ECO316: Applied Game Theory

Lecture 1: Strategic Games & Nash Equilibrium

Instructor: Stanton Hudja

University of Toronto

September 3, 2023

Based on materials by Martin J. Osborne & previous ECO316 instructors

1 / 32
Course Information

I Lectures: Monday 1:00-3:00 PM and Thursday 4:00-5:00 PM

MP 202
I Instructor Office Hours: Wed 9:00-11:00am, Max Gluskin
House 254
I Teaching Assistants (TA): Yi-Tsung Hsieh and Alex Ballyk
I TA Office Hours: Will be updated
I Optional Textbook: An Introduction to Game Theory by
Martin J. Osborne (Oxford University Press, New York, 2003),
ISBN-13 978-0195128956

2 / 32
Evaluation Scheme

I Problem sets (10%):

• 10 problem sets, each worth 1% of final grade, graded on
completion basis, submitted via Quercus
• Two highest problem sets replace two lowest problem sets
I Class Participation (10%):
• We will run classroom experiments for participation (average is
your grade)
• Your are allowed to miss two classroom experiments
I Midterm (35%): Later Date, 4-6pm
• Coverage: Weeks 1 - 6, inclusive
I Final exam (45%): 3-hour cumulative final exam, during final
assessment period

3 / 32
What is Game Theory?

A field of economics that provides mathematical/analytical tools to

model and analyze strategic interactions between decisionmakers.

I Used beyond economics as well (e.g., political science)

I This course covers theory and applications
I Typical game-theoretic analysis of an economic question
1. Formulate the question
2. Formulate a model that captures essence of situation
3. Analyze model using appropriate tools (typically look for an
“equilibrium”)
4. Extract insights about economic question

4 / 32
Some Topics We’ll Cover

I Firm competition (Lecture 2)

• Does having a larger number of firms in a market lead to
better outcomes for consumers?
I Electoral competition (Lecture 3)
• How do parties’ positions depend on voters’ preferences?
I Auctions (Lecture 9)
• What type of auction produces an efficient outcome?
Maximizes the auctioneer’s revenue?
I Signalling (Lecture 11)
• When can individuals credibly signal that they possess a
desirable trait?

5 / 32
Preference Relations

I A decisionmaker has preferences over a set of alternatives X

summarized by a preference relation % that compares pairs
of alternatives
I The preference relation % induces the strict relation  and
the indifference relation ∼

Preference notation English translation

x %y “x is at least as good as y ”
x y “x is better than y ”
x ∼y “x is as good as y ”

6 / 32
Rationality and Utility Functions

I A preference relation % is rational if it is complete, i.e.,

x % y or y % x or both, for all x and y in X

and transitive, i.e.,

x % y and y % z implies x % z

I A decisionmaker is rational if her preference relation is

rational and she behaves optimally with respect to her
preferences
I A utility function, which assigns a number u(x) to each
alternative, represents a rational preference relation % if

x % y if and only if u(x) ≥ u(y )

7 / 32
Ordinal Utility Functions

For now, utility functions are ordinal: only the ranking of

alternatives matters.

Stated otherwise:
I If a utility function u(x) represents a preference relation %,
then % is also represented by every increasing function of u(x)
I Reminder: A function f (a) is increasing if a > b implies
f (a) > f (b)

Example: If u(x) represents %, then log (u(x)) also represents %

8 / 32
Strategic Games

An action profile is a list of actions taken by each player.

I In a game where each of n players has a possible actions,
there are an possible action profiles.

A strategic game consists of

1. a set of players;
2. for each player, a set of actions;
3. for each player, preferences over the set of action profiles.

9 / 32
Example of a Strategic Game: The Prisoner’s Dilemma

I Two suspects in a major crime are held in separate cells.

There is enough evidence to convict each of them with a
minor offense, but not enough to convict either of them with
a major offense unless one of them acts as an informer against
the other.
I The suspects must (simultaneously and independently) decide
whether to Cooperate (remain silent) or Defect (“rat out” the
other suspect).
• If they both Cooperate, each will be convicted of the minor
offense and spend one year in prison.
• If one and only one of them Defects, she will be freed and used
as a witness against the other, who will spend four years in
prison.
• If they both Defect, each will spend three years in prison.

10 / 32
Example of a Strategic Game: The Prisoner’s Dilemma

Players: Suspect 1 and Suspect 2

Actions: {Cooperate, Defect}, for each Suspect

I For simplicity, denote Cooperate as “C” and Defect as “D”

Preferences:
I Suspect 1: (D, C )  (C , C )  (D, D)  (C , D)
I Suspect 2: (C , D)  (C , C )  (D, D)  (D, C )
where action profiles are in the form (Suspect 1’s action, Suspect
2’s action).

