STICKY INFORMATION VERSUS STICKY PRICES:

A PROPOSAL TO REPLACE THE NEW

KEYNESIAN PHILLIPS CURVE*

N. GREGORY MANKIW AND RICARDO REIS

This paper examines a model of dynamic price adjustment based on the

assumption that information disseminates slowly throughout the population.
Compared with the commonly used sticky-price model, this sticky-information
model displays three related properties that are more consistent with accepted
views about the effects of monetary policy. First, disinflations are always contrac-
tionary (although announced disinflations are less contractionary than surprise
ones). Second, monetary policy shocks have their maximum impact on inflation
with a substantial delay. Third, the change in inflation is positively correlated
with the level of economic activity.

The dynamic effects of aggregate demand on output and

inflation remain a theoretical puzzle for macroeconomists. In
recent years, much of the literature on this topic has used a model
of time-contingent price adjustment. This model, often called "the
new Keynesian Phillips curve," builds on the work of Taylor
[1980], Rotemberg [1982], and Calvo [1983]. As the recent survey
by Clarida, Gali, and Gertler [1999] illustrates, this model is
widely used in theoretical analysis of monetary policy. McCallum
[1997] has called it "the closest thing there is to a standard
specification."
Yet there is growing awareness that this model is hard to
square with the facts. Ball [1994a] shows that the model yields
the surprising result that announced, credible disinflations cause
booms rather than recessions. Fuhrer and Moore [1995] argue
that it cannot explain why inflation is so persistent. Mankiw
[2001] notes that it has trouble explaining why shocks to mone-
tary policy have a delayed and gradual effect on inflation. These
problems appear to arise from the same source: although the
price level is sticky in this model, the inflation rate can change
quickly. By contrast, empirical analyses of the inflation process
(e.g., Gordon [1997]) typically give a large role to "inflation
inertia."

* We are grateful to Alberto Alesina, Marios Angeletos, Laurence Ball, Wil-

liam Dupor, Martin Eichenbaum, Christopher Foote, Xavier Gabaix, Mark
Gertler, Bennett McCallum, Kenneth Rogoff, Julio Rotemberg, Michael Woodford,
and anonymous referees for comments on an earlier draft. Reis is grateful to the
Fundacao Ciencia e Tecnologia, Praxis XXI, for financial support.

co 2002 by the President and Fellows of Harvard College and the Massachusetts Institute of
Technology.
The Quarterly Journal of Economics, November 2002

1295
Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955
by Universidade Federal da Bahia user
on 11 May 2018
 1296 QUARTERLY JOURNAL OF ECONOMICS

This paper proposes a new model to explain the dynamic

effects of aggregate demand on output and the price level. The
essence of the model is that information about macroeconomic
conditions diffuses slowly through the population. This slow dif-
fusion could arise because of either costs of acquiring information
or costs of reoptimization. In either case, although prices are
always changing, pricing decisions are not always based on cur-
rent information. We call this a sticky information model to con-
-

trast it to the standard sticky price model on which the new

-

Keynesian Phillips curve is based.

To formalize these ideas, we assume that each period a
fraction of the population updates itself on the current state of the
economy and computes optimal prices based on that information.
The rest of the population continues to set prices based on old
plans and outdated information. Thus, this model combines ele-
ments of Calvo's [1983] model of random adjustment with ele-
ments of Lucas' [1973] model of imperfect information.
The implications of our sticky-information model, however,
are closer to those of Fischer's [1977] contracting model. As in the
Fischer model, the current price level depends on expectations of
the current price level formed far in the past. In the Fischer
model, those expectations matter because they are built into
contracts. In our model, they matter because some price
setters are still setting prices based on old decisions and old
information.'
After introducing the sticky-information model in Section I,
we examine the dynamic response to monetary policy in Section
II. In contrast to the standard sticky-price model, which allows
for the possibility of disinflationary booms, the sticky-information
model predicts that disinflations always cause recessions. In
some ways, the dynamic response in the sticky-information model
resembles Phillips curves with backward-looking expectations.

1. We should also note several other intellectual antecedents. Gabaix and

Laibson [2001] suggest that consumption behavior is better understood with the
assumption that households update their optimal consumption only sporadically;
it was in fact a presentation of the Gabaix-Laibson paper that started us working
on this project. Another related paper is Ball [2000], who tries to explain price
dynamics with the assumption that price setters use optimal univariate forecasts
but ignore other potentially relevant information. In addition, Rotemberg and
Woodford [1997] assume a one-period decision lag for some price setters Finally,
after developing our model, we became aware of Koenig [1997]; Koenig's model of
aggregate price dynamics is motivated very differently from ours and is applied to
a different range of questions, but it has a formal structure that is similar to the
model explored here.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1297

Yet there is an important difference: in the sticky-information

model, expectations are rational, and credibility matters. In par-
ticular, the farther in advance a disinflationary policy is antici-
pated, the smaller is the resulting recession.
In Section III we make the model more realistic by adding a
simple yet empirically plausible stochastic process for the money
supply. After calibrating the model, we examine how output and
inflation respond to a typical monetary policy shock. We find that
the sticky-price model yields implausible impulse response func-
tions: according to this model, the maximum impact of a mone-
tary shock on inflation occurs immediately. By contrast, in the
sticky-information model, the maximum impact of monetary
shocks on inflation occurs after seven quarters. This result more
closely matches the estimates from econometric studies and the
conventional wisdom of central bankers.
Section IV then examines whether the models can explain
the central finding from the empirical literature on the Phillips
curve—namely, that vigorous economic activity causes inflation
to rise. The standard sticky-price model is inconsistent with this
finding and, in fact, yields a correlation of the wrong sign. By
contrast, the sticky-information model can explain the widely
noted correlation between economic activity and changes in
inflation.
The sticky-information model proposed here raises many
questions. In Section V we examine the evidence that might be
brought to bear to evaluate the model, and we discuss how one
might proceed to give the model a more solid microeconomic
foundation. In Section VI we conclude by considering how the
model relates to the broader new Keynesian literature on price
adjustment.

I. A TALE OF TWO MODELS

We begin by deriving the two models: the standard sticky-
price model, which yields the new Keynesian Phillips curve, and
the proposed sticky-information model.
I.A. A Sticky-Price Model: The New Keynesian Phillips Curve
Here we review the standard derivation of the new Keynes-
ian Phillips curve, as based on the Calvo model. In this model,
firms follow time-contingent price adjustment rules. The time for
price adjustment does not follow a deterministic schedule, how-

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1298 QUARTERLY JOURNAL OF ECONOMICS

ever, but arrives randomly. Every period, a fraction X of firms

adjust prices. Each firm has the same probability of being one of
the adjusting firms, regardless of how long it has been since its
last price adjustment.
We start with three basic relationships. The first concerns
the firm's desired price, which is the price that would maximize
profit at that moment in time. With all variables expressed in
logs, the desired price is
*_
Pt - Pt ah.

This equation says that a firm's desired price p* depends on the

overall price level p and output y. (Potential output is normalized
to zero here, so y should be interpreted as the output gap.) A
firm's desired relative price, p* p, rises in booms and falls in
—

recessions.
Although we will not derive this equation from a firm's profit
maximization problem, one could easily do so, following Blan-
chard and Kiyotaki [19871. Imagine a world populated by identi-
cal monopolistically competitive firms. When the economy goes
into a boom, each firm experiences increased demand for its
product. Because marginal cost rises with higher levels of output,
greater demand means that each firm would like to raise its
relative price.
In this model, however, firms rarely charge their desired
prices, because price adjustment is infrequent. When a firm has
the opportunity to change its price, it sets its price equal to the
average desired price until the next price adjustment. The ad-
justment price x is determined by the second equation:

x, = E (1 -
.J=0

According to this equation, the adjustment price equals a

weighted average of the current and all future desired prices.
Desired prices farther in the future are given less weight because
the firm may experience another price adjustment between now
and that future date. This possibility makes that future desired
price less relevant for the current pricing decision. The rate of
arrival for price adjustments, X, determines how fast the weights
decline.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1299

The third key equation in the model determines the overall

price level p:

Pt = x E (1- x)Jxt_,.
,=0

According to this equation, the price level is an average of all

prices in the economy and, therefore, a weighted average of all the
prices firms have set in the past. The rate of arrival for price
adjustments, X, also determines how fast these weights decline.
The faster price adjustment occurs, the less relevant past pricing
decisions are for the current price level.
Solving this model is a matter of straightforward algebra. We
obtain the following:

a t = [aX 2/( 1X)]Yt Et7rt+i,

where iT t = p t — p t _ i is the inflation rate. Thus, we obtain the

new Keynesian Phillips curve. Inflation today is a function of
output and inflation expected to prevail in the next period. This
model has become the workhorse for much recent research on
monetary policy.

