. All rights reserved.
the History of Economic Tought.” Review of Economics and Statistics 28:187–
194.
Hager, Sandy. 2016. Public Debt, Inequality, and Power: Te Making of a Modern Debt
State. Oakland, Calif.: University of California Press.
Hall, Robert. 1978. “Stochastic Implications of the Life Cycle-Permanent Income
Hypothesis: Teory and Evidence.” Journal of Political Economy 86:971–987.
Hansen, Alvin. 1938. Full Recovery or Stagnation? New York: Norton.
———. 1941. Fiscal Policy and Business Cycles. New York: Norton.
———. 1953. Guide to Keynes. New York: McGraw-Hill.
Harris, Donald. 1978. Capital Accumulation and Income Distribution. Stanford, Calif.:
Stanford University Press.
Harrod, Roy. 1937. “Mr. Keynes and Traditional Teory.” Econometrica 5:74–86.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
868 References
———. 1939. “An Essay in Dynamic Teory.” Economic Journal 49:14–33.
———. 1948. Towards a Dynamic Economics: Some Recent Developments of Economic
Teory and Teir Application to Policy. London: Macmillan.
Hausman, Joshua K., Paul W. Rhode, and Johannes F. Wieland. 2019. “Recovery from
the Great Depression: Te Farm Channel in Spring 1933.” American Economic
Review 109:427–472.
Hawtrey, Ralph. 1937. Capital and Employment. London: Longmans, Greene.
Henderson, David. 2009. “Why Everyone Read Samuelson.” Wall Street Journal, De-
cember 14.
Henderson, H. D. 1938. “Te Signifcance of the Rate of Interest.” Oxford Economic
Papers 1:1–13.
Herndon, Tomas, Michael Ash, and Robert Pollin. 2014. “Does High Public Debt
Consistently Stife Economic Growth? A Critique of Reinhart and Rogof.” Cam-
bridge Journal of Economics 38:257–279.
Hicks, John. 1937. “Mr Keynes and the Classics.” Econometrica 5:147–159.
———. 1946. Value and Capital. Oxford: Clarendon Press. [First published in 1939.]
———. 1967. “Monetary History and Teory—An Attempt at Perspective.” In J.
Hicks, Critical Essays in Monetary Teory, pp. 155–173. Oxford: Clarendon Press.
———. 1974. Te Crisis in Keynesian Economics. Oxford: Basil Blackwell.
Historical Statistics of the United States, Millennial Edition Online. n.d. Susan Carter
and Richard Sutch, General Editors. Cambridge University Press. https://hsus
.cambridge.org/HSUSWeb/HSUSEntryServlet (accessed various dates).
HM Treasury and Te Rt Hon George Osborne. 2015. Repayment of £2.6 Billion His-
torical Debt to Be Completed by Government, March 27. https://www.gov.uk/gov
ernment/news/repayment-of-26-billion-historical-debt-to-be-completed-by-gov
ernment (accessed April 25, 2018).
Hoover, Herbert. 1952. Te Memoirs of Herbert Hoover. Vol. 3, Te Great Depression
1929–1941. New York: Macmillan.
House of Representatives, Committee on the Budget. 2011. Path to Prosperity: Restor-
ing America’s Promise. Washington, D.C.: United States Congress.
Houthakker, Hendrik, and Lester Taylor. 1966. Consumer Demand in the United States,
1929–1970: Analyses and Projections. Cambridge, Mass.: Harvard University Press.
Copyright © 2021. Harvard University Press. All rights reserved.
Hume, David. n.d. “Of Money.” In Hume’s Political Discourses: With An Introduction
By William Bell Robertson, pp. 27–39. London and Felling-on-Tyne: Walter Scott.
[First published as Political Discourses in 1752.]
Humphrey, Tomas. 1992. “Te Real Bills Doctrine.” Economic Review, Federal Re-
serve Bank of Richmond (September–October): 3–13.
Innes, A. Mitchell. 1913. “What is Money?” Banking Law Journal 31 (May):
377–408.
———. 1914. “Te Credit Teory of Money.” Banking Law Journal 32 (February):
151–168.
Intergovernmental Panel on Climate Change. 2018. Global Warming of 1.5°C: Sum-
mary for Policymakers. http://www.ipcc.ch/ (accessed October 26, 2018).
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 869
International Energy Agency. 2015. Energy and Climate Change. Paris: International
Energy Agency.
Irons, John, and Josh Bivens. 2010. “Government Debt and Economic Growth: Over-
reaching Claims of Debt ‘Treshold’ Suffer from Teoretical and Empirical
Flaws.” Economic Policy Institute Briefng Paper #271, July.
Johnson, Harry. 1955. “Economic Expansion and International Trade.” Manchester
School of Economic and Social Studies 23:95–112.
———. 1961. “Te General Teory afer Twenty-Five Years.” American Economic Re-
view 51:1–17.
Jonas, Jiri. 2012. “Great Recession and Fiscal Squeeze at U.S. Subnational Govern-
ment Level.” International Monetary Fund Working Paper WP/12/184. Washing-
ton: International Monetary Fund.
Kahn, Richard. 1931. “Te Relation of Home Investment to Unemployment.” Eco-
nomic Journal 41:173–198.
Kahneman, Daniel, and Amos Tversky. 1979. “Prospect Teory: An Analysis of Deci-
sion under Risk.” Econometrica 47:263–291.
Kaldor, Nicholas. 1934. “A Classifcatory Note on the Determination of Equilibrium.”
Review of Economic Studies 1:122–136.
———. 1966. “Marginal Productivity and the Macro-Economic Teories of Distribu-
tion.” Review of Economic Studies 33:309–319.
Kalecki, Michał. 1944. “Professor Pigou on ‘Te Classical Stationary State,’ A Com-
ment.” Economic Journal 54:131–132.
———. 1971. “Outline of a Teory of the Business Cycle.” In Selected Essays on the
Dynamics of the Capitalist Economy, pp. 1–14. Cambridge: Cambridge University
Press. [Originally published in Polish in 1933.]
———. 1982. “Some Remarks on Keynes’s General Teory.” In Ferdinando Targetti
and Bogulslawa Kinda-Hass, “Kalecki’s Review of Keynes’s General Teory.” Aus-
tralian Economic Papers, pp. 244–260. [First published in Polish in Ekonomista,
1936.]
Kaletsky, Anatole. 2014. “Now May Not Be the Time to Buy Bonds.” Reuters, June 6.
http://blogs.reuters.com/anatole-kaletsky/2014/06/06/now-may-not-be-the-time
-to-buy-bonds/ (accessed August 19, 2014).
Copyright © 2021. Harvard University Press. All rights reserved.
Kane, Edward. 1970. “Te Term Structure of Interest Rates: An Attempt to Reconcile
Teaching with Practice.” Journal of Finance 25:361–374.
Kemmerer, Edwin. 1911. “Money and Prices: Discussion.” American Economic Re-
view 1:52–58.
Keynes, John Maynard. 1914. “Review of A Mitchell Innes, ‘What is Money?’” Eco-
nomic Journal 24:419–421.
———. 1919. Te Economic Consequences of the Peace. London: Macmillan.
———. 1921. A Treatise on Probability. London: Macmillan.
———. 1930. A Treatise on Money. London: Macmillan.
———. 1931a. “A Programme for Expansion.” In Essays in Persuasion, pp. 118–134.
London: Macmillan. [Condensed from “Can Lloyd George Do It?, a campaign
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
870 References
tract written jointly with H. D. Henderson in support of the Liberal Party in
1929.]
———. 1931b. “Economic Possibilities for Our Grandchildren.” In Essays in Persua-
sion, pp. 358–373. London: Macmillan. [First published in Te Nation and Ath-
enaeum, October 11 and 18, 1930.]
———. 1933. Te Means to Prosperity. London: Macmillan. http://www.gutenberg.ca/
ebooks/keynes-means/keynes-means-00-h.html (retrieved July 23, 2015).
———. 1934. “Is the Economic System Self-Adjusting?” Listener (November 21):
850–851. Reprinted, with editorial changes, as “A Self-Adjusting Economic Sys-
tem?” New Republic (February 20, 1935): 35–37.
———. 1936. Te General Teory of Employment, Interest and Money. London: Mac-
millan.
———. 1937a. “Te General Teory of Employment.” Quarterly Journal of Economics
51:209–223.
———. 1937b. “Alternative Teories of the Rate of Interest.” Economic Journal
47:241–252.
———. 1939. “Relative Movements of Real Wages and Output.” Economic Journal
49:34–41.
———. 1940. How to Pay for the War: A Radical Plan for the Chancellor of the Exche-
quer. London: Macmillan.
———. 1973a. Collected Writings of John Maynard Keynes. Vol. 13, Te General Te-
ory and Afer, Part 1, Preparation. Ed. Donald Moggridge. London and New York:
Macmillan and Cambridge University Press.
