t e u s e s d y e m l s o ry g ry r - koç...
TRANSCRIPT
Introduction to Economic and Strategic BehaviorIntroduction
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Introduction 1 / 19
Description and Objectives
Description
Analysis of strategic interactions among individuals using game theory
Applications to economics, business, politics, law, biology, history
Evolution of social norms and other social institutions
Learning Objectives
Recognize real strategic situations to which game theory can beapplied
Model strategic situations as games
Apply tools of game theory to gain insights about real life situations
View social and economic issues from a completely new perspective
Levent Kockesen (Koc University) Introduction 2 / 19
Game Theory: Definition and Assumptions
Game theory studies strategic interactions within a group ofindividuals
◮ Actions of each individual have an effect on the outcome◮ Individuals are aware of that fact
Individuals are rational◮ have well-defined objectives over the set of possible outcomes◮ implement the best available strategy to pursue them
Rules of the game and rationality are common knowledge
Levent Kockesen (Koc University) Introduction 3 / 19
Example
10 people go to a restaurant for dinner
Order expensive or inexpensive fish?◮ Expensive fish: value = 18, price = 20◮ Inexpensive fish: value = 12, price = 10
Everbody pays own bill◮ What do you do?◮ Single person decision problem
Total bill is shared equally◮ What do you do?◮ It is a GAME
Levent Kockesen (Koc University) Introduction 4 / 19
Split or Steal
Levent Kockesen (Koc University) Introduction 5 / 19
Split or Steal
Levent Kockesen (Koc University) Introduction 6 / 19
Split or Steal
Van den Assem, Van Dolder, and Thaler, “Split or Steal? CooperativeBehavior When the Stakes Are Large” Management Science, 2012.
Individual players on average choose “split” 53 percent of the time
Propensity to cooperate is surprisingly high for consequential amounts
Less likely to cooperate if opponent has tried to vote them offpreviously
◮ Evidence for reciprocity
Little evidence that contestants propensity to cooperate dependspositively on the likelihood that their opponent will cooperate
◮ Little evidence for conditional cooperation
Young males are less cooperative than young females
Old males are more cooperative than old females
Levent Kockesen (Koc University) Introduction 7 / 19
Guess the Average
We will play a game
I will distribute slips of paper that looks like this
Round 1
Name:
Your guess:
Write your name and a number between 0 and 100
We will collect them and enter into Excel
The number that is closest to half the average wins
Winner gets 5TL (in case of a tie we choose randomly)Click here for the EXCEL file
Levent Kockesen (Koc University) Introduction 8 / 19
Another Example: A Single Person Decision Problem
Ali is an investor with $100
StateGood Bad
Bonds 10% 10%Stocks 20% 0%
Which one is better?
Probability of the good state p
Assume that Ali wants to maximize the amount of money he has atthe end of the year.
Bonds: $110
Stocks: average (or expected) money holdings:
p× 120 + (1− p)× 100 = 100 + 20× p
If p > 1/2 invest in stocks
If p < 1/2 invest in bonds
Levent Kockesen (Koc University) Introduction 9 / 19
An Investment Game
Ali again has two options for investing his $100:◮ invest in bonds
⋆ certain return of 10%◮ invest it in a risky venture
⋆ successful: return is 20%⋆ failure: return is 0%
◮ venture is successful if and only if total investment is at least $200
There is one other potential investor in the venture (Beril) who is inthe same situation as Ali
They cannot communicate and have to make the investment decisionwithout knowing the decisions of each other
Ali
BerilBonds Venture
Bonds 110, 110 110, 100Venture 100, 110 120, 120
Levent Kockesen (Koc University) Introduction 10 / 19
Games We Play
Group projects: How much effort should you put?◮ Coordination
Who is going to clean the kitchen?◮ War of attrition
Bargaining: How low the seller is willing to go?◮ Information, commitment, signaling
Poker: Bluff or not◮ Randomizing
Grading on the curve: Work hard or not?◮ Prisoners’ Dilemma
Levent Kockesen (Koc University) Introduction 11 / 19
Why Should We Learn Game Theory?
Media says so:
Managers have much to learn from game theory provided they use itto clarify their thinking, not as a substitute for business experience
◮ The Economist, 15 June 1996
Game Theory, long an intellectual pastime, came into its own as abusiness tool
◮ Forbes, 3 July 1995
Game theory is hot◮ The Wall Street Journal, 13 February 1995
Levent Kockesen (Koc University) Introduction 12 / 19
Why Should We Learn Game Theory?
Managers too...
Game theory forces you to see a business situation over many periodsfrom two perspectives: yours and your competitors
◮ Judy Lewent - CFO, Merck
To help predict competitor behavior and determine optimal strategy,our consulting teams use techniques such as pay-off matrices andcompetitive games
◮ John Stuckey & David White - McKinsey
At Bell Atlantic, we’ve found that the lessons of game theory give usa wider view of our business situation and provide us a more nimbleapproach to corporate planning
◮ Raymond Smith - CEO, Bell Atlantic
Levent Kockesen (Koc University) Introduction 13 / 19
Why Should We Learn Game Theory?
More Importantly
Life is sequence of strategic situations
Game theory does not provide recipes for them
But provides a method and toolbox to understand them better
You can even change the game for the better if you understand it well
Levent Kockesen (Koc University) Introduction 14 / 19
Why Should We Learn Game Theory?
It also helps us understand how and why certain norms andinstitutions arise throughout history and why some disappear
Used in◮ Economics◮ Political science◮ Law◮ Diplomacy◮ Business◮ Philosophy◮ Biology◮ History
Levent Kockesen (Koc University) Introduction 15 / 19
Strategic Thinking: Change the Game
I have 26 black cards while each one of you has one red card.
If one of us has one black and one red card, we will get paid 100 TL
How much are you willing to accept to give me your card?
Suppose now I throw away 3 black cards
How much are you willing to accept now?
Levent Kockesen (Koc University) Introduction 16 / 19
Correct Strategic Thinking
You have to understand the game correctly
You should try to predict what the other players are going to do◮ Who are the other players?◮ What are their possible moves?◮ What do they want from the game?◮ What do they know about the game?
Modeling the strategic interaction correctly is very important
Levent Kockesen (Koc University) Introduction 17 / 19
Types of Games
Games differ along two main dimensions:
1. Order of Moves◮ Simultaneous: sealed-bid auction◮ Sequential: ascending-bid auction
2. Information◮ Complete Information: valuations known◮ Incomplete Information: valuations unknown
Levent Kockesen (Koc University) Introduction 18 / 19
Game Forms
Information
Moves
Complete Incomplete
Simultaneous
Sequential
Strategic FormGames withCompleteInformation
Strategic FormGames withIncompleteInformation
Extensive FormGames withCompleteInformation
Extensive FormGames withIncompleteInformation
Levent Kockesen (Koc University) Introduction 19 / 19
Method
Look at examples and play games to get intuition
Understand general principles by studying theory
Study real life cases
Levent Kockesen (Koc University) Introduction 20 / 19
Introduction to Economic and Strategic BehaviorStrategic Form Games
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Strategic Form Games 1 / 1
Contribution Game
Everybody starts with 10 TL
You decide how much of 10 TL to contribute to joint fund
Amount you contribute will be doubled and then divided equallyamong everyone
I will distribute slips of paper that looks like this
Name:
Your Contribution:
Write your name and an integer between 0 and 10
We will collect them and enter into Excel
We will choose one player randomly and pay herClick here for the EXCEL file
Levent Kockesen (Koc University) Strategic Form Games 2 / 1
Contribution Game
Who are the other players?
What are their possible moves?
What do they want from the game?
What do they know about the game?
Levent Kockesen (Koc University) Strategic Form Games 3 / 1
Strategic Form Games
It is used to model situations in which players choose strategieswithout knowing the strategy choices of other players
Also known as normal form games
A strategic form game is composed of
1. Set of players
2. A set of strategies for each player
3. A payoff function for each player
An outcome is a collection of strategies, one for each player.◮ Also known as a strategy profile
Levent Kockesen (Koc University) Strategic Form Games 4 / 1
Price Competition
You will be matched randomly with another student
Each will independently choose High or Low price
If you both choose High: Each gets 10
If both Low: Each gets 5
If one High the other Low: High gets 2, Low gets 15
Make you choice and write on a piece of paper
Levent Kockesen (Koc University) Strategic Form Games 5 / 1
Price Competition
Player 1
Player 2High Low
High 10, 10 2, 15Low 15, 2 5, 5
Bimatrix
This way of writing the game is known as a bimatrix
Levent Kockesen (Koc University) Strategic Form Games 6 / 1
Price Competition
Player 1
Player 2High Low
High 10, 10 2, 15Low 15, 2 5, 5
What should Player 1 play?
Does that depend on what she thinks Player 2 will do?
Low is an example of a dominant strategy
It is optimal independent of what other players do
How about Player 2?
(Low, Low) is a dominant strategy equilibrium
Levent Kockesen (Koc University) Strategic Form Games 7 / 1
Dominant Strategies
A strategy A strictly dominates another strategy B if it yields a strictlyhigher payoff irrespective of how the other players play.
A weakly dominates B if it never does worse than B and sometimes doesstrictly better.
A strategy A is strictly dominant if it strictly dominates every otherstrategy of that player. It is called weakly dominant if it weakly dominatesevery other strategy.
Levent Kockesen (Koc University) Strategic Form Games 8 / 1
Dominant Strategy Equilibrium
If every player has a (strictly or weakly) dominant strategy, then thecorresponding outcome is a (strictly or weakly) dominant strategyequilibrium.
Levent Kockesen (Koc University) Strategic Form Games 9 / 1
Dominant Strategy Equilibrium
T
WH L
H 10, 10 2, 15L 15, 2 5, 5
L strictly dominates H
(L,L) is a strictly dominantstrategy equilibrium
T
WH L
H 10, 10 5, 15L 15, 5 5, 5
L weakly dominates H
(L,L) is a weakly dominantstrategy equilibrium
Levent Kockesen (Koc University) Strategic Form Games 10 / 1
Dominant Strategy Equilibrium
A reasonable solution concept
It only demands the players to be rational
It does not require them to know that the others are rational too
But it does not exist in many interesting games
Levent Kockesen (Koc University) Strategic Form Games 11 / 1
Prisoners’ Dilemma
Prisoners’ Dilemma
Player 1
Player 2Confess Deny
Confess −5,−5 0,−10Deny −10, 0 −1,−1
How would you play?
Levent Kockesen (Koc University) Strategic Form Games 12 / 1
Case Study: Advertising Ban in Tobacco Industry
1960s: tobacco companies advertise heavily on TV and radio◮ TV ads accounted for more than 80% of total advertising budget
1964: US Surgeon General’s 1964 report finds cigarette smokinghazardous to health
1970: Public Health Cigarette Smoking Act banned all cigarette TVand radio ads
Question
What is the effect on advertising expenditures? On profits?
Levent Kockesen (Koc University) Strategic Form Games 13 / 1
Case Study: Advertising Ban in Tobacco Industry
Model Assumptions
Symmetric oligopolistic industry
Firms choose how much to spend on advertising
Total demand is perfectly inelastic
Advertising Game
PM
RJRAd No
Ad 30, 30 60, 20No 20, 60 50, 50
Levent Kockesen (Koc University) Strategic Form Games 14 / 1
Case Study: Advertising Ban in Tobacco Industry
What Happened:
Advertising expenditures fell sharply in 1971
Profits rose as stock returns for the major tobacco companies reachedabnormal heights
Aggregate sales remained unchanged
After 5 years: Advertising spending began to recover and even exceedpre-ban levels
See Shi Qi, “The Impact of Advertising Regulation on Industry TheCigarette Advertising Ban of 1971,” RAND J. (2013)
Levent Kockesen (Koc University) Strategic Form Games 15 / 1
Rationality
Players have consistent preferences over outcomes◮ Can rank outcomes◮ Payoffs represent rankings
They have beliefs regarding future events◮ Other players’ moves◮ Chance events
On the basis of these beliefs they choose the best strategy◮ The strategy that maximizes their expected payoffs
Levent Kockesen (Koc University) Strategic Form Games 16 / 1
Rationality
Suppose there are 4 possible outcomes: a, b, c, d
Corresponding payoffs: u(a), u(b), u(c), u(d)
and probabilities p(a), p(b), p(c), p(d)
Expected payoff is
p(a)u(a) + p(b)u(b) + p(c)u(c) + p(d)u(d)
Levent Kockesen (Koc University) Strategic Form Games 17 / 1
Risk Aversion
Choose between◮ S: I will give you $100◮ R: I will flip a fair coin. If heads I will give you $200; If tails I will give
you nothing
If you prefer S over R you are risk averse
If you don’t care you are risk neutral
If you prefer R over S you are risk lover
An individual is risk averse is he prefers to receive the expected value of alottery to playing the lottery
Levent Kockesen (Koc University) Strategic Form Games 18 / 1
Risk Aversion
Payoff function (of certain outcome) is
Linear if risk neutral◮ u(x) = x
Concave if risk averse◮ u(x) =
√x
Convex if risk lover◮ u(x) = x2
Levent Kockesen (Koc University) Strategic Form Games 19 / 1
Risk Preferences
b
b
bb
x
u(x)
x1 x2E[x]
Risk averse: u(E[x]) > E[u(x)]
b
b
b
x
u(x)
x1 x2E[x]
Risk neutral: u(E[x]) = E[u(x)]
b
b
b
b
x
u(x)
x1 x2E[x]
Risk lover: u(E[x]) < E[u(x)]
Levent Kockesen (Koc University) Strategic Form Games 20 / 1
Rationality
Rationality does not mean selfishness◮ Payoff functions must include all relevant considerations such as
altruism
It is a consistency requirement
Should be thought as an approximation to reality
In evolutionary games we do not assume rationality◮ Genes govern behavior◮ Those that are successful reproduce faster◮ An evolutionary stable state is the eventual outcome of this dynamics
Levent Kockesen (Koc University) Strategic Form Games 21 / 1
Common Knowledge
We also assume that the game as well as the rationality is commonknowledge
A fact X is common knowledge if everybody knows it
If everybody knows that everybody knows it
If everybody knows that everybody knows that everybody knows it
... ad infinitum
Levent Kockesen (Koc University) Strategic Form Games 22 / 1
A Puzzle
Three girls sitting in a circle
Each is wearing a hat: Red or White
Each can see the others’ hat but not her own
Teacher asks each to guess her hat color◮ This is done sequentially◮ Assume a wrong guess is very costly
In fact all have red
Could any one of them guess correctly?
Suppose now that the teacher says
At least one of you is wearing a red hat
How about now?
Levent Kockesen (Koc University) Strategic Form Games 23 / 1
Price Matching
Wal-Mart: Our Ad Match Guarantee
We’re committed to providing low prices every day. On everything. So ifyou find a lower advertised price on an identical product, tell us and we’llmatch it. Right at the register.
Levent Kockesen (Koc University) Strategic Form Games 24 / 1
Price Matching
Sounds like a good deal for customers
How does this change the game?
Toys“R”us
Wal-MartHigh Low Match
High 10, 10 2, 15 10, 10Low 15, 2 5, 5 5, 5
Match 10, 10 5, 5 10, 10
Is there a dominant strategy for any of the players?
There is no dominant strategy equilibrium for this game
So, what can we say about this game?
Levent Kockesen (Koc University) Strategic Form Games 25 / 1
Price Matching
Toys“R”us
Wal-MartHigh Low Match
High 10, 10 2, 15 10, 10Low 15, 2 5, 5 5, 5
Match 10, 10 5, 5 10, 10
High is weakly dominated and Toys“R”us is rational◮ Toys“R”us should not use High
High is weakly dominated and Wal-Mart is rational◮ Wal-Mart should not use High
Each knows the other is rational◮ Toys“R”us knows that Wal-Mart will not use High◮ Wal-Mart knows that Toys“R”us will not use High◮ This is where we use common knowledge of rationality
Levent Kockesen (Koc University) Strategic Form Games 26 / 1
Price Matching
Therefore we have the following “effective” game
Toys“R”us
Wal-MartLow Match
Low 5, 5 5, 5Match 5, 5 10, 10
Low becomes a weakly dominated strategy for both
Both companies will play Match and the prices will be high
The above procedure is known as Iterated Elimination of DominatedStrategies (IEDS)
To be a good strategist try to see the world from the perspective of yourrivals and understand that they will most likely do the same
Levent Kockesen (Koc University) Strategic Form Games 27 / 1
Dominated Strategies
A “rational” player should never play an strategy when there isanother strategy that gives her a higher payoff irrespective of how theothers play
We call such a strategy a dominated strategy
A strategy A is strictly dominated by another strategy B if B does strictlybetter than A irrespective of how other players play.
A is weakly dominated by B if B never does worse than A and sometimesdoes strictly better.
Levent Kockesen (Koc University) Strategic Form Games 28 / 1
Iterated Elimination of Dominated Strategies
Common knowledge of rationality justifies eliminating dominatedstrategies iteratively
This procedure is known as Iterated Elimination of DominatedStrategies
If every strategy eliminated is a strictly dominated strategy◮ Iterated Elimination of Strictly Dominated Strategies
If at least one strategy eliminated is a weakly dominated strategy◮ Iterated Elimination of Weakly Dominated Strategies
Remember Guessing Game?
Pick a number between 0 and 100
You win if your number is closest to half the average
What do you pick?
Levent Kockesen (Koc University) Strategic Form Games 29 / 1
Iterated Elimination of Dominated Strategies
Half the average can be at most 50
If everybody is rational◮ Nobody should choose higher than 50
If everybody knows that everybody is rational◮ Nobody should choose higher than 25
If everybody knows that everybody knows that everybody is rational◮ Nobody should choose higher than 12.5
Can eliminate all down to zero
Keynes’ Beauty Contest
Number closest to the average wins.
Shiller’s NYT article
Levent Kockesen (Koc University) Strategic Form Games 30 / 1
Envelope Paradox
Ali and Berna has each an envelope with 5, 10, 20, 40, 80, or 160 TL.
One of the envelopes has twice the amount in the other
Each looks into their envelope and independently decide whether theywant to exchange envelopes
If both want to do so, they exchange envelopes
Should they exchange?
Suppose Berna has 20 TL. Should she exchange?
Levent Kockesen (Koc University) Strategic Form Games 31 / 1
Chairman’s Paradox
A committee of 3 members, 1, 2, 3 must choose among 3 options,A,B,C
The alternative that gets two votes is chosen
In case of a tie the chairman, player 3, chooses
Preferences1 2 3
A B CB C AC A B
Levent Kockesen (Koc University) Strategic Form Games 32 / 1
Chairman’s Paradox
1 2 3
A B CB C AC A B
1
2A B C
A 2, 0, 1 2, 0, 1 2, 0, 1B 2, 0, 1 1, 2, 0 0, 1, 2C 2, 0, 1 0, 1, 2 0, 1, 2
A
2A B C
A 2, 0, 1 1, 2, 0 0, 1, 2B 1, 2, 0 1, 2, 0 1, 2, 0C 0, 1, 2 1, 2, 0 0, 1, 2
B
2A B C
A 2, 0, 1 0, 1, 2 0, 1, 2B 0, 1, 2 1, 2, 0 0, 1, 2C 0, 1, 2 0, 1, 2 0, 1, 2
C
What happens if everyone votes sincerely?◮ There is a tie◮ Chairman (3) chooses C
Is this a reasonable outcome?
Levent Kockesen (Koc University) Strategic Form Games 33 / 1
Chairman’s Paradox
1
2A B C
A 2, 0, 1 2, 0, 1 2, 0, 1B 2, 0, 1 1, 2, 0 0, 1, 2C 2, 0, 1 0, 1, 2 0, 1, 2
A
2A B C
A 2, 0, 1 1, 2, 0 0, 1, 2B 1, 2, 0 1, 2, 0 1, 2, 0C 0, 1, 2 1, 2, 0 0, 1, 2
B
2A B C
A 2, 0, 1 0, 1, 2 0, 1, 2B 0, 1, 2 1, 2, 0 0, 1, 2C 0, 1, 2 0, 1, 2 0, 1, 2
C
Are there dominated strategies?◮ Player 1: C is weakly dominated by A◮ Player 2: A and C are weakly dominated by B◮ Player 3: A and B are weakly dominated by C
The reduced game
2B
A 0, 1, 2B 1, 2, 0
C
Final outcome: (B,B,C) and B wins◮ Worst outcome for the chairman!
