IO+solutions

i
September 18, 2000

ii
Solutions Manual
Industrial Organization:
A Strategic Approach
Jeffrey Church
University of Calgary
and
Roger Ware
Queen’s University
c 2000 Jeffrey Church and Roger Ware
September 18, 2000

ii
Contents
1 Introduction 1
2 The Welfare Economics of Market Power 3
3 Theory of the Firm 17
4 Market Power and Dominant Firms 21
5 Non-Linear Pricing and Price Discrimination 25
6 Market Power and Product Quality 27
7 Game Theory I: Static Games of Complete Information 29
8 Classic Models of Oligopoly 31
9 Game Theory II: Dynamic Games of Complete Information 39
10 Dynamic Models of Oligopoly 43
11 Product Differentiation 49
12 Identifying and Measuring Market Power 55
13 An Introduction to Strategic Behavior 59
14 Entry Deterrence 65
15 Strategic Behavior: Principles 71
16 Strategic Behavior: Applications 73
17 Advertising and Oligopoly 75
18 Research and Development 79
19 The Theory of the Market 85
20 Exclusionary Strategies I: Raising Rivals’ Costs 89
iii
iv CONTENTS
21 Exclusionary Strategies II: Predatory Pricing 91
22 Vertical Integration and Vertical Restraints 99
23 Horizontal Mergers 103
24 Rationale for Regulation 107
25 Optimal Pricing for Natural Monopoly 111
26 Issues in Regulation 121

Chapter 1
Introduction
Welcome to the solutions manual for Industrial Organization: A Strategic Approach (IOSA) by
Jeffrey Church and Roger Ware. This manual contains the solutions to the end of chapter problems
found in IOSA. The solutions should have sufficient detail that students can follow the derivations
and replicate the solutions. The solutions are available by chapter in three different configurations:
all problems, odd problems, and even problems at
http://www.econ.ucalgary.ca/iosa/IM/
In addition there are three versions of this solutions manual, corresponding to all, odd, or even
problems.
The solutions were prepared with the assistance of David Krause at the University of Calgary
and Andrea Wilson and Alexendra Lai both at Queen’s University.
Warning
Instructors may not post these pdf files on their websites unless they are able to restrict access to only
their students. Please be considerate of other instructors who may use these problems for exams
and assignments and will not appreciate their students having access to the solutions. Resist the
temptation to post the answers and impose—potentially significant—costs on other IO instructors.
Comments
If you have questions, comments, and/or find errors please contact either
Jeffrey Church ([email protected]) for chapters 2, 3, 4, 7, 8, 9, 10, 11, 12, 13, 14, 19, 24,
25, and 26 or
Roger Ware ([email protected]) for chapters 5, 6, 15, 16, 17, 18, 20, 21, 22, and 23.
1
2 Church and Ware Solutions Manual
Chapter 2
The Welfare Economics of Market

Power
1. Profit maximization by a monopolist requires producing where
P m (1 1) = ( Q m ):
" MC
1 (1 ) 1
On the inelastic portion of demand, " < which implies that =" > . So on the inelastic
portion of demand MR is negative and since MC is not, the monopolist can increase profits
by reducing output. Intuitively, on the inelastic portion of the demand curve a 1% increase
in price leads to a reduction in demand less than 1%, so total revenue increases as price
increases and quantity decreases. Profits will increase because total revenue rises and total
cost decreases. This will be true until demand becomes elastic, in which case an increase in
price (or equivalently a reduction in output) will decrease both total revenue and total cost.
Provided MC > MR, the reduction in output will increase profits.
2. Deadweight loss is equal to DWL = (1 2)(P m MCm)(Qc Qm). Multiply through by
=
( )( )
P m =P m Qm =Qm , rearrange and let K = (Qc Qm)=Qm . Then DWL equals
(1=2)KLP mQm:
If MC is constant then Qc = 2Qm and K = 1. For K < 1, Qc < 2Qm which is true if
marginal cost is increasing.
3. The welfare effects of lowering the price of long distance service (market one) to its marginal
cost and raising the price of local service (market two) to its marginal cost are shown in Figure
2.1. Consumers of local service suffer a loss of surplus equal to the sum of areas A and B,
so the move is not a Pareto improvement. It is however a PPI since total surplus increases by
F in market one and by G in market two. If the firm was breaking even at the initial prices,
quasi-rents in market one of E would equal the sum of its operating losses in market two (the
sum of areas G, A, and B) and its fixed costs.
4. (a) Since the number of cabs is unstated, it is reasonable to infer that this is a question about
the long-run equilibrium and that the number of cabs is variable. If so zero economic
profits requires that the equilibrium price be 5. At this price the number of rides is 1000,
so if each taxi cab operates at capacity, the minimum number of taxi cabs is 50.
3
D
P1
C E F
MC
G
A B
P2
D2 D1
Q2 Q1
Figure 2.1: Problem 2.3

Chapter 2. The Welfare Economics of Market Power 5
D2 SSR(n=50)
SSR(n=70)
D1
25 (b)
(a) (c)
5
Q
1000 1400
5
(b) For P the supply of rides is perfectly inelastic and equal to 1000. The price that
clears the market by setting the quantity demanded equal to 1000 is 25. So P = 25
,
Q = 1000 , and the profit per taxi cab is 400.
(c) Free entry insures that the price is restored to 5. At this price 1400 rides are demanded
and the minimum number of taxi cabs will be 70. Profits of 400 attracts entrants, and
price falls as supply increases until profits become zero.
(d) See Figure 2.2
5. 5
(a) For p the short-run supply is 1900 rides. Demand equals supply when p =5 and
the profit per taxi cab is zero. The taxi cab market is in long-run equilibrium because
economic profits are zero.
(b) In the short-run capacity is fixed and the price must increase to 55 to reduce demand to
1900 rides. At this price the profit per taxi cab is 1000. The long-run competitive price
is 5, so the new long-run equilibrium number of rides is 2900, provided by 145 taxi cabs,
and the profit per taxi cab is zero.
(c) In aggregate at p =5
the taxi cab drivers are willing to supply up to 2900 rides. At p=5
demand is 1400 rides, so there is no capacity constraint and the short-run equilibrium is
q = 1400, p = 5, and = 0. It is reasonable to assume that each cab will operate at

capacity in the long run, so there will be exit and the number of cabs will fall to 70 and
the long-run equilibrium will be q = 1400 = 5 = 70
,p ,n , and =0
.
6. (a) Setting price equal to marginal cost and solving for y, the supply function of a competi-
y(p) = 2pw
tive firm is
if f is a sunk expenditure since then marginal cost is always greater than average avoid-
able cost. If f is not sunk then this is the supply function provided p > pmin . Setting
MC =8 = 4 + 100
y equal to AC y =5
=y and solving for y yields ymin , so pmin = 40
.
If p = p then the firm is indifferent between supplying 0 and 5. For p < pmin the
min
firm will supply 0.
Since there are n identical firms, the market supply function is
ny(p) = S (p) = 2npw

if f is sunk. If f is avoidable then this is the supply function provided p > 40. For
p < 40 supply is zero.
(b) The competitive equilibrium price is found by setting supply equal to demand, D(p) =
S (p), or
A p = 2npw :
Solving for p,
pc (n) = 2w2wA + n:
Substituting this into either D(p) or S (p) the equilibrium quantity is
Qc (n) = 2wnA+ n :
Substituting the equilibrium price into the supply function for a firm, each firm produces
yc (n) = 2w A+ n :
The profits of each firm are defined as = py C (y). Substituting in the expressions
for p, y, and C (y) as a function of n, equilibrium profits as a function of n are
2
c (n) = wA 2 f
(2w + n)
if f is avoidable and equilibrium quasi-rents are
2
c (n) = wA 2
(2w + n)
if f is a sunk expenditure.
(c) For f avoidable p = 80, Q = 20, y = 10, and = 300. The market equilibrium is
illustrated in Figure 2.3 and the representative firm in Figure 2.4. For f sunk the market
equilibrium is illustrated in Figure 2.5 and the representative firm is illustrated in Figure
2.6 in the short-run and Figure 2.4 in the long-run.
P
S
80
Q
20
Figure 2.3: Problem 2.6—Market f avoidable
MC
AC
80
40
Q
5 10
Figure 2.4: Problem 2.6—Firm f avoidable
P
S
80
Q
20
Figure 2.5: Problem 2.6—Market f sunk
MC AAC
80
Q
10 20
Figure 2.6: Problem 2.6—Firm f sunk
(d) The equilibrium in part (c) is a short-run equilibrium since an entrant would anticipate
positive economic profits.
(e) The long-run equilibrium is p = 40 = 5 = 60 = 0
,y ,Q , , and n = 12. To find this
we know that in the long-run the price must equal minimum long-run average cost. This
=5
is 40, at which price demand is 60. Since qmin , this requires 12 firms.
7. If the monopolist operated one plant, then ym is defined by setting marginal revenue (MR =
100 2 8
y) equal to marginal cost ( y), so ym = 10, pm = 90 , and m = 400 . However,
this should be suspicious, since the two competitive firms in 6(c) earned in aggregate profits
of 600! The paradox is resolved by observing that marginal cost is increasing, so that using
only one plant is not going to be profit-maximizing. Instead the monopolist will be willing to
incur the fixed costs associated with additional plants in order to reduce its production costs—
by producing less in each plant. When the monopolist uses n plants it minimizes costs by
allocating production equally, so at the level of the firm marginal cost as a function of n is
MC (n) = 8ny :
n is found by equating MC(n) to MR or
Its profit- maximizing quantity as a function of
ym = 2n100+n 8 :
Substituting this into the definition of profits, the monopolist’s profits as a function of its
number of plants n are
m (n)

