Solution to Problem Set Five - Demand:

Consumer Choice
Lecture Notes in Microeconomic Theory by Ariel Rubinstein
Last Update: Oct 30, 2007

P
1. Calculate the demand function for the utility function k αk ln(xk ).
The consumer’s problem is:
X X
max αk ln(xk ) st pk xk ≤ w and xk ≥ 0 for all k ∈ {1, . . . , K}.
{xk }
k k

Since xk = 0 ⇒ u(x) = −∞ for every k = 1, ..., K, the non-negativity

constraints will never bind since. The FOCs of the problem imply
αk αl αl
= ⇒ pl xl = pk xk for all k, l ∈ {1, . . . , K}.
p k xk pl xl αk
By substituting this result into the budget constraint for l 6= k, it follows
p k xk X w αk
αi = w ⇒ xk (p, w) = P for all k ∈ {1, . . . , K}.
αk i pk i αi

2. Verify that when preferences are continuous, the demand function x(p, w)
is continuous in prices and in wealth (and not only in p).
Choose an arbitrary convergent sequence of price/wealth pairs (pn , wn ) →
(p, w) such that (p, w)  (0, 0). Since x(p, w) is homogeneous of degree
zero, this implies  n 
n n p
x(p , w ) = x ,1 .
wn
Since demand is continuous in p, then
 n   
p p
x ,1 → x , 1 = x(p, w),
wn w

where the second equality follows from demand being homogeneous of

degree zero.

1
3. Show that if a consumer has a homothetic preference relation, then his
demand function is homogeneous of degree one in w.
Assume that % is homothetic, and let λ > 0 and (p, w) ∈ <K+1 ∗
++ . Let y ∈
∗ 1 ∗ 1 ∗
B(p, λw), ie py ≤ λw. Clearly, λ y ∈ B(p, w), and thus x(p, w) % λ y .
By homotheticity, it follows λx(p, w) % y ∗ .
Since y ∗ was arbitrarily chosen from B(p, λw), it follows that λx(p, w) % y
for every y ∈ B(p, λw). Moreover, λx(p, w) ∈ B(p, λw), and therefore
λx(p, w) = x(p, λw) since it is the optimal bundle in B(p, λw).

4. Consider a consumer in a world with K = 2, who has a preference relation

that is quasi-linear in the first commodity. How does the demand for the
first commodity change with w?
I will further assume that preferences satisfy strict convexity. Let p be a
price vector where p1 is normalized to 1. There exists an α ∈ <+ ∪ ∞
such that 
0 if w ≤ α
x1 (p, w) =
w − p2 x2 (p, α) otherwise.
In words, after a certain wealth threshold α, a change in wealth is absorbed
by commodity 1.

• x1 (p, w) = 0, w0 < w ⇒ x1 (p, w0 ) = 0.

Proof: By contradiction, assume w0 < w and x1 (p, w0 ) > 0. Define
another bundle
 
0 0 0
y = x1 (p, w ) + [w − w ], x2 (p, w ) .

Note that y is on the frontier of B(p, w). Consequently, there exists

0
a convex combination of x(p, w) and y, say z, where z = (w − w0 , w
p2 ).
Graphically, we have

r6x(p, w)
@
@
@
@
@r z
@
@ @
@ r x(p, w0 ) @r y
@ @
@ @
@ @-
Clearly, x(p, w) % y, and thus by strict convexity z  y. But then
0
0
(0, w
p2 )  x(p, w ) by quasi-linearity, a contradiction.

2
 • x1 (p, w) > 0, w < w0 ⇒ x1 (p, w0 ) = x1 (p, w) + [w0 − w].
Proof: By contradiction, assume x(p, w0 ) 6= x(p, w) + e1 [w0 − w].
Since x(p, w) % y for all y ∈ B(p, w), then
   
0 0
x1 (p, w) + [w − w], x2 (p, w) % y1 + [w − w], y2

for all y ∈ B(p, w) by quasi-linearity. Consequently, it must be that

x2 (p, w0 ) > pw2 . Graphically, we have

6
@
@
r z @ r x(p, w0 )
@
@
@ Preferred to bold line directly by q-linearity
@ @@
@ r x(p, w) @@r x(p, w) + e1 [w0 − w]
@ @@
@ @
@
@ @-

By quasi-linearity, it must be that z = (0, x2 (p, w0 )) is optimal in

B(p, w0 − x1 (p, w0 )). But w < w0 − x1 (p, w0 ), and thus by the first
claim x1 (p, w) = 0, a contradiction.

