Economics 201b

Spring 2010
Problem Set 4 Solutions

1. Robinson-Crusoe in U.S.S.R. Consider a “Robinson-Crusoe” economy with

two-goods, one consumer and one firm. Firm is labor-oriented: it maximizes
the profits per unit of labor, given the wage rate w and price of potatoes p (i.e.
it maximizes π(p,w)
`
where ` is an amount of labor).

(a) What is the definition of competitive equilibrium in this case? Give a

formal definition. Call it P (“Proletariat”) equilibrium. (Yes, P equilibria
of the whole world, unite!).
Solution. One can define P equilibrium in the same way as it was defined
in a familiar Robinson-Crusoe economy in Lecture 4. The only distinction
is in the firm’s maximization problem: rather then maximize profits firm
would maximize profits, per unit of labor. So, define P equilibrium as a
tuple ((p∗ , w∗ ), (x∗1 , x∗2 ), (`∗ , q ∗ )) such that

1. (x∗1 , x∗2 ) ∈ argmax U (x1 , x2 ) s.th. p∗ x2 ≤ w∗ (L − x1 ) + π(p∗ , w∗ ),

x1 ,x2
p∗ f (`)−w∗ `
2. `∗ ∈ argmax `
,
`
3. x∗1 ∗
+ ` = L.

(b) When production function f (z), where z is labor input, is strictly concave
what is the set of all Pareto optimal allocations?
Solution. Again, as before, firms have no say in the determination of the
Pareto optimal allocations. So, this set is just the solution to the following
problem: argmax U (x1 , f (L − x1 )) (and clearly x2 = f (L − x1 )).
x1 ∈[0,L]
Notice that under assumption of strict quasi-concavity of U , this set is just
a singleton (i.e. there is a unique Pareto optimal allocation).

(c) Continue to assume that f (z) is strictly concave, under what condition on
utility function does the P equilibrium exist? Give the description of the
equilibrium in this case.
Solution. First, make standard assumption f (0) = 0 and without any
loss of generality set p = 1. Second, observe that under strict concavity
of f (z) assumption, the only solution to the firms maximization problem
defined in (a), is `∗ = 0. To see this, note that if f 0 (`) = ∞ as ` → 0
f (`) f 0 (`)
then l’Hospital rule gives lim = lim = +∞ by our assumption,
l→0 ` l→0 1
so `∗ = 0 is, indeed, the maximizer. Now, if f 0 (`) 6= ∞ as ` → 0, then FOC
Economics 201b Spring 2010 Problem Set 4 Solution

d(`f 0 (`) − f (`))

implies `f 0 (`) − f (`) = 0. Since = f 00 (`) < 0, `f 0 (`) − f (`)
d`
is strictly decreasing, and, thus, the only maximizer is `∗ = 0. In other
word, the ratio f (`)`
is always decreasing.
Consequently, the only possibility for P equilibrium to exist is
(L, 0) ∈ argmax U (x1 , x2 ) s. th. x2 ≤ 0.
x1 ,x2

The easiest way to achieve that is to set U (x1 , x2 ) = U (x1 ), (i.e. Robinson
does not care about consumption good, only leisure.) Aside from that
possibility, non-convex utility function will generate (L, 0) corner solution
for range of prices.

(d) For an arbitrary production function check whether P equilibrium is Pareto

efficient.
Solution. For the reasons described in (c), with a well-behaved utility
function P equilibrium(-a) will not be Pareto optimal. The same holds
true for strictly convex f (z), with the only distinction that `∗ = L in this
case.
But, with CRS technology, i.e. f (z) = kz, there will be at least one P
equilibrium (more if utility function is not strictly quasi-concave) which is
Pareto optimal. It is obtained by setting w = k1 , because with CRS any
x1 ∈ [0, L] is a maximizer.

