Você está na página 1de 21

Repeated Games

Economics 302 - Microeconomic Theory II: Strategic Behavior

Instructor: Songzi Du
compiled by Shih En Lu
Simon Fraser University

February 26, 2015

ECON 302 (SFU)

Lecture 7

February 26, 2015

1 / 18

Introduction to Repeated Games

One class of much studied sequential games is repeated games.


These are games formed by playing a stage game over and over
again.
Applications: competition in an oligopoly, customer relations,
cleaning a shared apartment, etc.
We will look at cases where this stage game has simultaneous moves,
and where actions are perfectly observed after each stage.
We will focus on SPEs.

ECON 302 (SFU)

Lecture 7

February 26, 2015

2 / 18

Discount Factors
Recall that if a player has discount factor and a payoff stream
u0 , u1 , u2 , ... in periods 0, 1, 2, ..., then the value of that payoff stream
from period 0s point of view is the time-discounted sum:

t ut = u0 + u1 + 2 u2 + ...

t=0

We usually assume [0, 1].


People are impatient.
You have a small chance of dying each day.
Firms (whose payoffs are usually assumed to be their profits) can
invest $1 today to get more than $1 (on average) tomorrow.
Aside: there is evidence that on top of this exponential discounting,
people place a premium on present payoffs. Search for hyperbolic
discounting and behavioral economics for more information.
ECON 302 (SFU)

Lecture 7

February 26, 2015

3 / 18

If Stage Game Has One NE

Recall the prisoners dilemma:


C
D
C -2, -2 -5, -1
D -1, -5 -3, -3
Unique NE is (D, D).
Suppose the game is played twice, and players have discount factor
= 1. Lets draw the game tree.
What are the subgames? What are the strategies? Which one is the
tit-for-tat strategy?
In SPE, what must happen in the last stage?
So what will players do in the first stage?

ECON 302 (SFU)

Lecture 7

February 26, 2015

4 / 18

Twice-repeated prisoners dilemma ( = 1)

ECON 302 (SFU)

Lecture 7

February 26, 2015

5 / 18

Twice-repeated prisoners dilemma ( = 1)


Both Player 1 and Player 2 play (D2, D3, D4, D5) in stage 2:

ECON 302 (SFU)

Lecture 7

February 26, 2015

6 / 18

Twice-repeated prisoners dilemma ( = 1)

Both Player 1 and Player 2 play (D2, D3, D4, D5) in stage 2:

(C1, D2, D3, D4, D5)


(D1, D2, D3, D4, D5)

ECON 302 (SFU)

(C1, D2, D3, D4, D5)


(5, 5)
(4, 8)

Lecture 7

(D1, D2, D3, D4, D5)


(8, 4)
(6, 6)

February 26, 2015

7 / 18

Twice-repeated prisoners dilemma ( = 1)

Both Player 1 and Player 2 play (D2, D3, D4, D5) in stage 2:

(C1, D2, D3, D4, D5)


(D1, D2, D3, D4, D5)

(C1, D2, D3, D4, D5)


(5, 5)
(4, 8)

(D1, D2, D3, D4, D5)


(8, 4)
(6, 6)

Both players playing (D1, D2, D3, D4, D5) is the only subgame
perfect equilibrium.

ECON 302 (SFU)

Lecture 7

February 26, 2015

7 / 18

If Stage Game Has One NE (II)


This reasoning applies whenever a stage game has just one NE, and
whenever the number of stages is finite.
In SPE, in the last stage, the unique stage-game NE must be played
no matter what happened in earlier stages.
But given that, what players do in the second-to-last stage does not
impact what happens in the last stage.
Therefore, the stage games NE must be played in the second-to-last
stage: no player has a reason not to play a best response.
Can keep going with this reasoning.
Conclusion: When a stage game with a unique NE is repeated
finitely many times, the unique SPE of the repeated game is to
play the stage games NE at each stage.

ECON 302 (SFU)

Lecture 7

February 26, 2015

8 / 18

If Stage Game Has Multiple NEs

Things get more interesting!


Its still true that an NE must be played in the last stage.
But now, which NE is played can depend on what has happened
before.
Sometimes, this can sustain a non-NE outcome in stages before the
last one!
(Note that this is allowed in SPE: any stage game but the last one,
taken by itself, is NOT a subgame.)

ECON 302 (SFU)

Lecture 7

February 26, 2015

9 / 18

Twice-repeated Polite Rude Game

Suppose that the following stage game is played twice, and the
discount factor is 1:
Polite
Rude

Nice
5,0
6,-10

Mean
0,0
1,-10

In the stage game, Rude is player 1s dominant strategy.

