Classical welfare economics

Ec9a1

January, 2024
 Individuals, i ∈ I = {1, . . . , I}, produce, exchange and consume com-
modities l ∈ L = {1, . . . , L}.
An individual is described by his utility function, u, over consumption
plans, x = (. . . , xl , . . .)′ , with domain X , the consumption set. The util-
ity function represents an underlying preference relation over consumption
plans, and, as a consequence, is ordinal: strictly monotonically increasing
transformations of the utility function represent the same preference rela-
tion; properties of the utility function not invariant to strictly monotonic
transformations are not ordinal.
Across individuals, the profile of utility functions is ⃗u = (. . . , ui , . . .).
Production is carried out by firms f ∈ F = {. . . , f, . . .}.
The technology of a firm is described by a production set, Y, of production
plans, y = (. . . , yl , . . .)′ . A commodity is employed as an input if yl < 0 or is
produced as an output if yl > 0.
Across firms, the profile of production P sets is Y⃗ = {. . . , Y f , . . .}.
The aggregate production set is Y a = f Y f .
The aggregate endowment in commodities is ea : (. . . , eal , . . .)′ .
An allocation of production plans is ⃗y = (. . . , P y f , . . .)′ , with y f ∈ Y f , for
every firm; the aggregate consumption plan y a = f y f .
An allocation of consumption plans is ⃗x = (. . . , xiP , . . .)′ , with xi ∈ X i , for
every individual; the aggregate consumption is x = i xi . a

An allocation is (⃗y , ⃗x), a pair of an allocation of production plans and an

allocation of consumption plans.
Associated with an allocation or an allocation of consumption plans, there
is an allocation of levels of utility, u(⃗y , ⃗x) = u(⃗x) = (. . . , ui (xi ), . . .)′ .
In an economy of pure exchange, Y a = {0}, and reference to production
is unnecessary.
A feasible allocation satisfies xa = y a + ea . The set of feasible allocations
is F.
A production set allows for inaction if 0 ∈ Y.
If a production set is convex and it allows for inaction, then, production
displays constant or decreasing returns to scale: y ∈ Y ⇒ λy ∈ Y, for all
0 ≤ λ ≤ 1.
A production plan y ≤ 0 disposes freely of commodities; a production set
allows for free disposal if y ≤ 0 ⇒ y ∈ Y.
In an economy of pure exchange, a consumption set allows for the free
disposal of commodities if x ∈ X whenever x ≥ x1 , for some consumption
plan, x1 ∈ X . If the consumption set allows for the free disposal of commodi-
ties, the utility function allows for the free disposal of commodities if it is
weakly monotonically increasing: ui (x) ≥ ui (x̃) whenever x ≥ x̃.

1
 A weakly feasible allocations satisfies xa ≤ y a + ea , and the set of weakly
feasible allocations is F+ . If the aggregate production set is closed, convex,
and it allows for free disposal, then, y1a ∈ Y a whenever y1a ≤ y2a , for some
y2a ∈ Y a , and, as a consequence, feasible and weakly feasible allocations
essentially coincide.
A production set does not allow the reversal of production if y ∈ Y and
−y ∈ Y only for y = 0; it does not allow for free production if y ≥ 0 ⇒ y = 0.
Even if the production sets of firms are closed, the aggregate production
set need not be; if the production sets of firms are closed and convex, then,
if the aggregate production set allows for inaction but does not allow for the
reversal of production it is closed.
A consumption set is bounded from below if there exists a bundle of
commodities, x, such that x ≥ x, for every x ∈ X .
If the consumption sets of individuals are closed and bounded from below,
while the aggregate production set is closed and convex and it does not
allow for free production, then, if there is only one firm (F = 1) or if the
aggregate production set does not allow for the reversal of production, the
set of (weakly) feasible allocations is compact.

Definition 1 (Pareto dominance and optimality). An allocation, (⃗y1 , ⃗x1 ),

pareto dominates (or is pareto superior to) another, (y2a , ⃗x2 ), if u(⃗x1 ) > u(⃗x2 );
strictly if u(⃗x1 ) ≫ u(⃗x2 ).
A feasible allocation, (⃗y , ⃗x), is pareto optimal if no feasible allocation
pareto dominates it; it is weakly pareto optimal if no feasible allocation
pareto dominates it strictly.

Pareto comparisons are ordinal.