I.B. A Sticky - Information Model

This subsection proposes an alternative model of price dy-
namics. In this model, every firm sets its price every period, but
firms gather information and recompute optimal prices slowly
over time. In each period, a fraction X of firms obtains new
information about the state of the economy and computes a new
path of optimal prices. Other firms continue to set prices based on
old plans and outdated information. We make an assumption
about information arrival that is analogous to the adjustment
assumption in the Calvo model: each firm has the same probabil-
ity of being one of the firms updating their pricing plans, regard-
less of how long it has been since its last update.
As before, a firm's optimal price is
_
Pt — Pt ciY t-

A firm that last updated its plans j periods ago sets the price

xt = ErrPr

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1300 QUARTERLY JOURNAL OF ECONOMICS

The aggregate price level is the average of the prices of all firms
in the economy:

Pt = K E (1 - x)ix',.
j=0

Putting these three equations together yields the following equa-

tion for the price level:

Pt = K E (1 - x).'Et-j (pt + ayt)•

J =0

The short-run Phillips curve is apparent in this equation: output

is positively associated with surprise movements in the price
level.
With some tedious algebra, which we leave to the Appendix,
this equation for the price level yields the following equation for
the inflation rate:

'rrt [ - x lYt E (1 - x) 3Et_i_,(7rt + aAyt),

j=0

where Ay, = y, – y,_ 1 is the growth rate of output. Inflation

depends on output, expectations of inflation, and expectations of
output growth. We call this the sticky-information Phillips curve.
Take note of the timing of the expectations. In the standard
sticky-price model, current expectations of future economic con-
ditions play an important role in determining the inflation rate.
In this sticky-information model, as in Fischer [1977], expecta-
tions are again important, but the relevant expectations are past
expectations of current economic conditions. This difference
yields large differences in the dynamic pattern of prices and
output in response to monetary policy, as we see in the next
section.
One theoretical advantage of the sticky-information model is
that it survives the McCallum critique. McCallum [1998] has
criticized the standard sticky-price model on the grounds that it
violates a strict form of the natural rate hypothesis, according to
which "there is no inflation policy—no money creation scheme—
that will keep output high permanently." Following Lucas [1972],
McCallum argues that "it seems a priori implausible that a nation

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1301

can enrich itself in real terms permanently by any type of mone-

tary policy, by any path of paper money creation." The sticky-
price model fails this test because a policy of permanently falling
inflation will keep output permanently high. By contrast, the
sticky-information model satisfies this strict version of the natu-
ral rate hypothesis. Absent surprises, it must be the case that
p, = E t _jp„ which in turn implies y, = O. Thus, the McCallum
critique favors the sticky-information Phillips curve over the
more commonly used alternative.

II. INFLATION AND OUTPUT DYNAMICS IN THE

STICKY-INFORMATION MODEL
Having presented the sticky-information Phillips curve, we
now examine its dynamic properties. To do this, we need to
complete the model with an equation for aggregate demand. We
use the simplest specification possible:

m P+Y
where m is nominal GDP. This equation can be viewed as a
quantity-theory approach to aggregate demand, where m is in-
terpreted as the money supply and log velocity is assumed con-
stant at zero. Alternatively, m can be viewed more broadly as
incorporating the many other variables that shift aggregate de-
mand. We take m to be exogenous. Our goal is to examine how
output and inflation respond to changes in the path of m. 2
As we proceed, it will be useful to compare the dynamics of
our proposed sticky-information Phillips curve with more famil-
iar models. We use two such benchmarks. The first is the sticky-
price model presented earlier, which yields the standard new
Keynesian Phillips curve:

1Tt = r3Yt EtiTt+1,

where (3 = [aX 2/(1 — X)] and the expectations are assumed to be
formed rationally. The second is a backward-looking model:
ITt = 13Yt 7 t-1.•

2. There are other, perhaps more realistic, ways to add aggregate demand to
this model. One possibility would be to add an IS equation together with an
interest-rate policy rule for the central bank Such an approach is more compli-
cated and involves more free parameters. We believe the simpler approach taken
here best illustrates the key differences between the sticky-information model and
more conventional alternatives.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1302 QUARTERLY JOURNAL OF ECONOMICS

This backward-looking model resembles the equations estimated

in the empirical literature on the Phillips curve (as discussed in,
e.g., Gordon [1997]). It can be viewed as the sticky-price
model together with the assumption of adaptive expectations:
Etlr t+1 = 1Tt- 1.
When we present simulated results from these models, we
try to pick plausible parameter values. Some of these parameters
depend on the time interval. For concreteness, we take the period
in the model to equal one quarter. We set a = .1 and X = .25 (and,
thus, (3 = .0083). This value of X means that firms on average
make adjustments once a year. The small value of a means that
a firm's desired relative price is not very sensitive to macroeco-
nomic conditions. Note that the firm's desired nominal price can
now be written as
13'; = (1 - a)p t + am t .
If a is small, then each firm gives more weight to what other firms
are charging than to the level of aggregate demand.'
We now consider three hypothetical policy experiments. In
each experiment, we posit a path for aggregate demand m. We
then derive the path for output and inflation generated by the
sticky-information model and compare it with the paths gener-
ated by the two benchmark models. The details of the solution are
presented in the Appendix. Here we discuss the dynamic paths
followed by output and inflation.
ILA. Experiment 1: A Drop in the Level of Aggregate Demand
The first experiment we consider is a sudden and permanent
drop in the level of aggregate demand. The demand variable m t is
constant and then, at time zero, unexpectedly falls by 10 percent
and remains at this new level.
The top graph in Figure I shows the path of output predicted
by each of the three models. In all three models, the fall in
demand causes a recession, which gradually dissipates over time.

3. In the backward-looking model, the parameter 13 determines the cost of

disinflation. According to this model, if output falls 1 percent below potential for
one quarter, then the inflation rate falls by [3 if measured at a quarterly rate, or
413 if annualized. If output falls by 1 percent below potential for one year, then the
annualized inflation rate falls by 1613. Thus, the sacrifice ratio—the output loss
associated with reducing inflation by one percentage point—is 1/(16(3). Our pa-
rameters put the sacrifice ratio at 7.5. For comparison, Okun's [1978] classic study
estimated the sacrifice ratio to be between 6 and 18 percent; Gordon [1997,
footnote 8] puts it at 6.4. Thus, our backward-looking model is in the ballpark of
similar models used in the previous literature.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1303

FIGURE I
Dynamic Paths after a 10 Percent Fall in the Level of Aggregate Demand at Time 0

The impact of the fall in demand on output is close to zero at

sixteen quarters. The backward-looking model generates an os-
cillatory pattern, whereas the other two models yield monotonic
paths. Otherwise, the models seem to yield similar results.
Differences among the models become more apparent, how-
ever, when we examine the response of inflation in the bottom of
Figure I. In the sticky-price model, the greatest impact of the fall

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1304 QUARTERLY JOURNAL OF ECONOMICS

in demand on inflation occurs immediately. The other two models

show a more gradual response. In the sticky-information model,
the maximum impact of the fall in demand on inflation occurs at
seven quarters. Inflation could well be described as inertial.
The inertial behavior of inflation in the sticky-information
model requires the parameter a to be less than one. Recall that
the firm's desired price is
0 = (1 — a)pt + am t .
If a = 1, then the desired price moves only with the money supply.
In this case, firms adjust their prices immediately upon learning
of the change in policy; as a result, inflation responds quickly
(much as it does in the sticky-price model). By contrast, if a < 1,
then firms also care about the overall price level and, therefore,
need to consider what information other firms have. For small a,
even an informed firm will not adjust its price much to the change
in aggregate demand until many other firms have also learned of
it. A small value of a can be interpreted as a high degree of real
rigidity (to use the terminology of Ball and Romer [1990]) or a
high degree of strategic complementarity (to use the terminology
of Cooper and John [1988]). In the sticky-information model, this
real rigidity (or strategic complementarity) is a source of inflation
inertia.