(e) Now, suppose that in recognition of such great management innovation,

firm receives an award from Politburo of the economy (i.e. local social
planner). Thus, the firm maximizes now π(p,w)+a
`
where a > 0 is the fixed
award amount. If the utility function U is quasi-concave, continuous and
strictly monotone, are there any conditions on the production function
such that P equilibrium exist? Prove or give counterexample.
Solution. Following similar arguments as in (c) and (d), maximizing
π(p, w) + a
yields `∗ = 0 as unique maximizer for the case when f (z) is
`
either concave or linear. When f (z) is strictly convex the unique maxi-
π(p, w) + a
mizer is either `∗ = L or `∗ = 0 (because is a sum of strictly
`
increasing function π(p,w)
`
and strictly decreasing function a` , thus, it has
only a unique minimizer on the interior and is maximized at the end-
points.) Consequently, in general, there are no condition on f (z) such
that P equilibrium exists.
2. Robinson-Crusoe: back to Berkeley. Consider again following “Robinson-
Crusoe” economies with two-goods, one consumer and one firm. For each case,
compute all Pareto optimal allocations and check whether or not the Second
Welfare Theorem holds. Justify your answer.

2
Economics 201b Spring 2010 Problem Set 4 Solution

(a) U (x1 , x2 ) = log x1 + log x2 , ω = (24, 0),

Y = {(−y1 , y2 ) : y2 ≤ ey1 −1 , y1 ≥ 0}
Solution. Notice that the production set is non-convex. The Pareto set
is given by
max log x1 + log x2 , s.th. x2 ≤ e23−x1
x1 ,x2

The solution is given by

max log x1 + 23 − x1
0≤x1 ≤24

∂U
Since ∂x i
→ −∞ as xi → 0, the solution will always be interior. The unique
maximizer is x∗1 = 1, thus, the Pareto optimal allocation is (x∗1 , x∗2 ) =
(1, e2 2). However, this allocation can’t be sustained as a competitive equi-
librium, because profit maximization will never yield an interior solution
with non-convex production set.
(b) U (x1 , x2 ) = log x1 + log x2 , ω = (24, 0),
 3
y
4 1
if 0 ≤ y1 ≤ 20
Y = {(−y1 , y2 ) : y2 ≤
(y1 − 20)2 + 15 if 20 < y1
Solution. Notice that production set is linear over some range and
then starts to exhibit increasing returns to scale. To find Pareto optimal
allocation we need solve two maximization subproblems and then compare
utility levels. For 0 ≤ x1 ≤ 20, the problem is
3
max log x1 + log x2 , s.th. x2 ≤ (24 − x1 )
x1 ,x2 4
The solution is given by
 
72 − 3x1
max log x1 + log .
4≤x1 ≤24 4
Again, the solution is a unique interior maximizer x1 = 12 with U (12, 9) =
log 108. For x1 > 20, the problem is
max log x1 + log x2 , s.th. x2 ≤ (4 − x1 )2 + 15
x1 ,x2

The solution is given by

max log x1 + log (4 − x1 )2 + 15 .


0≤x1 <4

There is no interior solution, and, thus, we only need to check one corner
(4, 15). But, since U (4, 15) = log 60 < log 108 = U (12, 9), the unique
Pareto optimal allocation is (x∗1 , x∗2 ) = (12, 9). However, this alloca-
tion can’t be sustained as a competitive equilibrium, because of the non-
convexities in the production set. Notice, that it does not matter that
Pareto optimal allocation lies in the region where production frontier is
linear. As long as there are increasing returns to scale for y1 > 20, the
Second Welfare Theorem will fail to hold.

3
Economics 201b Spring 2010 Problem Set 4 Solution

(c) U (x1 , x2 ) = 3x21 + ex2 , ω = (24, 0),

Y = {(−y1 , y2 ) : y2 ≤ log (y1 + 1), y1 ≥ 0}
Solution. Notice that the utility function is non-convex. The Pareto set
is given by
max 3x21 + ex2 , s.th. x2 ≤ log (25 − x1 )
x1 ,x2

The solution is given by

max 3x21 + 25 − x1
0≤x1 ≤24

Notice that x∗1 = 24 is the maximizer, thus, the Pareto optimal allocation
is (x∗1 , x∗2 ) = (24, 0) and it can be sustained as a competitive equilibria if
p1
p2
is sufficiently low.
Notice that non-convexities in the utility function will, in general, lead
to the failure of the Second Welfare Theorem. However, the latter only
provides sufficient conditions for any Pareto optimal allocation to be a
competitive equilibria with transfers and in this case the unique Pareto
optimal allocation (24, 0) can be sustained as a competitive equilibria.
Finally, observe that firm maximization problem implies y11+1 = w. Con-
sequently, setting p∗ = 1, w∗ = 1 generates a competitive equilibrium
allocation (24, 0).

