____________________________________________________________________________________________________

Subject ECONOMICS

Paper No and Title Paper 1: Quantitative Methods-I (Mathematical methods)

Module No and Title Module 40: Game Theory- II

Module Tag ECO_P1_M40

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II
____________________________________________________________________________________________________

TABLE OF CONTENTS

1. Learning Outcomes
2. Introduction
3. Mixed Strategy Nash equilibrium
4. Examples
5. Game: Its various classes
6. Economic Applications
7. Summary

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II
____________________________________________________________________________________________________

1. Learning Outcomes

After reading this module, you shall be able to:

1. Follow the concept of mixed strategy Nash equilibrium for a two player game.
2. Apply the concept of mixed strategy technique using the probability approach.
3. See to the application of the said approach by means of various examples and
problems in economics.

2. Introduction

Game theory is the technique or mathematical tool that deals with the general analysis of
strategic interdependence of players playing with their defined strategies. A game is
described by means of payoff matrix describing the payoffs to each of the players in view
of strategic moves they make. In many games associated with real life problems, pure
strategy Nash equilibrium may exist but there are numerous situations where no such
Nash equilibrium exists.
In this module, we discuss in detail the alternative method to find an equilibrium for such
problems. Since no pure strategy works for both or either of the players, a combination of
the strategies available to a particular player may work out in his search to attain
equilibrium. Next question that comes up is: as to what type of combination of strategies
should be made by a player so as to attain the equilibrium. For this, the concept of
probability is associated with each strategy. Each strategy is assigned a probabilistic
value such that sum of all associated probabilities is 1. Using the logical argument of
indifferent attitude of a player’s payoff with respect to each of his strategies, the assigned
probability is determined and the final payoff can be computed. Such an equilibrium is
referred to as mixed strategy Nash equilibrium.
The concept of game theory in models of learning theory are much simpler to handle
because of the underlying assumption that players have only fixed number of fixed
strategies. However this requires to include mixed strategy approach in order to ensure
the existence of equilibrium. So the concept of mixed strategy Nash equilibrium is
comparatively more general.
In this module we discuss this concept and see to its applications by means of various
examples and illustrations. An attempt has been made to address to an approach of
applying this concept to problems in economics

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II
____________________________________________________________________________________________________

3. Mixed Strategy Nash Equilibrium

In our first module on game theory, we had studied the concept of Nash equilibrium
which is a pair of strategies where each player is making the optimal choice given the
other person’s choice or it is a pair of expectations about each person’s choice such that
when the other person’s choice is revealed, neither individual wants to change his
strategy. However it should be observed that for a game, Nash equilibrium of the type
discussed above may or may not exist. Moreover even if Nash equilibrium exists, it may
or may not be unique i.e. a game may have more than one Nash equilibrium.
For example, let us again consider a simple game which had been already discussed in
the previous module on game theory. Suppose there are two players A and B in a game
where player A will write one of the two words from ‘left’ and ‘right’ on a piece of paper
and player B will write one of the two words from ‘up’ and ‘down’ on a piece of paper .
Both players will write simultaneously but none of them can see what the other player
has written. Once they have written their choices, papers will be read and they will get
the pay off according to the pay off matrix given below:

A B Up Down

Left 3,2 1,1

Right 1,1 2,3

In this matrix the payoff to A is the first entry in the box and pay off to B is the second
entry. Now as discussed regarding the Nash equilibrium it can be clearly seen that have
(Left, Up) and (Right, Down) both are Nash equilibrium. This is because if A chooses
Left B will choose Up and also if B chooses Up, A will choose Left. Similarly if A
chooses Right , B will choose Down and if B choose Down , A will choose Right. So by
this case it can be seen that Nash equilibrium of a game may not be unique.
Now let us consider another game of the same type where pay off matrix is given by :
A B Up Down

Left 1,1 1,0

Right 2,1 0,4

Here it can be seen that, Nash Equilibrium of the type described above does not exist.
This is because if player A chooses Left, player B would play Up , but if player B plays
Up , player A would choose Right . Similarly, if player A chooses Right, B would play
Down but if B plays Down A would choose Left. So this is an example of a game for
which Nash equilibrium (of the type discussed above) does not exist.

