ECON5190 Games and Strategic Behavior

Lecture 1
Elements of a Game, Thinking Strategically,
Sequential-move Games

Wooyoung Lim

The Hong Kong University of Science and Technology

Spring 2024
 “Human beings are social creatures.” -Aristotle-
Social?
▶ Many people (at least two)

▶ Interaction
 “Human beings are social creatures.” -Aristotle-
Social?
▶ Many people (at least two)

▶ Interaction
We will conceive game theory broadly as the theory of interactive
behavior.
Examples of Games

1. From economics: internet auctions

2. From Macroeconomics: notion of money
3. From biology: mate selection
4. From sports: bike racing
5. From politics: voting
6. From sociology: marriage behavior
7. From physics: interacting particles
8. From parlor games: poker
9. Recent phenomena: why do people bulk-buy toilet papers and
surgical masks?
Decisions vs. Games: Pick a number

▶ One number between 1 and 100 will be randomly selected.

▶ Your goal is to guess the number.

▶ I will give you five guesses, and after each wrong guess, I will
also tell you if you are too high or too low.

▶ If you guess correctly on the first try, you will get HK$25. On
the second try, the payment is down to HK$20, HK$15 for the
third guess, HK$10 for the fourth guess, then HK$5 for the
fifth try. If it takes more than five guesses, the game is over
and you get nothing.

▶ Are you ready to play?

Decisions vs. Games: Pick a number

▶ Microsoft CEO Steven Ballmer is said to have used this

“game” as a job interview question.

▶ What is the correct answer for him?

50, 25, 37, 42,....

▶ But, technically speaking, this is NOT a “game”!!!

Decisions vs. Games: Pick a number

▶ Microsoft CEO Steven Ballmer is said to have used this

“game” as a job interview question.

▶ What is the correct answer for him?

50, 25, 37, 42,....

▶ But, technically speaking, this is NOT a “game”!!!

Decisions vs. Games: Pick a number

▶ I will pick a number between 1 and 100.

▶ All other things remain the same.

▶ This is a game!!! Why?

Golden Ball

Golden Ball
A Language of Competition and Cooperation

“Game theory is a mathematical language for investigat-

ing the phenomena of competition and cooperation. These
phenomena arise at many levels. [...] A virtue of game the-
ory is its expressive power. It can describe, in a formal way,
a wide range of interactions among players: interactions
among players with fully opposed interests (purely compet-
itive situations); fully aligned interests (purely cooperative
situations); or interests that are neither fully opposed nor
fully aligned.”
- Adam Brandenburger
Three components of a game

1. The players participating in the game

2. The set of strategies available to the players of the game

3. The payoffs that the players receive as a function of all

players’ strategy choices.
Strategy

▶ The concept of a strategy is central to game theory.

▶ A strategy is a fully described behavioral disposition.

▶ A player’s strategy is a

Complete Contingent Plan

Why Complete Contingent Plan?

Think about yourself being

▶ A commender in the Allies who needs to pick the location for
the landing operation in 6 June, 1944. (The Normandy
landing plan began in 1943 which includes the famous
deception plans.)

▶ A president of a country who decides whether to join the Paris

agreement to reduce the CO2 emission.

▶ A CEO of a company working on merge and acquisition plans.

▶ Tim Cook making a decision about the next generation

iPhone.
Example 1: Guessing Game

Let’s play a simple game:

1. No talking

2. Choose a number in the interval [0,100] inclusive.

3. The average of all numbers will be calculated.

4. The person(s) whose number choice is closest to 2/3 times

the average of all numbers chosen will be declared the winner.
Guessing Game

▶ How many players?

▶ What is the set of strategies for each player?

▶ What are the payoffs?

Simultaneous vs. Sequential Move Games

▶ Simultaneous-move Games: Games where players choose

actions simultaneously.

▶ Sequential-move Games: Games where players choose actions

in a particular sequence.

▶ Is the guessing game a simultaneous-move game or not?

Example 2
Let’s play a simple game:
1. There are two players: Wooyoung and a
■ ⇓ Wooyoung
volunteer.
2. Wooyoung and a volunteer are facing
each other and there are 10 cells between
two players.
3. Wooyoung and the volunteer alternately
take turns moving their piece forward.
4. The player who starts moving his/her
piece is determined by a coin toss.
5. When it is a player’s turn, he/she is
allowed to move its piece either one or
□ ⇑ Volunteer
two cells forward, no more no less.
6. The player who is unable to move loses.
▶ This game is an example of the sequential-move game.

▶ A winning strategy for the first mover:

“After each round in which the other player moved his/her
piece two cells forward, move its piece one cell forward.
Otherwise, move its piece two cells forward.”

