1

Game Theory

PG

Dr. Dipankar Das

Email: [email protected]

Normal form game:

(1) A finite set of players, N= {1,2,3, 4, …, n}

(2) A set of actions for each player.

(3) A set of payoff functions for each player that give a payoff

value to each combination of the players’ chosen actions.

(4) A collection of sets of pure strategies, { S S , S , S

1 2 3 4 .. . , S}
n

Strategy:

A strategy is a complete contingent plan for a

player in the game.

A strategy is often defined as a plan of action intended to

accomplish a specific goal.

Pure strategy:

A pure strategy for player i is a deterministic plan of action.

 2

The set of all pure strategies for player i is denoted si i.e

s=s 1 , s 2 , s 3 , .. . , s n
Strategy profile: A strategy profile is a vector of strategies, one for

each player. In other words, a strategy profile describes strategies for

all of the players in the game.

For example, suppose we are studying a game with n players. A

typical strategy profile then is a vector

s=s 1 , s 2 , s 3 , .. . , s n where si is

the strategy of player i, for i = 1, 2, 3, n. Let S denote the set of

strategy profiles. Mathematically, we write S = S 1 × S 2 ×.. . .× S n . . Note

that the symbol “× ” denotes the Cartesian product.

A profile (or combination) of pure strategies s={s 1 , s 2 , s 3 ,... , s n } for all

i=1,2,3,..,n describes a particular combination of pure strategies

chosen by all n players in the game.

 3

Example

If say, you measure time in intervals of 15 minutes, then there are a

total of 12 time units in the three-hour window.

Let you have an examination in three hours and you must decide

how long to study for the exam and how long to just relax knowing

that your classmates are facing the same choice.

Here pure means deterministic action

The set of pure strategies is then

S = {1,2,….3,12}
i

Where, si ∈ S i
determines how many 15 units you will spend

{ }
S1 S 2 S 3 S 4 . .. .. ..... .. .... S7 . .. S 12
Si = 1 2 3 4 . .. .. ..... .. ... 7 . .. 12
↓ ↓ ↓ ↓ . .. .. ..... .. ..... ↓ . .. ↓
studying for the exam. 15 30 45 60 . .. .. ..... .. ...... 105 . .. 180

For example, if you chose si =7 then you will spend 1 hour and 45

minutes studying and 1 hour 15 minutes relaxing.

 4

Cartesian Product:

A={a,b,c,d}

B= {e,f,g,h}

Then, Cartesian product of these sets as A X B = {(a,e),(a,g),(a,h),)

(b,e),….,(d,h)}

(5) A set of payoff functions,{ v 1 ,v 2 ,v 3 ,v 4 ,...v n },each assigning a

payoff value to each combination of chosen strategies , that is

a set of functions v i : S 1 XS 2 XS 3 X . .. XS n → R for each i ∈ N

Where, (s 1 ,s 2 , s 3 , s 4 ,... ., sn ) ∈ R is the strategy profile that was selected

by the agents.

⟨ N , {S i }ni=1 , {v i (.)}ni=1 ⟩
The normal form game will be a triple of sets:

N= Set of players

{ Si }i=1 = Set of all players’ strategy i.e. { S S , S , S S}

n
1 2 3 4 .. . , n
 5

n
{ v i (.) }i=1= Set of all players’ payoff functions over the strategy

profiles of all the players, i.e. { v 1 ,v 2 ,v 3 ,v 4 ,...v n }.

*Given a single player i , we often need to speak of the strategies

chosen by all of the other players in the game. As a matter of

notation, it will be convenient to use the term − i to refer to these

players. Thus, s−i is a strategy profile for everyone except player i:

s−i = (s1, s2, c, si−1, si+1, c, sn). Separating a strategy profile s into

the strategy of player i and the strategies of the other players, we write
−i

s = ( s , s ). For example, in a three-player game with the strategy

i

profile s = (B, X, Y), we have s−2 = (B, Y).2 Incidentally, sometimes

I refer to the “− i ” players as player i’s opponents, but note that this

expression is not literally descriptive for all strategic situations

because many games have cooperative elements.