ILB. Experiment 2: A Sudden Disinflation

The second experiment we consider is a sudden and perma-
nent shift in the rate of demand growth. The demand variable m t
is assumed to grow at 10 percent per year (2.5 percent per period)
until time zero. In period 0, the central bank sets in, the same as
it was in the previous period and, at the same time, announces
that m t will thereafter remain constant. Figure II shows the path
of output and inflation predicted by the three models.
According to the sticky-price model, inflation falls immedi-
ately to the lower level. Price setters, realizing that disinflation is
underway, immediately respond by making smaller price adjust-
ments. Prices are sticky in the sticky-price model, but inflation
exhibits no inertia. The response of output, of course, is the other
side of the coin. Because inflation responds instantly to the fall in
money growth, output does not change. As in Phelps [1978],
disinflation is costless.
By contrast, the sticky-information model predicts a gradual
reduction in inflation. Even after the disinflationary policy is in

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1305

FIGURE II
Dynamic Paths Given an Unanticipated Fall in the Growth Rate of Aggregate
Demand at Date 0

place, most price setters are still marking up prices based on old
decisions and outdated information. As a result of this inertial
behavior, inflation is little changed one or two quarters after the
disinflation has begun. With a constant money supply and rising
prices, the economy experiences a recession, which reaches a
trough six quarters after the policy change. Output then gradu-

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1306 QUARTERLY JOURNAL OF ECONOMICS

ally recovers and is almost back to normal after twenty quarters.

These results seem roughly in line with what happens when
nations experience disinflation. 4
ILC. Experiment 3: An Anticipated Disinflation
Now suppose that the disinflation in our previous experiment
is announced and credible two years (eight periods) in advance.
Let us consider how this anticipated disinflation affects the path
of output and inflation.
Figure III shows output and inflation according to the three
models. The predictions for the backward-looking model are ex-
actly the same as in Experiment 2: the assumption of adaptive
expectations prevents the announcement from having any effect.
But the results are different in the other two models, which posit
rational expectations.
In the sticky-price model, the announced disinflation causes
a boom. As Ball [1994a] emphasizes, inflation in this model moves
in anticipation of demand. When price setters anticipate a slow-
down in money growth, inflation falls immediately. This fall in
inflation, together with continued increases in the money supply,
leads to rising real money balances and higher output.
By contrast, the sticky-information model does not produce
booms in response to anticipated disinflations. In this model,
there is no change in output or inflation until the disinflationary
policy of slower money growth begins. Then, the disinflation
causes a recession.
It would be wrong to conclude, however, that the announce-
ment has no effect in the sticky-information model. Because of the
announcement, many price setters have already adjusted their
plans in response to the disinflationary policy when it begins. As
a result, an announced slowdown in money growth leads to a
quicker inflation response and a smaller output loss than does a
sudden slowdown in money growth. For these parameters a dis-
inflation announced and fully credible eight quarters in advance
has a cumulative cost that is about one-fifth the size of the
surprise disinflation.
In a way, the sticky-information model combines elements of

4. Ball [1994b1 examines disinflation for a number of countries. For the nine
countries for which quarterly data are available, he identifies 28 periods of
disinflation. In 27 of these cases, the decline in inflation is associated with a fall
in output below its trend level. This finding is related to the acceleration phenom-
enon we document and discuss below.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1307

FIGURE III
Dynamic Paths Given an Announcement at Date —8 of a Fall in the Growth
Rate of Aggregate Demand at Date 0

the other two models. Like the backward-looking model (but

unlike the sticky-price model), disinflations consistently cause
recessions rather than booms. Like the sticky-price model (but
unlike the backward-looking model), expectations, announce-
ments, and credibility matter for the path of inflation and output.
These features of the sticky-information model seem consistent
with how central bankers view their influence on the economy.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1308 QUARTERLY JOURNAL OF ECONOMICS

III. THE RESPONSE TO REALISTIC MONETARY SHOCKS

So far, we have compared how output and inflation respond

to hypothetical paths for aggregate demand. We now take a step
toward greater realism. In particular, we assume a plausible
stochastic process for the money supply and then examine the
implied processes of output and inflation. As Christiano, Eichen-
baum, and Evans [1999] discuss, economists have a good sense of
the dynamic effects of monetary policy shocks. One way to gauge
a model's empirical validity is to see whether it can generate
plausible responses to such shocks.
III.A. The Stochastic Process for the Money Supply
We model the growth in the demand variable m as following
a first-order autoregressive process: Am t = p Am t _ i + E t . In this
environment, the price level is nonstationary, but the inflation
rate is stationary.
To calibrate p, we looked at quarterly U. S. data from 1960 to
1999. The variable m can be interpreted as a measure of money
supply, such as M1 or M2, or more broadly as a measure of
aggregate demand, such as nominal GDP. The first autocorrela-
tions for these time series are 0.57 for M1 growth, 0.63 for M2
growth, and 0.32 for nominal GDP growth. Based on these num-
bers, we set p = 0.5. The standard deviation of the residual is
0.009 for Ml, 0.006 for M2, and 0.008 for nominal GDP, so we
assume a standard deviation of 0.007 (although this choice affects
only the scale and not the shape of the dynamic paths).
The positive value of p means that a monetary shock builds
over time. That is, a positive shock E causes m to jump up and
then to continue to rise. With p = 0.5, the level of m eventually
asymptotes to a plateau that is twice as high as the initial shock.
This pattern for monetary shocks is broadly consistent with that
found in empirical studies. 5
III.B. Impulse Responses
Figure IV show the response of output and inflation to a
one-standard-deviation contractionary monetary policy shock. In
all three models, output exhibits a hump-shaped response. The
impact on output at first increases because demand is building

5. For example, Christiano, Eichenbaum, and Evans [1998] conclude that an

AR(1) process offers a good description of monetary policy shocks when using M2
as the measure of money.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1309

FIGURE IV
Dynamic Paths after a Negative One Standard Deviation (-0.007) Shock
to the AR(1) Aggregate Demand

over time. It eventually decays because prices adjust. The back-

ward-looking model yields oscillatory dynamics, whereas the
other two models yield a monotonic recovery from the recession.
The impulse responses for inflation to the monetary shock
show the differences between the sticky-price and sticky-informa-
tion models. In the sticky-price model, inflation responds quickly

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1310 QUARTERLY JOURNAL OF ECONOMICS

to a monetary policy shock. In fact, the largest impact on inflation

occurs immediately. By contrast, the sticky-information model
displays some of the inflation inertia that is built into the back-
ward-looking model. For these parameters, a monetary policy
shock in the sticky-information model has its maximum impact
on inflation after seven quarters.
The impulse response function for the sticky-information model
is far more consistent with conventional views about the effects of
monetary policy. Economists such as Friedman [1948] have empha-
sized the long lags inherent in macroeconomic policy. In particular,
a long lag between monetary policy actions and inflation is accepted
by most central bankers and confirmed by most econometric stud-
ies. 6 Figure IV shows that the sticky-information model can explain
a long lag between monetary policy shocks and inflation, whereas
the standard sticky-price model cannot.
III. C. Inflation Persistence
Fuhrer and Moore [1995] argue that the standard sticky-
price model "is incapable of imparting the persistence to inflation
that we find in the data" [p. 127]. In the model, they claim, "the
autocorrelation function of inflation . . . will die out very rapidly"
[p. 152]. This contradicts the empirical autocorrelations of infla-
tion, which decay slowly.
Motivated by these arguments, we calculated the implied
autocorrelations of inflation in our three models. We maintain the
empirically realistic process for money growth used above: Am t =
0.56n t _ 1 + E t . Table I presents the first eight autocorrelations of
inflation implied by the models, as well as the actual autocorre-
lations of inflation using the GDP deflator, the consumer price
index, and the core CPI. (The core CPI is the index excluding food
and energy.) Notice that inflation is highly autocorrelated in all
three models. That is, given the empirically realistic process for
the money supply, all the models deliver plausible persistence in
inflation.
In the end, we agree with Taylor [1999, p. 1040], who re-
sponds to Fuhrer and Moore by observing that "inflation persis-
tence could be due to serial correlation of money." This is why all
three models deliver high autocorrelations in Table I. Yet we also
agree with Fuhrer and Moore's deeper point: the standard sticky-

6. See, for example, Bernanke and Gertler [1995] or Christiano, Eichenbaum,

and Evans [1999].

