Game Theory

Reshad Ahsan
Lecture 22

Introductory Microeconomics

Lecture 22 (ECON10004) 1 / 23
Learning objectives

Today’s learning objectives are to

Examine the profit-maximising behaviour of firms in an oligopoly with
quantity competition
Learn about interactions where players’ decisions have cross effects
Examine how to derive an equilibrium in games where decisions are
interdependent.

Lecture 22 (ECON10004) 2 / 23
Cournot quantity competition

We will consider a Cournot model with two firms producing a

homogeneous product with full information on the market demand
curve.
Each firm must decide how much to produce at the same time. Thus,
this is a simultaneous game.
When making its production decision, each firm takes its competitor’s
production decision into account.
This is important because the competitor’s production decision will
affect the market price.

Lecture 22 (ECON10004) 3 / 23
Firm 1’ production problem

Suppose the inverse demand for a product is

P(QT ) = 100 − Q1 − Q2 , (1)

where Q1 and Q2 are Firm 1 and Firm 2’s output, respectively, and
QT = Q1 + Q2 is the total industry output.
If Firm 2 increases its output, the market price will fall, and adversely
affect Firm 1.
Firm 1 seeks to maximise: π1 = P(QT )Q1 − TC (Q1 ).
Profit maximisation (dπ1 /dQ1 = 0) allows us to derive Firm 1’s
reaction function.1

1
Deriving Cournot reaction functions is easier as you can use the MR = MC equilibrium
condition. I have skipped this step here, but you are encouraged the derive the reaction function
in the next slide from Firm 1’s profit maximisation problem with MC = 25. Next week’s tutorial
will give you another opportunity to practice this.
Lecture 22 (ECON10004) 4 / 23
Firm 1’ reaction function

Suppose the strategic interaction in the previous slide results in the

following reaction function for Firm 1:
75 Q2
Q1 = − . (2)
2 2

Q1

This tells us Firm 1’s profit-

maximizing output, Q1 , in
response to United’s output, Q2 .
Notice that the optimal Q1 is
declining in Q2 .
R1
Q2
Lecture 22 (ECON10004) 5 / 23
Firm 2’s reaction function

Similarly, suppose Firm 2’s reaction function is:

75 Q1
Q2 = − . (3)
2 2

Q1
This tells us Firm 2’s profit-
maximizing output in response
R2
to Firm 1’s output.
Notice that the optimal Q2 is
also declining in Q1 .
To plot the above, we can
rearrange it to write Q1 =
75 − 2Q2 . R1
Q2

Lecture 22 (ECON10004) 6 / 23
Cournot equilibrium

In the Cournot equilibrium (point A below), each firm’s output is

found at the intersection of the two reaction functions.
Q1
What if Firm 1 decides to
unilaterally move to point B?
R2
Both firms will adjust until we
are back at point A.
In a Cournot equilibrium, neither
B C firm finds it profitable to change
Q1∗
A its output once it discovers its
R1 rivals’ output choice.
Q2∗ Q2

We can solve for the Cournot equilibrium by substituting one reaction

function into the other, as in the Bertrand case.

Lecture 22 (ECON10004) 7 / 23
Outline

Introduction to game theory

Nash equilibrium
Dominant strategy
Using iterated deletion of dominated strategies

Lecture 22 (ECON10004) 8 / 23
Political advertising

Australia’s political parties spent a total of $418 million in political

advertising before the 2022 federal election.
International evidence suggests that the effectiveness of political
advertisements is, at best, mixed.2
So, why spend $418 million on advertising without a clear return?
Political parties are trying to gain an advantage with such advertising.
But this gets nullified if all major parties spend somewhat equally.
Thus, we are potentially stuck in a bad equilibrium where both major
parties spend large sums of money on ineffective advertising.

2
In the U.S. context, political advertising is more effective because it influences turnout, an issue that is less important in
Australia.
Lecture 22 (ECON10004) 9 / 23
Game theory

The previous slide describes a situation where political parties are

stuck in a prisoner’s dilemma.
Other examples of these situations are
International trade agreements
Environmental agreements
Competition vs. collusion.
We will analyse these issues using tools from game theory, which is
the study of decision-making in strategic situations.
In these situations, each player’s decision has cross effects on other
players, which they take into account.

Lecture 22 (ECON10004) 10 / 23
Important concepts

Player: A decision-maker in a strategic situation.

