Claudia Vogel
Mathematics for IBA
Winter Term 2009/2010
Outline
5. Game Theory
• Introduction & General Techniques
• Sequential Move Games
• Simultaneous Move Games with Pure Strategies
• Combining Sequential & Simultaneous Moves
• Simultaneous Move Games with Mixed Strategies
• Discussion
Claudia Vogel: Game Theory and Applications 2
Motivation: Why study Game Theory?
• Games are played in many situation of every days life
– Roomates and Families
– Professors and Students
– Dating
• Other fields of application
– Politics, Economics, Business
– Conflict Resolution
– Evolutionary Biology
– Sports
Claudia Vogel: Game Theory and Applications 3
The beginnings of Game Theory
1944 “Theory of Games and Economic Behavior“
Oskar Morgenstern & John Neumann
Claudia Vogel: Game Theory and Applications 4
Decisions vs Games
• Decision: a situation in which a person chooses from different
alternatives without concerning reactions from others
• Game: interaction between mutually aware players
– Mutual awareness: The actions of person A affect person
B, B knows this and reacts or takes advance actions; A
includes this into his decision process
Claudia Vogel: Game Theory and Applications 5
Sequential vs Simultaneous Move Games
• Sequential Move Game: player move one after the other
– Example: chess
• Simultaneous Move Game: Players act at the same time and
without knowing, what action the other player chose
– Example: race to develop a new medicine
Claudia Vogel: Game Theory and Applications 6
Conflict in Players‘ Interests
• Zero Sum Game: one player’s gain is the other player’s loss
– Total available gain: Zero
– Complete conflict of players’ interests
• Constant Sum Game: The total available gain is not exactly
zero, but constant.
• Games in trade or other economic activities usually offer
benefits for everyone and are not zero-sum.
– Example: Joint ventures
Claudia Vogel: Game Theory and Applications 7
One-shot vs Repeated Games
• One-shot Game: only one interaction between the players
– No knowledge about the opponent
• Repeated Game: repeated interactions
– With the same opponent: Reputation
Example: Long-Term Business Relations
– With changing opponents: information about usual
behavior
Example: Price setting in the Turkish Bazaar
Claudia Vogel: Game Theory and Applications 8
Information
• Perfect information: all information is available to all players
– Example: chess
• Imperfect Information
– External uncertainty: uncertainty about relevant variables
(e.g. weather)
– Strategic uncertainty: about opponents past moves
• Asymmetric Information: some information is only available to
one player
– Example: job market
Claudia Vogel: Game Theory and Applications 9
Cooperative vs Non-Cooperative Games
• Cooperative Game: players are able to enforce contracts
– Example: The European Union
• Non-cooperative Game: cooperation has to be self-enforcing
as it is not enforceable by an outside party
– Example: The European Union and non-member countries
Claudia Vogel: Game Theory and Applications 10
Terminology
• Strategy: (choices) action available to the players
• Payoff: the number associated with each possible outcome for
each player
– Expected Payoff: probability weighted average payoff
• Rationality: A player has a consistent set of payoffs over all
possible outcomes and calculates the strategy that best
serves his interests
• Equilibrium: Each player is using the strategy that is the best
response to the strategies of the other players
Claudia Vogel: Game Theory and Applications 11
Dynamics and Evolutionary Games
• Evolutionary Games allow dynamic processes in which
strategies that proved to be better are more likely to be
chosen in future games.
• Each player has a particular “programmed” strategy and
meets players with the same or other “programmed”
strategies better strategies multiply faster; worse strategies
decline
• Questions for the analysis: Does the dynamic process
converge to an evolutionary stable state? Does just one
strategy survive in the end or can several strategies coexist?
Claudia Vogel: Game Theory and Applications 12
Observation and Experiment
• Theory always should relate to reality in two ways:
– Reality should help structure theory.
– Reality should provide a check on the results of theory.
• Reality of strategic interactions can be found out by
– Observation
– Special experiments
Claudia Vogel: Game Theory and Applications 13
Introduction to Probabilities
• Probability: The probability of a random event is a quantitative
measure of the likelihood of ist occurence. For events that can
be observed in repeated trials, it is the long-run frequency
with which it occurs. For unique events or other situations
where uncertainty may be in the mind of a person, other
measures are constructd, such as subjective probability.
• Addition Rule: If you devide a set of events into a number of
subsets, none of which overlap, then the probabilities of each
subset occuring must sum to the probability of the full set of
events; if that full set includes all possible outcomes, then ist
probability is 1.
