Final Examination with Answers: Economics 210A

December, 2016, Ted Bergstrom, UCSB

I asked students to try to answer any 7 of the 8 questions. I intended the
exam to have some relatively easy parts and some quite challenging parts.
As I prepared the answers, I realized that the exam was longer and more
difficult than I expected it to be. Nevertheless, some of the students got
almost everything and most students got a lot correct. As a study aid, I
have written fairly elaborate answers. It is quite likely that there remain
some typos and some mistakes and ambiguities in my answers. I welcome
questions and corrections.

Question 1.
A) State the weak axiom of revealed preference, the strong axiom of revealed
preference and the generalized axiom of revealed preference.

Let xi be a commodity bundle that is chosen when the price vector is pi .

Bundle xi is said to be directly revealed preferred to bundle x if pi xi ≥ pi x.
The WARP states that if one bundle is directly revealed preferred to another,
then the second will never be directly revealed preferred to the first.
Another way of stating WARP is: If p1 x1 ≥ p1 x0 , then p0 x1 > p0 x0 .
A bundle xi is revealed preferred to x if there is a sequence of bundles
xn , . . . , x1 such that xt is revealed preferred to xt−1 for t = n, . . . , 2.
SARP requires that if x0 is revealed preferred to x, then x is not directly
revealed preferred to x0 .
A bundle xi that is selected at price vector pi is said to be revealed strictly
preferred to a bundle x if pi xi > pi x.
GARP requires that if x0 is revealed preferred to x, then x is not strictly
revealed preferred to x0 .

B) Give an example of preferences that satisfy the generalized axiom of re-

vealed preferences but do not satisfy the strong axiom of revealed preference.
Show that your example satisfies GARP but not WARP.

Consider preferences on <2+ that are represented by the utility function

x1 + x2 . Suppose that p1 = p0 = (1, 1), and that x1 = (1, 1) is chosen at p1
and x0 = (0, 2) is chosen at p1 . These choices are consistent with maximizing
x1 + x2 subject to the budget constraint. We see that p1 x1 ≥ p1 x0 and
p0 x0 ≥ p0 x1 . Thus we see that this data does not satisfy WARP.

1
 If choices maximize x1 +x2 subject to a budget constraint, it must be that
if (x01 , x02 ) is directly revealed preferred to (x1 , x2 ) then x01 + x02 ≥ x1 + x2 .
By the transitivity of the ordering of the real numbers it follows that if
(x01 , x02 ) is revealed preferred to (x1 , x2 ) either directly or indirectly, then
x01 + x02 ≥ x1 + x2 . Therefore it cannot be that (x01 , x02 ) is revealed preferred
to (x1 , x2 ) and x1 + x2 > x01 + x02 . So GARP is satisfied.

Question 2. Wilbur is an expected utility maximizer with a von Neumann-

Morgenstern utility function u(x) where x is a vector goods. Wilbur is con-
sidering moving to one of two cities. He is unsure about his future income
and about future prices. The cities are equally attractive to Wilbur in all
respects other than the probability distribution of prices and income.

A) Wilbur knows the probability distribution of possible price-income com-

binations in each city, and has von Neumann-Morgenstern utility function
u(x) for consumption bundles. How would you go about finding which city
Wilbur would prefer?

Find Wilbur’s indirect utility function corresponding to u(x). Wilbur

would prefer the city with the higher expected indirect utility.
B) Suppose that Wilbur consumes just two goods and that his von Neumann
Morgenstern utility function is
1
u(x1 , x2 ) = min{x1 , x2 }1/2 .
2
Wilbur believes that if he moves to City A, there are two possible out-
comes, each of which has a probability of 1/2. One of these outcomes is that
his income will be 125 and prices of goods 1 and 2 will be p1 = 1 and p2 = 2.
The other possible outcome is that his income will be 196 and prices of goods
1 and 2 will be p1 = 2 and p2 = 1.

• Calculate Wilbur’s expected utility if he moves to City A. Wilbur be-

lieves that in City B, the prices of goods 1 and 2 are certain to be p1 = 1
and p2 = 1.

• If in City B he would have an income of M with certainty, for what

values of M would he prefer City B to City A? Explain your answer.

