Scribd Logo
Upload

Welcome to Scribd!

  • 22 views

    Lecture 8

    Uploaded by

    Jessica Chen
    This document summarizes key concepts from a game theory lecture on voting models and location games. For the votes-candidate model, it notes there can be multiple equilibria but candidates won't be too far apart, like outside 1/6 to 5/6, or a center candidate could enter and win. The location game shows how segregation can emerge from individual preferences for diversity, and the integrated equilibrium is not stable while segregated ones are. Randomization, like busing policies, could help maintain an integrated equilibrium.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Lecture 8

    Uploaded by

    Jessica Chen
    22 views18 pages
    This document summarizes key concepts from a game theory lecture on voting models and location games. For the votes-candidate model, it notes there can be multiple equilibria but candidates won't be too far apart, like outside 1/6 to 5/6, or a center candidate could enter and win. The location game shows how segregation can emerge from individual preferences for diversity, and the integrated equilibrium is not stable while segregated ones are. Randomization, like busing policies, could help maintain an integrated equilibrium.

    Original Title

    lecture 8

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document summarizes key concepts from a game theory lecture on voting models and location games. For the votes-candidate model, it notes there can be multiple equilibria but candidates won't be too far apart, like outside 1/6 to 5/6, or a center candidate could enter and win. The location game shows how segregation can emerge from individual preferences for diversity, and the integrated equilibrium is not stable while segregated ones are. Randomization, like busing policies, could help maintain an integrated equilibrium.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    22 views18 pages

    Lecture 8

    Uploaded by

    Jessica Chen
    This document summarizes key concepts from a game theory lecture on voting models and location games. For the votes-candidate model, it notes there can be multiple equilibria but candidates won't be too far apart, like outside 1/6 to 5/6, or a center candidate could enter and win. The location game shows how segregation can emerge from individual preferences for diversity, and the integrated equilibrium is not stable while segregated ones are. Randomization, like busing policies, could help maintain an integrated equilibrium.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 18

    Game Theory

    Lecture 8

    Leong Kaiwen

    Assistant Professor in Economics


    Nanyang Technological University

    2021

    Leong Kaiwen (NTU) Game Theory NTU 2021 1 / 16


    Votes-Candidate Model

    Many NE, not all ”at the center” (candidates)


    Entry can lead to a more distant candidate winning

    This is not an equilibrium, why?


    Anyone near the center can stand up & run and win.
    Lesson: If the candidates are too far apart we are going to see some
    center entry which is going to win.

    Leong Kaiwen (NTU) Game Theory NTU 2021 2 / 16


    Votes-Candidate Model

    In equilibrium, just how far apart can the candidate be?

    Claim: If candidates (2 of then) are not outside 1/6 & 5/6, it will be
    an equilibrium.

    Leong Kaiwen (NTU) Game Theory NTU 2021 3 / 16


    Votes-Candidate Model
    Both candidates are vulnerable to someone entering at the center.
    Someone will enter at the center if they think they can win.
    1
    L& R candidates will get slightly more than of the votes. So center
    3
    candidates will not win.

    Leong Kaiwen (NTU) Game Theory NTU 2021 4 / 16


    Votes-Candidate Model

    1
    Lesson: If the 2 candidates are too extreme (in this example < & >
    6
    5
    ), someone will enter.
    6
    Lesson: Guess and check effective

    Leong Kaiwen (NTU) Game Theory NTU 2021 5 / 16


    Location Model
    Two towns: East (E) and West (W). Each town can hold 100,000
    people.
    Two types of people (players): Tall (T) and Short (S), 100,000 of each
    type of people.
    Strategy: People will choose where they live (E or W).

    Interpretation: People like to live


    in mixed towns. But if the town
    is not evenly mixed, they prefer
    to be in the majority.
    Rules:
    Simultaneous choice
    If there is no room, then ran-
    domize to ration.

    Leong Kaiwen (NTU) Game Theory NTU 2021 6 / 16


    Location Model

    Leong Kaiwen (NTU) Game Theory NTU 2021 7 / 16


    Location Model

    After 2 or 3 repeated plays, row 1-7 choose E & 8-14 choose W. We


    ended up with segregation.
    What is the equilibria in this game? (guess and check)

    Leong Kaiwen (NTU) Game Theory NTU 2021 8 / 16


    Location Model

    After 2 or 3 repeated plays, row 1-7 choose E & 8-14 choose W. We


    ended up with segregation.
    What is the equilibria in this game? (guess and check)
    Guess
    Two segregated NE: (T in E, S in W) and (S in E, T in W)
    1
    Integrated NE: ( of each in both towns)
    2

    Leong Kaiwen (NTU) Game Theory NTU 2021 8 / 16


    Location Model

    Check
    If all the short people are in E and all the tall people are in W, what’s a
    deviation?
    No one can deviate profitably, what we have are NE.
    1
    Another equilibrium: ( of each in both towns).
    2
    What is wrong about this equilibrium?