———. 1973b. Te Collected Writings of John Maynard Keynes. Vol. 14, Te General
Teory and Afer, Part II, Defence and Development. Ed. Donald Moggridge. Lon-
don and New York: Macmillan.
———. 1980. Collected Works of John Maynard Keynes. Vol. 27, Activities 1940–6:
Shaping the Post-War World: Employment and Commodities. Ed. Donald
Moggridge. London and New York: Macmillan and Cambridge University
Press.
———. 1981. Collected Works of John Maynard Keynes. Vol. 19, Activities 1922–9: Te
Return to Gold and Industrial Policy. Ed. Donald Moggridge. London and New
Copyright © 2021. Harvard University Press. All rights reserved.
York: Macmillan and Cambridge University Press.
———. 1982. Te Collected Writings of John Maynard Keynes. Vol. 21, Activities 1931–
9; World Crises and Policies in Britain and America. Ed. Donald Moggridge. Lon-
don and New York: Macmillan and Cambridge University Press.
Khan, Moshin, and Abdelhak Senhadji. 2001. “Treshold Efects in the Relationship
Between Infation and Growth.” IMF Staf Papers 48:1–21.
Klein, Lawrence. 1947. Te Keynesian Revolution. New York: Macmillan.
Knapp, Georg Friedrich. 1924. Te State Teory of Money. Trans. H. M. Lucas and
J. Bonar. London: Macmillan. [First published as Staatliche Teorie des Geldes.
Leipzig: Duncker and Hublot, 1905.]
Knight, Frank. 1921. Risk, Uncertainty and Proft. Boston: Houghton Miffl in.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 871
———. 1937. “Unemployment: And Mr. Keynes’s Revolution in Economic Teory.”
Canadian Journal of Economics and Political Science 3:100–123.
Koesler, Simon, and Michael Schymura. 2015. “Substitution Elasticities in a Constant
Elasticity of Substitution Framework—Empirical Estimates Using Nonlinear
Least Squares.” Economic Systems Research 27:101–121.
Koopmans, Tjalling. 1957. Tree Essays on the State of Economic Science. New York:
McGraw-Hill.
Krishnamurthy, Arvind, and Annette Vissing-Jorgensen. 2012. “Te Aggregate De-
mand for Treasury Debt.” Journal of Political Economy 120:233–267.
Krugman, Paul. 2009. Te Return of Depression Economics and the Crisis of 2008.
New York: Norton.
———. 2014. “Four Observations on Secular Stagnation.” In Secular Stagnation:
Facts, Causes, and Cures, ed. C. Teulings and R. Baldwin, pp. 61–68. London:
CEPR Press.
Krugman, Paul, and Robin Wells. 2018. Economics. 5th ed. New York: Worth.
Kuznets, Simon, assisted by Lillian Epstein and Elizabeth Jenks. 1941a. National In-
come and Its Composition, 1919–1938. New York: National Bureau of Economic
Research.
Kuznets, Simon. 1941b. “Capital Formation, Past and Present.” Paper presented at the
meeting of the American Statistical Association, December 1941.
Kydland, Finn, and Edward Prescott. 1982. “Time to Build and Aggregate Fluctua
tions.” Econometrica 50:1345–1370.
Lafer, Arthur. 2004. “Te Lafer Curve: Past, Present, and Future.” Heritage Founda-
tion Backgrounder #1765. http://www.heritage.org/research/reports/2004/06/
the-lafer-curve-past-present-and-future (accessed July 23, 2015).
Laidler, David. 1999. Fabricating the Keynesian Revolution. Cambridge: Cambridge
University Press.
Laughlin, J. Laurence. 1911. “Causes of the Changes in Prices Since 1896.” American
Economic Review 1:26–36.
Lavoie, Marc. 2014. Post-Keynesian Economics: New Foundations. Cheltenham, UK:
Edward Elgar.
Leduc, Sylvain, and Daniel Wilson. 2017. “Are State Governments Roadblocks to
Copyright © 2021. Harvard University Press. All rights reserved.
Federal Stimulus? Evidence on the Flypaper Efect of Highway Grants in the 2009
Recovery Act.” American Economic Journal: Economic Policy 9:253–292.
Leijonhufvud, Axel. 1968. On Keynesian Economics and the Economics of Keynes.
New York: Oxford University Press.
Lerner, Abba. 1941. “Te Economic Steering Wheel: Te People’s New Clothes.” Uni-
versity [of Kansas City] Review (June): 2–8.
———. 1943. “Functional Finance and the Federal Debt.” Social Research 10:38–51.
———. 1944. Te Economics of Control: Principles of Welfare Economics. New York:
Macmillan.
———. 1951. Economics of Employment. New York: McGraw-Hill.
———. 1978. “Keynesianism: Alive, If Not So Well.” In Fiscal Responsibilities in a
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
872 References
Constitutional Democracy, ed. James Buchanan and Richard Wagner. Boston:
Martinus Nijhof.
Lewis, W. Arthur. 1954. “Economic Development with Unlimited Supplies of La-
bour.” Manchester School of Economic and Social Studies 22:139–191.
Lo, Stephanie, and Ken Rogof. 2015. “Secular Stagnation, Debt Overhang and Other
Rationales for Sluggish Growth, Six Years On.” BIS Working Papers No 482, Janu-
ary.
Lowenstein, Roger. 2015. America’s Bank: Te Epic Struggle to Create the Federal Re-
serve. New York: Penguin.
Lucas, Robert. 1977. “Understanding Business Cycles.” Carnegie-Rochester Conference
Series on Public Policy 5:7–29.
———. 1981. “Tobin and Monetarism: A Review Article.” Journal of Economic Litera-
ture 19:558–567.
———. 2013. “Te Death of Keynesian Economics.” (A lecture delivered to the an-
nual management conference held at the Graduate School of the University of
Chicago in 1979.) In Lucas, Collected Papers on Monetary Teory, ed. Max Gill-
man. Cambridge, Mass.: Harvard University Press.
Madsen, Jakob. 2001. “Agricultural Crises and the International Transmission of the
Great Depression.” Journal of Economic History 61:327–365.
———. 2004a. “Price and Wage Stickiness During the Great Depression.” European
Review of Economic History 8:263–295.
———. 2004b. “Te Length and the Depth of the Great Depression: An International
Comparison.” Research in Economic History 22:239–288.
Magliocca, Gerard. 2011. “Te Gold Clause Cases and Constitutional Necessity.”
Florida Law Review 63:1–34.
Malinvaud, Edmond. 1977. Te Teory of Unemployment Reconsidered. Oxford: Basil
Blackwell.
———. 1980. Proftability and Unemployment. Cambridge: Cambridge University
Press.
Malthus, Tomas. 1798. An Essay on the Principle of Population, as It Afects the Fu-
ture Improvement of Society. London: Printed for J. Johnson, in St. Paul’s Church-
Yard.
Copyright © 2021. Harvard University Press. All rights reserved.
Mandeville, Bernard. 1988. Fable of the Bees, or Private Vices, Publick Benefts. Vol. 1.
Ed. F Kaye. Indianapolis: Liberty Classics. [First published in 1714; the Kaye edi-
tion was frst published in 1924.]
Mankiw, N. Gregory. 1997. “Comment.” In Reducing Infation: Motivation and Strat-
egy, ed. Christina Romer and David Romer, pp. 65–69. Chicago: University of
Chicago Press.
———. 2018. Principles of Economics. 8th ed. Boston: Cengage.
Marcuzzo, Maria Cristina. 2010. “Whose Welfare State? Beveridge versus Keynes.” In
No Wealth but Life: Welfare Economics and the Welfare State in Britain, 1880–
1945, ed. R. Backhouse and T. Nishizawa, pp. 189–206. Cambridge: Cambridge
University Press.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 873
Marglin, Stephen. 1970. “A Reformulation and Extension of Keynesian Teory.” Eco-
nomic Journal 80:910–931.
———. 1984. Growth, Distribution, and Prices. Cambridge, Mass.: Harvard University
Press.
———. 1990. “Losing Touch: Te Cultural Conditions of Worker Accommodation
and Resistance.” In Dominating Knowledge: Development, Culture, and Resistance,
ed. F. Apfel-Marglin and Stephen A. Marglin, pp. 217–282. Oxford: Clarendon
Press.
———. 2008. Te Dismal Science: How Tinking Like an Economist Undermines Com-
munity. Cambridge, Mass.: Harvard University Press.
———. 2017. “Wages, Prices, and Employment in a Keynesian Long Run.” Review of
Keynesian Economics 5 (Autumn): 360–425.
Marglin, Stephen, and Amit Bhaduri. 1990. “Proft Squeeze and Keynesian Teory.”
In Te Golden Age of Capitalism, ed. S. Marglin and J. Schor, pp. 153–186. Ox-
ford: Clarendon Press.
Marglin, Stephen, and Peter Spiegler. 2013. “Unpacking the Multiplier: Making Sense
of Recent Assessments of Fiscal Stimulus Policy.” Social Research 80:819–854.