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1311

TABLE I
AUTOCORRELATIONS FOR INFLATION: PREDICTED AND ACTUAL

Sticky- Sticky- Backward- Actual: Actual:

information price looking GDP Actual: core
model model model deflator CPI CPI

1 0.99 0.92 0.99 0.89 0.76 0.76

2 0.95 0.85 0.98 0.83 0.72 0.71
3 0.89 0.78 0.96 0.81 0.73 0.69
4 0.82 0.71 0.94 0.78 0.62 0.59
5 0.74 0.65 0.90 0.71 0.57 0.55
6 0.66 0.59 0.86 0.65 0.51 0.54
7 0.57 0.54 0.81 0.61 0.44 0.46
8 0.48 0.50 0.75 0.58 0.33 0.38

The first three columns of this table show the autocorrelations of inflation predicted by three models.
These calculations assume that money growth follows the process Am t = 0.5 Am, + Et. The model
parameters are set to o = .1 and X = .25. The last three columns show the actual autocorrelations of quarterly
inflation rates.

price model does not deliver empirically reasonable dynamics for

inflation and output. The key empirical fact that is hard to match,
however, is not the high autocorrelations of inflation, but the
delayed response of inflation to monetary policy shocks.'

IV. THE ACCELERATION PHENOMENON

When economists want to document the Phillips curve rela-

tionship in U. S. data from the last few decades of the twentieth
century, a common approach is to look at a scatterplot of the
change in inflation and some level of economic activity, such as
unemployment or detrended output. This scatterplot shows that
when economic activity is vigorous, as represented by low unem-
ployment or high output, inflation tends to rise. We call this
correlation the acceleration phenomenon.'
Panel A of Table II demonstrates the acceleration phenome-
non using U. S. quarterly data from 1960 to 1999. For these

7. Fuhrer and Moore also emphasize the persistence of inflation in response

to "inflation shocks," which could be interpreted as supply shocks. The model we
have developed here cannot address this issue, because there are no supply
shocks. In Mankiw and Reis [2001] we take a step toward remedying this
omission.
8. For some examples of economists using such a scatterplot to demonstrate
the acceleration phenomenon, see Abel and Bernanke [1998, p. 4571, Blanchard
[2000, p. 155], Dornbusch, Fischer and Startz [2001, p. 1091, Hall and Taylor
[1993, p. 2171, and Stock and Watson [1999, p. 48].

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1312 QUARTERLY JOURNAL OF ECONOMICS

TABLE II
THE ACCELERATION PHENOMENON

COST (.Yt/iTt+2 Trt —2) COST (.YtOrt+4 1Tt-4)

A. Actual
GDP deflator .48 .60
CPI .38 .46
Core CPI .46 .51
B. Predicted
Backward-looking model .99 .99
Sticky-price model —.13 —.11
Sticky-information model .43 .40

Panel A shows the correlation between output and the change in inflation in U. S. quarterly data from
1960 to 1999. The variable y is measured as log real GDP detrended with the Hodrick-Prescott filter. Panel
B shows the correlation between output and the change in inflation predicted by three models. These
correlations assume that money growth follows the process: Am, = 0.5 Am, 1 + Et. The model parameters
are set to a = .1 and X = .25.

calculations, output y t is the deviation of log real GDP from trend,

where trend is calculated using the Hodrick-Prescott filter. We
use three measures of inflation: the GDP deflator, the CPI, and
the core CPI. We use two timing conventions: we correlate y t with
irt+2 the one-year change in inflation centered around
the observation date, and with Tr t± 4 — Tr t - 4 , the two-year change
in inflation. All six correlations are positive and statistically
significant. In U. S. data, high output is associated with rising
inflation.
We now consider whether the models can generate the posi-
tive correlation between output and the change in inflation. We
assume the same stochastic process for the money supply as in
the previous section (Am t = 0 . 56 n t-1 + E t ) and the same
parameters (a = .1 and X = .25). Then, as explained in the
Appendix, we compute the population correlation between output
and the change in inflation.
Panel B of Table II shows the correlation predicted by the
models. Not surprisingly, the backward-looking model predicts a
high correlation. Because 7r, - Tr t _ i = Py t in this model, the
correlation is perfect for the one-period change in inflation and
only slightly lower for longer changes. In essence, the model
builds in the acceleration phenomenon through the assumption of
adaptive expectations. This is hardly a major intellectual victory:
the appeal of the backward-looking model comes not from its
theoretical underpinnings but from its ability to fit this
phenomenon.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1313

TABLE III
THE ACCELERATION PHENOMENON: SENSITIVITY ANALYSIS

A. Sticky-price model
a = .05 a = .1 a = .5 a= 1.0
X = .1 -0.08 -0.09 -0.12 -0.13
X = .25 -0.12 -0.13 -0.15 -0.15
X = .5 -0.15 -0.15 -0.13 -0.11
B. Sticky-information model
a = .05 a = .1 a = .5 a = 1.0
X = .1 0.49 0.39 0.05 -0.04
X = .25 0.51 0.43 0.12 0.02
X = .5 0.52 0.44 0.21 0.13

This table shows the correlation between output y, and the change in inflation arr t+2 - These
correlations assume money growth follows the process: Am, = 0.5 + Et.

We next look at the two models with better foundations.

Table II shows that the sticky-price model yields no association
between output and the change in inflation. For the one-year
change, the correlation between these variables is -0.13, which is
small and the wrong sign. By contrast, the sticky-information
model yields a strong, positive association. According to this
model, the correlation between output and the change in inflation
is 0.43. 9
To understand these results, recall the impulse response func-
tions. In the sticky-price model, when the economy experiences a
contractionary monetary shock, output falls for a while. Inflation
falls immediately, and then starts rising. Thus, low output coincides
with falling inflation at first, but then coincides with rising inflation
for a long period. This generates the small, negative correlation.
By contrast, in the sticky-information model, inflation ad-
justs slowly to a monetary shock. When a contractionary shock
lowers output, it also leads to a prolonged period of falling infla-
tion. This generates the positive correlation between output and
the change in inflation.
Table III presents a sensitivity analysis of this correlation to
alternative parameter values. Panel A of the table shows the

9. Our finding that the calibrated sticky-price model predicts a negative

correlation between y and Dar (in contrast to the positive correlation in the data)
is related to Gali and Gertler's finding [1999] that econometric estimation of this
model yields a coefficient on output of the wrong sign. Gali and Gertler's proposed
fix to the sticky-price model, however, differs substantially from the sticky-
information model proposed here.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1314 QUARTERLY JOURNAL OF ECONOMICS

correlation produced by the sticky-price model for different pa-

rameter values. Panel B shows the correlation produced by the
sticky-information model. The sticky-price model consistently
generates a small correlation of the wrong sign, whereas the
sticky-information model typically yields a positive correlation
between output and the change in inflation. 1°

