Game Theory

Lecture 3 - Analysis of Static Games

Young Joon Park

HSBC Business School

Peking University

March/2023

Young Joon Park (PHBS) Game Theory March/2023 1 / 54

 Game Theory
Lecture 3 - Analysis of Static Games

Young Joon Park

HSBC Business School

Peking University

March/2023

Young Joon Park (PHBS) Game Theory March/2023 1 / 54

Prediction

So far, we’ve discussed how to describe games

Prediction of games follows naturally as the next topic
Prediction comes in many forms
I Equilibrium models and Non-equilibrium models

Young Joon Park (PHBS) Game Theory March/2023 2 / 54

Static Game

In static games (one-shot games) all of the players’ actions are taken
simultaneously and independently
Although Extensive form representation can describe static games,
normal form specification can be more intuitive in summarizing the
key features
Thus we mainly use the normal form in analysis of static games
Some of the key points of the lecture includes:
I Dominance and Iterated Dominance
I Best Responses
I Rationalizable Strategies
I Nash Equilibrium

Young Joon Park (PHBS) Game Theory March/2023 3 / 54

Dominance
Rational players act in order to maximize their expected payo↵s, given
their beliefs
Thus players should not choose a strategy if it is clear that the
strategy is “inferior” to another one
We often call these kinds of inferior strategies “Dominated” strategies
I Strict Dominance
I Weak Dominace
Rational player would not play (strictly) dominated strategies
Common Knowledge of Rationality provides a restriction of players’
belief by ruling out irrational strategies of other players
By eliminating the possibility of playing such strategies, we can better
predict the result of the games
Moreover, it turns out that we can repeat the procedure to narrow
down our prediction further

Young Joon Park (PHBS) Game Theory March/2023 4 / 54

Example of Dominance
Prisoners’ Dilemma

Player 2
C D
C 1, 1 10, 0
Player 1
D 0, 10 5, 5
Prisoners’ Dilemma

Strategy D gives Player 1 a strictly higher payo↵ regardless of what

Player 2 may do
Rationality: Player 1 rationally must choose to play D, regardless of
her beliefs
We say D (strictly) dominates C
Same for Player 2

Young Joon Park (PHBS) Game Theory March/2023 5 / 54

Strict Dominance

Definition
A pure strategy si 2 Si is (strictly) dominated if there exists a strategy
i 2 Si such that
ui ( i , s i ) > ui (si , s i )
for all s i 2S i

Note that we only care about pure strategies since our object is to
eliminate strategies which will be non-factors to the games

Young Joon Park (PHBS) Game Theory March/2023 6 / 54

Strict Dominance
Practice

Player 2
L C R
T 5, 8 2, 0 2, 3
Player 1 M 4, 2 0, 3 0, 1
B 1, 1 3, 6 1, 0

M is dominated by T for player 1

R is dominated by L for player 2

Young Joon Park (PHBS) Game Theory March/2023 7 / 54

Strict Dominance
by Mixed Strategy

Player 2 Player 2
L R L R
T 5, 1 0, 2 T 6, 1 0, 2
Player 1 M 0, 0 5, 0 Player 1 M 2, 0 5, 0
B 1, 3 1, 2 B 5, 3 1, 2
Example 1 Example 1

Example 1: B is dominated by the mixed strategy, 1 = ( 12 , 12 , 0)

Example 2: B is dominated by the mixed strategy, 1 = ( 31 9
40 , 40 , 0)
Note that there are more than one strategy that dominate strategy B
in these two examples

Young Joon Park (PHBS) Game Theory March/2023 8 / 54

Weak Dominance

Definition
A pure strategy si 2 Si is (weakly) dominated if there exists a strategy
i 2 Si such that
ui ( i , s i ) ui (si , s i )
for all s i 2S i and
ui ( i , s i ) > ui (si , s i )
for some s i 2S i

Young Joon Park (PHBS) Game Theory March/2023 9 / 54

Weakly Dominated Strategies

Player 2
L R
Player 1 T 1, 1 0, 0
B 0, 0 0, 0

B and R are called weakly dominated strategies as they are weakly

inferior to T and L, respectively.
Although they won’t get ruled out by the assumption of rationality, it
is reasonable to avoid them in real practice.

