Module – 1

Syllabus: Introduction to Strategic Games: What is game theory? The theory of rational choice,
Strategic games; Examples: The prisoner’s dilemma, Bach or Stravinsky, Matching pennies;
Nash equilibrium; Examples of Nash equilibrium; Best response functions; Dominated actions.

What Is Game Theory?

• GAME THEORY aims to help us understand situations in which decision-makers
interact.
• A game in the everyday sense—“a competitive activity in which players contend with
each other according to a set of rules”.

Applications Of Game Theory

• Firms competing for business.
• Political candidates competing for votes.
• Jury members deciding on a verdict.
• Animals fighting over prey.
• Bidders competing in an auction.
• Evolution of siblings’ behavior toward each other.
• Competing experts’ incentives to provide correct diagnoses.
• Legislators’ voting behavior under pressure from interest groups.
• The role of threats and punishment in long-term relationships.

History Of Game Theory

1920s – First Major Development

​ Emile Borel

​ John von Neumann

1944 – Decisive Publication

 ​ Theory of Games and Economic Behavior

​ Authors: von Neumann and Morgenstern

Early 1950s – John Nash

​ Introduction of Nash Equilibrium

​ Game-theoretic study of bargaining

1994 – Nobel Prize in Economic Sciences

​ John C. Harsanyi (1920–2000) → Bayesian games (Harsanyi doctrine)

​ John F. Nash (1928– ) → Nash equilibrium

​ Reinhard Selten (1930– ) → Bounded rationality and extensive games

Modeling Process
• Step 1: Selecting aspects of a given situation (that appear to be relevant) and
incorporating them into a model. This step is mostly an “art”.
• Step 2: Model analysis (using logic and mathematic).
• Step 3: Studying model’s implications to determine whether our ideas make sense.
This may point towards a revision of the model’s assumptions in order to better
capture “stylized facts”.

Theory Of Rational Choice

“the action chosen by a decision-maker is at least as good, according to her preferences, as

every other available action.”

· The theory of rational choice is a component of many models in game theory.

· Briefly, this theory is that a decision-maker chooses the best action according to her
preferences, among all the actions available to her.

· No qualitative restriction is placed on the decision-maker’s preferences.

· Her “rationality” lies in the consistency of her decisions when faced with different sets of
available actions, not in the nature of her likes and dislikes.

Actions

· Set A consisting of all actions that, under some circumstances, are available to the
decision-maker.

· In any given situation, the decision-maker knows the subset of available choices, and takes it
as given (the subset is not influenced by the decision-maker preferences).

· The set A could, for example, be the set of bundles of goods that the decision-maker can
possibly consume; given her income at any time, she is restricted to choose from the subset
of A containing the bundles she can afford.

Preferences and payoff functions

· We assume that the decision-maker, when presented with any pair of actions, knows which
of the pair she prefers

· We assume further that these preferences are consistent (if a > b and b > c, then a > c).

· Preferences representation: preferences can be represented by a payoff function:

· The payoff function associates a number with each action in such a way that actions with
higher numbers are preferred.

· More precisely:

· u(a) > u(b) if and only if the decision-maker prefers a to b

· (Economists often speak about utility function)

EXAMPLE 5.2 (Payoff function representing preferences) A person is faced with the choice of
three vacation packages, to Havana, Paris, and Venice. She prefers the package to Havana to the
other two, which she regards as equivalent. Her preferences between the three packages are
represented by any payoff function that assigns the same number to both Paris and Venice and a
higher number to Havana. For example, we can set u(Havana) = 1 and u(Paris) = u(Venice) = 0,
or u(Havana) = 10 and u(Paris) = u(Venice) = 1, or u(Havana) = 0 and u(Paris) = u(Venice) = −2.

Solution:

The person prefers the vacation package to Havana over the packages to Paris and Venice, and
she is indifferent between Paris and Venice.

To represent these preferences with a payoff function u:

●​ Havana must have a higher payoff than Paris and Venice.
●​ Paris and Venice must have the same payoff.

One possible payoff representation is:

u(Havana) = 1

u(Paris) = 0

u(Venice) = 0

This payoff function correctly represents the preferences because Havana has a higher payoff
than Paris and Venice, and Paris and Venice have equal payoffs, reflecting the person's
indifference between them.

EXERCISE 5.3 (Altruistic preferences) Person 1 cares both about her income and about person
2’s income. Precisely, the value she attaches to each unit of her own income is the same as the
value she attaches to any two units of person 2’s income. How do her preferences order the
outcomes (1, 4), (2, 1), and (3, 0), where the first component in each case is person 1’s income
and the second component is person 2’s income? Give a payoff function consistent with these
preferences.

Person 1 values her own income and also values person 2’s income, but each unit of her own
income is worth the same as two units of person 2’s income. This means one unit of person 2’s
income is worth half as much as one unit of her own income.

So her utility can be written as:​

u1(x1, x2) = x1 + (1/2)x2

Now evaluate each outcome:

 ●​ For (1, 4):​
u1 = 1 + (1/2)(4) = 1 + 2 = 3
●​ For (2, 1):​
u1 = 2 + (1/2)(1) = 2 + 0.5 = 2.5
●​ For (3, 0):​
u1 = 3 + (1/2)(0) = 3

Thus, the preference ordering is:​

(1, 4) ~ (3, 0) > (2, 1)

So person 1 is indifferent between (1, 4) and (3, 0), and both are preferred to (2, 1).

EXERCISE 6.1 (Alternative representations of preferences) A decision-maker’s preferences over

the set A = {a, b, c} are represented by the payoff function u for which u(a) = 0, u(b) = 1, and
u(c) = 4. Are they also represented by the function v for which v(a) = −1, v(b) = 0, and v(c) = 2?
How about the function w for which w(a) = w(b) = 0 and w(c) = 8?

Given:​
u(a) = 0, u(b) = 1, u(c) = 4​
Preference order: c > b > a

For function v:

v(a) = −1, v(b) = 0, v(c) = 2​

Order: c > b > a

This is the same as u

v represents the same preferences

For function w:

w(a) = 0, w(b) = 0, w(c) = 8​

Order: c > a = b

Here, a and b are indifferent, but in u we had b > a

Ordering is different

w does not represent the same preferences

Final Answer:

●​ v: Yes, represents the same preferences

 ●​ w: No, does not represent the same preferences

Strategic game with ordinal preferences

A strategic game (with ordinal preferences) consists of

• a set of players

• for each player, a set of actions

• for each player, preferences over the set of action profiles.

A very wide range of situations may be modeled as strategic games.

For example,

●​ The players may be firms, the actions prices, and the preferences a reflection of the firms’
profits.
●​ The players may be candidates for political office, the actions campaign expenditures,
and the preferences a reflection of the candidates’ probabilities of winning.
●​ The players may be animals fighting over some prey, the actions concession times, and
the preferences a reflection of whether an animal wins or loses.

Prisoner’s Dilemma

Players:

●​ Player 1 (Suspect 1)
●​ Player 2 (Suspect 2)

Actions:

●​ Quiet (remain silent)

●​ Fink (betray)

Payoff Matrix:

Quiet Fink

Quiet (2, 2) (0, 3)

 Fink (3, 0) (1, 1)

Preferences:

For each player:

●​ Best outcome: Fink while other plays Quiet → payoff 3

●​ Next best: both play Quiet → payoff 2
●​ Next: both play Fink → payoff 1
●​ Worst: Quiet while other plays Fink → payoff 0

So preference ordering:

Fink vs Quiet > Quiet vs Quiet > Fink vs Fink > Quiet vs Fink

Conclusion:

●​ Fink is the dominant strategy for both players

●​ Nash equilibrium: (Fink, Fink)
●​ Outcome is not socially optimal (since (Quiet, Quiet) is better for both)

EXERCISE 14.1 (Working on a joint project) Formulate a strategic game that models a situation
in which two people work on a joint project in the case that their preferences are the same as
those in the game in Figure 14.1 except that each person prefers to work hard than to goof off
when the other person works hard. Present your game in a table like the one in Figure 14.1.

Players:

●​ Player 1
●​ Player 2

Actions:

●​ Work Hard (H)

●​ Goof Off (G)

Payoff Matrix:
 Work Hard Goof Off
(H) (G)

Work Hard (H) (2, 2) (0, 3)

Goof Off (G) (1, 0) (1, 1)

Explanation:

●​ The original game is modified so that:

○​ Each player prefers working hard when the other works hard
○​ Hence (H, H) is preferred over (G, H)
●​ So the payoff (3, 0) is reduced to (1, 0) to reflect this preference

Preferences:

For each player:

●​ If the other works hard:

○​ Prefer H over G
●​ If the other goofs off:
○​ Prefer G over H

Conclusion:

●​ (H, H) and (G, G) are Nash equilibria

●​ (H, H) is the better (efficient) outcome

Consider a duopoly market where two firms produce an identical good. Both firms
simultaneously decide whether to set a High price or a Low price. Each firm aims to maximize
its own profit. If both firms set a High price, each earns a profit of $1000. If both set a Low
price, each earns a profit of $600. However, if one firm sets a High price while the other sets a
Low price, the high-pricing firm attracts no customers and incurs a loss of $200, whereas the
low-pricing firm earns a profit of $1200 due to high sales volume. Based on this scenario answer
the following questions a)Construct the strategic game (normal form) by clearly identifying the
players, their strategies, and the payoff matrix. b)Analyze the game to find all possible Nash
Equilibria.
Players:

●​ Firm 1
●​ Firm 2

Strategies:

●​ High Price (H)

●​ Low Price (L)

Payoff Matrix:

H L

H (1000, 1000) (−200, 1200)

L (1200, −200) (600, 600)

Check best responses:

●​ If Firm 2 chooses H:
○​ Firm 1: H → 1000, L → 1200 → best is L
●​ If Firm 2 chooses L:
○​ Firm 1: H → −200, L → 600 → best is L

Firm 1 prefers L in both cases

Similarly for Firm 2:

●​ If Firm 1 chooses H:
○​ Firm 2: H → 1000, L → 1200 → best is L
●​ If Firm 1 chooses L:
○​ Firm 2: H → −200, L → 600 → best is L

Firm 2 also prefers L in both cases

Nash Equilibrium
 ●​ Both firms choose Low Price (L, L)
●​ Payoff: (600, 600)

Conclusion
●​ Low price (L) is a dominant strategy for both firms
●​ Unique Nash equilibrium: (L, L)
●​ However, (H, H) gives higher profits (1000, 1000) but is not stable

EXERCISE 16.1 (Hermaphroditic fish) Members of some species of hermaphroditic fish choose,
in each mating encounter, whether to play the role of a male or a female. Each fish has a
preferred role, which uses up fewer resources and hence allows more future mating. A fish
obtains a payoff of H if it mates in its preferred role and L if it mates in the other role, where H >
L. (Payoffs are measured in terms of number of offspring, which fish are evolved to maximize.)
Consider an encounter between two fish whose preferred roles are the same. Each fish has two
possible actions: mate in either role, and insist on its preferred role. If both fish offer to mate in
either role, the roles are assigned randomly, and each fish’s payoff is 12 (H + L) (the average of
H and L). If each fish insists on its preferred role, the fish do not mate; each goes off in search of
another partner, and obtains the payoff S. The higher the chance of meeting another partner, the
larger is S. Formulate this situation as a strategic game and determine the range of values of S,
for any given values of H and L, for which the game differs from the Prisoner’s Dilemma only in
the names of the actions.

