Unit 2: Exchange Equilibrium

Andrew Caplin and Srijita Ghosh

Fall 2023

1 / 62
Unit 2: Exchange Equilibrium

I Next three lectures on exchange equilibrium

I Consider all markets together: no production yet
I In equilibrium,
I everyone chooses what’s best for them or no one wants to
deviate
I all parameters are chosen within the model
I Goal is to solve for prices
I Diamond-water paradox as motivation

2 / 62
Unit 2: Exchange Equilibrium

I First two sections covered in lecture

I 2.1: The Edgeworth Box
I 2.2: Equilibrium and Pareto Optimality
I 2.3*: Technical Proofs of Walras’ Law and Pareto Optimality of
Equilibrium
I Technical Proofs at the end (time permitting) for those who are
curious (not in tests)

3 / 62
2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1. The Edgeworth Box

2.1.1 Endowment Economy

4 / 62
 2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1.1 Endowment Economy

I First new element today: income generated by sale of

endowments e1 , e2 ≥ 0,
I At least one ei strictly positive for strictly positive income
I Labor often in practice: goods 1 and 2 for abstract treatment
I With endowment, income is a function of prices of two goods
p1 , p2 > 0 :
m (p1 , p2 ) ≡ p1 e1 + p2 e2 > 0.
with p1 , p2 > 0.

5 / 62
 2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1.1 Endowment Economy

I Price change pivots the budget line
I Endowment bundle always affordable

Figure: price of good 1 decreases 6 / 62

 2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1.1 Endowment Economy

I Formal optimization for individual with utility function u (x1 , x2 )

max u (x1 , x2 ) subject to p1 x1 + p2 x2 ≤ m (p1 , p2 )

x1 ,x2

I Demand: function only of the two prices not income, x̂1 (p1 , p2 )
.
I With u strictly monotone and strictly concave, solution exist
and is unique for p1 , p2 > 0

7 / 62
 2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1.1 Endowment Economy

I Constraint set unaffected by multiplying all prices by k > 0 :

scales up income and prices in proportion,

p1 x1 + p2 x2 ≤ p1 e1 + p2 e2 ⇐⇒ kp1 x1 + kp2 x2 ≤ kp1 e1 + kp2 e2

I Hence optimal solution that defines demand independent of

k>0.
I Can always divide by k = p +1 p to make prices add to 1.
1 2

8 / 62
 2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1.1 Endowment Economy

I Given p1 ∈ (0, 1) set p2 = 1 − p1 ∈ (0, 1) and write optimum

as,

x̂1 (p1 ) = x̂1 (p1 , 1 − p1 );

x̂2 (p1 ) = x̂2 (p1 , 1 − p1 )

I With endowment income also as a function of p1 ∈ (0, 1) as:

m (p1 ) ≡ p1 e1 + (1 − p1 )e2 > 0.

9 / 62
 2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1.1 Endowment Economy

I Workhorse is Cobb-Douglas as always,

u (x1 , x2 ) = α1 ln x1 + (1 − α1 ) ln x2

I Given p1 ∈ (0, 1) and α1 ∈ (0, 1) demand solves:

max α1 ln x1 + (1 − α1 ) ln x2 subject to p1 x1 + (1 − p1 )x2 = m (p1 )

x1 ,x2

10 / 62
 2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1.1: Endowment Economy

I Introduce Lagrangian λ on budget equality and define,

L(x1 , x2 , λ) = α1 ln x1 + (1 − α1 ) ln x2 -λ [p1 x1 + (1 − p1 )x2 − m (p

I Solve unconstrained problem:

max L(x1 , x2 , λ).

x1 ,x2 ,λ

11 / 62
 2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1.1 Endowment Economy

I Set all partial derivatives to zero to solve for optimum x̂1 (p1 ),
x̂2 (p1 ), and λ̂ :
α1
L1 = 0 ⇐⇒ = λ̂p1 ;
x̂1 (p1 )
(1 − α1 )
L2 = 0 ⇐⇒ = λ̂(1 − p1 );
x̂2 (p1 )
Lλ = 0 ⇐⇒ p1 x̂1 (p1 ) + (1 − p1 )x̂2 (p1 ) = m (p1 ).
I Rearrange to get:
α1
p1 x̂1 (p1 ) =;
λ̂
(1 − α1 )
(1 − p1 )x̂2 (p1 ) = ;
λ̂
12 / 62
 2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1.1 Endowment Economy