Margo, Robert. 1993. “Employment and Unemployment in the 1930s.” Journal of
Economic Perspectives 7:41–59.
Marschak, Jacob. 1951. Income, Employment, and the Price Level: Notes on Lectures
Given at the University of Chicago, Autumn, 1948 and 1949. New York: Kelley.
Marshall, Alfred. 1920. Principles of Economics: An Introductory Volume. 8th ed. Lon-
don: Macmillan. [First published in 1890.]
Martin, Percival. 1924. Te Flaw in the Price System. London: King.
Marx, Karl. 1959. Capital. Vol. 1, A Critical Analysis of Capitalist Production. Mos-
cow: Foreign Languages Publishing House. [Originally published in German in
1867.]
Mas-Colell, Andreu, Michael Whinston, and Jerry Green. 1995. Microeconomic Te-
ory. New York: Oxford University Press.
McCulloch, J. Huston. 1971. “Measuring the Term Structure of Interest Rates.” Jour-
nal of Business 44:19–31.
———. 1975. “Te Tax-Adjusted Yield Curve.” Journal of Finance 30:811–830.
Copyright © 2021. Harvard University Press. All rights reserved.
Meade, J. E. 1937. “A Simplifed Model of Mr. Keynes’ System.” Review of Economic
Studies 4:98–107.
Meade, J. E., and P. W. S. Andrews. 1938. “Summary of Replies to Questions on Ef-
fects of Interest Rates.” Oxford Economic Papers 1:14–31.
Mill, John Stuart. 1909. Principles of Political Economy, With Some of Teir Applica-
tions to Social Philosophy. Ed., with an introduction, W. J. Ashley. London: Long-
mans Green. [First published in 1848.]
Miron, Jefrey, and Christina Romer. 1990. “A New Monthly Index of Industrial Pro-
duction, 1884–1940.” Journal of Economic History 50:321–337.
Mishkin, Frederic. 1978. “Te Household Balance Sheet and the Great Depression.”
Journal of Economic History 38:918–937.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
874 References
Mittelstaedt, Martin. 2012. “Why CP’s Old-Time Bondholders Have a Big Say in the
Future.” Toronto Globe and Mail, May 9. http://www.theglobeandmail.com/
globe-investor/why-cps-old-time-bondholders-have-a-big-say-in-the-future/
article4105862/ (accessed December 9, 2014).
Modigliani, Franco. 1944. “Liquidity Preference and the Teory of Interest and
Money.” Econometrica 12:45–88.
Modigliani, Franco, and Richard Brumberg. 1954. “Utility Analysis and the Con-
sumption Function: An Interpretation of Cross-Section Data.” In Post Keynesian
Economics, ed. Kenneth K. Kurihara. New Brunswick, N.J.: Rutgers University
Press.
Modigliani, Franco, and Richard Sutch. 1966. “Innovations in Interest Rate Policy.”
American Economic Review 56:178–197.
Nakamoto, Michiyo, and David Wighton. 2007. “Citigroup Chief Stays Bullish on
Buy-Outs.” Financial Times, July 9. http://www.f.com/intl/cms/s/0/80e2987a-2e
50-11dc-821c-0000779fd2ac.html#axzz3cQ8HaX00
National Association of State Budget Of ficers (NASBO). 2009. Te Fiscal Survey of
States, December. Washington, D.C.
———. 2012. Te Fiscal Survey of States, Spring. Washington, D.C.
National Commission on Fiscal Responsibility and Reform. 2010. Moment of Truth:
Report of the National Commission on Fiscal Responsibility and Reform. Washing-
ton, D.C.: Te White House.
New York Times. 2010. “Te Truth About the Defcit.” An editorial from February 7.
http://www.nytimes.com/2010/02/07/opinion/07sun1.html?scp=2&sq=+budget
&st=nyt (accessed February 19, 2010).
Nord, Mark, and Mark Prell. 2011. “Food Security Improved Following the 2009
ARRA Increase in SNAP Benefts.” ERR-116, U.S. Department of Agriculture,
Economic Research Service, April. www.ers.usda.gov (accessed March 31, 2016).
O’Donnell, Rod. 1999. “Keynes’s Socialism: Conception, Strategy and Espousal.” In
Keynes, Post Keynesianism and Political Economy: Essays in Honour of Geof Har-
court, Volume III, ed. Claudio Sardoni and Peter Kriesler, pp. 149–175. New York:
Routledge.
O’Driscoll, Gerald. 1976. “Te Ricardian Nonequivalence Teorem.” Journal of Politi
Copyright © 2021. Harvard University Press. All rights reserved.
cal Economy 85:207–210.
Parker, Jonathan A. 2011. “On Measuring the Efects of Fiscal Policy in Recessions.”
Journal of Economic Literature 49:703–718.
Pasinetti, Luigi. 1962. “Rate of P
roft and Income Distribution in Relation to the Rate
of Economic Growth.” Review of Economic Studies 29:267–279.
Patinkin, Don. 1948. “Price Flexibility and Full Employment.” American Economic
Review 38:543–564.
———. 1949. “Te Indeterminacy of Absolute Prices in Classical Economic Teory.”
Econometrica 17:1–27.
———. 1951. “Price Flexibility and Full Employment.” In Readings in Monetary Te-
ory, ed. F. A. Lutz and L. W. Mints. Homewood, Ill.: Irwin. [Revised version of
Patinkin (1948).]
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 875
———. 1965. Money, Interest, and Prices: An Integration of Monetary and Value Te-
ory. 2nd ed. New York: Harper and Row. [First published in 1956.]
Paulson, Henry. 2010. On the Brink: Inside the Race to Stop the Collapse of the Global
Financial System. New York: Business Plus.
Phelps, Edmund. 1968. “Money-Wage Dynamics and Labor-Market Equilibrium.”
Journal of Political Economy 76:678–711.
Phillips, A. William. 1958. “Te Relation between Unemployment and the Rate of
Change of Money Wage Rates in the United Kingdom, 1861–1957.” Economica
25:283–299.
Pigou, Arthur. 1927. “Wage Policy and Unemployment.” Economic Journal 37:353–368.
———. 1933. Te Teory of Unemployment. London: Macmillan.
———. 1943. “Te Classical Stationary State.” Economic Journal 53:343–351.
———. 1947. “Economic Progress in a Stable Environment.” Economica 14:180–188.
Piketty, Tomas. 2014. Capital in the Twenty-First Century. Trans. Arthur Goldham-
mer. Cambridge, Mass.: Harvard University Press.
———. 2015. “About Capital in the 21st Century.” American Economic Review
105:48–53.
———. 2020. Capital and Ideology. Trans. Arthur Goldhammer. Cambridge, Mass.:
Harvard University Press.
Piketty, Tomas, and Emmanuel Saez. 2015. “Wealth and Inheritance in the Long
Run.” In Handbook of Income Distribution, vol. 2, ed. Anthony Atkinson and
François Bourguignon. Amsterdam: North Holland.
Plassard, Romain. 2018. “Clower’s Volte-Face Regarding the ‘Keynesian Revolution.’”
History of Political Economy 50:261–287.
Pryor, Frederic. 2007. “Immiserizing Growth as Seen by Bhagwati, Samuelson, and
Others.” Journal of Economic Education 38:208–214.
Ramey, Valerie A. 2011. “Can Government Purchases Stimulate the Economy.” Jour-
nal of Economic Literature 49:673–685.
Ramsey, Frank. 1980. “Truth and Probability.” In Studies in Subjective Probability, 2nd
ed., ed. Henry Kyburg, Jr. and Howard Smokker, pp. 23–52. Huntington, N.Y.:
Krieger. [First published in 1926.]
Reinhart, Carmen, and Kenneth Rogof. 2010. “Growth in a Time of Debt.” NBER
Copyright © 2021. Harvard University Press. All rights reserved.
Working Paper 15639. [Condensed in American Economic Review 100 (2010):
573–578.]
Reinhart, Carmen, Vincent Reinhart, and Kenneth Rogof. 2012. “Public Debt Over-
hangs: Advanced-Economy Episodes since 1800.” Journal of Economic Perspec-
tives 26:69–86.
Ricardo, David. 1810. “Essay on the Funding System.” Supplement to the 4th, 5th,
and 6th editions of Te Encyclopedia Britannica. Online edition downloaded
from www.amazon.com, August 17, 2015.
———. 1951. On the Principles of Political Economy and Taxation. Vol. 1 of Te Works
and Correspondence of David Ricardo, ed. Piero Srafa with the collaboration of
M. H. Dobb. Cambridge: Cambridge University Press. [On the Principles of Politi
cal Economy and Taxation was frst published in 1817.]
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
876 References
Robertson, Dennis. 1935. Personal correspondence, to Keynes, February 3, 1935,
commenting on a draf of Te General Teory. In Keynes 1973a.
