Game Theory

Lecture 2 Basics of Game Theory

Young Joon Park

HSBC Business School

Peking University

March/2023

Young Joon Park (PHBS) Game Theory March/2023 1 / 53

 Game Theory
Lecture 2 Basics of Game Theory

Young Joon Park

HSBC Business School

Peking University

March/2023

Young Joon Park (PHBS) Game Theory March/2023 1 / 53

Basic Lessons of Game Theory

Define Game Theory

Get used to the language of Game Theory
Transform a story to a game (and also from a game to a story)
Learn some Important definitions:
I Players
I Strategies
I Payo↵ functions
I Static Games and Normal Form Representation
I Dynamic Games and Extensive Form Representation
I Belief and Rationality
I Common Knowledge
I Efficiency

Young Joon Park (PHBS) Game Theory March/2023 2 / 53

Example 1
Matching Pennies I

Amy and Bob are playing a game called “matching pennies I”

Amy chooses one side of her coin (either H or T)
Bob observes Amy’s choice and chooses one side of his coin (again
either H or T)
If Amy and Bob choose the same side, Bob gets Amy’s coin
If Amy and Bob choose a di↵erent side, Amy gets Bob’s coin
Whoever wins the coin becomes happier by 1
The one who loses the coin becomes happier by -1

Young Joon Park (PHBS) Game Theory March/2023 3 / 53

Describing a Game

Participants
The actions they can choose
What they know when they choose
How actions lead to outcomes
Preferences over outcomes
How they form beliefs about each other
How they make decisions

Young Joon Park (PHBS) Game Theory March/2023 4 / 53

Strategic Situations

Participants can a↵ect each other through their actions

I Perfectly Competitive market vs. Oligopoly market
The participants know this
I Thus they choose their actions accordingly
The participants’ problem: how to choose actions, considering what
actions others are likely to choose

Young Joon Park (PHBS) Game Theory March/2023 5 / 53

What is Game Theory?

A language for discussing strategic situations

An abstract framework for modeling strategic situations
A tool to make predictions about outcomes in strategic situations

Young Joon Park (PHBS) Game Theory March/2023 6 / 53

What is Game Theory?
Started as a sub-field of Mathematics but developed into a vital
methodology in Economics, Business, Political Science, Biology, Law,
and others
As people (or animals and plants) interact with each other regularly,
the tool to analyze these situations is critical (compared to the
deterministic case)
Some Examples:
I Microeconomics: oligopoly, pricing strategy, most competitive
economic interactions
I Corporate finance: internal structure of firms, incentive schemes, M&A
Strategies
I Macroeconomics: monetary and fiscal policy
I Political economy: voting, legislative bargaining
I Sports: roster management, tactics
I Computer science: resource allocation, network routing
I Biology: evolution

Young Joon Park (PHBS) Game Theory March/2023 7 / 53

Using Game Theory

Four steps of using game theory

1 Specify a strategic situation from a real-world example
2 Transform the strategic situation into a game theoretic framework
3 Solve the game
4 Apply the solution back to the actual problem

Young Joon Park (PHBS) Game Theory March/2023 8 / 53

Describing a Game

Participants
I “players”
The actions they can choose
I “strategies”
What they know when they choose
I “information sets” and “history”
How actions lead to outcomes
I “structure”
Preferences over outcomes
I “payo↵s”
How they form beliefs about each other
I “beliefs - common knowledge”
How they make decisions
I The main goal of learning game theory

Young Joon Park (PHBS) Game Theory March/2023 9 / 53

Game Form Representation
Extensive Form vs. Normal Form

There are two ways to describe a game - Normal Form and Extensive
Form
Normal Form
I Normal form representation contains essential information of the game
I Best used to describe simultaneous move games
Extensive Form
I Extensive form representation includes more information of the game
including the order of the players’ move
I Sometimes it is not very efficient to describe the game in the extensive
form (too much information)
I Most suitable to describe sequential move games

Young Joon Park (PHBS) Game Theory March/2023 10 / 53

Extensive Form Representation
Let’s draw a game tree - Matching Pennies I

Amy and Bob are playing a game called “matching pennies I”

Amy chooses one side of her coin (either H or T)
Bob observes Amy’s choice and chooses one side of his coin (again
either H or T)
If Amy and Bob choose the same side, Bob gets Amy’s coin
If Amy and Bob choose a di↵erent side, Amy gets Bob’s coin
Whoever wins the coin becomes happier by 1
The one who loses the coin becomes happier by -1

Young Joon Park (PHBS) Game Theory March/2023 11 / 53

Extensive Form Representation

(-1,1)
h
Bob

H
t
(1,-1)
Amy

h’ (1,-1)
T
Bob
t’ (-1,1)

