Beruflich Dokumente
Kultur Dokumente
• Auctions
• Voting systems
6
Basic Concepts
• All games have three basic elements:
– Players
– Strategies
– Payoffs
7
Players
• A player is a decision maker and can be anything
from individuals to entire nations.
10
Equilibrium Concepts
• In the theory of markets an equilibrium occurred
when all parties to the market had no incentive
to change his or her behavior.
11
Rules and Actions
• The rules of the game determine the timing of
players’ moves and the actions players can
make at each move.
• S1 = {U;D}and S2 = {L;R}.
• If a player has a dominant strategy, his problem is simple: he should clearly play
that strategy.
• If each player has a dominant strategy, the equilibrium of the game is obvious:
each player plays his own dominant strategy.
An Overview of Game Theory
Assumptions:
Beta
Air Lion Low High
Low 3,000, 3,000 1,000, 6,000
High 6,000, 1,000 4,000, 4,000
• A strategy that works at least as well as any other
one, no matter what the other player does, is
known as a dominant strategy.
• Each firm must play a best response to what the other firm
is doing.
• It is no coincidence that
Beta’s strategy is credible.
What about Air Lion’s strategy?
• Since Beta is playing
a dominant strategy,
we know that no
matter what Air Lion
does,
• Beta’s equilibrium
strategy does at
least as well as any
other.
• In summary, the dominant strategy
equilibrium that we found satisfies both the
Nash condition and the credibility condition.
The Prisoner’s Dilemma
• The Prisoner’s Dilemma is a game in which
the optimal outcome for the players is
unstable.
41
The Prisoner’s Dilemma
– The DA separates the suspects and tells each, “If you
confess and your companion doesn’t, I can promise
you a six-month sentence, whereas your companion
will get ten years.
42
The Prisoner’s Dilemma
B
Confess Not confess
A: 3 years A: 6 months
Confess
B: 3 years B: 10 years
A A: 10 years A: 2 years
Not confess
B: 6 months B: 2 years
43
The Prisoner’s Dilemma
• The normal form of the game is shown in Next
table
– The confess strategy dominates for both players so
it is a Nash equilibria.
45
Nash Equilibrium
• A Nash equilibrium is a pair of strategies (a*,b*)
in a two-player game such that a* is an optimal
strategy for A against b* and b* is an optimal
strategy for B against A*.
H 5,4
• Next, B chooses
L either H or L, but the
large oval
A surrounding B’s two
L 6,4
B decision nodes
H indicates that B does
H 6,3 not know what
choice A made.
48
An Illustrative Advertising Game
• The numbers at the end of
each branch, measured in
7,5 thousand or millions of
dollars, are the payoffs.
L
B
– For example, if A chooses H
and B chooses L, profits will
H 5,4 be 6 for firm A and 4 for firm
L B.
A
• The game in normal (tabular)
L 6,4 form is shown in Table where
B A’s strategies are the rows
H
and B’s strategies are the
H 6,3 columns.
49
Table: The Advertising Game in Normal
Form
B’s Strategies
L H
L 7, 5 5, 4
A’s Strategies H 6, 4 6, 3
7,5
L
B
H 5,4
L
A L 6,4
B
H
H 6,3 50
Dominant Strategies and Nash
Equilibria
• A dominant strategy is optimal regardless of the strategy adopted by an opponent.
– As shown in Table or Figure , the dominant strategy for B is L since this yields a larger
payoff regardless of A’s choice.
• If A chooses H, B’s choice of L yields 5, one better than if the choice of H was made.
• If A chooses L, B’s choice of L yields 4 which is also one better than the choice of H.
B’s Strategies
L H
L 7, 5 5, 4
A’s Strategies H 6, 4 6, 3
51
Dominant Strategies and Nash
Equilibria
• A will recognize that B has a dominant strategy and choose the strategy
which will yield the highest payoff, given B’s choice of L.
– A will also choose L since the payoff of 7 is one better than the payoff from
choosing H.
B’s Strategies
L H
L 7, 5 5, 4
A’s Strategies H 6, 4 6, 3
52
Dominant Strategies and Nash
Equilibria
• In other words, the payoff of 𝑖 from playing 𝑠𝑖 is always (i.e. under all
possible choices of other players) lower than the payoff from playing
𝑠𝑖′ .