Players:

●​ Fish 1
●​ Fish 2

Strategies:

●​ Offer (O): agree to mate in either role

●​ Insist (I): insist on preferred role

Payoff Matrix:

O I
 O ((H+L)/2, (H+L)/2) (L , H)

I (H , L) (S , S)

Analysis
Compare payoffs (given H > L):

●​ If opponent plays O:
○​ O → (H+L)/2
○​ I → H → better
●​ If opponent plays I:
○​ O → L
○​ I → S

So:

●​ I is better than O if S > L

●​ O is better than I if S < L

Condition for Prisoner’s Dilemma Structure

For Prisoner’s Dilemma, we need:

Temptation > Reward > Punishment > Sucker

Mapping:

●​ Temptation (T) = H
●​ Reward (R) = (H+L)/2
●​ Punishment (P) = S
●​ Sucker (S') = L

So condition becomes:

H > (H+L)/2 > S > L

The game is equivalent to the Prisoner’s Dilemma when:

L < S < (H+L)/2

Conclusion
●​ If S lies between L and (H+L)/2, the game has:
○​ Incentive to insist (defect)
○​ But mutual cooperation (offer) is better than mutual insist

⇒ Hence, it behaves like a Prisoner’s Dilemma

Bach or Stravinsky (Battle of the Sexes)

Players:

●​ Player 1
●​ Player 2

Actions:

●​ Bach (B)
●​ Stravinsky (S)

Payoff Matrix:

B S

B (2, 1) (0, 0)

S (0, 0) (1, 2)

Analysis

●​ Both players prefer to be together rather than apart

●​ Player 1 prefers Bach
●​ Player 2 prefers Stravinsky
Best Responses

●​ If Player 2 chooses B:
○​ Player 1: B → 2, S → 0 ⇒ best is B
●​ If Player 2 chooses S:
○​ Player 1: B → 0, S → 1 ⇒ best is S
●​ If Player 1 chooses B:
○​ Player 2: B → 1, S → 0 ⇒ best is B
●​ If Player 1 chooses S:
○​ Player 2: B → 0, S → 2 ⇒ best is S

Pure Nash Equilibria

●​ (B, B) with payoff (2, 1)

●​ (S, S) with payoff (1, 2)

Conclusion

●​ There are two pure Nash equilibria

●​ Both involve coordination (choosing the same option)
●​ The conflict is over which equilibrium to select

Stag Hunt Game

Players:

●​ Player 1
●​ Player 2

Actions:

●​ Stag
●​ Hare

Payoff Matrix:

Stag Hare
Stag (2, 2) (0, 1)

Hare (1, 0) (1, 1)

Analysis

●​ Stag requires both players to cooperate

●​ Hare gives a safe payoff even if the other player does not cooperate

Best Responses

●​ If Player 2 chooses Stag:

○​ Player 1: Stag → 2, Hare → 1 ⇒ best is Stag
●​ If Player 2 chooses Hare:
○​ Player 1: Stag → 0, Hare → 1 ⇒ best is Hare
●​ If Player 1 chooses Stag:
○​ Player 2: Stag → 2, Hare → 1 ⇒ best is Stag
●​ If Player 1 chooses Hare:
○​ Player 2: Stag → 0, Hare → 1 ⇒ best is Hare

Pure Nash Equilibria

●​ (Stag, Stag) with payoff (2, 2)

●​ (Hare, Hare) with payoff (1, 1)

Conclusion

●​ There are two pure Nash equilibria

●​ (Stag, Stag) is the efficient outcome
●​ (Hare, Hare) is the safe outcome

Nash Equilibrium

A Nash equilibrium is a situation in a game where no player can improve their payoff by
changing their strategy alone, given the strategies chosen by the other players.
Example

Consider two competing shops deciding whether to set a high price or a low price.

●​ If both set a high price, they earn good profits

●​ If one sets a low price while the other sets a high price, the low-price shop attracts more
customers
●​ If both set low prices, profits are lower but stable

In this situation, both shops may end up choosing low price.

Why this is Nash Equilibrium

●​ Given that the other shop is setting a low price, no shop can increase its profit by
switching to a high price
●​ So neither shop has an incentive to change its decision

A Nash equilibrium occurs when every player’s strategy is the best response to others, and no
one benefits from changing their choice alone.

Given Game

L C R

T (2, 2) (1, 3) (0, 1)

M (3, 1) (0, 0) (0, 0)

B (1, 0) (0, 0) (0, 0)

Player 1’s best responses:

●​ To L → M
●​ To C → T
●​ To R → T, M, B (all give same payoff)
Player 2’s best responses:

●​ To T → C
●​ To M → L
●​ To B → L, C, R (all give same payoff)

Nash Equilibria

●​ (M, L)
●​ (T, C)
●​ (B, R)

Conclusion

There are three Nash equilibria. Two give positive payoffs: (M, L) and (T, C), and one gives zero
payoff: (B, R).

Each of two players has two possible actions, Quiet and Fink; each action pair results in the
players’ receiving amounts of money equal to the numbers corresponding to that action pair .
(For example, if player 1 chooses Quiet and player 2 chooses Fink, then player 1 receives
nothing, whereas player 2 receives $3.) The players are not “selfish”; rather, the preferences of
each player i are represented by the payoff function mi(a) + αmj(a), where mi(a) is the amount of
money received by player i when the action profile is a, j is the other player, and α is a given
nonnegative number. Player 1’s payoff to the action pair (Quiet, Quiet), for example, is 2 + 2α. a.
Formulate a strategic game that models this situation in the case α = 1. Is this game the
Prisoner’s Dilemma? b. Find the range of values of α for which the resulting game is the
Prisoner’s Dilemma.

(a) Strategic Game for α = 1

Original monetary payoffs:

●​ (Quiet, Quiet) = (2, 2)

●​ (Quiet, Fink) = (0, 3)
●​ (Fink, Quiet) = (3, 0)
●​ (Fink, Fink) = (1, 1)

New payoff function:​

ui = mi + α mj, with α = 1

So each player’s payoff = own payoff + opponent’s payoff

New Payoffs:

●​ (Quiet, Quiet):​
u1 = 2 + 2 = 4, u2 = 2 + 2 = 4
●​ (Quiet, Fink):​
u1 = 0 + 3 = 3, u2 = 3 + 0 = 3
●​ (Fink, Quiet):​
u1 = 3 + 0 = 3, u2 = 0 + 3 = 3
●​ (Fink, Fink):​
u1 = 1 + 1 = 2, u2 = 1 + 1 = 2

Payoff Matrix:

Quiet Fink

Quiet (4, 4) (3, 3)

Fink (3, 3) (2, 2)

Conclusion:

●​ Quiet strictly dominates Fink

●​ Unique Nash equilibrium: (Quiet, Quiet)
●​ This is not a Prisoner’s Dilemma

(b) Range of α for Prisoner’s Dilemma

Compute general payoffs:

●​ (Quiet, Quiet): 2 + 2α
●​ (Quiet, Fink): 3α
●​ (Fink, Quiet): 3
●​ (Fink, Fink): 1 + α

For Prisoner’s Dilemma, we need:

Fink strictly dominates Quiet

If opponent plays Quiet:

●​ Quiet → 2 + 2α
●​ Fink → 3

Need:​
3 > 2 + 2α​
⇒ 1 > 2α​
⇒ α < 1/2

If opponent plays Fink:

●​ Quiet → 3α
●​ Fink → 1 + α

Need:​
1 + α > 3α​
⇒ 1 > 2α​
⇒ α < 1/2

Final Answer

●​ For α = 1 → Not a Prisoner’s Dilemma

●​ The game is a Prisoner’s Dilemma when:

α < 1/2

Conclusion

●​ As α increases (players care more about others), cooperation increases

●​ When α ≥ 1/2, the dilemma disappears

Consider the variants of the n-hunter Stag Hunt in which only m hunters, with 2≤m≤n, need to
pursue the stag in order to catch it (continue to assume that there is a single stag). Assume that a
captured stag is shared only by the hunters who catch it. nder each of following assumptions on
the hunters’ preferences, find the Nash equilibria of the strategic game that models the situation.
Analyze if,

a. Each hunter prefers the fraction 1/m of the stag to a hare;

b. Each hunter prefers a fraction 1/k of the stag to a hare, but prefers a hare to any smaller
fraction of the stag, where k is an integer with m≤k≤n.

Each player chooses:

●​ Stag (S)
●​ Hare (H)

A stag is caught only if at least m hunters choose S, where 2 ≤ m < n

(a) Case: each prefers 1/n of stag to hare

●​ If at least m hunters choose S → stag is caught

●​ Those hunters share the stag, each gets at least 1/n
●​ Since 1/n > hare, all prefer stag when enough hunters choose S

Nash Equilibria:

●​ All players choose Hare (H, H, …, H)

●​ Any profile where at least m players choose Stag and no one wants to deviate

So equilibria are:

●​ All H
●​ Any outcome with at least m players choosing S

(b) Case: prefers 1/k of stag to hare (m ≤ k ≤ n)

●​ A player prefers stag only if number of stag hunters ≥ k

●​ If fewer than k hunters choose S → payoff is worse than hare

Nash Equilibria:

●​ All players choose Hare (H, H, …, H)

●​ Any profile where exactly k or more players choose S, and no one benefits from
deviating

Conclusion

●​ In both cases, (H, H, …, H) is always a Nash equilibrium

●​ Cooperation (choosing S) is stable only when enough players choose it
●​ The threshold (m or k) determines when cooperation is sustainable
Best Response Function
A best response function gives the set of strategies that provide a player with the highest
payoff, given the strategies chosen by the other players.

Example

Consider two firms choosing High price (H) or Low price (L).

●​ If Firm 2 chooses H:
○​ Firm 1 gets higher profit from choosing L
●​ If Firm 2 chooses L:
○​ Firm 1 still gets higher profit from choosing L

So Firm 1’s best response is always L

Example

Consider two firms choosing High price (H) or Low price (L).

●​ If Firm 2 chooses H:
○​ Firm 1 gets higher profit from choosing L
●​ If Firm 2 chooses L:
○​ Firm 1 still gets higher profit from choosing L

So Firm 1’s best response is always L

A best response function shows optimal choices against others’ strategies Nash equilibrium
occurs when both players are playing best responses to each other Two individuals are involved
in a synergistic relationship. If both individuals devote more effort to the relationship, they are
both better off. For any given effort of individual j, the return to individual i’s effort first
increases, then decreases. Specifically, an effort level is a nonnegative number, and individual i’s
preferences (for i=1,2) are represented by the payoff function ai (c+aj-ai), where ai is i effort
level, aj is the other individual’s effort level, c>0 is a constant.

a)Model the situation as a strategic game

b)Find players best response functions

c)Find the Nash equilibrium

(a) Strategic Game

Players:

●​ Player 1
●​ Player 2

Strategies:

●​ Each chooses effort ai ≥ 0

Payoff Functions:

●​ u1(a1, a2) = a1(c + a2 − a1)

●​ u2(a2, a1) = a2(c + a1 − a2)

(b) Best Response (Quadratic Approach)

For Player 1:

u1 = a1(c + a2 − a1)​
= a1c + a1a2 − a1²

This is a quadratic in a1:

u1 = −a1² + (c + a2)a1

Maximum of a quadratic ax² + bx occurs at:

a1 = −b / (2a)

Here:

●​ a = −1
●​ b = (c + a2)

So,

a1 = (c + a2)/2

Similarly for Player 2:

u2 = −a2² + (c + a1)a2

⇒ a2 = (c + a1)/2

(c) Nash Equilibrium

Solve simultaneously:

a1 = (c + a2)/2​
a2 = (c + a1)/2

Substitute:

a1 = (c + (c + a1)/2)/2

Simplify:

a1 = (3c + a1)/4

Multiply both sides by 4:

4a1 = 3c + a1

⇒ 3a1 = 3c

⇒ a1 = c

Similarly:

a2 = c

Final Answer

●​ Best response functions:​

a1 = (c + a2)/2​
a2 = (c + a1)/2
●​ Nash equilibrium:​
(a1, a2) = (c, c)

Conclusion

●​ Payoffs are quadratic (concave), so maximum occurs at vertex

●​ Both players choose effort equal to c at equilibrium

Consider a two-player strategic game where each player chooses an action from the set of
nonnegative numbers. The payoff functions are given as: u1(a1,a2)=a1(a2-a1) and
u2(a1,a2)=a2(1-a1-a2)
(a) Develop the strategic game model by clearly specifying the

players, their action sets, and the payoff functions.

(b) Find Nash Equilibrium of this game..

(a) Strategic Game

Players:

●​ Player 1
●​ Player 2

Actions:

●​ Player 1 chooses a1 ≥ 0
●​ Player 2 chooses a2 ≥ 0

Payoff functions:

●​ u1(a1, a2) = a1(a2 − a1)

●​ u2(a1, a2) = a2(1 − a1 − a2)

(b) Nash Equilibrium (Quadratic Approach)

For Player 1:

u1 = a1 a2 − a1^2

This is a quadratic in a1, so maximum occurs at:

a1 = a2 / 2

For Player 2:

u2 = a2 − a1 a2 − a2^2

This is quadratic in a2, so maximum occurs at:

a2 = (1 − a1) / 2

Best Response Functions:

●​ a1 = a2 / 2
 ●​ a2 = (1 − a1) / 2

Solve:

Substitute a2 into a1:

a1 = (1 − a1) / 4

Multiply both sides by 4:

4a1 = 1 − a1

5a1 = 1

a1 = 0.2

Now find a2:

a2 = (1 − 0.2) / 2 = 0.8 / 2 = 0.4

Final Answer:

Nash equilibrium:

(a1, a2) = (0.2, 0.4)

Conclusion:

Both players choose effort levels 0.2 and 0.4 respectively, which are best responses to each other.
Module – 2
Steady state

In a steady state, every player’s behavior is the same whenever she plays the game, and no player
wishes to change her behavior, knowing (from her experience) the other players’ behavior. In a
steady state in which each player’s “behavior” is simply an action and within each population all
players choose the same action, the outcome of every play of the game is the same Nash
equilibrium.

