Monopolistic

Competition

Block 3
Imperfect Competition-II


 Oligopoly
UNIT 5 OLIGOPOLY

Structure

5.0 Objectives
5.1 Introduction
5.2 Oligopoly
5.2.1 Equilibrium in an Oligopolistic Market
5.3 The Cournot Model
5.3.1 Equilibrium using Residual Demand Curve
5.3.2 Equilibrium using Reaction Curves
5.3.3 Cournot Equilibrium with Different Costs
5.4 Collusion and Cartels
5.5 The Bertrand Model
5.5.1 Bertrand Paradox
5.5.1.1 Condition to be satisfied for the Bertrand Paradox
5.5.2 Bertrand Equilibrium using Reaction Curves
5.5.3 Comparison between the Cournot and the Bertrand Model
5.6 The Stackelberg Model
5.7 The Dominant Firm Model
5.8 Let Us Sum Up
5.9 Some Useful References
5.10 Hints or Answers to Check Your Progress Exercises

5.0 OBJECTIVES
After reading and studying the unit, you should be able to:

x understand the nature of strategic interaction in an Oligopolistic market

structure in terms of various models;
x find the equilibrium under various oligopoly models;
x compute Cournot equilibrium using the Residual demand curve and by
way of Reaction curves;
x throw light on the possibility of Collusions and Cartels among the rivals
in an Oligopoly;
x appreciate the Bertrand model of oligopoly as an alternative to the
Cournot approach;
x find the Stackelberg equilibrium; and
x explain the Dominant firm model.


Imperfect
Competition-II 5.1 INTRODUCTION
We have so far considered three different market structures, viz. Perfect
competition, Monopoly and Monopolistic competition. Perfect competition
and monopoly are the two extreme forms of market structures. In a perfect
competitive market, there is free entry and exit, many buyers and sellers,
perfect information, etc. Such an idealised market does not exist in reality.
Moreover, except for some natural monopolies, one hardly finds any
example of a monopolistic market structure. A step closer to the reality is
the monopolistic competition market structure which assumes no barriers
to entry and non-price competition (in the form of product differentiation)
between firms. But firms under monopolistic competition are naïve and
therefore do not involve in strategic interactions. This makes the market
structure under monopolistic competition rarely experienced in reality. The
present unit is a move towards a relatively more realistic market structure.
Most markets are better described as oligopolies. These are markets where
there exist more than one market player, yet where each firm is large
enough to enjoy some monopoly power. There are barriers to entry which
result in a market with a small number of dominant players dependent
strategically on each other’s decisions about output or price. The prominent
example of an oligopolistic market structure is that of soft drink suppliers,
Pepsi and Coca-cola. For instance, if price of Pepsi is lowered to attract more
sales, it will necessarily attract a reaction from Coca-cola suppliers whose
customers will get lured away by the lower Pepsi price. Unless Coca-cola is
willing to bear the revenue loss (which is unlikely to happen), it will find
itself obligated to respond to Pepsi’s price cut, which in turn would affect
Pepsi sales. Considering rival’s reaction thus becomes significantly important
in such a market condition. How do firms behave in an oligopolistic market
is what will be covered in this unit.

5.2 OLIGOPOLY
In oligopoly market structure a few firms account for most or all of the total
production. Barriers to entry and exit result in prevalence of a small number
of dominant players that remain strategically dependent on each other. By
strategically dependence it is meant, a firm while taking its optimising
decisions must consider the expected reaction of its rival. This is because,
output and/or price decision by one impacts output sales and hence the
revenue earnings of the other players in the market. Like in perfect
competition firms in oligopoly cannot take market price as it comes, neither
do they have the price-setting power like that with a monopolist who can
set the price and worry about no prospective retaliation. In oligopoly, firms
possess certain degree of price-setting power (or the monopoly power, i.e.,
the rate at which it can set the price above its marginal cost) depending on
their market structure, as they have to worry about retaliation from
 competitors. Any action by a firm in oligopolistic market is followed by a
counter reaction from other firms in that market. This compels each firm to Oligopoly

consider its rivals’ expected reactions as it decides about output and pricing.
It follows that accurately portraying the interactions between firms across a
wide range of possibilities is the critical task of any model of oligopoly.
There are two different ways in which firms might interact with each other
in an oligopolistic market structure. They may cooperate or choose not to
cooperate. If they collude, that is, choose to cooperate with each other so as
to maximise their joint profits, they form a cartel. For example, the
Organisation of Petroleum Exporting Countries (OPEC) is a cartel of oil
producing countries that cooperate in how much oil to produce in an
attempt to move the price of oil up or down in the market. On the other
hand, if they behave non-cooperatively, acting in their own self-interest,
they take into account the actions of other firms. Examples of a non-
collusive oligopoly model include— the Cournot model, the Bertrand model,
the Stackelberg model and the Dominant firm model. We will be covering
these models in the present Unit.

5.2.1 Equilibrium in an Oligopolistic Market

A firm’s equilibrium position refers to its profit optimising price and output
decisions in different market situations. Under perfect competition and
monopoly, outcomes are more or less certain with the former optimising at
the point where market determined price equals the marginal cost of
production; in the case of the latter, however, the firm possessing the
market power optimises by setting marginal revenue equal to the marginal
cost of production. However, under oligopoly no such certainty exists due to
the presence of a small number of dominant players who control the major
share of the market and who are strategically interdependent in terms of
pricing and output decisions. There exist several ways in which individual
oligopolists may respond to rivals' price and output decisions. Consequently,
several different models of oligopoly, viz. the Cournot model, the Cartel, the
Bertrand model, the Stackelberg model, the Dominant firm model, have
been developed, underpinned by different analytical approaches and
assumptions about the nature of oligopolistic market behaviour. The
discussion of these models is provided in the subsequent sections. However
before that, let us try to understand the nature of monopoly power of the
oligopolistic firm. Let us assume there are only two symmetric firms in the
market having the same constant marginal cost C (and zero fixed cost) and
the firms are involved in quantity competition. The demand function faced
by a firm ሺ݅ ൌ ͳǡ ʹሻ: ܲ ൌ ‫ܦ‬௜ ሺܳଵ ǡ ܳଶ ሻ, where P represents the market price
and Q1, Q2 are the quantity produced by firm 1 and 2, respectively. Let the
profit function of the firm ݅ሺ݅ ൌ ͳǡ ʹሻ be:

ߨ௜ (ܳଵ ǡ ܳଶ ሻ ൌ ܲ ൈ ܳ௜ െ ‫ܥ‬ሺܳ௜ ሻ ൌ  ‫ܦ‬௜ ሺܳଵ ǡ ܳଶ ሻܳ௜ െ ‫ܥ‬ሺܳ௜ ሻ


Imperfect Where C(Qi) is the cost function faced by firm i. First-order condition for
Competition-II
profit maximisation for any of the firm ݅ሺ݅ ൌ ͳǡ ʹሻ involves differentiating
the profit function with respect to ܳ௜ and put it equal to 0.
߲ߨ௜ ሺܳଵ ǡ ܳଶ ሻ ߲‫ܦ‬௜ ሺܳଵ ǡ ܳଶ ሻ ߲ܳ
ฺ ൌ Ͳ ֜ ܳ ൅  ‫ܦ‬௜ ሺܳଵ ǡ ܳଶ ሻ െ ‫ ܥ‬ൌ Ͳ
߲ܳ௜ ߲ܳ ߲ܳ௜ ௜
డ஽೔ ሺொభ ǡொమ ሻ డொ
ฺ ܳ௜ ൅ ܲ െ ‫ ܥ‬ൌ Ͳ
డொ డொ೔

[‫ ܲ ׶‬ൌ ‫ܦ‬௜ ሺܳଵ ǡ ܳଶ ሻሿ

Multiplying and dividing the first term on the LHS with Q×P, we get
డ஽೔ ሺொభ ǡொమ ሻ ொ ௉ொ೔
ฺ  ቂ ቀ ቁ ൅ ܲ െ ‫ܥ‬ቃ ൌ Ͳ
డொ ௉ ொ
డொ
     ሾ‫ ׶‬డொ ൌ ͳሿ
೔

Taking first term of the LHS on the RHS and dividing both the sides by P
ܲെ‫ܥ‬ ߲‫ܦ‬௜ ሺܳଵ ǡ ܳଶ ሻ ܳ ܲܳ௜ ͳ
ฺ ൌ ቆെ ቇ൬ ൰
ܲ ߲ܳ ܲ ܳ ܲ
௉ି஼ ଵ
ฺ ൌെ ‫ݏ‬
௉ ఌ೏ ௜

ฺThis is the equation for Lerner' s Index for Oligopoly.

௉ொ೔
Here, ‫ݏ‬௜ ൌ ௉ொ
is the share of firm ݅ in the total value of output; the
డ஽೔ ሺொభ ǡொమ ሻ ொ ଵ
reciprocal of market elasticity of demand is given by : డொ ௉
ൌ  ఌ . The
೏
ଵ th
term ‫ ݏ‬measures
ఌ೏ ௜
how far the i firm can raise the price above the
marginal cost C. Recall we have derived a similar equation for Lerner’s Index
௉ି஼ ଵ
in the case of Monopoly (Unit 3) which is ௉ ൌ െ ఌ . Given that ‫ݏ‬௜ is the
೏
th
share of the i firm in the total value of market output
௉ି஼
ฺ Ͳ ൑ ‫ݏ‬௜ ൑ ͳǤThus the monopoly power (as measured by ௉
ሻof an
oligopolistic firm is less than that of the Monopoly.

5.3 THE COURNOT MODEL

A Cournot model is an Oligopoly model in which all the firms decide on their
profit maximising output simultaneously with each firm assuming that its
rivals will continue producing their current output levels. Given the rival’s
quantity, each firm attains equilibrium by producing the quantity where
marginal revenue equals marginal cost, i.e. MR = MC. Introductory
Microeconomics course (BECC 101) of Semester 1 introduced the Cournot
model by considering two firms in the market. In such case both the firms
decide simultaneously about their profit-maximising level of output. Each
firm considers its rival’s output as given while making its output decision.
The relationship between a firm’s profit-maximising output and the given

output of its rival’s is summarised by a reaction function. These functions Oligopoly

are first obtained separately for each firm, and then solved simultaneously
to obtain Nash equilibrium. The market price is then determined using the
total output of both firms.

5.3.1 Equilibrium using Residual Demand Curve

A Residual demand curve for the first firm is that portion of the market
demand curve that remains for this firm assuming that the second firm
supplies a fixed amount of quantity Q2 in the market. From the Fig. 5.1,
residual demand curve facing the first firm is ascertained by shifting the
vertical axis to the right by the amount of output assumed to be sold by the
second firm (i.e. Q2). Let the market demand curve (DD’ in Fig. 5.1) be given
by P = A – BQ, where Q = Q1 + Q2 with Q1 and Q2 be the respective quantities
supplied by firm 1 and 2, P be the market price, and A and B any constants.
Then residual demand curve faced by the first firm will be given by

P = A – BQ
P = A – B(Q1 + Q2)
P =[A – BQ2] – BQ1
where [A – BQ2] represent the price intercept of the residual demand curve
facing the first firm. The MR curve will be given by first finding the total
revenue (TR) function of the first firm, which equals quantity × price, that is,

TR1: P × Q1= [AQ1– BQ2 Q1] – BQ12

and then taking the first derivative of this function with respect to Q1 to get
MR1
MR1 = A – BQ2– 2BQ1 (1)
In the Fig. 5.1, MR1 originates from the point of intersection of the vertical
line Q1 = 0 with the market demand curve, P = A – BQ. At point A, MR1
equals 0 which should be the case as at that point firm 1 produces no
output. Now, on assuming marginal cost to be 0, we get the profit
maximising quantity of firm 1 at point E where MR1 = MC. Using (1) and that
MC = 0, in equilibrium, we get

A – BQ2– 2BQ1 = 0
Now, we solve for profit maximising level of Q1 (let it be denoted by Q1*)
from the above equation, which we get as a function of given quantity of Q2
(also known as the reaction curve for firm 1):
஺ ଵ
Q1*= െ ܳଶ
ଶ஻ ଶ


Imperfect
Competition-II
D Q1 = 0

Price
A
A – bQ2

DZ1 E D’
0 ࡭ ૚ Quantity
Y1* = ૛࡮
െ ࡽ
૛ ૛
Y2 Y1
Fig. 5.1 : Equilibrium under Cournot Model using Residual Demand Curves

Similarly, profit maximising level of Q2 can be obtained (assuming first firm

supplies a fixed amount of quantity Q1 in the market) as a function of Q1.
Both the functions can then be solved to get the optimising quantities
supplied by each firm in the market.

5.3.2 Equilibrium using Reaction Curves

Reaction curves show relationship between an Oligopolist’s profit
maximising output and the amount it thinks its rivals will produce. If there
are only two firms in the model, then firm 1st profit maximising output will
depend upon what it thinks firm 2nd will produce, and this relationship will
derive firm 1’s reaction curve. Similarly firm 2nd reaction curve shows its
output as a function of how much it thinks firm 1st will produce. The
equation of the reaction curve is given by the profit maximising condition,
MR = MC. In the above case, the reaction curve equation for firm 1,
considering MC = 0, will be given by,
஺ ଵ
R1(Q2): MR1 = MC ֜Q1 = െ ଶ ܳଶ (2)
ଶ஻

where, R1 (Q2) represents the reaction function of firm 1 which gives the
optimal amount of output supplied by firm 1 as a function of given quantity
supplied by firm 2 (Q2). Similarly, the reaction curve equation for firm 2 will
be given by,
஺ ଵ
R2 (Q1): MR2 = MC ֜Q2 = ଶ஻ െ ଶ ܳଵ (3)

To note here is that if both demand and cost functions are linear, reaction
function will be linear as well. In Fig. 5.2, we plot these reaction functions by
marking output supplied by firm 1 and firm 2 on the vertical and the
horizontal axis, respectively. Reaction curve of firm 1, given by Equation (2)
஺
has the vertical intercept given by Q1 = ଶ஻ when Q2 = 0, and horizontal
஺
intercept given by Q2 = ஻
when Q1 = 0. Similarly, we plot reaction curve of
firm 2.