100 2
= 2n + 8 [n2 + 4n] 100n:
Since n must be an integer, we can progressively substitute in n = 1, n = 2, n = 3, etc. until
profits start to decline. The monopolist maximizes profits by building 6 plants, each of which
produces 5 units. The profit-maximizing quantity for the firm is 30, the profit-maximizing
price is 70 and monopoly profits are 900.
Calculus can be used to solve this problem more directly. The expression for profits as a
function of n can be maximized with respect to n. Or alternatively, the firms profits as a
function of y and n,
= (100 y)y 4 ny N 400n;
2
can be jointly maximized with respect to y and n.
8. (a) See Figure 2.7. The socially optimal amount is produced because at the monopoly quan-
=
tity the willingness to pay is equal to the social marginal cost ( P m SMC). This result
does not generalize since we would not expect the difference between price and marginal
revenue (the loss on inframarginal units) at the monopoly output to equal the difference
between social and private marginal cost.
(b) See Figure 2.8. Too little cantaloupe is produced, because the monopoly price is greater
than social marginal cost ( P m > SMC).
(c) See Figure 2.9. Too much cantaloupe is produced because the competitive price is below
social marginal cost (P c < SMC).
Pm SMC
PMC
MR D
Q
Qs=Qm
Figure 2.7: Problem 2.8(a)

Pm
SMC
PMC
MR D
Q
Qm Qs
Figure 2.8: Problem 2.8(b)

SMC
Pc PMC
MR D
Q
Qs Qc
Figure 2.9: Problem 2.8(c)

CS
65
Profits
DWL
MC=AC
MR D
Q
35
(d) Breaking up the monopoly to promote output expansion might exasperate the problem
associated with the negative externality. Taxing the monopolist to reduce the impact of
the negative externality may well reduce output, potentially increasing the deadweight
loss associated with market power. This problem is an application of the Theory of
Second Best, since the presence of two distortions (a negative externality and market
power) means that correcting only one of them may make things worse.
9. The assertion is false. A profit-maximizing monopolist will set its output ( Qm ) such that
MR Qm ( )= ( )
MC Qm . A reduction in the sales tax, will to the extent it was borne by the
monopolist, reduce its marginal cost, and to the extent it was borne by consumers its elimi-
nation will increase demand and hence marginal revenue. In response, a profit-maximizing
monopolist will increase its output and lower its price.
10. (a) MR(Q) = 100 2Q. See Figure 2.10.

(b) MC = AC = 30.
(c) Qm = 35, P m = 65 and m = 1225. The maintenance of monopoly profits requires
that entry be restricted. In the absence of barriers to entry, competition from entrants
will reduce or eliminate the monopolist’s profits.
(d) The efficient level of production occurs where P = MC = 30 . At this price, demand
is70 so Qs = 70 : The missing surplus is the sum of deadweight loss and monopoly
profits. Monopoly profits are a transfer from consumers to the firm. The deadweight
loss is surplus that is not realized because the monopolist restricts output such that units
for which willingness to pay exceeds marginal cost are not produced.
Chapter 3
Theory of the Firm
1. If average cost decrease as quantity increases then there will be economies of scale. For
0 1
< a < economies of scale exist if
AC (aq) > AC(q);
or
f + c(aq) > f + c(q) :
aq aq q q
This is true for a close to zero provided c(q) is bounded, that is it does not go to infinity as q
goes to zero.
2. Economies of scale are global if S (q) = MC(q)

AC(q)
> 1 for all q. Since
f + 1 > 1;
S (q) = qc
economies of scale are global.
3. S = MC
AC(q) , MC(q) = 2cq, and AC(q) = f=q + cq. So
(q)
f + 1:
S = 2cq 2 2
q
(a) Economies of Scale exist if S > 1 or q < fc .
q
(b) Constant Returns to Scale exist if S = 1 or q = fc .
q
(c) Diseconomies of Scale exist if S < 1 or q > fc .
If f =0 1
then S < for all q and there are always diseconomies of scale since production
costs are increasing and there are no fixed costs to spread out over more units of output to act
as a countervailing force on average cost.
17
4. (a) If e is observable then for the band to be just willing to exert the high level of effort,
( 2) = 0
U yh ; , so yh =1. For the band to be just willing to exert the low level of effort
( 2) = 0
U yl ; , so yl = 0 25
: . The firm’s expected profits if it offers the following contract
(which induces high effort)
If e = eh = 2 then y = 1
If e =
6 eh = 2 then y = 0.
( = ) = 7:75. Its expected profits if it offers the following contract (which
are E e eh
induces low effort)
If e = el = 1 then y = 0:25
If e =
6 el = 2 then y = 0
( = ) = 4 75
are E e el : . When effort is observable the profit-maximizing contract
induces high effort. Since the owner of VooDoo Records can verify the effort of the
band, she needs to pay only enough income to insure that the band reaches its reservation
utility level.
(2 = = 2) = 0 and U (2; e = el ) = 1: Therefore
(b) If effort is unobservable, then: U ; e eh
= 2, but has an incentive to provide low
the band promises high effort in return for yh
effort.
(c) The optimal incentive compatible contract conditions the band’s income on realized
profits. The two income levels satisfy the band’s individual rationality constraint (IRC)
and its incentive compatibility constraint (ICC). The IRC is
(
EU yl ; yh ; el = 1) = 0
and the ICC is
(
EU yl ; yh ; eh = 2) = EU(yl ; yh; el = 1);
where EU is expected utility and the band is paid yl when low profits are realized and
yh when high profits are realized. The ICC is
0:75(2pyh ) + 0:25(2pyl ) 2 = 0:25(2pyh ) + 0:75(2pyl ) 1
0:25(2pyh) + 0:75(2pyl ) 1 = 0:
and the IRC is
Solving these two equations for the two unknowns ( yl and yh ), the optimal incentive
contract is
If = 10 then y = 1625
If = 5 then y = 161 .
The expected profits of the firm when this contract is offered and effort is unobservable
are 7.5625. When effort is unobservable the optimal incentive contract imposes risk on
the band. In order to provide the band with an incentive to exert high effort, they are
paid more in the good state and less in the bad state.
(d) The expected profit of VooDoo under full information is 7.75 and its expected profit
under asymmetric information is 7.5625. Asymmetric information means that the profits
of VooDoo are reduced by 0.1875. The expected profits are lower because inducing high
effort requires imposing risk on the band, for which they must be compensated (since
they are risk averse) through higher expected income. Higher expected income for the
band means that VooDoo’s expected costs are higher and its expected profits lower.
Chapter 3. Theory of the Firm 19
5. (a) The full information contract is:

If e = eh = 2 then y = 2
If e =
6 eh = 2 then y = 0.
(b) By exerting e = 1, but promising e = 2 utility increases to 1.
(c) The optimal incentive contract is:
If = 10 then y = yh = 5=2.
If = 5 then y = yl = 1=2.
(d) The expected income with and without asymmetric information is 6.75. Risk neutral-
ity means that incentives can be provided by imposing risk without having to increase
expected average compensation.
6. (a) The efficient effort levels maximize the income of the team. The income of the team
equals output less the costs of effort:
= 16(e1 + e2 ) 2e12 2e22:

The efficient level of effort for each team member (ei ) is found by setting its marginal
benefit equal to its marginal cost. The marginal benefit of another unit of effort is the
increase in team income which is 16. The efficient level (ei ) is defined by:
16 = 4ei
or ei = 4. At the efficient effort levels x = 128, mi = 64, and ui = 32.
(b) In the partnership, each team member will select its effort level to maximize its income:
i = 16(e12+ e2 ) 2ei2:
The marginal benefit to an individual team member is 8. The utility-maximizing choice
for i sets the marginal cost of effort equal to its marginal benefit:
8 = 4ei
or epi = 2. At this level of effort, xp = 64, mpi = 32, and upi = 24.
(c) The maximum amount the two team members are willing to pay is determined by
ui P = upi
since they would then be just as well off with the monitor as without. Solving, each team
member is willing to pay 8, for a payment in total to the monitor of 16.
Chapter 4
Market Power and Dominant Firms
1. (a) Constant unit costs in the long-run imply constant returns to scale.
(b) For q k where k is capacity, MC (q) = w; q > k is not possible in the short run.
(c) (i) No entry since r would not be recovered.
(ii) Yes.
(iii) Depends on the nature of competition and capacity constraints post- entry.
(d) The incumbent would not recover r either.
(e) Capacity would be such that p = w + r so that all firms recover their capital costs.
2. (a) Since
f +1>1
S (q) = cq
there exist global economies of scale. This also means that AC (q) > MC(q) = c for all
output levels.
(b) If the incumbent can commit to charge a price equal to marginal cost, then there will not
be entry. An entrant would not be able to cover its fixed costs and would earn negative
profits. This result does not depend on whether f is sunk or not. Either way, the entrant
will not be able to recover f if they produce.
3. (a) The profits of the monopolist who leases in period t are
t = (1000 Qt )Qt:
= = 500, and its profit-maximizing outputs
Its profit-maximizing lease rates are r1 r2
=
are Q1 Q2 = 500 . Its maximized profits are
L = (1 + )25; 000:
= =
(b) If the monopolist is able to commit to prices, then Q1 Q2 Q—the stocks in the two
=
periods are the same—and q1 Q1 and q2 =0 . The willingness to pay of consumers
if the monopolist sells in the first period and is able to commit to its output —the first
period inverse demand curve—is
WTP = P1 + P2 (4.1)
21
= 1000 Q + (1000 Q)
= (1 + )1000 (1 + )Q
Setting MR equal to MC, Q = 500

. In the second period, the firm produces q2 =0
and
the profit-maximizing price profile is P1 = (1 + )500
and P2 = 500
. Its maximized
profits are again equal to
L = (1 + )25; 000:
(c) In the second period, the profit-maximizing output is q2 = (1=2)(1000 Q1): So P2 =
(1 2)(1000
= )
Q1 and first-period demand is
P1 = 1000 Q1 + (1=2)(1000 Q1) (4.2)
= (1000 Q1)(2 + ) ;
2
and first-period profits are
1 = (1000 Q12)(2 + )Q1 + (1000 4 Q1) :