• Expected value: The probability-weighted average of the
outcomes of a random variable, that is, ist statistical mean or
expectation.
Claudia Vogel: Game Theory and Applications 14
Summary
• Strategic game situations are distinguished from individual
decision-making situation by the presence of significant
interactions among the players.
• Players have strategies that lead to different outcomes with
different associated payoffs.
• Payoffs incorporate everything that is important to a player
and are calculated by using probalistic averages or
expectations if outcomes are random and include some risk.
• Game Theory may be used for explanation, prediction, or
prescription in various circumstances
Claudia Vogel: Game Theory and Applications 15
Outline
5. Game Theory
• Introduction & General Techniques
• Sequential Move Games
• Simultaneous Move Games with Pure Strategies
• Combining Sequential & Simultaneous Moves
• Simultaneous Move Games with Mixed Strategies
• Discussion
Claudia Vogel: Game Theory and Applications 16
Game Tree 1/3
• Game Tree: Representation of a game in the form of nodes,
branches, and terminal nodes and their associated payoffs.
• Node: is a point from which branches emerge, or where a
branch terminates.
• Branch: Each branch emerging from a node represents one
action that can be taken at that node.
• Terminal node: represents an end point in a game tree, where
the rules of the game allow no further moves, and payoffs for
each player are realized.
• Move: an action at one node of a game tree.
• Strategy: A complete plan of action for a player in a game,
specifying the action he would take at all nodes where it is
turn to act.
Claudia Vogel: Game Theory and Applications 17
Game Tree 2/3
Ann
Go Stop
Chris Bob
3 1
safe risky 2
3 Nature 10 Ann
5 6 Deb
Bad Good up down
3 50% 50% 1 low high
1 1
2 6 0 1.3 2 1
8 3 -2.7 2 7 -2
-1 4 0 -11 4 3
2 0 0 3 1 0
Claudia Vogel: Game Theory and Applications 18
Game Tree 3/3
Smoking Decision Smoking Game
Today's
try not Carmen
0 try not
continue not
Future 0
-1 +1 Carmen
continue not
-1 +1
+1 -1
Claudia Vogel: Game Theory and Applications 19
Solving Game Trees 1/3
• Rollback: Analyzing the choices that rational players will make
at all nodes of a game, starting at the terminal node and
working backward to the initial node.
– prune: identify and eliminate from a game tree those
branches that will not be chosen when the game is
rationally played
• Rollback equilibrium: The strategies (complete plans of
action) for each player that remain after rollback analysis has
been used to prune all the branches that can be pruned.
Claudia Vogel: Game Theory and Applications 20
Solving Game Trees 2/3
Today's
Carmen
try not
Future 0
Carmen
continue not
-1 +1
+1 -1
Claudia Vogel: Game Theory and Applications 21
Solving Game Trees 3/3
Ann
Go Stop
Chris Bob
3 1
safe risky 2
3 Nature 10 Ann
5 6 Deb
Bad Good up down
3 50% 50% 1 low high
1 1
2 6 0 1.3 2 1
8 3 -2.7 2 7 -2
-1 4 0 -11 4 3
2 0 0 3 1 0
Claudia Vogel: Game Theory and Applications 22
Order Advantages
• First-mover advantage: it is beneficial to set the stage for a
later player
• Second-mover advantage: beneficial to react to the moves of
the first player
– Example: price setting
• In some games the result is determined by the setup of the
game and the order of moves doesn’t matter
– Example: removing matchsticks from a pile
Claudia Vogel: Game Theory and Applications 23
Experimental Results
• Ultimatum Bargaining: Player A has the possibility to divide
an amount of 5 Euro between himself and player B. After the
proposal of the allocation, player B decides whether to accept
or not.
− If player B accepts both receive the money according to
the offer.
− If player B does not accept, both get nothing.
• Assuming that player A can only offer in steps of 1 Euro coins,
he can either offer 0,1,2,3,4 or 5 Euro.
Claudia Vogel: Game Theory and Applications 24
Game Tree Ultimatum Bargaining
A
0 1 2 3 4 5
B B B B B B
y n y n y n y n y n y n
5 0 4 0 3 0 2 0 1 0 0 0
0 0 1 0 2 0 3 0 4 0 5 0
Claudia Vogel: Game Theory and Applications 25
Summary
• Sequential move games require players to consider the future
consequences of their current moves before choosing their
actions.
• Game trees illustrate the games. They consist of nodes and
branches that show all possible actions of each player and the
associated outcomes.
