EconS 503 - Microeconomic Theory II

Homework #4 - Answer Key

1. [Temporary punishments in GTS] Consider the setting in Example 8.15 of the

Advanced Microeconomic Theory textbook (MIT Press). Assume a collusive agreement
in which …rms use a punishment scheme where they revert to the Nash equilibrium of
the stage game during T consecutive periods, and after that punishment phase they
return to cooperation. Under which conditions on the discount factor can collusion be
sustained as a SPNE of the in…nitely repeated game when …rms rely on this temporary
punishment? Interpret.

After a history of cooperation, every …rm i’s discounted stream of payo¤s from
continuing its cooperation is

(a c)2 (a c)2 2 (a c)2 T (a c)2

+ + + ::: + + :::
8b 8b 8b 8b
If, instead, …rm i deviates from the GTS, its discounted stream of payo¤s becomes

9 (a c)2 (a c)2 T (a c)2 T +1 (a c)2

+ + ::: + + + :::
64b | 9b {z 9b } | 8b
{z }
Punishment for T periods Back to cooperation

Therefore, …rm i prefers to cooperate if

(a c)2 (a c)2 2 (a c)2 T (a c)2

+ + + ::: +
8b 8b 8b 8b
2 2 2
9 (a c) (a c) T (a c)
+ + ::: +
64b 9b 9b
where we simpli…ed both sides eliminating the stream of payo¤s that coincide
after the punishment phase is over (i.e., after period T ). The above expression
can be further simpli…ed to

2 T (a c)2 9 (a c)2 2 T (a c)2

(1 + + + ::: + ) + (1 + + + ::: + )
8b 64b 9b
or, after canceling (a c)2 and b on both sides,

2 T 1 9 2 T 1
(1 + + + ::: + ) + (1 + + + ::: + )
8 64 9
At this point, note that (1 + + 2 + : : : + T ) is a …nite geometric progression
XT
T +1
t
which can be expressed as = 11 , which helps us simplify the above
t=0
inequality as
T +1 T
11 9 11
+
8 1 64 9 1

1
and further rearranging we obtain
T +1
9 9 8 + 8 T +1 9
72 (1 ) 64
T +1
)8 9 8 81 (1 )
T +1
)8 17 + 9 0

In similar exercises, we solve for discount factor to …nd the minimal discount
factor that support cooperation in the in…nitely repeated game. However, in
this case, our expression is highly non-linear in , and does not allow for such
approach. We can nonetheless gain some intuition of our results by solving for T ,
which represents the lenght of the punishment phase, as follows
17 9
ln
T Tb 8
1
ln
Plotting cuto¤ Tb in the vertical axis and the discount factor 2 (0; 1) in the
horizontal axis, we obtain the following …gure. Intuitively, the punishment phase
must be long enough and …rms must care enough about their future pro…ts (as
indicated by T and pairs on the northwest of the …gure) for the GTS with
temporary punishment to be sustained as a SPNE of the in…nitely repeated game.

For illustration purposes, the …gure also includes a dotted line at a height of
T = 1, which does not cross with the curve representing cuto¤ Tb. This indicates
that, when punishment only last one period, cooparation cannot be sustained
in the in…nitely repeated game even if players assign full weight to their future
payo¤s (i.e., even if = 1). When the punishment phase lasts two periods, as
indicated by the dotted line at a height of T = 2, cooperation can be sustained
for discount factors satisfying 0:67, graphically represented by the range of
to the right-hand side of = 0:67 in the …gure. A similar argument applies when
the punishment phase lasts T = 3 periods, where we obtain that cooperation can
be supported as long as 0:58; con…rming our results in Exercise #27 of the
book.

2
 After a history in which at least one …rm deviated from cooperation, the GTS
prescribes that every …rm i implements the punishment during T rounds. This
is …rm i’s best response to …rm j implementing the punishment, so there are
no further conditions on the discount factor, , or the lenght of the punishment
phase, T , that we need to impose.

2. Exercises from Tadelis:

(a) Exercises from Chapter 10: 10.3, 10.6, 10.9, and 10.11.
(b) Exercises from Chapter 11: 11.2, 11.3, 11.4, 11.7.
See answer key at the end of this handout.