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II
____________________________________________________________________________________________________

For such situations, a new type of Nash Equilibrium can be found based on mixed
strategies which would be referred to as Mixed Strategy Nash Equilibrium. As seen in
the example considered above , since no pure strategy can be an optimal choice for a
player , so the player might choose to play each of his choices a part of the time only and
hence playing with mixed strategy. In this case the player can randomize their strategies
by assigning a probability to each of their choice and then play their choices according to
the assigned probabilities.
We first consider a game in general and discuss the computation of mixed strategy Nash
equilibrium . Later we can apply it to particular examples of the type discussed above
where pure strategy Nash equilibrium failed to exist. Let us consider a game where pay
off matrix is given by :
A B b1 b2

a1 u11 ,v11 u12 ,v12

a2 u21 ,v21 u22 ,v22

Here uij ,vij (i,j=1,2) are integers

Here player A has two choices of action given by a1 and a2 and B can also play with two
strategies given by b1 and b2 . Suppose A chooses to play his strategy a1 with probability
p and (1-p) be the probability with which player A chooses to play with strategy a2 . Let
q be the probability with which player B chooses to play his strategy b 1 and (1-q) for
strategy b2 . Clearly 0≤p ,q≤1.
Let us denote by αi (x) the probability assigned by ith player to his action x where i=
A,B,x= ai or bi ,i=1,2……..
So αA(a1)=p, αA(a2)=1-p, αB(b1)=q, αB(b2)=1-q
As the players choices are taken to be independent , so when players use mixed
strategies αA and αB ,the probability of any action pair say (ai , bj ) (i,j=1,2) is given by the
product αA(ai) αB(bj). Hence the probabilities of mixed strategy pair over the four possible
outcomes is given by the probability distribution generated by mixed strategy pair in the
following table:
A B b1 b2

a1 Pq p(1-q)
a2 (1-p)q (1-p)(1-q)

Now the expected pay off of a player to the mixed strategy is the sum of the products of
pay offs urs ,vrs (r,s= 1,2) of the player with the corresponding probability of mixed
strategies . So expected pay off for player A is given by u11 pq + u12 p(1-q)+ u21(1-p)q +

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II
____________________________________________________________________________________________________

u22(1-p)(1-q). Here coefficient of p represents A’s expected pay off where he uses a pure
strategy assigning probability 1 to strategy a1 with player B using mixed strategy αB.
Similarly coefficient of (1-p) represents A’s expected payoff if it assigns probability 1 to
strategy a2 with player B using mixed strategy αB .
On the same lines ,player B’s expected pay off is given by
q[v11 p + v21 (1-p) ]+(1-q)[ v12 p+ v22 (1-p)]
Again in case of player B the interpretations for coefficients of q and (1-q) can be made
as done above for player A .
Now to a given mixed strategy of player B , player A has three possibilities for his best
response :
(i) A’s unique best response is strategy a1 which is true if
u11q+ u12 (1-q)> u21 q+ u22 (1-q).
(ii) A’s unique best response is strategy a2 which is true if
u11q+ u12 (1-q)< u21 q+ u22 (1-q).
(iii) All mixed strategies of player A will yield same expected pay off
if u11q+ u12 (1-q)= u21 q+ u22 (1-q).-------------------------(1)

However for 0<p<1 , the mixed strategy (p,1-p) is never the unique best response. In
general all mixed strategies are best responses.
Similarly for player B , the best response to a given mixed strategy of player A is again
given by v11p+ v12 (1-p)= v21 p+ v22 (1-p) for the case (2) when all mixed strategies of B
will yield same expected pay off.
So mathematically, we need to find probability p and q ( if they exist ) which satisfy
equation (1) and (2) .
Solving (1) we get the value of q and solving (2) gives the value of p . This is based on
the fact that in equilibrium A must be indifferent between the strategies available to him
to be mixed for B’s given mixed strategy. Similar arguments hold for indifference with
player B as well .
We now illustrate this concept of mixed strategy Nash equilibrium by considering some
simple games for which pure strategy Nash Equilibrium does not exist.