Young Joon Park (PHBS) Game Theory March/2023 10 / 54

Iterated (Strict) Dominance

If players have common knowledge of rationality, they will not stop

after one round of elimination of strictly dominated strategy - why?
Player 2
L C R
T 5, 1 0, 4 1, 0
Player 1 M 3, 1 0, 0 3, 5
B 3, 3 4, 4 2, 5
Algorithm of iterated dominance
1 Delete strictly dominated strategies for each player
2 Restrict attention to the resulting reduced game (Note this is possible
due to common knowledge of rationality)
3 If any strategies are strictly dominated in this game, go to step 1
4 End

Young Joon Park (PHBS) Game Theory March/2023 11 / 54

Notes on Iterated Dominance

The order of the iteration does not matter for strict dominance
The order, however, matters for the weak dominance case
I The order of elimination might change the result of the game
Although very useful in some games (dominance-solvable games),
iterated dominance is a limited tool to predict the result of games in
general
There are, in fact, a very limited number of games that can be solved
by applying iterated dominance
I Often, there are too many strategies left after applying iterated
dominance

Young Joon Park (PHBS) Game Theory March/2023 12 / 54

Best Response Correspondence

Best response correspondence indicates the strategies that maximize

expected payo↵, given my opponents’ strategies
For any particular profile of opponents’ strategies, I may have several
best responses
A correspondence is like a function, but it can take multiple values

Player 2
L C R
T 5, 8 2, 0 1, 3
Player 1 M 4, 2 1, 3 0, 2
B 1, 1 3, 6 1, 0

Young Joon Park (PHBS) Game Theory March/2023 13 / 54

Best Response

Definition
A strategy i 2 Si is a best response to the strategy profile i 2 S i
if
ui ( i , i) ui (si , i)

for all si 2 Si .

Note that there is no restriction in the belief

Also note that the comparison is made against the set of pure
strategies

Young Joon Park (PHBS) Game Theory March/2023 14 / 54

Some Motivational Stories for Game Theory
Fight against Terrorism

CAAC (Civil Aviation Administration of China) conducted thorough

historical research on terrorism attacks in the past few decades and
reported that 80% of such attacks (including attempted attacks) were
committed by male with age between 30 and 55.
Following this investigation, CAAC adjusted the field manual to
mainly focus on fortifying search practice for male passengers with
age between 30 and 55.

Young Joon Park (PHBS) Game Theory March/2023 15 / 54

Toy Experiment I
Overview

Experiments have become popular in economics setting recently

I’d like to run several experiments in class since
I It will help you deepen your understanding of the material by actually
put you into the (somewhat) more realistic situation
I I also found it useful to myself as I am curious of some of the results
I We will further use the opportunity to encourage more lively discussion
in class
The proper incentive is known to be vital in many experimental
settings

Young Joon Park (PHBS) Game Theory March/2023 16 / 54

Toy Experiment I
Guessing Game

Every member of the class is making a guess, simultaneously, between

limits 0 and 100.
The member whose guess is closest to 2/3 the class average guess
wins 200 RMB worth of Starbucks gift card, with ties broken
randomly.
What is your guess?
Write down your guess with your name and student id

Young Joon Park (PHBS) Game Theory March/2023 17 / 54

Some Motivational Stories for Game Theory
Kaynes’s Beauty Contest
“...professional investment may be likened to those newspaper
competitions in which the competitors have to pick out the six prettiest
faces from a hundred photographs, the prize being awarded to the
competitor whose choice most nearly corresponds to the average
preferences of the competitors as a whole; so that each competitor has to
pick, not those faces which he himself finds prettiest, but those which he
thinks likeliest to catch the fancy of the other competitors, all of whom are
looking at the problem from the same point of view. It is not a case of
choosing those which, to the best of one’s judgment, are really the
prettiest, nor even those which average opinion genuinely thinks the
prettiest. We have reached the third degree where we devote our
intelligences to anticipating what average opinion expects the average
opinion to be. And there are some, I believe, who practice the fourth, fifth
and higher degrees.”
-John Maynard Keynes, The General Theory of Employment, Interest, and
Money
Young Joon Park (PHBS) Game Theory March/2023 18 / 54
Guessing Game
Result 2022S