Prisoner’s Dilemma (PD) Game:

Two suspects in a major crime are held in separate cells. There is

enough evidence to convict each of them of a minor offense, but not
enough evidence to convict either of them of the major crime unless
one of them acts as an informer against the other (finks). If they both
stay quiet, each will be convicted of the minor offense and spend one
 6

year in prison. If one and only one of them finks, she will be freed and
used as a witness against the other, who will spend four years in
prison. If they both fink, each will spend three years in prison. This
situation may be modelled as a strategic game:
Players The two suspects.
Actions Each player’s set of actions is {Quiet(Mum), Fink}.

Players: N= {1,2}

Strategies: S = {M,F} i.e. S = {M,F} , S

i 1 2= {M,F}

M=Mum/Quiet, F= Finks

Cartesian product of S1 XS 2= {(M,M),(M,F),(F,M),(F,F)}

Where, u (F,M)= Freed=-1, u (M,M)= One year in prison=-2,

1 1

u (F,F)=Three years in prison=-4, u (M,F)= Four years in prison

1 1

=-5.

Here, the payoff values have been decided considering the concept

that if freed (F,M), then no jail, but jail for one year if (M, M),

hence the difference is one; that is why (-2-(-1)=-1). Thereafter, jail

for three years, so the difference between the no jail value and the

jail for three years should be three (-1+(-3)=-4). In the end, the
 7

difference between no Jain and jail for four years should be four (-

1+(-4)=-5).

The suspect 1’s ordering of the action profiles, from best to worst,

is u 1( F , M )> u ( M , M )>u ( F , F )>u ( M , F )

1 1 1

For suspect 2

u2 ( M , F )=−1>u2 ( M , M )=−2>u2 ( F , F )=−4>u ( F , M )=−5 2

Player 2
M F

[
Player 1 M (−2 ,−2 ) (−5 ,−1)
F (−1 ,−5 ) (−4 ,−4 ) ]
Solution concept:
Iterated elimination of strictly dominated strategies.
Solution by iterated elimination of strictly dominated strategies.
¿
In an n- player game, strategy si ,for player i is strictly dominated if
there exists some strategy si// such that,
Π i ( s1 , s1 , ,s1 , ... s''i ,...s n, )>, Π i ( s1 , s1 , ,s1 , ... s'i ,...s n , ),

Nash equilibrium
¿ ¿
Π i ( si , s−1 )≥Π i (s 'i , s−i )
¿ s ' ∈ Si
For all i and for all N players.
Nash equilibrium in games with continuous strategies:
Continuous variables : For example si ∈ S i could be a price, which in
theory could take any positive value.
¿ ¿ ¿
Π i ( si , s−1 )≥Π i (s 'i , s−i )
 8

s ' ∈ Si
For all i and for all N players. In other words each player
s Π ( s ) max Π i ( si , s−1 ) '

choses to maximize
i i s i

First Order Condition (FOC)

∂ Π i ( si , s−i )
=0
∂s i for i=1,2, 3, n.
Solution of the PD game: (F, F).
Application of PD game in advertisement competition between two
firms:
Firm 1
Low High
Firm 2 Low 300,300 0,400
High 400,0 100,100
Low advertising and high advertising. Answer is (H, H)
Application of PD game: Working on a joint project
You are working with a friend on a joint project. Each of you can
either work hard or goof off. If your friend works hard then you prefer
to goof off (the outcome of the project would be better if you worked
hard too, but the increment in its value to you is not worth the extra
effort). You prefer the outcome of your both working hard to the
outcome of your both goofing off (in which case nothing gets
accomplished), and the worst outcome for you is that you work hard
and your friend goofs off (you hate to be “exploited”). If your friend
has the same preferences then the game that models the situation you
face is given in Figure, which, as you can see, differs from the
Prisoner’s Dilemma only in the names of the actions.
 9

player 1
Work hard Goof off
player 2 Work hard 2,2 0,3
Goof off 3,0 1,1
Application of PD game: Duopoly
In a simple model of a duopoly, two firms produce the same good, for
which each firm charges either a low price or a high price. Each firm
wants to achieve the highest possible profit. If both firms, choose
High then each earns a profit of $1000. If one firm chooses High and
the other chooses Low then the firm choosing High obtains no
customers and makes a loss of $200, whereas the firm choosing Low
earns a profit of $1200 (its unit profit is low, but its volume is high).
If both firms, choose Low then each earns a profit of $600. Each firm
cares only about its profit, so we can represent its preferences by the
profit it obtains, yielding the game in Figure.