• The rollback equilibrium is a complete plan of actions of all
players.
• Different types of games entail advantages for different
players, e.g. first-mover advantages.
Claudia Vogel: Game Theory and Applications 26
Outline
5. Game Theory
• Introduction & General Techniques
• Sequential Move Games
• Simultaneous Move Games with Pure Strategies
• Combining Sequential & Simultaneous Moves
• Simultaneous Move Games with Mixed Strategies
• Discussion
Claudia Vogel: Game Theory and Applications 27
Simultaneous Move Games
• Players choose their actions at exactly the same time.
• No information about what the other player has done or what
he will do.
• Also called games of imperfect information or imperfect
knowledge.
• Many situation can be described by such games: producers in
an industry usually make their decisions about product design
and features without knowing, what their competitors are
doing, voters in elections vote simultaneously.
• Strategies cannot be made contingent on the other player’s
action, the player can only reason through the game from the
perspective of the opponent.
Claudia Vogel: Game Theory and Applications 28
Illustration of Simultaneous Move Games
• Game matrix: A table whose dimension equals the number of
players in the game; the strategies available to each player
are arrayed along one of the dimensions (row, column); and
each cell shows the payoffs of all players in a specified order,
corresponding to the configuration of strategies which yield
that cell. Also called game table or payoff table.
• Game matrix is called the normal form or strategic form of a
simultaneous move game.
• Pure Strategies: actions that are available to each player.
• Mixed Strategy: a random choice to be made with the
specified probabilities, from a player’s originally specified pure
strategies.
Claudia Vogel: Game Theory and Applications 29
Example of a Game Table
Column
Left Middle Right
Top 3,1 2,3 10,2
High 4,5 3,0 6,4
Row
Low 2,2 5,4 12,3
Bottom 5,6 4,5 9,7
Claudia Vogel: Game Theory and Applications 30
Nash equilibrium
• Nash equilibrium: A configuration of strategies (one for each
player) such that each player’s strategy is best for him, given
those of the other players. (Can be in pure or mixed
strategies.)
• No player should want to change his strategy once he has
seen what his rivals did, i.e. each action chose is the best
response to the other players’ actions.
John Nash, mathematician, economist,
Nobel Prize winner
Claudia Vogel: Game Theory and Applications 31
Finding the Nash equilibrium
Column
Left Middle Right
Top 3,1 2,3 10,2
High 4,5 3,0 6,4
Row
Low 2,2 5,4 12,3
Bottom 5,6 4,5 9,7
Claudia Vogel: Game Theory and Applications 32
Dominance
• Dominant strategy: A strategy X is dominant if, for each
permissible strategy configuration of the other player, X gives
him a higher payoff than any of his other strategies. That is,
his best response is always X.
• Dominated strategy: A strategy Y is dominated for a player if
there is another strategy X such that, X always gives a higher
payoff.
• Successive (or iterated) elimination of dominated strategies:
Considering the players in turns and repeating the process in
rotation, eliminating all strategies that are dominated for one
at a time, and continuing doing so until no such further
elimination is possible.
• If the process yields a unique outcome the game is
dominance solvable.
Claudia Vogel: Game Theory and Applications 33
The Prisoners’ Dilemma 1/2
Consider a story line from a typical television crime program: A
husband and wife are under suspicion that they were conspirators in
a murder. The detectives place them in separate detention rooms
and interrogate them one at a time. There is little concrete evidence
linking the pair to murder, although there is some evidence they
were involved in kidnapping the victim. The detectives explain to
each suspect that they are both looking at jail time for the
kidnapping charge (3 years) even if there is no confession from
either of them. In addition they are told that the detectives “know”
how one had been coerced by the other to participate in the crime; it
is implied that jail time for a solitary confessor will be significantly
reduced (to only 1year instead of 25 for murder) if the whole story is
committed. If both confess, jail terms could be negotiated down (10
years) but not as much as in one confession and one denial.