2
So the profit-maximizing production profile and prices are

Q1 = 2000(4 + ) 1
q2 = 500(2 + )(4 + ) 1
P1 = 500(2 + )2
P2 = 500(2 + )(4 + ) 1
(d) The impact of a discount factor less than 1 is to increase period 1 production because
second period profits are worth less than first period profits.
4. (a) From Exercise 4.2, the profits from selling the durable are $450,000. The per period
profits from selling a nondurable as a function of its marginal cost are
= (10004 c) :
2
Since there are two periods and the discount factor is 1, the non-durable is more prof-
itable if
2 (10004 c) > 450; 000:
2
This is true if c : .51 32

(b) From Exercise 4.2, the profits from leasing the durable are $500,000, which are always
greater than the profits from selling the nondurable provided its marginal cost is positive.
(c) The efficient solution is to produce 1000 units of the durable product, since it is costless
and at that level of output willingness to pay equals zero—marginal cost. The monopolist
reduces durability—engages in planned obsolescence—to avoid the Coase Conjecture.
5. (a) The dominant firm will be an unconstrained monopolist if its monopoly price
pM = (1=2)(100 + cD )
is less than pf , or
cD < 2pf 100:
Chapter 4. Market Power and Dominant Firms 23
(b) pD = pf .
(c) pD = pM = 50.
(d) The dominant firm is a price taker: by reducing its output the price will not rise. pD=
pf = 60
, D =0, but the firm earns Ricardian rents of 1800 and each unit of capacity
would command a price of 60.
(e) In (c) yes, since the firm can exercise market power. In (d) no, since the firm cannot
exercise market power.
6. (a) For P p0 = 20, the supply function for the fringe is Qf (P ) = P .
(b) The residual demand function for the dominant firm is QD (P ) = 100 2P .
(c) P = 26.
(d) Monopoly profits are 2401; the profits of the dominant firm 1152 Total surplus under
monopoly is 3601.5, and the total surplus when there is a dominant firm is 4028. In this
case a dominant firm is more efficient than a monopolist as it results in a larger total
surplus.
Chapter 5
Non-Linear Pricing and Price

Discrimination
1. The marginal price is p = 0:20. The monopolist is able to extract all of the surplus in this
case, so CS =0
, and PS = 0:32.
2. (a) In order to make this type of arbitrage unprofitable, the price differential must be less
than the transportation cost, v2 v1 < t.
(b) The manufacturer will successfully make arbitrage unprofitable if v1 (1 )v2 .
3. (a) The uniform monopoly price is pm = 3.
(b) The price charged to the under 25 group is p = 2:80, and the price charged to the over
25 group is p = 3 50
: .
(c) Each group should pay the marginal cost of each drink, plus a cover charge equal to
2 00
their consumer’s surplus at that point. Both types pay : per drink. Those under 25
pay a cover charge of CS p ( = 2 00) = 6 40
: : , and those over 25 pay a cover charge of
( = 2 00) = 9 00
CS p : : .
(d) At a price of 2:00 per drink, students obtain a surplus of 6:40. Thus, this is the maximum
cover charge that will still attract both types of consumers.
(e) After midnight all consumers are students, so the monopolist should charge the optimal
student price of p = 2 80 3 15
: . Before midnight the optimal price is : . The monopolist
$6 40
is practicing partial third degree price discrimination. The profits for part (d) are :
$3 78
and the profits for part (e) are : .
4. a < 13. Type 2’s have a lower valuation of network television while type 1’s have a lower
valuation of sports. Bundling takes advantage of this negative correlation by “averaging” the
reservation prices.
5. $4 00
(a) The manufacturer should charge Type 1 individuals : , their valuation for the good.
$1 25
Since their time cost is : , the coupon would have to offer at least this large a discount
before the Type 1’s are tempted to use them.
$4 00
(b) At a price of : , Type 2 individuals will only buy cereal if they receive a discount of
$1 00
: , the difference between their valuation and the price that they must pay.
25
$4 00
(c) Using parts (a) and (b), the manufacturer should set a price of : and have a coupon
$1 00 $1 50
discount of : . This means that he is making a profit of : from the Type 1’s,
$0 50 $0 50
and : from the Type 2’s. An increase in manufacturing costs by : would reduce
profits from the Type 2 individuals to zero, so there is no longer any incentive to offer
coupons.
6. (a) The optimal two-part tariff has a variable price of p = 1+ , and a fixed fee of T =
4 1 2
.
(b) Since (1+ ) 2 1
> whenever < , the optimal price is above marginal cost as long as
there are two types of consumers in the market. It is only possible to raise the price
of a competitively supplied good above marginal cost if it is tied to the purchase of the
associated primary good.
(c) The optimal two-part tariff has a variable price of p = 2, and a fixed fee of T = 32.
(d) If = 1 , then the profits from the pricing scheme in part (a) are 9.5, and using the
2
pricing scheme in part (c) are 16. Thus, the pricing scheme in part (c) yields greater
profits. If = 3 , then the profits from the pricing scheme in part (a) are 9.1, and using
4
the pricing scheme in part (c) are 8. In this case, profits are higher using the pricing
scheme in part (a). Note that if the proportion of low demand consumers is sufficiently
low, then profits are maximized at prices which deter them from entering the market.
Chapter 6
Market Power and Product Quality
1. (a) Consumers are willing to pay their expected valuation pvH + (1 p)vL .
(b) A high quality firm will only produce if they receive non-negative profits, pvH + (1
)
p vL cH . This implies that p 45 . Thus, only values of p greater than (or equal to) 45
could be consistent; below this, the price is not high enough to induce any high quality
producers to produce, so all goods would be of low quality.
2. Firm 1 should choose the highest quantity that would not be profitable for a low quality firm
at price vH ; i.e. q such that the low quality producer is indifferent between p vL ; ( = and 10)
( )
vH ; q . Since the producer can only sell up to 10 units, his profit maximizing strategy is to
restrict supply to 2.5 units and set a price of vH .
3. ( ) ( )
(a) Firm 1’s total profits are 32 p cH , and firm 2’s total profits are 12 p cL , where p
is the common price. The adverse selection problem could arise if firm 1 realizes that it
would be more profitable to switch to a low quality product; i.e. if
p < 32 cH 1 cL:
2
(b) A high quality firm would choose the lowest level of advertising expenditure that sig-
naled high quality (i.e. would be unprofitable for the low quality firm); thus expenditure
is A = ( )
1 p cL . At this level of expenditure, consumers would buy everything from
2
firm 1, realizing that it must be the high quality firm; thus, profits are 32 p 2 +
cH 12 cL ,
which is lower than the profit with no advertising. Since firm 1 chooses not to advertise
(and makes higher profits than firm 2 by assumption), there is also no incentive for firm
2 to advertise.
4. Using the expression in the text, expected lifetime cost equals 200.
5. (a) Note: The probability of a high quality good breaking down is not p as stated in
the first printing of the book. The consumer chooses e to maximize expected utility,
(1 ( )) + ( ) =
e V e V e V e. Since expected valuation of the good (under full
warranty) is independent of effort, the consumer will choose e =0 to minimize the
=
disutility of effort. In this case l, so the high quality manufacturer has no incentive
to offer a warranty (the question states that this is true for low quality goods).
27
(b) At e= 0, = l, expected consumer utility is given by (1 l)V + l(1 a)V = V laV .

At e= 1, = h, expected consumer utility is given by (1 h)V + h(1 a)V 1 =
V haV 1. Thus, the consumer will choose e = 1 if a > (l 1h)V .
6. (a) The monopolist chooses sm = 13 , qm = 13 , and pm = 29 .
(b) Consumer’s surplus evaluated at qm = 13 is 18 1 s. Profits evaluated at qm = 1 are
3
2 1 2
9s 3s .
(c) At qm = 13 , the socially efficient level of quality is s = 12
5 . Thus, given the monopolist’s
choice of quantity, he chooses an inefficiently low level of quality.
7. (a) With observable type, ticket prices should be set to extract all surplus. For the business
travelers qH = 60 and TH = 3000. For the vacation travelers qL = 20 and TL =
600 . These qualities are socially efficient, since they satisfy Vi0 q ()=c (so marginal
benefit equals marginal cost). Note that qH =3 =5
qL; while TH TL : This is because
the monopolist is practicing perfect price discrimination and business travellers have a
greater willingness to pay for quality.
(b) With unobservable type and the above pricing scheme, business travelers would also
purchase economy tickets, since V H qL ( ) TL= (80)(20) 1
2(20) 600 = 800 0
2 >
(recall that the airline extracts all surplus with the intended packages, so V H qH( )
TH = 0)_
. Note that profits are currently much higher on the first class seats ($1800
vs $200), so the airline would like to separate types with self-selecting tariffs. For all
() ( );
quality levels above 15, V H q is higher than V L q thus, the only way to prevent
business travelers from mimicking vacation travelers is to chose a quality level qL ; 15
the optimum is then qL = 15 , since in this range of quality levels, profits from economy
tickets are still increasing in q: Ticket prices should still be set so as to extract all surplus,
and there is no reason to distort qH : Thus, the new optimum is qL = 15 =
, TL V L ql( )=
$487 50
: , qH = 60 =
and TH V qH H ( ) = $3000 .
Chapter 7
Game Theory I: Static Games of

Complete Information
1. The Nash equilibrium strategy profile is s = (M; M ).