Firm 2
q=0 (1-q)=1
Low price High price
Firm 1. p=0 Low 1000,100 -200,1200 price
(1-p)=1 High 0 price
1200,-200 600,600
PSNE = (HP, HP)=(600,600)
EU_1(Low)=1000*q +-200*(1-q)=1000q-200+200q=1200q-200
EU_1(High)=1200*q +600*(1-q)= 1200q-600q+600= 600q+600
EU_1(Low)= EU_1(High)
1200q-200=600q+600
 10

600q=800
q = 1.33 (q \in [0,1]); q>1

Apple
Samsung

In a simple model of a duopoly, two firms produce the same good for
which each firm charges either a low price or a high price. Each firm
wants to achieve the highest possible profit.
Battle of the Sexes (BOS):
Player 1
Opera Football
Player 2 Opera 2,1 0,0
Football 0,0 1,2
For the battle of the sexes; however, there is no strategy that is
always best: Playing F is the best if your opponent plays F and
playing O is the best if your opponent plays O.
NE: two
Bach or Stravinsky (cooperation):
Conflict and cooperation both are present in both the PD and BOS.
The next game is purely conflictual.

Matching of Pennies:
Two people choose, simultaneously, whether to show the Head or the
Tail of a coin. If they show the same side, person 2 pays person 1 a
dollar; if they show different sides, person 1 pays person 2 a dollar.
Each person cares only about the amount of money she receives, and
(naturally!) prefers to receive more than less. A strategic game that
 11

models this situation is shown in Figure 17.1. (In this representation

of the players’ preferences, the payoffs are equal to the amounts of
money involved. We could equally well work with another
representation—for example, 2 could replace each 1, and 1 could
replace each −1.)

Player 1
Head Tail
Player 2 Head 1,-1 -1,1
Tale -1,1 1,-1
In this game the players’ interests are diametrically opposite. This is
called “Strictly Competitive”. Player 1 wants to take the same action
as the other player, whereas player 2 wants to take the opposite
action.
NE: no
Economic example:
Game between new firm and established firm.
The established producer prefers the newcomer’s product to look
different from its own, so that its customer’s will not be tempted to
buy the newcomer’s product, whereas the newcomer prefers that the
products look alike.
Stag Hunt:
Each of a group of hunters has two options: she may remain attentive
to the pursuit of a stag, or catch a hare. If all hunters pursue the stag,
they catch it and share it equally; if any hunter devotes her energy to
catching a hare, the stag escapes, and the hare belongs to the defecting
hunter alone. Each hunter prefers a share of the stag to a hare. The
strategic game that corresponds to this specification is:
 12

Players The hunters.

Actions Each player’s set of actions is {Stag, Hare}.
Preferences” For each player, the action profile in which all players
choose Stag (resulting in her obtaining a share of the stag) is ranked
highest, followed by any profile in which she chooses Hare (resulting
in her obtaining a hare), followed by any profile in which she chooses
Stag and one or more of the other players chooses Hare (resulting in
her leaving empty-handed).
For each player, stag is ranked highest followed by any profile in
which she chooses hare.
Player 1
Stag Hare
Player 2 Stag 2,2 0,1
Hare 1,0 1,1
NE: two
If one player remains attentive to pursuit of the stag, then the other
player prefers to remain attentive; if one player chooses a hare, the
other one prefers to chase a hare.
Competitive equilibrium and imperfect (duopoly) equilibrium:
Problem:
Players (firms) 1 and 2 produce some goods. Assuming that
there are no fixed costs of production, and let the variable cost
to each firm i of producing quantity q ≥0 be given by the cost
i

function,
C ( q )=q
2

i i i for i∈ {1 , 2}

Demand is given by the function

 13

q =100− p
where, q=q 1 +q 2
Competitive equilibrium:
MC=2 qi , equating this to the price gives us each first supply
function.
2 qi = p
p p
or , q i= i. e . q 1=q2 =
2 2