▶ Our goal is to find a fool-proof method for finding the

solution of this game and of similar games.
A Simple Case: 2 Cells

N
1 1
2 2
W V
■ ⇓ Wooyoung
1 2 1 2
V W
⎛1⎞ ⎛ −1⎞ □ ⇑ Volunteer
1 ⎜ ⎟ 1 ⎜ ⎟
⎝ −1⎠ ⎝1⎠
⎛ −1⎞ ⎛1⎞ ⎛ Wooyoung
W ff ⎞
' s payoff
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝1⎠ ⎝ −1⎠ ⎝ Volunteer ' s payoff ⎠
The way of representing a game via a game tree is known as the
extensive form.
A game tree consists of
1. An initial node (in our example the node marked with N).

2. Decision nodes (in our example indicated by solid circles).

3. Terminal nodes (in our example indicated by open circles).

4. An assignment of players (in our example N, W and V) to

decision nodes.

5. Branches that start at decision nodes and which indicate the

decisions available to the player acting at that node.

6. An assignment of payoffs to players at each terminal node.

Nature

The player N in our game tree is a special player, called Nature.

Nature chooses according to exogenously fixed probabilities (in our
example indicated by the labels “1/2” that are assigned to
nature’s two choices). In our example, the remaining (genuine)
choice nodes are labeled by the number of cells a player moves
with his/her piece.
Solving the game

1. A player who moves with only one cell left has no effective
choice and wins the game.

N
1 1
2 2
W V
1 2 1 2
V W
1  1
1   1  
 1 1
 1 1  Wooyoung ' s payoff 
     
1  1  Volunteer ' s payoff 
Solving the game
2. A player who moves with two cells left anticipates that if he
leaves the other player with one cell, the other player will win the
game.

N
1 1
2 2
W V
1 2 1 2
V W
1  1
1   1  
 1 1
 1 1  Wooyoung ' s payoff 
     
1  1  Volunteer ' s payoff 
Two Solution Principles

1. The principle that we applied in finding our solution is

anticipation.

2. We will later identify an additional principle, inference, that

will help us with solving more general games.
Steps for finding the solution

For now anticipation suffices:

1. For each node at which all choices lead to a terminal node, we
can identify optimal choices.

2. Fix these choices.

3. For each node at which all choices either lead to a terminal

node or to a node at which choices have been fixed, we can
identify optimal choices.

4. Repeat the last step, until all nodes are associated with
choices.
Backward Induction

The application of the anticipation principle historically had been

referred to as backward induction. This process of solving games
backwards is sometimes also called rollback.
A Game with 3 cells

A slightly more complicated game, Three cells, with the

simplification that we eliminate nature.

W
1 2
V V ■ ⇓ Wooyoung
1 2 1
W
 1  1
1     □ ⇑ Volunteer
1 1

1  Wooyoung ' s payoff 

   
 1  Volunteer ' s payoff 
Step 1 of the solution of the three-cell game:

W
1 2
V V
1 2 1
W
 1  1
1    
1 1

1  Wooyoung ' s payoff 

   
 1  Volunteer ' s payoff 
Step 2 of the solution of the three-cell game:

W
1 2
V V
1 2 1
W
 1  1
1    
1 1

1  Wooyoung ' s payoff 

   
 1  Volunteer ' s payoff 
Step 3 of the solution of the three-cell game:

W
1 2
V V
1 2 1
W
 1  1
1    
1 1

1  Wooyoung ' s payoff 

   
 1  Volunteer ' s payoff 
▶ Once the solution has been obtained, the outcome that is
induced by the solution can be found by following the path
induced by the solution from the initial node.

▶ For example, in the three-cell game, the second mover always

wins the game no matter what the first mover chose.
Exercise

■ ⇓ Wooyoung

4 cells?

□ ⇑ Volunteer
Order Advantages

▶ Similarly, in any games with number of cells that is multiple of

3, the second mover always wins the game: Second-mover
advantage

▶ However, in any other games, the first mover always wins the
game: First-mover advantage
Market Entry Game

▶ Consider the rivalry between Airbus and Boeing to develop a

new commercial jet aircraft. Suppose Boeing is ahead in the
development process and Airbus is considering whether to
enter the competition. If Airbus stays out, it earns zero profit,
whereas Boeing enjoys a monopoly and earns a profit of $1
billion. If Airbus decides to enter and develop the rival
airplane, then Boeing has to decide whether to accommodate
Airbus peaceably or to wage a price war. In the event of
peaceful competition, each firm will make a profit of $300
million. If there is a price war, each will lose $100 million
because the prices of airplanes will fall so low that neither firm
will be able to recoup its development costs.
Draw the tree for this game. Find the rollback equilibrium and
describe the firms’ equilibrium strategies.
Airbus vs. Boeing