2
 Wilbur’s indirect utility function is
!1/2
m
v(p, m) =
p1 + 2p2
In city A, his expected utility would be
1 1 1 1
v(1, 2, 125) + v(2, 2, 196) = 5 + 7 = 6
2 2 2 2
In city B, his expected utility would be
1/2
m

v(1, 1, m) = .
3
So he will prefer city B to city A if
1/2
m

> 6.
3
This is the case if m > 108.
C) Wilbur’s friend Charlotte has a von Neumann Morgenstern utility func-
tion
1
u(x1 , x2 ) = min{x1 , x2 }.
2
If Charlotte and Wilbur face the same prices and have the same incomes,
how will their consumptions differ? If Charlotte and Wilbur both believe that
the probability distributions over their prices and incomes in Cities A and B
are as in Part B, will they ever have different preferences about which city
to live in? Explain.
If Wilbur and Charlotte face the same prices and have the same incomes,
they will consume the same bundles. This is seen because their vNM utility
functions, which are monotone transformations of each other, represent the
same preferences over bundles consumed with certainty.
They have different attitudes toward risk, however. Wilbur is more risk
averse than Charlotte. Charlotte’s expected utility from City A would be
1
2
25 + 12 49 = 37. Her expected utility from City B would be m/3. So she
would prefer City A to City B if m < 111 and woud prefer B to A if m > 111.

Question 3. Consider a consumer who consumes three goods and has utility
function
U (x1 , x2 , x3 ) = (x1 + b1 )α (x2 + b2 )β (x3 + b3 )γ

3
where bi ≥ 0 for i = 1, . . . 3 and where α > 0, β > 0, and γ > 0.
A) Why can you assume that α + β + γ = 1 without loss of generality?
1
The function V (x1 , x2 , x3 ) = U (x1 , x2 , x3 ) α+β+γ is a monotone increasing
transformation of U and thus represents the same preferences. Let α0 =
β γ
α
α+β+γ
, β 0 = α+β+γ , and γ 0 = α+β+γ . We see that
0 0 0
V (x1 , x2 , x3 ) = (x1 + b1 )α (x2 + b2 )β (x3 + b3 )γ

where α0 + β 0 + γ 0 = 1.
B)Does this consumer have homothetic preferences? Explain.

This consumer does not have homothetic preferences. Preferences are

homothetic if and only if U (x) = U (x0 ) implies that U (tx) = U (tx0 ) for all
scalars t > 0. It is easily verified that U does not have this property unless
b1 = b2 = b3 = 0.

C) Does this consumer have additively separable preferences? Explain.

This consumer has additively separable preferences. We see that

V (x1 , x2 , x3 ) = log U (x1 , x2 , x3 ) = α ln(x1 + b1 ) + β ln(x2 + b2 ) + γ ln(x3 + b3 )

represents the same preferences as U and V is additively separable.

D)Find this consumer’s Marshallian demand function. Don’t forget to ac-

count for corner solutions if they exist.

Since preferences are continuous and strictly convex, the demand function
is a well-defined, single valued function. We found in Part A that there is no
loss of generality in assuming that α + β + γ = 1.
A shortcut for finding a solution if one exists is to define zi = xi + bi . We
can rewrite the problem as Maximize

U (z1 , z2 , z3 ) = (z1 )α (z2 )β (z3 )γ

subject to the constraints that p1 z1 + p2 z2 + p3 z3 = p1 x1 + p2 x2 + p3 x3 +

p1 b1 + p2 b2 + p3 z3 = m + p1 b1 + p2 b2 + p3 z3 and zi ≥ bi for i = 1, . . . 3. This
states the problem as a standard Cobb-Douglas demand problem in terms of