11 / 32
Example of a Strategic Game: The Prisoner’s Dilemma
I Any utility functions u1 and u2 that satisfy the following
inequalities represent Suspect 1 and 2’s preferences,
respectively:
u1 (D, C ) > u1 (C , C ) > u1 (D, D) > u1 (C , D)
u2 (C , D) > u2 (C , C ) > u2 (D, D) > u2 (D, C )

I For example, a Suspect’s utility from an action profile could

be the negative of the amount of years they spend in prison if
that action profile is played:
u1 (D, C ) = u2 (C , D) = 0
u1 (C , C ) = u2 (C , C ) = −1
u1 (D, D) = u2 (D, D) = −3
u1 (C , D) = u2 (D, C ) = −4

12 / 32
Example of a Strategic Game: The Prisoner’s Dilemma
Using these utility functions, we can represent the Prisoner’s
Dilemma as:
Suspect 1’s utility Suspect 2’s utility
Suspect 2 Suspect 2
C D C D
C −1 −4 C −1 0
Suspect 1 Suspect 1
D 0 −3 D −4 −3

Combining the above tables:

Suspect 2
C D
C −1, −1 −4, 0
Suspect 1
D 0, −4 −3, −3

13 / 32
Nash Equilibrium (NE)

An action profile is a Nash equilibrium of a strategic game if

every player’s action is optimal, given other players’ actions

or, equivalently, if

no player can change her action to make herself strictly better off,
given other players’ actions.

14 / 32
Finding Nash Equilibria: Method 1

Method 1 (Checking Action Profiles): Check every action profile

to see if a player can deviate from her action to make herself
strictly better off, holding other players’ actions constant.
I If a profitable deviation exists, the action profile is not a Nash
equilibrium.
I If a profitable deviation does not exist, the action profile is a
Nash equilibrium.

15 / 32
Example of Applying Method 1: The Prisoner’s Dilemma
Suspect 2
C D
C −1, −1 −4, 0
Suspect 1
D 0, −4 −3, −3

I (C , C ) is not a NE: given that one Suspect Cooperates, the

other Suspect can make herself strictly better off by Defecting.
I (C , D) is not a NE: given that Suspect 2 Defects, Suspect 1
can make herself strictly better off by Defecting.
I (D, C ) is not a NE: given that Suspect 1 Defects, Suspect 2
can make herself strictly better off by Defecting.
I (D, D) is a NE: given that one Suspect Defects, the other
Suspect cannot make herself better off by Cooperating.

The unique Nash equilibrium of the Prisoner’s Dilemma is

(Defect, Defect).
16 / 32
Pareto Efficiency

An action profile is Pareto efficient if

there is no alternative action profile in which every player is at

least as well off and at least one player is strictly better off

or, equivalently, if

no player can be made strictly better off without making at least

one player strictly worse off.

17 / 32
Example of Pareto Efficiency: The Prisoner’s Dilemma

Suspect 2
C D
C −1, −1 −4, 0
Suspect 1
D 0, −4 −3, −3

I (C , C ), (C , D), and (D, C ) are Pareto efficient: at least one

Suspect is strictly worse off in every other action profile,
relative to the action profile under consideration.
I (D, D) is not Pareto efficient: both Suspects are strictly
better off if they instead play (C , C ).

The unique Nash equilibrium of the Prisoner’s Dilemma is not

Pareto efficient.

18 / 32
Experiment 1

We are about to start Experiment #1.

19 / 32
Other Two-Player Games: Bach or Stravinsky?

I Daphne and Simon wish to go out together.

I Two concerts are available: one of music by Bach, and one of
music by Stravinsky.
I Daphne prefers Bach and Simon prefers Stravinsky. Both
players prefer attending a concert together to attending a
concert by themselves. If they go to different concerts, each
of them is equally unhappy listening to the music of either
composer.

20 / 32
Other Two-Player Games: Bach or Stravinsky?

Players: Daphne and Simon

Actions: {Bach, Stravinsky }, for each player

I Denote Bach as “B” and Stravinsky as “S”

Preferences:
I Daphne: (B, B)  (S, S)  (B, S) ∼ (S, B)
I Simon: (S, S)  (B, B)  (B, S) ∼ (S, B)

21 / 32
Other Two-Player Games: Bach or Stravinsky?

