EconS 501 - Microeconomic Theory I

Homework #2 - Answer Key

1. Exercise from FMG. Chapter 2, exercise 26. Consider an individual facing price
vector p = (p1 ; p2 ) >> 0 and income w > 0. If, after solving his UMP, his indirect
1
utility function is v(p; w) = p1 p21 w, show that his utility function u(x) must
have a Cobb-Douglas representation, where x = (x1 ; x2 ).

Proof (EMP approach). First, recall from duality that

u(x) min v(p; p x) (1)
p2R2+

Since v(p; w) is homogeneous of degree zero, we can divide both arguments by

p x to obtain that the indirect utility function v(p; w) is una¤ected, i.e., v(p; w) =
v ppx ; 1 . Let p p
px
, and thus v(p; w) = v (p; 1). As a consequence, if price
vector p minimizes v(p; p x) for an income level p x = w, then price vector p
minimizes v (p; 1) for an income level p x = 1. That is, we can rewrite program
(1) as
u(x) min2 v(p; 1) subject to p x = 1 (2)
p2R+

We can now …nd the price vector p = (p1 ; p2 ) that solves program (2). Plugging
them afterwards in the indirect utility function v(p; 1) will yield the original utility
function u(x) that this consumer maximized in his UMP (as stated in (2)). Since
program (2) is a constrained minimization problem, we set up the Lagrangian
L = p1 p21 (p1 x1 + p2 x2 1)
Taking …rst-order conditions with respect to p1 and p2 yields, respectively
dL 1 1
= p1 p2 x1 = 0
dp1
dL
= (1 )p1 p2 x2 = 0
dp2
and
dL
= p 1 x1 p 2 x2 + 1 = 0
d
and simultaneously solving for p1 and p2 we obtain
1
p1 = and p2 =
x1 x2
We can …nally plug these two prices, which solve (2), into the indirect utility
funciton v(p; 1), yielding
! 1
1
1
v(p1 ; p2 ; 1) = (1) = (1 ) 1 x1 x21 .
x1 x2 | {z }
constant, A

which is clearly of the Cobb-Douglas type. For instance, labeling A (1

) 1 yields v(p1 ; p2 ; 1) = Ax1 x21 , thus taking a more familiar format.

1
 Proof (Roy’s identify). Using Roy’s identity, we …nd that the Walrasian demand
for good 1 is
@v(p;w)
@p1 ( ) p 1 1 p2 1
w
x1 (p; w) = = = w
@v(p;w)
@w
p1 p2 1 p1

which, solving for p1 , gives us the indirect demand function p1 = x1

w. Similarly,
the Walrasian demand of good 2 is
@v(p;w) 2
@p2 ( 1) p1 p2 w 1
x2 (p; w) = = = w
@v(p;w)
@w
p1 p2 1 p2

which, solving for p2 , gives us the indirect demand function p2 = 1x2 w. Inserting
these indirects demands into the utility function, we obtain
! 1
1
1
v(p1 ; p2 ; w) = w w (w) = (1 ) 1 x1 x21 .
x1 x2 | {z }
constant, A

which is clearly of the Cobb-Douglas type. For instance, labeling A (1

) 1 yields v(p1 ; p2 ; w) = Ax1 x12 , thus taking a more familiar format.

2. Short proofs. Consider an individual with X = R2+ and budget line p1 x1 + p2 x2 = w,

where p1 , p2 , and w are all strictly positive.

(a) If the preference relation is continuous then the consumer’s problem has a solution.
If the preference relation is continuous then it has a continuous utility rep-
resentation. Given that both prices are positive, the budget set is compact,
so by Weirstrass’theorem the continuous utility function has a maximizer in
the budget set, which is a solution of the consumer’s problem.
(b) If the preference relation is strictly convex then the consumer’s problem has at
most one solution.
By contradiction, assume that a preference relation is strictly convex but there
are two bundlex x and y, where x 6= y, that are both solutions to a consumer’s
UMP. Then, the linear combination between x and y, x + (1 )y where
2 [0; 1], also lies in the budget set (which is convex). Since x + (1 )y is
a more balanced bundle than x and y alone. Because preferences are strictly
convex, the individual strictly prefers bundle x + (1 )y to bundle x and
y. As a result, bundles x and y cannot be solutions to his UMP.
(c) If the preference relation is monotone, then every solution of the consumer’s prob-
lem must be on the budget line.
By contradiction, suppose that the bundle x = (x1 ; x2 ) solves the consumer’s
UMP, but lies below the budget line, i.e., p1 x1 + p2 x2 < w. Then, there must
be another bundle y = (x1 + "; x2 + "), which is in the " -ball around x, that
also lies strictly below the budget line, that is,
p1 (x1 + ") + p2 (x2 + ") < w

2
 for a small enough " > 0. Because preferences satisfy monotonicity, bundle y
is strictly preferred to x, y x, implying that bundle x cannot be a solution
to the UMP. (We found another bundle y that was still a¤ordable and strictly
preferred to x.)
(d) The demand function of a rational consumer whose marginal rate of substitution,
M RS(x1 ; x2 ), is increasing in x2 for every value of x1 has the property that good
1 is normal.
Consider the price pair (p01 ; p02 ) and let wealth increase from w to w0 , where
w0 > w. Let bundle a be a solution to the UMP when the consumer faces
(p01 ; p02 ; w), as depicted in the …gure below. Consider now an increase in wealth
from w to w0 , without changing any of the prices, and let a0 denote a bundle
that lies on the new budget line, not necessarily a solution to the UMP, with
the property that he consumes the same amount of good 1 in both bundles,
x1 = x01 ; also depicted in the …gure.