 Q1 Oligopoly
࡭
࡮

࡭
૛࡮

࡭
Y1* = ૜࡮

O ࡭ ࡭ ࡭ Q2
Y2* = ૜࡮
૛࡮ ࡮

Fig. 5.2: Equilibrium under Cournot Model using Reaction Curves

Now, equilibrium output for firm 1 and firm 2 will be given by the
intersection point of the two reaction curves. This is also called “Cournot-
Nash Equilibrium”, as each firm is doing the best it can given the behaviour
of rival firms. This can be ascertained by substituting reaction curve for Firm
2 into the reaction curve for firm 1:
஺ ଵ ஺ ଵ
Q1 = ଶ஻ െ ଶ ቂଶ஻ െ ଶ ଵ ቃ
஺ ஺ ଵ ஺
Q1 = ଶ஻ െ ସ஻ ൅ ସ ܳଵ ֜ Q1*= ଷ஻

Inserting the equilibrium value of Q1* in equation (3), we get

஺ ଵ ஺ ஺
Q2 = ଶ஻ െ ଶ ቀଷ஻ቁ֜Q2*= ଷ஻
ଶ஺
Total equilibrium quantity, ܳ ‫ = כ‬Q1* + Q2*֜ܳ ‫= כ‬ ଷ஻
. Insert this in market
demand curve to get Cournot price: P = A – BQ
ଶ஺ ஺
P = A – Bቀଷ஻ቁ֜P =ଷ

Cournot equilibrium profit of the firm 1 and 2 :

஺ ஺ ஺మ
ߨଵ ሺܳଵ ǡ ܳଶ ሻ ൌ ܲܳଵ ൌ  ߨଶ ሺܳଵ ǡ ܳଶ ሻ ൌ ܲܳଶ = ሺଷ ൈ ଷ஻ሻ = ଽ஻
ଶ஺మ
Cournot equilibrium profit of the industry: ߨଵ ሺܳଵ ǡ ܳଶ ሻ ൅ ߨଶ ሺܳଵ ǡ ܳଶ ሻ= ଽ஻

If firm act as a monopoly, then the optimising output will be given by setting
MR = MC:

Total revenue (TR): P × Q = (A – BQ) × Q;

ௗ்ோ
Marginal Revenue (MR): ௗொ = A – 2BQ
஺
For monopoly equilibrium, we set MR = MC ֜A –2BQ = 0 ֜QM= , where
ଶ஻
QM gives the profit maximising monopoly output. Inserting this in the

Imperfect ஺
demand curve equation, we get monopoly price PM= ଶ . So the Monopoly
Competition-II
஺ ஺ ஺మ
profit : PM × QM = ଶ஻ × ଶ = ସ஻

Hence, we find that Cournot duopoly model has a higher total quantity and
a lower price as compared to Monopoly quantity and price, respectively.
Moreover firms under the Cournot competition earns less profit (even at the
஺మ ஺మ ஺మ ଶ஺మ
industry level) as compared to the monopolyฺ ସ஻ ൐ ଽ஻ and ସ஻ ൐ ଽ஻
.

Example

Consider a duopoly of firm 1 and 2 producing a homogenous product, the

demand of which is described by the following demand function:
ଵ
Q = ଶ (100 – P)

Where Q is total production of both firms (i.e., Q = QІ н QЇͿ͘ ůƐŽ ůĞƚ ƚŚĞ
marginal cost of production faced by both firms be Rs. 40, i.e.͕DІсDЇс
40. Calculate the residual demand function for both the firms. Using them
ascertain their reaction curves and the Cournot-Nash equilibrium quantity
produced by each firm?

Solution

Residual demand function of a firm is ascertained by fixing the quantity

produced by the other firm.

We will first be finding the inverse demand function from the given demand
ଵ
function. Given, Q = ଶ (100 – P), the inverse demand function would be

P = 100 – 2Q֜P = 100 – 2(Q1 + Q2)

The residual demand function faced by firm 1 will be given by assuming
quantity Q2 produced by firm 2 as fixed:
P =[100–2Q2] – 2Q1
where [100–2Q2] represent the price intercept. Now, MR1 will be given
డ்ோభ
by డொభ

డ்ோభ
TR1: P × Q1= 100Q1–2Q2Q1 – 2Q12 ; MR1: డொభ
= 100–2Q2– 4Q1

For reaction function, we make use of the condition, MR1 = MC

ଵ
100–2Q2– 4Q1 = 40 ֜Q1 = 15 – ଶ Q2 [Reaction curve for firm 1]

Similarly, residual demand function faced by firm 2 will be given by:

P =[100–2Q1] – 2Q2
ଵ
And the corresponding reaction curve for firm 2 will be: Q2 = 15 – ଶ Q1

We solve both the reaction curves to get the Cournot-Nash equilibrium,


Q1* = Q2* = 10, and considering the demand function, we can derive the Oligopoly

equilibrium price, P* = 60
Profit by firm 1 = Profit by firm 2 = Total Revenue – Total Cost

= (60 × 10) – (40 × 10) = 200

Hence total industry profit = 400

Let us compare this result with the monopoly outcome. For the monopolist,
the aggregate inverse demand curve will be P = 100 – 2Q.

Total revenue (TR): P × Q = 100Q –2Q2;

డ்ோ
Marginal Revenue (MR): డொ
= 100 – 4Q

For monopoly equilibrium, we set MR = MC ֜100–4Q = 40 ֜ QM = 15,

where QM gives the profit maximising monopoly output. Inserting this in the
demand curve equation, we get monopoly price PM = 70.
Monopoly profit = total revenue – total cost
= (15 × 70) – (15 × 40) = 450

Hence, we find that Cournot duopoly model has a higher total quantity (=
20) and a lower price (= 60) as compared to Monopoly quantity (= 15) and
price (= 70), respectively. Also the total industry profit in case of Cournot
model (= 400) is less than that of a monopolistic industry profit (= 450).

5.3.3 Cournot Equilibrium with Different Costs

In the above example we assumed that the two firms are symmetric i.e.,
having the same marginal cost of production as 40. However, cost
conditions may vary from firm to firm. That is, a firm may be more cost
efficient as compared to the other firm.

Considering the demand function PсϱϬоϮQ. Let the marginal cost faced by
firm 1 be Rs. 2, and that faced by firm 2 be Rs. 8. Find the Cournot-Nash
equilibrium in such a case.

We make use of the optimising condition to find the reaction curves for
each of the firms.
డ்ோభ
TR1: P × Q1= 50Q1–2Q2Q1 – 2Q12 ; MR1: డொభ
= 50–2Q2– 4Q1

For reaction function, we make use of the condition, MR1 = MC

ଵ
50–2Q2– 4Q1 = 2 ֜Q1 = 12 – ଶ Q2 [Reaction curve for Firm 1]
ଶଵିொభ
Similarly, reaction function for firm 2 is given by: Q2 =
ଶ

On solving the two reaction functions, we get the Cournot-Nash solution as

Q1= 9, Q2 = 6, Q (total quantity) = 15, P (Cournot price) = 20, with profit of
Firm 1 = 152, profit of Firm 2= 60. 
Imperfect Hence, when firms have different costs, they choose different output levels,
Competition-II
with the firm having low-cost (here firm 1) enjoying higher share of the
market and making higher profits than the firm incurring high-cost (here
firm 2).
Check Your Progress 1

1) Consider a Cournot duopoly of firm 1 and 2 producing output Q1 and Q2,
respectively. Let market demand curve be P = 60 – Q, where Q = Q1 +
Q2. Assume marginal cost of production is null for both the firms.
Calculate Cournot-Nash equilibrium output of both the firms.
……………………………………………………………………………………………………………..

……………………………………………………………………………………………………………..

……………………………………………………………………………………………………………..

……………………………………………………………………………………………………………..

……………………………………………………………………………………………………………..
2) Let market demand curve be P = a – bQ, where Q = Q1 + Q2. Let there be
two firms 1 and 2 producing output Q1 and Q2, respectively at a
constant marginal cost of ‘c’. Calculate output of both the firms under
the Cournot model.

……………………………………………………………………………………………………………..
……………………………………………………………………………………………………………..

……………………………………………………………………………………………………………..

……………………………………………………………………………………………………………..

……………………………………………………………………………………………………………..

5.4 COLLUSION AND CARTELS

Collusion is said to take place when rival firms enter into an agreement in
terms of price, market share, etc. in an attempt to realise higher profits than
what they would have realised individually in an independent scenario.
Firms may explicitly enter into an agreement resulting in an explicit
collusion, or the collusion may take place without a formal agreement,
which then is referred to as a Tacit collusion. Cartel results when few firms
formally agree upon a certain level of output or a certain price of the goods
in order to maximise joint profits of the industry, which they can share
among themselves on a mutually binding agreement. Let us consider a case
of two firms under Cournot market structure as well as under collusion.
Coming back to the earlier example where demand curve is as follows:
P = A – BQ, with Q = QІнQЇ

 along with the assumption that MC = 0.

If both the firms produce individually under Cournot market structure then Oligopoly

(we know from our earlier section) the equilibrium outputs of each firm:
஺
QІсQЇсଷ஻

If the two firms collude and form a cartel, then total Revenue of the cartel
will be given by:
డ்ோ
TR = (A – BQ) Q = AQ – BQ² ; Marginal Revenue (MR): డொ
= A – 2BQ

For cartel equilibrium, the two firms jointly behave like a monopolist and
jointly control the entire market share. So from the equilibrium condition:
஺
MR = MC ֜A –2BQ = 0 ֜ Q* = ଶ஻
(same as the monopoly solution).
஺
Therefore total equilibrium quantity is Q = ଶ஻
. Using the demand function,
஺
we get the equilibrium price as P = ଶ .

The goal of the cartel is to set the industry output at a level that maximises
industry profits. A rule governing the cartel behaviour specifies how the
industry output and profits must be shared among the cartel members. In
our example, we assume the two firms are producing homogeneous goods
and are sharing similar cost structure which enabled the firms to equally
஺
share the industry output and thus profits, which means QІсQЇсସ஻. This is
஺
less than the Cournot equilibrium output of ଷ஻ units produced by each firm.
By independently maximising their own profits, firms produce more total
output than they would if they collusively maximised industry profits. This
pursuit of self interest does not typically maximise the well being of the
industry as well as their individual profits. In terms of our numerical
example, we saw that the total industry output of the duopoly was higher
than that of the monopolistic industry, but total industry profit of
monopolistic industry (Cartel) was higher as compared to the duopoly.

Conditions in oligopolistic industries tend to promote collusion since the

number of firms is small and firms recognise their interdependence. The
advantages can be more profits and decreased uncertainty. However it is
hard to retain the collusion and firms tend to cheat each other and thus
collusive arrangement often breaks down. Consequently, as long as a cartel
is not maintained by legal provisions, there is a constant threat to its
existence.

5.5 THE BERTRAND MODEL

In the Cournot model, each firm takes quantity as a strategic variable, and
the resulting total output determines the market price. However there
exists an alternative model, the Bertrand model where firms take price as
the strategic variable, with each firm selecting a price and standing ready to
meet all the demand it faces given the prices chosen by all the other firms.
However unlike Cournot competition, each firm faces separate demand

Imperfect function based on the price charged by it and its rivals. Let a homogeneous
Competition-II
good is produced by ‘n’ firms in the industry competing on price, with each
firm producing Qi units of a good at a cost of Ci (Qi), where i = 1,2...n. The
model further assumes that if firms set different prices, all demand shifts to
the firm charging the lowest price, which in turn produces enough output to
meet this demand. Further price rationing rule states that with more than
one firm charging the lowest price, output demand is shared between those
firms equally. Any firm charging above the lowest price charged in the
market, receives no market share.

Let us assume a Bertrand model of duopoly where there are two firms, 1
and 2, producing homogenous product at a constant marginal cost “C”. The
ĨŝƌŵƐ ĐŚŽŽƐĞƉƌŝĐĞƐ WІ ĂŶĚ WЇ͕ ƐŝŵƵůƚĂŶĞŽƵƐůLJ͘ ůů ƐĂůĞƐ ŐŽ ƚŽƚŚĞĨŝƌŵ ǁŝƚŚ
the lowest price. Sales are split equally if PІ = PЇ͘/ŶĐŽŶƚƌĂƐƚƚŽƚŚĞŽƵƌŶŽƚ
or Stackelberg models, the only Nash Equilibrium is the perfectly
competitive outcome i.e., PІ = PЇ = C. This results from the fact that each
firm in the Bertrand model has an incentive to undercut price as long as
production remains profitable. If one firm cut its price than the rival’s price,
then even for a slightest undercutting, it can appropriate the entire market
share, thereby, inducing the rival out of the market. Similarly, the same firm
may face zero market share if its rival outcompetes it by a slightest
undercutting of their price. Given that each firm has an incentive to
undercut its price in order to grab the market share, the firms would engage
in reciprocal price undercutting until the price gets pushed down to the level
of marginal cost “C”. At this price, economic profits would be zero. Hence
firms would have no incentive to undercut its price further. This would be
Nash equilibrium as there would be no incentive for either firm to change its
price once ଵ ൌ  ଶ ൌ . If either firm lowers their price below marginal cost,
they would incur losses. If either raised their price, then it would be no
better off, because it would lose the entire market share to the rival. Thus,
quantity sold by each firm depends on both prices. If P1< P2, then firm 1
serves the entire market demand D(P1); the opposite happens, that is, firm 2
serving the entire market demand D(P2) when P1> P2; and when P1 = P2¸ each
஽ሺ௉ሻ
firm supplies half of the total market demand, that is, . If ܲଵ ൌ  ܲଶ ൐
ଶ
‫ܥ‬ǡeven though the firms share the market equally, it is not an equilibrium
strategy. In this case either of the firm has an incentive to undercut the price
and grab the entire market and still enjoy positive profit.

Assuming ȫ௜ denote the profit earned by firm i, where i = 1, 2. Then, Nash

equilibrium will be given by the pair of price (P1*, P2*) such that,
ȫଵ (P1*, P2*) шȫଵ (P1, P2*) ; ‫ ׊‬P1
5.1
ȫଶ (P1*, P2*) шȫଶ (P1*, P2) ; ‫ ׊‬P2

where symbol ‫׊‬stands for “for all”.


 Oligopoly
5.5.1 Bertrand Paradox
The unique pair of prices satisfying condition 5.1 is given by P1 = P2 = C (the
given constant marginal cost) at which ȫଵ = ȫଶ = 0. This situation is referred
to as the condition of Bertrand Paradox. The paradox results from the fact
that just two firms are sufficient to dissipate the market power and yield an
outcome similar to the perfectly competitive market (which usually assumes
many sellers). In other words, with the number of firms rising from one to
two, the price decreases from the monopoly price to the competitive price
and stays at the same level even when the number of firms increases
further. This is contrary to our perception, where we think that markets with
a small number of firms possessing market power typically charge a price
above the marginal cost.

