Economics 401

8th April 2011

Instructions: Answer 7 of the following 9 questions. All questions are of equal weight.
Indicate clearly on the first page which questions you want marked.
1. Answer both parts.
(a) What does it mean to say that a utility function, u(), represents a preference relation
on some choice set X? Prove that if u() represents preference relation %, this preference
relation must be complete and transitive.
(b) Now suppose X = R2+ and (x11 , x12 )  (x21 , x22 ) when x11 > x21 , or x11 = x21 and x12 > x22 .
Is this preference relation complete, transitive and continuous? Defend your answers.
ANSWER
(a) A utility function u : X R represents a preference relation % on the choice set X
if
x, y X, x % y u(x) u(y).
Completeness: We need to show that x, y X, either x % y or y % x or both. Now
note that u(x) and u(y) are real numbers so either u(x) u(y) in which case x % y or
u(y) u(x) in which case y % x; if u(x) = u(y) we know both are true.
Transitivity: Suppose x, y, z X, and x % y and y % z. We need to show x % z. Since
u represents %, u(x) u(y) and u(y) u(z). Since these are real numbers u(x) u(z)
x % z.
(b) This is an example of lexicographic preferences where good 1 is the dominant good.
Lexicographic preferences are complete and transitive but not continuous.
Completeness: Consider any two distinct points in R2+ (x11 , x12 ) (point a) and (x21 , x22 )
(point b). If x11 > x21 , a  b. If x21 > x11 , b  a. If x11 = x21 , since a and b are distinct, it must
be that either x12 > x22 in which case a  b or x22 > x12 in which case b  a.
Transitivity: Let a, b, c R2+ where a = (x11 , x12 ), b = (x21 , x22 ) and c = (x31 , x32 ). Suppose
a  b and b  c. We want to prove a  c. Now notice that a  b and b  c x11 x31 .
1

There are two cases (i) and (ii). In case (i), x11 = x31 and then we know that x11 = x21 = x31 in
which case it must be that x12 > x22 > x32 , and thus x12 > x32 which means a  c. In case (ii)
x11 > x31 and then immediately we know a  c.
Continuity: Suppose
{xn } R2+ and

lim
x =x
n n

{yn } R2+ and

lim
y =y
n n

and

and xn  yn , n. Then if  were continuous we would be able to deduce that x  y. Here is

one counterexample. Let xn = (1/n, 0) and yn = (0, 1). Then x = (0, 0), y = (0, 1), xn  yn
n but y  x. So lexicographic preferences are complete and transitive but they violate
continuity.
2. This question applies what we have done in class this term to think about a carbon
tax. Suppose a typical Canadian household currently buys 200 litres of gasoline per month
at $1 per litre and spends $1800 per month buying goods (and services) other than gasoline. Assume the households choices are consistent with maximizing a Cobb-Douglas utility
function in the two goods gasoline and dollars spent on goods other than gasoline.
Now suppose the government is considering a carbon tax that will raise the price of gasoline
from $1 to $2 per litre. According to the EV measure of welfare change how much would
the government have to pay this household to exactly compensate the household for having
to pay the carbon tax? What would the answer be with CV?
ANSWER
Denote the price of gas by pg and gas consumed by G. The price of a dollar spent on
other goods is 1 and let m be the number of dollars spent on goods other than gasoline;
denote income/wealth by w. With CD preferences we know the indirect utility function and
the expenditure functions for
u (g, m) = g a m1a , 0 < a < 1
are
(1a)
V (pg , 1, w) = aa (1 a)1a pa
w and
g 1

e (pg , 1, u0 ) = aa (1 a)(1a) pag 11a u0

The level of utility with pg = 2 is clearly lower than the initial level of utility where
pg = 1. EV will be a negative number in this instance because utility goes down with the
price increase. Since its easier to work with positive numbers think of EV as a positive

number. Then EV uses initial prices to measure the number of dollars it would take to move
from the lower to the higher level of utility. Thus
EV = e (1, 1, V (1, 1, 2000)) e (1, 1, V (2, 1, 2000))
= 2000 2a 2000