5. Let % be a continuous preference relation (not necessarily strictly convex)

and w a number. Consider the set G = {(p, z) ∈ <K K
++ × <+ | z is optimal
in B(p, w)}. Note that for some price vectors, there could be multiple
(p, x) ∈ G. Calculate G for the case of K = 2 and preferences represented
by x1 + x2 .
Note that the agent is indifferent between consuming α units of x1 and α
units of x2 . He will thus allocate his entire wealth to the cheapest good:

w



 p, p1 , 0
 if p1 < p2
p, α, w−p 1α
for any α ∈ 0, pw1


 
G(w) = p2 if p1 = p2
 p, 0, w



if p1 > p2
p2

Show that, in general, G is a closed set.

Let w > 0 and the sequence {(pn , xn )} be such that (pn , xn ) ∈ G(w) for
every n, and let (pn , xn ) → (p, x). We must show (p, x) ∈ G(w) to prove
that G is closed. In other words, we must show that x ∈ B(p, w) and that
x % y for every y ∈ B(p, w).

3
 Since (pn , xn ) ∈ G(w), then pn xn ≤ w for every n. It readily follows that

px = lim pn xn ≤ w ⇒ x ∈ B(p, w).

n→∞

By way of contradiction, assume x is not optimal in B(p, w). Then there

exists a y ∈ B(p, w) such that y  x. Then there exists an  > 0 such that
B (y)  B (x) by continuity, and thus there exists a bundle z ∈ B (y)
such that z < y and z  x. Moreover, since z < y and pn → p, then
pn z ≤ w for n large enough. But since xn → x, then xn ∈ B (x) for n
large enough, and thus z  xn , a contradiction to xn being the optimal
bundle in B(pn , w). Therefore x % y for every y ∈ B(p, w), and thus
(p, x) ∈ G(w).

6. Determine whether the following behavior patterns are consistent with the
consumer model:
 
2w w
(a) The consumer’s demand function is x(p, w) = 2p1 +p2 , 2p1 +p2 .
Yes, x(p, w) can be rationalized by the monotonic preference rela-
tion represented by u(x) = min{x1 , 2x2 }. Since % is monotonic, the
consumer will always set
i. p1 x1 + p2 x2 = w by Walras’ Law, and
ii. x1 = 2x2 by the functional form of u(x).
Substituting (ii) into (i), it follows
w 2w
2p1 x2 + p2 x2 = w ⇒ x2 (p, w) = ⇒ x1 (p, w) = .
2p1 + p2 2p1 + p2

(b) The consumer consumes up to quantity 1 of x1 and spends his excess

wealth on x2 .
Yes, the behavior is rationalizable by the utility function

x1 if x1 < 1
u(x) =
1 + x2 if x1 ≥ 1

(c) The consumer chooses a bundle (x1 , x2 ) which satisfies xx21 = pp12 and
costs w. Does the utility function u(x) = x21 + x22 rationalize the
consumer’s behavior?
No, u(x) = x21 +x22 does not rationalize the behavior. Since u(x) is not
quasi-concave, the maximization approach that we used in Question
1 is not appropriate. If 0 < p2 < p1 , then a consumer maximizing
u(x) would set x(p, w) = (0, pw2 ). Nevertheless, the consumer actually
chooses
x1 p1
= > 0 ⇒ x1 (p, w) > 0.
x2 p2

4
 Moreover, the behavior violates the WA and therefore is not ratio-
nalizable. Consider the choices from the following budget sets:
x((2, 1), 5) = (2, 1) and x((1, 2), 5) = (1, 2).
Note that each of the bundles is affordable at the other bundle’s
prices:
(1, 2) · x((2, 1), 5) = (1, 2) · (2, 1) = 4 < 5
(2, 1) · x((1, 2), 5) = (2, 1) · (1, 2) = 4 < 5,
which is a violation of the WA.

