ECON 6090: Section VII

TA: Nahim Bin Zahur

NOTE: I have included solutions for the final part of Question 2 and Question 3. We will also
revisit the Cobb-Douglas problem during this Friday’s section.

1. Cost minimization

a. You have estimated the following function relating a firm’s cost to its output (y) and the prices
it pays for labor (w) and capital (r) : C(w, r, y) = w1/2 r3/4 y 1/2 . With this information, can you
conclude that the firms technology exhibits increasing returns to scale?

b. You are the CEO of a multinational company with factories in two neighbouring countries 1 and
2. You produce some pre-determined level of output 2y, with y units produced in 1 and y units
produced in 2. Wages and capital rental rates in the two countries are given by {w1 , w2 , r1 , r2 }.
Labor is cheaper in country 1 and capital is cheaper in country 2 (w1 < w2 , r1 > r2 ).

A left-wing politician proposes opening up the border between 1 and 2. The result of this change
will be to equalize wages and capital rental rates in the two countries, so that the new wages and
capital rental rates will be given by w̄ = (w1 +w
2
2)
and r̄ = (r1 +r
2
2)
.

Should you support the change?

2. More cost minimization

Consider a firm that uses labor (l) and capital (k) to produce a single good (y). The input prices
for labour and capital are w and r respectively. Find the cost functions C(w, r, y) for the following
production functions. How does the cost vary with output y?

a. f (l, k) = ak + bl.

b. f (l, k) = min {ak, bl}.

c. f (l, k) = lα k β .

Answer

We first write the cost minimization problem as a maximization problem and take first-order nec-
essary conditions:

1
C(y, w, r) = − max −(wl + rk)
s.t. lα k β = y
l, k ≥ 0

Notice that if l = 0 or k = 0, then lα k β = 0 and the output constraint will not be met for any
y > 0. Therefore l, k > 0 in any solution and we can ignore those constraints when writing the
Lagrangian.

L = −(wl + rk) + λ(lα k β − y)

The first-order conditions are:

(1) (l) w = λαlα−1 k β

These conditions are sufficient as long as the input requirement set is convex, which is the case
as long as the production function is quasiconcave. We can check this by computing the principal
minors of the Bordered Hessian matrix, but there is a much simpler method here. Notice that
g(l, k) = logf (l, k) = α log(l) + β log(k) is a strictly concave function for any α, β. Since g(l, k) is a
strictly monotone transformation of f (l, k) and is strictly concave, it follows that f (l, k) is strictly
quasi-concave. (A monotone transformation of a concave function is quasi-concave).

Dividing (1) by (2) yields the following condition:

(4) w/r = αk/βl ⇒ k = wβl/rα

We can plug (4) into (3) to solve for k(y, w, r) and l(y, w, r):

(5) k(y, w, r) = y 1/(α+β) ( wr )α/(α+β) ( αβ )α/(α+β)

(6) l(y, w, r) = y 1/(α+β) ( wr )−β/(α+β) ( αβ )−β/(α+β)

The cost function can be computed by plugging the conditional input demand functions (5) and
(6) into the expression (wl + rk):

C(y, w, r) = wl(y, w, r) + rk(y, w, r)

3. Profit Maximization

You run a firm that uses labor (l) and capital (k) to produce a single good (y). Your sole objective
in life is to maximize profits. The production function is given by f (l, k) = lα k β , where α + β ≤ 1.
The input prices for labour and capital are w and r respectively. The output price is p.

a. Find the optimal level of output supplied.

2
Answer: The profit maximization problem is:

Π(p, w, r) =max(y,k,l) py − wl − rk

s.t. y = lα k β .

We could solve this directly. Notice, however, that if you are choosing y ∗ , k ∗ and l∗ to maximize
profits, k ∗ and l∗ must also minimize the cost of producing y ∗ (otherwise there would be some other
input combination that would give you the same output at a lower cost and thereby increase your
profits). So we can rewrite the profit maximization problem in terms of the cost function:
1 α β −β α
Π(p, w, r) = max(y) py − C(y, w, r) = max(y) py − y α+β w α+β r α+β [( αβ ) α+β + ( αβ ) α+β ]

The first-order necessary condition is:

p = M C, or:
1−α−β α β −β α
(8) p = 1
α+β
y α+β w α+β r α+β [( αβ ) α+β + ( αβ ) α+β ]

Since this is an unconstrained maximization problem, the first-order condition is sufficient as long
as the objective function py −C(y, w, r) is concave, which is true as long as α+β ≤ 1. We separately
consider the cases where α + β < 1, and α + β = 1.

