Game Theory Correlated equilibrium 1

Similar documents
Game Theory Lecture 7 Refinements: Perfect equilibrium

Game Theory Lecture 10+11: Knowledge

Game Theory. Monika Köppl-Turyna. Winter 2017/2018. Institute for Analytical Economics Vienna University of Economics and Business

Econ 618: Correlated Equilibrium

Game Theory and Algorithms Lecture 2: Nash Equilibria and Examples

Game Theory. Wolfgang Frimmel. Perfect Bayesian Equilibrium

SF2972 Game Theory Exam with Solutions March 15, 2013

On revealed preferences in oligopoly games

Static Models of Oligopoly

Chapter 9. Mixed Extensions. 9.1 Mixed strategies

Advanced Microeconomics

Supermodular Games. Ichiro Obara. February 6, 2012 UCLA. Obara (UCLA) Supermodular Games February 6, / 21

Basics of Game Theory

6.254 : Game Theory with Engineering Applications Lecture 8: Supermodular and Potential Games

Oligopoly Theory 2 Bertrand Market Games

Theory of Auctions. Carlos Hurtado. Jun 23th, Department of Economics University of Illinois at Urbana-Champaign

Calculus in Business. By Frederic A. Palmliden December 7, 1999

Government 2005: Formal Political Theory I

Game Theory. 2.1 Zero Sum Games (Part 2) George Mason University, Spring 2018

Introduction to Game Theory

Game theory lecture 4. September 24, 2012

Virtual Robust Implementation and Strategic Revealed Preference

Solution to Tutorial 9

4. Partial Equilibrium under Imperfect Competition

EC3224 Autumn Lecture #04 Mixed-Strategy Equilibrium

Outline for today. Stat155 Game Theory Lecture 17: Correlated equilibria and the price of anarchy. Correlated equilibrium. A driving example.

Lecture Notes on Game Theory

Moral Hazard in Teams

Game Theory and Algorithms Lecture 7: PPAD and Fixed-Point Theorems

Economics 3012 Strategic Behavior Andy McLennan October 20, 2006

On the relation between Sion s minimax theorem and existence of Nash equilibrium in asymmetric multi-players zero-sum game with only one alien

Game Theory: Spring 2017

EC319 Economic Theory and Its Applications, Part II: Lecture 7

Introduction to Game Theory

Wireless Network Pricing Chapter 6: Oligopoly Pricing

Zero-Sum Games Public Strategies Minimax Theorem and Nash Equilibria Appendix. Zero-Sum Games. Algorithmic Game Theory.

4. CONTINUOUS VARIABLES AND ECONOMIC APPLICATIONS

Cournot and Bertrand Competition in a Differentiated Duopoly with Endogenous Technology Adoption *

Introduction to Game Theory Lecture Note 2: Strategic-Form Games and Nash Equilibrium (2)

Mechanism Design. Christoph Schottmüller / 27

Some forgotten equilibria of the Bertrand duopoly!?

Economics 385: Suggested Solutions 2

Classic Oligopoly Models: Bertrand and Cournot

C31: Game Theory, Lecture 1

Bayesian Games and Mechanism Design Definition of Bayes Equilibrium

6.254 : Game Theory with Engineering Applications Lecture 7: Supermodular Games

Static Information Design

Correlated Equilibrium in Games with Incomplete Information

Industrial Organization II (ECO 2901) Winter Victor Aguirregabiria. Problem Set #1 Due of Friday, March 22, 2013

The WhatPower Function à An Introduction to Logarithms

Mathematical Economics - PhD in Economics

Informed Principal in Private-Value Environments

Algorithmic Game Theory and Applications. Lecture 4: 2-player zero-sum games, and the Minimax Theorem

Negotiation: Strategic Approach

MS&E 246: Lecture 4 Mixed strategies. Ramesh Johari January 18, 2007

NTU IO (I) : Auction Theory and Mechanism Design II Groves Mechanism and AGV Mechansim. u i (x, t i, θ i ) = V i (x, θ i ) + t i,

On the Unique D1 Equilibrium in the Stackelberg Model with Asymmetric Information Janssen, M.C.W.; Maasland, E.