Young Joon Park (PHBS) Game Theory March/2023 12 / 53

Extensive Form Representation
Let’s draw a game tree - Matching Pennies II

Amy and Bob are playing a game called “matching pennies II”
Amy chooses one side of her coin (either H or T)
Without observing Amy’s choice, Bob chooses one side of his coin
(again either H or T)
If Amy and Bob choose the same side, Bob gets Amy’s coin
If Amy and Bob choose a di↵erent side, Amy gets Bob’s coin
Whoever wins the coin becomes happier by 1
The one who loses the coin becomes happier by -1

Young Joon Park (PHBS) Game Theory March/2023 13 / 53

Extensive Form Representation

(-1,1)
h

H
t
(1,-1)
Amy Bob

h (1,-1)
T

t (-1,1)

Young Joon Park (PHBS) Game Theory March/2023 14 / 53

Game Tree Elements

Nodes: states of the world/histories of actions

Special nodes:
I initial node: the start of the game
I terminal nodes: outcomes
Branches: actions
Information sets: decision points and available information when the
decision is made
Basic Rules for nodes:
I One initial node
I Each non-terminal node is associated with a player
I Each branch is associated with an action
I Each terminal node is associated with a vector of payo↵s
I Each non-initial node has a unique immediate predecessor node

Young Joon Park (PHBS) Game Theory March/2023 15 / 53

Game Tree Elements

Basic Rules for Information sets:

I All the nodes in any particular information set:
1 are associated with the same player
2 are followed by the same number of branches, which are associated
with the same actions
I Assumption: Perfect recall
Labeling convention
I Label information sets by their players
I Label branches by their actions
I Label terminal nodes by their payo↵s

Young Joon Park (PHBS) Game Theory March/2023 16 / 53

Extensive Form Representation
Matching Pennies III

Amy and Bob are playing a game called “matching pennies III”
Amy and Bob choose one side of their coins (either H or T)
simultaneously
If Amy and Bob choose the same side, Bob gets Amy’s coin
If Amy and Bob choose a di↵erent side, Amy gets Bob’s coin
Whoever wins the coin becomes happier by 1
The one who loses the coin becomes happier by -1

Young Joon Park (PHBS) Game Theory March/2023 17 / 53

Extensive Form Representation

(-1,1)
h

H
t
(1,-1)
Amy Bob

h (1,-1)
T

t (-1,1)

Young Joon Park (PHBS) Game Theory March/2023 18 / 53

Extensive Form Representation
Basic Entry I

Entrant decides whether to enter, at cost c

If Entrant does not enter:
I Incumbent chooses a quantity, Q1 , to supply
If Entrant enters:
I Incumbent and Entrant simultaneously choose their quantities, Q2 and
Q3 , to supply, respectively
Demand curve: Price = 10 - Total Quantity
Zero production costs

Young Joon Park (PHBS) Game Theory March/2023 19 / 53

 Basic Entry
Extensive Form Representation
Basic Entry I

Young Joon Park (PHBS) Game Theory March/2023 20 / 53

Extensive Form Representation
3-player Coordination Game

3 players (Amy, Bob, and Chris) decide where to meet

Each player can choose to go to either “Window of the World” or
“Coastal City”
They cannot communicate where to meet
They like to hang out together
I If all 3 of them happen to choose the same spot, they all receive payo↵
of 3
I If 2 of them choose the same spot, those who can meet receive a payo↵
of 2, but the one who is alone receives a payo↵ of 0 for being lonely

Young Joon Park (PHBS) Game Theory March/2023 21 / 53

Main Features of the “Game”

Game Theory is the language to describe strategic situations

Definition for some terms should be vital
We can define those main features of the game

Young Joon Park (PHBS) Game Theory March/2023 22 / 53

Player

Players are the participants of the game

Each player has strategies that (in general) alter the outcome of the
game
For each outcome (at the terminal node), each player has payo↵
assigned to him or her
i signifies a certain player
i signifies all players except player i

Young Joon Park (PHBS) Game Theory March/2023 23 / 53

Strategy

Strategy is the most important concept in the theory of games

It denotes the collection of available actions players can choose in the
game
A strategy is a complete contingent plan for a player
in the game
Complete contingent plan means a full specification of a player’s
behavior, which describes the actions the player will take at each of
her possible information sets
A player’s strategy describes what she will do at each of her
information sets even if the information sets are not reached as the
result of the game

Young Joon Park (PHBS) Game Theory March/2023 24 / 53

 TheContingent
Example for Complete Grab GamePlan
The Grab Game

Young Joon Park (PHBS) Game Theory March/2023 25 / 53

Why Complete Contingent Plan?