5.5.1.1 Condition to be Satisfied for the Bertrand Paradox

For the Bertrand model to generate the Bertrand paradox, i.e., a situation
when a perfect competitive outcome results with just two firms (engaged in
reciprocal price undercutting above their marginal cost) are: 1. Firms
involved in price competition must possess unlimited capacities. If initially
the price condition is given by, P1 = P2 = P (say) > C, with each firm sharing
market equally, then either firm would be tempted to undercut its price
slightly (say by Ԫ൐Ͳ) and grab the entire market. In order to throw the rival
out of the market and solely cater to the market demand, either firm needs
a drastic expansion of capacity. The firm can satisfy this increased demand
only when it faces no capacity constraints. If the firm faces a capacity
constraint, price undercutting would not be profitable and it would not be
able to drive its rival out of the market. Rather, that would leave some
residual demand for the higher-priced rival firm and would decrease the
incentive to undercut. Moreover Bertrand paradox prevails when firms are
assumed to be in a homogeneous product industry i.e., the products of each
firms are close substitutes, so that consumer cannot distinguish one firm’s
product from the other firms. If the firms are in the differentiated product
industry (where consumer can exercise a brand preference as the products
are similar but not exactly close substitutes) each firm enjoys some
monopoly power and hence P1 = P2 = C is not the Nash equilibrium. Rather
the Nash equilibrium may be P1‫س‬P2 > C, depending upon their brand values.

5.5.2 Bertrand Equilibrium using Reaction Curves

In case of a Bertrand model, reaction function of a firm will be in price
terms. In other words, a reaction function will give the optimum price at
which firm chooses to supply its output, given the price of its rival’s. Firm 1’s
reaction curve equation will be given by R1(P2), giving the optimal price
charged by firm 1 given that the price set by firm 2 is P2. Similar description
goes for the reaction curve equation of firm 2, that is, R2(P1). Refer Fig. 5.3,
where we represent Bertrand equilibrium using reaction curves assuming
same marginal cost C is faced by both the firms.

Imperfect
Competition-II R2(P1)
P1
45ι

PM R1(P2)

A
C

O PM P2
C
Fig. 5.3: Bertrand Equilibrium using Reaction Curves

In the above figure, firm 1’s reaction function R1(P2) shows when firm 2’s
price is below the marginal cost, i.e., when P2< C, it is optimal for firm 1 to
set its price at the marginal cost, i.e., P1 = C as shown by the horizontal
segment CA of the firm 1’s reaction curve R1(P2). The 45ιline represents
prices where P1 = P2. PM represent the monopoly prices charged for the
good when both the firms collude and act as a monopolist. Now, again look
at the reaction curve of firm 1. When firm 2’s price is above the marginal
cost (C) but below the monopoly price (PM), then it is optimal for firm 1 to
set price just below firm 2, i.e. P1< P2. Whereas, when firm 2’s price is above
the monopoly price, it is optimal for firm 1 to charge the monopoly price, i.e.
P1 = PM. With both facing the same marginal cost (C), reaction function of
firm 2, R2 (P1) will be symmetrical with respect to the 45ι line. Symbolically,
we can represent a reaction function as follows:

୑ ‹ˆଶ ൐ ୑
R1 (P2) = ൝ ଶ െ ɂ‹ˆ ൏
ଶ ൑ ୑ and
‹ˆଶ  ൑ 

୑ ‹ˆଵ ൐ ୑
R2 (P1) = ൝ ଵ െ ɂ‹ˆ ൏
ଵ ൑ ୑
‹ˆଵ  ൑ 
where, Ԫ represents a small positive number. Bertrand equilibrium is given
by the intersection of both the reaction curves at Point A, where P1 = P2 = C,
which is mutually best response for both the firms. At any other price above
the marginal cost, either firm would always find it is in their best interest to
undercut its rival’s price a little (say by Ԫ> 0) and serve the entire market
itself. It is only at the price equals to the marginal cost that firms have no
incentives to deviate from the equilibrium prices.


 Oligopoly
5.5.3 Comparison between the Cournot and the Bertrand
Model
In the Cournot model, firms engage in quantity competition, so quantity is
the strategic variable of the firm whereas in the Bertrand model firms
engage in price competition, so price is the strategic variable. In the Cournot
model the equilibrium price is generally above the marginal cost and the
quantity approaches the perfectly competitive market situation only when
the number of firms becomes large. In contrast to this, in the Bertrand
model even with two firms, price competition dissipates the monopoly
power and ensures competitive price which is equal to the marginal cost.
Another difference is that in the Cournot model, the firms take their rival’s
output as given and cannot “steal” any consumers away from their rivals by
lowering their prices. Whereas, a Bertrand rival believes that it can lure
customers from its rivals by small cuts in price.

5.6 THE STACKELBERG MODEL

Similar to firms in Cournot model, in the Stackelberg model of oligopoly,
firms produce homogeneous product and engage in quantity competition.
The principal difference between the two models is that instead of
simultaneous output or quantity choice of the rival firms (as in the Cournot
model) the firms in Stackelberg model is based on sequential output or
quantity choice. So instead of a static-move game like that of Cournot
model, the firms in the Stackelberg model are engaged in the dynamic-move
game. In case of a Stackelberg model of duopoly, we assume one of the firm
moves first, followed by it the other firm. Thus, the model basically becomes
a model of two periods, where first mover (say firm 1) decides about its
quantity choice in period 1, then the second mover (say firm 2) after
observing the firm 1’s move, decides about its optimal quantity choice in
period 2, and there the game ends with the appropriation of profits by both
the firms. How does this affect the equilibrium of this game? We analyse the
result using the example considered above in case of the Cournot model of
duopoly. Now, let the firm 1, also referred to as the Stackelberg leader,
moves first to produce quantity Q1. Firm 2, the Stackelberg follower,
observes firm 1’s quantity choice Q1, and then chooses to produce quantity
Q2. The quantity chosen by the follower must therefore be along its reaction
function. Since, firm 1 knows that firm 2 will take firm 1’s output as given
and optimally decides its output along its reaction function, firm 1 (the
Stackelberg leader) being the first mover enjoys a strategic advantage over
firm 2 (the Stackelberg follower) by knowing the reaction function of the
firm 2. Firm 1 can influence the behaviour of firm 2 by altering its own
output and it takes into account the effects of its own output on firm 2’s
behaviour.

In sequential games, the optimal solution is obtained by the backward

induction techniques, where, we first solve the firm 2’s optimal quantity

Imperfect choice problem in the second period and then proceed backward to solve
Competition-II
the firm 1’s optimal quantity choice problem in the first period. Such
approach is called backward induction as it is a process of reasoning
backwards in time. In period 2, firm 2 chooses Q2 given firm 1 has chosen Q1
in period 1. This gives us the reaction curve of firm 2 as a function of Q1,
R2(Q1). This reaction function of firm 2 is then considered by firm 1 in period
1 as given to decide its own profit maximising quantity Q1. We illustrate this
below:
Consider the same demand function that we considered in the earlier
example of Cournot model. Here we assume firm 1 to be the leader and firm
2, to be the follower. The solution is worked out as follows:

P = A – BQ [where, Q = QІнQЇ΁

'ŝǀĞŶŵĂƌŐŝŶĂůĐŽƐƚŝƐDІсDЇсϬ͘

Backward Induction: Consider period 2 first. Here, firm 2 will choose its
optimal output according to its own Reaction function i.e.
஺ ଵ
ܴଶ ሺܳଵ ሻǣQ2 = ଶ஻ െ ଶ ܳଵ (4)

(as derived in the earlier example)

Now consider period 1: Firm 1 will choose its output by maximising its own
profit function and considering the reaction function of firm 2, ܴଶ ሺܳଵ ሻ. The
profit function of firm 1 (Ȉଵ ) is :
ȫଵ (ܳଵ ǡ ܳଶ ͿсdZІʹdІ
ǁŚĞƌĞdZІŝƐƚŽƚĂůƌĞǀĞŶƵĞŽĨĨŝƌŵϭĂŶĚdІŝƐƚŽƚĂůĐŽƐƚŽĨĨŝƌŵϭ
ȫଵ (ܳଵ ǡ ܳଶͿсdZІʹdІс΀A – B(Q1 + Q2)] Q1 – 0 × Q1
֜ȫଵ (ܳଵ ǡ ܳଶ ሻ= AQ1 – BQ12– BQ2 Q1
= AQ1 – BQ12 – BQ1ൈ ܴଶ ሺܳଵ ሻ
Firm 1 being the first mover knows the reaction function of firm 2 (as shown
in Eq. 4) and therefore incorporate it in its own profit function in order to
solve for its own optimal output choice:
஺ ଵ
ȫଵ (ܳଵ ǡ ܳଶ ) = AQ1 – BQ12– BQ1ቀଶ஻ െ ଶ ܳଵ ቁ
஺ொభ ஻ொభమ
= AQ1 – BQ12 – ଶ
൅ ଶ

So the first order condition for the optimisation of firm 1’s profit is obtained
by setting the first - order partial derivative with respect to QІĞƋƵĂůƚŽnjĞƌŽ͘
డஈభ ሺொభ ǡொమ ሻ ஺ ଶ஻ொభ
డொభ
= 0 ֜A – 2BQ1 – ଶ
൅ ଶ
=0
஺
Solving for optimal Q1, gives QІсଶ஻.


Now, substituting QІ ďĂĐŬ ŝŶ Ƌ͘ ;ϰͿ͕ ǁĞ ŐĞƚ ƚŚĞ ŽƉƚŝŵĂů ŽƵƚƉƵƚ ŽĨ Ĩŝƌŵ Ϯ͗ Oligopoly
஺
Q2 = ସ஻. So the equilibrium output choice of firm 1 and 2 are:
஺ ஺
[QІсଶ஻ , Q2 = ସ஻ሿ.
ଷ஺ ஺
Equilibrium price under Stackelberg model : P = [A – B(Q1 + Q2Ϳ΁с΀оସ஻ሿ= ସ

Now profit of the leader (firm 1) : ȫଵ (ܳଵ ǡ ܳଶ)= [A – B(Q1 + Q2)] Q1 =

ଷ஺ ஺ ஺ ஺ ஺మ
΀оସ஻ሿ ଶ஻ ֜ ȫଵ (ܳଵ ǡ ܳଶ )= ସ ଶ஻ ൌ  ଼஻

Profit of the follower (firm 2): ȫଶ (ܳଵ ǡ ܳଶ) = [A – B(Q1 + Q2)]ܳଶ ൌ΀ о
ଷ஺ ஺ ஺ ஺ ஺మ
Bସ஻ሿ ସ஻ ֜ ȫଶ ሺܳଵ ǡ ܳଶ ሻ ൌ ସ ସ஻
ൌ  ଵ଺஻ ֜ ȫଵ(ܳଵ ǡ ܳଶ ሻ ൐ ȫଶ ሺܳଵ ǡ ܳଶ ሻ ֜ Thus
the leader enjoys higher market share and higher profit than the follower in
equilibrium.

Unlike the Cournot outcome which was symmetric, that is both the firms
produced the same level of output, in the Stackelberg model, leader firm
i.e., firm 1 enjoys greater market share (produces more output) as
compared to the follower i.e., ܳଵ ൐ ܳଶ in the equilibrium. One may
compare the Stackelberg equilibrium outcome with that of the Cournot
outcome and may check that the leader (firm 1) produces more than
produced by a firm in Cournot equilibrium whereas the follower produces
less than Cournot equilibrium quantity. Thus, there is an advantage of being
the first mover.
ଷ஺
Now, the total industry output will be Q1 + Q2 = ସ஻, which is more than the
ଶ஺
total industry output under Cournot equilibrium of ଷ஻
for a given value of A
and B. Profit of the Stackelberg leader is higher than the firm under Cournot
஺మ ஺మ
competition ฺ  ȫଵ ሺܳଵ ǡ ܳଶ ሻௌ >ȫଵ ሺܳଵ ǡ ܳଶ ሻ஼ ฺ  ଼஻ ൐ ଽ஻. Moreover profit of
the Stackelberg follower is lower than the firm under Cournot competition
஺మ ஺మ
ฺ  ȫଶ ሺܳଵ ǡ ܳଶ ሻௌ ൏ ȫଶ ሺܳଵ ǡ ܳଶ ሻ஼ ฺ  ଵ଺஻ ൏ ଽ஻. Industry profit under
Stackelberg model is greater than that of the Cournot competition

ฺ ȫଵ ሺܳଵ ǡ ܳଶ ሻௌ ൅ ȫଶ ሺܳଵ ǡ ܳଶ ሻௌ ൐ ȫଵ ሺܳଵ ǡ ܳଶ ሻ஼ ൅ ȫଶ ሺܳଵ ǡ ܳଶ ሻ஼

ଷ஺మ ஺మ ஺మ ஺మ
ฺ ଵ଺஻ ቀൌ  ଼஻ ൅ ଵ଺஻ቁ> ଽ஻

5.7 THE DOMINANT FIRM MODEL

In some oligopolistic models, one large firm dominates the market share and
many small fringe firms are the followers catering the residual demand and
acting competitively. A Dominant firm (also known as the leader), typically
having a larger share in market, behaves as a price-setter that faces smaller
price-taking firms (also known as Fringe firms) each having a very small
share in the market. The leader or the dominant firm sets the price for the
commodity that maximises its own profits and assumes that its rivals will
behave as competitive firms or in other words as price-takers that will take 
Imperfect the price set by the dominant firms as given in determining their output in
Competition-II
that price. Let us explain how this market structure operates with the help
of a diagram below.