7. In this question, we consider a consumer who behaves differently from the

classic consumer we talked about in the lecture. Once again we consider a
world with K commodities. The consumers choice will be from budget sets.
The consumer has in mind a preference relation that satisfies continuity,
monotonicity, and strict convexity; for simplicity, assume it is represented
by a utility function u.
The consumer maximizes utility up to utility level u0 . If the budget set
allows him to obtain this level of utility, he chooses the bundle in the budget
set with the highest quantity of commodity 1 subject to the constraint that
his utility is at least u0 .

(a) Formulate the consumer’s problem.

The agent’s objective is
maxx∈B(p,w) u(x) if maxx∈B(p,w) u(x) < u0 , and
maxx∈B(p,w) x1 s.t. u(x) ≥ u0 if maxx∈B(p,w) u(x) ≥ u0 .

(b) Show that the consumer’s procedure yields a unique bundle.

Case 1: maxx∈B(p,w) u(x) < u0
In this instance, the consumer acts as in the standard framework.
Since preferences are continuous, monotonic and strictly convex,
then the problem has a unique solution (see the lecture notes).
Case 2: maxx∈B(p,w) u(x) ≥ u0
Suppose, by way of contradiction, that x and y both solve the
problem. Then x, y ∈ B(p, w), u(x), u(y) ≥ u0 and x1 = y1 .
Define z = αx + (1 − α)y for some α ∈ (0, 1). Note that z1 =
x1 = y1 , z ∈ B(p, w) and u(z) > min{u(x), u(y)} ≥ u0 .
Note there exists a j = 2, ..., K such that zj > 0, and by conti-
nuity there exists an  > 0 such that u(z − ej ) > u0 . With the
p
pj the consumer is saving, he can afford to purchase p1j more
p
units of commodity 1. Define z 0 = z − ej + p1j e1 , and note that
z 0 ∈ B(p, w), u(z 0 ) > u0 and z10 > x1 , a contradiction to x being
a solution to the problem.

5
(c) Is this demand procedure rationalizable?
Yes, it can be rationalized by the monotonic utility function

if u(x) < u0

u(x)
v(x) =
u0 + x1 if u(x) ≥ u0 .

(d) Does the demand function satisfy Walras Law?

Yes. If maxx∈B(p,w) u(x) < u0 , Walras Law is implied by monotonic-
ity. If maxx∈B(p,w) u(x) ≥ u0 , then px = w; otherwise, the consumer
could purchase more of x1 and obtain a better bundle.

(e) Show that in the domain of (p, w) for which there is a feasible bundle
yielding utility of at least u0 the consumer’s demand function for
commodity 1 is decreasing in p1 and increasing in w.
Let (p, w) be such that maxx∈B(p,w) u(x) ≥ u0 , and define p0 = p−γe1
for some γ > 0. Since p0 x(p, w) < w, then x1 (p, w) < x1 (p0 , w), as
the agent can afford strictly more of commodity 1 under B(p0 , w)
while maintaining utility u0 . Consequently, x1 (p.w) is decreasing in
p1 .
Similarly, define w0 = w + γ, and note px(p, w) < w0 . As before,
it must be that x1 (p, w) < x1 (p.w0 ), as the agent can afford strictly
more of commodity 1 under B(p, w0 ), and thus x1 (p.w) is increasing
in w.

(f) Is the demand function continuous?

Yes. Since demand is homogeneous of degree zero in (p, w), it is
sufficient to show that x(p, w) is continuous in p by Question 2. By
contradiction, assume that x(p, w) is not continuous in p, ie there
exists a sequence of prices {pn } such that pn → p and kx(p, w) −
x(pn , w)k ≥  for every n. For notational ease, define xn = x(pn , w)
and x = x(p, w). First, we must show that {xn } converges. As in
the lecture notes, define

m = inf pni | i ∈ {1, . . . , K} and n ∈ N > 0,



w
and note that xni ≤ m for every i = 1, ..., K and every n. Conse-
n w K
quently, {x } is contained in the compact hypercube [0, m ] . There-
n
fore, without loss of generality we can assume that x → y. Moreover,
note that py = limn→∞ pn xn ≤ w, and thus y ∈ B(p, w). There are
two cases to consider:
Case 1: u(y) < u0
Since x is unique by (b), then u(y) < u(x). By continuity, there
exists a point z  x such that pz < w and u(y) < u(z). Then