Proposition 1. If the set of feasible allocations is non-empty and compact,

and the utility functions of individuals are continuous, then, pareto optimal
allocations exist.

Proof. Since the set of feasible allocations, F, is compact and non-empty:

there exists a feasible allocation (⃗y0 , ⃗x0 ) ∈ F, and since the utility functions
of individuals are continuous, the set of allocations F0 = {(⃗y , ⃗x) ∈ F :
u(⃗x) ≥ u(⃗x0 )} is non-empty and compact. Defined inductively, the sets Fi =
arg max{ui (xi ) : ⃗x ∈ Fi−1 } are non-empty and compact; by construction,
F0 ⫆ F1 ⫆ . . . Fi ⫆ . . . FI .
Allocations (⃗y ∗ , ⃗x∗ ) ∈ F1 are weakly pareto optimal.
Allocations (⃗y ∗ , ⃗x∗ ) ∈ FI are pareto optimal: if a feasible allocation,
(⃗y , ⃗x), pareto dominates (⃗y ∗ , ⃗x∗ ), then for some individual, uk (xk ) > uk (xk∗ ),
which implies that (y a , ⃗x) ̸∈ Fk−1 , since (⃗y ∗⃗x∗ ) ∈ Fk ; then, (⃗y , ⃗x) ̸∈ Fk−n−1 ,

2
since uk (xk−n ) ≥ uk (xk−n∗ ), and, as a consequence, (y a , ⃗x) ̸∈ F0 , a contradic-
tion.

For different initial allocations, (⃗y0 , ⃗x0 ), and different orderings of the set
of individuals, the procedure in the proof of proposition 1 yields all (weakly)
Pareto optimal allocations.
Under the premises of proposition 1, the set of weakly pareto optimal
allocations is compact; the set of pareto optimal allocations need not be
closed.
A utility function displays non-satiation if, for every consumption bundle,
i
the set of strictly preferred consumption bundles is not empty: U++ (x) = {x̃ :
i i
u (x̃) > u (x)} = ̸ ∅; it displays local non-satiation if every neighborhood, Vx ,
i
of a consumption bundle contains a strictly preferred bundle: U++ (x) ∩ Vx ̸=
∅.
If the utility functions of individuals are continuous and display local
non-satiation, then a weakly pareto optimal allocation with the consumption
plans of individuals in the interior of their consumption sets is pareto optimal;
this may fail if the consumption plan of some individual is on the boundary
of the consumption set.
A utility function with domain a convex consumption set, X , is quasi-
concave if, for every consumption bundle, the set of preferred or indifferent
consumption plans, U+ (x) = {x̃ : u(x̃) ≥ u(x)}, is convex: if x1 ∈ U+ (x) and
x2 ∈ U+ (x), then (λx1 + (1 − λ)x2 ) ∈ U+ (x), for every 0 ≤ λ ≤ 1; it is strictly
quasi-concave if (λx1 + (1 − λ)x2 ) ∈ U++ (x), whenever x1 ∈ U+ (x), x2 ∈
U+i (x), and 0 < λ < 1 or U+ (x) is strictly convex.
Quasi-concavity is an ordinal property, weaker than concavity, that is not
ordinal.
If a utility function is continuous, quasi-concave, and it displays local
non-satiation, then Bd U+ (x) = U (x) = {x̃ : u(x̃) = u(x)}, and Int U+ (x) =
U++ (x), for every consumption plan.
Proposition 2. If the aggregate production set and the consumption sets of
individuals are convex and the utility functions of individuals are concave,
then, a weakly pareto optimal allocation is a solution to the utilitarian opti-
misation problem P i i i
max i a u (x ),
s.t. ⃗x ∈ F,
for some a = (. . . , ai , . . .) > 0.
Proof. Since, the aggregate production set as well as consumption sets are
convex and utility functions are concave, the set of utility allocations Φ =

3
{ϕ : ϕ ≤ u(⃗x), (⃗y , ⃗x) ∈ F}, is convex. If a feasible, in particular a weakly
pareto optimal allocation, (⃗y ∗ , ⃗x∗ ), exists, the set Φ is not empty, and it
contains all ϕ ≤ u(⃗x∗ ).
If (⃗y ∗ , ⃗x∗ ) is a weakly pareto optimal allocation, then ϕ∗ = u(⃗x∗ ) is a
boundary point of Φ.
By the supporting hyperplane theorem, there exists a = P (. . . , ai , . . .) ̸=
∗ i i i
0,
P such that aϕ ≤ aϕ , for every ϕ ∈ Φ. It follows that i a u (x ) ≤
i i i∗
i a u (x ), for every feasible allocation ⃗ x and, as a consequence, ⃗x∗ is a
solution to the utilitarian optimisation problem for a = (. . . , ai , . . .). By con-
struction, ϕ ∈ Φ, whenever ϕ ≤ ϕ∗ , and, as a consequence, a > 0.

