SOLUTION TO SAMPLE EXAM 1

Note: This file contains a solution to the sample exam as if I were answering
it. This means that it should give you an idea of what I expect from your answers.
Two hints: 1) If I ask you to define something, give a clear one-two-liner definition,
do not write paragraphs; writing a lot of unrelated stuff will ABSOLUTELY NOT
increase your grade; 2) when finding Nash equilibria of games: do not write too
much; be precise; make sure I can READ WELL what the equilibrium is; I want to
see that you understand the intuition behind the game: you can get partial credit for
having CORRECT INTUITION and failing to solve the task completely.
Algebraic mistakes WILL NOT decrease your grade (most likely). This class is
about your understanding modern microeconomics, not algebra. So, you MIGHT
get full credit if you solve the problem correctly but arrive to the wrong numerical
answer due to the algebraic mistake.

1. (10 points) Explain briefly the essence of Westinghouse-GE Case. Describe a game that
models the situation with Westinghouse and GE.
Westinghouse and GE were producing turbines in 1950’s and 60’s. For 20 years
they maintained monopoly prices without explicit agreement (tacit collusion).
This regime was broken once for 2 years when two European firms entered the
market. They started to price undercut both Westinghouse and GE. This lead
to price war and lowering of the prices. After 2 years Westinghouse and GE
lobbied the law according to which foreign companies could not sell turbines in
the US. This lead to the re-establishment of the monopoly collusion pricing. In
the end, US antitrust authorities stopped all this in 70’s.
The model of this situation is infinitely repeated Bertrand duopoly. In this game
two firm collude to price their goods at monopoly level. No one firm wants to
deviate from this because the strategy of the other firm is to start the price war
once deviation happens. Thus deviation leads to the competitive pricing and
zero profits.

2. (10 points) Explain the difference between Cournot and Stackelberg duopoly. How do
profits of firms compare in these games? How do total profits compare to the profit of a
monopoly?
In Cournot duopoly firms choose quantities simultaneously. In Stackelberg
duopoly first firm 1 chooses quantity, then firm 2 observes it and chooses quan-
tity. Given inverse demand P(q) = a − q and fixed marginal cost c for each

1
 ( a − c )2
firm, the Nash equilibrium profit in a Cournot game is 9 each. In Stackel-
( a − c )2
berg game, profit of the leader firm is 8 and the profit of the follower firm
( a − c )2
is 16 . So, leader firm in Stackelberg case makes half the monopoly profits
as would the follower firm if they could collude. However, strategic interaction
leads to inefficiency (from the firms’ perspective) in which follower firm gets a
quarter of the monopoly profits instead of a half.
I don’t mind if you remember profits in these games by heart as long as you
specify the parameters of the market like inverse demand and marginal cost.
You also might derive them: that might give you additional credit.

3. (10 points) Give an example of a commitment device. What is credible and incredible
threat?
Doomsday device in the Dr. Strangelove movie is a perfect commitment device.
It creates credible threat because its creators cannot stop it if they are attacked.
Since the creators of Doomsday device removed the possibility of them stopping
it, the threat is credible and other party is not willing to attack. Incredible threat
by a player is a situation such that when it actually gets to threatening the other
party it is not in player’s interests to do that. So, threatened party is not “afraid”
of it.

4. (10 points) Find Nash Equilibria of the following game:

A B C
N 0,2 3,2 0,4
W 3,0 2,3 1,0
S 5,0 4,2 3,5
E 2,4 0,0 2,0

There is one Nash equilibrium (S,C) (Underline best responses in the game ma-
trix. That way I see how you got the solution.).
Note: Nash equilibrium consists of THE ACTIONS that players choose NOT
PAYOFFS. If you specify it the way I just did you are going to be fine, but if you
say that Nash equilibrium is (3, 5) you will be wrong because (3,5) might also
happen somewhere else in the game.

2
 5. Consider Prisoner’s Dilemma with additional action:
C D N

C 4, 4 0, 7 0, 0

D 7, 0 2, 2 0, 0

N 0, 0 0, 0 0, 0

a) (10 points) Suppose this game is repeated finitely many times. Is it possible to
sustain cooperative outcome (C, C ) in some equilibrium for at least 1 period? If
Yes, choose appropriate number of periods that you need to repeat the game and
specify equilibrium strategies. If No, explain.
Yes, you can sustain cooperation (C, C ) in an equilibrium of a finitely re-
peated game. For example, if this game is repeated three times and the
strategy of both players is
– Play C, D, D if other player does the same thing
– play N if other player deviates from C, D, D.
This strategy is indeed Nash equilibrium of the repeated game because in
periods 2 and 3 no one wants to deviate since both players are already in
NE of the stage game and the other NE of the stage game gives strictly less
for either player. In period 1 if a player wants to deviate she gets 7 + 0 +
0 (since other player will punish her according to the prescribed strategy)
and on the equilibrium path she will get 4 + 2 + 2 > 7 + 0 + 0. Thus no
deviation will occur.
b) (10 points) Suppose this game is repeated infinitely many times. For which dis-
counting factor δ can you sustain cooperation (C, C ) in equilibrium? Specify
equilibrium strategies and calculate for which δ these strategies constitute an equi-
librium.
Consider the following strategy for both players: play C as long as other
player plays C. If deviation occurs, play D (or N) forever.1

1 Ininfinitely repeated game punishing with D or N is all fine. You will just get different deltas.
Both solutions are correct.

3
D Punishment. Equilibrium path gives discounted payoff of 4/(1 − δ).
Best deviation in first (or for that sake any other) period gives

δ
7 + δ2 + δ2 2 + δ3 2 + ... = 7 + 2
1−δ

So no deviation will occur if

δ
4/(1 − δ) ≥ 7 + 2
1−δ
δ
4 ≥ (1 − δ)[7 + 2]
1−δ
4 ≥ 7 − δ7 + δ2
4 ≥ 7 − δ5
3
δ ≥
5

You do not have to be this specific about derivations. Believe me, I

can see what you are doing. Specifying the original condition and the
answer is enough. Keep in mind that I DO WANT TO SEE THE ORIG-
INAL CONDITION that gave you the answer. The same logic: COR-
RECT INTUITION is more important to me than numerically correct
answer. Though of course an answer is still required.
N Punishment. Equilibrium path gives discounted payoff of 4/(1 − δ).
Best deviation in first (or for that sake any other) period gives

7 + δ0 + δ2 0 + δ3 0 + ... = 7

So no deviation will occur if

4/(1 − δ) ≥ 7
4 ≥ (1 − δ )7
4 ≥ 7 − δ7
3
δ ≥
7