Stochastic steady states

steady state, in which each player chooses her actions probabilistically; such a steady state is
called stochastic (“involving probability”).

NASH EQUILIBRIUM

NASH EQUILIBRIUM of a strategic game is an action profile in which every player’s action is
optimal given every other player’s action.

Matching Pennies: Proof of the Unique Stochastic Steady State

Players’ Mixed Strategies

Let​
p = probability that Player 1 chooses Head​
1 − p = probability that Player 1 chooses Tail​
q = probability that Player 2 chooses Head​
1 − q = probability that Player 2 chooses Tail

Probability that Player 1 Wins

Player 1 wins $1 when outcomes are:​

(Head, Head)​
(Tail, Tail)

Thus​
P(Win) = pq + (1 − p)(1 − q)

Simplifying:​
P(Win) = pq + 1 − p − q + pq​
P(Win) = 1 − q + p(2q − 1)
Probability that Player 1 Loses

Player 1 loses $1 when outcomes are:​

(Head, Tail)​
(Tail, Head)

Thus​
P(Lose) = p(1 − q) + (1 − p)q

Simplifying:​
P(Lose) = q + p(1 − 2q)

Player 1’s Best Response

Case 1: q < 1/2

Then​
2q − 1 < 0

So​
P(Win) = 1 − q + p(2q − 1)

is decreasing in p.

Thus Player 1 maximizes the probability of winning by choosing​

p=0

i.e., always choose Tail.

Case 2: q > 1/2

Then​
2q − 1 > 0

So​
P(Win)

is increasing in p.

Thus Player 1 maximizes winning by choosing​

p=1

i.e., always choose Head.

Player 2’s Best Response

If Player 1 plays a pure strategy:

If Player 1 plays Head, Player 2 should play Tail​

If Player 1 plays Tail, Player 2 should play Head

Thus Player 2 also chooses a pure strategy with certainty.

Implication for Steady State

If q ≠ 1/2:

Player 1 plays a pure strategy​

Player 2 then changes to the opposite pure strategy

Hence strategies keep changing and cannot stabilize.

Therefore no steady state exists when q ≠ 1/2.

Unique Stochastic Steady State

The only situation where neither player wants to change is:

p = 1/2, q = 1/2

Thus both players randomize equally.

Final Result

The unique stochastic steady state is

p = q = 1/2

i.e., each player chooses Head and Tail with probability 1/2.

Matching Pennies: No Steady State (Mathematical Form)

Players’ Mixed Strategies

Let​
p = probability that Player 1 chooses Head​
1 − p = probability that Player 1 chooses Tail​
q = probability that Player 2 chooses Head​
1 − q = probability that Player 2 chooses Tail
Probability that Player 1 Wins

Player 1 wins when outcomes are:​

(Head, Head), (Tail, Tail)

P(Win) = pq + (1 − p)(1 − q)

Simplifying:​
P(Win) = pq + 1 − p − q + pq​
P(Win) = 1 − q + p(2q − 1)

Player 1’s Best Response

Case 1: q < 1/2

Then​
2q − 1 < 0

So​
P(Win) = 1 − q + p(2q − 1)

is decreasing in p

Thus optimal choice:​

p=0

Player 1 chooses Tail with certainty

Case 2: q > 1/2

Then​
2q − 1 > 0

So​
P(Win) is increasing in p

Thus optimal choice:​

p=1

Player 1 chooses Head with certainty

Case 3: q = 1/2

Then​
2q − 1 = 0
 So​
P(Win) = 1/2 (independent of p)

Thus any p ∈ [0,1] is optimal

Player 2’s Best Response

If Player 1 chooses a pure strategy:

If p = 1 (Player 1 chooses Head):​

Player 2 payoff:​
Head → −1​
Tail → +1​
Best response: choose Tail

If p = 0 (Player 1 chooses Tail):​

Player 2 payoff:​
Head → +1​
Tail → −1​
Best response: choose Head

No Steady State in Pure Strategies

Suppose a steady state exists with pure strategies:

If (H,H): Player 2 deviates → (H,T)​

If (H,T): Player 1 deviates → (T,T)​
If (T,T): Player 2 deviates → (T,H)​
If (T,H): Player 1 deviates → (H,H)

Thus every pure outcome admits a profitable deviation

Conclusion

If q ≠ 1/2, Player 1 chooses a pure strategy (p = 0 or p = 1)​

Then Player 2 responds with the opposite pure strategy​
This leads to continuous switching

Hence no fixed pair (p,q) with p ∈ {0,1}, q ∈ {0,1} can satisfy mutual best responses

Final Result

There is no steady state in pure strategies for Matching Pennies.

vNM preferences

Preferences over lotteries that can be represented by the expected value of a payoff function over
deterministic outcomes, as developed by John von Neumann and Oskar Morgenstern.

Eg: A person prefers a lottery that gives ₹100 with probability 0.6 over one that gives ₹100 with
probability 0.4 because it has a higher expected utility.

Another person is indifferent between receiving ₹50 for sure and a lottery that gives ₹100 with
probability 0.5 and ₹0 with probability 0.5 because both yield the same expected utility.

Bernoulli payoff function

A payoff function over deterministic outcomes whose expected value represents such
preferences, named after Daniel Bernoulli.

Eg: If a person’s utility function is u(x) = x, then the expected utility of a lottery is computed
using this function rather than the monetary value.​
For example, using a Bernoulli payoff function, a person may prefer ₹50 for sure over a 50–50
lottery between ₹0 and ₹100 because u(50) > 0.5u(100) + 0.5u(0).

Consider the following lottery choices:

Choice 1:
Lottery 1 ($2 million with certainty) vs.
Lottery 2 ($10 million with prob. 0.1, $2 million with prob. 0.89, $0 with prob. 0.01)
Choice 2: Lottery 3 ($2 million with prob. 0.11, $0 with prob. 0.89) vs.
Lottery 4 ($10 million with prob. 0.1, $0 with prob. 0.9)
Experimental subjects often prefer Lottery 1 over Lottery 2 and Lottery 4 over Lottery 3.
Explain why these preferences cannot be represented by an expected utility function.

First pair of lotteries

You are asked to choose between:
Lottery 1: $2 million for sure.
Lottery 2: $10 million with probability 0.1, $2 million with probability 0.89, $0 with
probability 0.01.
Many people say that Lottery 1 is better than Lottery 2 because Lottery 1 is safe, while
Lottery 2 has a small chance of getting nothing. People dislike that tiny risk.
Second pair of lotteries
Now you are asked to choose between:
Lottery 3: $2 million with probability 0.11, $0 with probability 0.89.
Lottery 4: $10 million with probability 0.1, $0 with probability 0.9.
Many people say that Lottery 4 is better than Lottery 3 because $10 million is attractive.
The problem
People’s choices are:
Lottery 1 > Lottery 2​
Lottery 4 > Lottery 3
But according to Expected Utility Theory, this combination cannot happen.
If someone prefers Lottery 1 over Lottery 2, the mathematics implies they must also
prefer Lottery 3 over Lottery 4. However, experiments show that people often prefer
Lottery 4 over Lottery 3 instead.
So people’s behavior contradicts the expected utility model.
From
u(2) > 0.1u(10) + 0.89u(2) + 0.01u(0)
it becomes
0.11u(2) + 0.89u(0) > 0.1u(10) + 0.9u(0)
This second inequality represents:
Left side = expected utility of Lottery 3​
Right side = expected utility of Lottery 4
So mathematically, preferring Lottery 1 over Lottery 2 implies that one must prefer Lottery 3
over Lottery 4.
But people do not behave like that.
5.​ What experiments show
Experiments show that many people choose:
Lottery 1 over Lottery 2​
Lottery 4 over Lottery 3
This pattern is called the Allais paradox.
It shows that real human decisions do not always follow expected utility theory.
Why economists still use expected utility
 Even though it is not perfect, it is simple, works reasonably well, and makes models
easier to analyze. Therefore, game theory still assumes expected utility most of the time.
Conclusion: The example shows that real people often make choices that violate the predictions
of expected utility theory, but economists still use the theory because it is mathematically
convenient.

Strategic games in which players may randomize

A strategic game (with vNM preferences) consists of

• a set of players
• for each player, a set of actions
• for each player, preferences regarding lotteries over action profiles that may
be represented by the expected value of a (“Bernoulli”) payoff function over
action profiles.

Consider two different payoff matrices, both representing the Prisoner's Dilemma when
preferences are ordinal. Explain why these two games are considered the same under ordinal
preferences but become different strategic games when preferences are treated as vNM (von
Neumann-Morgenstern) preferences. Provide a justification for this distinction.

Q F

Q 2, 2 0, 3

F 3, 0 1, 1

Q F
 Q 3, 3 0, 4

F 4, 0 1, 1

Under ordinal preferences, only the ranking of outcomes matters.

For Player 1 in both games:

Best: FQ → payoff 3 (Game 1), 4 (Game 2)​

Second: QQ → payoff 2 (Game 1), 3 (Game 2)​
Third: FF → payoff 1 (both games)​
Worst: QF → payoff 0 (both games)

Thus the ranking is:

FQ ≻ QQ ≻ FF ≻ QF

Similarly for Player 2:

QF ≻ QQ ≻ FF ≻ FQ

Since the ranking is identical in both matrices, the two games are the same under ordinal
preferences.

Now consider vNM preferences.

For two payoff matrices to represent the same vNM preferences, there must exist constants a > 0
and b such that:

u₂ = a · u₁ + b

Check if such a transformation exists for Player 1:

From Game 1 → Game 2:

2a + b = 3​
0a + b = 0 ⇒ b = 0

Substitute b = 0:

2a = 3 ⇒ a = 3/2
Now check another payoff:

3a + b = 4​
3 × (3/2) = 4.5 ≠ 4

Contradiction.

Thus no affine transformation exists.

Hence, under vNM preferences:

Expected utility depends on actual numbers.

For example, suppose Player 2 plays Q with probability q.

In Game 1:

U₁(Q) = 2q + 0(1 − q) = 2q​

U₁(F) = 3q + 1(1 − q) = 2q + 1

In Game 2:

U₁(Q) = 3q​
U₁(F) = 4q + 1(1 − q) = 3q + 1

Although both give dominance of F, the expected utilities are numerically different and affect
behavior in more general settings such as mixed strategies and risk.

Final conclusion:

The two games are identical under ordinal preferences because rankings are the same, but they
are different under vNM preferences because no positive affine transformation maps one payoff
matrix to the other, and expected utilities differ.

Define the term Mixed Strategies. Mention the notations used to represent Mixed Strategies.

A mixed strategy is a strategy in which a player chooses a probability distribution over her
available actions, generating a lottery over outcomes.
The notations used are: αᵢ to denote a mixed strategy of player i, α* to denote a mixed strategy
profile, α*₋ᵢ to denote the strategies of all players other than player i, and Uᵢ(α) to denote the
expected payoff to player i from the mixed strategy profile α.

1)Best response functions in two-player two-action games

Players: Player 1 and Player 2

Actions:
We have:

Player 1: T or B

Player 2: L or R

But instead of choosing one action, they randomize:

Player 1:

T with probability p

B with probability 1 − p

Player 2:

L with probability q

R with probability 1 − q

Payoff Matrix:

L R

T pq p(1-q)

B (1-p)q (1-p)(1-q)

Player 1 doesn’t get one payoff — she gets an average payoff.

Expected payoff = pq · u₁(T,L) + p(1−q) · u₁(T,R) + (1−p)q · u₁(B,L) + (1−p)(1−q) · u₁(B,R)

Expected payoff = p[q·u₁(T,L) + (1−q)·u₁(T,R)] + (1−p)[q·u₁(B,L) + (1−q)·u₁(B,R)]

Expected payoff = p · E₁(T, α₂) + (1−p) · E₁(B, α₂)

Player 1 is basically choosing between:

T gives payoff = E₁(T, α₂)

B gives payoff = E₁(B, α₂)
and mixing between them.

