ProblemBank203 2

Professor B.
Douglas Bernheim
Economics 203
Winter 2013-2014
Problems
Strategic Environments
1. Consider the following game. There are two players, 1 and 2. They will
play either
• matching pennies version A (as defined in class, where 1 moves first),

• matching pennies version A, modified so that player 2 moves first
and player 1 moves second, or
• matching pennies version B.
“Nature” selects randomly among these alternatives; each game is chosen

with equal likelihood. Nature’s choice is observed by both players.
(a) Draw the extensive form of this game.

(b) Determine the strategy sets for each player in this game. (You are
not asked to draw the normal form for this game, as it is extremely
large.)
2. Consider the following game. There are two players, 1 and 2. Player 1
moves first, and names either “Heads”, “Tails”, or “Middle”. If he chooses
either “Heads” or “Tails”, the game ends with probability 0.5. If it ends
and player 1 has said “Heads,” payoffs are (3,2) (the convention here is
that player 1’s payoff is listed first, and player 2’s payoff is listed second).
If it ends and player 1 has said “Tails,” payoffs are (4,3). If the game
does not end, then player 2 gets to make a choice. The game necessarily
terminates after this choice. If player 1 has named “Heads”, player 2 can
choose either “Up” yielding payoffs (9,6), “Down” yielding payoffs (3,1),
or “Sideways” yielding payoffs (1,5). Player 2 cannot tell the difference
between player 1 choosing “Tails” and player 1 choosing “Middle”. In
either case, his alternatives are to choose either “Up” or “Backwards”.
If player 1 has chosen “Tails” then “Up” yields (1,1) and “Backwards”
yields (2,6). If player 1 has chosen “Middle” then “Up” yields (1,1) and
“Backwards” yields (6,2).

(b) Determine the strategy sets for each player in this game.
(c) Draw the normal form for this game.
1
3. Consider a game in which a particular player has N information sets,
indexed by n = 1, 2, ... Suppose that he has Mn possible actions at infor-
mation set n. How many strategies does he have in all?
Dominance, Rationalizability, and Pure Strategy

Nash Equilibrium
4. Consider the following game. Two players, 1 and 2, bargain to determine
the split of $v. Bargaining works as follows. Player 1 offers a split, s,
which must take oone of the following three values: 0, v/2, or v. Player
2 observes this offer, and either accepts it or rejects it. If 2 accepts the
offer, 1 gets a payoff of s, and 2 gets a payoff of v-s. If 2 rejects the offer,
both players get 0, and the game ends.

(b) Determine the strategy sets for each player in this game.
(c) Draw the normal form for this game.
(d) Identify any strictly dominated strategies. What is left after iterative
deletion of strictly dominated strategies?
(e) Identify any weakly dominated strategies.
5. Consider the following game in normal form, where player 1 chooses rows
and player 2 chooses columns (in each cell, player 1’s payoff is listed first
and player 2’s second).
a b c d e
A 2,1 4,2 1,0 10,3 2,4
B 6,10 2,20 0,10 7,15 0,18
C 5,0 0,0 3,3 5,0 0,1
D 8,1 2,4 0,0 5,1 4,2
E 1,3 2,6 0,4 8,5 1,3
(a) Does either player have strictly dominated strategies? If so, what are
they? Does either player have weakly dominated strategies? If so,
what are they?
(b) What strategies are eliminated through iterative deletion of strictly
dominated choices? Is the game dominance solvable?
(c) Completely characterize the set of pure strategy Nash equilibria for
this game.
(d) Completely characterize the set of rationalizable strategies for this
game.
2
6. Consider the following bargaining situation. Two individuals are consider-
ing undertaking a business venture that will earn them $100 in profit, but
they must agree how to split the $100. Bargaining works as follows: The
two individuals simultaneously make a demand (which you can assumer
to be a nonnegative number). If their demands sum to more than $100,
then they fail to agree and each gets nothing. If their demands sum to
less than $100, they do the project, each gets his demand, and the rest
goes to charity (which neither values).
(a) What are each player’s strictly dominated strategies?

(b) What are each player’s weakly dominated strategies?
(c) What are the pure strategy Nash equilibria of this game?
7. In class, we constructed a pure strategy Nash equilibrium of the centipede

game in which player 1 says “stop” at the initial node. Prove that player
1 says “stop” at the root node in every pure strategy Nash equilibrium.
8. Consider a finite game, and suppose that it is dominance solvable. Prove

that it has a unique Nash equilibrium and that this coincides with the
iterated dominance solution.
9. Consumers are uniformly distributed along a boardwalk of length 1 mile.

The price of ice-cream is fixed by a regulator, so consumers go to the
nearest firm because they dislike walking (assume that at the regulated
price all consumers will purchase an ice-cream cone even if they have to
walk a full mile). If more than one vendor is at the same location, they
evenly split the business.
(a) Consider a game where two ice-cream vendors pick their locations
simultaneously. Show that there exists a unique pure strategy Nash
equilibrium and that it involves both vendors locating at the mid-
point of the boardwalk.
(b) Show that with three vendors, no pure strategy Nash equilibrium
exists.
10. Consider the Cournot oligopoly model with inverse demand function P (Q),
where all firms have identical cost functions C(q). Suppose that there are
N firms. Suppose moreover that the functions P and C are both differ-
entiable, that C is strictly increasing and convex, and that P is strictly
decreasing and concave. Prove: there exists a unique Nash equilibrium
for this game, and that this equilibrium is symmetric (all firms name the
same quantity).
3
11. Imagine that consumers are located uniformly around a circle of unit cir-
cumference. There are J firms at locations that are equally spaced around
the circumference, each of which sells beanie babies. All firms have the
same constant production technology, and produce beanie babies at con-
stant unit cost c. Each consumer wants at most 1 beanie baby and derives
a gross benefit of v from its consumption. The total cost of buying a
beanie baby from firm i for a consumer located a distance d from firm i
is pi + td2 , where t is a transportation cost parameter. Assume that firms
compete by simultaneously naming price. Determine the symmetric Nash
equilibrium prices and sales levels. You may assume that v is sufficiently
large to assure that all customers purchase a beanie baby in equilibrium.
When happens as J grows large? As t falls to zero?
12. Consider the following trading game between two players, 1 and 2. First,
nature selects one of three states, ω 1 , ω 2 , or ω 3 , with probabilities p,
p, and (1 − 2p) , respectively, where p ∈ (0, 12 ). Each player imperfectly
observes the resulting state as follows. Player 1 knows for sure whether
or not state ω 1 occurred, but cannot distinguish between states ω 2 and
ω 3 . Player 2, on the other hand, knows for sure whether or not state ω 2
occurred, but cannot distinguish between states ω 1 and ω 3 .Second, both
players simultaneously select whether to “trade” or “not trade”, after
which the game ends and payoffs take place as follows. If at least one
player selects “not trade” payoffs are zero for each player. If both players
select “trade” payoffs depend on the particular state of nature and are
given by the matrix below (which is NOT the normal form of the game).
state player 1 player 2
ω1 1 −2
ω2 −2 1
ω3 4 4
That is, in state ω 1 player 1 benefits from trade but player 2 is hurt, in
state ω 2 player 2 benefits from trade but player 1 is hurt, and in state s3
both benefit.
(a) Identify the sets of strictly and weakly dominated pure strategies for
each player.
(b) Suppose p ∈ (0, 52 ]. Identify the set of pure strategy Nash equilibria
that do not involve the use of weakly dominated strategies. What are
the equilibrium payoffs, states in which trade occurs, and expected
total surplus (defined as the sum of player’s payoffs)?
(c) Suppose .p ∈ ( 25 , 12 ). Identify the set of pure strategy Nash equilibria
that do not involve the use of weakly dominated strategies. What
are the equilibrium payoffs, states in which trade occurs, and total
surplus?
4
(d) Ignoring strategic considerations, what are the socially optimal strat-
egy profiles for each level of p, i.e., those that maximize total surplus?
What are the states in which trade occurs and expected total surplus
under such strategies? Do these socially optimal strategies constitute
an equilibrium? Provide intuition.
13. Consider a weekly work-consumption-savings plan of a household consist-

ing of two individuals. Each of the individuals, i = 1, 2, decides on his/her
own work level wi and level of consumption of a frivolous good fi . Assume
that wi ∈ [0, 1600] and fi ∈ [0, 5000]. Given their work and consumption
decisions, the amount of money the household can put in the bank (the
household’s savings) is given by b = (w1 + w2 ) − (f1 + f2 ).
(a) Suppose that both members of the household have the same utility
function which is given by
u(w1 , w2 , f1 , f2 ) = 2 min{b, 500} − (w1 + w2 ) + 0.1(f1 + f2 )
i. Demonstrate that all (wi , fi ) with wi > 0 and fi > 0 are strictly
dominated for player i. What other strategies, if any, are strictly
dominated for player i?
ii. Which pure strategies survive iterated elimination of strictly
dominated strategies?
iii. Identify the pure strategy Nash equilibria of this game.
(b) Continue to assume that the payoff of player 1 is as above, but that
the payoff of player 2 is instead
u2 (w1 , w2 , f1 , f2 ) = 2 min{b, 500 + ε} − (w1 + w2 ) + 0.1(f1 + f2 )
for ε > 0.
i. What is the set of pure strategy Nash equilibria? What is the set
of strategies for each player that survives iterated elimination of
strictly dominated strategies?
ii. Explain what happens in the limit as ε approaches 0. Consider
the correspondence which for each ε gives the set of Nash equilib-
ria in part (d). Is this an upper hemicontinuous correspondence?
14. The governorship of Freedonia is up for grabs. Freedonia has a large class
of political insiders (politicians). Only politicians can hold office. Each
politician is characterized by a quality level q ∈ [0, 1]. For simplicity,
assume there are infinitely many politicians of every quality level. A
politician can choose to run for office at a cost k. All politicians know
each other well, and in particular know each other’s quality level. Voters,
however, do not know the quality of any politician. Consequently, if N
5
politicians decide to run for office, each wins with probability N1 . (Note:
do NOT deviate from this simplifying assumption. In particular, do NOT
treat the voters as players in the game; the election here is simply a lottery
with equal probability on all candidates.) If the governor turns out to
be of quality q, non-candidates (politicians who did not run) receive a
payoff of q, while losing candidates receive a payoff of q − k. The winner,
however, receives a payoff of q − k + s, where s > 0 represents the rents
from holding office (consisting of salary and ego-rents). Politicians decide
simultaneously whether to run. If no one runs, there is no governor and
anarchy prevails, so that all politicians receive a payoff of −∞.
Some Freedonians argue that setting higher compensation, s, and reducing
the costs of running for office, k, will attract more high quality candidates
and thereby improve the quality of government. Others insist that those
measures will only attract riff-raff. You have been hired to determine
which position is theoretically correct.
(a) How does one describe a strategy for a politician of quality q? What
is the strategy space for that politician?
(b) We will begin by looking for pure strategy equilibria with a single
candidate.
i. Write down a condition representing the requirement that no
non-candidate wishes to enter against a single candidate of qual-
ity q. Must any further conditions be met to have a single-
candidate equilibrium?
ii. Using what you have shown so far, provide a necessary and suf-
ficient condition for the existence of a single-candidate equilib-
rium. Your condition should involve only s and k. (Hint:
Think about completing the following sentence: “There exists
an equilibrium with one candidate if and only if there exists an
equilibrium with one candidate and q =...”)
iii. Assuming that your necessary and sufficient condition is satisfied,
how is the quality of the governor in the best (highest quality)
and worst (lowest quality) single-candidate equilibrium affected
by an increase in the governor’s compensation, s? By an increase
in the costs of running for office, k?
(c) Next we look at pure strategy equilibria with multiple candidates.
i. Write down a condition representing the requirement that no
non-candidate wishes to enter against N candidates of average
quality q.
ii. Write down a condition representing the requirement that no
candidate wishes to drop out of an election with N candidates.
Write the condition in terms of q, the average quality of the
candidates, qmin , the quality of the worst candidate, s, k, and
N . (Remember that you can write an expression for average
quality without a candidate of quality q 0 in terms of q 0 and q.)
6
iii. Using your answers to parts (i) and (ii), derive a necessary and
sufficient condition (in terms of k, s, and N alone) for the exis-
tence of a pure strategy N -candidate equilibrium. (Hint: Start
by completing the following sentence: “There exists a set of N
candidates with average quality q satisfying the candidate incen-
tive constraint if and only if, with the same N and q, such a set
also exists for qmin =...” Then complete the following sentence:
“There exists a set of N candidates with average quality q satis-
fying the non-candidate incentive constraint if and only if such
a set exists for q =...”).
(d) Using your answers to (b-iii) and (c-iii), prove that a pure stategy
equilibrium exists for all k. (Do NOT try to invoke one of our stan-
dard theorems, as this is a setting where the set of players is uncount-
ably infinite.) Explain how the number of candidates varies with k
and s, indicating when that number is not unique.
(e) Using what you’ve shown so far, explain what happens both to the
number of candidates and to the range of average candidate qual-
ity as k → 0 (barriers to running are minimized) and as s → ∞
(compensation becomes very large).
15. For the first portion of this question, imagine that two firms, n = 1, 2,
compete in an industry, selling a perfectly homogenous product. Firm n
chooses quantity qn ≤ 0, which it produces at a cost of cqn . The firms
make these quantity choices simultaneously, and all output is then sold at
the price p = a − b(q1 + q2 ).
So far, this is just the linear Cournot model. Heres the twist. There are
two investors, A and B. Investor A owns the share α1 ≤ 0.5 of firm 1
and the share 1 − α2 ≤ 0.5 of firm 2; investor B owns the complementary
shares. Important: assume that each firm acts in the interests of its
majority shareholder (so that firm 1 maximizes the payoff to investor A
and firm 2 maximizes the payoff to investor B).
(a) Derive each firm’s best response function.

(b) Solve for the Nash equilibrium quantities.
(c) For the special case where α1 = α2 = α, how do equilibrium quan-
tities vary with the value of α? Interpret your answer and explain
intuitively why the effect operates in the direction it does.
(d) Use your formulas to identify the solution when α1 = 0.5 and α2 >
0.5. Interpret your answer and explain intuitively why the firms
produce the resulting quantities.
(e) Identify all Nash equilibria for the case where α1 = α2 = 0.5. Again,
interpret your answer and explain intuitively why the firms produce
the resulting quantities.
7
Now imagine that there are N firms and N investors. Investor n owns the
1−α
share α of firm n and the share N −1 of each other firm.
(f) Derive each firm’s best response as a function of the total production
of other firms.
(g) Solve for symmetric Nash equilibrium quantities.
(h) What ownership structure (that is, value of α) will yield the same
outcome with N firms as a conventional Cournot equilibrium with
M < N firms?
Mixed Strategy Nash Equilibria

16. Recall the game described in question 5. Completely characterize the set
of mixed strategy Nash equilibria for this game.
17. Consider the following game in normal form, where player 1 chooses rows,
player 2 chooses columns, and player 1’s payoff is listed first.
a b
A u, v m, n
B w, x y, z
Suppose that u > w, y > m, n > v, and x > z.
(a) Prove that there does not exist a pure strategy Nash equilibrium for
this game.
(b) Solve for the mixed strategy Nash equilibria in terms of the param-
eters.
18. Firms compete to sell some homogeneous good. There are N firms and Q
consumers. Each firm has a fixed capacity K, and can produce quantity
up to K at zero cost (production beyond quantity K is impossible). Each
consumer has a reservation value of v: if p < v, the consumer buys one
unit, but if p > v, the consumer buys nothing. Assume that N K > Q,
and that (N − 1)K < Q. Assume that firms are Bertrand competitors in
the sense that they simultaneously announce prices. Consumers purchase
the good from the firm with the lowest price, splitting equally among the
lowest-priced firms in the event of a tie.
Prove that a pure strategy Nash equilibrium does not exist. Compute a
symmetric mixed strategy Nash equilibrium. How do the properties of the
equilibrium change with N ?
8
19. Consider the following war game between two nations: A and B. Initially,
nature chooses if A is “strong” or “weak;” each option occurs with equal
probability, and only A observes this choice. After nature’s choice, A
decides either to “attack” nation B, or to “not attack.” If A chooses
“not attack” the game ends, and payoffs are (0,10) (the first entry is the
payoff for A). If A chooses “attack,” then B has three options: “retaliate,”
“yield,” and “destroy the world.” The following matrix provides payoffs
for the cases in which A chooses to “attack” (in each cell, A’s payoff is
listed first). Note: this matrix is not the normal form of the game.
If A is “strong” If A is “weak”
If B plays “retaliate” 6, −6 −6, 10
If B plays “yield” 10, 0 10, 0
If B plays “destroy the world” −100, −100 −100, −100

(b) Derive the normal form of this game.
(c) Find the set of mixed strategy Nash equilibria (recall that a pure
strategy is a special case of a mixed strategy).
player 2 chooses columns, and player 1’s payoff is listed first:
LL L M R
U 100, 2 −100, 1 0, 0 −100, −100
D −100, −100 100, −49 1, 0 100, 2
(a) If you were player 2 in this game and you were playing it once without
the ability to engage in preplay communication with player 1, what
strategy would you choose?
(b) What are all the Nash equilibria (pure and mixed) of this game?
(c) Is your strategy choice in part A a component of any Nash equilibrium
strategy profile? Is it a rationalizable strategy?
(d) Suppose now that preplay communication were possible. Would you
expect to play something different from your choice in A?
21. Two risk neutral firms, i = 1, 2, produce a homogeneous product at a

unit cost c > 0. There is a single consumer who buys either one unit of
the good or nothing. The consumer has a reservation value v > c, and
resolves indifference in favor of purchasing at the price p = v. Suppose
the firms simultaneously decide whether to announce prices pi ≥ 0. If
a firm chooses to quote a price, it incurs a cost k > 0 (e.g. to pay a
saleperson to visit the buyer), but it has the option not to quote a price
(saving the cost k). Suppose 0 < k < v − c. The consumer buys the good
at the lowest available price (provided it does not exceed v), and chooses
between the firms with equal probabilities in the event of a tie.
9
(a) Does this game have a pure strategy Nash equilibrium? Justify your
answer.
(b) Consider a symmetric mixed strategy equilibrium in which each firm
names a price, incurring the cost k, with probability π ∈ (0, 1); con-
ditional on naming a price, the firm prices according to the atomless
CDF F . Explain why F puts no weight on prices below c + k or
higher than v. Solve for F and π as functions of p, k, v, and c
(you may assume, without proof, that the support of F is [c + k, v]).
What does the distribution function F look like?
22. Two risk netural firms, i = 1, 2, produce a homogeneous product at a unit

cost c > 0. There are M = N + 2K consumers, each of whom buys one
unit of the good or nothing. Each consumer has a reservation value v > c,
and resolves indifference in favor or purchasing at the price p = v. N
consumers are willing to purchase from either firm, and choose the firm
with the lowest price (splitting equally if the prices are the same). K
consumers are loyal to each firm. They but from that firm when it’s price
doesn’t exceed v, and never buy from the other firm. Assume throughout
that N, K > 0.
(a) In any Nash equilibrium (pure or mixed), does each firm earn zero
profits, or strictly positive profits? Justify your answer.
(b) Is there a pure strategy Nash equilibrium for this game? If so,
describe it, and explain why it’s an equilibrium. If not, why not?
(c) Consider a symmetric mixed strategy equilibrium in which each firm
randomly selects a price based on an atomless CDF F . You may
assume, without proof, that the support of F is some interval [p0 , v]
for p0 < v. Solve for F as a function of p, c, v, N , and K. What
does F look like? How does the mixed strategy equilibrium change
as K K
N → 0? As N → ∞?
23. Consider an asymmetric all-pay auction between two bidders whose values
are common knowledge. The object is worth v to Bidder 1 and v 0 to
Bidder 2 where v 0 > v. All bidders simultaneously submit sealed-bids bi
which must be non-negative; the highest bidder wins the object and all
bidders pay their bids. In the event of a tie, the good is given to Bidder
2.
(a) What bidding strategies are strictly dominated? Can you iteratively
eliminate any more strictly dominated strategies?
(b) Are there any pure-strategy Nash equilibria? If so, characterize the
set of PSNE. If not, explain why not.
We now consider a mixed-strategy Nash equilibrium where both bid-
ders randomize.
10
(c) At a MSNE, would Bidder 1 ever bid strictly higher than v with
positive probability? Would Bidder 2 ever bid strictly higher than v
with positive probability?
(d) Consider a MSNE where Bidder i randomly selects a bid based on
the CDF Fi . You can assume (without proof) that the support of
Fi is [0, v].
i. Write an expression for 2’s payoff from bidding x in [0, v], fixing
F1 .
ii. Use the fact that x = v is in the support of Bidder 2’s strategy
to solve for 2’s payoff.
iii. Use answers from the last two parts to solve for F1 (x). What is
F1 (0)?
i. Write an expression for 1’s payoff from bidding x in [0, v], fixing
F2 . Use the fact that x = v is in the support of Bidder 1’s
strategy to solve for 1’s payoff.
ii. Use answers from the previous part to solve for F2 (x). What is
F2 (0)?
(e) What is the expected revenue? As v 0 varies in the set (v, ∞), is
expected revenue increasing, constant, or decreasing in v 0 ?
24. Consider a race between 2 Stanford Economics graduate students—1 and