———. 1937. [Alternative Teories of the Rate of Interest, II.] In Bertil Ohlin, Dennis
Robertson, and Ralph Hawtrey, “Alternative Teories of the Rate of Interest:
Tree Rejoinders.” Economic Journal 47:423–443.
Robinson, Joan. 1933a. “A Parable on Savings and Investment.” Economica (39): 75–
84.
———. 1933b. “Te Teory of Money and the Analysis of Output.” Review of Eco-
nomic Studies 1:22–26.
———. 1962. Essays in the Teory of Economic Growth. London: Macmillan.
———. 1966. Te Accumulation of Capital. 2nd ed. London: Macmillan. [First pub-
lished in 1956.]
———. 1969. Te Economics of Imperfect Competition. 2nd ed. London: Macmillan.
[First published in 1933.]
———. 1980. “History vs Equilibrium.” Collected Economic Papers. Vol. 5. Cam-
bridge: Basil Blackwell; and Cambridge, Mass.: MIT Press.
Rogof, Kenneth. 2013. FAQ on Herndon, Ash and Pollin’s Critique of “Growth in a
Time of Debt,” October 1. Harvard University. https://scholar.harvard.edu/fles/
rogof/fles/faq_on_hap_update.pdf (accessed January 15, 2018).
Romer, Christina. 1986. “Spurious Volatility in Historical Unemployment Data.”
Journal of Political Economy 94:1–37.
Romer, Christina, and Jared Bernstein. 2009. “Te Job Impact of the American Re-
covery and Reinvestment Plan.” White House Council of Economic Advisers.
Washington, D.C. http://otrans.3cdn.net/45593e8ecbd339d074_l3m6bt1te.pdf
(accessed August 8, 2012).
Roosevelt, Franklin D. 1933. “Franklin D. Roosevelt’s First Inaugural Address,”
March 4. National Archives Catalog. https://catalog.archives.gov/id/197333
(accessed September 28, 2018).
Rowthorn, Bob. 1982. “Demand, Real Wages, and Economics Growth.” Studi Econ-
omici 18:3–53.
Saez, Emmanuel. 2014. Income and Wealth Inequality: Evidence and Policy Implica-
tions. Neubauer Collegium Lecture, University of Chicago. https://eml.berkeley
Copyright © 2021. Harvard University Press. All rights reserved.
.edu/~saez/lecture_saez_chicago14.pdf (accessed October 27, 2019).
Samuelson, Paul. 1941. “Stability of Equilibrium: Comparative Statics and Dynam-
ics.” Econometrica 9:97–120.
———. 1943. “Full Employment afer the War.” In Postwar Economic Problems, ed.
S. Harris, pp. 27–54. New York: McGraw-Hill.
———. 1946. “Lord Keynes and the General Teory.” Econometrica 14:187–200.
———. 1947. Foundations of Economic Analysis. Cambridge, Mass.: Harvard Univer-
sity Press.
———. 1948. Economics: An Introductory Analysis. New York: McGraw-Hill.
———. 1955. Economics: An Introductory Analysis. 3rd ed. New York: McGraw-Hill.
———. 1964. “A Brief Post-Keynesian Survey.” In Keynes’ General Teory: Reports of
Tree Decades, ed. Robert Lekachman. New York: St Martins.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 877
———. 1980. Economics. 11th ed. New York: McGraw-Hill.
Sargent, Tomas. 1972. “Rational Expectations and the Term Structure of Interest
Rates.” Journal of Money, Credit, and Banking 4:74–97.
Savage, Leonard J. 1954. Te Foundations of Statistics. New York: Wiley.
Schor, Juliet. 1998. Te Overspent American: Upscaling, Downscaling, and the New
Consumer. New York: Basic Books.
Schumpeter, Joseph. 1934. “Depressions.” In Brown, Chamberlin et al. 1934,
pp. 3–21.
———. 1942. Capitalism, Socialism and Democracy. New York: Harper and Brothers.
———. 1949. “Science and Ideology.” American Economic Review 39:346–359.
———. 1954. History of Economic Analysis. New York: Oxford University Press.
———. 1961. Te Teory of Economic Development: An Inquiry into Profts, Capital,
Credit, Interest, and the Business Cycle. Trans. Redvers Opie. Oxford: Oxford Uni-
versity Press. [First English publication in 1934. Originally published as Teorie
der Wirtschaflischen Entwicklung, Leipzig: Duncker and Hublot, 1912.]
Scitovsky [de Scitovszky], Tibor. 1940. “A Study of Interest and Capital.” Economica,
new series, 7:293–317.
Securities Industry and Financial Markets Association. 2018. Statistics. http://www
.sifma.org/research/statistics/ (accessed October 6, 2018).
———. 2020. Statistics. http://www.sifma.org/research/statistics/ (accessed May 12,
2020).
Shiller, Robert. 1997. “Why Do People Dislike Infation?” In Reducing Infation: Moti-
vation and Strategy, ed. Christina Romer and David Romer, pp. 13–65. Chicago:
University of Chicago Press.
———. 2017. “Narrative Economics.” American Economic Review 107:967–1004.
Shoag, Daniel. 2010. “Te Impact of Government Spending Shocks: Evidence on the
Multiplier from State Pension Plan Returns.” Unpublished working paper.
———. 2013. “Using State Pension Shocks to Estimate Fiscal Multipliers since the
Great Recession.” American Economic Review 103:121–124.
Sims, Christopher. 1980. “Macroeconomics and Reality.” Econometrica 48:1–48.
———. 2007. “On the Fit of New Keynesian Models: Comment.” Journal of Business
and Economic Statistics 25:152–154.
Copyright © 2021. Harvard University Press. All rights reserved.
Skidelsky, Robert. 2000. John Maynard Keynes vol. 3: Fighting for Britain, 1937–1946.
London: Macmillan.
———. 2009. Keynes: Te Return of the Master. New York: Public Afairs, 2009.
Smith, Adam. 1937. Te Wealth of Nations. Ed. E. Cannan. New York: Modern Li-
brary. [First published in 1776.]
———. 1982. Te Teory of Moral Sentiments. Ed. D. Raphael and A. Macfe. India-
napolis: Liberty Classics. [First published in 1759.]
Social Security Board. 1937. Unemployment Compensation: What and Why?, Publica-
tion No. 14. Washington, D.C.: United States Government Printing Of fi ce. http://
www.larrydewitt.net/SSinGAPE/UI1937book2.htm (accessed February 4, 2010).
Solow, Robert. 1956. “A Contribution to the Teory of Growth.” Quarterly Journal of
Economics 70:65–94.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
878 References
———. 1979. “Alternative Approaches to Macroeconomic Teory: A Partial View.”
Canadian Journal of Economics / Revue canadienne d’Economique 12:339–354.
———. 1990. Te Labor Market as a Social Institution. Cambridge, Mass.: Blackwell.
Solow, Robert, and Joseph Stiglitz. 1968. “Output, Employment, and Wages in the
Short Run.” Quarterly Journal of Economics 82:537–560.
Stack, Carol. 1975. All Our Kin: Strategies for Survival in a Black Community. New
York: Harper Colophon.
Statistical Abstracts of the United States. various dates. https://www.census.gov/
library/publications/time-series/statistical_abstracts.html (accessed various
dates).
Steindl, Josef. 1952. Maturity and Stagnation in American Capitalism. Oxford: Black-
well.
Stiglitz, Joseph. 1970. “A Consumption-Oriented Teory of the Demand for Finan
cial Assets and the Term Structure of Interest Rates.” Review of Economic Studies
37:321–351.
Summers, Lawrence. 2014. “Refections on the ‘New Secular Stagnation Hypothesis.’”
In Secular Stagnation: Facts, Causes, and Cures, ed. C. Teulings and R. Baldwin,
pp. 27–46. London: CEPR Press.
———. 2015. “Demand Side Secular Stagnation.” American Economic Review 105:60–
65.
———. 2016. “Te Age of Secular Stagnation: What It Is and What to Do About It.”
Foreign Afairs 95 (March–April): 2–9.
Surowiecki, James. 2011. “State of the U nions.” New Yorker, January 17. http://www
.newyorker.com/talk/fnancial/2011/01/17/110117ta_talk_surowiecki (accessed
February 10, 2011).
Sutch, Richard. 2018. “Reading Keynes at the Zero Lower Bound: Te Liquidity Trap,
the Great Depression, and Unconventional Policy.” Journal of the History of Eco-
nomic Tought 40:301–334.
Tarshis, Lorie. 1939. “Changes in Real and Money Wages.” Economic Journal 49:150–
154.
———. 1947. Te Elements of Economics. Boston: Houghton Miffl in.
Taylor, John. 1979. “Staggered Wage Setting in a Macro Model.” American Economic
Copyright © 2021. Harvard University Press. All rights reserved.
Review 69:108–113.
———. 1980. “Aggregate Dynamics and Staggered Contracts.” Journal of Political
Economy 88:1–23.
———. 1993. “Discretion Versus Policy Rules in Practice.” Carnegie-Rochester Confer-
ence Series on Public Policy 39:195–214.
