Oligopoly

Bruno Salcedo∗
Winter 2020

An oligopoly is market operated by a few firms. The term “a few” is to be

interpreted as follows. First, it means that there is more than one firm. Second, it
also means that there are not that many firms so that all firms act as price takers.
At least some firms know that they can have a significant effect on the market
price. Oligopolies cannot be analyzed using our perfect competition model, nor
our monopoly model. These notes provide some models that are better suited to
capture the strategic aspects of oligopolistic markets.

1. Benchmark models

All the models in these notes are presented in the context of a specific market.
The total market demand is given by:

D(p) = 20 − 2p. (1)

All firms operating in this market have the same cost function given by:

C(q) = 2q. (2)

As a benchmark, let us start by solving our competitive and monopolistic models

in the context of this market.
∗
Department of Economics, Western University · brunosalcedo.com · [email protected]
Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported license cbna.

1
 p
q = D(p)

pC b
p = MC(q)

q
qC
Figure 1 – Competitive equilibrium

Suppose that all firms are price takers. The market equilibrium would be
determined by the intersection of supply and demand. See Figure 1. The equi-
librium quantity would be q C = 16 and the equilibrium price would be pC = 2.
Firms would make zero profits (this is because of the constant marginal cost).
And the total market surplus would be the area of the blue triangle in the figure,
8 × 16/2 = 64.
Suppose instead that the market is operated by a monopolist. To solve the
monopolistic problem, we need the inverse market demand, which is given by:
q
P (q) = 10 − . (3)
2
The monopolist’s profits are given by

1
π(q) = qP (q) − C(q) = 8q − q 2 . (4)
2
The optimal quantity chosen by the monopolist is given by the first-order condi-
tion: 8 − q M = 0. Consequently, the monopolist would produce q M = 8. The cor-
responding market price is given by the inverse demand function pM = P (q M ) = 6.
The monopolist profits would be 8 × (6 − 2) = 32. This would result in a dead-
weight loss of 16. See Figure 2.

2
 p
q = D(p)

pM b

pC p = MC(q)

q
qM qC
Figure 2 – Monopoly

2. Cournot duopoly

Now suppose that there are two firms, Firm 1 and Firm 2. Each firm chooses
how much it produces. These choices are made independently and simultaneously.
This model is called Cournot duopoly. The quantity produced by firm j is denoted
by qj ≥ 0. The price at which firms can sell their production is determined by the
inverse demand function and the total production. That is, p = P (q1 + q2 ). The
revenue of each firm equals the market price times the quantity sold by the firm.
Firm j’s profit function is thus given by:

1 1
 
πj = qj P (qj + q−j ) − C(qj ) = 8 − q−j qj − qj2 . (5)
2 2
Let us compare this profit function with that of a monopolist, given in equation
(4). The difference is in the first term. The duopolist faces a smaller market size,
because part of the market is serviced by another firm. The duopolist optimal
choice depends on the quantity produced by its competitor. This optimal choice
is characterized by the first order condition (8 − q−j /2) − qj = 0. Firm i’s best
response function is thus

1
BRi (q−i ) = 8 − q−j . (6)
2

3
2.1. Rationalizability

Equation (6) is not enough to predict the output when there are two firms,
but it is a start. Note that the firms’ best response functions never take values
greater than 8. See the first panel of Figure 3. In fact, we could show (but we
won’t), that every quantity greater than 8 is strictly dominated. Rational firm
would never choose quantities greater than 8.
Now suppose that Firm 2 know that Firm 1 is rational. Then, Firm 2 would
know that q1 would be between 0 and 8. Firm 2’s best responses would thus be
between BR2 (8) = 4 and BR 2 (0) = 8. See the second panel of Figure 3. If there is
mutual knowledge of rationality, then each firms will produce a quantity between
4 and 8.
Now suppose that Firm 1 knows that Firm 2 knows that Firm 1 is rational.
Then, Firm 2 would know that q1 would be between 4 and 8. Firm 1’s best
responses would thus be between BR2 (8) = 4 and BR2 (4) = 6. See the third
panel of Figure 3. If there is 2nd mutual knowledge of rationality, then each firms
will produce a quantity between 4 and 6.
We can continue this process. If there is 3rd mutual knowledge of rationality,
then each firms will produce a quantity between BR i (6) = 5 and BRi (4) = 6.
If there is 4th mutual knowledge of rationality, then each firms will produce a
quantity between BR i (6) = 5 and BR i (5) = 5.5. If there is 5th mutual knowledge
of rationality, then each firms will produce a quantity between BRi (5.5) = 5.25
and BRi (5) = 5.5. The upper and lower bounds approach one another. In the
limit, they coincide, which means that only one quantity survives. See the fourth
panel of Figure 3.

Claim 1 In Cournot duopolies, the only rationalizable outcome is the intersection

of the best response functions.