Price
SF

MC D
P1
P*
D
C
B
P2 A

DZD D’
O YD` YF YT Quantity
Fig. 5.4: Dominant Firm Model

In Fig. 5.4 DD’ curve represents the market demand which is served by the
dominant firm and the fringe firms collectively. Fringe firms are assumed to
behave as competitive firms supplying where their marginal cost equals the
price set by the dominant firm. This way fringe firms collectively behave as a
competitive industry whose output can be determined by the supply curve
SF. The dominant firm’s problem is to find a price which maximises its own
profits. To solve this problem we need demand curve of the dominant firm.
The extent of demand that the dominant firm gets to serve is equal to the
total quantity demanded at that price (given by DD’) minus the quantity the
fringe firms supply at that price (given by SF).This is represented by the
residual demand curve P1AD’. In other words, P1AD’ represents the
horizontal difference between the fringe firm’s supply curve SF and the
market demand curve DD’. At price P1, dominant firm produces nothing with
fringe firms serving the entire market demand at that price. On the other
hand, at a price as low as P2 dominant firm gets to serve the entire market
demand with supply by fringe firms falling to zero. The dominant firm’s
residual demand curve between these two extreme prices is given by P1A.
Corresponding to the residual demand curve of the dominant firm, we have
a marginal revenue curve represented by MRD. MCD likewise represents the
marginal cost curve of the dominant firm.
The dominant firm optimise at point B where marginal revenue equals
marginal cost resulting in profit maximising output QD sold at the market
price P*. At this price fringe firms will supply QF amount of output as will be
given by their supply curve. Total output at P* equals QT which is the sum of
QD and QF. Total output generated in this model will be less than
competitive output. This is clear by the figure above. Dominant firm
optimise at a point where its price is above the marginal cost it faces. An
efficient optimum in this case will be given by the point C where MCD
intersects the residual demand curve resulting in higher total output as is

given by the point D. In this case, not only fringe firms but also the dominant Oligopoly
firm is producing where MC equals price.
Check Your Progress 2
1) Explain the difference between the Bertrand model and the Stackelberg
model of Oligopoly.
……………………………………………………………………………………………………………..

……………………………………………………………………………………………………………..

……………………………………………………………………………………………………………..

……………………………………………………………………………………………………………..

2) For the following demand curve P = a – 2b and constant marginal cost
curve ‘c’, find equilibrium price, quantity and profit according to:
i) Stackelberg Model

ii) Bertrand Model

……………………………………………………………………………………………………

……………………………………………………………………………………………………
……………………………………………………………………………………………………

……………………………………………………………………………………………………

5.8 LET US SUM UP

The present unit is an attempt to move relatively closer to the real market
conditions. The extreme market structures like perfect competition and
monopoly have already been taken into account in order to discuss a
relatively real market situation of that of an oligopolistic market. Oligopoly is
a form of market structure where only a few firms account for most or all of
production. The market also assumes barriers to entry which allows a few
firms to act as dominant players. These few firms are characterised by
strategic interdependence on output and pricing decisions. The unit
discussed several oligopolistic models with varying assumptions, decision
parameters and consequently the equilibrium outcome. In oligopolistic
market, the concept of Nash equilibrium is more appropriate as it gives due
consideration to the strategic interdependence among the firms in their
decision-making. A Nash-equilibrium is a situation in which each firm adopts
the best response strategy, given the strategy of its rival firm. In the Cournot
model, each firm set its profit maximising quantity assuming its rival’s
output as given. On the other hand, in Bertrand model, each firm set the
profit maximising price assuming its rival’s price as given. Then after these
two simultaneous-move models, we discussed a sequential-move model, i.e.
the Stackelberg models, where we have a leader and a follower, with leader
deciding first about the optimum output giving due consideration to the

Imperfect reaction function of the follower. Subsequently, the Dominant firm model is
Competition-II
described, where a Dominant firm (the leader), behaves as a price-setter,
with the smaller price-taking firms (Fringe firms) taking the price set by the
dominant firms as given in determining their output in that price. Each
model has its own significance with respect to the assumptions and the
respective equilibrium outcome.

5.9 SOME USEFUL REFERENCES

x Hal R. Varian, Intermediate Microeconomics, a Modern Approach, W.W.
Norton and company/Affiliated East- West Press (India), 8th Edition, 2010.
x Pindyck R. S and Rubinfeld D. L, Microeconomics, Pearson India, 2009
x C. Snyder and W. Nicholson, Fundamentals of Microeconomics, Cengage
Learning (India), 2010.

5.10 ANSWERS OR HINTS TO CHECK YOUR PROGRESS

EXERCISES
Check Your Progress 1
1) Firm 1
dZІсWYІс;ϲϬʹYͿYІ
сϲϬYІʹ;YІнYЇͿYІ
сϲϬYІʹYІϸʹYЇYІ 
‫׵‬DZІсϲϬʹϮYІʹYЇ
^ĞƚƚŝŶŐDZІсDІ
ଵ
&ŝƌŵϭ͛ƐZĞĂĐƚŝŽŶĐƵƌǀĞŝƐ͗YІсϯϬʹଶYЇ
ଵ
^ŝŵŝůĂƌůLJǁĞĐĂŶŐĞƚ&ŝƌŵϮ͛ƐZĞĂĐƚŝŽŶĐƵƌǀĞĂƐYЇсϯϬʹଶYІ
dŚĞĞƋƵŝůŝďƌŝƵŵůĞǀĞůƐŽĨYІĂŶĚYЇĂƌĞŐŝǀĞŶďLJƐŽůǀŝŶŐƚŚĞƚǁŽƌĞĂĐƚŝŽŶ
curves.
‫׵‬ŽƵƌŶŽƚͲEĂƐŚĞƋƵŝůŝďƌŝƵŵYІсYЇсϮϬ͕ĂŶĚŽƵƌŶŽƚƉƌŝĐĞ͕WсϮϬ
2) Solve the question like the above question.
a–c 2(a–c) a+2c
YІсYЇс , total industry output, Q = , Cournot price, P =
3b 3b 3
Check Your Progress 2
1) Bertrand model is based on price competition, where the only Nash
ĞƋƵŝůŝďƌŝƵŵ ŝƐ ǁŚĞŶ WІ с WЇ с ͘ /Ŷ ƚŚĞ ^ƚĂĐŬĞůďĞƌŐ ŵŽĚĞů͕ ƚŚĞ ĨŝƌƐƚ
mover, firm 1 has an advantage, it has a higher market share, hence
higher profits than firm 2, price is also above marginal cost.
2) Market form Firms output Price
a–c a–c a+3c
^ƚĂĐŬĞůďĞƌŐ  YІс 2b YЇс 4b P= 4
a–c
ĞƌƚƌĂŶĚ   YІсYЇс 2b P=C

 Game Theory and
UNIT 6 GAME THEORY AND ITS APPLICATIONS its Applications

Structure

6.0 Objectives
6.1 Introduction
6.2 The Game Theory
6.3 A Game
6.4 Types of Games
6.4.1 Non-cooperative versus Cooperative Games
6.4.2 Games of Complete and Incomplete Information
6.4.3 Zero-sum Game versus Non-zero Sum Games
6.4.4 Simultaneous-move versus Sequential-move Games
6.5 Alternative Forms of Representing a Game
6.5.1 Normal-form Representation of a Game
6.5.2 Extensive-form Representation of a Game
6.6 Solving a Game theory Problem
6.6.1 Dominated Strategies
6.6.2 Dominant Strategies
6.6.3 Dominant Strategy Equilibrium
6.6.4 Iterated Elimination of Strictly Dominated Strategies
6.6.5 Nash Equilibrium
6.6.6 Multiple Equilibria
6.7 Mixed Strategies
6.8 Sequential Games
6.8.1 Sub-game Perfect Nash Equilibrium (SPNE)
6.8.2 Backward Induction
6.9 Application of the Game Theory
6.10 Let Us Sum Up
6.11 Some Useful References
6.12 Answers or Hints to Check Your Progress Exercises

6.0 OBJECTIVES
After reading the unit, you should be able to:
x define the concept of a Game in the Economic theory;
x describe basic elements of a Game, viz. (1) players, (2) strategies and (3)
payoffs;
x discuss different types of Games;
x appreciate two different forms of representing a Game; 
Imperfect x appreciate various basic concepts required to solve a Game theoretic
Competition-II
problem;
x elucidate different equilibrium concepts related to the Game theory;
x explain the notion of a sequential game; and
x discuss the application of the Game theory.

6.1 INTRODUCTION
In Unit 5 we discussed the market structure of oligopoly where firms’
decision problem exhibit strategic interdependence. In such a market,
output or price decision by one firm impacts output (sales) or price decision
and hence the profit of the other firms in the market. As a result,
optimisation by firms in this market involves taking care of their strategic
interdependence. Game theory is nothing but the study of interactive
decision-making. A group of people (or players, teams, firms, countries) are
in a game if their decision-making problem is interdependent. Thus in a
multi-agent framework the behaviour of individual agents are contingent to
the fact that action of one agent affects the payoff of the other agent(s). In
the previous unit we discussed game theory in the context of firm
competition. One can come across many real life situations involving
strategic interdependence where more than one agent is involved in
decision-making (like football, soccer, baseball are games). Economists apply
various general tools to arrive at the solution of the problems involving
strategic situations. In fact game theory has been used in many fields in
recent decades (apart from Economics) like Political Science, Sociology,
Computer Science, Biology. The present unit will be elaborating upon that.

6.2 THE GAME THEORY

The early beginnings of Game Theory can be traced to analysis of
imperfectly competitive market by Augustine Cournot (1838). The first
systematic attempt is Neumann’s and Morgenstern’s Theory of games and
Economic Behaviour published in 1944. The next great advance is by John
Nash who introduced the concept of ‘Nash Equilibrium’. Game theory was
invented as an attempt to find a theoretical solution to the problems posed
by uncertainty in games of chance where rational players take decision in an
interdependent set up. Basically it comprises a formal methodology and a
set of techniques to study the interaction of rational agents in strategic
settings. By ‘rational agent’ it is meant an individual ZKR is assumed to take
into account all the available information, probabilities of events, and
potential costs and benefits to perform the action with the optimal
expected outcome for itself from among all feasible actions. A ‘strategic
scenario’ is defined as the one where actions of one individual affects the
payoff (or reward) or utility of other individuals. Game theory can be used to
model a wide variety of human behaviour in economic, political, and social

settings. Game theory models seek to portray complex strategic situations in Game Theory and
its Applications
a simplified setting. Like in case of Perfect competition, Monopoly and
Imperfect competition, game theory involves generalising the methods used
earlier and with the help of some specific language arrive at a mathematical
representation of the strategic situation. Let us now discuss some
preliminaries related to the topic.

6.3 A GAME
A ‘Game’ is an abstract of a strategic situation involving interdependence. A
simplest form of game is defined by: players, actions or strategies and
payoffs. In game theory, players are the agents who are involved in the
decision-making. Each player has a number of strategies or action to choose
from. The strategies chosen by each player determine the outcome of the
game, with each possible outcome resulting in a payoff to each player.
i) Players: Players are the agents playing the game. They may be firms,
individuals, countries, or just about anything else that is capable of
executing a strategy. In the duopoly game, for instance, the players are
the two firms. Generally, in an n-player game, players are numbered
from 1 to n, with any arbitrary player to be called player i.
ii) Strategies: Strategies are the actions or the set of actions available to
the players. For instance, in case of Cournot game, each firm’s strategy
is to choose its quantity, taking as given the quantity of its rivals. In case
of an n-player game we assume any player i has a strategy set Si
consisting of different strategies, with si referring to any arbitrary
element of the set Si, i.e. si੣^i.
iii) Payoffs: Payoffs are the returns to the players at the conclusion of the
game. For instance, payoffs are the profits in case of profit maximising
firms. In an n-player game, payoff function for any arbitrary player i will
be given by Ui (s1,s2, s3....sn), where (s1, s2, s3....sn) represents the
combination of strategies (one for each player) chosen by the n-players.
Here Ui is the payoff to player i as a function of (s1, s2, s3....sn).
Besides the above set of three elements, a game is modelled by specifying
‘information’ that each player has when it decides its course of action. It is
usually assumed that there is common knowledge, that is, each player
knows not only about the ‘‘rules of the game’’ called common knowledge of
the game, but also what the other player knows, and so forth called the
common knowledge of rationality. Other aspects of information vary from
game to game, depending on timing of moves and other issues.

A Game can be static (called simultaneous-move game) or dynamic (called

sequential-move game) depending upon the information base of the players
or the time of move of the players. In simultaneous-move games, neither
player knows the other’s action when she(he) decides her own move. In
sequential-move games, the players move in sequence and the first-mover 
Imperfect does not know the second’s action but the second-mover knows what the
Competition-II
first did and the payoffs are decided at the end.

6.4 TYPES OF GAMES

6.4.1 Non-cooperative versus Cooperative Games
There are two branches of the game theory, viz. cooperative and non-
cooperative game theory. Under the cooperative game theory, groups or
sub-sets of the players make a binding agreement to reach an outcome that
is best for the group as a whole and is shared equally among the members.
In contrast to this, under non-cooperative game theory, players cannot
write binding contract. Players are guided by self-interest, each player acts
as an individual who is normally assumed to maximise his own utility
without caring about the effects of his choice on other players in the game.
The outcome of the game, however, is jointly determined by the strategies
chosen by all players in the game. As a result, each player's welfare
depends, in part, on the decisions of other players in the game. An example
of cooperative game is two firms negotiating a joint investment to develop a
new technology. An example of non-cooperative game is two competing
firms taking into account each other’s behaviour when setting their prices
independently.

Self-interested behaviour does not always lead to an outcome that is best

for the players as a group. This we will come across when we discuss
different illustrations of the games. Non-cooperative game theory is more
widely used by economist; nevertheless, cooperative game theory has been
used to model bargaining games and political processes.

6.4.2 Games of Complete and Incomplete Information

In the games of complete information, the payoffs, strategies and types of
players are common knowledge. Complete information is the concept that
each player in the game is aware of the sequence, strategies, and payoffs
throughout the game. Given this information, the players have the ability to
plan accordingly based on the information to maximise their own rewards or
payoff at the end of the game. The equilibrium solution concepts are Nash
equilibrium or Sub-game perfect Nash equilibrium depending upon whether
the game is simultaneous-move (static) or sequential-move (dynamic). We
will introduce these concepts in the subsequent sections.

Inversely, in a game with incomplete information, players do not possess full

information about their opponents. Some players possess private
information, a fact that the others should take into account when forming
expectations about how those players will behave. A typical example is an
auction: each player knows his own utility function (valuation for the item),
but does not know the utility function of the other players. The equilibrium
solution concepts are Bayesian Nash equilibrium or Perfect Bayesian

equilibrium depending upon whether the game is simultaneous-move Game Theory and
its Applications
(static) or sequential-move (dynamic).