Claudia Vogel: Game Theory and Applications 34
The Prisoners’ Dilemma 2/2
Wife
CONFESS DENY
(Defect) (Cooperate)
CONFESS
10yr, 10yr 1yr, 25yr
(Defect)
Husband
DENY
25yr, 1yr 3yr, 3yr
(Cooperate)
Claudia Vogel: Game Theory and Applications 35
One player Dominance
Federal Reserve
Low interest High interest
rates rates
Budget
3, 4 1, 3
Balance
Congress
Budget
4, 1 2, 2
Deficit
Claudia Vogel: Game Theory and Applications 36
Weak Dominance
Player 2
a b
x 1,1 1,0
Player 1
y 0,1 1,1
Claudia Vogel: Game Theory and Applications 37
Classes of Games
• one Nash-equilibrium in pure strategies
− Prisoner’s Dilemma
• multiple Nash-equilibria in pure strategies
− Pure coordination
− Assurance
− Battle of Sexes
− Chicken
• no equilibrium in pure strategies
Claudia Vogel: Game Theory and Applications 38
The Prisoners’ Dilemma
Wife
CONFESS DENY
(Defect) (Cooperate)
CONFESS
10yr, 10yr 1yr, 25yr
(Defect)
Husband
DENY
25yr, 1yr 3yr, 3yr
(Cooperate)
Each player has two strategies, Cooperate and Defect,
such that for each player Defect dominates Cooperate
and the outcome (Defect, Defect) is worse for both than
the outcome (Cooperate, Cooperate)
Claudia Vogel: Game Theory and Applications 39
Pure coordination
Sally
Starbucks Local Latte
Starbucks 1,1 0,0
Harry
Local Latte 0,0 1,1
All players are indifferent among all Nash equilibria, and
coordination is needed only to ensure avoidance of a
non-equilibrium outcome.
Claudia Vogel: Game Theory and Applications 40
Assurance
Sally
Starbucks Local Latte
Starbucks 1,1 0,0
Harry
Local Latte 0,0 2,2
All players prefer the outcome at Local Latte, as this
leads to higher payoffs. This equilibrium can act as a
Focal Point.
Claudia Vogel: Game Theory and Applications 41
Battle of Sexes
Sally
Starbucks Local Latte
Starbucks 2,1 0,0
Harry
Local Latte 0,0 1,2
Each player has a Hard and a Soft strategy. Each player
prefers the outcome where he is Hard and the other is
Soft, but both prefer the Nash equilibria to the other two
possibilities
Claudia Vogel: Game Theory and Applications 42
Chicken
Dean
Swerve Straight
(Chicken) (Tough)
Swerve
0, 0 -1, 1
(Chicken)
James
Straight
1, -1 -2, -2
(Tough)
Each player has a tough and a weak strategy. Each
player prefers the strategy where he is tough and the
other is weak, but the outcome (tough, tough) is worst
for both.
Claudia Vogel: Game Theory and Applications 43
No equilibrium in pure strategies
Navratilova
DL CC
DL 50,50 80,20
Evert
CC 90,10 20,80
Claudia Vogel: Game Theory and Applications 44
Summary
• Simultaneous move games differ from sequential move games
in that players make decisions without knowing their rivals’
actions.
• The games are illustrated in game tables.
• The solution is the Nash equilibrium, which exists when each
player chooses that strategy that is best for him, given that all
other players are using their equilibrium strategies.
• Nash equilibria can be found by successive elimination of
dominated strategies or cell-by-cell inspection
• Specific Games include Prisoners’ Dilemma, Coordination
games such as assurance, chicken and battle of sexes and
matching pennis.
Claudia Vogel: Game Theory and Applications 45
Outline
5. Game Theory
• Introduction & General Techniques
• Sequential Move Games
• Simultaneous Move Games with Pure Strategies
• Combining Sequential & Simultaneous Moves
• Simultaneous Move Games with Mixed Strategies
• Discussion
Claudia Vogel: Game Theory and Applications 46
Changing the order of moves in a game
• Sequential-move games become simultaneous if the players
cannot observe moves made by their rivals
− Analysis: search for Nash equilibrium rather than
rollback!