Ordinal properties guarantee the existence of a concave representation.

The weights, ai , in the utilitarian optimisation problem cannot be inter-
preted independently of the utility function ui . Variations in ai can compen-
sate for variations in the utility representation that do not reflect variations
in the underlying preferences.
A solution to the utilitarian optimisation problem is a weakly pareto
optimal allocation; if a ≫ 0, the allocation is pareto optimal.

Definition 2 (Irreducibility). At a feasible allocation, (⃗y0 , ⃗x0 ), the economy

is irreducible if, for any non-trivial partition, {I1 , I2 }, of the set of individu-
als, there exists a feasible allocation, (⃗y1 , ⃗x1 ), such that ui (xi1 ) ≥ ui (xi0 ), for
every i ∈ I1 , with strict inequality, ui (xi1 ) > ui (xi0 ), for some i ∈ I1 .

At a solution to the utilitarian optimisation problem with ai = 0, for some

individual, the economy is not irreducible. Equivalently, if the economy is
irreducible at a solution to the utilitarian optimization problem, then, a ≫ 0,
and the allocation is pareto optimal.
Example 1. Individuals are i = 1, 2, and commodities are l = 1, 2. The utility
function of the first individual is u1 = x1 and of the second u2 = x2 , both
with domain the set of non-negative consumption bundles. The aggregate
endowment is e = (1, 1), and there is no production. There is a unique pareto
optimal allocation, with x1 = (1, 0), and x2 = (0, 1). Weakly pareto optimal
allocations satisfy max{x11 , x22 } = 1. A feasible allocation, ⃗x = (x1 , x2 ), with
x1 = (1, x12 ), and x2 = (0, x22 ), where x12 > 0, is a solution to the utilitarian
optimization problem for a2 = 0, but it is not pareto optimal. The economy
is not irreducible at ⃗x.
Private ownership, distributes ownership of the aggregate endowment in
commodities and the profits of firms production across individuals.

4
P The endowment of an individual in commodities is ei = (. . . , eil , . . .), and
i a i i,f
Pi ei,f= e ; his endowment in shares in firms is θ = (. . . , θ , . . .), and
iθ = 1, for every firm.
Across individuals, the allocation of endowments in commodities is ⃗e =
(. . . , ei , . . .) and the allocation of endowments in shares in firms is θ⃗ =
(. . . , θi , . . .).
With production distributed across firms, an allocation of P production
plans is ⃗y = (. . . , y , . . .); the aggregate production plan is y = f y f .
f a

An allocation is, alternatively, a pair, (⃗y , ⃗x), of allocations of production

and consumption plans.
Autarky is feasible if 0 ∈ Y a and, for every individual, ei ∈ X i ; in partic-
ular, the allocation (0, ⃗e) is feasible.
In a market economy, commodities are exchanged at competitive prices.
Prices of commodities are p = (. . . , pl , . . .); the value of a bundle of com-
modities, x = (. . . , xl , . . .), at prices of commodities p = (. . . , pl , . . .) is
X
px = p l xl .
l

The profit maximization problem of a firm at prices p is

max py,
s.t. y ∈ Y.

The solution to the profit maximization is y(p), and it defines the supply
correspondence of the firm.
The profit of the firm at prices p is

π(p) = py(p),

which is well defined even if y(p) is not a singleton; if y(p) = ∅, then π(p) =
−∞.
The dividend income of an individual is
X
π i (p) = θi,f π f (p).
f

The aggregate supply is y a (p) = f y f (p), and it defines the aggregate

P
supply correspondence.
An aggregate production plan, y aP
, maximizes profit over the aggregate
production set Y if and only if y = f y f and y f ∈ y f (p), for every firm.
a a

Revenue for an individual is τ.