Airbus

Enter Stay Out

Boing
 0 
Fight  
Accommodate 1000 

 300   100   Airbus ' s payoff 

     
 300   100   Boing ' s payoff 

Figure: Airbus vs. Boeing I

Airbus vs. Boeing

Airbus

Enter Stay Out

Boing
 0 
 
Accommodate Fight 1000 

 300   100   Airbus ' s payoff 

     
 300   100   Boing ' s payoff 
Airbus vs. Boeing

Airbus

Enter Stay Out

Boing
 0 
Fight  
Accommodate 1000 

 300   100   Airbus ' s payoff 

     
 300   100   Boing ' s payoff 
Airbus vs. Boeing II

It might be interesting to enrich our description of the rivalry

between Airbus and Boeing. If Boeing fights entry, then Airbus has
two options: Airbus can decide to hold out in order to eventually
acquire some market share, with a profit of $100 million for Airbus
and a profit of $200 million for Boeing. Airbus can decide to cave
in and leave the market with a loss of $100 million for Airbus and a
profit of $600 million for Boeing. The game might look as follows:
Solution of Airbus vs. Boeing II

Airbus

Enter Stay Out

Boing
 0 
Fight  
Accommodate 1000 
Airbus
 300 
   Airbus ' s payoff 
 300  Hold out Cave in  
 Boing ' s payoff 

 100   100 
   
 200   600 

Figure: Airbus vs. Boeing II

Solution of Airbus vs. Boeing II

Airbus

Enter Stay Out

Boing
 0 
Fight  
Accommodate 1000 
Airbus
 300 
   Airbus ' s payoff 
 300  Hold out Cave in  
 Boing ' s payoff 

 100   100 
   
 200   600 

Figure: Airbus vs. Boeing II

Solution of Airbus vs. Boeing II

Airbus

Enter Stay Out

Boing
 0 
Fight  
Accommodate 1000 
Airbus
 300 
   Airbus ' s payoff 
 300  Hold out Cave in  
 Boing ' s payoff 

 100   100 
   
 200   600 

Figure: Airbus vs. Boeing II

Solution of Airbus vs. Boeing II

Airbus

Enter Stay Out

Boing
 0 
Fight  
Accommodate 1000 
Airbus
 300 
   Airbus ' s payoff 
 300  Hold out Cave in  
 Boing ' s payoff 

 100   100 
   
 200   600 

Figure: Airbus vs. Boeing II

A few words about strategies in Airbus vs Boeing II.
1. Boeing has only two strategies in Airbus vs Boeing II:
“accommodate” and “fight.”

2. Airbus has a more interesting set of strategies. For example, it

can “enter” at its first node and “hold out” at its second
node. We will represent this strategy by the pair (enter, hold
out). An alternative strategy is (enter, cave in), i.e. to enter
at its first node and to leave once Boeing fights.

3. In total, Airbus has for strategies: (enter, hold out), (enter,

cave in), (stay out, hold out), and (stay out, cave in).
▶ Are there redundant strategies in our list?

▶ Once Airbus chooses to stay out, it does not matter how it

would behave had it entered.

▶ Therefore, can we ignore the distinction between (stay out,

hold out) and (stay out, cave in)?

▶ The answer is: NO. A strategy will always completely specify

behavior, even at contingencies that are ruled out by the
strategy itself.
Why do we distinguish between strategies that induce identical
behaviors?
1. The principal reason is that players reason about each
other: We want all players to be able to reason through all
the circumstances in which they may find themselves. Without
a complete specification of Airbus’s strategy, Boeing may be
unable to evaluate its own strategies. Even if Boeing expects
“stay out,” it wants to be prepared for the event “enter.”

2. At some point, we will want to examine how robust strategies

are to error. Then Airbus cannot ignore the possibility that it
chooses “enter” by mistake, even if it intended to choose
“stay out.”
Counting Strategies

Remember that a player’s strategy is a

Complete Contingent Plan

Counting Strategies: an exercise

Dixit, Skeath, and Reiley p.81

S2 In each of the following game, how many pure strategies
(complete plans of action) are available to each player? Give
me an example of the pure strategies for each player.
(c) A Scarecrow’s strategy: {N, (N|Nb, S|Sd)} i.e.
I will choose N in the first decision node and in the second
decision node,
1. I will choose N if my previous choice was N and Tinman’s
choice was b, and
2. I will choose S if my previous choice was S and Lion’s choice
was d.
▶ How many strategies does Scarecrow have? 2 × 2 × 2 = 8.
Counting Strategies: an exercise

Dixit, Skeath, and Reiley p.81

S2 In each of the following game, how many pure strategies
(complete plans of action) are available to each player? Give
me an example of the pure strategies for each player.
(c) A Scarecrow’s strategy: {N, (N|Nb, S|Sd)} i.e.
I will choose N in the first decision node and in the second
decision node,
1. I will choose N if my previous choice was N and Tinman’s
choice was b, and
2. I will choose S if my previous choice was S and Lion’s choice
was d.
▶ How many strategies does Scarecrow have? 2 × 2 × 2 = 8.
Counting Strategies: Two-move Chess

1. In chess the first mover can choose among 20 possible actions.

2. The second mover, regardless of the first mover¡¯s actions

choice, also has 20 available action choices.