V. RESPONSES TO SKEPTICS
A skeptic of the sticky-information model might naturally
ask two questions. What is the evidence for the model? What are
the model's microeconomic foundations? At this point, we cannot
give definitive answers, but we can offer some suggestive insights.
V.A. Evidence
We were motivated to explore the sticky-information model
by a set of empirical anomalies. As we have discussed, the canoni-
cal sticky-price model of inflation-output dynamics cannot be
reconciled with the conventional wisdom about the effects of
monetary policy, whereas the sticky-information model is consis-
tent with the conventional wisdom. This fact is itself evidence in
favor of this model compared with the leading alternative.
Our skeptic might say that the backward-looking model, with
its assumption of adaptive expectations, can also be reconciled with
the conventional wisdom. Or he might claim that the sticky-infor-
mation model is little more than a revival of adaptive expectations.
There are, however, at least two key differences between the sticky-
information model and the backward-looking model, and they argue
in favor of the sticky-information model. Both of these differences
arise from the fact that agents in the sticky-information model form
expectations rationally, even though they do not do so often.
One difference relates to changes in regime. As Barsky [1987]
and Ball [2000] point out, inflation has been close to a random
walk in the period since World War II, whereas before World War
I, when the gold standard was in effect, it was close to white noise.
The sticky-information model predicts that the reduced-form

10. These simulated correlations are computed under the assumption that all
fluctuations are due to demand shocks. If we were to append supply shocks to this
model, the predicted correlations would likely be driven down, because such
shocks push inflation and output in opposite directions. Thus, supply shocks
would make it even harder for the sticky-price model to match the positive
correlation in the data.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1315

Phillips curve should shift in response to this regime change. In

the recent period, expected inflation should roughly equal past
inflation, and output should be related to changes in inflation;
that is, the data should conform with the accelerationist Phillips
curve. In the early period, expected inflation should be roughly
constant, and output should be related to the level of inflation;
that is, the data should conform with the classic Phillips curve.
Ball [2000] reports that these two predictions hold true in the
data, which is inconsistent with the backward-looking model
strictly construed as a structural relationship.'
A second difference between the sticky-information model
and the backward-looking model concerns the role of credibility.
In the model with backward-looking expectations, credibility in
monetary policy has no role. By contrast, in the sticky-informa-
tion model, credibility can reduce the costs of disinflation. Most
central bankers believe that credibility is important, but this
belief is hard to confirm empirically. One intriguing study is that
of Boschen and Weise [2001], which examines 72 disinflationary
episodes from 19 OECD countries. This study measures credibil-
ity as the probability of success conditional on economic and
political variables known at the start of the disinflation. They
report that credibility lowers the output cost of reducing inflation.
This finding is consistent with the sticky-information model but
not with the backward-looking model.
Our skeptic might also ask for evidence on whether price
setters actually respond to information slowly. One piece of evi-
dence comes from Zbaracki et al. [2000], a study of the costs
associated with changing prices at a large manufacturing firm. In
this extensive case study, the authors find that only a small
percentage of these costs are the physical costs of printing and
distributing price lists. Far more important are the "managerial
and customer costs," which include the costs of information-
gathering, decision-making, negotiation, and communication.
Whether our sticky-information model captures the effects of
such costs is an open question, but this microeconomic evidence
suggest that macroeconomists need to think broadly about the
frictions that impede price adjustment.

11. Although the sticky-information model is consistent with Ball's findings,

other models may be as well. Ball proposes his own explanation: agents are
assumed to follow optimal univariate forecasts but to ignore information in other
variables. In Ball's theory, the optimal univariate forecasting rule changes when
the monetary regime changes.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1316 QUARTERLY JOURNAL OF ECONOMICS

In a recent paper Carroll [2001] reports some direct evidence

on the slow dissemination of information about inflation. Moti-
vated in part by our sticky-information model, Carroll compares
the inflation expectations from surveys of two groups: profes-
sional forecasters and the general public. Not surprisingly, he
finds that the professional forecasters are better at forecasting
inflation than the general public is. More important, he finds that
the general public's expectations respond to the professionals'
expectations with a lag. Based on the assumption that profession-
als do not suffer from sticky information, he estimates a parame-
ter similar to our X that measures how quickly the public's ex-
pectations catch up. Remarkably, the estimated value of X is
almost exactly the 0.25 that we assumed above. 12
Carroll [2001] also reports two related pieces of evidence that
cannot be explained with the sticky-information model as pre-
sented here. He finds that the professional's and the public's
expectations are closer on average when there are more news
stories about inflation. In addition, when there are more news
stories about inflation, the public's expectations adjust more rap-
idly to the professional's expectations. Thus, although Carroll's
study is consistent with the hypothesis that the public's inflation
expectations adjust slowly, it suggests that the rate of informa-
tion acquisition X is not constant over time.

V.B. Microfoundations
The starting point of this paper is the premise that some price
setters respond to information about monetary policy with a lag.
This hypothesis raises many questions. Why, exactly, do people set
prices based on outdated information? What set of constraints on the
process of decision-making leads to this outcome? How can econo-
mists model these imperfections in human understanding?
One approach to answering these questions is to use the tools
of information theory, as exposited, for instance, in the textbook
by Cover and Thomas [1991]. Drawing on these tools, Sims [2001]
suggests modeling humans as having a limited channel for ab-
sorbing information. Woodford [2001] uses this idea to build a
model of inflation-output dynamics In his model, because price
setters learn about monetary policy through a limited-informa-
tion channel, it is as if they observe monetary policy with a

12. In Mankiw and Reis [2001] we econometrically implement a model closely

related to the one developed here. We also find that X = 0.25 fits the data well.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1317

random error and have to solve a signal-extraction problem along

the lines of Lucas [1973].
As Woodford [2001] notes, his noisy-information model is pro-
posed in the same spirit as the sticky-information model explored in
this paper. The difference between the models is how information
arrives. In Woodford's model, price setters get a noisy signal about
monetary policy in every period. In our model, price setters obtain
perfect information about monetary policy with probability X in
every period. This difference in information arrival leads to some
differences in the dynamic response to monetary policy. 13
Information theory, however, may not be the best approach to
microfoundations. For most people, it is easy to find out what the
monetary authority is doing, but it is much harder to figure out what
it means. As Begg and Imperato [2001] emphasize, the real cost is
the cost of thinking One interpretation of the sticky-information
model is that because thinking is costly, people do it only once in a
while and, at other times, continue with outdated plans.
At best, this time-contingent approach to thinking is only an
approximation. How much a person thinks about an issue de-
pends on the benefit of doing so. Most people spend little time
thinking about monetary policy, but circumstances can affect the
allocation of their mental resources. This might explain Carroll's
finding that the public's expectations of inflation adjust more
rapidly when there are more news stories about inflation. Simi-
larly, the ends of hyperinflations (as studied by Sargent [1982])
may be different than more typical disinflations (as studied by
Ball [199413]) because the major institutional reforms that end
hyperinflations are exceptionally newsworthy events.
In the end, microfoundations for the Phillips curve may re-
quire a better understanding of bounded rationality. But until
those foundations are established, the sticky-information model
as described here may offer a useful tool for the study of inflation-
output dynamics.

VI. CONCLUSION
This paper has explored a dynamic model of price adjust-
ment. In particular, we have proposed a model to replace the

13. Our model, like Woodford's, starts by simply assuming the nature of the
information flow. Alternatively, one could start by assuming some cost of using an
information channel and then derive the optimal flow of information. See Mosca-
rini [2001] for an exploration of this issue in a different context.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1318 QUARTERLY JOURNAL OF ECONOMICS

widely used "new Keynesian Phillips curve." In this model, prices

are always changing, but decision-makers are slow to update
their pricing strategies in response to new information.
Although the choice between the sticky-information model
and the standard sticky-price model is ultimately an empirical
issue, three of our findings suggest that the sticky-information
model is more consistent with accepted views of how monetary
policy works. First, in the sticky-information model, disinfla-
tions are always contractionary (although announced disinfla-
tions are less costly than surprise ones). Second, in the sticky-
information model, monetary shocks have their maximum
effect on inflation with a substantial delay. Third, the sticky-
information model can explain the acceleration phenomenon
that vigorous economic activity is positively correlated with
rising inflation.
The dynamic patterns implied by the sticky-information
model resemble those from the Fischer [1977] contracting model,
although long-term contracts have no role. In both models, past
expectations of the current price level play a central role in
inflation dynamics In a sense, the slow dissemination of infor-
mation in our model yields a nominal rigidity similar to the one
Fischer assumed in his contracts.
Critics of the Fischer contracting model (e.g., Barro [1977])
have noted that it is hard to rationalize signing such contracts ex
ante or enforcing them ex post in light of the obvious inefficiencies
they cause. Such critiques do not apply to the model proposed
here. The assumption of slow information diffusion, perhaps due
to costs of acquiring or acting on new information, leaves no
apparent, unrealized gains from trade. Thus, sticky information
offers a more compelling rationale for this type of nominal rigidity.
Moving the theory of price adjustment away from sticky
prices toward sticky information may seem like a radical sugges-
tion, but we temper it with an important observation: many
lessons from the "new Keynesian" literature on price adjustment
apply as well to our sticky-information model as they do to the
standard sticky-price model.
An early lesson about price adjustment by firms with some
degree of monopoly power is that the private losses from sticky
prices are only second order, even if the social losses are first
order [Mankiw 1985; Akerlof and Yellen 1985]. Thus, if firms face
small costs of price adjustment or are only near rational, they
may choose to maintain sticky prices, even if the macroeconomic