Adding up the two supply functions yields the market supply,

p p
. q1 + q 2 = +
2 2
Or ,q= p[ ∵ q=q1 +q2 ]

Equating this to the demand (i.e. equating market demand and

market supply) yields
(P=100-P), or P=50
and and plugging this into the supply function yields
q = p2 =502 =25 . for i = 1, 2
i
.

profit
Π i =pq−c =(100−q )q i−q2i =(100−50)25−252 = Rs.625.

Duopoly equilibrium (introduced by Augustin Cournot

(1838)):
According to Cournot, the competitive equilibrium is naive
(inexperienced) because rational firms should understand that
the price is not given; but rather determined by their actions.
From above example if firm 1 realizes its effect on the market
price, and produces q 1= 24 instead q 1= 25 (competitive
equilibrium), then the price will have to increase to
P(24+25)=100-49=51.(quantity of commodity 24 for firm 1
and 25 for firm 2)
 14

Π 1 =51×24−24 2=625 .

This means firm 1 instead look for the best choice it can
make. Its best choice depends on the quantity that firm 2 will
produce- what will be that?
The actions are choices of quantity, and payoffs are the
profits.
N= {1, 2}
Strategy sets: S1=[ 0 , α ] for , i∈ {1 , 2} and firms choose quantities si ∈ S i .

Payoffs: For i,j∈ {1,2} i¿ j

v i ( s i , s j )=
{
( 100−si −s j ) s i−s 2i ifsi + s j <100
−s 2i ifsi + s j ≥100

Beliefs, Mixed Strategies, and Expected Payoffs:

We use the term belief for a player’s assessment about the strategies
of the others in the game.
For example, consider the prisoners’ dilemma game, in which each
player chooses between strategies M and F. Player 1 may say to
himself, “I think the other guy is likely to play strategy F” or “I think
the other guy will surely play strategy M.” These are two alternative
beliefs that player 1 could have. Unfortunately, a statement such as “I
think the other guy is likely to play strategy F” is somewhat
ambiguous. Because we aim to model decision making
mathematically, we need a precise way of representing a player’s
beliefs. The key is to use probabilities. Here is an illustration.
Continuing with the prisoners’ dilemma example, suppose we let p
represent the likelihood that player 1 thinks the other player will
select strategy M in the game. Formally, p is a probability—a number
between zero and one—where p = 1 means that player 1 is certain that
player 2 will select strategy M, and p = 0 means that player 1 is sure
player 2 will not choose M. The probability that player 1 believes the
other player will select F is 1 − p. Thus, if p = 1>2, then player 1
believes that the other player is equally likely to select M and F. The
 15

numbers p and 1 − p constitute a probability distribution over the set

{M, F}.

Player 2
M F

[
Player 1 M (−2 ,−2 ) (−5 ,−1)
F (−1 ,−5 ) (−4 ,−4 ) ]
Note that if p ∈ (0, 1), then player 1 thinks it is possible that player 2
will play strategy M and also thinks it is possible that player 2 will
play F. Understand that player 1 might not believe that player 2
actually randomizes (by, say, flipping a coin). It’s just that player 1
may be uncertain about what strategy player 2 will choose, and so he
associates probabilities with player 2’s strategies. Furthermore, player
l’s belief may not be accurate; he could be certain that player 2 will
select M when in fact player 2 selects F.
Mathematically, a belief of player i is a probability distribution over
the strategies of the other players. Let us denote such a probability
distribution θ and writeθ ∈ Δ S , where Δ S is the set of
−i −i −i −i

probability distributions over the strategies of all the players except

player i . (We often use Greek letters, such as theta, here, to represent
probability distributions.)
For example, take a two-player game (so that − i = j ) and suppose
each player has a finite number of strategies. The belief of player i
about the behaviour of player j is a functionθ j ∈ Δ S j such that, for
each strategy θ j ∈ Δ S j of player j, θ j ( s j ) is interpreted as the probability
that player i thinks player j will play s j . As a probability distribution,
θ has the property that θ ∈ Δ S ¿ 0 for eachθ ∈ Δ S , and
j j j j j