I Hence final equation says:

α1 (1 − α1 )
+ = m (p1 ); or,
λ̂ λ̂
1 = λ̂m (p1 ); or,
1
λ̂ =
m ( p1 )
I So we get in the end,

p1 x̂1 (p1 ) = α1 m (p1 );

(1 − p1 )x̂2 (p1 ) = (1 − α1 )m (p1 );

13 / 62
 2.1. The Edgeworth Box 2.1.1 Endowment Economy

2.1.1 Endowment Economy

I Back to the standard expenditure shares solution.

I As for the demand functions,

α1 m (p1 )
x̂1 (p1 ) = ;
p1
(1 − α1 )m (p1 )
x̂2 (p1 ) =
(1 − p1 )

14 / 62
2.1. The Edgeworth Box 2.1.2 Edgeworth Box

2.1.2 Edgeworth Box

15 / 62
 2.1. The Edgeworth Box 2.1.2 Edgeworth Box

2.1.2: Edgeworth Box

I This individual demand lays groundwork for study of equilibrium.

I Now two individuals, A, B and two goods; 1, 2

I Superscript agent and subscript good
I Utility functions u A (x1 , x2 ) and u B (x1 , x2 ) nice: CD in all
examples
I Endowment of A , B respectively ẽA = (e1A , e2A ) ,
ẽB = (e1B , e2B ) .
I They choose how much to consume x̃A = (x1A , x2A ) ≥ 0 ,
x̃B = (x1B , x2B ) ≥ 0 .

16 / 62
 2.1. The Edgeworth Box 2.1.2 Edgeworth Box

2.1.2: Edgeworth Box

I Pair x̃A , x̃B ≥ 0 called an allocation.

I An allocation is feasible if,

x1A + x1B = e1A + e1B ≡ E1 ;

x2A + x2B = e2A + e2B ≡ E2 .

I Note that ẽA , ẽB is a feasible allocation.

17 / 62
 2.1. The Edgeworth Box 2.1.2 Edgeworth Box

2.1.2: Edgeworth Box

I Next slide Edgeworth box with an interior endowment point

I Clever origin and reversal of viewpoint
I Reversal of viewpoint: both allocation inside the Edgeworth box

18 / 62
 2.1. The Edgeworth Box 2.1.2 Edgeworth Box

2.1.2: Edgeworth Box

Figure: Edgeworth box: endowment 19 / 62

 2.1. The Edgeworth Box 2.1.2 Edgeworth Box

2.1.2: Edgeworth Box

I indifference curves reverses for the two agents

I cigar of improvement from endowment point
I This is a case of possible gains from trade
I We are going to look to understand prices and their role in
relation to gains from trade

20 / 62
 2.1. The Edgeworth Box 2.1.2 Edgeworth Box

2.1.2: Edgeworth Box

Figure: Edgeworth box: cigar of improvement

21 / 62
 2.2 Equilibrium and Pareto Optimality

2.2 Equilibrium and Pareto Optimality

22 / 62
2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.2.1 Exchange Equilibrium

23 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.2.1: Exchange Equilibrium

I Key is to study market demand: sums demand from both A and
B .
I At some prices may be excess demand for good 1:
x̂1A (p1 ) + x̂1B (p1 ) > E1 ,
or excess supply,
x̂1A (p1 ) + x̂1B (p1 ) < E1
I Interest is equilibrium price p1∗ : no excess demand or supply in
either market,
x̂1A (p1∗ ) + x̂1B (p1∗ ) = E1 ;
x̂2A (p1∗ ) + x̂2B (p1∗ ) = E2 .
I Refer to x̂ A (p1∗ ), x̂ B (p1∗ ) as equilibrium allocation


24 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.2.1: Exchange Equilibrium

Figure: excess demand: no equilibrium

25 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.2.1: Exchange Equilibrium

I Can also write down excess supply and excess demand for each
consumer
I For consumer A

z1A = x̂1A − e1A ; z2A = x̂2A − e2A

I For consumer B

z1B = x̂1B − e1B ; z2B = x̂2B − e2B

I In equilibrium,

z1A + z1B = 0
z2A + z2B = 0

26 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.2.1: Exchange Equilibrium

