0% found this document useful (0 votes)
37 views6 pages

PS3 Sol

This document discusses bargaining games and correlated equilibria. It presents a mathematical economics course which includes two problems. The first problem examines alternating offer bargaining with constant costs of delay. The second problem defines a game in strategic form and asks to find Nash equilibria, characterize correlated equilibria, and explain why correlated equilibrium does not expand outcomes for this game.

Uploaded by

Antonio Aguiar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
37 views6 pages

PS3 Sol

This document discusses bargaining games and correlated equilibria. It presents a mathematical economics course which includes two problems. The first problem examines alternating offer bargaining with constant costs of delay. The second problem defines a game in strategic form and asks to find Nash equilibria, characterize correlated equilibria, and explain why correlated equilibrium does not expand outcomes for this game.

Uploaded by

Antonio Aguiar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Course: Mathematical Economics

Lecturer: Peter Wagner, Ph.D.


Autumn 2021

Problem Set 3

Problem 1 (Alternating offer bargaining with constant cost of delay)


In his “Economic and Philosophic Manuscripts” of 1844, Karl Marx writes that > “Wages are
determined by the antagonistic struggle between capitalist and worker. Victory goes necessarily
to the capitalist. The capitalist can live longer without the worker than can the worker without
the capitalist.”
Perhaps he has in mind the variant of the bargaining game of alternating offers in which
each player i loses ci during each period of delay (rather than discounting her payoff), as in
Exercises 1 & 2.
1. Show that if c1 < c2 , then this game has a subgame perfect equilibrium in which player
1 always proposes (1, 0). (In this case, in fact, the game has no other subgame perfect
equilibrium.)
2. Show also that if c1 = c2 = c, then for every value of z1 with c ≤ z1 ≤ 1 the game has a
subgame perfect equilibrium in which player 1 always proposes (z1 , 1 − z1 ). (In both
cases, a strategy pair is a subgame perfect equilibrium if and only if it satisfies the
one-deviation property.)

SOLUTION:
1. First suppose that c1 < c2 .
• A reasonable guess is that in a subgame perfect equilibrium in which player 1
always proposes (1, 0), player 2 always accepts all proposals.
• Another reasonable guess is that the game has such an equilibrium in which player
2 always makes the same offer, say (z1 , z2 ), and player 1 always uses the criterion
“accept a proposal (x1 , x2 ) if and only if x1 ≥ z1 ” to respond to a proposal.
Is there a value of (z1 , z2 ) such that this strategy pair is a subgame perfect equilibrium?
A strategy pair is a subgame perfect equilibrium if and only if it satisfies the one-
deviation property, so consider the conditions imposed on (z1 , z2 ) by the one-deviation
property.
Examine each type of subgame in turn:
1. Subgame starting with proposal by player 1 : If player 1 follows her strategy
she obtains all the pie, so she cannot profitably deviate.

1
2. Subgame starting with response by player 2: Denote by (x1 , x2 ) the proposal
to which player 2 is responding.
• Her strategy calls for her to accept the proposal, yielding her the payoff x2 .
• If she rejects the proposal, she proposes (z1 , z2 ), which player 1 accepts,
yielding her the payoff z2 − c2 .
• Thus for equilibrium we need x2 ≥ z2 − c2 for all x2 , which means that
0 ≥ z2 − c2 , or z2 ≤ c2 .
3. Subgame starting with proposal by player 2
• If player 2 follows her strategy she obtains the payoff z2 . If she offers player 1
more than z1 , player 1 accepts, and player 2 is worse off.