———. 2011. “An Empirical Analysis of the Revival of Fiscal Activism in the 2000s.”
Journal of Economic Literature 49:686–702.
Tcherneva, Pavlina. 2007. “Chartalism and the Tax-Driven Approach to Money.” In A
Handbook of Alternative Monetary Economics, ed. Philip Arestis and Malcolm
Sawyer. Cheltenham, UK: Edward Elgar.
Temin, Peter. 1976. Did Monetary Forces Cause the Great Depression? New York: Nor-
ton.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
References 879
Temin, Peter, and Barrie Wigmore. 1990. “Te End of One Big Defation.” Explora-
tions in Economic History 27:483–502.
Tompson, E. P. 1963. Te Making of the English Working Class. New York: Random
House.
Tobin, James. 1947. “Money Wage Rates and Employment.” In Te New Economics,
ed. Seymour Harris. New York: Knopf.
———. 1956. “Te Interest-Elasticity of Transactions Demand for Cash.” Review of
Economics and Statistics 38:241–247.
———. 1958. Liquidity Preference as Behavior towards Risk.” Review of Economic
Studies 25:65–86.
———. 1969. “A General Equilibrium Approach to Monetary Teory.” Journal of
Money Credit and Banking 1:15–29.
———. 1975. “Keynesian Models of Recession and Depression.” American Economic
Review 65:195–202.
———. 1980. Asset Accumulation and Economic Activity: Refections on Contemporary
Macroeconomic Teory. Chicago: University of Chicago Press.
———. 1992. “An Old Keynesian Counterattacks.” Eastern Economic Journal 18 (Fall):
387–400.
———. 1993. “Price Flexibility and Output Stability: An Old Keynesian View.” Jour-
nal of Economic Perspectives 7:45–65.
Tucker, Paul. 2018. Unelected Power: Te Quest for Legitimacy in Central Banking and
the Regulatory State. Princeton: Princeton University Press.
Tversky, Amos, and Daniel Kahneman. 1981. “Te Framing of Decisions.” Science
211:453–458.
U.S. Department of the Treasury. n.d. “Corporate Bond Yield Curve Papers and
Data.” https://home.treasury.gov/data/treasury-coupon-issues-and-corporate
-bond-yield-curve/corporate-bond-yield-curve (accessed September 7, 2014).
———. 2008. “Treasury Announces Temporary Guarantee Program for Money Mar-
ket Funds,” September 29. https://www.treasury.gov/press-center/press-releases/
Pages/hp1161.aspx (accessed December 2, 2018).
U.S. Government Printing Of fi ce. 2008. “Te Financial Crisis and the Role of Federal
Regulators.” October 23. http://www.gpo.gov/fdsys/pkg/CHRG-110hhrg55764/
Copyright © 2021. Harvard University Press. All rights reserved.
html/CHRG-110hhrg55764.htm (accessed July 11, 2015).
Veritas Foundation. 1960. Keynes at Harvard. New York: Veritas Foundation.
Viner, Jacob. 1933a. Balanced Defation, Infation, or More Depression. Lecture deliv-
ered on February 20, 1933, Minneapolis, Minnesota. Minneapolis: University of
Minnesota Press.
———. 1933b. “Infation as a Possible Remedy for the Depression.” Proceedings of the
Institute of Public Afairs, University of Georgia, 7th Annual Meeting, May 8–16,
1933, Athens, Ga.
———. 1936. “Mr. Keynes on the Causes of Unemployment.” Quarterly Journal of
Economics 51:147–167.
Volcker, Paul. 2013. “Te Fed & Big Banking at the Crossroads.” New York Review of
Books, August 15.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
880 References
von Neumann, John, and Oskar Morgenstern. 1944. Te Teory of Games and Eco-
nomic Behavior. Princeton, N.J.: Princeton University Press.
Walras, Léon. 1954. Elements of Pure Economics, or Te Teory of Social Wealth.
Trans. W Jafé from the Édition Défnitive, Éléments de l’Économie Politque Pure,
Homewood, Ill., 1926. [First published in 1874.]
Watkins, Myron. 1933. “Te Literature of the Crisis.” Quarterly Journal of Economics
47:504–532.
Weber, Max. 1930. Te Protestant Ethic and the Spirit of Capitalism. Trans. Talcott
Parsons. New York: Scribner. [Originally published as “Die Protestantische Ethik
und Der ‘Geist’ des Kapitalismus.” Archiv fur Sozialwissenschaf und Sozialpolitik
20 (1905): 1–54; and 21 (1905): 1–110.]
Weir, David. 1992. “A Century of US Unemployment, 1890–1990: Revised Estimates
and Evidence for Stabilization.” Research in Economic History 14:301–346.
Wells Fargo. 2013. “How Much of Your Net Worth Should You Keep in Cash?”
https://conversations.wfmagazines.com/create_wealth/investment_planning/
article/how_much_of_your_net_worth_should_you_keep_in_cash/ (accessed
August 5, 2013).
———. 2019. “How Much Cash Should I Hold in My Portfolio?” https://www08
.wellsfargomedia.com/assets/pdf/personal/investing/investment-institute/how
%20much%20cash.pdf (accessed May 12, 2020).
Wicksell, Knut. 1936. Interest and Prices: A Study of the Causes Regulating the Value
of Money. Trans. Richard Kahn, London: Macmillan. [Originally published as
Geldzins und Güterpreise. Jena: Gustav Fischer, 1898.]
———. 1958. “Te Infuence of the Rate of Interest on Commodity Prices.” Trans.
Sylva Gethin. In Wicksell, Selected Papers on Economic Teory, ed. Erik Lindahl.
Cambridge, Mass.: Harvard University Press. [Originally published as “Pen-
ningeräntans Infytande på Varuprisen.” Nationalekonmiska Föreningens Förhand
lingar under Förra Halväret 1898, pp. 47–70. Stockholm, 1899.]
Wildavsky, Aaron. 1964. Te Politics of the Budgetary Process. Boston: Little Brown.
Wilson, Daniel. 2012. “Fiscal Spending Jobs Multipliers: Evidence from the 2009
American Recovery and Reinvestment Act.” American Economic Journal: Eco-
nomic Policy 4:251–282.
Copyright © 2021. Harvard University Press. All rights reserved.
World Bank. 2016. World Development Indicators. Washington, D.C.: World Bank.
http://databank.worldbank.org/data/reports.aspx?source=world-development-in
dicators (accessed November 27, 2016).
Wray, L. Randall. 2014. “From the State Teory of Money to Modern Money Teory:
An Alternative to Economic Orthodoxy.” Levy Economics Institute of Bard Col-
lege, Working Paper 792. Annandale on Hudson, N.Y.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
acknowledgments
My frst debt is to the students to whom I presented my argument as it was
developing. Teir questions at some points, their incomprehension at others,
and their encouragement at virtually every step were essential to this project.
One of these students, Sam Harland, became an invaluable research assistant
in my collaboration with Peter Spiegler assessing the Obama stimulus.
Duncan Foley read the entire manuscript in draf, and his questions and
comments refected the keen mind he has brought to economics for more
than half a century. Te book has benefted from his attention. Bob Rowthorn
also read the manuscript, and made excellent points, unfortunately, for the
most part too late in the process for me to incorporate them. Ben Friedman
read the chapters on liquidity-preference theory, chapters 11 and 12. John
Campbell read the empirical appendix to chapter 12. Both made helpful ob-
servations that saved me misunderstanding and error. Christine Desan prod-
ded me, both in conversation and in her own writings, to explore the assump-
tions about money in Te General Teory, and chapter 13 is much improved
as a result. Te assessment of the Obama stimulus in chapter 15 c omes out of
a project undertaken jointly with Peter Spiegler. It refects his skepticism
about the customary use of regression analysis to torture the data and the
consequent need to supplement it with more humane methods of interro
gation. Michael Ash made very helpful criticisms of an early draf of this
Copyright © 2021. Harvard University Press. All rights reserved.
chapter. I am also grateful to Professor Ash for pushing me to justify a funda-
mental assumption of this book: the importance of confronting mainstream
economics on its most favorable turf, the feld of perfect competition, rather
than accepting that the ills of capitalism, particularly unemployment, are due
to departures from the norms of perfect competition—frictions, rigidities,
and the like.
My editor at Harvard University Press, Ian Malcolm, his assistant, Olivia
Woods, and Cheryl Lincoln at Technologies ’N Typography helped transform
a large and complicated manuscript into what we all hope is a readable book.
Michael Aronson, my longtime editor at Harvard University Press, retired
881
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
882 Acknowledgments
in the middle of this project, but even afer his retirement has given me un-
failing support in many ways.
All of these people, I hope, have saved me from the fate that Keynes ob-
served (in the preface to Te General Teory) is too ofen the lot of theorists
operating in isolation: “It is astonishing,” he wrote, “what foolish things one
can temporarily believe if one thinks too long alone.”