4. Examples

(1) We again consider the same example, we discussed in the beginning for which no
pure strategy Nash Equilibrium existed. The pay off matrix is given by :

A B Up Down

Left 1,1 1,0

Right 2,1 0,4
ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)
Module 40: Game Theory- II
____________________________________________________________________________________________________

Let p and (1-p) be the probability with which player A chooses strategy left and right and
q and (1-q) be the probability with which player B chooses strategy Up and down with 0
≤p ,q ≤1. Now probability distribution table is formed as q
(1-q)
A B b1 b2

a1 Pq p(1-q)
a2 (1-p)q (1-p)(1-q)

Now A’s expected pay off against (q,1-q) is obtained by equating coefficients of p and
(1-p) in p[1.q+1.(1-q)]+(1-p)[2q+0.(1-q)]

ie. q+1-q=2q

or 2q=1 or q=1/2
Similarly B’s expected pay off against (p,1-p) is obtained by equating coefficients of q
and (1-q) in
q[1.p+1.(1-p)]+(1-q)[0.p+4.(1-p)]
ie. 1=4(1-p)
or 4p=3 or p=3/4
So player A mixes his strategies with weight (3/4,1/4) for Left and Right and player B
mixes his strategies with weight (1/2,1/2) for Up and Down .
Hence the mixed strategy Nash Equilibrium is given by [(3/4,1/4),(1/2,1/2)].

(2)Penny Matching Game

Consider a simple game between two players A and B. Each of them has a penny with
which they will secretly turn to heads or tails. Their choices would be revealed
simultaneously. If both the pennies match ie. both heads or both tails , player A keeps
both pennies which will give +1 for A and -1 for B. On the other hand if the pennies do
not match , player B keeps both pennies +1 for B and -1 for A. Clearly it is a case of two
person zero sum game. We present the pay off by means of the following pay off matrix
A B Head Tails

Heads +1,-1 -1,+1

Tails -1,+1 +1,-1

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II
____________________________________________________________________________________________________

We first check for pure strategy Nash equilibrium. If player A chooses Heads, player B
would choose tails but if player B choose Tails , A would choose Tails. Similarly if A
chooses tails B would choose Heads but if B choose Heads then A would choose Heads
again. So pure strategy Nash Equilibrium fails to exist in the case. Hence each of the
players will have to randomize or mix their strategies in order to attain equilibrium.
Again let p and (1-p) be the probabilities with which A mixes his strategies of choosing
Heads and Tails and q and (1-q) be the probabilities in case of player B. As seen earlier,
the probability distribution table can be formulated.
Now for A’s expected pay off from the Heads against (q,1-q)
=q.1+(1-q)(-1)
=2q-1
And A’s expected pay off from tails against (q,1-q)
=q(-1)+(1-q)(1)
=1-2q
Now when A mixes both of his strategies , then both must yield the same expected pay
off ( in order to attain Nash Equilibrium).
ie. 2q-1=1-2q
or 4q=2
or q=1/2
This is because in equilibrium A must be indifferent in his strategies involved to solve for
B’s equilibrium mixed strategy.
Now for player B ,we find A’s equilibrium.
So B’s expected pay off from Heads against (p,1-p)
=p(-1) +(1-p)(1)
=1-2p
And B’s expected pay off from tails against (p,1-p)
=1.p+(1-p)(-1)
=2p-1
Again in equilibrium, B must be indifferent between both of his strategies. Therefore
1-2p=2p-1
or 4p=2
or p=1/2
So both players choose both strategies with probabilities ½ each. Hence the Nash
equilibrium is given by ((1/2,1/2),(1/2,1/2))

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II
____________________________________________________________________________________________________

5. Game: Its various classes

Amongst various classes of game available in literature, we discuss a few in this module.
They are coordinate games , competition games , co existence games etc. which we
illustrate by means of a particular examples.
(i) Coordinate games :Assurance game

Going back to year 1950, when both the countries U.S and U.S.S.R were in a race of
arms. Each country could build nuclear missiles or refrain itself from building missiles.
So each country had two strategies available with it –to build or to refrain. Following
their respective strategies, the pay off matrix is of the following type.