The strategic game “Bach or Stravinsky?” can be represented as

Simon
B S
B 2, 1 0, 0
Daphne
S 0, 0 1, 2

Nash equilibria: (B, B) and (S, S)

Pareto efficient action profiles: (B, B) and (S, S)

22 / 32
Other Two-Player Games: Stag Hunt

I Each of two hunters has two options: she may retain attentive
to the pursuit of a Stag, or she may catch a Hare.
I If both hunters pursue the Stag, they catch it and share it
equally; if any hunter devotes her energy to catching a Hare,
the stag escapes, and the hare belongs to the defecting hunter
alone. Each hunter prefers a share of the Stag to a Hare.

23 / 32
Other Two-Player Games: Stag Hunt

Players: Hunter 1 and Hunter 2

Actions: {Stag , Hare}, for each player

I Denote Stag as “S” and Hare as “H”

Preferences:
I Hunter 1: (S, S)  (H, S) ∼ (H, H)  (S, H)
I Hunter 2: (S, S)  (S, H) ∼ (H, H)  (H, S)

24 / 32
Other Two-Player Games: Stag Hunt

The strategic game “Stag Hunt” can be represented as

Hunter 2
S H
S 2, 2 0, 1
Hunter 1
H 1, 0 1, 1

Nash equilibria: (S, S) and (H, H)

Pareto efficient action profile: (S, S)

25 / 32
Other Two-Player Games: Matching Pennies

I Ted and Barney choose, simultaneously, whether to show the

head or tail of a coin.
I If they show the same side, Barney pays Ted a dollar; if they
show different sides, Ted pays Barney a dollar.
I Each person cares only about the amount of money he
receives.

26 / 32
Other Two-Player Games: Matching Pennies

Players: Ted and Barney

Actions: {Head, Tail}, for each player

I Denote Head as “H” and Tail as “T”

Preferences:
I Ted: (H, H) ∼ (T , T )  (H, T ) ∼ (T , H)
I Barney: (H, T ) ∼ (T , H)  (H, H) ∼ (T , T )

27 / 32
Other Two-Player Games: Matching Pennies

The strategic game “Matching Pennies” can be represented as

Barney
H T
H 1, −1 −1, 1
Ted
T −1, 1 1, −1

Nash equilibria: None (!)

Pareto efficient action profiles: All action profiles

28 / 32
Finding NEs in Games with > 2 Players

I In a game where each of n players has a possible actions,

there are an possible action profiles.
I For large a and/or n, applying Method 1 (checking every
single action profile) becomes difficult.
• For instance, a = 2, n = 10 ⇒ 210 = 1, 024 action profiles (!)
I In such games, it is helpful to group together similar action
profiles when applying Method 1.

29 / 32
Example of Grouping: Public Good Contribution

I Each of n people chooses whether to contribute a fixed

amount toward the provision of a public good.
I The good is provided if and only if at least k people
contribute, where 2 ≤ k ≤ n.
I Each person ranks outcomes from best to worst as:
1. Any outcome in which the good is provided and she does not
contribute
2. Any outcome in which the good is provided and she contributes
3. Any outcome in which the good is not provided and she does
not contribute
4. Any outcome in which the good is not provided and she
contributes

30 / 32
Example of Grouping: Public Good Contribution

Players: The n people

Actions: {Contribute, Don0 t Contribute}, for each player

I Denote Contribute as “C” and Don’t Contribute as “dC”

Preferences: For person 1 (others’ preferences are symmetric),

(dC , C , ..., C , dC , ..., dC )  (C , C , ..., C , dC , ..., dC )

| {z } | {z }
≥k ≥k−1
 (dC , C , ..., C , dC , ..., dC )
| {z }
<k
 (C , C , ..., C , dC , ..., dC )
| {z }
<k−1

31 / 32
Example of Grouping: Public Good Contribution
Preferences: (dC , C , ..., C , dC , ..., dC )  (C , C , ..., C , dC , ..., dC ) 
| {z } | {z }
≥k ≥k−1
(dC , C , ..., C , dC , ..., dC )  (C , C , ..., C , dC , ..., dC )
| {z } | {z }
<k <k−1

I > k people contributing: Not a NE. Any contributor can

profitably deviate to not contributing.
I k people contributing: NE. If any contributor stops
contributing, the good is not provided; if any non-contributor
switches to contributing, she is worse off.
I m people contributing, where 0 < m < k: Not a NE. Any
contributor can profitably deviate to not contributing.
I 0 people contributing: NE. If any person switches to
contributing, she is worse off.
The Nash equilibria are the set of action profiles where k people
contribute, plus the action profile where no one contributes.
32 / 32