4
the zi ’s, with an inequality constraint. Thus it must be that at an interior
maximum,
α
z1 = (m + p1 b1 + p2 b2 + p3 b3 )
p1
β
z2 = (m + p1 b1 + p2 b2 + p3 b3 )
p2
γ
z3 = (m + p1 b1 + p2 b2 + p3 b3 )
p3
Then
α
x1 = z1 − b1 = (m + p1 b1 + p2 b2 + p3 b3 ) − b1 (1)
p1
β
x2 = z2 − b2 = (m + p1 b1 + p2 b2 + p3 b3 ) − b2 (2)
p2
γ
x3 = z3 − b3 = (m + p1 b1 + p2 b2 + p3 b3 ) − b3 (3)
p3
These three equations must be satisfied if there is an interior solution
where x1 > 0, x2 > 0 and x3 > 0.
I gave full credit to those who found this necessary condition for there
to be an interior solution, whether or not they properly characterized the
corner solutions.
A full characterization of corner solutions is a little complicated and I did
not expect everyone to carry this out in detail during the exam. It is not
true, as some exam-takers claimed, that if Equation 3 tells us that x3 < 0,
then there will be a solution with x3 = 0 and x1 and x2 given by Equations
1 and 2.
But, for your enjoyment, I will sketch a full description of corner solutions
and how you would find them. If there is no interior solution, there must be a
corner solution. In general, with three goods, there are 6 possible corner so-
lutions. In three of these, two goods are consumed in positive quantities and
one is not consumed. Since the demand function is single-valued, we know
that exactly one of these corner solutions will obtain. Equations 1-3 don’t
give us an immediate answer to which of them it is. A corner solution must
satisfy the Kuhn-Tucker conditions. These involve a system of 3 inequalities,
along with complementary slackness conditions for each of them.
In general, solving a system of 3 Kuhn-Tucker inequalities can be a pain in
the neck, because there are 6 possible places to look for a corner solution. It

5
could be that the solution has any two of the variables at interior values and
one at a boundary value, or it could be that it has only one of the variables
at an interior value and the others at boundary values. For this problem, our
search is easier, because there is a fairly simple condition that tells us where
a solution could possibly be.
Explaining the answer is a little easier i we use the notation α1 = α, α2 =
β and α3 = γ. The consumer’s maximization problem can be represented as
Maximize
U (z1 , z2 , z3 )) = α1 ln z + 1 + α2 ln z2 + α3 ln z3
subject to the constraints
p1 z1 + p2 z2 + p3 z3 = m + p1 b1 + p2 b2 + p3 z3
and zi ≥ bi for i = 1, 2, 3.
Where the Lagrangean for this problem is

3
X
L(z1 , z2 , z3 , λ) = αi ln zi +λ ((m + p1 b1 + p2 b2 + p3 z3 ) − (p1 z1 + p2 z2 + p3 z3 )) ,
i=1

the Kuhn-Tucker conditions require that for each i, either zi > bi and the
partial derivative of the Lagrangean with respect to zi is zero or zi = bi and
the partial derivative of the Lagrangean with respect to zi is negative.
The interpretation of the Kuhn-Tucker conditions is as follows. If zi > bi ,
then the bang-per-buck from i is zi /pi = λ. If zi = bi , then the bang-per-buck
from i is less than λ.
In seeking a solution, one might look first for an interior solution where
there is equal bang per buck from all three goods and zi ≥ bi for all i. If you
don’t find such a solution, you can simplify your search for a corner solution
by calculating bi /pi for each of the i’s. This is the bang per buck at the
boundary. Suppose that we have bi /pi > bj /pj > bk /pk . Then there are only
two possibilities for a corner solution. One of them has zk = bk , zi > bi , and
zj > bj . The other has zj = bj , zk = bk and zi > bi . (I will leave it as an
exercise for you to show that this is the case.)
Since strict convexity implies that there is exactly one solution, it is easy
find out which one is the equilibrium by checking which one satisfies the
Kuhn-Tucker conditions.
Having found solutions for the zi ’s, one can then find the xi ’s since xi =
zi − bi .

6
Question 4. For the consumer described in the previous question:
A) Find this consumer’s indirect utility function and verify that Roy’s iden-
tity holds in this case.

(I didn’t expect you to work this out for the corner solutions. You got
full credit if you did it correctly for the interior solutions.)
Let the direct utility function be xα1 xβ2 xγ3 where α + β + γ = 1. If there is
an interior solution, then the indirect utility function is
−β −γ
V (p1 , p2 , p3 , M ) = (M + p1 b1 + p2 b2 + p3 b3 ) αα β β γ γ p−α
1 p2 p3

According to Roy’s law, where xi (p, m) is the Marshallian demand func-

tion, we must have
! !
∂V (p1 , p2 , p3 , M ) ∂V (p1 , p2 , p3 , M )
xi (p, m) = − ÷ . (4)
∂pi ∂M
Let’s verify this for good 1. Very similar calculations do the trick for
goods 2 and 3.

