G5212 (Game Theory & Info Econ): Introduction

Guillaume Haeringer

Spring 2025

1/30
Why Game Theory?

Basic microeconomics (consumer, firm, competitive eq.) does not

consider interactive decision making:

the outcome for each person depends on their actions and the
actions of the other person

I People bidding for the same item on eBay

I People working together on a joint project

I Generals deciding where to position their armies

I Oligopolistic firms setting prices for similar products

2/30
Why Game Theory?

For interactive settings, optimization (like for consumer or the

firm) will not get us very far:
I Best bid of bidder 1 depends on bid of bidder 2
I Best bid of bidder 2 depends on bid of bidder 1

→ Need some way of solving both problems together

This (basically) is what game theory studies

Applied to: mating displays of birds, banking crises, spectrum

auctions, kidney exchanges, insurance, school choice, political
platforms, sport, war, and more

3/30
Why mathematical models?

Empiricism (descriptive models):

I May want to make predictions about the world
I Math can do heavy lifting, as for physicists

Advice (normative models):

I Modeling others’ behavior helps optimize yours
I Math can be a parsimonious language

Storytelling (rhetorical models):

I Model can highlight hypothetical cause and effect
I Math forces a story to be internally consistent

For all purposes: Forced to make implicit assumptions explicit

4/30
The Plan (For The Course)

Part 1: Game Theory (first half of course)

I Focus on tools
I Static games of complete information
I Dynamic games of complete information
I Games of incomplete information
I Solution concepts and refinements

I With a few applications along the way

I Bargaining
I Auctions
I Voting

5/30
The Plan (For The Course)
Part 2: Information Economics
Focus on applications in the face of asymmetric information
I Example 1: Signaling
I Companies want to recruit people of high ability
I But they cannot observe ability directly
I Can education be used by high ability candidates to signal that
they have high ability?
I Example 2: Moral Hazard
I A boss wants to encourage their worker to work hard
I But they cannot observe effort directly, only outcomes (which
have a random component)
I How should they design their incentive scheme?
I Example 3: Optimal selling mechanisms
I Seller makes a product, doesn’t know how much it’s worth to
prospective buyers
I How to design auction rules?
6/30
The Plan (for today)

I A gentle introduction!
I Talk through some ‘classic’ games
I Formal definition of a game
I Mixed strategies

7/30
Example: Matching Pennies

Bob
H T
H 1, −1 −1, 1
Alice
T −1, 1 1, −1

I Two players (Alice and Bob) each reveal a penny showing

heads or tails
I If the pennies match then Bob pays Alice a dollar
I If not, Alice pays Bob a dollar
I This is the matrix form of this game

8/30
Example: Prisoners’ Dilemma

Bob
Defect Cooperate
Defect −6, −6 0, −9
Alice
Cooperate −9, 0 −1, −1

I Classic story of two prisoners who must decide whether or not

to confess
I But many other (more economically interesting) applications:
e.g. pollution, traffic, pricing (collusion)

9/30
Example: partnership

I Effort E produces an output of 6 at a cost of 4, with output

shared equally;
I shirking S produces an output of 0 at a cost of 0.

Bob
S E

6
S 0 − 0, 0 − 0 2 − 0, 62 − 4
Alice

6
E 2 − 4, 62 − 0 2× 6
2 − 4, 2 × 6
2 −4

That’s also a Prisoner’s dilemma

10/30
Example: partnership

After some heavy and tedious calculations we get a simpler version

of the game:

Bob
S E
Alice S 0, 0 3, −1
E −1, 3 2, 2

11/30
Prisoner’s dilemma

The “general” form of the prisoner’s dilemma is given by

Bob
Defect Cooperate
Defect P, P T,S
Alice
Cooperate S, T R, R

where

T(emptation) > R(eward) > P(unishment) > S(ucker)

12/30
Defining a Game
A simultaneous-move game (a.k.a strategic form game, normal
form game) consists of 3 elements:

1. The players

2. The actions that each player can take

3. The payoffs associated with each set of actions

We are initially making some ‘hidden’ assumptions:

I Players move at the same time
I All payoffs are known to all players

Later in the course we will relax these assumptions, and so a

description of the game will also include
I The sequence of play
I Who knows what
13/30
Defining a Game
Definition
An n-player normal form (or strategic form) game G is an n-tuple
{(S1 , u1 ) , ..., (Sn , un )} , where for each i

(1) Si is a nonempty set, called i’s strategy space, and

(2) ui : nk=1 Sk → R is called i’s payoff function.

Q

Notation
I S := nk=1 Sk
Q

I s := (s1 , ..., sn ) ∈ S

I S−i := k6=i Sk
Q

I (si0 , s−i ) := (s1 , ..., si−1 , si0 , si+1 , ..., sn )

I u : S → Rn
14/30
What are we assuming about individuals?

Preferences are transitive

I The cornerstone of rationality

Preferences are complete

I Sometimes without loss (but not always!)

Preferences are quantifiable

I Essentially without loss

Players only care about....

I ... money? NOT assuming that!
I ... themselves? NOT assuming that!

15/30
Payoffs in games

A player’s payoff (function) in a game comes from her preferences

over outcomes.

Those preferences may depend on:

I The player’s ‘direct’ outcome (the object of money she gets)
I Other players’ outcomes (which may include subjective
elements).

So, a player
I only looks at her own payoff to evaluate an outcome
I BUT may want to look at the others’ payoffs to ‘guess’ what
they’ll do.

16/30
Example: Second-Price Sealed-Bid Auction

I A seller has one indivisible object.

