Lecture 2 Slides

Download as pdf or txt
Download as pdf or txt
You are on page 1of 57

LECTURE 2: NORMAL

FORM GAMES AND


DOMINANCE
DR VERA TE VELDE

1
LAST WEEK

• Decisions under certainty: completeness and transitivity mean that preferences can
be represented by utility numbers.
• Decision under uncertainty: we can select dominant options and eliminate dominated
ones
• Expected utility: we can compare options based on the average of the possibilities it
leads to, weighted by the probability of each possibility

2
FROM DECISIONS TO INTERACTIONS

• We have looked at decision under risk


• From now on we are going to look at strategic interactions
• But what we have seen will help us a lot

3
Go to www.menti.com and enter the code:

XXXX XXXX
2 questions

4
LET’S PLAY:
THE BRIEFCASE EXCHANGE

Carmen San Diego Crackle


5
PAYOFFS OF THE BRIEFCASE EXCHANGE

Crackle
Diamonds Nothing

Money 10 , 10 -5 , 20
Carmen
Nothing 20 , -5 0 ,0

Carmen’s payoffs Crackle’s payoffs

6
Go to www.menti.com and enter the code:

XXXX XXXX
2 questions

7
THE PRISONERS'
DILEMMA

This puzzle is one of the most


famous and important games in
economics, business, law, politics,
evolutionary science, social
psychology...

Silent Confess

Silent -1 , -1 -10 , 0

Confess 0 , -10 -5 , -5

8
PAYOFF MATRIX FOR PRISONERS' DILEMMA

Silent Confess

Silent -1 , -1 -10 , 0

Confess 0 , -10 -5 , -5

Diamonds Nothing

Money 10 , 10 -5 , 20

Nothing 20 , -5 0 ,0

9
PRISONERS’ DILEMMA IN REAL LIFE

The Prisoners' Dilemma occurs in many fields in the real world:


1. In climate change (cutting emissions)
2. In international politics (stockpile of nuclear weapons)
3. In business (advertising budget)
4. In law (should you hire a lawyer, or not?)
5. In sport (doping)
6. In fashion and dating (make-up, cosmetic surgery)
7. In your studies (group assignments)
8. In finance (bank runs)
10
PAYOFF MATRIX FOR TWO-PERSON GAMES:
GENERAL
• Payoff matrix is as easy as it can get, just remember the conventions:
1. Player 1 is always the “row player” (i.e., she chooses the row)
2. Player 2 is always the “column player” (i.e., he chooses the column)
3. The first payoff number belongs to Player 1, second to Player 2

• My own convention: Player 1 is a “she” and Player 2 a “he”

11
WHAT IF WE HAVE THREE PLAYERS?

• Three students are meeting to discuss their group project


• Each can choose (without knowing others' choices) whether to prepare for the
meeting or not
• The meeting will go well only if all of them have prepared
• If at least one of them comes unprepared, the meeting will be a waste of time
• Preparing, however, takes effort and is costly (regardless of how the meeting goes)
• Each prefers a good meeting, to coming unprepared to a bad meeting (“shirking” your
responsibilities), to coming prepared to a bad meeting:

Good meeting ≻ Bad meeting, no effort ≻ Bad meeting, effort


12
PAYOFF MATRICES FOR THREE-PERSON GAMES

Player 2 Player 2
Prepare Shirk Prepare Shirk

Prepare 3, 3, 3 0, 1, 0 Prepare 0, 0, 1 0, 0, 1
Player 1 Player 1
Shirk 1, 0, 0 1, 1, 0 Shirk 1, 0, 1 1, 1, 1

Prepare Shirk

If Player 1 chooses Shirk


‘Shirk’, verb: …and Player 2 chooses Prepare
To avoid or neglect a duty …and Player 3 chooses Shirk,
or responsibility. Player 3 chooses the …then Player 1 gets 1, Player 2
payoff matrix gets 0, Player 3 gets 1. 13
THREE-PERSON GAMES: CONVENTIONS