I last figure:

x̂1A (p1 ) + x̂1B (p1 ) < E1

x̂2A (p1 ) + x̂2B (p1 ) > E2

I excess supply for good 1

I excess demand for good 2
I p1 needs to go down for equilibrium: good 1 becomes cheaper ,
good 2 more expensive

27 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.2.1: Exchange Equilibrium

Figure: no excess demand or supply: equilibrium

28 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.2.1: Exchange Equilibrium

I Back to C-D example with identical utility function

1 1
u A,B (x1 , x2 ) = ln x1 + ln x2 .
2 2
I Endowment: (ẽA , ẽB ) = ((2, 0), (0, 2))
I Since both want both the goods, possibility of gains from trade

29 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.2.1 Exchange Equilibrium

I Demand for consumer A would be
mA (p1 ) 2p1
x̂1A = = = 1;
2p1 2p1
mA (p1 ) 2p1 p1
x̂2A = = =
2 ( 1 − p1 ) 2(1 − p1 ) 1 − p1
I Demand for consumer B would be
mB (p1 ) 2 ( 1 − p1 ) 1 − p1
x̂1B = = = ;
2p1 2p1 p1
m B ( p1 ) 2(1 − p1 )
x̂2B = = =1
2(1 − p1 ) 2(1 − p1 )
I Note that a feature of equilibrium is tangency between
indifference curves
30 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.2.1: Exchange Equilibrium

I In equilibrium no excess demand or supply in any market

I Also known as market clearing condition
I Walras’ Law: equilibrium in one market implies equilibrium in
the other market

31 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.1.2: Exchange Equilibrium

I In market for good 1 no excess supply or demand

x̂1A (p1 ) + x̂1B (p1 ) = E1 = e1A + e1B = 2

1 − p1
1+ =2
p1
1 − p1 = p1
p1 = 0.5

I Allocation at p = 0.5 ((1, 1), (1, 1))

I Equilibrium p = 0.5, (x̃A , x̃B ) = ((1, 1), (1, 1))

32 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.1 Exchange Equilibrium

2.2.1: Exchange Equilibrium

I Lets confirm Walras’ Law (only once ever!): this price equates
demand and supply of good 2
I Equilibrium condition in market 2 is,

x̂2A (p1∗ ) + x̂2B (p1∗ ) = E2 = e2A + e2B = 2.

I Note that,
p1 1 − 0.5
x̂2A (p1 ) = =⇒ x̂2A (0.5) = = 1;
1 − p1 0.5
x̂2B (p1 ) = 1.
I Hence indeed,
x̂2A (0.5) + x̂2B (0.5) = 2;
so that market 2 is in equilibrium also.
33 / 62
2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2 Examples

34 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2: Examples

I Next example: Diamond-Water logic

I Solve with asymmetric corner endowments e A = (2n, 0) ,
e B = (0, 2) for general n > 0 and equal expenditure shares
Cobb-Douglas utility function,
1 1
u A,B (x1 , x2 ) = ln x1 + ln x2 .
2 2
I Note n > 1 better to illustrate.
I What do you expect for equilibrium?

35 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2: Examples

I Algebra:

p1 x̂1A (p1 ) = np1 = (1 − p1 )x̂2A (p1 );

p1 x̂1B (p1 ) = (1 − p1 ) = (1 − p1 )x̂2B (p1 ).

I Hence,
(1 − p1 )
x̂1A (p1 ) + x̂1B (p1 ) = n + .
p1

36 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2: Examples

I Equilibrium condition in market 1 is,

(1 − p1∗ )
x̂1A (p1∗ ) + x̂1B (p1∗ ) = 2n =⇒ n + = 2n
p1∗
(1 − p1∗ )
=⇒ =n
p1∗
1
=⇒ p1∗ =
n+1
n
⇐⇒ p2 = 1 − p1∗ =
∗
.
n+1

37 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2: Examples

I Generalizes case in which n = 1 and p1∗ = 12 .

I In general each gets income n2n
+1
I Price of common good low as there is more of it in the economy.