Game Theory. Strategic Form Games with Incomplete Information. Levent Koçkesen. Koç University. Levent Koçkesen (Koç University) Bayesian Games 1 / 15

EC3224 Autumn Lecture #03 Applications of Nash Equilibrium

Entry under an Information-Gathering Monopoly Alex Barrachina* June Abstract

Robust Predictions in Games with Incomplete Information

Lecture December 2009 Fall 2009 Scribe: R. Ring In this lecture we will talk about

On Renegotiation-Proof Collusion under Imperfect Public Information*

Sion s minimax theorem and Nash equilibrium of symmetric multi-person zero-sum game

Mixed Strategies. Krzysztof R. Apt. CWI, Amsterdam, the Netherlands, University of Amsterdam. (so not Krzystof and definitely not Krystof)

Msc Micro I exam. Lecturer: Todd Kaplan.

A Primer on Strategic Games

Katz and Shapiro (1985)

Equilibrium Refinements

Correlated Equilibria: Rationality and Dynamics

MS&E 246: Lecture 12 Static games of incomplete information. Ramesh Johari

Lecture 6. Xavier Gabaix. March 11, 2004

Lecture Note II-3 Static Games of Incomplete Information. Games of incomplete information. Cournot Competition under Asymmetric Information (cont )

Dynamic Bertrand and Cournot Competition

ANSWER KEY 2 GAME THEORY, ECON 395

UC Berkeley Haas School of Business Game Theory (EMBA 296 & EWMBA 211) Summer 2016

Profitability of price and quantity strategies in a duopoly with vertical product differentiation

Introduction to Game Theory

University of Warwick, Department of Economics Spring Final Exam. Answer TWO questions. All questions carry equal weight. Time allowed 2 hours.

Research and Development

Volume 35, Issue 2. Cournot Equilibrium Uniqueness in Case of Concave Industry Revenue: a Simple Proof

Lecture Notes on Game Theory

Information Design. Dirk Bergemann and Stephen Morris. Johns Hopkins University April 2017

Overview. Producer Theory. Consumer Theory. Exchange

Lecture 6: April 25, 2006

In the Name of God. Sharif University of Technology. Microeconomics 1. Graduate School of Management and Economics. Dr. S.

Welfare consequence of asymmetric regulation in a mixed Bertrand duopoly

Static (or Simultaneous- Move) Games of Complete Information

Computation of Efficient Nash Equilibria for experimental economic games

On Hotelling s Stability in Competition

Ph.D. Preliminary Examination MICROECONOMIC THEORY Applied Economics Graduate Program June 2016

A technical appendix for multihoming and compatibility

Ü B U N G S A U F G A B E N. S p i e l t h e o r i e

Contents. Set Theory. Functions and its Applications CHAPTER 1 CHAPTER 2. Preface... (v)

The Impact of Advertising on Media Bias. Web Appendix

Bayesian Persuasion Online Appendix

Design Patent Damages under Sequential Innovation

GS/ECON 5010 Answers to Assignment 3 W2005

Bayes Correlated Equilibrium and Comparing Information Structures

Topics of Algorithmic Game Theory

Transcription:

Game Theory Correlated equilibrium 1 Christoph Schottmüller University of Copenhagen 1 License: CC Attribution ShareAlike 4.0 1 / 17

Correlated equilibrium I Example (correlated equilibrium 1) L R U 5,1 0,0 D 4,4 1,5 Table: correlated equilibrium Find the three Nash equilibria! 2 / 17

Correlated equilibrium II Example (correlated equilibrium 1) L R U 5,1 0,0 D 4,4 1,5 Table: correlated equilibrium pure strategy NE give high aggregate but very unequal payoff mixed strategy equilibrium gives equal but low payoff 3 / 17

Correlated equilibrium III Can players get equal and high payoffs? flip a coin: if tails (U, L), if head (D, R) with unfair coins any payoff in the convex hull of the NE payoffs is attainable can players do even better? Yes: randomization device with three equally likely states A,B and C P1 gets a message iff state is A P2 gets a message iff state is C Then the following is an equilibrium P1 plays U when he gets a message and D otherwise P2 plays R when he gets a message and L otherwise Check that no deviation is profitable! 4 / 17

Correlated equilibrium IV expected payoff 1/3(5, 1) + 1/3(4, 4) + 1/3(1, 5) = (3.33, 3.33) is outside of the convex hull of the Nash payoffs correlated equilibrium can lead to higher payoffs than NE interpretation correlated equilibrium: both players first communicate and construct a correlation machine together each player sees output of the machine before taking action impartial mediator gives (privately!) recommendations a i to each player according to some probability distribution Recommendations are self-enforcing 5 / 17