    Leong Kaiwen (NTU) Game Theory NTU 2021 9 / 16


    Location Model

    Check
    If all the short people are in E and all the tall people are in W, what’s a
    deviation?
    No one can deviate profitably, what we have are NE.
    1
    Another equilibrium: ( of each in both towns).
    2
    What is wrong about this equilibrium?
    It is a ”weak” equilibrium because there is no strict incentive to deviate,
    there is no strict incentive not to deviate as well.
    At the integrated equilibrium, I am exactly indifference about where I
    live, both towns look the same to me.
    Whereas at the segregated equilibria, I strictly prefer to go to the town
    in which I am the majority.

    Leong Kaiwen (NTU) Game Theory NTU 2021 9 / 16


    Location Model

    Why is the integrated equilibrium not stable?


    At the integrated equilibria, if we move away from it a little bit, let’s say
    one town has 5% more short people and the other town has 5% more
    tall people, then we are in trouble. All the short people will prefer the
    town with more short people and all the tall people are going to prefer
    the town with more tall people. We are back in segregation again.
    Conversely, for segregated equilibria, they are very stable. If we start
    close to 100% short people in East ans close to 100% tall in West, and
    we move some people in W to E and play the game, those I moved
    from W to E will go back to W and we will have the same segregated
    equilibria.

    Leong Kaiwen (NTU) Game Theory NTU 2021 10 / 16


    Location Model

    Summary
    Two segregated NE: (T in E, S in W) and (S in E, T in W). Integrated
    1
    NE: ( of each in both towns)
    2
    Segregated equilibria: ”stable” and you strictly prefer not to deviate.
    Integrated equilibria: ”not stable” and ”weak equilibria”
    1 1
    Key: There are 2 strict equilibria and if we got beyond the vs mix
    2 2
    slightly one way, we will end up all the way with a segregated equilibrium.
    1 1
    This game has a tipping point. you can push people from the vs
    2 2
    equilibrium in each town, people will keep coming back. But if you
    1
    push them too far beyond , they will go off to the other equilibrium
    2
    (segregated). This was found by Schelling who won the Nobel prize for
    this.

    Leong Kaiwen (NTU) Game Theory NTU 2021 11 / 16


    Location Model

    Notice: This is not a prisoners dilemma.


    There is actually another equilibrium.
    1 1
    Suppose everyone chooses E. the players will be in E and in W.
    2 2
    Check for deviations
    If everyone is choosing E and allowing the randomisation device to place
    you, then you have no incentive to deviate.
    Lesson:
    The seemingly irrelevant details can matter
    Having society randomize for you, ended up better than ”active” choice.

    Leong Kaiwen (NTU) Game Theory NTU 2021 12 / 16


    Location Model

    More Lessons
    ”Sociology” , seeing segregation ⇏ preference for segregation
    ”Policy”, randomisation, bussing
    Randomization: government and individual randomisation.

    Leong Kaiwen (NTU) Game Theory NTU 2021 13 / 16


    Randomized Mixed Strategy

    It is a randomisation over pure strategies. the strategies we have dealt


    with in course up to now are pure strategies.
    We are going to expand the actual available choices to include all
    randomisation over pure strategies.
    Example (Rock, paper, scissors)
    Player 2
    R S P
    R (0,0) (1,-1) (-1,1)
    Player 1
    S (-1,1) (0,0) (1,-1)
    P (1,-1) (-1,1) (0,0)
    Claim: No NE in pure strategies, pure strategy={R,P,S}
    R is BR to S, S is BR to P, P is BR to R. Any attempt to look for BR
    that are BR to each other will lead to cycle. So no pure strategy NE.

    Leong Kaiwen (NTU) Game Theory NTU 2021 14 / 16


    Randomized Mixed Strategy

    1 1 1
    Claim: NE: each player ( , , )
    3 3 3
    1 1 1 1 1 1
    Expected payoff R vs ( , , )= (0)+ (1)+ (-1)=0
    3 3 3 3 3 3
    1 1 1 1 1 1
    Expected payoff S vs ( , , )= (-1)+ (0)+ (1)=0
    3 3 3 3 3 3
    1 1 1 1 1 1
    Expected payoff P vs ( , , )= (1)+ (-1)+ (0)=0
    3 3 3 3 3 3
    All expected payoffs are 0.

    Leong Kaiwen (NTU) Game Theory NTU 2021 15 / 16


    Randomized Mixed Strategy

    1 1 1 1 1 1
    Expected payoff of ( , , ) vs ( , , )
    3 3 3 3 3 3
    1 1 1 1
    = [Expected payoff of R vs ( , , )]
    3 3 3 3
    1 1 1 1
    + [Expected payoff of S vs ( , , )]
    3 3 3 3
    1 1 1 1
    + [Expected payoff of P vs ( , , )]
    3 3 3 3
    =0
    1 1 1
    I will get the same payoff if I played ( , , ) or any pure strategy
    3 3 3
    1 1 1 1 1 1
    against ( , , ). Hence, ( , , ) is a best response weakly against
    3 3 3 3 3 3
    1 1 1 1 1 1 1 1 1
    ( , , ). So [( , , ),( , , )] is a NE.
    3 3 3 3 3 3 3 3 3

    Leong Kaiwen (NTU) Game Theory NTU 2021 16 / 16

    You might also like