Game Theory (5en254)

Download as pdf or txt
Download as pdf or txt
You are on page 1of 52

Game theory

ZS 2020/2021
Klára Žemlová
1

Lecture 1
Example 1.1
• Same position
• Do not know each other, will never meet again
• Enough time to think
• Held in separate cells
• Want to minimize the years in Gulag
• Not important whether guilty or not
• Get explained consequences (punisments)

Game
• Game = strategic situation
o Payoffs mutually depend
o Cooperative/non-cooperative
• Types:
o Static games
▪ Simulatanous/order does not matter
o Repeated static games
▪ Same situation again and again
o Dynamic games
▪ Structure/order of decisions matter
• Elements of games
o Players
▪ Decision makers
▪ Ex. People, groups, firms, governments, genes…
o Actions
▪ What players can do
▪ Action set = collection of all actions of a player
▪ Action profiles
• Situations that may happen
• List with one action/decision per each player
▪ Outcomes = anything of value for players
• Ex. Money, prestige, food, corporate profit
o Prefecences & payoff-functions:
▪ Over outcomes associated with action profiles
▪ Only ordinal meaning for now
• Definition: Normal-form Game
o I players, for each i: Actions, preferences/utility function
o Normal-form = table (matrix), typically:
▪ Row-player
▪ Column-player
▪ Actions
▪ Cells with utility for players

Example 1.1 – elements


• Players
2

o Conductor, Tchaykovskij
• Actions:
o A=(Confess – C, Not confess – NC)
o Action profiles: {(C,C),(NC,NC), (C,NC),(NC,C)}
o Outcomes: years in prison
• Preference relations:
o (Over outcomes of action profiles)
o Conductor: {C,NC} > {C,C} > {NC, NC} > {NC,C}
o Tchaykovskij: {NC, C} > {NC, NC} > {C,C} > {C, CN}

Additionally…
• Information structure
o Who knows what
o For now, everybody knows everything
• Assumptions on behavior
o Make life easier
o Proceed from simple to complicated
• No cheating
• No cooperation

Assumptions – “nice” preferences


Preferences: (we want “nice” utility function)
• For each pair of actions, player knows which action is better (or indifferent between)
o Completeness
• If action A preferred to B & B is preferred to C => A is preferred to C
o Transitivity
• Action chosen is at least as good as every other available action, acording to preferences
o Rational choice – utility maximization
o Most preferred bundle of all possible bundles
• If {A,B} => A, than not {A,B,C} => B
o Consistency

Assumptions
• Players care only about own utility/payoff
o Rational selfishness
o Not harming others
• Rationality of players
o Irrelevance of unimportant information
o No emotins
o Cognitive cooperation, memory, iteration…
• Common knowledge of rationality
o P1 is rational.
o P2 is rational.
o P1 knows that P2 is rational.
o P2 knows that P1 is rational.
o P1 knows that P2 knows that P1 is rational.
o …
3

Game theory – definition


• “Game theory is the science of strategic thinking.”
o ST – trying to outsmart the others
o Conductor outsmartinf Tchaykovskij
• Strategic thinking in complex situations remains an art
o But! Its foundations consist of some simple basic principles embodied in the game
theory
• Mathematical language to describe strategic situations – model

Models
• Model = simplification of reality that represents certain features of reality
o Map vs globe vs the Earth
• Assumptions
o Simplity reality, while serving a purpose
o Capture the essence, not irrelevant details => good enough approximation
o Shorter than verbal description
o Ass. Of basic microeconomics : homo-economicus (often unrealistic, non-existent
later on)

Economic models
• Why models?
o Intuition unreliable
▪ Analyze complex situations in a tractable way
▪ Help to undestand the realitionships, recognize patterns
▪ Provide non-trivial insights and common language
o Help with predictions
▪ Practical implications
o Give a bechmark toward real behavior
• Real-life: homo sapiens sapiens
o (sub)conscious psychological drives
o Cognitive limitations
o Not well-defined goals
o Lack of ability to achieve them, not maximizing utility..
o When is this important?

Important types of 2x2 games


Example: roommates
Albert and Victoria are roommates. Each of them prefers a clean room to dirty room, but neither
likes to clean the room. If both clean the room, they each get a payoff of 5. If one cleans and the
other does not, the person who does cleaning has an utility of 0, and the person who does not clean
the room has utility of 8. If neither cleans the
room, the room stays a mess, and each has a
utility of 1.
4

Example: Angelina and Brad 1


Brad and Angelina go out tonight
• Angelina prefers a boxing match
• Brad prefers shopping
• Both want to be together than at a prefered
event
• If not together, no utility
Set of players: Brad and Angelina
Set of actions: Boxing, Shoping
Action Profiles
• {BB}, {BS}, {SB}, {SS}
Preferences/payoff functions
• Angelina: {BB} > {SS} > {BS} > {SB}
• Brad: {SS} > {BB} > {SB} > {BS}
This game is called battle of sexes

Example: Halloween party


Oprah and Cindy go out for Halloween party
• Only 2 types of masks:
o Harry Potter (H)
o Teenage Mutant Ninja Turtle (T)
• Cindy wants to copy Oprah
• Oprah wants to be original
• Payoff: win or lose
Set of players: Oprah and Cindy
Set of actions: Harry, Turtle
Action profiles:
• {HH}, {HT}, {TH}, {TT}
Preferences/payoff functions
• Cindy: {HH} = {TT} > {HT} = {TH}
• Oprah: {HT} = {TH} > {HH} = {TT}
Example: Hunters
Two friends go for hunt
• They can hunt either a stag or a hare
• Stag
o If killed, equal split
o The 50 % split bigger than Hare
o Will get only if cooperate in hunt
• Hare
o Each hunter can get it no matter the other
Set of players: Winnetou, Old Shatterhand
Set of actions: Stag, Hare
Profiles of action
• {SS}, {SH}, {HS}, {HH}
Preferences/payoff functions
• Winnetou: {SS} > {HS} = {HH} > {SH}
• Old Shatterhand: {SS} > {SH} = {HH} > {HS}
5

Lecture 2
How will the game likely end?
Method 2: leave out “bad” decisions
• Rational players would not play them
1. Compare actions – find out which are bad
2. Cross out “bad” ones
3. Repeat in a reduced game

Bad decisions
• “Bad” action x = if another action y is “always” better
o Y brings higher utility than x no matter what other players do
o Action x is strictly dominated by y
▪ Dominated = bad
▪ Dominating = good

Assumptions
• We assume rational players will not play dominated strategies
o Rationally
• Other players know this
o Assume: common knowledge of rationality
• Rephrase: There is no belief that a player can hold such that it would be optimal to play
dominated action
o Cannot rationalize the action
▪ Rationalize = defend that based od my beliefs, this was the best I could do
• => we can leave them out (the bad decisions)

Party game
Angelina and Brad can come to a party Early or late
• If both come early, both enjouy each other’s company = 10 h
• If both come late, both enjoy each other’s company = 5 h
• If one comes early and has none to speak to, gets drunk and does not have fun (0 h) while
the other later makes fun of him/her, but this is less fun than other’s company (3 h)
Any dominated strategies?