The guess varies from 0 to 66

The average guess is 24.48
The 2/3 of the average is 16.32
The winning guess is 15
Liu Yuchen
The past five winning guesses were: 15.63, 12, 18, 15.01, and 17

Young Joon Park (PHBS) Game Theory March/2023 21 / 54

(Pure Strategy) Nash Equilibrium

Definition
A strategy profile s ⇤ 2 S is a Pure Strategy Nash equilibrium if

ui (si⇤ , s ⇤ i ) ui (si , s ⇤ i )

for all si 2 Si and for all i = 1, · · · , n.

Young Joon Park (PHBS) Game Theory March/2023 22 / 54

Three ways of interpreting Nash Equilibrium

A strategy profile from which no player, unilaterally, can profitably

deviate
An intersection of best response correspondences
Mutual best responses with correct beliefs

Young Joon Park (PHBS) Game Theory March/2023 23 / 54

Remarks on Nash Equilibrium

Unlike with our earlier solution concepts of dominance, Nash

equilibrium applies to a profile of strategies rather than any
individual’s strategy.
I When people say “Nash equilibrium strategy”, what they mean is “a
strategy that is part of a Nash equilibrium profile.”
The term equilibrium is used because it connotes that if a player knew
that his opponents were playing the prescribed strategies, then she is
playing optimally by following her prescribed strategy.
In a sense, this is like a “rational expectations” equilibrium, in that in
a Nash equilibrium, a player’s beliefs about what his opponents will
do get confirmed (where the beliefs are precisely the opponents’
prescribed strategies).

Young Joon Park (PHBS) Game Theory March/2023 24 / 54

Remarks on Nash Equilibrium

Rationality only requires a player play optimally with respect to some

‘conjecture about the opponents’ play.
On the other hand, Nash requires that a player play optimally with
respect to what his opponents are actually playing. That is to say, the
conjecture she holds about her opponents’ play is correct.
The above point makes clear that Nash equilibrium is not simply a
consequence of (common knowledge of) rationality and the structure
of the game.

Young Joon Park (PHBS) Game Theory March/2023 25 / 54

Finding (Pure Strategy) Nash Equilibrium

Player 2 Employee
H D W S
H 5, 5 2, 2 M 0, 2 2, 0
Player 1 Manager
D 2, 2 0, 0 N 2, 0 0, 2
Hawk and Dove Monitoring Game

Player 2 Male
C D A B
C 1, 1 10, 0 A 2, 1 0, 0
Player 1 Female
D 0, 10 5, 5 B 0, 0 1, 2
Prisoners’ Dilemma Battle of the Sexes

Young Joon Park (PHBS) Game Theory March/2023 26 / 54

Finding (Pure Strategy) Nash Equilibrium in a Normal
Form Game

Player 2
F G H I J
A 1, 3 2, 0 0, 5 7, 5 4, 1
B 4, 2 0, 0 5, 9 2, 5 5, 4
Player 1 C 0, 4 0, 7 4, 8 4, 3 8, 4
D 5, 7 2, 5 1, 1 5, 5 3, 3
E 2, 4 1, 3 4, 3 3, 3 4, 6

Young Joon Park (PHBS) Game Theory March/2023 27 / 54

Finding (Pure Strategy) Nash Equilibrium in a Normal
Form Game

Player 2
F G H I J
A 1, 3 2, 0 0, 5 7, 5 4, 1
B 4, 2 0, 0 5, 9 2, 5 5, 4
Player 1 C 0, 4 0, 7 4, 8 4, 3 8, 4
D 5, 7 2, 5 1, 1 5, 5 3, 3
E 2, 4 1, 3 4, 3 3, 3 4, 6

Young Joon Park (PHBS) Game Theory March/2023 28 / 54

Finding (Pure Strategy) Nash Equilibrium in a Normal
Form Game

Player 2
F G H I J
A 1, 3 2, 0 0, 5 7, 5 4, 1
B 4, 2 0, 0 5, 9 2, 5 5, 4
Player 1 C 0, 4 0, 7 4, 8 4, 3 8, 4
D 5, 7 2, 5 1, 1 5, 5 3, 3
E 2, 4 1, 3 4, 3 3, 3 4, 6