Copyright © 2021. Harvard University Press. All rights reserved.
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
index
Abdelhak, Senhadji, 684 41, 664, 688–712, 771, 788–789, 791; main-
absolute needs, 324, 616, 847n12 stream view, 32, 54–55, 781–782; in
Achen, Christopher, 563, 575–576 Marschak’s model, 804n13; in Modigliani’s
Adrian, Tobias, 835–836n9 model, 33–34, 130–131, 224–226, 255–257,
Aga Khan, 327 276–277, 502; and money demand, 90,
aggregate demand (AD), 1–6; and adjustment 119–120, 518–519; and money supply,
process, 179–180, 210–211, 213, 220; and 133–135, 193–194; and money wage, 133–
Cambridge saving theory, 330–336, 368– 135, 502, 777–781; in neoclassical synthe-
371, 698–701; and capital deepening, 356– sis, 36; and New Classical theory, 36–37; in
359, 728–730; and capital widening, 349– New Keynesian theory, 37, 41; optimizing
356, 366–371, 699–701; and change in the the composition of, 594–602, 612–623,
price of energy, 691–695, 715–717, 718– 788; and paradox of thrif, 39; Patinkin on,
719, 724–726; Dennison on, 817n8; as a 26–27, 216; and price level, 90, 104–107,
determinant of equilibrium, 517, 519, 523; 145, 223, 227, 230, 497; and real-balance
in Domar’s model, 656–658, 661–662; in efect, 36, 145–153, 154–155, 234–244, 253,
Dos Santos Ferreira and Michel’s model, 257–259; in Robinson’s model, 670–673,
218; empirical relationship with infation, 676; Samuelson on, 205; and Say’s Law, 60–
755–764, 771, 790; as equality of expendi- 64, 68–69, 81, 378, 784; in second-pass
ture and income, 81–85, 183–185; Euro- model, 31, 99–108, 120–121; in Solow’s
pean Central Bank (ECB) on, 48; in frst- model, 658–662; Temin on, 816–817n5; in
pass model, 81–86, 90–92, 497; and fscal third-pass model, 194, 203–204, 208–212,
policy, 12–13, 42–43, 531–543, 785–788, 219–221; as understood in the 1950s and
790, 792; and Fisher efect, 250–252, 253– 1960s, 40–41; in World War II and post-
254, 259–260; in fxprice adjustment, 181– War planning, 46, 312–313
182, 204–211, 334–335, 354–356, 357–358, aggregate supply (AS), 39–41, 47, 71, 90,
497, 691–692, 702–711; in fexprice adjust- 129–131, 150, 521. See also goods supply;
Copyright © 2021. Harvard University Press. All rights reserved.
ment, 182–183, 197–204, 208–211, 334– labor supply
335, 354, 357–358, 497, 691–692; and Agricultural Adjustment Administration
fractional-reserve banking, 135–145, 226– (AAA), 289
230, 257; and government debt, 635–638; agriculture: in Great Depression, 8, 261–263,
in Great Depression, 274; in Harrod’s 266, 278–292, 300–304, 781–782, 797–
model, 647–656, 661–662; and income dis- 798n2, 817–819nn7–15, 819nn17–18,
tribution, 39, 330, 790–791; indeterminacy, 819n20; and measurement of unemploy-
519–522, 523, 784–785; and interest rate, ment, 540–541, 797n2; and reserve army,
28, 78, 92, 517–519; and investment de- 677–678, 851–852n2
mand, 110–118, 358–359; as locus of sta Ait-Sahalia, Yacine, 831n14
tionary labor:capital ratio (l), 664; as locus Akerlof, George, 325–326
of stationary output, 204–205; as locus of Allais, Maurice, 363
stationary prices, 198; in long run, 13–14, Amazon, 349
883
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
884 Index
American Recovery and Reinvestment Act Besomi, Daniele, 850n2
(ARRA). See Obama stimulus Bewley, Truman, 26, 843n14
Ando, Albert, 70, 78, 787, 827n4 Bhaduri, Amit, 686–687, 704–705, 790–791
Andrews, P. W. S., 843n14 Bhagwati, Jagdish, 819n17
animal spirits. See under investment Bibow, Jörg, 848–849n17
Arrow, Kenneth, 161, 164, 168, 774, 823– Biden, Joe, 544
824n10 Bivens, Josh, 846–847n10
Ash, Michael, 846–847n10 Blanchard, Olivier, 852n4
asset demand for money, 78, 79, 95, 105, 120, Blinder, Alan, 26, 51, 71, 163, 557, 768, 776,
131, 136–138, 145, 250, 403, 498, 506, 802–803n7, 824n1, 842n6, 843n14, 855n5
804n1, 806n11, 808n3 Boehner, John, 544
automatic stabilization, 12, 539–542, 633, 787 Boianovsky, Mauro, 71, 850–851n8
average maturity: of corporate bonds in bond efect, 106, 145, 226, 230, 257, 498
Moody’s index, 828n10; of Treasury bonds, bond price as random walk, 403
477–483, 492, 496, 831n17 bond-price fuctuations, 389–393, 402, 404–
406, 415, 423, 424, 430, 432–441, 456, 461,
Backhouse, Roger, 71, 850–851n8 469, 477–478, 482, 486, 490–491, 500–501,
Bakija, Jon, 848n14 503, 508–509, 513, 826n12. See also vol
balanced-budget multiplier, 535–537, 598, atility
608, 634. See also multiplier Brainard, William, 9, 346–348
Ball, Laurence, 815n2, 835–836n9 Buckley, William, 40
banana parable, 82–83, 198 budget defcit, 12–13, 39, 42–46, 235, 242,
bank holiday, 269 492, 531, 534–543, 556, 594–644, 664, 786–
bank reserves, 140–147, 227, 230, 233–235, 788, 820n2, 848–849n17; and crowding
253, 497, 833–834n6, 836n11; in Great out, 551–552, 559, 603–608, 609, 623, 626,
Depression, 270, 277–278, 295–298, 815– 641–644, 841–842n3; B. Friedman on,
816n2 (chap. 8), 834n6; afer Great Reces- 611–616, 620, 621–622, 631, 847n11,
sion, 515–519, 837n12. See also excess 848n14; in Great Depression, 272; Keynes
reserves on, 42, 97, 531, 805–806n9, 837–839n3;
bankruptcy, 236–243, 253, 258–259, 613, Lerner on, 12–13, 42, 531, 594, 595, 636–
835n9; in Great Depression, 268, 275, 284, 637, 837–839n3, 849n19; primary, 602, 626,
815n2 629–631, 634–636, 849n19; Ryan on, 609,
Baqaee, David Rezza, 661 622; Samuelson on, 820n2; Viner on, 56–59,
Barens, Ingo, 785, 807n12 529, 531. See also automatic stabilization
Barro, Robert, 636–638, 684, 775, 841n3 Bufett, Warren, 348
Baruch, Bernard, 44 bulls and bears, 398–400, 402–403
Bauer, Peter, 803n9 Bush, George H. W., 628
Baumol, William: on adjustment process, 163; Bush, George W., 540, 542, 628, 787, 842n7,
on determination of employment and out- 844n6
Copyright © 2021. Harvard University Press. All rights reserved.