Levent Kockesen (Koc University) Strategic Form Games 34 / 1
Effort Game
You choose how much effort to expend for a joint project◮ An integer between 1 and 7
The quality of the project depends on the smallest effort: e◮ Weakest link
Effort is costly
If you choose e your payoff is
6 + 2e− e
We will play this twice
We will randomly choose one round and one student and pay her
Enter your name and effort choice for Round 1Click here for the EXCEL file
Levent Kockesen (Koc University) Strategic Form Games 35 / 1
Effort Game
Two effort level - two player version
Low HighLow 7, 7 7, 1High 1, 7 13, 13
Is there a dominated strategy?
If player 1 expects player 2 to choose Low, what is her best strategy(best response)?
If player 2 expects player 1 to choose Low what is its best response?
(Low, Low) is an outcome such that◮ Each player best responds, given what she believes the other will do◮ Their beliefs are correct
It is a Nash equilibrium
Is (Low,Low) the only Nash equilibrium?
Levent Kockesen (Koc University) Strategic Form Games 36 / 1
Nash Equilibrium
Nash Equilibrium
Nash equilibrium is a strategy profile (a collection of strategies, one foreach player) such that each strategy is a best response (maximizes payoff)to all the other strategies
Nash equilibrium is self-enforcing: no player has an incentive todeviate unilaterally
One way to find Nash equilibrium is to first find the best responsecorrespondence for each player
◮ Best response correspondence gives the set of payoff maximizingstrategies for each strategy profile of the other players
... and then find where they “intersect”
Levent Kockesen (Koc University) Strategic Form Games 37 / 1
Nash Equilibrium
Low HighLow 7, 7 7, 1High 1, 7 13, 13
1’s best response to Low is Low
Her best response to High is High
Similarly for player 2
Best response correspondences intersect at (Low, Low) and (High,High)
These two strategy profiles are the two Nash equilibria of this game
We would expect in the long-run one of these outcomes to prevail
Risk dominance
Payoff dominance
Levent Kockesen (Koc University) Strategic Form Games 38 / 1
Stag Hunt
Jean-Jacques Rousseau in A Discourse on Inequality
If it was a matter of hunting a deer, everyone well realized thathe must remain faithful to his post; but if a hare happened topass within reach of one of them, we cannot doubt that he wouldhave gone off in pursuit of it without scruple...
Stag HareStag 2, 2 0, 1Hare 1, 0 1, 1
Levent Kockesen (Koc University) Strategic Form Games 39 / 1
The Bar Scene
Levent Kockesen (Koc University) Strategic Form Games 40 / 1
The Bar Scene
Blonde BrunetteBlonde 0, 0 2, 1
Brunette 1, 2 1, 1
See S. Anderson and M. Engers: Participation Games: Market Entry,Coordination, and the Beautiful Blonde, Journal of EconomicBehavior and Organization, 2007
Levent Kockesen (Koc University) Strategic Form Games 41 / 1
John Nash
Levent Kockesen (Koc University) Strategic Form Games 42 / 1
Nash Demand Game
Each of you will be randomly matched with another student
You are trying to divide 10 TL
Each writes independently how much she wants (in multiples of 1 TL)
If two numbers add up to greater than 10 TL each gets nothing
Otherwise each gets how much she wrote
Write your name and demand on the slips
I will match two randomly
Choose one pair randomly and pay themClick here for the EXCEL file
Levent Kockesen (Koc University) Strategic Form Games 43 / 1
Some Other Commonly Used Games
Hawk-Dove
Player 1
Player 2D H
D 3, 3 1, 5H 5, 1 0, 0
Coordination
Player 1
Player 2L R
L 1, 1 0, 0R 0, 0 1, 1
Player 1
Player 2L R
L 1, 1 0, 0R 0, 0 2, 2
Levent Kockesen (Koc University) Strategic Form Games 44 / 1
Some Other Commonly Used Games
Battle of the Sexes
Player 1
Player 2B S
B 2, 1 0, 0S 0, 0 1, 2
Chicken
Jim
BuzzQuit Stay
Quit 0, 0 −1, 1Stay 1,−1 −2,−2
Levent Kockesen (Koc University) Strategic Form Games 45 / 1
Some Other Commonly Used Games
Matching Pennies
Player 1
Player 2H T
H 1,−1 −1, 1T −1, 1 1,−1
Levent Kockesen (Koc University) Strategic Form Games 46 / 1
Nash Equilibrium: Discussion
Nash equilibrium as steady-state
Nash equilibrium as self-enforcing agreement
Nash equilibrium as stable norm or institution
Levent Kockesen (Koc University) Strategic Form Games 47 / 1
Evidence from Experiments
Davis and Holt, Experimental Economics◮ In simple games with unique Nash equilibrium, that outcome has
considerable drawing power
Multiple equilibria: Importance of focal points
Interdependent preferences
Levent Kockesen (Koc University) Strategic Form Games 48 / 1
Difficulties with Nash Equilibrium
Why should players play Nash equilibrium?◮ Roger Myerson: Why not?
How would you play the following?
Player 1
Player 2A B C
A 2, 2 3, 1 0, 2B 1, 3 2, 2 3, 2C 2, 0 2, 3 2, 2
What would Myerson say?◮ If you really think the other will play C, you should play B◮ But if the other goes through this reasoning, she should play A◮ in which case, you might as well play a best response: A
Levent Kockesen (Koc University) Strategic Form Games 49 / 1
Introduction to Economic and Strategic BehaviorStrategic Form Games: Applications
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Applications 1 / 1
Auctions
Many economic transactions are conducted through auctions
treasury bills
foreign exchange
publicly owned companies
mineral rights
airwave spectrum rights
art work
antiques
cars
houses
government contracts
Also can be thought of as auctions
takeover battles
queues
wars of attrition
lobbying contests
Levent Kockesen (Koc University) Applications 2 / 1
Auction Formats
1. Open bid auctions1.1 ascending-bid auction
⋆ aka English auction⋆ price is raised until only one bidder remains, who wins and pays the
final price
1.2 descending-bid auction⋆ aka Dutch auction⋆ price is lowered until someone accepts, who wins the object at the
current price
2. Sealed bid auctions2.1 first price auction
⋆ highest bidder wins; pays her bid
2.2 second price auction⋆ aka Vickrey auction⋆ highest bidder wins; pays the second highest bid
Levent Kockesen (Koc University) Applications 3 / 1
Auction Formats
Auctions also differ with respect to the valuation of the bidders
1. Private value auctions◮ each bidder knows only her own value◮ artwork, antiques, memorabilia
2. Common value auctions◮ actual value of the object is the same for everyone◮ bidders have different private information about that value◮ oil field auctions, company takeovers
Levent Kockesen (Koc University) Applications 4 / 1
Strategically Equivalent Formats
!
!
Open Bid Sealed Bid
English Auction
Dutch Auction
Second Price
First Price
We will study sealed bid auctions
For now we will assume that values are common knowledge◮ value of the object to player i is vi dollars
For simplicity we analyze the case with only two bidders
Assume v1 > v2 > 0
Levent Kockesen (Koc University) Applications 5 / 1
Second Price Auctions
Highest bidder wins and pays the second highest bid
In case of a tie, the object is awarded to player 1
Suppose player 1’s value is 5 and player 2’s is 3
Assume they can bid 1, 2, 3, 4, 5, 6
1 2 3 4 5 61 4, 0 0, 2 0, 2 0, 2 0, 2 0, 22 4, 0 3, 0 0, 1 0, 1 0, 1 0, 13 4, 0 3, 0 2, 0 0, 0 0, 0 0, 04 4, 0 3, 0 2, 0 1, 0 0,−1 0,−15 4, 0 3, 0 2, 0 1, 0 0, 0 0,−26 4, 0 3, 0 2, 0 1, 0 0, 0 −1, 0
Levent Kockesen (Koc University) Applications 6 / 1
Second Price Auctions
1 2 3 4 5 61 4, 0 0, 2 0, 2 0, 2 0, 2 0, 22 4, 0 3, 0 0, 1 0, 1 0, 1 0, 13 4, 0 3, 0 2, 0 0, 0 0, 0 0, 04 4, 0 3, 0 2, 0 1, 0 0,−1 0,−15 4, 0 3, 0 2, 0 1, 0 0, 0 0,−26 4, 0 3, 0 2, 0 1, 0 0, 0 −1, 0
For player 1: bidding 4 and 5 weakly dominates all others◮ What if player 2 could also bid 4.5?◮ 5 would weakly dominate 4
For player 2: bidding 3 and 4 weakly dominates all others◮ If player 1 could also bid between 3 and 4?◮ 3 would weakly dominate 4
If players could bid any number, everybody bidding own value is a weaklydominant strategy equilibrium.
Levent Kockesen (Koc University) Applications 7 / 1
Second Price Auctions
I. Bidding your value weakly dominates bidding higher
Suppose your value is $10 but you bid $15. Three cases:
1. The other bid is higher than $15 (e.g. $20)◮ You loose either way: no difference
2. The other bid is lower than $10 (e.g. $5)◮ You win either way and pay $5: no difference
3. The other bid is between $10 and $15 (e.g. $12)◮ You loose with $10: zero payoff◮ You win with $15: loose $2
5
10 value
12
15 bid
20
Levent Kockesen (Koc University) Applications 8 / 1
Second Price Auctions
II. Bidding your value weakly dominates bidding lower
Suppose your value is $10 but you bid $5. Three cases:
1. The other bid is higher than $10 (e.g. $12)◮ You loose either way: no difference
2. The other bid is lower than $5 (e.g. $2)◮ You win either way and pay $2: no difference
3. The other bid is between $5 and $10 (e.g. $8)◮ You loose with $5: zero payoff◮ You win with $10: earn $2
Weakly dominant strategy equilibrium = (v1, v2)
There are many Nash equilibria. For example (v1, 0) 2
10 value
8
5 bid
12
Levent Kockesen (Koc University) Applications 9 / 1
First Price Auctions
Highest bidder wins and pays her own bid
In case of a tie, the object is awarded to player 1
Suppose player 1’s value is 5 and player 2’s is 3
Assume they can bid 1, 2, 3, 4, 5, 6
1 2 3 4 5 61 4, 0 0, 1 0, 0 0,−1 0,−2 0,−32 3, 0 3, 0 0, 0 0,−1 0,−2 0,−33 2, 0 2, 0 2, 0 0,−1 0,−2 0,−34 1, 0 1, 0 1, 0 1, 0 0,−2 0,−35 0, 0 0, 0 0, 0 0, 0 0, 0 0,−36 −1, 0 −1, 0 −1, 0 −1, 0 −1, 0 −1, 0
Levent Kockesen (Koc University) Applications 10 / 1
First Price Auctions
1 2 3 4 5 61 4, 0 0, 1 0, 0 0,−1 0,−2 0,−32 3, 0 3, 0 0, 0 0,−1 0,−2 0,−33 2, 0 2, 0 2, 0 0,−1 0,−2 0,−34 1, 0 1, 0 1, 0 1, 0 0,−2 0,−35 0, 0 0, 0 0, 0 0, 0 0, 0 0,−36 −1, 0 −1, 0 −1, 0 −1, 0 −1, 0 −1, 0
Player 2: bidding 2 is weakly dominant
Player 1’s best response to 2 is 2
(2, 2) is a Nash equilibrium◮ What if player 2 could bid between 2 and 3?
Other Nash equilibria? (3, 3), (4, 4), (5, 5)
Levent Kockesen (Koc University) Applications 11 / 1
Nash Equilibria of First Price Auctions
Assume players can bid any number
What are the Nash equilibria?
We can compute the best response correspondences
or we can adopt a different approach◮ First find the necessary conditions for a Nash equilibrium
⋆ If a strategy profile is a Nash equilibrium then it must satisfy theseconditions
◮ Then find the sufficient conditions⋆ If a strategy profile satisfies these conditions, then it is a Nash
equilibrium
Levent Kockesen (Koc University) Applications 12 / 1
Necessary ConditionsLet (b∗1, b
∗2) be a Nash equilibrium. Then,
1. Player 1 wins: b∗1 ≥ b∗2
Proof
Suppose not: b∗1 < b∗2. Two possibilities:
◮ b∗2 ≤ v2: Player 1 could bid v2 and obtain a strictly higher payoff
◮ b∗2 > v2: Player 2 has a profitable deviation: bid zero
Contradicting the hypothesis that (b∗1, b∗2) is a Nash equilibrium.
2. b∗1 = b∗2
Proof
Suppose not: b∗1 > b∗2. Player 1 has a profitable deviation: bid b∗2
3. v2 ≤ b∗1 ≤ v1
Proof
ExerciseLevent Kockesen (Koc University) Applications 13 / 1
Sufficient Conditions
So, any Nash equilibrium (b∗1, b∗2) must satisfy
v2 ≤ b∗1 = b∗2 ≤ v1.
Is any pair (b∗1, b∗2) that satisfies these inequalities an equilibrium?
Set of Nash equilibria is given by
{(b1, b2) : v2 ≤ b1 = b2 ≤ v1}
Levent Kockesen (Koc University) Applications 14 / 1
Price Competition Models
Quantity (or capacity) competition: Cournot Model◮ Augustine Cournot (1838)
Price Competition: Bertrand Model◮ Joseph Bertrand (1883)
Two main models:
1. Bertrand Oligopoly with Homogeneous Products
2. Bertrand Oligopoly with Differentiated Products
Levent Kockesen (Koc University) Applications 15 / 1
Bertrand Duopoly with Homogeneous Products
Two firms, each with unit cost c ≥ 0
They choose prices◮ The one with the lower price captures the entire market◮ In case of a tie they share the market equally
Total market demand is equal to one (not price sensitive)
Both firms are profit maximizers
Levent Kockesen (Koc University) Applications 16 / 1
Nash Equilibrium
Suppose P ∗1 , P
∗2 is a Nash equilibrium. Then
1. P ∗1 , P
∗2 ≥ c. Why?
2. At least one charges c◮ P ∗
1 > P ∗2 > c?
◮ P ∗2 > P ∗
1 > c?◮ P ∗
1 = P ∗2 > c?
3. P ∗2 > P ∗
1 = c?
4. P ∗1 > P ∗
2 = c?
The only candidate for equilibrium is P ∗1 = P ∗
2 = c, and it is indeed anequilibrium.
The unique Nash equilibrium of the Bertrand game is (P ∗1 , P
∗2 ) = (c, c)
Levent Kockesen (Koc University) Applications 17 / 1
Models of Spatial Competition
There is a population of individuals (consumers, voters, etc.)characterized by some trait (type)
◮ income◮ taste◮ political inclination
Individuals have preferences (which may depend on their types) overoutcomes
◮ tax rates◮ product characteristics◮ environmental regulations
Players (firms, political parties, etc.) choose outcomes
Given players’ choices, each individual chooses a player
Players have preferences over outcomes and the choices of theindividuals
◮ Firms choose product characteristic and care about how much they sell◮ Political parties choose policy positions and care about whether they
win the election and which position gets implemented
Levent Kockesen (Koc University) Applications 18 / 1
Electoral Competition: Downs Model
Originally due to Hotelling (1929)◮ Applied to electoral competition by Downs (1957)
There is a large number of voters each with an ideal policy
Assume equal number of voters for each ideal policy◮ Uniform distribution
Each voter has single peaked preferences over policies◮ They prefer policies that are closer to their ideal policy
Levent Kockesen (Koc University) Applications 19 / 1
Electoral Competition: Downs Model
Candidates announce a policy◮ 10% tax rate vs. 25% tax rate◮ pro-EU vs anti-EU
Voters vote sincerely◮ each voter votes for the candidate whose policy is closest to her ideal
policy◮ if indifferent they vote for each with equal probabilities
Candidate with the highest amount of votes wins
Candidates care only about the outcome of the election◮ preferences: win ≻ tie ≻ lose
Say the two candidates choose 0 < p1 < p2 < 1
0 1p1 p2p1+p22
Vote for 1 Vote for 2
Levent Kockesen (Koc University) Applications 20 / 1
Nash Equilibrium
Suppose p∗1, p∗2 is a Nash equilibrium. Then
1. Outcome must be a tie◮ Whatever your opponent chooses you can always guarantee a tie
2. p∗1 6= p∗2?
0 11/2p1 p2
3. p∗1 = p∗2 6= 1/2?
0 11/2p1
p2
The only candidate for equilibrium is p∗1 = p∗2 = 1/2, which is indeed anequilibrium.
The unique Nash equilibrium of the game is (p∗1, p∗2) = (1/2, 1/2)
Levent Kockesen (Koc University) Applications 21 / 1
Extensions
Median Voter Theorem
Consider the above model but assume that the voters are distributedaccording to any distribution function. Then the unique Nash equilibriumis to choose the median voter’s ideal policy for both candidates.
Other Models
Models with participation costs
Models with more than two players
Models with multidimensional policy space
Models with ideological candidates: Wittman (1973) Model
Levent Kockesen (Koc University) Applications 22 / 1
Game Theory and Traffic
4 cars to go from A to D
If n cars travel, it takes n hours from A to B and C to D
Takes 6 hours from A to C and C to D
A
B
C
D
n
6
6
n
What is the Nash equilibrium?
Levent Kockesen (Koc University) Applications 23 / 1
Game Theory and Traffic
A
B
C
D
nB
6
6
nC
nB + 6 > nC + 6 ⇒ switch to C
nC + 6 > nB + 6 ⇒ switch to B
This implies nB = nC = 2 in any equilibrium
nB = nC = 2 is indeed a Nash equilibrium
Travel time is 8 hours
Levent Kockesen (Koc University) Applications 24 / 1
Game Theory and Traffic
A fast highway is built from B to C
A
B
C
D
n
6
6
n
1
How does the travel time change?
Levent Kockesen (Koc University) Applications 25 / 1
Game Theory and Traffic
A
B
C
D
nB
6
6
nC
1
What is the Nash equilibrium nB and nC?No car will go to C directly because going through B takes at most 5hrsTherefore, all the cars must go through BOnce at B no car will go directly to D because going through C takesat most 5 hrsnB = nC = 4Total travel time is 9 hours!
Levent Kockesen (Koc University) Applications 26 / 1
Braess’s Paradox
German mathematician Dietrich Braess in a 1969 paper
In South Korea, traffic sped up after a motorway was removed
42nd Street in NYC was closed to traffic on Earth Day (1990) ⇒ lesscongestion
Levent Kockesen (Koc University) Applications 27 / 1
Game Theory of Legal Regimes
A driver and a pedestrian choose care level◮ Due care: Socially optimal◮ No care: Less than socially optimal
Due care costs 10
Accident inflicts injury on pedestrian with cost 100
An accident happens for certain unless both choose due care
It happens with one-in-ten chance if both choose due care
Levent Kockesen (Koc University) Applications 28 / 1
No Liability
If driver is never liable for accident
Pedestrian
DriverNo Care Due Care
No Care −100, 0 −100,−10Due Care −110, 0 −20,−10
What do you expect to happen?
Iterated elimination of dominated strategies → (No Care, No Care)
Driver has too little incentive to take care
Levent Kockesen (Koc University) Applications 29 / 1
Pure Strict Liability
Driver pays for all injuries
Pedestrian
DriverNo Care Due Care
No Care 0,−100 0,−110Due Care −10,−100 −10,−20
What do you expect to happen?
Iterated elimination of dominated strategies → (No Care, No Care)
Pedestrian has too little incentive to take care
Levent Kockesen (Koc University) Applications 30 / 1
Negligence with Contributory Negligence
Driver pays only if he is negligent and pedestrian is not
Pedestrian
DriverNo Care Due Care
No Care −100, 0 −100,−10Due Care −10,−100 −20,−10
Prevailing principle of Anglo-American tort law
What do you expect to happen?
