Econ.

C103, 2003 Daniel McFadden

Final Exam Solutions

1. Consider the following simultaneous-move strategic-form game:

Find all the Nash equilibria (pure and mixed) of this game.

Let u be the probability that 1 plays U, and a,c be the probabilities that player 2 plays A and C, respectively. Then,
the expected payoffs are

B1 = (1-2a)(1-2u) and B2 = (3u-2)a + (1-2u)c.

Firm 1 maximizes its response to a by u = 1 when a > , u = 0 when a < , and any u when a = , Firm 2
maximizes its response to u by a = 1 when u > 2/3, by c = 1 when u < , and by a= c = 0 when < u < 2/3, by
c = 0 and any a when u = 2/3, and by a = 0 and any c when u = . Then, u = a = 1 is one Nash equilibrium, u =
a = 0 is another. The corresponding pure strategies are (U,A) and (D,C). Finally, a = , c = 0, and u = 2/3 is a
mixed strategy Nash equilibrium.

2. Firm 1 has to decide whether to enter a market in which Firm 2 is the incumbent. If Firm
1 enters, then the firms have to decide simultaneously whether to fight (F or Firm 2/f for Firm
1) or accommodate (A for Firm 2/a for Firm 1), with the payoffs given in the graph below.

(A) Suppose there can be no communication between the firms before the game. Find
all the subgame-perfect Nash equilibria, pure or mixed.

(B) Suppose Firm 2 can legally pre-commit itself to fight if Firm 1 enters, say by
contracting with each of its customers that it will strictly beat any price posted by Firm
1. How does this change the game, and the subgame-perfect Nash equilibria?
Do backward induction. If entry, the firms play the following game in strategic form:

Player 2

F A

f (-1,-3) (-1,-2)
Player 1
a (-2,-1) (1,3)

This game has a unique pure strategy Nash equilibrium at (a,A), so the extensive game can be reduced to a
decision by Firm 1 to enter with payoff (1,3) or stay out with payoff (0,2). Then, entry and accommodation is the
unique subgame-perfect Nash equilibrium. If the game is changed so that Firm 2 can pre-commit to F, then a is
a dominant strategy in the final strategic form game, and Firm 1 faces the effective payoffs (-2,-1) with entry and
(0,2) with no entry. Then, choosing no entry is a dominant strategy for Firm 1 in the reduced game.

3. Two firms FN and FOare said to be Cournot duopolists if they produce a common product
in quantities qN and qO at a marginal cost m > 0, and then face price p = A - B(qN + qO) in the
market, where A > m and B > 0 are constants known to both firms.

(A) Find a Nash equilibrium when the two firms choose quantities simultaneously.

(B) How would the Nash equilibrium change if firm FN first chooses qN, and then knowing
this, firm FO chooses qO? Which firm gains and which loses relative to (A)?

Firm FN has payoff BN = (A - B(qN + qO) - m)qN, maximized at qN = (A-m)/2B - qO/2. Symmetrically, firm 2 has the
optimal reaction function qO= (A-m)/2B - qN/2. These two equations are solved at qN = qO = (A-m)/3B, yielding price
(A+2m)/3 and a profit of (A-m)2/9B to each firm. This is the Nash equilibrium in the simultaneous move game.
Alternately, suppose firm FO chooses knowing qN. Its optimal choice is the same reaction function, qO= (A-m)/2B -
qN/2. Knowing that firm FO will make this choice, firm FN will maximize BN = (A - B(qN + (A-m)/2B - qN/2) - m)qN at
qN = (A-m)/2B. Then, qO = (A-m)/4B. The profit of FN is (A-m)2/8B and the profit of firm FO is (A-m)2/16B. Then,
firm FN has a higher payoff and firm FO has a lower payoff in the sequential move case. .
4. Consider a first-price sealed bid auction of a single object with two bidders j = 1,2 and no
reservation price. Bidder 1s v1 = 2, and bidder 2's valuation is v2 = 5. Both v1 and v2 are
known to both bidders. Bids must be in whole dollar amounts. In the event of a tie, the object
is awarded by a flip of a fair coin. Is there a Nash equilibrium? What is it? Is it unique? Is it
efficient?

The complete payoff matrix is given in the following strategic game:

Player 2

0 1 2 3 4 5

0 (1,5/2) (0,4) (0,3) (0,2) (0,1) (0,0)

2 (0,0) (0,0) (0,3/2) (0,2) (0,1) (0,0)

Bid 1 by player 1 and bid 3 by player 2 are one Nash equilibrium. However, bids 0 and 2 for player 1 are weakly
dominated by bid 1, and If player 1 bids 1, then the dominant response for player 2 is to bid 2. This is another
Nash equilibrium. Since player 1 has a weakly dominant strategy and player 2 has a unique best response to it,
there is no need to look for mixed strategies. Both Nash equilibria are efficient, assigning the object to the higher
value bidder.

5. Suppose an exchange economy has J consumers and 2 goods. Consumer j has a utility function
of the form uj = zj + tj@vj, where zj is consumption of the first commodity, which is divisible, with each
consumer having an initial endowment of one unit, and tj is consumption of the second commodity,
which can be consumed only in integer units, with consumer 1 having an endowment of one unit and
all others having an endowment of zero. The vj are values of the indivisible good which are drawn
independently for each consumer from the uniform distribution on [0,1]. Each consumer knows her own
vj, but knows only that the vs of all others are drawn from the uniform distribution..Suppose an auction
mechanism is used to assign the indivisible good, with consumer 1 participating as a possible buyer
as well as the seller. Suppose this mechanism leads to a resource allocation that is in the core of the
economy.

(A) From the properties of the core, prove that the allocation produced by this auction must
have resulted in the indivisible good going to the consumer with the highest vj, with a payment
by the buyer no less than the second highest vj, and no greater than the highest vj.

(B) Discuss the relationship of this result to the revenue equivalence theorem of auction theory.

An allocation in the core cannot be blocked by any coalition. Consumer 1 on her own is guaranteed utility 1 + v1
without trading, and every other consumer is guaranteed utility 1 without trading. Individual rationality says they
will block any allocation that fails to deliver these utility levels. Then, if consumer 1 trades with another consumer
j, the amount of the divisible goods going to the seller can be no less than v1 and no more than the highest vj.
Further, if consumers 1 and, say 2, trade the indivisible good for z units of the divisible good, with z less than the
highest value of a player not receiving the good, then this player and the seller have a blocking coalition that trades
for this players value of the good. This will necessarily happen unless the highest value player gets the indivisible
good and the seller gets at least the second-highest value. Therefore, any auction that delivers an allocation that
is in the core (i.e., efficient) implies that the final buyer has highest value and pays a price at least equal to the
second highest value and no more than the highest value.

The revenue equivalence theorem says that all auction mechanisms that are efficient and individually rational yield
the same expected revenue to the seller as an efficient second-price sealed bid auction. This result is consistent
with, but sharper than, the implications of an allocation being in the core, establishing that the seller can get no
more than the second value when bidder behavior is taken into account.