3. Quasi-equilibrium to equilibrium in economy with production. In lec-

ture we have shown that with strict monotonicity of preferences any price quasi-
equilibrium is also a price equilibrium in pure exchange economy. Now, you
need to prove that under our assumptions on preferences (continuous, convex
P strongly monotone) as well as an additional assumption that ∃yj ∈ Yj :
and
j yj + ω̄  0 this claim is also true in the Arrow-Debreu economy with pro-
duction.
L
Solution. Let {%}P be continuous and strongly monotone, X = R+ and
assume ∃yj ∈ Yj : j yj + ω̄  0. Define quasi-equilibrium withP transfers
(x , yP, p , T ) as an assignment of wealth levels (w1 , ..., wI ) with i wi = p∗ ·
∗ ∗ ∗

ω̄ + i p∗ · yi∗ such that

1. xi %i x∗i =⇒ p∗ · xi ≥ wi ∀i,
2. ∀yj ∈ Yj : p∗ · yj∗ ≥ p∗ · yj ∀j,
P ∗ P ∗
3. i xi = ω̄ + i yi

Step 1. Show that p ≥ 0. Notice that p 6= 0 by assumption (in the proof

of the Second Welfare Theorem it is a result of the application of Minkowski
Separating Hyperplane Theorem). To see this, assume by contradiction that
∃l : pl < 0.
 Without any  loss of generality set l = 1, i.e. p1 < 0. Now, define
xi = xi + − p1 , 0, . . . , 0 . By strict monotonicity x0i i x∗i . Now consider the
0 ∗ 1

4
Economics 201b Spring 2010 Problem Set 4 Solution

 
cost of this bundle p · x0i = p · x∗i + p1 · (− p11 ) . Since p · x0i < wi this is a
contradiction, thus, p ≥ 0.
Step 2. Show wi > 0 for some i. Observe that profit maximization by firms
implies
∀yj ∈ Yj : p∗ · yj∗ ≥ p∗ · yj ∀j
summing it up over j X X
p · yj∗ ≥ p · yj
j j

which is true if X X
p · yj∗ + p · ω̄ ≥ p · yj + p · ω̄
j j
∗
P
and by our assumption j p · yj + p · ω̄ 
P 0. Recall that wi = p · wi + p ·
∗
P
θ
j ij · yj + T i and by our previous result i wi > 0 =⇒ ∃i such that wi > 0.

Step 3. Show p  0. Since we know that p ≥ 0 and p 6= 0 by our assumption,

all we need to show that pl = 0 for some l leads to a contradiction. But, suppose
not and there exists at one good (without loss of any generality, the first one),
such that p1 = 0. Consider x0i = x∗i +(ε, 0, . . . , 0), ε > 0. By strict monotonicity
x0i i x∗i and the cost of this bundle is the same p·x0i = p·x∗i = wi . By continuity
of preferences:
if x0i i x∗i ∃¯ s.th. yi i x∗i ∀yi ∈ B¯(x0i ).
But this implies that there exists some bundle in the budget set that costs as
much as x∗i , but is strictly preferred, which is a contradiction.
Step 4. Finally, we show that if some bundle is strictly preferred to x∗i then
it will have to cost more then x∗i (which is equivalent to say that it is a price
equilibrium). Notice that if wi = 0 then both demand and quasi-demands are
empty sets and we are done. If wi > 0 then p · x∗i > 0. Now, suppose that
x0i i x∗i but p · x0i = p · x∗i = wi . Now, by continuity of the preferences we have
∃¯ s.th. yi i x∗i ∀yi ∈ B¯(x0i ).
Observe that B¯(x0i ) ∩ Bi 6= ∅, where Bi is a budget set of the consumer i. But
this is a contradiction and we are done.