Iterated elimination of dominated strategies → (Due Care, Due Care)
Levent Kockesen (Koc University) Applications 31 / 1
Strict Liability with Contributory Negligence
If both exercise care driver pays
Pedestrian
DriverNo Care Due Care
No Care −100, 0 −100,−10Due Care −10,−100 −10,−20
Same prediction: (Due Care, Due Care)
Only the costs of accident are shared differently
Levent Kockesen (Koc University) Applications 32 / 1
Political Reform
Victorious parties rewards their followers with government jobs◮ Patronage
Leads to corruption and inefficiencies
Meritocracy is better◮ A reforms is needed to achieve it◮ Has proved to be difficult◮ Why?
Model by Geddes (1991)
Levent Kockesen (Koc University) Applications 33 / 1
Political Reform
Both patronage and reform have advantages in elections
Majority party and minority party
With no patronage: Chances of winning (p, 1− p)
Patronage increases chances by v1 and v2
NP PNP p, 1− p p− v2, 1− p+ v2P p+ v1, 1− p− v1 p+ v1 − v2, 1− p− v1 + v2
P: Patronage, NP: No Patronage
Row player: Majority party, Column player: Minority Party
Patronage is dominant strategy equilibrium
Levent Kockesen (Koc University) Applications 34 / 1
Political Reform
Supporting reform may help during campaign
Suppose reform passes if and only if majority party supports it
If majority party supports reform, reform laws pass and neither partycan use patronage
If majority party opposes reform, then both parties use patronage◮ This is from the first model: Patronage is dominant strategy without
reform
However, minority may still support reform during the campaign
There is a benefit of supporting reform when the other does not: e
Levent Kockesen (Koc University) Applications 35 / 1
Political Reform
R NRR p, 1− p p+ e, 1− p− e
NR p+ v1 − v2 − e, 1− p− v1 + v2 + e p+ v1 − v2, 1− p− v1 + v2
Supporting reform is dominant for minority party◮ Reform is a common theme with challengers
Reform happens if
v1 − v2 < e
Reform is more likely if v1 − v2 is small◮ If parties are relatively equal in electoral strength◮ If majority party is much stronger, it may expect greater benefits from
patronage◮ If majority party has had disproportionate access to patronage in the
past it will have more entrenched workers and more to benefit frompatronage
Reform is also more likely if there is a strong support for reform◮ That is e is high◮ Likely to be case after scandals
Levent Kockesen (Koc University) Applications 36 / 1
Financial Crises
Consumers demand liquidity
Banks provide them: Demand deposits
But they make loans that are less liquid
This makes them vulnerable to bank runs
Diamond-Dybvig Model (1983)◮ See also: Diamond (2007)
September 2007
Levent Kockesen (Koc University) Applications 37 / 1
Bank Runs
Consumers demand liquidity because they are uncertain about whenthey need liquid funds
Assume three dates: t = 0, 1, 2
Each consumer need to consume either at date 1 or date 2
At date 0, they don’t know when◮ probability of date 1: p◮ probability of date 2: 1− p
At date 0 each has 1 TL that they can invest◮ If they liquidate on date 1: r1◮ If they liquidate on date 2: r2◮ r1 < r2 ⇒ the asset is illiquid
Expected utility
pu(r1) + (1− p)u(r2)
Levent Kockesen (Koc University) Applications 38 / 1
Bank Runs
Banks can invest 1 TL in an asset that returns◮ 1 if liquidated on date 1◮ R > 1 if liquidated on date 2
They offer a contract to consumers that promise◮ r1 > 1 if withdrawn on date 1◮ r2 < R if withdrawn on date 2◮ This is more liquid than banks’ asset
Banks create liquidity◮ Without them there may be no investment in illiquid assets
Why do the consumers prefer deposit in the bank?◮ Because they are risk averse. For example
u(c) = 1− 1
c
Levent Kockesen (Koc University) Applications 39 / 1
Bank Runs
Expected utility of less liquid asset
p(1− 1
1) + (1− p)(1 − 1
R)
Expected utility of bank deposit
p(1− 1
r1) + (1− p)(1− 1
r2)
If R = 2, p = 0.25, r1 = 1.28, r2 = 1.813◮ Expected utility of illiquid asset is 0.375◮ Expected utility of bank deposit is 0.391
Would a bank be willing to offer these rates?
Levent Kockesen (Koc University) Applications 40 / 1
Bank Runs
Suppose 100 consumers deposit 1 TL each on date 0
There is no aggregate uncertainty◮ A fraction p of all consumers is definitely date 1 type
Fraction p withdraws on date 1 gets r1◮ Date 1 payout is 100pr1
Remaining 100(1 − pr1) brings 100(1 − pr1)R on date 2
Out of that 100(1 − p)r2 is paid to the depositors
Banks’ profit is
100(1 − pr1)R− 100(1 − p)r2
Levent Kockesen (Koc University) Applications 41 / 1
Bank Runs
Break-even r2 is
r2 =(1− pr1)R
1− p
1 < r1 < 1/p ⇒ r2 < R
R = 2, p = 0.25, r1 = 1.28, r2 = 1.813◮ Gives zero profits to banks and consumers prefer bank deposits
At date 1 bank has 100 TL
If 25 of them withdraws, it pays out 25 × 1.28 = 32
Remaining 68 TL brings 68× 2 = 136 on date 2
Each of 75 depositor receives
136
75= 1.813
So everything works out fine if only 25 withdraw at date 1
Levent Kockesen (Koc University) Applications 42 / 1
Bank Runs
At date 1 all type 1 consumers choose to withdraw
Remaining fraction 1− p has to decide whether to withdraw or not◮ They can withdraw at date 1 and consume at date 2
This defines a game with 100 players
What are the Nash equilibria?
Let the equilibrium fraction who withdraw at date 1 be f
We must have f ≥ p
If fr1 > 1, the bank becomes insolvent and does not survive to date 2
If fr1 ≤ 1, at date 2 the bank pays out
r2(f) =(1− fr1)R
1− f
Levent Kockesen (Koc University) Applications 43 / 1
Bank Runs
Type 2 players have two options:◮ Withdraw at date 1 and get
r1(f) =
{r1, fr1 ≤ 11f , fr1 > 1
◮ or withdraw at date 2 and get
r2(f) =
{(1−fr1)R
1−f , fr1 ≤ 1
0, fr1 > 1
In any Nash equilibrium◮ They choose date 1 if r1(f) > r2(f)◮ and choose date 2 otherwise
Levent Kockesen (Koc University) Applications 44 / 1
Bank Runs
Assume 1 < r1 < 1/p, R > 1, and
r2 =(1− pr1)R
1− p
Then r2 > r1
We can show that
f = p is a Nash equilibrium.
f = p and pr1 < 1 imply
r2(f) = r2 > r1 = r1(f)
So, all type 2 consumers prefer to withdraw at date 2
This is an equilibrium without a bank run
Levent Kockesen (Koc University) Applications 45 / 1
Bank Runs
Claim
f = 1 is a Nash equilibirum.
r1 > 1 and f = 1 imply fr1 > 1
Therefore
r1(f) =1
f= 1 > 0 = r2(f)
So, all type 2 consumers prefer to withdraw at date 1
This is a bank-run equilibrium
f = 1 is a self-fulfilling prophecy
Levent Kockesen (Koc University) Applications 46 / 1
Bank Runs
f > f ⇒ r1(f) > r2(f) ⇒ Everybody withdraws at date 1
f < f ⇒ r1(f) < r2(f) ⇒ Only fraction p withdraws at date 1
f = 1 and f = p are the only Nash equilibria
f11/r1fp
r2(f)
r1(f)
Rr2
r1
1
Levent Kockesen (Koc University) Applications 47 / 1
Bank RunsA simpler model:
4 depositors, each with 1 TL, half need to consume on date 1
R = 4, r1 = 1.5, r2 = 2
Suppose on date 1, 2 depositors withdraw
Bank is left with 1 TL
This brings 4 TL on date 2, which the bank returns to the other twodepositorsWhat happens if 3 depositors withdraw on date 1?
◮ The bank goes bankrupt◮ They each get 4/3, the other gets nothing
What happens if 4 depositors withdraw on date 1?◮ The bank goes bankrupt◮ They each get 1
The game between two date 2 depositors
W NW 1, 1 4/3, 0N 0, 4/3 2, 2
Levent Kockesen (Koc University) Applications 48 / 1
Bank Runs
Starting from good equilibrium a rumor that the bank is not doingwell could cause a run
This is so even if most know it not to be true
You only need people believe that the others will withdraw
Levent Kockesen (Koc University) Applications 49 / 1
Bank Runs
How do you prevent bank runs?
Suspend convertibility of deposits to cash: force f ≤ p◮ before deposit insurance, banks regularly suspended convertibility to
stop runs
Deposit insurance: Government promises (r1, r2)◮ Does it with taxpayers’ money◮ May create incentive problems for banks
Levent Kockesen (Koc University) Applications 50 / 1
page.1
Introduction to Economic and Strategic BehaviorMixed Strategies
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Mixed Strategies 1 / 20
page.2
Matching Pennies
Player 1
Player 2H T
H −1, 1 1,−1T 1,−1 −1, 1
How would you play?
No solution?
You should try to be unpredictable
Choose randomly
Levent Kockesen (Koc University) Mixed Strategies 2 / 20
page.3
Audit
Tax authority wants to prevent tax evasion with minimal cost.
Decides whether to audit or not◮ Audit always catches a cheater◮ But costs 1
The tax payer decides whether to cheat or not◮ Cheating saves 20◮ If you get caught you pay a fine of 10◮ There is also non-monetary cost equivalent to 30
Tax Payer
TreasuryAudit No Audit
Cheat −20, 9 20, 0Not Cheat 0, 19 0, 20
Levent Kockesen (Koc University) Mixed Strategies 3 / 20
page.4
Audit
What is the set of Nash equilibria?
Tax Payer
TreasuryAudit No Audit
Cheat −20, 9 20, 0Not Cheat 0, 19 0, 20
Both parties want to be unpredictable
Levent Kockesen (Koc University) Mixed Strategies 4 / 20
page.5
Mixed Strategy Equilibrium
Tax Payer
TreasuryAudit No Audit
Cheat −20, 9 20, 0Not Cheat 0, 19 0, 20
A mixed strategy is a probability distribution over the set of actions.
Suppose the treasury audits half the cases◮ Probability of audit is 1/2
What should the tax payer do?◮ He is indifferent between cheating and not cheating◮ He is also indifferent over cheating with any probability
Suppose in general 10% of tax payers cheat◮ Probability of cheating is 1/10
What should the treasury do?◮ It is indifferent between audit or not◮ Also indifferent over auditing with any probability
Levent Kockesen (Koc University) Mixed Strategies 5 / 20
page.6
Mixed Strategy Equilibrium
Auditing with probability 1/2 and cheating with probability 1/10 is amixed strategy equilibrium
Both players are best responding to each other
Levent Kockesen (Koc University) Mixed Strategies 6 / 20
page.7
Mixed Strategy Equilibrium
In a mixed strategy equilibrium every action played with positiveprobability must be a best response to other players’ mixed strategies
In particular players must be indifferent between actions played withpositive probability
Tax payer’s probability of cheating p
Treasury’s probability of auditing q
Tax payer’s expected payoff to◮ Cheating is q × (−20) + (1− q)× 20 = 20− 40q◮ Not cheating is 0
Indifference condition0 = 20− 40q
or q = 1/2
Levent Kockesen (Koc University) Mixed Strategies 7 / 20
page.8
Mixed Strategy Equilibrium
The treasury’s expected payoff to◮ Audit is p× 9 + (1− p)× 19 = 19− 10p◮ Not is p× 0 + (1− p)× 20 = 20− 20p
Indifference condition
19 − 10p = 20− 20p
or p = 1/10
(p = 1/10, q = 1/2) is a mixed strategy equilibrium
Since there is no pure strategy equilibrium, this is also the unique Nashequilibrium
Levent Kockesen (Koc University) Mixed Strategies 8 / 20
page.9
Audit
What determines cheating rate?
Let auditing cost be c
Fine if caught cheating f
Other costs of cheating r
Tax Payer
TreasuryAudit No Audit
Cheat 20− f − r, f − c 20, 0Not Cheat 0, 20 − c 0, 20
Levent Kockesen (Koc University) Mixed Strategies 9 / 20
page.10
Audit
Tax Payer
TreasuryAudit (q) No Audit (1-q)
Cheat (p) 20− f − r, f − c 20, 0Not Cheat (1-p) 0, 20 − c 0, 20
f + r > 20◮ Otherwise always cheats
c < f◮ Otherwise never audits
Equilibrium condition I
p(f − c) + (1− p)(20 − c) = (1− p)20 ⇒ p =c
f
Equilibrium condition II
q(20 − f − r) + (1− q)20 = 0 ⇒ q =20
f + r
Levent Kockesen (Koc University) Mixed Strategies 10 / 20
page.11
Audit
Cheating rate is c/f◮ Increase the fine◮ Decrease the cost of audit◮ Other costs of getting caught has no effect!
Audit rate is 20/(f + r)◮ Increase the total penalty of cheating◮ Increase audit rate as the amount of cheating increases
Levent Kockesen (Koc University) Mixed Strategies 11 / 20
page.12
Mixed and Pure Strategy Equilibria
How do you find the set of all (pure and mixed) Nash equilibria?
In 2× 2 games we can use the best response correspondences interms of the mixed strategies and plot them
Consider the Battle of the Sexes game
Player 1
Player 2B S
B 2, 1 0, 0S 0, 0 1, 2
Denote Player 1’s strategy as p and that of Player 2 as q (probabilityof choosing B)
Levent Kockesen (Koc University) Mixed Strategies 12 / 20
page.13
B SB 2, 1 0, 0S 0, 0 1, 2
What is Player 1’s best response?
Expected payoff to◮ B is 2q◮ S is 1− q
If 2q > 1− q or q > 1/3◮ best response is B (or equivalently p = 1)
If 2q < 1− q or q < 1/3◮ best response is S (or equivalently p = 0)
If 2q = 1− q or q = 1/3◮ he is indifferent◮ best response is any p ∈ [0, 1]
Player 1’s best response correspondence:
B1(q) =
{1} , if q > 1/3
[0, 1], if q = 1/3
{0} , if q < 1/3
Levent Kockesen (Koc University) Mixed Strategies 13 / 20
page.14
B SB 2, 1 0, 0S 0, 0 1, 2
What is Player 2’s best response?
Expected payoff to◮ B is p◮ S is 2(1− p)
If p > 2(1− p) or p > 2/3◮ best response is B (or equivalently q = 1)
If p < 2(1− p) or p < 2/3◮ best response is S (or equivalently q = 0)
If p = 2(1− p) or p = 2/3◮ she is indifferent◮ best response is any q ∈ [0, 1]
Player 2’s best response correspondence:
B2(p) =
{1} , if p > 2/3
[0, 1], if p = 2/3
{0} , if p < 2/3
Levent Kockesen (Koc University) Mixed Strategies 14 / 20
page.15
B1(q) =
{1} , if q > 1/3
[0, 1], if q = 1/3
{0} , if q < 1/3
B2(p) =
{1} , if p > 2/3
[0, 1], if p = 2/3
{0} , if p < 2/3
q
p1
1
0 2/3
1/3
b
b
b
B1(q)
B2(p)
Set of Nash equilibria
{(0, 0), (1, 1), (2/3, 1/3)}
Levent Kockesen (Koc University) Mixed Strategies 15 / 20
page.16
Dominated Actions and Mixed Strategies
Up to now we tested actions only against other actions
An action may be undominated by any other action, yet bedominated by a mixed strategy
Consider the following game
L RT 1, 1 1, 0M 3, 0 0, 3B 0, 1 4, 0
No action dominates T
But playing M and B with 1/2 probability each strictly dominates T
A strictly dominated action is never used with positive probability in amixed strategy equilibrium
Levent Kockesen (Koc University) Mixed Strategies 16 / 20
page.17
Dominated Actions and Mixed Strategies
An easy way to figure out dominated actions is to compare expectedpayoffsLet player 2’s mixed strategy given by q = prob(L)
L RT 1, 1 1, 0M 3, 0 0, 3B 0, 1 4, 0
u1(T, q) = 1
u1(M, q) = 3q
u1(B, q) = 4(1 − q)
0
1
2
3
4
0 14/7
12/7
q
u1(., q)
u1(T, q)
u1(M, q)
u1(B, q)An action is a never bestresponse if there is no beliefthat makes that action a bestresponseT is a never best responseAn action is a NBR iff it isstrictly dominated
Levent Kockesen (Koc University) Mixed Strategies 17 / 20
page.18
What if there are no strictly dominated actions?
L RT 2, 0 2, 1M 3, 3 0, 0B 0, 1 3, 0
Denote player 2’s mixed strategy by q = prob(L)u1(T, q) = 2, u1(M, q) = 3q, u1(B, q) = 3(1 − q)
q
u1(., q)
1/2
3/2
2/31/3 1
3
2
0
u1(T, q)
u1(M, q)u1(B, q)
Pure strategy Nash eq. (M,L)
Mixed strategy equilibria?◮ Only one player mixes? Not possible◮ Player 1 mixes over {T,M,B}? Not possible◮ Player 1 mixes over {M,B}? Not possible◮ Player 1 mixes over {T,B}? Let p = prob(T )
q = 1/3, 1− p = p → p = 1/2◮ Player 1 mixes over {T,M}? Let p = prob(T )
q = 2/3, 3(1− p) = p → p = 3/4
Levent Kockesen (Koc University) Mixed Strategies 18 / 20
page.19
Real Life Examples?
Ignacio Palacios-Huerta (2003): 5 years’ worth of penalty kicks
Empirical scoring probabilities
L RL 58, 42 95, 5R 93, 7 70, 30
R is the natural side of the kicker
What are the equilibrium strategies?
Levent Kockesen (Koc University) Mixed Strategies 19 / 20
page.20
Penalty Kick
L RL 58, 42 95, 5R 93, 7 70, 30
Kicker must be indifferent
58p + 95(1 − p) = 93p + 70(1 − p) ⇒ p = 0.42
Goal keeper must be indifferent
42q + 7(1− q) = 5q + 30(1 − q) ⇒ q = 0.39
Theory Data
Kicker 39% 40%Goallie 42% 42%
Also see Walker and Wooders (2001): Wimbledon
Levent Kockesen (Koc University) Mixed Strategies 20 / 20
Introduction to Economic and Strategic BehaviorStrategic Form Games with Incomplete Information
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Bayesian Games 1 / 1
Games with Incomplete Information
Some players have incomplete information about some components ofthe game
◮ Firm does not know rival’s cost◮ Bidder does not know valuations of other bidders in an auction
We could also say some players have private information
Levent Kockesen (Koc University) Bayesian Games 2 / 1
Games with Incomplete Information
Does incomplete information matter?
Suppose you make an offer to buy out a company
If the value of the company is V it is worth 1.5V to you
The seller accepts only if the offer is at least V
If you know V what do you offer?
You know only that V is uniformly distributed over [0, 100]. Whatshould you offer?
Enter your name and your bid
Click here for the EXCEL file
Levent Kockesen (Koc University) Bayesian Games 3 / 1
Bayesian Games
We will first look at incomplete information games where playersmove simultaneously
◮ Bayesian games
Later on we will study dynamic games of incomplete information
Levent Kockesen (Koc University) Bayesian Games 4 / 1
Bayesian Games
What is new in a Bayesian game?
Each player has a type: summarizes a player’s private information
Players’ payoffs depend on types
Incomplete information can be anything about the game◮ Payoff functions◮ Actions available to others◮ Beliefs of others; beliefs of others’ beliefs of others’...
Harsanyi (1967-1968) showed that introducing types in payoffs isadequate
Each player has beliefs about others’ types
Levent Kockesen (Koc University) Bayesian Games 5 / 1
Bayesian Equilibrium
Since payoffs depend on types different types of the same player mayplay different strategies
We assume that players maximize their expected payoffs.