5
 The optimization problem of an individual at prices p and revenue τ i is
max u(x),
s.t. px ≤ τ.

The solution to the optimization problem is x(p, τ ), and it defines the

demand correspondence of the individual.
At a solution, x0 ∈ x(p, τ ), to the optimization problem,

u(x) > u(x) ⇒ px > τ i ;

if the utility function displays local non-satiation, px0 = τ i .

Across individuals, a distribution of revenue is ⃗τ = (. . . , τ i , . . .); aggregate
revenue is τ a = i τ i .
P
The aggregate demand is xa (p, ⃗τ ) = i xi (p, τ i ), and it defines the ag-
P
gregate demand correspondence.
Definition 3 (Competitive equilibrium). A competitive equilibrium is (p∗ ,
⃗y ∗ , ⃗x∗ ), prices and a feasible allocation, such that y f ∗ ∈ y f (p∗ ), for every firm,
and xi∗ ∈ xi (p∗ , p∗ ei + π i (p∗ )), for every individual.
A competitive equilibrium with a distribution of revenue is (p∗ , ⃗τ ∗ , ⃗y ∗ , ⃗x∗ ),
prices, a distribution of revenue and a feasible allocation, such that y f ∗ ∈
y f (p∗ ), for every firm and xi∗ ∈ xi (p∗ , τ i∗ ), for every individual.
At a competitive equilibrium, p∗ (y a (p∗ ) + ea ) ≤ τ a∗ ; if the utility function
of every individual displays local non-satiation, p∗ (y a (p∗ ) + ea ) = τ a .
If the aggregate production set or, alternatively, the utility function of
some individual allow for the free disposal of commodities, then, competitive
equilibrium prices are non-negative: p∗ ≥ 0.
If in addition, the utility function of some individual displays non-satiati-
on, then, competitive equilibrium prices are positive: p∗ > 0.
Proposition 3 (The first theorem of welfare economics). A competitive equi-
librium allocation with a distribution of revenue is weakly pareto optimal.
Proof. If a feasible allocation, (⃗y , ⃗x) = (. . . , y f , . . . , xi , . . .), strictly pareto
dominates a competitive equilibrium allocation, (⃗y ∗ , ⃗x∗ ) = (. . . , y f ∗ , . . . , xi∗ ,
. . .), then, since xi∗ is a solution to the optimization problem at prices p∗
and revenue τ i∗ , while ui (xi ) > ui (xi∗ ), for every individual, p∗ xi > τ i∗ ≥
p∗ xi∗ . Aggregation yields p∗ xa > p∗ xa∗ = p∗ (y a∗ + ea ). Since y a∗ maximizes
profit over the aggregate production set at competitive equilibrium prices,
p∗ y a∗ ≥ p∗ y a . It follows that p∗ xa > p∗ xa∗ ≥ p∗ (y a + ea ), which contradicts
xa = y a + ea , the feasibility of the allocation (⃗y , ⃗x).

6
 A competitive equilibrium allocation need not be pareto optimal.
At prices p, and a level of utility, u, the expenditure minimization problem
of an individual is
min px,
s.t. u(x) ≥ u.
The solution to the expenditure minimization problem is xc (p, u), and it
defines the compensated demand correspondence.
A consumption plan, x, is a minimum wealth point for the individual at
prices p if x ∈ arg min{px : x ∈ X i .}
At a solution, , x0 ∈ xc (p, u), to the expenditure minimization problem,
ui (x) ≥ ui (xi ) ⇒ px ≥ pxi ;
if the consumption set is convex, the utility function is continuous, and x0 is
not a minimum wealth point, then u(x0 ) = u.
Lemma 1. If the utility function of an individual displays local non-satiation,
then, a solution, x0 , to the optimization problem at prices of commodities and
revenue (p, τ ) is a solution to the expenditure minimization problem, with
u = u(x0 ).
Proof. If not, there exists a bundle of commodities, x1 , such that u(x1 ) ≥
u(x0 ), and px1 < τ. Since the utility function displays local non-satiation,
there exists a bundle, x2 , such that u(x2 ) > u(x1 ) ≥ u(x1 ) and px2 ≤ τ
— it suffices to choose x2 in an appropriate neighborhood of x1 . This is a
contradiction, since x1 is a solution to the optimization problem.