= 2000 1 2a
CV (again, thought of as a positive number) is the same except that it uses the new price
of gas.
CV = e (2, 1, V (1, 1, 2000)) e (2, 1, V (2, 1, 2000))
= 2a 2000 2a 2a 2000
= 2000 (2a 1)
Determining a is easy. We know
G = aw/pg ,
so with the initial values of G = 200, w = 2000, and pg = 1 a must be 1/10.
3. Prove the following statements.
(a) Assume a two-good model. Prove the expenditure function, e (p1 , p2 , u0 ), must be
concave in prices;
(b) In the choice-based approach, with two goods, the Slutsky matrix, S, must be symmetric.
ANSWER
(a) Consider two sets of prices (p11 , p12 ) and (p21 , p22 ) and the convex combination of them
(pt1 , pt2 ) = t(p11 , p12 )+(1t)(p21 , p22 ) for 0 t 1. Then e(p1 , p2 , u0 ) is concave in prices (p1 , p2 )
if
e(pt1 , pt2 , u0 ) te(p11 , p12 , u0 ) + (1 t)e(p21 , p22 , u0 ).
Let (ht1 , ht2 ) be the expenditure-minimizing bundle for prices (pt1 , pt2 ) and utility level u0 .
Then
e(pt1 , pt2 , u0 ) =
=
=

pt1 ht1 + pt2 ht2

(tp11 + (1 t)p21 )ht1 + (tp12 + (1 t)p22 )ht2
t(p11 ht1 + p12 ht2 ) + (1 t)(p21 ht1 + p22 ht2 )
te(p11 , p12 , u0 ) + (1 t)e(p21 , p22 , u0 ).
3

The last line follows because the expenditure function minimizes the cost of buying
utility.
(b) The choice-based approach assumes the demand system is homogeneous of degree
zero and satisfies Walrass Law. With two goods this means that the Slutsky substitution
matrix, S, satisfies these identities for all positive prices:


 S11 S12



p1 p2
= 0 0
S21 S22
and


S11 S12
S21 S22



p1
p2


=

0
0

Thus from the first identity we know

p1 S11 + p2 S21 = 0
and from the second identity
S11 p1 + S12 p2 = 0
Putting these together we know that for any positive value of p2
p2 S21 = S12 p2 or
S21 = S12 .

4. I have talked with you about the importance of microeconometrics and we have spent
considerable time on tax policy. This question combines these themes. Microdata on male
labour supply show that men work pretty much the same hours per week whatever the wage
rate, and whatever the marginal tax rate. These facts are consistent with two very different
models. In model A, using standard notation and with proportional commodity tax rates in
effect, the typical persons budget constraint is
(1 + t1 ) p1 x1 + (1 + t2 ) p2 x2 + wl = wT,

lT

and preferences are represented by the following utility function

U (l, x1 , x2 ) = la xb1 x1ab
,
2

a > 0, b > 0, a + b < 1.

In model B, the firms people work for fix the hours worked per week for each employee,
and given the choice of working at these fixed hours, or no job, people choose to work the
hours specified by the employer, that is, time at work, T l, is fixed in model B. Suppose you
have been hired by the government to advise them on commodity taxation in particular,
for economic efficiency, should t1 > t2 or t2 > t1 or t1 = t2 ? Would your advice depend on
whether model A or model B is the better characterization of reality? Defend your answer.
4

ANSWER
Advice based on model A:
Here we are in a second-best world where taxation is distortionary. Is it efficient for the
government to set t1 = t2 ? We know that
t1 = t2 is efficient if and only if

d (l, x1 , x2 , u0 ) /x1
is independent of l,
d (l, x1 , x2 , u0 ) /x2

.
where d (l, x1 , x2 , u0 ) is the distance function that corresponds to U (l, x1 , x2 ) = la xb1 x1ab
2
To find the distance function apply the definition of the distance function and obtain



l x1 x2
U
, ,
= u0 or
d d d
 a    
l
x1 b x2 1ab
= u0 or
d
d
d
a b 1ab
d (l, x1 , x2 , u0 ) = u1
0 l x1 x2
Then
a b1 1ab
bx2
bu1
d (l, x1 , x2 , u0 ) /x1
0 l x1 x2
=
=
,
ab
1
d (l, x1 , x2 , u0 ) /x2
(1 a b) x1
(1 a b) u0 la xb1 x2

which is independent of l. Thus setting t1 = t2 is (second-best) efficient in model A.