put, 51, 802–803n7; on infation and wealth business failures in Great Depression, 266–
distribution, 768, 855n8; on inventory- 268, 275
theoretic model of money holdings, 511–
513, 524–526, 834–835n8; on loanable Calvo, Guillermo, 162
funds, 824n1; on Pigou’s forgetfulness, Cambridge saving theory, 9, 330–335, 337,
812–813n7; on Say’s Law, 802n4 338–339, 352, 368–371, 604, 670, 698–705,
behavioral economics, 329, 823 720–727, 811n12, 821n12; in Pasinetti–
Béraud, Alain, 802n14 Piketty model, 733–745
Berkshire Hathaway, 348 Campbell, John, 328, 432, 457–458, 480,
Bernanke, Ben, 51, 64–69, 111, 179, 240, 514, 827n2, 829n1, 831n14
626, 802–803n6, 835–836n9, 839–840n7, capacity-augmenting investment. See capital
850–851n8 widening
Bernstein, Jared, 843n6 capital controversies, 661
Beveridge, William, 856 capital deepening, 9–10, 13, 349–352, 356–
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Index 885
359, 372–375, 687–688, 705–712, 727–733, Coates, Ta-Nehisi, 615
762, 764, 791–792, 822nn4–5 cobweb theorem, 808n1
capital widening, 9–10, 13, 349–355, 364, Cochrane, John, 841n3
367–371, 604, 606–608, 642–644, 670, Cogan, John, 12, 547–548, 552–553, 561–564,
687, 698–705, 707–708, 710–712, 720–727, 570–571, 572–574, 577, 582, 584, 788, 841–
731, 736–745, 761, 791–792, 822nn4–5, 842n3
855n5 Colander, David, 33, 34, 40, 222–223, 691,
capitalism lef to its own devices, 11, 99–108, 779, 818n11, 837–839n3, 849n20
214, 521, 523. See also invisible hand; self- collateralized debt obligations (CDOs), 140,
regulating economy 197, 508
Cato Institute, 841n2 collective goods, 595, 599, 602, 612, 614, 620,
central bank: and commitment, 426–429; and 623, 625, 845–846n2
confict between Main Street and Wall commercial, industrial, agricultural, and con-
Street, 772; dot plot, 486–487; Eichengreen sumer (CIAC) loans in Great Depression,
on, 816n4; and fractional-reserve banking, 271–272, 275, 297–298
833n5; in Great Depression, 278, 812n4; in commercial loans, 138–146, 151, 193, 232–
Great Recession, 426–429, 514–515, 839– 233, 238–239, 505, 507–508, 512, 515,
840n7; and helicopter money, 785; inde 808n3, 833n3; in Great Depression, 298–
pendence, 772; and interest-rate setting, 299. See also commercial, industrial, agri-
104, 135, 194, 295–296, 497, 504, 523, 668, cultural, and consumer (CIAC) loans in
817n6, 831–832n19; limits to ability to Great Depression
control interest rates, 24, 95–97, 108, 278, commercial paper, 140, 412–413, 423, 505,
516–517, 536, 785; and liquidity prefer- 507–508, 512, 515, 808n3, 833n3, 834–
ence, 377; and the money supply, 504–505, 835nn8–9; in Great Depression, 298–299
522, 812n2; and New Classical economics, Commercial Paper Funding Facility (CPFF),
47; providing anchor to interest-rate 835–836n9, 839–840n7
spreads, 4, 11, 417, 426, 429–430, 514, 522, commodity money, 11, 193, 223–233, 237,
792; and quantitative easing, 426–428, 239, 253, 497, 503–509, 514, 522–523, 785,
444–446, 836n10; reaction function, 752; 808n3, 812n2. See also metallic money
and real-bills doctrine, 812n4; and reserve comparative statics, 3, 6–7, 25, 36, 88, 105,
scarcity, 515; in the United States, 816n4; 139, 140, 145, 150, 153, 191–192, 196, 222,
and views of future rates, 486–488. See also 223, 226, 227, 230, 233, 235, 237, 238–239,
European Central Bank 252–254, 257, 296, 303, 384, 538, 653, 686,
Chamberlin, Edward, 19, 670 691, 708, 715, 718, 731, 749, 773, 775, 777,
Chang, Winston, 827n4 780, 783, 792, 806n10, 812–813n13, 856n2
Chayanov, Alexander, 283 compensatory fnance, 511. See also func-
Chirinko, Robert, 744 tional fnance
Chodorow-Reich, Gabriel, 557, 564, 565, 578, consistent-forecasting condition, 467, 469
788 consols, 88, 343, 383, 390–394, 458–460, 464,
Copyright © 2021. Harvard University Press. All rights reserved.
Chou En-Lai, 619 636, 825n5, 826nn11–13, 829–830nn3–5,
Citigroup, 360 837n13. See also rentes
class: and Cambridge saving theory, 330–337, constant-elasticity of substitution production
670, 720, 727, 811n12, 821–822n13, function, 352, 357, 372–373, 666, 728, 741,
851n10; capitalist, 734, 766; and consump- 744, 745, 823n6, 850n5
tion, 8, 9, 147, 308, 328, 336–337, 787–788, conventional wage, 13–14, 680–685, 689–691,
821n11, 852–853n6; creditor, 769–770; 696–698, 704, 708–710, 711, 712, 713, 715,
division, 695; power, 13, 679–680, 761, 762, 718, 720, 726, 727, 731, 732–733, 746, 754,
789–790. See also patrimonial middle class; 761–764, 770, 771, 789–791, 852nn4–5
rentier; working class Cornwall, John, 851n1
classical dichotomy, 28, 54, 72, 109, 125, 193, correspondence principle, 26
210–211, 247, 683, 814n16 cost-cutting investment. See capital deep
climate change, 617–618 ening
Clower, Robert, 213, 216–217, 775, 811n11 cost-push infation, 693, 696
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
886 Index
coupon yield, 343, 390–401, 404–406, 418– elasticity of substitution. See constant-
420, 422, 450, 454, 459–462, 500, 829– elasticity of substitution production func-
830nn4–5, 837n13 tion
Crotty, James, 855–856n1, 856n3 Ellsberg, Daniel, 363, 823n9
crowding out. See under budget defcit; stabil- equilibrium vs. equilibrating, 64–70. See also
ity of equilibrium market-clearing equilibrium; price mecha-
nism; stability of equilibrium
Davidson, Paul, 35 ergodic and nonergodic systems, 35
Davis, Chester, 289 European Central Bank (ECB), 5, 47–48, 516,
de Finetti, Bruno, 362 836n10, 848–849n17
Debreu, Gerard, 161, 774, 823–824n10 euthanasia of the rentier, 43, 664
debt, foreign, 620–625, 626–627 excess reserves, 143, 230, 233, 506, 515–516,
debt relief: in Great Depression, 292; in Great 837n12; in Great Depression, 270–272,
Recession, 548 277–278, 297–298, 833–834n6
debt-defation theory of depression, 8, 236– expectations: in downturn, 233; I. Fisher on,
240, 254, 276, 783–784 814–815n16; hypothesis, 11, 457–458,
default risk, 10, 343, 388, 389, 401, 402–403, 829n1; of infation, 472, 484–486, 827n3;
409, 412, 413, 423–425, 427–428, 429–430, and interest rates, 248; Keynes on, 347,
456, 457, 486, 488–496, 500, 508–509, 513, 360–361, 365, 813–814n13; of normal
514, 517–518, 522, 808n3, 826n15, reversion, 393–400, 414–415, 426–429,
828nn9–10, 837–839n3 430, 446–455, 457, 469–470, 472, 480,
defcit. See budget defcit; trade defcit 501, 826n13, 830n12, 831n15; Patinkin
defation, 233, 234, 354–356, 357, 367, 517, on, 26–27, 216; in permanent-income
666–668, 674, 680, 764, 771, 782, 790, hypothesis, 321; Solow on, 222. See also
813–814n13, 849n19, 853n8; I. Fisher on, rational expectations
8, 236–240, 254, 276, 783–784; in Great expenditure smoothing, 548, 553–555, 561,
Depression, 243, 266, 276–277; and invest- 562, 564, 567, 570, 584–587, 787–788
ment demand, 248–249, 252; Klein on, Ezekiel, Mordecai, 309–310
215–216, 217; Mill on, 802n5; Viner on,
56–59, 529. See also debt-defation theory family farm. See agriculture
of depression Fannie Mae, 836n10, 837n12
demand-pull infation, 693 Farhi, Emmanuel, 661
Dennison, Henry, 817n8 Federal Funds rate, 410, 412–413, 423, 486–
Desan, Christine, 504, 810n2 488, 510, 515–516, 831n18, 832n20,
Diamond, Peter, 823–824n10 837n12, 839–840n7
Dickens, Charles, 846n6 Federal Open Market Committee (FOMC),
disequilibrium, 7, 23–27, 33–34, 57, 85, 145, 486–488, 538, 839–840n7
162–190; Keynes on, 23 Federal Reserve: in Great Depression, 297–
disequilibrium hypothesis of saving, 323–324, 299, 816n4; in Great Recession, 487, 492,
Copyright © 2021. Harvard University Press. All rights reserved.