U.S Refrain Build

U.S.S.R
Refrain 4,4 1,3
Build 3,1 2,2

From the above table it is quite clear that the best pay off for both the countries is to
refrain from building missiles which results in a pay off of (4,4) ie. 4 units to both the
countries. However if one country refrains and the other goes for building the missile, the
pay off would be three units to the country building the missile whereas just one unit to
the refrainer. And if both the countries build the missile each gets 2 units as its pay off.
Now if we examine the pay off matrix carefully it can be clearly seen that this game
which in particular is also called assurance game is a case of pure strategy Nash
equilibrium where (refrain, refrain) and (build, build) are both the Nash equilibrium.
However the better option for the both the countries is (refrain, refrain).But in case, for
instance, U.S refrains U.S.S.R builds the missiles, then pay off to U.S would be
comparatively, less than U.S.S.R and will not be preferred by U.S. Similarly if U.S.S.R
decides to refrain itself from building the missiles but U.S goes for building it, again
U.S.S.R would be getting a lower pay [Link],the country that opts to refrain might end up
in earning lower pay off if the other country goes for the building option, which can very
well be the case in practical terms. So although refrain is better for both the countries
neither country knows the choice the other country is going to make, so before
committing to refrain, the country in action wants clean assurance from the other country
that it will refrain too. One of the ways discussed in literature were that when the country
to play first announces its choice of refrain and convinces the other country with
sufficient evidence of its action of refrain, then the other country would definitely refrain
where it is assumed that the sole interest to each country is to maximize its payoff. So in
that case the Nash equilibrium would be taken as (refrain, refrain)
ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)
Module 40: Game Theory- II
____________________________________________________________________________________________________

(ii) Competition Games: Penalty point in game of football

Let us consider the game of football at a point where player A is the shooter for a penalty
point and player B is the goalkeeper defending the penalty shot. Now player A will kick
to right or left and so can player B to prevent the goal. Now in practice, A may be better
at kicking in one direction and so may be B in defending better in one of the two
directions, i.e. each of them may have his strong side in terms of kicking left or right or
defending left or right. Now as a particular case we assume that A will score a goal 80
percent of the times if he kicks to the left and B jumps to the right but only 50 percent of
times if B jumps to left. Also if A kicks to the right he succeeds in scoring a goal 90
percent of the times if B jumps to the left and 20 percent of times if B jumps towards the
right. So the payoff matrix is as follows
B
Defend Left Defend right
A
Shoot Left 50, -50 80,-80
Shoot Right 90, -90 20,-20

Since the objective of each of the players is to maximize his own payoff and minimize
the other player’s payoff so gain of one player would result in an equivalent loss of the
other player. This can be clearly seen in the above payoff matrix. Clearly it is a case of
two person zero sum game.
Now the strategy of player A would be broadly focused to his objective so as to keep
player B guessing all the time. So A would keep kicking to left and right sides both,
though his strong side may be any one of them. As a result, Player A will have to follow
a mixed strategy. Suppose A kicks left with probability p and right with probability (1-p).
Now the probability distribution table can be formed in the similar way as discussed in
earlier articles. Now A’s expected payoff when B jumps to left = -50p -90(1-p) and A’s
expected payoff when B jumps to the right = -80p -20(1-p). Now A must be indifferent
between both of his strategies and should get the same payoff by mixing both of his
strategies.
i.e. 50p + 90(1-p) = 80p +20(1-p)
or 100p = 70
or p = 0.7
So A chooses to kick to the left 70 percent of times and 30 percent of the times he kicks
to the right. In that case, A’s expected pay off is 50 × 0.7 + 90 × 0.3 = 62.
Similarly we analyze the strategic behavior of player B, his expected payoff is given by
50q + 80(1-q) = 90q+20(1-q) in equilibrium