1. Player 1 chooses row


Player 2 chooses column
Player 3 chooses matrix

2. The first number in any cell is the payoff to player 1


The second number in any cell is the payoff to player 2
The third number in any cell is the payoff to player 3

14
NOT EVERY GAME CAN BE REPRESENTED BY A
PAYOFF MATRIX
But how about. . .
• Games with more than 3 players?
• Games with 2 or 3 players, but each player has many actions?
 Practically, more than 5 actions per player is bad enough
 Not to say there may be infinitely many actions (e.g., firm choosing level of output)

15
DEFINING A GAME

In general, if you would like to describe a game (i.e., a strategic interaction), all you need is to
specify 3 elements:

• Players Who are making decisions in this game?


• Strategies What can a player choose at each decision under all possible circumstances?
• Payoffs What is the preference (over outcomes) of each player?

16
DESCRIBING THE BRIEFCASE EXCHANGE GAME

• Players Carmen; Crackle

• Strategies Carmen: {Money , Nothing}; Crackle: {Diamonds , Nothing}

• Payoffs Corresponding payoff matrix:

17
DESCRIBING A GAME: PLAYERS

• A player is someone who can make a choice in a strategic interaction


• For this course they will usually be named Player 1, Player 2, and so on
• A generic player will usually be called Player i

18
DESCRIBING A GAME: ACTIONS

• Something that Player i can do is called an Action of Player i


 E.g., In Prisoners’ Dilemma, “Confess” is an action of each player
 So is “Silent”
 Generic actions of Player i are usually denoted as 𝑎𝑎𝑖𝑖 , 𝑎𝑎𝑖𝑖 ′, etc.

• The set of all actions that Player i can take is the action set of Player i
 E.g., In Prisoners' Dilemma, the action set of each player is

{Confess; Silent}
• Note: Action sets do not have to be the same across players

19
DESCRIBING A GAME: ACTION PROFILES

• Suppose each player chooses an action available to him/her


• Then I can write down a list of actions:
(action by player 1; action by player 2; … ; action by player N)
or
(𝑎𝑎1 , 𝑎𝑎2 , … , 𝑎𝑎𝑁𝑁 )
• Lists like this are known as Action Profiles
 E.g., In Prisoners’ Dilemma, there are 4 possible action profiles:
(Confess, Confess), (Confess, Silent)
(Silent, Confess), (Silent, Silent)
 Action profiles are typically denoted as 𝑎𝑎, 𝑎𝑎 ′ , etc.

20
DESCRIBING A GAME: PAYOFFS

• Each action profile is an Outcome of the game


• A Payoff Function of Player i is a (vNM) utility function over the set of outcomes
• E.g., in the Prisoners’ Dilemma, I have assigned the following utility:

21
A NOTE ON PAYOFF FUNCTIONS

• You may wish to use other numbers for the payoffs – which is fine
• In exercises or exams, if I do not give you the numbers for the payoffs, then you can
come up with any number as long as it respect the ordinal comparisons between
outcomes

22
NORMAL FORM GAME

• A Normal Form Game (also known as Strategic Form Game) consists of


1. A set of players
2. An action set for each player
3. A payoff function for each player

• If there are finitely many players, and each has a finite action set, the normal form
game is a finite game

23
EXAMPLE: SECOND-PRICE SEALED-BID AUCTION

• N bidders are bidding in an auction for one indivisible object


• Bidder i has a private value 𝑣𝑣𝑖𝑖 < 100 for the object
• Each of them can place a sealed bid (For simplicity, assume they can only bid in
whole dollars and cannot bid over 100)
• The highest bid wins the object (ties are broken uniformly randomly)
• Winner pays the second highest bid (i.e., the highest losing bid)
• Losers pay nothing and get nothing

24
SECOND-PRICE AUCTION AS A GAME

Players Bidder 1, Bidder 2, . . . , Bidder N


Actions Bidder i’s action set is
{0, 1, 2, …, 100}

25
SECOND-PRICE AUCTION AS A GAME: PAYOFFS

• An action profile in a Second-Price Auction is a list of bids


𝑏𝑏 = (𝑏𝑏1 , 𝑏𝑏2 , … , 𝑏𝑏𝑁𝑁 )
• where 𝑏𝑏𝑖𝑖 is the bid submitted by Bidder i
• Given a list of bids b, let p(b) be the second highest bid (i.e., the highest losing bid)
• Bidder i gets:

𝑣𝑣𝑖𝑖 − 𝑝𝑝 𝑏𝑏 if 𝑏𝑏𝑖𝑖 is the only highest bid


• 𝑢𝑢𝑖𝑖 𝑏𝑏 = �1 (𝑣𝑣𝑖𝑖 − 𝑝𝑝 𝑏𝑏 ) if 𝑏𝑏𝑖𝑖 ties with 𝑘𝑘 > 1 other bids
𝑘𝑘
0 otherwise

26
TWO-PERSON GAME VS. DECISION UNDER
UNCERTAINTY

• The two look pretty much the same, except that


1. There are 2 numbers in each cell for a game (We need to care about interactions)
2. Probabilities on the states of nature are known (and are strictly positive)

27
STRONGLY DOMINANT ACTION IN TWO-PERSON
GAMES

Definition
An action of player i is Strongly Dominant if and
only if,
for each and every action of the other player,
it is the unique best action among player i 's
actions.

• Note: A strongly dominant action has to be the best for each and every action of the
opponent – being the best for just one action of the opponent does not qualify.

28
Go to www.menti.com and enter the code:

XXXX XXXX
2 questions *

29
STRONGLY DOMINANT ACTION: TWO-PLAYER
EXAMPLE

L R
T 2, 2 7, 1
M 1, 0 0, 1
B 0, 1 3, 2

• Is there a strongly dominant action for Player 1?

• Is there a strongly dominant action for Player 2?

30
HOW ABOUT MORE THAN TWO PLAYERS?

• If there are more than two players, a player can't just consider the action of one
opponent
• But still s/he can ponder on the combinations of actions by his/her opponents
• For example, if Player 2 can choose between L and R and Player 3 can choose
between U and D, then to Player 1, there are 4 possible combinations of actions by
her opponents: Note that if there are 2 opponents with
2 actions each, there are exactly 22 = 4
(L,U), (L,D), (R,U), (R,D) opponent action profiles.
• 2 opponents, 3 actions each: 32 = 9.
• 3 opponents, 2 actions each: 23 = 8.

• These combinations are called Opponent Action Profiles

31
Go to www.menti.com and enter the code:

XXXX XXXX
2 questions *

32
STRONGLY DOMINANT ACTION: THREE-PLAYER
EXAMPLE
L R L R

T 6, 3, 6 5, 1, 5 T 5, 0, 5 4, 1, 4

B 5, 0, 5 4, 1, 4 B 4, 0, 4 3, 1, 3

U D

• Is there a strongly dominant action for Player 1?


 Yes, T

• Is there a strongly dominant action for Player 2?


• Is there a strongly dominant action for Player 3?



33
STRONGLY DOMINANT ACTION IN N-PERSON
GAMES

Definition
An action of player i is Strongly Dominant if and
only if,
for each and every opponent action profile,
it is the unique best action among player i 's
actions.

• Note: A strongly dominant action has to be the best for each and every opponent
action profile – being the best for just one opponent action profile does not qualify.

34
WEAKLY DOMINANT ACTION IN N-PERSON GAMES

Definition
An action of player i is Weakly Dominant if and only if:

1. For each and every opponent action profile, it is one


of the best actions to player i, and
2. For each and every other action of player i, there is an
opponent action profile against which it is strictly better.

• Note 1: A weakly dominant action has to be one of the best for each and every opponent
action profile – being the best for just one opponent action profile does not qualify
• Note 2: To satisfy requirement 2, we can use different opponent action profiles for
different alternative actions; that is, there need not be a single opponent action profile
against which the weakly dominant action is strictly better than every other action of
player i
35
Go to www.menti.com and enter the code:

XXXX XXXX
2 questions *

36
WEAKLY DOMINANT ACTION: EXAMPLE

L R

T 2, 2 3, 1

M 2, 0 0, 1

B 0, 2 3, 2

• Is there a weakly dominant action for Player 1?