Two likely outcomes: (EE) and (LL)


Can be argued using beliefs:
• If Angelina belives Brad comes early (with 100 % chance), she should come also early.
• If Angelina belives Brad comes late (with 100 % chance), she should come also late.
6

Dominant equilibrium
Definition:
• Action profile a* if for every player her action is strictly dominating all other actions, given
other players also play stricly dominating strategies
Must be unique if strict domination is used.

IESDA – Interative elimination of strictly dominated actions


Example 3.1
Any dominated actions?
Equilibrium?

Example 3
Any strickly bad (dominated) actions? For P1? P2?
Equilibrium?

Lecture 3
Practice homework
Solve the matrix using IEDS
Dominant equilibrium: {D,M}

How will the game likely end?


Intuition
• Equilibrium = an action profile where nobody wants to change decision alon, given behavior
of others
7

o A “steady” state
o A plausible final situation of game with rational agents
o Holding decisions of others constant, can one single player improve? NO
equilibrium
o No regret principle
▪ If having chance to come back and repeat my decision as the onli one from
all players, a player does not want to
o Self-enforcing contract
▪ If you know others will make their equilibrium action, you want to also make
your eq. Action
o Players have “coordinated” beliefs
• Solution of a game: a theory/way how to find an equilibrium

Nash Equilibrium – definition 1


• Definition: action profile a* is a NE if, for every player i and every action bi of player i, a* is at
least as good according to player i’s preferences as the action profile (bi, a*-i) in which player
i chooses bi while every other player chooses a*-i
• Intuition: in NE, the action I choose brings me higher (or the same) utility than my other
action I can play, given other players play their such action, and this hold for all players
• Method 1: find it directly and reason

Example – Evolutionary game theory Baldwin and Meese (1979)


• Two pigs in one cage
o Dominant and submissive
• Pressing a lever on one side -> food on other side
o Easily found out by pigs
o But costs some effort
• Dominant can force submissive out from food
• Repeated interaction
Setting the table (a model):
• Actions: push lever, wait for food
• WLOG pushing lever is effort – 2
• Eating all food is 10
• Small can eat 3
• Set the table
• Dominations?
Applications of GT in biology
• Numerous examples
• Fights, conflicts, cooperation, competition of roots, size of blossoms, number and gender
ration of children…
Actually, biology best place to use GT
8

Level-K thinking
Level O (L0): non-strategic thinking
• Think of number between 0 and 100
• Expected average? 50
• Eliminate numbers above 50
Level 1 (L1): think all others are L0 and responds
• Think the average guess is 50
• Writes down 33
Level 2 (L2): thinks all others are L1 and responds to
• Writes down 22..
Full rationality?
Perfect rationality vs common knowledge of rationality

Keynesian beuty contest


J. M. Keynes (General Theory of Employment Interest and Money, 1936)
Aim: explain price fluctations in equity markets.
Analogy based on a contest that was run by a London newspaper:
• Entrants asked to choose a set of six faces from 100 photographs of women that were the
“most beautiful”.
• Everyone who picked the most popular face was entered into a lottery for a prize.
• Naive strategy:
o Choose the six faces that you think are the most beautiful
• More sophisticated:
o Wishing to maximize the chance of winning a price
o Think about what the majority perception of beauty is
o Choose six faces that you think fulfill the public perception of beuty
Message: Investors do not always care about real (fundamental) value of assets, but try to guess
what others will do and act accordingly

IESDA – not always perfect


Example 3.6 (IEDS)
2 players
P1: top, middle and P2: left, centerm right
Dominated strategies? C or R? It is C? Or is it R?
Path dependency!
9

Weak dominance
If action z almost always better, but sometimes yields the same payoff as x
• = action x i weakly dominated by z
• Z and x could be the same
Formally (definition):
• Player i’s action a weakly dominates action b if ui(a, a-i) >/= u-i (b, a-i) for every list a-i of other
players’ actions
• And for some actions, ui(a, a-i) > u-i (b, a-i)

Iterated dominance: summary


Pros:
• Simple – just compare all pairs of actions and you find if some are dominated
• If many actions, elimination simplifies game
Cons:
• Not always gives solution
• Weak dominance: path dependency
• Strong dependence: order does not matter
Need for stronger concept

Lecture 4
Best response
Core idea: if one player knows what the others are going to do, her decision is easy

Example 3.6
Method 3: Best response intersection
BR1(L) = T, M
BR1(C) = M
BR1(R) = T

BR2(T) = L
BR2(M) = L

Example 3.7
Can you solve the matrix using IEDS?
Write down the BR of both playes and
indicate them in matrix.
BR1(A) = Y BR2(X) = B, C
BR1(B) = Y BR2(Y) = B
BR1(C) = Z BR2(Z) = A
Do the BR cross? Yes, crossing by playing
actions {Y, B}.
Is X dominated? No.
• Never a best response =/=> strictly
dominated
• Strictly dominated => never a best response
10

Best response function


Definition (BR): a set of actions (strategies) that produce the outcome with highest utility for player,
given the actions of other players.
Every action in set BRi(a-i) is a best response of player i to a-i*
If each of the other players adheres to a-i then player i can do no better than choose a member of
BRi(a-i)

Nash Equilibrium – definition (2)


• Definition (NE2): the action profile a* is NE if and only if () every player’s action is a best
response to the other player’s actions
o Intuition: in NE, nobody can unilaterally (=acting alone) improve their payoff,
everybody is playing the best they can
• How players actually reach NE? Do not know! -> Beliefs or coordination device

Nasch Equilibrium – understanding


Each player has prior belief (stereotype) about other players’ actions, ex. Buyers and sellers
• Imagine playing over and over
• Derived from past experience playing the game
• Experience sufficient to know how opponents will behave
• Does not know action of her particular opponents
Idealized circumstances:
• For each player – population of many such players
• Players are selected randomly from each population
• Players gain experience about “typical” opponents, but not anyspecific set of opponents
Coordination game – which side of sidewalk to chose
• If walking towards a Japanese or a Belgian?

Example – Battle of sexes again


• Angelina and Brad go out tonight
• Find BR and NE
BR1(B) = B BR2(B) = B
BR2(S) = S BR2(S) = S
• Two NE = {(BB), (SS)}
• Cannot decide which is better
o “coordinated” beliefs
o Need for coordination device – a phone call or 3rd person

Example – matching pennies again


• Halloween party story (dressing same or not)
• Find BR and NE
BR1(H) = H BR2(H)= T
BR1(T) = T BR2(T) = H
• No Nash Equilibria in current definition
11

More about Nash Equilibrium


• Current NE definition does not imply
o That a game always has a NE
o That a game has at most one NE
• Existence of NE:
o There may be none to many NE
o In pure strategies, will explain later
• Dominant equilibrium is NE
o But not all NE are dominant E
• Dominated action/strategy can never be best response
o No belief would justify playing it

Example 4.1
1. Find NE
2. Find DE using weak elimination
• If we eliminate T which is weakly
dominated by B, and then eliminate
L which is dominated by R, we lose
nonstrict NE {TL}

Strict and non-strict NE


Strict NE:
• Equilibrium action is strictly better than any other action (given that all other players stick to
NE actions)
Non-strict NE:
• Equilibrium action is not worse than any other action (given that all other players stick to NE
actions)
• Best response includes more than one element
• Let’s assume that if a player cannot be strictly better off by changing equilibrium action, no
change occurs.