ECON 302 (SFU)

Lecture 7

February 26, 2015

10 / 18

Twice-repeated Polite Rude Game

ECON 302 (SFU)

Lecture 7

February 26, 2015

11 / 18

Easy SPEs in Twice-repeated Polite Rude Game


Player 1 plays (R2, R3, R4, R5) in stage 2, Player 2 plays (N2, N3, N4,
N5) in stage 2. ( = 1)

ECON 302 (SFU)

Lecture 7

February 26, 2015

12 / 18

Easy SPEs in Twice-repeated Polite Rude Game


Player 1 plays (R2, R3, R4, R5) in stage 2, Player 2 plays (M2, M3, M4,
M5) in stage 2. ( = 1)

ECON 302 (SFU)

Lecture 7

February 26, 2015

13 / 18

Easy SPEs in Twice-repeated Polite Rude Game


Player 1 plays (R2, R3, R4, R5) in stage 2, Player 2 plays (M2, M3, M4,
M5) in stage 2. ( = 1)

It is a SPE to play an NE in the first stage, followed by a fixed


NE in the second stage.
ECON 302 (SFU)

Lecture 7

February 26, 2015

13 / 18

A More Efficient SPE in Twice-repeated Polite Rude Game

A more efficient SPE:


Player 1: In stage 1, Polite; in stage 2, Rude
Player 2: In stage 1, Nice; in stage 2, Nice if player 1 was polite in stage 1, Mean
if player 1 was rude in stage 1

What if the discount factor is 0.1?

ECON 302 (SFU)

Lecture 7

February 26, 2015

14 / 18

A More Efficient SPE in Twice-repeated Polite Rude Game


Player 1 plays (R2, R3, R4, R5) in stage 2, Player 2 plays (N2, N3, M4,
M5) in stage 2. ( = 1)

ECON 302 (SFU)

Lecture 7

February 26, 2015

15 / 18

A More Efficient SPE in Twice-repeated Polite Rude Game


Player 1 plays (R2, R3, R4, R5) in stage 2, Player 2 plays (N2, N3, M4,
M5) in stage 2.

(P1, R2, R3, R4, R5)


(R1, R2, R3, R4, R5)

ECON 302 (SFU)

(N1, N2, N3, M4, M5)


(5, 0) + (6, 10)
(6, 10) + (1, 10)

Lecture 7

(M1, N2, N3, M4, M5)


(0, 0) + (6, 10)
(1, 10) + (1, 10)

February 26, 2015

16 / 18

A More Efficient SPE in Twice-repeated Polite Rude Game


Player 1 plays (R2, R3, R4, R5) in stage 2, Player 2 plays (N2, N3, M4,
M5) in stage 2.

(P1, R2, R3, R4, R5)


(R1, R2, R3, R4, R5)

(N1, N2, N3, M4, M5)


(5, 0) + (6, 10)
(6, 10) + (1, 10)

(M1, N2, N3, M4, M5)


(0, 0) + (6, 10)
(1, 10) + (1, 10)

(P1, R2, R3, R4, R5) and (N1, N2, N3, M4, M5) is a subgame
perfect equilibrium when = 1.
(P1, R2, R3, R4, R5) and (N1, N2, N3, M4, M5) is not a subgame
perfect equilibrium when = 0.1.

ECON 302 (SFU)

Lecture 7

February 26, 2015

16 / 18

If Stage Game Has Multiple NEs

Playing an NE in each stage, without conditioning on previous play, is


still subgame-perfect.
There are also SPE where in some stages, the stage games NE
is not played.
The stage-game best response may not be the overall best response,
because it may lead to lower payoffs in the future.
But this can only happen when players care sufficiently about future
payoffs.

ECON 302 (SFU)

Lecture 7

February 26, 2015

17 / 18

Recap of Finitely Repeated Games


In a SPE of a repeated game, does a stage game NE have to be
played every period?
No: sometimes, it is possible for players to incentivize each other to
play something else (usually an efficient outcome) by promising
higher payoffs in later periods.
Example: if a NE of the stage game Pareto dominates another one,
then players might play the good NE in a later period if they did
what they were supposed to do, and the bad NE otherwise.
This is more likely to work when players are more patient.
However, a stage game NE must be played in the last period.
Takeaway message: repetition promotes efficiency/cooperation.

ECON 302 (SFU)

Lecture 7

February 26, 2015

18 / 18

Você também pode gostar