2—to become the next Social Chair of the Economics Grad Students Asso-
ciation. The only way to become the Social Chair is to bribe the ex-Social
Chair; the graduate student who offers the biggest bribe receives the im-
mense prestige of the position, which carries a utility v. Therefore, after
paying a bribe of bi , if Student i is selected, his utility is v − bi , and oth-
erwise, his utility is simply −bi . The bribes are paid regardless of who
is selected. If both students offer the same bribe, the ex-Social Chair
randomly picks one of the two with probability 21 .
(a) What bribing strategies are strictly dominated? Can you iteratively
eliminate any more strictly dominated strategies?
(b) Are there any pure-strategy Nash equilibria? If so, characterize the
set of PSNE. If not, explain why not.
Consider a mixed-strategy Nash equilibrium (MSNE) where the stu-
dents randomize.
(c) At a MSNE, would a student ever pay a bribe strictly more than v
with positive probability?
Consider an MSNE where students randomly select a bribe according
to the CDF F . You can assume that the support of F is [0, v] (you
do not have to prove this).
(d) Write an expression for i’s payoff from paying bribe bi in [0, v] con-
ditional on the other bribing according to F .
11
(e) Use the fact that bi = v lies in the distribution of i’s choices to solve
for i’s payoff.
(f) Use answers from the last two parts to solve for F (bi ).
(g) How much money does the ex-Social Chair expect to make from all
the bribes?
Now imagine that Economics Department Chair imposes a spending
cap on the bribes of b̄ ∈ v2 , v so that b1 , b2 ≤ b̄. We shall look

for an MSNE of the following form: a student bids exactly b̄ with

probability α, and with probability (1 − α), he bids according to cdf
G whose support is [0, b0 ].
(h) Write an expression for i’s payoff from paying a bribe bi < b0 , and
from paying a bribe b̄.
(i) Use the fact that bi = 0 lies in the support to solve for i’s payoff.
(j) Use answers from the last two parts to solve for α, G (bi ), and b0 .
(k) How much money does the ex-Social Chair expect to make from the
bribes now? Did the spending cap increase, maintain constant, or
decrease the total amount of bribing?
25. For years, Professor Bernheim has been running an illicit all-pay auction
in Econ 203, in which he sells a $10 bill to poor unsuspecting first-year
students. (Remember the structure of an all-pay auction: everyone sub-
mits a bid; the object goes to the highest bidder, with the winner chosen
at random with equal probabilities in the event of ties; all bidders pay
their bids regardless of whether they win or lose.) The Department Chair
has become worried that Bernheim is taking advantage of the students.
He allows Bernheim to continue to run his auction, but imposes a “bid
cap,” bmax , on allowable bids, so that no student can bid more than bmax .
Throughout this problem, assume that bmax ∈ (5, 10) and that students
are risk-neutral. Let N ≥ 2 denote the number of first-year students.
(a) What is the pure strategy set for each player? What is the mixed
strategy set?
(b) Is there a pure strategy Nash equilibrium of the all-pay auction with
the bid cap? (Be sure to argue carefully why or why not.)
(c) Consider the case where N = 2. Find a non-degenerate mixed strat-
egy Nash equilibrium of this auction with the following structure:
each student bids bmax with probability α and with probability (1−α)
bids according to an atomless cdf F whose support is [0, b0 ]. (Hint:
throughout, remember the mixed strategy indifference condition.)
(d) According to theory, does the Chair’s intervention reduce the amount
of revenue Bernheim raises from the students? Justify your answer
completely, and explain intuitively.
12
26. The Department Chair despairs of successfully regulating Bernheim’s all-
pay auction (described in the last problem), and gives up. In the new
unregulated environment, Bernheim tries to increase his profits by impos-
ing a “bid floor,” bmin , on allowable bids, so that no student can bid less
than bmin . Regardless of the floor, a student can always choose not to bid,
in which case she simply receives a payoff of zero; let φ denote “no-bid.”
Let N ≥ 2 denote the number of first-year students. Assume students
are risk-neutral.
(a) What is the pure strategy set for each player? What is the mixed
strategy set?
(b) Is there a pure strategy Nash equilibrium of the all-pay auction with
the bid floor? (Be sure to argue carefully why or why not.)
(c) Explain (ideally prove) why, in a symmetric equlibrium, F cannot
have an atom at either bmin or 10.
(d) Now suppose N = 2. Find a symmetric mixed strategy equilibrium
involving an atomless cdf F whose support is [bmin , 10], along with a
probability π for φ. (Hint: throughout, remember the mixed strategy
indifference condition.)
(e) According to theory, does Bernheim’s gambit increase the amount of
revenue he raises from the students? Justify your answer completely,
and explain intuitively.
27. Palo Alto Elementary school is holding a science fair for its students.
N students are enrolled at the school. To encourage participation, the
principal decides to offer a prize d, which will be rewarded randomly to one
of the entrants. Each student must pay an entry fee K, with d > K > 0,
to participate in the science fair.
(a) Identify the students’ strategy sets.

(b) Characterize all of the pure-strategy Nash equilibria. How many
students enter the science fair?
(c) Let N =2. Are there any mixed-strategy Nash equilibria, other than
the ones you identified in part 2? If so, characterize them. If not,
explain why not.
Some of the teachers worry that students won’t work hard enough on
their science fair projects and suggest giving the prize to the best project
instead of rewarding it randomly. Assume students can choose to put any
effort e ≥ 0 into their project and that the quality of the projects are
strictly increasing in effort. (So the student that puts in the most effort
wins the science fair.) The prize is split in case of ties. There is a constant
13
marginal cost c to exerting effort. Entry and effort decisions are made
simultaneously, so the number of entrants is not known when choosing an
effort level. (Consider the case of any N > 1 students for the following
questions.)
(d) Identify the students’ strategy sets.

(e) Are any strategies strictly dominated? Can you iteratively eliminate
any more strictly dominated strategies?
(f) Are there any pure-strategy Nash equilibria? If so, characterize them.
If not, explain why not.
(g) Consider a symmetric mixed-strategy equilibrium in which students
enter with probability π ∈ (0, 1) and choose effort level according to
an atomless cdf F with support [e, e].
i. What are expected payoffs in such an equilibrium?
ii. Given your answer to part (i), what is e? Can we have e > d−K
c ?
d−K
Can we have e < c ?
iii. Write an expression for the cdf F . What is e?
iv. With what probability do students enter the science fair?
v. How does the probability of entry change as N gets large?
Equilibrium Refinements for Games in Normal

Form
28. Consider the following three player game, in which player 1 chooses rows,
player 2 chooses columns, and player 3 chooses boxes (within each cell,
player 1’s payoff is listed first, player 2’s second, and player 3’s third).
If player 3 chooses A:
L R
U 1, 1, 1 1, 0, 1
D 1, 1, 1 0, 0, 1
If player 3 chooses B:
L R
U 1, 1, 0 0, 0, 0
D 0, 1, 0 1, 0, 0
(a) Calculate the set of pure strategy Nash equilibria.

(b) Does any pure strategy Nash equlibrium involve a weakly dominated
strategy?
(c) Is every pure strategy Nash equilibrium trembling-hand perfect?
14
Correlated Equilibria
player 2 chooses columns, and player 1’s payoff is listed first.
a b c
A 5, 3 5, 5 3, 4
B 4, 10 9, 9 4, 11
C 3, 3 11, 4 5, 5
Completely characterize the set of correlated equilibria for this game. Are
there any correlated equilibria that are not Nash equilibria? Explain your
answer. (Hint: is the game dominance solvable?)
30. Consider the following two-player game in normal form (where player 1
selects rows and player 2 selects columns, and where, in each cell, player
1’s payoff is listed first):
L R
T 6, 6 2, 7
B 7, 2 0, 0
(a) Derive the set of Nash equilibrium payoffs.

(b) Draw the convex hull of such payoffs.
(c) Is there any correlated equilibrium that yields payoffs outside such
convex hull?
31. Consider a game with N players sitting in a circle. At first, nature selects
a card for each player with an integer number between 1 and M . The
random process is independent for each player, and each card has a positive
probability of occurring. Each player observes only his card, after which
he decides upon one of two actions: “trade” or “not trade”. All these
choices are simultaneous, after which the game ends. Payoffs are as follows.
Whenever a player selects “not trade”, he receives the number on his card
as a payoff. Whenever a player selects “trade”, he receives the number on
the card of the player closest to his right of those who selected “trade”
(notice that this player is himself if nobody else selected “trade”.)
Derive the set of mixed strategy Nash equilibria. Derive the set of corre-
lated equilibrium outcomes.
Static Games of Incomplete Information

32. Consider the following strategic situation. Two opposed armies are poised
to seize an island. Each army’s general can choose either “attack” or “not-
attack.” In addition, each army is either “strong” or “weak” with equal
15
probability (the draws for each army are independent), and an army’s type
is known only to its general. Payoffs are as follows: The island is worth
M if captured. An army can capture the island either by attacking when
its opponent does not or by attacking when its rival does if it is strong
and its rival is weak. If two armies of equal strength both attack, neither
captures the island. An army also has a “cost” of fighting, which is s if it
is strong and w if it is weak, where s < w. There is no cost of attacking
if its rival does not. Identify all pure strategy Bayesisn Nash equilibria of
this game.
33. Consider the linear Cournot model. Now, however, suppose that each firm
has probability q of having unit costs of c and (1 − q) of having costs d,
where d > c. Solve for the Bayesian Nash equilibrium.
34. The Chair of the Economics Department would like to improve the quality
of life around the Landau building by turning the third floor into a jazz
club. He has decided to finance this project through private contributions
from students. Moreover, he has decided that he will only do the project
if every student contributes $10.
Suppose each student i’s valuation for the jazz club is independently drawn
from a uniform distribution on [0, 45]. Eachs tudent observes his or
her own valuation, but not that of other students. There are N ≥ 2
risk neutral students, all of whom decide to contribute (y) or not (n)
simultaneously. If all choose y, the project goes forward and no money
is returned to the students. If one or more chooses n, the project does
not go forward; below you are asked to consider two variants which differ
according to whether the money is returned in this case. Throughout,
assume each student resolves indifference in favor of contributing o the
project.
(a) Suppose each student’s contribution is returned to him if the project

does not go forward. Are there Bayesian Nash equilibria in which the
project definitely does not go forward? If so, describe one. Is there
a Bayesian Nash equilibrium in which the project goes forward with
some positive probability? Is there more than one such equilibrium?
(b) Suppose the Chair keeps each student’s contribution even if the project
does not go forward (he uses it to fund an espresso machine for the
faculty, which the students don’t care about). Suppose for the mo-
ment that N = 2 (it’s a small class).
i. Is there a Bayesian Nash equilibrium in which the project never
goes forward? If so, describe it. If not, explain.
ii. Are the symmetric Bayesian Nash equilibra in which the project
goes forward with some probability? If so, describe them. If
not, explain.
16
iii. Are there any asymmetric Bayesian Nash equilibria in which the
project goes forward with some probability? If so, describe
them. If not, explain.
(c) Repeat (i) and (ii) from part (b), for the case where N > 2. As N
gets large, what happens to the robability the project goes forward?
You may restrict attention to symmetric equilibria.
35. Imagine an economy consisting of two consumers, labeled 1 and 2. Con-

sumer i chooses xi in [0, 1]. Consumer i’s preferences are given by:
x2j θi
Ui (xi , xj ) = xi − + ti
2
(j not equal to i), where ti is a transfer from the government. Nature
selects each θi randomly and independently from the interval [1, 2]. The
distribution of each θi over the interval [1, 2] is given by some CDF, F .
(a) Suppose that consumers choose the xi simultaneously and indepen-

dently. Assume there are no transfers. Solve for a Bayesian Nash
equilibrium. Is it unique?
(b) Suppose that the government can observe the preference parameters
θi (i = 1, 2), and that it seeks to maximize U1 + U2 . Setting transfers
equal to zero, compute the first best allocation as a function of these
parameters.
(c) Now suppose that the government cannot observe the preference pa-
rameters. Design a mechanism (wherein each consumer announces
its preference parameter, and the government determines actions and
transfers as functions of these announcements) satisfying the follow-
ing two properties: (1) the mechanism implements the first-best allo-
cation in dominant strategies, and (2) the mechanism always results
in a balanced budget (t1 = −t2 ) for all possible announcements.
Note: Do not give a general formula; give the specific mechanism
that works for this particular problem.
36. Lilo and Tim are in the woods picking berries. They separately see a genie
that gives each of them a box that contains some money. The genie tells
Lilo that Tim also got such a box. Lilo is then informed how much money
her box contains, but not how much Tim’s box contains. Similarly, the
genie gives Tim a box and tells him that Lilo has received a box too. The
genie also reveals Tim the amount in his box, but not in Lilo’s. Moreover,
the genie tells each of them that the amount in the other box can be
represented by a random variable which is uniformly distributed between
0 and 100 (as an approximation, we will assume that units of money can
be represented by a continuous random variable). The genie tells Tim and
17
Lilo that they have to play a game where they will make simultaneous bids.
If the sum of their bids is less than or equal to the actual total amount
contained in the boxes, they get half of the sum of their bids, otherwise the
money disappears and they each get zero. Describe the players’ strategy
sets, and completely describe one Bayesian Nash Equilibrium.
37. Two individuals are bidding for a single object. The value of the object,
v, is the same for both of them (this is a “common value” auction), but
uncertain. Before bidding, the individuals receive signals, s1 for bidder
1 and s2 for bidder 2, each of which is distributed uniformly on [0, 1].
These signals are private; each bidder observes his own signal, but not his
opponent’s. The value of the item is v = s1 + s2 for each bidder. Assume
for the entire problem that the bids have to be non-negative.
Suppose the two bidders compete in a second-price sealed bid auction.
That is, the one with the higher bid wins the item, and pays the bid of his
opponent. In the case of ties, suppose a winner is randomly chosen with
equal probabilities and the winner pays the tying bid. The loser does not
pay for anything.
(a) Describe a bidder’s pure strategy set. Give an example of a pure

strategy.
(b) Give an example of a bidder’s strategy that is weakly dominated.
Show why. Is this strategy also strictly dominated? Show why.
(c) Suppose bidder 2 adopts the bid function b2 (s) = αs. Solve for
bidder 1’s expected payoff as a function of bidder 1’s signal and bid.
(d) Using the expression derived in part (c), determine bidder 1’s best
bid as a function of bidder 1’s signal.
(e) You have learned that, in a second price auction, it is weakly domi-
nant for players to bid their valuations. How do you reconcile that
conclusion with your answer to part (d)?
(f) Using your answer to part (d), solve for a symmetric Bayesian Nash
equilibrium with linear bid functions – that is, an equilibrium in
which both bidders use the same bid function, b(s) = αs.
(g) Show there is a continuum of asymmetric equilibria. (Hint: If the
bid function b1 (s1 ) = αs1 is a best response to the bid function
b2 (s2 ) = βs2 , under what conditions is the bid function b2 (s2 ) = βs2
a best response to the bid function b1 (s1 ) = αs1 ?)
(h) Consider the asymmetric equilibrium above where bidder 1’s equilib-
rium strategy is b1 (s1 ) = αs1 . For bidder 1 with signal s1 , compute
his (expected) equilibrium payoff π 1 (s1 ), as a function of α. How
does it vary with α? Provide intuition for the direction of this effect.
18
38. Imagine that your final exam has been returned to you with a numerical
grade in the set {0, 1, ..., 10}. Unfortunately, your TA has forgotten to
record your name next to your numerical grade; all that your TA knows is
the distribution of grades, and he makes the distribution public. For each
grade from 0 to 10, there are 5 people who received that grade. The TA
asks you to disclose your grade to him by returning the exam to him; in
the event that you do not disclose your grade, he will give you the average
of all those grades that remain unmatched to any name. (For example,
if the undisclosed grades are {0, 2, 3, 3}, each of those 4 individuals would
receive 2.) The game is as follows:
Stage 1: You observe your grade.
Stage 2: You decide whether to reveal your grade or not. Revelation is
necessarily truthful.
Stage 3: The TA awards a grade to you. The TA is a machine, and not a
player, and will mechanically award you the disclosed grade if you return
the exam to him, or the average grade of all those who have not disclosed
their exams.
You would like to act in a manner that would get you the highest (ex-
pected) grade possible.
(a) Is there an equilibrium where every individual discloses her grade?

If so, describe why it is an equilibrium, and if not, show why not.
(b) Is it rationalizable for a student who receives 10 to not disclose her
grade?
(c) Is there any pure strategy equilibrium where individuals with strictly
positive grades choose not to disclose their grades? If so, describe
why it is an equilibrium, and if not, show why not.
(d) Is there any mixed strategy equilibrium where individuals with strictly
positive grades do not disclose their grades with positive probability?
39. Two friends, Thelma and Louise, are starting a new technology company in
Thelma’s garage. The success of the company will depend on the amount
of the time each spends working on it. Their efforts are complementary, so
effort by one increases the other’s marginal product. However, both have
other attractive ways to spend their time. The payoff of player i (where i
equals either T for Thelma or L for Louise) is given by:
U (ti , tj ) = f (h − ti ) + g(ti , tj ), (1)
where ti denotes the number of hours spend on the company, and h denotes
the total time budget, so that h−ti is leisure time. The function f captures
19
the utility from leisure, and the function g captures the payoff from the
company. We will assume in what follows that:
f (h − ti ) = (h − ti ) − β(h − ti )2 (2)
and
g(ti , tj ) = αti (1 + tj ) (3)
You should assume throughout this problem that the values of all param-
eters are consistent with interior solutions.
(a) Let E(tj ) denote the expected value of tj , given i0 s information. Solve
for i0 s optimal choice.
(b) Suppose Thelma and Louise share the same value of α. Solve for the
symmetric Nash equilibrium.
(c) Suppose that Thelma and Louise have different values of α, call them
αT and αL . Solve for the Nash equilibrium. (Suggestion: check to
see whether your formulas coincide with those in part 2 for the case
where αT = αL .)
(d) Now suppose that, for both Thelma and Louise, the parameter α
takes on the value αH with a probability λ and the value αL with
probability 1 − λ. The realizations are independent. Each of them
knows the value of their own parameter, but not the value of the
other’s parameter. Solve for the symmetric Bayesian Nash equilib-
rium.
40. Party S (for seller) owns an umbrella, to which he attaches a value of c (for
the “cost” of parting with it). Party B (for buyer) would like the umbrella,
to which she attaches a values of v. If v > c, there are potential gains
from trade. S and B will try to negotiate a deal through the following
procedure. Simultaneously, S will name a price, a ≥ 0 (the “ask”), and
B will name a price, b ≥ 0 (the “bid”). if b ≥ a, the transaction will take
place: B will pay S the average of the bid and the ask, b+a 2 , in return for
the umbrella. If b ≤ a, no transition will occur.
(a) Assume the game involves complete information (both players know
v and c). Also assume v ≥ c.
i. Characterize all pure strategy Nash equilibria (if any) in which
trade occurs. (Throughout, when the question ask you to charac-
terize a set of equilibria, it means that you should identify all the
strategy profiles and outcomes for equilibria in that set.) Justify
your answer completely.
ii. Characterize all pure strategy Nash equilibria (if any) in which
trade does not occur. Justify your answer completely.
(b) Still assuming the game involves completely information, now assume
v < c.
20
i. Characterize all pure strategy Nash equilibria (if any) in which
trade occurs. Justify your answer completely.
ii. Characterize all pure strategy Nash equilibria (if any) in which
trade does not occur. Justify your answer completely.
(c) Now we will introduce incomplete information. The parameters v and
c are drawn from independent uniform distributions on the interval
[0, 1]. S observes c but not v; B observes v but not c. The steps
below walk you through the solution of a Bayesian Nash equilibrium
wherein the strategies are linear. Specifically, S will adhere to an
asking strategy of the form a = δ + αc, while B will adhere to a
bidding strategy of the form b = γ + βv. Our object is to solve for
α, β, γ, and δ.
i. Suppose S learns that c takes on a particular values, and believes
B is adhering to a bidding strategy b = γ + βv.
A. Write the S’s payoff as a function of his ask α, his valuation
c, and B’s valuation v, given B’s bidding strategy (and re-
membering that no trade occurs unless the bid exceeds the
ask).
B. Using your answer to (i), write an expression for the expected
payoff to S when asking α, given his valuation c and B’s
bidding strategy.
C. Maximize the expression you derived in (ii), to determine
S’s optimal ask for each value of c. [Hint: we are solving for
an equilibrium in linear strategies, so check that S’s optimal
policy function is linear.] (Notice that you are recovering
implied values of α and δ as functions of β and γ)
D. The optimal policy function you solved for in part (iii) may
have the property that the ask is less than c for some values
of c in the interval [0, 1]. Does that make sense? [Hint: what
is the support of the bid distribution?]
ii. Suppose B learns that v takes on a particular value, and believes
S is adhering to the asking strategy α = δ + αc.
A. Write the B’s payoff as a function of his bid b, his valua-
tion v, and S’s valuation c, given S’s asking strategy (and
remembering that no trade occurs unless the bid exceeds the
ask).
B. Using your answer to (i), write an expression for the expected
payoff to B when bidding b, given his valuation v and S’s
asking strategy.
C. Maximize the expression you derived in (ii), to determine
B’s optimal bid for each value of v. [Same hint as above.]
(Notice that you are recovering implied values of β and γ.)
D. The optimal policy function you solved for in part (iv) may
have the property that the bid is greater than v for some
21
values of v in the interval [0, 1]. Does that make sense? [Hint:
what is the support of the ask distribution?]
iii. Use your answers in parts (a−iii) and (b−iii) to solve for α, β, γ,
and δ.
iv. Finally, we will identify the efficiency loss from asymmetric in-
formation.
A. Using a graph, show the values of c and v for which there
are gains from trade.
B. Solve for the values of c and v for which trade occurs. Show
those values on the same graph.
C. Compute the probability that trade occurs, both overall and
conditional upon the existence of gains from trade. (Hint:
the distribution of c and v is uniform, so you can do this just
by computing areas in your picture.)
Subgame Perfect Nash Equilibria

41. Two firms, E (for entrant) and I (for incumbent) play the following game.
First, E chooses “In” or “Out”. If E chooses “Out” then payoffs are
(0, 2) (the convention here is that E’s payoff is given first). If E chooses
“In”, then E and I play a simultaneous move game where each chooses
to either “fight” (F ) or “accommodate” (A). (F, F ) yields payoffs of
(−3, −1), (F, A) yields payoffs of (1, −2), and (A, F ) yields payoffs of
(−2, −1). In the event that the firms choose (A, A) at this stage, the
game continues. Each firm simultaneously chooses either “here” (H) or
“there” (T ). Payoffs after these choices are given as follows:
H T
H 3, 1 0, 0
T 0, 0 x, 3

(b) Solve for the subgame perfect equilibria of this game as a function of
x. Be sure to consider mixed strategies.
42. Consider a legislature constituted of three (3) individuals, A, B, and C.