∂V −β −γ α −β −γ
= b1 αα β β γ γ p−α
1 p2 p3 − b1 αα β β γ γ p−α
1 p2 p3 (M + p1 b1 + p2 b2 + p3 b3 )
∂p1 p1
(5)
∂V −β −γ
= αα β β γ γ p−α
1 p2 p3 (6)
∂M
Substituting Equations 5 and 6 into the right side of Equation 4 and
cancelling, we have
α
xi (p, m) = (m + p1 b1 + p2 b2 + p3 b2 ) − b1 .
p1
Looking back at the answer to Question 3, we see that this is the same
answer we found by calculating Marshallian demand directly.
For the masochists among you, we could go on to calculate indirect util-
ities at the corner solutions.
At a corner solution, where x1 = 0 x2 > 0 and x3 > 0, the indirect utility
must be !β !γ
β γ
α
(M + p2 b2 + p3 b3 ) b1 p−β −γ
2 p3 .
β+γ β+γ

7
 At a corner solution where x1 = 0, x2 = 0 and x3 > 0, indirect utility
must be
(M + p3 b3 ) bα1 bβ2 p−γ
3

.
B) Find this consumer’s expenditure function.
A fairly quick way to find this is to use the fact that where V is the
indirect utility function,
V (p, e(p, u)) = u.
Using the indirect utility function that we found previously, we then have
−β −γ
(e(p, u) + p1 b1 + p2 b2 + p3 b3 ) αα β β γ γ p−α
1 p2 p3 = u. (7)

From Equation 7 it follows that

e(p, u) = upα1 pβ2 pγ3 α−α β −β γ −γ − p1 b1 − p2 b2 − p3 b3 (8)

C) Find this consumer’s Hicksian demand function and the substitution ma-
trix.

The Hicksian demand functions are the partial derivatives of the expenditure
function.
α α β γ −α −β −γ
h1 (p, u) = u p p p α β γ − b1
p1 1 2 3
α α β γ −α −β −γ
h2 (p, u) = u p p p α β γ − b2
p2 1 2 3
α α β γ −α −β −γ
h3 (p, u) = up p p α β γ − b3
p3 1 2 3
The substitution matrix is the Hessian of the expenditure function. the
ijth element of this matrix is the partial derivative of hi (p, u) with respect
to pj .

Question 5. Consider an economy with two types of people. While their

utility functions are all of the form described in Question 3, they have differ-
ent values of the bi ’s. Type 1 has b1 = b2 = b3 = 10 and Type 2 has b1 = 10,
and b2 = b3 = 0. Persons of Type 1 have incomes M1 and persons of type 2
have incomes M2 .

8
A)If there are 100 people of each type, write an expression for the aggregate
demand for good 1 as a function of the prices and incomes.

At interior solutions it must be that the demand functions of Type 1’s

are given by

α (M1 + 10(p1 + p2 + p3 ))
x11 (p, m) = − b1 .
p1
and Type 2’s by

α(M2 + 10p1 )
x21 (p, m) = − b1 .
p1
Then aggregate demand for good 1 is

!
  M1 + M2 + 20p1 + 10p2 + p3
100 x11 (p, m) + x21 (p, m) = 100 − b1 . (9)
p1

B)Suppose that initially prices and income are such that everybody buys pos-
itive amounts of all goods. Income is redistributed from type 2’s to type 1’s
and after the income redistribution everyone is still buying positive amounts
of all goods. What can you say about the change in aggregate demand for
good 1? Explain your answer.

There will be no change in aggregate demand for good 1. We see this from
Equation 9, since aggregate demand for Good 1 is determined by aggregate
income.
C) Suppose that the prices are p1 = p2 = p3 = 1. Find incomes M1 and
M2 such that a transfer of income from type 1’s to type 2’s must change the
aggregate demand for good 1. Explain your answer.

This will happen if enough income is transferred from Consumer 1 to

consumer 2 so that Consumer 1 is at a corner solution, consuming no Good
1. If this is the case, further transfers from 1 to 2 will increase consumer 2’s
consumption of Good 1, but will not change Consumer 1’s. Therefore such
transfers will increase aggregate demand for Good 1.