Young Joon Park (PHBS) Game Theory March/2023 29 / 54

Finding (Pure Strategy) Nash Equilibrium in a Normal
Form Game

Player 2
F G H I J
A 1, 3 2, 0 0, 5 7, 5 4, 1
B 4, 2 0, 0 5, 9 2, 5 5, 4
Player 1 C 0, 4 0, 7 4, 8 4, 3 8, 4
D 5, 7 2, 5 1, 1 5, 5 3, 3
E 2, 4 1, 3 4, 3 3, 3 4, 6

Young Joon Park (PHBS) Game Theory March/2023 30 / 54

Problems of Nash Equilibrium

How to reach the equilibrium?

What if there are multiple equilibria?

Young Joon Park (PHBS) Game Theory March/2023 31 / 54

Problems of Nash Equilibrium

Initial Responses
I People do not usually play the NE strategy at the first attempt
I Especially when the circumstances is uncommon and/or the structure
is complicated
Learning
I There is evidence that players’ strategies converge to the equilibrium if
they play the same game repeatedly
I In the real world, people do not always play repeated games
Equilibrium selection
I When there are multiple equilibria, the explanation of which equilibrium
should be played is unsatisfactory
I Which equilibrium is reached in the long-run seems to depend on the
history of the initial results

Young Joon Park (PHBS) Game Theory March/2023 32 / 54

Nash Equilibrium in Infinite Games

When strategy set contains infinite value (typically real value) the
conventional way is not very useful.
Instead use the following:
1 First, write down the best response correspondences
F Use calculus if possible
F Otherwise use reasoning
2 Then, look for intersections of best response correspondences
F Eliminate strategies to which there is no well-defined best response
F A graph may help

Young Joon Park (PHBS) Game Theory March/2023 33 / 54

Application of Nash Equilibrium I
Cournot Duopoly

Two firms with zero costs

Each firm simultaneously chooses its quantity to supply, between 0
and 10
Q = q1 + q2
Demand curve: P = 10 Q

Young Joon Park (PHBS) Game Theory March/2023 34 / 54

Application fo Nash Equilibrium II
Bertrand Duopoly

Two firms with mc = 10

Each firm simultaneously chooses its price (any real number)
Whoever sets the lower price sells to the entire market at that price
I If they set the same price, each sells to half the market at that price
Consumers will buy 100 units at any price

Young Joon Park (PHBS) Game Theory March/2023 35 / 54

Application of Nash Equilibrium III
Cleaning the Neighborhood

There are n 1 neighbors

Each can spend from 0 to 2 hours cleaning the neighborhood
They decide simultaneously
The payo↵ for player i is just the sum of all the time spent cleaning by
all the players, divided by 2, and then minus the time spent by player i

Young Joon Park (PHBS) Game Theory March/2023 36 / 54

Application of Nash Equilibrium IV
Three-way Voting

Three players (1,2,3) simultaneously vote for three candidates (A,B,C)

Player 1 prefers A to B to C
Player 2 prefers B to C to A
Player 3 prefers C to A to B
Majority rules; ties are broken randomly
For simplicity, assume each player’s payo↵ is
I 5 if his/her best candidate wins
I 2 if his/her second best candidate wins
I 1 if his/her third best candidate wins

Young Joon Park (PHBS) Game Theory March/2023 37 / 54

Application of Nash Equilibrium V
Hi-Tech Joint Venture

Two hi-tech firms, i = 1, 2, consider a joint venture

Each firm simultaneously chooses an investment level xi 2 [0, 5]
The cost of investment xi is c(xi ) = 14 xi2
The revenue for each firm, given xi and x i , is
8
< 0 if xi < 1
Ri (xi , x i ) = 2 if xi 1 and x i <2
:
xi ⇤ x i if xi 1 and x i 2

Young Joon Park (PHBS) Game Theory March/2023 38 / 54

Mixed Strategy Nash Equilibrium

Find a Nash Equilibrium of the following game

Player 2
H T
H 1, 1 1, 1
Player 1
T 1, 1 1, 1
Matching Pennies
There is no (pure strategy) Nash equilibrium.
How should “equilibrium” look like if you actually play this game
We can check whether there is a “mixed strategy Nash equilibrium”

Young Joon Park (PHBS) Game Theory March/2023 39 / 54

Nash Equilibrium
Pure and Mixed

Definition
A strategy profile ⇤ 2 S is a Nash equilibrium if
⇤ ⇤ ⇤
ui ( i, i) ui (si , i)

for all si 2 Si and for all i = 1, · · · , n.