2. For Janet to be willing to mix over Ballet and Hockey she must be indifferent—her payoff
must be the same. Janet’s expected payoff from playing Hockey is
(H )J = pBH 3 + (1 pBH )0

where pB
H is the probability Bob plays Hockey. Janet’s expected payoff from Ballet is
(B )J = pBH 0 + (1 pBH ):
Set
(H )J = (B )J
H : pH = 1=4 and therefore pB = 3=4. Bob plays Hockey with probability 1/4,
and solve for pB B B
and Ballet with probability 3/4. Similarly, Janet plays Hockey with probability 3/4 and Ballet
with probability 1/4.
4
3. The set of rationalizable strategies for Player 1 is fR g and for Player 2 it is fC 3g. The
unique Nash equilibrium strategy profile is s = ( 4 3)
R ;C .
4. (a) All strategies for both players are rationalizable, since every strategy is a best response
to N. The normal form of this game, where A is the row player and B is the column
player, is shown in Figure 7.1.
LO TO LT N
LO -10, 30 0, 0 -10, 30 0, 0
TO 0, 0 30, -10 30, -10 0, 0
LT -10, 30 30, -10 20, 20 0, 0
N 0, 0 0, 0 0, 0 0, 0
Figure 7.1: Normal Form for Problem 7.4
29
(b) If A and B could cooperate they would both choose Library and Theater for a payoff of
20 each.
(c) Both playing Library and Theater is not a Nash equilibrium since both players have an
incentive to deviate. Supporting only the preferred option would increase utility by 10.
5. 2
(a) In order for R to be strictly dominated by R1, it must be the case that b < 2.
(b) Since 2(R2; C 2) > 2(R2; C 1) C 1 can never strictly dominate C 2.
(c) For (R1; C 1) to be a Nash equilibrium a 2 and c 3.
(d) For (R1; C 1) to be a unique Nash equilibrium d > 4 and c < 3 OR e < 1 and c < 3.
6. (a) For each play to mix, they must be indifferent over the pure strategies that comprise
their mixed strategy. This means that the expected payoff from any pure strategy in the
mixed strategy is the same. Given the probabilities set the expected payoff for Player
1 of playing T equal to the expected payoff from paying B. Similarly, set the expected
payoff for Player 2 from playing L equal to the expected payoff from playing R. This
creates two equations in two unknowns—a and c. Solving, a = 1 and c = 3 .
(T; L) and (B; R).
(b) The two pure strategy Nash equilibria are
7. The two Nash equilibria are (D; L) and (M; M ).
Chapter 8
Classic Models of Oligopoly
1. (a) A firm’s best-response function gives its profit-maximizing quantity for any belief it has
50 2
regarding its rival’s level of output. Setting firm i’s marginal revenue ( 4
q j qi )
equal to its marginal cost (2) and solving for qi, i’s best-response function is
qi = 24 2 qj :
(b) Since the firms are symmetric (identical costs and output homogenous) q1C =q2C =
C = = C
q . Setting qi qj q in i’s best-response function and solving for q : q C C =8.
Aggregate output is 16 . Substituting QC = 16into the demand curve, P C = 18
:
The profits of each firm are 1C =2C = 128
. The equilibrium is not efficient since
P C = 18 > MC .=2
2. (a) q1C = q2C = qC = 60
1C = 2C = C = 3600
P C = 80
(b) See Figure 8.1.
(c) Because of symmetry and constant unit costs, the joint profit-maximizing price is the
monopoly price. The profit-maximizing monopoly output sets its marginal revenue
(200 2 Q) equal to marginal cost (20): Qm= 90 = 110
. The monopoly price is P M
and monopoly profits are M = 8100 . Any combination of q1 and q2 that satisfies
+ = 90
q1 q2 implements the monopoly outcome. Letting the share of firm 1 be
0 1 = 8100
(where < < ) then the profits of the two firms also sum to 8100 with 1
and 2 = (1 )8100
: For instance if the two firms split the monopoly output equally
( =1 2 = = 90
= ) then q1 q2 = = 4050
and 1 2 .
(d) The marginal revenue for firm i is MRi = 200 qj 2qi: Substituting in qi = 90
and qj = (1 )90, MRi = 110 90 which is greater than marginal cost (20) for
0 < < 1. For instance when they split the monopoly output equally, = 1=2 and
MRi (45; 45) = 65 > MCi = 20. The optimal defection for each firm to produce on its
best-response function. If qj = (1 )90 then
qiD = 180 (12 )90 :

31
q2
180
Firm 1's Best Response
90
Cournot Equilibrium
60
Firm 2's Best Reponse

q1
60 90 180
Chapter 8. Classic Models of Oligopoly 33
If =12
= , then qiD = 67 5
: . Since each firm has an incentive to increase output, the
collusive agreement is unstable.
(e) The free-entry number of firms is N C = 8.
3. Setting the marginal revenue of firm i (30 qj 2qi ) equal to its marginal cost ( ci ) and solving
for qi, the best-response function for firm i is:
qi = 30 c4i 2qj :
Of course the best- response function for firm j is:
qj = 30 c4j 2qi :
This is a system of 2 equations in 2 unknowns. Substituting qi into the best-response function
for qj and solving for qj , the equilibrium quantity for firm j is
qjC = 30 2cj + ci :
6
Similarly,
qiC = 30 2ci + cj :
6
Since QC = qiC + qjC ,
QC = 60 c6i cj ;
and after substituting this into the demand curve,
P C = 30 + c3i + cj :
The profits of firm i are iC = (P C ci)qiC and since
P C ci = 30 23ci + cj ;
iC = (30 218ci + cj )2 :
4. Using the formulas from the preceding question, in the premerger Cournot equilibrium:
QC = 55
, PC = 45 = 400 = 1225
, 1C , 2C = 1512 5
, CSC : , and TSC = 3137 5
: , where CS
is consumer surplus and TS is total surplus. Postmerger the monopoly outcome is:
QM = 45 , PM = 55
, 1M= 2025 = 1012 5
, CSM : , and TSM = 3037 5
: . On efficiency
grounds the merger should be blocked since total surplus falls.
5. (a) In the no-trade (NT) equilibrium the single firm in Country A is a monopolist:
qANT = (1 2)(40
= )
cA , PANT = (1 2)(40 + )
= cA , and ANT = (1 4)(40
= cA 2 . )
In the no-trade equilibrium the Cournot equilibrium in Country B is:
qBNT = (1 3)(40
= )
cB , PBNT = (1 3)(40 + 2 )
= cB , and BNT = (1 9)(40
= cB 2 . )
(b) Demand in Country A is QA = 40 P and demand in Country B is QB = 40 P , so

the world demand curve (Q A + = 80 2
B
Q ) is Q P . Solving for P, the inverse demand
curve for the world is P = 40 2 Q= . The world trade equilibrium is found by solving
for the 3 firm Cournot equilibrium. Setting marginal revenue equal to marginal cost for
the single firm in A,
40 qBi =2 qBj =2 qA cA = 0;
and for firm i in Country B,
40 qBj =2 qA=2 qBi cB = 0:

Recognizing that qB i = qj = qB , we have two equations in two unknowns which
B T = (1=2)(40 3cA + 2cB ) and qT =
when solved give equilibrium production: qA B
(1=2)(40 2cB + cA). Worldwide production equals QS = qAT + 2qBT or
QS = 120 c2A 2cB :
Since Country A consumes half of world supply, the imports of country A are MA =
QS =2 qAT or
MA = 40 + 5c4A 6cB ;
which is greater than 0 if
cA > 8 + 65 cB :
If cA = 8 and cB = 10, then country A is an importer. This pattern of trade is inefficient
since A is the low cost country. If cA = 2 and cB = 10 then country A is an exporter
and the pattern of trade is globally efficient.
(c) For these marginal costs, A is an importer and the non-tariff or free trade equilibrium (F)
is:
QF = 46 , PF = 17 = 18F
, qA =5 ; MA , qBF = 14
, CSA F= 264 5 : , CSBF = 264 5 :,
BF = 98
, and AF= 162 = 426 5
. TSA F F
: , and TSB = 460 5
:.
With a marginal cost of 12 for Country B, Country A remains an importer. The tariff
equilibrium (T) is:
QT = 44 = 18
, PT T
, qA = 20 , MA =2 , qBT = 12
, CSAT = 242 = 242
, CSB T T
, B= 72 ,
and AT = 200 . Since tariff revenue is 4, TSA T = 446
, and TSB= 386
T . Total surplus
increases in Country A by 19.5. Total surplus worldwide decreases by 55.
6. (a) If there is no trade then each firm acts as a monopolist in its own country. The equilib-
rium quantities are QM = 18 and P M = 22.
(b) KJ’s marginal revenue on the first unit that they sell to the United States is equal to the
monopoly price in the United States: MRKJ ; (0 18) = M
PUS = 22 , the monopoly price
since it does not have any inframarginal units.
(c) In the trade case there is Cournot duopoly in each country. The equilibrium in each
country is: qiC = 12 = 16
, PC , i = 144 , and CS C = 288
. In the no trade case, iM =
324 and CS M = 162 . Consumers prefer the trade equilibrium since consumer surplus
is greater. While, the profits of each firm fall from 324 to 288 (144 in each market),
total surplus increases with trade by 90 in each country. The trade equilibrium without
arbitrage is an example of market segmentation or third degree price discrimination.
(d) Australia and U.S. with transportation costs of $3: cKJ =4 and cNW =7 . Using
the results from Problem 3, the equilibrium in Australia is:
qKJ = 13 = 10
, qNW = 17
, PA , KJ= 169 , NW = 100 , CSA = 264 5 : and
TSA = 533 5 : . The United States equilibrium is symmetric.
Australia with $3 per unit transportation cost and $3 per unit tariff: cKJ =4 and
cNW = 10 . Using the results from Problem 3, the equilibrium in Australia is:
qKJ = 14 , qNW =8 , PA= 18 , KJ= 196 , NW= 64 , CSA = 242 and
TSA = 562 .
US with $3 per unit transportation cost: cNW =4 =7
and cKJ . The equilibrium
in the U.S. is:
qNW = 13 = 10
, qKJ = 17
, PUS , NW = 169 , KJ= 100 , CSUS = 264 5 : and
TSUS = 533 5 :.
With the tariff Australian consumers’ welfare is reduced by 22.5. However, the
profits of KJ increase (summing across the two countries) by 27 and the Australian
government earns revenues of 24. Without retaliation the total gains from trade for
Australia increase by 28.5.
Australia and U.S. with a tariff of $3 per unit transportation cost and $3 per unit
tariff: cKJ =4 = 10
and cNW . Using the results from Problem 3, the equilibrium
in Australia is:
qKJ = 14 , qNW = 8 = 18
, PA , KJ= 196 , NW = 64 , CSA = 242 and
TSA = 562 . The United States equilibrium is symmetric. The total gains from
trade received by Australia equal 526, less than the 533.5 earned when there is no
trade protection.
(e) Even with retaliation, the gains from trade for Australian’s will increase by approxi-
mately 5 (4.89) by imposing a tariff of $1 per unit. There is a trade off involved. Im-
posing a tariff raises the domestic price, decreasing consumers surplus and retaliation
reduces profits in the export market. However, there is a resource saving (reduced trans-
portation costs) and a gain in profits from switching to low-cost domestic producers and
away from high-cost foreign producers.
7. (a) Using (8.44) in the text and observing that dqi=dpi = 5: pi = (1=5)(12 + pj ).
(b) Since the equilibrium will be symmetric set p1 = p2 = pB in the best-response for i.
Solving, pB =3.
(c) Use (8.51) and observe that dqi=dpi = 5 and dqj =dpi = 2. Then by symmetry set
= =
p1 p2 pM . Solving, pM . =4
(d) See Figure 8.2.
(e) iM = 48 and iB = 45.
(f) Substituting pj = 4 into i’s best-response function pD
i = 3:2 and thus i = 51:2.
D
8. Consider the incentives for firm 1 to charge a price higher than ph . A price greater than ph
ensures that it will be the high price firm. In firm 1’s case, given that it is going to be the high
price firm and thus that q2 = = ( )
k2, firm 1’s profit-maximizing choice is to sell q1 R1 k2
units which it can by charging ph . Hence it does not want to charge a price greater than ph . By
construction, firm 1’s profits are lower if it charges a price less than pl . Recall the derivation
of pl .
In order to see that firm 2 does not have an incentive to deviate, we need to derive firm 2’s
equilibrium profits. By construction, its expected profit from playing any price in the interval
p2
R1(p2)
4 M
B
R2(p1)
3
p1
3 4
[pl; ph] is the same, otherwise it would not be willing to mix over prices in the interval. Its
profits if it plays pl will be pl k2 since it will be the low price firm and it is capacity constrained.
Clearly, firm 2 also does not want to charge a price less than pl . Doing so will continue to
guarantee that it is the low price firm but since it is capacity constrained at pl its profits will be
reduced as its price falls. Moreover, firm 2 will not find it profitable to charge a price higher
than ph . Doing so will guarantee that it is the high priced firm. The best that firm 2 can do if
( + ( )) ( )
it is the high-price firm is earn profits of P k1 R2 k1 R2 k1 which are its Cournot profits
( + ( ))
if firm 1 makes sales of k1 . However, P k1 R2 k1 is less than P R1 k2 ( ( )+ ) k2 since
k1 > k2. Firm 1 charges a higher price than firm 2 when it acts as a monopolist on its residual
demand curve because its residual demand is larger than that of firm 2 since the capacity of
firm 1 is larger. As a result, firm 2 will not find it profitable to charge a price greater than ph .
Chapter 9
Game Theory II: Dynamic Games of

