Problem Set 9 - Graduate Microeconomics II
Prof. Leandro Arozamena
TA: Matías Cersosimo
2017
Exercise 1 (M.W.G. 9.B.10) B
Reconsider the game in Example 9.B.3...
Example 9.B.3: Consider the following predation game. Firm E (for entrant) is considering entering a
market that currently has a single incumbent rm (I) If it does so (playing "in"), rms I and E now play a
simultaneous-move game after entry, each choosing either "accommodate", giving up some of its sales but
causing no change in the market price, or " ght", engaging in a costly war of predation that dramatically
lowers the market price.
...but now change the post-entry game so that when both players choose "accommodate", instead of receiving
the payos (uE , uI ) = (3, 1), the players now must play the following simultaneous-move game:
1
Firm I
` r
U (3, 1) (0, 0)
Firm E
D (0, 0) (x, 3)
What are the SPNEs of this game when x ≥ 0? When x < 0?
Exercise 2 (M.W.G. 9.B.7) B
Consider the nite horizon bilateral bargaining game in Appendix A (Example [Link].1); but instead of
assuming that players discount future payos, assume that it costs c < v to make an oer. (Only the player
making an oer incurs this cost, and players who have made oers incur this cost even if no agreement is
ultimately reached.) What is the (unique) SPNE of this alternative model? What happens as T approaches
∞?
Example [Link].1 : Finite Horizon Bilateral Bargaining, Stahl (1972). Two players, 1 and 2, bargain to
determine the split of v dollars. The rules are as follows: the game begins in period 1; in period 1, player
1 makes an oer of a split (a real number between 0 and v ) to player 2, which player 2 may then accept
or reject. If she accepts, the proposed split is immediately implemented and the game ends. If she rejects,
nothing happens until period 2. In period 2, the players' roles are reversed, with player 2 making an oer to
player 1 and player 1 then being able to accept or reject it. Each player has a discount factor of δ ∈ (0, 1),
so that a dollar received in period t is worth δ t−1 in period 1 dollars. However, after some nite number
of periods T , if an agreement has not yet been reached, the bargaining is terminated and the players each
receive nothing.
Exercise 3 (M.W.G. 9.B.11) B
Two rms, A and B, are in a market that is declining in size. The game starts in period 0, and the rms
can compete in periods 0, 1, 2, 3, . . . (i.e., inde nitely) if they so choose. Duopoly pro ts in period t for rm
A are equal to 105 − 10t, and they are 10.5 − t for rm B. Monopoly pro ts (those if a rm is the only one
left in the market) are 510 − 25t for rm A and 51 − 2t for rm B.
Suppose that at the start of each period, each rm must decide either to "stay in" or "exit" if it is still active
(they do so simultaneously if both are still active). Once a rm exits, it is out of the market forever and
earns zero in each period thereafter. Firms maximize their (undiscounted) sum of pro ts.
What is this game's subgame perfect Nash equilibrium outcome (and what are the rms' strategies in the
equilibrium)?
Exercise 4 (F.T. 3.7)* (Eckel & Holt (1989))
Consider a voting game in which three players, 1, 2 and 3, are deciding among three alternatives. A, B and
C. Alternative B is the "status quo" and alternatives A and C are "challengers". At the rst stage, players
choose which of the two challengers should be considered by casting votes for either A or C, with the majority
2
choice being the winner and abstentions not allowed. At the second stage, players vote between the status
quo B and whichever alternative was victorious in the rst round, with majority rule again determining
the winner. Players vote simultaneously in each round. The players care only about the alternative that is
nally selected, and are indierent as to the sequence of votes that leads to a given selection. The payo
functions are u1 (A) = 2, u1 (B) = 0, u1 (C) = 1; u2 (A) = 1, u2 (B) = 2, u2 (C) = 0; u3 (A) = 0, u3 (B) = 1,
u3 (C) = 2.
(a) What would happen if at each stage the players voted for the alternative they would most prefer as
the nal outcome?
(b) Find the subgame-perfect equilibrium outcome that satis es the additional condition that no strategy
con be eliminated by iterated weak dominance. Indicate what happens if dominated strategies are
allowed.
(c) Discuss whether dierent "agendas" for arriving at a nal decision by voting between two alternatives
at a time would lead to a dierent equilibrium outcome.
Exercise 5 (F.T. 4.3)*
In the model of Rubinstein 1982, two players must agree on how to share a pie of size 1. In periods 0, 2, 4,
etc., player 1 proposes a sharing rule (x, 1 − x) that player 2 can accept or reject. If player 2 accepts any
oer, the game ends. If player 2 rejects player 1's oer in period 2k, then in period 2k + 1 player 2 can
propose a sharing rule (x, 1 − x) that player 1 can accept or reject. If player 1 accepts one of player 2's
oers, the game ends; if he rejects, then he can make an oer in the subsequent period, and so on. This
is an in nite-horizon game of perfect information. Note that the "stages" in our de nition of a multi-stage
game are not the same as "periods" -period 1 has two stages, corresponding to player 1's oer and player
2's acceptance or refusal.
We will specify that the payos if (x, 1 − x) is accepted at date t are (δ1t x, δ2t (1 − x)), where x and 1 − x are
the players' share of the pie, and δ1 and δ2 are the two players' discount factors.
Consider a variant of Rubinstein's in nite-horizon bargaining game where partitions are restricted to be
integer multiples of 0.01, that is, x can be 0, 0.01, 0.02, . . . , 0.99, or 1. Characterize the set of subgame-
perfect equilibria for δ = 21 and for δ very close to 1.
Exercise 6 (F.T. 4.4)* (Dutta and Gevers)
Consider the I - player version of Rubinstein's game. PI At dates 1, I + 1, 2I + 1, . . ., player 1 oers a division
(x1 , . . . , xI ) of the pie with xi ≥ 0 for all i, and i=1 xi ≤ 1. At dates 2, I + 2, 2I + 2, . . ., player 2 oers
a division, and so on. When player i oers a division, the other players simultaneously accept or veto the
division. If all accept, the pie is divided; if at least one vetoes, player i + 1 (player 1 if i = I ) oers a division
in the following period. Assuming that the players have common discount factor δ , show that, for all i player
i oering division
δ I−1
1 δ
, , . . . ,
1 + . . . + δ I−1 1 + . . . + δ I−1 1 + . . . + δ I−1
for players i, i + 1, . . . , i − 1 at each date (kI + i) and the other players' accepting is a subgame-perfect
equilibrium outcome.