Advanced Microeconomic Theory EC104

Problem Set 1

1. Each of n farmers can costlessly produce as much wheat as she chooses. Suppose that
the kth farmer produces W k , so that the total amount of what produced is
W  W 1  W 2 . . . W n . The price p at which wheat sells is then determined by the
demand equation p  e −W .
1a. Show that the strategy of producing one unit of wheat strongly dominates all of
a profit-maximizing farmer’s other strategies. Check that the use of this strategy yields a
profit of e −n for a farmer.
The payoff for a single (i th ) farmer is:
 i W i , W −i   e −W 1 ...W n  W i
First-order condition gives:
∂ i  0  e −W  W −1e −W   0
i
∂W i
and we easily obtain:
W ∗i  1
Notice that W ∗i  1 is the maximizing value irrespective of the actual value of W (or more
specifically, W −i ). Hence, W i  1 must be the dominant strategy. If all n farmers set
W i  1, then W  1 . . . 1  n  1  n. Hence,  i  e −n  1  e −n .

1b. Explain why the best of all agreements that treat each farmer equally requires
each to produce only 1n units of wheat. Check that a farmer ’s profit would then be e1n .
Why would such an agreement need to be binding (that is, signed as a legally binding
contract) for it to be honored by a profit-maximizing farmers?

To find the “best of all agreements”, we need to maximize the social surplus with
respect to W 1 , . . . , W n , i.e., we need to solve:
max W 1 . . . W j . . . W n e −W 1 ...W j ...W n 
W 1 ,...,W n

The First-order condition (FOC) with respect to the representative farmer’s quantity choice
w j is:
−1W 1 . . . W j . . . W n e −W 1 ...W j ...W n   e −W 1 ...W j ...W n   0
By imposing the symmetry condition W j  W i  W S for any i ≠ j, we have
−1NW s  1  0
Hence,
W s  1/n
It follows that:
−1
 i  1 ne  n1e
Notice, however, that if each farmer other than i produces 1n ( 1!!!), farmer i ′ s best
response will be to produce...W i  1!!! That means each farmer will have an incentive to
produce more. Hence we would need a binding contract to enforce this agreement

1c. Confirm that xe −x is largest when x  1. Deduce that all the farmers would
make a larger profit if they all honored the agreement rather than each producing one unit
and so flooding the market.

We need to maximise
max
x
xe −x 
The F.O.C. is given by:
− xe −x  e −x  0
which implies that
x1
To maximize the social surplus, we can also maximise We −W with respect to the total
quantity W. The socially optimum total output has to be 1 and since all farmers have equal
share, each produces 1n . Again,
−1
 i  1 ne  ne 1  e −n

1d. You would have realized what the exercise went through was version of the
“tragedy of the commons”. Why is such an n-player game a generalization of the Prisoners’
Dilemma?

This wheat production game may considered a generalization of the Prisoner’s

Dilemma (PD) game because just as in a PD game the equilibrium strategy is a dominant
strategy and yet the equilibrium outcome is clearly Pareto dominated (i.e., is inefficient).

2. Consider the following case of a differentiated good Cournot model. Firm i produces
type i widgets at a constant unit cost of c i , i  1, 2. If q 1 and q 2 are the quantities of the
two varieties produced, the respective prices for the two goods are determined by the
demand equations p 1  M − 2q 1 − q 2 and p 2  M − q 1 − 2q 2 . Adapt Cournot’s
duopoly model to this new situation, and find:
2a. The firms’ reaction functions.

The profit function of each firm i  1, 2 is equal to:

 i q i , q j   p i q i  q j  − c i q i
 M − 2q i  q j − c i q i
 Mq i − 2q 2i − q i q j − c i q i
To find the reaction function of firm i simply set
 ∂ i  0
∂q i
 M − 4q i − q j − c i  0
M − qj − ci
 q i q j  
4

2b. The quantities produced in equilibrium and prices at which the goods are sold
and the equilibrium profits.

