Scribd
Upload
33 views

Lecture4 With Notes

This document provides an overview of game theory concepts including beliefs, best responses, and rationalizability. It defines a player's belief as their assessment of what strategies other players will choose. A best response is the strategy that maximizes a player's payoff given their belief about others. Rationalizability refers to the set of strategies that survive iterative elimination of strategies that are never best responses, based on the assumption of common knowledge of rationality among players. An example of applying rationalizability to the Guessing 2/3 game is provided.

Uploaded by

Manan Jindal
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
33 views

Lecture4 With Notes

This document provides an overview of game theory concepts including beliefs, best responses, and rationalizability. It defines a player's belief as their assessment of what strategies other players will choose. A best response is the strategy that maximizes a player's payoff given their belief about others. Rationalizability refers to the set of strategies that survive iterative elimination of strategies that are never best responses, based on the assumption of common knowledge of rationality among players. An example of applying rationalizability to the Guessing 2/3 game is provided.

Uploaded by

Manan Jindal
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 27

Economics 409:

Game Theory

Beliefs, Best Response, and Rationalizability


Example of IESDS: Soda Vendors at the Beach

I Two soda vendors must simultaneously choose their locations


on the beach
I Five customers are distributed evenly across five locations on
the beach
I Customers patronize the vendor nearest them
I Each patron yields profit of $1 per day
Which locations would the soda vendors choose?
IESDS: Soda Vendors at the Beach

1 2 3 4 5
1 2.5, 2.5 1, 4 1.5, 3.5 2, 3 2.5, 2.5
2
3
4
5
IESDS: Soda Vendors at the Beach

1 2 3 4 5
1 2.5, 2.5 1, 4 1.5, 3.5 2, 3 2.5, 2.5
2 4, 1 2.5, 2.5 2, 3 2.5, 2.5 3, 2
3 3.5, 1.5 3, 2 2.5, 2.5 3, 2 3.5, 1.5
4 3, 2 2.5, 2.5 2, 3 2.5, 2.5 4, 1
5 2.5, 2.5 2, 3 1.5, 3.5 1, 4 2.5, 2.5
The Election Game

I Voters are located on a left-to-right ideological scale


I Two political candidates; each chooses a policy promise from
left to right
I Each voter votes for the candidate closest to her ideological
preference; majority rules

Median Voter Theorem: If the candidates care only about vote


shares, the unique IESDS outcome is for both to cater to the
policy preference of the median voter.
Overview

I A formal definition of rationality based on beliefs and best


responses

I Rationalizability: a consequence of common knowledge of


rationality
Belief
We can ask a player
“what were you thinking when you made your choice?”
For example, in a PD (prisoners’ dilemma), a player may say
“I think the other player will play D”
Belief
We can ask a player
“what were you thinking when you made your choice?”
For example, in a PD (prisoners’ dilemma), a player may say
“I think the other player will play D”

I Formally, a belief of player i is a possible profile of her


opponents’ strategies, s−i ∈ S−i .
Belief
We can ask a player
“what were you thinking when you made your choice?”
For example, in a PD (prisoners’ dilemma), a player may say
“I think the other player will play D”

I Formally, a belief of player i is a possible profile of her


opponents’ strategies, s−i ∈ S−i .

I Here the belief is “deterministic” (we will generalize this later


to allow probabilistic beliefs)
I # of player i’s possible beliefs = # of elements in S−i

I Note: A player’s belief may not be the same as what her


opponents will actually choose
Best Response: Definition

I The strategy si ∈ Si is player i’s best response to his belief


about her opponents’ play s−i ∈ S−i if

vi (si , s−i ) ≥ vi (si0 , s−i ), for all si0 ∈ Si .

I We will often say si is a best response to (her belief) s−i


Best Response: Definition

I The strategy si ∈ Si is player i’s best response to his belief


about her opponents’ play s−i ∈ S−i if

vi (si , s−i ) ≥ vi (si0 , s−i ), for all si0 ∈ Si .