3. Consider a version of chess that ends after each player has

made one move. To make this a well-specified game, we need
payoffs. For our purposes the exact specification of payoffs is
irrelevant.

Question: How many strategies does the second mover have in

two-move chess?
▶ To answer this question it is worth recalling that a strategy is
a complete contingent plan.

▶ So, how many possible contingencies does the second mover

face?

▶ The answer is: all the possible moves of the first mover. That
would be 20.

▶ Now recall that for each of those 20 contingencies the

second mover has 20 actions.
▶ In this example, we can think of a strategy as a list with 20
entries. Each entry itself can take all values from 1 to 20.
First-mover’s choice 1 2 3 4 5 ...... 17 18 19 20
Second-mover’s choice 7 5 1 2 6 ...... 4 6 17 20

▶ There are 20 × 20 × .......... × 20 = 2020 strategies.

| {z }
20 times
Exercies 1: A Simple Version of Two-move Chess

1. In chess the first mover can choose among 3 possible actions.

2. The second mover, regardless of the first mover¡¯s actions

choice, has 20 available action choices.

3. Consider a version of chess that ends after each player has

made one move. To make this a well-specified game, we need
payoffs. For our purposes the exact specification of payoffs is
irrelevant.

Question: How many strategies does the second mover have in

two-move chess?
Exercies 1: A Simple Version of Two-move Chess

1. In chess the first mover can choose among 3 possible actions.

2. The second mover, regardless of the first mover¡¯s actions

choice, has 20 available action choices.

3. Consider a version of chess that ends after each player has

made one move. To make this a well-specified game, we need
payoffs. For our purposes the exact specification of payoffs is
irrelevant.

Question: How many strategies does the second mover have in

two-move chess?
▶ To answer this question it is worth recalling that a strategy is
a complete contingent plan.

▶ So, how many possible contingencies does the second mover

face?

▶ The answer is: all the possible moves of the first mover. That
would be 3.

▶ Now recall that for each of those 3 contingencies the

second mover has 20 actions.
▶ To answer this question it is worth recalling that a strategy is
a complete contingent plan.

▶ So, how many possible contingencies does the second mover

face?

▶ The answer is: all the possible moves of the first mover. That
would be 3.

▶ Now recall that for each of those 3 contingencies the

second mover has 20 actions.
▶ To answer this question it is worth recalling that a strategy is
a complete contingent plan.

▶ So, how many possible contingencies does the second mover

face?

▶ The answer is: all the possible moves of the first mover. That
would be 3.

▶ Now recall that for each of those 3 contingencies the

second mover has 20 actions.
▶ In this example, we can think of a strategy as a list with 3
entries. Each entry itself can take all values from 1 to 20.
First-mover’s choice 1 2 3
Second-mover’s choice 1 1 1

▶ This strategy is different from the following strategy:

First-mover’s choice 1 2 3
Second-mover’s choice 2 1 1

▶ There are 20 × 20 × 20 = 203 strategies.

▶ In this example, we can think of a strategy as a list with 3
entries. Each entry itself can take all values from 1 to 20.
First-mover’s choice 1 2 3
Second-mover’s choice 1 1 1

▶ This strategy is different from the following strategy:

First-mover’s choice 1 2 3
Second-mover’s choice 2 1 1

▶ There are 20 × 20 × 20 = 203 strategies.

▶ In this example, we can think of a strategy as a list with 3
entries. Each entry itself can take all values from 1 to 20.
First-mover’s choice 1 2 3
Second-mover’s choice 1 1 1

▶ This strategy is different from the following strategy:

First-mover’s choice 1 2 3
Second-mover’s choice 2 1 1

▶ There are 20 × 20 × 20 = 203 strategies.

▶ In this example, we can think of a strategy as a list with 3
entries. Each entry itself can take all values from 1 to 20.
First-mover’s choice 1 2 3
Second-mover’s choice 1 1 1

▶ This strategy is different from the following strategy:

First-mover’s choice 1 2 3
Second-mover’s choice 2 1 1

▶ There are 20 × 20 × 20 = 203 strategies.