Problem Set 1

1. For the following bargaining games, nd the Nash bargaining solution by (a) solving the
maximization problem and by (b) using only the four axioms.
(a) R is the closed region inside the triangle with vertices (0, 0), (9, 0), (0, 6) and the
disagreement point is u0 = (0, 0)
(b) R is the closed region inside the triangle with vertices (0, 0), (9, 0), (0, 6) and the
disagreement point is u0 = (3, 2)
(c) R is the closed region inside the triangle with vertices (0, 0), (8, 0), (0, 8) and the
disagreement point is u0 = (2, 1)
(d) R is the closed region inside the quadrilateral with vertices (0, 0), (0, 6), (6, 3), (8, 0)
and the disagreement point is (0, 0)
(e) R is the closed region inside the quadrilateral with vertices (0, 0), (0, 6), (6, 3), (8, 0)
and the disagreement point is (2, 2)

2. Proof of the Existence and Uniqueness of the Nash Bargaining Solution

Nashs Theorem
There is only one solution f : B <2 that satises Pareto optimality, Symmetry, Preservation under Strictly Increasing Ane Transformation and Independence of Irrelevant
Alternatives. Moreover, for any (R, u0 ), f (R, u0 ) solves
max{(u1 u01 )(u2 u02 )|(u1 , u2 ) R, u1 u01 , u2 u02 }
and the solution (u1 , u2 ) to the above maximization problem is unique. This f is called
the Nash bargaining solution
Let B be the set of bargaining problems (R, u0 ) such that
R is a convex and compact subset of <2
u0 = (u01 , u02 ) R
There is a (u1 , u2 ) R such that u1 > u01 , u2 > u02
(Proof)
(a) Let f be a function such that for each (R, u0 ), f (R, u0 ) is the solution to the maximization problem above. To show that f above is well-dened as a function (i.e.
f (R, u0 ) is single-valued for each (R, u0 ))
Let H(u1 , u2 ) = (u1 u01 )(u2 u02 ) and let R0 = {u R|u1 u01 , u2 u02 }
Because R is compact, R0 is also compact
Because H is a continuous function on R0 , H attains a maximum on R0
(Problem) Prove the following statements.
i. If s = (s1 , s2 ) is a maximizer for H on R0 , then s1 > u01 and s2 > u02
ii. R0 is convex
iii. There is only one such s = (s1 , s2 ); therefore f is a well-dened function
(Hint)Suppose there is another maximizer t = (t1 , t2 ) in R0 , that is dierent
from s; dene r = (r1 , r2 ) = ((s 1 +t1 )/2, (s 2 +t2 )/2)
Show that H(r1 , r2 ) > H(s1 , s2 ) and (r1 , r2 ) R0 , which contradicts the
maximality of s
1

(b) (Problem) Show that f satises Pareto optimality, Symmetry, Preservation under
Strictly Increasing Ane Transformation and Independence of Irrelevant Alternatives.
(c) To show that f is the unique solution that satises the four axioms:
Let g : B <2 be another solution that satises Pareto optimality, Symmetry,
Preservation under Strictly Increasing Ane Transformation, and Independence of
Irrelevant Alternatives.
It is sucient to show that for each(R, u0 ), f (R, u0 ) = g(R, u0 )
Take any (R, u0 ) and let u = f (R, u0 )
i. Consider the following ane transformation and let R0 be the set of (u01 , u02 )
dened below ((u1 , u2 ) R)
u0
u01 = 2(uu1u0 ) 2(u 1u0 )
1

u02

u2
2(u2 u02 )

u02
2(u2 u02 )

ii. (Problem) Show that under the transformation dened above,

(u1 , u2 ) is transformed to (1/2, 1/2)
(u01 , u02 ) is transformed to (0, 0)
iii. Therefore, f (R0 , (0, 0)) = (1/2, 1/2) and by axiom 3 (Preservation under Strictly
Increasing Ane Transformation), it is sucient to show g(R0 , (0, 0)) = (1/2, 1/2)
iv. For each u0 = (u01 , u02 ) R0 it can be shown that u01 + u02 1 has to hold.
Suppose u01 + u02 > 1 for some (u01 , u02 )
For a small , 0 1, consider (1 )(1/2, 1/2) + (u01 , u02 )
(Problem) Show that this point lies in R0
(Problem) Show that for suciently small the product of the two coordinates of this point exceed 1/4
This contradicts f (R0 , u0 ) = (1/2, 1/2).
v. Let T be any triangle that is symmetric with respect to the 450 line and contains
R0 and that (1/2, 1/2) is Pareto optimal within T . Because R is bounded, such
T must exist. By Pareto optimality and symmetry, g(T, (0, 0)) = (1/2, 1/2).
R0 T and (0, 0), (1/2, 1/2) R0 , which implies (by independence of irrelevant
alternatives g(R0 , (0, 0)) = (1/2, 1/2).
3. A Non-cooperative Approach to the Nash Bargaining Solution
(Refer to Problem 2 on page 68 of Introduction to Game theory.)
(a) Bargaining Game
Feasible set R = {(u1 , u2 ) <2 |u1 + u2 100}
Disagreement point u0 = (u01 , u02 ) = (0, 0)
The Pareto optimality and the symmetry give the Nash bargaining solution u =
(u1 , u2 ) = (50, 50)
(b) Nashs idea
Players 1 and 2 simulataneously and independently announce their demandsu1 , u2
0, respectively. If u1 + u2 100, they obtain their demands. If u1 + u2 > 100, then
they get their payos at the disagreement point.
In this game, there are many Nash equilibria including the Nash bargaining solution.
ProblemFind all Nash equilibria of this game.
2