I We will often say si is a best response to (her belief) s−i


I There can be more than one best responses to some belief
L C R
U 4, 3 5, 1 6, 2
M 2, 1 8, 4 3, 6
D 4, 0 9, 6 2, 8
Best Response: Properties

I A rational player with belief s−i will always choose a best


response to s−i
I rationality = “maximize payoff given belief”

I A strictly dominated strategy cannot be a best response to


any belief
I a weakly (but not strictly) dominated strategy can be a best
response to some belief

I If a game has a dominant strategy equilibrium s ∗ = (si∗ , s−i


∗ ),

then si∗ is a best response to s−i


∗
Best Response Correspondence

I More systematically, for each player, we can keep track of her


best responses to her all possible beliefs

I The best response correspondence of player i selects for


each s−i ∈ S−i a subset BRi (s−i ) ⊂ Si where every
si ∈ BRi (s−i ) is a best response to s−i
Best Response Correspondence

I More systematically, for each player, we can keep track of her


best responses to her all possible beliefs

I The best response correspondence of player i selects for


each s−i ∈ S−i a subset BRi (s−i ) ⊂ Si where every
si ∈ BRi (s−i ) is a best response to s−i

I The best response correspondence BRi (·) is a function if it is


always a singleton set for every s−i

I “Player i is rational” = BRi (·)


Practice: I

L C R
U 3, 3 5, 1 6, 2
M 4, 1 8, 4 3, 6
D 4, 0 9, 6 6, 8
What is player 1’s best response correspondence?

What is player 2’s best response correspondence?


Practice: II

L C R
U 5, 1 0, 4 1, 0
M 3, 1 0, 0 3, 5
D 3, 3 4, 4 2, 5
Rationalizability

I Rationality of player i: BRi (·)


Rationalizability

I Rationality of player i: BRi (·)


I A strategy si ∈ Si is never a best response if there is no
belief s−i ∈ S−i such that si ∈ BRi (s−i )
Rationalizability

I Rationality of player i: BRi (·)


I A strategy si ∈ Si is never a best response if there is no
belief s−i ∈ S−i such that si ∈ BRi (s−i )
I If player j knows that player i is rational, then j knows that i
will not play strategies that are never a best response
Rationalizability

I Rationality of player i: BRi (·)


I A strategy si ∈ Si is never a best response if there is no
belief s−i ∈ S−i such that si ∈ BRi (s−i )
I If player j knows that player i is rational, then j knows that i
will not play strategies that are never a best response
I Like IESDS, when we assume common knowledge of
rationality, we can build an iterative process
I in each step, eliminate strategies that are never a best response
I stop when we cannot further eliminate any strategies
I The set of strategy profiles that survive this process is called
the set of rationalizable strategy profiles
Application: The Guessing 2/3 Game

I Ai = {0, 1, . . . , 20}
Application: The Guessing 2/3 Game

I Ai = {0, 1, . . . , 20}
I IESDS is not very useful in this game
I for example, 20 is not strictly dominated by 19!
Application: The Guessing 2/3 Game

I Ai = {0, 1, . . . , 20}
I IESDS is not very useful in this game
I for example, 20 is not strictly dominated by 19!
I Let’s consider best response and rationalizability
I 20 is never a best response
Application: The Guessing 2/3 Game

I Ai = {0, 1, . . . , 20}
I IESDS is not very useful in this game
I for example, 20 is not strictly dominated by 19!
I Let’s consider best response and rationalizability
I 20 is never a best response
I we can eliminate 20 based on rationalizability
Application: The Guessing 2/3 Game

I Ai = {0, 1, . . . , 20}
I IESDS is not very useful in this game
I for example, 20 is not strictly dominated by 19!
I Let’s consider best response and rationalizability
I 20 is never a best response
I we can eliminate 20 based on rationalizability
I we are left with {0, 1, . . . , 19} for each player
I now 19 is never a best response
Application: The Guessing 2/3 Game

I Ai = {0, 1, . . . , 20}
I IESDS is not very useful in this game
I for example, 20 is not strictly dominated by 19!
I Let’s consider best response and rationalizability
I 20 is never a best response
I we can eliminate 20 based on rationalizability
I we are left with {0, 1, . . . , 19} for each player
I now 19 is never a best response
I iterating this process, we end up with {0, 1}
I The set of rationalizable strategies for player i is {0, 1}
Summary

I Rationality can be formally described as playing a best


response to some belief

I The implication of common knowledge of rationality is


rationalizability

You might also like