Welfare Economics: Lecture 12

Ram Singh

Course 001

October 20, 2014

Ram Singh: (DSE) Welfare Economics October 20, 2014 1 / 16

Fair Vs Efficient

Question
1 What is a fair allocation?
2 Is a ‘fair’ allocation also an efficient allocation?
3 Is a ‘fair’ allocation Pareto efficient?
4 Does a competitive market lead to ‘fair’ allocations?

Ram Singh: (DSE) Welfare Economics October 20, 2014 2 / 16

Fairness: Two Definitions
Consider a N × M pure exchange economy. Let,
(e1 , ..., eN ) be the vector of initial endowments.

Definition
Allocation x = (x1 , ..., xn ) is ‘Fair’ if it is equal division of endowments. That is,
if " PN i #
e
(∀i, j ∈ N) xi = xj = i=1 .
N

Definition
Allocation x = (x1 , ..., xn ) is ‘Fair’ if it is Non-envious/envy-free. That is, if

(∀i, j ∈ N)[xi Ri xj ], i.e.,

(∀i, j ∈ N)[u i (xi ) ≥ u i (xj )].

Are these definitions equivalent to each other?

Ram Singh: (DSE) Welfare Economics October 20, 2014 3 / 16

Pareto Optimal Allocations I

Theorem
An allocation x̂ = (x̂1 , ..., x̂I ) is PO iff: For some utility levels Ū 2 , ..., Ū I ,
x̂ = (x̂1 , ..., x̂I ) is a solution of the following

max {u 1 (x1 )}
x1 ,...,xI

s.t. xi ≥ 0, and

u i (xi ) ≥ Ū i , for all i > 1

I
X I
X
xi = ei
i=1 i=1

See MWG, Section 16.F

Ram Singh: (DSE) Welfare Economics October 20, 2014 4 / 16

Fair Vs Pareto Efficient

Question
1 Is an ‘Equal division’ allocation ‘Non-envious’ ?
2 Is a Non-envious allocation also an Equal division allocation?
3 Is an ‘Equal division’ allocation Pareto Efficient?
4 Is a Non-envious allocation Pareto Efficient?

Ram Singh: (DSE) Welfare Economics October 20, 2014 5 / 16

Fair but Not Efficient

Consider a 3 × 3 exchange economy. Let

u 1 = 3x1 + 2y1 + z1
u 2 = 2x2 + y2 + 3z2
u 3 = x1 + 3y1 + 2z1
e1 = e2 = e3 = (1, 1, 1)

Consider an allocation y = (y1 , y2 , y3 ), where

y1 = (3, 23 , 0), y2 = (0, 0, 2) and y3 = (0, 73 , 1)

Ram Singh: (DSE) Welfare Economics October 20, 2014 6 / 16

Can Fair be Efficient? I

Consider a 2 × 2 pure exchange economy:

The goods are; x and y .
1
u 1 (.) = x α .y 1−α and u 2 (.) = x β .y 1−β , and α = β = 2

the total initial endowment vector is (x̄, ȳ ) >> (0, 0).

Let

x1 and x2 denote the amounts of good x allocated to individuals 1 and 2,

resp.
y1 and y2 denote the amounts of good y allocated to individuals 1 and 2,
resp.

Ram Singh: (DSE) Welfare Economics October 20, 2014 7 / 16

Can Fair be Efficient? II

Clearly

y1
MRS1 =
x1
y2 ȳ − y1
MRS2 = =
x2 x̄ − x1

So the set of PO allocations is solution to

y1 y2 ȳ − y1
= =
x1 x2 x̄ − x1
But,    
y1 ȳ − y1 y1 ȳ
= ⇒ =
x1 x̄ − x1 x1 x̄

Ram Singh: (DSE) Welfare Economics October 20, 2014 8 / 16

Can Fair be Efficient? III

Question
Is fair (equal) division of endowments a P.O allocation?

Consider a 2 × 2 exchange economy:

The goods are; x and y .
u 1 (x1 , y1 ) = x1 .y1 and u 2 (x2 , y2 ) = x2 .y2
the total initial endowment vector is (x̄, ȳ ) >> (0, 0).
So the set of PO allocations is solution to

MRS1 = MRS2 , i.e.,

∂u(.) ∂u(.)
∂x1 ∂x2
∂u(.)
= ∂u(.)
∂y1 ∂y2

Ram Singh: (DSE) Welfare Economics October 20, 2014 9 / 16

Can Fair be Efficient? IV

Under ‘equal division’ allocation, i.e., when x1 = x2 and y1 = y2 , we have

∂u(.) ∂u(.)
∂x1 ∂x2
∂u(.)
= ∂u(.)
∂y1 ∂y2

Ram Singh: (DSE) Welfare Economics October 20, 2014 10 / 16

Fairness under Markets

Question
Is competitive equilibrium allocation fair?