6.4.3 Zero-sum versus Non-Zero Sum Games

A zero-sum game is the one in which the gain of one player comes at the
expense of the other player and is exactly equal to the loss of the other
player. In other words, the sum of the payoffs of the two players always
adds to zero. An economic application can be the transaction between a
buyer and a seller at the cost price. A non-zero sum game is when gain or
loss does not come at the expense of the other player. An example of this
might arise if increased advertisement leads to higher profits for both the
firms.

6.4.4 Simultaneous-move versus Sequential-move Games

The order of moves is significant in the game theory. Players in a game may
move simultaneously or sequentially which in turn results in different
outcomes of the game. A simultaneous-move game is a game in which
neither player knows the other’s action when moving, that is, players take
their action simultaneously without knowing the action that have been
chosen by the other player(s). For instance, in Cournot model of oligopoly,
each firm decides its profit maximising levels of output simultaneously. In
contrast, in sequential-move games, the order of moves comes into picture.
In this case, one player moves first which is then observed by his opponent.
The player(s) who moves afterwards gets to observe and learn information
about the course of the game up to that point, including what actions other
players have chosen. These observations can then be used by that player to
decide his (her) own optimal strategies than simply choosing an action. This
way, strategies of the players depend on what the other player(s) before
have done already.

6.5 ALTERNATIVE FORMS OF REPRESENTING A GAME

There are two principal representations of the rules of a game, i) the Normal
or Strategic form, and ii) the Extensive form. The normal form of
representation of a game is by using a payoff matrix in the form of rows and
columns. By convention, rows correspond to represent player 1 and
columns correspond to player 2, whenever we have a two player game and
an entry in a cell shows the payoff of two players for that specific
combination of strategies chosen. The extensive form is the pictorial
representation of the rules. The main pictorial form is called the game tree.
The game starts from player1 from a node and the choices available are
represented by branches emanating from that node. At the end of a branch,
player 2 makes his choice and the branch will split into further branches. The
game specifies the order in which the players make choices, how many
times each gets to choose and the eventual payoffs to each player for any

Imperfect sequence of choices. Normal form is usually used for simultaneous move
Competition-II
games, whereas, an extensive form through a game tree may be used to
represent either a simultaneous-move or a sequential-move game. Let us
illustrate both of these presentation forms using the very famous example
of Prisoner’s Dilemma game.

Prisoner’s Dilemma
A well known example of a non-cooperative and a game of complete
information is the Prisoners’ Dilemma game. Consider the following set-up
of the game: A crime is committed for which there is no eye witness.
Suspects 1 and 2 are caught and imprisoned in two separate cells. Thus
each prisoner is in a solitary confinement with no means of communicating
with each other. The magistrate speaks to each prisoner separately, and
asks them to act as an informer. If one of them confesses the crime, he will
be freed but the other one will spend 4 years in prison. If both confess, each
will spend 3 years in prison. If both stay quiet and do not confess, the crime
cannot be probed, so they will get nominal punishment by spending only
one year in prison. Thus each player then has two possible strategies: Not
confess (N) or Confess (C) and they decide simultaneously.

6.5.1 Normal-form Representation of a Game

The situation may be modelled as a strategic (normal) form game with the
following elements:

Players: Two suspects, prisoner 1 and 2.

Strategy: each player’s strategy set is {Not confess (N), Confess (C)}
Payoff: Number of years of prison sentence
Player 2
Not Confess Confess
Player 1 EŽƚŽŶĨĞƐƐ ;оϭ͕оϭͿ ;оϰ͕ϬͿ
 ŽŶĨĞƐƐ ;Ϭ͕оϰͿ ;оϯ͕оϯͿ

Fig. 6.1: Payoff Matrix

Entry in each cell of the above figure represents (Player 1’s, Player 2’s)
payoff in terms of number of years in jail (a negative payoff with a negative
sign) from each of the two strategies. In other words, the first numerical
figure in each cell of the matrix corresponds to payoff of player 1 and the
second figure corresponds to payoff of player 2. For example, if we look at
ƚŚĞ ĂĐƚŝŽŶ ΀EŽƚ ŽŶĨĞƐƐ͕ EŽƚ ĐŽŶĨĞƐƐ΁ ƚŚĞ ƉĂLJŽĨĨƐ ĂƐƐŽĐŝĂƚĞĚ ĂƌĞ ;оϭ͕ оϭͿ͘
This means when both the players simultaneously decide not to confess,
both will be spending one year in jail. The next entry in first row, second
column is (Not confess, Confess), this means player 1 will get 4 years of
imprisonment and player 2, freedom.


 Game Theory and
6.5.2 Extensive-form Representation of a Game
its Applications
A simultaneous game can also be represented by the extensive form using a
game tree. In Fig. 6.2 each dark dot is called the decision node for the player
indicated there. The first move belongs to Player 1, who can choose to
confess or not confess. The next move belongs to Player 2, who can also
choose among the two options (confess or not confess). Payoffs are decided
at the end of the tree towards the extreme right. To reflect the fact that the
Prisoners’ Dilemma is a simultaneous move game, a dotted oval is drawn
around Player 2’s decision nodes to reflect the fact that Player 2 does not
know which of the two decision nodes he is at since he does not observe
which action Player 1 has chosen, that is, he does not know whether the
first decision by Player 1 was to confess or not confess. This dotted oval
around the two nodes of Player 2 indicates his lack of specific information.
An information set of a Player is a collection of nodes such that the same
player (here Player 2) moves at each of these nodes; and the same moves
(here Not Confess, Confess) are available at each of these nodes.

 Not Confess
(оϭ͕оϭͿ
Player 2

Not Confess Confess

(оϰ͕ϬͿ
Player 1 Not Confess
(0, оϰͿ
Confess
Confess
Player 2
(оϯ͕оϯͿ
Fig. 6.2
Fig. 6.2 is consistent with the game being a simultaneous game. If we were
to use the game tree to illustrate the above game as a sequential one in
which Player 1 moves first which then is observed and reacted upon by
Player 2, then the game would be more correctly drawn without the ellipse
as:

Not Confess
(оϭ͕оϭͿ
Player 2
A
Not Confess Confess
(оϰ͕ϬͿ
Player 1 Not Confess
B (0, оϰͿ
Confess
Confess
Player 2
(оϯ͕оϯͿ

Fig. 6.3

The two nodes (say A and B) which signifies the move of Player 2 represent
the information set of Player 2. Clearly the information base of Player 2 at A

Imperfect and at B are not same in this case of sequential-move game. The Prisoner’s
Competition-II
Dilemma is widely studied throughout the Social Sciences. It is a compelling
scenario because the tensions it portrays between an individual player’s
self-interest and the group’s self-interest shows up in many different ways
around us. Understanding this shows us the reality of counter-productive
outcomes in a non-cooperative situation.

6.6 SOLVING A GAME THEORY PROBLEM

Having covered different representation forms of a game theory problem in
the previous section, let us now discuss how to find solution of game-
theoretic problem. We start with discussing some basic concepts regarding
different kinds of strategies.

6.6.1 Dominated Strategies

We consider the same illustration of the Prisoners’ Dilemma. When we
consider the payoffs associated with different decision options, we realise, if
one player chooses to confess, then the other player would prefer to
confess and be in the prison for three years, rather than opting to not
confess and suffer imprisonment for four years. In the same lines, when one
chooses to not confess, the other would prefer to confess and so get
released immediately rather than opting to not confess and suffer
imprisonment for a year. In both the situations and for both the players we
observe that option “not confess” is dominated by the option “confess”.
That is, for each strategy that Player 1 (or Player 2) opts, payoff to Player 2
(or Player 1) from not confessing would be less than payoff to him from
confessing. Hence, ‘Not Confess’ strategy is dominated by the ‘Confess’
strategy. Rational players in an attempt to maximise their positive payoffs
will not opt for a strictly dominated strategy. In Prisoners’ dilemma game,
rational players will choose to confess, and hence the equilibrium outcome
ŽĨ ƚŚĞ ŐĂŵĞ ǁŝůů ďĞ ;ŽŶĨĞƐƐ͕ ŽŶĨĞƐƐͿ ƌĞƐƵůƚŝŶŐ ŝŶ ƚŚĞ ƉĂLJŽĨĨ ŽĨ ;оϯ͕ оϯͿ͘
^ƵĐŚ Ă ƌĞƐƵůƚ ŝŶ ŶŽƚ ŽƉƚŝŵĂů ǁŚĞŶ ĐŽŵƉĂƌĞĚ ǁŝƚŚ ƚŚĞ ƉĂLJŽĨĨ ;оϭ͕ оϭͿ
associated with the strategy profile (Not Confess, Not Confess). In other
words, both players will suffer only one year of imprisonment when they opt
for the strategy ‘not confess’, but they end up opting for the strategy
‘confess’ and suffer imprisonment of three years each. We will discuss this
issue later in some more detail when we will come across the concept of
nash equilibrium.

Generally, in an n-player game, among strategies s’i and s’’i each belonging
to strategy set Si of an arbitrary player i, strategy s’i is dominated by strategy
s’’i if player i’s payoff from playing s’i is lower than the payoff he attains
from playing s’’i for each feasible combination of other players’ strategies
(i.e., whatever be the strategy played by the other players).


 Game Theory and
6.6.2 Dominant Strategies
its Applications
A dominant strategy for a player is the one that yields best payoff for that
player no matter what the other player does. That is, a dominant strategy is
a player’s best response to all the feasible strategies of the other player.
Generally, strategy si* will be referred to as the dominant strategy for
player i if he is strictly better off playing si* rather than any other strategy
regardless of what his opponent plays.
Please note: If a player has a dominant strategy then for this player all other
feasible strategies become dominated by this dominant strategy. But in
many situations, a player may not have a dominant strategy and yet may
have dominated strategies.

6.6.3 Dominant Strategy Equilibrium

A rational player is expected to play his dominant strategy, and thus, if all
players have a dominant strategy, then it is rational for them to choose the
dominant strategies and this way we reach at the dominant strategy
equilibrium. In terms of the Prisoners’ Dilemma, both players have ‘confess’
as a dominant strategy, that is, in that case, each player’s best response
would be to play confess regardless of what the other player might opt for.
dŚƵƐ͕ ǁĞ ŚĂǀĞ ;ŽŶĨĞƐƐ͕ ŽŶĨĞƐƐͿ ƌĞƐƵůƚŝŶŐ ŝŶ Ă ƉĂLJŽĨĨ ŽĨ ;оϯ͕ оϯͿ ĂƐ ƚŚĞ
dominant strategy equilibrium in case of Prisoners’ Dilemma.

6.6.4 Iterated Elimination of Strictly Dominated Strategies

Now, suppose players (one or both) have no dominant strategy. In that case,
we come to a method of reaching an equilibrium position of a game-
theoretic problem by way of elimination of strictly dominated strategies. We
discussed what is meant by dominated strategies in Sub-section 6.6.1.
Further this dominance could be strict or weak. Among strategies s’i and s’’i
each belonging to strategy set Si of an arbitrary player i, strategy s’i is strictly
dominated by strategy s’’i if for every strategy choice of the opponent,
player i’s payoff from choosing s’’i is strictly greater than player i’s payoff
from choosing s’i. On the other hand, strategy s’i will be said to be weakly
dominated by strategy s’’i if, i’s payoff from choosing s’’i is at least as good as
i’s payoff from choosing s’i. Here, s’i is called dominated strategy while s’’i is
referred to as dominant strategy.

Iterated elimination of strictly dominated strategies involves elimination of

strategies which are dominated relative to opponents’ strategies which have
not yet been eliminated until there are no strictly dominated strategies left.
The principle behind this method is that rational players of the game never
play a dominated strategy since they can get higher payoffs by playing
another strategy, no matter what the other players are playing. Hence, a
dominated strategy can be safely discarded to play a reduced form of game
with a smaller number of strategies. Consider the following payoff matrix

Imperfect where we have two players 1 and 2 with their respective strategies P, Q (for
Competition-II
player 1) and A, B, C (for player 2) and associated payoffs.

Player 2
A B C
Player 1 P (6,6) (1, 11) (4, 5)
Q (4,1) (3,3) (5, 6)

We can begin eliminating the dominated strategy of either of the player. For
instance, for Player 2, there are three strategies A, B and C. Irrespective of
Player 1’s strategy (either P or Q), strategy A for Player 2 is strictly
dominated by strategy B. This is because if Player 1 plays P, Player 2 gets 6
from playing A whereas 11 from playing B (6 < 11). Similarly if Player 1 plays
Q, Player 2 gets 1 from playing A whereas 3 from playing B (1 < 3). Hence,
strategy A can be eliminated for Player 2 so that the game reduces to:

Player 2
B C
Player 1 P (1, 11) (4, 5)
Q (3, 3) (5, 6)

Now irrespective of Player 2’s strategy (either B or C), strategy P for Player 1
is strictly dominated by Q. If Player 2 plays B, Player 1 gets 1 from playing P
and 3 from playing Q (1 < 3). Similarly, if Player 2 plays C, Player 1 gets 4
from playing P and 5 from playing Q (4 < 5). So player 1 being rational won’t
play the strictly dominated strategy P, so P can be eliminated so that the
game to reduce to:

Player 2
B C
Player 1 Q (3, 3) (5, 6)

Now Player 1 is left with one strategy Q. Player 2 will choose to play strategy
C for a payoff of 6 which is higher than the payoff of 3 from strategy B. Thus,
we arrive at the equilibrium solution strategy profile of (Q, C) yielding a
payoff of (5, 6) through iterated elimination of dominated strategy.