The legislature is considering a bill. If the bill doesn’t pass, payoffs are
(0, 0, 0) (where A’s payoff is listed first, B’s second, and C’s third). If the
bill does pass, payoffs are (1, 1, −1). The outcome is determined by a vote,
where each player indicates “for” or “against” (“abstain” is not a choice).
(a) Suppose that players cast their votes simultaneously.

i. Identify the (pure) strategy sets for each player.
ii. Draw the extensive form of the game.
22
iii. Depict the normal form of the game. (Hint: with three players,
you can’t depict the normal form in a two dimensional matrix.
Remember, however, that the normal form is defined as a map-
ping from strategy profiles into payoff vectors. The object here
is simply to depict that mapping.)
iv. Identify all of the pure strategy Nash equilibria for this game.
For each, indicate whether the bill passes.
v. Identify all of the pure strategy Nash equilibria for which no
player chooses a weakly dominated strategy. For each, indicate
whether the bill passes.
vi. Modify the game as follows. Before voting, players must decide
simultaneously whether to show up. Only those showing up can
vote. Showing up involves some very tiny cost, ε > 0. Players
can observe who has shown up before voting. A strict majority
is required for passage (ties defeat the bill). Is there a pure
strategy subgame perfect Nash equilibrium? Justify your answer
completely.
(b) Suppose that players cast their votes sequentially in the order A, B,
C, and that they can observe the votes cast by those who vote before
them.
i. Identify the strategy sets for each player.
ii. Draw the extensive form of the game.
iii. Reasoning directly from the extensive form, is there a pure strat-
egy subgame perfect Nash equilibrium in which the bill passes?
If so, identify one. If not, explain why not.
iv. Reasoning directly from the extensive form, is there a pure strat-
egy subgame perfect Nash equilibrium in which the bill does not
pass? If so, identify one. If not, explain why not.
43. Consider a legislature consisting of four members – three legislators (i =

1, 2, 3) who vote on proposals, and one ”agenda setter” who can’t vote.
There are three possible public projects (also indexed i = 1, 2, 3). Project
i provides a benefit bi > 0 to legislator i at a cost ci > 0 to all other
players (both of the other legislators and the agenda setter). Assume
c1 > c2 > c3 . Benefits and costs are additive over projects. So, for
example, when projects 1 and 2 are adopted, legislator #1’s payoff is
b1 − c2 , legislator #2’s payoff is b2 − c1 , legislator #3’s payoff is −c1 − c2 ,
and the agenda setter’s payoff is −c1 − c2 . Assume that bi > cj + ck for
j, k 6= i, which means that adopting all three projects is mutually beneficial
for the three legislators (compared with adopting nothing). Obviously,
the agenda setter would like to block all of these projects. We can describe
an outcome here is a triplet, z = (z1 , z2 , z3 ) where zi ∈ {0, 1}, with zi = 1
indicating that project i is undertaken, and zi = 0 indicating that it is
23
not undertaken. For example, (0, 1, 0) indicates that only project #2 is
undertaken. Notice that there are eight possible outcomes.
(a) Suppose the legislature works as follows. We start with some status
quo, z. The agenda setter makes a proposal, z 0 . The legislators
vote either for z or z 0 . A majority carries the day; if two or more
legislators vote for z, then z is implemented; if two or more vote for
z 0 , then z 0 is implemented.
For each of the eight possible configurations of the status quo z, solve
for the pure strategy subgame perfect Nash equilibrium in which
the legislators vote “sincerely” (that is, each one votes for his or
her preferred alternative). Provide a table listing, for each of the
eight possible status quos, the proposals that would pass, and the
outcome.
(b) Now imagine the legislature operates in two rounds. We start with
some initial status quo, z1 . The agenda setter makes a proposal z 0 .
The legislators vote either for z1 or z 0 . Let’s call z2 the victor from
this first round (it’s either z1 or z 0 ). This becomes the status quo
for the second round. The agenda setter now makes a new proposal,
z 00 . The legislators vote either for z2 or z 00 . The victor in this round
is implemented.
For each of the eight possible configurations of the initial status quo,
z1 , solve for the pure strategy subgame perfect Nash equilibrium in
which legislators vote “sincerely” (that is, each legislator votes for his
or her preferred alternative, ranking alternatives by the outcomes to
which they ultimately lead, given your answer to part (a)). Provide
a table listing, for each of the eight possible initial status quos, the
resulting outcome, and the proposal or proposals that the agenda
setter might make in equilibrium during the first round.
(c) Now imagine the legislature operates in three rounds. We start with
some initial status quo, z1 . The agenda setter makes a proposal z 0 .
The legislators vote either for z1 or z 0 . Let’s call z2 the victor from
this first round (it’s either z1 or z 0 ). This becomes the status quo
for the second round. The agenda setter now makes a new proposal
z 00 . The legislators vote either for z2 or z 00 . The victor, call it z3
(which is either z2 or z 00 ) becomes the status quo for the third round.
The agenda setter makes another proposal, z 000 . The legislators vote
either for z3 or z 000 . The victor in this round is implemented.
For each of the eight possible configurations of the initial status quo,
z1 , solve for the pure strategy subgame perfect Nash equilibrium in
which legislators vote ”sincerely” (that is, each legislator votes for his
or her preferred alternative, ranking alternatives by the outcomes to
which they ultimately lead, given your answer to part b). Provide
a table listing, for each of the eight possible initial status quos, the
24
resulting outcome, and the proposal or proposals that the agenda
setter might make in equilibrium during the first round.
(d) We could, of course, continue adding rounds. A student claims:
“With three or more rounds, there is always an equilibrium in which
the agenda setter proposes (0, 0, 0) in the first round, everyone votes
for it, and it’s the final outcome (it isn’t changed in subsequent
rounds). This is true even if the initial status quo is (1, 1, 1), despite
the fact that all three legislators prefer (1, 1, 1) to (0, 0, 0).” True or
false? Explain your answer.
44. Two firms, A and B, are in a market that is declining in size. The game
starts in period 0, and the firms can compete in periods 0,1,2,3,... (i.e.
indefinitely) if they so choose. Duopoly profits in period t for firm A are
equal to 101-10t and they are 10-t for firm B. Monopoly profits (which are
earned if a firm is the only one left in the market) are 500-25t for firm A
and 50-2t for firm B.
Suppose that at the start of each period the two firms each must decide
to “stay in” or to “exit” if it is still active (they do so simultaneously if
both are still active). Once a firm exits, it is out of the market forever and
earns zero in each period thereafter. Firms maximize the (undiscounted)
sum of their profits.
Characterize the subgame perfect Nash equilibria for this game.
45. Consider a game played by three agents: a parent (P ), and two children
(1 and 2). Suppose that each player simultaneously takes some action
(ai ≥ 0, i = P, 1, 2) that affects the earnings of every family member
(yj ≥ 0, j = P, 1, 2). In particular, earnings are given as follows:
a2P a2 a2
yP = fP (aP , a1 , a2 ) = 6 + aP − − 1− 2
6 6 6
a2P a2 a2
y1 = f1 (aP , a1 , a2 ) = 2 + a1 − − 1− 2
6 6 6
2 2
a a a2
y2 = f2 (aP , a1 , a2 ) = 2 + a2 − P − 1 − 2
6 6 6
In other words, by increasing ai , party i can increase his or her own
earnings, but imposes a negative externality on the earnings of others.
Unless stated otherwise, assume that the utility of each party is just equal
to that party’s income, and that actions (ai ) do not affect utility directly.
(a) Assuming that a central planner can make lump sum redistributions
to achieve any desired distribution of resources across the three par-
ties, what are the socially optimal actions? What is the resulting
total income for the family unit?
25
(b) Assume that the parties independently and simultaneously choose
their actions. Do players have dominant strategies? If so, what are
they? What is the Nash equilibrium? Is it unique? What is the total
income of the family unit in equilibrium?
(c) Imagine that the government can impose a tax/subsidy scheme to ad-
dress the externality. What is the optimal corrective tax (known as a
Pigouvian tax)? Verify your answer by recalculating the equilibrium.
(d) Now imagine that the parent, P , cares about both children. In par-
ticular, suppose that the parent’s utility function is
1 1
u(cP , c1 , c2 ) = ln cP + ln c1 + ln c2
2 2
where ci is the consumption of party i. Consumption may differ
from income because the parent is allowed to make monetary trans-
fers: cP = yP − t1 − t2 , and ci = yi + ti for i = 1, 2. To keep the
analysis simple, assume that the parent can make positive or nega-
tive transfers (i.e. there is no non-negativity constraint). However,
only the parent can make transfers.
The game unfolds as follows: first, all parties simultaneously choose
actions ai ; second, the parent chooses transfers, having observed ac-
tions and realized incomes.
Solve for the subgame perfect equilibrium of this game. (Hint: there’s
an easy way to do this, and a hard way. Think about how the con-
sumption of each party varies with income on the continuation path.
What does that tell you about each party’s objectives in stage 1?)
46. Nation A plans to attack nation B. The attack can occur at one of two
locations, C and D. The success or failure of the attack depends on three
factors: where troops are massed prior to the attack (near C or near D),
where the attack occurs, and which location is defended. Let x denote
the fraction of troops massed near C, and let 1 − x denote the fraction of
troops massed near D. The game is played as follows: simultaneously, A
chooses to attack either C or D, and B decides to defend either C or D.
Payoffs are determined as follows. If the attack is successful, the payoff to
A is 1, and the payoff to B is zero. If the attack is unsuccessful, the payoff
to A is zero, and the payoff to B is 1. The probability of a successful
attack is in turn determined as follows. Let z denote the fraction of A’s
troops massed near the location where A attacks (so z = x if A attacks at
C, and z = 1 − x if A attacks at D). Then the probability of a successful
attack is z if B does not defend the location where A attacks, and z2 if B
does defend the location where A attacks.
(a) Depict the normal form of this game. Are there ranges of x for
which either or both players have dominant strategies? If so, indicate
26
which strategies are dominant over which ranges. Are there ranges
of x for which pure strategy Nash equilibria exist? If so, indicate
the ranges, and characterize the Nash equilibria (specify equilibrium
strategies and indicate payoffs). Are there ranges of x for which
mixed strategy Nash equilbria exist? If so, indicate the ranges and
characterize the mixed strategy Nash equilibria (specify equilibrium
strategies and indicate payoffs). Draw a graph representing A’s
expected equilibrium payoff as a function of x.
(b) Now imagine that, instead of being fixed, x (the deployment of
troops) is under A’s control. Events occur in the following order:
first, A decides how many troops to mass near each location (that is,
A chooses x); next, B observes the deployment of A’s troops (that is,
B observes x); finally, A chooses to attack either C or D, and simul-
taneously B decides to defend either C or D. Solve for the subgame
perfect equilibria of this game. How does A deploy its troops?
47. Consider a population of N people labeled i = 1, 2, ..., N . Each chooses

between two actions, IN , which corresponds to adopting a new technology,
and OU T , which corresponds to staying with the status quo technology.
If n others choose IN , the payoff to person i is
S(i, n) = ai − bn if i chooses OU T
B(i, n) = c + dn − ei if i chooses IN , making the total of IN -choosers n + 1
Assume that b, d > 0. This implies that the new technology and the old
technology are both characterized by “network externalities,” in the sense
that the benefits to using a given technology are greater when the number
of other people using it is larger. You are asked to consider three different
versions of this game.
Version #1: People make their choices of technology simultaneously, all
individuals have the same preferences (a = e = 0), c < 0, and c + d(N −
1) > −b(N − 1).
Version #2: People make their choices of technology simultaneously,
both a and e are strictly positive (so that people are ordered by increasing
reluctance to adopt the new technology), c − e > a, and d − e ≥ a − b.
Version #3: Same as version #2, but people make their decisions in
some arbitrary, fixed order (with each individual choosing once), with all
choices observable to subsequent decision makers.
(a) With respect to version #1: Do any players have dominant strate-
gies? Solve for all pure strategy Nash equilibria. Does the set of
equilibrium depend upon whether S(i, 0) ≷ B(i, N − 1)? What does
this tell you about the tendency for the group to adopt an efficient
technology?
27
(b) With respect to version #2: Is the game dominance solvable? If
so, identify the solution. Identify the following for this game: the
set of pure strategy Nash equilibria, the set of mixed strategy Nash
equilibria, the set of rationalizable strategies, and the set of correlated
equilibria.
(c) With respect to version #3: Identify the subgame-perfect Nash equi-
libria of this game for the case of N = 3.
48. Players 1 and 2 each begin with an endowment of one dollar. Player i
decides whether to contribute this dollar to a public good, for i = 1, 2. The
contribution yields benefit a for each player. If both players contribute,
both receive benefit 2a. Contribution decisions are made simultaneously.
Assume that 0 < a < 1 < 2a.
(a) Assume that players care only about their own payoffs. Identify all
pure strategy Nash equilibria.
(b) Now suppose that after the contribution stage, actions are observed
and each player has a chance to punish his opponent. By paying
p ≥ 0 dollars, he reduces his opponent’s monetary payoff by Kp
where K > 1. Punishment choices are made simultaneously. p is not
restricted by stage 1 payoffs. Any p ≥ 0 is allowed. Solve for the
pure strategy subgame perfect Nash equilibria of this game.
(c) Now suppose that instead of caring solely about monetary outcomes,
the payoff of player i is given by
gi (xi , xj ) = xi − αi max[xj − xi , 0] − β i max[xi − xj , 0]
where αi ≥ β i , β i ∈ [0, 1) for i = 1, 2 and xi and xj are the monetary
outcomes for i and j. αi and β i are common knowledge.
i. Interpret these payoffs.
ii. Repeat part (a) using these payoffs. How do your answers depend
on the parameters? Interpret.
iii. Repeat part (b) using these payoffs. When considering the pun-
ishment subgames, restrict attention to equilibria in which at
most one player punishes. Interpret the equilibria.
49. It’s the last lecture of the course in introductory game theory. To en-
sure applause, the professor offers a prize of d dollars to the last student
clapping. To model this, suppose that time is discrete. In each period
t = 1, 2, 3 . . ., each player decides (simultaneously) whether or not to con-
tinue clapping. Once a player has stopped clapping, he may not start
clapping again. There is a cost c > 0 per period of clapping. If during
any period, only one player chooses to clap, the game ends and he earns
the whole prize. If no players choose to clap, the game ends and the prize
goes unclaimed.
28
(a) First, assume there are only two students in the class. Further assume
that there are only two periods. If at the end of the second period,
both students are still clapping, the prize d is split evenly.
i. Provide the extensive form for this game. Identify strategy sets
and derive the normal form.
ii. Suppose d ∈ (2c, 4c). Identify all pure strategy Nash equilibria.
iii. Continue to assume d ∈ (2c, 4c). Identify all subgame perfect
Nash equilibria (pure and mixed) and the equilibrium payoffs
associated with each.
(b) Assume there are still only two students in the class, but now there
are three periods.
i. Suppose d ∈ (4c, 6c). Identify all subgame perfect Nash equilib-
ria (pure and mixed) and the equilibrium payoffs associated with
each.
ii. Repeat for d ∈ (2c, 4c).
(c) Now assume there is an infinite time horizon t = 1, 2, 3 . . ., and two
players, A and B.
i. Suppose that in period T , both players remain. Suppose A ex-
pects that B will clap with probability p this period and that in
future periods, both players will choose to clap with probability
p. What is A’s payoff from choosing to clap in T ?
ii. Identify a (mixed strategy) symmetric subgame perfect Nash
equilibrium. What are equilibrium payoffs?
iii. Identify an asymmetric subgame perfect Nash equilibrium. What
are equilibrium payoffs?
50. There are 3 political candidates. Each candidate has a budget sufficient to
launch an attack to discredit only one of his two opponents. For i = 1, 2, 3,
candidate i’s attack is successful with probability pi (regardless of whom
i chooses to attack). Suppose 0 < pi < 1 for all i = 1, 2, 3, and pi 6= pj
for i 6= j. If an attack against candidate j is successful, j is eliminated
from the game. All candidates’ moves and nature’s moves (i.e. whether
an attack is successful or not) are perfectly observable.
(a) Each candidate, when it is his turn to attack, may choose to attack
any opponent that is still in the game. The timing of attacks is as
follows. First, candidate 1 attacks (can choose either candidate 2 or
3), then candidate 2 attacks (if he is still in), and then candidate 3 (if
he is still in) and the game ends. Each candidate wants to maximize
his probability of survival; among the outcomes in which his survival
probability is the same, he wants to minimize the potential danger
(not modeled explicitly) posed by any other survivors (i.e. given the
same survival probability for himself, he wants to minimize the (sum
of) pi (’s) of the surviving opponent(s)).
29
i. Describe each player’s strategy set.
ii. Solve for the subgame perfect equilibrium (equilibria) of the
game. (Hint: You do not need to draw a complete extensive
form, as it can be messy; but note that all subgames following
histories in which candidate 1’s attack is not successful are the
same.)
iii. Show that in the above SPNE, candidate 3’s survival probability
is higher when p3 < p2 than when p3 > p2 .
(b) Now suppose that each candidate has an additional option when it is
his turn to attack: attack no one an donate his attack fund to charity.
The candidates do not care about the charity cause. Everything else
is the same as in the previous case. Further suppose p1 < p2 . Solve
for the subgame perfect equilibrium (equilibria) of this game. How
does it differ from the SPNE you found in part (1)? Explain.
51. Two firms have simultaneously introduced new drugs, which are perfect
substitutes for each other. They produce the drugs at a cost of c per
unit, and sell them to doctors, who use the drugs to treat patients. Firm
i sets a list price, pi per unit, and a discount, di ; the doctors pay pi − di
per unit. There is a fixed upper bound, p > c, on the list prices that
the firms are allowed to set. For simplicity, we will also assume that
the firms are legally precluded from setting net prices below cost; that
is, they must respect the constraint pi − di ≥ c. The total number of
patients is fixed at Q, and the dosage per patient is fixed at one unit;
doctors are obliged to treat all patients, and the distribution of patients
among doctors is exogenous. Doctors are reimbursed for firm i’s drug
by Medicare at the rate of qi per unit of the drug used. Since the drugs
are therapeutically equivalent, doctors choose between the drugs based on
the doctor’s profit. If doctors are indifferent between the drugs, their
indifference can be resolved as needed to sustain an equilibrium.
(a) Consider a static version of the game described above. There is one
period; firms choose list prices and discounts simultaneously, then
each doctor selects a drug. Assume that qi = αpi , where α > 1.
i. Solve for a Nash equilibrium of this game. In this equilibrium,
what are the list prices, discounts, and net prices?
ii. Are there any other Nash equilibria? Provide a proof.
(b) Now consider a two-period version of the same game. Assume the
firms simultaneously set list prices and discounts in the first period;
then, after observing the first period outcome, they simultaneously
set list prices and discounts in the second period. The two-period
game has the following twist (modeled after the procedures actu-
ally used for drug reimbursement under Medicare Part B). In pe-
riod 1, the market operates exactly like the static model described
30
above, with reimbursement based on list price, qi1 = αp1i (where su-
perscripts indicate the period). However, in period 2, reimbursement
is determined as follows: (i) if firm i has made sales in the first pe-
riod, reimbursement is based on the first period net selling price:
qi2 = α(p1i − d1i ); (ii) if firm i has made no sales in the first period, re-
imbursement is based on the second period list price: qi2 = αp2i . The
firms discount future profits at the rate δ ∈ (0, 1). Doctors always
choose the drug that maximizes their current period profits; they do
not consider the future consequences of their choices (equivalently,
for doctors, δ = 0).
i. Identify the strategy sets for the firms.
ii. Fixing first period list prices, discounts, and sales, solve for the
second period continuation equilibrium. Is it unique? Provide
a proof.
iii. Is there a subgame perfect equilibrium in which both firms make
first period sales? Explain your answer (ideally, provide a proof).
iv. Solve for a subgame perfect equilibrium of the two period model.
Indicate list prices, discounts, net prices, and reimbursement
rates as functions of δ. Does competition reduce Medicare re-
imbursement costs under this system?
52. A team of K individuals must perform some task before the end of time
period T , where T ≤ K. At each point in time, t = 1, ..., T , there is a
team leader (one of the K individuals). Each period has a different team
leader. The leader for each period is assigned in advance of the game and
known to all team members. In each period, the leader decides whether
to have the team perform the task or wait. If he decides to perform the
task, the game ends. If he decides to wait, the game continues to the next
period. Completing the task creates an immediate gross value of vt at a
cost of ct for each team member; however, the cost to the team leader is
βct , with β > 1; thus, the net payoff is vt − ct for team members other
than the period t leader, who receives a net payoff of vt − βct . If the task
has not been performed prior to period T , the leader for period T must
have the team perform it. Team members do not discount future payoffs.
Assume that vt − ct is strictly decreasing in t.
Portions of this problem ask you to consider the following parameterized
example: K = 4, T = 4, v1 = 12, v2 = 10, v3 = 8, v4 = 5, ct = 4, and
β = 2.
(a) For the paramterized example: draw the extensive form for this game;
identify the set of pure strategies for each player; provide the normal
form representation.
(b) Suppose the team members play a subgame perfect Nash equilibrium.
31
i. For the parameterized example, solve for the subgame perfect
Nash equilibrium.
ii. Is the equilibrium described in part (i) Pareto efficient?
iii. More generally, will the equilibria for this model (and not just
the example) be Pareto efficient? Prove your answer. (You can
assume for this part that the game is generic; that is, no player
is indifferent between any two outcomes.)
(c) Now make the following behavioral assumption: for each t, the leader
leader for t has naive (and optimistic) expectations: that leaders for
t0 > t will collectively choose to perform the task at the point in time
that maximizes leader t’s net payoff.
i. For the parameterized example, solve for the outcome.
ii. How well do the team members fair, relative to the subgame
perfect Nash equilibrium? Is the outcome Pareto efficient?
iii. In the example, is the task performed sooner or later than in a
subgame perfect Nash equilibrium? More generally (not just in
the parametric example), does the team perform the task earlier
or later if its members are naively optimistic, than in a subgame
perfect Nash equilibrum? Prove your answer.
53. In a game of “hid and seek,” the hider (H) chooses h ∈ [0, 1], and the
seeker (S) chooses s ∈ [0, 1]. The payoffs are: (h − s)2 for H, and
−(h − s)2 for S.
(a) Suppose H moves first, and then S moves having observed H’s choice.
i. Describe a typical pure strategy for each player, as well as their

strategy sets.
ii. Exhibit a pure strategy Nash equilibrium.
iii. Identify the set of all possible pure strategy Nash equilibrium
outcomes (not the strategies, but the choices made).
iv. Does subgame perfection reduce the set of pure strategy equilib-
rium outcomes?
(b) Suppose S moves first, and then H moves having observed S’s choice.
strategy sets.
ii. Exhibit a pure strategy Nash equilibrium.
iii. Identify the set of all possible pure strategy Nash equilibrium
outcomes (not the strategies, but the choices made).
iv. Does subgame perfection reduce the set of pure strategy equilib-
rium outcomes?
32
(c) Suppose H and S make their choices simultaneously.
strategy sets.
ii. Prove that there exists no pure strategy Nash equilibrium.
iii. Suppose H’s choice is governed by some arbitrary CDF, F . Solve
for S’s best response as a function of the characteristics of F .
iv. Suppose that S’s choice is governed by some arbitrary CDF, G.
Solve for H’s best response as a function of the characteristics
of G.
v. Using your answers to parts (iii) and (iv), solve for a mixed
strategy Nash equilibrium.
vi. Using your answers to parts (iii) and (iv), prove that the mixed
strategy Nash equilibrium is unique.
(d) Suppose that H and S make their choices simultaneously, but that
each now chooses a point on the unit circle. The payoffs are d2 for H
and −d2 for S, where d is the shortest distance between their chosen
points around the circle’s circumference. Exhibit a mixed strategy
Nash equilibrium.
54. Players 1 and 2 have decided to go on a road trip together. The trip
begins at position 0, and they can go either to the Far East or to the Far
West. The Far East is x days from 0 traveling east and the Far West is x
days from 0 traveling west. There is one single straight route connecting
the Far East and the Far West going through 0. So for example, if the
players have traveled m days east and decide to turn around, it would
take them x + m days to get to the Far West (m days to get back to 0
and x additional days traveling west to get from 0 to the Far West).
The game proceeds as follows:
• Player 1 starts driving on day 1, and from there on they alternate.