Bayesian Equilibrium
A strategy profile is a Bayesian equilibrium if each type of each playermaximizes her expected payoff given other players’ strategies.
Levent Kockesen (Koc University) Bayesian Games 6 / 1
An Example: Hawk-Dove
Two animals are trying to share a prey
They can be hawkish (H) or dovish (D)
Value of the prey = 2
There is a fight if both of them choose to be hawkish
If they get into a fight they both suffer injuries◮ Cost to player 1= 2◮ Cost to player 2= c2
If only one is hawkish, he gets the entire prey
Otherwise they share
H DH −1, 1 − c2 2, 0D 0, 2 1, 1
c2 can be 0 (tough type) or 2 (weak type)◮ Pl. 2 knows◮ Pl. 1 believes it is 2 with probability q
Levent Kockesen (Koc University) Bayesian Games 7 / 1
Hawk-Dove
H DH −1,−1 2, 0D 0, 2 1, 1
Weak (q)
H DH −1, 1 2, 0D 0, 2 1, 1
Tough (1− q)
Player 1 chooses H or D
Player 2’s strategy may depend on type◮ If both do the same: pooling◮ If different types play differently: separating
Four possible types of pure strategy Bayesian equilibria◮ Separating: (W : H,T : D) or (W : D,T : H)◮ Pooling: (W : H,T : H) or (W : D,T : D)
H is strictly dominant for tough type → only two possibilities
Levent Kockesen (Koc University) Bayesian Games 8 / 1
Hawk-Dove: Separating EquilibriumH D
H −1,−1 2, 0D 0, 2 1, 1
Weak (q)
H DH −1, 1 2, 0D 0, 2 1, 1
Tough (1− q)
(W : D,T : H)
For Weak to play D, Pl. 1 must be playing HExpected payoff to H
q × 2 + (1− q)× (−1) = 3q − 1
Expected payoff to D
q × 1 + (1− q)× 0 = q
H is optimal if
3q − 1 ≥ q or q ≥ 1
2
1 plays H and 2 plays (W : D,T : H) is a Bayesian equilibrium iff q ≥ 1/2.
Levent Kockesen (Koc University) Bayesian Games 9 / 1
Hawk-Dove: Pooling Equilibrium
H DH −1,−1 2, 0D 0, 2 1, 1
Weak (q)
H DH −1, 1 2, 0D 0, 2 1, 1
Tough (1− q)
(W : H,T : H)
For Weak to play H, Pl. 1 must be playing D
Expected payoff to H is −1
Expected payoff to D is 0
So D is optimal
1 plays D and 2 plays (W : H,T : H) is a Bayesian equilibrium.
Levent Kockesen (Koc University) Bayesian Games 10 / 1
Auctions
Many economic transactions are conducted through auctions
treasury bills
foreign exchange
publicly owned companies
mineral rights
airwave spectrum rights
art work
antiques
cars
houses
government contracts
Also can be thought of as auctions
takeover battles
queues
wars of attrition
lobbying contests
Levent Kockesen (Koc University) Bayesian Games 11 / 1
Auction Formats
1. Open bid auctions1.1 ascending-bid auction
⋆ aka English auction⋆ price is raised until only one bidder remains, who wins and pays the
final price
1.2 descending-bid auction⋆ aka Dutch auction⋆ price is lowered until someone accepts, who wins the object at the
current price
2. Sealed bid auctions2.1 first price auction
⋆ highest bidder wins; pays her bid
2.2 second price auction⋆ aka Vickrey auction⋆ highest bidder wins; pays the second highest bid
Levent Kockesen (Koc University) Bayesian Games 12 / 1
Auction Formats
Auctions also differ with respect to the valuation of the bidders
1. Private value auctions◮ each bidder knows only her own value◮ artwork, antiques, memorabilia
2. Common value auctions◮ actual value of the object is the same for everyone◮ bidders have different private information about that value◮ oil field auctions, company takeovers
Levent Kockesen (Koc University) Bayesian Games 13 / 1
Independent Private Values
Each bidder knows only her own valuation
Valuations are independent across bidders
Bidders have beliefs over other bidders’ values
Risk neutral bidders◮ If the winner’s value is v and pays p, her payoff is v − p
Levent Kockesen (Koc University) Bayesian Games 14 / 1
Second Price Auctions
Only one 4G license will be sold
There are 10 groups
I generated 10 random values between 0 and 100
I will now distribute the values: Keep these and don’t show it toanyone until the end of the experiment
I will now distribute paper slips where you should enter your name,value, and bid
Highest bidder wins, pays the second highest bid
I will pay the winner her net payoff: value - priceClick here for the EXCEL file
Levent Kockesen (Koc University) Bayesian Games 15 / 1
Second Price Auctions
I. Bidding your value weakly dominates bidding higher
Suppose your value is $10 but you bid $15. Three cases:
1. There is a bid higher than $15 (e.g. $20)◮ You loose either way: no difference
2. 2nd highest bid is lower than $10 (e.g. $5)◮ You win either way and pay $5: no difference
3. 2nd highest bid is between $10 and $15 (e.g. $12)◮ You loose with $10: zero payoff◮ You win with $15: loose $2
5
10 value
12
15 bid
20
Levent Kockesen (Koc University) Bayesian Games 16 / 1
Second Price Auctions
II. Bidding your value weakly dominates bidding lower
Suppose your value is $10 but you bid $5. Three cases:
1. There is a bid higher than $10 (e.g. $12)◮ You loose either way: no difference
2. 2nd highest bid is lower than $5 (e.g. $2)◮ You win either way and pay $2: no difference
3. 2nd highest bid is between $5 and $10 (e.g. $8)◮ You loose with $5: zero payoff◮ You win with $10: earn $2
2
10 value
8
5 bid
12
Levent Kockesen (Koc University) Bayesian Games 17 / 1
English Auction
Suppose you value the item at 100 TL
What is your optimal strategy?
Stay in bidding until the price exceeds 100 TL
This is a dominant strategy
If everyone plays this strategy what happens?◮ The bidder with highest value wins◮ Pays something close to second highest value
Levent Kockesen (Koc University) Bayesian Games 18 / 1
First Price Auctions
Only one 4G license will be sold
There are 10 groups
I generated 10 random values between 0 and 100
I will now distribute the values: Keep these and don’t show it toanyone until the end of the experiment
I will now distribute paper slips where you should enter your name,value, and bid
Highest bidder wins, pays her bid
I will pay the winner her net payoff: value - priceClick here for the EXCEL file
Levent Kockesen (Koc University) Bayesian Games 19 / 1
First Price Auctions
Would you bid your value?
What happens if you bid less than your value?◮ You get a positive payoff if you win◮ But your chances of winning are smaller◮ Optimal bid reflects this tradeoff
Bidding less than your value is known as bid shading
Choose your bid b to maximize
π = (v − b) prob(win)
Probability of winning depends on your bid and others’ bids
Given what you believe about the others, it is increasing in your bid
Levent Kockesen (Koc University) Bayesian Games 20 / 1
Example
Suppose your value is 6 and the highest possible value is 10
b v − b prob π
10 -4 1 -49 -3 0.9 -2.78 -2 0.8 -1.67 -1 0.7 -0.76 0 0.6 05 1 0.5 0.54 2 0.4 0.83 3 0.3 0.92 4 0.2 0.81 5 0.1 0.5
Levent Kockesen (Koc University) Bayesian Games 21 / 1
Example
0
1
−1
−2
−3
−4
−5
1 2 3 4 5 6 7 8 9 10
bb b b
b
b
b
b
b
b
Levent Kockesen (Koc University) Bayesian Games 22 / 1
Equivalent Formats
English auction has the same equilibrium as Second Price auction
This is true only if values are private
Stronger equivalence between Dutch and First Price auctions
Open Bid Sealed Bid
Dutch Auction First Price
EnglishAuction
Second Price
Strategically Equivalent
Same Equilibrium
in Private Values
Levent Kockesen (Koc University) Bayesian Games 23 / 1
Which One Brings More Revenue?
Second Price◮ Bidders bid their value◮ Revenue = second highest bid
First Price◮ Bidders bid less than their value◮ Revenue = highest bid
Which one is better?
Turns out it doesn’t matter
Levent Kockesen (Koc University) Bayesian Games 24 / 1
Revenue Equivalence Theorem
Any auction with independent private values with a common distributionin which
1. the number of the bidders are the same and the bidders arerisk-neutral,
2. the object always goes to the buyer with the highest value,
3. the bidder with the lowest value expects zero surplus,
yields the same expected revenue.
Levent Kockesen (Koc University) Bayesian Games 25 / 1
Common Value Auctions
Value of 4G license = v◮ Uniform between 0 and 100
You observe v + x◮ x ∼ N(0, 10)◮ x is between −20 and 20 with 95% prob.
I generated a signal for each of 10 groups
I will now distribute the values: Keep these and don’t show it toanyone until the end of the experiment
I will now distribute paper slips where you should enter your name,value, and bid
Highest bidder wins, pays the second highest bid
I will pay the winner her net payoff: value - priceClick here for the EXCEL file
Levent Kockesen (Koc University) Bayesian Games 26 / 1
Common Value Auctions and Winner’s Curse
If everybody bids her estimate winning is “bad news”
This is known as Winner’s Curse
Optimal strategies are complicated
Bidders bid much less than their value to prevent winner’s curse
To prevent winner’s curse
Base your bid on expected value conditional on winning
Auction formats are not equivalent in common value auctions
Open bid auctions provide information and ameliorates winner’s curse◮ Bids are more aggressive
Sealed bid auctions do not provide information◮ Bids are more conservative
Levent Kockesen (Koc University) Bayesian Games 27 / 1
Auction Design: Failures
New Zeland Spectrum Auction (1990)◮ Used second price auction with no reserve price◮ Estimated revenue NZ$ 240 million◮ Actual revenue NZ$36 million
Some extreme cases
Winning Bid Second Highest BidNZ$100,000 NZ$6,000
NZ$7,000,000 NZ$5,000NZ$1 None
Source: John McMillan, “Selling Spectrum Rights,” Journal of Economic Perspectives, Summer 1994
Problems◮ Second price format politically problematic
⋆ Public sees outcome as selling for less than its worth
◮ No reserve price
Levent Kockesen (Koc University) Bayesian Games 28 / 1
Auction Design: FailuresAustralian TV Licence Auction (1993)
◮ Two satellite-TV licences◮ Used first price auction◮ Huge embarrasment
High bidders had no intention of paying
They bid high just to guarantee winning
They also bid lower amounts at A$5 million intervalsThey defaulted
◮ licences had to be re-awarded at the next highest bid◮ those bids were also theirs
Outcome after a series of defaults
Initial Bid Final PriceA$212 mil. A$117 mil.A$177 mil. A$77 mil.
Source: John McMillan, “Selling Spectrum Rights,” Journal of Economic Perspectives, Summer 1994
Problem: No penalty for default
Levent Kockesen (Koc University) Bayesian Games 29 / 1
Auction Design: Failures
Turkish GSM licence auction
April 2000: Two GSM 1800 licences to be auctioned
Auction method:
1. Round 1: First price sealed bid auction2. Round 2: First price sealed bid auction with reserve price
⋆ Reserve price is the winning bid of Round 1
Bids in the first roundBidder Bid AmountIs-Tim $2.525 bil.
Dogan+ $1.350 bil.Genpa+ $1.224 bil.Koc+ $1.207 bil.Fiba+ $1.017 bil.
Bids in the second round: NONE!
Problem: Facilitates entry deterrence
Levent Kockesen (Koc University) Bayesian Games 30 / 1
Auction Design
Good design depends on objective◮ Revenue◮ Efficiency◮ Other
One common objective is to maximize expected revenue
In the case of private independent values with the same number ofrisk neutral bidders format does not matter
Auction design is a challenge when◮ values are correlated◮ bidders are risk averse
Other design problems◮ collusion◮ entry deterrence◮ reserve price
Levent Kockesen (Koc University) Bayesian Games 31 / 1
Auction Design
Correlated values: Ascending bid auction is better
Risk averse bidders◮ Second price auction: risk aversion does not matter◮ First price auction: higher bids
Collusion: Sealed bid auctions are better to prevent collusion
Entry deterrence: Sealed bid auctions are better to promote entry
A hybrid format, such as Anglo-Dutch Auction, could be better.
Anglo-Dutch auction has two stages:
1. Ascending bid auction until only two bidders remain
2. Two remaining bidders make offers in a first price sealed bid auction
Levent Kockesen (Koc University) Bayesian Games 32 / 1
page.1
Introduction to Economic and Strategic BehaviorExtensive Form Games
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Extensive Form Games 1 / 23
page.2
Extensive Form Games
Strategic form games are used to model situations in which playerschoose strategies without knowing the strategy choices of the otherplayers
In some situations players observe other players’ moves before theymove
Removing Coins:◮ There are 21 coins◮ Two players move sequentially and remove 1, 2, or 3 coins◮ Winner is who removes the last coin(s)◮ We will determine the first mover by a coin toss◮ Volunteers?
Levent Kockesen (Koc University) Extensive Form Games 2 / 23
page.3
Extensive Form Games
Strategic form has three ingredients:◮ set of players◮ sets of actions◮ payoff functions
Extensive form games provide more information◮ order of moves◮ actions available at different points in the game◮ information available throughout the game
Easiest way to represent an extensive form game is to use a game tree
Levent Kockesen (Koc University) Extensive Form Games 3 / 23
page.4
Entry Game
Kodak is contemplating entering the instant photography market andPolaroid can either fight the entry or accommodate
K
P
Out In
F A
0, 20
−5, 0 10, 10
Levent Kockesen (Koc University) Extensive Form Games 4 / 23
page.5
Game Trees
What’s in a game tree?
nodes◮ decision nodes◮ initial node◮ terminal nodes
branches
player labels
action labels
payoffs
information sets◮ to be seen later
K
P
Out In
F A
0, 20
−5, 0 10, 10
Levent Kockesen (Koc University) Extensive Form Games 5 / 23
page.6
Extensive Form Game Strategies
A pure strategy of a player specifies an action choice at each decision nodeof that player
K
P
Out In
F A
0, 20
−5, 0 10, 10
Kodak’s strategies◮ SK = {Out, In}
Polaroid’s strategies◮ SP = {F,A}
Levent Kockesen (Koc University) Extensive Form Games 6 / 23
page.7
Backward Induction Equilibrium
What should Polaroid do ifKodak enters?
Given what it knows aboutPolaroid’s response to entry,what should Kodak do?
This is an example of abackward induction equilibrium
K
P
Out In
F A
0, 20
−5, 0 10, 10
At a backward induction equilibrium each player plays optimally atevery decision node in the game tree
◮ plays a sequentially rational strategy
(In,A) is the unique backward induction equilibrium of the entrygame
Levent Kockesen (Koc University) Extensive Form Games 7 / 23
page.8
Grab the Money
Two piles of money on the table◮ One has 20 Kr. the other 0
Player 1 begins◮ Grab one pile (the other goes to Player 2)◮ Pass
Grab → game ends
Pass → I add 10 Kr. to BOTH piles
Player 2 moves: Grab or Pass
This continues 4 rounds
Two volunteers?
Levent Kockesen (Koc University) Extensive Form Games 8 / 23
page.9
Grab the Money
1 2 1 2 1 2 1 210, 8
2, 0 1, 3 4, 2 3, 5 6, 4 5, 7 8, 6 7, 9
G
P
G
P
G
P
G
P
G
P
G
P
G
P
G
P
How many strategies does player 1 have?
What is the backward induction equilibrium outcome?◮ G
What is the backward induction equilibrium?◮ (GGGG,GGGG)
Levent Kockesen (Koc University) Extensive Form Games 9 / 23
page.10
Backward Induction
Backward Induction Algorithm
1. Go to the last decision node(s)◮ Find the optimal strategies there
2. Go to penultimate decision node(s)◮ Find the optimal strategies there
3. Repeat until you reach the first decision node
Levent Kockesen (Koc University) Extensive Form Games 10 / 23
page.11
Pirates
5 pirates are trying to split 100 gold coins
Oldest proposes an allocation
All vote for or against
If at least half vote for it, it is shared that way
Otherwise the proposing pirate is thrown overboard
Repeated with the pirates that remain
They like to live and like gold
If you are the oldest, what do you propose?
Levent Kockesen (Koc University) Extensive Form Games 11 / 23
page.12
Pirates
1 2 3 4 5
100 099 0 1
99 0 1 098 0 1 0 1
Levent Kockesen (Koc University) Extensive Form Games 12 / 23
page.13
Strategic Form of an Extensive Form Game
If you want to apply a strategic form solution concept◮ Nash equilibrium◮ Dominant strategy equilibrium◮ IEDS
Analyze the strategic form of the game
Strategic form of an extensive form game
1. Set of playersand for each player
2. Strategies
3. Payoffs
Levent Kockesen (Koc University) Extensive Form Games 13 / 23
page.14
Strategic Form of an Extensive Form Game
K
P
Out In
F A
0, 20
−5, 0 10, 10
1. Players: K and P
2. Strategies:SK = {Out, In}, SP = {F,A}
3. Payoffs in the bimatrix
K
PF A
Out 0, 20 0, 20In −5, 0 10, 10
Levent Kockesen (Koc University) Extensive Form Games 14 / 23
page.15
Nash Equilibria of an Extensive Form Game
K
PF A
Out 0, 20 0, 20In −5, 0 10, 10
K
P
Out In
F A
0, 20
−5, 0 10, 10
Set of Nash equilibria = {(In,A), (Out, F )}(Out, F ) is sustained by an incredible threat by Polaroid
Backward induction equilibrium eliminates equilibria based uponincredible threats
Nash equilibrium requires rationality
Backward induction requires sequential rationality◮ Players must play optimally at every point in the game
Levent Kockesen (Koc University) Extensive Form Games 15 / 23
page.16
Extensive Form Games with Imperfect Information
We have seen extensive form games with perfect information◮ Every player observes the previous moves made by all the players
What happens if some of the previous moves are not observed?
Consider the following game between Kodak and Polaroid
Kodak doesn’t know whetherPolaroid will fight oraccommodate
The dotted line is aninformation set:
◮ a collection of decision nodesthat cannot be distinguishedby the player
We cannot determine theoptimal action for Kodak atthat information set
K
P
K K
Out In
F A
F A F A
0, 20
−5,−5 5, 15 15, 5 10, 10
Levent Kockesen (Koc University) Extensive Form Games 16 / 23
page.17
Subgame Perfect EquilibriumWe will introduce another solution concept: Subgame Perfect Equilibrium
Definition
A subgame is a part of the game tree such that
1. it starts at a single decision node,
2. it contains every successor to this node,
3. if it contains a node in an information set, then it contains all thenodes in that information set.
This is a subgame
P
K K
F A
F A F A
−5,−5 5, 15 15, 5 10, 10
This is not a subgame
P
K
A
F A
15, 5 10, 10
Levent Kockesen (Koc University) Extensive Form Games 17 / 23
page.18
Subgame Perfect Equilibrium
Extensive form game strategies
A pure strategy of a player specifies an action choice at each informationset of that player
Definition
A strategy profile in an extensive form game is a subgame perfectequilibrium (SPE) if it induces a Nash equilibrium in every subgame of thegame.