Action: A single decision in the game.
Strategy: A complete set of actions in response to the other players’
actions in every contingency.
If you write down your strategy, someone else can play just like you
would, no matter what other players do.
Payoff: Utility/benefit a player gets from an outcome of the game.

Lecture 22 (ECON10004) 11 / 23
Nash equilibrium

The interdependence of players’ actions in game theory requires a

new equilibrium concept.

Nash equilibrium
A situation in which players each
choose their best strategy given the
strategies that all other players have
chosen.

Once at a Nash equilibrium,

neither player has an incentive
to deviate.

Lecture 22 (ECON10004) 12 / 23
A payoff matrix

Consider the payoff matrix (or game table) below that lists each
player’s payoffs from various advertising scenarios.

Player 2
(column player)
Advertise Don’t advertise
Advertise (250, 250) (750, 0)
Player 1 (row player)
Don’t advertise (0, 750) (500, 500)

Important: The first number in the parentheses is Player 1’s payoff

while the second number is Player 2’s payoff.

Lecture 22 (ECON10004) 13 / 23
Considering Nash equilibrium candidates

Player 2
Advertise Don’t advertise
Advertise (250, 250) (750, 0)
Player 1
Don’t advertise (0, 750) (500, 500)

Consider the (Don’t advertise, Don’t advertise) cell. Is this a Nash

equilibrium?
No, as Player 1 can do better by advertising.
Player 2 can also do better by advertising.

Lecture 22 (ECON10004) 14 / 23
Considering other Nash equilibrium candidates

Player 2
Advertise Don’t advertise
Advertise (250, 250) (750, 0)
Player 1
Don’t advertise (0, 750) (500, 500)

Suppose Player 1 deviates first to (Advertise, Don’t advertise). What

is Player 2’s best response? Advertise.
Similarly, suppose Player 2 deviates first to (Don’t advertise,
Advertise). What is Player 1’s best response? Also, advertise!

Lecture 22 (ECON10004) 15 / 23
Nash equilibrium

Player 2
Advertise Don’t advertise
Advertise (250, 250) (750, 0)
Player 1
Don’t advertise (0, 750) (500, 500)

Thus, the only cell where neither player has an incentive to deviate is
(Advertise, Advertise). This is the Nash equilibrium.
This is an equilibrium with a lower combined and individual payoff
than (Don’t advertise, Don’t advertise).
This is an illustration of a prisoner’s dilemma, where cooperation is
mutually beneficial but difficult to maintain.

Lecture 22 (ECON10004) 16 / 23
Worked example

Lecture 22 (ECON10004) 17 / 23
Worked example: solution

Lecture 22 (ECON10004) 18 / 23
Outline

Introduction to game theory

Nash equilibrium
Dominant strategy
Using iterated deletion of dominated strategies

Lecture 22 (ECON10004) 19 / 23
Dominant strategy

The previous solution method we used is known as cell-by-cell

inspection.
This is a time-consuming approach, particularly for more complicated
games, e.g. a 3 × 3 payoff matrix.
A more efficient solution method involves isolating dominant
strategies.
A dominant strategy provides a player with a better outcome
regardless of the other players’ actions.
If a dominant strategy exists, then we can drop the other
(“dominated”) strategies.

Lecture 22 (ECON10004) 20 / 23
Iterated deletion of dominated strategies in a 3 × 3 payoff
matrix

Player 2
Low Med. High
Low (1, 2) (0, 1) (2, 0)
Player 1 Med. (2, 3) (1, 2) (3, 1)
High (4, 4) (5, 5) (6, 6)

We will delete dominated strategies using a step-by-step or iterated

process.
For Player 1, High dominates both Low and Medium. Thus, High is
Player 1’s dominant strategy, and it will not choose Low or Medium.

Lecture 22 (ECON10004) 21 / 23
Nash equilibrium in a 3 × 3 payoff matrix

Player 2
Low Med. High
Low (1, 2) (0, 1) (2, 0)
Player 1 Med. (2, 3) (1, 2) (3, 1)
High (4, 4) (5, 5) (6, 6)

If Player 1 chooses High, then Player 2 knows that High is also its
dominant strategy.
It follows that (High, High) is the unique Nash equilibrium.

Lecture 22 (ECON10004) 22 / 23
Additional materials

This week’s optional exercise asks you to consider a game where there
is more than one Nash equilibrium.
Recommendations for further reading (optional)
An analysis of political donations before 2022 federal elections.

Lecture 22 (ECON10004) 23 / 23