0% found this document useful (0 votes)
13 views

Section 7

This document contains the solutions to problems from a mathematics for economists course. It addresses questions about determining if functions are supermodular, optimal actions as a function of parameters for maximization problems, separating equilibria in signaling games, properties of Nash equilibria in Bertrand competition models, and the single crossing property. The solutions analyze the functions and models using concepts like supermodularity, increasing differences, and indifference curves.

Uploaded by

Victor Rudenko
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
13 views

Section 7

This document contains the solutions to problems from a mathematics for economists course. It addresses questions about determining if functions are supermodular, optimal actions as a function of parameters for maximization problems, separating equilibria in signaling games, properties of Nash equilibria in Bertrand competition models, and the single crossing property. The solutions analyze the functions and models using concepts like supermodularity, increasing differences, and indifference curves.

Uploaded by

Victor Rudenko
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Mathematics for Economists-1

Sections 7 (Problem set 6), October 18, 2023


Alexander Tonis, Kirill Savin

0. (Ch. 10.5, Exercise 7) Let x ∈ S = [0, 1] and θ ∈ Θ = R+ . In each of the following cases, determine
if the given function is supermodular in (x, θ). In each case, determine also if the optimal action
correspondence in the problem max{f (x, θ) | x ∈ S} is nondecreasing in θ.
(a) f (x, θ) = xθ − x2 θ2 .
Solution
∂ 2 f (x, θ)
= 1 − 4xθ ⇒ f is strictly supermodular in (x, θ) (for all x ∈ [0, 1]) if θ ∈ [0, 1/4] and is
∂x∂θ
not supermodular if θ can be higher than 1/4.
Let D∗ (θ) := arg max{f (x, θ) | x ∈ S}. If θ > 0, then f is strictly concave in x and the FOC for
∂f (x, θ)
an interior solution is = θ − 2xθ2 = 0 ⇒ D∗ (θ) = min(1/(2θ), 1) and D∗ (0) = [0, 1].
∂x

The optimal actions are weakly increasing in θ for θ ∈ [0, 1/2]. Thus, supermodularity is sufficient
but not necessary for monotonicity.
(b) f (x, θ) = xθ − x2 .
Solution
∂f (x, θ)
= θ − 2x is increasing in θ and decreasing in x ⇒ f is strictly supermodular in (x, θ)
∂x
and strictly concave in x ⇒ D∗ (θ) = min(θ/2, 1). The optimal action is weakly increasing in θ.

(c) f (x, θ) = x/(1 + θ).


Solution
∂f (x, θ)
= 1/(1+θ) > 0 is decreasing in θ ⇒ f is strictly supermodular1 in (x, −θ) and D∗ (θ) ≡ 1.
∂x
1
To deal with decreasing parametric monotonicity, one can redefine parameters or variables, e.g. use θ0 = −θ instead of θ.
(d) f (x, θ) = x(1 + θ).
Solution
∂f (x, θ)
= 1 + θ > 0 is increasing in θ ⇒ f is strictly supermodular in (x, θ) and D∗ (θ) ≡ 1.
∂x
(e) This task is identical to (b), so probably there is a typo. I suppose that the following function is
meant: f (x, θ) = x(x − θ).
Solution
∂f (x, θ)
= 2x − θ is decreasing in θ and increasing in x ⇒ f is strictly supermodular in (x, −θ)
∂x

 1 if θ < 1,
and strictly convex in x ⇒ the solution is at x = 0 or x = 1 ⇒ D∗ (θ) = {0, 1} if θ = 1,
0 if θ > 1.

The optimal actions are weakly decreasing in θ.

1. (Signaling) Let S := {1, 2} be the set of actions available to a set of economic agents. Each agent
has a privately known type θ which belongs to the set Θ := {1, 2}. The action x selected by the agent
is observed in the labor market and rewarded at the wage rate w(x). In turn, c(x, θ) is the cost of
action x for an agent of type θ, leading to a net payoff π(x, θ) := w(x) − c(x, θ). Say that the wage
scheme w(·) separates the high type agent from the low type if each type θ has a unique optimal action
x∗ (θ) such that x∗ (2) > x∗ (1).