Conclusion: Player 1 mixes between T and B, and her total payoff is just the weighted average of
what she would get from each pure strategy.
—--------------------------------------------------------------------------------------------------------------------
2)Apply mixed strategy algorithm to find the expected payoffs for each player in the game of
Matching Pennies

Players: Player 1 and Player 2

Actions: A = {H,T}
Both players choose: Head (H) or Tail (T)
Player 1:

●​ Wins (+1) if choices match

●​ Loses (−1) if different

Mixed Strategies:

p = probability Player 1 plays Head

q = probability Player 2 plays Head

Payoff Matrix:
H T

H +1,-1 -1,+1

T -1,+1 +1,-1

Player 1’s expected payoffs:

If Player 1 plays Head:

Player 2 plays Head (q) → payoff = +1

Player 2 plays Tail (1−q) → payoff = −1

So: E₁(H) = q(1) + (1−q)(−1) = 2q − 1

If Player 1 plays Tail:

Player 2 plays Head → −1

Player 2 plays Tail → +1

E₁(T) = q(−1) + (1−q)(1) = 1 − 2q

Now compare:

If q < 1/2:

E₁(T) > E₁(H)

Best choice = Tail

If q > 1/2:

E₁(H) > E₁(T)

Best choice = Head

If q = 1/2:

Both equal

Player 1 is indifferent → can mix any way

We summarize:

B₁(q) =

- 0 if q < 1/2 (play Tail)

- [0,1] if q = 1/2 (any mix)

 - 1 if q > 1/2 (play Head)

Meaning:

0 = probability of Head → always Tail

1 = always Head

Player 2 is symmetric:

B₂(p) =

- 1 if p < 1/2 (play Head)

- [0,1] if p = 1/2 (any mix)

- 0 if p > 1/2 (play Tail)

The only point where both are best responding:

p = 1/2, q = 1/2

Conclusion: Each player mixes 50-50 so that the opponent cannot predict or gain
advantage.

—---------------------------------------------------------------------------------------------------------

3)Apply mixed strategy algorithm to find the expected payoffs for each
player in BoS model
Players: Player 1 and Player 2
Actions: A ={S,B}
B -> Go to Bach Concert
S -> Go to Stravinsky concert
Mixed strategies:
Let:
p = probability Player 1 plays B
q = probability Player 2 plays B

Payoff Matrix:
B S

B (2,1) (0,0)

S (0,0) (1,2)

Player 1’s expected payoff

If Player 1 plays B:

E₁(B) = 2q + 0(1−q) = 2q

If Player 1 plays S:

E₁(S) = 0·q + 1(1−q) = 1−q

Compare (find best response of Player 1)

We compare:

2q vs 1−q

Solve:

2q > 1−q ⇒ 3q > 1 ⇒ q > 1/3

So:

If q > 1/3 → best response = B → p = 1

If q < 1/3 → best response = S → p = 0

If q = 1/3 → indifferent → any p

Player 1’s best response function

B₁(q) =

- 0 if q < ⅓

- [0,1] if q = ⅓

- 1 if q > ⅓

Player 2’s expected payoff

If Player 2 plays B:

E₂(B) = 1·p + 0(1−p) = p

If Player 2 plays S:

E₂(S) = 0·p + 2(1−p) = 2(1−p)

Compare (Player 2)

p vs 2(1−p)

Solve:

p > 2(1−p) ⇒ p > 2 − 2p ⇒ 3p > 2 ⇒ p > 2/3

So:

If p > 2/3 → best response = B → q = 1

If p < 2/3 → best response = S → q = 0

If p = 2/3 → indifferent → any q

Player 2’s best response function

B₂(p) =

- 1 if p > ⅔

- [0,1] if p = ⅔

- 0 if p < ⅔

Find Nash equilibria (intersection points)

We look for points where both are best responding

Pure strategy equilibrium

(p,q) = (1,1) → (B,B)

Both best responding Therefore, Nash equilibrium

Pure strategy equilibrium

(p,q) = (0,0) → (S,S)

Both best responding Therefore, Nash equilibrium

Mixed strategy equilibrium

At indifference:

Player 1 indifferent → q = ⅓

Player 2 indifferent → p = 2/3

So:

(p,q) = (2/3, 1/3)

Both best responding Therefore, Nash equilibrium

 Conclusion: Two pure equilibria (coordination) + one mixed equilibrium where players
randomize to balance preferences.

4)Find all the mixed strategy Nash equilibria of the following strategic games

i)

L R

T 6,0 0,6

B 3,2 6,0

Players: Player 1, Player 2

Actions: Player 1 : {T,B}, Player 2 : {L,R}

Let:

​ Player 1 plays T with probability p, B with (1 − p)

​ Player 2 plays L with probability q, R with (1 − q)

Player 1’s indifference condition

Player 1 mixes when payoff(T) = payoff(B)

payoff(T) = 6q

payoff(B) = 3q + 6(1 − q) = 6 − 3q

Set equal:

​ 6q = 6 − 3q

​ 9q = 6

​ q = 2/3
Player 2’s indifference condition

Player 2 mixes when payoff(L) = payoff(R)

​ payoff(L) = 0·p + 2(1 − p) = 2 − 2p

​ payoff(R) = 6p

Set equal:

​ 2 − 2p = 6p

​ 2 = 8p

​ p = 1/4

Mixed Strategy Nash Equilibrium:

Player 1 plays T with probability 1/4 and B with probability 3/4

Player 2 plays L with probability 2/3 and R with probability 1/3

Conclusion: The mixed strategy Nash equilibrium is: (p, q) = (1/4, 2/3)

—--------------------------------------------------------------------------------------------------------------

ii)

L R

T 0,1 0,2

B 2,2 0,1

Players: Player 1, Player 2

Actions: Player 1 : {T,B}, Player 2 : {L,R}

Let:

Player 1 plays T with probability p, B with (1 − p)

Player 2 plays L with probability q, R with (1 − q)

Player 1’s indifference condition

​ payoff(T) = 0·q + 0·(1 − q) = 0

​ payoff(B) = 2q + 0·(1 − q) = 2q

Set equal:

​ 0 = 2q

​ q=0

Player 2’s indifference condition

​ payoff(L) = 1·p + 2·(1 − p) = 2 − p

​ payoff(R) = 2·p + 1·(1 − p) = 1 + p

Set equal:

​ 2−p=1+p

​ 1 = 2p

​ p = 1/2

Mixed Strategy Nash Equilibrium:

Player 1 plays T with probability 1/2 and B with probability 1/2

Player 2 plays L with probability 0 and R with probability 1

Conclusion: The mixed strategy Nash equilibrium is: (p, q) = (1/2, 0)

5)Two people can perform a task if, and only if, they both exert effort. They are both
better off if they both exert effort and perform the task than if neither exerts effort
(and nothing is accomplished); the worst outcome for each person is that she exerts
effort and the other person does not (in which case again nothing is accomplished).
Specifically, the players’ preferences are represented by the expected value of the
payoff functions in the following figure, which c is a positive number less than 1
than can be interpreted as the cost of exerting effort. Find all the mixed strategy
Nash equilibria of this game. How do the equilibria change as c increase? Explain
the reasons for the changes.

No Effort Effort

No Effort 0,0 0,-c

Effort -c,0 1-c,1-c

Players: Player 1, Player 2

Actions: A: {No Effort, Effort}

Let:

Player 1 plays Effort with probability p, No Effort with (1 − p)

Player 2 plays Effort with probability q, No Effort with (1 − q)

Player 1’s indifference condition

​ payoff(No Effort) = 0·(1 − q) + 0·q = 0

​ payoff(Effort) = (−c)(1 − q) + (1 − c)q

​ = −c + cq + q − cq

​ =q−c

Set equal:

​ 0=q−c

​ q=c

Player 2’s indifference condition

​ payoff(No Effort) = 0·(1 − p) + 0·p = 0

​ payoff(Effort) = (−c)(1 − p) + (1 − c)p

​ = −c + cp + p − cp

​ =p−c

Set equal:

​ 0=p−c

​ p=c

Mixed Strategy Nash Equilibrium:

Player 1 plays Effort with probability c

Player 2 plays Effort with probability c

So,

(p, q) = (c, c)

Pure Strategy Nash Equilibria:

(No Effort, No Effort)

(Effort, Effort)
Effect of increase in c:

​ As c increases, the probability of playing Effort (p = c, q = c) increases

​ So players exert effort more often in the mixed equilibrium

Explanation:

​ Higher c means higher cost of effort

​ When effort becomes costly, players need stronger belief that the other will also exert
effort

​ This changes the balance point (indifference), shifting probabilities

Conclusion:

Mixed equilibrium: (c, c)

As c increases, players adjust probabilities to maintain indifference

The strategic tension comes from coordination + cost of effort

—--------------------------------------------------------------------------------------------------------------

6) Analyze whether a mixed strategy (3/4,0, 1/4) for player 1 and (0, 1/3, 2/3) for
player 2 in the following game is a mixed strategy nash equilibrium

L(0) C(1/3) R(2/3)

T(3/4) .,2 3,3 1,1

M(0) .,. 0,. 2,.

B(1/4) .,4 5,1 0,7

Given Strategies:

Player 1: (T, M, B) = (3/4, 0, 1/4)

Player 2: (L, C, R) = (0, 1/3, 2/3)

Check Player 1’s best response

Using Player 2’s strategy (0, 1/3, 2/3):

payoff(T) = 0·0 + 3·(1/3) + 1·(2/3) = 5/3​

​ payoff(M) = 0·0 + 0·(1/3) + 2·(2/3) = 4/3​
​ payoff(B) = 0·0 + 5·(1/3) + 0·(2/3) = 5/3

So,

payoff(T) = payoff(B) = 5/3 > payoff(M)

Player 1 is best responding by mixing between T and B​

Given strategy is optimal

Check Player 2’s best response

Using Player 1’s strategy (3/4, 0, 1/4):

payoff(L) = 2·(3/4) + 0·0 + 4·(1/4) = 5/2​

​ payoff(C) = 3·(3/4) + 0·0 + 1·(1/4) = 5/2​
​ payoff(R) = 1·(3/4) + 0·0 + 7·(1/4) = 5/2

So,

payoff(L) = payoff(C) = payoff(R) = 5/2

Player 2 is indifferent between all strategies​

Given strategy is optimal

Final Answer:

Yes, the given strategies form a mixed strategy Nash equilibrium

Conclusion:

Player 1 mixes between T and B (both give same payoff)

Player 2 is indifferent across all strategies

No player has incentive to deviate

Hence, it is a Nash equilibrium

7) Analyze whether the mixed strategy (0,1/2,1/2) yields a better payoff to Player 1
than pure strategy.

P S

P 1,0 1,0

S 4,1 0,1

R 0,1 3,1

Given Mixed Strategy (Player 1):

Player 1 plays (P, S, R) = (0, 1/2, 1/2)

Compute expected payoff of mixed strategy

Case 1: Player 2 plays P

payoff = 0·1 + (1/2)·4 + (1/2)·0​

=2

Case 2: Player 2 plays S

payoff = 0·1 + (1/2)·0 + (1/2)·3​

= 3/2

Compare with pure strategies

If Player 1 plays:

●​ P:
○​ vs P → 1
○​ vs S → 1
 ●​ S:
○​ vs P → 4
○​ vs S → 0
●​ R:
○​ vs P → 0
○​ vs S → 3

Compare results

●​ Against P:
○​ Mixed = 2
○​ Best pure = 4 (strategy S)
●​ Against S:
○​ Mixed = 3/2
○​ Best pure = 3 (strategy R)

Final Answer:

The mixed strategy (0, 1/2, 1/2) does not yield a better payoff than pure strategies.

Conclusion:

There exists a pure strategy that gives strictly higher payoff in both cases

Hence, the mixed strategy is not better than pure strategies

—---------------------------------------------------------------------------------------------------------

8) Determine whether each of the following statements is true of false:

a)A mixed strategy that assigns positive probability to a strictly dominated action is strictly
dominated.

b)A mixed strategy that assigns positive probability only to actions that are not strictly
dominated is not strictly dominated

a) Answer: True

Reason:

●​ A strictly dominated action is always worse than some other strategy (pure or mixed), no
matter what the opponent does.
 ●​ If a mixed strategy puts positive probability on such a bad action, we can improve it by
shifting that probability to a better action.
●​ This strictly increases payoff in all cases.

Hence, the mixed strategy is strictly dominated.

b)Answer: False

Reason:

●​ Even if individual actions are not strictly dominated, a combination of them (mixed
strategy) can still be strictly dominated by another mixed strategy.
●​ Dominance applies to strategies as a whole, not just individual actions.