Solving the reaction functions simultaneously, we easily obtain for i  1, 2:

c j − 4c i
q ∗i  M 
5 15
and thus
q ∗1  M  c 2 − 4c 1
5 15
q ∗2  M  c 1 − 4c 2
5 15
The prices are then given by:
p ∗1  M − 2q ∗1 − q ∗2
 2M  7c 1  2c 2
5 15 15
p ∗2  M − q ∗1 − 2q ∗2
 2M  2c 1  7c 2
5 15 15
Finally, the equilibrium profits are:
 ∗1  p ∗1 − c 1 q ∗1
 2 3M − 4c 1  c 2 3M − 4c 1  c 2 
225
 ∗2  p ∗2 − c 2 q ∗2
 2 3M  c 1 − 4c 2 3M  c 1 − 4c 2 
225

3. Two firms set prices in a market whose demand curve is given by Q  Dp, where
Dp is a downward-sloping function and p is the lower of the two prices.The lower
priced firm meets all of the demand; if the two firms post the same price, then they
each get half the market. Assume costs of production are zero and that prices can only
be quoted in dollar discrete units (0, 1, 2...).
3a. Show that if the rival firm charges a price above the monopoly price p m , then
the best response is to charge the monopoly price.

Call the price that the rival firm charges p (where p ≥ p m ). Any price below p captures
the market. By definition of a monopoly price, p m dollars yields a higher profit than any of
the alternatives. Therefore the best response to p is p m .

3b. Show further that if the rival firm charges a price p 1 at or below the
monopoly price, then the best response is to charge a price below p.

Dp
The profits at p are 2 p whereas the profits at p − 1 are Dp − 1p − 1. Since
p
demand is downward sloping, we have: Dp − 1  Dp while p − 1 ≥ 2 since p ≥ 2.
Thus the best response is to charge a price below p.

3c. Conclude from the preceding arguments that the unique Nash Equilibrium price
must be for each firm to price at 1 dollar.

So long as the rival charges a price p ≥ 2, it will be a best response to undercut, and the
rival’s best response will be to further undercut. However, when p  1, then the best
response must be a price at $1 (since at $0, profits are zero). Thus, a price of $1 for each
firm is a Nash Equilibrium.

3d. What would be the Nash Equilibrium if there were 3 firms in this market? More
generally, if there were n firms? Explain.

There is no change because the benefit to undercutting the rivals remain the same since
the entire market is captured by the lowest price. But the profits from not undercutting are
Dp
now N p. Hence, each firm would undercut if
Dp
p  Dp − 1p − 1
N
p
As before, Dp − 1  Dp and p − 1 ≥ N as long as p ≥ N−1 N
. Thus, the Nash
Equilibrium will still be for every firm to set a price equals to $1.

4. (Osborne, Exercise 60.1) (Variant of Cournot’s game, with market-share maximizing

firms) Consider a standard Cournot game with two firms 1 and 2 competiting in
quantities q 1 and q 2 , where the marginal cost of production is constant and the same
for both firms and denoted by c and the (inverse) market demand is given by:
 − Q if Q ≤ 
PQ 
0 if Q  
where   0 and c ≥ 0 are constant and Q  q 1  q 2 is the market demand.
4a. Find the Nash equilibrium (equilibria?) of a variant of the standard Cournot’s
duopoly game (linear inverse demand, constant unit cost) that differs only in that one of the
two firms chooses its output to maximize its market share subject to not making a loss,
rather than to maximize its profit.