Because the MRS is increasing in x2 and x1 = x01 , we must have that

M RS(x01 ; x02 ) > M RS(x1 ; x2 ). Hence, the solution of the UMP when the
consumer faces (p01 ; p02 ; w0 ), which we denote as bundle b on the …gure, must
satisfy contain more units of good 1, that is, b1 > x1 = x01 . Overall, an
increase in wealth, lead to an increase in the consumer’s demand for good 1.

3. UMP in several utility functions. Consider a consumer with budget line p1 x1 +

p2 x2 = w, where p1 , p2 , and w are all strictly positive. Find her Walrasian demand in
the following utility functions, and explain how the consumer distributes her wealth,
w, across both goods.
p
(a) Quasi-linear utility function: u(x1 ; x2 ) = x1 + x2 .
The marginal rate of substitution is
M U1 1 p
M RS = = 0:5 = 2 x2 :
M U2 0:5x2
p q
As x2 decreases, the M RS = 2 x2 decreases from 2 pw2 (where x2 = pw2
p
indicates the vertical intercept of the indi¤erence curve) to 2 0 = 0 when

3
 x2 = 0. This indicates that two corner solutions can arise, depending on the
price ratio:
q
p1
– If 2 pw2 p2
holds, there is a corner solution to the UMP, (x1 ; x2 ), where
p p2
M RS = pp21 , or 2 x2 = pp12 . Solving for x2 , yields x2 = 4p12 . His demand for
2
good 1 can be found by inserting x2 into the budget line, p1 x1 +p2 x2 = w,
that is,
p21
p 1 x1 + p 2 =w
4p22
| {z }
x2

p2
which simpli…es to p1 x1 + 4p12 = w. Solving for x1 , we obtain that x1 =
w p1
p1 4p2
.
q
– If, instead, 2 pw2 < pp12 holds, there is a corner solution where the individ-
w
ual spends all his wealth on the second good, x2 = p2
and purchases zero
units of good 1, x1 = 0.
q
Intuitively, the consumer purchases positive units of both goods when 2 pw2
p1
,which implies that good 1 (that entering linearly in his utility function) is
p2 q
relatively inexpensive, i.e., price ratio pp21 must be low for condition 2 pw2 p1
p2
q
to hold. If, instead, good 1 is relatively expensive, condition 2 pw2 < pp12 holds,
inducing the consumer to spend all his income on the good that enters non-
linearly. This is a common feature of quasi-linear utility functions.
(b) "Max" utility function: u(x1 ; x2 ) = maxfx1 ; x2 g.
First, we depict an indi¤erence map.

This utility function gives rise to three cases:

– If pp21 < 1 (p1 < p2 ), the solution will be at the lower right corner, where
x1 = pw1 and x2 = 0. The consumer uses all her wealth on x1 .
– If pp21 = 1 (p1 = p2 ), both the lower right and upper left corners are
solutions. The consumer evenly splits her wealth between both goods,
w
that is, x1 = x2 = 2p .

4
 – If pp12 > 1 (p1 > p2 ), the solution will be at the upper left corner, where
x1 = 0 and x2 = pw2 . The consumer uses all her wealth on x2 .
Intuitively, the consumer spends all her wealth on the cheaper good if prices
di¤er, but evenly splits her wealth between both goods if their prices coincide.
Other utility functions such as that in part (c), u(x1 ; x2 ) = x21 + x22 , and
u(x1 ; x2 ) = x1 + x2 , generate similar Walrasian demands.
(c) Utility function with both goods entering quadratically: u(x1 ; x2 ) = x21 + x22 .
As described above, this utility function generates a Walrasian demand like
the utility function in part (b), where all the wealth is spend on the cheaper
good. As a practice, we include some explanation below.
First, note that this individual’s indi¤erence curves are strictly concave in
the (x1 ; x2 )-quadrant. Indeed, solving this individual’s utility function for x2 ,
we obtain the equation
p of an indi¤erence curve (evaluated at a generic utility
level u) is x2 = u x21 . Di¤erentiating with respect to x1 , yields

