UNIVERSITY CARLOS III

Master in Economics Master in Industrial Economics and Markets

Game Theory. Final Exam
January 24th, 2018

TIME: 2 hours. Write your answers in this booklet.

NAME:

Problem 1:
Problem 2: Consider the following normal form game.

A B C
A 7,7 1,10 1,1
B 10, 1 4,4 1,1
C 1, 1 1, 1 2, 2

(1) Find the Nash equilibria in pure strategies.

Solution: There are two NE in pure strategies (B, B) and (C, C).

(2) Suppose that the above game is played two times. Can you find a sub-game perfect NE
of the repeated game with the following properties?
(a) no mixed strategies of the stage game are used.
(b) (A, A) is played in the first stage.

Solution: We try the trigger strategy. Each player i = 1, 2 plays the following
• At t = 1 play A.
• At t = 2 play B if (A, A) was played at stage t = 1. Otherwise, play C.

If both players follow the grim strategy their payoffs are u∗1 = u∗2 = 7 + 4 = 11. We
check if the above strategy constitutes a NE of the whole game. No player has incentives
to deviate at t = 2, since the strategy proposed is a NE of the stage game.
If say player i = 1, 2 deviates at t = 1 and the other player j ̸= i follows the grim
strategy, the payoff for player i is ui = 10 + 2 = 12. Hence, both players have incentive
to deviate. Therefore, the grim strategy is not a NE of the repeated game. The answer
to the question is no.

(3) Suppose that the above game is played three times. Can you find a sub-game perfect NE
of the repeated game in which (A, A) is played in the first stage?

Solution: We try the trigger strategy. Each player i = 1, 2 plays the following
• At t = 1 play A.
• At t = 2 play B if (A, A) was played at stage t = 1. Otherwise, play C.
• At t = 3 play B if (A, A) was played at stage t = 1 and (B, B) was played at stage
t = 2. Otherwise, play C.

1
2

If both players follow the grim strategy their payoffs are u∗1 = u∗2 = 7 + 4 + 4 = 15. We
check if the above strategy constitutes a NE of the whole game. No player has incentives
to deviate at t = 1, 2, since the strategy proposed plays a NE of the stage game in each
period.
If say player i = 1, 2 deviates at t = 1 and the other player j ̸= i follows the grim
strategy, the payoff for player i is at most ui = 10 + 2 + 2 = 14. Hence, no player has
incentives to deviate. Therefore, the grim strategy is a SPNE of the repeated game. The
answer to the question is yes.

Problem 3:
Problem 4: Consider the following signalling game. There are two types of player 1, t = t1
and t = t2 .

(2, 1) Player 1 (5, 4)

u t = t1 u
L R
a c
(2, 0) d d (3, 1)
0.4

Player 2 Nature Player 2

(4, 0) 0.6 (2, 0)

u u
b L R d

Player 1
(2, 1) d d (3, 1)
t = t2

(1) One of the types of player 1 has a dominating strategy. Which one?

Solution: If player t1 plays L his maximum payoff would be 2. If he plays R his

minimum payoff is 3. Hence, R is a dominant strategy for player t1 . We will assume
that in any PBNE , σ(t1 ) = R.

(2) Compute all the separating perfect Bayesian Nash equilibria (in pure strategies) of the
above game. Write the separating PBNE, including the beliefs of player 2. Justify your
answer.

Solution: By part (1) the only candidate for a separating PBNE is σ(t1 ) = R,
σ(t2 ) = L. In this equilibrium the beliefs of player 2 are

µ2 (c|R) = 1, µ2 (b|L) = 1

Given these beliefs, the best reply of player 2 is

BR2 (R|µ2 ) = {u}, BR2 (L|µ2 ) = {d}

Graphically,
 3

(2, 1) Player 1 (5, 4)

u t = t1 u
L R
a c
(2, 0) d d (3, 1)
0.4

Player 2 Nature Player 2

(4, 0) 0.6 (2, 0)

u u
b L R d

Player 1
(2, 1) d d (3, 1)
t = t2

Now, anticipating that player 2 will play σ2 (R) = {u} and σ2 (L) = {d}, the optimal
strategy for player 1 is to choose σ(t1 ) = R, σ(t2 ) = {L, R}. We conclude that the
following is a PBNE.

σ(t1 ) = R, σ(t2 ) = L
σ2 (R) = u, σ2 (L) = d
µ2 (a|L) = 0, µ2 (b|L) = 1
µ2 (c|R) = 1, µ2 (d|R) = 0

(3) Compute all the pooling perfect Bayesian Nash equilibria (in pure strategies) of the above
game. Write the pooling PBNE, including the beliefs of player 2. Justify your answer.
Solution: By part (1) the only candidate for a pooling PBNE is σ(t1 ) = R, σ(t2 ) = R.
In this equilibrium the beliefs of player 2 are

µ2 (a|L) = x, µ2 (b|L) = 1 − x
µ2 (c|R) = 0.4, µ2 (d|R) = 0.6

Given these beliefs, the expected utilities of player 2, given R, are

u2 (u|R) = 0.4 × 4 + 0.6 × 0 = 1.6

u2 (d|R) = 0.4 × 1 + 0.6 × 1 = 1

Hence, BR2 (R|σ1 , µ2 ) = {u}. Graphically,

(2, 1) Player 1 (5, 4)
u t = t1 u
L R
a c
(2, 0) d d (3, 1)
0.4

Player 2 Nature Player 2

(4, 0) 0.6 (2, 0)

u u
b L R d

Player 1
(2, 1) d d (3, 1)
t = t2
4

On the other hand, the expected utilities of player 2 , given L, are

u2 (u|L) = x × 1 + (1 − x) × 0 = x
u2 (d|R) = x × 0 + (1 − x) × 1 = 1 − x
Hence, 
u
 if x > 1/2
BR2 (R) = u, d if x = 1/2

d if x < 1/2

Suppose x > 1/2. Then BR2 (R) = u and player t2 would like to deviate from the
strategy σ(t2 ) = R to σ(t2 ) = L. We conclude that there there is no pooling equilibrium
with x > 1/2.
On the other hand, if x ≤ 1/2, then d is a best reply for player 2 and player 1 has no
incentives to deviate from the strategy σ(t1 ) = R, σ(t2 ) = R. We see that the following
is are PBNE for each value 0 ≤ x ≤ 1/2,

σ(t1 ) = R, σ(t2 ) = R
σ2 (R) = u, σ2 (L) = d
µ2 (a|L) = x, µ2 (b|L) = 1 − x
µ2 (c|R) = 0.4, µ2 (d|R) = 0.6

(4) Imagine player t = t1 follows the mixed stragey xL + (1 − x)R and player t = t2 follows
the mixed strategy yL + (1 − y)R with 0 < x, y < 1. What are the consistent beliefs of
player 2?

Solution: Note that

p(a) = 0.4x p(b) = 0.6y, p(c) = 0.4(1 − x), p(d) = 0.6(1 − y)
Therefore
p(a) 2x p(p) 3y
µ2 (a|L) = = µ2 (b|L) = =
p(a) + p(b) 2x + 3y p(a) + p(b) 2x + 3y

p(c) 2(1 − x) p(d) 3(1 − y)

µ2 (c|R) = = µ2 (d|R) = =
p(c) + p(d) 5 − 2x − 3y p(c) + p(d) 5 − 2x − 3y