• Is there a weakly dominant action for Player 2?



37
SOLVING A GAME USING DOMINANT ACTIONS

• If Player i has a (strongly or weakly) dominant action, she can never do strictly better
than playing that action, regardless of what her opponents are doing
• We would expect her to play her dominant action
• So would her opponents
• Each of her opponents can ask: what is the best action for me, given that i will play
her dominant action?
• Sometimes this method may narrow us down to a single action profile

38
SOLVING A GAME USING DOMINANT ACTIONS:
EXAMPLE
L R

T 2, 2 7, 1

M 1, 0 0, 1

B 0, 0 3, 2

• T is strictly dominant for Player 1


• Given that Player 1 plays T, L is the best option for Player 2
• We would expect (T, L) to be the outcome
39
STRONGLY DOMINATED ACTION IN N-PERSON
GAMES
Definition
An action of player i, 𝑎𝑎𝑖𝑖 , is Strongly Dominated by
another action 𝑎𝑎𝑖𝑖′ if and only if:
𝑎𝑎𝑖𝑖′ is strictly better than 𝑎𝑎𝑖𝑖 against each and every
opponent action profile.

• Note 1: The dominating action 𝑎𝑎𝑖𝑖 ′ has to be strictly better than 𝑎𝑎𝑖𝑖 against each and
every opponent action profile – being better for just one opponent action profile does
not qualify
• Note 2: It is important to fix the dominating action 𝑎𝑎𝑖𝑖 ′ – you cannot use different
dominating actions for different opponent action profiles
40
STRONGLY DOMINATED ACTION: EXAMPLE

L R

T 3, 1 0, 0

M 0, 0 1, 3

B 2, 2 2, 2

• Does Player 1 have any strongly dominated action? If yes, which action(s) dominate(s) it/them?
 Yes. M is strongly dominated by B

• Does Player 2 have any strongly dominated action? If yes, which action(s) dominate(s) it/them?
 No. Player 2 does not have any strongly dominated action

41
WEAKLY DOMINATED ACTION IN N-PERSON
GAMES
Definition
An action of player i, 𝑎𝑎𝑖𝑖 , is Weakly Dominated by another
action 𝑎𝑎𝑖𝑖′ if and only if:

1. 𝑎𝑎𝑖𝑖′ is no worse than 𝑎𝑎𝑖𝑖 against each and every


opponent action profile, and
2. there is at least one opponent action profile against
which 𝑎𝑎𝑖𝑖′ is strictly better than 𝑎𝑎𝑖𝑖 .

• Note 1: The dominating action 𝑎𝑎𝑖𝑖 ′ has to be no worse than 𝑎𝑎𝑖𝑖 against each and every
opponent action profile – being better against just one opponent action profile does not
qualify
• Note 2: It is important to fix the dominating action 𝑎𝑎𝑖𝑖 ′ – you cannot use different
dominating actions for different opponent action profiles
42
Go to www.menti.com and enter the code:

XXXX XXXX
2 questions *

43
WEAKLY DOMINATED ACTION: EXAMPLE

L R

T 3, 1 0, 0

M 0, 0 2, 3

B 2, 2 2, 2

• Does Player 1 have any weakly dominated action? If yes, which action(s) dominate(s) it/them?

• Does Player 2 have any strongly dominated action? If yes, which action(s) dominate(s) it/them?

44
BUT DECISIONS AND GAMES ARE DIFFERENT. . .

• All this ‘dominates’, ‘is dominated’, etc - didn't we already do this last lecture?
• Not quite! Nature does not react to the decision-maker’s choice
• But opponents in a game do!
• In particular, if my opponent thinks that I am going to avoid a dominated action, they
can respond accordingly

45
ITERATIVE ELIMINATION OF STRONGLY
DOMINATED STRATEGIES
L C R

T 3, 2 2, 1 0, 0

M 1, 2 1, 1 2, 0

B 0, 0 0, 2 0, 3

• B is strongly dominated by M
• We (and Player 2) would not expect Player 1 to play B. Let’s eliminate it.