4. “Tricky” Boundary Conditions. A common misconception about the bound-

ary condition on excess demand is to think that it says that if the price of a
good goes to zero, then excess demand for that good goes to infinity. Although
intuitively plausible, this is false even for very well-behaved preferences, since
relative prices matter. Working this problem should help you avoid this mis-
conception.
Consider the preference relation on R3+ represented by the utility function
√ √ x3
U (x1 , x2 , x3 ) = x1 + x2 +x2 + 1+x 3
, and let the consumer’s initial endowment
be ω = (1, 1, 1).

5
Economics 201b Spring 2010 Problem Set 4 Solution

(a) Show that U is strongly monotone, strictly concave, and continuous.

∂U
Solution. Strong monotonicity requires showing that ∂xl
> 0 for each
good l. We have
∂U 1 −1/2
= x > 0 ∀x1 ∈ R++
∂x1 2 1
∂U 1 −1/2
= x + 1 > 0 ∀x2 ∈ R++
∂x2 2 2
∂U 1
= > 0 ∀x3 ∈ R++
∂x3 (1 + x3 )2

Strict concavity requires that the Hessian matrix of second derivatives be

negative definite. We have:
 −1
0 0

3/2
4x
 1 −1
D2 U (x1 , x2 , x3 ) =  0 3/2
4x2
0  
−2
0 0 (1+x3 )2

√ √
which is negative definite ∀x ∈ R3++ . Continuity. Note that x1 , x2 ,
x3
x2 , and 1+x 3
are all continuous functions, and that the sum of continuous
functions is continuous, so U is continuous.

(b) If (x1 , x2 , x3 ) ∈ R3+ and x3 > 0, show that U (x1 , x2 + x3 , 0) > U (x1 , x2 , x3 )
√ √
Solution. U (x1 , x2 + x3 , 0) = x 1 + x + x3 + x2 + x3 . Note that
x3 √ √ 2
x3 > 1+x3 ∀x3 ∈ R+ , and x2 + x3 > x2 ∀x2 , x3 ∈ R+ . Thus if
(x1 , x2 , x3 ) ∈ R3+ and x3 > 0, then U (x1 , x2 + x3 , 0) > U (x1 , x2 , x3 ).

(c) If p = (p1 , p2 , p3 ) >> 0 and p2 = p3 , show that x3 (p) = 0.

Solution. Proceed by contradiction and suppose
x3 (p) > 0. From part

(b) above, we know that U x1 (p), x2 (p)+x3 (p), 0 > U x1 (p), x2 (p), x3 (p) .
Also, because p2 = p3 , we have



p · x1 (p), x2 (p) + x3 (p), 0 = p · x1 (p), x2 (p), x3 (p) .


These two facts contradict x1 (p), x2 (p), x3 (p) being demanded at price
p, so we must have x3 (p) = 0.

(d) For each n, let pn = (1 − n2 , n1 , n1 ). Show that x3 (pn ) = 0 for each n (and
thus that demand for x3 remains bounded even though pn3 → 0).
Solution. For this sequence, we have pn  0 and pn2 = pn3 ∀n. Thus, by
part (c), x3 (pn ) = 0 for each n.

6
Economics 201b Spring 2010 Problem Set 4 Solution

(e) Show that lim x2 (pn ) = ∞.

n→∞

Solution. From what we learned about preferences in part (a) and

the proof of question 4, we know that the boundary condition on excess
demand must hold. That is, |z(pn )| → ∞. Demand for at least one of the
goods must be blowing up. And since x3 (pn ) = 0 ∀n, it must be either
good 1 or good 2 that blows up.
If x1 (pn ) → ∞, Then we have that pn · x(pn ) → ∞ which violates Walras’
Law (which must hold given strong monotonicity). Thus, we must have
lim x2 (pn ) = ∞.
n→∞

5. Importance of Assumptions. Consider a two good economy, and illustrate

graphically four examples of functions z : ∆o → R2 which demonstrate that
if any one of the conditions: continuity; Walras’ Law; boundary condition;
bounded below fails, then there may not be a solution to z(p) = 0. That is,
each function you draw should violate only one of the four conditions, and have
the property that @p s.t. z(p) = 0.