• Simultaneous-move games become sequential if one player
is able to observe the other’s move before choosing her own
• Any changes to the rules of the game can change its
outcome
Claudia Vogel: Game Theory and Applications 47
Example 1: No change in outcome 1/2
Wife
CONFESS DENY
(Defect) (Cooperate)
CONFESS
10yr, 10yr 1yr, 25yr
(Defect)
Husband
DENY
25yr, 1yr 3yr, 3yr
(Cooperate)
Claudia Vogel: Game Theory and Applications 48
Example 1: No change in outcome 2/2
Sequential play:
Husband moves first Wife moves first
Claudia Vogel: Game Theory and Applications 49
Example 2: First-mover advantage 1/2
Dean
Swerve Straight
(Chicken) (Tough)
Swerve
0, 0 -1, 1
(Chicken)
James
Straight
1, -1 -2, -2
(Tough)
Claudia Vogel: Game Theory and Applications 50
Example 2: First-mover advantage 2/2
Sequential play:
James moves first Dean moves first
Claudia Vogel: Game Theory and Applications 51
Example 3: Second-mover advantage 1/2
Navratilova
DL CC
DL 50,50 80,20
Evert
CC 90,10 20,80
Claudia Vogel: Game Theory and Applications 52
Example 3: Second-mover advantage 2/2
Sequential play:
Evert moves first Navratilova moves first
Claudia Vogel: Game Theory and Applications 53
Example 4: Both players do better 1/2
Federal Reserve
Low interest High interest
rates rates
Budget
3, 4 1, 3
Balance
Congress
Budget
4, 1 2, 2
Deficit
Claudia Vogel: Game Theory and Applications 54
Example 4: Both players do better 2/2
Sequential play:
Congress moves first Fed moves first
Claudia Vogel: Game Theory and Applications 55
Illustrating simultaneous-move games by
using game trees
Claudia Vogel: Game Theory and Applications 56
External Uncertainty
Claudia Vogel: Game Theory and Applications 57
Analyzing sequential-move games in strategic
form
Claudia Vogel: Game Theory and Applications 58
Further Definitions 1/2
• Subgame: A portion of a larger game, starting at a non initial
node of the larger game
• Off-equilibrium path: a path of play that does not result from
the players´ choices of strategies
• Off-equilibrium subgame: a subgame starting at a node, that
does not lie on the equilibrium path of play
• Continuation: the remaining part of the plan of action of that
strategy, applicable to the subgame that starts at this node
• Subgame-perfect equilibrium: A configuration of strategies
(complete plans of action) such that their continuation in any
subgame remains optimal (part of a roll back equilibrium),
whether that subagme is on- or off-equilibrium.
Claudia Vogel: Game Theory and Applications 59
Further Definitions 2/2
Claudia Vogel: Game Theory and Applications 60
Games with sequential and simultaneous
moves
A
e r
2 B
2
x c
A A
b t b t
3 0 0 1
1 0 0 3
Claudia Vogel: Game Theory and Applications 61
Summary
• When illustrating simultaneous games with a game tree
information sets contain those nodes in which players do not
have information about previous actions.
• Changing the rules of a game to alter the timing of actions can
alter the equilibrium outcome of the game:
– A sequential move game has one rollback equilibrium, but
the same game played simultaneously may have several
Nash equilibria.
– A sequential game analyzed from its strategic form may
have many possible Nash equilibria.
– By using the criterion of credibility, one can eliminate some
strategies; this leads to the subgame-perfect equilibrium of
the sequential game.
Claudia Vogel: Game Theory and Applications 62
Outline
5. Game Theory
• Introduction & General Techniques
• Sequential Move Games
• Simultaneous Move Games with Pure Strategies
• Combining Sequential & Simultaneous Moves
• Simultaneous Move Games with Mixed Strategies
• Discussion
Claudia Vogel: Game Theory and Applications 63
Mixed Strategies
• pure strategy: nonrandom plan of action for each player
• mixed strategy: the actual move is chosen randomly from the
set of pure strategies with some specific probabilities
• Under general conditions every simultaneous move game has
a Nash equilibrium in mixed strategies.