46
ITERATIVE ELIMINATION OF STRONGLY
DOMINATED STRATEGIES
L C R

T 3, 2 2, 1 0, 0

M 1, 2 1, 1 2, 0

• Now C and R are strongly dominated by L. Let's eliminate them.

47
ITERATIVE ELIMINATION OF STRONGLY
DOMINATED STRATEGIES
L

T 3, 2

M 1, 2

• Now Player 1 is sure that Player 2 will play L. Her best action is then T. Outcome is
(T, L).
• This method is known as Iterative Elimination of Strongly Dominated Strategies

48
ORDER INDEPENDENCE OF STRONG
ELIMINATION
L C R
T 3, 2 2, 1 0, 0
M 1, 2 1, 1 2, 0

L C L R
• The order of eliminating strongly
T 3, 2 2, 1 T 3, 2 0, 0
dominated strategies does not
M 1, 2 1, 1 M 1, 2 2, 0
matter.

L
T 3, 2
M 1, 2
49
ITERATIVE ELIMINATION OF WEAKLY DOMINATED
STRATEGIES
• One may also conduct iterative elimination of weakly dominated strategies
• Unlike strongly dominated strategies, however, the order of elimination can affect
the outcome predicted.

50
ORDER DEPENDENCE OF WEAK ELIMINATION

L R L R
T 3, 3 2, 3 T 3, 3 2, 3

M 1, 0 0, 1 M 1, 0 0, 1

B 0, 1 1, 0 B 0, 1 1, 0

• Eliminate M (strongly dominated by T) • Eliminate B (strongly dominated by T)

51
ORDER DEPENDENCE OF WEAK ELIMINATION

L R L R
T 3, 3 2, 3 T 3, 3 2, 3

B 0, 1 1, 0 M 1, 0 0, 1

• Eliminate M (strongly dominated by T) • Eliminate B (strongly dominated by T)


• Now R is weakly dominated by L • Now L is weakly dominated by R

52
ORDER DEPENDENCE OF WEAK ELIMINATION

L R
T 3, 3 T 2, 3

B 0, 1 M 0, 1

• Eliminate M (strongly dominated by T) • Eliminate B (strongly dominated by T)


• Now R is weakly dominated by L • Now L is weakly dominated by R
• Next, eliminate B • Next, eliminate M

53
ORDER DEPENDENCE OF WEAK ELIMINATION

L R
T 3, 3 T 2, 3

• Eliminate M (strongly dominated by T) • Eliminate B (strongly dominated by T)


• Now R is weakly dominated by L • Now L is weakly dominated by R
• Next, eliminate B • Next, eliminate M
• Outcome is (T, L) with payoff (3, 3) • Outcome is (T, R) with payoff (2, 3)

54
WHAT’S WRONG?

L R
T 3, 0 1, 1

B 3, 1 0, 0

• There is another difference between Nature and an opponent:


 Nature puts strictly positive probability on each state
 An opponent may never play a certain action
• Against a particular opponent action profile, a weakly dominated action may be one of
the best actions (above: B is one of the best choice against L)
• If Player 1 is very sure that Player 2 will play L, choosing B is not a bad idea
• Elimination of a weakly dominated action is less tenable in games than in decisions
55
WHAT CAN WE DO?

• Since the result of iterative elimination of weakly dominated strategies depends on


the order of elimination, in most applications we can either:
1. Use only iterative elimination of strictly dominated strategies;

or
2. Do full elimination – that is, eliminate all weakly dominated strategies for all players at
each round of elimination

56
SUMMARY

1. Representation of Strategic Interaction: Normal Form Game


 Two-Person Game: Payoff Matrix
 Three-Person Game: Payoff Matrices
 General Game: Players, Actions and Payoffs

2. Dominance in Games
 Strongly and Weakly Dominant Action
 Strongly and Weakly Dominated Action
 Iterative Elimination of Dominated Actions

57

You might also like