(i) What conditions on the cost function c(·, ·) would be sufficient for the existence of a wage scheme
w(·) that separates the high type agent from the low type? Explain your answer.
Solution
A wage scheme w(·) given by w(1) and w(2) separates the high type agent from the low type if
x∗ (1) = 1 and x∗ (2) = 2 ⇔ π(1, 1) > π(2, 1) and π(2, 2) > π(1, 2) ⇔ w(1)−c(1, 1) > w(2)−c(2, 1)
and w(2) − c(2, 2) > w(1) − c(1, 2) ⇔ c(2, 2) − c(1, 2) < w(2) − w(1) < c(2, 1) − c(1, 1). There
exists a wage scheme satisfying these inequalities if −c(·, ·) is strictly supermodular.
(ii) Assuming c(·, ·) is a C 2 function on R2 , suggest an alternative answer to question (i) in terms of
the derivatives of c.
Solution
Using the second-order characterization of supermodularity, we obtain the following sufficient
∂ 2 c(x, θ)
condition: there exists a separating wage scheme if < 0 for (x, θ) ∈ (1, 2)2 .
∂x∂θ

2. (Bertrand price competition with linear demand) Consider 2 firms competing in prices. If the
vector of prices chosen by the firms is p = (p1 , p2 ) ∈ R2+ , the total demand for firm i’s product is given
by

qi (p) = 2 − pi + pj
where pi is the firm’s own price and pj is the other firm’s price.(For simplicity, we are allowing for the
case qi (p) < 0.) The profit of firm i is given by

πi (p) = (pi − ci )qi (p).

where ci ≥ 0 is a parameter that represents the cost of producing one unit of output for firm i.
A Nash equilibrium is a price vector p∗ = (p∗1 , p∗2 ) such that p∗i solves the following problem for each
firm i:

max πi (pi , p∗j )


pi ∈R+

Without using the first order conditions, show that there cannot exist two Nash equilibria p∗ and p̂∗
such that p∗1 < p̂∗1 and p∗2 > p̂∗2 .
Solution
2
Firstly, let us show that πi (p) is strictly supermodular: ∂∂pπi ∂p
i (p)
j
= 1 > 0 ∀pi , pj , i 6= j. By symmetry,
the same is true for πj (p). Strict supermodularity implies strict increasing differences (in this case
these two properties are actually equivalent).
Now, suppose by contradiction that ∃p∗ , p̂∗ such that p∗1 < p̂∗1 and p∗2 > p̂∗2 .
As p∗ and p̂∗ are the solutions ⇒ π1 (p∗1 , p∗2 ) ≥ π1 (p̂∗1 , p∗2 ) and π1 (p̂∗1 , p̂∗2 ) ≥ π1 (p∗1 , pˆ∗2 ).
Strictly increasing differences property of π1 (p) and p∗ , p̂∗ being the solutions imply:

0 ≥ π1 (p̂∗1 , p∗2 ) − π1 (p∗1 , p∗2 ) > π1 (p̂∗1 , p̂∗2 ) − π1 (p∗1 , pˆ∗2 ) ≥ 0.

A contradiction. Hence, @p∗ , p̂∗ satisfying these inequalities.

3. Consider a function π : R3 → R defined as

π(w, x, θ) = w + u(x, θ) ∀(w, x, θ) ∈ R3 .


2
Here, u : R2 → R is a C 2 function such that ∂ ∂x∂θ u(x,θ)
> 0 for every (x, θ) ∈ R2 . Pick some numbers
θ1 , θ2 , N1 , N2 ∈ R such that θ1 6= θ2 . Show that there can be at most one pair (w, x) ∈ R2 that satisfies
both of the following equations:

π(w, x, θ1 ) = N1 , π(w, x, θ2 ) = N2 . (1)

That is, if this system of equations has a solution (w, x), then there exists no other solution (w0 , x0 ).
(Intuitively, this means that the indifference curves associated with π(·, ·, θ1 ) and π(·, ·, θ1 ) can intersect
at most one point. This is the source of the term single crossing property.)
Solution
∂ 2 u(x,θ)
Note that ∂x∂θ > 0 implies strict supermodularity of u.
Suppose that there exists another solution (w0 , x0 ). Then

π(w0 , x0 , θ1 ) = N1 , π(w0 , x0 , θ2 ) = N2 . (2)

Subtract (2) from (1):


w + u(x, θ1 ) − w0 − u(x0 , θ1 ) = w + u(x, θ2 ) − w0 − u(x0 , θ2 ) = 0 ⇒
u(x, θ1 ) − u(x0 , θ1 ) = u(x, θ2 ) − u(x0 , θ2 )
which is not consistent with supermodularity of u if x 6= x0 . Hence x = x0 ⇒ w = N1 − u(x, θ1 ) =
N1 − u(x0 , θ1 ) = w0 . Thus, (w, x) = (w0 , x0 ).

You might also like