So, such a mixed strategy can still be strictly dominated.

8) Analyze the game theoretical model – Expert diagnosis model and hence find the pure
and mixed Nash equilibrium.

Players:

●​ Player 1: Expert (Doctor/Mechanic)

●​ Player 2: Client (Patient/Customer)

Actions:

●​ Expert:
○​ H → Honest diagnosis
○​ D → Dishonest (over-treatment)
●​ Client:
○​ T → Trust
○​ N → Not trust (seek second opinion / refuse)

Payoff Structure (standard form):

T (Trust) N (Not Trust)

H (Honest) (2, 2) (0, 1)

 D (Dishonest) (3, −1) (0, 0)

Step 1: Find Pure Strategy Nash Equilibria

Check best responses:

1.​ If Client plays T:

○​ Expert: H → 2, D → 3 ⇒ best = D
2.​ If Client plays N:
○​ Expert: H → 0, D → 0 ⇒ indifferent
3.​ If Expert plays H:
○​ Client: T → 2, N → 1 ⇒ best = T
4.​ If Expert plays D:
○​ Client: T → −1, N → 0 ⇒ best = N

Pure Nash Equilibria:

●​ (D, N)

Mixed Strategy Nash Equilibrium

Let:

●​ Expert plays H with probability p

●​ Client plays T with probability q

Client’s indifference:

payoff(T) = 2p + (−1)(1 − p) = 3p − 1​
​ payoff(N) = p

Set equal:

3p − 1 = p​
​ 2p = 1​
​ p = 1/2
Expert’s indifference:

payoff(H) = 2q​
​ payoff(D) = 3q

Set equal:

2q = 3q​
​ q=0

Mixed Equilibrium:

●​ Expert: (H, D) = (1/2, 1/2)

●​ Client: (T, N) = (0, 1)

Final Conclusion:
●​ Only pure NE: (D, N)
●​ Mixed NE: (1/2, 0)

This is a trust breakdown problem:

●​ Client cannot trust because expert has incentive to cheat

●​ So equilibrium collapses to no trust

—------------------------------------------------------------------------------------------------------------------

Practice Problem:

Players 1 and 2 each choose a positive integer up to K. If the players choose the same number,
then player 2 pays $1 to player 1; otherwise no payment is made. Each player’s preferences are
represented by her expected monetary payoff.

a)Show that the game has a mixed strategy Nash equilibrium in which each player chooses each
positive integer up to K with probability 1/K

b)Show that the game has no other mixed strategy Nash equilibria (Deduce from the fact that
player 1 assigns positive probability to some action k that player 2 must do so; then look at the
implied restriction on player 1’s equilibrium strategy)
 Module – 3
Extensive games with perfect information;
Strategies and outcomes;
Nash equilibrium;
Sub-game perfect equilibrium;
Finding sub-game perfect equilibria of finite horizon games: Backward induction;
Illustrations: The ultimatum game, Stackelberg’s model of duopoly.

●​ A strategic (normal-form) game ignores the sequence of moves and assumes players
choose a complete plan once and cannot revise it later.
●​ An extensive game explicitly models the sequential nature of decisions, allowing
players to adjust actions as the game progresses.
●​ The current model assumes perfect information, meaning every player knows all prior
actions when making decisions.
●​ A more general model (introduced later) allows for imperfect information, where
players may not fully observe earlier actions.

EXAMPLE 152.1 (Entry game) An incumbent faces the possibility of entry by a

challenger. (The challenger may, for example, be a firm considering entry into an industry
currently occupied by a monopolist, a politician competing for the leadership of a party,
or an animal considering competing for the right to mate with a congener of the opposite
sex.) The challenger may enter or not. If it enters, the incumbent may either acquiesce or
fight.

●​ The challenger moves first: choose In or Out.

●​ If In, the incumbent chooses Acquiesce or Fight.

Final outcomes:

●​ (In, Acquiesce), (In, Fight), Out

●​ A history = actions taken so far (like ∅, In, (In, Fight))
●​ A subhistory = any part of a history
●​ A proper subhistory = part of it, but not the full sequence

Important rule:
●​ A partial sequence (like In) cannot be a final outcome if more moves follow.

DEFINITION 153.1 (Extensive game with perfect information) An extensive game with
perfect information consists of

●​ a set of players
●​ a set of sequences (terminal histories) with the property that no sequence is a proper
subhistory of any other sequence
●​ a function (the player function) that assigns a player to every sequence that is a
proper subhistory of some terminal history
●​ for each player, preferences over the set of terminal histories.

An incumbent faces the possibility of entry by a challenger. (The challenger may, for
example, be a firm considering entry into an industry currently occupied by a
monopolist, a politician competing for the leadership of a party, or an animal
considering competing for the right to mate with a congener of the opposite sex.)
The challenger may enter or not. If it enters, the incumbent may either acquiesce or
fight. In the situation described above, suppose that the best outcome for the
challenger is that it enters and the incumbent acquiesces, and the worst outcome is
that it enters and the incumbent fights, whereas the best outcome for the incumbent
is that the challenger stays out, and the worst outcome is that it enters and there is a
fight. Model this situation as an extensive game with perfect information.

Players: The challenger and the incumbent.

Terminal histories: (In, Acquiesce), (In, Fight), and Out.

Player function: P(∅) = Challenger and P(In) = Incumbent.

Preferences:

The challenger’s preferences are represented by the payoff function

u1 for which

u1(In, Acquiesce) = 2,

u1(Out) = 1

u1(In, Fight) = 0,

The incumbent’s preferences are represented by the payoff function

 u2 for which

u2(Out) = 2,

u2(In, Acquiesce) = 1,

​ ​ u2(In, Fight) = 0.

Final Answer (Subgame Perfect Nash Equilibrium):

●​ Challenger: In
●​ Incumbent: Acquiesce

Conclusion: Equilibrium outcome: (In, Acquiesce) with payoff (2, 1)

Strategies and outcomes

Strategies

A key concept in the study of extensive games is that of a strategy. A player’s strategy
specifies the action the player chooses for every history after which it is her turn to move.

DEFINITION 157.1 (Strategy) A strategy of player i in an extensive game with

perfect information is a function that assigns to each history h after which it is
player i’s turn to move (i.e. P(h) = i, where P is the player function) an action in A(h)
(the set of actions available after h).

Example Explanation

Player 1:
 ●​ Player 1 moves only once (at the start)
●​ Choices: C or D

So Player 1 has 2 strategies:

●​ Choose C
●​ Choose D

Player 2:

●​ Player 2 may move after:

○​ C → choose E or F
○​ D → choose G or H

A strategy must specify actions for both situations, even if one is not reached.

Player 2’s possible strategies:

1.​ (E after C, G after D)

2.​ (E after C, H after D)
3.​ (F after C, G after D)
4.​ (F after C, H after D)

So Player 2 has 4 strategies

Conclusion

●​ Strategy = complete plan

●​ Player 1 → 2 strategies
●​ Player 2 → 4 strategies

Outcome of a Strategy Profile

A strategy profile sss tells us what each player will do at every decision point.
This determines the terminal history (final outcome) of the game.

Process to find outcome O(s)

●​ At the start (empty history ∅), player P(∅) moves

●​ She follows her strategy s_P(∅) and chooses action:​
a1 = s_P(∅)(∅)
●​ If a1 is not terminal:
○​ Next player is P(a1)
○​ She chooses:​
a2 = s_P(a1)(a1)
●​ If (a1, a2) is not terminal, continue similarly
●​ This process continues until a terminal history is reached

Final Outcome

The final sequence of actions is called the outcome: O(s)

Nash equilibrium

As for strategic games, we are interested in notions of equilibrium that model the players’
behavior in a steady state. That is, we look for patterns of behavior with the property that if every
player knows every other player’s behavior, she has no reason to change her own behavior.

One way to find the Nash equilibria of an extensive game in which each player has finitely many
strategies is to list each player’s strategies, find the outcome of each strategy profile, and analyze
this information as for a strategic game. That is, we construct the following strategic game,
known as the strategic form of the extensive game.

Players: The set of players in the extensive game.

Actions: Each player’s set of actions is her set of strategies in the extensive game.

Preferences: Each player’s payoff to each action profile is her payoff to the terminal history
generated by that action profile in the extensive game.

From Definition, we see that

the set of Nash equilibria of any extensive game with perfect information is the set of Nash
equilibria of its strategic form.

Consider the following strategic-form game:

 Acquiesce Fight

In (2, 1) (0, 0)

Out (1, 2) (1, 2)

Player 1: Challenger → strategies: {In,Out}

Player 2: Incumbent → strategies: {Acquiesce,Fight}

a)Find all Nash equilibria of the game.

b)Explain why one equilibrium is not credible in the extensive-form interpretation.

c)Explain the role of commitment and credibility.

Check best responses:

●​ If Player 2 chooses Acquiesce:

○​ Player 1: In → 2, Out → 1 ⇒ best is In
●​ If Player 2 chooses Fight:
○​ Player 1: In → 0, Out → 1 ⇒ best is Out
●​ If Player 1 chooses In:
○​ Player 2: Acquiesce → 1, Fight → 0 ⇒ best is Acquiesce
●​ If Player 1 chooses Out:
○​ Player 2: Acquiesce → 2, Fight → 2 ⇒ both are best responses

Nash Equilibria:

●​ (In, Acquiesce)
●​ (Out, Fight)

(b) Non-credible Equilibrium

●​ (Out, Fight) is not credible in the extensive-form game

Reason:
 ●​ If entry (In) actually occurs, the incumbent would choose:
○​ Acquiesce → 1
○​ Fight → 0

⇒ Acquiesce is better than Fight

So the threat to Fight is not rational when the time comes

(c) Role of Commitment and Credibility

●​ A strategy is credible if it is optimal when actually required

●​ In this game, the incumbent cannot credibly commit to Fight
●​ Because once entry happens, it is better to Acquiesce

Implication:

●​ Challenger anticipates this

●​ So chooses In

Conclusion

●​ Nash equilibria: (In, Acquiesce) and (Out, Fight)

●​ Only (In, Acquiesce) is credible (subgame perfect)
●​ Credibility ensures only rational, believable strategies are followed

EXERCISE 161.2 (Voting by alternating veto) Two people select a policy that affects them both
by alternately vetoing policies until only one remains. First person 1 vetoes a policy. If more than
one policy remains, person 2 then vetoes a policy. If more than one policy still remains, person 1
then vetoes another policy. The process continues until only one policy has not been vetoed.
Suppose there are three possible policies, X, Y, and Z, person 1 prefers X to Y to Z, and person 2
prefers Z to Y to X. Model this situation as an extensive game and find its Nash equilibria.

Extensive Game Model

Players:

●​ Player 1
●​ Player 2

Policies:

●​ X, Y, Z
Preferences:

●​ Player 1: X > Y > Z

●​ Player 2: Z > Y > X

Game Structure:

●​ Player 1 first vetoes one policy

●​ From the remaining two policies, Player 2 vetoes one
●​ The remaining policy is the final outcome

Strategies

Player 1 must specify:

●​ Which policy to veto first

●​ (No further moves since game ends after Player 2’s veto)

Player 2 must specify:

●​ Which policy to veto for each possible pair:

○​ If (Y, Z) remain
○​ If (X, Z) remain
○​ If (X, Y) remain

Backward Induction

Consider Player 2’s choices:

●​ If (Y, Z): prefers Z > Y ⇒ veto Y → outcome Z

●​ If (X, Z): prefers Z > X ⇒ veto X → outcome Z
●​ If (X, Y): prefers Y > X ⇒ veto X → outcome Y

Player 1’s Decision

Anticipating Player 2:

●​ If Player 1 vetoes X → remaining (Y, Z) → outcome Z

●​ If Player 1 vetoes Y → remaining (X, Z) → outcome Z
●​ If Player 1 vetoes Z → remaining (X, Y) → outcome Y

Player 1 prefers: X > Y > Z

So best outcome among these is Y

Player 1 vetoes Z

Final Outcome

●​ Remaining policy: Y

Conclusion

●​ Unique Nash equilibrium (subgame perfect) leads to outcome Y

●​ The procedure results in a compromise outcome, not the top choice of either player

What is a subgame in an extensive game with perfect information. Explain how subgames
are identified and state the relationship between the number of nonterminal histories and
the number of subgames. Provide an illustration to support your explanation.

What is a subgame?

In an extensive game with perfect information, a subgame is a portion of the game that can be
viewed as a complete game starting from some decision point (history), without losing any of the
original structure.