6
 for n large enough, pn z ≤ w and u(xn ) < u(z), a contradiction
to xn being the optimal bundle in B(pn , w).
Case 2: u(y) ≥ u0
Since x is unique by (b), then u(x) ≥ u0 and x1 > y1 . Define
z = 21 x + 12 y, and note that z1 > y1 , z ∈ B(p, w) and u(z) > u0
by strict convexity. By continuity, there exists an  > 0 such
that z1 −  > y1 , p[z − e] < w and u(z − e) > u0 . Then for n
large enough, pn [z − e] ≤ w and z1 −  > xn1 , a contradiction to
xn being the optimal bundle in B(pn , w).

8. A common practice in economics is to view aggregate demand as being

derived from the behavior of a “representative consumer.” Give two ex-
amples of “well-behaved” consumer preference relations that can induce
average behavior that is not consistent with maximization by a “represen-
tative consumer.” (That is, construct two “consumers,” 1 and 2, who
choose the bundles x1 and x2 out of the budget set A and the bundles y1
and y2 out of the budget set B so that the choice of the bundle x1 +x
2
2
from
y1 +y2
A and of the bundle 2 from B is inconsistent with the model of the
rational consumer.)
Consider the following sets of preferences:
 
x1 if x1 < 4 x2 if x2 < 4
u1 (x) = u2 (x) =
4 + x2 if x1 ≥ 4. 4 + x1 if x2 ≥ 4.

and consider the price/wealth pairs (pA , wA ) = ((1, 2), 8) and (pB , wB ) =
((2, 1), 8). Note that

xA
1 = (4, 2) xB
1 = (4, 0) xA
2 = (0, 4) xB
2 = (2, 4).

Taking averages across the price/wealth pairs, note that

xA = (2, 3) ⇒ pB xA = 7 < 8 ⇒ xB  xA , and

xB = (3, 2) ⇒ pA xB = 7 < 8 ⇒ xA  xB , a contradiction.

9. Let  be an acyclic binary relation on a finite set X. Show that there is

a complete, asymmetric and transitive relation ∗ which extends  (that
is, if a  b then a ∗ b.)

Lemma: If X is finite and  is an acyclic relation on X, then there exists

at least one x ∈ X such that for every other y ∈ X, either x  y or
x and y are not compared by .

7
Proof of Lemma: Let |X| = N . By contradiction, assume that for every
x ∈ X, there is a y ∈ X such that y  x. Choose any element in X,
say x1 . Then there exists another element in X, say x2 , such that
x2  x1 . Again, there exists an x3 ∈ X such that x3  x2 and so
on. Thus we can find a sequence such that xN +1  xN  ...  x1 .
Since |X| = N , then for some i ∈ {1, ..., N − 1}, xN +1 = xi . But
then xN +1  ...  xi+1 and xi+1  xN +1 , a contradiction to  being
acyclic. 

Since X is finite, then there exists an element in X, say x1 , such that no

other element in X is better than x1 by the Lemma. For all other y ∈ X,
let x1 ∗ y. Again, the Lemma implies that there exists an element in
X \{x1 }, say x2 , such that no element in X \{x1 } is better than x2 . Define
x2 ∗ y for all other y ∈ X \ {x1 }. Continue this process inductively, ie
for all n < N , there is an xn+1 that is one of the “best” elements in
X \ {x1 , ..., xn }; let xn+1 ∗ y for all other y ∈ X \ {x1 , ..., xn }.

Extension: If a  b, then a is defined as an element in the sequence

{x1 , ..., xN } before b, and hence a ∗ b.
Completeness: Implied by construction.
Asymmetry: Implied by construction.
Transitivity: Let a ∗ b and b ∗ c. Then a is defined as an element in
the sequence {x1 , ..., xN } before b, and b is defined before c. Conse-
quently, a ∗ c.