Let firm 1 be the market-share maximizing firm. If q 2   − c, there is no output of

firm 1 for which its profit is nonnegative. Thus its best response to such an output of firm 2
is q 1  0. If q 2 ≤  − c, then firm 1 wants to choose its output q 1 large enough so that the
price is c (and hence its profit is zero). Thus firm 1’s best response to such a value of q 2 is
q 1   − c − q 2 . We conclude that firm 1’s best response function is:
 − c − q 2 if q 2 ≤  − c
b 1 q 2  
0 if q 2   − c
Firm 2’s best response function is the same as in the standard Cournot model (linear
inverse demand, constant unit cost)
 − c − q 2 /2 if q 2 ≤  − c
b 2 q 1  
0 if q 2   − c
These best response functions are shown in the following figure.
q2

BR 1 (q2)
a-c

(a - c)/2

NE
BR2 (q 1)
q1
0 a-c

The game has a unique Nash equilibrium, q ∗1 , q ∗2    − c, 0, in which firm 2 does not
operate. (The price is c, and firm 1’s profit is zero.)

4b. What happens if each firm maximizes its market share?

If both firms maximize their market shares, their reaction functions are:
  − c − q 2 if q 2 ≤  − c
b 1 q 2  
0 if q 2   − c

 − c − q 1 if q 1 ≤  − c
b 2 q 1  
0 if q 1   − c

This means that the downward-sloping parts of their best response functions coincide in
the analogue of the figure above.
q2

BR 1(q2)
a-c

BR2(q1)
q1
0 a-c

Thus every pair (q 1 , q 2 ) with q 1  q 2   − c is a Nash equilibrium.

5. (Hotelling voting game with 3 candidates) An election has 3 candidates and takes place
under the plurality rule. Voters are uniformly spread along an ideological spectrum
from left to right whose extreme points are 0 (extreme left) and 1 (extreme right). Each
voter votes for the candidate whose declared position is closest to the voter’s own
position. The candidates have no ideological attachment and take up any position along
the line, each seeking only to maximize her share of votes.
Suppose you are one of the three candidates. The leftmost of the other two is at point x,
and the rightmost is at point (1 − y), where x  y  1, so the rightmost candidate is a
distance y from 1. The following figure illustrates this situation:
 y

0 x 1-y 1

5a. Show that your best response is to take up the following positions under the
given conditions:
i just slightly to the left of x if x  y and 3x  y  1.
ii just slightly to the right of 1 − y if y  x and x  3y  1.
iii exactly halfway between the other candidates if 3x  y  1 and x  3y  1.

Observe firs that:

i If you locate just to the left of the leftmost candidate, your vote total equals x.
ii If you locate just to the right of the rightmost candidate, your vote total equals y.
iii If you locate between the other two candidates, your vote total equals 1 − y − x/2.
The conditions given in the question determine which of x, y, and (1 – y – x)/2 is
biggest and thus which of these locations gives you the most votes.
i Note that x  y is given. So to prefer locating just left of x, it has to be that it gives
the most votes, that is:
1−y−x
x
2
This is equivalent to:
3x  y  1
Thus, locating just to the left of the leftmost candidate gives the most votes.
ii Note that y  x is given. We need to verify now that
1−y−x
y
2
This is equivalent to:
x  3y  1
Thus, locating just right of 1 − y gives the most votes.
iii We have to check that:
1−y−x
x
2
which is equivalent to:
3x  y  1
We have also to check that
1−y−x
y
2
which is equivalent to:
 x  3y  1
Thus, locating between the other two candidates gives the most votes.

5b. In a graph with x and y along the axes (y is on the vertical axis and x on the
horizontal axis), show the areas (the combination of x and y values) where each of the
reponse rules i, ii, and iii of question (5a), is best for you.

The graph is as follows:

1
y

x=y
(ii)

x + 3y = 1
(i)
1/3

3x + y = 1
(iii)
0 1/3 x 1

5c. From your analysis, what can you conclude about the Nash equilibrium of the
game where the three candidates each choose positions?