2.2. Cournot equilibrium

The intersection of the best response functions is called the Cournot equilib-
rium. In general Cournot competition models, it can be found by solving a system
of n variables and n equations of the form qi = BRi (q−i ), where n is the number
of firms. When all firms have exactly the same utility function the equilibrium

4
q2 q2

q1 = BR1 (q2 )

8 8

q2 = BR2 (q1 ) 4

8 q1 4 8 q1

q2 q2

8 8
6 6 b

4 4

4 6 8 q1 4 6 8 q1

Figure 3 – Iterated dominance in Corunot competition

5
 p
q = D(p)

pM

pD b

π1D π2D

pC p = MC(q)

q
qiD qM qD qC
Figure 4 – Cournot

price quantity profits surplus dwl

Competitive 2 16 0 64 0
Monopoly 6 8 32 48 16
Cournot 4.6̄ 10.6̄ 28.4̄ 28.4̄ 7.1̄

Table 1 – Comparison of different market arrangements

is symmetric, in that all firms produce the same quantity. In such cases, we can
find the Cournot equilibrium by solving a single equation qiD = BR i (qiD ).
Our example is symmetric. Using equation (6) we get qiD = 8 −qiD /2. Hence, if
there is common knowledge of rationality, both firms would produce qiD = 16/3 =
5.3̄. The total quantity produced would be q D = 2qiD = 32/3. The corresponding
market price is given by the inverse demand function pD = P (q D ) = 14/3 = 4.6̄.
The profits of each firm would be (16/3) × (14/3 − 2) = 128/9. The total dead-
weight loss would be 64/9. See Figure 4.
Note that the Cournot equilibrium is in the middle ground between perfect
competition and a monopoly. This is in terms of price, quantity, firm profits,
and total market surplus. See table. The same is true about most models of
oligopolistic competition. In the problem set you are asked to show that, when
the number of firms in an oligopoly grows, the outcome gets closer to that of a
competitive market.

6
 3. Stackelberg leadership

The model of Cournot competition assumes that firms in the oligopoly make
their choices simultaneously and independently. That is not always the case. In
reality, firms choices are often asynchronous, and the dynamic structure of choices
can affect the equilibrium outcome.
Suppose for instance that firms make choices sequentially as follows. First,
firm 1 chooses how much to produce and firm 2 observes q2 . Afterwards, firm
2 chooses how much to produce. This dynamic structure is called Stackelberg
leadership. The first firm to choose is called the leader, and the second one the
follower. As we shall see, the outcome of this game is different from the Cournot
equilibrium.
The way to solve this dynamic game is via backwards induction. The follower
is rational, and observes q1 before choosing q2 . Hence, it will choose the quantity
that maximizes its profits given q1 , that is, it will choose to produce q2 = BR2 (q1 ).
Note that this optimal choice depends on the quantity chosen by the leader.
The leader knows that the follower is rational, and thus anticipates this de-
pendence. Instead of taking the choice of the follower as given, the leader treats
the choice of the follower as a function of its own. The relevant profit function for
the leader is thus
 
π1 = q1 P q1 + BR2 (q1 ) − C(q1 ) (7)

Substituting with (2), (3) and (6) yields

1 1 1
  
π1 = q1 10 − q1 + 8 − q1 − 2q1 = 4q1 − q12 . (8)
2 2 4
The optimal quantity chosen by the leader is given by the first-order condition:
4 − q1S /2 = 0. Consequently, the monopolist would produce q1S = 8. The follower
would produce q2S = BR2 (q1S ) = 4. The corresponding market price is given
by the inverse demand function pS = P (q1S + q2S ) = 4. The profits would be
π1S = 8 × (4 − 2) = 16 and π2S = 4 × (4 − 2) = 8.

Claim 2 The leader in a Stackelberg quantity duopoly makes higher profits than
in would in a Cournot duopoly with the same demand and cost functions.

7
 4. Collusion

In a Cournot duopoly, the total industry profits are less than in a monopoly.
See Table 1. If firms could agree to each produce half of the monopolistic quantity
each, they could each make half the monopolistic profit. Both firms would be
strictly better off than under the Cournot equilibrium. However, this agreement
would not be incentive compatible. If we ignore consumer welfare and analyze an
oligopoly exclusively from the perspective of the firms, the situation is a social
dilemma. When each firm tries to maximize their profits independently, they end
up in a situation in which each firm is making suboptimal profits.
We can capture this idea with a much simpler model in which firms can only
choose between a high and a low quantities. The corresponding profits are given
in Figure 5. If both firms choose a low quantity, the market price will be high
and the firms will share high profits. However, if a firm deviates by increasing its
quantity, it will enjoy higher profits resulting from high volume sales at a high
price. The game has the structure of a prisoners’ dilemma. Choosing H is strictly
dominated by choosing L, but the outcome (L,L) is Pareto dominated (from the
perspective of the firms) by the outcome (H,H).

