final_re-take_march_solutions
final_re-take_march_solutions
final_re-take_march_solutions
March 2011
a) first-price
b) second-price
d) all-pay
Solution discussed many times before... player 3 will win in every case. Under
first-price, given risk neutrality and independent values of n players, uniformly
distributed on [0, 1] the formula is bi = n−1
n vi so the winning bid b3 and hence
seller’s revenue will be .6. Under second-price, truth-telling (bi = vi ) is weakly
dominant so the revenue will be .7 with no reserve price and .8 when reserve
price is there. Under all-pay the formula is bi = n−1 n
n vi so after some math we
can establish that the revenue (sum of payments by 3 players) will be 0.722 if I
recall correctly...
Problem 2 (13 points) There are 6 tokens on the table. Two players alternate
removing some of the tokens. In each move any player can either remove exactly
one or exactly three tokens. Whoever removes the last token is the winner
(tokens are worthless per se – it does not matter who removed how many of
them).
2. Which player is going to win? Find the subgame perfect Nash equilibrium
using backward induction. Precisely describe full strategies.
1
Solution Any action involves removal of an odd number of tokens. Thus it
takes an even number of moves to remove 6 tokens. Player 2 will win no matter
what. A concise description of SPNE strategies is thus (whatever, whatever).
p2 = 2 p2 = 3
p1 = 2 (8*,8*,8*) (10*,.6.10*)
p2 = 3 (6,10*,10*) (9,9,12)
For p3 = 2 we have
p2 = 2 p2 = 3
p1 = 2 (10*,10*,6) (12*,9,9)
p2 = 3 (9,12*,9) (12*,12*,12*)
There are two pure strategy NE: everyone choosing p = 2 and everyone
choosing p = 3. Note: expressions such as “NE for players 1 and 2” that I
have seen a few times don’t make any sense and only prove that you don’t
understand the concept of NE... NE means that EVERY player is best
responding. For player 3 you need to compare across matrices because we
can’t have a 3d matrix. For example, for (8,8,8) in the first matrix to be
a NE we require that player 3 cannot profitably switch to the upper left
cell of the second matrix.
2. Turn this game into a coalitional game (determine what each coalition can
jointly surely get no matter what others do).
Solution As explained in the class, the standard way of turning a non-
coop game into a coop one, is to answer the question what is the minimum
payoff that any coalition can guarantee by coordinated action, when others
are trying to hurt it. Any individual firm can only guarantee the payoff
2
of 8 by choosing pi = 2. Firms 1 and 2 can guarantee a joint payoff of 18
by choosing p1 = 3, p2 = 3. Same for other 2-firm coalitions. The grand
coalition can get 36. Thus we have.
xi ≥ 8, (2)
xi + xj ≥ 18 for i 6= j and thus xi ≤ 18
x1 + x2 + x3 = 36
4. Now, for each pure strategy Nash equilibrium consider payoff profile in-
duced by it. Which inequalities defining the core are satisfied, which are
violated? Comment on your results.
Solution Inequalities associated with one-players coalitions are necessar-
ily satisfied because in equilibrium no player can be getting less than what
he could have as guaranteed. The other inequalities are satisfied in the
“friendly” Pareto-efficient pi = 3 equilibrium only.