Complete Information
1. Imperfect Information:
There are two subgames, the game itself and a second that begins at player 1’s decision
node. The normal form for the subgame that begins at 1’s decision node is shown in
Figure 9.1.
( ) ( )
There are two Nash equilibria to this subgame: a1 ; B1 and a2 ; B2 . This means that
there are also two SPNE. They are: (i) for player 2 b2 , if b1 then B1 ; for player 1 a1 ;
and (ii) for player 2 b1 and then B2 ; for player 1 a2. If player 1 gets to move, it should
anticipate that player 2 expects the Nash equilibrium to the subgame that follows to be
( )
a2 ; B2 since her payoff from playing b2 is 4. She would only play b1 if she expected
to do better.
Perfect Information:
By backward induction, the unique SPNE strategies are player 2 goes b2, if a1 then B1 ,
if a2 then B2 , and if a3 then B1 . For player 1 a1.
2. In order for the tit-for-tat strategy to be a SPNE it must induce a Nash equilibrium in every
subgame. Suppose that player 1 played Clam in the first period. The tit-for-tat strategy means
( )
that the outcome in the second period will be C; C . However, this is not a Nash equilibrium
to the second-period subgame since both players have an incentive to deviate and play Rat. In
general we know that regardless of the history in the first period, the Nash equilibrium in the
(
second period will be R; R . )
3. (a) s = (R1; C 1).
B1 B2
a1 5, 3 1, 2
a2 0, 4 4, 6
a3 3, 5 2, 1
39
(b) Player 2 has an incentive to deviate to C 3 and increase her payoff to 4. The net present
value of cooperating for player 2 is
V C = 1 3 :
The net present value of deviating for player 2 is
V R = 4 + 1 2 :
> V R or > 1=2. Since
There is no incentive to deviate if V C
= 11 + p ;
this is true if (1=2)(1 ) > p where is the rate of time preference.
(c) (1=5)(4 ) > p.
4. (a) Using Figure 9.18, there are 5 subgames. These subgames are:
(i) The entire twice repeated game

( )
(ii) C; C in period 1 and Figure 9.18 in period 2
( )
(iii) C; R in period 1 and Figure 9.18 in period 2
( )
(iv) R; C in period 1 and Figure 9.18 in period 2
( )
(v) R; R in period 1 and Figure 9.18 in period 2
(b) The SPNE is player 1 plays R in both periods and player 2 plays R in both periods.
(c) If =1, then the players value the future the same as they value today. If a player
deviates then their payoff today increases by 4, but their payoff tomorrow and every
period thereafter is reduced by 5. An individual with a discount factor equal to 1 would
not exchange a stream of payoffs of 6 every period for one of 10 today and 1 every
period thereafter. The gain in the current period is more than offset by the loss in the
next period, to say nothing of the continual reduction of 5.
(d) The smallest value of the discount factor for which grim strategies support the coopera-
tive outcome as a SPNE is =4 9
=.
5. (a) The strategies for player 1 are:
(1a) d1, u2
(1b) d1, d2
(1c) u1, u2
(1d) u1, d2
The strategies for player 2 are:

(2a) if d1 then U , if u1 then U
(2b) if d1 then D, if u1 then U
(2c) if d1 then D, if u1 then D
(2d) if d1 then U , if u1 then D
Chapter 9. Game Theory II 41
2a 2b 2c 2d
1a 6, 6 0, 0 6, 6 0, 0
1b 6, 6 0, 0 6, 6 0, 0
1c 3, 8 3, 8 2, 5 2, 5
1d 2, 1 2, 1 1, 1 1, 1
U D
u2 6, 6 2, 2
d2 2, 2 1, 1
(b) The normal form of the game is given by Figure 9.2.

The five Nash equilibria are s1 = (1 2 )
c; d , s2 a; = (1 2a), s3 = (1a; 2b), s4 = (1b; 2a),

and s5 = (1 2 )
b; b .
(c) There are three subgames. The subgames are the entire game, the game following u1
and the game following d1.
(d) The unique SPNE strategies are 1a for player 1 and 2a for player 2.
(e) The SPNE provides a better prediction because it is able to filter out non-credible threats.
The interesting non-credible threat supports the first Nash equilibrium s1 in which player
2 threatens to play D if player 1 plays d1. However, if player 1 chooses d1 then player 2
would rather play U than D.
6. If player 1 plays d1 then the normal form of the resulting subgame is shown in Figure 9.3.
The unique Nash equilibrium is (u2; U ) with payoff (6, 6).
If the path is u1 then U , player 1 goes u2. If the path is u1 then D, player 1 will play d2. This
means that if player 1 goes u1, player 2 will go U and the payoffs will be (3, 8). Consequently
player 1 chooses to play d1. To summarize, player 2’s SPNE is to always go U . Player 1’s
SPNE is to go d1 and then U2 ; if the path is u1 then U , then u2 . If the path is u1 then D, then
d2 .
Chapter 10
Dynamic Models of Oligopoly
1. (a) The monopolist profit maximizes by producing where its marginal revenue ( 20 2
Q)
2
equals its marginal cost ( Q). Setting MR equal to MC and solving for Q the profit-
maximizing level of output is QM =5
. At this level of output profits are M .= 50
(b) The conditions which must be satisfied for industry profits to be maximized when there
are two firms are:
( )= ( )
(i) MR Q MC1 q1
and
( )= ( )
(ii) MR Q MC2 q2
or
MR Q( )= ( )=MC1 q1 ( ) = +
MC2 q2 ; where industry output Q q1 q2 . Because marginal
= =
cost is increasing, profit maximization means that q1 q2 q and then Q =2 q and
=
MC1 MC2 Substituting Q =2q and MR Q( ) = 20 2 Q into (i) gives 20 4 = 2
q q .
Solving, q
Q
= 10 3 = 100 3
= . The profits of each firm are i = , and industry profits are
= 2 = 200 3
i = .
(c) Increasing marginal costs mean that costs are reduced and profits larger if output can be
split between two plants.
(d) For it to be a Nash equilibrium neither firm can have an incentive to unilaterally deviate.
However, both have an incentive to deviate since at the output (10 3 10 3)
= ; = the marginal
revenue of a firm is 30/3, but its marginal cost is only 20/3.
2. For this specification of demand and costs, i = 1800, ic = 1600; and ir = 2025: The
present value of cooperating is
V = 11800 :
The present value of cheating on the agreement is
V r = 2025 + 2025 + 11600

since cheating profits are earned for two periods. Setting V =V r and solving, the crit-
ical discount factor that sustains collusion using grim punishment strategies if detection of
deviation requires two periods is = 0 7276
: .
43
3. (a) If competition is Bertrand, then i = 1800