(c) Rubinsteins idea

i. bargaining process
1st period
One player (player 1 in the following) oers two players payos (u11 , u12 ). Then
player 2 decides whether to accept 1s oer or not. If player 2 accepts the oer,
the game ends, and players 1 and 2 obtain u11 and u12 , respectively. If player 2
rejects, they go into the 2nd period.
2ne period
Player 2 oers (u21 , u22 ). If player 1 accepts the oer, the game ends, and players
1 and 2 obtain u21 and u22 , respectively. If player 1 rejects, they go into the 3rd
period.
3rd period
Player 1 oers (u31 , u32 ). Then player 2 decides whether to accept 1s oer or not.
Repeat this procedure until one of the players accepts another players oer.
Both players payos are 0 when the game never ends. Introduce the discount
factor , 0 < < 1, and consider the discounted payos. Thus the discounted
payos when player 1 accepts the oer in the second period are (u21 , u22 )and
the payos when the game ends in the third period are ( 2 u31 , 2 u32 )
ii. Claim
The subgame perfect equilibrium outcome of the game converges to the Nash
bargaining solution (50, 50) when 1 converges to 1.
iii. (Proof)
Suppose there exists a subgame perfect equilibrium and let the payos of the
two players be u1 , u2 The whole game and the subgame starting from the third
period have the same structure except the payos. The payos of the subgame
are given by the payos of the whole game multiplied by 2 . Thus we focus on
the subgame perfect equilibrium in which the equilibrium stategies of the whole
game and the subgame starting from the third period are the same.
If the two players obtain u1 and u2 in the subgame starting from the third stage,
their dicounted payofs in the whole game are 2 u1 , 2 u2 . Hereafter all the payos
are discounted ones.
The subgame starting from the period 2If player 1 accepts player 2s oer (u21 , u22 ),
then his payo is u21 ; and if he rejects, they go into the third stage and his payo
is 2 u1 . Thus if u21 2 u1 , i.e., u21 u1 , then player 1 accepts the oer: the
minimum oer that player 1 accepts is u1 . Therefore the maximum amount
that player 2 can obtain is 100 u1 .
If player 1 rejects, then they go into the third period; thus player 2 obtains
2 u2 . Since u2 is at most 100 u1 , tha maximum payo that player 2 can
get when player 1 rejects his oer is 2 (100 u1 ). Since 0 < < 1, we have
(100 u1 ) > 2 (100 u1 ). Thus in the subgame perfect equilibrium in the
subgame starting from the second period, player 2 oers (u1 , 100 u1 ), and
player 1 accepts 2s oer.
The whole game
ProblemRead carefully the analysis above in the subgame starting from the
second period; and show the following.
3

A. The minimum amount u12 that player 2 accepts 1s oer is (100 u1 ).

B. Show u1 5.
4. An applicationNegotiation between management and labor
A manager and a labor union negotiates about wages w and the number of employees k.
The labor union represents K workers. Each worker earns w0 when not employed by this
rm. This rm produces f (k) unist of the good when it employes k workers. One unit of
the good is sold at a price p. f (k) satises the following three properties.
f (k) is strictly concaveThat is, for any s, t and any , 0 < < 1,
f (s + (1 )t) > f (s) + (1 )f (t)
f (0) = 0
There exists k, 1 k K, such that pf (k) > wk
For each pair (w, k), payos to the manager and the labor union are given by
u1 (w, k) = pf (k) wk, u2 (w, k) = wk + w0 (K k)
respectively. the disagreement point is(0, w0 K)
The sum of the payos to the manager and the labor union is
u1 (w, k) + u2 (w, k) = pf (k) wk + wk + w0 (K k) = pf (k) + w0 (K k)
and thus depends only on k.

(a) Find the Preto optimal k. That is, k maximizing the total prot of manager and
labor union.
(b) Find the Nash bargaining solution.