3
 10. Repeated Games 181

3. Not so Grim Trigger: Consider the infinitely repeated Prisoner’s Dilemma

with discount factor   1 described by the following matrix:

Player 2
 
 4 4 −1 5
Player 1
 5 −1 1 1

Instead of using “grim trigger” strategies to support a pair of actions (1  2 )

other than ( ) as a subgame perfect equilibrium, assume that the player
wish to choose a less draconian punishment called a “length  punishment”
strategy. Namely, if there is a deviation from (1  2 ) then the players will play
( ) for  periods, and then resume playing (1  2 ). Let  be the critical
discount factor so that if     then the adequately defined strategies will
implement the desired path of play with length  punishment as the threat.

(a) Let  = 1. What is the critical value  1 to support the pair of actions
( ) played in every period?
Answer: The proposed one period punishment means that instead of
getting 4 for the period after deviation, the players will get 1, and after-
wards will resort to getting 4 forever. Hence, the punishment is of size
3 and the discounted value is 3. The gain from deviating in one period
is getting 5 instead of 4 so this will be deterred if 1 ≤ 3 or  ≥ 13 .1 ¥

(b) Let  = 2. What is the critical value   to support the pair of actions
( ) played in every period?2
Answer: The proposed two period punishment means that instead of
getting 4 for the two periods after deviation, the players will get 1,

1 To see this using the whole stream of payoffs, sticking to ( ) yields 4
1−
while deviating with the threat
4 4 4
of a one period punishment will yield 5 + 1 +  2 1− and this is not profitable if 1− ≥ 5 + 1 +  2 1− , which
4 4
can be rewritten as 4 + 4 +  2 1− ≥ 5 + 1 +  2 1−  which in turn reduces to 3 ≥ 1.
2 Helpful hint: You should encounter an equation of the form  3 − ( + 1) + 1 = 0 for which it is easy to see

that  = 1 is a root. In this case, you know that the equation can be written in the form ( − 1)( 2 +  − 1) = 0
and solve for the other relevant root of the cubic equation.
182 10. Repeated Games

and afterwards will resort to getting 4 forever. Hence, the discounted

punishment is ( + 2 )3. The gain from deviating in one period is getting
√ √
5 instead of 4 so this will be deterred if 1 ≤ (+ 2 )3, or  ≥ 16 3 7− 12 ≈
026376.3 ¥

(c) Compare the two critical values in parts (a) and (b) above. How do they
differ and what is the intuition for this?
Answer: The punishment in part b. last for two periods which is more
severe than the one period punishment in part a. This means that it
can be supported with a lower discount factor because the intensity of
the punishment is increasing either in the length or when we have less
discounting. ¥

4. Trust off-the-equilibrium-path: Recall the trust game depicted in Figure

10.1. We argued that for  ≥ 12 the following pair of strategies is a subgame
perfect equilibrium. For player 1: “in period 1 I will trust player 2, and as as
long as there were no deviations from the pair ( ) in any period, then I
will continue to trust him. Once such a deviation occurs then I will not trust
him forever after.” For player 2: “in period 1 I will cooperate, and as as long
as there were no deviations from the pair ( ) in any period, then I will
continue to do so. Once such a deviation occurs then I will deviate forever
after.” Show that if instead player 2 uses the strategy “as long as player 1
trusts me I will cooperate” then the path ( ) played forever is a Nash
equilibrium for  ≥ 12 but is not a subgame perfect equilibrium for any value
of .
Answer: It is easy to see that this is a Nash equilibrium: the equilibrium
path is followed because neither player benefits from deviating as they both
believe that a deviation will call for the continuation of grim trigger. To
see that it is not subgame perfect consider the subgame that follows after

3 To 4
see this using the whole stream of payoffs, sticking to ( ) yields 1− while deviating with the threat of
2 3 4 4 4
a two period punishment will yield 5 + 1 +  1 +  1− and this is not profitable if 1− ≥ 5 + 1 +  2 1 +  3 1− .
2 3 4 2 3 4
This can either be solved as a cubic inequality or can be rewritten as 4 + 4 +  4 +  1− ≥ 5 + 1 +  1 +  1− 
which in turn reduces to ( +  2 )3 ≥ 1.
Homework #4 – EconS 503, Answer key
Homework #4 – EconS 503, Answer key
Homework #4 – EconS 503, Answer key
Homework #4 – EconS 503, Answer key
 Homework #4 – EconS 503, Answer key