Claudia Vogel: Game Theory and Applications 64
Example
Player 1
x y z
a 4,10 1,0 1,3
Player 2
b 7,0 0,10 10,3
Claudia Vogel: Game Theory and Applications 65
Best responses 1/3
Player 1
p x y (1-p)
a q 4,10 1,0
Player 2
b (1-q) 7,0 0,10
Expected value of player 1:
E1=10pq +0(1-p)q +0p(1-q) +10(1-p)(1-q) =20pq-10p-10q+10
Expected value of player 2:
E2=4pq+1(1-p)q+7p(1-q) +0(1-p)(1-q) = - 4pq+7p+q
Claudia Vogel: Game Theory and Applications 66
Best responses 2/3
Best response of player 1 to q of player 2:
E1=p(20q-10)-10q+10
q>0.5 (20q-10)>0 best response: p=1
q=0.5 (20q-10)=0 best response: p any
q<0.5 (20q-10)<0 best response: p=0
Best response of player 2 to p of player 1:
E2=q(- 4p+1)+7p
p<0.25 (- 4p+1)>0 best response: q=1
p=0.25 (- 4p+1)=0 best response: q any
p>0.25 (- 4p+1)<0 best response: q=0
Claudia Vogel: Game Theory and Applications 67
q
Best responses 3/3
1
Nash equilibrium
0,5
0,25 0,5 1 p
Claudia Vogel: Game Theory and Applications 68
Mixed Strategies in Chicken
Dean
swerve straight
p (1-p)
swerve q 0,0 -1,1
James
straight (1-q) 1,-1 -2,-2
Expected value of DEAN:
ED=0pq +1(1-p)q -1p(1-q) -2(1-p)(1-q) =p(1-2q)+3p-2
Expected value of JAMES:
EJ=0pq -1(1-p)q +1p(1-q) -2(1-p)(1-q) = q(1-2p)+3p-2
Claudia Vogel: Game Theory and Applications 69
Best responses 1/2
Best response of DEAN to q of JAMES:
ED= p(1-2q)+3p-2
q<0.5 (1-2q)>0 best response: p=1
q=0.5 (1-2q)=0 best response: p any
q>0.5 (1-2q)<0 best response: p=0
Best response of JAMES to p of DEAN:
E2=q(1-2p)+3p-2
p<0.5 (1-2q)>0 best response: q=1
p=0.5 (1-2q)=0 best response: q any
p>0.5 (1-2q)<0 best response: q=0
Claudia Vogel: Game Theory and Applications 70
q
Best responses 2/2
1
0,5
0,5 1 p
Claudia Vogel: Game Theory and Applications 71
General case
Player 1
p x y (1-p)
a q α,a β,b
Player 2
b (1-q) γ,c δ,d
Expected value of player 1:
E1=apq+b(1-p)q+cp(1-q)+d(1-p)(1-q)
=p[qa-qb+(1-q)c-(1-q)d]+qb+(1-q)d
Claudia Vogel: Game Theory and Applications 72
Best response of player 1 for (q,1-q) of player 2
E1=p[qa-qb+(1-q)c-(1-q)d]+qb+(1-q)d
1 for […]>0
p= any for […]=0
0 for […]<0
Analog process for player 2.
Claudia Vogel: Game Theory and Applications 73
Indifference Criteria
A mixed strategy equilibrium requests for both players to be
indifferent between their strategies […]=0
Solving […]=0 for p respectively q results in the formulas
δ-β d-c
p= q=
α-β-γ+δ a-b-c+d
Claudia Vogel: Game Theory and Applications 74
Summary
• If players want to be unpredictable, they do so by using a
mixed strategy that specifies a probability distribution over
their set of available pure strategies.
• Every simultaneous move game where each player has a
finite number of pure strategies has a Nash equilibrium in
mixed strategies.
• Equilibrium mixtures for two players must be best responses
to each other.
• Graphs of best-response curves for both players show all
equilibria of the game at the intersection(s) of the curves.
• Good use of mixed strategy requires that there be no
predictable system of randomization.
Claudia Vogel: Game Theory and Applications 75
Outline
5. Game Theory
• Introduction & General Techniques
• Sequential Move Games
• Simultaneous Move Games with Pure Strategies
• Combining Sequential & Simultaneous Moves
• Simultaneous Move Games with Mixed Strategies
• Discussion
Claudia Vogel: Game Theory and Applications 76
Empirical evidence concerning Nash
equilibrium
Do people actually play Nash equilibrium strategies in
experiments?
•Yes, in simple, single-move games with a unique Nash
equilibrium after some repetition
•Mixed evidence, for
– more complex or repeated situations
– coordination among multiple equilibria
– games with complex calculations of the Nash equilibrium
Claudia Vogel: Game Theory and Applications 77
Critical Discussion of the Nash equilibrium
concept 1/3
Column
A B C
A 2,2 3,1 0,2
Row B 1,3 2,2 3,2
C 2,0 2,3 2,2
Claudia Vogel: Game Theory and Applications 78
Critical Discussion of the Nash equilibrium
concept 2/3
B
Left Right
Up 9,10 8,9.9
A
Down 10,10 -1000,9.9
Claudia Vogel: Game Theory and Applications 79
Critical Discussion of the Nash equilibrium
concept 3/3
1. The Nash equilibrium is too imprecise.
2. Players in actual games do not play Nash equilibrium
strategies.
3. Rationality does not itself imply Nash equilibrium.
Have considerable confidence in the Nash equilibrium
concept if it is played frequently by players from a reasonably
stable population and under relatively unchanging rules and
conditions.
Be cautious if the game is new, played just once and players
are inexperienced.
Treat games that are played repeatedly by the same group of
players as a game in its own right.
Claudia Vogel: Game Theory and Applications 80
