Homework #3

General Equilibrium and Welfare Economics

Econ 500a, Yale University
Due Date: Thursday 11/14/2024

Question 1 Consider a I = 3, J = 2 exchange economy where V i = ci1 ci2 for i ∈ {1, 2, 3}, and ȳ 11 , ȳ 12 = (6, 1),





ȳ 21 , ȳ 22 = (1, 6), ȳ 31 , ȳ 32 = (1, 1). Now consider the proposed allocation c11 , c12 = (3, 3), c21 , c22 = (3, 3),


c31 , c32 = (2, 2).

1. Show that this proposed allocation is Pareto optimal.

2. Notice that each individual i strictly prefers the proposed allocation to their endowment. Despite this, give an
argument (not involving prices or equilibrium) as to why it might be unlikely for our exchange economy to end
up with the proposed allocation.
Hint: Try to think if there can be a situation in which two of the agents get together and choose an allocation
that makes them better off relative to the competitive equilibrium.

Question 2 Consider a I = 4, J = 2 exchange economy with endowments ȳ 11 , ȳ 12 = ȳ 21 , ȳ 22 = (10, 10) and





ȳ 31 , ȳ 32 = ȳ 41 , ȳ 42 = (10, 30) and the same utility function V i = log ci1 + log ci2 , for i ∈ {1, 2, 3, 4}. For each
allocation given below show whether

i) it is Pareto optimal,

ii) it is in the core (if not, provide a blocking coalition); and

iii) it can be supported as a competitive equilibria with transfers for some price vector. Briefly explain your reasoning.





1. c11 , c12 = c21 , c22 = (7.5, 15) and c31 , c32 = c41 , c42 = (12.5, 25).


√ √


√ √

2. c11 , c12 = c21 , c22 = 50, 2 50 and c31 , c32 = c41 , c42 = 20 − 50, 40 − 2 50 .





3. c11 , c12 = (8, 12), c21 , c22 = (9, 11), c31 , c32 = (12, 23) and c41 , c42 = (11, 29).

Question 3 Consider an Edgeworth box economy with two individuals i ∈ {1, 2} who consume two goods j ∈ {1, 2}.
Individual i has preferences given by

ζ i
1−ζ i
V i = ci1 ci2 .

Resource constraints are

X X
ci1 = ȳ 1 and ci2 = ȳ 2 .
i i
In a competitive equilibrium, individuals maximize utility subject to a budget constraint of the form

p1 1 + τ i1 ci1 + p2 1 + τ i2 ci2 ≤ p1 ȳ i1 + p2 ȳ i2 + T i ,

1
where ȳ ij denotes individual i’s endowment of good j, τ ij are individual-specific taxes, and T i are transfers given by

T i = p1 τ i1 ci1 + p2 τ i2 ci2 .

Individual excess demands are defined as

z ij = cij − ȳ ij ,

where the aggregate excess demand of good j is defined as z j = i z ij .

P
1 3
Assume that ζ 1 = 2 and ζ 2 = 5 for the individual Cobb-Douglas preferences. Assume that τ 12 = τ 21 = τ 22 = 0
and that τ 11 = τ . We consider the following three scenarios . This exercise studies the welfare implications of moving

Table 1: Equilibrium Scenarios

Scenario τ ȳ 11 ȳ 12 ȳ 21 ȳ 22

A 0 0.5 3 2.5 0.5

B 1 0.5 3 2.5 0.5
C 0 0 3 2.5 0.5

from A to B, and from C to A.

Part I. Edgeworth Box and Pareto Efficiency Questions 1-3 should be answered analytically. Question 5
requires that you use the computer. Questions 4 and 6 should be answered without any computations.