9
 Consumer 1 will demand a positive amount of good 1 only if α(M1 +30) >
10, which is equivalent to M1 > 10
α
− 30. So if M1 < 10
α
− 30, transfers from
1 to 2 will increase aggregate demand for Good 1.

Question 6. Harry consumes just one commmodity and he will live for T
periods. His current preferences over consumption streams are represented
by a utility function of the form
T
X
U (x1 , . . . , xT ) = βt u(xt )
t=1

where xt is the amount of the commodity that he will consume in year t and
where the function u(·) is increasing, strictly concave and twice continuously
differentiable. Harry knows that his income stream will be (w1 , . . . , wT ) where
wt is the income that he will receive in period t. Harry is able to borrow or
save at the constant interest r. At time 1, Harry is able to commit himself
to any time path of consumption that satisfies his budget constraint. His
budget constraint is that the present value of his lifetime consumption does
not exceed the present value of his lifetime income stream.
A) Suppose that for some α where 0 < α < 1, and for all t = 1, . . . T ,
βt = αt−1 . At what interest rate will Harry will choose to consume the same
amount of goods in every period of his life? Explain your answer. Does this
interest rate depend on the time path of his income stream? At this interest
rate, what can you say about the way in which his borrowing and saving
behavior depend on the time path of his income stream?

Harry’s intertemporal budget constraint is

T
X 1 X 1
xt t
= wt .
t=1 (1 + r) t=1 (1 + r)t
Harry would choose a consumption profile such that his marginal rate of
substitution between consumption in period t + 1 and t is equal to their
relative prices. That is
t−1
αt u0 (xt+1 )
u0 (xt ) = (1 + r).
α
If Harry chooses constant consumption, it must be hat u0 (xt+1 ) = u0 (xt ) and
1
therefore α = 1+r . His time path of consumption does not depend on the

10
time path of his income but only on the present value of income. Where P V
is the present value of his income and c is his constant consumption, it must
be that
c(1 + r + r2 . . . + rT ) = P V
Since
1 − rT +1
1 + r + r2 + . . . rT =
1−r
it follows that
PV
c = (1 − r)
1 − rT +1
He will spend less than his income in periods where
PV
wt > (1 − r)
1 − rT +1
and spend more than his income when this inequality is reversed.

B) Suppose that β2 = β1 = 1 and that for t = 3, . . . T , βt = αt−2 . Suppose

that at time 1, Harry can commit himself to a time path of future consump-
tion. Qualitatively, how does his time path of consumption depend on the
interest rate? For example, at what if any interest rates r > 0 is his consump-
tion first increasing, then constant, at what interest rates is his consumption
always increasing, at what interest rates is his consumption first increasing,
then decreasing, etc, etc.
If he can commit himself to a plan for future consumption and the interest
rate is constant at r, Harry will choose a consumption path such that
βt+1 u0 (xt+1 ) 1
= .
βt u0 (xt ) 1+r
If his preferences are convex, this means that xt+1 is greater or less than xt
depending on whether ββt+1 t
1
is greater or less than 1+r . In this example, Harry
would increase his consumption between periods 1 and 2. His consumption
in period 3 would be less than his consumption in period 2 and would be
increasing or decreasing over the remaining course of his life, depending on
whether 1 + r > α or 1 + r < α.
C) Suppose that T = 3 and Harry’s utility function is
√ √ √
U (x1 , x2 , x3 ) = x1 + x2 + α x3 .

11
Harry earns income W > 0 in period 1, while wt = 0 for t > 1. Suppose also
that
1
= α.
1+r
If at time 1, Harry can choose his consumption for each period, subject to
his budget constraint, solve for his choice of x1 , x2 , and x3 as a function of
the parameters W and r.