Formally, a subgame:

●​ Starts at a nonterminal history (a decision node),

●​ Includes all future actions and outcomes from that point,
●​ Contains no “cut” information sets (this condition is automatically satisfied in perfect
information games, since each information set is a single node).

How to identify subgames

To identify subgames, follow a simple rule:

1.​ Pick any nonterminal history (node) in the game tree.

2.​ Consider the entire subtree starting from that node.
3.​ If the game has perfect information (which means no overlapping information sets), then:
Every such subtree is a valid subgame.

So in perfect information games, identifying subgames is straightforward:

Every decision node defines a subgame.

Relationship between nonterminal histories and subgames

Here is the key result: In an extensive game with perfect information, the number of subgames is
equal to the number of nonterminal histories.

Illustration:

●​ Nonterminal histories (decision nodes):

○​ Root node (Player 1)
○​ Node after action A (Player 2)
○​ Node after action B (Player 2)
●​ Subgames:
○​ Subgame starting at root → entire game
○​ Subgame starting after A → left subtree
○​ Subgame starting after B → right subtree

Total:

●​ Nonterminal histories = 3
●​ Subgames = 3

Backward Induction:

Backward induction is a method to solve finite extensive-form games with perfect information
by working from the end of the game back to the beginning.

Consider the following game:

Step1:

Step 2:
Step3:

Subgame Perfect Equilibrium (SPE)

A Subgame Perfect Equilibrium is a refinement of Nash equilibrium for extensive-form games.

Definition:​
A strategy profile is a Subgame Perfect Equilibrium if it constitutes a Nash equilibrium in every
subgame of the original game.

How to Find Subgame Perfect Equilibrium

The standard method is backward induction.

Steps:

1.​ Start from the last decision nodes of the game.

2.​ At each node, choose the action that gives the highest payoff to the player making the
decision.
3.​ Replace that node with the resulting payoff.
4.​ Move backward step by step.
5.​ Continue until the initial node is reached.

In extensive games with perfect information: Backward induction always produces a

Subgame Perfect Equilibrium.
EXERCISE 171.4 (Burning a bridge) Army 1, of country 1, must decide whether to attack army
2, of country 2, which is occupying an island between the two countries. In the event of an
attack, army 2 may fight, or retreat over a bridge to its mainland. Each army prefers to occupy
the island than not to occupy it; a fight is the worst outcome for both armies. Model this situation
as an extensive game with perfect information and show that army 2 can increase its subgame
perfect equilibrium payoff (and reduce army 1’s payoff) by burning the bridge to its mainland,
eliminating its option to retreat if attacked.

Model as an Extensive Game with Perfect Information

Players:

●​ Army 1 (country 1)
●​ Army 2 (country 2)

Sequence of moves:

1.​ Army 1 decides whether to Attack (A) or Not Attack (N).

2.​ If Army 1 attacks, then Army 2 decides whether to:
○​ Fight (F), or
○​ Retreat (R) across the bridge.

Preferences:

●​ Each army prefers occupying the island to not occupying it.

●​ A fight is the worst outcome for both.

Assign payoffs consistent with preferences:

●​ If Army 1 does not attack: Army 2 occupies island​

→ (Army 1, Army 2) = (0, 2)
●​ If Army 1 attacks and Army 2 retreats: Army 1 occupies island​
→ (2, 0)
●​ If Army 1 attacks and Army 2 fights: worst outcome​
→ (−1, −1)

Step 1: Solve without burning the bridge

Use backward induction.

At Army 2’s decision node (after attack):

●​ Compare payoffs:
 ○​ Fight → −1
○​ Retreat → 0
●​ Army 2 chooses Retreat (R)

Now Army 1 anticipates this:

●​ If Attack → outcome (2, 0)

●​ If Not Attack → outcome (0, 2)

So Army 1 chooses Attack (A)

Subgame Perfect Equilibrium (without burning bridge)

●​ Army 1: Attack
●​ Army 2: Retreat if attacked

Outcome: (2, 0)

Step 2: Burning the Bridge

If Army 2 burns the bridge:

●​ Retreat is no longer available

●​ Only option after attack is Fight

Solve again using backward induction

At Army 2’s node:

●​ Only option: Fight → (−1, −1)

Now Army 1 compares:

●​ Attack → (−1, −1)

●​ Not Attack → (0, 2)

So Army 1 chooses Not Attack (N)

Subgame Perfect Equilibrium (with burned bridge)

●​ Army 1: Not Attack

●​ Army 2: Fight (forced)

Outcome: (0, 2)
Conclusion

By burning the bridge:

●​ Army 2 removes its option to retreat

●​ This makes its threat to fight credible
●​ Army 1 is deterred from attacking

Ticktacktoe has subgame perfect equilibria in which the first player puts her first X in a corner. The
second player’s move is the same in all these equilibria. What is it?
1.​ If the first player places X in a corner, the optimal response for the second player is to place O in
the center.
2.​ The center is the most strategically important position because it is part of the maximum number
of winning lines.
3.​ Playing in the center prevents the first player from creating a fork (a position with two
simultaneous winning threats).
4.​ Any move other than the center allows the first player to gain a strategic advantage and
potentially force a win.
5.​ Therefore, in all subgame perfect equilibria, the second player’s move is always the center.
The ancient game of “Three Men’s Morris” is played on a ticktacktoe board. Each player has three
counters. The players move alternately. On each of her first three turns, a player places a counter on
an unoccupied square. On each subsequent move, a player may move a counter to an adjacent
square (vertically or horizontally, but not diagonally). The first player whose counters are in a row
(vertically, horizontally, or diagonally) wins. Find a subgame perfect equilibrium strategy of player
1, and the equilibrium outcome.

Subgame Perfect Equilibrium Strategy (Player 1)

In Three Men’s Morris, each player has three counters and first places them, then moves them to
adjacent squares. The objective is to form a straight line.

A subgame perfect equilibrium strategy for Player 1 is as follows. Player 1 begins by placing a
counter in the center square, which is the most strategically valuable position since it connects to
the maximum number of lines. On subsequent placement moves, Player 1 places counters in
positions that either contribute to forming a line or block Player 2 from creating one. After all
counters are placed, Player 1 adopts a defensive strategy: at every stage, she blocks any
immediate threat by Player 2 and avoids creating opportunities for Player 2 to form a line. If an
opportunity arises to complete a line without allowing a counter-response, Player 1 takes it;
otherwise, she prioritizes blocking.

Subgame Perfect Equilibrium Strategy (Player 2)

Player 2 responds optimally by also prioritizing the center if available; otherwise, choosing
positions that block Player 1’s potential lines. During the movement phase, Player 2 continuously
blocks Player 1’s attempts to form a line and avoids moves that allow Player 1 to create a fork or
immediate win.

Equilibrium Outcome

If both players follow these strategies, neither can force a win. Every attempt to form a line can
be countered by the opponent in subsequent moves. Thus, the game results in a draw. This
outcome is a subgame perfect equilibrium because at every stage (subgame), both players are
playing optimally and have no incentive to deviate.
Toetacktick is a variant of ticktacktoe in which a player who puts three marks in a line loses
(rather than wins). Find a strategy of the first-mover that guarantees that she does not lose. (If
fact, in all subgame perfect equilibria the game is a draw.)

1.​ The first player should start by placing her mark in the center.
2.​ After that, she follows a mirror (symmetry) strategy, playing in the square opposite to
the opponent’s move.
3.​ This keeps the board balanced and prevents the opponent from creating a forced situation.
4.​ She must always avoid completing three in a row herself, since that leads to losing.
5.​ By following this strategy, the first player guarantees that she does not lose, and the
game ends in a draw.

●​ Player 1 moves first: chooses D or U.

●​ Then Player 2 moves at each of the two nodes:
○​ After D: Player 2 chooses L or R
○​ After U: Player 2 chooses L or R

Payoffs:

●​ After D:
○​ L → (0,0)
○​ R → (3,1)
●​ After U:
○​ L → (2,5)
○​ R → (5,2)

Find All Subgames

A subgame starts at any decision node and includes all its successors.
Subgames are:

1.​ Entire game (starting at Player 1)

2.​ Subgame after D (Player 2 chooses L or R)
3.​ Subgame after U (Player 2 chooses L or R)

Solve Using Backward Induction

Subgame after D (Player 2 chooses):

●​ L → payoff to Player 2 = 0
●​ R → payoff to Player 2 = 1

Player 2 chooses R

So outcome after D → (3,1)

Subgame after U (Player 2 chooses):

●​ L → payoff to Player 2 = 5
●​ R → payoff to Player 2 = 2

Player 2 chooses L

So outcome after U → (2,5)

Player 1’s Decision

Now Player 1 compares:

●​ If D → payoff = 3
●​ If U → payoff = 2

Player 1 chooses D

Subgame Perfect Equilibrium (SPE)

Strategies:

●​ Player 1: D
●​ Player 2:
○​ After D → R
○​ After U → L
Final Answer

Subgames:

●​ Whole game
●​ Subgame after D
●​ Subgame after U

Subgame Perfect Equilibrium:

●​ Player 1 plays D
●​ Player 2 plays R after D, L after U

Equilibrium outcome:

●​ (3,1)

Game Structure

●​ Player 1 moves first: chooses C, D, or E

●​ Then Player 2 moves at each branch:

After C:

●​ F → (3,0)
●​ G → (1,0)

After D:

●​ H → (1,1)
●​ I → (2,1)
After E:

●​ J → (2,2)
●​ K → (1,3)

Subgames

Subgames are:

1.​ Entire game (starting at Player 1)

2.​ Subgame after C (Player 2 chooses F or G)
3.​ Subgame after D (Player 2 chooses H or I)
4.​ Subgame after E (Player 2 chooses J or K)

Backward Induction

Subgame after C (Player 2):

●​ F → payoff to Player 2 = 0
●​ G → payoff to Player 2 = 0

Player 2 is indifferent; both F or G are optimal

Subgame after D (Player 2):

●​ H → payoff = 1
●​ I → payoff = 1

Player 2 is indifferent; both H or I are optimal

Subgame after E (Player 2):

●​ J → payoff = 2
●​ K → payoff = 3

Player 2 chooses K

Step 4: Player 1’s Decision

Now consider Player 1’s payoffs:

●​ If C → payoff = 3 (if F) or 1 (if G)

●​ If D → payoff = 1 (if H) or 2 (if I)
●​ If E → payoff = 1 (since Player 2 chooses K)
Best outcome for Player 1 is:

→ C with F gives payoff 3

Subgame Perfect Equilibrium (SPE)

Strategies:

●​ Player 1: choose C
●​ Player 2:
○​ After C → F (or G, both optimal)
○​ After D → H or I (both optimal)
○​ After E → K

Final Answer

Subgames:

●​ Whole game
●​ Subgame after C
●​ Subgame after D
●​ Subgame after E

Subgame Perfect Equilibrium:

●​ Player 1: C
●​ Player 2:
○​ After C → F (or G)
○​ After D → H or I
○​ After E → K

Equilibrium outcome:

●​ (3,0) (if Player 2 chooses F after C)

In the ultimatum game:

Person 1 proposes a split of a fixed amount (say $1), offering x to person 2 and keeping 1−x
Person 2 either accepts (both receive the proposed amounts) or rejects (both receive 0)
Find the values of x for which there is a Nash equilibrium of the ultimatum game in which person 1
offers x.
Ultimatum Game – Nash Equilibrium

Game Description​
​ Player 1 proposes a split of 1 by offering x to Player 2 and keeping 1 − x.​
​ Player 2 observes x and chooses either Accept or Reject.

​
​ If Player 2 accepts, the payoffs are (1 − x, x).​
​ If Player 2 rejects, the payoffs are (0, 0).

Player 2’s Best Response​

​ If Player 2 accepts, she gets x.​
​ If Player 2 rejects, she gets 0.

Therefore:​
​ If x > 0, Player 2 prefers Accept since x > 0.​
​ If x = 0, Player 2 is indifferent between Accept and Reject since both give 0.

Player 1’s Best Response​

​ Player 1 knows Player 2’s behavior.​
​ If Player 2 accepts, Player 1 gets 1 − x.​
​ To maximize payoff, Player 1 wants to make x as small as possible.

Nash Equilibrium Analysis

Case 1: x > 0​
​ Player 2 will accept.​
​ However, Player 1 can deviate and offer a smaller positive amount x′ < x.​
​ This increases Player 1’s payoff.​
​ Hence, no Nash equilibrium exists for any x > 0.