Proposition
When preferences are strongly monotonic and initial allocation is ‘Equal’, the
competitive equilibrium is fair (non-envious)

Suppose, e1 = e2 = ... = en , and (x∗ 1 , ..., x∗ n ) is a Walrasian equilibrium

allocation. Suppose,

(∃i, j ∈ {1, ..., n})[u i (x∗ i ) < u i (x∗ j )].

Then, (x∗ 1 , ..., x∗ n ) not a WEA. So, the following holds.

(∀i, j ∈ {1, ..., n})[u i (x∗ i ) ≥ u i (x∗ j )].

Ram Singh: (DSE) Welfare Economics October 20, 2014 11 / 16

Effect of Endowments I

Question
Does an increase in endowment make the person better off?

Answer is ‘Yes’, for well behaved utilities. Consider a 2 × 2 pure exchange

economy:
The goods are; x and y .
u 1 (.) = x α .y 1−α and u 2 (.) = x β .y 1−β , α, β ∈ (0, 1) and α 6= β
the initial endowments are e1 (.) = (x̄1 , ȳ1 ) >> (0, 0) and
e2 (.) = (x̄2 , ȳ2 ) >> (0, 0).

Ram Singh: (DSE) Welfare Economics October 20, 2014 12 / 16

Effect of Endowments II
Question
Find out the competitive equilibrium prices and allocations.
How do the competitive equilibrium allocations change with initial
endowments
For any given price vector p = (p1 , p2 ) >> (0, 0), you can check that
demands are:
α(p1 x̄1 + p2 ȳ1 )
x1 (p) =
p1
(1 − α)(p1 x̄1 + p2 ȳ1 )
y1 (p) =
p2
β(p1 x̄2 + p2 ȳ2 )
x2 (p) =
p1
(1 − β)(p1 x̄2 + p2 ȳ2 )
y2 (p) =
p2

Ram Singh: (DSE) Welfare Economics October 20, 2014 13 / 16

Effect of Endowments III

Market Clearance for 1st good implies:

x1 (p) + x2 (p) = x̄1 + x̄2

α(p1 x̄1 + p2 ȳ1 ) β(p1 x̄2 + p2 ȳ2 )
+ = x̄1 + x̄2
p1 p1
αp1 x̄1 + αp2 ȳ1 + βp1 x̄2 + βp2 ȳ2 = p1 x̄1 + p1 x̄2

Re-arranging the last equality, we get

p2 (αȳ1 + β ȳ2 ) = [(1 − α)x̄1 + (1 − β)x̄2 ]p1 , i.e.,

p1∗ (αȳ1 + β ȳ2 )

∗ = (1)
p2 (1 − α)x̄1 + (1 − β)x̄2

Ram Singh: (DSE) Welfare Economics October 20, 2014 14 / 16

Effect of Endowments IV

Let p2∗ = 1. In view of (1), the equilibrium demands can be written as

 
(1 − α)x̄1 + (1 − β)x̄2
x1∗ (p∗ ) = α x̄1 + ȳ1
(αȳ1 + β ȳ2 )
 
(αȳ1 + β ȳ2 )
y1∗ (p∗ ) = (1 − α) x̄1 + ȳ1
(1 − α)x̄1 + (1 − β)x̄2
 
(1 − α)x̄1 + (1 − β)x̄2
x2∗ (p∗ ) = β x̄2 + ȳ2
(αȳ1 + β ȳ2 )
 
(αȳ1 + β ȳ2 )
y2∗ (p∗ ) = (1 − β) x̄2 + ȳ2
(1 − α)x̄1 + (1 − β)x̄2

From (1)
p∗
∂( p1∗ ) (αȳ1 + β ȳ2 )(1 − α)
2
=− < 0.
∂ x̄1 [(1 − α)x̄1 + (1 − β)x̄2 ]2

Ram Singh: (DSE) Welfare Economics October 20, 2014 15 / 16

Effect of Endowments V

Now, you can verify the following:

∂x1∗ (1 − α)
= α + ȳ1 >0
∂ x̄1 [(αȳ1 + β ȳ2 )]
∂y1∗ (αȳ1 + β ȳ2 )(1 − β)x̄2
= (1 − α) >0
∂ x̄1 [(1 − α)x̄1 + (1 − β)x̄2 ]2
∂x2∗ (1 − α)
= β ȳ2 >0
∂ x̄1 (αȳ1 + β ȳ2 )
∂y2∗ (1 − α)(αȳ1 + β ȳ2 )
= −x̄2 (1 − β) <0
∂ x̄1 [(1 − α)x̄1 + (1 − β)x̄2 ]2

Ram Singh: (DSE) Welfare Economics October 20, 2014 16 / 16