Correlated equilibrium V We now use the second interpretation: take a strategic form game G = N, (A i ), (u i ) a probability distribution p over A leads to the game G (p): 1 mediator draws an action profile a = (a 1,..., a n ) from A according to probability distribution p 2 mediator reveals a i to each player i (but does not reveal a i ) 3 each player chooses an action a i A i 4 payoff for each player i is u i (a 1,..., a n) pure strategy for player i in G (p) is function s i : A i A i (action as function of recommendation) belief of player i when getting recommendation a i : p(a i a i ) = p(a i, a i ) b i A i p(a i, b i ) 6 / 17

Correlated equilibrium VI Lemma (MSZ Thm 8.5) All players following the recommendation, i.e. s i (a i ) = a i for all players i, is an equilibrium of G (p) if and only if a i A i p(a i, a i )u i (a i, a i ) for all players i and all actions a i, a i A i. a i A i p(a i, a i )u i (a i, a i ) (1) 7 / 17

Correlated equilibrium VII Proof. Expected utility of player i when a i after receiving recommendation a i is EU i (a i ) = a i A i p(a i, a i) b i A i p(a i, b i ) u i(a i, a i ). Expected utility of playing a i after receiving recommendation a i is EU i (a i) = p(a i, a i) b i A i p(a i, b i ) u i(a i, a i ). a i A i EU i (a i ) EU i (a i) if and only if (1) holds. For this proof, we use the convention b i A i. p(a i,a i ) = 0 if p(a i, b i ) = 0 for all p(a i,b i ) b i A i 8 / 17

Correlated equilibrium VIII Definition (correlated equilibrium) A probability distribution p over A is a correlated equilibrium in the strategic form game G = N, (A i ), (u i ) if s i (a i ) = a i for all players i is an equilibrium of G (p). 9 / 17

Correlated equilibrium IX Proposition Let α be a mixed strategy equilibrium. Then the distribution p α defined by is a correlated equilibrium. Proof. p α (a 1,..., a n ) = Π n i=1α i (a i ) 10 / 17

Correlated equilibrium X Corollary A correlated equilibrium exists in all finite games. 11 / 17

Correlated equilibrium XI Example (correlated equilibrium 2) L R U 0,1,3 0,0,0 D 1,1,1 1,0,0 A L R U 2,2,2 0,0,0 D 2,2,0 2,2,2 B L R U 0,1,0 0,0,0 D 1,1,0 1,0,3 C Table: correlated equilibrium 2 12 / 17

Correlated equilibrium XII Example (correlated equilibrium 2 continued) unique NE is (D,L,A) (check!) getting the (2, 2, 2) payoff from (U, L, B) or from (D, R, B) would be nice for all players but P3 has an incentive to deviate (either to A or to C) this example: limiting your own information can be beneficial what could a mediator do to solve this problem? 13 / 17

Correlated equilibrium XIII Proposition (MSZ Thm 8.9) Let G = N, (A i ), (u i ) be a strategic form game. The set of correlated equilibria of G is convex. That is, if p A and p A are correlated equilibria, then p = αp + (1 α)p is a correlated equilibrium for any α [0, 1]. Proof. 14 / 17

Review Questions What is a correlated equilibrium and how can it be interpreted? Why can correlating recommendations (sometimes) help players to achieve a higher payoff than in any Nash equilibrium? Explain why every mixed strategy Nash equilibrium can be interpreted as a correlated equilibrium. reading: MSZ ch. 8 *reading: Jann and Schottmüller Correlated equilibria in homogenous good Bertrand competition, Journal of Mathematical Economics, Vol. 57, March 2015, pp. 31-37 15 / 17

Exercises I 1 Determine all pure and mixed Nash equilibria of the following game. Find a correlated equilibrium in which the sum of the players payoff is higher than in any Nash equilibrium. L R U 0,0 6,2 D 2,6 5,5 2 repeat the previous exercise with the following game L C R U 1,1 2,4 4,2 M 4,2 1,1 2,4 D 2,4 4,2 1,1 3 Show that all the actions that are played with positive probability in a correlated equilibrium are rationalizable. 16 / 17

Exercises II 4 *Think of a Bertrand game (2 firms with zero costs set price; one consumer buys from the firm with lowest price if this price is less than his valuation 1$). Assume that prices have to be in whole cents. Show that there is no correlated equilibrium that leads to higher total payoffs than the pure strategy Nash equilibrium (0.02, 0.02). Hint: Think of the highest recommendation given with positive probability in a correlated equilibrium. 17 / 17