So Player 2 drives on day 2, Player 1 drives on day 3, Player 2 on
day 4 ...and so on. In general Player 1 drives on odd days and Player
2 drives on even days. The player who drives on a given day, decides
in what direction they move on that day: east (e) or west (w).
• The game stops when they reach the Far East (F E) or the Far West
(F W ). So for example, if x = 3, one complete history of play could
look like e1 , w2 , w1 , w2 , e1 , e2 , e1 , e2 , e1 . —where I am using the sub-
script to emphasize which player is moving that day (redundantly
given that we have already stated the way in which they must alter-
nate).
• Player 1 values the Far East at v and the Far West at v, where
0 < v < v. Player 2 values the Far East at v and the Far West at v.
They each discount the future according to δ, 0 < δ < 1, and each
33
player only cares about his own enjoyment of the destination. So for
example, if they reach the Far East after 37 days (assuming the last
player moves at t = 37), Player 1 gets δ 37 v and Player 2 gets δ 37 v. If
the players never arrive at a some destination then they both get 0.
(a) What are the strategies of the players? Throughout the question we
will be focusing on pure strategies only.
(b) Let’s first suppose that either final destination is just 1 day away from
0; namely x = 1. What is the subgame perfect Nash equilibrium of
the game?
(c) Now assume that x = 2.
i. Is there a Nash equilibrium in which the players never arrive at a
destination? If your answer is yes, provide an example; otherwise
provide a proof of nonexistence.
ii. Is there a Nash equilibrium in which F E is reached?
iii. Is there a subgame perfect equilibrium in which F E is reached?
(Hint: Consider the cases v ≤ δ 2 v, and v > δ 2 v)
iv. What is the largest T such that there is a Nash equilibrium in
which the Far East is reached at time T ?
v. What is the largest T such that there is a Subgame Perfect Nash
equilibrium in which the Far East is reached at time T ?
(d) Repeat part b(i) through b(v) with F W instead of F E.
55. Consider a game played by a sequence of players, t = 0, ..., T . Each player

makes a single choice, with player t choosing a number xt from the set
R− = (−∞, 0]. (Note that −∞ is not a feasible choice; each player must
choose a finite number, though one that can be arbitrarily large in absolute
value.) Player t observes (x0 , ..., xt−1 ) before making his own choice. All
players have Pexactly the same preferences: they rank outcomes according
T
the value of k=0 δ k xk , with δ ∈ (0, 1). (Note that player k cares about
the choices made both before and after his own choice. If you like, think
about this as a single individual who at any given point in time is playing
a game with his past and future “selves.”) Thus, all players agree that
the best outcome is for all players to choose 0.
Let’s call (x0 , ..., xT ) a Nash path if the players choose (x0 , ..., xT ) on the
equilibrium path of some Nash equilibrium, and let’s call (x0 , .., xT ) a sub-
game perfect Nash path if the players choose (x0 , .., xT ) on the equilibrium
path of some subgame perfect Nash equilibrium. In this question, you
will be asked to characterize the sets of Nash paths and subgame perfect
Nash paths. (Note: This is easier than characterizing the sets of Nash
equilibria and subgame perfect Nash equilibria, because you don’t have to
list all possibilities for behavior off the equilibrium path.)
34
(a) How does one describe a strategy for player t? What is the strategy
space for player t?
(b) Suppose T = 1.
i. Characterize the set of Nash paths for this game. Demonstrate
that paths in this set can indeed be sustained in a Nash equi-
librium. (Hint: start with x1 , and then think about how the
equilibrium might provide incentives for choosing a given x0 , in
light of the fact that subgame perfection is not required.)
ii. Characterize the set of subgame perfect Nash paths for this game.
Demonstrate that paths in this set can indeed be sustained in a
subgame perfect Nash equilibrium.
iii. How do your answers to parts (i) and (ii) change if there is a
finite lower bound xmin on x0 and x1 ?
(c) Repeat questions (b-i) and (b-ii) for finite T > 2.
(d) Now consider T = +∞.
i. For any arbitrary x ∈ R − , exhibit a subgame perfect Nash equi-
librium for which (x, 0, 0, 0, ...) is the equilibrium path. Demon-
strate that it is a subgame perfect Nash equilibrium. (Hint: use
punishments that generate continuation paths belonging to this
same class.)
For any arbitrary path (x0 , x1 , ...) with xt ∈ R− for all t and
ii. P
∞ t
t=0 δ xt finite, exhibit a subgame perfect Nash equilibrium for
which (x0 , x1 , ...) is the equilibrium path. Demonstrate that it
is a subgame perfect Nash equilibrium. (Hint: when thinking
about how to construct punishments for deviations by each t,
remember part (d-i).)
iii. Suppose there is a finite lower bound xmin on the choices x0 , x1 , ...
What would that imply about the set of subgame perfect Nash
paths?
56. The Department of Defense (DoD) would like to develop a weapons system
with completely novel capabilities. If the system is developed successfully,
the total benefit will be B. The DoD does not have the requisite R&D
capabilities, and therefore must contract out the research effort. It does
this by offering a reward R for successful development. √ Defense contrac-
tors that invest x in R&D succeed with probability µ = x; equivalently,
the resource cost of succeeding with probability µ is µ2 .
(a) First suppose the DoD decides to work with only one contractor
from the outset (in other words, only one contractor is eligible for
the reward).
i. Solve for the contractor’s probability of success as a function of
the reward R.
35
ii. Solve for the DoD’s optimal choice of R, as well as the probability
of success µ, as functions of B.
(b) Now suppose the DoD decides to offer the reward to two contractors.
The contractors make their R&D investments simultaneously. If
only one is successful, it receives the reward R. If they are both
successful, they each receive R2 .
i. Are research investments strategic substitutes or strategic com-
plements?
ii. Solve for the symmetric Nash equilibrium of the game played by
the contractors. As a function of R, what is the probability of
success for each firm? Is it monotonic in R? For what value of
R, if any, does it reach unity?
iii. In the aforementioned equilibrium, what is the overall probability
of success for the DoD?
iv. Solve for the DoD’s optimal choice of R, as well as each firm’s
probability of success µ and the probability of at least one suc-
cess, as functions of B. Is there a B beyond which success will
be certain?
(c) Does the DoD achieve a higher probability of success working with
one contractor or two contractors? Does it achieve a higher payoff
working with one contractor or two contractors? If your answer is
ambiguous, on what does it depend, and how?
Bayesian Perfect and Sequential Equilibria

57. Consider the following two player game in extensive form:
Nature
1 1
2 2
1
x y z z y x
2
(2,10) (0,7) l r l r l r (2,10)
(0,10) (-10,2) (6,0) (-10,7) (3,5) (3,5)
(a) Characterize the set of Perfect Bayesian Equilibria.

(b) Characterize the set of Sequential Equilibria.
36
58. Consider the following game played by two parties, A and B. First, “na-
ture” chooses either C or D. C is chosen with probability 0.7, and D is
chosen with probability 0.3. Second, party A chooses either E or F . Party
A does not observe nature’s choice when it makes this choice. Next, party
B chooses either G or H. Prior to making this choice, party B observes
the choice of party A; party B also observes nature’s choice if A has cho-
sen E, but does not observe nature’s choice if A has chosen F . Payoffs
are determined as follows: A and B always receive the same payoff; the
payoff is 0 if A chooses E and B chooses G, regardless of nature’s choice;
the payoff is 5 if A chooses E and B chooses H, regardless of nature’s
choice; the payoff is 0 if nature chooses C, A chooses F , and B chooses
G; the payoff is 10 if nature chooses C, A chooses F , and B chooses H;
the payoff is 10 if nature chooses D, A chooses F , and B chooses G; and
the payoff is 0 if nature chooses D, A chooses F , and B chooses H.

(b) Identify the strategy set for each player. Write the normal form of
the game.
(c) Identify all pure strategy Nash equilibria for this game.
(d) Identify all pure strategy subgame perfect Nash equilibria for this
game.
(e) Identify all pure strategy Bayesian perfect equilibria for this game.
(f) Identify all pure strategy sequential equilibria for this game.
59. Two parties, the defendant (D) and the plaintiff (P), are involved in a
dispute where the plaintiff is seeking compensation from the defendant. If
the parties do not settle, there will be a trial, which will cost each party
C > 0, and which will result in a judgement of v (which the defendant
will have to pay to the plaintiff). The parties may, however, settle before
going to trial. Before settling, they may (at the defendant’s discretion)
also hold a mock trial to resolve uncertainty concerning v.
The order of decisions is as follows. First, nature selects v (but this
isn’t revealed to either party). Nature’s choice can be represented by a
random variable which has a positive probability density over the unit
interval [0,1] and expectation E(v) > C.. Second, the defendant decides
whether to hold a mock trial. If the defendant chooses to hold a mock
trial, both the defendant and the plaintiff incur a cost D > 0, and both
immediately learn the true value of v. Regardless of whether a mock trial
is held, the game then proceeds as follows: The plaintiff decides whether
to continue the suit or abandon it. If the plaintiff decides to abandon the
suit, the game ends immediately, and there are no further costs, benefits,
or transfers. If the plaintiff decides to continue the suit, the defendant
makes a settlement offer q to the plaintiff, and the plaintiff either accepts
37
or rejects this offer. If the offer is accepted, defendant pays q to plaintiff
and the game ends. If the offer is rejected, the dispute is resolved at trial.
Throughout, assume that this process consumes very little time, so that
there is no discounting of future payoffs.

(b) Suppose the defendant chooses the mock trial. What is the subgame
perfect equilibrium outcome of the bargaining process? Given this
outcome, when will the plaintiff choose to withdraw the suit?
(c) Suppose the defendant chooses to forego the mock trial. What is the
outcome of the bargaining process, assuming that we are studying
Sequential Equilibria? (Remember, you will need to think about
the plaintiff’s beliefs about nature’s choice for offers not chosen in
equilibrium.)
(d) Describe the Sequential Equilibrium outcome of the whole game as
a function of the parameters.
(e) What difference, if any, would it have made if you had been asked to
study Perfect Bayesian Equilibria instead of Sequential Equilibria?
60. There is an owner of a company, and a potential buyer. The company has
the same monetary value (and no other form of value) to the owner and
to the potential buyer. The monetary value of the company is V withe
probability π, and 0 with probability 1 − π. The value is known to the
seller, but not to the buyer.
(a) Consider the following game: The owner proposes a non-negative

price, which the potential buyer then either accepts or rejects. Both
players are concerned only with their net monetary payoffs.
ii. Is there a perfect Bayesian equilibrium in which a sale is made
when the company’s value is V ? If so, exhibit it. If not, explain
(as precisely as possible) why not.
iii. Is there a perfect Bayesian equilibrium in which no sale is made,
regardless of the compay’s value? If so, exhibit it. If not,
explain (as precisely as possible) why not.
iv. How would your answers to parts (ii) and (iii) change if the
monetary value of the company to the buyer in the “good” state
were βV instead of V , with β > 1?
(b) Repeat part a (i through iv), but change the game as follows: the
buyer proposes a non-negative price, which the owner then either
accepts or rejects.
38
iv. How would your answers to parts (ii) and (iii) change if the
monetary value of the company to the buyer in the “good” state
were βV instead of V , with β > 1?
(c) As in part (b), assume that the buyer proposes a non-negative price,
which the owner either accepts or rejects. Now, however, assume
that there is a probability p that the owner has a liquidity problem
and must accept any offer. Only the owner knows whether he has
a liquidity problem. Assume that this event is independent of the
realization of the firm’s value.
61. On the first day of class, the professor announces whether or not there
will be a final exam. The possible announcements are Y and N , where Y
stands for the announcement: “Yes, there will be an exam.” N stands for
the announcement: “No, there will not be an exam.”
Suppose that after the announcement, the professor decides secretly (with-
out telling the students), whether or not he will come to the final exam
with a stack of exams (e) or a stack of pizzas for a party (p). Thus, the
students only hear the professor’s announcement (Y or N ), but do NOT
know whether they’ll have an exam or a party (e or p). The students must
decide during finals week whether or not to study (s for study, n for not
study). For simplicity, assume there is one student.
(a) Draw the extensive form of this game, without any payoffs at the
terminal nodes.
(b) What is each player’s strategy set?
Suppose that the utilities of the professor and student depend only on
whether the student studied and whether the professor wrote an exam.
39
The announcement itself does not affect the utilities of either player. Sup-
pose that they have the following payoffs for the given combinations of
choosing whether or not to study and choosing whether or not to show up
with a test:
(e, s) (e, n) (p, s) (p, n)

Professor 3 1 4 2
Student 3 1 2 4
(c) Add the payoffs to the terminal nodes of the extensive form.
(d) Does there exist a pure-strategy Nash equilibrium of this game in
which the professor plays e at some information set? If so, describe
such an equilibrium. If not, explain why not.
(e) Does there exist a subgame perfect Nash equilibrium of this game in
which the professor plays e at some information set? If so, describe
such an equilibrium. If not, explain why not.
(f) Give an example of a mixed-strategy Nash equilibrium in which the
professor puts strictly positive weight on Y and N .
(g) Characterize the set of perfect Bayesian equilibria of this game.
(h) Characterize the set of sequential equilibria of this game.
Now suppose that the chair of the economics department frowns upon
professors lying to students and will fire the professor if he lies; thus, the
professor gets a payoff of 0 if his action does not match his announcement
(regardless of whether or not the student studies). If the professor tells
the truth, then his payoff is determined using the table above. The stu-
dent’s payoff is determined by the table above regardless of whether the
professor has told the truth or lied.
(i) Draw the extensive form again, but modify the payoffs at the termi-
nal nodes where appropriate. Note, the shape of the extensive form
diagram hasn’t changed.
(j) Now, using the new diagram, characterize the set of perfect Bayesian
equilibria.
62. There are N > 1 firms selling a homogeneous good. They compete for
one consumer’s business by naming prices. However, unlike in the stan-
dard Bertrand model, the consumer cannot costlessly observe prices. To
discover a firm’s price, the consumer must pay a search cost s > 0 to visit
its store. When the consumer purchases a quantity q at price p, his utility
40
2
is U (p, q) = vq − q2 − pq. All firms have identical constant marginal cost
c. Assume v > c > 0 and (v + c)2 > 8s. The game proceeds as follows:
Stage 1: Firms simultaneously name prices. The consumer does not ob-
serve these prices.
Stage 2: The consumer can obtain prices from as many firms as he wishes,
at cost s per firm. Each time he decides to search, a firm is chosen at
random (by Nature) from those remaining. Searches happen sequen-
tially. After each search, the consumer observes the price and the
identity of the firm. Then the consumer can choose to either search
again (if any firms remain) or continue to Stage 3.
Stage 3: The consumer chooses one of the firms from which he obtained
a price in stage 2. Then he chooses any nonnegative quantity to
purchase from that firm at the firm’s specified price.
(a) Identify the strategy sets for the firms. Are any of the firms’ strategies
strictly dominated?
(b) Is there a Nash equilibrium in which the consumer chooses not to
search at all (thereby receiving a payoff of 0)? Describe one or explain
why such an equilibrium does not exist.
(c) We will now look for pure-strategy sequential equilibria in which the
consumer purchases a positive quantity of the good (and therefore
conducts at least one search).
i. Suppose the consumer has obtained exactly k prices for some k ∈
{1, . . . , N } and that he is now in Stage 3 of the game. Describe
his optimal purchasing decision (quantity and firm).
ii. Prove that throughout Stage 2 (that is, initially and after each
search), the consumer must have correct beliefs about the firms’
prices in any sequential equilibrium, both on and off the equilib-
rium path. For this part only, you may prove this for the case
of 2 firms, each with finite strategy set {p, p0 }. Does the same
property also hold in a perfect Bayesian equilibrium?
iii. Given beliefs from part (ii), suppose the consumer is in Stage 2
and has observed some k prices. When must his strategy call for
him to search again and when must his strategy call for him to
continue to Stage 3?
iv. Are there symmetric equilibria in which all firms name the same
price and the consumer conducts at least one search? If so, what
are the possible equilibrium prices?
v. Are there asymmetric equilibria in which at least two firms name
different prices and the consumer conducts at least one search?
Characterize them or explain why they do not exist.
41
63. Firms A and B compete in an industry. A third firm, C, does not compete
in that industry, but has invented a technology that can give either firm A
or firm B a competitive advantage. The profits for firms A and B (prior to
subtracting any payments to firm C for the use of the technology) depend
upon which of them adopts the technology, as follows:
Firm B - not adopt Firm B - dopt
Firm A - not adopt X, X W, Z
Firm A - adopt Z,W Y, Y
In this table, Firm A’s profit appears first, and tim B’s appears second.
We will assume throughout that Z > X and Y > W (so that adoption is
always unilaterally beneficial), but W < X and Y < Z (so that adoption
always negatively impacts the other firm). If a firm decides to adopt the
technology for a licensing fee P, the amount P is simply subtracted from
its payoff (i.e., leaving it with either Z −P or Y −P , depending on whether
the other firm has adopted).
(a) First we’ll suppose that firm C has an opportunity to license the
technology to firm A, but not to firm B. The game proceeds as follows:
firm firm C offers to license the technology to firm A for the price P;
then firm A decides whether or not to adopt. Solve for the sub game
perfect equilibria. What price will C charge? Indicate the resulting
payoffs for each firm.
Throughout the rest of the question, assume that firm C has the
opportunity to license the technology to both firms. Focus only on
pure strategy equilibria.
(b) Now we’ll suppose that firm C can try to license the technology to
both firms A and B by making offers to them sequentially. The game
proceeds as follows: first, firm C offers to license the technology to
firm A for the price PA ; then firm A decides whether or not to adopt;
then firm C offers to license the technology to firm B for the price
PB ; then firm B decides whether or not to adopt, knowing whether A
has adopted. Solve for the sub game perfect equilibria. What price
will C charge? Indicate the resulting payoffs for each firm. Does firm
C do better or worse selling to both firms, compared with selling only
to A?
(c) Now we’ll assume that firm C can try to license the technology to
both firms A and B by making offers to them simultaneously instead
of sequentially. For the moment, also assume that Z − X > Y − W .
The game proceeds as follows: in the first state, firm C offers to
license the technology to firm A for the price PA and to firm B for
the price PB ; then, in the second stage, firms A and B simultaneously
decide whether or not to adopt. Here, assume that A and B observe
both prices before making their decisions.
i. Solve for equilibria of the second-stage, conditional on PA and
PB . (Hint: draw yourself a grid dividing the possibilities into
42
categories, and then think about what happens within each cat-
egory.)
ii. What price will C charge in a sub game perfect equilibrium?
Can this game have more than one equilibrium? Indicate the
resulting payoff for each firm. Does firm C do better or worse
selling to both firms, compared with parts (a) and (b)?
(d) Repeat part (c), assuming Z − X < Y − W .
(e) As in parts (c) and (d), we’ll assume that firm C can try to license
the technology to both firms A and B by making offers to them
simultaneously. The structure of the game is the same, except we
assume here that each firm can only observe the price it is offered, and
not the price the other firm is offered. Because there are no proper
sub games, we will apply the notion of Perfect Bayesian Equilibrium
rather than sub game perfection. Is there any PBE in which neither
firm adopts? In which one and only one firm adopts? In which
both firms adopt? Explain your answer completely. What is the
highest level of profits firm C earns in any PBE? How do your answers
compare to part (d)?
64. There is an agenda setter who choses a proposal x ∈ [0, 2] and a legislator
who accepts the proposal x or rejects it. If the proposal of x is accepted,
the agenda setter gets a payoff of x and the legislator gets a payoff of
2 − (x − θ)2 for some fixed θ ∈ (0, 2). If the proposal of x is rejected, each
player gets a payoff of zero. Finally, assume that the value of θ is known
to both the legislator and the agenda setter.
(a) First, assume that the legislator observes the proposal x.