To find SPE
1. Find the Nash equilibria of the “smallest” subgame(s)
2. Fix one for each subgame and attach payoffs to its initial node
3. Repeat with the reduced game
Levent Kockesen (Koc University) Extensive Form Games 18 / 23
page.19
Subgame Perfect Equilibrium
Consider the following game
K
P
K K
Out In
F A
F A F A
0, 20
−5,−5 5, 8 8, 5 10, 10
Levent Kockesen (Koc University) Extensive Form Games 19 / 23
page.20
Subgame Perfect Equilibrium
The “smallest” subgame
P
K K
F A
F A F A
−5,−5 5, 8 8, 5 10, 10
Its strategic form
K
PF A
F −5,−5 8, 5A 5, 8 10, 10
Nash equilibrium of the subgame is (A,A)
Reduced subgame isK
Out In
0, 20 10, 10
Its unique Nash equilibrium is (In)
Therefore the unique SPE of the game is ((In,A),A)
Levent Kockesen (Koc University) Extensive Form Games 20 / 23
page.21
Role of Chance
In some games chance also plays a role◮ Games like poker, bridge, etc.◮ But also in many economic or social situations
We introduce chance as another “player”
Chooses her action randomly◮ Using exogenously given probabilities
Levent Kockesen (Koc University) Extensive Form Games 21 / 23
page.22
Role of Chance: An Example
Elites choose income distribution: Equitable (e) or Unequitable (u)
If unequitable, People decide whether to revolt (r) or not (n)
Exogenous factors ⇒ revolution is successful (s) or a failure (f)◮ Probability of success is p
Revolution is costly◮ Destroys half the income available
See Acemoglu and Robinson (2006) book
Levent Kockesen (Koc University) Extensive Form Games 22 / 23
page.23
Role of Chance: An Example
Backward Induction
Last real decision node is P
Expected payoff to r is 5p◮ r if p > 0.2◮ n otherwise
Move up to E◮ e if p > 0.2◮ u otherwise
E
5, 5
P
C
9, 1
0, 5 5, 0
u e
r n
s f
Backward Induction Equilibrium
1. If p > 0.2 (e, r)
2. If p ≤ 0.2 (u, n)
Levent Kockesen (Koc University) Extensive Form Games 23 / 23
page.1
Introduction to Economic and Strategic BehaviorExtensive Form Games: Applications
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Extensive Form Games: Applications 1 / 47
page.2
Entry Game
Consider the following entry game
K
P
0, 100
−50, 20 50, 50
n e
f a
Levent Kockesen (Koc University) Extensive Form Games: Applications 2 / 47
page.3
Commitment
K
P
0, 100
−50, 20 50, 50
n e
f a
Backward induction equilibrium (e, a)◮ Polaroid’s payoff is 50
Suppose Polaroid commits to fight (f)if entry occurs.
What would Kodak do?
Kodak would not enter and Polaroid would be better off
Is this commitment credible?
Levent Kockesen (Koc University) Extensive Form Games: Applications 3 / 47
page.4
A U.S. air force base commander ordersthirty four B-52’s to launch a nuclearattack on Soviet Union
He shuts off all communications withthe planes and with the base
U.S. president invites the Russianambassador to the war room andexplains the situation
They decide to call the Russianpresident Dimitri
Levent Kockesen (Koc University) Extensive Form Games: Applications 4 / 47
page.5
Dr. Strangelove
What is the outcome if the U.S. doesn’tknow the existence of the doomsdaydevice?
What is the outcome if it does?
Commitment must be observable
What if USSR can un-trigger thedevice?
Commitment must be irreversible
US
USSR
0, 0
−4,−4 1,−2
Attack Don’t
Retaliate Don’t
Levent Kockesen (Koc University) Extensive Form Games: Applications 5 / 47
page.6
Commitment
Thomas Schelling
The power to constrain an adversary depends upon the power to bindoneself.
Levent Kockesen (Koc University) Extensive Form Games: Applications 6 / 47
page.7
Credible Commitments: Burning Bridges
In non-strategic environments having more options is never worse
Not so in strategic environments
You can change your opponent’s actions by removing some of youroptions
1066: William the Conqueror ordered his soldiers to burn their shipsafter landing to prevent his men from retreating
1519: Hernan Cortes sank his ships after landing in Mexico for thesame reason
Sun-tzu in The Art of War, 400 BC
At the critical moment, the leader of an army acts like one who hasclimbed up a height, and then kicks away the ladder behind him.
Levent Kockesen (Koc University) Extensive Form Games: Applications 7 / 47
page.8
Commitment
How can you achieve credible commitment in the entry game?
Change the game◮ Change payoffs◮ Remove accommodate as an option
Change others’ beliefs about your payoffs or options◮ Build reputation for toughness◮ Delegate to a tough CEO
Levent Kockesen (Koc University) Extensive Form Games: Applications 8 / 47
page.9
Change Payoffs
Suppose you can invest in a new production facility
Profitable only if you sell enough
P
K
P
0, 100
−50, 20 50, 50
n e
f a
K
P
0, 90
−50, 10 50, 5
n e
f a
N I
What is the backward induction equilibrium?
Levent Kockesen (Koc University) Extensive Form Games: Applications 9 / 47
page.10
Delegation
Delegate the decision to a CEO who is known to be aggressive
Or tie his compensation to market share◮ Pay 10 if market share is above 50%◮ Otherwise pay 5
K
CEO
0, 90
−50, 10 50, 5
n e
f a
What is the BIE?
Contract must be observable, non-renegotiable
Levent Kockesen (Koc University) Extensive Form Games: Applications 10 / 47
page.11
Reputation
You might be playing this game repeatedly in different markets
You may want to build reputation for being tough
Proper analysis requires a model with incomplete information
But intuition is clear
Suppose other firms believe you are tough with probability p
And if you are tough you always fight entry
Payoff to entry is
−50p+ 50(1 − p) = 50 − 100p
If they believe you are more likely to be tough they will not enter
You might fight initially to increase the belief that you are tough evenif you are not
Acting “irrational” could be rational
Levent Kockesen (Koc University) Extensive Form Games: Applications 11 / 47
page.12
A Simple Game
You have 10 TL to share
A makes an offer◮ x for me and 10− x for you
If B accepts◮ A’s offer is implemented
If B rejects◮ Both get zero
Half the class will play A (proposer) and half B (responder)◮ Proposers should write how much they offer to give responders◮ I will distribute them randomly to responders
⋆ They should write Yes or No
Click here for the EXCEL file
Levent Kockesen (Koc University) Extensive Form Games: Applications 12 / 47
page.13
Ultimatum Bargaining
Two players, A and B, bargainover a cake of size 1
Player A makes an offerx ∈ [0, 1] to player B
If player B accepts the offer(Y ), agreement is reached
◮ A receives x◮ B receives 1− x
If player B rejects the offer (N)both receive zero
A
B
x, 1− x 0, 0
x
Y N
Levent Kockesen (Koc University) Extensive Form Games: Applications 13 / 47
page.14
Subgame Perfect Equilibrium of Ultimatum BargainingWe can use backward induction
B’s optimal action◮ x < 1 → accept◮ x = 1 → accept or reject
1. Suppose in equilibrium B accepts any offer x ∈ [0, 1]◮ What is the optimal offer by A? x = 1◮ The following is a SPE
x∗ = 1
s∗B(x) = Y for all x ∈ [0, 1]
2. Now suppose that B accepts if and only if x < 1◮ What is A’s optimal offer?
⋆ x = 1?⋆ x < 1?
Unique SPE
x∗ = 1, s∗B(x) = Y for all x ∈ [0, 1]
Levent Kockesen (Koc University) Extensive Form Games: Applications 14 / 47
page.15
Bargaining
Bargaining outcomes depend on many factors◮ Social, historical, political, psychological, etc.
Early economists thought the outcome to be indeterminate
John Nash introduced a brilliant alternative approach◮ Axiomatic approach: A solution to a bargaining problem must satisfy
certain “reasonable” conditions⋆ These are the axioms
◮ How would such a solution look like?◮ This approach is also known as cooperative game theory
Later non-cooperative game theory helped us identify critical strategicconsiderations
Levent Kockesen (Koc University) Extensive Form Games: Applications 15 / 47
page.16
BargainingTwo individuals, A and B, are trying to share a cake of size 1If A gets x and B gets y,utilities are uA(x) and uB(y)If they do not agree, A gets utility dA and B gets dBWhat is the most likely outcome?
uA
uB
1
1
dA
dB
Levent Kockesen (Koc University) Extensive Form Games: Applications 16 / 47
page.17
BargainingLet’s simplify the problem
uA(x) = x, and uB(x) = xdA = dB = 0A and B are the same in every other respectWhat is the most likely outcome?
uA
uB
1
1
(dA, dB)
45◦
0.5
0.5
Levent Kockesen (Koc University) Extensive Form Games: Applications 17 / 47
page.18
BargainingHow about now? dA = 0.3, dB = 0.4
uA
uB
1
1
0.3
0.4
45◦
0.45
0.55
Let x be A’s share. Then
Slope = 1 =1− x− 0.4
x− 0.3
or x = 0.45
So A gets 0.45 and B gets 0.55
Levent Kockesen (Koc University) Extensive Form Games: Applications 18 / 47
page.19
Bargaining
In general A gets
dA +1
2(1− dA − dB)
B gets
dB +1
2(1− dA − dB)
But why is this reasonable?
Two answers:
1. Axiomatic: Nash Bargaining Solution2. Non-cooperative: Alternating offers bargaining game
Levent Kockesen (Koc University) Extensive Form Games: Applications 19 / 47
page.20
Bargaining: Axiomatic Approach
John Nash (1950): The Bargaining Problem, Econometrica1. Efficiency
⋆ No waste
2. Symmetry⋆ If bargaining problem is symmetric, shares must be equal
3. Scale Invariance⋆ Outcome is invariant to linear changes in the payoff scale
4. Independence of Irrelevant Alternatives⋆ If you remove alternatives that would not have been chosen, the
solution does not change
Levent Kockesen (Koc University) Extensive Form Games: Applications 20 / 47
page.21
Nash Bargaining Solution
What if parties have different bargaining powers?
Remove symmetry axiom
Then A gets
xA = dA + α(1− dA − dB)
B gets
xB = dB + β(1− dA − dB)
α, β > 0 and α+ β = 1 represent bargaining powers
If dA = dB = 0
xA = α and xB = β
Levent Kockesen (Koc University) Extensive Form Games: Applications 21 / 47
page.22
Alternating Offers Bargaining
Two players, A and B, bargain over a cake of size 1
At time 0, A makes an offer xA ∈ [0, 1] to B◮ If B accepts, A receives xA and B receives 1− xA
◮ If B rejects, then
at time 1, B makes a counteroffer xB ∈ [0, 1]◮ If A accepts, B receives xB and A receives 1− xB
◮ If A rejects, he makes another offer at time 2
This process continues indefinitely until a player accepts an offer
If agreement is reached at time t on a partition that gives player i ashare xi
◮ player i’s payoff is δtixi
◮ δi ∈ (0, 1) is player i’s discount factor
If players never reach an agreement, then each player’s payoff is zero
Levent Kockesen (Koc University) Extensive Form Games: Applications 22 / 47
page.23 A
B
xA, 1− xA
B
xA
Y N
A
δA(1− xB), δBxB
A
xB
Y N
B
δ2AxA, δ2B(1− xA)
A
xA
Y N
Levent Kockesen (Koc University) Extensive Form Games: Applications 23 / 47
page.24
Alternating Offers Bargaining
Stationary No-delay Equilibrium
1. No Delay: All equilibrium offers are accepted
2. Stationarity: Equilibrium offers do not depend on time
Let equilibrium offers be (x∗A, x∗B)
What does B expect to get if she rejects x∗A?◮ δBx
∗B
Therefore, we must have
1− x∗A = δBx∗B
Similarly1− x∗B = δAx
∗A
A
B
xA, 1− xA
B
xA
Y N
A
δA(1 − xB), δBxB
A
xB
Y N
B
δ2AxA, δ2B(1− xA)
A
xA
Y N
Levent Kockesen (Koc University) Extensive Form Games: Applications 24 / 47
page.25
Alternating Offers Bargaining
There is a unique solution
x∗A =1− δB
1− δAδB
x∗B =1− δA
1− δAδB
There is at most one stationary no-delay SPE
Still have to verify there exists such an equilibrium
The following strategy profile is a SPE
Player A: Always offer x∗A, accept any xB with 1− xB ≥ δAx∗A
Player B: Always offer x∗B, accept any xA with 1− xA ≥ δBx∗B
Levent Kockesen (Koc University) Extensive Form Games: Applications 25 / 47
page.26
Properties of the Equilibrium
Bargaining Power
Player A’s share
πA = x∗A =1− δB
1− δAδB
Player B’s share
πB = 1− x∗A =δB(1− δA)
1− δAδB
Share of player i is increasing in δi and decreasing in δj
Bargaining power comes from patience
Example
δA = 0.9, δB = 0.95 ⇒ πA = 0.35, πB = 0.65
Levent Kockesen (Koc University) Extensive Form Games: Applications 26 / 47
page.27
Properties of the Equilibrium
First mover advantage
If players are equally patient: δA = δB = δ
πA =1
1 + δ>
δ
1 + δ= πB
First mover advantage disappears as δ → 1
limδ→1
πi = limδ→1
πB =1
2
Levent Kockesen (Koc University) Extensive Form Games: Applications 27 / 47
page.28
Voting
We looked at voting in the context of candidate competition inelections
We assumed voters vote sincerely
If the voting body is small there may be strategic issues
Levent Kockesen (Koc University) Extensive Form Games: Applications 28 / 47
page.29
Voting Procedures
How should the preferences of voters be aggregated?
Two alternatives (candidates, options): Easy◮ Majority Rule
More than two alternative: Problematic◮ A popular procedure: pairwise voting◮ Condorcet method: each alternative against each of the others◮ The alternative that defeats every other in pairwise voting wins
⋆ Condorcet winner
◮ Many other methods:⋆ Borda count⋆ Majority runoff, etc.
Levent Kockesen (Koc University) Extensive Form Games: Applications 29 / 47
page.30
Example
Three voters: 1, 2, 3
Three alternatives: a, b, c
Preferences1 2 3
a c bb a cc b a
Between a and b: a wins
Between a and c: c wins
Between b and c: b wins
a ≻ b ≻ c ≻ a
Aggregate preferences are intransitive
No Condorcet winner (Condorcet Paradox)
Levent Kockesen (Koc University) Extensive Form Games: Applications 30 / 47
page.31
Pairwise Voting
Alternatives are paired according to some order: Agenda
Winner of pairwise voting proceeds to next round
Winner of the entire election is the winner of the last round
Consider again
1 2 3
a c bb a cc b a
Agenda:1. a vs. b2. winner vs. c
Who is the winner?1. a vs. b: a wins2. a vs. c: c wins3. c is the winner
Is this an equilibrium?
Levent Kockesen (Koc University) Extensive Form Games: Applications 31 / 47
page.32
Strategic Voting
What is the subgame perfect equilibrium?
Two possible outcome of the first round: a and b
Suppose the winner of the first round is a
Second round game:
1
2a c
a 2, 1, 0 2, 1, 0c 2, 1, 0 0, 2, 1
a
2a c
a 2, 1, 0 0, 2, 1c 0, 2, 1 0, 2, 1
c
Three Nash equilibria: (a, a, a), (a, c, c, ), (c, c, c)
Weakly dominant strategy equilibrium: (a, c, c) → c wins
Levent Kockesen (Koc University) Extensive Form Games: Applications 32 / 47
page.33
Strategic Voting
1 2 3
a c bb a cc b a
Suppose the winner of the first round is bWeakly dominant strategy equilibrium: (b, c, b) → b wins
a wins → c
b wins → b
Weakly dominant strategies in first round:◮ Pl. 1: b◮ Pl. 2: a◮ Pl. 3: b
b wins in both roundsDifferent from sincere votingSuppose pl. 2 is the agenda setterWhat is the best she can do?
Levent Kockesen (Koc University) Extensive Form Games: Applications 33 / 47
page.34
Agenda Setting
1 2 3
a c bb a cc b a
Round 1 Winner Round 2 Winner
(a, b) b (b, c) b(a, c) a (a, b) a(b, c) c (a, c) c
Can make any alternative a winner!
Levent Kockesen (Koc University) Extensive Form Games: Applications 34 / 47
page.35
Political InstitutionsAcemoglu & Robinson (2006): Economic Origins of Dictatorship andDemocracy
Democracy typically emerged as a result of voluntary extension offranchise by elites
After democratization: redistribution, pro-poor policies
Why do elites democratize?
In response to social unrest
The British 1832 Reform Act extended franchise◮ In response of revolutionary threat◮ Riots in early 1800’s
Still, why democratize instead of introducing reforms?
Acemoglu & Robinson’s answer:◮ Political institutions are more durable and therefore provide credible
commitment to future policies◮ Democracy as a commitment to pro-poor policies
Levent Kockesen (Koc University) Extensive Form Games: Applications 35 / 47
page.36
Political InstitutionsA simplified version of Acemoglu & Robinson model:
δ fraction is rich with income yr
1− δ is poor with income yp < yr
Mean income
y = δyr + (1− δ)yp
Divide by y
1 =δyr
y+
(1− δ)yp
y
Measure of income inequality
θ =δyr
y
yp < yr ⇒ θ > δ
Levent Kockesen (Koc University) Extensive Form Games: Applications 36 / 47
page.37
Political Institutions
Main political issue is taxation and redistribution
Tax rate τ
Government budget constraint
T = (τ − C(τ))y
T : transfer, C(τ) distortion cost of taxation
Assume C(τ) = 0
Net incomes:
ur(τ) = (1− τ)yr + τy = yr + τ(y − yr)
up(τ) = (1− τ)yp + τy = yp + τ(y − yp)
Ideal tax rates
τ r = 0, τp = 1
Levent Kockesen (Koc University) Extensive Form Games: Applications 37 / 47
page.38
Non-Democratic Politics
Elites (rich) govern
Citizens (poor) get a chance to revolt
If it succeeds a fraction µ is destroyed, the rest divided among poor◮ Poor gets
(1− µ)y
(1− δ)
◮ Rich gets 0
Two possibilities:
1. Elites can commit to tax rate2. Elites cannot commit
Levent Kockesen (Koc University) Extensive Form Games: Applications 38 / 47
page.39
Elites can commit E
C
(1−µ)y(1−δ)
, 0 yp + τ(y − yp), yr + τ(y − yr)
τ
R N
Assume no revolution if indifferent
R ⇔ (1− µ)y
(1− δ)> yp + τ(y − yp)
A little algebra
R ⇔ µ < θ − τ(θ − δ)Levent Kockesen (Koc University) Extensive Form Games: Applications 39 / 47
page.40
Elites can commit
R ⇔ µ < θ − τ(θ − δ)
µ < δ ⇒ even when τ = 1 revolt
µ ≥ θ ⇒ even when τ = 0 do not revolt
δ ≤ µ < θ ⇒ there is a tax rate τ∗ such that
µ = θ − τ∗(θ − δ)
Elites choose tax rate
τ∗ =θ − µ
θ − δ
and revolution is prevented
Payoffs
up =(1− µ)y
(1− δ), ur =
µy
δ
Levent Kockesen (Koc University) Extensive Form Games: Applications 40 / 47
page.41
Elites can commit
θ1
τ∗
1
Inequality increases tax rate
Higher the inequality, higher the need for redistribution to dissuadepeople from revolting
Levent Kockesen (Koc University) Extensive Form Games: Applications 41 / 47
page.42
Elites cannot commitC
(1−µ)y(1−δ)
, 0
E
yp + τ(y − yp), yr + τ(y − yr)
R N
τ
If no revolution elites set best tax rate τ = 0
R ⇔ (1− µ)y
(1− δ)> yp
Revolution iff µ < θ
Levent Kockesen (Koc University) Extensive Form Games: Applications 42 / 47
page.43
Non-Democratic Politics
Assume that the cost of revolution is not too high or low
When elites are in power they have a commitment problem:◮ They would like to commit to some redistribution to poor◮ But their promise is not credible◮ Once the threat of revolution passes, they will renege
Acemoglu & Robinson Hypothesis: Democracy is a commitmentdevice
◮ It transfers power from elites to citizens and solves the commitmentproblem.