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1319

effect of doing so is significant. When we move from sticky prices

to sticky information, this lesson applies in somewhat modified
form. If there are small costs of acquiring information or recom-
puting optimal plans, firms may choose not to update their pric-
ing strategies. The private loss from maintaining old decisions,
like the cost of maintaining old prices, is second order.
Another lesson from the literature on price adjustment is
that real rigidities amplify monetary nonneutralities [Ball and
Romer 1990]. Real rigidity is defined as a lack of sensitivity of
desired relative prices to macroeconomic conditions. Here, this
translates into a small value of the parameter a. Real rigidities
also play a role in our sticky-information model. Price setters who
are updating their decisions are aware that other price setters are
not, and this knowledge limits the size of their adjustments,
especially when a is small. As a result, real rigidity tends to
exacerbate the effects of monetary policy.
An advantage of sticky-information over sticky-price mod-
els is that they more naturally justify the widely assumed
time-contingent adjustment process. If firms have sticky prices
because of menu costs but are always collecting information
and optimizing in response to that information, then it is
natural to assume state-contingent adjustment. Dynamic mod-
els of state-contingent adjustment, however, often yield empir-
ically implausible results; Caplin and Spulber's [1987] conclu-
sion of monetary neutrality is a famous example. By contrast,
if firms face costs of collecting information and choosing opti-
mal plans, then it is more natural to assume that their adjust-
ment process is time-contingent. Price setters cannot react
between scheduled adjustments, because they are not collect-
ing the information and performing the calculations necessary
for that purpose.
Yet we must admit that information processing is more com-
plex than the time-contingent adjustment assumed here. Models
of bounded rationality are notoriously difficult, but it seems clear
that when circumstances change in large and obvious ways, peo-
ple alter the mental resources they devote to learning and think-
ing about the new aspects of the world. Developing better models
of how quickly people incorporate information about monetary
policy into their plans, and why their response is faster at some
times than at others, may prove a fruitful avenue for future
research on inflation-output dynamics.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1320 QUARTERLY JOURNAL OF ECONOMICS

APPENDIX: DETAILS OF SOLUTIONS

This Appendix explains the solutions of the three models
presented in the text.
A. The Derivation of the Sticky - Price Phillips Curve
From the equations for the adjustment price x t and the price
level p t , breaking the sum and using the law of iterated expecta-
tions, we obtain
(Al) xt = X/3-k (1 —
(A2) Pt = Xxt+ ( 1 — X)Pt-i•
But then solving for x t in (A2) and replacing in (Al) for x t and
x t±1 , together with the definition of pt = p t + ay e , yields the
desired expression for inflation presented in the text.
B. The Derivation of the Sticky - Information Phillips Curve
Begin with the equation for the price level derived in the text:

(A3) Pt = x E (1 — x)3Et-,(pt + etyt)•

,=0

By taking out the first term and redefining the summation index,
this equation can be written as

(A4) Pt = X(Pt + ayt) + A E (1 — + aye).

j =0

Analogous to equation (A3), the previous period's price level can

be written as

(A5) Pt 1 = X E (1 — + ayt_D•
,=0

Subtracting (A5) from (A4) and rearranging yields the following

equation for the inflation rate:

(A6) Trt= X(Pt+ otYt) + X E (1 — x + otAyi)

„=0

— x2 E (1 — x)JEt_i-,(Pt + an).
J=0

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1321

Now equation (A4) can be rearranged to show that

(A7) Pt [1_ xlYt = X E (1 — x)'Et-i-j(Pt + elyt)•

j =0

We now use equation (A7) to substitute for the last term in

equation (A6). After rearranging, we obtain

r aX
(A8) art =
[1_ jYtx x JE=0 ( 1 — (•rt + otA.Yt)•

This is the sticky-information Phillips curve presented in the text.

C. The Response of Output and Inflation in the Policy
Experiments
The three policy experiments we undertake can be described
as follows:
(El) An unexpected fall in the level of aggregate demand by
10 percent at date O. Thus, m t = —log (0.9) for t < 0 and
m t = 0 for t O.
(E2) An unexpected drop in the rate of money growth Am t at
date 0, from 2.5 percent per period to 0 percent. Thus,
m t = 0.025(1 + t) for t —1, m t = 0 for t O.
(E3) Same as (E2) but announced at date t = —8.
We focus on finding solutions for Pt as a function of m t . The
solution for y t then follows from the aggregate demand equation.
For the backward-looking model, the solution follows imme-
diately once the aggregate demand equation is used to substitute
out for y:
(A9) + R) = 2Pt-i — Pt-2 + Pnt.
This is a second-order difference equation. The associated roots
are [1 ± (-13) 1/2 ]/(1 + 13), which are complex (since R > 0),
generating the oscillatory behavior.
For the sticky-price model, rewrite the Phillips curve, using
the aggregate demand equation, as
(A10) EtPt+i — ( 2 + R)Pt + Pt-i = — Rmt.
This is an expectational difference equation, which can be solved
by the methods explained in Sargent [1987]. First, take expecta-
tions at t and express all expectations at t variables with an
asterisk. Then using the lag operator L, such that LE t p t =

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1322 QUARTERLY JOURNAL OF ECONOMICS

E t p t _ l and its inverse, the forward operator, F = L -1 such that

FE t p t = E t p t , i , reexpress (A10) as
(All) (F2 — (2 + (3)F + 1)Lp =
The quadratic (x 2 — (2 + (3)x + 1), has two positive roots: 0 and
1/0, such that (1 — 0) 2/0 = p. Pick 0 to correspond to the smaller
of the roots. Then (All) becomes
(Al2) (1 — = (1 — 0) 2 (1 — F0) -1 m1'.
But using the fact that 0 < 1, the inverse on the right-hand side
is well defined and can be expanded. Finally, because p t and p t _ l
are part of date t information set, we obtain the final solution:

(A13) Pt = OPt-i + ( 1 — 0) 2 E ORtmt+,•

i=o
For the policy experiment El, up to date 0, p t = m t = —log
(0.9). From 0 onwards, E t m t+ , = m t+ , = 0, so the price level is
given by the recursion p t = Op t - 1 with initial condition ;3_ 1 =
—log (0.9). For E2, p t = m t until t = —1, and after again p t =
Op t _ i , but now the initial condition is p = 0. Thus, p t = 0, t
0, and so y t = 0 at all t. For E3, in the period —8 t —1, then
the terms of the sum in the right-hand side of (A13) are E t m t+ , =
0.025(1 + t) for — 8 t + i —1 and E t m t± , = 0 for t +
0. After that, for t 0, p t = Opt-i.
Finally, consider the sticky-information model, as captured
by the equation:

(A14) Pt = X E (1 - x)JEt_,[(1- Opt+ unlit

j=0

The price level at time t 0 can then be broken into two

components, where the first includes price setters aware of the
new path for aggregate demand, and the second those agents who
decided on their prices before the change:

(A15) Pt = X E (1 - x).Et_ - a)Pt + cunt]

j=0

+ X E (1 - x)'Et_.,[(1 - a)Pt + am t].