Example 4.2 (strict and non-strict NE)


Find BR of both players, NE
Best responses:
BR1(L) = Z BR2(X) = L
BR1(C) = Y, T BR2(Y) = C, R
BR1(R) = X BR2(Z) = L
BR2(T) = C
12

Example – party game again


Find BR and NE of the game
BR1(E) = E BR2(E) = E
BR1(L) = L BR2(L) = L
NE = {(EE), (LL)}
• Not sure which one will happen
• Both know which is better
• Need for coordination device – a phone call or a 3rd person
From point of view of any oserver, any action profile better?
• Probably {EE}
Let’s formalize it
Can not say which player is more important
• Like payoffs in different currencies when exchange not know

Pareto optimality (efficiency)


When outcome o of an action profile is not worse in terms of utility than a different outcome o’ for
all players, while it is strictly better for at least one player, we say o Pareto-dominates o’
• An outcome o* is Pareto-optimal if it is not Pareto-dominated (default)
Possibly more Pareto-optimal outcomes
• At least one P-O outcome

Example 3.8
Underline the BR of both players
Do the BR cross? Yes, the cross by playing
action {DM}

Example 3.9
Underline the BR of both players
Do the BR cross? Yes, they cross by playing actions {(XA),(ZC)}
Can you solve the matrix using IEDS – strict or weak? No

Lecture 5
Example 5.1 – hand feeding squirrel
Situation: matching pennies
What is the bes decision here?
No Nash Equilibria in deterministic decisions
• No pair of actions is complatible with a steady
state
1) If squirrel goes always left:
13

• my best response – go always left


• Squirrel can improve
2) If squirrel goes always right
• My best response – go always right
• Squirrel can improve
Squirrel won’t always go L or R! – mixing over actions!

How to describe a plan of play? What about payoffs?


Strategy (intuition) – plan of actions
• E.g. Squirrel has actions {left, right} but may choose to play each with some probability
• Plan of what probabilities give to each action = mixed strategy
Payoffs: number from payoff matrix insufficient
• We won’t end-up always in the same cell
Instead, expected utility/payoff of the mixed strategy
• EU = weighted average of expected payoffs in the mix
• Weight = how often I end-up with payoff in the mix

Probability distribution
Giving probabilities to outcomes
Example: tossing a fair coin
• Outcomes: {HT}
• Probabilities: (1/2, ½)
Example: playing fair die
• Outcomes: {1, 2, 3, 4, 5, 6}
• Probabilities (1/6, 1/6, 1/6, 1/6, 1/6, 1/6)
Example: playing two fair dice, focus on sum
• Outcomes: {2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12}
• Probabilities: (1/36, 1/18, 1/12, 1/9, 5/36, 1/6, 5/36, 1/9, 1/12, 1/18, 1/36)

Expected payoffs
EP = sum (probabilityi * payoffi)
Example: tossing a fair coin
• Payoffs: H => +1, T => -1
• Probabilities: (1/2, ½)
• EP = ½*1 + ½*(-1) = 0
Example: playing fair die
• Payoffs: {1, 2, 3, 4, 5, 6}
• Probabilities: 1/6, 1/6, 1/6, 1/6, 1/6, 1/6
• EP = 1/6*1 + 1/6*2 + … + 1/6*6 = 3,5
Example: playing two fair dice, focus on sum
• Payoffs: {2, 3, 4, 4, 6, 7, 8, 9, 10, 11, 12}
• Probabilities: (1/36, 1/18, 1/12, 1/9, 5/36, 1/6, 5/36, 1/9, 1/12, 1/18, 1/36)
• EP = 1/36*2 + 1/18*3 + … + 1/36*12 = 7

Introducing mixed strategies


Definition: Mixed strategy si a probability distribution (p1, p2, …, pn) over set of actions
14

• E.g. si = ( ½, ¼, ¼) is mixed strategy where player i plays “left” with probability ½ and “middle”
and “right” with probability ¼
• Probabilities have to sum up to 1!
• Randomization over actions
Mixed strategy may assign probability 1 to a single action – pure strategy
• E.g. si = (0, 1) is pure strategy where player i always plays “R”
Support of mixed strategy – set of actions played with positive probability.

Graph

Best response function – redefined


Definition (best response): Strategy (a set of) that produces the outcome with highest EXPECTED
utility for player, given the strategies of other players
• We re-defined utility from now on to expected utility, so we can write onli u instead of EU
• Same definiton as before, but only “s”s instead of “a”s

Graph – continue
15

Payoffs
Formally:
• Expected utility of the mixed strategy profile for player y:

o Sum (over all cells in the matrix) of utility of the cell weighted by the probability of
reaching the cell undet given mixed strategy profile

o Probability of reaching the cell is given by the product of probabilities of each player
playing his action that leads him to this cell

Graph – squirrel goes 90 % left

Graph – squirrel goes 100 % right0


16

Graph – squirrel goes 90 % right

Graph – solution of my BR

Graph – solution of squirrels BR


17

Graph – final solution and MSNE

Mixed strategy Nash Equilibrium


• Definition: Mixed strategy Nash Equilibrium (MSNE) is a mixed-strategy
profile s* such that no player i has a mixed-strategy si such that she prefers
(si, s*-i) to s*
• I.e. expected payoff of s* is at least as large as expected payoff of (si, s*-i) for every si
• Re-definiton: s* is a MSNE if and only if s*i is in Bi(s*-i)
for every player. The old definition with “a”s defines a “pure
strategy” NE

So far it looks all the same


• But to mixed strategies, we have a theorem to apply:
o Theorem (Nash, 1950): Every finite game has a Nash Equilibrium
▪ Finite game: takes a finite amount of space to write
down (finite no. Of players, actions)
▪ Then there exists at least one NE, possibly in mixed strategies
• Very powerful
▪ Important for game theory (Nobel prize)
▪ Does not tell us how to find it

Von-Neumann Morgenstern prefs


So far, payoff function had only ordinal meaning, now there is more…
• VNM Prefs: preferences regarding lotteries can be represented by the
expected value of the payoff function
o Now, cardinal meaning for comparison of expected values
• Each player prefers lottery with higher expected value of a payoff funcion
• (So far – players were maximizing their payoff, now – maximize expected value)
18

Computing MSNE
Example 5.3 – Battle of sexes again
• Every Thursday after work they meet
• Even though they may have decided upon where to go, Brad often forgets the decision
• So Angelina is uncertain about what Brad will do: will he assume they go shopping or that
they go to a boxing game?
o (Or, Brad is fed-up with always going boxing and would like to have a fair
relationship)
• Brad can not communicate with Angelina
o This would show her that he forgot
• What should they do?
o They should somehow best-respond to that uncertainty

1. Guess the support:


o Which actions are rationalizable? (At least
once BR)
o All actions will be played with some
probability -> both are mixing
2. Assign “some” probabilities
o Say, p and q
3. Expected utility/payoff
o In equilibrium, Angelina must be indifferent and randomize b/w playing B & S
▪ Otherwise she could do better
▪ Trick: Brad knows this and adjust his play p to that
▪ Brad wants Angelina randomize
o Expected utility of Angelina must be the same from both actions

o Similiarly for Brad


▪ Must be indifferent b/w playing B & S

The MSNE is {(2/3,1/3), (1/3,2/3)} = {(q, 1-q), (p, 1- p)}


In addition to {(1,0), (1,0)} and {(0,1), (0,1)}
• Pure strategy NE
What if p >1?
• No way in mixed strategies with this support