Each player has beliefs on the other player

To grab $1, Player 1 should believe that Player 2 would grab $2
Player 2 has beliefs about Player 1’s strategy
To grab $2, Player 2 should believe that Player 1 would grab $3
To choose rationally, Player 1 should develop beliefs about Player 2’s
beliefs about Player 1, and so on
In reality, a player may not require a complete contingent plan if she
can eliminate some part of the game by her own choices
But the analyst needs to examine her reasons for making choices and
needs to know her beliefs about what other players believe about her
Therefore, it is necessary to know players’ plan for every contingency

Young Joon Park (PHBS) Game Theory March/2023 26 / 53

Notation of (Pure) Strategy

Si is player i’s strategy space; i = 1, · · · , n

si signifies a strategy for player i; si 2 Si
s signifies a strategy profile; s = (s1 , · · · , sn )
S is the space of strategy profiles; s 2 S
S = S1 ⇥ S2 ⇥ · · · ⇥ Sn
s i signifies a strategy profile for players except i

Young Joon Park (PHBS) Game Theory March/2023 27 / 53

Payo↵ Functions

Players have preferences over outcomes of the game

Each strategy profile leads to a particular outcome
Define payo↵s as a function of strategy profiles
ui : S ! R is player i’s payo↵ function
Although the convention in game theory is to call the number
assigned to the preference the “payo↵” it is often safe to interpret
this as the utility of the player in many cases

Young Joon Park (PHBS) Game Theory March/2023 28 / 53

Matching Pennies I

(-1,1)
h
Bob

H
t
(1,-1)
Amy

h’ (1,-1)
T
Bob
t’ (-1,1)

Young Joon Park (PHBS) Game Theory March/2023 29 / 53

Matching Pennies II

(-1,1)
h

H
t
(1,-1)
Amy Bob

h (1,-1)
T

t (-1,1)

Young Joon Park (PHBS) Game Theory March/2023 30 / 53

 Basic Entry
Basic Entry I

Young Joon Park (PHBS) Game Theory March/2023 31 / 53

Normal Form Representation

Normal Form is another way to represent a game

The normal form representation of a game contains three kinds of
information:
I Players
I Their strategy spaces
I Their payo↵ functions
General Normal Form representation of a game;
< {1, · · · , n}, S, {u1 (·), · · · , un (·)} >

Young Joon Park (PHBS) Game Theory March/2023 32 / 53

Matching Pennies I

(-1,1)
h
Bob

H
t
(1,-1)
Amy

h’ (1,-1)
T
Bob
t’ (-1,1)

Young Joon Park (PHBS) Game Theory March/2023 33 / 53

Matching Pennies II

(-1,1)
h

H
t
(1,-1)
Amy Bob

h (1,-1)
T

t (-1,1)

Young Joon Park (PHBS) Game Theory March/2023 34 / 53

 Rock,
Rock, Paper, Scissors Paper, Scissors

Young Joon Park (PHBS) Game Theory March/2023 35 / 53

Normal Form Representation for RPS

Players: 1,2
Strategies: (S1 = {R, P, S}, S2 = {R, P, S})
Payo↵s:

U1 (R, R) = 0 U1 (R, P) = 1 U1 (R, S) = 1

U1 (P, R) = 1 U1 (P, P) = 0 U1 (P, S) = 1
U1 (S, R) = 1 U1 (S, P) = 1 U1 (S, S) = 0

U2 (R, R) = 0 U2 (R, P) = 1 U2 (R, S) = 1

U2 (P, R) = 1 U2 (P, P) = 0 U2 (P, S) = 1
U2 (S, R) = 1 U2 (S, P) = 1 U2 (S, S) = 0

Young Joon Park (PHBS) Game Theory March/2023 36 / 53

Normal Form (Matrix) Representation

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 37 / 53

Matching Pennies II

(-1,1)
h

H
t
(1,-1)
Amy Bob

h (1,-1)
T

t (-1,1)

Young Joon Park (PHBS) Game Theory March/2023 38 / 53

Normal Form Representation for Matching Pennies II

Bob
H T
H 1, 1 1, 1
Amy
T 1, 1 1, 1

Young Joon Park (PHBS) Game Theory March/2023 39 / 53

Matching Pennies I

(-1,1)
h
Bob

H
t
(1,-1)
Amy

h’ (1,-1)
T
Bob
t’ (-1,1)