(a) Continuity
Solution. Consider following z(p)
 " #
αpω
 p1
− ω1
, if p 6= p∗

 (1−a)pω
− ω

 2
z (p) = " p2 #
k
p∗1
 ∗
 − k∗ 6= 0 ∀k > 0, if p = p



p 2

Notice that z(p) as defined above for p 6= p∗ represent excess demand

for a Cobb-Douglas ”representative consumer,” so that there is a unique
p∗ ∈ ∆o such that z(p) 6= 0 (for Cobb-Douglas preferences AD curves are
always slope downwards). So, if one redefines excess demand in just one
point, p = p∗ , the result is discontinuous excess demand function z(p) that
satisfies all other properties and, clearly, no Walrasian equilibrium in that
economy.
Note that z(p) satisfies Walras’ Law
α·p·ω (1−a)·p·ω
− p2 · ω2 if p 6= p∗

p1 − p1 · ω1 + p2
p · z(p) = p1 p2
=0 if p = p∗

which implies

p · ω − p1 · ω1 − p2 · ω2 = 0 if p 6= p∗

p · z(p) =
=0 if p = p∗

7
Economics 201b Spring 2010 Problem Set 4 Solution

is bounded below
(ω1 , ω2 )o if p 6= p∗
(
− max n
z (p) ≥ =⇒
− max pk∗ , pk∗ if p = p∗
1 2
 
k k
z (p) ≥ − max ω1 , ω2 , ∗ , ∗
p 1 p2

and satisfies boundary conditions: if pl → ∞ zl (p) → ∞ since

α·p·ω
lim − ω1 → ∞
p1 →∞ p1
(1 − α) · p · ω
lim − ω2 → ∞
p2 →∞ p2

(and we do not need to check for boundary conditions when p = p∗ ).

See figure 5(a).

(b) Walras’ Law

Solution. Again, consider the excess demand function z (p) for a Cobb-
Douglas ”representative consumer” and translate the curve so that it does
not pass through the origin
" α·p·ω # 
− ω

p1 1 c1
z (p) = (1−a)·p·ω +
p
− ω2 c2
2

Note that z (p) as defined above is: continuous as a sum of two continuous
functions; bounded below and satisfies boundary conditions (see the argu-
ments given above). Also, one can check that p · z (p) = −p1 c1 − p2 c2 6=
0 whenever p1 c1 6= p2 c2 .
See figure 5(b).

(c) boundedness below (∃x ∈ R s.t. z(p) ≥ x ∀p ∈ ∆o ),

Solution. (∃ x ∈ R s.th. z(p) ≥ x ∀p ∈ ∆o ). Consider
 1 
p1
z (p) =
− p12

Note that it is clearly a continuous function because p1l is a continuous

function for ∀p ∈ ∆o , satisifies Walras Law and boundary conditions, but
is unbounded below because.

6 ∃M > 0 s. th. zl (p) > −M ∀l, ∀p.

See figure 5(c).

8
Economics 201b Spring 2010 Problem Set 4 Solution

(d) boundary condition (if pn → p ∈ ∆\∆o , then |zl (pn )| → ∞)

Solution. Consider following z (p)

p·a p·b
z (p) = b− a
k k
where a, b ∈ R2++ , k ∈ R++ and p ∈ ∆. It is easy to check that continuity,
Walras’ Law and bounded below conditions are satisfied but boundary
conditions not.
See figure 5(d).

6. Continuity of correspondences. Let ψ : ∆ → 2∆ is a correspondence. Show

that ψ is uhc if it has a closed graph. Demonstrate graphically an example of
correspondence ψ : X → 2X such that ψ has a closed graph but ψ is not uhc.

Solution. Let ψ : ∆ → 2∆ has a closed graph. Then by the definition

M.H.3 in MWG we only need to show that the image of compact set under ψ
is bounded. So, let K ⊂ ∆ be compact. Since ψ(K) ⊂ ∆, ψ(K) is bounded.
Thus, ψ is uhc.
Please see figure 6 for the example of correspondence that has a closed graph
but not uhc.

9
Economics 201b Spring 2010 Problem Set 4 Solution
Economics 201b Spring 2010 Problem Set 4 Solution
Economics 201b Spring 2010 Problem Set 4 Solution
Economics 201b Spring 2010 Problem Set 4 Solution
Economics 201b Spring 2010 Problem Set 4 Solution