MSC.

IN ECONOMICS
GAMES & CONTRACT THEORY
MID-TERM
19 November 2015
Time allowed for the exam: 1h20m
Teachers: Henrique Monteiro
Group I 5 pts. (20m)
Correct answers: 0.5 pts.;
Wrong answers: -0.25 pts. (T/F) or -0.125 pts. (multiple choice).

True-False Questions:
1. A mixed strategy is a probability distribution over the feasible pure strategies.
2. A continuation game is a particular case of a subgame.
3. The correct solution concept for a dynamic game with complete information is
the Bayesian equilibrium.
4. If a base game has a unique Nash equilibrium, in a finite set of repetitions of the
stage game, the Nash equilibrium is implemented in each stage.
Complete the following sentences: (wrong answers have no point deduction penalty)
5. A fact is ______________________ among the players of a game if the players know
the fact, they know that the other players also know, they know that the other players
know that they know it and so on.
6. A players ______________________ can be found by maximizing across his
strategies the minimum payoff he can achieve depending other players actions.

Multiple-choice questions:
7. How do you call the strategy which consists in cooperating until cooperation is
breached by another player and not to cooperate ever again from that moment onwards?

a) Dominant strategy
b) Tit-for-tat strategy
c) Mixed strategy
d) Trigger strategy

8. When all players have the same probability distribution over the states of the world,
we call that distribution the:

a)
Common prior
b)

Common knowledge

c)

uniform

d)

Posterior

9. In a static game of incomplete information, strategies and actions:

a)
Differ
b)

Do not exist

c)

Coincide

d)

Are opposite

10. Nash equilibriums are never:

a)
The intersection of best response functions
b)

Pareto optimums

c)

The intersection of payoff functions

d)

A combination of strategies

Group II 5 pts. (20m)

1. (2.5 pts.) Consider the following pollution game1:
U.S. Steel

Oxy Steel

Strategies

Low pollution

High pollution

Low pollution

($100,$100)

(-$30,$120)

High pollution

($120,-$30)

($100, $100)

Find the Nash equilibrium of the game. Would the actions chosen by the steel
companies change if one of them had a chance to decide first, and the other only
decided after observing what the first one did? Justify your answer.
2. (2.5 pts.) What is a strategy? For static games, how do strategies differ between
complete and incomplete information settings?

Samuelson, Paul A. And William D. Nordhaus (1993), Economics, 4th edition, McGraw-Hill, chapter
12.

Group III 10 pts. (40m)

1. (5 pts.). Consider the following information about a 2-player game (players A

and B):
- Action space for both players: Ai = {Rock , Paper , Scissors} ;
-

1 if

Payoff functions: u i = 0 if
1 if

v(ai ) > v(a j )

a i = a j , i, j A, B
v(ai ) < v(a j )

v(Rock ) > v(Scissors )

where the following intransitive (circular) relations hold: v(Scissors ) > v(Paper )
v(Paper ) > v(Rock )

a) (2 pts.) Represent this game in normal form for the case where both players
choose simultaneously and everything is common knowledge. Does this game
have any pure-strategies Nash equilibriums? If so, which? Does this game have
mixed-strategies Nash equilibriums? Justify your answer.
b) (1.5 pts.) Assume now that player A chooses first and player B observes As
choice and chooses next. Represent this game in extensive form and find the
outcome of the game. Does player A enjoy a first-mover advantage in this case?
Justify your answer.
c) (1.5 pts.) Assume again that both players choose simultaneously, but that player
A is uncertain whether player Bs payoff in case of victory is 1 or 2, believing
that both situations are equally likely to occur. Is there a Bayesian Nash
equilibrium in this game? If so, which? Justify your answer.

2. (5 pts.) Consider the following oligopoly formed by firms 1, 2 and 3. The

inverse demand function for the good they produce is P = 30 Q . Total costs
for each of the three firms are Ci (qi ) = 10qi , i = 1, 2 and 3 , where qi is the
quantity produced by firm i .
a) (2.5 pts.) Compute the Nash equilibrium of this oligopoly.
b) (2.5 pts.) Suppose firms 2 and 3 merge. Call the new firm 4. There are now only
two firms in the market, firms 1 and 4. Assume again that both firms decide
simultaneously and also that firm 1 does not know firm 4s cost function,
believing there is an equal chance that the costs are given by each of the
following functions C 4L (q4 ) = 5q4 , C 4M (q4 ) = 10q4 or C4H (q4 ) = 15q4 . Find the
Bayesian Nash equilibrium.