P21

No Brand Brand
Name Name

P1 t P1t

N T N T

P2t P2t
0
0 C D 0 C
0 D

1 – 1 –
1 2 1+p 2
Player 2 creates a brand name since + p > 2

P2 t

No Brand Brand
Name Name

P1 t P1t

N T N T

P2t P2t
0
0 C D 0 C
0 D

1 – 1 –
1 2 1 2–p
Player 2 does not cheat (honors the contract)
 This is page 197
Printer: Opaque t

11
Strategic Bargaining

1. Disagreement: Construct a pair of strategies for the ultimatum game ( =

1 bargaining game) that constitute a Nash equilibrium, which together sup-
port the outcome that there is no agreement reached by the two players and
the payoffs are zero to each. Show that this disagreement outcome can be
supported by a Nash equilibrium regardless of the number of bargaining pe-
riods.

Answer: Consider the following strategies: player 1 offers nothing to player

2 ( = 0) and player 2 only accepts if he is offered all of the surplus ( = 1).
In this case both players are indifferent (player 1 is indifferent between any
offer and player 2 is indifferent between accepting and rejecting), and both
receive zero. It is easy to see that repeating these strategies for any length of
the game will still constitute a Nash equilibrium. ¥

2. Hold Up: Considering an ultimatum game ( = 1 bargaining game) where

before player 1 makes his offer to player 2, player 2 can invest in the size of the
pie. If player 2 chooses a low level of investment () then the size of the pie
is small, equal to  while if player 2 chooses a high level of investment ()
then the size of the pie is large, equal to  . The cost to player 2 of choosing
198 11. Strategic Bargaining

 is  , while the cost of choosing  is   Assume that     0,

    0 and  −    −  .

(a) What is the unique subgame perfect equilibrium of this game? Is it

Pareto Optimal?
Answer: Solving this game backward, we know that the ultimatum
game has a unique equilibrium in which player 1 will offer nothing to
player 2 and player 2 will accept the offer. Working backwards, if player 2
first chooses the low level of investment then his payoff will be − , while
he will be worse off if he chooses the high level of investment because
−  − . Hence, the unique subgame perfect equilibrium has player
2 first choose the low level of investment, then player 1 offering to keep
all the value  to himself, and finally player 2 accepting the offer and
getting − . ¥

(b) Can you find a Nash equilibrium of the game that results in an outcome
that is better for both players as compared to the unique subgame
perfect equilibrium?
Answer: Consider the following strategy for player 2: first choose the
high level of investment, and then accept any offer that gives himself at
least  − − for  small. Given this strategy, player 1’s best response
is to offer to keep  +  for himself and  −  −  for player 2. Player
2’s payoff is then  −  −  −   − for small enough , and player
1’s payoff is  +    so the players are both better off. ¥

3. Even/Odd Symmetry: In section ?? we analyzed the alternating bargain-

ing game for a finite number of periods when  was odd. Repeat the analysis
for  even.
Answer: Consider the case with an even number of rounds   ∞, implying
that player 2 has the last mover advantages. The following backward induc-
tion argument applies:
- In period  , player 1 accepts any offer, so player 2 offers  = 0 and payoffs
 11. Strategic Bargaining 199

are 1 = 0; 2 =   −1
- In period  −1 (odd period — player 1 offers), by backward induction player
2 should accept anything resulting in a payoff of 2 ≥   −1 . If player 2 is of-
fered  in period  − 1 then 2 =   −2 (1 − ); This implies that in period
 − 1 player 2 will accept any (1 − ) ≥  and by backward induction player
1 should offer  = 1 − , which yields player 1 a payoff of 1 = (1 − )  −2
and 2 =   −1 
- In period  −2 (even period), conditional on the analysis for  −1, player 1’s
best response is to accept any  that gives him   −3  ≥ (1 − )  −2  Player
2’s best response to this is to offer the smallest  that satisfies this inequal-
ity, and solving it with equality yields player 2’s best response:  =  −  2 
This offer followed by 1’s acceptance yields 1 =   −3  =  −2 −   −1 and
2 =   −3 (1 − ) =   −3 −   −2 +   −1 .
We can continue with this tedious exercise only to realize that a simple pat-
tern emerges. If we consider the solution for an even period  −  ( being
even because  is assumed to be even) then the backward induction argument
leads to the sequentially rational offer,

 − =  − 2 + 3 · · · −  

while for an odd period  −  ( being odd) then the backward induction
argument leads to the sequentially rational offer,

 − = 1 −  +  2 · · · −  .