i. What is the strategy set of the agenda setter? What about the
legislator?
ii. Find a subgame perfect Nash equilibrium.
iii. Is there a Nash equilibrium in which the agenda setter proposes
x = θ? Is it subgame perfect?
(b) Now suppose that the legislator only observes whether or not x = θ
(i.e., he cannot distinguish between other values of x).
i. What are the strategy sets of the agenda setter and the legislator?
ii. Is there a subgame perfect Nash equilibrium in which the agenda
setter proposes x = θ? Is there a subgame perfect Nash equi-
librium in which the agenda setter proposes some other value of
x? Do your answers depend on the value of θ?
iii. Determine whether the subgame perfect Nash equilibria you de-
scribed in the part (ii) are perfect Bayesian equilibria. Do your
answers depend on θ?
43
Now suppose that there are two types of legislators: a “moderate”
(who’s payoff is as described above) and a “partisan.” For the parti-
san, the payoff when the bill is passed is 1 if x = θ and −1 otherwise.
When not passed, the partisan’s payoff is also 0. Suppose that the
agenda setter’s prior belief that the legislator is moderate is µ ∈ (0, 1).
Finally, again assume that the legislator observes the proposal x and
the agenda setter observes θ. The game proceeds exactly as above:
the agenda setter chooses x, after which the legislator votes to accept
of reject the proposal.
(c) What is the subgame perfect Nash equilibrium? How does it depend
on µ and θ?
65. A student must decide whether to do her homework or to ask her professor
for an extension. If she continues to work and finishes her homework on
time, her payoff is 1 and the professors is 2. If she stops working and asks
for an extension, the professor can either grant it or refuse. Her payoff
will be 2 if he grants it and 0 if he refuses. In the event he is asked for an
extension, the professors payoff depends on whether he is demanding or
lenient. If he is demanding, his payoff will be −1 if he grants the extension
and 0 if he refuses. If he is lenient, his payoff will be 1 if he grants the
extension and 0 if he refuses.
Note: Throughout your answer, focus on pure strategy equilibria.
(a) Suppose first that the student knows whether or not her professor is
demanding.
i. What is the strategy set of the student? What about the profes-
sor?
ii. Solve for the subgame perfect Nash equilibria (conditional on the
professor’s type). Are there other Nash equilibria? If so, identify
one; if not, explain.
(b) Now suppose the student does not know whether her professor is
demanding or lenient.
i. What is the student’s strategy set? What is the professor’s?
ii. Let µ be the probability that the professor id demanding. Solve
for the subgame perfect Nash equilibrium, as a function of µ.
(c) Now suppose that, to begin the quarter, the professor announces
whether she is demanding or not. The professor does not need to tell
the truth, and she experiences no costs if she lies.
i. What is the student’s strategy set? What is the professor’s?
ii. Suppose µ > 21 . Solve for all PBE strategy profiles. For each
such profile, describe beliefs that sustain it.
iii. Suppose µ < 12 . Solve for all PBE strategy profiles. For each
such profile, describe beliefs that sustain it.
44
Applications of Dynamic Games
66. Consider an industry consisting of two firms producing a single homoge-
neous product. Decisions are taken as follows: first, firm 1 chooses price
p1 ; having observed this choice, firm 2 then selects p2 . Industry demand,
Q(p), is continuous and downward sloping. The firms have access to the
same constant-returns production technology, with c denoting the unit
cost of production. Consumers purchase the good from the firm quoting
the lowest price, and split equally in the event of ties. Finally, prices must
be quoted in whole units (“pennies”). Solve for all pure strategy subgame
perfect Nash equilibria.
67. Complete the exercise on p. 121 of the lecture notes.
68. Suppose that there are I consumers, each with demand given by qi (p) =
a − bp (a, b > 0). Firms incur a cost K when they enter the industry,
and a variable cost c for each unit produced. There are infinitely many
(identical) potential firms.
(a) Suppose that all firms make entry and production choices (quantities)
simultaneously in a single stage. What happens to equilibrium prices,
output levels, and consumer surplus as I goes to infinity?
(b) How do your answers in part (a) change if firms first make entry
decisions simultaneously, and then, having observed each others’ en-
try decisions, choose quantities (assuming that the entry cost, K, is
sunk)?
69. Consider the standard Hotelling spatial location model with product char-
acteristics on the interval [0, 1], two firms with identical linear production
costs, and quadratic transportation costs. Firm 1 enters the market first
and can choose to set up either one plant at one end of the interval or two
plants, one at each end of the interval. Each plant has a setup cost of F .
Firm 2 observes firm 1’s plant construction decisions, and then chooses
whether to enter; if it enters, it chooses one end of the interval and builds
only one plant at the cost F . Determine the equilibrium of this model.
How is it affected by the underlying parameter values? Suppose the gov-
ernment granted firm 1 a statutory monopoly before any entry decisions
were made. Would social welfare go up or down? Suppose that the gov-
ernment passed a law preventing any firm from building more than one
plant. Would social welfare go up or down?
70. Crush The Weak Enterprises, Inc. (henceforth, CTWE) produces two
goods, “stuff” (S) and “junk” (J). It is currently the only producer of
45
these goods. New entry into the production of stuff is impossible; however,
CTWE faces the possibility that Upstart Corp. may soon start to produce
junk.
Demand is determined as follows. There is a single consumer. The con-
sumer buys either one unit of stuff, or none at all. Similarly, she buys
either one unit of junk, or none at all. Consumption of the two goods is
not mutually exclusive; she may buy one unit of each, and may purchase
these units from different vendors. The consumer receives value RS from
consuming stuff, and value RJ from consuming junk. The consumer’s
utility is given by
U (S, J) = SRS + JRJ − P
where the quantity variables S and J take on the values of either 0 or 1,
and P is the total of all payments made to vendors.
CTWE produces each unit of stuff at a cost of cS < RS , and it produces
each unit of junk at a cost of cJ < RJ . CTWE has no fixed costs. Upstart
has a technology that will allow it to produce a unit of junk at a cost of
cJU < cJ .
Throughout this question, make the following parametric assumption:
RS − cS > cJ − cJU
(i.e. the social surplus from producing stuff is greater than the increased
social surplus from having Upstart, rather than CTWE, produce junk).
(a) Imagine that the game unfolds as follows:

First, Upstart decides whether or not to enter. If it enters, it pays
some small set-up cost, K. Assume that K < cJ − cJU , so that
Upstart can provide junk at lower total cost than CTWE.
Second, firms name prices. If Upstart has entered in the first stage,
prices are announced simultaneously. CTWE names two prices, one
for stuff, pS , and one for junk, pJ . If Upstart enters, it names a price
for junk, pJU .
Third, the consumer makes purchases to maximize utility.
Describe a pure strategy subgame perfect Nash equilibrium for this
game. In constructing an equilibrium, use strategies that do not
involve any firm setting a price below its costs. You may resolve the
consumer’s indifference however you like.
(b) Imagine that the game unfolds exactly as in part (a), except that
CTWE “ties” its products together by selling them ONLY as a bundle
for a single price, pB , which it names in the second stage. Describe a
subgame perfect pure strategy Nash equilibrium for this new game,
again using strategies that do not involve any firm setting a price
below its costs. Resolve the consumer’s indifference however you
like.
46
(c) Now imagine a game that unfolds exactly as in part (a), except that
CTWE chooses whether or not to “tie” its products into a single
indivisible bundle prior to the first stage. (Assume that it can commit
itself to this policy). Describe a subgame perfect pure strategy Nash
equilibrium. Again, avoid strategies that involve any firm setting a
price below its cost. Does tying stop Upstart from entering? If the
government banned tying arrangements, what would be the effect on
consumer surplus and on social welfare?
71. Complete the exercise on page 159 of the lecture notes.
72. Consider the bargaining model discussed in class, but instead of assuming
that players discount future payoffs, assume that it costs c < 1 to make
each offer. (Only the player making the offer incurs this cost, and players
who have made offers incur this cost even if no agreement is ultimately
reached.) Assume first the horizon is finite, of length T .
(a) What is the (unique) SPNE of this alternative model?

(b) What happens as T approaches infinity?
(c) Now suppose that the horizon is infinite. What are the subgame
perfect Nash equilibria of this game?
73. Consider the standard Cournot setting. Inverse demand is given by p(Q) =
1 − Q. Marginal and fixed costs are both zero.
(a) Suppose there are two firms. The board of firm 1 considers splitting
the company in two. The two ensuing firms, 1a and 1b will be run
by different managers, and thus will compete as separate entities
along with firm 2. That is, manager a cares about maximizing π 1a ,
manager b cares about π 1b , and the board cares about π 1a + π 1b .
Compute Cournot quantities, prices and profits before and after the
split. Should the board split the company?
(b) Now suppose that the game consists of two stages. In the first
stage, board i (i = 1, 2) decides to split its company Ni ways, where
Ni ∈ {1, 2, 3}. In the second stage, the N1 +N2 managers compete in
quantity. Compute the (pure strategy) subgame perfect Nash equi-
libria of this “spinoff” game.
74. Consider a multilateral bargaining game where Agents 1, 2, and 3 bargain

over a pie of size 1. There are T periods where T < ∞. In each period,
one of the 3 agents is selected randomly to be a proposer with probability
1
3 , and this selection process is i.i.d. across periods. In each period, the
47
proposer proposes a split (x1 , x2 , x3 ) of the pie (where x1 + x2 + x3 ≤ 1),
and this proposal is voted on in a fixed sequential order by the two other
agents in order of their name (i.e, if i and j are the voters and i < j,
i votes first and this vote is observable by j). If all voters accept the
proposal (x1 , x2 , x3 ) at t, then the proposal passes and the game ends
with a realized utility of δ t−1 (x1 , x2 , x3 ) where δ ∈ (0, 1). Otherwise,
the game proceeds to the next round, unless the proposal was rejected at
the T th round in which case all agents receive 0. [NOTE: In ALL the
questions below, do not worry about describing in full detail the off-the-
equilibrium path actions; simply restrict your attention to offers that are
made on the equilibrium path and whether they are accepted]
(a) Consider T = 2 (i.e., there are only 2 rounds of offers).

i. If agent i were selected to be the proposer in round 2, what
proposal would be made? Would this proposal be accepted by
other agents?
ii. What is the expected utility at the beginning of stage 2 before
the proposer is selected?
iii. Restricting attention to SPE, if agent i were the proposer in
round 1, what proposal would be made, and would it be ac-
cepted? How much of the surplus does the proposer in Round
1 obtain?
iv. What is the expected utility of each agent at the beginning of
the game, before the selection of the proposer at round 1?
(b) Now consider any arbitrary T < ∞. Restricting attention to the
equilibrum path, what is the first proposal that is made? Is it ac-
cepted? (Hint: What did you learn from parts (ii) to (iv) of (a)?)
(c) Now consider weakening the voting rule to majority-rule: for a pro-
posal to pass, only one agent must accept the proposer’s proposal.
Consider T = 2.
i. Does this change anything for the proposal that would be made
in round 2?
ii. Restrict attention to symmetric SPE where at the first round,
the proposer randomizes evenly between which of the other two
agents to offer a positive share of the surplus. What offer would
the proposer make at the first round, and how much of the sur-
plus does the proposer in round 1 obtain?
iii. Based on your answers to above, and restricting attention to
symmetric SPE, calculate each agent’s expected utility at the
beginning of the game, before the proposer is selected in round
1.
iv. Can you construct an SPE where expected payoffs before the 1st
round are not the same for all agents?
48
75. A buyer B wishes to buy an indivisible object from one of two sellers, S1
and S2 . This object can be “good,” in which case it is worth v > 0 to
the buyer, or “bad,” in which case it is worth nothing.
(a) Events unfold as follows.

Stage 1: Sellers simultaneously offer purchase contracts. The con-
tract for Si specifies a purchase price p. It also requires the buyer
to purchase the object, unless he can prove that the quality is bad
(more on this below).
Stage 2: Buyer B decides whether to accept a seller’s offered con-
tract. If B rejects both offers, all players receive zero payoffs and
the game ends. Otherwise the game continues between B and the
selected seller, call it S ∗ .
Stage 3: Seller S ∗ selects one of two effort levels, e ∈ {0, 1}. This
determines the object’s quality, ev.
Stage 4: Buyer costlessly obtains evidence of quality. The evidence
establishes with certainty whether quality is good or bad. (This is
automatic; there is no decision in this stage.)
Stage 5: If buyer has evidence that quality is bad, he can either void
the transaction or complete it. If he does not have evidence that
quality is bad, he must complete the transaction. If the transaction
is voided, B receives a payoff of zero, and S ∗ receives a payoff of −ce,
where 0 < c < v. If the transaction is completed, B receives a payoff
of ev − p, and S ∗ receives a payoff of p − ce.
Solve for a subgame perfect Nash equilibrium of this game. You may
focus on pure strategies. Make sure you describe the equilibrium
completely – that is, in each state, describe what happens both on
and off the equilibrium path. Does the usual Bertrand result hold?
That is, does competition between the sellers drive the price down to
their costs? why or why not?
(b) Now modify the game by substituting the following for stage 4:
Stage 4∗ : Buyer does not observe the seller’s effort level, but can
spend M to obtain evidence of quality. The evidence, if obtained,
establishes with certainty whether quality is good or bad. Suppose
that 0 < c < M < v4 .
i. Consider any subgame beginning in stage 3 (identified by a price
p for an accepted contract). Solve for the equilibrium of this
subgame as a function of p. (Hint: first reduce the game by
substituting best choices in stage 5. Then consider the normal
form of the subgame.) Make sure to consider both pure and
mixed strategies. Express the equilibrium payoffs for both B
and S ∗ as a function of p.
ii. Using your answer to B(i), solve for a subgame perfect equilib-
rium of the entire game. Does the usual Bertrand result hold?
49
That is, does competition between between the sellers drive the
price down to their costs? Why, or why not?
(c) How would your answers to parts (a) and (b) change if there was
only one monopolistic seller quoting a price in stage 1?
76. A game is played by an infinite sequence of overlapping generations. Gen-

eration t is born at the start of period t, where t = 1, 2, ... Each generation
is alive for two periods. When discussing period t, we refer to generation
t − 1 as the parent, and to generation t as the child.
(a) Suppose that each generation is endowed with W units of water in

the first period of life, and none in the second period. Lifetime
utility is given by u(w1 ) + δu(w2 ), where wi is the amount of water
consumed in the ith period of life, and u is strictly increasing and
strictly concave. Suppose that water isn’t storable, and that there
are no markets. Consider the following social convention: each gen-
eration, when young, gives g units of water to its parent.
i. Under what conditions does there exist g > 0 such that the con-
vention makes every generation better off (than with autarky)?
Ignoring the preferences of generation 0 (who is only alive when
old), characterize the optimal social convention.
ii. Is it possible to sustain the optimal social convention in a sub-
game perfect Nash equilibrium? Does your answer depend on
δ?
iii. Consider any social convention that makes every generation bet-
ter off (than with autarky). Which of these conventions are
sustainable in subgame perfect Nash equilibria? Does your an-
swer depend on δ?
iv. Consider any social convention that makes every generation worse
off (than with autarky). Which of these conventions are sus-
tainable in subgame perfect Nash equilibria? Does your answer
depend on δ?
v. How does your answer change if the horizon is finite (that is, if
there is a last generation)?
(b) Suppose that each generation is endowed with F units of food in the
second period of life, and none in the first period. Lifetime utility is
given by v(f1 ) + γv(f2 ), where fi is the amount of food consumed in
the ith period of life, and v is strictly increasing and strictly concave.
Suppose that food isn’t storable, and that there are no markets. Con-
sider the following social convention: each generation, when old, gives
b units of food to its child. Which social conventions ar sustainable
as subgame perfect Nash equilibria? Which are not sustainable?
Does your answer depend on γ?
50
(c) Suppose that each generation is endowed with W units of water in
the first period of life, and none in the second period; also, each
generation is endowed with F units of food in the second period
of life, and none in the first period. Lifetime utility is given by
w1 + f1 + δw2 + γf2 . Assume that δ > 1 and that γ < 1. Consider
the following social convention: each generation t gives g units of
water to its parent in the second half of period t, and gives b units
of food to its child in the first half of period t + 1 (so that, in each
period, there is a transfer of food from parent to child, followed by a
transfer of water from child to parent).
i. Which social conventions are sustainable in subgame perfect Nash
equilibria, and which are not? How does your answer depend
on δ and γ (remembering that δ > 1 and γ < 1)? With respect
to transfers of food, how do you explain any differences between
your answer here and in response to part (b)?
ii. What is the optimal social convention (ignoring generation 0),
subject to the constraint that it must be sustainable as a sub-
game perfect Nash equilibrium? How does your answer depend
on δ and γ?
77. Two (and only two) companies, A and B, produce automobiles, each at a
cost c per unit. There is one consumer, who purchases exactly one auto-
mobile, or none (cars are indivisible, and a second car is worth nothing).
The consumer receives a benefit of vA > c from buying A’s car, and a
benefit of vB > c from buying B’s car.
(a) Suppose the companies are Bertrand competitors (they name prices
simultaneously, and the consumer picks the best deal). What is
the equilibrium outcome? You may ignore any equilibria in which
a car maker names a price below its cost, and you may resolve the
consumer’s indifference in whatever way is necessary to create the
equilibrium. Explain your answer completely (a proof is best but
not required). Also, without loss of generality, you may assume that
vA ≥ vB .
(b) Now consider a two-stage game. In the first stage, firm A picks ad-
vertising expenditures, d. Without advertising, the consumer views
the cars as identical: vA = vB . Advertising causes the consumer to
place a higher value on A’s car: vA = vB + f (d), where f is strictly
concave and f 0 (0) > 1. The second stage is the same as the game
described in part (a). Restrict attention to pure strategy subgame
perfect Nash equilibria.
i. For each level of d, write an expression for A’s profits, given the
second-stage equilibrium.
51
ii. What level of advertising expenditure does A choose in the first
stage? Provide a general √ characterization, and solve explicitly
for the case where f (d) = d.
(c) Now modify the game described in part (b) as follows. In the first
stage, A and B simultaneously pick advertising expenditures, dA and
dB . Advertising causes the consumer to place a higher value on i’s
car: vi = v0 + f (di ), where v0 > c, and f 0 (0) > 1. The second
stage is again the same as the game described in part (a). Restrict
attention to pure strategy subgame perfect Nash equilibria.
i. Given the second-stage equilibrium, describe each firm’s profits
as functions of the advertising levels.
ii. Is there a pure strategy subgame perfect Nash equilibrium in
which both firms choose the same level of advertising? If so,
identify it, and indicate each firm’s profit
√ level (do this generally
and for the special case where f (d) = d). If not, explain why
not.
iii. Is there a pure strategy subgame perfect Nash equilibrium in
which the firms choose different levels of advertising? If so,
identify it, and indicate each firm’s profit
√ level (do this generally
and for the special case where f (d) = d). If not, explain why
not.
78. A monopolist sells a “network” good to I consumers. When N consumers

buy the good at price p, each of the buyers recieves a net benefit of v +
sN − p. Note that v + s is the value a consumer receives if she alone
consumes the good, and pays nothing for it. We will assume s > 0, so
that the benefit of consumption increases with network size. The unit
cost of producing the good is constant, and equal to c. Assume that v > c.
(a) Now consider the following game. First, the monopoloist names a
price p. Then all I consumers simultaneously decide to BUY or
NOT BUY.
i. For what prices p, if any, is buying a dominant action for a
consumer? For what prices p, if any, is not buying a dominant
action for a consumer?
ii. Now consider the set of prices for which neither buying, nor not
buying, is a dominant action. Is there a continuation equilibrium
where everyone buys? Is there a continuation equilibrium where
no one buys? Are there any other pure-strategy continuation
equilibria for these prices?
iii. Solve for all pure-strategy subgame perfect Nash equilibria. What
prices can arise in these equilibria?
52
(b) Now suppose that the benefit consumer i derives from the good is
vi + sN − p, where vi is a random variable taking on values in the
set {w1 , ..., wM } (where wk is strictly increasing in k) with probabili-
ties {λ1 , ..., λM }, where the draws are independent across consumers.
Each consumer’s type is private information (neither the monopolist
nor other consumers observe it). The game is otherwise the same.
Each choice of p induces a game of incomplete information among the
consumers; through the following series of steps, you will examine the
properties of the symmetric pure strategy Bayesian equilibriaP of these
M
games. The following notation is used below: Λk = (I − 1) j=k λj
(that is, in a group of I − 1 people, the expected number with values
of wk or greater), and Wk (p) = wk + s(Λk + 1) − p.
i. Fixing p and focusing just on the game played amongst the con-
sumers, describe each consumer’s pure strategy set. Give an
example of a pure strategy.
ii. Again, fix p. Show that, given any strategies for other con-
sumers, if it is optimal for a consumer to buy with value wk ,
then it is also optimal for her to buy with value wr for r > k.
iii. Again, fix p. Assume that Wk (p) ≥ 0, and that Wk−1 (p) ≤
0. Show that there exists a symmetric pure strategy Bayesian
equilibrium in which each consumer i buys iff vi ≥ wk .
iv. Again, fix p. Using step iii, show that, if Wk (p) ≥ 0, then
the game has a symmetric pure strategy Bayesian equilibrium in
which each consumer i buys iff vi ≥ wr for some r ≤ k.
v. Again, fix p. Using step iv, show that a symmetric pure strategy
Bayesian equilibrum exists.
vi. Suppose that each choice of p leads to the the symmetric pure
strategy Bayesian equilibrium involving the largest possible num-
ber of buyers. Using step v, show that the expected quantity
sold, x(p), is non-increasing. In other words, show that, for any
two prices p and p0 with p > p0 , and for any symmetric pure
strategy Bayesian equilibrium associated with p, there is a sym-
metric pure strategy Bayesian equilibrium associated with p0 for
which the expected quantity sold is at least as large.
vii. Now consider the monopolist’s decision. Show that the price
charged by the monopolist is non-decreasing in c. (Hint: you
only need to use the fact that x(p) is non-increasing. But be
careful, as x(p) is not continuous.)
79. Consider the following Hotelling spatial location model. The product
space is [0, 1]. There are two firms, A and B. The characteristics of
the products produced by these firms are fixed at xA = 0 (for A) and
xB = 1 (for B). Let pi denote the price of a unit of firm i’s product.
53
The production cost is c per unit for both firms. Consumers are differ-
entiated by a characteristic θ, which has a uniform distribution on [0, 1];
for convenience, normalize the total mass of consumers to unity. Each
consumer buys one unit or nothing. When a consumer buys from firm i,
his payoff is v − pi − ti |θ − xi |. Each consumer purchases the product
that yields the highest payoff, and those who are indifferent split evenly
between the firms. Throughout, you may assume that v is sufficiently
large so that all consumers buy a unit from one of the firms, and you may
confine attention to pure strategy equilibria.
Notice the new twist: the “transportation cost,” ti , differs depending on
which product is purchased. Think of ti as measuring the degree to which
the product is tailored to particular tastes. Later in this problem, you
will be asked to treat ti as a choice variable for firm i. In other words,
you will use the Hotelling model to ask whether firms will tend to produce
specialized products with narrow appeal, or adaptable ones with broad
appeal.
(a) For the moment, TREAT tA AND tB AS FIXED, and consider the
static game in which the firms simultaneously choose prices only.
Solve for the following (in each case making sure your answer covers
the case in which tA = tB = 0):
i.sales as a function of prices and transportation costs,
ii.payoffs as a function of prices and transportation costs,
iii.best response functions,
iv. Nash equilibrium prices as a function of transportation costs,
and
v. Nash equilibrium payoffs as a function of transportation costs.
(b) Now consider a game in which the firms choose the pairs (pi , ti )
simultaneously, where ti ∈ [0, t] (where t is finite). Assume the
choice of ti has no implications for costs – a low ti is “free.” Solve
for the Nash equilibrium.
(c) Now consider a two stage game. In stage 1, the firms simultaneously
select ti , where the only options are ti = 0 or ti = t > 0 (where t
is finite). In stage 2, they simultaneously select pi , having observed
(tA , tB ). Assume again that the choice of ti has no implications for
costs – a low ti is “free.”
i. Identify the strategy set for each firm.
ii. Solve for the subgame perfect Nash equilibrium. (Hint: part (a)
gives you the continuation payoffs for any (tA , tB ).)
(d) Now consider the same game as in part (c), but assume that ti can
be any number in the interval [0, t].
i. Identify the strategy set for each firm.
ii. Solve for the subgame perfect Nash equilibrium.
54
(e) Provide an intuitive explanation for any similarity or difference you
notice in comparing your answers for part (b) with parts (c) and (d).
80. Suppose two players play the following prisoners dilemma two times in
succession:
C F
C x,x -1,4
F 4,-1 0,0
The players observe the choices from the first play of the game before play
of the second. Payoffs are the (undiscounted) sum of their payoffs in the
two plays of this game. For parts (a)-(f), assume that x = 3. Also, focus
exclusively on pure strategies.
(a) Identify the players strategy sets. How many strategies does each
player have? How many strategy profiles are there altogether?
(b) Do the players have any strictly dominated strategies? How many?
What are they?
(c) Do the players have any weakly dominated strategies? How many?
What are they?
(d) Identify all pure strategy Nash equilibria of this game. How many
are there?
(e) Which of the equilibria you identified in part (d) are normal-form
trembling hand perfect?
(f) Which of the equilibria you identified in part (d) are subgame perfect?
(g) Now we modify the game as follows: a player must pay a finking tax
of T > 1 if he or she plays F twice. Suppose as well that T > 4 − x
and that x ∈ (0, 4) . A players payoff is the (undiscounted) sum of
his or her payoffs in the two plays of this game less any finking tax he
or she must pay. What are the pure strategy subgame perfect Nash
equilibria of this modified game? How do they depend on x?
Repeated Games
55
82. Consider the following game in normal form (where player 1 chooses rows,
player 2 chooses columns, and player 1’s payoff is given first in each cell).
a b c
A 5,3 5,5 3,4
B 4,10 9,9 4,11
C 3,3 11,4 5,5
(a) Suppose that this game is repeated a finite number of times, and
that the discount factor (δ) is unity. Characterize the set of subgame
perfect equilibria.
(b) Suppose that the game is repeated infinitely, and that δ is less than
unity. Suppose that the players try to sustain the cooperative choice
(B, b) in every period through Nash reversion. For what values of δ
is this an equilibrium? Is it subgame perfect?
84. Consider an infinitely repeated Bertrand duopoly with identical firms,

discount factor δ < 1, and constant unit costs of c > 0. Determine the
conditions under which Nash reversion sustains the monopoly price in each
of the following cases.
(a) Market demand in period t is Q(p) = γ t Q(p), where γ > 0.