I Each consumes precisely half of all goods in the economy,

x̂1A (p1∗ ) = x̂1B (p1∗ ) = n;

x̂2A (p1∗ ) = x̂2B (p1∗ ) = 1.

I Next slide: Illustrate equilibrium in Edgeworth Box

38 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2: Examples

Figure: example: diamond-water paradox

39 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2: Examples

I Another case in which taste differences give rise to gains from

trade with equal endowments
I Impact of Differences in Taste
I Solve in easiest case of equal endowments e A = e B = (1, 1)
but distinct CD utility functions,
2 1
u A ( x1 , x2 ) = ln x1 + ln x2 ;
3 3
1 2
u B ( x1 , x2 ) = ln x1 + ln x2 .
3 3

40 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2: Examples

Figure: difference in taste: cigar of improvement

41 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2: Examples

I Algebra of equilibrium: with the simple endowment and

normalized prices wealth of both is always 1.
I Hence by Cobb-Douglas expenditure shares property,
2
p1 x̂1A (p1 ) = (1 − p1 )x̂2B (p1 ) = ;
3
1
(1 − p1 )x̂2A (p1 ) = p1 x̂1B (p1 ) = .
3
I Hence,
2 1
x̂1A (p1 ) + x̂1B (p1 ) = + .
3p1 3p1

42 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2: Examples

I The equilibrium condition in market 1 is,

2 1
∗ + ∗ = 2 ⇐⇒ 3 = 6p1∗ ⇐⇒ p1∗ = 0.5.
3p1 3p1
I Hence the equilibrium allocation is
4
x̂1A (p1∗ ) = x̂2B (p1∗ ) = ;
3
2
x̂2A (p1∗ ) = x̂1B (p1∗ ) = .
3
I next slide: solution in Edgeworth box

43 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.2 Examples

2.2.2: Examples

Figure: difference in taste: equilibrium

44 / 62
2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3 Pareto Optimality

45 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

I Define Pareto efficient allocations: no way to make both strictly

better off
I In our two individual economy, allocation (x̃A , x̃B ) ≥ 0 is
Pareto Optimal (Efficient) if there is no feasible allocation (ỹA ,
ỹB ) such that,

u A (ỹA ) > u A (x̃A ) and u B (ỹB ) > u B (x̃B ) .

I Note that, any P.O. will always be feasible under standard

preferences

46 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

I For non P.O. (X ) one can envisage cigar of gains from trade (go
to Y )

Figure: X not Pareto optimal, Y makes both better

47 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

I Corresponds to tangency of ICs

I for concave, monotonic utility function, P.O. implies

MRS A = MRS B

I If MRS A > MRS B : A wants to give up more good 2 than B, for

a given amount of good 1
I A exchange some good 1 to for some good 2
I make A slightly better off
I B’s utility increase, happy with less good 2 (in exchange for
good 1) than A wants to give up
I Contract curve comprises all tangencies

48 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

Figure: contract curve

49 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

I Note that says nothing about distribution

I Example: same equal expenditure shares Cobb-Douglas utility
function,
1 1
u A,B (x1 , x2 ) = ln x1 + ln x2 .
2 2
I Unequal endowment (ẽA , ẽB ) = ((1.5, 1), (0.5, 1))

50 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

I Demand functions same as before

mA (p1 ) 1.5p1 + 1 − p1 1 + 0.5p1
x̂1A = = =
2p1 2p1 2p1
mA (p1 ) 1 + 0.5p1
x̂2A = =
2(1 − p1 ) 2(1 − p1 )
B
m (p1 ) 0.5p1 + 1 − p1 1 − 0.5p1
x̂1B = = =
2p1 2p1 2p1
B
m (p1 ) 1 − 0.5p1
x̂2B = =
2(1 − p1 ) 2(1 − p1 )

51 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

I For equilibrium in market for good 1
1 + 0.5p1 1 − 0.5p1
+ = E1 = e1A + e1B = 2
2p1 2p1
2
= 2 =⇒ p1 = 0.5
2p1
I Equilibrium allocation ((1.25, 1.25), (.75, .75))
I At these values
x2A x2B
MRS A = A
= 1 = B
= MRS B
x1 x1