, iB = 1600
; and ir = 3600
: The values of
the discount factor where grim punishment strategies will support an equal division of
monopoly profits when non- cooperative behaviour is Bertrand are : . 0 50
0 53
(b) For Cournot : . Bertrand competition is more likely to sustain the collusive agree-
ment since under Bertrand competition punishment is harsher and more than sufficient
to make up for the increased incentive to deviate.
4. For this specification of demand and costs:
i = (A 8 c) ;
2
ic = (A 9 c) ;
2
ir = 9(A64 c) :
and
2
Using these values and (10.7), the critical value of the discount factor required to sustain a
collusive agreement specifying an equal sharing of monopoly profits through grim punishment
strategies is = 0 53
: .
5. (a) As a function of q :
i = (a 2q )q ;

2
c = A ;
i 9
and
ir = (A 4 q ) ;
2
since qr = (1=2)(A q ): Use (10.7) and solve for q as a function of :
qi = A3 (5( 9)9) :

This is the minimum industry output that is sustainable assuming the two firms agree to
produce the same amount.
(b) As increases, qi is reduced indicating that with more patient firms, the greater the
0
increase in market power from colluding. To show that dqi =d < use calculus to
dqi = 36A < 0
show that
d 3( 9)2
or graph qi as a function of :
6. (a) In this case as a function of n:
i = (A 4nc) ;
2
ic = ((An + 1)
c)2 ;
2
Chapter 10. Dynamic Models of Oligopoly 45
and
ir = (n + 1)16(nA2 c) :
2 2
Using these values and (10.7), the critical value of the discount factor is
= n2(n++6n1)+ 1 :
2
(b) As the number of firms increases, the discount rate required to maintain collusion in-
creases. The reason that the discount rate is increasing in the number of firms is that the
incentive to deviate increases as a firm’s share of monopoly profits decreases.
7. From Problem 8.7:

i = 48
, iB = 45
; and ir = 51:2. Using (10.7), = 0:516.
8. (a) The best-response function for firm i is
pi = a +2bcpj :
The Bertrand equilibrium price is
pBi = pBj = pB = 2b a c :
The symmetric prices that maximize industry profits are
pi = pj = p = 2(b a c) :

The optimal deviation is
pr = a4(2 b c)
b(b c) :
Using these, derive the following:
2
iB = jB = (2ba bc)2 ;
2
i = 4(ba c) ;
and
ir = (2b c)2a2 :
16b(b c)2
Letting r = c=b:
2
iB = jB = b(2 a r)2 ;
2
i = 4b(1a r) ;
and
(2 r)2a2 :
ir = 16b(1 r)2
Using (10.7) the critical value of the discount factor that supports joint-profit maximiza-
tion with grim punishment strategies is:
= r2 (2 r)2 :
(2 r)4 16(1 r )2
(b) As the degree of product differentiation decreases (r increases) increases—to show
this take the derivative of with respect to r or graph as a function of r. If r =1
, then
the two goods are perfect substitutes. If r =0
then each firm is a monopolist over its
own product.
9. (a) Because B = 0. A firm can always earn at least profits of zero by exiting.
(b) The critical value of the discount factor is:
r
r i = :
i
i
and r =M since by undercutting slightly a firm can capture the entire market and
monopoly profits.
(c) In the symmetric case i = M =N . Substitute into the expression for in (b) and
rearrange.
(d) It disappears since the profits from cheating change one for one with collusive profits,
unlike the Cournot case where profits from cheating fall faster than collusive profits
when the extent of collusion is reduced.
(e) N (1 ) 1 3 :
(f) 1 > N (1 ) + t+2:
10. We must show that the following two conditions
Sustainability:
r (q ) (q ) = [(q ) (qp )]:
Credibility:
r (q p ) (qp ) = [(q ) (qp )]:
must hold for the following strategies to be SPNE:
At t play q if all firms played q at t 1.

If a firm deviated from q at t 1, play qp at t.
If all firms played qp at t 1, play q at t.
If a firm deviated from qp at t 1, play qp at t.
Chapter 10. Dynamic Models of Oligopoly 47
Sustainability:
Following the equilibrium strategy the payoff for firm i is:
V = (q ) + (q ) + 2(q ) + :::

The optimal deviation followed by punishment yields the following payoff for firm i:
V r = r (q) + (qp ) + 2(q ) + :::

Subtracting V r from V and rearranging gives (10.10).
Credibility:
Following the equilibrium strategy the payoff for firm i is:
V p = (qp ) + (q ) + 2 (q ) + :::

The optimal deviation followed by punishment yields the following payoff for firm i:
V rp = r (qp ) + (qp ) + 2 (q ) + :::

Subtracting V rp from V p and rearranging gives (10.11).
Chapter 11
Product Differentiation
1. For an equilibrium in location two conditions must be satisfied:

No firm’s market length is less than any other firm’s half-market length.
The two peripheral firms must be paired. The peripheral firms must be adjacent to an
interior firm.
When there are three firms it is not possible for both of the peripheral firms to be paired and
the full market length of the middle firm not to be less than the half-market lengths of the
peripheral firms.
2. There is a unique equilibrium where firms 1 and 2 are paired at 1/6, firm 3 is located at 3/6,
and firms 4 and 5 are paired at 5/6. The equilibrium market lengths of firms 1, 2, 4, and 5 is
1/6, while the equilibrium market length of firm 3 is 2/6.
3. (a) Firms 1 and 2 are paired at 7/48, firm 3 is at 21/48, firm 4 locates at 27/48, and firms 5
and 6 are paired at 41/48.
(b) Firms 1 and 2 are paired at 1/8, firm 3 is at 3/8, firm 4 locates at 5/8, and firms 5 and 6
are paired at 7/8.
(c) Firms 1 and 2 are paired at 1/6, firms 3 and 4 are paired at 3/6, and firms 5 and 6 are
paired at 5/6.
4. There are two extremes when there are six firms:
(i) the distance between firms 3 and 4 is minimized and equal to 0. Firms 1 and 2 are
paired at 1/6, firms 3 and 4 are paired at 3/6, and firms 5 and 6 are paired at 5/6.
(ii) the distance between firms 3 and 4 is maximized and equal to 1/4. Firms 1 and 2 are
paired at 1/8, firm 3 is at 3/8, firm 4 locates at 5/8, and firms 5 and 6 are paired at 7/8.
When firms 3 and 4 are paired all firms have the same market length. Call this length l. Then
2
the two peripheral firms have a total market of l. The distance between a paired interior firm
2
and firm 3 or 4 must be l so that it (and firm 3 or 4, whichever is closest) has market length
4 2 +4 = 1
of l. In total this is l. Solving l =1 6
l for l gives l = , the market length of each firm.
The upper bound is found by observing that the distance between firms 3 and 4 cannot exceed
2l or a firm would relocate between them. Hence the upper bound is found by setting d = 2l
and solving 2l + 4l + 2l = 1 or l = 1=8.
49
There is an equilibrium for any d such that 0 d 1=4 where d is the distance between firms
3 and 4. We can use the formula l 2 + 4 + d = 1 to solve for l given any 0 d 1 = 4.
l
5. Both broadcast hockey and the outcome is inefficient, since broadcasting both the hockey
game and the ballet would result in a perfect match between programming and the preferences
of viewers.
6. (a) The Nash equilibrium for this location game is the set of locations such that firms do not
have an incentive to relocate given the location of their rivals.
(b) The two drinking establishments are both located half-way between the two towns.
(c) In this case we would expect the drinks to taste the same. This also means that the market
in this case does not produce any variety: product differentiation is minimized. Of course
this result depends on government regulation that eliminates price competition, so it is
not just the market that is responsible for the lack of variety.
(d) No, the Nash equilibrium is not the socially optimal one. In the Nash equilibrium 2/3
of the population has to travel half the distance between the two towns. The socially
optimal location would minimize aggregate travel costs by locating one bar in each town,
in which case 2/3 of the population incurs 0 travel costs.
(e) The Nash equilibrium set of locations has firms 1 and 2 paired at 1/6 and firms 3 and 4
paired at 5/6. The market share for each firm is 1/4.
7. (a) N = 4, located at 1/8, 3/8, 5/8, and 7/8.
(b) N = 8, located at 1/16, 3/16, 5/16, 7/16, 9/16, 11/16, 13/16, and 15/16.
(c) N = 2, located at 1/4 and 3/4.
(d) The critical value for k is 8 for part (a), 16 for part (b), and 4 for part (c).
(
8. Since prices are fixed such that p c )=1 and the number of consumers has been normalized
to 1, density is 1 and profits equal market length. There are multiple equilibria: there can be
either 4, 5, or 6 firms in equilibrium. With four firms the equilibrium locations are for firms
1 and 2 to be paired at 1/4 and firms 3 and 4 paired at 3/4. The profits for each firm are 1/12.
If a fifth firm entered, the locations of the firms would be 1 and 2 paired at 1/6, firm 3 at 3/6,
and firms 5 and 6 paired at 5/6. With entry of five firms, the profits of firms 1, 2, 4 and 5 are
zero, but the middle firm (3) earns profits of 1/6. If six firms were to enter, then the symmetric
equilibrium locations would be 1 and 2 paired at 1/6, 3 and 4 paired at 3/6, and 5 and 6 paired
at 5/6. In this case the profits of all firms are zero.
For an equilibrium number of firms there are also multiple equilibria. A firm’s number does
not imply an entry order, just a name. For example in the case of 6 firms, another equilibrium
is for firms 5 and 6 to be paired at 1/6, 1 and 2 to be paired at 3/6, and 3 and 4 to be paired at
5/6.
9. The marginal consumer is indifferent between buying from the monopolist (utility equal to
V kt2 pm ) and exercising her outside option p (utility equal to 0). Equating these two
options, the marginal consumer is located at t= ( )
V pm =k. The monopolist’s profits are
p
m = (V pm )=k pm :
Setting dm =dpm
m
=0and solving for pm , the profit-maximizing price is pm V= . =2 3
Substitute p into t and set t equal to 1—its value when the monopolist serves the entire
Chapter 11. Product Differentiation 51
market. Solve for the relationship required between V and k for t to equal one. For equal
3
or greater values (V k) the monopolist will find it profit maximizing to serve the entire
market.
10. A monopolist with a single product would serve the entire market, charging pm (1) = V k
and earning profits of m = V k f . A two product monopolist would set its prices
to make the consumer at 1/2 indifferent between buying either product or its outside option
= = (2) =
(utility of 0). Its profit-maximizing prices are p1 p2 pm 4
V k= and its profits are
m (2) = V k= 4 2 34
f . For = k > f the monopolist would find it profitable to introduce
the second product.
From footnote 33 on page 410 we know that costs when there is a single product equal k= f 3+
and k=12+2 4
f when there are two products. When k= > f the social planner would introduce
the second product. For the indicated range of f , a monopolist would introduce two products,
the social planner only one. The monopolist has an excessive incentive to introduce the second
product since she is not concerned about the savings in transportation (disutility costs) from a
better match between preferences and products. The advantage to her is that she can charge
higher prices and extract more surplus with the introduction of a second product. That is, her
proliferation decision is based both on the savings in transportation costs and her ability to
extract a larger share of total surplus from consumers.
11. Suppose the entrant enters at T and the incumbent has previously expanded, then Bertrand
competition at the right-end point results in marginal cost pricing: pB R =0 . As a result
2
the price of the incumbent’s product at the left-end point will equal k= and the marginal
consumer is at 1/4. The profit of the monopolist if it produces at both end points equals k= . 8
2
If it abandons its product at the right-end point (exits) its profits increase to k= (duopoly
3 8
equilibrium where the market is shared equally). If its costs of exit are less than k= it will
pay to exit. Anticipating this equilibrium, the incumbent will not preempt and the entrant will
enter at T —the socially efficient entry time.
12. (a) The density of consumers around the circle is 200 (=population/distance). There are
100 1 (2 )
=N consumers in a circle segment of length = N miles (density x length).
(b) The socially optimal number of products minimizes the total cost to supply every con-
sumer with one unit. The costs per product are: C N s ( )= + + ( )
F qN s T N s where
( )s
T N are total transportation costs as a function of the number of firms. Cost mini-
mization means that the products will be distributed symmetrically around the circle and
thus q = 200 =N s where N s is the number of products. The transportation costs for a
firm are Z 1=(2N )
s
T (N s ) = 4M t dt;
0
T (N s ) = 100
or
:
( N s )2
C (N s) = F + 200 100
So
N s + (N s )2 :
The total costs for N s firms equals N sC (N s ) or
TC (N s) = N sF + 200 + 100
(N s)2 :
( )
Set dTC N s =dN s =0 and solve for N s : N s=5 .
(c) If F =0 then each consumer would get their own product or N S = 200. That is,
every person would have their own personalized product that matches exactly their most
preferred product. This suggests—since we do not observe it—that it is very unlikely
that F =0 . There is a trade-off between the incentive to increase the number of firms
in order better match consumers and products (and minimize transport costs) and the
incentive to decrease the number of firms in order to minimize set-up costs. The optimal
number trades these two off appropriately.
(d) The N firms are assume to be distributed equally. Firm i is in competition with firm j
1 1
located =N away on one side and firm h located =N away on the other side. On the j
side the consumer indifferent between i and j is defined by:
pi + 2x = pj + ( N1 x)2
where x is the distance from firm i. Solving for x the demand curve for firm i on this side
is
q(pi; pj ) = 50(pj pi + N2 ):
The profits of firm i are
i = (pi 1)(50(pj pi + N2 )) + (pi 1)(50(ph pi + N2 ));

