Java Python Problem set 4
Advanced Microeconomics,
Ec 992, Spring 2024
Extensive form games and subgame perfection II
[and Nash bargaining (optional)]
1. Exercise 9.B.11 of the MWG textbook.
2. Exercise 9.B.13 of the MWG textbook. (Note: you should try to find / charac- terize at least one subgame perfect Nash equilibrium.)
3. Consider the following variation of the infinite horizon alternating offers bar- gaining game discussed in the lecture. Suppose δ = 1. In any period, if an offer is rejected, the game terminates with probability α ∈ (0, 1). For Player i = 1, 2, let ui (xi ) denote the Bernoulli utility from payoff xi and let bi ∈ [ui (0), ui (1)] denote the Bernoulli utility in the event of a termination. Assume that ui is strictly increasing and concave in xi. Let (ˆ(x)1 , ˆ(x)2 ) and (ˆ(y)1 , yˆ2 ) satisfy
u1 (ˆ(y)1 ) = αb1 + (1 − α)u1 (ˆ(x)1 )
u2 (ˆ(x)2 ) = αb2 + (1 − α)u2 (ˆ(y)2 )
where bi ≤ ui (ˆ(x)i ) and bi ≤ ui (ˆ(y)i ), i = 1, 2. Does the strategy profile (σ1(*),σ2(*)) below constitute a subgame perfect equilibrium? Explain your answer.
● σ 1(*): Player 1 always offers (ˆ(x)1 , ˆ(x)2 ) and accepts an offer (y1 , y2 ) if and only if y1 ≥ ˆ(y)1 .
● σ2(*): Player 2 always offers (ˆ(y)1 , yˆ2 ) and accepts an offer (y1 , y2 ) if and only if y2 ≥ ˆ(x)2 .
4. Exercise 12.D.1 of the MWG textbook.
(Erratum: “...the market ceases to exist with probability 1 − γ.”)
5. Exercise 12.D.3 of the MWG textbook.
(Erratum: “...with q ∈ [4b/a−c, 3b/a−c) are sustainable... ...δ(q) is a decreasing, differentiable function of q...)
6. Exercise 12.D.5 of the MWG textbook.
7. Table 1 depicts the payoff matrix with payoff parameter y > 0 of a two-player game where Player 1 chooses the row A or B and Player 2 chooses the column A or B . Suppose this is a stage game which is repeated infinitely. Let δ ∈ (0, 1) denote the common discount factor for the players. For what values of δ can the action profile (A, A) be sustained in a subgame perfect Nash equilibrium of the infinitely repeated game with a strategy profile that calls for (A, A) to be played in period 1 and thereafter if neither player has yet deviated and for a stage-game Nash equilibrium to be played otherwise? Explain.
A | B | |
A | 1 1 | −1, 1 + y |
B | 1 + y, −1 | 0, 0 |
Table 1: Stage game. Payoff parameter y > 0.
8. Consider the following infinite-horizon game between one (long-lived) player labeled F who is present in every period and a sequence of (short-lived) players, each of whom lives for one period only. Periods and the short-lived players are both indexed by t = 1, 2, ....
Within each period, the short-lived player first chooses between actions e and ne. Player F observes the choice of action by the short-lived player and then chooses an action w ∈ R+ . If the short-lived player chooses e and player F chooses w, then the payoff to the short-lived player is w − c, where c > 0 and the& Ec 992 Spring 2024 Advanced Microeconomics Problem set 4 nbsp;payoff to F is y − w, where y > c > 0. If the short-lived player choose ne and player F chooses w then the payoff to the short-lived player is w and the payoff to F is −w.
In the infinite-horizon problem, at the beginning of each period t, the history of the game through to period t − 1 is observed by the short-lived player who plays in period t and by F. Let the discount factor of F be δ ∈ (0, 1).
(a) Consider the game within each period between F and the short-lived player of that period. What is the subgame-perfect equilibrium of this (one- period) game? Explain.
(b) Consider the infinite-horizon problem. For what values of δ does there exist a sub-game perfect equilibrium of the infinite-horizon game where the short-lived player in each period chooses e? Explain.
9. Consider the following two-player repeated game with periods indexed by t = 1, 2, 3, . . . and a common discount factor δ ∈ (0, 1). The stage game action set Ai is finite for each player i = 1, 2 and for any action profile a = (a1 , a2 ), ui (a) denotes player i’s payoff in the stage game. For player i = 1, 2, let ui denote the (stage game) minmax payoff, which is defined as follows with j i.
ui = a i j [a(m)i i ui (a)] (1)
Denote by ˆ(a)j the action of player j which solves the minimisation problem in (1), i.e. the action that holds player i down to their minmax payoff ui. Letˆ(a) = (ˆ(a)1 , ˆ(a)2 ) denote the punishment action profile. Note that ui (ˆ(a)) need not be equal to ˆ(u)i. Additionally, let ¯(u)i denote the highest payoff that player i can obtain in the stage game.
Let a* = (a1(*), a2(*)) be such that ui (a* ) > ui for all i = 1, 2.
Consider the strategy profile s = (s1 , s2 ) in the repeated game which is com- prised as follows. Let K ≥ 1.
Player i chooses ai(*) in t = 1 and every period thereafter if the outcome of t − 1 is a* ; otherwise player i chooses ˆ(a)i. If the outcome in each of t−1, t − 2, . . . , t−K is ˆ(a) then player i chooses ai(*) in t. If the outcome in t − 1 is (ˆ(a)i ,aj ), where aj ˆ(a)j , then player i chooses ˆ(a)i for K periods.
(a) Suppose t = 1 or the outcome of t − 1 is a* . Provide an expression (in- equality) involving δ, ui (a* ), ¯(u)i , ui (ˆ(a)),K such that it is optimal for player i to follow the strategy si.
(b) Suppose the outcome of each of t − ℓ is ˆ(a), where 1 ≤ ℓ ≤ K. Provide an expression (inequality) involving δ, ui (a* ), ui , ui (ˆ(a)),K such that it is optimal for player i to follow si.
(c) Given the expressions obtained above, does there exist a value δ < 1 such that for any δ > δ the strategy profile s is a subgame perfect Nash equilibrium of the repeated game for some value of K? Explain your answer.
10. (Optional exercise on Nash Bargaining) Formulate the following as a Nash Bar- gaining problem and characterize the symmetric Nash Bargaining solution. A union V representing L > 0 workers bargains with a firm F. Each worker can obtain w0 ≥ 0 if not employed by the firm. If the firm hires ℓ ∈ [0, L] workers, it produces output f(ℓ), where f′ > 0, f′′ < 0 with f(0) = 0 and f(ℓ) ≥ ℓw0 for some ℓ ∈ [0, L]. An agreement is a wage employment pair (w,ℓ) such that w0 ≤ w ≤ f(ℓ)/ℓ with utility to F given by f(ℓ) − ℓw and to V given by ℓw + (L − ℓ)w0 . Disagreement means that F gets 0 and U gets Lw0 . Note that Pareto efficiency requires that the choice of ℓ maximizes the sum of the firm and union utilities