Advice based on model B:
In model B, leisure is fixed so earnings are an endowment and then we know setting
t1 = t2 is first-best efficient.
To summarize, both models support equal proportional taxation of consumption goods.
The only difference between the models is that taxation will cause a deadweight loss in model
A, and, as usual, the higher the consumption tax rate, the higher the ratio of DWL to tax
revenue.
5. You are given the following information about a consumers purchases. Goods 1 and
2 are the only goods consumed.
Year 1
Quantity Price
Good 1
100
100
Good 2
100
100

Year 2
Quantity Price
120
p21
2
80
x2

For what values of p21 will there exist values of x22 that contradict the weak axiom of
revealed preference? Justify your answer carefully.
ANSWER
The year 1 bundle is revealed preferred to the year 2 bundle if
(100) (100) + (100) (100) (100) (120) + (100) x22 .
The year 2 bundle is revealed preferred to the year 1 bundle if
p21 (120) + (80) x22 p21 (100) + (80) (100) .
These two inequalities can be written as
80 x22
1
x22 100 p21
4
For WARP to be contradicted these two intervals must overlap so
1
80 100 p21 or
4
2
p1 80
Given that the person buys more of good 1 in year 2, if the relative price of good 1 goes
up in year 2 then its possible that the data could contradict WARP.
6. Consider a price-taking consumer in a two-good world where the consumer is endowed
with w dollars of wealth. Write the indirect utility function as the maximized value of the
appropriate Lagrangian expression and show that
V (p1 , p2 , w)
0.
p1
Now consider the same consumer in a world where she is endowed with quantities of the
two goods (e1 , e2 ) and no money. Fix p2 , e1 and e2 and describe as precisely as you can the
relationship between the modified indirect utility function V (p1 , p2 , e1 , e2 ) and p1 .
ANSWER
We know
V (p1 , p2 , w) =

Max
u (x1 , x2 ) + (w p1 x1 p2 x2 )
x1 , x2 ,

Applying the envelope theorem

V (p1 , p2 , w)
= (p1 , p2 , w) x1 (p1 , p2 , w) .
p1
Since (p1 , p2 , w) 0 and x1 (p1 , p2 , w) 0
V (p1 , p2 , w)
0.
p1
Now
V (p1 , p2 , e1 , e2 ) =

Max
u (x1 , x2 ) + (p1 e1 + p2 e2 p1 x1 p2 x2 )
x1 , x2 ,

Applying the envelope theorem

V (p1 , p2 , e1 , e2 )
= (p1 , p2 , e1 , e2 ) [e1 x1 (p1 , p2 , e1 , e2 )] .
p1
Since (p1 , p2 , e1 , e2 ) 0,
Sign

V (p1 , p2 , e1 , e2 )
= Sign .x1 (p1 , p2 , e1 , e2 ) e1
p1

Given the fixed p2 , e1 and e2 define p1 by

x1 (p1 , p2 , e1 , e2 ) = e1 .
Then a graph of V against p1 will have a U-shape with the bottom of the U at p1 .
7. Consider the expected-utility maximizing individual discussed in class who must decide
whether and for how much to insure his car. Assume the probability that he will not have an
accident is . Denote his initial wealth by w0 and denote his loss in the event of an accident
by L. Suppose that insurance is available at a price p for $1 worth of insurance coverage.
(a) Prove that if he is offered fair insurance he will choose to fully insure, that is, his
wealth will be the same whether he has an accident or not.
(b) At the point where p is set to deliver fair insurance, does the amount of car insurance
purchased increase or decrease with w0 ? Justify your answer carefully.
ANSWER
(a) The individuals expected utility is
f (x, p, , w0 , L) u (w0 px) + (1 ) u (w0 px L + x)
If the insurance is priced fairly the firms expected profit on each dollar of insurance
is zero, so
p + (1 ) (p 1) = 0 or
p = 1 .
7