553–554, 564 515–516, 835–836nn9–10, 837n12, 839–
Domar, Evsey, 13, 656–658, 661–663, 668– 840n7
669, 837–839n3, 850nn3–4 Federico, Giovanni, 280–281, 282
Dos Santos Ferreira, Rodolphe, 217–218, 775 Ferber, Robert, 313
Draghi, Mario, 48 Field, Alexander, 810n4
Duesenberry, James, 324, 327 fne tuning, 47. See also fscal policy
Dunlop, John, 211 frst-pass model, 6, 29–32, 41, 46, 49, 73, 79–
dynamic stochastic general equilibrium 95, 112, 119–120, 192, 194; adjustment in,
(DSGE) models, 5, 162, 556–557 179–183, 197–211; and aggregate demand,
97–99, 105–106; monetary policy in, 95–
ecological limits to growth, 616–620 97; and money wage, 123
Eggertsson, Gauti, 288, 290 fscal federalism, 546, 844–845n1
Eichengreen, Barry, 278, 815n2, 816– fscal policy: countercyclical, 12, 42–43, 47,
817nn4–5 97, 330, 529–543, 594–595, 664, 773, 785–
Marglin, Stephen A.. Raising Keynes : A Twenty-First-Century General Theory, Harvard University Press, 2021. ProQuest Ebook
Index 887
787, 792, 847n11, 848–849n17; discretion- Frank, Robert, 51, 64, 67, 111, 179, 205, 324,
ary, 12, 539–540, 542, 787; implementation 802–803nn6–7, 824n1
lags, 538; Keynes on, 42, 97, 501, 530–531, Freddie Mac, 837n12, 839–840n7
805–806n9, 820–821n7, 837n1; New York Friedman, Benjamin, 515, 612–625, 626, 631,
Times on, 43–45; recognition lags, 538. See 827n4, 836n11, 847n11, 848n14
also automatic stabilization Friedman, Milton, 3, 5, 8–9, 27, 36, 37–38,
Fisher, Frank, 25, 163 48, 56, 70, 71, 72, 78, 79, 152, 234, 240,
Fisher, Irving, 7, 8, 9, 78, 236–237, 240, 254, 242–243, 253, 307–308, 538, 548, 598, 683,
276, 342, 359, 381, 517, 519, 783, 814– 754, 764, 776, 778, 779, 782, 785, 798n5,
815n16, 833–834n6 820–821n7, 823n9; and Anna Schwartz on
Fisher efect, 7, 8, 217, 223, 245–252, 253– Great Depression, 277–278, 293–299, 816–
254, 259–260; in Great Depression, 266, 817n5, 835–836n9; on permanent-income
276–278, 279, 783 hypothesis, 147–148, 313–314, 320–326,
fxprice adjustment, 175, 179, 190, 198, 204– 329–330, 552–555, 557, 558, 787. See also
211, 217, 220–221, 300, 771, 839n4, under real-balance efect
851n11; and Cambridge saving theory, Frost, Robert, 282
331–335, 338–339, 368–371, 698–705; and functional fnance, 4, 12–13, 42, 529–531,
You might also like
- Introduction To Keynesian Theory and Keynesian Economic PoliciesNo ratings yetIntroduction To Keynesian Theory and Keynesian Economic Policies20 pages
- Combined Solved Intermediate Macroeconomics PaperNo ratings yetCombined Solved Intermediate Macroeconomics Paper6 pages
- (Ebook PDF) Macroeconomics 4th Edition by Charles I. Jones Download100% (7)(Ebook PDF) Macroeconomics 4th Edition by Charles I. Jones Download58 pages
- (Ebook PDF) Macroeconomics 4th Edition by Charles I. Jones Instant Download100% (5)(Ebook PDF) Macroeconomics 4th Edition by Charles I. Jones Instant Download57 pages
- Macroeconomics - Third Edition (2012), Charles Jones, W.W. Norton & Company, New York.14% (7)Macroeconomics - Third Edition (2012), Charles Jones, W.W. Norton & Company, New York.8 pages
- Unit 3: Exchange Rates: Keynesian Models (ISLM) 4/9/2012No ratings yetUnit 3: Exchange Rates: Keynesian Models (ISLM) 4/9/201250 pages
- Classical Theory of Employment II-outputNo ratings yetClassical Theory of Employment II-output9 pages
- Unit 4 Class 4 Powerpoints Correct VersionNo ratings yetUnit 4 Class 4 Powerpoints Correct Version29 pages
- Final Notes For Macro Eco2214 Topic One and TwoNo ratings yetFinal Notes For Macro Eco2214 Topic One and Two25 pages
- Keynesian Economics I: The Keynesian System (I) : The Role of Aggregate DemandNo ratings yetKeynesian Economics I: The Keynesian System (I) : The Role of Aggregate Demand29 pages
- ECON 302 Syllabus, Pocedure & Sample Quesntionnaire For Final ExamNo ratings yetECON 302 Syllabus, Pocedure & Sample Quesntionnaire For Final Exam14 pages
- (Ebook PDF) Macroeconomics 10th Edition by N. Gregory Mankiw Sample100% (2)(Ebook PDF) Macroeconomics 10th Edition by N. Gregory Mankiw Sample150 pages
- Formulas Under Each Topuc in ECO 203 PDFNo ratings yetFormulas Under Each Topuc in ECO 203 PDF7 pages
- Macroeconomics - (Chapter 11 Aggregate Demand I Building The is-LM Model)No ratings yetMacroeconomics - (Chapter 11 Aggregate Demand I Building The is-LM Model)37 pages
- (Brian Snowdon, Howard R Vane) An Encyclopedia ofNo ratings yet(Brian Snowdon, Howard R Vane) An Encyclopedia of742 pages
- Andrew B. Abel, Ben S. Bernank Global Edition-Pearson (2023) 94No ratings yetAndrew B. Abel, Ben S. Bernank Global Edition-Pearson (2023) 941 page
- Money and Interest Rates: 7.1 Asset Equilibrium, Reviewed 7.2 IS-LM Model 7.3 Price and Aggregate DemandNo ratings yetMoney and Interest Rates: 7.1 Asset Equilibrium, Reviewed 7.2 IS-LM Model 7.3 Price and Aggregate Demand48 pages
- Introduction To Fluctuations Keynesian Economics: Master Epp - Macro 2 2010 Yann AlganNo ratings yetIntroduction To Fluctuations Keynesian Economics: Master Epp - Macro 2 2010 Yann Algan54 pages
- Macroeconomics (From Greek Prefix "Makros-" Meaning "Large" + "Economics") Is A Branch ofNo ratings yetMacroeconomics (From Greek Prefix "Makros-" Meaning "Large" + "Economics") Is A Branch of5 pages
- Notes On General Equilibrium Econ 304 FALL 2013No ratings yetNotes On General Equilibrium Econ 304 FALL 20136 pages
- Lectures - Pinciples of Macro EconomicsNo ratings yetLectures - Pinciples of Macro Economics22 pages
- What Is Macroeconomics?: Inflation Economic Growth Gross Domestic Product (GDP) UnemploymentNo ratings yetWhat Is Macroeconomics?: Inflation Economic Growth Gross Domestic Product (GDP) Unemployment9 pages
- Davidson - 1982 - International Money and The Real WorldNo ratings yetDavidson - 1982 - International Money and The Real World336 pages
- Kaldor - 1982 - The Scourge of MonetarismNo ratings yetKaldor - 1982 - The Scourge of Monetarism132 pages
- The Past and Future of The Social Sciences (Definitivo, Rivisto H.)No ratings yetThe Past and Future of The Social Sciences (Definitivo, Rivisto H.)29 pages
- Moore - 1979 - The Endogenous Money StockNo ratings yetMoore - 1979 - The Endogenous Money Stock1 page
- Thomas Chermack - Foundations of Scenario Planning - The Story of Pierre Wack (Routledge International Studies in Business History) - Routledge (2017)No ratings yetThomas Chermack - Foundations of Scenario Planning - The Story of Pierre Wack (Routledge International Studies in Business History) - Routledge (2017)318 pages
- Immanuel Wallerstein O Sistema Mundial Moderno Vol. 188% (8)Immanuel Wallerstein O Sistema Mundial Moderno Vol. 178 pages
- Container Logistics in Paraguay, Terport CaseNo ratings yetContainer Logistics in Paraguay, Terport Case5 pages
- Sundaram Multi-Factor Fund - Product NoteNo ratings yetSundaram Multi-Factor Fund - Product Note2 pages
- Admin & Project Management Associate (Part-Time)No ratings yetAdmin & Project Management Associate (Part-Time)2 pages
- The Adoption Process by J. Intong For RB Cagmat Review CenterNo ratings yetThe Adoption Process by J. Intong For RB Cagmat Review Center102 pages
- Co-Operative Banks Details of The Zoroastrian Co-Operative Bank Limited Central OfficeNo ratings yetCo-Operative Banks Details of The Zoroastrian Co-Operative Bank Limited Central Office10 pages
- Behavioral Finance Biases in Investment Decision Making PresentationNo ratings yetBehavioral Finance Biases in Investment Decision Making Presentation26 pages
- Digital Marketing Excellence Planning, Optimizing ... - (Cover)50% (2)Digital Marketing Excellence Planning, Optimizing ... - (Cover)5 pages
- Stock Market Index: Financial Markets & InstitutionsNo ratings yetStock Market Index: Financial Markets & Institutions23 pages
- Assigment Marketing Mix 4 P - All Nippon AirwaysNo ratings yetAssigment Marketing Mix 4 P - All Nippon Airways41 pages
- Introduction To Globalization: Presented By: CB Ronie E. Sugarol, MPBMNo ratings yetIntroduction To Globalization: Presented By: CB Ronie E. Sugarol, MPBM10 pages
- Memanfaatkan Marketplace Sebagai Media Promosi Untuk Meningkatkan Penjualan Di Tengah Pandemi Covid-19No ratings yetMemanfaatkan Marketplace Sebagai Media Promosi Untuk Meningkatkan Penjualan Di Tengah Pandemi Covid-198 pages
- Chapter-2-Partnership-Operations-and-Financial-Reporting-Problems 3No ratings yetChapter-2-Partnership-Operations-and-Financial-Reporting-Problems 35 pages
- Current Italian Universities Open For Admission 2025-26AYNo ratings yetCurrent Italian Universities Open For Admission 2025-26AY1 page