If the utility function of an individual displays local satiation, a solution

to the utility maximization problem need not be a solution to the expenditure
minimization problem.
Corollary 1. If the preferences of individuals display local non-satiation,
then, a competitive equilibrium with a distribution of revenue is pareto opti-
mal.
Proof. If a feasible allocation, (⃗y , ⃗x), pareto dominates a competitive equi-
librium allocation, (⃗y ∗ , ⃗x∗ ), then, for every individual, p∗ xi ≥ τ i∗ , since
preferences display local non-satiation and xi∗ is a solution to the expen-
diture minimization problem, while ui (xi ) ≥ ui (xi∗ ); if, for some individ-
ual, ui (xi ) > ui (xi∗ ), then p∗ xi > τ i . Since, for some individual preference
is indeed, strict, aggregation and profit maximization by firms imply that
p∗ xa > p∗ (y a + ea ), which contradicts the feasibility of the allocation (⃗y , ⃗x).

7
 The pareto optimality of competitive equilibrium allocations is of rele-
vance if competitive equilibria exist.

Proposition 4. If

1. production sets allows for inaction,

2. the aggregate production set is closed and convex, it allows for free
disposal, but it does not allow for the reversibility of production or for
free production,

3. consumption sets are closed, convex and bounded from below,

4. utility functions are continuous, quasi-concave and display local non-

satiation,

5. endowments in commodities belong to the consumption sets of individ-

uals, while their endowments in shares in firms are non-negative,

then, competitive equilibria exist if the consumption set of every individual

contains a consumption plan strictly inferior to the initial endowment: xi ≪
ei .

The aggregate, compensated demand correspondence, xa , is defined by

xac (p, ⃗u) = i xic (p, ui ).
P
A compensated equilibrium is (p∗ , ⃗y ∗ , ⃗x∗ ), prices, p∗ ̸= 0, and a feasible
allocation, such that y f ∗ ∈ y f (p∗ ), for every firm, and xi∗ ∈ xic (p∗ , ui (xi∗ )),
for every individual.
If prices are allowed to vanish, p = 0, a compensated equilibrium is of no
interest: any feasible allocation is a compensated equilibrium allocation with
p = 0.
If the utility function of some individual allows for the disposal of com-
modities, compensated equilibrium prices are positive: p∗ > 0.
The condition on the endowments of individuals that guarantees the ex-
istence of competitive equilibria is restrictive.

Corollary 2. Under the premises of proposition 5, if the consumption sets

of individuals allow for the free disposal of commodities, then, compensated
equilibria exist.

The economy is irreducible if autarky is feasible and the economy is irre-

ducible at every feasible allocation, (⃗y , ⃗x), such that ⃗u(⃗x) ≥ ⃗u(⃗e).

Corollary 3. Under the premises of proposition 4, if

8
 1. consumption sets allow for the free disposal of commodities,

2. the endowments of individuals are positive: ei > 0 and the consumption

set contains a plan xi = λei , for some 0 > λi < 1

3. there is a feasible (⃗y0 , ⃗x0 ) such that all commodities are in positive sup-
ply ea + y0a ≫ 0,

then, if the economy is irreducible, competitive equilibria exist.

Competitive equilibria or compensated equilibria support (weakly) pareto

optimal allocations.

Proposition 5 (The second theorem of welfare economics). If the aggre-

gate production set and the consumption sets of individuals are convex, and
the utility functions of individuals are quasi-concave and display local non-
satiation, then, a weakly pareto optimal allocation is a compensated equilib-
rium allocation.

Proof. For an allocation, (⃗y0 , ⃗x0 ), the set of aggregate endowments suffi-
cient to allow for pareto superior allocations is E a (⃗y0 , ⃗x0 ) = {ẽa0 : u(⃗x) ≫
u(⃗x0 ), for some allocation with xa = ea0 + y a }.
Since the aggregate production set and the consumption sets are convex,
while the utility functions are quasi-concave, the set E a (⃗y0 , ⃗x0 ) is convex.
If a feasible allocation, (⃗y ∗ , ⃗x∗ ), is weakly pareto optimal, then, ea ̸∈
E a (⃗y ∗ , ⃗x∗ ).
By the supporting hyperplane theorem, there exists p∗ ̸= 0, such that
p e0 ≥ p∗ ea , for every ea0 ∈ E a (⃗x∗ ).
∗ a