6.6.5 Nash Equilibrium

Presence of dominated strategies allows iteratively eliminating such
strategies from the game and successively reaching the equilibrium solution.
In many games, however, there are no dominated strategies or no
dominant-strategy, and hence ruling out any outcome is not an option.
Hence deriving the equilibrium is not possible. A more general approach of
deriving the equilibrium solution is a Nash Equilibrium. Nash equilibrium
refers to a set of mutually best response strategies, where each player

chooses his best (optimal) strategy given the strategy of the other players. Game Theory and
its Applications
Thus at Nash equilibrium no player has incentive to deviate from the chosen
strategy, given the Nash equilibrium strategy of its rivals. In other words, no
player can obtain a higher payoff by switching to a different strategy given
that all the other players stick to their Nash equilibrium strategy. For
instance, if two firms are selling a particular commodity, Nash equilibrium
will be an array of production levels, one for each firm, such that neither
firm can raise its profits by unilaterally deviating and making a different
choice, in other words, no firm will be able to gain by changing its
production level, given the strategy choice of the other player.
Generally, in an n-player game, strategy profile (s1*, s2*, s3*...., sn*) is said to
be a Nash equilibrium if for each player i, where si* is player i’s best
ƌĞƐƉŽŶƐĞ ƚŽ ƚŚĞ ŐŝǀĞŶ ƐƚƌĂƚĞŐŝĞƐ ŽĨ Ăůů ƚŚĞ ŽƚŚĞƌ Ŷ о ϭ ƉůĂLJĞƌƐ͘ ,ĞŶĐĞ͕
(•୧ ‫ כ‬ǡ •ି୧ ‫ כ‬ሻ is the Nash equilibrium strategy if: ܷ௜ ሺ•୧ ‫ כ‬ǡ •ି୧ ‫ כ‬ሻ ൒ ܷ௜ (•୧ ǡ •ି୧ ‫ כ‬ሻ; for
all •୧ ‫א‬Si, where ሺ•ି୧ ‫ כ‬ሻ is the Nash equilibrium strategy of all the (n – 1)
players and Ui is the payoff function. This shows that if player 1 deviates
from the Nash equilibrium strategy given the remaining (n – 1) players’
strategy, she (he) won’t gain. Same applied for other (n – 1) players. So it is
a mutually best response strategy.

Let us attempt to find the Nash equilibrium in case of the Prisoners’

Dilemma game. Consider the normal form of the game as given below:

Player 2
Not Confess Confess
Player 1 EŽƚŽŶĨĞƐƐ ;оϭ͕оϭͿ ;оϰ͕ϬͿ
 ŽŶĨĞƐƐ ;Ϭ͕оϰͿ ;оϯ͕оϯͿ

Now we proceed by underlining the best response of each player to a given

strategy of the other player. For instance, if Player 2 believes that Player 1
will choose not confess, then Player 2’s best response will be to play confess
ďĞĐĂƵƐĞǁŝƚŚƚŚĞƐƚƌĂƚĞŐLJ;EŽƚĐŽŶĨĞƐƐ͕EŽƚĐŽŶĨĞƐƐͿƚŚĞƉĂLJŽĨĨŽĨ;оϭ͕оϭͿ
implies Player 2 will suffer 1 years of imprisonment, while with the strategy
;EŽƚ ĐŽŶĨĞƐƐ͕ ĐŽŶĨĞƐƐͿ ƉĂLJŽĨĨ ŽĨ ;оϰ͕ϬͿ ŝŵƉůŝĞƐ WůĂLJĞƌ Ϯ ǁŝůů ďĞ ƐĞƚ ĨƌĞĞ ĨŽƌ
telling truth. We mark this best response of Player 2 to the strategy (not
confess) of Player 1 by underlining 0. This way we underline best responses
of each player to the given strategy of the other player as shown below:

Player 2
Not Confess Confess
Player 1 EŽƚŽŶĨĞƐƐ ;оϭ͕оϭͿ ;оϰ͕0)
Confess (0͕оϰͿ ;оϯ͕оϯ)

From the above payoff matrix, mutually best response strategy of both the
players is the Nash equilibrium in pure strategy i.e., each move involved is
the best response to the other moves. In other words, a cell in the normal

Imperfect form is a Nash equilibrium in pure strategy if each entry is marked
Competition-II
(underlined) as being the best response to the other moves. This way,
;ŽŶĨĞƐƐ͕ŽŶĨĞƐƐͿLJŝĞůĚŝŶŐĂƉĂLJŽĨĨŽĨ;оϯ͕оϯͿǁŝůůďĞƚŚĞEĂƐŚĞƋƵŝůŝďƌŝƵŵŝŶ
the above game.
But one may also note that both the players would have been better off if
both played ‘not confess’. When both play ‘not confess’ the resulting payoff
is one year of imprisonment. This is certainly better than the three year of
imprisonment that they suffer when both choose to play ‘confess’. Thus,
(Not Confess, Not Confess) is the Pareto optimal outcome of this game.
However, both players playing ‘not confess’ is not an equilibrium, because
given (say) Player 1 plays Not Confess, Player 2 always has an incentive to
deviate towards the strategy ‘confess’ and vice versa in the absence of
binding agreement. If binding agreement is possible, then both would agree
on the (Not confess, Not confess) combination, reaching a higher payoff.
Thus, rules of the games that do not allow binding agreements may induce
the players towards the strategy that results in lower payoffs for both of
them. Game-theoretic conditions like Prisoners’ Dilemma arise in many real
world settings. For instance, in a cartel agreement among suppliers of steel
to restrict output would lead to higher prices and profits if it could be
sustained, but such an agreement may be unstable because it may be too
tempting for an individual steel supplier to sell more output at the high
price.
Remember:

1) In game theory, every dominant strategy equilibrium is a Nash

Equilibrium. However, a Nash Equilibrium may or may not be a
dominant strategy equilibrium.

2) With just two strategies for each player if one strategy is dominant, the
other must be dominated. However with more than two strategies
available, a player might have dominated strategies but no dominant
strategy. If neither player has a dominant strategy, we can deduce
equilibrium by iteratively eliminating the dominated strategies and
successively moving towards the reduced form of game.

3) Nash equilibrium is widely used as an equilibrium definition because it

exists for all games. For games that at first appear not to have a Nash
equilibrium in pure strategy will end up having one in mixed strategies
(we will discuss this in the subsequent section).

6.6.6 Multiple Equilibria

Nash equilibrium is useful for it being stable and existing for all games.
However, an unpleasant situation arises when a game has multiple Nash
equilibria. The problem arises due to the fact that a unique outcome cannot
be predicted in such a case. To illustrate the possibility of multiple equilibria
we consider yet another classic game, the Battle of the Sexes.

Battle of the Sexes Game Theory and
its Applications
The game involves two players, a husband and a wife, planning an evening
out with both preferring to go together rather than going alone. The wife
wants to listen to an Opera performance, while the husband wants to watch
a Boxing match. The normal form for the game is given below:

Husband

Opera Boxing

Wife Opera (3, 1) (0, 0)

Boxing (0,0) (1, 3)

When both wife and husband end up at Opera, woman receives a payoff of
3 while her husband 1, on the other hand, when they end up attending a
boxing match, woman receives a payoff of 1 while her husband receives a
payoff of 3. When they both go to different locations, their payoff reduces
to 0. There is no dominant (or dominated) strategy for any of the player in
this game. A player would rather go for Opera if the other player chooses to
go for Opera, while the former would go for a boxing match if the later
chooses to go for a boxing match. It illustrates the fact that every game
cannot be solved using the iterated elimination of dominated strategies. For
Nash equilibrium we mark the best responses of the husband and the wife.

Husband
Opera Boxing
Wife Opera (3, 1) (0, 0)
Boxing (0,0) (1, 3)

For both of them, best response is to play the same action as the other. This
results in multiple pure strategy Nash equilibria, (Opera, Opera) and (Boxing,
Boxing) yielding payoffs of (3, 1) and (1, 3), respectively. It is not possible to
say which is pareto superior as they are symmetric.

6.7 MIXED STRATEGIES

The game of Prisoner’s Dilemma we considered so far involved playing a
single strategy with certainty or with a probability of 1, known as a pure
strategy. Some games involve playing more complicated strategies in the
form of mixed strategies than simply choosing a single strategy with
certainty. In mixed strategies, players’ choice of strategies follows a
probability distribution from among several possible strategies. Consider a
classic game known as the Matching Pennies to understand the concept of
mixed strategies.


Imperfect Matching Pennies
Competition-II
This is a game where two players 1 and 2, each possessing a coin (penny)
decides to simultaneously display their coins with either heads or tails facing
up. If the face of the coins matches with either both head or both tail faced
up, Player 2 gives his penny to Player 1; if face of the coins do not matches,
Player 1 gives his penny to Player 2. The normal form of the game is as
follows:

Player 2

Head Tail
Player 1 ,ĞĂĚ ;ϭ͕оϭͿ ;оϭ͕ϭͿ
 dĂŝů ;оϭ͕ϭͿ ;ϭ͕оϭͿ

Such a game is called a zero-sum game due to the fact that sum of the
payoffs of the players in each box add to zero. We will employ the same
method of underlining the best responses we applied in case of Prisoner’s
Dilemma to solve for the Nash equilibrium. We get the following result:

Player 2
Head Tail
Player 1 Head (1͕оϭͿ ;оϭ͕ϭ)
 dĂŝů ;оϭ͕ϭ) (1͕оϭͿ

There is no Nash equilibrium in pure strategies in this case as whatever

Player 2 opts for, Player 1 would want to choose the same option, but then
Player 2 would want to deviate to the other option, and this process of
moving to a different option would be endless. Thus, as long as one player
knows where the other player is, the latter gets a bad payoff. Therefore,
each would want to be unpredictable with their choices. Such a strategy is
referred to as a mixed strategy, and the resulting strategy profile with
associated probability for actions is called mixed strategy Nash equilibrium.

In a mixed strategy equilibrium, each of the player equate their expected

payoff from each of the strategy (Head and Tail) in equilibrium. In other
words, the players must be indifferent between the actions which they
choose to play. If a player was not indifferent between the available actions,
this would imply that one particular action yields a higher payoff than the
others, and the player would play that action with probability 1 rather than
mixing strategies with certain probability distribution.

To solve the Mixed Strategy Nash equilibrium, suppose that Player 2 opts for
,ĞĂĚ ǁŝƚŚ ƉƌŽďĂďŝůŝƚLJ Ɖ͕ ĂŶĚ ĐŚŽŽƐĞƐ dĂŝůƐ ǁŝƚŚ ƉƌŽďĂďŝůŝƚLJ ϭоƉ͘ ^ŝŵŝůĂƌůLJ
ƐƵƉƉŽƐĞWůĂLJĞƌϭƉůĂLJƐƚŚĞŐĂŵĞǁŝƚŚĂƉƌŽďĂďŝůŝƚLJĚŝƐƚƌŝďƵƚŝŽŶƋĂŶĚ;ϭоƋͿ
for Head and Tail. Now, the expected payoff of Player 1 for playing a pure


strategy of Head when Player 2 plays a mixed strategy with a probability Game Theory and
its Applications
ĚŝƐƚƌŝďƵƚŝŽŶƉĂŶĚ;ϭоƉͿŝƐ͗

‫ܧ‬ଵ ሺ‫ܪ‬ሻ ൌ ‫݌‬ሺͳሻ ൅ ሺͳ െ ‫݌‬ሻሺെͳሻ ൌ ʹ‫ ݌‬െ ͳ

Similarly, the expected payoff of Player 1 for playing a pure strategy of Tail
when Player 2 plays a mixed strategy with a probability distribution p and
;ϭоƉͿŝƐ͗

‫ܧ‬ଵ ሺܶሻ ൌ ‫݌‬ሺെͳሻ ൅ ሺͳ െ ‫݌‬ሻሺͳሻ ൌ െʹ‫ ݌‬൅ ͳ

If the Player 1 plays a mixed strategy, in the equilibrium, he should be
indifferent between the two expected payoffs from Head and Tail. Thus:
ͳ ͳ
‫ܧ‬ଵ ሺ‫ܪ‬ሻ ൌ ‫ܧ‬ଵ ሺܶሻ ฺ ʹ‫ ݌‬െ ͳ ൌ െʹ‫ ݌‬൅ ͳ ฺ ‫ ݌‬ൌ  ฺ ሺͳ െ ‫݌‬ሻ ൌ 
ʹ ʹ
Thus, Player 2 will play the game with a mixed strategy for Head and Tail
ଵ ଵ
with a probability distribution ‫ ݌‬ൌ ଶ ƒ†ሺͳ െ ‫݌‬ሻ ൌ ଶ ǡ ‹Ǥ ‡Ǥǡto keep Player 1
guessing, Player 2 must opt for Head with a probability of 1/2 and Tail with
the probability of 1/2. Similarly, we can find Player 1’s equilibrium mixed
strategy. Now, the expected payoff of Player 2 for playing a pure strategy of
Head when Player 1 plays a mixed strategy with a probability distribution q
ĂŶĚ;ϭоƋͿŝƐ͗

‫ܧ‬ଶ ሺ‫ܪ‬ሻ ൌ ‫ݍ‬ሺെͳሻ ൅ ሺͳ െ ‫ݍ‬ሻሺͳሻ ൌ െʹ‫ ݍ‬൅ ͳ

Similarly, the expected payoff of Player 2 for playing a pure strategy of Tail
when Player 1 plays a mixed strategy with a probability distribution q and
;ϭоƋͿŝƐ͗

‫ܧ‬ଶ ሺܶሻ ൌ ‫ݍ‬ሺͳሻ ൅ ሺͳ െ ‫ݍ‬ሻሺെͳሻ ൌ ʹ‫ ݍ‬െ ͳ

If Player 1 plays a mixed strategy, in the equilibrium, he should be
indifferent between the two expected payoffs from Head and Tail. Thus:
ͳ ͳ
‫ܧ‬ଶ ሺ‫ܪ‬ሻ ൌ ‫ܧ‬ଶ ሺܶሻ ฺ  െʹ‫ ݍ‬൅ ͳ ൌ ʹ‫ ݍ‬െ ͳ ฺ ‫ ݍ‬ൌ  ฺ ሺͳ െ ‫ݍ‬ሻ ൌ 
ʹ ʹ
Thus, Player 1 will play the game with a mixed strategy for Head and Tail
ଵ ଵ
with a probability distribution ‫ ݍ‬ൌ ଶ ƒ†ሺͳ െ ‫ݍ‬ሻ ൌ ଶ ǡ ‹Ǥ ‡Ǥǡto keep Player 1
guessing, Player 2 must opt for Head with a probability of ½ and Tail with
the probability of 1/2.
Thus, in mixed strategy Nash equilibrium, Player 1 would opt for Heads with
probability 1/2, and Tails with probability 1/2 and Player 2 would also play
Heads and Tails with probabilities 1/2 and 1/2 respectively. Such an
equilibrium can be represented as p*=1/2 and q*=1/2.
Note that, suppose ‫ܧ‬ଵ ሺ‫ܪ‬ሻ ‫ܧ ش‬ଵ ሺܶሻǡthat is, expected payoff of the Player 1
from Head is greater (lesser) than Tail. Obviously in that case a rational
player will not play mixed strategy but will play Head (Tail) with certainty.


Imperfect Thus if the player has to play Mixed strategy, in equilibrium the expected
Competition-II
payoffs from the different strategies should be equal.
Check Your Progress 1
1) Following is the payoff matrix in strategic form in which two players
have two strategies each.

