First Welfare Theorem

Econ 2100 Fall 2018

Lecture 17, October 29

Outline
1 First Welfare Theorem
2 Preliminaries to Second Welfare Theorem
 Past De…nitions
A feasible allocation (^x ; y^ ) is Pareto optimal if there is no other feasible
allocation (x; y ) such that
xi %i x^i for all i and xi i x
^i for some i:
L
An allocation (x ; y ) and a price vector p 2 R+ form a competitive
equilibrium if
1 for each j = 1; :::; J: p yj p yj for all yj 2 Yj ;
2 for each i = 1; :::; I : P
xi %i xi for all xi 2 fxi 2 Xi : p xi p ! i + j ij p yj g ; and
P P P P P P
3
i xi i !i + j yj and if Ii =1 xli < Ii =1 ! li + Jj=1 ylj then pl = 0.

What is the relationship between competitive equilibrium and Pareto

e¢ ciency?
Is any competitive equilibirum Pareto e¢ cient? First Welfare Theorem.
This is about excluding something can Pareto dominate the equilibrium
allocation.
Is any Pareto e¢ cient allocation (part of) a competitive equilibrium? Second
Welfare Theorem.
This is about …nding prices that make the e¢ cient allocation an equilibirum.
First Welfare Theorem: Examples
Things seem easy
First Welfare Theorem: Edgeworth Box Example
Things seem easy
O2

x*

O1
First Welfare Theorem: Robinson Crusoe Example
Things seem easy
 First Welfare Theorem: Counterexample
An Edgeworth Box Economy
Consider a two-person, two-good exchange economy.
a’s utility function Ua (x1a ; x2a ) = 7;
b’s utility function Ub (x1b ; x2b ) = x1b x2b .
The initial endowments are ! a = (2; 0) and ! b = (0; 2).

CLAIM: xa = (1; 1), xb = (1; 1), and prices p = (1; 1) form a competitive
equilibirum.
a’s utility is maximized.
b’s utility when her income equals 2 is maximized (this is a Cobb-Douglas utility
function with equal exponents, so spending half her income on each good is
optimal).
xa + xb = (2; 2) = ! a + ! b .
Is this allocation Pareto optimal? No:
x^a = (0; 0) and x^b = (2; 2) Pareto dominates xa , xb since consumer a has the
same utility while consumer b’s utility is higher.

How do we rule examples like this out?

Need consumers preferences to be locally non satiated (there is always
something nearby that makes the consumer better o¤).
 Local Non Satiation
De…nition
A preference ordering %i on Xi is satiated at y if there exists no x in Xi such that
x i y.

De…nition
The preference relation %i on Xi is locally non-satiated if for every x in Xi and for
every " > 0 there exists an x 0 in Xi such that kx 0 xk < " and x 0 i x.
qP
L 2
Remember: ky zk = l =1 (yl zl ) is the Euclidean distance between
two points.

Remark
If %i is continuous and locally non-satiated there exist a locally non-satiated
utility function; then, any closed consumption set must be unbounded (or
there would be a global satiation point).
 Local Non Satiation and Walrasian Demand
Lemma
Suppose %i is locally non-satiated, and let xi be de…ned as:
xi %i xi for all xi 2 fxi 2 Xi : p xi wi g :
Then
x i %i x i implies p xi wi
and
xi i xi implies p xi > wi

If a consumption vector is weakly preferred to a maximal consumption bundle

(i.e. an element of the Walrasian demand correspondence), it cannot cost
strictly less.
If a consumption vector is strictly preferred to a maximal bundle, it must not
be a¤ordable
If not the consumer would have chosen it and been better-o¤.
Make sure you prove this as an exercise (easy, but still worth doing).
 First Welfare Theorem

Theorem (First Fundamental Theorem of Welfare Economics)

Suppose each consumer’s preferences are locally non-satiated. Then, any allocation
x ; y that with prices p forms a competitive equilibrium is Pareto optimal.

The invisible hand is Pareto e¢ cient.