Harry’s budget constraint is

x2 x+3
x1 + + =W
1 + r (1 + r)2

Setting marginal rates of substitution equal to price ratios, we find that

x3 = x2 = (1 + r)2 x1 . Together with the budget constraint, this implies that

x1 (1 + (1 + r) + 1) = W

and hence
W
x1 = ,
3+r
and
W (1 + r)2
x2 = x3 = .
3+r

D Suppose that Harry can save money in period 1 but he must leave the choice
of allocation between periods 2 and 3 to his future self. Harry is aware of
this and knows that in Period 2 his utility function for consumptions periods
2 and 3 will be
√ √
U (x2 , x3 ) = x2 + x3 .
He also knows that the interest rate will continue to satisfy the equation
1
= α.
1+r
If Harry consumes x1 in period 1, what consumptions will his period 2 self
choose for periods 2 and 3? Write down an expression for Harry’s utility as
a function of x1 , taking into account the fact that he knows that his period
2 self will determine the division of income between his period 2 self and his

12
period 3 self. Find the optimal choice of x1 for Harry. Is this the same as
the amount of x1 that he would choose in Part C above?

Harry’s budget in Period 2 will be

1
x2 + x3 = W − x1 .
1+r
Setting his marginal rate of substitution equal to the price ratio, we see that

x3 = (1 + r)2 x2 .

Substituting into the budget, we find that x2 (1 + 1 + r) = W − x1 . Hence

W − x1
x2 =
2+r
and
x3 = (1 + r)x2 .
Then Harry’s lifetime utility evaluated in period 1 if he chooses x1 will
be !1/2
1/2
W − x1 (W − x1 )(1 + r)

1/2
x1 + + . (10)
2+r 2+r
Expression 10 simplifies to
1 + (1 + r)1/2
!
1/2 1/2
x1 + (W − x1 ) (11)
(2 + r)1/2
To find Harry’s optimal choice of x1 , set the derivative of Expression 10
with respect to x1 equal to zero. This implies that

1 + (1 + r)1/2
!
−1/2
x1 = (W − x1 )−1/2 (12)
(2 + r)1/2
and hence
W − x1 2+r
= (13)
x1 1 + (1 + r)1/2
Looking back at the answer to Part C, we find that when Harry can
commit future consumptions, he chooses x1 such that
W − x1
= 2 + r. (14)
x1

13
 Comparing Equation 13 with Equation 14, we see that Harry chooses to
save less and spend more in the case where he can not commit than in the
case where he can commit future behavior.

Question 7. A pure exchange economy has 1000 Type 1 consumers and

1000 Type 2 consumers. There are two commodities. Everyone has a utility
function of the form
1
U (x1 , x2 ) = (xα1 + xα2 ) α .
Type 1’s have an initial endowment of ω1 units of good 1 and zero units of
good 2. Type 2’s have an initial endowment of ω2 units of good 2 and zero
units of good 1. Suppose that there is competitive trading where good 1 is the
numeraire, the price of good 2 is p and each consumer has wealth equal to
the value of his or her initial endowment.

A) Suppose that α = −1. At the price p, how much of good 2 will Type 2’s
demand for themselves and how much will they be willing to sell? Does the
amount that they are willing to sell increase or decrease as p increases?

The budget equation for Type 2’s is

x1 + px2 = pω2 . (15)

Setting p equal to the ratio of the marginal utility of good 2 to the marginal
utility of good 1, we have
xα−1
2
p = α−1 . (16)
x1
Where α = −1, this implies that

x21
p=
x22
and hence
√
x1 = px2 . (17)
Let x22 (p) be the amount that a Type 2 will demand for her own con-
sumption when the price is p. Substituting from Equation 17 into Equation
15, we have
pω2
x22 = √ . (18)
p+ p

14
 The amount that a Type 2 will be willing to sell will be
√ ! !
2 p 1
ω2 − x2 (p) = ω2 √ = ω2 √ (19)
p+ p 1+ p

We see from Equation 19 that Type 2’s will offer to sell less of good 2
when its price rises.

B) At the price p, how much of good 2 will Type 1’s demand?

Type 1’s have budget

x + 1 + px2 = ω1
and their demand function for good 2 is
ω1
x12 (p) = √ .
p+ p

C) Find the competitive equilibrium price p.

There are (at least) two ways to solve this.

Brute force Method
Aggregate demand for Good 2 equals aggregate supply when
√ !
ω1 pω2
1000 √ + √ = 1000ω2 (20)
p+ p 1+ p
Simplifying Equation 20, we find that
2
w1

p= .
w2

Quick method
All consumers have identical homothetic preferences. So if they all face
the same prices, they must all consume the two goods in the same ratio.
Since supply must equal demand in both markets, the ratio in which each
consumer consumes the two goods must be the same as the ratio in which
the goods are available, that is ω2 /ω1 .