since it makes sales on each side. Maximizing i with respect to pi requires:
@ i = (50(p p + 2 )) + (50(p p + 2 )) 2(p 1) = 0:

@ pi j i N h i N i
In the symmetric Nash pricing equilibrium pj = pi = ph = p, so this reduces to
4 2(p 1) = 0:
N
Solving,
p = ( N2 + 1)
200=N and
and each firm’s sales are
(N ) = 400
N 2 F:
Setting F = 4 and setting (N ) = 0 the free-entry number of firms is 10 and the
equilibrium price is 1.2.
(e) If F is sunk and we assume that zero profits means that a firm does not enter, the subgame
perfect number of firms with sequential entry is N =5 , each 1/5 around the circle, and
the price is p=14 : . Incumbent’s earn profits of 12, but an entrant would not enter, since
1 10
the best it could do would be to capture market share of = and its equilibrium price
would be 1.2 so its profits would be zero.
Alternatively, if a firm with zero profits is assumed to enter, then the equilibrium would
have 6 firms, each 1/6 around the circle, and the equilibrium price is 4/3. The next
entrant would anticipate a market share of 1/12, price of 7/6 and profits of -11/9.
Chapter 11. Product Differentiation 53
(f) Same as (d).

13. (a) The profits of firm i are
= (pi c)xi (11.1)
= (pi c) pI2q
i
Profit-maximization requires firm i to set its price such that
d i = x + (p c) dx i
d pi i i dp = 0;
i
where (for those who care) we have abused notation by using derivative notation instead
of partial derivatives. Making the substitutions for dxi=dpi and xi and solving, pi =2
c.
= =
In the symmetric equilibrium pi pj pm =2 c.
(b) The free-entry number of firms is determined by the zero-profit condition. In equilibrium
= = =2 = = (2 )
pi pj pm c, so q N=p N= c and xi xj x I= cN . So= = = (2 )
(N ) = 2cI
cN f:
Setting (N ) = 0 and solving, N m = I=(2f ).
(c) The equilibrium price in the small group case is
pB = (2(N
N 1)c
1)
and the equilibrium quantity in the small group case is
qB = NcI ((2NN 1)1) :

(d) The free-entry number of firms in the small group case is N B = I=(2f ) + 1=2.
(e) Higher prices and more firms in the small group (Bertrand) case.
14. (a) Set the marginal benefit of another firm ( = 1 (2p )
N ) equal to its marginal cost (f) and
solve for N. The socially optimal number of firms is N S = 1 (4 )
= f2 .
(b) As a function of the number of firms N ( ) = pq f . Since the equilibrium is symmetric
=1 =1
q =N and in equilibrium p =N , so (N ) =p1=N 2 f . Set (N ) = 0 and solve
for N. The free-entry number of firms is N f = 1= f .
(c) There is excess variety when the free-entry number of firms is greater than the socially
1 16
optimal amount, or when f 3 > = . There is insufficient variety when f 3 < = . 1 16
For large values of the fixed cost, the business stealing effect dominates and there is
excessive variety. For small value of the fixed cost, the inability to appropriate surplus
dominates and there is insufficient variety.
15. (a) P C = 0:80 and N C = 1:25.
(b)
CSBG = 1:12(k)1=2 P1 0:32 :
If CSBG > 0 then the buyers group would accept.
(c) PSSG = (P1 0:8k).

(d) W BG+SG = CS BG + PS SG = (1:12k1=2 0:8k 0:32). kmin = 0:15804.
If k > kmin then W BG+SG > 0.
(e) Profits outside of the coalition are (1 )P C =N C f = (:01)(0:80)=1:25 0:64 =
0:6336. The free-entry number of firms is such that the profits of a firm outside of the
coalition are 0. In the long-run the new equilibrium number of firms is N C = 0:125,
which since k = 0:1 means that only firms in the sellers group stay in the market in the
long-run.
(f) PSSG = 0:99P1.
(g) Short Run: W BG+SG = 0:0440.
Long Run: W BG+SG = 0:0352.
(h) TS = 0:0445.
Chapter 12
Identifying and Measuring Market

Power
1. See Table 12.1.

Start with the condition for profit-maximization for firm i:
P MC
= si" :
P
For price taking behaviour P = MC, so = 0. Since 1 + v = , v = 1.
For Cournot = 1 and since 1 + v = , v = 0. Sum over all firms and observe that
in the symmetric equilibrium si = 1=n for all firms to derive the Lerner index for the
market.
For Cartel behavior si i = 1, but si = 1=n, so = n and v = n 1:
2. The variance of market share is defined by
n
X
2 = (si s)2;
i=1
where s is the average market share

n
X si :
s=
i=1 n
Behaviour v Market Lerner Index

Price Taking 0 1 0
Cournot 1 0 1=(n")
Cartel n n 1 1="
Table 12.1: n firms: Relationship between and v
55
Expanding it
n
X
2 = s2i 2s + ns2
i=1
P
since si = 1. Substitute in s = 1=n and simplify to show that
n
s2i n1 :
X
2 =
i=1
=
3. The monopolist maximizes profit by setting MR MC. The perfectly competitive equilibrium
occurs where supply is equal to demand. This means that
QM = QC = 10015+ Y :
On the basis of differential responses to changes in Y it is not possible to identify the degree of
market power. Changes in Y always result in identical changes in output, making it impossible
to determine from the change in industry output if there is a monopolist with marginal costs
5
of Q or a perfectly competitive industry with supply curve equal to Q. 10
4. The initial equilibrium outputs are QM QC= =8
and the initial equilibrium prices are
P M = = 60
P C . When demand increases, the monopoly equilibrium output is QM = 18
and the monopoly equilibrium price is P M = 110
. The new competitive equilibrium output
is QC = 28and the new competitive equilibrium price is P M . = 60
If the market is competitive and unit costs constant then reestablishing P =
MC when de-
mand increases requires that the change in total revenue equal the change in total cost and
thus industry profits remain unchanged. However, if there is a monopoly the change in total
revenue in response to the change in demand will be greater than its change in costs since the
increase in demand will result in an output response that increases industry profits. See Figure
12.1 where demand increases from D to D’.
5. (a) 1967: CR4 = 0:95, HHI = 0.650
1977: CR4 = 0:90, HHI = 0.225
1987: CR4 = 0:95, HHI = 0.275
(b) The HHI, unlike the CR4 , adjusts to reflect the equalization of market shares over time.
(c) High market shares are often thought to be necessary, but not sufficient for the existence
of market power. In this case, the expansion by small firms, and entry and growth of a
sixth firm suggest that market shares are not indicative of market power.
6. In all of these cases the use of market share as an indicator of market power depends on
two things: (i) whether the market is correctly identified to include all products that exerts a
significant competitive constraint and exclude those that do not; (ii) the conditions of entry. If
the market is too narrowly defined—excludes substitutes of significance—then the four-firm
concentration ratio will overstate market power. If the market is too broadly defined—includes
products that are not good substitutes—then the four-firm concentration ratio will understate
market power. High market shares—in correctly defined—markets are probably necessary
for market power, but they are not sufficient. Market power requires high market shares and
protection from entry.
Chapter 12. Identifying and Measuring Market Power 57
PM’
PM=PC MCC
D’
MR’
MCM
D
MR
Q
QM=QC QM’ QC’

(a) Might be indicative, depending on the extent of barriers to entry.