• If she offers player 1 less than z1 , player 1 rejects her offer and proposes (1, 0),
which she accepts, yielding her the payoff −c2 Thus for equilibrium we need
z2 ≥ −c2 .
4. Subgame starting with response by player 1: Denote by (y1 , y2 ) the proposal
to which player 1 is responding.
• If y1 ≥ z1 her strategy calls for her to accept the proposal, yielding her
the payoff y1 . If instead she rejects the proposal, she proposes (1, 0), which
player 2 accepts, yielding her the payoff 1 − c1 . Thus for equilibrium we need
y1 ≥ 1 − c1 whenever y1 ≥ z1 , and hence z1 ≥ 1 − c1 .
• If y1 < z1 her strategy calls for her to reject the proposal, in which case she
proposes (1, 0), which player 2 accepts, yielding player 1 the payoff 1 − c1 . If
instead she accepts the proposal she obtains y1 . Thus for equilibrium we need
1 − c1 ≥ y1 whenever y1 < z1 , and hence 1 − c1 ≥ z1 .
From the analysis of the last subgame, we have z1 = 1 − c1 , so that z2 = c1 . Given
c1 < c2 , the one-deviation property is satisfied in every subgame, so that the strategy
pair is a subgame perfect equilibrium.
In summary, the following pair of strategies is a subgame perfect equilibrium:
• player 1 always proposes (1, 0) and accepts a proposal (y1 , y2 ) if and only if
y 1 ≥ 1 − c1
• player 2 always proposes (1 − c1 , c1 ) and accepts all proposals.
2. Now suppose that c1 = c2 = c. Let c ≤ z1 ≤ 1 and consider the pair of strategies in
which
• player 1 always proposes (z1 , 1 − z1 ) and accepts a proposal (y1 , y2 ) if and only if
y1 ≥ z1 − c
• player 2 always proposes (z1 − c, 1 − z1 + c) and accepts a proposal (x1 , x2 ) if and
only if x2 ≥ 1 − z1 .
I argue that this strategy pair is a subgame perfect equilibrium, by showing that it
satisfies the one-deviation property. I consider each of the four distinct subgames in

2
turn.
1. Subgame starting with proposal by player 1: If player 1 follows her strategy
she obtains the payoff z1 . If she increases her offer to player 2 , her offer is accepted
and she is worse off. If she reduces her offer to player 2, her offer is rejected,
and player 2 proposes (z1 − c, 1 − z1 + c), which she accepts. Thus no deviation
increases her payoff.
2. Subgame starting with response by player 2: Denote by (x1 , x2 ) the proposal
to which player 2 is responding. Her strategy calls for her to accept this proposal if
and only if x2 ≥ 1 − z1 . If she rejects a proposal, she proposes (z1 − c1 , 1 − z1 + c),
which player 1 accepts, yielding her the payoff 1 − z1 . Thus no deviation increases
her payoff.
3. Subgame starting with proposal by player 2: If player 2 follows her strategy
she obtains the payoff 1 − z1 + c. If she offers player 1 more than z1 − c, player 1
accepts, and player 2 is worse off. If she offers player 1 less than z1 − c, player
1 rejects her offer and proposes (z1 , 1 − z1 ), which she accepts, yielding her the
payoff 1 − z1 − c. Thus no deviation increases her payoff.
4. Subgame starting with response by player 1: Denote by (y1 , y2 ) the proposal
to which player 1 is responding. Her strategy calls for her to accept this proposal
if and only if y1 ≥ z1 − c. If she rejects a proposal, she proposes (z1 , 1 − z1 ), which
player 2 accepts, yielding her the payoff z1 − c. Thus no deviation increases her
payoff.

Problem 2 (Correlated Equilibrium)

x y
x 0, 1 1, 0
y a, 0 0, b
1. Suppose a, b ≥ 0. Find the unique Nash equilibria as a function of the parameter a and
b, calculate the expected payoffs for each player, and determine the probability with
which the players play each action pair in this equilibrium .
2. Suppose a,b > 1. Explain how the players can use contracts and public randomization
to coordinate on an outcome in which their expected payoffs exceed that of the Nash
equilibrium in Part 1.
3. Write down the system of linear inequalities that characterise the set of correlated
equilbria.
4. Show that there is only one correlated equilibrium that generates the same outcome as
the mixed strategy Nash equilibrium.

3
5. Explain in your own words why in this game, correlated equilibrium does not expand
the set of equilibrium outcomes.