Levent Kockesen (Koc University) Extensive Form Games: Applications 43 / 47
page.44
Democratization
Initially Elites hold power
They may democratize or not◮ Democratization: Citizens hold power
⋆ Choose tax rate⋆ Afterwards choose whether to revolt
◮ No Democratization: Elites hold power⋆ Promise tax rate⋆ Citizens choose whether to revolt⋆ If no revolution⋆ With probability q Elites keep promise, otherwise choose tax rate again
Levent Kockesen (Koc University) Extensive Form Games: Applications 44 / 47
page.45
DemocratizationE
C
C
D
τ
(1−µ)y(1−δ)
, 0
R
y, y
N
N
E
C
τ
(1−µ)y(1−δ)
, 0
R
Chance
N
yp + τ(y − yp), yr + τ(y − yr)
q
yp, yr
1− q
Levent Kockesen (Koc University) Extensive Form Games: Applications 45 / 47
page.46
Democratization
Let us assume Elites’ promises are not credible: q = 0
In the previous model elites had commitment problem whenδ ≤ µ < θ
Assume this is the case
µ < θ implies
(1− µ)y
(1− δ)> yp
So, if Elites do not democratize, Citizens revolt
µ ≥ δ implies
(1− µ)y
(1− δ)≤ y
Under democracy there is no revolution
It is best for Elites to democratize
Levent Kockesen (Koc University) Extensive Form Games: Applications 46 / 47
page.47
Democratization
Democracy
If δ ≤ µ < θ, Elites democratize and there is no revolution.
If µ ≥ θ, Elites do not democratize and there is no revolution.
If µ < δ, there is always revolution.
Exercise
What happens if q > 0?
Levent Kockesen (Koc University) Extensive Form Games: Applications 47 / 47
page.1
Introduction to Economic and Strategic BehaviorRepeated Games
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Repeated Games 1 / 29
page.2
A Simple Repeated Game
Everybody is matched with somebody else
Choose C or N
1. N → You gain 1 TL2. C → Other gains 2 TL
Will repeat as long as Excel draws a number ≤ 0.6Click here for the EXCEL file
Levent Kockesen (Koc University) Repeated Games 2 / 29
page.3
Prisoners’ Dilemma
Cooperate DefectCooperate 2, 2 −1, 3
Defect 3,−1 0, 0
It is a parable for many social interactions◮ Contributing to a joint project◮ Common pool resources◮ Arms’ race
The prediction is quite bleak
But humanity seems to have solved at least some of these problems◮ Many social and economic interactions depend on trust and
cooperation
We could not have survived otherwise
Levent Kockesen (Koc University) Repeated Games 3 / 29
page.4
Institutions
Institutions may establish cooperative outcomes◮ Formal institutions:
⋆ State⋆ Law⋆ Police, etc.
◮ Informal Institutions⋆ Social norms⋆ Religion⋆ Ethics, etc.
How do these institutions arise?
What kind of institutions survive?
Levent Kockesen (Koc University) Repeated Games 4 / 29
page.5
Institutions
Institutions
Any institution must be a Nash equilibrium of the game of life.
Otherwise some actor is not acting optimally
In the long-run such non-Nash behavior will be eliminated◮ We will see later that such behavior is not evolutionary stable
Levent Kockesen (Koc University) Repeated Games 5 / 29
page.6
Cooperation
Cooperation in PD type situations is a commonly observedphenomenon
But it is not a Nash equilibrium
How did it survive as an institution?
In early societies PD type games were used to be played in smallgroups in repeated manner
Could cooperation be sustained in such environments?
Levent Kockesen (Koc University) Repeated Games 6 / 29
page.7
Repeated Interactions
In repeated games, reward-punishment mechanisms can be used
Tit-for-Tat◮ Start with cooperating◮ Then do what the other has done last period
Trigger Strategy◮ Start with cooperating◮ Cooperate as long as the other has cooperated◮ After a defection, play defect forever◮ Also known as Grim-Trigger Strategy
Both are forms of reciprocal altruism◮ It is one of the universal norms◮ Known as the Golden Rule
Observed in animal world too◮ Vampire bats, stickleback fish
Levent Kockesen (Koc University) Repeated Games 7 / 29
page.8
Golden Rule
Levent Kockesen (Koc University) Repeated Games 8 / 29
page.9
Reciprocal Altruism
Cooperate DefectCooperate 2, 2 −1, 3
Defect 3,−1 0, 0
Is Tit-for-Tat or Trigger strategy a subgame perfect equilibrium?
Suppose you play it once
Twice?
Need infinite repetition◮ But is this realistic?◮ Perhaps more realistic than finite repetition◮ Every period there is a positive probability that there will be another
period of interaction
Levent Kockesen (Koc University) Repeated Games 9 / 29
page.10
Repeated Games
Players play a stage game repeatedly over time
If there is a final period: finitely repeated game
If there is no definite end period: infinitely repeated game◮ We could think of firms having infinite lives◮ Or players do not know when the game will end but assign some
probability to the event that this period could be the last one
Levent Kockesen (Koc University) Repeated Games 10 / 29
page.11
Repeated Games
Today’s payoff of $1 is more valuable than tomorrow’s $1◮ This is known as discounting◮ Think of it as probability with which the game will be played next
period◮ ... or as the factor to calculate the present value of next period’s payoff◮ or some combination
Denote the discount factor by δ ∈ (0, 1)
In PV interpretation: if interest rate is r
δ =1
1 + r
Levent Kockesen (Koc University) Repeated Games 11 / 29
page.12
Payoffs
If starting today a player receives an infinite sequence of payoffs
u1, u2, u3, . . .
The payoff consequence is
(1− δ)(u1 + δu2 + δ2u3 + δ3u4 · · · )
Example: Period payoffs are all equal to 2
(1− δ)(2 + δ2 + δ22 + δ32 + · · · ) = 2(1 − δ)(1 + δ + δ2 + δ3 + · · · )= 2(1 − δ)
1
1− δ= 2
Levent Kockesen (Koc University) Repeated Games 12 / 29
page.13
Equilibria of Infinitely Repeated Games
There is no end period of the game
Cannot apply backward induction type algorithm
We use One-Shot Deviation Property to check whether a strategyprofile is a subgame perfect equilibrium
One-Shot Deviation Property
A strategy profile is a SPE of a repeated game if and only if no player cangain by changing her action after any history, keeping both the strategiesof the other players and the remainder of her own strategy constant
Take an history
For each player check if she has a profitable one-shot deviation (OSD)
Do that for each possible history
If no player has a profitable OSD after any history you have an SPE
If there is at least one history after which at least one player has aprofitable OSD, the strategy profile is NOT an SPE
Levent Kockesen (Koc University) Repeated Games 13 / 29
page.14
Reciprocal Altruism
Cooperate DefectCooperate 2, 2 −1, 3
Defect 3,−1 0, 0
Consider the following trigger strategy◮ Start with Cooperate◮ Cooperate as long as everyone has always cooperated◮ Defect otherwise
Is it a subgame perfect equilibrium?
Suppose you think the other player will play according to that rule
Do you have an incentive to deviate from it?
There are two types of histories:1. Cooperative Phase
⋆ Nobody has ever defected ⇒ Cooperate
2. Punishment Phase⋆ Somebody has defected at some point ⇒ Defect
Levent Kockesen (Koc University) Repeated Games 14 / 29
page.15
Reciprocal AltruismCooperate Defect
Cooperate 2, 2 −1, 3Defect 3,−1 0, 0
Cooperative Phase
If you cooperate you expect to receive 2 forever◮ Payoff is
(1− δ)(2 + δ2 + δ22 + δ32 · · · ) = 2
If you defect◮ You get 3 today and 0 from tomorrow onwards◮ Payoff is
(1− δ)3
Cooperate is optimal if 2 ≥ (1− δ)3 or
δ ≥ 1
3
Levent Kockesen (Koc University) Repeated Games 15 / 29
page.16
Reciprocal Altruism
Cooperate DefectCooperate 2, 2 −1, 3
Defect 3,−1 0, 0
Punishment Phase
Defecting gets you 0
If you cooperate◮ You get −1 today◮ At most 0 starting tomorrows
Defect is optimal
Trigger is an equilibrium if and only if δ ≥ 1/3.
Cooperation can be sustained if players are patient enough and futureinteraction is likely enough.
Levent Kockesen (Koc University) Repeated Games 16 / 29
page.17
Reciprocal Altruism
More general payoffs
Cost of Cooperation c > 0
Benefit to the other b > 0
Cooperate DefectCooperate b− c, b− c −c, b
Defect b,−c 0, 0
Condition for equilibrium
b− c ≥ (1− δ)b
or
δ ≥ c
b
Levent Kockesen (Koc University) Repeated Games 17 / 29
page.18
Reciprocal Altruism
Reciprocal Altruism is an equilibrium norm if
1. Future interaction is likely enough
2. Cost of altruism is small enough
3. Benefit of altruism is large enough
This provides testable predictions◮ In times of instability we should see more defection◮ If short term gain is too large, cooperation is harder to sustain
Also important:◮ Must be able to identify cheaters◮ Trigger strategy is inefficient
⋆ Limited punishment may be enough
Levent Kockesen (Koc University) Repeated Games 18 / 29
page.19
Reciprocal Altruism
David Hume (1740): A Treatise of Human Nature
I learn to do service to another, without bearing him any real kindness,because I foresee, that he will return my service in expectation of anotherof the same kind, and in order to maintain the same correspondence ofgood offices with me and others. And accordingly, after I have served himand he is in possession of the advantage arising from my action, he isinduced to perform his part, as foreseeing the consequences of his refusal.
Levent Kockesen (Koc University) Repeated Games 19 / 29
page.20
Theory of Repeated Games
Other equilibria?
Many!
Folk Theorem
If the likelihood of future interaction is high enough, any individuallyrational payoff can be sustained as an equilibrium payoff.
Institutions have another role: Selecting equilibrium◮ Efficiency◮ Fairness
Societies whose institutions select efficient and fair allocations thrive,others perish.
Levent Kockesen (Koc University) Repeated Games 20 / 29
page.21
Evidence: Stickleback Fish
When a potential predator appears, one or more sticklebacksapproach to check it out
This is dangerous but provides useful information◮ If hungry predator, escape◮ Otherwise stay
Milinski (1987) found that they use Tit-for-Tat like strategy◮ Two sticklebacks swim together in short spurts toward the predator
Cooperate: Move forward
Defect: Hang back
Levent Kockesen (Koc University) Repeated Games 21 / 29
page.22
Evidence: Stickleback Fish
Milinski also run an ingenious experiment
Used a mirror to simulate a cooperating or defecting stickleback
When the mirror gave the impression of a cooperating stickleback◮ The subject stickleback move forward
When the mirror gave the impression of a defecting stickleback◮ The subject stickleback stayed back
Levent Kockesen (Koc University) Repeated Games 22 / 29
page.23
Evidence: Vampire Bats
Vampire bats (Desmodus rotundus) starve after 60 hours
They feed each other by regurgitatingIs it kin selection or reciprocal altruism?
◮ Kin selection: Costly behavior that contribute to reproductive successof relatives
Wilkinson, G.S. (1984), Reciprocal food sharing in the vampire bat,Nature.
◮ Studied them in wild and in captivation
Levent Kockesen (Koc University) Repeated Games 23 / 29
page.24
Evidence: Vampire Bats
If a bat has more than 24 hours to starvation it is usually not fed◮ Benefit of cooperation is high
Primary social unit is the female group◮ They have opportunities for reciprocity
Adult females feed their young, other young, and each other◮ Does not seem to be only kin selection
Unrelated bats often formed a buddy system, with two individualsfeeding mostly each other
◮ Reciprocity
Also those who received blood more likely to donate later on
If not in the same group, a bat is not fed◮ If not associated, reciprocation is not very likely
It is not only kin selection
Levent Kockesen (Koc University) Repeated Games 24 / 29
page.25
Evidence: Medieval Trade Fairs
In 12th and 13th century Europe long distance trade took place infairs
Transactions took place through transfer of goods in exchange ofpromissory note to be paid at the next fair
Room for cheating
No established commercial law or state enforcement of contracts
Fairs were largely self-regulated through Lex mercatoria, the”merchant law”
◮ Functioned as the international law of commerce◮ Disputes adjudicated by a local official or a private merchant◮ But they had very limited power to enforce judgments
Has been very successful and under lex mercatoria, trade flourished
How did it work?
Levent Kockesen (Koc University) Repeated Games 25 / 29
page.26
Evidence: Medieval Trade Fairs
What prevents cheating by a merchant?
Could be sanctions by other merchants
But then why do you need a legal system?
What is the role of a third party with no authority to enforcejudgments?
Levent Kockesen (Koc University) Repeated Games 26 / 29
page.27
Evidence: Medieval Trade Fairs
If two merchants interact repeatedly honesty can be sustained bytrigger strategy
In the case of trade fairs, this is not necessarily the case
Can modify trigger strategy◮ Behave honestly iff neither party has ever cheated anybody in the past
Requires information on the other merchant’s past
There lies the role of the third party
Levent Kockesen (Koc University) Repeated Games 27 / 29
page.28
Evidence: Medieval Trade Fairs
Milgrom, North, and Weingast (1990) construct a model to show howthis can work
The stage game:
1. Traders may, at a cost, query the judge, who publicly reports whetherany trader has any unpaid judgments
2. Two traders play the prisoners’ dilemma game3. If queried before, either may appeal at a cost4. If appealed, judge awards damages to the plaintiff if he has been
honest and his partner cheated5. Defendant chooses to pay or not6. Unpaid judgments are recorded by the judge
Levent Kockesen (Koc University) Repeated Games 28 / 29
page.29
Evidence: Medieval Trade Fairs
If the cost of querying and appeal are not too high and players aresufficiently patient the following strategy is a subgame perfectequilibrium:
1. A trader querries if he has no unpaid judgments2. If either fails to query or if query establishes at least one has unpaid
judgement play Cheat, otherwise play Honest3. If both queried and exactly one cheated, victim appeals4. If a valid appeal is filed, judge awards damages to victim5. Defendant pays judgement iff he has no other unpaid judgements
This supports honest trade
An excellent illustration the role of institutions◮ An institution does not need to punish bad behavior, it just needs to
help people do so
Levent Kockesen (Koc University) Repeated Games 29 / 29
page.1
Introduction to Economic and Strategic BehaviorExtensive Form Games with Incomplete Information
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Ext. Form Inc. Info 1 / 38
page.2
Extensive Form Games with Incomplete Information
We have seen extensive form games with perfect information◮ Entry game
And strategic form games with incomplete information◮ Auctions
Many incomplete information games are dynamic
There is a player with private information
Levent Kockesen (Koc University) Ext. Form Inc. Info 2 / 38
page.3
Used Car Market
Two types of used cars:◮ Bad (Lemon)◮ Good (Peach)
Seller knows
Potential buyer does not◮ Believes a fraction q is peach
Willingness to pay (or be paid)
Peach LemonBuyer 16,000 6,000Seller 12,500 3,000
Levent Kockesen (Koc University) Ext. Form Inc. Info 3 / 38
page.4
Used Car Market
Peach LemonBuyer 16,000 6,000Seller 12,500 3,000
If there is complete information
Both types are brought to the market and sold◮ This is the efficient outcome
Price?
Depends on bargaining power◮ Assume seller has all the bargaining power
Lemon’s price = 6,000
Peach’s price = 16,000
Levent Kockesen (Koc University) Ext. Form Inc. Info 4 / 38
page.5
Used Car Market
Peach LemonBuyer 16,000 6,000Seller 12,500 3,000
Back to incomplete information
Is there an efficient equilibrium?
If both are brought to the market expected value to the buyers
q × 16, 000 + (1− q)× 6, 000 = 6, 000 + 10, 000q
This is the most they are willing to pay for a car
Peach owner must be willing to sell at that price:
6, 000 + 10, 000q ≥ 12, 500
or q ≥ 0.65
Levent Kockesen (Koc University) Ext. Form Inc. Info 5 / 38
page.6
Lemon’s Problem
There is an efficient equilibrium iff fraction of lemons is small
Otherwise only lemons are brought to the market
First analyzed by George Akerlof◮ Winner of 2001 Nobel in Economics◮ Akerlof, G. (1970), The Market for Lemons, Quarterly Journal of
Economics.
Levent Kockesen (Koc University) Ext. Form Inc. Info 6 / 38
page.7
Adverse Selection
This is also known as adverse selection
Comes from the insurance industry
Suppose the average risk of accident is 10%
Cost of accident is 100,000
Insurance company breaks even if premium is 10,000
But then only those whose risk is greater than 10% will buy insurance
Insurance company will make losses
Increase premium?◮ Say 15,000
What happens?
Levent Kockesen (Koc University) Ext. Form Inc. Info 7 / 38
page.8
Signaling and Screening
Two possible but usually costly solutions◮ Signal◮ Screen
Signaling Games: Informed player moves first◮ Warranties◮ Education
Screening Games: Uninformed player moves first◮ Insurance company offers different contracts◮ Price discrimination
Levent Kockesen (Koc University) Ext. Form Inc. Info 8 / 38
page.9
Signaling Examples
Used-car dealer◮ How do you signal quality of your car?◮ Issue a warranty
An MBA degree◮ How do you signal your ability to prospective employers?◮ Get an MBA
Entrepreneur seeking finance◮ You have a high return project. How do you get financed?◮ Retain some equity
Stock repurchases◮ Often result in an increase in the price of the stock◮ Manager knows the financial health of the company◮ A repurchase announcement signals that the current price is low
Limit pricing to deter entry◮ Low price signals low cost
Peacock tails
Levent Kockesen (Koc University) Ext. Form Inc. Info 9 / 38
page.10
An Example: Beer or Quiche
A stranger comes to the bar for breakfast
A native is also at the bar trying to decide whether to pick a fight
The stranger is either tough or weak◮ Native doesn’t know
Stranger’s payoff to◮ Fight is 2◮ No fight is 4
Native’s payoff to◮ Fight is 2 against weak, −1 against tough◮ No fight is 0
Stranger decides on breakfast◮ Quiche: no cost◮ Beer: costs
⋆ 1 to tough⋆ 3 to weak
Levent Kockesen (Koc University) Ext. Form Inc. Info 10 / 38
page.11
Beer or Quiche
Nature
1
1
2, 2
4, 0
2,−1
4, 0
−1, 2
1, 0
1,−1
3, 0
W〈0.5〉
T〈0.5〉
Q
Q
2
f
r
f
r
B
B
2
f
r
f
r
How would you play?
Backward induction or subgame perfect equilibrium?
Beliefs are important
We need a new solution concept
Levent Kockesen (Koc University) Ext. Form Inc. Info 11 / 38
page.12
Beer or Quiche
Nature
1
1
2, 2
4, 0
2,−1
4, 0
−1, 2
1, 0
1,−1
3, 0
W〈0.5〉
T〈0.5〉
Q
Q
2
f
r
f
r
B
B
2
f
r
f
r
Given beliefs we want players to play optimally at every informationset
◮ sequential rationality
We want beliefs to be consistent with chance moves and strategies◮ Bayes Law gives consistency
Levent Kockesen (Koc University) Ext. Form Inc. Info 12 / 38
page.13
Beer or Quiche
Nature
1
1
2, 2
4, 0
2,−1
4, 0
−1, 2
1, 0
1,−1
3, 0
W〈0.5〉
T〈0.5〉
Q
Q
2
f
r
f
r
B
B
2
f
r
f
r
If you are tough can you avoid a fight?
Drink beer?
Is this an equilibrium?
Levent Kockesen (Koc University) Ext. Form Inc. Info 13 / 38
page.14
Strategies and Beliefs
A solution in an extensive form game of incomplete information is acollection of
1. A behavioral strategy profile
2. A belief system
We call such a collection an assessment
A behavioral strategy specifies the play at each information set of theplayer
◮ This could be a pure strategy or a mixed strategy
A belief system is a probability distribution over the nodes in eachinformation set
Levent Kockesen (Koc University) Ext. Form Inc. Info 14 / 38
page.15
Perfect Bayesian Equilibrium
Sequential Rationality
At each information set, strategies must be optimal, given the beliefs andsubsequent strategies
Weak Consistency
Beliefs are determined by Bayes Law and strategies whenever possible
The qualification “whenever possible” is there because if an informationset is reached with zero probability we cannot use Bayes Law to determinebeliefs at that information set.