Best responses:
• We found MSNE point {(2/3,1/3), (1/3,2/3)}
o Or {1/3, 2/3} in this plane
• What if Angelina knows brad goes to Box less often than p?
19

o = Brad goes more often shopping


o E.g. never goes to boxing: p=0 => Angelina should go there as well!
o i.e. S is then always better than B if p<1/3
o q=0
• What if Angelina knows Brad goes to box more often than p?
o = Brad goes less often shopping
o E.g. always bowing: p=1 => Angelina should go there as well!
o i.e. b is then always better than B if p>1/3
o q=1
• Best response: Angelina (P1)
o If p < 1/3: S is better than B
o If p > 1/3: B is better than S
o If p = 1/3 B is as good as S
• Best response: Brad (P2)
o If q < 2/3: S is better than B
o If q > 2/3: B is better than S
o If q = 2/3: B is as good as S

Best response
• Intersection of BR
o  MSNE
• Check it is really NE
o By deviating, can anybody strictly
gain?
▪ Any pure strategies better?
• 3 intersection = 3 equilibria
o Two pure strategy
o One mixed strategy

Computing MSNE
1. Guess the support
o Support – the actions played with positive probabilities
2. Assign “some” probabilities to actions
o Assume players are mixing and reacting to others
3. Calculate the expected payoffs based on probabilities
o Assume indifference
4. Construct the best-response functions
20

5. Intersect the BR to find equilibrium

Example 5.4 – Penalty kicks


• Can football players calculate MSNE?
• Palacios-Huerta (2003)
• Analysis of 1417 penalty kicks of FIFA games
• Actions: shoot to / jump to {left, right}
• Success (payoffs) in matrix
• Task: calculate MSNE
• MSNE = {(23/60, 37/60), (5/12, 7/12)} = {(0.38, 0.62),
(0.42, 0.58}
• What really happened?
o Do players randomize optimally? Yes
o Do they know GT? We do not know
▪ Does squirrel know GT
o Analysis of player’s behavior important

Interpretations of mixed strategy equilibria


Randomize to confuse your opponent
• Squirrel and me
• Make yourself unpredictable – brings benefits
What might happen in a repeated play
• If you play it over and over, then how many times play pure strategies
• Brad would get des up with all the boxing so they sometimes have to go shopping to make it
fair in the long-run
• Sometimes they may miscoordination
Randomize when uncertain about other’s action
• MSNE sometimes unhappy – ending up uncoordinated action profile
• Angelina does not want to confuse Brad, rather uncertain about the other’s action
• If they are unpredictable in a precise way, they can be in balance
• Solution – a coordination device (flip a coin)
o Result: correlated equilibrium – a randomized assignment of actions such that
nobody wants to deviate

Revision 1
Venus vs Serena
• Serena is at the net
• Venus has the ball
• Where will Venus send the ball?
o To Serena’s left or right?
o Serena has to guess
• Pure-strategy NE? What are BR?

Rock Paper Scissors


• Write down the payoff matrix
• What will be the optimal mix?
• What are the payoffs? PSNE? MSNE?
• What if she plays (1/3, 1/3, 1/3):
o If I play R (100%), what is my EP? Can
21

she do better?
o If I play P, what is my EP?
o If I play S, what is my EP?
o If I play (1/3, 1/3, 1/3), what is my EP?
• MSNE = {(1/3, 1/3, 1/3); (1/3, 1/3, 1/3)}

Conclusion
To be able to compare expected values we need different type of preferences – cardinal
• Von Neumann – Morgenstern preferences
Re-definition of Best-Response, NE, game
• Mixed strategies
Hint: dominated strategy never used in mixing
Interpretation of mixing:
• Reaction to uncertainty
• Uncoordinated repeated play
• Confusing your opponent

Example – Centipede game (McKelvey, 1992)


Experimental results
• Only 37/662 games ends at the first stage
• 23/662 games end in the last stage
• People usually pass until few stages from the end
o Probability to take increases to the end
• Not a single strategy – heterogenous behavior
Why?
• Reputation effect
• Altruism expectations
o (Very small belief -> lot of passing)
• Error/learning
• K-level thinking (not thinking too far)

Sequential games
Famous example: prisoner’s dilemma
• What if player 2 gets info about move of player 1?
• Time dimension: sequential game
• Representation
o Extensive form = game “tree”
o Capture the order of decisions
o P1 moves, P2 observes decision and moves…
22

Game tree

Elements of games – extensive form


• N – set of players
• A – set of actions
• Choice nodes
o When players take an action
• Terminal nodes
o Where the game ends
o Not a choice node
• Action function
o Which actions in which choice nodes possible
• Player function
o Which player makes a decision at the choice node
• Utility/payoff functions
o Assigns utility to each terminal node
• Assumptions:
o Information structure
▪ Perfect information = players know actions taken by other players
▪ Complete information = actions and payoff functions are known to all
players

Prisoner’s dilemma: game tree diagram


How can players decide to play?
S1 = {C, NC}
S2 = {C if C, NC if NC), (NC if C, NC if NC), (NC if C, C if NC),
(C if C, C if NC)}
23

Elements of games – extensive form


Then:
• Definition: Pure strategy in extensive form games
o Complete specification how a player moves
o For every choice node there has to be specification of move
• Intuition
o Compelete “plan” of what to do in any situation player may take turn
o Like programming a robot that will do moves for you in any situation that may
happen (every choice node of player)
• Even for situation that most likely will never happen? Yes! Need to know opportunity costs of
decision!
• How many? Cross-products of action-sets of the choice-nodes.

Example – game tree diagram


What are the pure strategies of P2?
S2 = {(C if A, E if B), (D if A, F if B), (C if A, F if B), (D
if A, E if B)} = {(C,E), (D,F), (C,F), (D,F)}

What are the pure strategies of P1?


S1 = {(A,G), (A,H), (B,G), (B,H)}
 Even if P1 thinks game will never end in
choice-node 4!

Compare (A,C) and (AG, CE).


 Without specification of moves in 3 and 4
we can not evaluate if there are better
alternatives!

Re-using old definitions


The old definitions can be reused with the new meaning of “pure strategy”.
Mixed strategy
• Probability distributions over pure strategies
• Pure strategies now different (plans)
Best-Response
• Strategy that maximizes expected utility given strategies of other players
Nash equilibrium
• All players are best-responding to all other players
• By changing strategy alone, no player can increase payoff

Converting extensive form into normal form games


Are normal-form and extensive-form games connected?
• Yes, extensive form can be transformed into normal form
• Some information lost
Because a normal-form exists, then NE (MSNE) exist.
How? Cross the strategies of players and find corresponding payoffs in the payoff matric
24

Conversion to normal form

Pure strategies:
S1 = {(A,G), (A,H), (B,G), (B,H)}
S2 = {(C,E), (C,F), (D,E), (D,F)}

What are the NE?


NE = {(A,G), (C, F); (A,H), (C,F); (B,H), (C,E)}
SPNE? You have to check all subgames
SPNE = {(A,G), (C, F)}

What about (B,H), (C,E)?