H L

H 20, 20 5, 60

L 60, 5 10, 10

Figure 5 – Simplified duopoly model

Firms often try different mechanisms to coordinate their choices in order to

escape this social dilemma. Any attempt from part of the firms to do so is called
collusion. Collusion helps to increase the firms profits by bringing the market
outcome closer to the monopolistic outcome. This comes at the cost of dead-
weight loss. Once we take consumer welfare into account, collusion is inefficient
for society. Hence, there are anti-collusion laws in most countries in the world.
However, these laws are difficult to implement and firms often collude despite it
being illegal.
One possible collusion mechanism is to exploit the long-lasting nature of

8
oligopolistic competition. Most firms are long-lived and interact repeatedly over
a series of periods. Coke and Pepsi have competed with one another for over a
hundred years, since 1898, and will probably continue to do so for a very long time.
When agents interact repeatedly, they can use the promise of future reciprocity
to generate incentives for cooperation. Each firm might be willing to produce a
low quantity today in exchange for the promise that the other firm will continue
to produce low quantities in the future, thus securing high profits for both firms.

4.1. Discounted present value

In order to formally model that idea, we need to understand how firms value
streams of cash flows over time. Assume that firms have access to a financial
institution where they can invest money and received a constant risk-free interest
rate r > 0. If a firm invested x dollars at time t = 0, it would receive (1 + r)x
dollars the next period. If it reinvested both the initial investment and the first
period return, it would have (1 + r)2 x dollars at t = 2. Similarly, if it continue
to reinvest for t consecutive periods, it would receive (1 + r)t x. Hence, if a firm
wanted to have y dollars at period t, it would have to invest

x = δty (9)

dollars at period t = 0, where δ is the number given by δ := 1/(1 + r). The

number δ is called the discount factor. Equation (9) allows us to compare the
value of money in the future, with the value of money today.

Definition 1 The discounted present value of y dollars in period t is δ t y.

Note that the discount factor δ is always between 0 and 1. Suppose that we
want to compute the present value of receiving y on period t, and let us consider
two extreme cases. If δ = 1, then the present value would be equal to 1t y = y. In
that case, the value of money does not depend when it is received. The DM is
very patient, in that it does not mind waiting to receive a payment far ahead into
the future. If δ = 0, then the present value would be 0t y = 0 if t ≥ 1. In that
case, the DM doesnot value receiving money in the future. DM is ver y impatient
in that it only values money if it receives it right now. In general, the discount

9
factor can be interpreted as a measure of patience. High discount factors mean
that the DM values the future almost as much as the present, and low discount
factors means that the DM cares much more about the present.
Suppose that a firm expects to receive x0 dollars today, x1 dollars tomorrow,
x2 dollars the period after tomorrow, xt dollars on period t, and so on and so
forth. This is called a stream of cash flows. One way to assign a value to the
whole stream, is to compute the discounted present value of the cash flow in each
period, and add these values up.

Definition 2 The discounted present value of a stream of cash flows x1 , x2 , x3 , . . .

is given by
X
v= δ t xt . (10)
t

For example, suppose that a firm expects to receive x dollars each and every
period forever after. The discounted present value of this stream of cash flows
would be

v = x + δx + δ 2 x + δ 3 x + δ 4 x + . . . . (11)

Let us multiply both sides of equation (11) by the discount factor to get

δv = δx + δ 2 x + δ 3 x + δ 4 x + δ 5 x + . . . . (12)

We can subtract equation (12) from (11). When we do so, a lot of terms cancel
out on the right hand side and we are left with

✯0
✟
✟ x
v − δv = x + lim ✟t
✟δ x
t→∞
✟
⇒ v=
1−δ
. (13)

This is a useful formula to compute the present discounted value of a stream of

constant cash flows.

4.2. Grim-trigger strategies

Suppose that the firms agree to the following dynamic strategies

10
 • Both firms will produce the low quantity as long as no-one has deviated
from the agreement

• If someone ever deviates from the agreement at least once, then both firms
will proceed to produce high quantities forever after.

This kind of strategies are called grim trigger strategies. We know that, if the
firms played the game a single period, H is the dominant action for each firm and
this agreement wouldn’t work. What if firms where in a long-term relation?
Suppose that firm j expects its competitor to stick to the agreement. If firm
j were to also comply with the agreement, then both firms would make a profit
of 20 every period for ever after. The present discounted value of this stream of
profits is

20
v(comply) = . (14)
1−δ
Instead, suppose that firm j deviates at some period. That period, firm j
would make a profit of 60. However, in every subsequent period, j’s competitor
would choose H, which means that j’s profits would be at most 10. The present
discounted value from deviating is thus

v(deviate) = 60 + δ10 + δ 2 10 + δ 3 10 + . . .
  δ
= 60 + δ 10 + δ10 + δ 2 10 + . . . = 60 + 10. (15)
1−δ

The agreement is incentive compatible if the value from complying is no less

than the value from defecting. Using (14) and (15), the agreement is incentive
compatible if

20 δ
≥ 60 + 10 ⇔ 20 ≥ 60(1 − δ) + 10δ = 60 − 50δ
1−δ 1−δ
4
⇔ δ≥ . (16)
5
That is, only if the firms are patient enough so that the temptation of making
a short gain at the expense of the other firm does not trump over the value of
maintaining a profitable long run relation.

Ü///

11