General equilibrium analysis:
Welfare results in pure exchange
Lectures in Microeconomic Theory
Fall 2010, Part 12
07.07.2010
G.B. Asheim, ECON4230-35, #12
Pareto efficiency
Edgeworth box
The initial endowment is
not Pareto-efficient,
since both consumers
can be made better off
by moving to x (x1 , x 2 )
21
22 x ( x , x )
1
2
x (x1 , x 2 )
(1 , 2 )
12
1
is Paretoefficient, since no consumer can be made better
off without making the
other worse off.
11
07.07.2010
2
G.B. Asheim, ECON4230-35, #12
Illustrating the first welfare theorem
First welfare theorem: Assume that
Any Walrasian equilibriumxis (x1 , x 2 )
Pareto-efficient. is a Walrasian
x2
x2
x1
p ( p1 , p2 )
x1
equilibrium.
Then it is not
feasible to make
both consumers
better off.
Why?
07.07.2010
G.B. Asheim, ECON4230-35, #12
3
1
Illustrating the second welfare theorem
Assume that
Second welfare theorem:
x (x1 , x 2 )
Any Pareto-efficient allocation
is Pareto efficient.
can be implemented as a
If utility functions
Walrasian equilibrium.
x
are quasi-concave,
i
then x can be
implemented as
a Walrasian
equilibrium
How?
07.07.2010
4
G.B. Asheim, ECON4230-35, #12
Why is quasi-concavity assumed in
the second welfare theorem?
x ( x1 , x 2 )
is still Pareto
efficient, but the
allocation
ll ti cannott be
b
implemented as a
Walrasian equilibrium because of
consumer 1’s nonconvex preferences.
x
Kinks — no problem.
07.07.2010
5
G.B. Asheim, ECON4230-35, #12
Weak and strong Pareto efficiency
A feasible allocation ( x 1 , x 2 )
is weakly Pareto - efficient
u2
if there is no other feasible
allocation ( x 1 , x 2 ) such that
u i ( x i ) u i ( x i ) for both i .
(u(1u(1x(1x),1 ),
u 2u(2x(2x))2 ))
A feasible allocation ( x 1 , x 2 )
is strongly Pareto - efficient
(u1 (x1 ), u2 (x 2 ))
if there is no other feasible
u1
allocation ( x 1 , x 2 ) such that
u i ( x i ) u i ( x i ) for both i , and u i ( x i ) u i ( x i ) for one i .
07.07.2010
G.B. Asheim, ECON4230-35, #12
6
2
Equivalence of weak and strong PE
Assumption : For each i , u i is (1) continuous
(2) strictly quasi - concave, and
(3) monotone (i.e., x x implies u i ( x ) u i ( x ) ).
Result. Under the assumption on the utility functions,
an allocation
c
iss w
weaklyy Pareto efficient
c
if and
d onlyy if
it is strongly Pareto efficient.
Proof.
If ( x 1 , x 2 ) is strongly PE, then ( x 1 , x 2 ) is weakly PE.
If ( x 1 , x 2 ) is not strongly PE, then there is ( x 1 , x 2 ) s.t.
u i ( x i ) u i ( x i ) for both i , and u i ( x i ) u i ( x i ) for one i .
By transferri ng from i, both can be made better off.
07.07.2010
7
G.B. Asheim, ECON4230-35, #12
First welfare theorem
Result. If ( x 1 , x 2 , p ) is a Walrasian equilibr.,
then ( x 1 , x 2 ) is weakly Pareto efficient.
Proof. Suppose ( x 1 , x 2 , p ) is a Walrasian equilibr.,
where ( x 1 , x 2 ) is not weakly Pareto efficient.
Then there is a feasible allocation ( x 1 , x 2 ) such that
u i ( x i ) u i ( x i ) for both i . Hence,
1 2 x 1 x 2 and p x i p i for both i .
This leads to a contradict ion :
p 1 2 p x 1 x 2 p x 1 p x 2 p 1 p 2 .
07.07.2010
8
G.B. Asheim, ECON4230-35, #12
Second welfare theorem
Result. Suppose the assumption on the utility functions
holds. If the feasible allocation ( x 1 , x 2 ) is PE, then there
is a p such that ( x 1 , x 2 , p ) is a Walrasian equilibr.
x12 x22
Set of aggregate consumption vectors that
lead to a Pareto
improvement.
P
x1 x2
p ( p1 , p2 )
x11 x12
07.07.2010
G.B. Asheim, ECON4230-35, #12
9
3
Second welfare theorem — sketch of proof
The set P is convex since the utility
functions are quasi-concave.
x12 x22
The monotonicity of utility functions
combined with strict quasi-concavity
leads to the existence of p 0 such
that px p x 1 x 2 for all x P .
P
x1 x 2
p ( p1 , p2 )
The continuity of utility functions leads
to px p x 1 x 2 for all x P .
x11 x12
Suppose ( x 1 , x 2 , p ) is not a Walrasian equilibrium with ( x 1 , x 2 )
as initial endowments. Then there exists i and x i such that
u i ( x i ) u i ( x i ) and px i px i .
Contradiction: p x i x i p x i x i and x i x i P .
07.07.2010
G.B. Asheim, ECON4230-35, #12
10
Welfare maximization
Pareto efficiency does not yield a specific allocation.
This can be resolved by assuming the existence of a
social welfare function, W ( u1 , u 2 ) , that is
increasing in each of its arguments.
Result If ( x 1 , x 2 ) maximizes a social welfare
Result.
function, then ( x 1 , x 2 ) is strongly Pareto efficient.
Proof. If it would have been possible to increase u i
without decreasing the utility of the other consumer,
then it would have been possible to increase W ( u1 , u 2 ).
07.07.2010
G.B. Asheim, ECON4230-35, #12
11
Supporting a strongly PE allocation
u2
(u1 (x1 ), u2 (x2 ))
U
(a1 , a2 )
u1
07.07.2010
Question: If ( x 1 , x 2 ) is
strongly PE, does there
exists a social welfare function a1 u1 ( x 1 ) a 2 u 2 ( x 2 )
such that maximization of
this social welfare function
leads to ( x 1 , x 2 ) ?
Answer: The utility possibility set U must be convex.
Concave utility functions
ensure this in pure exchange.
G.B. Asheim, ECON4230-35, #12
12
4
Determining the supporting welfare weights
a i
1
i
1
v i ( p , px i )
mi
u2
v 2 (p , m 2 )
a1 2
m 2
slope :
v1 ( p , m1 )
a 2 1
m1
a1
07.07.2010
v1 ( p , m1 )
m1
a 2
(u1 (x 1 ), u 2 (x 2 ))
U
( a1 , a2 )
u1
Intuition?
v 2 (p , m 2 )
m 2
G.B. Asheim, ECON4230-35, #12
13
To recapitulate:
Competitive equilibria are always Pareto efficient.
Pareto efficient allocations are competitive equilibria
with quasi-concave
w
q s co c ve u-functions
c o s & endowm.
e dow . redistrib.
ed s b.
Welfare maxima are always Pareto efficient.
Pareto efficient allocations are welfare maxima with
concave u-functions for some choice of welfare weights.
07.07.2010
G.B. Asheim, ECON4230-35, #12
14
5