• We can conclude that Player 1 will never play U, and so our game reduces to the
matrix.
• But Player 2 should know that Player 1 will never play U, and
• If Player 1 never plays U then some of Player 2’s strategies are strictly
dominated!
Player 2 Player 2
Player 1 C R
Player 1 C
M 1,5 4,4
D 3,6
D 3,6 3,0
Example
Player 2
Player 1
• R is strictly dominated by C
• How do we proceed?
Pure and mixed strategies
• The Prisoners' Dilemma game, each player has the strategies C and D.
• A player could also do the following: play C with probability 𝑝 and play D
with probability (1 − 𝑝), where 𝑝 is some number between 0 and 1.
• This is why Nash equilibrium is the central solution concept in game theory.
• A strategy profile (𝑠1∗ , 𝑠2∗ , … . . , 𝑠𝑛∗ ) is a Nash equilibrium (in pure strategies) if it is a
mutual best response.
• In other words, for every player 𝑖, the strategy 𝑠𝑖∗ is a best response to 𝑠−𝑖
∗
. (as
explained above, this is the strategy profile of players other than 𝑖).
• In other words, if (𝑠1∗ , 𝑠2∗ , … . . , 𝑠𝑛∗ ) is a Nash equilibrium, it must satisfy the
∗
property that given the strategy profile 𝑠−𝑖 of other players.
• Player 𝑖 cannot improve his payoff by replacing 𝑠𝑖∗ with any other strategy.
• To find out the Nash equilibrium of the game,
we must look for the mutual best responses.
• Let p be the probability that Serena chooses DL (down the line), so 1‐p is the
probability that she chooses CC (Cross court).
• Let q be the probability that Venus positions herself for DL, so 1‐q is the
probability that she positions herself for CC.
50p+10(1‐p) = 20p+80(1‐p), or
50p+10‐10p = 20p+80‐80p, or
40p+10 = 80‐60p, or
100p = 70, so
p = 70/100 = .70.
• If Serena plays DL with probability p=.70 and CC with probability 1‐p=.30, then Venus’s success rate
from
• Since this is a constant sum game, Venus’s success rate is 100%‐Serena’s success rate =
100‐62=38%.
Column Player’s Optimal Choice of q
• Choose q so as to equalize the payoff your
opponent receives from playing either pure
strategy.
• This requires understanding how you opponent’s
payoff varies with your choice of q.
• Graphically, in our example:
• A strictly mixed strategy Nash equilibrium in a 2
player, 2 choice (2x2) game is a p > 0 and a q > 0
such that p is a best response by the row player to
column player’s choices, and q is a best response by
the column player to the row player’s choices.
• However, as you will see below, the game does have a Nash
equilibrium in mixed strategies.
• A Nash equilibrium can then be defined in the
usual way: a profile of mixed strategies that
constitute a mutual best response.
• and
• Player 2's choice of q must be such that player 1 is indifferent between playing A1 or B1.
• In other words, in equilibrium q must be such that 1(A1) = 1(B1), i.e. we have:
3𝑞 + 2(1 − 𝑞) = 2𝑞 + 3(1 − 𝑞)
• But if player 2 is going to choose q = 1/2 it must be that he is indifferent between A2 and
B2 (otherwise player 2 would not want to mix, but would play a pure strategy).
1 = 3p
which implies p = 1/3.
• Therefore, the mixed strategy Nash
equilibrium is as follows.
• The best-response functions. They cross only at E, which is the mixed strategy
Nash equilibrium.
• If he goes across, he
would get 2; if he goes
down, he will get 3.
• He will go down.
• Therefore, we reduce
the game as follows:
• Therefore, he goes
down.
Matching Pennies (Perfect
Information)
• If player 1
chooses Head,
player 2 will
Head; and if 1
chooses Tail,
player 2 will
prefer Tail, too.
• Information set is
represented as in
Figure by connecting
the two nodes by a
dotted line
Imperfect information: information
sets
• Player 2 knows he is at the
information set (after player
1 moves),
• Initially player 1
decides whether to
come in (and play
some game with
player 2) or stay out
(in which case player
2 plays no role).