j=t+1

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1323

Because the agents represented by the second term are still

setting prices based on their old information sets, their expecta-
tions are given by E t _fp t = E t _i m t = —log (0.9). As a result, the
second term reduces to —log (0.9) (1 — X)'. The agents repre-
sented by the first term have responded to the new path of
aggregate demand, so E t _i m, = 0, and because there is no
further uncertainty, E t _jp t = p t . Collecting terms and rearrang-
ing, we obtain the solution:
(A16) Pt = [—log (0.9)(1 — X)']/[1 — (1 —14(1 — (1 — X)')].
This equation gives the solution for the price level in the sticky-
information model under policy experiment El.
We can find the outcome under policy experiment E2 with
similar steps. Under E2, however, E t t = E t _j m t = 0.025(1 +
t) for t — j < O. Thus, the solution is
(A17) P t = [0.025(1 + t)(1 — X)']/[1 — (1 — a)(1 — (1 — X)')].
This equation gives the price level in the sticky-information
model under policy experiment E2.
Finally, for E3, for t < 0, the path is the same as expected by
all agents, so p, = m t = 0.025(t + 1) and y, = O. After date 0,
p, is given by (note the limit of the sums)
t+8

(A18) Pt = X E (1 - X)/Et-,[(1- a)Pt + ant]

j =o

+ X E (1 - x)'.E,_,[(1- a)p t + am t].

j=t+9
All else is the same as in E2. The solution follows as
(A19) P t = [0.025(1 + t)(1 — X)']/[1 — (1 — a)(1 — (1 — X)')].
This equation gives the path of the price level for the sticky-
information model under policy experiment E3.
D. Output and Inflation When Money Growth Is AR(1)
Suppose that Am, = pAm t _ i + E t , where E t is white noise
and Ipl < 1. It will prove convenient to write this in MA(so) form:
CO W.

(A20) Amt = E piE t _i or m t = EE pj€,_; _ k •

j=0 k=0 j=0

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1324 QUARTERLY JOURNAL OF ECONOMICS

Consider first the backward-looking model. First-differenc-

ing both sides of (A9), multiplying through by (1 — pL) and
rearranging yields the following AR(3) for the inflation rate:

(A21) = [1 + 13] 12 + PthTt-i

-

— (2p + 1) ,Tr t _ 2 + pir t _ 3 + 130.

From this equation we can calculate impulse responses and all

moments of inflation.
Consider now the sticky-price model. We find the general
solution of these rational expectation models by the method of
undetermined coefficients as outlined in Taylor [1985]. Because
the money growth rate is stationary, it is a reasonable conjecture
that the inflation rate is also stationary and so can be expressed
in the MA(00) general form:

(A22) =E or Pt — E E (t);Et_i_k,
j=0 k=0 j=0

where the (1).7 are coefficients to be determined Then realize that

E t [E t±i _j _ k } = E t+ ,_j _ k for i — j k and is zero otherwise. Using
the solution of the model in (A13),
CO 00 00 00

(A23)E E 1 j -j-k = 0
() E E 43JEt 1 k
k=0 j=0 k=0 j=0

+ (1— 0) 2 E et E E P +,-,-k•
,=0 j=0 k=max {i-j,0}

But then, because this must hold for all possible realizations of
E t matching coefficients on both sides of this equation yields for
the coefficient on E t :

1—0
(A24) (i)c, = (1 — 0) 2 E 0i E P = 1 — Op •
i=0 j=0

And for a general v, the coefficient on s t _ v :

v-1 i-Ev

(A25) E4 ; =0 E + ( 1— 0) E 0i E P i . 2
j=0 j=0 i=0 j=0

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1325

This yields

(1 - 0) 2 ] [ 1pv+i
(A26) (1),,_ (0 - 1) E [ 1 _p 1— 1 — Op] •
J =o

Equations (A22), (A24), and (A26) fully describe the stochastic

process of inflation. The impulse response of inflation for a unit
shock to aggregate demand is given by {kJ. The autocorrelation
coefficients of order j are then given by (see Hamilton [1994],
p. 52)

(A27) E E (1) 02 •
v=j o=0

Consider now the sticky-information model. Similarly to

(A22), conjecture the solution: ar t E -y i E t _, orp t = - - "Yz€t-z-k,
where the sums go from 0 to infinity Taking the relevant expec-
tations and substituting in (A8), the equation for the Phillips
curve, we obtain

00 CXD 00 CO CC

ctx
(A28) E = — EP EEt—k—i
i E E Et—k—i
i=0 xil i=0 k=0 i=0 k=0

+xE - x)i - a) E -yiE t _ i + a

j=0 i=j+i i =j+1

So, again matching coefficients,

(A29) 'Yo = aX/[1 - X(1 -

k
1-1
(A30) [
.yk = ax 1- x(i - E
0 a-
0
XY

k-1 k
x E ,yiE pi + pk E - xyl.
i=0 i =1 i =1

This provides the full characterization of the stochastic process

for inflation. Impulse responses, autocorrelations, and cross cor-
relations can be easily calculated.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1326 QUARTERLY JOURNAL OF ECONOMICS

E. Impulse Responses of Output and Population Correlations

between Output and the Change in Inflation
For the backward-looking model, corr (ir t+2 1Tt-2/ Yt )

corr [iTt+2 1Tt-2/ ( 1Tt ITt--1)/13] which we can evaluate using

(A21). Corr (,ITt+4 1t-4, Yt) follows likewise.
For the sticky-price model, note that output growth is given
from the quantity theory: Ay, = Am, - ar t = E (p" -
From this, we can obtain the MA(co) for output: y, = wi €,_i with
the recursion o = 0).7 _ 1 + p 3 - 4:1:n/ , initiated by co o = 1 - 4 0 . The
impulse response to a unit shock is given by the sequence {co y }. To
solve for the change in inflation 7r, - 7r,_ 4 , start with 7r, = E
4)j €,-j ; the coefficients in the MA(00) representation for the change
in inflation 7r, - art-4 = E are then given by ti = 4 - (130j_4
with t o = 4/ 0) 1 = 41) S 2 = 412, and t3 = 413. Given these results,
the population cross correlation between the change in inflation
and output, corr ('t+2 1Tt-2/Yt), is

(A31) E i Av +2
()

v=0 \ v =0 v=0

The cross-correlation corr ( iTt+ 4 — Irt-4,Yt) is derived in the same

fashion.
The derivation of the population cross correlations in the
sticky-information model is precisely the same, except we start
with 7r, = -y,E t _ i as the process for inflation.

DEPARTMENT OF ECONOMICS, HARVARD UNIVERSITY

REFERENCES
Abel, Andrew B., and Ben S. Bernanke, Macroeconomics, 3rd ed. (Reading, MA:
Addison-Wesley, 1998).
Akerlof, George A., and Janet L. Yellen. "A Near-Rational Model of the Business
Cycle with Wage and Price Inertia," Quarterly Journal of Economics, C
(1985), 823-838.
Ball, Laurence, "Credible Disinflation with Staggered Price Setting," American
Economic Review, LXXXIV (1994a), 282-289.
—, "What Determines the Sacrifice Ratio?" in Monetary Policy, N. G. Mankiw,
ed. (Chicago, IL: University of Chicago Press, 1994b), pp. 155-182.
, "Near Rationality and Inflation in Two Monetary Regimes," NBER Working
Paper No. 7988,2000.
Ball, Laurence, and David Romer, "Real Rigidities and the Non-Neutrality of
Money," Review of Economic Studies, LVII (1990), 183-203.
Barro, Robert, "Long-Term Contracting, Sticky Prices, and Monetary Policy,"
Journal of Monetary Economics, III (1977), 305-316.
Barsky, Robert B., "The Fisher Effect and the Forecastability and Persistence of
Inflation," Journal of Monetary Economics, XIX (1987), 3-24.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 STICKY INFORMATION VERSUS STICKY PRICES 1327