(b) At the end of each period, the market ceases to exist with probability
λ.
(c) It is possible to observe the other firm’s price only after a delay of K
periods.
85. Consider the linear Cournot oligopoly model where price p is given as
a linear function of total output Q, p = a − bQ, and where output is
produced by each firm i at constant marginal cost c (same for all firms).
Suppose there are N firms. Suppose also that the stage game will be
repeated infinitely, and that firms discount future payoffs at the rate δ.
Let δ N be the minimum discount factor for which it is possible to sustain
complete collusion in subgame perfect Nash equilibrium, using Nash re-
version. Solve for δ N as a function of N . How does it vary with N ? Is
this what you would expect? Why, or why not?
86. Consider the infinitely repeated oligopoly game for two firms. Suppose
that demand and costs are linear, as in problem 85.
56
(a) Suppose that firms choose quantities as the strategic variable (i.e. we
have the infinitely repeated Cournot game). Characterize the highest
symmetric payoffs sustainable through Nash reversion in a subgame
perfect Nash equilibrium, as a function of the discount factor.
(b) Repeat part (a) under the assumption that firms choose prices (i.e.
we have the infinitely repeated Bertrand game).
87. Two firms, 1 and 2, are the only suppliers of a completely homogeneous
good. There is an infinite sequence of discrete periods. In each period,
the firms compete by simultaneously naming prices, pi ∈ R+ . Consumers
purchase the good from the firm naming the lowest price, and are indif-
ferent between the firms if the prices are identical. (You can resolve
consumer indifference however you like to sustain an equilibrium.) Total
demand is equal to one (infinitely divisible) unit when price is less than
or equal to some reservation level v, and equals zero if price exceeds v.
The unit cost of production for firm i is ci , with c1 < c2 ; production is
characterized by constant returns to scale. The choices in each period are
observed by both firms prior to the next period, and the two firms share a
constant discount factor, δ. For the following questions, focus exclusively
on pure strategy subgame perfect equilibria with the following stationarity
property: within any subgame, the same outcome occurs in every period
on the equilibrium path.
(a) Show that there is an equilibrium that yields zero profits for both
firms. Is there an equilibrium that yields lower payoffs to either firm?
Determine the set of sustainable prices (that is, prices observed in
the aforementioned class of equilibria) as a function of the discount
factor.
(b) Consider the same problem, only now imagine that the two firms
operate in two markets that are identical except that the roles of
the firms are reversed (one firm is the low cost producer in one mar-
ket, while the other is the low cost producer in the other market).
Focus again on pure strategy subgame perfect equilibria with the sta-
tionarity property described above, but in this case further restrict
attention to equilibria in which each market is served exclusively by
the low cost producer, and where the price is the same in both mar-
kets. Determine the set of sustainable prices (that is, prices observed
in the aforementioned class of equilibria) as a function of the discount
factor. Is it possible to sustain some price above c2 for all δ > 0?
How does the highest sustainable price change as δ increases?
88. Consider the following T period game between the proprietor of a business
and N potential managers. At the beginning of each period t = 1, 2, ..., T
the proprietor, P, MUST elect one of the managers to run the business.
57
The manager selected then decides what fraction of the business income (of
that period) she will expropriate for herself. Let mt ∈ {1, 2, ...N } denote
the manager selected in period t, and let xt ∈ [0, 1] denote the expropriated
fraction. Total business income per period is constant through time and
normalized to one. The fraction of business income not expropriated, i.e.,
1 − xt , is consumed by P. A manager’s per-period payoff is given by xt
if selected that period, and zero otherwise. The proprietor’s per-period
payoff is given by (1 − xt ). All parties (the proprietor and every manager)
discount future payoffs according to discount factor δ ∈ (0, 1).
(a) Assume that N = 2 (two potential managers) and T = 2 (two peri-

ods).
i. What is the highest possible payoff that P can obtain in a sub-
game perfect Nash equilibrium? Provide an equilibrium strategy
profile that sustains such a payoff, and explain why it is impos-
sible for P to obtain a higher payoff.
ii. What is the lowest possible payoff that P can obtain in a sub-
game perfect Nash equilibrium? Provide an equilibrium strategy
profile that sustains such a payoff, and explain why it is impos-
sible for P to obtain a lower payoff.
iii. Is the set of PURE strategy subgame perfect equilibrium payoffs
for the proprietor convex?
(b) Now assume T = ∞ and N = ∞. Does there exist a subgame perfect
Nash equilibrium such that, on the path-of play, xt < 1 for all t? If
so, provide an example. If not, explain why not. How would your
answer change with finite N ? How would your answer change for
N = 2?
89. Consider a game of mutual gains where players 1 and 2 can choose to
jointly hunt a stag (S) or individually hunt a hare (H). If they both
choose to hunt stag (S), their payoffs are (6, 6), but if they both choose
to hunt hare (H), their payoff is (0, 0). If one chooses stag and the other
chooses hare, the one who chooses stag receives 2 and the other receives
7. In a simultaneous move game the payoff matrix is as follows:
S H
S 6, 6 2, 7
H 7, 2 0, 0
(a) In a one-shot simultaneous-move game, characterize the set of all

Nash Equilibria, and draw the convex hull of Nash equilibrium pay-
offs.
(b) Now imagine that the simultaneous-move stage game is repeated
twice, and that at the end of each round, actions in the previous
58
round are observable. If each player seeks to maximize the sum of
his two payoffs, is there an SPE where (S, S) is played in the 1st
round? If so, construct it, and if not, explain why not.
(c) Now imagine that the game is repeated twice, but at every round,
player 1 moves first, 2 observes 1’s move, and this is common knowl-
edge. Solve for the sub-game perfect equilibrium outcome.
90. Consider the infinitely repeated Bertrand competition game with two firms
and homogeneous goods. Assume that both firms discount future payoffs
at the rate δ. Suppose when both firms set the same price, the demand
is split equally between them. Each firm has unit cost c, the demand
function is D(p), and suppose the profit function π(p) = (p − c)D(p) has a
unique maximizer pm and π(p) is increasing for all p ≤ pm . Suppose that
for i = 1, 2, firm i cannot detect a deviation until ki ≥ 1 periods after the
deviation (so ki = 1 for both i is the standard case with no detection lag
in class). k1 and k2 are common knowledge.
(a) Find, as a function of δ, the highest symmetric profit (or equiv-

alently, highest symmetric price) sustainable in a subgame perfect
equilibrium.
(b) Now suppose that at the beginning of the game, each firm i simul-
taneously chooses any positive integer ki or φ (meaning no detection
at all). Choosing φ incurs no cost, while the cost of choosing detec-
tion after k periods is given by function c(k) for each firm, where
c > 0, c0 < 0. The choices of both firms are perfectly observable.
Solve for the pure strategy subgame perfect equilibrium (equilibria)
of this game.
91. Consider an infinitely repeated game with 2 firms. In each period, two
firms simultaneously choose a characteristic of their products from a bi-
nary set {a, b}, if they choose the same characteristic, they split the market
equally and each gets stage payoff of π2 , otherwise firm 1 always gets the
entire market (so firm 1 gets stage payoff of π and firm 2 gets 0).
(a) What is the minmax payoff for each firm in the stage game?
(b) What is the highest payoff for each firm that can be an outcome of
a Nash equilibrium of the infinitely repeated game?
92. Consider the infinitely repeated version of the 3×3 game illustrated below.
(a) Find all of the pure strategy Nash equilibria of the stage game.
(b) Determine the lowest discount factor, δ ∗ , for which it is possible
to sustain the symmetric cooperative outcome, (A, a), using Nash
reversion.
59
(c) Does there exist any other subgame perfect Nash equilibrium in which
it is possible to sustain (A, a) at a lower discount rate, δ < δ ∗ ?
Provide a proof.
Player 2
a b c
Player 1 A 8,8 -5,9 1,5
B 9,2 -6,-3 0,6
C 4,1 3,3 2,4
93. Consider the infinitely repeated Cournot competition game with two firms.
Each firm has unit cost c, the demand function is p = a − bQ, a >
c > 0, b > 0. Consider generalized “stick-and-carrot” strategies based
on two quantities, qL and qH , where qL = a−c 4b (the “carrot”) is half of
the monopoly quantity, and qH = a−c 2b (the “stick”) is half of the perfect-
competition quantity as in class. For these strategies, after and deviation,
the stick is applied for k consecutive periods, rather than for a single
period.
(a) Write each firm’s equilibrium incentive constraint for each period in
which the carrot is prescribed by the stick-and-carrot strategies, and
for each period in which the stick is prescribed. (Hint: there should
be k + 1 separate constraints.)
(b) For which values of δ does the k-period version of the stick-and-carrot
strategies constitute a subgame perfect equilibrium? How does your
answer depend on k? Which value of k allows the firms to sustain
the monopoly outcome with the lowest discount factor? Explain.
94. A society consists of N liberals and M conservatives. All of these individu-

als are concerned about a policy x. All citizens attempt to influence policy
by making contributions to lobbying. Let bi ≥ 0 denote the contribution
to lobbying by the ith liberal, and let gj ≥ 0 denote the contribution to
lobbying by the jth conservative. Let B denote total contributions by
liberals, and let G denote total contributions by conservatives. Suppose
the policy outcome is x = B − G. For a liberal making a contribution of
b, payoff is −(x − c)2 − b2 . For a conservative making a contribution of
g, payoff is −(x + c)2 − g 2 . Thus, from the liberals’ perspective, the ideal
policy is x = c > 1, and from the conservatives’ perspective, the ideal
policy is x = −c.
(a) Suppose the game is played once.

i. Solve for the optimal contribution of a conservative, and the
optimal contribution of a liberal, as a function of the total net
contributions by all other players.
60
ii. Using your answer to (a.i), show that there is no Nash equilib-
rium in which two liberals make different contributions. Sim-
ilarly, show that no two conservatives make different contribu-
tions.
iii. Consider a Nash equilibrium where every liberal contributes b∗
and every conservative contributes g ∗ . Solve for b∗ and g ∗ , and
determine the policy that is selected.
N
iv. Fix the ratio of liberals to conservatives (that is, M = k), and
let the population get large. What happens to contributions per
capita? Do they rise or fall? To what values do they converge?
What happens to the chosen policy?
v. In the typical public goods problem, per capita contributions
converge to zero as the population gets large. Explain intuitively
why this result does or does not hold here.
vi. Considering the payoffs of just the liberals and conservatives (and
not those of the policy maker), is the equilibrium Pareto efficient?
Focusing on the case of M = N , what is the most efficient way
to achive the same policy produced by the equilibrium?
(b) Now suppose that the game is repeated an infinite number of times.
Each player discounts the future at the rate δ. Suppose that all
actions in round t are known to all players as of the start of round
t + 1. Confine attention to the case of N = M .
i. In part (a.vi), you were asked to identify the most efficient way
to achieve the same policy outcome that prevails in the one-shot
equilibrium. Using Nash reversion to punish deviations, for what
discount factors is this outcome sustainable?
ii. As the population size rises, does it become harder or easier to
sustain this efficient outcome? Explain your answer.
95. N families send one child each to a private school. All of the families are
identical. The representative family’s preferences correspond to the utility
function U (e, c),where e is the level of educational spending (per pupil),
and c is the consumption of other goods. The family has gross income y.
It pays an income tax equal to the fraction t of its taxable income. Taxable
income equals gross income minus contributions to charitable causes (and
the school is a charitable cause). So if the family contributes g ≥ 0 to
the school, its taxable income is y − g, and its tax liability is t(y − g).
The familyalso pays tuition f ≥ 0. Its consumption of other goods, c, is
less than or equal to whatever is left over after contributions, taxes, and
tuition (in other words, it can either spend this residual income on the
private consumption good, or throw it away).
(a) Suppose all the families can cooperate with the school to set the
per-pupil spending level and the representative family’s contribution
61
level and the tuition level. Characterize the solution. Solve for it
explicitly for the case where U (e, c) = ec.
(b) Now consider the following game. First, the school names per pupil
spending, e. Its objective is to maximize the sum of the utilities of
all the families. Second, families simultaneously choose their con-
tributions. Third, the school sets the per pupil tuition fee required
to make up the difference between the promised level of per-pupil
spending, and per-pupil contributions received. (The promised level
of educational spending is binding.)
i. Identify the strategy set for each player.
ii. Characterize the symmetric subgame perfect Nash equilibria as
a function of the model’s parameters. Solve for it explicitly for
the case of U (e, c) = ec.
iii. How do the cooperative and non-cooperative solutions compare?
Explain why there is or is not a difference.
(c) Now suppose the game in (b) is repeated an infinite number of
times (the kids never graduate), and that the representative fam-
ily discounts future utility at the rate δ. Focus on the case where
U (e, c) = ec.
i. Using Nash reversion to punish deviations, for what discount fac-
tors is the cooperative solution sustainable as a subgame perfect
Nash equilibrium?
ii. Take the limit as N goes to infinity. What is the threshold
discount factor for which cooperation is sustainable? How does
this vary with the tax rate?
(d) Now suppose the school’s promised level of educational spending is
non-binding, and that it can select a different level in the third stage.
i. Identify the strategy set for each player.
ii. For large N , characterize the symmetric subgame perfect Nash
equilibrium of this model.
iii. How does the solution compare to the case where the school’s
promise is binding? Is there a difference? Why or why not?
96. There are n people in a group. Each person has friends and enemies in
the group (everyone’s friends and enemies are common knowledge). We
can illustrate these relationships with the following type of graph:
1 2 3
Here, a “link” indicates that people are friends (people are automatically
linked with themselves). No link indicates that they are enemies. So,
62
player 1 is friends with player 2 (and player 2 is friends with player 1),
but player 1 is enemies with player 3 (and vice versa). Player 2 is friends
with player 3 (and vice versa).
We will analyze a game that involves passing on a secret. The secret is
“hard information”: players simply choose whether or not to pass it on;
they cannot lie about it. At first, only player 1 knows the secret. The
game is played as follows:
1) Player 1 chooses a subset of her friends (possibly empty), to whom she
passes on the secret. If the chosen subset is empty, the game ends. If it
is non-empty, the game continues, as follows.
2) Those who just received the information choose subsets of their unin-
formed friends (possibly empty), to whom they pass on the secret. If the
chosen subsets are all empty, the game ends. If it is non-empty, the game
continues by repeating this step.
Let xi = 1 if person i learns the secret by the end of the game, and
xi = 0 if she does not. P Each person
P has the following utility function:
ui (x1 , x2 , x3 , ...) = α xj − β xk where α ≥ 0 and β ≥ 0, and Fi
j∈Fi k∈Ei
and Ei are the sets of i’s friends and enemies, respectively.
For parts (a)-(d) and (f), assume that α = 1 and β = 2.
(a) Suppose the group consists of only two people, who are friends. Note
that player 2 has no choices. What is player 1’s strategy set? Does
player 1 have a dominant strategy? What is the SPNE of this game?
(Warning: don’t look for a trick or something deep here. It’s just a
simple first step.)
(b) Suppose the group consists of 3 people whose relationships corre-
spond to the graph below. What are the SPNE of this game? What
are the NE of this game?
1 2 3
(c) Suppose the group consists of 4 people whose relationships corre-

spond to the graph below. What are the SPNE of this game? What
are the NE of this game?
1 2 3 4
(d) Suppose the group consists of n people arrayed in a line, as above

(see the graph below). What are the SPNE of this game? How does
it depend on n?
63
1 2 n−1 n
(e) Solve question (d) for arbitrary α and β. Show how the SPNE of
the game depend on n, α and β. Your answer does not need to
be completely formal, but try to characterize the way in which the
equilibrium depends on the parameters.
(f) Now, once again, assume that α = 1 and β = 2. Consider a repeated
version of the game in problem (c) with 3 players:
1 2 3
(a) So, at every time t, player 1 learns a new secret and must decide
whether to tell player 2 (who, if told, must decide whether to tell
player 3). Each players’ utility is the sum of the payoffs in each
stage game discounted at the geometric rate δ.
i. Suppose the number of repetitions is finite. Is there a SPNE
in which player 1 tells player 2 the secret? If so, describe the
equilibrium. If not, explain why.
ii. Suppose the number of repetitions is infinite. Also suppose
player 1 CANNOT verify whether player 2 tells player 3 the
secret (Even so, player 1’s utility still depends on whether player
3 learns the secret.). Is there a SPNE in which player 1 tells
player 2 the secret? If so, what is the lowest discount factor for
which such an equilibrium exists?
iii. Suppose again that the number of repetitions is infinite. If player
2 passes the secret on to player 3, player 1 discovers that fact with
probability p; moreover, player 2 always knows whether player 1
has made that discovery. Is there a SPNE in which player 1 tells
player 2 the secret? If so, what is the lowest discount factor for
which such an equilibrium exists (as a function of p)?
97. A large tree sits on the border of two neighboring countries, A and B. The
countries can choose, simultaneously, to either fight (F) to win the fruit
from the tree or to surrender (S). Fighting, however, destroys the fruit.
If both countries surrender, they can share the fruit from the tree. The
payoffs are shown in the following matrix, where Country A chooses rows
and Country B chooses columns.
64
S F
S 3,3 1,4
F 4,1 0,0
(a) Find all of the Nash equilibria of this game, pure and mixed. What
are the associated payoffs?
(b) What is the highest symmetric payoff that can be achieved in a cor-
related equilibrium?
Now imagine this game is repeated every year when the tree blooms and
that the countries discount future payoffs at rate δ ∈ (0, 1). Let T denote
the life span of the tree.
(c) If T = 2, is there a subgame perfect equilibrium in which both coun-

tries surrender in the first period? If so, describe the strategies that
generate cooperation. If not, explain why not. Does your answer
depend on δ?
(d) If T = 3, is there a subgame perfect equilibrium in which both coun-
tries surrender in the first period? If so, describe the strategies that
generate cooperation. If not, explain why not. Does your answer
depend on δ?
(e) Suppose T = ∞. Is there a subgame perfect equilibrium in which
both countries surrender in every period on the equilibrium path? If
so, describe the strategies that generate this cooperation. What is the
lowest δ for which there exists a subgame perfect Nash equilibrium
in which both countries surrender in every period on the equilibrium
path?
98. Two firms compete to sell pencils. The (downward-sloping, continuous)