This is true under pretty mild conditions on each preference relation.
Local non-satiation has bite: there is always a more desirable commodity
bundle nearby.
There is another assumption implicit in our framework: lack of externalities
(more later).
 Proof of the First Welfare Theorem (by contradiction)
Suppose not: there exists a feasible allocation x; y such such that xi %i xi for all i,
and xi i xi for some i.
P
xi %i xi implies p xi p ! i + j ij (p yj )
By local non satiation, P
xi i xi implies p xi > p ! i + j ij (p yj )
Therefore, summing over consumers !
X I XI XI X J XI X J
accounting
p xi > p !i + ij (p yj ) = p !i + p yj
i =1 i =1 i =1 j =1 i =1 j =1
Since each yj maximizes pro…ts at prices p , we also have
XJ XJ
p yj p yj :
j =1 j =1
Substituting this into the previous
! inequality: !
X I XI XJ I
X J
X
p xi > p !i + p yj p !i + p yj
i =1 i =1 j =1 i =1 j =1
But then x; y cannot be feasible since if it were we would have !
XI XI XJ X I XI J
X
xi !i + yj ) p xi p !i + p yj
i =1 i =1 j =1 i =1 i =1 j =1
which contradicts the expression above.
 First Welfare Theorem

Theorem (First Fundamental Theorem of Welfare Economics)

Suppose each consumer’s preferences are locally non-satiated. Then, any allocation
x ; y that with prices p forms a competitive equilibrium is Pareto optimal.

The theorem says that as far as Pareto optimality goes the social planner
cannot improve welfare upon a competitive equilibrium.
Conjecture
The theorem needs only a seemingly weak assumption to obtain a pretty strong
conclusion.
On the other hand, the important assumption of absence of externalities is
implicit in the way we set up the model.
An externality is present when preferences or pro…t depend on more than one’s
choices.
 Externalities: An Example
An Edgeworth Box Economy
Consider a two-person, two-good exchange economy.
uA (x1A ; x2A ; x1B ) = x1A x2A x1B ; uB (x1B ; x2B ) = x1B x2B .
A su¤ers from B ’s consumption of the …rst good.

CLAIM: xA = (1; 1), xB = (1; 1), and p = (1; 1) is a competitive equilibirum.

A cannot choose x1B so that is a constant in her utility function. Then A’s
utility is maximized by xA at prices p since this is a Cobb-Douglas utility
function with equal exponents and spending half your income on each good is
optimal.
B ’s utility when her income equals 2 is maximized (this is a Cobb-Douglas
utility function with equal exponents, so spending half her income on each good
is optimal).
xa + xb = (2; 2) = ! A + ! B .
xA ; x^B ) = ( 54 ; 23 ); ( 34 ; 43 ) is a feasible
Is this allocation Pareto optimal? No: (^
Pareto improvement:
52 3 1
UA (^xA ; x^1B ) = = > UA (xA ; x1B ) = 1 1 = 0
43 4 12
34
UB (^ xB ) = = 1 = UB (xB )
43
 First Welfare Theorem: Externalities

In the previous example, the …rst welfare theorem fails because A’s utility
depends on B’s consumption.
This is called a (negative) externality: the more B consumes of the good, the
worse-o¤ A becomes.
Among the assumptions implicit in our de…nition of preferences, one is
important for the …rst welfare theorem: there are no externalities in
consumption.
There can be also externalities in production.
Also, externalities can also be positive.
 Competitive Equilibrium and the Core

Theorem
Any competitive equilibrium is in the core.

Proof.
Homework. This is very very similar to the proof of the First Welfare Theorem.

A ‘converse’can be established in cases in which the economy is “large”, that

is, it contains many individuals.
That is called the core convergence theorem and I do not think we will have
time for it.
 Second Welfare Theorem: Preliminaries
This is a converse to the First Welfare Theorem.
The statement goes something like this: under some conditions, any Pareto
optimal allocation is part of a competitive equilibrium.
Next, we try to understand what these conditions must be. We state and prove
the theorem next class.
In order to prove a Pareto optimal allocation is part of an equilibrium one
needs to …nd the price vector that ‘works’for that allocation, since an
equilibrium must specify and allocation and prices.
First, we see a simple sense in which this cannot work: Pareto optimality
disregards the budget constraints.
This is …xed by adjusting the de…nition of equilibrium.
Then we see two counterexamples that stress the need for convexities.
These are …xed by assuming production sets and better-than sets are convex.
Finally, we see an example showing that boundary issues can pose problems.
This is …xed by, again, adjusting the de…nition of equilibrium.
 Second Welfare Theorem: Need Transfers