• If the choice is to come in,
player 1 decides between A
and B.
• However, subgame
perfection still works.
Pure strategy subgame perfect Nash
equilibria.
• Consider, for instance, the centipede game in Figure 11.1, where the
equilibrium is drawn in thick lines. This game has three subgames.
One of them is:
• Here is another subgame:
• Any subgame other than the entire game itself is called proper.
Matching penny game
• This game has three subgames:
• 2 possible actions at
the right node (after
a2).
Actions and strategies
• So there are 2 X 2 = 4 possible
strategies for player 2. These
are:
Player 1 b1 b2
a1 3,1 1,0
a2 4,1 0,1
Player 1 b1 b2
a1 3,1 1,0
a2 4,1 0,1
• (a2; b1b1) no player can do better by deviating. Therefore, (a2; b1b1) is a Nash equilibrium.
• (a1; b1b2). This is indeed a Nash equilibrium. It is not enough to write (a1; b1).
• Unless we know what player 2 would have played in the node that was not reached.
• These are the three pure strategy Nash equilibria of the game.
Repeated Prisoners’ Dilemma
• In a one-shot game, rational players
simply play their dominant
strategies. So (D;D) is the only
possible equilibrium.
• Each player plays an action (in this case either C or D) in each period.
• So in each period, the players end up playing one of the four possible
action profiles (C;C), (C;D), (D;C) or (D;D). Let at denote the action
profile played in period t.
• This can also arise from a simple rate of interest calculation, in which
𝛿 case can be interpreted as 1/(1 + 𝑟), where r is the rate of interest.
• Higher values of indicate that players are more patient (i.e. value future
payoffs more).
• If 𝛿 is very low, the situation is almost like a one-shot game, since players
only value today's payoff, and place very little value on any future payoff.
• Given such discounting, the payoff of player i in
the repeated game is:
Cooperation through trigger
strategies
• The history at t is the action profile played in
every period from period 0 𝑡𝑜 𝑡 ≥ 1.
• In period 𝑡 ≥ 1:
• In period t:
If (C;C) is played in t - 1, play (C;C)
Otherwise play (D;D).
• which implies:
• Thus if the players are patient enough,
cooperation can be sustained in equilibrium.
• The result about the large set of payoffs that can be sustained as
equilibrium outcomes is known as the `folk theorem'.
• Folk theorem is adapted to the repeated Prisoners' Dilemma
game.
• At any period t > 0, if (C;C) has been played in the last period,
play (C;C). Otherwise, switch to (D;D).
4
• which implies 𝛿 ≥
5
(b) Under what conditions is there an
equilibrium in the infinitely-repeated game in
which players alternate between (C,C) and (D,D),
starting with (C,C) in the first period?
1 3 6
3∗ +1 = = 1.5
4 4 4
1 3
• 𝑎2 𝑤𝑖𝑡ℎ 𝑝𝑟𝑜𝑏𝑎𝑏𝑙𝑖𝑡𝑦𝑜𝑓 𝑏1 𝑎𝑠 & 𝑏2 𝑎𝑠
4 4
1 3 8
2∗ +2∗ = =2
4 4 4
3 1 10
3∗ +1∗ = = 2.5
4 4 4
1 3
• 𝑎2 𝑤𝑖𝑡ℎ 𝑝𝑟𝑜𝑏𝑎𝑏𝑙𝑖𝑡𝑦𝑜𝑓 𝑏1 𝑎𝑠 & 𝑏2 𝑎𝑠
4 4
3 1 8
2∗ +2∗ = =2
4 4 4
• If player 1 chooses ‘𝑷𝟏 ’, Player 2 then chooses between ‘𝑷𝟐 ’ and ‘𝑺𝟐 ’.
• The payoffs are written as (Payoff to 1, Payoff to 2). Identify any subgame
perfect Nash equilibrium.
3rd
• Two firms face the following strategic pricing
problem.
• Each firm can set a high price H or a low price
L.
The payoffs are as follows:
• Suppose this pricing game between the firms is repeated
infinitely.
• Your answer must specify the strategies that the firms should
follow in the repeated game to sustain cooperation.