demand for pencils is given by Q? = max(A − Bp, 0), where p is price.
Production involves a fixed cost, F, that is incurred only if a positive
amount is produced, and a constant variable cost-per-unit of c < A. As-
sume (A − Bp)(p − c)/2 − F > 0 for some price p (so that both firms can in
principle operate profitably). The firms compete by simultaneously nam-
ing price. Consumers purchase pencils from the firm naming the lowest
price. If the firms name the same price, consumers are willing to buy from
either one, and you may resolve their indifference as needed to support
equilibria.
65
(a) Suppose this game is played once. Identify the set of pure strategy
Nash equilibria, and the associated profit level.
Now suppose the firms play this game repeatedly (and indefinitely),
discounting future payoffs at the geometric rate δ. They are consid-
ering two modes of collusion, as follows:
Mode 1: Set some profitable price, splitting the market equally in
each period.
Mode 2: Take turns producing positive amounts in alternate periods.
They flip a coin to determine which firm goes first; they are both risk
neutral.
(b) What is the lowest discounted payoff achievable in a sub game perfect
equilibrium? Exhibit an equilibrium that achieves it, and explain
why there is no equilibrium that achieves a lower payoff. Does your
answer depend on the discount factor?
(c) Using the equilibrium you identified in a part (b) as a punishment,
determine what the firms can achieve through Model1, as follows:
i. Focusing on stationary equilibrium paths (along with the market
price p is the same every period), write the incentive constraint
that governs a firm’s decision to comply with the equilibrium or
deviate in any given period (conditional on that value of p).
ii. For any given price p, for what discount factors is it possible to
sustain that price on a stationary subgame perfect equilibrium
path?
iii. Determine the highest firm-level discounted profits sustainable
in a stationary subgame perfect Nash equilibrium as a function
of δ.
(d) Using the equilibrium you identified in part (b) as a punishment,
determine what the firms can achieve through Mode 2, as follows:
i. Focusing on a stationary equilibrium paths (along with the mar-
ket price p is the same in every period), write the incentive con-
straints that govern both the decision to comply with the equilib-
rium or deviate in any given period for both the currently active
firm and the currently inactive firm (conditional on that value of
p).
ii. For any given price p, for what discount factors is it possible to
sustain that price on a stationary subgame perfect equilibrium
path?
iii. Determine the highest firm level discount profits sustainable in
a subgame perfect Nash equilibrium as a function of δ. (Here,
compute an expectation based on the coin flip that determines
which firm goes first).
(e) Is one form of collusion necessarily better (easier, more profitable)
than the other? Upon what does your answer depend? (Note: a
qualitative answer will suffice, but refer to your results.)
66
99. In the market for new economists, there are two departments i ∈ {1, 2}
and three recent graduates j ∈ {1, 2, 3}. When recent grad j goes to
department i, the department increases their reputational payoff by i · j.
Thus, department i’s payoff for being matched with graduate j, when
offering wage p is i · j − p. Assume throughout that firms have to offer
p ≥ 0.
To save costs, the economics profession institutes a centralized matching
system to pair recent Ph.D. graduates with their future employers. Instead
of offering personalized wages, each department submits a single wage
and whichever department submits the highest wage gets the best new
graduate (here j = 3); the other firm gets the next best graduate (here
j = 2). If they offer the same price, department 2 gets the higher quality
graduate.
(a) What are the strategy sets of the two departments if Department
1 has to submit their price before Department 2? What if the two
departments submit their wage simultaneously?
For the rest of the problem, assume that that wages are submitted
simultaneously.
(b) Is there a pure strategy Nash equilibrium? If so, characterize it. If
not, explain why not.
(c) Now we will look for a mixed strategy Nash equilibrium. Let Fi
denote the CDF governing department i’s choice of wage. You may
assume that the support of Fi is an interval [0, p0 ] for some p0 > 0
(where p0 is the same for both departments), that F1 has an atom at
0 but nowhere else, and that F2 is atomless.
i. Write an expression for the expected payoff of department 2 as a
function of the wage it sets, given F1 . Also write an expression
for the expected payoff of department 1 as a function of the wage
it sets, given F2 .
ii. What is Department 1’s equilibrium profit? Using your answer,
solve for p0 and department 2’s equilibrium CDF, F2 (p).
iii. What is Department 2’s equilibrium profit? Using your answer,
solve for department 1’s equilibrium CDF, F1 (p)?
iv. Explain why, in any mixed strategy equilibrium, neither depart-
ment uses a CDF with an atom at any wage greater than zero.
(d) Suppose that the departments repeat the game an infinite number
of times (with new students each time). If the departments use the
average payoff criterion, what is the set of possible payoffs for the
two departments?
Now suppose that graduate three has an outside offer of p. Thus, if
the highest wage offered is less than p, then the department that offers
the highest wage is matched with Graduate 2 (and thus receives a
reputational benefit of 2 · i) and the department that offers the lowest
67
wage is matched with Graduate 1. On the other hand, if the highest
wage offered is greater than or equal to p, the matching process is
the same as above: the department with the highest wage offer gets
Graduate 3 and the other department gets Graduate 2.
(e) Suppose p ∈ [1, 2]. Is there a pure strategy Nash equilibrium? If
so, identify it; if not, explain why not.
(f) Now suppose that p ∈ (0, 1). Solve for a mixed strategy equilibrium.
You may assume that both departments choose prices in (0, p) with
zero probability, that department 1 (and only department 1) chooses
p = 0 with positive probability, that department 2 (and only de-
partment 2) chooses p with strictly positive probability, and that the
departments otherwise randomize over an interval (p, p0 ].
(g) Suppose that before the firm’s offer their wages, worker 3 can choose
p. In other words, she can commit to rejecting the offer if the highest
wage is below p for any p of her choice. In a SPNE, what p will
worker 3 chose?
Dynamic Games with Incomplete Information

100. Consider the following game of incomplete information:
1 c c 1 c
(2,6)
s s s
A
(1,1) (0,4) (3,3)
λ
N 2
1−λ
(0,1) (1,4) (2,3)
B
s s s
(3,6)
1 c c 1 c
The top half of this is an abbreviated version of the centipede game. The
bottom half is similar, except that player 1 believes that continuation is
always better for him. One might interpret this game as follows: Player 2
is playing the centipede game against an opponent, but is not certain that
his opponent understands the game. The opponent is aware of player 2’s
uncertainty in this regard.
(a) Solve for the subgame perfect equilibria of this game.

(b) Solve for the sequential equilibria of this game. Does behavior un-
ravel, as in the simple centipede game with complete information?
68
What happens as λ approaches 1? What does this tell you about how
people might play the centipede game when they are unsure about
each others’ motives?
101. A buyer and a seller are bargaining. The seller owns an object for which
the buyer has value v > 0 (the seller’s value is zero). This value is known
to the buyer, but not to the seller. Its prior distribution is common knowl-
edge. There are two periods of bargaining. The seller makes a take-it-or-
leave-it offer at the start of each period (a price) which the buyer may
accept or reject. The game ends when an offer is accepted, or after two
periods, whichever comes first. Both players discount period-two payoffs
with a discount factor of δ < 1.
Assume throughout that the buyer always accepts the seller’s offer when-
ever the buyer is indifferent.
(a) Characterize the (pure strategy) perfect Bayesian equilibria for a case
where v can take two values, {vL , vH }, with 0 < vL < vH . Use λ to
denote the probability that v = vH .
(b) Do the same for the case where v is uniformly distributed on the
interval [vL , vH ].
102. A plaintiff, Ms. P , files a suit against Ms. D (the defendant). If Ms. P
wins, she will collect π dollars in damages from Ms. D. Ms. D knows
the likelihood that Ms. P will win, λ ∈ [0, 1], but Ms. P does not (for
example, Ms. D may know if she was actually at fault). They both have
strictly positive costs of going to trial, ci (i = P, D). The prior distribution
of λ has density f (λ) (which is common knowledge).
Suppose pretrial settlement negotiations work as follows: Ms. P makes
a take-it-or-leave-it settlement offer (a dollar amount) to Ms. D. If Ms.
D accepts, she pays Ms. P and the game is over. If she does not accept,
they go to trial.
(a) What are the (pure strategy) weak perfect Bayesian equilibria of this
game?
(b) What effects do changes in the parameters cP , cD , and π have?
(c) Now allow Ms. P , after having her offer rejected, to decide not to go
to court after all. What are the (pure strategy) weak perfect Bayesian
equlibria? What are the effects of changes in the parameters now?
103. Repeat exercise 102, assuming that Ms. P knows λ and Ms. D does not.
69
104. Student/workers are of two types, L and H. There are at least two
potential employers. The incremental profits earned from hiring a worker
of type i is π i , with π H > π L . Events unfold as follows: (1) Students
observe their types (employers do not). (2) Students choose education,
e, which is costly and observed by employers, but is not productive. (3)
Employers name wages for prospective workers (currently students), w,
contingent on observed education (Bertrand competition). (4) Students
choose employers and become workers. The payoff for a student/worker
of type i receiving a wage w is given by w − ci (e). The payoff for the
employer of a type i student/worker paying a wage w is given by π i − w.
Assume: cH (e) = e2 , and cL (e) = e. Throughout, focus on pure strategy
equilibria.
(a) Does this model satisfy the Spence-Mirrlees single-crossing property?

Justify your answer.
(b) Suppose that π H − π L < 1. Are there equilibria with complete
separation of types? If so, characterize the class of equilibria with
complete separation. If not, justify your answer. What if π H −
π L > 1? (You may focus on equilibria in which type L receives no
education.)
For the remainder of this question, suppose that, in addition to
obtaining education, student/workers can also signal their type by
spending summer vacations in Europe. Suppose that the amount
spent on these vacations, d, is observable by employers (e.g. it is
proportional to the number of months spent in Europe, which is re-
ported on the student/worker’s resume). Apart from this monetary
cost, student/workers are entirely indifferent about taking European
vacations. Moreover, the cost of vacations does not depend upon
the student/worker’s type. Thus, the student/worker’s payoff is
now given by w − ci (e) − d. Imagine that student/workers choose
e and d simultaneously. Otherwise, the problem is unchanged from
above.
(c) Show that, for arbitrary π H − π L > 0, is it possible to have com-
plete separation of types in equilibrium, using a combination of the
signals e and d. Characterize the class of equilibria with complete
separation. That is, what combinations of e and d will suffice to
separate the two types? (You can restrict attention to equilibria in
which eL = dL = 0.) Among this class, which is most favorable to
type H? For this equilibrium, what happens to the payoff of type
H as the productivity of type H, π H , increases? Why?
105. Recall question 49. Suppose there are only two players, A and B, and an
infinite time horizon. However, there is a probability λ ∈ (0, 1) that player
B is crazy and plans to clap forever, and this is common knowledge. For
this problem, use the sequential equilibrium concept.
70
(a) Show that there is no sequential equilibrium in which in each period,
players clap with probability in (0,1).
(b) Show that in no period will B choose the pure action “stop clapping.”
(c) Show that in any sequential equilibrium, B wins the prize in the first
period with certainty.
106. “Seller” is the only producer of some product. It produces the product at
zero cost, and sells it at a fixed price p = 1, which is fixed by regulation.
Let v denote the product’s quality. We’ll assume that v ∈ {v1 , v2 , v3 },
with v3 > v2 > v1 > 1. Each value of v is equally likely. “Buyer” is
the only purchaser of this product. Let q denote the quantity purchased.
Seller’s payoff is q; buyer’s payoff is (v − 1)q − q 2 .
Events unfold as follows.
Stage 1: Seller observes v (buyer does not).
Stage 2: Seller decides whether to disclose v. Disclosure involves the
costless release of non-falsifiable studies, which means buyer learns v with
certainty. (In disclosing quality, seller can neither overstate not understate
v.)
Stage 3: Buyer decides how much to purchase.
(a) Draw the extensive form of this game. For each player, how do you
describe a strategy? What are the strategy sets?
(b) Suppose that, when making its purchase decision, buyer believes that
quality is v1 with probability a, quality is v2 with probability b, and
quality is v3 with probability 1 − a − b. Solve for buyer’s optimal
quantity. Derive an expression for the seller’s payoff as a function of
the expected value of v.
(c) Consider any sequential equilibrium in which a particular quality,
call it v 0 , is disclosed with zero probability on the equilibrium path.
Explain why, upon observing the disclosure of v 0 (off the equilibrium
path), buyer must believe that the quality is actually v 0 with prob-
ability one. Would the same statement hold in a Perfect Bayesian
Equilibrium?
(d) In a sequential equilibrium, does seller disclose if quality is v3 ? What
about v2 ? What about v1 ? Justify your answer completely. (Hint:
use your answers to parts (b) and (c).)
(e) Now modify the game as follows: In the first stage, there is a prob-
ability λ that the seller learns v, and a probability 1 − λ that she
does not; moreover, only the seller knows whether she has learned v.
Assume that v1 = 2, v2 = 4, and v3 = 9. In a sequential equilibrium,
does the seller disclose if quality is v3 ? What about v2 ? What
71
about v1 ? How does your answer depend on λ? Justify your answer
completely.
107. Student/workers are of two types, L and H. There are at least two
potential employers. The incremental profits earned from hiring a worker
of type i is π i , with π H > π L . Events unfold as follows: (1) Students
observe their types (employers do not). (2) Students choose education,
e, which is costly and observed by employers, but is not productive. (3)
Employers name wages for prospective workers (currently students), w,
contingent on observed education (Bertrand competition). (4) Students
choose employers and become workers. The utility for a student/worker
of type i receiving a wage w is given by U (w, e) = w − ci (e). The payoff
for the employer of a type i student/worker paying a wage w is given by
π i − w. Assume: c0H (e) < c0L (e) for all e > 0. Throughout, focus on pure
strategy equilibria.
(a) Does this model satisfy the Spence-Mirrlees single-crossing property?

Justify your answer.
(b) What level of education is obtained by a type L worker in a sep-
arating equilibrium? Derive expressions indicating the minimum
and maximum levels of education obtained by a type H worker in
separating equilibria.
(c) Derive expressions indicating the minimum and maximum levels of
education obtained by all workers in pooling equilibria.
(d) Which equilibrium is likely to prevail? Justify your answer.
(e) Now suppose the government imposes a tax on education, t (so that
the amount collected from an individual who obtains e years of edu-
cation is te). Also suppose that the government keeps all of the tax
revenue.
i. How does this tax alter the conditions for separating equilibria
and pooling equilibria that you derived in parts (b) and (c)?
(Derive new conditions.)
ii. Focus on the most efficient separating equilibrium. Derive an
expression for the derivative of the utility of a type H worker with
respect to t, the tax rate. Sign the derivative. Do the same for
a type L worker. How does the tax affect the well-being of the
workers?
(f) Suppose as in part (e) that the government imposes a tax on educa-
tion, t (so that the amount collected from an individual who obtains
e years of education is te). Now, however, suppose that the govern-
ment distributes all revenue back to the workers. Because it cannot
distinguish a type L worker from a type H worker, all workers must
72
receive the same lump-sum payment. (Assume there are many work-
ers so that each worker ignores the impact of his decision on the size
of the lump-sum.)
i. How does this tax-transfer system alter the conditions for sepa-
rating equilibria and pooling equilibria that you derived in parts
(b) and (c)? (Derive new conditions.)
ii. Focus on the most efficient separating equilibrium. Derive an
expression for the derivative of the utility of a type H worker
with respect to t, the tax rate. Sign the derivative. Do the
same for a type L worker. How does the tax affect the well-
being of the workers? Are there conditions under which the tax
yields a Pareto improvement?
108. Consider a monopolist selling a good to a buyer. The buyer only demands
1 unit of the good and has a valuation, v, for the good that is known to
her, but not to anyone else. It is however commonly known that v is
distributed U [0, 1]. The cost of producing the good is 0. The monopolist
can offer a price, p1 , to the buyer. If the buyer accepts the offer, her utility
is v − p1 and the seller’s utility is p1 . Otherwise, if the buyer rejects the
offer, the good is destroyed and both parties receive 0 utility.
(a) For a given price p1 , what is the probability with which the buyer
accepts the offer? What is the optimal price for the monopolist?
(b) Now imagine that if the buyer rejects the offer in period 1, the mo-
nopolist can offer p2 at t = 2. If the buyer accepts this offer, her
utility is δ (v − p2 ) and the seller’s utility is δp2 . Let us first consider
the case where the monopolist can commit to a sequence of prices
(p1 , p2 ). Assume throughout that δ, p1 , p2 ∈ (0, 1) .
i. If p1 < p2 < 1, what is the probability of sale in the 1st period?
What is the probability of sale in the 2nd period?
ii. If p1 ≥ p2 , under what conditions is there a buyer type, ṽ, that
is indifferent between buying in the 1st period and buying in the
2nd period? Show that if ṽ ≤ 1, then all types v ∈ (ṽ, 1) strictly
prefer to buying the product in period 1 rather than period 2,
and that all types v ∈ (p2 , ṽ) strictly prefer buying the product
in period 2. Show that if ṽ ≥ 1, then all types, v ∈ (p2 , 1)
strictly prefer buying the product in period 2.
iii. Given your answers above, for a sequence of prices (p1 , p2 ), com-
pute the probabilities of sale in the 1st and 2nd periods and the
expected revenue for the seller.
iv. Given your answer above, what are the optimal choices for (p1 , p2 )?
How does the buyer respond to these prices? (Hint: First find
the optimal p2 for any given p1 , and then compute the optimal
p1 .)
73
v. Given these optimal prices, how does the seller’s payoff compare
to that in the 1 period game?
Now imagine that the monopolist cannot commit to the 2nd period price
and will set p2 to maximize profits in period 2.
c. If the buyer expects the monopolist to set prices (p1 , p2 ) from your
answer to (b)(iv) and behaves accordingly, will the monopolist wish
to stick to the posted price p2 , or will he do better by deviating to a
different price at t = 2?
d. We shall now proceed to find a sequential equilibrium of this game.
i. Given a price p1 < 1, suppose the monopolist believes that,
conditional on the good not being purchased at t = 1, the buyer’s
valuation v is distributed U [0, v ∗ ] where v ∗ < 1. What is the
optimal (sequentially rational) price p2 for the monopolist to set
at t = 2?
ii. Assume that type v ∗ is indifferent between buying the good at
price p1 at t = 1 or the price p2 that you derived in the question
above (you don’t need to prove this). What must p1 equal (as
a function of v ∗ ) for this indifference to hold?
iii. Given your answers from above, what is the expected revenue of
the seller in terms of v ∗ ?
iv. Solve for the optimal choice of v ∗ from the perspective of the
monopolist. Calculate the expected revenue. Is this revenue
greater or less than those from (a) and (b)(v)?
109. Consider the dynamic game of incomplete information shown in the figure
below. There are two players. Nature moves first by choosing player 1’s
type (the “sender”), type 1 (t1 ) or type 2 (t2 ), both of which are equally
likely (which is common knowledge). After learning his type, player 1
chooses left (L) or right (R). After observing player 1’s choice (but not
his type), player 2 (the “receiver”) chooses up (u) or down (d). Payoffs
correspond to the numbers next to the terminal nodes in the figure; player
1’s payoff appears first, and player 2’s second.
74
(a) Are there any pure strategy separating perfect Bayesian equilibria?
If so, find all of them. Justify your answer completely.
(b) Is there a perfect Bayesian equilibrium in which both types of player
1 pool on L? Is there a sequential equilibrium in which both types
of player 1 pool on L? Justify your answer completely.
(c) Is there a subgame perfect Nash equilibrium in which both types of
player 1 pool on R? Is there a perfect Bayesian equilibrium in which
both types of player 1 pool on R? Justify your answer completely.
(d) Are there any perfect Bayesian equilibria in which t1 plays a mixed
strategy placing strictly positive probability on both alternatives,
while t2 plays a pure strategy? If so, find all of them. Justify your
answer completely.
(e) Does the Intuitive Criterion eliminate any of the perfect Bayesian
equilibria identified above? Explain.
(f) How do your answers to parts (a) and (b) change when type 1 is cho-
sen with arbitrary probability µ ∈ (0, 1) rather than with probability
0.5?
110. The President is deciding between two alternatives: enact a new educa-
tional program (E) or not enact and stay with the status quo (N ). The
status quo will give the president a (commonly known) payoff of 0. The
payoff from the new education program depends on the state of the world.
75
With probability 14 , the new program will be good (G) and give the Pres-
ident a payoff of 1. With probability 43 , it will be bad (B) and provide a
payoff of -1.
Hoping to obtain sage advice, the President hires an Economist. He is
uncertain, however, of the Economist’s competance. The Economist is
knowledgeable with probability 14 , and ignorant with probability 34 . If
the Economist is knowledgeable (K), he is aware of the state of the world.
If the Economist is ignorant (I), he does not know the state of the world
(and therefore believes that the program will be good with probability 14 ).
The Economist knows his own type.
The game proceeds as follows:
Stage 1: Nature chooses the state of the world (G or B) and the type of
Economist (K or I). The knowledgable Economist observes the state of
the world.
Stage 2: The Economist advises the President about the state of the
world. Specifically, he sends one of two messages, either g (for good) or
b (for bad). The Economist cannot profess ignorance, and is constrained
to send one of those two messages.
Stage 3: The President, viewing the advice, chooses either E or N .
After viewing his payoff (either −1, 0, or +1), the President, who is a good
Bayesian, infers that the Economist is knowledgable with probability µ.
The Economist’s payoff is simply µ (he cares about his future reputation).
All players are risk-neutral. Throughout, apply the PBE equilibrium
concept.
(a) Identify each player’s strategy set. (You may treat the Economist
as either one player who learns his type, or as two players, one for
each type.)
(b) Now you will explore the possibility that there is an equilibrium in
which the knowledgeable Economist gives informative advice based
on the state of the world (g when G and b when B), and the ignorant
Economist always sends the message b.
i. In such an equilibrium, what would be the probability that the
President receives the message g? What would be his belief
about the state of the world given he receives g? What will the
President do when he receives the message g?
ii. In such an equilibrium, what would be the probability that the
President receives the message b? What would be the probability
that he receives the message b and that the state of the world
is G? What is his belief about the state of the world given he
receives the message b? What will the President do when he
receives the message b?
76
iii. In such an equilibrium, what would be the President’s Bayesian
beliefs about the probability that the Economist is knowledge-
able given each combination of message and outcome? (For
nodes off-the-equilibrium path, you should report the beliefs as
a range.)
iv. Is there an equilibrium of the form we have been examining? If
so, characterize it completely. If not, explain why not. (Re-
member that the President’s final beliefs about the Economist’s
type determine the Economist’s payoff.)
(c) How do your answers to part (b) change if the probability of the
Economist being knowledgable is 21 ? Interpret your results.
111. Consider the following two-stage two-player game.