SOLUTION:
1. Denote by p the probability of Player 1 choosing x, and let q denote the probability of
Player 2 choosing x.
• Player 1 is indifferent between x and y if
1
q · 0 + (1 − q) · 1 = q · a + (1 − q) · 0 ⇒ q = .
1+a
• Player 2 is indifferent between x and y if
b
p · 1 + (1 − p) · 0 = p · 0 + (1 − p) · b ⇒ p = .
1+b
Thus, there is a mixed strategy Nash equilibrium
1 1
!  !
b a
, , , .
1+b 1+b 1+a 1+a

With these strategies, the probability distribution over action pairs is given as follows:

x y
b ab
x 1+a+b+ab 1+a+b+ab
1 a
y 1+a+b+ab 1+a+b+ab

2. Use the following contract:


 
• If only player 1 signs the contract, then she plays the mixed strategy: b
, 1
1+b 1+b
.
 
• If only player 2 signs, then she plays the mixed strategy: 1
, a
1+a 1+a
.
• If both players sign the contract, then player 1 plays y with certainty, and player
2 plays x and y with equal probability.
If player 1 does not sign, then player 2 plays the equilibrium strategy without contracts,
and player 1’s payoff is a/(1 + a). Similarly, if player 2 does not sign, player 1 plays the
equilibrium strategy without contracts, and player 1’s payoff is b/(1 + b). When they
both sign the contract, then player 1 receives the payoff a/2 and player 2 receives b/2.
Since a, b > 1, we have
a a a b b b
< = , and < = ,
1+a 1+1 2 1+b 1+1 2
which proves the claim!

4
3. The system of inequalities representing the incentive-compatibility constraints are the
following:

(0 − a)µ(x, x) + (1 − 0)µ(x, y) ≥ 0,
(a − 0)µ(y, x) + (0 − 1)µ(y, y) ≥ 0,
(1 − 0)µ(x, x) + (0 − b)µ(y, x) ≥ 0,
(0 − 1)µ(x, y) + (b − 0)µ(y, y) ≥ 0.

4. Rearranging the system of inequalities above, we obtain:


µ(x, y) ≥ aµ(x, x),
aµ(y, x) ≥ µ(y, y),
µ(x, x) ≥ bµ(y, x),
bµ(y, y) ≥ µ(x, y).
We can combine these inqualities into a sequence as follows:
µ(x, y) ≥ aµ(x, x) ≥ abµ(y, x) ≥ bµ(y, y) ≥ µ(x, y).
Note that the first and the last part of the sequence are the same, and hence all of
these must be equal:
µ(x, y) = aµ(x, x) = abµ(y, x) = bµ(y, y) = µ(x, y).
Note further that since the 4 probabilities must add up to 1, we have: µ(x, y) =
1 − µ(x, x) − µ(y, x) − µ(y, y). Using this identity and substituting the various equalities
from above, we get:
1 1 1
µ(x, y) = 1 − µ(x, y) − µ(x, y) − µ(x, y).
a ab b
or equivalently:
1
!
b a ab
µ(x, y) 1 + + + = 1 ⇒ µ(x, y) = .
ab ab ab 1 + a + b + ab
We can now use this result and the above chain of identities to find all relevant
probabilities:
ab
µ(x, y) = ,
1 + a + b + ab
b
µ(x, x) = ,
1 + a + b + ab
a
µ(y, y) = ,
1 + a + b + ab
1
µ(y, x) = .
1 + a + b + ab
We have shown that the incentive-compatibility constraints permit a single solution
characterised by the above equalities. The distribution over action pairs is the same as
for the mixed strategy Nash equilibrium.

5
5. In order for correlated equilibrium to have any effect in terms of expanding the set
of outcomes, there has to be a “coordination problem” in the sense that there is a
combination of strategies that is good for both players, but in itself not an equilibrium.
The above game, however, is strictly competitive, in the sense that each outcome that
is good for one player is bad for the other, and vice versa.

You might also like