We can use this Pattern to solve for the subgame perfect equilibrium offer in
the first period, 1  which by backward induction must be accepted by player
2, and it is equal to

1 = 1 −  +  2 −  3 +  4 · · · −  −1 =
= (1 + 2 + 4 + · · · +  −2 ) − ( +  3 +  5 + · · · +   −1 )
1 −   −  +1
= −
1 − 2 1 − 2
1 − 
= 
1+
200 11. Strategic Bargaining

and this in turn implies that

1 −   + 
1∗ = 1 =  and 2∗ = (1 − 1 ) = 
1+ 1+
¥

4. Constant Delay Cost: Consider a two player alternating bargaining game

where instead of the pie shrinking by a discount factor   1, the players
each pay a cost   0,  ∈ {1 2} to advance from one period to another. So,
if player  receives a share of the pie that gives him a value of  in period
 then his payoff is  =  − ( − 1) . If the game has  periods then a
sequence of rejections results in each player receiving  = −( − 1) .

(a) Assume that  = 2. Find the subgame perfect equilibrium of the game
and show in which way it depends on the values of 1 and 2 .
Answer: In the last period player 2 makes the offer in an ultimatum
game and will offer to keep the whole pie: 1 = 0 and 2 = 1 and
player 1 is will accept (he’s indifferent). Payoffs would be 1 = −1
and 2 = 1 − 2 . Going backwards to period 1, player 1 has to offer at
least 2 = 1 − 2 to player 2 for him to accept, so the unique subgame
perfect equilibrium has player 1 offering 1 − 2 to player 2, and player
2 accepts anticipating that he will offer and get 2 = 1 in the second
period following rejection. Payoffs are 1 = 2 and 2 = 1 − 2 . Payoffs
therefore do not depend on 1 . ¥

(b) Are there Nash equilibria in the two period game that are not subgame
perfect?
Answer: Yes. Just like in the game we studied with a discount factor ,
any split can be supported by a Nash equilibrium. Consider the following
strategy by player 2: reject anything but the whole pie in the first period
and offer to keep the whole pie in the second. Player 1’s best response in
the first period is to offer exactly the whole pie to player 2 because that
way he is guaranteed 0, while if he believes that player 2 will follow the
 11. Strategic Bargaining 201

proposed strategy and he offers anything else then he will get 1 = −1 .
¥

(c) Assume that  = 3. Find the subgame perfect equilibrium of the game
and show in which way it depends on the values of 1 and 2 .

Answer: In the last period player 1 makes the offer in an ultimatum

game and will offer to keep the whole pie: 1 = 1 and 2 = 0, and
player 2 is will accept (he’s indifferent). Payoffs would be 1 = 1 − 21
and 2 = −22 . Going backwards to period 2, player 2 has to offer at
least 1 = 1 − 21 to player 1 for him to accept, so the payoffs starting
from the second period are 1 = 1 − 31 and 2 = 21 − 2 (player 1 gets
a piece of the pie equal to 1 − 21 and because this is the second period
he incurs the cost 1 from the first period.) Finally, in period 1 player 1
must offer player 2 at least 21 − 2 so he will offer exactly that, player
2 will accept the offer, and the payoffs will be 1 = 1 − 21 + 2 and
2 = 21 − 2 . ¥

5. Asymmetric Patience 1: Consider a 3-period sequential (alternating) bar-

gaining model where two players have to split a pie worth 1 (starting with
player 1 making the offer). Now the players have different discount factors,
 1 and  2 .

(a) Compute the outcome of the unique subgame perfect equilibrium.

Answer: In the third period player 1 will get the whole pie and hence
the payoffs will be 1 =  21 and 2 = 0. Moving back to the second period,
player 2 will offer player 1  1 and player 1 will accept, so the payoffs are
1 = 21 and 2 =  2 (1− 1 ). Moving back to the first period, player 1 will
offer to keep  such that player 2 will receive 2 = (1 − ) =  2 (1 −  1 )
implying that player 1 gets 1 =  = 1 −  2 (1 −  1 ) = 1 −  2 +  1  2 . ¥

(b) Show that when  1 =  2 then player 1 has an advantage.