15
 The equilibrium price p must equal the marginal rate of substitution of a
consumer who consumes the goods in this ratio. So we must have
−2 2
ω2 ω1
 
p= = .
ω1 ω2

D) Suppose that a natural catastrophe reduces the endowment of each Type

2 by 50% and the price moves to the new equilibrium price. What happens to
the value of the endowment of type 2’s? Are the type 2’s better off or worse
off after the catastrophe than they were before? What about the type 1’s?

It is easy to see that Type 1’s are made worse off. Their income does not
change and the price of good 2 has increases.
It is less easy to see whether type 2’s are better off. Their income doubles,
but the cost of Good 2 quadruples. (In fact, when I wrote the question, I
didn’t realize that the answer was as difficult as it is.)
One way to check this out is to note that the indirect utility function for
type 2’s when good 1 is numeraire is
m
V (p, m) = √ . (21)
(1 + p)2
In equilibrium, Type 2’s have income
ω12
m = pω2 = . (22)
ω2

√ ω1 2
 
2
(1 + p) = 1 + . (23)
ω2
Let us define F (ω1 , ω2 to be the utility of Type 2’s in competitive equi-
librium when endowments are such that type 1’s have ω1 units of good 1 and
no good 2 and type 2’s have ω2 units of good 2 and no good 1.
Substituting from equations 22 and 23 into equation 21 and arranging
terms, we have
ω12 ω2
F (ω1 , ω2 ) = (24)
(ω1 + ω2 )2
Now we can check whether reducing the endowment of good 2 by 1/2
benefits or harms Type 2’s. Calculations are just a bit easier if we look at
the effect of doubling ω2 .

16
 We have
F (ω1 , 2ω2 ) ≥ F (ω1 , ω2 )
if and only if
2ω2 ω2
2
≥ .
(ω1 + 2ω2 ) (ω + 1 + ω2 )2
This is the case if and only if

2ω2 (ω1 + ω2 )2 ≥ ω2 (ω1 + 2ω2 )2 ,

which is equivalent to

2ω12 + 4ω1 ω2 + 2ω22 > ω12 + 4ω1 ω2 + 4ω22 .

But this implies that

ω12 ≥ 2ω22 ,
or equivalently √
ω1 ≥ 2ω1 .
Since we showed√ that doubling the endowments of ω2 makes type 2’s
better off if ω1 > 2ω1 , it follows
√ that a natural disaster that reduces ω2 by
50% benefits Type 2’s if ω1 < 2ω2 and harms Type 2’s if the inequality is
reversed.

Question 8.
A) A consumer consumes n goods and his utility function is u(x) where x is
his consumption vector. Define this consumer’s expenditure function.

The consumer’s expenditure function e(p, u) is the minimum income with

which the consumer can achieve utility u when the price vector is p.

B)If the utility function u(·) is homogeneous of degree k, what special struc-
ture does the expenditure function have?

If the utility function is homogeneous of degree k, the expenditure func-

tion is homogeneous of degree 1/k.

C) Assuming that the expenditure function corresponding to utility function

u(x) is well-defined, show that it is a concave function.

17
 We show that e(p, u) is concave in p. Let p and p0 be two different
price vectors. Then e(p, u) = ph(p, u) and e(p0 , u) = p0 h(p0 , u). Let p(λ) =
λp + (1 − λ)p0 where 0 ≤ λ ≤ 1. Then e(p(λ), u) = p(λ)h (p(λ)) , u).
Since h(p, u) is the cheapest bundle at prices p with utility u, it must be
that
ph(p(λ)) ≥ ph(p) = e(p, u). (25)
Likewise, since h(p., u) is the cheapest bundle at prices p0 with utility u,
it must be that
p0 h(p(λ)) ≥ p0 h(p0 ) = e(p0 , u). (26)
Now

e(p(λ), u) = p(λ)h(p(λ)
= λph(p(λ)) + (1 − λ)p0 h(p(λ)) (27)

Then it follows from Inequalities 25 and 26 and Equation 27, that

e(p(λ, u) ≥ λe(p, u) + (1 − λ)e(p0 , u). (28)

Therefore e(p, u) is concave in p.

18