• Is it NE? = can anybody improve payoff (5,5)?
• Will change BH -> BG improve payoff P1?
• Will change BH -> AH improve payoff P1?
• Will change CE -> DE improve payoff P2?
• Will change CE -> CF improve payoff P2?
• Why would P1 ever choose H?
o G dominates H!
o Intuitively wrong!
• P1 by playing H threatens P2 not to play F
o However, is that incredible? No.
o So, even in this sub-game there should be NE

Induced normal form to the extensive form


• Many repetitions: more cells than leaf-nodes.
• In practice, this may get difficult.
• Reverse transformation not always possible (matching pennies) in perfect information game.
• Theorem: every perfect information game in extensive form has a pure strategy NE.

Lecture 7
Subgame perfection
Subgame
• Remaining portion of game three after some moves
• Restriction of the original game
• Can be a game on its own
• Full game is its own subgame
“Subgame-perfect” Nash Equiibrium
• In no subgame can any player i do better by choosing a strategy different from s*i, given that
every other player j plays s*j
25

• Nash Equilibrium for each subgame


• Refinement of NE
o Every SPNE is NE, not always other way round
• -> we rule out non-credible threats
Every extensive form perfect information game has SPNE

Subgame

How to find a SPNE?


Backward induction
• Assume people best-respond in each subgame
Solve game from the end
• For each subgame, eliminate non-credible threats
Reduce game to find the outcome (prunning)
• Replace subgame by its credible outcome

Backward induction
26

Backward induction as comics

Example 7.4
Find SPNE. => SPNE = {(L,R,R), (U,D); (L, R, R),(D,D)}
27

Example 7.4 comics

Example 7.5 – firm competition


• Billa is considering opening a new supermarket across Tesco on Narodni trida
• If it does, Tesco can either
o Do nothing (accept – A)
o Or start a price war to force Billa to close (fight – F)
• If Billa does not enter, Albert will open at the same place
• Tesco can either
o Do nothing (accept – A)
o Or start a price war to force
Albret to close (fight – F)
• Tesco can fight even if Albert decides not
to enter
o Do nothing (accept – A)
o Or start a price war to force
Albert to close (fight – F)

There are two NE in this game: (OUT, OUT, AAF)


and (IN, IN, AFA)

Only the first one based on reasonable behavior


in every part of the game. SPNE: (OUT, OUT,
AAF).

(IN, IN, AFA) is NE but it is an empty threat from


tesco.
Note, that SPNE is a refinement (subset) of NE.
28

Example 7.5 – firm competition comics

Example 7.6 – 2 firm competition


• Two companies in market, Alpha and Beta
• Alpa moves first
o Can invest in FDI in a country
o Or not invest and just export goods
• Beta moves second
o Can decide to export or not

Example 7.7
Two players
• Incumbent monopolist
• A potentional entrant
Entrant chooses between entering the market or not.
• If the entrant enters, the monopolist profit will be lower.
• The monopolist may try to deter enter by threatening to fight
o Engage in an expensive advertising war, predatory pricing, preventing the entrant
from making positive profit.
29

• Such fighting will, however, be costly for the monopolist as well as the entrant.

Two NE:
{Enter (not fight, not fight)} – also SPNE
{not enter (fight, not fight)}

Conclusion
• Extensive representation of games
o Extension of definition of games
o New definition of strategy
• Problem of NE: incredible threat
• Definition – subgame
• Subgame-perfect Nash Equilibrium

Lecture 8
Recap from previous topic
• Sequential games
o Game tree
• New definition of strategy (!)
• Transformation of game from seq to normal form
• Subgame
• Subgame-perfect Nash Equilibrium
• (In-)credible threat
• Backward induction

Introduction: incomplete information


So far, we assumed all players know the game
• Who is playing
• Actions for each player
• Payoffs for each action profile
Now relax that:
• At least one player is not certain about full game => incomplete information about game
o => more than one game for her
o OR => more than one type of the agent
• Same number of agents
• Same strategy space differing only in payoffs
• Add beliefs about which game is being played
Example 1 – Battle of sexes
Angelina and Brad – Mr. & Mrs. Smith
What if Angelina suffers from a bi-polar disorder?
30

• Two possible types of one player


• Brad is not sure whether
o Angelina wants to have a drink
o Angelina wants to avoid him
• As before, Angelina knows Brad’s preferences
• Brad still wants to be with her, during Boxing or Shopping
• Brad (based on experience):
o With probability ½ Angelina wants to meet him —> playing left game (normal B&S)
o With probability ½ Angelina wants to avoid him —> playing right game

• Angelina knows her type


o If she wants to meet or avoid brad
o Depends on her mood after waking up
• Brad does not know Angelina’s type
• To make rational decision, Brad has to form beliefs about the action of each type of Angelina
o + He knows probabilities of each type happening
• Given these beliefs he computes expected payoff of each action and chooses optimally
o E.g. beliefs: Nice A -> B, Bad A -> S, EUB(B) = ?

Bayesian Nash Equilibrium


For this type of game, in equilibrium holds:
• Brad’s action is optimal, given…
o …the actions of the types of Angelina, and
o …B’s beliefs about the state of the world = what each type does.
o E.g. Nice A -> B, Bad A -> S => EUB -> optimal action of Brad
• Action of each type of A is optimal, given the action and beliefs of Brad
o E.g. If nice A knows Brad with these beliefs -> optimal reaction
• Brad’s beliefs are in equilibrium correct
• “Each player’s beliefs about the other player’s actions are correct, and each player acts
optimally, given her beliefs”
o Principle of regret holds
Example 1 – Brad and Angelina – continue
• We analyze this type of Bayesian game as if there were three players:
o Brad
o Angelina – nice
o Angelina – avoid
• Angelina knows her type, but Brad does not => B needs to specify optimal action given both
types
o For all combinations of A’s types, B should find his expected payoff and optimal
response
• Trick: game can be represented in one table instead of two
31

• Interpretation of NE (B,(B,S)):
o Given Brad’s beliefs and actions of both types of Angelina, Brad is playing the best
response
o Given Brad’s action, both types types of Angelina are playing best response
• Interpretation of NE if Angelina is the 1st type:
o Angelina wants to meet Brad and chooses B
o Brad chooses B and believes that 1st type Angelina chooses B and the 2nd type
Angelina chooses S
• Interpretation of NE if Angelina is the 2nd type:
o Angelina wants to avoid brad and chooses S
o Brad chooses B and believes that 1st type Angelina chooses B and the 2nd type
Angelina chooses S
Definition of a Bayesian game – Battle of sexes
• Players
o Brad and Angelina
• States
o Meet, avoid
• Actions
o B and S
• Signals and signal function
o B: one signal both states
o A: two signals
• Beliefs
o Probabilities of states after observing signals
o B: ½ to each state after receiving signal
o A: 1 to either state after signal
o Associated expected action of other player
• Payoffs
o Given in the table

Example 2 – prisoner’s dilemma revisited


What if Tchajkovskij was a nice guy?
• P1 knows that:
o With probability 2/3 P2 is rational => playing left game
o With probability 1/3 P2 is super nice => playing right game
32

Example – Bayesian Nash


Sheriff and suspect
• A sheriff caught a person in a bank in a place of robbery. This person may be a criminal with a
probability p. Both sheriff and the suspect are holding guns. They must now decide whether
to shoot at the other person or not. Their preferences are set as follows:
o The sheriff would rather shoot when the other shoots, not if the suspect does not
shoot.
o The criminal would rather shoot even when the sheriff does not, since the criminal
would be caught if he does not shoot.
o An innocent suspect would rather not shoot even if sheriff shoots.
• Will sheriff shoot if he thinks it is 50/50 the suspect is a criminal?