Case 2: x = 0​
​ Player 2 is indifferent between Accept and Reject.​

Consider the strategy where Player 2 accepts. Then:​

Player 1 gets payoff 1, which is the maximum possible, so there is no incentive to deviate.​
Player 2 gets 0, which is the same as rejecting, so there is no incentive to deviate.

Thus, both players are best responding.

Conclusion​
The only value of x for which a Nash equilibrium exists is x = 0, with Player 2 accepting the
offer.

Final Answer​
The unique Nash equilibrium of the ultimatum game is that Player 1 offers x = 0 and Player 2
accepts. No other value of x can be sustained in equilibrium because Player 1 would always
prefer to offer a smaller amount.

(Subgame perfect equilibria of the ultimatum game with indivisible units) Find the subgame perfect
equilibria of the variant of the ultimatum game in which the amount of money is available only in
multiples of a cent.

Subgame Perfect Equilibria of the Ultimatum Game (Indivisible Units)

Consider the ultimatum game where the total amount is $1 and offers can only be made in
multiples of a cent (i.e., x ∈ {0, 0.01, 0.02, …, 1}).

Player 2’s Strategy​

Player 2 observes the offer x and decides whether to accept or reject.

●​ If x > 0, then accepting gives payoff x > 0, while rejecting gives 0.​
So Player 2 strictly prefers Accept.
●​ If x = 0, then Player 2 gets 0 whether she accepts or rejects, so she is indifferent.
Thus, in any subgame perfect equilibrium:

●​ Player 2 must accept every offer x ≥ 0.01

●​ At x = 0, Player 2 can either accept or reject

Player 1’s Optimization​

Player 1 anticipates Player 2’s behavior.

●​ Since Player 2 accepts all positive offers, Player 1 wants to minimize x

●​ The smallest positive amount available is 0.01

So Player 1 maximizes payoff by offering:

x = 0.01 → payoff = 0.99

Subgame Perfect Equilibrium

A subgame perfect equilibrium consists of:

●​ Player 1 offering x = 0.01

●​ Player 2:
○​ Accepts all offers x ≥ 0.01
○​ At x = 0, can accept or reject (off-path behavior)

Equilibrium Outcome

●​ Player 1 gets 0.99

●​ Player 2 gets 0.01

Final Answer

The subgame perfect equilibrium is that Player 1 offers the smallest positive amount (one cent),
and Player 2 accepts all positive offers. The equilibrium outcome is (0.99, 0.01).
Module – 4
Introduction to Bayesian Games

In Nash equilibrium, we assume that each player knows everything about the game, including
other players’ preferences and actions.

However, in real life, this is often not true. Players may not have complete information. For
example:

●​ A buyer may not know the seller’s true valuation

●​ Firms may not know competitors’ costs
●​ People may not know others’ intentions

To handle such situations, we use the concept of a Bayesian game.

A Bayesian game is a model where players have incomplete information about other players,
but they have beliefs (probabilities) about them.

Motivating Example (Simple Idea)

Suppose Player 1 is unsure about Player 2’s preferences.

There are two possibilities:

●​ Player 2 wants to meet Player 1

●​ Player 2 wants to avoid Player 1

Player 1 believes:

●​ Each case happens with probability 1/2

Player 2 knows her own type, but Player 1 does not.

Key Idea: Types

In Bayesian games, a player can have different types.

Here:

●​ Player 2 has two types:

1.​ Type 1: wants to meet
2.​ Type 2: wants to avoid

Player 1 does not know which type Player 2 is.

How Player 1 Makes Decisions

Player 1 forms beliefs about what each type of Player 2 will do.

Using these beliefs, Player 1 calculates expected payoff for each action.

Example:

●​ If Player 1 thinks:
○​ “meet” type plays B
○​ “avoid” type plays S

Then Player 1 calculates expected payoff using probabilities:​

Expected payoff = (1/2 × payoff in first case) + (1/2 × payoff in second case)

Nash Equilibrium in Bayesian Games

In this setting, equilibrium includes:

●​ One action for Player 1

●​ One action for each type of Player 2

So we treat:

●​ Each type of Player 2 as a separate decision-maker

A Nash equilibrium must satisfy:

1.​ Player 1 chooses the best action based on beliefs

2.​ Each type of Player 2 chooses the best action given Player 1’s action

Important Insight

Even though Player 2 knows her own type, we still consider strategies for all types because:

●​ Player 1 must form beliefs about each possible type

●​ These beliefs must be correct in equilibrium

Final Equilibrium Example

An equilibrium is:

(B, (B, S))

Meaning:

●​ Player 1 chooses B
●​ Player 2:
○​ If she wants to meet → chooses B
○​ If she wants to avoid → chooses S
This is equilibrium because:

●​ Player 1 is best responding to beliefs

●​ Each type of Player 2 is best responding to Player 1

Define Bayesian Game and its Nash Equilibrium. Also list out the components of
the Bayesian model with an example.

A Bayesian game is a game in which players have incomplete information about some aspects
of the game, such as other players’ preferences or types, but they have beliefs (probability
distributions) about these uncertainties.

A Bayesian game consists of the following components:

• A set of players​
• A set of states

For each player:​

• A set of actions​
• A set of signals that the player may receive, along with a signal function that assigns a signal
to each state​
• For each signal, a belief about the states consistent with that signal, represented as a
probability distribution over those states​
• A Bernoulli payoff function defined over pairs (a, ω), where a is an action profile and ω is a
state; the expected value of this function represents the player’s preferences over lotteries
involving such pairs

A Nash equilibrium of a Bayesian game is a Nash equilibrium of the strategic game (with vNM
preferences) defined as follows.

Players: The set of all pairs (i, tᵢ), where i is a player in the Bayesian game and tᵢ is one of the
signals (types) that player i may receive.

Actions: The set of actions available to each player (i, tᵢ) is the same as the set of actions
available to player i in the Bayesian game.

Preferences: The preferences of each player (i, tᵢ) are represented by a Bernoulli payoff
function, whose expected value determines the player’s preferences over lotteries.
Analyze a variant of the Battle of the Sexes (BoS) model where player 1 is unsure
whether player 2 wants to meet or avoid him and hence find its Nash equilibrium.

Game Description​
There are two players, Player 1 and Player 2. Player 1 is unsure whether Player 2
wants to meet or avoid him.

Bayesian Game Representation

Players​
The pair of players: Player 1 and Player 2.

States​
The set of states is
S = {meet, avoid}.

Actions​
Each player has two possible actions:
A = {B, S}.

Signals​
Player 1 receives a single signal z and cannot distinguish between the states:​
​ τ₁(meet) = τ₁(avoid) = z

Player 2 receives:​
​ m if she wants to meet​
​ v if she wants to avoid​
​ So, τ₂(meet) = m and τ₂(avoid) = v

Beliefs​
Player 1 assigns probability 1/2 to each state after observing signal z.​
Player 2 knows her type exactly:

●​ If she receives m → state is meet with probability 1

●​ If she receives v → state is avoid with probability 1
Payoffs The payoff values 𝑢𝑖(a,meet) for each player i under all possible action
combinations are shown in the left panel of the figure below, while the payoff values 𝑢𝑖
(a,avoid) are shown in the right panel of the figure below.

Nash Equilibrium

Since:

●​ Player 1’s best response to (B, S) is B

●​ Player 2’s best response to B is (B, S)

The Nash equilibrium is:

(B, (B, S))

Analyze a variant of the Battle of the Sexes (BoS) model where both players are unsure
whether the other player wants to meet or avoid them and hence find its Nash
equilibrium.

Game Description​
Both players are unsure whether the other player wants to meet or avoid them. Each player
knows their own type but not the other’s type.

Bayesian Game Representation

Players​
The pair of players: Player 1 and Player 2.

States​
The set of states is
S = {yy, yn, ny, nn},
where:​
​ ​ yy: both want to meet​
​ ​ yn: Player 1 wants to meet, Player 2 wants to avoid​
​ ​ ny: Player 1 wants to avoid, Player 2 wants to meet​
​ ​ nn: both want to avoid

Actions​
Each player has two possible actions:
A = {B, S}.

Signals​
​ Player 1 receives one of two signals: y1 or n1​
​ ​ τ1(yy) = τ1(yn) = y1​
​ ​ τ1(ny) = τ1(nn) = n1

Player 2 receives one of two signals: y2 or n2​

​ ​ τ2(yy) = τ2(ny) = y2​
​ ​ τ2(yn) = τ2(nn) = n2

Beliefs​
​ Player 1 assigns:

●​ Probability 1/2 to each of yy and yn after receiving y1

●​ Probability 1/2 to each of ny and nn after receiving n1

Player 2 assigns:

●​ After y2: P(yy) = 2/3, P(ny) = 1/3

●​ After n2: P(yn) = 2/3, P(nn) = 1/3
Payoffs The payoffs 𝑢𝑖(a, ω) of each player i for all possible action pairs and states are given in
the following figure.

Nash Equilibrium

((B, S), (B, S))

All players are best responding given their beliefs, and no one has an incentive to deviate.
Given:

1
●​ Player 1 chooses T with probability 2
1
●​ Player 1 chooses B with probability 2
1
●​ There are two states 𝑤1​and 𝑤2​, each occurring with probability 2
1
●​ 0 ≤ ε ≤ 2

We calculate the expected payoff of Player 2 for each action L, M, and R.

Expected Payoff from L

In state 𝑤1​:

●​ If Player 1 chooses T, Player 2 payoff is 2ε

●​ If Player 1 chooses B, Player 2 payoff is 2

1 1
Expected payoff in state 𝑤1​: 2
(2ε) + 2
(2) = ε + 1

In state 𝑤2​:

●​ If Player 1 chooses T, Player 2 payoff is 2ε

●​ If Player 1 chooses B, Player 2 payoff is 2

1 1
Expected payoff in state 𝑤2​: 2
(2ε) + 2
(2) = ε + 1

1 1
Now combine both states: 𝑈2(𝐿) = 2
(ε + 1) + 2
(ε + 1)
= (ε + 1)
Expected Payoff from M

In state 𝑤1​:

●​ If Player 1 chooses T, Player 2 payoff is 0

●​ If Player 1 chooses B, Player 2 payoff is 0

1 1
Expected payoff in state 𝑤1​: 2
(0) + 2
(0) = 0

In state 𝑤2​:

●​ If Player 1 chooses T, Player 2 payoff is 3ε

●​ If Player 1 chooses B, Player 2 payoff is 3

1 1 3ε + 3
Expected payoff in state 𝑤2​:
2
(3ε) + 2
(3) = 2

1 1 3ε + 3
Now combine both states:
2
(0) + 2
( 2
)

3ε + 3
𝑈2(𝑀) = 4

Expected Payoff from R

In state 𝑤1​:

●​ If Player 1 chooses T, Player 2 payoff is 3ε

●​ If Player 1 chooses B, Player 2 payoff is 3

1 1 3ε + 3
Expected payoff in state 𝑤1​:
2
(3ε) + 2
(3) = 2

In state 𝑤2​:

●​ If Player 1 chooses T, Player 2 payoff is 0

●​ If Player 1 chooses B, Player 2 payoff is 0

1 1
Expected payoff in state 𝑤2​: 2
(0) + 2
(0) = 0

1 3ε + 3 1
Now combine both states:
2
( 2
)+ 2
(0)
 3ε + 3
𝑈2(𝑅) = 4

Comparison of Expected Payoffs

𝑈2(𝐿) = 1 + ε

3ε + 3
𝑈2(𝑀) = 4

3ε + 3
𝑈2(𝑅) = 4

3ε + 3
Compare L and M: 1 + ε> 4

Multiply by 4:

4 + 4ε > 3ε + 3

​ ​ ​ 1 + ε >0

Since,​ ​ ε≥0

the inequality is always true.

Thus:

𝑈2(𝐿) > 𝑈2(𝑀) = 𝑈2(𝑅)

Therefore, Player 2’s best response is:

Now determine Player 1’s best response.

If Player 2 chooses L:

●​ If Player 1 chooses T, payoff = 1

●​ If Player 1 chooses B, payoff = 2
Since:

2>1

Player 1 prefers:

Final Nash Equilibrium

(B,L)

with equilibrium payoffs:

(2,2)

Find The Nash Equilibrium of​

Players: Player 1 and Player 2

{L,R}

States

There are three states:

●​ State α
●​ State β
●​ State γ

Player 1 cannot distinguish between the three states (single information set).