If for some individual, uh (x) ≥ uh (xh∗ ), there exists a sequence of al-

locations, ((⃗y ∗ , ⃗xn ) : n = 1, . . .), modified consumption plans, such that
u(⃗xn ) ≫ u(⃗x∗ ), and limn→∞ xin = xi∗ , for i ̸= h, while limn→∞ xhn = x;
such a sequence exists, since the utility functions of individuals display lo-
cal non-satiation; the associated sequence of aggregate endowments is (ean =
xhn + i̸=h xin − f y f ∗ : n = 1, . . . .)
P P
Along the sequence, ean ∈ E a (⃗y ∗ , ⃗x∗ ), and, as a consequence, p∗ ean ≥ p∗ ea =
xh∗ + i̸=h xi∗ − f y f ∗ . Since limn→∞ xin = xi∗ , for i ̸= h, the consumption
P P

plan x satisfies p∗ x ≥ p∗ xi∗ .

With uh (x) ≥ uh (xh∗ ) ⇒ p∗ x ≥ pxh∗ , the consumption plan xh∗ is a
solution to the expenditure minimization problem of individual h at prices
of commodities p∗ and level of utility uh∗ = uh (xh∗ ).
The same argument applies to every individual, and, as a consequence,
for every individual,xi∗ ∈ xic (p∗ , ui (xi∗ )).

9
 Similarly, if, for some firm, p∗ y g > p∗ y g∗ , there exists a sequence of allo-
cations, ((⃗yn , ⃗xn ) : n = 1, . . .), with modified consumption plans as well as a
modified production plan, y g , for firm g, while the production plans of other
firms remain unchanged, such that u(⃗xn ) ≫ u(⃗x∗ ), and limn→∞ xin = xi∗ ,
for everyP individual; Pthe associated sequence of aggregate endowments is
a i g f∗
(en = i xn − y − f ̸=f y : n = 1, . . . .)
P Along the sequence, ean ∈ E a (⃗y ∗ , ⃗x∗ ), and, as a consequence, p∗ ean ≥ p∗ ea =
i∗ g∗
− f ̸=i y f ∗ . Since limn→∞ xin = xi∗ , the production plan y g
P
ix − y
satisfies −p y ≥ p∗ y g∗ , a contradiction.
∗ g

With p∗ y g∗ ≥ p∗ y for any production plan y ∈ Y g , the production plan

y g∗ is a solution to the profit maximization problem of firm g at prices p∗ .
The same argument applies to every firm, and, as a consequence, for every
firm,y f ∗ ∈ y f (p∗ ).
It follows that (p∗ , ⃗y ∗ , ⃗x∗ ) is a compensated equilibrium.

A pareto optimal allocation need not be a competitive equilibrium allo-

cation.
Example 2. Individuals are i = 1, 2, and commodities are l = 1, 2. the util-
ity function of the first individual is u1 = x1 , and of the second u2 = x2 ,
both with domain the set of non-negative consumption bundles. The aggre-
gate endowment is e = (1, 1). Weakly pareto optimal allocations, ⃗x∗ , satisfy
max{x1∗ 2∗
1 , x2 } = 1, while the unique pareto optimal allocation obtains for
x1∗ 2∗
1 = x2 = 1.
The only prices that support the consumption bundle of individual 1 at
∗ ∗
a weakly pareto optimal allocation with x1∗ 1 < 1 are p = (p1 , 0) > 0, where
the utility maximization problem of individual 2 has no solution.
The economy is not resource related at any weakly pareto optimal allo-
cation.
For a modification of the utility function of individual 1, a feasible alloca-
tion with x1∗ 2∗
1 = 1 and x2 < 1 to be pareto optimal, and still not a competitive
equilibrium allocation: it suffices that, along the indifference curve of individ-
ual 1 through (1, x1∗ 1 1
2 ), the marginal rate of substitution (∂u /∂x2 )/(∂u /∂x1 )
2∗ 2∗
be equal to 0, for x2 ≥ x2 , but positive for x2 < x2 .
A modification of the utility function of individual 1 to

 x1 if 0 ≤ x1 ≤ 21 ,
u1 (x) = 1
if 12 ≤ x1 ≤ 1,
 2 1
x − 2 if 1 ≤ x1 ,
introduces local satiation and allows for (weakly)pareto optimal allocations
that either competitive or quasi-equilibrium allocations.