I ((1.25, 1.25), (.75, .75)) Pareto optimal, not equal

52 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

I In general, for the same equal expenditure share C-D utility

function the Pareto optimality condition, for any endowment, is
given by

x2A x2B
MRS A = A
= B
= MRS B
x1 x1
I Since any P.O. is always feasible, we can rewrite the P.O.
condition as
x2A x2B E2 − x2A
= =
x1A x1B E1 − x1A

53 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

I Historic interest (Adam Smith) in relationship between Pareto

optimality and equilibrium prices
I First welfare theorem says that equilibrium is always optimal
I Geometrically clear that all equilibria lie on contract curve

54 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

I Second welfare theorem says that any PO allocation is an

equilibrium for some endowments.
I Next slide: Illustrate in Edgeworth box picture both for
symmetric equal expenditure shares Cobb-Douglas utility
function

55 / 62
 2.2 Equilibrium and Pareto Optimality 2.2.3 Pareto Optimality

2.2.3: Pareto Optimality

Figure: second welfare theorem 56 / 62

2.3*: Additional Proofs

2.3*: Additional Proofs

57 / 62
 2.3*: Additional Proofs

Proof that Equilibrium is Pareto Optimal

I Suppose that x̃A , x̃B represents equilibrium allocation for

some endowment specification ẽA , ẽB and that preferences

are strictly monotone

I Let equilibrium price vector be p̃∗ ,

x̂A ( p̃∗ ) = x̃A ;

x̂B ( p̃∗ ) = x̃B
I No restriction on number of goods.
I Note by strict monotonicity that:

p̃∗ .x̂A ( p̃∗ ) = p̃∗ . ẽA ;

p̃∗ .x̂B ( p̃∗ ) = p̃∗ . ẽB .

58 / 62
 2.3*: Additional Proofs

Proof that Equilibrium is Pareto Optimal

I Suppose that there exists (ỹA , ỹB ) with.

ỹA + ỹB = ẽA + ẽB ,

hence an allocation, such that,

u A (ỹA ) > u A (x̃A ) and u B (ỹB ) > u B (x̃B ) .

I Argue leads to contradiction.

59 / 62
 2.3*: Additional Proofs

Proof that Equilibrium is Pareto Optimal

I By optimality of demand, u A (ỹA ) > u A (x̃A ) and
u B (ỹB ) > u B (x̃B ) requires,

p̃∗ .ỹA > p̃∗ .x̂A ( p̃∗ ) = p̃∗ . ẽA ;

p̃∗ .ỹB > p̃∗ .x̂B ( p̃∗ ) = p̃∗ . ẽB .

I Addition yields,
h i h i
p̃∗ . ỹA + ỹB > p̃∗ . ẽA + ẽB ,

which contradicts,

ỹA + ỹB = ẽA + ẽB .

60 / 62
 2.3*: Additional Proofs

Proof of Walras Law

I With strict monotonicity each individual is spending total value
of endowment:

p1∗ x̂1A (p1∗ ) + (1 − p1∗ )x̂2A (p1∗ ) = p1∗ e1A + (1 − p1∗ )e2A ;
p1∗ x̂1B (p1∗ ) + (1 − p1∗ )x̂2B (p1∗ ) = p1∗ e1B + (1 − p1∗ )e2B ;

I Adding together the two budget equations,

h i h i
p1∗ x̂1A (p1∗ ) + x̂1B (p1∗ )) + (1 − p1∗ ) x̂2A (p1∗ ) + x̂2B (p1∗ ))
h i h i
∗ A B ∗ A B
= p1 e1 + e1 + (1 − p1 ) e2 + e2 ;
= p1∗ E1 + (1 − p1∗ )E2 .

61 / 62
 2.3*: Additional Proofs

Proof of Walras Law

I If market one is in equilibrium, we know that in value terms:

h i h i
p1∗ x̂1A (p1∗ ) + x̂1B (p1∗ )) = p1∗ e1A + e1B = p1∗ E1 .

I Subtraction from the last equation yields

h i
(1 − p1 ) x̂2 (p1 ) + x̂2 (p1 )) = (1 − p1∗ )E2 .
∗ A ∗ B ∗

I Division by (1 − p1∗ ) > 0 yields equilibrium in second market

x̂2A (p1∗ ) + x̂2B (p1∗ ) = E2 .

62 / 62