Perfect Bayesian Equilibrium
An assessment is a PBE if it satisfies
1. Sequentially rationality
2. Weak Consistency
Levent Kockesen (Koc University) Ext. Form Inc. Info 15 / 38
page.16
Bayes Law
Suppose you take a test for a disease
Test is 99% reliable◮ prob(+|D) = 0.99◮ prob(−|N) = 0.99
On average 0.5% of population have disease
You tested positive
What is the likelihood that you have the disease?
Levent Kockesen (Koc University) Ext. Form Inc. Info 16 / 38
page.17
Bayes Law
Suppose you test 1000 people
5 have disease◮ 5× 0.99 = 4.95 positive◮ 5× 0.01 = 0.05 negative
995 don’t have it◮ 995× 0.99 = 9.95 positive◮ 995× 0.01 = 985.05 negative
positive negative
Disease 5 4.95 0.05
No Disease 995 9.95 985.05
Disease prob. 33.22% 0.005%
Levent Kockesen (Koc University) Ext. Form Inc. Info 17 / 38
page.18
Nature
1
1
〈1〉
〈0〉
〈0〉
〈1〉
2, 2
4, 0
2,−1
4, 0
−1, 2
1, 0
1,−1
3, 0
W〈0.5〉
T〈0.5〉
Q
Q
2
f
r
f
r
B
B
2
f
r
f
r
Separating Equilibria
1. T : B,W : Q◮ Bayes rule (BR) ⇒ B → T,Q → W◮ Sequential rationality (SR) of 2 ⇒ Q → f,B → r◮ SR of 1 ⇒ Tough: OK, Weak: OK
Levent Kockesen (Koc University) Ext. Form Inc. Info 18 / 38
page.19
Separating Equilibria
Separating Equilibrium
The following is a PBE
Behavioral Strategy Profile:◮ Stranger: (T : B,W : Q)◮ Native: (B : r,Q : f)
Belief System◮ prob(T |B) = 1◮ prob(W |Q) = 1
Levent Kockesen (Koc University) Ext. Form Inc. Info 19 / 38
page.20
Nature
1
1
〈0〉
〈1〉
〈1〉
〈0〉
2, 2
4, 0
2,−1
4, 0
−1, 2
1, 0
1,−1
3, 0
W〈0.5〉
T〈0.5〉
Q
Q
2
f
r
f
r
B x
B
2
f
r
f
r
Separating Equilibria
2. T : Q,W : B◮ Bayes rule (BR) ⇒ B → W,Q → T◮ Sequential rationality (SR) of 2 ⇒ Q → r, B → f◮ SR of 1 ⇒ Tough: OK, Weak: NO◮ No such PBE
Levent Kockesen (Koc University) Ext. Form Inc. Info 20 / 38
page.21
Nature
1
1
〈〉
〈〉
〈0.5〉
〈0.5〉
2, 2
4, 0
2,−1
4, 0
−1, 2
1, 0
1,−1
3, 0
W〈0.5〉
T〈0.5〉
Q
Q
2
f
r
f
r
B
B
2
f
r
f
r
Pooling Equilibria
1. T : B,W : B◮ Bayes rule (BR) ⇒ prob(T |B) = 0.5, prob(T |Q) = free◮ Sequential rationality (SR) of 2 ⇒ B → f,Q →?◮ SR of 1 ⇒ Tough: NO, Weak: NO◮ No such PBE
Levent Kockesen (Koc University) Ext. Form Inc. Info 21 / 38
page.22
Nature
1
1
〈0.5〉
〈0.5〉
〈〉
〈〉
2, 2
4, 0
2,−1
4, 0
−1, 2
1, 0
1,−1
3, 0
W〈0.5〉
T〈0.5〉
Q
Q
2
f
r
f
r
B
B
2
f
r
f
r
Pooling Equilibria
2. T : Q,W : Q◮ Bayes rule (BR) ⇒ prob(T |Q) = 0.5, prob(T |B) = free◮ Sequential rationality (SR) of 2 ⇒ Q → f,B →?◮ But SR of player 1 type T ⇒ B → f◮ SR of 2 ⇒ prob(W |B) = 1
Levent Kockesen (Koc University) Ext. Form Inc. Info 22 / 38
page.23
Pooling Equilibria
Pooling Equilibrium
The following is a PBE
Behavioral Strategy Profile:◮ Stranger: (T : Q,W : Q)◮ Native: (B : f,Q : f)
Belief System◮ prob(W |B) = 1◮ prob(W |Q) = 0.5
Levent Kockesen (Koc University) Ext. Form Inc. Info 23 / 38
page.24
Beer or Quiche
What if the cost of beer is equal to 1 for weak stranger too?
Nature
1
1
2, 2
4, 0
2,−1
4, 0
1, 2
3, 0
1,−1
3, 0
W〈0.5〉
T〈0.5〉
Q
Q
2
f
r
f
r
B
B
2
f
r
f
r
Can Beer signal toughness?
Signal must be costlier for the weak
Known as the handicap principle in biology
Levent Kockesen (Koc University) Ext. Form Inc. Info 24 / 38
page.25
Peacock’s Tail
Why?
Levent Kockesen (Koc University) Ext. Form Inc. Info 25 / 38
page.26
Peacock’s Tail
Could peahen be the reason?
Levent Kockesen (Koc University) Ext. Form Inc. Info 26 / 38
page.27
Handicap Principle
Why do we observe traits that reduce fitness?
Because they are signals
Handicap Principle
Reliable signals must be costly to the signaler, costing the signalersomething that could not be afforded by an individual with less of aparticular trait
First proposed by biologist Amotz Zahavi◮ Zahavi, A. (1975) Mate selection - a selection for a handicap. Journal
of Theoretical Biology.
Grafen, A. (1990) Biological signals as handicaps. Journal ofTheoretical Biology
◮ Game theoretical modeling
Levent Kockesen (Koc University) Ext. Form Inc. Info 27 / 38
page.28
Peacock’s Tail
Peacocks have different genetic quality◮ High (H) or Low (L)
Peahens’ fitness higher if they mate with high quality
Peacocks’ fitness higher if they are chosen as mates
Peacocks may exhibit different sized tails◮ Short (S) or Long (L)
Long tail is costly◮ cH to high quality◮ cL to low quality
Is there an equilibrium in which peahens select long tailed peacocks asmates?
◮ Known as sexual selection
Levent Kockesen (Koc University) Ext. Form Inc. Info 28 / 38
page.29
Peacock’s Tail
Nature
Pc
Pc
1, 1
0, 0
1, 0
0, 1
1− cH , 1
0, 0
1− cL, 0
0, 1
H〈0.5〉
L〈0.5〉
S
S
Ph
m
r
m
r
L
L
Ph
m
r
m
r
Levent Kockesen (Koc University) Ext. Form Inc. Info 29 / 38
page.30
Peacock’s Tail
Chance
Pc
Pc
1, 1
0, 0
1, 0
0, 1
1− cH , 1
0, 0
1− cL, 0
0, 1
H〈0.5〉
L〈0.5〉
S
S
Ph
m
r
m
r
L
L
Ph
m
r
m
r
Equilibrium with only High quality “choosing” Long tail?
Long must be interpreted as High quality
Short must be interpreted as Low quality
cH < 1
cL > 1
Levent Kockesen (Koc University) Ext. Form Inc. Info 30 / 38
page.31
Alternative Theories of Education
Human Capital Theory◮ Education enhances productivity and increases earnings◮ Individuals decide comparing marginal benefit and cost of additional
education
Signaling Theory◮ Higher education signals higher ability and increases earnings◮ It pays only more able to get higher education
Levent Kockesen (Koc University) Ext. Form Inc. Info 31 / 38
page.32
Signaling Theory of Education
First analyzed by Michael Spence◮ Winner of Economics Nobel in 2001◮ Spence, A. M. (1973). Job Market Signaling. Quarterly Journal of
Economics.
There are two types of workers◮ high ability (H): proportion q◮ low ability (L): proportion 1− q
Output is equal to◮ H if high ability◮ L if low ability
Workers can choose to have High education (H) or Low education (L)
Low education costs nothing but High costs◮ cH if high ability◮ cL if low ability
There are many employers bidding for workers◮ Wage of a worker is equal to her expected output
High education is completely useless in terms of worker’s productivity!
Levent Kockesen (Koc University) Ext. Form Inc. Info 32 / 38
page.33
A Theory of Education
If employers can tell worker’s ability wages will be given by
wH = H,wL = L
Nobody gets High education
Best outcome for high ability workers
If employers can only see worker’s education, wage can only depend oneducation
Employers need to form beliefs about ability in offering a wage
wH = pH ×H + (1− pH)× L
wL = pL ×H + (1− pL)× L
where pH (pL) is employers’ belief that worker is high ability if shehas High (Low) education
Levent Kockesen (Koc University) Ext. Form Inc. Info 33 / 38
page.34
Separating EquilibriaOnly High ability gets High education
What does Bayes Law imply?
pH = 1, pL = 0
What are the wages?wH = H,wL = L
What does High ability worker’s sequential rationality imply?
H − cH ≥ L
What does Low ability worker’s sequential rationality imply?
L ≥ H − cL
Levent Kockesen (Koc University) Ext. Form Inc. Info 34 / 38
page.35
Separating Equilibria
Only High ability gets High education
CombiningcH ≤ H − L ≤ cL
Again the same principle as in Beer or Quiche and Peacock’s Tail
There is an equilibrium in which only High ability gets High education iff
cH ≤ H − L ≤ cL
High education is a waste of money but High ability does it just tosignal her ability
The result would be even stronger if education increases productivity◮ This is an alternative but complementary theory of education◮ Known as human capital theory
Levent Kockesen (Koc University) Ext. Form Inc. Info 35 / 38
page.36
Evidence
Two papers
Bedard, K. (2001), “Human Capital Versus Signalling Models:University Access and High School Drop-outs,” Journal of PoliticalEconomy 109, 749-775.
Land, K. and D. Kropp (1986), “Human Capital Versus Sorting: TheEffects of Compulsory Attendance Laws,” Quarterly Journal ofEconomics, 101, 609-624.
Levent Kockesen (Koc University) Ext. Form Inc. Info 36 / 38
page.37
Bedard (2001)Tests signaling model of education against the human capital modelUses ease of access to university (presence of a local university) as aninstrumentHuman capital model:
◮ Education enhances productivity and increases earnings◮ Individuals decide comparing marginal benefit and cost of additional
education◮ Easier access to university ⇒ More high-school graduates go to
universitySignaling model
◮ Easier access to university ⇒ High ability high-school graduates go touniversity
◮ Average ability of high-school graduates becomes lower◮ Incentives to pool with them become weaker◮ Low ability students drop out of high school
The paper finds that easier university access1. increases the probability to dropout from high school2. increases the average skill of dropouts
Rejects human capital model in favor of the signaling modelLevent Kockesen (Koc University) Ext. Form Inc. Info 37 / 38
page.38
Signaling: Lang and Kropp (1986)
Tests signaling model of education against human capital model
Uses compulsory school attendance law (CAL) as an instrumentSignaling model:
◮ Suppose CAL increases minimum education from s− 1 to s years◮ Lower ability workers who would have left school after s− 1 will now
remain in school through s◮ Average ability and therefore wage of workers with s years of education
decrease◮ For the more able workers who previously stopped at s years, obtaining
s+ 1 years of education becomes more attractive, and so on◮ Prediction: CAL increases educational attainment of high-ability
individuals who are not directly affected by the law
Human capital model:◮ Prediction: No change in educational attainment of those who are not
directly affected by the law
They find CAL indeed increases the education received by those whoare not directly affected by the law and hence reject human capitalmodel in favor of the signaling model
Levent Kockesen (Koc University) Ext. Form Inc. Info 38 / 38
page.1
Introduction to Economic and Strategic BehaviorExtensive Form Games with Incomplete Information
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Ext. Form Inc. Info 1 / 30
page.2
Screening
A contracting problem with Hidden Information
Uninformed party (principal) offers contract to informed party (agent)
Examples◮ Insurance
⋆ Insuree knows her risk, insurer does not⋆ Insurer offers several packages with different premiums and deductibles
◮ Finance⋆ Borrower knows the risk of project, lender does not⋆ Lender offers several packages with different interest rates and
collateral requirements
◮ Hiring⋆ Applicants know their ability, employer does not⋆ Employer offers different packages varying wages, bonuses, etc.
◮ Pricing⋆ Buyer knows her valuation of the product, seller does not⋆ Seller offers different qualities at different prices, or quantity discounts
Levent Kockesen (Koc University) Ext. Form Inc. Info 2 / 30
page.3
Screening by a Monopolist: A Very Simple Example
You are a wine producer
A fraction c of your customers are connoisseurs while the rest arenovice
You can produce high quality and/or low quality wine
The following table gives the cost of producing the two qualities ofwine as well as the willingness to pay for them by the two types ofcustomers
Willingness to payCost Connoisseur Novice
Low Quality 2 10 6High Quality 10 30 12
You choose which qualities to produce and their prices to maximizeyour expected profit
Levent Kockesen (Koc University) Ext. Form Inc. Info 3 / 30
page.4
Screening by a Monopolist
Suppose you can tell your customer’s type
What should you do?
Produce both qualities
Sell high quality to connoisseur at $30 and low quality to novice at $6
Your profit is
(30− 10)× c+ (6− 2)× (1− c) = 20c + 4(1− c)
Maximizes the total surplus:◮ Extra benefit from high quality is bigger (smaller) than the extra cost
for connoisseur (novice)
This is first degree (or perfect) price discrimination
It is great if you can implement it
But usually you cannot observe types
You may try to ask them their types. What would they tell you?
Levent Kockesen (Koc University) Ext. Form Inc. Info 4 / 30
page.5
You cannot observe types
What happens if you offer high quality at $30 and low quality at $6?
Here are the net payoffs of the two types from purchasing the twoqualities
Net PayoffsConnoisseur Novice
Low Quality 10− 6 = 4 6− 6 = 0High Quality 30− 30 = 0 12− 30 = −18
High type wants to buy low quality
Everybody will buy low quality
Your profit will be 6− 2 = 4 only
Less than if you could observe types and perfectly discriminate
What is the best you can do in this case?
Levent Kockesen (Koc University) Ext. Form Inc. Info 5 / 30
page.6
You cannot observe types
There are several things that you can do
1. Offer only low quality wine
2. Offer only high quality wine
3. Offer both so that Connoisseur (C) chooses high and Novice (N)chooses low quality
Levent Kockesen (Koc University) Ext. Form Inc. Info 6 / 30
page.7
Offer only low quality
Willingness to payCost Connoisseur Novice
Low Quality 2 10 6High Quality 10 30 12
What price should you set?
If you set price at $6 both types buy◮ Your profit is 6− 2 = 4
If you set price at $10 only type C buys◮ Your profit is c(10− 2) = 8c
Therefore your best pricing policy and resulting profit is
Price Profitc ≤ 1/2 6 4c > 1/2 10 8c
Levent Kockesen (Koc University) Ext. Form Inc. Info 7 / 30
page.8
Offer only high quality
Willingness to payCost Connoisseur Novice
Low Quality 2 10 6High Quality 10 30 12
What price should you set?
If you set price at $12 both types buy◮ Your profit is 12− 10 = 2
If you set price at $30 only type C buys◮ Your profit is c(30− 10) = 20c
Therefore your best pricing policy and resulting profit is
Price Profitc ≤ 1/10 12 2c > 1/10 30 20c
Levent Kockesen (Koc University) Ext. Form Inc. Info 8 / 30
page.9
You offer both qualities
You want C to buy high and N to buy low quality wine
You want to choose PH and PL to maximize your expected profit
c(PH − 10) + (1− c)(PL − 2)
C must prefer high to low quality
30− PH ≥ 10− PL (ICC)
N must prefer low to high quality
6− PL ≥ 12− PH (ICN )
Called Incentive Compatibility (IC) constraints
And you want them to prefer to buy
30− PH ≥ 0 (IRC)
6− PL ≥ 0 (IRN )
Called Individual Rationality (IR) (or participation) constraints
Levent Kockesen (Koc University) Ext. Form Inc. Info 9 / 30
page.10
Solving the problem1. (IRN ) holds with equality: PL = 6
Suppose instead that 6− PL > 0. Then
30− PH ≥ 10 − PL > 6− PL > 0
◮ We could increase PH and PL by the same amount without violatingIC constraints and increase our profit
◮ This also shows that we can neglect (IRC)
2. (ICC) holds with equality: 30 − PH = 10− PL
Suppose instead that 30− PH > 10− PL. Then
30− PH > 10 − PL > 6− PL = 0
◮ We could increase PH without violating any constraints and increaseour profit
3. Therefore, we have
PL = 6, PH = PL + (30 − 10) = 30− (10− 6) = 26
Levent Kockesen (Koc University) Ext. Form Inc. Info 10 / 30
page.11
You offer both quantities
So prices are PL = 6 and PH = 26
Notice that you extract the full surplus from the Novice but leave $4surplus to the Connoisseur
Why?Remember what happens if you set prices at perfect informationlevels: i.e. PL = 6 and PH = 30
◮ Novice will happily buy his low quality wine◮ Too bad: Connoisseurs want to buy low quality as well
The amount of surplus you leave to Connoisseur is the minimum youhave to give her to buy high quality
◮ This is called information rent
What is your profit?
(26− 10)× c+ (6− 2)× (1− c) = 16c + 4(1− c)
Compare this with your profit under observable types
(30− 10)× c+ (6− 2)× (1− c) = 20c + 4(1− c)
Levent Kockesen (Koc University) Ext. Form Inc. Info 11 / 30
page.12
What is the best?
Low Only High Only BothPrice Profit Price Profit Price Profit
c ≤ 1/10 6 4 12 2 (26, 6) 16c+ 4(1− c)1/10 < c ≤ 1/2 6 4 30 20c (26, 6) 16c+ 4(1− c)
c > 1/2 10 8c 30 20c (26, 6) 16c+ 4(1− c)
Both is always better than Low only
If c ≤ 1/2 producing both is the best
If c > 1/2 producing High only is the best
Getting surplus from Novice customers is costly◮ You have to leave the Connoisseurs some surplus◮ As the fraction of Connoisseurs increases the benefit decreases and the
cost increases
Therefore, if there are enough Connoisseurs around you don’t caremuch about getting the surplus from the Novice
Levent Kockesen (Koc University) Ext. Form Inc. Info 12 / 30
page.13
Insurance
Similarly in insurance markets
Different policies with different premiums and coverage
High Premium High Coverage → High risk chooses
Low Premium Low Coverage → Low risk chooses
Levent Kockesen (Koc University) Ext. Form Inc. Info 13 / 30
page.14
Insurance
Why are there insurance firms?
They only shift risk from one party to another
Insurance could be beneficial for both the insurer and the insuree
Suppose cost of accident is 100
Probability is 10%
If individual is risk averse◮ She is willing to pay a little bit more than 10, say 12
If the insurance firm is risk neutral◮ It is willing to sell full insurance at little bit more than 10
At 11 they are both better off
Levent Kockesen (Koc University) Ext. Form Inc. Info 14 / 30
page.15
Insurance
This is a very general principle
Efficient Risk Sharing
More of the risk should be borne by less risk averse party. In particular, allthe risk should be borne by the risk neutral party.