1 1
x02 (x1 ) = (u x22 ) 2 ( 2x1 ) < 0
2
and di¤erentiating again with respect to x1 , we …nd that
1 3 1
x002 (x1 ) = (u x22 ) 2 ( 2x1 )( 2x1 ) + (u x22 )( 2) < 0
4 2
When indi¤erence curves are not strictly convex, either strictly concave (like
those in part b or in part c) or linear (perfect substitutes), we can anticipate
that the consumer’s Walrasian demands will be corner points.
If we compare the utility levels associated to each of these bundles, we obtain
that the consumer prefers ( pw1 ; 0) to (0; pw2 ) if and only if u( pw1 ; 0) > u(0; pw2 ),
2 2
which implies wp2 > wp2 or, after rearranging, p2 > p1 . We can, then, summa-
1 2
rize the Walrasian demand as follows:
8
>
> ( w ; 0) if p2 >p1
>
< p1
(x1 (p; w); x2 (p; w)) = ( w ; w ) if p = p
>
> 2p 2p 2 1
>
:(0; w ) if p <p
p2 2 1

p p
(d) Utility function with both goods entering as a square root: u(x1 ; x2 ) = x1 + x2 .
As a practice, note that, after solving for x2 , we …nd the equation of an
p 2
indi¤erence curve (evaluated at a generic u), to be x2 = u x1 . Di¤er-
u
entiating this expression with respect to x1 , yields 1 px1 , and di¤erentiating
u
again with respect to x1 , we …nd 3=2 , which is unambiguously positive (i.e.,
x1
positive for any value of x1 and u). As a result, we can claim that indi¤er-
ence curves are strictly convex (bowed in towards the origin). This property
helps us anticipate interior solutions, as opposed to the previous part of the
exercise.

5
 We now set the price ratio equal to the marginal rate of substitution, as
follows,
M U x1 0:5x1 0:5 p1
= 0:5 =
M U x2 0:5x2 p2
which, after rearranging, yields
2
p1
x2 (p; w) = x1 :
p2
Inserting this expression into the budget line, p1 x1 + p2 x2 = w, yields
2
p1
p 1 x1 + p 2 x1 = w
p2
| {z }
x2

After simplifying and solving for x1 , we obtain the Walrasian demand for
good 1,
wp2
x1 (p; w) =
p1 (p2 + p1 )
Similarly, the Walrasian demand function for good x2 is
wp1
x2 (p; w) = .
p2 (p2 + p1 )
If p1 > p2 , Walrasian demands satisfy x2 (p; w) > x1 (p; w) since p2 (pwp 1
2 +p1 )
>
wp2 p1 p2 p1 p2
p1 (p2 +p1 )
simpli…es to p2 > p1 , which holds true because p2 > 1 > p1 given that
p1 > p2 . If, instead, p1 < p2 , Walrasian demands satisfy x2 (p; w) < x1 (p; w)
using a similar argument. Finally, if prices coincide, p1 = p2 = p, these
w
Walrasian demands coincide as well and simplify to x1 (p; w) = x2 (p; w) = 2p .
(e) Utility function with intervals: u(x1 ; x2 ) = x1 x2 if x1 10 units, but u(x1 ; x2 ) =
10x2 otherwise.
The next …gure depicts the consumer’s problem. The consumer would never
purchase more than 10 units of good 1, as doing so would not help him increase
his utility. Graphically, we have well-behaved (Cobb-Douglas) indi¤erence
curves for all x1 10 units, but a kink at x1 = 10 and a ‡at segment for all
x1 > 10 (as with Leontie¤ preferences).

6
Case 1. If x1 < 10, we set the price ratio equal to the marginal rate of
substitution, obtaining
M U x1 x2 p1
= =
M U x2 x1 p2

which, solving for x2 , yields

p1
x2 (p; w) = x1 .
p2

Inserting this expression into the budget line, p1 x1 + p2 x2 = w, we …nd that

p1
p 1 x1 + p 2 x1 = w
p2
| {z }
x2

which simpli…es to p1 x1 + p1 x1 = w, and yields the Walrasian demand for

good 1, as follows
w
x1 (p; w) = :
2p1
Note that, for the initial condition x1 < 10 to hold, we need that 2pw1 < 10, or
w
20
< p1 , intuitively indicating that good 1 is relatively expensive. Otherwise,
this case cannot be supported, and Case 2 below is the only possible solution.
Using the tangency condition in this setting, x2 (p; w) = pp21 x1 , we can …nd the
Walrasian demand for good 2, that is,
p1 w w
x2 (p; w) = = .
p2 2p1 2p2
Then, the indirect utility function of the consumer in this case is
w w w2
v 1st (x1 ; x2 ) = x1 x2 = = .
2p1 2p2 4p1 p2
Case 2. If x1 = 10, the Walrasian demand function for good 2 can be found
by inserting x1 = 10 into the budget line, that is
p1 10 + p2 x2 = w
w 10p1
Solving for x2 , yields x2 (p; w) = p2
. In this case, the consumer’s indirect
utility function becomes
w 10p1
v 2nd (x1 ; x2 ) = 10x2 = 10 .
p2
w
Comparison. If 20 < p1 holds and, thus, Case 1 can be supported, we can
compare the indirect utility functions in each case, v 1st (x1 ; x2 ) and v 2nd (x1 ; x2 ),
as follows
w2 w 10p1
v 1st (x1 ; x2 ) v 2nd (x1 ; x2 ) = 10
4p1 p2 p2
2
(w 20p1 )
= ;
4p1 p2

7
which is always positive. Therefore, the consumer chooses bundle (x1 ; x2 ) =
w
; w .
2p1 2p2