Player 2:
 ●​ Cannot distinguish between β and γ
●​ Knows when the state is α

Probabilities:

P(β) = 3/4​
P(γ) = 1/4

Strategies

Each player chooses either:

L or R

Player 2’s Expected Payoffs

Suppose Player 1 chooses L.

If Player 2 chooses L:

●​ In β: payoff = 2
●​ In γ: payoff = 2

Expected payoff:

U₂(L) = (3/4)(2) + (1/4)(2)

= 3/2 + 1/2

=2

If Player 2 chooses R:

●​ In β: payoff = 0
●​ In γ: payoff = 0

Expected payoff:

U₂(R) = 0

Thus, if Player 1 chooses L, Player 2 prefers L.

Suppose Player 1 chooses R.

If Player 2 chooses L:

●​ In β: payoff = 0
●​ In γ: payoff = 0

Expected payoff:

U₂(L) = 0

If Player 2 chooses R:

●​ In β: payoff = 1
●​ In γ: payoff = 1

Expected payoff:

U₂(R) = (3/4)(1) + (1/4)(1)

= 3/4 + 1/4

=1

Thus, if Player 1 chooses R, Player 2 prefers R.

Player 1’s Best Responses

If Player 2 chooses L:

●​ In state α:
○​ L → 2
○​ R → 3

So Player 1 prefers R.

If Player 2 chooses R:

●​ In state α:
○​ L → 0
○​ R → 1

Again, Player 1 prefers R.

Thus, R strictly dominates L for Player 1.

Nash Equilibrium

Since:

●​ Player 1 chooses R
●​ Best response of Player 2 to R is R

the Nash equilibrium is:

(R, R)

Final Answer

The Nash equilibrium of the game is: (R, R) with equilibrium payoff: (1, 1)

Develop a model in case of Cournot's game with imperfect information

about cost.

Players Firm 1 and firm 2.

States {L, H}.

Actions Each firm’s set of actions is the set of its possible outputs (nonnegative
numbers).

Signals

Firm 1’s signal function τ1 satisfies τ1(H) = τ2(L) (its signal is the same in both
states);

Firm 2’s signal function τ2 satisfies τ2(H) = τ2(L) (its signal is perfectly informative
of the state).

Beliefs The single type of firm 1 assigns probability θ to state L and probability
1− θ to state H.

Each type of firm 2 assigns probability 1 to the single state consistent with its
signal.

Payoff functions The firms’ Bernoulli payoffs are their profits; if the actions
chosen are (q1, q2) and the state is I (either L or H) then
firm 1’s profit is q1(P(q1 + q2) − c) and

firm 2’s profit is q2(P(q1 + q2) − cI ),

where P(q1 + q2) is the market price when the firms’ outputs are q1 and q2.

Develop a model in case of ‘Cournet’s game with imperfect information’

about both cost and information.

Players Firm 1 and firm 2.

States {L0, L1, H0, H1}, where the first letter in the name of the state indicates
firm 2’s cost and the second letter indicates whether (1) or not (0) firm 1 knows
firm 2’s cost.

Actions Each firm’s set of actions is the set of its possible outputs (nonnegative
numbers).

Signals

Firm 1 gets one of the signals 0, L, and H, and her signal function τ1 satisfies
τ1(L0) = τ1(H0) = 0,
τ1(L1) = L, and
τ1(H1) = H.

Firm 2 gets the signal L or H and her signal function τ2 satisfies

τ2(L0) = τ2(L1) = L and

τ2(H0) = τ2(H1) = H.
Beliefs

Firm 1: type 0 assigns probability θ to state L0 and probability 1 − θ to state H0;

type L assigns probability 1 to state L1; type H assigns probability 1to state H.

Firm 2: type L assigns probability π to state L1 and probability 1− π to state L0;

type H assigns probability π to state H1 and probability 1 − π to state H0.

Payoff functions The firms’ Bernoulli payoffs are their profits; if the actions
chosen are (q1, q2), then firm1’sprofit is q1(P(q1 +q2)−c) and firm 2’s profit is
q2(P(q1 +q2)−cL) in states L0 and L1, and q2(P(q1 +q2)−cL) in states H0 and
H1.
Module 5
Find the Nash equilibrium of finitely repeated Prisoner's Dilemma.

Finitely Repeated Prisoner’s Dilemma – Nash Equilibrium

Consider a Prisoner’s Dilemma repeated for a finite number T of periods. In each

period, both players simultaneously choose between:

●​ C : Cooperate
●​ D : Defect

The one-shot Prisoner’s Dilemma payoff matrix is:

C D

C (R,R) (S,T)

D (T,S) (P,P)

where:

T>R>P>S

In the one-shot Prisoner’s Dilemma, Defect (D) is the dominant strategy for both
players because defecting gives a higher payoff regardless of the opponent’s
action.

Backward Induction Analysis

To find the Nash equilibrium of the finitely repeated game, we use backward
induction.

Consider the final period, period T.

Since there are no future rounds after period T, players cannot reward
cooperation or punish defection later. Therefore, period T becomes a one-shot
Prisoner’s Dilemma.

Hence, both players choose:

(D,D)

in period T.

Now consider period T−1.

Both players know that period T will end with (D,D) regardless of what happens
in period T−1. Therefore, cooperation in period T−1 cannot affect future
outcomes.

Thus, period T−1 also reduces to a one-shot Prisoner’s Dilemma, and both
players defect.

The same reasoning applies repeatedly to every earlier period:

●​ Period T: both defect

●​ Period T−1: both defect
●​ Period T−2: both defect
●​ …
●​ Period 2: both defect
●​ Period 1: both defect

Therefore, by backward induction, rational players defect in every round.

Nash Equilibrium

The Nash equilibrium strategy is:

“Choose Defect in every period, irrespective of previous history.”

Thus, the unique Nash equilibrium of the finitely repeated Prisoner’s Dilemma is:

(D,D) in every period.

Subgame Perfect Equilibrium

Since the strategy prescribes optimal actions after every possible history of play,
it is also a Subgame Perfect Equilibrium.

Therefore, the unique Subgame Perfect Nash Equilibrium is:

(D,D)

in all rounds of the game.

Conclusion

In a finitely repeated Prisoner’s Dilemma, cooperation cannot be sustained

because the final period has no future consequences. By backward induction,
defection in the last period implies defection in all earlier periods. Hence, rational
players defect in every round.

Final Answer:

The unique Nash equilibrium and Subgame Perfect Equilibrium of the finitely
repeated Prisoner’s Dilemma is:

(D,D) in every period of the game.

Infinitely Repeated Prisoner’s Dilemma with Tit for Tat Strategy – Nash
Equilibrium

Consider the Prisoner’s Dilemma played infinitely many times. In each round,
both players choose between:

●​ Quiet (Q)
●​ Confess (C)

The payoff matrix is:

 Player 2: Player 2:
Quiet Confess

Player 1: (2,2) (0,3)

Quiet

Player 1: (3,0) (1,1)

Confess

In the one-shot Prisoner’s Dilemma, Confess is the dominant strategy because it

gives a higher payoff irrespective of the opponent’s action.

However, in an infinitely repeated game, future interactions become important.

Players may cooperate today to obtain larger future benefits.

Tit for Tat Strategy

The Tit for Tat strategy is defined as follows:

●​ In the first round, play Quiet.

●​ In every subsequent round, copy the opponent’s previous action.

Thus:

●​ If the opponent played Quiet in the previous round, play Quiet.

●​ If the opponent played Confess in the previous round, play Confess.

Analysis of Tit for Tat

Suppose both players follow Tit for Tat.

Round 1:

Both players choose Quiet.

Outcome:
(Q,Q)

Payoffs:

(2,2)

Since both players cooperated, Tit for Tat instructs both players to continue
cooperating.

Therefore, in every future round:

(Q,Q)

is played.

Each player receives payoff 2 in every period.

Total payoff:

2 + 2 + 2 + ...

Thus, mutual cooperation is sustained.

Incentive to Deviate

Now check whether a player can gain by deviating.

Suppose Player 1 deviates in one round.

Instead of choosing Quiet, Player 1 chooses Confess.

Outcome:

(C,Q)

Player 1 obtains payoff 3 in that round.

However, Tit for Tat causes Player 2 to retaliate in the next round by choosing
Confess.

The next outcome becomes:

(C,C)
with payoff:

(1,1)

Thus, deviation gives a short-term gain but leads to lower future payoffs because
of retaliation.

Therefore, when players value future payoffs sufficiently, deviation is not

profitable.

Nash Equilibrium

Under Tit for Tat:

●​ Both players start by playing Quiet.

●​ Cooperation continues as long as no player defects.
●​ Any deviation is punished by future defection.

Hence, neither player has an incentive to deviate from the cooperative path.

Therefore, the Nash equilibrium strategy is:

“Both players play Tit for Tat.”

Equilibrium path:

(Q,Q)

in every round.

Conclusion

In the infinitely repeated Prisoner’s Dilemma, future punishment makes

cooperation possible. Under Tit for Tat, players cooperate initially and then
imitate the opponent’s previous move. Since defection triggers future
punishment, sustained cooperation can be maintained.

Final Answer:

The Nash equilibrium in the infinitely repeated Prisoner’s Dilemma using Tit for
Tat strategy is:
Both players follow Tit for Tat, leading to:

(Q,Q)

in every round of the game.

Infinitely Repeated Prisoner’s Dilemma with Grim Trigger Strategy (Limited

Punishment) – Nash Equilibrium

Consider the infinitely repeated Prisoner’s Dilemma in which, in every round,

each player chooses between:

●​ Quiet (Q)
●​ Confess (C)

The payoff matrix is:

Player 2: Player 2:
Quiet Confess

Player 1: (2,2) (0,3)

Quiet
 Player 1: (3,0) (1,1)
Confess

In the one-shot Prisoner’s Dilemma, Confess is the dominant strategy because it

yields a higher payoff regardless of the opponent’s action.

However, in an infinitely repeated game, future consequences affect present

decisions.

Grim Trigger Strategy with Limited Punishment

The Grim Trigger strategy with limited punishment is defined as follows:

●​ In the first round, play Quiet.

●​ Continue playing Quiet as long as both players have always played Quiet.
●​ If any player defects by playing Confess, punish the defection by playing
Confess for a limited number of periods.
●​ After the punishment period ends, return to playing Quiet.

Thus, cooperation is rewarded, and deviation is followed by temporary

punishment rather than permanent punishment.

Analysis of the Strategy

Suppose both players follow Grim Trigger with limited punishment.

Initially:

Both players choose Quiet.

Outcome:

(Q,Q)

Payoffs:

(2,2)

Since no player deviates, cooperation continues.

Thus, every round produces:

(Q,Q)

with payoff:

(2,2)

for both players.

Incentive to Deviate

Suppose Player 1 deviates in some period.

Instead of choosing Quiet, Player 1 chooses Confess.

Outcome:

(C,Q)

Player 1 obtains:

in the current round.

However, Player 2 observes the deviation.

According to Grim Trigger with limited punishment, Player 2 responds by playing

Confess for a fixed number of future periods.

During punishment rounds, outcomes become:

(C,C)

with payoff:

(1,1)

After the punishment period ends, players return to:

(Q,Q)
and cooperation resumes.

Thus, deviation gives:

●​ a short-run gain of 3 instead of 2

●​ but causes several future rounds of lower payoff 1 instead of 2.

If the future loss from punishment exceeds the immediate gain from deviation,
deviation is not profitable.

Therefore, when players sufficiently value future payoffs, cooperation can be

sustained.

Nash Equilibrium

Under Grim Trigger strategy with limited punishment:

●​ Both players begin by playing Quiet.

●​ Cooperation continues while no player defects.
●​ Any deviation triggers temporary punishment.
●​ After punishment, cooperation resumes.

Since deviation causes future losses, neither player has an incentive to deviate.

Hence, the Nash equilibrium strategy is:

“Both players follow Grim Trigger with limited punishment.”

Equilibrium path:

(Q,Q)

in every round except during punishment phases following deviations.

Conclusion

In the infinitely repeated Prisoner’s Dilemma, Grim Trigger with limited

punishment can sustain cooperation. Players cooperate initially, punish
deviations temporarily by defecting for a fixed number of periods, and then return
to cooperation. Because future punishment discourages deviation, cooperative
behavior can be maintained as an equilibrium.
Final Answer:

The Nash equilibrium of the infinitely repeated Prisoner’s Dilemma with Grim
Trigger strategy and limited punishment is:

Both players follow Grim Trigger with limited punishment, resulting in sustained
cooperation:

(Q,Q)

along the equilibrium path.