Solution:
• One way – create a table, three players, p = 0.5…
• Shortcut:
o if you are good suspect, strictly dominant strategy NOT shoot
o if you are bad, dominant strategy is to shoot
▪ you can leave the dominated strategies out
• Expected utility for sheriff:
o E(shoot) = -1*(1-p) + 0*p = 0 – 2*p => p = 1/3
▪ If p > 1/3, sheriff shoots
▪ If less, does not shoot
▪ If equal, then mixing
Example 2 – too much information may hurt
• Single-person decision problem
o Player cannot be worse off if she has more information: if she wishes, she can ignore
the information
• Strategic game
o If a player has more information and the other players know that she has it, she may
33

be worse off
• Following game has two possible states, both players believe that states S1 and S2 are
equally likely
34

Can be complicated further: two Brads


• Can be modeled as 4 people
o Plan for both types of signal that may
happen
• Similar procedure as before
• Expected payoffs
• Best responses
• NE – 4 actions

Real world applications


The “Box”
• USA, ex-offenders may be asked for having been convicted
• HR manager could distinguish between two types of a person when not sure
Ban-the-box campaign
• Againts this right to ask
• Goal: improve employment prospects of ex-offenders
• Adopted across USA
Effects (Doleac & Hansen, 2020)
• Non-offenders with similiar charecteristics lower chance to get job
o Young afro-americans by 5.1 % less likely to be emloyed
• Statical discrimination in practice

Conclusion
Bayesian games – information is incomplete
• Same players, actions, different payoffs
How to find NE in static Bayesian games:
• Consider each type as a single player
• Given beliefs, compute expected payoffs
• Find NE in this game

Lecture 9
Revision
Incomplete information games
• Min one player does not know full game
• Uncertain about part of game, e.g. Payoff/type of other player
• Know probability of events happening
o Chance of meeting good/bad Angelina
Solution procedure
• Treat two types of one player as two players
• Create all combinations of beliefs about their actions
• Create a “big” matrix by calculating expected payoff of uncertain player and the two types
Equilibrium
• Actions optimal given beliefs and actions of other players; beliefs are correct
35

Prisoner’s dilemma:
Game tree – general
• For P2, action C different from E
• Can choose differently in each node
• Strategy set = {CE, CF, DE, DF}

What if player 2 does not know what P1 did? -> game of


imperfect information

Game of imperfect information


• P2 does not know whether P1 moved to the left
or to the right..
• Impossible to choose different actions within
one information set: NC (left) and NC (right) are
not different actions.
• (Strategically) equivalent to normal form
representation!
• For P2, strategy set = {CD}

Definition: imperfect information


Game with imperfect information:
• Players do not know full history of the game
• At least one information set is not singleton
Example: information sets
• P3 has two information sets:
o With single node (singleton) following 1-R, 2-R’
o With 3 nodes
• Game has two subgames:
o Whole game
o Game that follows 1-R, 2-R’

How to analyze these games?


Intuition 1
How to analyze this game?
ANS: regardless P1’s move, P2 plays X
• X dominates Y
Thus A for P1
• “No uncertainty”
• P2 beliefs P1 -> A
36

Could be transformed to a normal form game: NE: (A,X)

Intuition 2
How to analyze this game?
ANS: no domination for P2 this time.
But! Domination for P1
• (P1: if P2 Y -> A)
P2 “knows” this -> X
• Believes
Could be transformed to a normal form game
• NE: (A,X)

More complicated…
Example 1
How to analyze this game?
• Without R easy:
o T dominates B
o Thus P1 plays L
o Holds also with R
Subgames?
Finding Nash
• Let’s transform it to normal-form game
• Find NE
o Two NE: (L,T) and (R,B)
• Are these subgame perfect as well?
o Only one subgame, whole game (subgames can not cut
through information sets) => yes, they are SPNE as well
Main point 1
• What about (R,B)? Is it plausible?
o (R,B) is empty/incredible threat
▪ P2 will never play B
o We want to eliminate this, keep only (L,T)
• How to eliminate empty threat?
o Dynamic games with perfect information
▪ SPNE eliminates non-credible threats
o Dynamic games with imperfect information
▪ SPNE is not strong enough!
37

▪ Need a new concept: weak perfect Bayesian Nash Equilibrium “wPBNE”


o Idea: players then best-respond to their beliefs
o wPBNE: eq. Strategies and beliefs
▪ belief system to each information set a probability distribution over the
decision nodes in that information set
▪ => some implausible NE will be ruled out

Max expected utility given beliefs


Is there any p such that EP(B) > EP(T)?
• EPP2(T) = 1p + 2(1-p) = 2-p
• EPP2(B) = 0p + 1(1-p) = 1-p
• (2-p) always > (1-p)
=>T is always better than B irrespective of the beliefs (p
• = domination
Principle of sequential rationality requires that P2 chooses T => (R,B) is not wPBNE and is eliminated

Weak perfect Bayesian Nash Equilibrium


• P1 knows that P2 always plays T => for
P1, left is better than middle and right
• Consistency beliefs requires that p=1, 1-
p=0
• WPBE: (L,T) and beliefs: P2 believes that
if he is in his information set, then P1
chose left with probability 1 and middle
with probability 0
• WPBE: {(L,T), p=1}
• Why is {(R,B), p=1} not wPBE?
o If P2 gets to play in info node,
will never play B
o Violation of sequential
rationality
Weak perfect Bayesian Nash Equilibrium
Definition: a wPBE equilibrium consists of equilibrium strategies and beliefs systems satisfying
following conditions:
1. Sequential rationality
o Each players’ strategy is optimal whenever she has to move, given her belief and the
other players’ strategies
o P1 -> L, P2 -> T is optimal choice given P2’s beliefs
2. Consistency of beliefs
o With strategies if information set is reached – each player’s belief is consistent with
strategy profile (strategies of all players)
o P2 has beliefs consistent with action of P1 (P1 takes the turn that is expected by P2)

Practice example
Consider following 3-player game
• How many proper subgames it has? 3
• Find wPBNE.
• What about (A, L, L’, p=0)
38

Example 3
Pepsi is considering a new marketing war against Coca Cola. Pepsi can decide to stay out of the
market (O). If Pepsi decides to enter the market Coca Cola knows that there is a new increased
competition but does not observe whether Pepsi is ready (R) or unready (U). Coca Cola decides to
accept (A) the new competitor or to fight (F).
1. Write down normal form of the game.
2. Find Nash equilibria (optimal actions)
3. Check NE one by one to see if consistent beliefs can be found (if yes, we have wPBNE; if no,
no wPBNE)