Young Joon Park (PHBS) Game Theory March/2023 40 / 53

Normal Form Representation for Matching Pennies I

Bob
0 0
HH HT TH 0 TT 0
H 1, 1 1, 1 1, 1 1, 1
Amy
T 1, 1 1, 1 1, 1 1, 1

Young Joon Park (PHBS) Game Theory March/2023 41 / 53

Popular Normal Form Games

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 / 53

Rationality and Belief

Rationality: Players act to maximize their (expected) payo↵s subject

to their beliefs
Beliefs: What a player believes is the likelihood that the other players
will choose any particular strategies
Player 2
L R
T 1, 0 0, 1
Player 1
B 0, 1 1, 0

Young Joon Park (PHBS) Game Theory March/2023 43 / 53

Common Knowledge

Conventional assumption of the analysis of games is that players have

a shared understanding of the entire game
More specifically, we assume the game is common knowledge
between players
A particular fact F is said to be common knowledge between the
players if each player knows a particular fact, F, each player knows
that the others know F, each player knows that every other player
knows that each player knows F, and so on
It is as if before the game is played, players gather around at a table
and see the game
Also, in many cases, we assume common knowledge of rationality

Young Joon Park (PHBS) Game Theory March/2023 44 / 53

Notation for Beliefs

S i is the space of all possible probability distributions over the

strategies of players i
✓ i 2 S i is a belief for player i
✓ i (s i ) 2 [0, 1] is the probability that player i’s belief assigns to
strategy profile s i 2 S i
P
Properties of ✓ i : s i 2S i ✓ i (s i ) = 1

Young Joon Park (PHBS) Game Theory March/2023 45 / 53

Expected Utility

Payo↵s are in utility terms, not monetary terms

Therefore players
P want to maximize expected payo↵s
ui (si , ✓ i ) = s i 2S i ✓ i (s i )ui (si , s i )
Preferences over outcomes are una↵ected by adding a constant or
multiplying by a positive scalar

Young Joon Park (PHBS) Game Theory March/2023 46 / 53

Calculating Expected Utility

Suppose Player 1 believes that:

I Pr(D) = 1/2
I Pr(E ) = 1/3
I Pr(F ) = 1/6
What is the expected payo↵ of Player 1 for playing A, B, or C?

Player 2
D E F
A 8, 4 3, 2 2, 7
Player 1 B 5, 6 1, 0 0, 2
C 1, 1 3, 5 1, 4

Young Joon Park (PHBS) Game Theory March/2023 47 / 53

Mixed Strategy

Player i may randomize among her strategies

Usual assumption: players randomize independently
A “pure strategy” is a trivial mixed strategy: A particular strategy
played with probability 1

Young Joon Park (PHBS) Game Theory March/2023 48 / 53

Notation for Mixed Strategy

Si is the space of all possible probability distributions over of player

i’s strategies
i 2 Si is a mixed strategy for player i
i (si ) 2 [0, 1] is the probability that player i plays strategy si 2 Si
2 S1 ⇥ · · · ⇥ Sn is a mixed strategy profile
Independence: (s) = 1 (s1 ) ⇤ ··· ⇤ n (sn )

Young Joon Park (PHBS) Game Theory March/2023 49 / 53

Efficiency

Sometimes we will be interested in seeing whether the players, as a

group, can behave efficiently even though they are acting
non-cooperatively
Efficiency means that no player can be made better o↵ without
making any other player worse o↵
This is sometimes called “Pareto efficiency”

Young Joon Park (PHBS) Game Theory March/2023 50 / 53

Defining Efficiency

Definition
For s, s 0 2 S, s 0 is more efficient than s if ui (s 0 ) ui (s) for all i, with
uj (s 0 ) > uj (s) for some j

That is, all players weakly prefer s’ to s, with at least one player
strictly preferring s’

Definition
s 2 S is efficient if there does not exist any s 0 2 S that is more efficient
than s

Young Joon Park (PHBS) Game Theory March/2023 51 / 53

Notes on Efficiency

A game can have more than one efficient strategy profile

I Prisoners’ Dilemma: (C,C), (C,D), and (D,C) are efficient
Every efficient strategy profile maximizes a weighted sum of utilities,
for some positive weights
I (C,C) maximizes the sum with equal weights
I (D,C) maximizes the sum with almost all weight on Player 1

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

Young Joon Park (PHBS) Game Theory March/2023 52 / 53

Efficient Outcomes

Player 2 Employee
H T W S
H 1, 1 1, 1 M 0, 2 2, 0
Player 1 Manager
T 1, 1 1, 1 N 2, 0 0, 2
Matching Pennies Monitoring Game

Player 2 Male
H D A B
H 5, 5 2, 2 A 2, 1 0, 0
Player 1 Female
D 2, 2 0, 0 B 0, 0 1, 2
Hawk and Dove Battle of the Sexes

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