∑s ∈ s θ j ( s j )=1
j j

Example:
16
 17

Dominance and Best Response

Figure-6.1a
Strategy D is dominated by strategy U. Neither of player 2’s strategy
is dominated.
Figure-6.1b
Strategy D is dominated by strategy M. U is not dominated by M.
Figure-6.1.c
For player 1, no pure strategy is dominated by another pure strategy.
Obviously, neither U nor M dominates the other, and D
does not dominate these strategies. In addition, neither
U nor M dominates D.
For instance, although U is better than D when player 2
selects L, D performs better than U when player 2
selects R. However, a mixed strategy dominates D.
Consider player 1’s mixed strategy of selecting U with
probability 1/2, M with probability 1/2, and D with
probability zero. We represent this mixed strategy as
(1/2, 1/2, 0). If player 2 selects L, then this mixed
 18

strategy yields an expected payoff of 2 to player 1; that

is, 2 = 4(1/2) + 0(1/2) + 1(0). Player 1 does worse by
playing D. The same is true when player 2 selects R.
Therefore, strategy D is dominated by the mixed
strategy (1/2, 1/2, 0).

profiles s−i ∈ s of the other players.

−i

The First Strategic Tension and the Prisoners’

Dilemma:
Player 2
M F

Player 1
[
M (−2 ,−2 ) (−5 ,−1 )
F (−1 ,−5 ) (−4 ,−4 ) ]
Here M is dominated by F. So, we would predict that neither would
select M. Bothe players would be better off if they each selected M.
Major tension in strategic settings:
The clash between individual and group interests.
The Concept of Efficiency:

the prisoners’ dilemma, (M, M) is more efficient than (F, F).

Furthermore, (M, M), (M, F), and (F, M) are all efficient strategy
profiles. In the game pictured in Figure 6.1(c), (D, R) is more efficient
than both (M, L) and (U, R). In this game, (U, L) and (D, L) are the
only efficient strategy profiles.
Book: Joel Watson: Page:40; Questions:1,3.

Belief and Best Response:

 19

A belief of player i is a possible profile (combination) of his

opponent’s strategies, s−1 ∈ S−1 . Given that a player has a particular
belief about his opponent’s strategies he will be able to formulate a
best response to that belief.
Example: BOS
Alex
O F
Chris O 2,1 0,0
F 0,0 1,2

When Chris belief that Alex is going to the opera, his unique best
response is to go to the opera. Similarly, if he believes that Alex will
go to the Football game, then he should go to the game.
For every unique belief there is a best response.
Mapping beliefs into a choice of action and this choice of action must
be a best response to the beliefs.
This is called player i’s best response function.
Example:
For some beliefs a player will have more than one best response
strategy.
P2

L C R

[ ]
U 3,3 5,1 6,2
M 4 ,1 8,4 3,6
P1 D 4 ,0 9,6 6,8

For R the best responses are U, D

For L the best responses are M, D
Best Response Correspondence of player i selects for each s−1 ∈ S−1 . a
subset BRi ( s−i )∈ S i where each strategy si ∈ BRi ( s−i ) is a best response to
 20

s−i . If he has a unique best response to s−i the BRi ( s−i ) will contain only

one strategy from Si .

 21

Example: Cournot Duopoly:

Demand equation p(q)= 100-q

Cost C i ( q i )=10 q i for i∈ {1 , 2} i.e. for both firms.

Firm 1 maximizes its profits
Π 1 ( q1 , q 2 )=(100−q1 −q 2 )q 1−10 q1

F.O.C.
90−2 q1 −q 2=0
90−q2
Or ,q 1=
2

If firm 1 believes that firm 2 will choose the quantity q2 then it should
choose q 1 according to the best response function.

{
90−q2
BR 1 ( q 2 )= if , 0≤q2 ≤90
2
0 if , q 2 >90
 22

Example:
There may be more than one best response to a given belief. In a finite
game every belief has at least one best response.