This works because wPBNE is a subset of NE. There are two NE: (O,F) and (U,A)
1. (U,A) is NE. Is it wPBE?
o Option 1: information set is reached
o Beliefs have to be consistent with actions => p=0
and 1-p=1
o Yes, wPBE 1 is (U,A) and p=0, 1-p=1
▪ P1 behaves optimally given action of P2
▪ P2 behaves optimally given his beliefs
and P1’s action
▪ P2’s beliefs are consistent with P1’s
actions
2. (O,F) is NE. Is it wPBNE?
o Option 2: information set is not reached
o Beliefs have to be consistent with actions =>
beliefs have nothing to be consistent with set p
such that actions of all players are still optimal
o => F has to be optimal for Coca-Cola  EP(F) >/=
EP(A)
o WPBE 2 is (O,F) and any p > ½ (F is better than A)
▪ P1 behaves optimally given action of P2
▪ P2 behaves optimally given his beliefs and
P1’s action
▪ P2’s beliefs are consistent with P1’s action
39

o wPBE 3 is (OF) and p = ½ (F and A are equally good)


▪ if P2 mixes, it has to be such that 0 is still for P1:
EP(0) > EP(R); EP(0) > EP(U)

▪ P1 behaves optimally given action of P2


▪ P2 behaves optimally given his beliefs and
P1’s action
▪ P2’s beliefs are consistent with P1’s actions

This game has three wPBE:


• WPBE 1 is (U,A) and p=0, 1-p=1 (p < ½ )
• WPBE 2 is (O,F) and any p > ½
• WPBE 3 is (O,F) and p = ½ ; q = ½

Summary
Static games
• Complete information: NE
• Incoplete information: Bayesian NE
Dynamic games
• Perfect information: SPNE
• Imperfect information: weak perfect Bayesian equilibrium
Weak perfect Bayesian equilibrium
• Sequential rationality
• Consistency of beliefs
• Elimenates on-credible threats

Principle of signaling game


Example 1
1. Nature draws the type of P1 (employee)

2. P1 observes own type


o Smart or stupid

3. P1 moves
o Obtain education or no

4. P2 (employer) observes P1’s move and moves


o Employ or not
40

Good education can differentiate the type.


No education can not.
Payoffs realized.

Lecture 10
Revision: weak perfect Bayesian Nash Equilibrium
Definition: A wPBE equilibrium consists of equilibrium strategies and beliefs systems satisfying:
1. Sequential rationality
o Each players’ strategy is optimal whenever she has to move, given her belief and the
other players’ strategies
▪ Similar idea to subgame perfection
▪ Rules out incredible threat in an imperfect-information game
2. Consistency of beliefs
o With strategies – each player’s belief is consistent with strategy profile (strategies of
all players)
▪ With players believe others are doing is true
▪ Required if information set with uncertainty is reached

Signaling games
• Extensively used in Economics
o Also in biology – credibility enhancing displays
o Peakock’s tail
• Michael Spence – Nobel prize 2001
• Job-market signaling
• Sender of a signal/message
o Worker can signal productivity – get MBA
• Receiverer of a signal
o Employer observes signal and decides to hire or not
• Spence shows it may make sense to have MBA even though it may be useless
o Costs less high-ability employees than low-ones
o Signal can separete them
• Rephrase MBA and abilities to beer and quiche
41

Principle of signalling game: example 1


1. Nature draws the type of P1 (employee)
2. P2 observes own type
o Smart or stupid
3. P1 moves
o Obtain education or not
4. P2 (employer) observes P1’s move and moves
o Employ or not
Employer can not differentiate type of P2.
Payoffs realized.

Let’s use simpler notation.


• Leftie, rightie, up, down…
How to analyze it?
When P1 goes D or B, no uncertainty and we know result.
What about U or T?

What about dominated moves?


• P1L: U >/= D
• P2: X >Y
• P1L knows this => U
• P1R compares T and B
o P2 will play X
• P1R => B
• Equilibrium:
o ((U,B), (E,H,X), p=1)
• Separating equilibrium
o P1L and P1R play different strategies
o
Dynamic game with imperfect information – example 2
• Dominated strategies?
o PL1: D > U
o => if P1 goes up, it must be that P1 is rightie
▪ Uncertainty resolved
o => 1-p=1
o => P2 always play Y
• Again, separating equilibrium
o Equilibrium: ((D,T), (E,H,Y), p=0)
42

Dynamic game with imperfect information – example 3


Dominated strategies?
• P1L: has to go U
o Or D => 0
• P1R: has to go U
o Or B => 0 as well
• P2? Expected payoff!
o P2: larger EP from X
o EP(X) = ½ * 10 + ½ * 4 = 7
o EP(Y) = ½ * 2 + ½ * 8 = 5
• Equilibrium:
o ((U,T), (E,G,X), p = ½)
• Pooling equilibrium
o P1L and P1R play same strategies

Dynamic game with imperfect information – example 4


Dominated strategies?
• P2: E > F, H > G
• P1L: U > D
• P1R: ?
• Is there a chance P1R goes to T?
o If P2 belives P1 is R
o Or given p, which EP(Y) E(X) is higher?
o => odes p = ½ work?
• P2: EP(Y) > EP(X)
o => pooling equilibrium
o ((U,T),(E,H,Y), p = ½ )
• P1L: U > D
• P1R: ?
• P2: E > F, H > G
• Any other equilibrium?
o Is there a chance P1R goes B?
▪ Then BR of P2 is play X
▪ Is it an equilibrium? Check.
• Separating equilibrium:
o ((U,B), (E,H,X), p=1)

Dynamic game with imperfect information – example 5


• PL1: ?
• P1R: T > B
• P2: E > F, H > G
• Is ((D,T), (E,H,Y), p=0) equilibrium?
o No. (Regret for P1L).
• Is ((U,T), (E,H,X), p=0.6) equilibrum?
o No. (Regret for P1L).
• No pure-strategy equilibrium
• Mixed-strategy equilibrium exists (not shown)
43

Dynamic game with imperfect information – example 6


Dominated strategies?
• P1L: D > U
• P1R: B > T
• P2: Y > X
• P2: E > F
Dominant pooling equilibrium:
• ((D,B),(Y,E), q=0.5)

Dynamic game with imperfect information – example 7


Dominated strategies?
• P1: ?
• P2: Y > X
• P2: E > F
• => P1: D > U & P1: T > B
• Separating equilibrium:
o ((D,T), (E,Y), p=0, q=1)
o Actually, for any p

Signaling games
Beer or quiche
• 0. Nature chooses the type of P1:
o Weak, strong with certain probabilities
• 1. PA chooses his breakfast: Beer of Quiche
• 2. P2 (a bully) does not know the type of P1 but
observes breakfast.
o Then he decedes whether to fight P1.
• P1 gets 1 point if he had his favorite meal, and gets
additional 2 points if there was no fight.
• P2 gets 1 point if he fought the Weak type did not
fight the Strong
Equilibria:
• Separating equilibrium
o Each type (P1) chooses a different action
o Receiver (P2) knows the sender’s type
o E.g.: strong -> Beer; weak -> quiche
• Pooling equilibrium
o All types (P1) choose the same action
o Receiver (P2) has no clue about sender’s type
o E.g.: strong -> beer; weak -> beer
• Partially separating/pooling equilibrium
o Some types of sender send one message; some types of sender mix the messages
o E.g.: strong -> beer; weak -> mix
44

