Introduction to Game Theory
Introduction
Prita N Kusumawardhani and Chaikal Nuryakin
Why Do We Need Game Theory?
• GT has been applied to social sciences, humanities, engineering, and
natural sciences.
• We will focus on:
• Basic concepts
• Rationality
• Equilibrium strategies
• Ideas and philosophies that hidden behind the maths.
• We need to model social problems in which there is strategic
interdependence among agents.
• Examples: Traffics on roads, competition in markets, politics, and daily lives
(for example: whether to free ride or work hard in a group assignment).
What is Game Theory?
• GT concerns social problems
• In social setting, agents are interacting with each other.
• What is best for you depends on what others do
Strategic Situation
• GT constructs mathematical models to predict how people behave in
strategic situations.
Modelling Social Problems as A Game
Tol layang Jakarta – Cikampek
How does Tol layang affect the traffic on the existing road? How can you save
travelling time from Jakarta to Bandung?
Modelling Social Problems as A Game
• A famous example in political economics:
What determines the policies of Democrats and Republicans?
• An example from daily lives: Whether to work hard or to free ride
when working on a joint project.
GT provides a unified way of formulating and
solving ALL social problems.
GT provides a common language which facilitates
interdisciplinary research.
How to formulate a social problem as a
game?
• A game → a mathematical model of a game.
• Why maths? criticism
(1) People has free will. People can always deviate from the math prediction.
(2) We can just ask “why do you do that?” (a detailed interview).
(3) Unlike natural science, social science has no single theory that can explain everything.
• What is common between all those social problems?
• All social problems shared two features:
a) Individuals try to do their best against others
b) Under a certain set of rules
• Let’s formulate social problems which has those two features!
vNM
• Every social problem can be formulated to a game
• Game consists of three elements:
• Players: relevant agents i=1,2,…,N
• i’s Strategy: possible choice of actions
• i’s Payoff: utility for each possible action profile
Back to the Examples
• Political struggle between Democrats and Republicans.
• Set of strategies? Not so obvious!
• We need to simplify!
• Traffics between Jakarta – Bandung.
• Thousands of players!
• Strategies of each player {Route a, Route b, Route c}
• Payoff of each player = -(travelling time)
• Working on a joint project.
In search for a unified solution…
After formulating a social situation as a game, you have to solve it.
• vNM fails to develop a general solution concept of the game
Is there any general theory applicable to ALL social problems?
In search for the governing principle…
Let’s go back to the first feature of ALL social problems:
Each player try to do their best against others
• Let’s try to formulate using Maths!
• We need an assumption of Rationality
• Then, we need to assign payoffs
• Rational choice can be formulated as maximizing payoffs.
• Example: Choice between three vacation packages: Havana, Paris, and Venice. Preferences
represented by any payoff function. She prefers Havana to the other two.
u(Havana) = 1, u(Paris) = 0, u(Venice) = 0
Notice that utility function is a payoff function! It represents an agent’s
preferences.
Can we solve social problems by means of
rationality?
Poker game (man against man) Vs Roulette game (man against
machine)
Can we solve social problems by means of
rationality?
• Strategic thinking leads to the problem of infinite regress.
• Agents are trying to predict each other’s behaviour and maximize their payoff
(A’s belief about B’s belief about A’s belief about B’s belief ……)
• Rationality alone fails to pin down individual’s behaviour in social problems.
WE NEED A NEW THEORY!
Strategic Thinking
• Example: US vs. China
• US’s (optimal) action depends on how US predicts the
China’s action.
• China’s action depends on how China predicts the
US’s action.
• US’s action depends on how US predicts how China
predicts the US’s action.
• US’s action depends on how US predicts how China
predicts how US predicts the China’s action.
• and so on… (this is called “infinite regress”)
John Nash
• John Nash find a unifying solution concept.
• He found the answer in a cup of coffee
• 1994: Nobel Price in Economics
• Motion picture: A beautiful mind
John Nash
• Vortex point does not move.
• Players are moving towards the best replies.
• Suppose that the surface of coffee represents the set of all possible
human behaviour.
• Vortex (equilibrium) : All players are doing their best against others.
• Stable situation (equilibrium) in social settings.
John Nash
To sum up:
• Formulate a social situation as a mathematical model of a game (by
specifying players, strategies, and payoffs)
• Use maths to find equilibrium (a vortex point).
• You get a prediction in every social problem!
Nash Equilibrium
• John Nash develop a solution concept:
• In (Nash) equilibrium, no one can benefit if she unilaterally changes her
action.
• The solution always exists.
Nash Equilibrium: Traffic Problem
• What happens to the traffic flows when a new bypass is constructed?
Hotelling’s location game
• Ice cream vendors A and B on the street
• Customers are uniformly distributed
• Each customer goes to the nearest vendor
• A and B at the same location → they split customers equally.
• Vendors payoff = number of customers
Hotelling’s location game
Nash Equilibrium: Policies of Democrats and
Republicans
• Possible policies: Liberal and conservative
Nash Equilibrium: Policies of Democrats and
Republicans
• Median voter theorem
• Policies of the two parties tend to be very similar.
• Median voter’s opinion determines the policy .
Nash Equilibrium: Working on a joint project
• 2 players
• Strategies = {work hard, free ride}
• Payoffs
Player’s 1 ordering of the action profiles, from worst to best is:
(work hard, free ride) < (free ride, free ride) < (work hard, work hard) < and (free ride, work
hard)
We need a payoff such that the above preferences hold.
u(work hard, free ride) = 0
u(free ride, free ride) = 1
u(work hard, work hard) = 2
U(free ride, work hard) = 3
Nash Equilibrium: Working on a joint project
P2
Work hard Free ride
P1
Work hard 2, 2 0, 3
Free ride 3, 0 1, 1
Remarks
• Playing “free ride” is optimal no matter how the opponent takes
“work hard” or “free ride”.
• “Free ride” is an optimal (dominant) strategy.
• Combination of dominant strategies is Nash equilibrium.
• There are many games where no dominant strategy exists.
• Individually best decision ≠ Socially efficient outcome
• Optimality for individuals does not necessary imply optimality (Pareto
efficiency) for a group or society.
Two Frameworks
• Non-cooperative Game Theory
• examine individual decision making in strategic settings.
• assume a person decides her action on her own.
• does NOT rule out cooperative behaviors.
• Cooperative Game Theory
• examine group decision making in strategic settings.
• assume players can agree on their joint action, or can make
binding contracts.
• simplifies strategic analysis by NOT modeling the negotiation
process explicitly.
• The two tools are complements to one another, but this
lecture focuses mainly on Non-cooperative games.
Timing and Information
Complete Incomplete
Information Information
Nash Equilibrium (NE) Bayesian NE (BNE)
Static
Subgame Perfect Perfect Bayesian
Dynamic Equilibrium (SPNE) Equilibrium
Games in two forms
• Static games
Analyses in a normal form
• Dynamic games
Mostly analyze in an extensive-form
Normal form Games
30
Three elements/features:
1. Players.
2. Strategies available to every player
3. Payoff receive by each of player (for any
combinations of players strategies)
Static Games
31
• We consider static situations in which each player
simultaneously and independently chooses a strategy,
and the combination of strategies determines a
payoff for each player.
• The players need not literally act simultaneously.
• Each chooses her own action without knowing others’
choices.
• We will discuss it more formally later