Player 1’s believes are (1/3,1/2,1/6)

If player 1 selects strategy U he excepts a payoff of
1/3.2+1/2.0+1/6.4=8/6
If player 1 selects strategy M he excepts a payoff of
1/3.3+1/2.0+1/6.1=7/6
If player 1 selects strategy D, he excepts a payoff of
1/3.1+1/2.3+1/6.2=13/6.

Thus his best response is strategy D, the strategy that yields the
greatest expected payoff given his belief. Strategy D is his only best
response
 23

BR2 (1>2, 1>4, 1>4) = {L, R}.

Dominance and Best Response Compared

UDi = The set of strategies for player i that are not strictly dominated.
Let, Bi be the set of strategies for player i that are best responses, over
all of the possible beliefs of player i.
Mathematically,

Example:

For player 2:
R is the dominated strategy for player 2.

And
 24

For player:1
For player 1 neither U nor M is dominated. D is dominated by neither
U nor M.
Now calculate “Mixed Strategy”:
D is dominated by the mixed strategy with probability 1/3 for U, 2/3
for M and 0 for D.
(1/3,2/3,0).
Set of UD1 = {U,M}.
Now derive the set of best responses.
Expected payoff of playing U with beliefs {p,(1-p)}.
i.e. 6.p+0.(1-p)=6p
Expected payoff of playing M is
2p+(1-p)5
Expected payoff of playing D is
p.3+(1-p).3

Conclusion:
Strategies are best responses if and only if they are not strictly
dominated.
 25

One question remains: Is strategy Z an element of B1? This question is

answered by the third step of the procedure. Note that if a mixed
strategy were-

fies to p > 1/3. Because there is no value of p that simultaneously

Therefore, Z ∈ B1. In fact, you can verify that Z is a best response to

satisfies p < 1/4 and p > 1/3, we know that Z is not dominated.

the belief u2 = (3/5, 2/5).

 26

Weak dominance

Example:

If player 1 were to select Y, then player 2’s strategy M

delivers a strictly higher payoff than does L. If player 1
selects X, then strategies L and M yield the same payoff
for player 2. Thus, player 2 always weakly prefers M to
L, and she strictly prefers M in the event that player 1
picks Y. This means that M weakly dominates L.
In relation to best-response behaviour:

Problems & Solutions:

Book: Joel Watson: Page:63; Questions:1,2,4.

 27

Extensive form game:

Example:

In one, player 2 knows that player 1 did not select A

when player 2 has to decide between C and D. In the
other, player 2 has no such information.

Sequential move game:

BOS: From simultaneous move to sequential move.
Example: BOS: simultaneous move
Chris
O F
Alex O 2,1 0,0
F 0,0 1,2

Example: BOS: Sequential move

 28

Information set:
 29

Firm 1 decides whether to be aggressive in the market (A), to be

passive in the market (P), or to leave the market (O). If firm 1 leaves,
then firm 2 enjoys a monopoly. Otherwise the firms compete and firm
2 selects whether or not to assume an aggressive stance. Furthermore,
when firm 2 makes its decision, it knows only whether firm 1 is in or
out of the market; firm 2 does not observe firm l’s competitive stance
before taking its action. In this game, there is one information set for
firm 1 (the initial node) and one for firm 2. The strategy sets are S1 =
{A, P, O} and S2 = {A, P}.
 30

Case 1:
Every player i has a collection of information sets hi ∈ H i . When
h2 ={x i }, this means that the information set at x i is a singleton( it
 31

includes only the node x i . Hence player 2 has information that says “I
am at x i ’.
Case: 2
When player 2 has two information sets h2 ={x 1 }and , h2 ={x 2 }.

Case :3
A game in which player 2 does not know whether he is at x1 or x2 , the
it must be the case that his information is
“I know that I am at either x1 or x2 , but I do not know at which of the
two I am ”.
i.e. h2 ={x 1 , x 2 }
Player 1 choose from the action set A 1 = {O, F}, and player 2 chooses
from A2 ={o, f} without observing the choice of player 1.
32
33
34
35
36
37