I There are n bidders with respective valuations for the object

0 ≤ v1 ≤ · · · ≤ vn

(which are common knowledge).

I The bidders simultaneously submit bids.

I The highest bidder wins the object and pays the second
highest bid.
In the case of a tie all winning bidders are equally likely to
have their bid accepted.

17/30
Example: Second-Price Sealed-Bid Auction
Formally, the 2nd price auction is described as follows:

I Players: 1, ...., n

I Strategy set for each player i = 1, . . . , n:

Si = [0, ∞)

I Payoffs:
Given a profile of bids, s, the set of highest bidders is

W (s) ≡ {k : ∀j, sk ≥ sj }

Then we have

 vi − maxj6=i sj if si > maxj6=i sj
1
ui (si , s−i ) = (vi − si ) if si = maxj6=i sj
 |W (s)|
0 if si < maxj6=i sj .
18/30
Example: Cournot Duopoly

I There are two firms, call them 1 and 2, producing perfectly

substitutable products.
I The market demand is P(Q) = max {a − Q, 0}, Q = q1 + q2 .

I The cost of producing qi is given by C (qi ) = cqi , 0 < c < a.

I The two firms choose quantities simultaneously.

19/30
Example: Cournot Duopoly

Formally, we have
I Players: 1,2
I Strategies Si = [0, ∞).
I Payoffs
ui (q1 , q2 ) = [P (q1 + q2 ) − c] qi .

20/30
Example: Voting

I There are three players i = 1, 2, 3.

I There are two candidates a and b which they can vote for.

I The voting rule is the majority rule.

I Voters’ preferences are as follows (with preferred candidate

listed higher)
1 2 3
a b b
b a a

I A player receives a payoff of 1 if his favorite candidate wins; 0

else.

21/30
Example: Voting

I Players: 1,2,3

I Strategies Si = {a, b}

I Payoffs (for Player 1):

u1 (a, a, a) = 1 u1 (b, a, a) = 1
u1 (a, a, b) = 1 u1 (b, a, b) = 0
,
u1 (a, b, a) = 1 u1 (b, b, a) = 0
u1 (a, b, b) = 0 u1 (b, b, b) = 0

22/30
Mixed Strategies

Consider again the matching pennies game

Here is another action Bob could take: rather than put the coin
down H or T, he could flip it, and play whichever way the coin falls

This is a new strategy: it is not H or T, but a 50% chance of

H and a 50% chance of T

More generally, we might like to extend the player’s strategy space

to allow them to randomize between pure strategies

These are ‘Mixed Strategies’

They will be useful going forward. . .

23/30
Mixed Strategies

Definition
Suppose {(S1 , u1 ) , ..., (Sn , un )} is an n-player normal-form game.
A mixed strategy for player i is a probability distribution over
elements of Si , denoted by σi ∈ ∆ (Si ) .
Strategies in Si are called pure strategies.

Remarks
I In most cases, we assume Si is finite. Then σi : Si → [0, 1]
P
s.t. si ∈Si σi (si ) = 1.
I When Si is not countable, it’s more work/math to define
mixed strategies (mixed strategy sets become infinite
dimensional. . . )
→ We will not worry about this ;-)

24/30
Mixed Strategies
Qn
Extend ui to j=1 ∆ (Sj ) by taking expected values. If Si is finite:
X X
ui (σ1 , . . . , σn ) := ··· ui (s1 , ..., sn ) σ1 (s1 ) σ2 (s2 ) · · · σn (sn ) .
s1 ∈S1 sn ∈Sn

So, we’re simply using expected utility (i.e., we’re vNM utilities,
but does not mean we’re assuming risk neutrality).

Notation:
X Y
ui (si , σ−i ) := ui (si , s−i ) σj (sj )
j6=i
s−i ∈S−i
X
ui (σi , σ−i ) := ui (si , σ−i ) σi (si )
si ∈Si

25/30
Mixed Strategies

Games are usually described with pure strategy sets:

{(S1 , u1 ) , ..., (Sn , un )}

And then we say we’ll allow players to use mixed strategies:

{(∆(S1 ), u1 ) , ..., (∆(Sn ), un )}

26/30
Mixed Strategies

Do mixed strategies mean that all distributions over strategy

profiles can arise?

No, because we don’t allow for correlation:

Yn Y n 
∆ (Sj ) 6= ∆ Sj = ∆ (S)
j=1 j=1
| {z } | {z }
No correlation Correlation

We have the same for strategy profiles:

Y
(σ1 , ..., σi−1 , σi+1 , ..., σn ) = σ−i ∈ ∆ (Sj ) 6= ∆ (S−i )
j6=i

27/30
Correlated v. non-correlated strategies
Bob
L R
U α β
Alice
D γ δ

SA = {U, D} ⇒ ∆(SA ) is a probability distribution over {U, D}

SB = {L, R} ⇒ ∆(SB ) is a probability distribution over {L, R}.

So, we have

Yn
∆ (Si ) = ∆(SA ) × ∆(SB )
i=1
= a probability Alice plays U (and D)
AND a probability Bob plays L (and R)

⇒ Alice and Bob randomize independently (non-correlated). 28/30

Correlated v. non-correlated strategies

If instead we look at
Yn 
∆ Si = ∆(SA × SB )
i=1
= a probability Alice and Bob play (U, L),
and (U, R), and (D, L), and (D, R))

⇒ Alice and Bob randomize together (correlated).

29/30
Summary

Key things from today

I Understand what a game is

I Understand how to translate a story into a game

I Understand what mixed strategies are

I Understand how to translate a game in pure strategies into a

game with mixed strategies

30/30