Player 2
Left Right
Player 1 Top (7,3) (5,3)
 ŽƚƚŽŵ ;ϳ͕ϬͿ ;ϯ͕оϭͿ

A) Write the game in extensive form.

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….
……………………………………………………………………………………………………….

B) Find Nash equilibrium of the game.

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….
……………………………………………………………………………………………………….
C) Find dominant strategies of both players.

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….
……………………………………………………………………………………………………….

2) Following is a game in which the players have three strategies each.

Player 2
Left Centre Right
Player 1 Top (4,5) (1,6) (5, 6)
Middle (3,5) (2,5) (5, 4)
Bottom (2, 5) (2, 0) (7, 0)

a) Write the game in extensive form.

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….

 b) Find Nash equilibrium in pure strategy of the game. Game Theory and
its Applications
……………………………………………………………………………………………………….

……………………………………………………………………………………………………….
……………………………………………………………………………………………………….

3) Are dominant strategy always Nash equilibrium? Is the converse true?

…………………………………………………………………………………………………………….

…………………………………………………………………………………………………………….
…………………………………………………………………………………………………………….

…………………………………………………………………………………………………………….

…………………………………………………………………………………………………………….

4) From the following payoff matrix where the payoffs are the profits or
losses of the two firms, determine

Firm B

Low Price High Price

Firm A >ŽǁWƌŝĐĞ ;Ϯ͕ϮͿ ;ϰ͕оϭͿ
 ,ŝŐŚWƌŝĐĞ ;оϭ͕ϰͿ ;ϲ͕ϯͿ

a) Whether firm A has a dominant strategy?

……………………………………………………………………………………………………….
……………………………………………………………………………………………………….
……………………………………………………………………………………………………….

……………………………………………………………………………………………………….

b) Whether firm B has a dominant strategy?

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….
……………………………………………………………………………………………………….

……………………………………………………………………………………………………….

c) Find the Nash equilibrium in pure strategy, if there is one.

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….

Imperfect 5) From the following payoff matrix, where the payoffs refer to the profits
Competition-II
firms earn by cheating and not cheating in a cartel

Firm B
Cheat Don’t Cheat
Firm A Cheat (3,2) (9,1)
Don’t Cheat (2,6) (7,4)

a) Determine the Nash equilibrium in pure strategy.

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….
……………………………………………………………………………………………………….

b) What if we change the payoff of the bottom left cell to (4, 4)?

……………………………………………………………………………………………………….

……………………………………………………………………………………………………….
……………………………………………………………………………………………………….
7) In the Battle of Sexes game (given in the Section 6.6), two pure strategy
Nash equilibria are derived. Derive the mixed strategy Nash equilibrium
of the game (if there is any).
……………………………………………………………………………………………………………

……………………………………………………………………………………………………………

……………………………………………………………………………………………………………
……………………………………………………………………………………………………………
……………………………………………………………………………………………………………

8) If any of the player in a game is having a dominant strategy, do you

think that the player will ever play mixed strategy ? Why ?and Why not?

……………………………………………………………………………………………………………
……………………………………………………………………………………………………………

……………………………………………………………………………………………………………
……………………………………………………………………………………………………………

6.8 SEQUENTIAL GAMES

So far, we have been considering games in which players make decisions
simultaneously, that is, where all the players (possessing no prior
information about the actions of their opponents) act at the same time. In
 such games, time plays no role. So those games are often called Static
games. In many games, however, one player can move before the other Game Theory and
its Applications
players. Such games are referred to as sequential-move games where
players take action at well-defined turns (over time), and have perfect
knowledge about what the other player(s) did at previous turns. In other
words, in a two-player sequential-move game, one player (the first mover)
takes an action before another player (the second mover). The second
mover observes the action taken by the first mover before it decides what
action it should take. Now consider the same game of the Battle of the
Sexes with the same strategies and payoffs, but with sequential moves. We
assume, instead of them both moving simultaneously, the wife is the first to
move and makes a choice between Opera or Boxing, on observing which the
husband decides next between the two options. Illustration of the game is
presented in the following extensive form:

Opera (3, 1)
Husband
Boxing
Opera
A (0, 0)
Wife Opera
(0, 0)
Boxing
B Boxing
Husband
(1, 3)
Fig. 6.4

As one may notice, the oval around the decision nodes of Player 2 (that was
there in case of simultaneous-move game) has been removed. This is
because with sequential moves, husband (second mover) can observe his
wife’s (first mover) action and hence is aware of the decision node he is on
before he takes his action. The wife’s possible strategies remain unchanged
at Opera or Boxing. But the husband’s set of possible strategies has
expanded to four strategies; as for each of the wife’s two actions, he can
choose one of two actions. So the husband has the option to move either
from node A or node B, that is, he is not in the same information base. Note
that to define the husband’s strategies completely we must define an action
at each information set that he makes a choice at, even if that information is
not actually reached in the game. Thus all strategies in his strategy profile,
include an action for both the top and bottom information sets. To solve for
the Nash equilibria, we will consider the normal form of the game and use
the method of underlining payoffs for best responses:


Imperfect
Competition-II
Husband
Always Opera Follows Wife Opposite to Always Boxing
(Opera|Opera) (Opera|Opera) Wife (Boxing|Opera)
(Opera|Boxing) (Boxing|Boxing) (Boxing|Opera) (Boxing|Boxing)
(Opera|Boxing)

Wife Opera (3, 1) (3, 1) (0, 0) (0, 0)

Boxing (0,0) (1, 3) (0, 0) (1, 3)

Now we have a 2 × 4 matrix, where wife has a strategy profile SW = {Opera,

Boxing}, while husband has the strategy profile SH= {(Opera|Opera)
(Opera|Boxing), (Opera|Opera) (Boxing|Boxing), (Boxing|Opera)
(Opera|Boxing), (Boxing|Opera) (Boxing|Boxing)}. Here, (Opera|Opera)
(Opers|Boxing) for instance means that husband opts for Opera given that
wife has chosen Opera while husband opts for Opera after wife has chosen
to go for a Boxing match. Now the underlining of the best responses
ensuring we underline the payoffs for all the strategies that tie for the best
response as well is shown below:

Husband
Always Opera Follows Wife Opposite to Always Boxing
(Opera|Opera) (Opera|Opera) Wife (Boxing|Opera)
(Opera|Boxing) (Boxing|Boxing) (Boxing|Opera) (Boxing|Boxing)
(Opera|Boxing)
Wife Opera (3, 1) (3, 1) (0, 0) (0, 0)
Boxing (0,0) (1, 3) (0, 0) (1, 3)

This way, we get multiple Nash equilibria:

i) Wife opts Opera, husband opts (Opera|Opera) (Opera|Boxing)

ii) Wife opts Opera, husband opts (Opera|Opera) (Boxing|Boxing)

iii) Wife opts Boxing, husband opts (Boxing|Opera) (Boxing|Boxing)

Among the three equilibria let us analyse the plausibility of each. We start
with the first Nash equilibrium. Husband’s strategy (Opera|Opera)
(Opera|Boxing) involves a threat to wife that he will choose Opera when his
wife will go for a Boxing match. However, this is an empty threat. This
results from the fact that if the wife really were to choose a boxing match
first, then husband would be giving up a payoff of 3 by choosing Opera
rather than Boxing, which is certainly hard to believe. Thus, husband’s threat
to his wife to opt for Opera if the wife goes for a boxing match is non-
credible. On the similar lines, the husband’s strategy in the third Nash
equilibrium (Boxing|Opera) (Boxing|Boxing) involves an empty threat of him
choosing to go for a boxing match when his wife opts for Opera in the first

move. As for this he will be giving up a payoff of 1 by choosing Boxing rather Game Theory and
its Applications
than Opera.

6.8.1 Sub-game Perfect Nash Equilibrium (SPNE)

Nash equilibrium in sequential games may result in strategy profiles which
are not very plausible. A formal procedure for selecting a reasonable Nash
equilibrium strategy profile (in a way that rules out empty threats) is done
using the concept of Sub-game Perfect Nash equilibrium (SPNE) which
requires that for equilibrium to be rational it should be an equilibrium not
just for the game as a whole, but also for each subgame of the game. A
subgame is a part of the extensive form beginning from any node of the
game and including everything that branches out below it. A decision node
initiates a subgame if neither it nor any of its successors are in an
information set that contains nodes that are not successors to it. Such a
subgame is termed a proper subgame. This way, the sequential Battle of the
Sexes game has three subgames, viz. the game itself, and the two lower
subgames beginning from the nodes where the husband gets to decide. We
mark these subgames with a dashed rectangle as shown in the figure:

Opera (3, 1)
Husband (ii)
Boxing
Opera
(0, 0)
Wife Opera (i)
(0, 0)
Boxing
Boxing (iii)
Husband
(1, 3)

Fig. 6.5

A sub-game Perfect Nash Equilibrium is a set of strategies, one for each

player, that induce a Nash equilibrium in every proper subgame. As far as
the sequential Battle of the Sexes game is concerned, for a strategy profile
to be termed as a SPNE, in addition to it being a Nash equilibrium on the
whole game, it must be the Nash equilibrium of the two other proper
subgames. In the above set-up, given that wife chooses Opera, in the
subgame (ii) the husband has to decide between Opera (resulting in payoff
of 1) and Boxing (yielding a payoff of 0). His best response in this subgame
will be to go for Opera. Similarly, given that wife chooses to go for a Boxing
match, husband in the subgame (iii) is to decide between Opera (resulting in
payoff of 0) and Boxing (yielding payoff of 3). His best response would be to
go for a Boxing match. Thus, (Opera|Opera) (Boxing|Boxing) is the only
strategy profile of the husband that forms a part of the SPNE. The other two
strategy profiles like (Opera|Opera) (Opera|Boxing) and (Boxing|Opera)
(Boxing|Boxing) results in him playing something that is not a Nash

Imperfect equilibrium on some proper subgame. Thus, among the three Nash
Competition-II
equilibria we came across for the game in Section 6.8, only the second one is
subgame perfect Nash equilibrium while the first and the third are not.

6.8.2 Backward Induction

In the previous sub-sections, we solved for the equilibrium in the sequential
Battle of the Sexes game by finding the Nash equilibria using the normal
form and then look for a subgame perfect Nash equilibrium among them.
Another method providing a somewhat direct way of solving for the
subgame perfect Nash equilibrium in such a setting is the method of
Backward Induction. In this method we start with the subgames at the
bottom of the extensive form, and determine the Nash equilibrium of these
subgames. These subgames are then replaced by their respective Nash
equilibrium. This process of replacing a subgame with the associated Nash
equilibrium is then continued up to the next level of subgames till we reach
a subgame perfect Nash equilibrium. The process is illustrated below:

Opera (3, 1)
Husband (ii) Opera (3, 1)
Boxing
Opera Opera
(0, 0)
Husband
Wife Opera (i) Wife
(0, 0)
Boxing Boxing
Boxing (iii) Boxing (1, 3)
Husband
(1, 3)

Fig. 6.6

In Fig. 6.6 the method of Backward Induction involves first solving the two
subgames (ii and iii) for Nash equilibrium. In subgame (ii), given that wife
opts for Opera, husband’s best response would be to choose Opera for a
payoff of 1 rather than going for Boxing match resulting in payoff of 0.
Similarly, in subgame (iii), given that wife opts for Boxing, husband’s best
response will be to go for a Boxing match. Now, we replace the two
subgames with their respective Nash equilibrium strategies to get a simple
game where wife is to decide. Wife gets a payoff of 3 if she goes for Opera,
while Boxing results in a payoff of 1. Nash equilibrium strategy yielding the
higher payoff will be for her to go for Opera. So the wife’s best response is
to go for Opera. Thus, we get the subgame perfect Nash equilibrium
outcome as (Opera|Opera).


 Game Theory and
6.9 APPLICATIONS OF THE GAME THEORY its Applications

We also came across important equilibrium concepts and a few examples of

games in the previous sections. The purpose of this section is to give an
overview of its application in order to capture models of conflict and
cooperation in the field of Economics. Also this theory is widely applied in
the field of, Biology, Sociology, Political Sciences, etc. to predict important
trends. For instance, firms operating in an Oligopolistic market structure
face nothing but a game theoretic situation where firms are interdependent
on each other. The concept of Nash equilibrium which we came across in
the present unit, can help us to define the Cournot equilibrium, described in
Unit 6. The Cournot equilibrium has the property that each firm is choosing
its profit-maximising optimal output, given the choice of the other firm.
Similarly, we have the Bertrand equilibrium, as the Nash equilibrium in
pricing strategies, which is the mutually best response price strategy. Each
firm chooses the optimal price that maximises its profit, given the price that
it thinks the other firm will set. We consider a few other examples below:
Example 1: Consider a game that discusses Prisoner’s Dilemma in the
context of firms attempting to make decisions about how they should
operate without knowing about their competitor’s actions. The simplest
oligopoly model is by Cournot (1838), known as Cournot duopoly model,
where there are two competing firms, say i=1,2 producing qi amounts of
output each. There is a single homogeneous good being produced by the
two firms, with demand function p(Q), where p stands for price which is a
function of total quantity Q = q1 + q2, produced by the firms. According to
Cournot, each firm simultaneously takes output decision, without actually
knowing the output choice of rival firm. We seek the non-cooperative
equilibrium of the game, where firms cannot write binding contract and
uses the output as their strategy variable.
Example 2: Consider two symmetric firms, A and B, having same constant
average cost of Rs. 2 per unit. The firms face a total market demand of 100
and 180 units corresponding to two different prices i.e., either a high price
of Rs. 10 or at a low price of Rs. 5 respectively. In both these cases, the
market demand is splitted between the two firms. If one firm sets high price
and other low price, the low priced firm sells 150 units and high priced firm
sells only 20 units of the commodity. Now we calculate the profits of these
firms and derive the equilibrium set of strategies using Prisoner’s Dilemma
game.
We have the following possible strategy profiles for {Firm A, Firm B} and
their associated payoffs as (Firm A, Firm B):
a) {High price, High price}: In this case, total demand = 100, so each firm
will sell 50. Total revenue of an individual firm = 50 × 10= 500 and Total
ĐŽƐƚсϱϬпϮсϭϬϬ͘dŚĞƌĞďLJWƌŽĨŝƚсϱϬϬоϭϬϬсϰϬϬĨŽƌĞĂĐŚĨŝƌŵ͘

Imperfect b) {low price, low price}: Now, total demand = 180, so each firm will sell
Competition-II
90. Total revenue of an individual firm = 90 × 5 = 450 and Total cost = 90
пϮсϭϴϬ͘dŚĞƌĞďLJWƌŽĨŝƚсϰϱϬоϭϴϬсϮϳϬĨŽƌĞĂĐŚĨŝƌŵ͘
c) {High price, low price}: Suppose Firm A sells at high price, its total
demand is 20 units, so Revenue of Firm A = 20 × 10 = 200 and Total cost
сϮϬпϮсϰϬ͘dŚĞƌĞďLJWƌŽĨŝƚсϮϬϬоϰϬсϭϲϬĨŽƌĮƌŵ͘&ŝƌŵƐĞůůĂƚ
low price, its demand is 150, so Revenue of Firm B = 150 × 5 = 750 and
Total cost = 150 × 2 = 300. Thereby Profit = 750 – 300 = 450 for firm B.
d) {low price, high price}: Contrary to above point, assume firm A sells at
low price and B sells at high price. Then with similar calculation as
above, Firm A profit is Rs. 450 and firm B profit is Rs. 160.
The payoff (profit) matrix therefore is

Firm B’s Profit

High Price Low Price
Firm A’s Profit High Price (400,400) (160,450)
Low Price (450,160) (270, 270)

Each player has a dominant strategy of charging low price. The equilibrium
is therefore {low price, low price} yielding payoff (270, 270). It is a scenario
similar to Prisoner’s Dilemma game, where the equilibrium outcome is the
one which gives lowest joint-payoffs (= 270 + 270 = 540). The strategy {Low
Price, Low Price} is the Nash equilibrium as well as Dominant strategy
Equilibrium. However the outcome {high price, high price} with a highest
joint-payoffs (= 400 + 400 = 800) is a Pareto improvement over the Nash
equilibrium outcome.