To find the optimal value of x given the parameters of the problem set
f (x, p, , w0 , L)
=0
x
and solve for x at p = 1 . Thus
(p) u0 (w0 px) + (1 ) (1 p) u0 (w0 px L + x) = 0 or
u0 (w0 px) = u0 (w0 px L + x) and thus
w0 px = w0 px L + x or
x = L.
Thus his wealth is w0 (1 ) L whether he has an accident or not.
(b) We have just proved that when p = 1 , the demand for insurance, x (p, , w0 , L),
equals L, so
x (1 , , w0 , L) = L,
which holds for all sensible values of w0 at this point. Therefore
dx (1 , , w0 , L)
=0
dw0
8. In the two-type car insurance model discussed in class, when the insurance companies
cannot distinguish the good from the bad drivers, prove that a pooling equilibrium cannot
exist and explain why a separating equilibrium may not exist. If you use diagrams in your
answer make sure they are carefully drawn and clearly labeled.
ANSWER
Let the candidate for a pooling equilibrium be at point A on the market line (the mline). Since the good driver indifference curve through A must be steeper than the bad
driver indifference curve through A there must be points, say like B, below the bad driver
indifference curve and above the good driver indifference curve, southeast of A and below
the good-driver zero-profit line. If contract B were offered it would attract only the good
drivers and it would therefore make positive profits because it would attract only the good
drivers (it is below the g-line). The firms holding the contract at A will now lose money
because their pool of customers has disproportionately more bad drivers. Thus A cannot be
sustained.
The best chance of finding a separating equilibrium puts all the bad drivers at their
preferred point on the b-line, and the good drivers at the point where the bad driver indifference curve cuts the g-line; call this point X. If the m-line cuts the good driver indifference
curve through X there cannot be a separating equilibrium. The reason is that there will be
pooling contracts below the m-line and about the good driver indifference curve that will
8

attract both good and bad drivers and these contracts will also be profitable. These pooling
contracts knock out the best chance for a separating equilibrium. Since there cannot be a
pooling equilibrium, there is no trade all drivers, good or bad, end up at their endowment
point (w0 , w0 L).
9. (a) The AP mill in Dryden Ontario produces wood pulp with labour, N, capital, K,
and wood, W. Assume AP acts to maximize its profits and is a price-taker in all input and
output markets. Denote the input prices by PN , PK , PW and the output price by P. Using
data on profits, (which are always positive) suppose one of your colleagues has estimated
APs profit function using the following functional form:
ln = 0 + 1 ln PN + 2 ln PK + 3 ln PW + 4 ln P .
Since AP is a major employer in Dryden and there is a lot of unemployment in the
area the Ontario Ministry of Finance has decided to subsidize at a proportional rate, , the
price of labour to the firm. Assume APs data fit the above profit function perfectly and
no price changes as a consequence of the wage subsidy. Find a formula for in terms of
the parameters of the profit function if the governments objective is to increase APs use of
labour by 50 percent.
(b) Suppose a price-taking firm can produce output q with inputs z1 and z2 . When the
output price is p and input prices are w1 and w2 , the profit-maximizing levels of output and
inputs are q , z1 and z2 . Suppose in the short run z2 cannot be varied but z1 can be, but
0
in the long run both inputs are variable. Let the price of output rise to p > p, and let
input prices be constant. Prove that the increase in output in the long run is at least as
large as the increase in output in the short run.
ANSWER
(a) Since
(PN , PK , PW , P ) =

Max
P f (N, K, W ) (PN N + PK K + PW W ),
N, K, W

where f is the production function, we can use the envelope theorem to obtain the input
demand for N as a function of the input prices and the output price.
(PN , PK , PW , P )
PN
Taking the derivative of the profit function with respect to PN we obtain
N (PN , PK , PW , P ) =

1
PN

1
or
PN
1
N (PN , PK , PW , P ) =
.
PN
Since

3 4
P
= e0 PN1 PK2 PW

we can see that

3 4
P .
N (PN , PK , PW , P ) = 1 e0 PN1 1 PK2 PW

So to increase the firms use of N by 50 percent we need

3
= (1 + )1 1 ,
2
which can be solved for .
(b) Write short-run total cost as CS (q) and long-run total cost as CL (q). In CS (q) z2 is
fixed at z2 and so CS (q) CL (q) and we know the two are equal at q . So q is a minimizer
of g(q) CS (q) CL (q). Thus
g 0 (q ) = CS0 (q ) CL0 (q ) = 0
g 00 (q ) = CS00 (q ) CL00 (q ) > 0.
So at q the short-run and long-run marginal costs are equal to each other and the short0
run MC is steeper than the long-run MC. Given p0 > p, then, the q at p0 = CL (q) will exceed
0
the q at p0 = CS (q).

10