This is the general definition for Nash equilibrium that includes the
possible mixed strategy Nash equilibrium.

Young Joon Park (PHBS) Game Theory March/2023 40 / 54

Finding Strictly Mixed Strategy Nash Equilibrium
2X2 Game with 2 players

Choose Player 2’s mixture to make Player 1 indi↵erent between her

strategies
I This will make Player 1 willing to play any mixture
Choose Player 1’s mixture to make Player 2 indi↵erent
If there exist probabilities that solve this problem, you’ve found the
only strictly mixed equilibrium
If there is no strictly mixed equilibrium, you won’t be able to find the
probabilities that solve the problem

Young Joon Park (PHBS) Game Theory March/2023 41 / 54

Finding MSNE
2X2 Game with 2 players

Player 2 Employee
H D W S
H 5, 5 2, 2 M 0, 2 2, 0
Player 1 Manager
D 2, 2 0, 0 N 2, 0 0, 2
Hawk and Dove Monitoring Game

Player 2 Male
C D A B
C 1, 1 10, 0 A 2, 1 0, 0
Player 1 Female
D 0, 10 5, 5 B 0, 0 1, 2
Prisoners’ Dilemma Battle of the Sexes

Young Joon Park (PHBS) Game Theory March/2023 42 / 54

Finding MSNE
Larger Games

Player 2
F G H I J
A 1, 3 2, 0 0, 5 7, 5 4, 1
B 4, 2 0, 0 5, 9 2, 5 5, 4
Player 1 C 0, 4 0, 7 4, 8 4, 3 8, 4
D 5, 7 2, 5 1, 1 5, 5 3, 3
E 2, 4 1, 3 4, 3 3, 3 4, 6

Young Joon Park (PHBS) Game Theory March/2023 43 / 54

Facts about Nash Equilibrium

Every strategy played with positive probability in equilibrium must

survive the iterated dominance.
In a finite game, there is always at least one Nash equilibrium
(existence)
In a finite symmetric game, there is always at least one symmetric
Nash equilibrium (it may be mixed)
(Indi↵erence condition) If si is played with positive probability in a
NE, i⇤ , then
⇤
1 ui (si , i) = ui (si0 , ⇤
i) for all si , si0 2 Si⇤ ;
⇤
2 ui (si , i) ui (si0 , ⇤ ⇤ 0
i ) for all si 2 Si and si 2 Si .

Young Joon Park (PHBS) Game Theory March/2023 44 / 54

Finding MSNE in Large Games

Use iterated dominance first

Find all pure strategy Nash equilibria
If none, there must be a MSNE (finite games)
To find all equilibria, you must consider every combination of
strategies that survived the iterated dominance
If one player has only 2 strategies that survived the iterated
dominance, use a graph of the other player’s payo↵s
Guess & check also works well in games with symmetry

Young Joon Park (PHBS) Game Theory March/2023 45 / 54

Finding MSNE in Large Games
Graph Method

Player 2
L R
T 5, 1 1, 4
Player 1 M 4, 2 3, 3
B 3, 3 4, 0

Young Joon Park (PHBS) Game Theory March/2023 46 / 54

Finding MSNE in Large Games
Guess and Check

Player 2
R P S
R 0, 0 1, 1 1, 1
Player 1 P 1, 1 0, 0 1, 1
S 1, 1 1, 1 0, 0

Young Joon Park (PHBS) Game Theory March/2023 47 / 54

Notes on MSNE

When we first introduced mixed strategies, I suggested that the

easiest way to think about them was as though players were rolling
dice to determine which pure strategy to use.
While this is pedagogically true, some people find it a little
uncomfortable to think that agents are actually choosing their (pure)
strategies though an act of explicit randomization.
This may be particularly discomforting given that we have already
seen that in any Nash equilibrium, each player is indi↵erent over the
set of pure strategies that he is mixing over!
So why would a player then randomize, instead of just picking one
with certainty? Of course, if he does pick one with certainty, this
would in general destroy the indi↵erence of the other players over the
strategies they are randomizing over, and break the equilibrium
altogether.