Why this paradoxical situation arises in equilibrium? From the Dominant

strategy equilibrium perspective obviously none of the firm will choose the
dominated strategy {High Price} over the dominant strategy. Hence the
Dominant strategy equilibrium is {Low Price, Low Price}. Let us now analyse
it from the perspective of Nash equilibrium. From the payoff matrix, it is
evident that even though each of the firm has an incentive to choose High
Price strategy because the strategy {high price, high price} offers highest
joint profit, but each firm knows that if it chooses (High Price), the rival firm
has the incentive to deviate and its best response is to choose (Low Price)
and appropriate higher profit of Rs. 450 (450 > 400). Thus there is a mutual
threat for deviation for Low Price against the choice of High Price. This
credible threat actually compels the firms to choose the mutually best
response strategy, even though it is a sub optimal outcome {Low Price, Low
Price}.
Example 3: Now consider a non-price competition where two firms are
deciding whether to spend on an expensive advertising campaign. If none of
the firm advertise they will earn a normal profit of Rs. 50 million. If one firm

advertise, it will earn more profit say Rs. 75 million due to comparative Game Theory and
its Applications
advantage, whereas the other firm which did not advertise will have to face
loss and will earn only Rs. 25 million. If one firm advertises and the other
does not, the firm that advertised will gain profit despite advertising cost, as
advertising will help this firm’s to capture larger market size through
product popularity, over the other firm. If both firms advertise, they will be
in same situation in the market but have to bear extra loss of advertisement
cost and hence their profit will be Rs. 20 million each. So payoff matrix is
given by:

Firm B’s strategy

Advertise Do not Advertise
Firm A’s Advertise (20,20) (75,25)
strategy
Do not (25,75) (50,50)
Advertise

In this game, neither of the two firms will have a dominant strategy, hence
no dominant strategy equilibrium. But the game will be having two Nash
equilibria, viz. {Do not advertise, Advertise} and {Advertise, Do not
advertise}.

Example 4: Product Choice Problem

Following is a product choice problem faced by two firms, who have to
decide whether to produce a salty or a sweet snack. If both produce the
same variant, the market will have excess supply and both will end up
making losses. The payoff matrix is as follows:

Firm 2
Salty Sweet
Firm 1 ^ĂůƚLJ ;оϯ͕оϯͿ ;ϭϬ͕ϭϱͿ
 ^ǁĞĞƚ ;ϭϱ͕ϭϬͿ ;оϯ͕оϯͿ

Clearly in a static (or simultaneous move) game there are two Nash
equilibria in pure strategies (Sweet, Salty) and (Salty, Sweet). Let us consider
the sequential move (extensive form) of the game (see Fig. 6.7 and 6.8)

Salty (о3, о3)

Firm 2
Salty
(10, 15)
Sweet
Firm 1 Salty
(15, 10)
Sweet
Firm 2
Sweet (оϯ͕оϯͿ

Fig. 6.7

Imperfect
Competition-II

Salty
(10, 15)

Firm 1
Sweet (15, 10)
Fig. 6.8: (Reduced form of the Game)

Suppose firm 1 gets to move first, followed by firm 2. If it chooses the

strategy ‘sweet’, firm 2 after observing firm 1’s choice, will choose ‘salty’. If
firm 1 chooses the strategy ‘Salty’, firm 2 after observing firm 1’s choice will
choose ‘sweet’. Notice, firm 2’s choices actually arise from firm 1’s choices,
which the payoff matrix does not reveal. As it hides the fact that firm 2 gets
to know choice of firm 1. Through the backward induction the Subgame
perfect Nash equilibrium outcome of the game is (Sweet, Salty) and the
Subgame perfect Nash equilibrium strategy is (Sweet, Salty) , (Salty Sweet),
as shown in the reduced form of the game (Fig. 6.8)

Example 4: Business Game

Let there be a firm A which is trying to enter a business, either on a small
scale or large scale. Firm B now has to decide whether to accommodate or
start a price war. So the game essentially involves the two strategies of
whether a firm enters business on a large scale or, small scale. Refer the
following payoff (in millions of rupees) matrix for the game:

Firm B
Accommodate Price War
Small (2,10) (0,8)
Firm A
Large (4,5) (1, 12)

As per the above payoff matrix, for firm A ‘large’ is the dominant strategy,
given firm A choose this, firm B will respond by launching a ‘price war’. At this
the Nash Equilibrium will be (Large, Price War) where firm A will have a payoff
of 1 and firm B a payoff of 12. But firm A can do better if you turn this into a
sequential-move game. Here firm A can commit to a scale of operation in
advance. If firm A decide ‘large’, then firm B’s best response is to fight a ‘price
war’, resulting in firm A’s payoff to be 1. But if firm A choose ‘small’ then firm
B’s best response will be to accommodate yielding a payoff of 2 to firm A.
Hence Sub game Perfect Nash Equilibrium strategy in this sequential move
game is (Small, Accommodate, Large, Price War), where firm A enters ‘small’
and firm B ‘accommodates’ in the top information set, and engages in a price
war in the bottom information set. Hence, equilibrium payoffs is (2,10).The
Sub-game Perfect Nash Equilibrium outcome is (Small, Accommodate). This
shows moving first can have strategic value.

The extensive form can be depicted as follows: Game Theory and
its Applications
Accommodate (2, 10)
Firm B
Price War
Small
(0, 8)
Firm A Accommodate
(4, 5)
Large
Price War
Firm B
(1, 12)
Fig. 6.9

Threats, Commitment and Credibility

Firms in oligopoly often adopt strategies to gain a competitive advantage

over their rivals even if it means constraining their own behaviour. An
Oligopolist must have a commitment, so that the threat it makes is credible.
This can be explained by an example, let firm A produce cars and firm B
produce car seats, where production decisions depend heavily on firm A.

Firm A
Small Cars Big Cars

Firm B Small Car Seats (2,6) (2,0)

Big Car seats (1,1) (10,3)

We have a sequential game, in which firm A is the first mover (leader), it can
do best by making small cars and so firm B should make small car seats.

ĂŶĨŝƌŵƚŚƌĞĂƚĞŶĨŝƌŵƚŽƉƌŽĚƵĐĞďŝŐĐĂƌƐĞĂƚƐ‫خ‬EŽ͕ďĞĐĂƵƐĞŝĨŝƚŵĂŬĞƐ
big car seats and firm A makes small cars, firm B would earn a payoff of ‘1’
and not ‘2’. Therefore this threat won’t work out and it is called an empty
threat. In other words, this threat is not a credible threat. However it can
make its threat credible by shutting down small car seats factory. The new
payoff matrix is now:

Firm A
Small Cars Big Cars
Small Car Seats (0,6) (0,0)
Firm B
Big Car seats (1,1) (10,3)

Now firm A would be forced to produce big cars and firm B would produce
big car seats and earn a payoff of ‘10’. But here one may question, what if
ĨŝƌŵĨŝŶĚƐĂŶŽƚŚĞƌƉƌŽĚƵĐĞƌƚŽƉƌŽĚƵĐĞƐŵĂůůĐĂƌƐĞĂƚƐ‫خ‬,ĞƌĞ͕ƚŚĞƌŽůĞŽĨ
reputation becomes very important.
Check Your Progress 2
1) Two firms are planning to enter into a market. Firm 1 is contemplating
its capacity strategy, which can be “aggressive” or “accommodating”.
Firm 2 also has similar options. The payoff matrix is as follows:

Imperfect
Competition-II
Firm 2
Aggressive Accommodating
Aggressive (7,2) (10,3)
Firm 1
Accommodating (9,5) (11,4)

a) If both firms decide their strategies simultaneously, what is the

Nash Equilibrium in pure strategy ?
……………………………………………………………………………………………………..

……………………………………………………………………………………………………..

……………………………………………………………………………………………………..

b) Write the game in Extensive form and find the optimal strategy of
firm 1, if it could move first. What would firm 2 doǫ

……………………………………………………………………………………………………..

……………………………………………………………………………………………………..

……………………………………………………………………………………………………..
2) Two cereal manufacturers firm A and firm B are contemplating
manufacturing a cereal made from either wheat or rice.
The payoff matrix is as follows:

Firm 2
Wheat Rice
Firm 1 Wheat (10,12) (6,8)
Rice (1,6) (8,12)

a) If two firms choose simultaneously what is the Nash Equilibrium?

……………………………………………………………………………………………………..

……………………………………………………………………………………………………..

……………………………………………………………………………………………………..
b) If firm 1 chooses first, what is the optimal strategy of firm 2 ? Does
firm 1 have any advantage in moving first ?

……………………………………………………………………………………………………..

……………………………………………………………………………………………………..

……………………………………………………………………………………………………..

……………………………………………………………………………………………………..

……………………………………………………………………………………………………..


 Game Theory and
6.10 LET US SUM UP its Applications

Game theory is a set of tools that economists use to analyse conflict and
cooperation between firms or any other rational players or decision-making
agents. Each firm adopts a strategy or plan of action to compete with other
firms. The unit discusses various elements employed in representing a game.
Every game theory model includes players, strategies and payoffs. The
players are the decision-makers and the strategies are the potential choices.
The payoff is the reward or profit or the outcome of each combination of
strategies. The unit also discussed the two forms of representing a game,
viz. the normal and the extensive form. It continued in explaining various
basic concepts needed to solve a game theoretic problem. The dominant
strategy is the optimal choice for a player, no matter what the strategy of
the other player is. When both the players are having a dominant strategy,
we reach a dominant strategy equilibrium. Nash Equilibrium occurs when
player choose mutually best response strategy, given the strategy chosen by
the other player. Examples of various games including, the Prisoners’
Dilemma, the Battle of the Sexes, Matching Pennies, etc. were discussed in
order to illustrate the concepts of equilibrium. The concept of the Mixed
strategy Nash equilibrium arises when there is no Nash equilibrium in Pure
strategies. From the simultaneous-move games, we moved next to a
Sequential-move games where the time pattern of choices is important. The
two additional concepts of that of a Sub-game Perfect Nash equilibrium and
the method to reach such equilibrium known as the Backward induction
method, were also covered with illustration. The unit concluded with a
discussion on some application of the Game theory.

6.11 SOME USEFUL REFERENCES

x Varian, H. R. (2010). Intermediate Microeconomics: A modern
approach (8thed.). New York: W.W. Norton & Co.
x Gibbons, R. (1992). Game Theory for Applied Economists. Princeton,
New Jersey: Princeton University Press.

x Nicholson & Snyder. Intermediate Microeconomics and its Application

(11thed).South-Western Cengage Learning.
x “Notes 6: Game theory”. retrieved from
http://www2.econ.iastate.edu/classes/econ335/lewin/games.pdf


Imperfect
Competition-II 6.12 ANSWERS OR HINTS TO CHECK YOUR PROGRESS
EXERCISES
Check Your Progress 1

1) a)
Left (7, 3)

Top Right
(5, 3)
Player 1 Left (7, 0)
Bottom
Right

(3, о1)
Player 2

b) There are 3 Nash Equilibria

{Top, Left}, {Top, Right} and {Bottom, Left}

c) Dominant Strategy of player 1 is Top

Dominant Strategy of player 2 is Left

2) a)

Left (4, 5)
Center (1, 6)

Top Right (5, 6)

Left (3, 5)
Middle Center
Player 1 (2, 5)

Right (5, 4)

Bottom Left (2, 5)

Center (2, 0)

Player 2 Right (7, 0)

b) Nash Equilibrium of the game is (Middle center)

3) a) Yes, if all players are using their dominant strategies then it would
be the case that it is optimal given what the other players are doing
and therefore a Nash Equilibrium. No, the converse is not true.
4) a) Firm A does not have a dominant strategy.
b) Firm B has a dominant strategy of low price.
c) Nash Equilibrium is {Low price, Low price}
5) a) Nash Equilibrium is {Cheat, Cheat}
b) Nash Equilibrium would be {Don’t Cheat, Cheat}

Check Your Progress 2 Game Theory and
its Applications
1) a) Nash Equilibrium would be {Accommodating, Aggressive}

b) If firm 1 could move first it would opt to be ‘Aggressive’followed by

firm 2 to be choosing ‘Accommodating’.

Aggressive (7, 2)

Aggressive
(10, 3)
Accommodating
Firm 1
Aggressive (9, 5)
Accommodating

Accommodating (11, 4)

Firm2

2) a) The two Nash Equilibrium are {wheat, wheat} and {Rice, Rice}.
b) Firm 1 should choose wheat and firm 2 would also choose wheat,
giving firm 1 a payoff of 10. Yes firm 1 has an advantage of moving
first. If firm 2 were to move first, it would choose Rice,then firm 1
would also choose Rice giving firm 1 a payoff of only 8.