Stage 1: Player A decides whether to undertake a project. The project
requires A to make an up-front investment of $10 that will return $30.
However, the returns will flow entirely to player B.
Stage 2: After observing whether A has undertaken the project (and re-
ceiving any associated returns), B makes a non-negative transfer (t ≥ 0)
to A. Note: B has the opportunity to make such a transfer whether or
not A has undertaken the project.
Initially, player A is endowed with $10 and player B with $90. Using a and
b to denote the final amounts of money held by A and B (respectively) at
the end of the game, payoffs are as follows. Player A is selfish: UA (a, b) =
a. With probability π, player B is also selfish: UB (a, b) = b. However,
with probability 1 − π, player B is egalitarian: UB (a, b) = min{a, b}.
Player B’s type is known to B but not to A.
(a) Draw the extensive form of the game. Describe a typical strategy for
each player, as well as the players’ strategy sets. Solve for the pure
strategy subgame-perfect Nash equilibria of the game. For what
values of π does A undertake the project?
(b) Now modify the game to include the following initial stage:
Stage 0: B makes a non-negative gift (g ≥ 0) to A.
i. Draw the extensive form of the game. Describe a typical strategy
for each player, as well as the players’ strategy sets.
ii. Assume provisionally that A undertakes the project if and only
if B makes an initial gift of g ∗ . For what values of g ∗ would the
selfish B be willing to give g ∗ to A in stage 0? For what values
of g ∗ would the unselfish B be willing to give g ∗ to A in stage 0?
iii. Using your answer to part (ii), characterize the set of Perfect
Bayesian separating equilibria. That is, what levels of gifts will
permit the egalitarian B to signal her type? (Hint: remember
to check that player A retains the incentive to undertake the
project, having inferred from the gift that B is egalitarian.)
77
iv. Among the equilibria identified in part (iii), which is (are) most
efficient?
v. Characterize the set of Perfect Bayesian pooling equilibria. How
does that set depend upon π?
112. Two prisoners, 1 and 2, are being interrogated over T periods. We model
each interrogation stage as a simultaneous move game with payoffs given
by the matrix below. Throughout the question assume that the players
discount the future according to δ = 1, and you may restrict attention to
pure strategies.
P2
C D
P1 C 5, 5 −10, 10
D 10, −10 −5, −5
(a) What average payoff vectors are sustainable in subgame perfect Nash
equilibria? Consider separately the cases of T = +∞ and T finite.
(b) Now we introduce incomplete information. Let each player i = 1, 2
have private information concerning his type, ti . A player’s type ti
can take on one of two values, either A or B. The payoff of each
player depends only on her own type. All payoffs are collected at
the end of round T . If ti = A, the player i’s payoff are the same as
those presented in Figure ??. If ti = B, player i is altruistic, and
for each cell in Figure ?? receives the average payoff shown in that
cell. So, letting ui (ai , a−i , ti ) denote the payoff to i from the action
pair (ai , a−i ), we have the following payoffs:
ui (C, C, A) = 5, ui (C, D, A) = −10, ui (D, C, A) = 10, and ui (D, D, A) = −5

ui (C, C, B) = 5, ui (C, D, B) = 0, ui (D, C, B) = 0, and ui (D, D, B) = −5
The probability that a player is of type A is µ, and the probability

that he is of type B is 1 − µ.
i. Suppose T = 1. Solve for a Bayesian equilibrium and show that
it is unique.
ii. Suppose T = 2. Solve for a weak perfect Bayesian equilibrum
and show that it is unique.
iii. Consider any finite T > 2. Identify the set of weak perfect
Bayesian equilibria.
(c) Next we’ll introduce a different variant of incomplete information.
The setup will be exactly the same as in part (b) above, except that
we will change the preferences of type B. In particular, the payoffs
received by a type B player will depend not only on her own type,
but also on the other player’s type. (Remember that payoffs are not
78
received until after round T ends, so players cannot infer the other
player’s type during the game from a payoff received.) In particular,
if ti = tj = B, then i is altruistic toward j, precisely as in part (b).
But if ti = B and tj = A, then i is selfish, just as a type A player.
Thus, writing i’s preferences as ui (ai , aj , ti , tj ), we have:
ui (C, C, B, B) = 5, ui (C, D, B, B) = 0, ui (D, C, B, B) = 0, and ui (D, D, B, B) = −5
but for all (ti , tj ) 6= (B, B),
ui (C, C, ti , tj ) = 5, ui (C, D, ti , tj ) = −10, ui (D, C, ti , tj ) = 10, and ui (D, D, ti , tj ) = −5
i. Suppose the game is played only once. i.e T = 1. Derive a

condition under which there is a Bayesian equilibrium in which
type A players choose D and type B players choose C (call this
condition Y ). Derive another condition under which there is a
Bayesian equilibrium in which both players choose D (call this
condition Z). Are there any other Baysian perfect equlibria?
ii. Now suppose T = 2 and that condition Y is satisfied. Suppose
players following these strategies:
Type A plays C in period 1 and D in period 2 (regardless of
what transpires in period 1).
Type B plays C in period 1. In period 2, she plays C if the
other player chose C in the first period, and D otherwise.
A. Solve for the payoffs received by both types if both players
follow these strategies.
B. For each period 1 action pair, are there beliefs that justify
the prescribed second period choices? Can you choose those
beliefs so that they are derivable from the strategy profile
whenever Bayes rule is applicable?
C. As of period 1, what is a type A’s most attractive two-period
deviation (assuming the other player follows the strategy
identified above), and what is the resulting payoff?
D. As of period 1, what is a type B’s most attractive two-period
deviation (assuming the other player follows the strategy
identified above), and what is the resulting payoff?
E. Based on your answers to (A) through (D), under what condi-
tion do these strategies and beliefs constitute a weak perfect
Bayesian equilibrium?
F. When the WPBE equilibrium you’ve identified in part (E)
exists, is it a sequential equilibrium? Why or why not?
iii. Now suppose T is finite and that condition Y is satisfied. Sup-
pose players following these strategies:
Type A plays D in period T regardless of what has previously
transpired. In period period t < T , she plays C if neither player
has previously played D, and D otherwise.
79
Type B always plays C if neither player has previously played
D, and D otherwise.
A. Solve for the payoffs received by both types if both players
follow these strategies.
B. For each history of play up to (but not including) period
t > 1 wherein some player has previously chosen D, are
there beliefs that justify the prescribed choices from that
point forward (assuming each player believes the other will
follow its strategy)?
C. For each history of play up to (but not including) period t ≥ 1
wherein no player has previously chosen D, what beliefs can
be inferred from the strategies?
D. Starting in period t and assuming no prior deviations, what
is the most attractive point in time prior to T for type A
to deviate (assuming the other player follows the strategy
identified above), and what is the greatest available payoff
from such a deviation?
E. Starting in period t and assuming no prior deviations, given
the beliefs in part (C), what is the most attractive point in
time prior to T for type B to deviate(assuming the other
player follows the strategy identified above), and what is the
greatest available payoff from such a deviation?
F. Based on your answers to (A) through (E), under what condi-
tion do these strategies and beliefs constitute a weak perfect
Bayesian equilibrium?
G. When the WPBE equilibrium you’ve identified in part (f)
exists, is it a sequential equilibrium? Why or why not?
113. Doohicklies vary in quality q, which falls within the interval [0, 1]. Con-
sumers attach no value to doohicklies unless q is at least qT , and they
attach no additional value to doohicklies of quality greater than qT , where
qT < 1. Because it is more expensive to produce a higher quality doohick-
lie (see below), firms only produce doohicklies of quality qT .
To make doohicklies, firms have to hire technicians. There are two types
of technicians: bright, and dim. Throughout, technicians are assumed to
know their own types. A bright technician can produce a doohicklie of
quality q at a cost (to the firm) of q 2 . For a dim technician, the cost
is q. Remembering that q lies on the unit interval, we see that a bright
technician can make a doohicklie at strictly lower cost provided q lies in
the open interval (0, 1). Let π denote the fraction of technicians who are
bright.
Any technician hired by a doohicklie manufacturer will make a total of
K doohicklies before losing interest, leaving the firm, and enrolling in
an economics PhD program. A doohicklie of quality qT or greater sells
80
for a price of unity, and a doohicklie of quality less than qT sells for a
price of zero. In the market for technicians, doohicklie makers compete by
offering a fixed wage, w, for producing K units of quality qT (which they
can monitor). They announce levels of w simultaneously, and technicians
pick the firm offering the highest wage (or decide randomly if the wages
are equal).
The payoffs equal profits for doohicklie makers, and equal wages net of
the monetary costs of signaling (see below) for technicians.
(a) Suppose doohicklie manufacturers can observe technicians types be-

fore making offers. Solve for the wages paid in equilibrium to bright
and dim technicians.
For the rest of this question, assume that doohicklie makers CANNOT ob-
serve technicians types before making offers. However, bright technicians
can try to signal their types to the firms by making sample doohicklies.
The technicians bear the costs of making the samples, which are subse-
quently thrown away (they are not given to the doohicklie makers or sold
to consumers).
Suppose that technicians are required (by some certifying body) to make
S sample doohicklies, where S is some fixed number. They are free, how-
ever, to choose the quality of those samples. They show the samples to
prospective employers, to whom the quality of the samples is apparent.
Thus, technicians signal through their selection of q. Focus on Weak Per-
fect Bayesian Equilibria.
(b) Explain why this model does not satisfy the Spence-Mirrlees single-
crossing property. Justify your answer completely.
The single-crossing property is sufficient for the existence of separating

equilibria, but not necessary. In the next part of this question, you will
explore whether separating equilibria exist in this model.
(c) In a separating equilibrium (if one exists), what would be the quality
of the samples produced by dim technicians? Justify your answer
completely.
(d) Answer the following questions concerning separating equilibria.
i. What quality levels (if any) can be used to signal a technicians
brightness in a separating equilibrium? Show how your answer
depends on K and S.
ii. Does the set of feasible signals (when non-empty) always include
the highest quality level, q = 1? Why, or why not?
iii. What is the most efficient signal?
(e) Answer the following questions concerning pooling equilibria.
81
i. Which quality levels can arise in pooling equilibria? Show how
your answer depends on K and S.
ii. What is the most efficient pooling signal?
Now suppose technicians can also invest in fancy clothes for their job inter-
views. The amount spent on clothes, C, is observable to the interviewer.
(f) Answer the following questions concerning separating equilibria.

i. What combinations of C and q (if any) can be used to signal a
technicians brightness in a separating equilibrium? Show how
your answer depends on K and S.
ii. What is the most efficient signal? Provide intuition for the result
you obtain for the optimal q.
114. Consider the following “exclusive contracting game”: There is an incum-

bent (I), two buyers (B1 and B2), and a potential entrant who, when
present, competes with I. Firm I approaches the two buyers sequentially,
offering buyer Bi the opportunity to sign an exclusive contract in exchange
for a payment xi . Each buyer Bi may either accept the offer (Ai) or reject
the offer (Ri). If the buyer accepts an exclusive contract, or if he rejects
and no entry occurs, firm I monopolizes the buyer. If the buyer rejects the
exclusive offer and entry occurs, the buyer has a gain in consumer surplus
equal to 12. Firm I’s profit from monopolizing a buyer is 9, while firm I
earns nothing if entry occurs.
Entry occurs for sure if neither buyer accepts an exclusive contract. The
probability that entry occurs in the event only one buyer accepts an exclu-
sive contract is known by firm I but not by the buyers. With probability
1 1
2 this probability is p and with probability 2 this probability is p, where
2
p < p < 3.
Thus, the timing is: The probability p or p is determined and observed by
firm I. Firm I then makes an offer x1 to B1. B1 observes the offer and
chooses whether to accept or reject. Then having observed B1’s response,
firm I makes an offer x2 to B2 who, having observed this offer and the offer
to and response by buyer B1, chooses whether to accept or reject. Finally,
depending on how many buyers accepted exclusive contracts, entry occurs
with the probabilities and consequences stated above. (Note that we are
not explicitly modeling the potential entrant’s decision – just assume it is
as stated above.)
In what follows, an “equilibrium” refers to a pure strategy weak perfect
Bayesian equilibrium.
In your answer, let µ2 (x1 , x2 , d) where d ∈ {A1, R1} denote B2’s belief
that firm I’s type is p after observing offer x1 to B1 and B1’s response d,
when he himself receives offer x2 . Likewise, let µ1 (x1 ) denote B1’s belief
that firm I’s type is p after receiving offer x1 .
82
(a) Draw the extensive form of this game. (You do not need to draw
all of it, but be sure to draw representative decision nodes for each
possible history of acceptance and rejection by B1.)
(b) What is a strategy for each player in this game?
(c) Consider the part of the game following rejection of an offer by B1.
i. Show that B2’s strategy following rejection by B1 of an offer x1
takes a cut-off form: that is, there is a lowest offer, say x
b2 (x1 , R1 ),
that B2 is willing to accept, and he is willing to accept any offer
above that.
ii. Can there be an equilibrium in which B2 does not accept an
offer from firm I following rejection by B1 of an offer x1 ? If so,
describe the strategies and beliefs. If not, explain fully why.
(d) Consider the part of the game following acceptance of an offer by B1.
i. Must buyer B2’s strategy following acceptance by B1 of an offer
x1 take a cut-off form? Explain fully why or why not.
ii. Can there be an equilibrium in which B2 accepts an offer that
firm I makes following acceptance by B1 of an offer x1 ? If so,
describe the strategies and beliefs. If not, explain fully why.
(e) Consider firm I’s offer to B1 and B1’s response. Is there an equilib-
rium in which B1 rejects firm I’s offer on the equilibrium path? If
so, describe the strategies and beliefs that firm I and B1 employ in
this equilibrium. If not, explain fully why.
(f) Is there an equilibrium in which different types of firm I make dif-
ferent offers to B1? If so, describe it; if not explain fully why.
(g) Are there equilibria in which different types of firm I make the same
offer to B1? If so, describe all of them; if not explain fully why.
115. There is an entrepreneur who would like to take her company public by
selling shares of it to outside investors. The company may be of low
value, θL , or high value, θH > θL . The entrepreneur knows the value of
the company, but outside investors do not. Investors’ priors are that each
valuation is equally likely.
The entrepreneur, after observing the company’s value, puts a fraction
q ∈ [0, 1] of her company up for sale. The market observes q (but not θ),
and the shares (per unit) sell for p (an amount to be determined shortly).
The payoff for the outside investors is then θq − pq, and the entrepreneur’s
payoff is pq + θ2 (1 − q). Notice that the entire company is worth θ in the
hands of outsiders (q = 1), but only θ2 in the hands of the entreprenuer
(q = 0). This difference is what motivates the entreprenuer to sell the
firm. While the source of the difference is not important for our purposes,
83
you can think of it as reflecting either synergies with other business ac-
tivities of an acquirer, or the entrepreneur’s aversion to a lack of portfolio
diversification.
Play proceeds as follows. First, the entreprenuer announces q. Second,
multiple investors bid for the portion of the firm that is offered for sale.
Third, the entreprenuer sells the offered portion of the firm to the highest
bidder at the highest bid. Throughout, restrict attention to Weak Perfect
Bayesian Equilibria (WPBE) in which all investors have the same beliefs
conditional upon any q.
(a) Suppose that upon observing q, the market forms beliefs µ(q) that
the firm’s value if θL . What will be the market price of the offered
share, q?
(b) Now we will focus on separating equilibria. Let qi = q(θi ) and
pi = p(q(θi )) for i = H, L.
i. In any separating equilibrium, what must be the values of pL ,
qL , and pH ? Explain your answers completely.
ii. What values of qH could arise in a separating equlibrium?
iii. For any given separating equilibrium, explain how to construct a
belief mapping µ(q) (and an associated price mapping p(q)) that
is consistent with the requirements of WPBE.
iv. Explain in words why q can serve as a signal of firm value.
v. Which separating equilibria are ruled out by the intuitive crite-
rion (equilibrium dominance)? Which survive? Explain your
answer completely.
(c) Next we consider pooling equilibria, in which entreprenuers offer the
share q ∗ and receive price p∗ irrespective of value.
i. What must be the value of p∗ ?
ii. Is there a pooling equilibrium with q ∗ = 1? If so, what beliefs
support this equilibrium?
iii. For what other values of q ∗ are there pooling equilibria? Exhibit
market beliefs that support those equilibria.
iv. Which pooling equilibria are ruled out by the intuitive criterion
(equilibrium dominance)? Which survive? Explain your answer
completely.
116. There are N types of workers, n = 1, . . . , N . The marginal product of a

worker of type n is
n−1
M Pn = 1 + .
N −1
(Note that M P1 = 1 and M PN = 2.) A worker of type n can obtain a
level of education e ≥ 0 at a cost of

n−1
cn (e) = 1 − 0.5 e.
N −1
84
(Note that c1 = e and cN = 0.5e.) Education has no effect on a workers
marginal product. A type n workers utility is given by
u(w, e, n) = wn − cn (e),
where wn is the wage received. The payoff to a firm hiring a type n worker
is M Pn − wn . Notice that, as N increases, the two extreme types dont
change. Rather, we are just filling in more and more types between the
extremes.
There are (at least) two potential employers, who can observe workers’
educational levels but not their types. Workers can signal their types
through education. In particular, a worker announces her level of edu-
cation, then employers simultaneously bid for her services by announcing
wage rates, and then the worker chooses an employer.
Below, you will be asked to analyze signaling equilibria. Let en denote the
level of education that a worker of type n receives in a given equilibrium.
Throughout, restrict attention to perfect Bayesian separating equilibria.
Use eH L
n (N ) and en (N ) to denote, respectively, the highest and lowest levels
of education type n workers obtain in any separating equilibrium with N
types of workers. Define An (N ) = eH L
n (N ) − en (N ). Note that An (N )
measures the degree of indeterminacy with respect to the equilibrium value
of en . In this question, youll be examining how AN (N ) behaves as N
becomes large.
(a) Determine eH L
1 (N ) and e1 (N ).
Now assume that N = 2.
(b) Determine eH L
2 (2) and e2 (2).
(c) Which is the most efficient separating equilibrium? Which (if any)
equilibria satisfy the equilibrium dominance criterion?
Now assume that N = 3.
(d) Determine eH L
n (3) and en (3) for n = 2 and n = 3.
(e) Which separating equilibria (if any) satisfy the equilibrium domi-
nance condition?
(f) Which separating equilibria (if any) satisfy the D1 criterion?
Now consider arbitrary N .
(g) For n = 2, , N , compute ∆H H H

n (N ) = en (N ) − en−1 (N ).
(h) For n = 2, , N , compute ∆L L L
n (N ) = en (N ) − en−1 (N ).
85
(i) Using your answers to parts (g) and (h), evaluate limN →∞ AN (N ).
Based on your answer, what happens to the indeterminacy of sepa-
rating equilibria as N gets large? Are equilibrium refinements useful
in the limit?
117. A worker is either high quality (θH ) or low quality (θH ), where θH >
θL > 1. Let λ ∈ (0, 1) be the proportion of high quality workers in the
population. A worker of type j who receives education e ≥ 1 along with
a wage w enjoys a payoff of
1
u(e, w, θi ) = w − e.
θi
In contrast to the standard signaling model, education is productive: a
worker with education e produces θi log(e). Both types of workers have the
same outside option, which yields a very low payoff that you can assume
is so unattractive as to make it irrelevant.
We will assume that there are two identical firms, and that they simul-
taneously offer wages to the worker after observing her chosen level of
education. The worker then accepts the highest offer.
(a) Suppose first that the firms can also observe the worker’s type, θi .
i. What are the equilibrium wage offers to a high quality worker
who has education e? What are the offers to a low quality
worker?
ii. What is the SPNE? How does the equilibrium education level
depend upon the workers type?
(b) Now suppose the firms observe e but not the workers type.
i. Suppose the firms believe the worker is high quality with proba-
bility β. What are the equilibrium wage offers?
ii. What level of education does the low quality worker receive in
any WPBE involving full separation of types? Why?
iii. What are the lowest and highest levels of education a high quality
worker will receive in a WPBE involving separation? (Derive
equations that determine these levels implicitly. Do not try to
solve for them analytically.)
iv. What are the lowest and highest levels of education a worker will
receive in a WPBE involving pooling? (Again, derive equations
that determines these levels implicitly. Do not try to solve for
them analytically.)
v. Which of the equilibria described in the previous two sub-questions
satisfy equilibrium dominance? Explain your answer.
86

ProblemBank203 2

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

ProblemBank203 2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ProblemBank203 2

Uploaded by

Copyright:

Available Formats

Professor B.

• matching pennies version A (as defined in class, where 1 moves first),

“Nature” selects randomly among these alternatives; each game is chosen

(a) Draw the extensive form of this game.

(a) Draw the extensive form of this game.

Dominance, Rationalizability, and Pure Strategy

(a) Draw the extensive form of this game.

(a) What are each player’s strictly dominated strategies?

7. In class, we constructed a pure strategy Nash equilibrium of the centipede

8. Consider a finite game, and suppose that it is dominance solvable. Prove

9. Consumers are uniformly distributed along a boardwalk of length 1 mile.

13. Consider a weekly work-consumption-savings plan of a household consist-

u(w1 , w2 , f1 , f2 ) = 2 min{b, 500} − (w1 + w2 ) + 0.1(f1 + f2 )

u2 (w1 , w2 , f1 , f2 ) = 2 min{b, 500 + ε} − (w1 + w2 ) + 0.1(f1 + f2 )

(a) Derive each firm’s best response function.

Mixed Strategy Nash Equilibria

(a) Draw the extensive form of this game.

21. Two risk neutral firms, i = 1, 2, produce a homogeneous product at a

22. Two risk netural firms, i = 1, 2, produce a homogeneous product at a unit

24. Consider a race between 2 Stanford Economics graduate students—1 and

for an MSNE of the following form: a student bids exactly b̄ with

(a) Identify the students’ strategy sets.

(d) Identify the students’ strategy sets.

Equilibrium Refinements for Games in Normal

(a) Calculate the set of pure strategy Nash equilibria.

(a) Derive the set of Nash equilibrium payoffs.

Static Games of Incomplete Information

(a) Suppose each student’s contribution is returned to him if the project

35. Imagine an economy consisting of two consumers, labeled 1 and 2. Con-

(a) Suppose that consumers choose the xi simultaneously and indepen-

(a) Describe a bidder’s pure strategy set. Give an example of a pure

(a) Is there an equilibrium where every individual discloses her grade?

U (ti , tj ) = f (h − ti ) + g(ti , tj ), (1)

Subgame Perfect Nash Equilibria

(a) Draw the extensive form of this game.

42. Consider a legislature constituted of three (3) individuals, A, B, and C.

(a) Suppose that players cast their votes simultaneously.

43. Consider a legislature consisting of four members – three legislators (i =

47. Consider a population of N people labeled i = 1, 2, ..., N . Each chooses

B(i, n) = c + dn − ei if i chooses IN , making the total of IN -choosers n + 1

i. Describe a typical pure strategy for each player, as well as their

• Player 1 starts driving on day 1, and from there on they alternate.

55. Consider a game played by a sequence of players, t = 0, ..., T . Each player

Bayesian Perfect and Sequential Equilibria

(0,10) (-10,2) (6,0) (-10,7) (3,5) (3,5)

(a) Characterize the set of Perfect Bayesian Equilibria.

(a) Draw the extensive form of this game.

(a) Draw the extensive form of this game.

(a) Consider the following game: The owner proposes a non-negative

(e, s) (e, n) (p, s) (p, n)

(a) First, assume that the legislator observes the proposal x.

67. Complete the exercise on p. 121 of the lecture notes.

(a) Imagine that the game unfolds as follows:

71. Complete the exercise on page 159 of the lecture notes.

(a) What is the (unique) SPNE of this alternative model?

74. Consider a multilateral bargaining game where Agents 1, 2, and 3 bargain

(a) Consider T = 2 (i.e., there are only 2 rounds of offers).

(a) Events unfold as follows.