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II
____________________________________________________________________________________________________

or 100q = 60
or q = 0.6
So, Nash equilibrium is ((0.7,0.3),(0.6,0.4))
i.e. A should kick to the left with probability .7 and B should jump to the left to defend
with probability 0.6. In this case, if any one of them chooses his strategy with said
probability then whatever the other player does, A’s and B’s expected payoff will be the
same.
(iii) Coexistence game: Hawk dove game
This game is an example of animal interaction. In real life situations there cannot be any
game between Hawk and a dove. There will be always a single obvious outcome of any
game between Hawk and a dove. Here it is a game between two competent of equal
potential say two wild jackals which may have hawkish or dovish attitude or may differ
in attitude with each other. If they fight over a piece of food, it is a hawkish strategy
whereas if they agree on sharing it is dovish strategy. And it is all together possible that
one of them is hawkish and the other may be dovish. Now if both the jackals play dovish,
the payoff is say (2, 2). If one plays hawkish and other plays dovish, the one who plays
hawkish wins everything, i.e. (4, 0) or (0, 4) in that case. If both play hawkish, both will
be injured and payoff is (-2, -2) in that case. So the payoff matrix is given by
II
Hawk Dove
I
Hawk -2,-2 4,0
Dove 0,4 2,2
Where I and II represent first and second jackals.
Obviously, if the whole population of jackals plays hawk or whole as dove, it cannot be
equilibrium then, so there will be a mixture of hawk types and dove types in equilibrium.
Let p be the probability that jackal I would play hawk type and (1-p) for it to be dove
type. Similarly, let q be the probability of jackal II playing hawk and (1-q) for it to play
dove.
Now expected payoff of jackal II in equilibrium,
-2q+4(1-q) = 2(1-q)
or 4q =2
or q=½
Similarly, in equilibrium jackal I’s expected payoff is given by
-2p+4(1-p) = 2(1-p)
or 4p =2
or p=½

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II
____________________________________________________________________________________________________

So both the jackals play hawk and dove with probability ½. Hence there will be a 50-50
type doves and hawks in equilibrium.
Remark: This hawk and dove game can be related to economical problems when the
payoff matrix on similar lines can be written as
Firm II
Hawk Dove
Firm I
v−c v−c
Hawk , v,0
2 2
v v
Dove 0,v ,
2 2
Here v is the value of resources that one of the firms will possess. If both the firms shared
the resources, their values are halved (i.e. dove attitude). Hawk attitude means the firm
stands a tough go and does not give up the resources. However, if both the firm play
hawks, both end up fighting and incur a cost of 2c on themselves so value coming up as
v c
2
−
2
.
If one of the firms plays hawk and other as dove, the hawk firm gets the whole resources.
Depending on if v > c or v < c, the firms would like to play their strategy. Here for v < c,
the firm would like to play hawk if opponent plays dove and would play dove if the
opponent plays hawk.

6. Economic applications of game theory

Consider two firms A and B with a choice of entering or refraining from a market which
is not large enough for both to survive profitably. If they both enter they both will
experience losses in payoff of 5 units. If none of them enters their payoffs are 0 each. Of
the two whichever only firm enters the market first, will receive a payoff of 10 units. So
the payoff matrix is given as follows:
B
Enter Doesn’t Enter
A
Enter -5, -5 10, 0
Doesn’t Enter 0, 10 0, 0
Clearly two pure strategy Nash equilibrium are for one firm to enter and other not. The
one entering will receive a payoff of 10 units

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II
____________________________________________________________________________________________________

7. Summary

For a two person game a pure strategy Nash equilibrium may or may not exist.
1. For a finite strategy two person game, with no Nash equilibrium each strategy
is chosen with some probability such that sum of all probabilities is 1

2. For a two person (A,B) game each with two strategies (I,II), probabilities p
and (1-p) assigned to the strategies of one player and q and (1-q) are assigned
to the strategies of the other player, The probability distribution table is
formed as
B
I II
A
I Pq p(1-q)
II (1-p)q (1-p)(1-q)
3. For the pay
off matrix

B
I II
A
I u11v11 u12v12
II u21v21 u22v22
The payoff of each of players is same with respect to both of his strategies. So
equilibrium can be obtained by solving
u11q+u12(1-q)=u21q+u12(1-q) for q
and
v11p+v12(1-p)=v21p+v12(1-p) for p
Mixed strategy Nash equilibrium is then given by
((p,(1-p)),(q,(1-q)))
4. Various types of games like coordination games, competition games etc can
be solved by either pure strategy Nash equilibrium approach or mixed strategy
Nash equilibrium.

ECONOMICS Paper 1: Quantitative Methods-I (Mathematical methods)

Module 40: Game Theory- II