Picture
There is only one candidate price vector (the one tangent to both indi¤erence
curves).
However, this cannot work (B is not rich enough).
One can …x this by making B richer, giving her some money.
 Equilibrium With Transfers
De…nition
Given an economy fXi ; %i ; ! i gi =1 ; fYj gj =1 , an allocation x ; y and a price
I J

vector p constitute a price equilibrium with transfers if there exists a vector of

wealth levels
XI XI XJ
w = (w1 ; w2 ; :::; wI ) with wi = p !i + p yj
i =1 i =1 j =1
such that:
1 For each j = 1; :::; J: p yj p yj for all yj 2 Yj .
2 For each i = 1; :::; I : xi %i xi for all xi 2 fxi 2 Xi : p xi wi g
P
I P
I P
J
3 xi !i + yj , with pl = 0 if the inequlity is strict
i =1 i =1 j =1

Aggregate wealth is divided among consumers so that the budget constrants

are satis…ed.
How is the wealth of each consumer e¤ected? They get a positive or negative
transfer relative to the value of their endowment at the equilibrium
PJ prices.
A competitive equilibrium satis…es this: set wi = p ! i + j =1 ij (p yj ).
 Equilibrium With Transfers
Remark
The income transfers (across consumers) that achieve the budget levels in the
previous de…nition are:
2 3
XJ
Ti = wi 4p ! i + ij p yj 5
j =1

Summing over consumers, we get

2 3
X X I X I X
X J
Ti = wi 4 p !i + ij p yj 5
i i =1 i i =1 j =1
2 3
I
X J
X
= wi 4p !+ p yj 5
i =1 j =1

= 0
Transfers redistribute income so that the ‘aggregate budget’balances.
This is important: in a general equilibirum model nothing should be ‘outside’
the economy
Second Welfare Theorem: Need Convex Preferences
and Production Sets

Pictures
Counterexample I to Second Welfare Theorem
Need convex preferences for the Second Welfare Theorem

x*

Í1
Í2

x* is Pareto optimal, but one can see it is not an equilibrium at prices p

Counterexample II to Second Welfare Theorem
 Convexity
De…nition
A preference relation % on X is convex if the set % (x) = fy 2 X j y % xg is
convex for every x.

If x 0 and x 00 are weakly preferred to x so is any convex combination.

Convexity implies existence of an hyperplane that ‘supports’a consumer’s

better than set.

De…nition
In an exchange economy, an allocation x is supported by a non-zero price vector p
if: for each i = 1; :::; I
x 0 %i x i =) p x0 p xi

Convexity also yields an hyperplane that ‘supports’all producers’better than

set at the same time.
This hyperplane is the price vector that makes a Pareto optimal allocation an
equilibirum.
 Need Interior Allocations

Picture
Counterexample III to Second Welfare Theorem

x is Pareto optimal, hence we need prices to make it an equilibrium. The unique

price vector that can support it as equilibrium (normalizing the price of the …rst good
to 1) is p = (1; 0) .
The corresponding wealth is wB = (1; 0) (0; ! B ) = 0 .
However, xB is not maximial for consumer 1 at p since she would like
 Quasi-Equilibrium
To …x the ‘existence at the boundary’problem, we make a small change to the
de…nition of equilibrium.
De…nition
Given an economy fXi ; %i gi =1 ; fYj gj =1 ; !, an allocation x ; y and a price vector
I J

p constitute a quasi-equilibrium with transfers if there exists a vector of wealth

levels
X I XJ
w = (w1 ; w2 ; :::; wI ) with wi = p ! + p yj
i =1 j =1
such that:
1 For each j = 1; :::; J: p yj p yj for all yj 2 Yj .
2 For every i = 1; :::; I :
if x i xi then p x wi
P
I P
I P
J
3 xi = !i + yj
i =1 i =1 j =1

Make sure you see why this deals with the problem in the previous slide.
Any equilibrium with transfers is a quasi-equilibrium (make sure you check
 Next Class

Proof of the Second Welfare Theorem