Begg, David K. H., and Isabella Imperato, "The Rationality of Information Gath-
ering: Monopoly," The Manchester School, LXIX (2001), 237-252.
Bernanke, Ben S., and Mark Gertler, "Inside the Black Box: The Credit Channel
of Monetary Policy Transmission," Journal of Economic Perspectives, IX
(1995), 27-48.
Blanchard, Olivier, Macroeconomics, 2nd ed. (Upper Saddle River, NJ: Prentice
Hall, 2000).
Blanchard, Olivier, and Nobuhiro Kiyotaki, "Monopolistic Competition and the
Effects of Aggregate Demand," American Economic Review, LXXVII (1987),
647-666.
Boschen, John F., and Charles L. Weise, "The Ex Ante Credibility of Disinflation
Policy and the Cost of Reducing Inflation," Journal of Macroeconomics, XXIII
(2001), 323-347.
Calvo, Guillermo A., "Staggered Prices in a Utility Maximizing Framework,"
Journal of Monetary Economics, XII (1983), 383-398.
Caplin, Andrew, and Daniel Spulber, "Menu Costs and the Neutrality of Money,"
Quarterly Journal of Economics, CII (1987), 703-725.
Carroll, Christopher, "The Epidemiology of Macroeconomic Expectations," NBER
Working Paper No. 8695,2001.
Clarida, Richard, Mark Gertler, and Jordi Gali, "The Science of Monetary Policy:
A New Keynesian Perspective," Journal of Economic Literature, XXXVII
(1999), 1661-1707.
Christiano, Lawrence J., Martin Eichenbaum, and Charles L. Evans, "Modeling
Money," NBER Working Paper No. 6371,1998.
Christiano, Lawrence J., Martin Eichenbaum, and Charles L. Evans, "Monetary
Policy Shocks: What Have We Learned and to What End?" Handbook of
Macroeconomics, J. B. Taylor and M. Woodford, eds. (Amsterdam, The Neth-
erlands• Elsevier, 1999), pp. 65-148.
Cooper, Russell, and Andrew John, "Coordinating Coordination Failures in
Keynesian Models," Quarterly Journal of Economics, CIII (1988), 441-463.
Cover, Thomas M., and Joy A. Thomas, Elements of Information Theory (New
York: John Wiley and Sons, 1991).
Dornbusch, Rudiger, Stanley Fischer, and Richard Startz, Macroeconomics, 8th
ed. (Boston, MA: McGraw-Hill, 2001).
Fischer, Stanley, "Long-term Contracts, Rational Expectations, and the Optimal
Money Supply Rule, Journal of Political Economy, LXXXV (1977), 191-205.
Friedman, Milton, "A Monetary and Fiscal Framework for Economic Stability,"
American Economic Review, XXXVIII (1948), 279-304.
Fuhrer, Jeffrey, and George Moore, "Inflation Persistence," Quarterly Journal of
Economics, CX (1995), 127-160.
Gabaix, Xavier, and David Laibson, "The 6D Bias and the Equity Premium
Puzzle," NBER Macroeconomics Annual: 2001 (Cambridge, MA: MIT Press,
forthcoming).
Gali, Jordi, and Mark Gertler, "Inflation Dynamics: A Structural Econometric
Analysis," Journal of Monetary Economics, XLIV (1999), 195-222.
Gordon, Robert J., "The Time-Varying Nairu and Its Implications for Economic
Policy," Journal of Economic Perspectives, XI (1997), 11-32.
Hall, Robert E., and John B. Taylor, Macroeconomics, 4th ed. (New York: Norton,
1993).
Hamilton, James, Time Series Analysis (Princeton, NJ: Princeton University
Press, 1994).
Koenig, Evan F., "Aggregate Price Adjustment: The Fischerian Alternative,"
unpublished paper, 1997.
Lucas, Robert E., Jr., "Econometric Testing of the Natural Rate Hypothesis, in O.
Eckstein, ed., The Econometrics of Price Determination (Washington, DC:
Board of Governors of the Federal Reserve System, 1972).
, "Some International Evidence on Output-Inflation Tradeoffs," American
Economic Review, LXIII (1973), 326-334.
Mankiw, N. Gregory, "Small Menu Costs and Large Business Cycles: A Macro-
economic Model of Monopoly," Quarterly Journal of Economics, C (1985),
529-537.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018
 1328 QUARTERLY JOURNAL OF ECONOMICS

, "The Inexorable and Mysterious Tradeoff between Inflation and Unemploy-

ment," Economic Journal, CXI (2001), C45–C61.
Mankiw, N. Gregory, and Ricardo Reis, "Sticky Information: A Model of Monetary
Nonneutrality and Structural Slumps," NBER Working Paper No. 8614,
2001; forthcoming in Knowledge, Information, and Expectations in Modern
Macroeconomics: In Honor of Edmund S. Phelps, P. Aghion, R. Frydman, J.
Stiglitz, and M. Woodford, eds. (Princeton, NJ: Princeton University Press).
McCallum, Bennett, "Comment," NBER Macroeconomics Annual: 1997 (Cam-
bridge, MA: MIT Press), pp. 355-359.
, "Stickiness: A Comment," Carnegie-Rochester Conference Series on Public
Policy, XLVIII (1998), 357-363.
Moscarini, Giuseppe, "Limited Information Capacity as a Source of Inertia," Yale
University, unpublished paper, 2001.
Okun, Arthur M., "Efficient Disinflationary Policies," American Economic Review
Papers and Proceedings, LXVIII (1978), 348-352.
Phelps, Edmund S., "Disinflation without Recession: Adaptive Guideposts and
Monetary Policy," Weltwirtschaftliches Archiv, C (1978), 239-265.
Rotemberg, Julio, "Monopolistic Price Adjustment and Aggregate Output," Review
of Economic Studies, XLIV (1982), 517-531.
Rotemberg, Julio, and Michael Woodford, "An Optimization-Based Econometric
Framework for the Evaluation of Monetary Policy," NBER Macroeconomics
Annual: 1997 (Cambridge, MA: MIT Press, 1997), pp. 297-346.
Sargent, Thomas J., "The Ends of Four Big Inflations," in Inflation: Causes and
Consequences, R. Hall, ed. (Chicago, IL: University of Chicago Press, 1982),
pp. 41-97.
, Macroeconomic Theory, 2nd ed. (New York, NY: Academic Press, 1987).
Sims, Christopher A., "Implications of Rational Inattention," Princeton Univer-
sity, unpublished paper, 2001.
Stock, James H., and Mark W. Watson, "Business Cycle Fluctuations in U. S.
Macroeconomic Time Series," Handbook of Macroeconomics, J. B. Taylor and
M. Woodford, eds. (Amsterdam, The Netherlands: Elsevier, 1999), pp. 3-64.
Taylor, John B., "Aggregate Dynamics and Staggered Contracts," Journal of
Political Economy, LXXXVIII (1980), 1-22.
, "New Econometric Approaches to Stabilization Policy in Stochastic Models of
Macroeconomic Fluctuations," in Handbook of Econometrics, Vol. 3, Zvi Grill-
ches and M. D. Intriligator, eds. (Amsterdam• North-Holland, 1985), pp.
1997-2055.
—, "Staggered Price and Wage Setting in Macroeconomics," Handbook of Mac-
roeconomics, J. B. Taylor and M. Woodford, eds. (Amsterdam, The Nether-
lands: Elsevier, 1999), pp. 1009-1050.
Woodford, Michael, "Imperfect Common Knowledge and the Effects of Monetary
Policy," Princeton University, 2001, forthcoming in Knowledge, Information,
and Expectations in Modern Macroeconomics: In Honor of Edmund S. Phelps,
P. Aghion, R. Frydman, J. Stiglitz, and M. Woodford, eds. (Princeton, NJ:
Princeton University Press).
Zbaracki, Mark J., Mark Ritson, Daniel Levy, Shantanu Dutta, and Mark Bergen,
"The Managerial and Customer Costs of Price Adjustment: Direct Evidence
from Industrial Markets," unpublished paper, Emory University, 2000.

Downloaded from https://academic.oup.com/qje/article-abstract/117/4/1295/1875955

by Universidade Federal da Bahia user
on 11 May 2018