Solution
Group I
True-False Questions
1) T
2) F
3) F
4) T
Complete the sentences
5) common knowledge
6) maxmin value/security level
Multiple-choice questions
7) d
8) a
9) a
10) c

Group II
1) (2.5 pts.)
Nash equilibrium: (High pollution; High pollution)
U.S. Steel

Oxy Steel

Estratgias

Low pollution

High pollution

Low pollution

($100,$100)

(-$30,$120)

High pollution

($120,-$30)

($100, $100)

To see if it makes a difference whether the game is dynamic or static and whether the
order of play makes a difference, we represent the game in extensive form by using
game trees for both possible cases according to which company decides first.
If Oxy Steel decides first, the outcome will still be (High pollution; High pollution)

High

($100, $100)

U.S. Steel
High

Low
($120,-$30)

Oxy Steel
High

(-$30,$120)

Low

U.S. Steel

Low
($100,$100)

If U.S. Steel decides first the outcome becomes (High pollution; High pollution)

High

($100, $100)

Oxy Steel
High

Low
($120,-$30)

U.S. Steel
High

(-$30,$120)

Low

U.S. Steel

Low

($100,$100)
The dynamic nature of the game does not change the outcome no matter which
company decides first.
Note: In the static version of the game it could already be seen that both players had
dominant strategies (which returned a higher payoff no matter the choice of the
competitor). If both maximize payoffs independently of what the other chooses, the
ordering will not make a difference.

2) (2.5 pts.)
A strategy is a complete plan of action that assigns an action for the player to do in
every circumstance of the game.
In a Bayesian (incomplete information) game, a strategy for player is a function which
for every type in the type space assigns an action for the player to do from the action
space. In static games of complete information, there is no uncertainty about the type of
opponents the player faces, strategies are mere actions.

Group III
1) (6 pts.)

a. (2 pts.)

Normal-form representation of the game

Payoff matrix

Player A

Player B
Rock

Paper

Scissors

Rock

(0, 0)

(-1, 1)

(1, -1)

Paper

(1, -1)

(0, 0)

(-1, 1)

Scissors

(-1, 1)

(1, -1)

(0, 0)

This game has no Nash equilibriums in pure strategies.

By Nashs theorem we know that the game must have a Nash equilibrium, because it is
a static game of complete information with a finite number of players and strategies.
The Nash equilibrium must therefore exist in mixed strategies.

b. (1.5 pts.)
Extensive form representation of the game:
Rock

Paper

0, 0

-1, 1

Scissors

1, -1

Rock
Rock
A

Paper

Paper

1, -1

0, 0

Scissors

-1, 1
Scissors

Rock

Paper

-1, 1

1, -1

Scissors

0, 0

Solving the game by backwards induction:

Rock

Paper

0, 0

-1, 1

Scissors

1, -1

Rock
Rock
A

Paper

Paper

1, -1

0, 0

Scissors

-1, 1
Scissors

Rock

Paper

-1, 1

1, -1

Scissors

0, 0
Player B will always be able to choose an action that lets him win the game if he gets
the chance to observe player As choice before deciding. Player A has no first-mover
advantage because no matter what he chooses he will always lose. He will be indifferent
between every strategy.
c. (1.5 pts.)

Start by determining player Bs best-response function.

If player B has a high (2) payoff in case of victory the payoff matrix is:
Player B
Payoff matrix
Rock
Paper
Player A

Scissors

Rock

(0, 0)

(-1, 2)

(1, -1)

Paper

(1, -1)

(0, 0)

(-1, 2)

Scissors

(-1, 2)

(1, -1)

(0, 0)

If player B has a low (1) payoff in case of victory the payoff matrix is:
Player B
Payoff matrix
Rock
Paper
Player A

Scissors

Rock

(0, 0)

(-1, 1)

(1, -1)

Paper

(1, -1)

(-1, 1)

Scissors

(-1, 1)

(0, 0)
(1, -1)

(0, 0)

Notice that player Bs best response function does not differ in both cases, therefore it is
unaffected by the uncertainty described:
a A = Rock
Paper if

a B = Scissors if
a A = Paper , no matter the type of player B
Rock
if a A = Scissors

Player As best response function is symmetric:

a B = Rock
Paper if

a A = Scissors if
a B = Paper
Rock
if a B = Scissors

There are no intersections in pure strategies between the best-response functions,

therefore there are no Bayesian Nash equilibriums in pure strategies in this game.
2) (5 pts.)
a. (2.5 pts.)

Best-response functions:
Firm 1 Profit function:
1 = P (Q )q1 C1 (q1 ) = (30 Q ) q1 10q1 = (30 q1 q2 q3 ) q1 10q1 =
= 30q1 q12 q1q2 q1q3 10q1 = (20 q2 q3 )q1 q12
max 1
q1

(20 q2 q3 )q1 q12 = 0 (20 q2 q3 ) 2q1 = 0 q1 = 10 q2 + q3

=0
q1
q1
2

Firm 2 Profit function:

2 = P (Q )q2 C 2 (q2 ) = (30 Q ) q2 10q2 = (30 q1 q2 q3 ) q2 10q2 =
= 30q2 q1q2 q22 q2 q3 10q2 = (20 q1 q3 )q2 q22
max 2
q2

(20 q1 q3 )q2 q22 = 0 (20 q1 q3 ) 2q2 = 0 q2 = 10 q1 + q3

=0
q2
q2
2

Firm 3 Profit function:

3 = P (Q )q3 C3 (q# ) = (30 Q ) q3 10q3 = (30 q1 q2 q3 ) q3 10q3 =
= 30q3 q1q3 q2 q3 q32 10q3 = (20 q1 q2 )q3 q32

max 3
q3

(20 q1 q2 )q3 q32 = 0 (20 q1 q2 ) 2q3 = 0 q3 = 10 q1 + q2

=0
q3
q3
2

Nash equilibrium:
Intersect the best-response functions:
q2 + q3

q + q3

q1 = 10 2
20 2 q2 q3 = 10 2

q1 = f (q 2 , q3 )

q1 + q3

2 q2 = 20 q1 q3 q1 = 20 2 q 2 q3

q 2 = f (q1 , q3 ) q 2 = 10
2
q = f (q , q )

1
2
3
q1 + q 2

q
=
10

3
2

20 3q 2 = q3
40 4 q 2 2 q3 = 20 q 2 q 3

q1 = 20 2 q 2 (20 3q 2 ) q1 = q 2 q1 = q 2

q + q2

20 3q 2 = 10 1

20 3 5 = q 3
q3 = 5

q1 = q 2

q1 = 5
q1 = 5

10 = 2 q
q = 5
q = 5
q + q2
2

2
2
20 3q 2 = 10 2

b. (2.5 pts.)

Start by determining firm 4s best-response function.

If firm 4 has a low cost:
4 = P(Q )q 4 C 4L (q 4 ) = (30 Q ) q 4 5q 4 = (30 q1 q 4 ) q 4 5q 4 =

= 25q 4 q1 q 4 q 42 = (25 q1 )q 4 q 42
25 q1
4

(25 q1 )q 4 q 42 = 0 (25 q1 ) 2q 4 = 0 q 4 =
max 4
=0
q4
q 4
q 4
2

If firm 4 has a medium cost:

4 = P(Q )q 4 C 4L (q 4 ) = (30 Q ) q 4 10q 4 = (30 q1 q 4 ) q 4 10q 4 =

= 20q 4 q1 q 4 q 42 = (20 q1 )q 4 q 42
4
q

(
max 4
=0
20 q1 )q 4 q 42 = 0 (20 q1 ) 2q 4 = 0 q 4 = 10 1
q4
q 4
q 4
2

If firm 4 has a high cost:

 4 = P(Q )q 4 C 4L (q 4 ) = (30 Q ) q 4 15q 4 = (30 q1 q 4 ) q 4 15q 4 =

= 15q 4 q1 q 4 q 42 = (15 q1 )q 4 q 42
15 q1
4

(
max 4
=0
15 q1 )q 4 q 42 = 0 (15 q1 ) 2q 4 = 0 q 4 =
q4
q 4
q 4
2

Compute firm 1s best-response function taking uncertainty into account.

1
1
1
1 = P (Q L )q1 C1 (q1 ) + P (Q M )q1 C1 (q1 ) + P (Q H )q1 C1 (q1 ) =
3
3
3
1
1
= (30 q1 q 4* (c L )) q1 10q1 + (30 q1 q4* (c M )) q1 10q1 +
3
3
1
+ (30 q1 q4* (c H )) q1 10q1 =
3
1
= 20q1 q12 q 4* c L + q 4* c M + q 4* c H q1
3

[ ( )

( )

( )]

1

1

L
M
H
2
*
*
*
=0
20q1 q1 q4 (c ) + q4 (c ) + q4 (c ) q1 = 0
q1
q1
3

1
1
20 2 q1 q4* (c L ) + q4* (c M ) + q4* (c H ) = 0 q1 = 10 q4* (c L ) + q4* (c M ) + q4* (c H )
3
6

max 1
q1

Intersect the 4 best-response functions:

*
q 1

*
q 4

q *
4

q *
4

1
= 10 q 4* c L + q 4* c M + q 4* c H
6
*
1 60 3q1*
*
q1*
*
q
=
10

q
=
10

5
+
25

1
1
6
2
cL =
4

20 q1*
M

c =

15 q1*
H
c =
2
* 20
q 1 = 3
* 20

20
q1 = 3

25

3 *

* L
q1* = 6, (6 )
3
4 q1 = 5
q * c L = 55
q 4 c =
* L
2

q c = 9,1(6 )
6

4* M
20
20

* M
q * c M = 40
q 4 c = 6, (6 )
3
q 4 c =

4
q * c H = 4,1(6 )
6
4

25
*
H

q 4 c =
20
15
6

3
q 4* c H =

[ ( )

( )

( )]

( )

( )
( )

( )

( )

( )

( )

( )
( )
( )

( )

( )

Bayesian Nash equilibrium:

(q

*
1

( )

( )

( )

= 6, (6 ); q 4* c L = 9,1(6 ); q 4* c M = 6, (6 ); q 4* c H = 4,1(6 )