1. Characterize the Pareto set.

2. Characterize the individual indifference curves.

3. Characterize the competitive equilibrium. Recall that τ 12 = τ 21 = τ 22 = 0 and that τ 11 = τ .

4. Consider the three scenarios: A, B, and C. Discuss which scenario is Pareto efficient, which is not, and explain
why.

5. Now let’s explore the Edgeworth Box of these three scenarios in Matlab (or any other software that you are
comfortable with). Draw the box diagram while plotting the indifference curves, the Pareto set(s), the budget
constraints, and the competitive equilibria allocations for each of the three scenarios.

6. Use the box diagram to confirm your discussion in question 4. Discuss what happens to the Pareto set as the
total endowment of good 1 increases from C to A.

Part II. Welfare Decomposition Consider a utilitarian planner with the following Social Welfare Function
X
W = αi V i ,
i

where αi denotes the Pareto weight for individual i. We consider the a welfare numeraire that implies individual
normalizing factors of the form
∂ui 1 ∂ui
λi =
i1
ȳ + i2 ȳ 2 .
∂c ∂c
This part considers the direct perturbation approach to characterize the welfare decomposition. We index all
perturbations by θ.
dV i ∂ui
dV i|λ
1. Derive the normalized individual welfare gain dθ = dθ
λi . Use the definition of M RScij = ∂cij
λi in your answer.

2
 dW
dW λ
2. Derive the efficiency/redistribution welfare decomposition. Start from the definition dθ = 1
Pdθ
αi λi
.
I i

3. Show that the efficiency component can be decomposed into i) Cross-Sectional Consumption Exchange Efficiency
and ii) Good Endowment Change.

Part III. Taxes and Welfare We consider moving from scenario A to B by increasing τ 11 = τ from 0 to 1. Use
the Pareto weights αi = [1, 1] in your answer.
You’ll need to construct a grid for τ that starts at 0 and ends at 1 to answer questions 3-7. In your numerical
dχij
c 1
solution, you may use numerical gradients on Matlab to compute dτ .

1. Without any calculations, which terms of the welfare decomposition do we expect to be non-zero and which to
be zero? Why?

2. Use the competitive equilibrium derived in Part I to compute the normalized individual weights, ω i , in scenarios
A and B and report their numerical values for both individuals. Which individual has a higher weight? Explain.
p1
3. Plot p2 as a function of τ .

4. Plot χ11 21
c and χc as a function of τ .

5. Plot χ12 22
c and χc as a function of τ .

dV 1|λ dV 2|λ
6. Plot dτ and dτ as a function of τ .
dW λ
7. Plot dτ , ΞE , and ΞRD , as a function of τ .

(a) Why is efficiency, ΞE , zero at τ = 0?

dW λ
(b) Explain why efficiency, ΞE , and normalized welfare gains, dτ , are negative whenever τ > 0.

Part IV. Endowments and Welfare We consider moving from scenario C to A by increasing ȳ 11 from 0 to 0.5.
Use the Pareto weights αi = [1, 1] in your answer.
You’ll need to construct a grid for ȳ 11 that starts at 0 and ends at 0.5 to answer questions 3-7. In your numerical
dχij
solution, you may use numerical gradients on Matlab to compute dȳ 11 .
c

1. Show that the Cross-Sectional Consumption Exchange Efficiency component is zero. (Hint: Use the individual
or the planner F.O.Cs)

2. Report the numerical values of the normalized individual weights, ω i , at scenario C.

p1
3. Plot p2 as a function of ȳ 11 .

4. Plot χ11 21 11
c and χc as a function of ȳ .

5. Plot χ12 22 11
c and χc as a function of ȳ .

dV 1|λ dV 2|λ
6. Plot dȳ 11 and dȳ 11 as a function of ȳ 11 .
dW λ
7. Plot dȳ 11 , ΞE , and ΞRD , as a function of ȳ 11 .
dW λ
(a) Explain why efficiency, ΞE , and normalized welfare gains, dȳ 11 , are always positive.

1 Let’s dx
say we want to compute dθ
by using a numerical gradient. This can be achieved in Matlab by using gradient(x,theta). The finer
the grid, the more accurate is your result.