Case 1: α + β < 1.

We can then solve for y(p, w, r) from (8)

Case 2: α + β = 1.

The technology then exhibits constant returns to scale, so the marginal cost is constant. There are
three possibilities:
α β −β α
1
a. If p < α+β w α+β r α+β [( αβ ) α+β + ( αβ ) α+β ], at any positive level of output, the firm’s profits are
negative. So the firm will choose y = 0 (this is the case when price < marginal cost).
α β −β α
1
b. If p = α+β w α+β r α+β [( αβ ) α+β + ( αβ ) α+β ], at any level of output, the firm’s profits are zero. So any
non-negative y is a profit-maximizing solution (this is the case when price = marginal cost).
α β −β α
c. If p > α+β 1
w α+β r α+β [( αβ ) α+β + ( αβ ) α+β ], at any positive level of output, the firm’s profits are
positive. Morever profits are strictly increasing in the output level. So the firm will choose output
level that is arbitrarily high (this is the case when price > marginal cost), and a finite solution to
the profit maximization problem does not exist.

3
b. Find the optimal level of labor and capital demanded.

Answer: We can use the fact that the profit-maximizing input demands are equal to the cost-
minimizing conditional input demands at the profit-maximizing level of output. Thus, we can plug
in the profit-maximizing level of output y(p, w, r) into the conditional input demand functions given
in (5) and (6):

(10) k(p, w, r) = y(p, w, r)1/(α+β) ( wr )α/(α+β) ( αβ )α/(α+β)

(11) l(p, w, r) = y(p, w, r)1/(α+β) ( wr )−β/(α+β) ( αβ )−β/(α+β)

where y(p, w, r) is given by equation (9) when α + β < 1.

What if α + β = 1? Then we cannot uniquely solve for output (as argued in the previous part), so
we cannot uniquely solve for the input demands as well.

c. How would an increase in wages change the demand for labor?

Answer: (We focus on the case where α + β = 1). We can use the expression in (11) to answer this.
Notice that given a fixed y, l(p, w, r) is decreasing in w (substitution effect). Morever, from (9), we
can see that an increase in w leads to a reduction in the profit-maximizing level of output y(p, w, r),
which from (11) leads to a decrease in l(p, w, r) (output effect). The substitution and output effects
reinforce each other, and thus the net effect of an increase in wages is to reduce the demand for labor.

d. What is your profit? How does your profit depend on α and β?

Answer: We can find profits by plugging the profit-maximizing level of output y(p, w, r) into the
profit function py(p, w, r) − C(y(p, w, r), w, r). This yields the following (nasty-looking) profit func-
tion:
1 −α −β −β α 1 α+β 1
Π(p, r, w) = p 1−α−β [w α+β r α+β [( αβ ) α+β + ( αβ ) α+β ]−1 ] 1−α−β ([α + β] 1−α−β − [α + β] 1−α−β )

Inspecting Π(p, r, w), we can see that it is strictly positive as long as α + β < 1. With decreasing
returns to scale, firms make positive profits in equilibrium. An increase in α (and similarly β) will
increase profits. One way to show this is to differentiate the above function with respect to α. A
second, easier approach is to argue that if α increases, the firm can reduce l while continuing to
produce the same level of output, and thus strictly increase profits.

e. What would happen if α + β > 1?

Answer: This is the case when there are increasing returns to scale. In that case marginal cost is
decreasing with the output level (you can verify this from (8)). Therefore there is no interior solution
to the profit maximization problem: the firm has an incentive to produce an arbitrarily high level of
output. To see why, suppose there exists some level of output y ∗ which is profit-maximizing. Then
it must be true that p = M C (first-order necessary condition). But by increasing y, the firm can
reduce MC, and so for all additional units that it produces, p > M C. Thus, the firm can increase
profits by increasing y beyond y ∗ , so y ∗ cannot be profit-maximizing.