Given the goal of candidates (as stated in the question), a Nash equilibrium in locations
can exist only if each candidate is in the location that maximizes his or her share of the
vote, given the locations of the other candidates.
A candidate can always increase his vote share by moving. For example, any candidate
who doesn’t have another candidate located at a just slightly more moderate position can
always gain votes by moving toward the center of the line. Thus, there is no Nash
equilibrium in locations in the three-candidate game.

6. (Osborne Exercise 80.1) (Timing product release) Two firms are developing competing
products for a market of fixed size. The longer a firm spends on development, the
better its product. But the first firm to release its product has an advantage: the
customers it obtains will not subsequently switch to its rival. Once a person starts using
 a product, the cost of switching to an alternative, even one significantly better, is too
high to make a switch worthwhile.
A firm that releases its product first, at time t, captures the share ht of the market,
where h is a function that increases from time 0 to time T, with h0  0 and hT  1. The
remaining market share is left for the other firm. If the firms release their products at the
same time, each obtains half of the market. Each firm wishes to obtain the highest possible
market share.

6a. Model this situation as a strategic game and plot the utility function u i of firm i
as a function of t i for different values of t j (i.e. consider three cases: ht j   1/2,
ht j   1/2, and ht j   1/2).

A strategic game that models this situation is:

Players: The two firms.
Actions: The set of actions of each player is the set of possible release times, which we
can take to be the set of numbers t for which
Preferences: Each firm’s preferences are represented by its market share; the payoffs
are:
ht i  if ti  tj
u i t 1 , t 2   1/2 if t i  t j
1 − ht j  if t i  t j
 Payoffs for different values of t j
vi − t j
ui ui ui

1/2 1/2 1/2

0
tj ti tj ti tj ti

h(t j ) < 1/ 2 h(t j ) = 1/ 2 h(t j ) > 1/ 2

10/02/2010 1
 6b. Calculate the best-reponse functions of each player. Denote the time t for which
ht 1   ht 2   ht ∗   1/2 by t ∗ . Observe that when finding firm i’s best response to firm
j’s release time t j , three cases must be considered:
(1) if t j is such that ht j   1/2;
(2) if t j is such that ht j   1/2;
(3) if t j is such that ht j   1/2.

Based on the payoff functions, we can easily determine the BR functions. Indeed, from
the payoff function, we see that
(1) if t j is such that ht j   1/2, then the set of firm i’s best responses is the set of
release times after t j .
(2) if t j is such that ht j   1/2, then the set of firm i’s best responses is the set of
release times greater than or equal to t j .
(3) if t j is such that ht j   1/2, then firm i wants to release its product just before t j .
Since there is no latest time before t j , firm i has no best response in this case. (It has good
responses, but none is optimal.)
In summary, player i’s best response function is given by:
t i : t i  t j  if ht j   1/2
BR i t j   t i : t i ≥ t j  if ht j   1/2
No BR if ht j   1/2
Denote the time t for which ht 1   ht 2   ht ∗   1/2 by t ∗ , then we have:
t 1 : t 1  t 2  if t2  t∗
BR 1 t 2   t 1 : t 1 ≥ t 2  if t 2  t ∗
No BR if t 2  t ∗
Figure: Player 1’s BR function
„ Player 1’s BR
t2 t1 = t2

t1 < t2
t1 > t2

t*

t1

t*
10/02/2010 2
Similarly, for player 2, we have:
t 2 : t 2  t 1  if t1  t∗
BR 12 t 1   t 2 : t 2 ≥ t 1  if t 1  t ∗
No BR if t 1  t ∗
Figure: Player 2’s BR function
„ Player 2’s BR
t2

t1 < t2
t1 = t2

t*
t1 > t2

t1

0
10/02/2010
t* 3
 6c. Show that there is unique a Nash equilibrium in which both firms release their
product at time t ∗ where ht 1   ht 2   ht ∗   1/2.
 Unique Nash equilibrium (t*,t*)
BR1
BR2 t1 = t2
t1 < t2
„ NE
t2

t1> t2

t*

t1

t*
10/02/2010 4