Insurance companies are risk neutral because they hold many policieswith imperfectly correlated risks
Same as you holding a diversified portfolio
Levent Kockesen (Koc University) Ext. Form Inc. Info 15 / 30
page.16
Moral Hazard
But if all the risk is borne by the insurer, the insuree may act careless
This may increase accident probability◮ Known as moral hazard
If preventive care is unobservable, leave some risk on insuree
Moral hazard is a very pervasive phenomenon◮ Compensation◮ Development aid◮ Financial bail-outs
Levent Kockesen (Koc University) Ext. Form Inc. Info 16 / 30
page.17
Moral Hazard
There is moral hazard when
Agent takes an action that affects his payoff as well as the principal’s◮ How hard to work
Principal only observes an outcome, an imperfect indicator of theaction
◮ How many items the salesperson sold
Agent chooses an inefficient action◮ Salesperson slacks off
Levent Kockesen (Koc University) Ext. Form Inc. Info 17 / 30
page.18
Moral Hazard
Principal’s problem is to find a contract that induces high effort◮ Principal-Agent Problem
If the agent is risk neutral (and there is no limited liability) efficientoutcome can still be obtained
If he is risk averse then there is a trade off between◮ risk sharing: agent’s wage should not depend too strongly on outcome◮ incentives: to get high effort wage must depend on outcome
Levent Kockesen (Koc University) Ext. Form Inc. Info 18 / 30
page.19
Risk Aversion
An individual is risk averse is he prefers to receive the expected value of alottery to playing the lottery
Payoff function (of certain outcome) is
Linear if risk neutral◮ u(x) = x
Concave if risk averse◮ u(x) =
√x
Convex if risk lover◮ u(x) = x2
Levent Kockesen (Koc University) Ext. Form Inc. Info 19 / 30
page.20
Risk Preferences
b
b
bb
x
u(x)
x1 x2E[x]
Risk averse: u(E[x]) > E[u(x)]
b
b
b
x
u(x)
x1 x2E[x]
Risk neutral: u(E[x]) = E[u(x)]
b
b
b
b
x
u(x)
x1 x2E[x]
Risk lover: u(E[x]) < E[u(x)]
Levent Kockesen (Koc University) Ext. Form Inc. Info 20 / 30
page.21
An Example
The principal will hire a manager for a new project
Revenue is◮ 500K if successful◮ 0 if not
Probability of success depends on the manager’s effort◮ High effort: 0.9◮ Low effort: 0.1
Effort is costly for the manager◮ High effort: 250K◮ Low effort: 50K
Outside option of the manager is 0
Levent Kockesen (Koc University) Ext. Form Inc. Info 21 / 30
page.22
An Example: Risk Neutral Manager
Suppose the manager is risk neutral
If you can observe his effort level◮ Pay 250K for high and 0 for low effort◮ Manager expends high effort◮ Expected profit
0.9× 500− 250 = 200
◮ This is the first best profit for the principal
Levent Kockesen (Koc University) Ext. Form Inc. Info 22 / 30
page.23
An Example: Risk Neutral Manager
What is the optimal pay schedule if effort cannot be observed?
Constant wage?
Manager will choose low effort
Then the best wage is 50K
Expected profit0.1× 500− 50 = 0
Can the principal do better?
Levent Kockesen (Koc University) Ext. Form Inc. Info 23 / 30
page.24
An Example: Risk Neutral Manager
25K salary plus 250K bonus if successful
Manager:◮ Low effort: 25 + 0.1× 250− 50 = 0◮ High effort: 25 + 0.9× 250− 250 = 0◮ Chooses high effort
Expected profit
0.9× (500 − 250) − 25 = 200
Same as observable effort!
Levent Kockesen (Koc University) Ext. Form Inc. Info 24 / 30
page.25
An Example: Risk Averse Manager
If you offer 25K salary plus 250K bonus if successful
Expected wage of the manager: 0
But there is a downside: With positive probability negative payoff
Outside option is a sure 0 payoff
A risk averse manager would not accept it
Principal has to increase expected wage◮ How much depends on how risk averse the manager is
To induce high effort there has to be a bonus
Profit will be less than the first best
For an example see Dixit et al. p. 558-561
Levent Kockesen (Koc University) Ext. Form Inc. Info 25 / 30
page.26
Incentives
Tying compensation to performance may itself create problems
Sometimes it is not easy to measure performance◮ How do you measure a professor’s performance◮ Student evaluations?◮ Professor may become student serving
Sometime success is multi-dimensional◮ If compensation is tied to only one, other dimensions may suffer
Levent Kockesen (Koc University) Ext. Form Inc. Info 26 / 30
page.27
Incentives
In Soviet Union incentives were tied to production volume◮ Quality suffered
AT&T used to pay according to line of programs◮ Unnecessarily long programs
1980s American football player Ken O’Brian◮ Many of his passes were being intercepted◮ His club gave a new contract
⋆ Penalty for every intercepted pass
◮ It worked: His passing percentage increases⋆ But he was passing a lot less
Levent Kockesen (Koc University) Ext. Form Inc. Info 27 / 30
page.28
Incentives
Individual commissions may lead employees sabotage each other◮ See David Mamet film Glengarry Glen Ross
If measuring performance is difficult but can be subjectively assessed◮ Efficiency wage may work
⋆ If performance is good pay high wage⋆ If bad, fire
Levent Kockesen (Koc University) Ext. Form Inc. Info 28 / 30
page.29
Incentives
If task is multidimensional whether they are complements orsubstitutes matter
Farmer may be working on both apple orchard and corn field◮ Substitutes
Farmer works on both apple orchard and beehive◮ Complements
Levent Kockesen (Koc University) Ext. Form Inc. Info 29 / 30
page.30
Incentives
If complements◮ High powered incentives
If substitues◮ Low powered incentives
It is better to collect complementary tasks together
Tasks at an airport are complements◮ It is best if different airports are managed by different companies◮ and each airport managed by a single company
London airports (Heathrow, Gatwick, Stanstead) did exactly theopposite
◮ All three belong to British Airports Authority◮ Different companies manage different “tasks” in each
Outcome is not very good
Levent Kockesen (Koc University) Ext. Form Inc. Info 30 / 30
page.1
Introduction to Economic and Strategic BehaviorEvolutionary Game Theory
Levent Kockesen
Koc University
Levent Kockesen (Koc University) Evolutionary Game Theory 1 / 34
page.2
Evolution
Evolution: Change over time in one or more inherited traits in populationsof individuals
Four mechanisms of evolutionary change
1. Mutation
2. Gene Flow
3. Genetic drift
4. Natural selection
Levent Kockesen (Koc University) Evolutionary Game Theory 2 / 34
page.3
Genetic Variation
Drift and natural selection require genetic variation top operate
Sources
1. Mutations: Changes in DNA2. Gene flow: Movement of genes from one population to another3. Sex: Genetic shuffling
Levent Kockesen (Koc University) Evolutionary Game Theory 3 / 34
page.4
Natural Selection
Darwin’s biggest idea: Main mechanism behind evolution
Three conditions:
1. Variation2. Differential reproduction.3. Heredity
Fitter trait becomes more common in the population
Fitness refers to how good a particular genotype (set of genes) is atleaving offspring relative to other genotypes
◮ Survival, mate-finding, reproduction all matter
Levent Kockesen (Koc University) Evolutionary Game Theory 4 / 34
page.5
Natural SelectionMany examples:
Finches’ beaks on the Galapagos Islands◮ after droughts, the finch population has deeper, stronger beaks that let
them eat tougher seeds
Moths in 19th century Manchester
◮ Both light and dark peppered moths exists◮ Industrial revolution blackened trees◮ Being dark became advantageous in hiding from predators◮ In 50 years nearly all were black
Levent Kockesen (Koc University) Evolutionary Game Theory 5 / 34
page.6
Sexual Selection
Another important mechanism of evolution
Acts on ability to obtain or successfully copulate with a mate
Peacocks, male redback spider
Levent Kockesen (Koc University) Evolutionary Game Theory 6 / 34
page.7
Evolution
It is a fascinating subject but we don’t have the time to study itfurther
Check the web: Understanding Evolution1
One of the best books:◮ Richard Dawkins (1976): The Selfish Gene
1http://evolution.berkeley.eduLevent Kockesen (Koc University) Evolutionary Game Theory 7 / 34
page.8
Evolutionary Game Theory
Fitness of an individual sometimes depends on what others do◮ Success of defending the territory depends also on whether others
defend as well
This creates a game◮ Players: Members of a population◮ Strategies: Traits - behavior, color, etc.◮ Payoffs: Fitness
Can game theory help understand animal or plant behavior?
The answer has been an enthusiastic yes
Levent Kockesen (Koc University) Evolutionary Game Theory 8 / 34
page.9
Evolutionary Game Theory: History
C. Darwin (1871): The Descent of Man◮ Used game theory informally to explain equal sex ratio◮ Eliminated it from the 2nd ed.
R. A. Fisher (1930): The Genetic Theory of Natural Selection
Reformulated the theory:◮ Suppose less female births◮ Females have better mating prospects and leave more offspring◮ Parents who give birth to more females have more grandchildren◮ Genes for female birth spread
R. C. Lewontin (1961): Evolution and the Theory of Games◮ First explicit application
John Maynard Smith: Introduced evolutionarily stable strategy◮ 1972: Game Theory and the Evolution of Fighting◮ with Price 1973: The Logic of Animal Conflict◮ 1982: Evolution and the Theory of Games
Theory of evolution influenced social sciences too
Levent Kockesen (Koc University) Evolutionary Game Theory 9 / 34
page.10
The Model
A large population of organisms with different traits
Asexual reproduction◮ Each organism reproduces on its own◮ No gene reshuffling
Within species evolution◮ Strategies and payoffs are the same◮ Symmetric games
With high probability each member adopts its parent’s strategy
With small probability chooses another strategy (mutation)
Random matching in pairs
Payoffs represent (reproductive) fitness
Central idea: Strategies with higher fitness spread, those with lowerfitness diminish
Levent Kockesen (Koc University) Evolutionary Game Theory 10 / 34
page.11
The Model
We can look at two types of population composition
1. Monomorphic: Each individual plays the same strategy2. Polymorphic: Different individuals may play different strategies
Levent Kockesen (Koc University) Evolutionary Game Theory 11 / 34
page.12
Evolutionary Game Theory
Players: Fish defending their territory
Strategies:
1. Cooperate: Join in defending2. Defect: Do not defend
Payoffs:
C DC 2, 2 0, 3D 3, 0 1, 1
Can cooperation be evolutionarily stable?◮ A strategy is evolutionarily stable if it drives out any mutant
Levent Kockesen (Koc University) Evolutionary Game Theory 12 / 34
page.13
Prisoners’ Dilemma
Suppose everybody plays C
A small fraction ε of mutants appear◮ They play D
Large population ⇒◮ Probability of meeting C type: 1− ε◮ Probability of meeting D type: ε
What is the average payoff of a cooperating type?
(1− ε)× 2 + ε× 0 = 2(1− ε)
How about for the defecting type?
(1− ε)× 3 + ε× 1 = 3− ε
Defecting type has higher fitness for any ε
They will invade the population of cooperators
Cooperation is not evolutionarily stable
Levent Kockesen (Koc University) Evolutionary Game Theory 13 / 34
page.14
Prisoners’ Dilemma
Is D evolutionarily stable?
A small fraction ε of mutants play C
Probability of meeting D type: 1− ε
Probability of meeting C type: ε
What is the average payoff of a cooperating type?
ε× 2 + (1− ε)× 0 = 2ε
How about for the defecting type?
ε× 3 + (1− ε)× 1 = 1 + 2ε
Defecting type has higher fitness for any ε
Defecting is evolutionarily stable
Levent Kockesen (Koc University) Evolutionary Game Theory 14 / 34
page.15
Prisoners’ Dilemma
Lesson 1
A dominated strategy cannot be evolutionarily stable
Lesson 2
Evolution can be inefficient
Levent Kockesen (Koc University) Evolutionary Game Theory 15 / 34
page.16
Another Example: Hawk-Dove
H DH −1,−1 4, 0D 0, 4 2, 2
Is H ESS?
H → (1− ε)× (−1) + ε× 4 = −1 + 5ε
D → (1− ε)× 0 + ε× 2 = 2ε
A small population of D invades
So, H is not an ESS
What is wrong?
(H,H) is not a Nash equilibrium◮ There is a better response to H than H itself◮ This can invade a population of H players
⋆ Most of the time they meet H⋆ Rarely they meet each other
Levent Kockesen (Koc University) Evolutionary Game Theory 16 / 34
page.17
Another Example: Hawk-Dove
H DH −1,−1 4, 0D 0, 4 2, 2
D is not an ESS either◮ We will get back to this later◮ There maybe polymorphic equilibria
Remarkable fact
ESS vs. Nash
If a strategy s is an ESS, then (s, s) must be a Nash equilibrium
Only candidates for monomorphic (pure strategy) ESS are symmetricNash equilibria
Levent Kockesen (Koc University) Evolutionary Game Theory 17 / 34
page.18
Another Example
Is every Nash equilibrium ESS?
H DH 1, 1 2, 0D 0, 2 2, 2
(D,D) is a NE
Is D an ESS?
D → ε× 0 + (1− ε)× 2 = 2(1 − ε)
H → ε× 1 + (1− ε)× 2 = ε+ 2(1− ε)
H does as well as against D AND does better against H
D is not an ESS
What is special about (D,D)?
Every strict NE is ESS
Levent Kockesen (Koc University) Evolutionary Game Theory 18 / 34
page.19
Evolutionary Stability
NE vs. ESS
1. ESS ⇒ ESS
2. Strict NE ⇒ ESS
Evolutionarily Stable Strategy
A strategy s is an ESS if s yields a strictly higher expected payoff thandoes any mutant as long as the population of mutants is small enough.
Evolutionarily Stable Strategy
A strategy s is an ESS if there exists ε > 0 such that for any s′ 6= s
(1− ε)u(s, s) + εu(s, s′) > (1− ε)u(s′, s) + εu(s′, s′)
for all ε with 0 < ε < ε.
This doesn’t look easy to check.
Levent Kockesen (Koc University) Evolutionary Game Theory 19 / 34
page.20
Evolutionarily Stable Strategy
ESS
A strategy s is an ESS if
1. (s, s) is a Nash equilibrium
2. u(s′, s) = u(s, s) ⇒ u(s, s′) > u(s′, s′)
Intuitive
Easy to check
Levent Kockesen (Koc University) Evolutionary Game Theory 20 / 34
page.21
Evolutionarily Stable Strategy
H DH 1, 1 2, 0D 0, 2 2, 2
H is an ESS
1. It is a Nash equilibrium2. No other best response
D is not an ESS
1. It is a NE2. H is another best response and against H , D does not do strictly
better than H
Levent Kockesen (Koc University) Evolutionary Game Theory 21 / 34
page.22
Polymorphic Populations
Hawk-Dove again
H DH −1,−1 4, 0D 0, 4 2, 2
There is no (pure strategy) ESS
Could there be a mixed strategy monomorphic ESS?◮ Each organism plays randomly◮ Has to be a mixed strategy Nash equilibrium◮ Not sufficient
Another way of thinking about it: Polymorphic population◮ Is there a stable population composed of both Hawks and Doves?
Levent Kockesen (Koc University) Evolutionary Game Theory 22 / 34
page.23
Polymorphic Populations
H DH −1,−1 4, 0D 0, 4 2, 2
Suppose a fraction p is H
Average payoffs:
H → p× (−1) + (1− p)× 4 = 4− 5p
D → p× 0 + (1− p)× 2 = 2− 2p
4− 5p > 2− 2p ⇒ H ր4− 5p < 2− 2p ⇒ D ր
Stable only if
4− 5p = 2− 2p or p = 2/3
But this is nothing other than the mixed strategy equilibrium
Levent Kockesen (Koc University) Evolutionary Game Theory 23 / 34
page.24
Polymorphic Populations
How do we know it is ESS?
A phase diagram helps
p1
4
2
−1
Fitness
2/3
We are not going to do dynamics but intuitively◮ Any payoff monotonic dynamic leads to p = 2/3
This is the unique stable population composition
Polymorphic
Levent Kockesen (Koc University) Evolutionary Game Theory 24 / 34
page.25
Polymorphic Populations
H DH −1,−1 4, 0D 0, 4 2, 2
How can we check if p = 2/3 an ESS?
Definitions is as before
Is it a NE? Yes
Are there other best responses? Yes
Does p = 2/3 do strictly better against them than they do againstthemselves? Yes
You can calculate it. Intuitively:◮ If mutant plays p′ > 2/3, they are hurt when they face each other◮ If mutant plays p′ < 2/3, p = 2/3 strategy takes advantage
Levent Kockesen (Koc University) Evolutionary Game Theory 25 / 34
page.26
Hawk-Dove: General Payoffs
H DH (v − c)/2, (v − c)/2 v, 0D 0, v v/2, v/2
What does the incidence of fight depend on?
Is D an ESS?◮ No: (D,D) is not NE
How about H?
Must have v ≥ c
Is H a better response to D than D itself?
u(H,D) = v > v/2 = u(D,D)
So, a population of all Hawks is stable iff v ≥ c
Whenever the prize of the contest is at least as high as the cost ofresulting fight we should observe fights
Levent Kockesen (Koc University) Evolutionary Game Theory 26 / 34
page.27
Hawk-Dove: General Payoffs
H DH (v − c)/2, (v − c)/2 v, 0D 0, v v/2, v/2
What if v < c
Must have polymorphic population
pv − c
2+ (1− p)v = (1− p)
v
2⇒ p =
v
c
Incidence of fighting increases in prize, decreases in cost
Average payoff of the population
(1− p)v
2= (1− v
c)v
2
Payoff increases in cost of fighting
Levent Kockesen (Koc University) Evolutionary Game Theory 27 / 34
page.28
Rock, Paper, Scissors
R P SR a, a −1, 1 1,−1P 1,−1 a, a −1, 1S −1, 1 1,−1 a, a
In RPS a = 0 but we will allow a > 0
Unique Nash Equilibrium (1/3, 1/3, 1/3)
Levent Kockesen (Koc University) Evolutionary Game Theory 28 / 34
page.29
Rock, Scissors, Paper
R P SR a, a −1, 1 1,−1P 1,−1 a, a −1, 1S −1, 1 1,−1 a, a
Let a > 0
Is p = (1/3, 1/3, 1/3) an ESS?
Consider S
S against S yields a
p against S yields a/3
So, p is not an ESS
If the game is like RPS, we should not find a stable population
Levent Kockesen (Koc University) Evolutionary Game Theory 29 / 34
page.30
Side-Blotched Lizards
Common Side-blotched Lizard (Uta stansburiana) common in PacificCoast of North AmericaEach males follows a fixed heritable mating strategy1. Orange-throated: Strongest, do not form strong pair bonds2. Blue-throated: Middle-sized, form strong pair bonds3. Yellow-throated: Smallest, coloration mimics females
Orange beats Blue, Blue beats Yellow, and Yellow beats OrangeSinervo and Lively (1996):
◮ 1990-91: Blue dominated◮ 1992: Orange◮ 1993-94: Yellow
Evolutionary game theory seems to be more than a theoreticalcuriosity
Levent Kockesen (Koc University) Evolutionary Game Theory 30 / 34
page.31
Asymmetric Games
Games played between different species◮ Insects and plants◮ Predator-Prey
Games played by asymmetrically positioned members of the samespecies
◮ Current owner of a nesting ground, food, or mates versus intruder◮ Different sized animals, etc.
Many confrontations are resolved peacefully◮ Current owner of food source keeps it
Levent Kockesen (Koc University) Evolutionary Game Theory 31 / 34
page.32
Asymmetric Games
Owner
IntruderH D
H (V − c)/2, (v − c)/2 V, 0D 0, v V/2, v/2
An individual may come to play either role in different times
A strategy conditions action on the role◮ For example: Hawk if Owner, Dove if Intruder
What is ESS?
Result
A conditional strategy is an ESS iff it is a strict Nash equilibrium.
V < c and v < c ⇒ only two strict NE◮ (H,D): Known as bourgeois strategy◮ (D,H): Known as paradoxical strategy
Levent Kockesen (Koc University) Evolutionary Game Theory 32 / 34
page.33
Asymmetric Hawk-DoveFraction of Hawks in Owner: pFraction of Hawks in Intruder: q
q < V/c ⇒ p րp < v/c ⇒ q ր
Stable points: (1, 0) and (0, 1)
p1v/c
q
1
V/c
1
Levent Kockesen (Koc University) Evolutionary Game Theory 33 / 34
page.34
Evidence
A male Hamadryas baboon forms a long-lasting relationship withseveral females
Kummer et al. (1974) experiment
Male A is allowed to interact with the female
Male B is in a cage and watches
When released, B does not challenge A
The same males change roles over a different female
The same outcome
Escalated fights if each perceives himself as the owner
Levent Kockesen (Koc University) Evolutionary Game Theory 34 / 34