Pooling NE 1 Pooling NE 2

Separated NE 1 Separated NE 2

Separating equilibrium 1: guess


P1:
• Strong -> beer
• Weak -> quiche
Let’s check this: Separated NE 1 - is it possible?
P2: optimal response:
• If beer -> not fight
• If quiche -> fight
Not an equilibrium, weak wants to deviate
• If he did the same as strong, would get not fight

Separating equilibrium 2: guess


P1:
• Strong -> quiche
• Weak -> beer
Let’s check this.
P2: optimal response:
• Beer -> fight
• Quiche -> not fight
Not an equilibrium, weak wants to deviate.
• If he did the same as strong, would get not fight
45

Pooling equilibrium 1: guess


P1:
• Strong -> beer
• Weak -> beer
Let’s check this:

WBNE: P1S, P1W -> B, P2(Q) -> F, P2(B) -> NF, P > 1/2, q = 0.1

P2: optimal response to that:


• Not fight if beer
• Fight if quiche (so P1 weak does not deviate)
WPBE:
• If beer, Pr(weak) = 0.1; Pr(strong) = 0.9
• If quiche, Pr(weak) = p; Pr(strong) = 1 – p; where p > ½

Pooling equilibrium 2: guess


P1:
• Strong -> quiche
• Weak -> quiche
Let’s check this:

WBNE: P1S, P1W - > Q, P2(Q) -> NF, P2(B) -> F, q > 1/2 , p=0.1

P2: optimal response to that:


• Not fight if quiche
• Fight if beer (so P1 – stron does not deviate)
WPBE:
• If quiche, Pr(weak) = 0.1; Pr(strong) = 0.9
• If beer, Pr(weak) = q; Pr(strong) = 1 -q; where q < ½

Beer or quiche: summary


Two pooling WPB equilibria in this game:
• P1: strong -> quiche; weak -> quiche
• P2: if beer -> fight; if quiche -> not fight
If beer: Pr(strong) = 1 – p; Pr(weak) = p, p > ½
If quiche, Pr(strong) = 0.9; Pr(weak) = 0.1
• P1: strong -> beer; weak -> beer
• P2: not fight if beer, fight if quiche
If beer, Pr(weak) = 0.1; Pr(strong) = 0.9
If quiche, Pr(weak) = p; Pr(strong) = 1- p, p > ½

Lecture 11
Revision
Static games
• Perfect information: nash equilibrium
o Conductor and Tchaykovskij
• Mixed strategy NE
o Squirrel and me
• Imperfect information: (Bayesian) Ne
o Angry Angelina or a nice Angelina?
46

Dynamic games
• Perfect infromation: SPNE
o Billa, Tesco and Albert entering market
• Imperfect information: weak perfect Bayesian equilibrium
o Bully and Kung-fu master / weak guy

Another beer or quieche: modifed BoQ 1

Separation NE 1 - not possible Separatin NE 2 – possible

Pooling NE 1 – not possible


47

Beer of quiche
Analyze all possible equilibria:
• Separating equilibrium 1: weak-quiche, strong-beer
• Separating equilibrium 2: weak-beer, strong-quiche
• Pooling equilibrium 1: weak-beer, strong-beer
• Pooling equilibrium 2: weak-quiche, strong-quiche
In each case:
1. Start with player 1 actions
2. Determinate player 2’s beliefs
3. Find player 2’s optimal response
4. Check if player 1’s action is optimal
5. Pooling equilibrium: find beliefs s.t. P1’s action is optimal

Revision: Nash Equilibrium


What is it:
• One strategy (action) for each player = strategy (action) profile
• Every strategy is best response to all the other strategies
• Nobody can gain anything by unilaterally changing the strategy
How to find it:
• Construct a payoff table; find best responses
• Find situations (lists of strategies) where player is playing best response to other(s)
Problems:
• Low predictive power
o (we cab have too many NE in a game)
• Sometimes not sensible predictions – extensive form games (non-credible threats)

Example – dating site


• P1 (Brad): stay out (O), send a picture (P)
• P2 (Angelina): accept a date (A), reject (R)
• 1) SPNE?2) NE?
• What about OR?
• There are two NE in this game:
o (O,R) - > (0,2)
o (P,A) - > (2,1)
• Does (O,R) make sense?
o (O,R) does not contain a “credible threat”
o It is never optimal to Reject after player 1
participates
o Not SPNE

Subgame perfect Nash Equilibrium


What it is: a strategy profile
• Optimal actions of all players at every choice-node in the game tree (in every subgame)
without uncertainty
How to find it:
• Backward induction
Problems:
• SPNE concept can not be used in games with imperfect information (information sets)
Solution -> weak perfect Bayesian Equilibrium
48

Example – dating site – extension 1


• New action: What if Brad uses photoshop and
makes himself too handsome?
• P1:
o Stay out (O)
o Real picture (RP)
o Photoshop (P)
• 1) Angelina can find out
o P2:
▪ accept a date (A)
▪ reject (R)
• SPNE?
o (RP,(R,A’)) – specify action of each player in each node
• 2) What if Angelina can not distinguish a real from a photoshoped picture?
o What about (O,R)?

• There are two NE (and SPNE) in this game:


o (O,R) -> (0,2)
o (RP,A) -> (3,1)
• (O,R) is not a “credible threat”
o It is never optimal to Reject after P1 plays P or R (A is
dominant) = not sequentially optimal
o No systém of beliefs p based on which P2 chooses this
action => (O,R) is not WPBE
• WPBE: (RP,A), p(real) = 1

Economists at real dating sites


Dating site = matching market
• Like organs to patients, doctors and hospitals, students to schools…
• Nobel prize 2012 – Alvin Roth
• More on matchmaking
o Trick: write ordering of preferences
o “Deferred acceptance” principle
Problem: congestion
• Too many men writing to women (stereotype – M starts)
• Men span every women they would like to meet
• Women dislike this
Solution: cost of signal
• Meaning is sincere
49

• One signal per day – e.g. Propose with a rose


o Lee & Niederle, 2014
• One message per day
• Or, only woman initiates contact (Bumble)

Rich or poor Brad


• Angelina is not sure about the type of Brad
• Can be rich or poor
• Brad sends a signal – car or roses

Pooling NE 1 Pooling NE 2

Separated NE 1 Separated NE 2

Separating 1 – ok. Separating 2 – not OK


50

Pooling 1 – not OK Pooling NE 2 – not OK

Major – economic analysis/economics


Economics as in major world universities
• Macro or micro focus
• More game theory and analysis in general
Skills:
• Asking right (causal) questions & finding precise answers
• Understanding, analysing and interpreting data (thus human behavior)
• Analytic and critical thinking
Personal contact with faculty and research
Graduates demanded by:
• Advisory, consulting, banking & finance, policy, energy and other industries
Requirements: knowledge of math and stats
• Into math intensive course
• Passing this course = good start

Modified Beer or Quiche 2


Mistake in these graps – the chance is wrong; it should be 0.5 and 0.5 (instead of 0.3 and 0.7). I am
too lazy to rewrite this…
Separated NE 1 – not OK Separated NE 2 – not OK
51

Pooling NE 1 - possible

Pooling NE 2 – not OK

Pooling equilibrium:
P1W,S -> Q
P2(Q) - F
P2(B) – F
Q=0.3, p <13/20

You might also like