Young Joon Park (PHBS) Game Theory March/2023 48 / 54

Notes on MSNE

One response is to say that the player is indi↵erent, hence is happy to

randomize. But this won’t sit well with you if you don’t like the idea
that players randomize in practice.
Fortunately, it turns out that we do not need players to be actually
randomizing in a MSNE. All that matters is that as far as other
players are concerned, player i’s choice seem like a randomized choice.
That is, what matters is the uncertainty that other players have about
i’s strategy.

Young Joon Park (PHBS) Game Theory March/2023 49 / 54

Notes on MSNE
To give an example, consider an NBA basketball game where team A has
possession, is down by 2 points, and there is only time for one more play.
Team A is in a time-out, and has to decide whether to go for a 2-pointer to
tie the game, or a 3-pointer to win. Team B is obviously deciding whether
to focus its defense against a 2-point shot or a 3-point shot.
This is basically a generalized game of Matching Pennies, version II: team A
wants to mismatch; team B wants to match. It may be that Team A’s
coach has a deterministic way of deciding whether to go for the win or the
tie — for example, he uses his star shooter’s morning practice 3-point
accuracy as the critical factor.
So long as Team B did not observe the shooter’s morning practice accuracy,
it is a randomized choice as far as they are concerned. Hence, B’s belief
about A’s play is a non-degenerate one, even though A may actually be
playing a pure strategy based on on some private information not available
to B.
That is, when we talk about A’s mixed strategy, we are really talking about
B’s beliefs about A’s strategy.
Young Joon Park (PHBS) Game Theory March/2023 50 / 54
Note on MSNE

Male Male
A B A B
A 2, 1 0, 0 A 3, 1 0, 0
Female Female
B 0, 0 1, 2 B 0, 0 1, 2
Battle of the Sexes I Battle of the Sexes II

Note that the only di↵erence between two games is uF (A, A)

Player F’s MSNE strategy does not change
Player M’s MSNE strategy change in the direction that decreases the
probability of playing (A, A)

Young Joon Park (PHBS) Game Theory March/2023 51 / 54

Existence of Nash Equilibrium (technical)

Theorem (Kakutani’s Fixed Point Theorem)

Suppose that X ⇢ RN is a non-empty, compact, convex set, and that
f : X ) X is a non-empty and convex-valued correspondence with a
closed graph. Then there exists x ⇤ 2 X such that x ⇤ 2 f (x ⇤ ).

Understand X as a set of strategy profiles, S (or ⌃) and f as best

response correspondence BR : S ) S (or BR : ⌃ ) ⌃).
The existence of s 2 S (or 2 ⌃) such that s 2 BR(s) (or
2 BR( )) means there is a strategy profile (pure or mixed) that is
BR to all other strategies for all i.

Young Joon Park (PHBS) Game Theory March/2023 52 / 54

Existence of Nash Equilibrium (technical)

Theorem (Existence of PSNE)

Suppose each Si ⇢ RN is compact and convex (and non-empty); and each
ui : S ! R is continuous in s and quasi-concave in si . Then there exists a
PSNE.

The continuity and quasi-concavity of payo↵ function ensures the BR

correspondence to satisfy the conditions for Kakutani’s FPT.

Young Joon Park (PHBS) Game Theory March/2023 53 / 54

Existence of Nash Equilibrium (technical)

Theorem (Existence of NE)

Every finite game has a Nash equilibrium (possibly in mixed strategies).

The critical need to allow for mixed strategies is that in finite games,
the pure strategy space is not convex, but allowing players to mix
over their pure strategies “convexifies” the space.

Young Joon Park (PHBS) Game Theory March/2023 54 / 54