Answer: In this case 1 = 1 −  2 +  1  2 = 1 −  +  2 and 2 =  −  2 ,

 11. Strategic Bargaining 203

and
2 ≥ 1 − 1 1  (11.4)
Subtracting (11.2) from (11.1) yields

1 − 1 ≤  2 ( 2 −  2 )  (11.5)

and similarly, subtracting (11.4) from (11.3) yields

2 − 2 ≤  1 ( 1 −  1 )  (11.6)

But (11.5) and (11.6) together imply that

 1 −  1 ≤  2 (2 − 2 ) ≤  2  1 ( 1 − 1 ) 

and because  2  1  1 it follows that 1 =  1 (= 1 ) and 2 =  2 (= 2 ).

Revisiting the inequalities above, (11.1) and (11.2) imply that

1 = 1 −  2 2 ,

and (11.3) and (11.4) imply that

2 = 1 −  1 1 ,

and from these last two equalities we obtain that in the unique subgame
perfect equilibrium, in the first period player 1 receives
1 − 2
1∗ = 
1 − 12

and player 2 receives 1 − 1∗ = 1−

2 (1− 1 )
1 2
. Now let  1 = , and let  2 approach 
2 1
The denominator approaches 1− = (1−)(1+) and we get that 1∗ = 1+ ,
which is the solution we obtained is section 11.3 for a symmetric discount
factor. ¥

7. Legislative Bargaining (revisited): Consider a finite  period version of

the Baron and Ferejohn legislative bargaining game with an odd number 
of players and with a closed rule as described in section 11.4.1.
204 11. Strategic Bargaining

(a) Find the unique subgame perfect equilibrium for  = 1. Also, find a
Nash equilibrium that is not subgame perfect.
Answer: If  = 1 then following a failed vote (a majority rejects the
proposer’s proposal) all the players receive a payoff of 0. Hence, like in
the Rubinstein game, the proposer will ask for all the surplus and a ma-
jority of players will vote in favor. No other outcome can be supported
by a subgame perfect equilibrium. There are many Nash equilibria. For
example, some player  asks for at least ∗ ∈ [0 1] of the surplus while
all other players will settle for nothing. Then any player  6=  will offer
 the amount ∗ , and nothing to the other players, and all the players
will vote in favor of the proposal. ¥

(b) Find the unique subgame perfect equilibrium for  = 2 with a discount
factor 0   ≤ 1 Also, find a Nash equilibrium that is not subgame
perfect.
Answer: If the proposal is not accepted in period 1 then period 2 will
have the unique subgame perfect equilibrium described in part a. above.
This implies that in the first period, every player has an expected surplus
of  because they will be the proposer with probability 1 and will get
the whole surplus of 1. This means that the player who offers in the first
period must offer at least  to −1 2
other players to form a majority
and have the proposal accepted. Hence, the proposing player will keep
1− 2−1  to himself in the unique subgame perfect equilibrium. Just like
in part a. above, we can support an arbitrary division of the surplus
in a Nash equilibrium by having some players commit to incredible
strategies. ¥

(c) Compare what the first period’s proposer receives in the subgame per-
fect equilibrium you found in part (b) above to what a first period
proposer receives in the two-period two-person Rubinstein-Ståhl bar-
gaining game. What intuitively accounts for the difference?
Answer: In the two-period two-person Rubinstein-Ståhl bargaining
game the proposing player 1 gets 1 −  because player 2 can get the
 11. Strategic Bargaining 205

whole pie in the second period. Notice that the difference between the
payoff in the Baron-Ferejohn model and the Rubinstein-Ståhl model is,

 −1  ( + 1) 
1− − (1 − ) =  .
2  2 2
As we can see, the first proposer has a lot more surplus in the Baron-
Ferejohn model. This is because the responder is not one player who
plays an ultimatum game in the second period, but a group of player
from which a majority needs to be selected. This lets the proposer pit
the responders against each other and capture more surplus. ¥

(d) Compare the subgame perfect equilibrium you found in part (b) above to
the solution of the infinite horizon model in section ??. What intuitively
accounts for the similarity?
Answer: The share received by the first proposer is the same as what
we derived in equation (11.8). The intuition is that the same forces are
at work: the larger the discount factor the more the proposer needs to
give away, and the more people there are, the more he has to give away.
Still, he gets to keep at least 12 because of the competitive nature of the
situation in which the responders are put. ¥