10
Lemma 2. If the consumption set of an individual is convex and the utility
function is continuous, then a solution, xi , to the expenditure minimization
problem at prices of commodities and level of utility (p, ui ) is a solution to the
optimization problem with revenue τ i = pxi , as long as xi is not a minimum
wealth point at prices p.
Proof. Since xi is not a minimum wealth point, there exists a consumption
plan, x, such that px < τ i . If there exists a consumption plan, x̂, such that
ui (x̂) > ui (xi ), since the utility function is continuous and the consumption
set is convex, there exists 0 < λ, such that ui (xλ ) > ui (xi ), where xλ = (1 −
λ)x̂ + λx. Since pxλ < τ, while xi is a solution the expenditure minimization
problem, this is a contradiction.

At minimum wealth point or with a consumption set that is not convex,

a solution to the expenditure minimization problem need not be a solution
to the optimization problem of an individual.
Corollary 4. If, in addition to the premises of proposition 5,

1. consumption sets allow for the free disposal of commodities,

2. the endowments of individuals are positive: ei > 0 and the consumption
set contains a plan xi = λei , for some 0 > λi < 1
3. there is a feasible (⃗y0 , ⃗x0 ) such that all commodities are in positive sup-
ply ea + y0a ≫ 0,

then, a weakly pareto optimal allocation at which the economy is irreducible

is a competitive equilibrium allocation.
Proof. By proposition 2, a weakly pareto optimal allocation, ⃗x∗ is a com-
pensated equilibrium allocation; for every individual, xi∗ is a solution to the
expenditure minimization problem at prices p∗ , and utility level ui∗ = ui (xi∗ ).
Since the utility function of some individual allows for the disposal of
commodities, while (p∗ , ⃗x∗ ) is a compensated equilibrium, p∗ > 0.
The allocation ⃗x∗ is a competitive equilibrium allocation if, for every
individual, xi∗ is a solution to the expenditure minimization problem at prices
p∗ , and revenue τ i = p∗ xi∗ .
By lemma 2, it suffices that τ i ̸= min{p∗ x : x ∈ X ⟩ }; since p∗ > 0, while,
for every consumption bundle, λx ∈ X i , for some 0 < λ < 1, it suffices that,
for every individual, p∗ xi∗ = τ i > 0.
The partition {I1 , I2 } of the set of individuals is defined by I1 = {i :
p∗ xi∗ > 0}, and I2 = {i : p∗ xi∗ > 0; since ea ≫ 0, p∗ > 0, and, as a

11
consequence, τ a = p∗ ea > 0, some individual has positive revenue at ⃗τ and
I1 ̸= ∅. If I2 = ∅, no individual is at a minimum wealth point at (p∗ , ⃗τ ),
and utility maximization follows from expenditure minimization for every
individual. If I2 ̸= ∅, the partition is non-trivial, and, since the economy is
irreducible at the allocation ⃗x∗ , there exists a feasible allocation, ⃗x1 , such that
ui (xi ) ≥ ui (xi∗ ), for every individual i ∈ I1 , with strict preference for some.
For individuals i ∈ I1 , since xi∗ is a solution to the optimization problem at
prices p∗ and revenue τ i , if ui (xi1 ) > ui (xi∗ ), then p∗ xi > τ i while, since xi∗ is
a solution to the expenditure minimization problem at prices p∗ and utility
level ui∗ = ui (xi∗ ), if ui (xi ) ≥ ui (xi∗ ) then p∗ xi ≥ τ i , For individuals i ∈ I2 ,
since τ i = 0, while xi2 ≥ 0 and p∗ > 0, a fortiori p ∗ xi ≥ τ i . Aggregation over
individuals yields p∗ x̃a1 > τ a = p∗ ea , which contradicts the feasibility of the
allocation ⃗x1 .

12
References
G. Debreu. Theory of Value. Wiley, 1959.

T. C. Koopmans. Three Essays of the State of Economic Science. McGraw

Hill, 1957.

D. M. Kreps. Notes on the Theory of Choice. Westview Press, 1988.

D. M. Kreps. Microeconomic Foundations I: Choice and Competitive Mar-

kets. Princeton University Press, 2013.

A. Mas-Colell, M. Whinston, and J. R. Green. Microeconomic Theory. Ox-

ford University Press, 1995.

13