(b) Does not refer to an actual market: a flight from Chicago to Miami is not a very good
substitute for a flight from Seattle to Denver.
(c) Neglects regional nature of the market and substitutes.
(d) Neglects imports, secondary market, and substitutes.
(e) Drugs are usually ailment specific, hence the market definition is too broad.
(f) The market definition is too narrow because it excludes glass (and other types of) con-
tainers.
(g) Neglects the regional nature of the market and other sources of information.
(h) Market definition is too broad since some products are functionally distinct and others
are differentiated.
(i) Neglects regional nature of the market and product differentiation.
(j) Neglects availability of substitutes.
Chapter 13
An Introduction to Strategic
Behavior
1. (a) (Rat, Rat)

(b) (No Swerve, Swerve)
(c) Preferred match of the first mover.
2. Consider the three possible types of subgame in the second period:
Neither firm committed in the first period. In this case the second-period equilibrium is
for both to produce their Cournot output.
Both firms commit in the first period. Then both firms produce in the second period their
committed outputs.
In the first period one firm commits and one does not. The committed firm produces its
commitment, while the second responds optimally by producing according to its best-
response function.
The SPNE is for both to commit to their Cournot outputs in the first stage. Then neither has
an incentive to deviate in the first stage. If a firm does not commit in the first stage, it too has
an incentive to deviate and commit in the first stage. By doing so it can either negate its rival’s
first-mover advantage or create a first-mover advantage.
3. The result is monopolization by the first-mover, provided the marginal cost of the first mover
does not increase too rapidly. If 2 responds to a unit expansion by firm 1 by reducing its output
by more than one unit, then an expansion in output by firm 1 results in price increasing—
since market supply has decreased—and greater market share for firm 1. The equilibrium is
= =0
q1 q1M (which exceeds q1l ) and q2 . See Figure 13.1 for the constant unit cost case. If
the marginal cost of firm 1 rises too rapidly, then the effect is to simply make the first-mover
advantage more lucrative.
4. (a) The game tree is shown in Figure 13.2. Using backward induction the equilibrium is
that Bell L enters and LLR accommodates. The SPNE are In and Accommodate.
(b) The game tree is shown in Figure 13.3 Using backward induction the equilibrium is that
LLR acts aggressively by printing flyers and Bell L stays out. The SPNE are Out and
(Commit, Fight if In).
59
q2
q2=R2(q1)
q1=R1(q2)
q1
q1l q1M
(0, 15)
Out
(5, 5)
In
Accommodate
L
Fight
(−1, −1)
Figure 13.2: No Commitment: LLR vs. Bell L
Chapter 13. An Introduction to Strategic Behavior 61
(0, 15)
Out
(5, 5)
In
Passive Accommodate
L
Fight
L (−1, −1)
(0, 7)
Aggressive
Out
(5, -3)
In
Accommodate
L
Fight
(−1, −1)
Figure 13.3: Commitment: LLR vs. Bell L
(c) Part (b) demonstrates that the opportunity to transform avoidable costs into sunk expen-
ditures provides a first mover with a means to make its threats credible. The investment
in the flyers ex ante is a strategic move that changes the expectations of the entrant re-
garding the optimality to the incumbent of launching a price war. Credible threats severe
enough to make the expected profits of entry negative will result in entry deterrence.
5. The Stackelberg equilibrium is:
q1S = 9; q2S = 4:5; P S = 29; 1S = 81; 2S = 40:5:

The Cournot equilibrium is:
q1C = 6; q2C = 6; P S = 32; 1S = 72; 2S = 72:

The Stackelberg equilibrium has firm 1 producing more and earning greater profits; firm 2
producing less and earning lower profits. Aggregate output increases and price falls.
6. (a) The limit outputs are:
If f = 50, then q1l = 8 and 1 = 1M = 162 f .

If f = 32, then q1l = 10 and 1l = 160 : f
If f = 18, then q1l = 12 and 1l = 144 : f
If f = 2, then q1l = 16 and 1l = 64 :f
(b) In the Stackelberg equilibrium, firm 1 earns profits of 81- f. The SPNE are:
If f = 50
, then q1M =9 =0
and q2 —there is a blockaded monopoly.
If f = 32 = = 10
, then q1 q1l =0
and q2 and firm 1 produces the limit output
If f = 18 = = 12
, then q1 q1 l =0
and q2 and firm 1 produces the limit output.
If f =2, then q1 = =9
q1S =
and q2 q2S = 45
: and the firms produce at the
Stackelberg equilibrium.
7. (a) The Cournot equilibrium quantities are
qiC = (1=3)(30 2ci + cj ):

See the solution to Problem 8.3 for the derivation.
(b) Firm 2’s best-response function is
q2 = 30 q21 c2 :
Firm 1’s profits are
1 = (30 q1 q2)q1 c1q1 :
Substituting in 2’s best-response function for q2 this becomes
= (30 q1 + c2 ) q c q ;
1
2 1 1 1
Chapter 13. An Introduction to Strategic Behavior 63
(30 q1 + c2)
where
2
is Firm 1’s firm level demand curve. Setting firm 1’s marginal revenue
30 + c2 q
2 1
—double the slope of its demand curve—equal to its marginal cost ( c1) and solve for q1:
q1S = (1=2)(30 2c1 + c2)

and so
q2S = (1=4)(30 3c2 + 2c1):
(c) TS C = 277:78 and TS S = 292:97. Since the firms are equally efficient, the increase
in production by firm 1 increases total surplus. Thus the Stackelberg equilibria is more
efficient than the Cournot equilibria.
(d) TS C = 249:94 and TS S = 253:97. Even though firm 1 has higher costs, the increase
in quantity produced offsets this inefficiency and the Stackelberg equilibria is more effi-
cient.
(e) TS C = 237:50, and TS S = 236:72. In this case, the Cournot equilibria is more
efficient than the Stackelberg equilibria because even though output expands, too much
production is shifted to the high cost firm.
(f) There is a trade-off between allocative and cost efficiency. Output expansion and im-
proved allocative efficiency arises from the first-mover advantage of Stackelberg, but
because firm 1 is inefficient, it also results in cost inefficiency.
8. (a) The Cournot equilibrium firm outputs are q1C = q2C = 89:6.
(b) The Stackelberg equilibrium firm outputs are q1S = 96 and q2S = 88.
Chapter 14
Entry Deterrence
1. (a) For q < 4, MC = 2 and for q > 4 MC = 20. Firm 1’s marginal revenue function is
MR = 56 2q2 4q1: See Figure 14.1.
(b) R21 = (1=2)(27 q2),
1 = (1=2)(18 q2),
R20
R2 = (1=2)(18 q1).
See Figure 14.2.
(c) The equilibrium quantities are q1 =8 and q2 =5. Capacity less than 6 will result in
( ( )) 20
MR1 k1; R2 k1 > so firm 1 will want to expand its capacity. Capacity less than
( ( )) 2
12 will result in MR1 k1; R2 k1 < and it is not profit maximizing to produce to
capacity. Firm 1 will have excess capacity that it had to pay for, but does not use.
(d) Firm 2’s profits at T are72 f and its profits at V are 18 f .
If f = 75 then there is a blockaded monopoly so k1 = q1 = q1M = 9 and q2 = 0.
If f = 50 then q1l = 8 < q1M , so this is another case of blockaded monopoly and
k1 = q1 = q1M = 9 and q2 = 0.
If f = 32 then the limit output is 9 and it is credible to produce if k1 = 9. Firm
1’s profits from strategically deterring entry are 160, which exceed the 81 from
strategically accommodating. So k1 = q1 = q1l = 10 and q2 = 0.
If f = 15, the profits for firm 2 at V are positive and the only option for firm 1 is to
strategically accommodate entry. So k1S = q1S = 9 and q2S = 4:5.
(e) In this case, the SPNE would be at T where q1 = 6 and q2 = 6. The reason is that firm
1’s marginal cost is now 20 (since it incurs an opportunity cost of 18 if it uses a unit of
capacity) and its threat to produce output beyond 6 where its marginal revenue is less
than 20 is non-credible even if it has invested in a greater capacity level.
2. Marginal revenue must be increasing in rival’s output near the monopoly output, so that firm
1’s best-response is increasing in q2: See Figure 14.3 where B is the post-entry equilibrium,
k1 is sufficient to deter entry, but when it is deterred firm 1 produces q1M < k1 and thus has
excess capacity.
3. (qS1V ; q2V ) 0,Sthen k1 = q1M and Vq2 = 0. If 2VV (q1V ; q2V ) > 0, then if q1S q1V ,
(a) If 2V
k1 = q1 and q2 = q2 ; otherwise k1 = q1 and q2 = q2 .
65
20
MR(23)
MR(6)
MR(15)
2
q1
4
Chapter 14. Entry Deterrence 67
q2
q1=R12(q2)
q1=R120(q2)
V
q2=R220(q1)
q1
8
q2
q1=R1(q2)
q2=R2(q1)
q1
M
q1 k1
Chapter 14. Entry Deterrence 69
(b) As k1 =q1l > q1M , the incumbent would never expand its output beyond k1, since its
marginal revenue is less than w r. +
4. The key is that decreases in the incumbent’s marginal cost shift its price best-response func-
tion down and to the right—for every price of its rival, the incumbent will have an incentive to
charge a lower price. As a result, equilibrium prices will also fall. The incumbent still has an
incentive to overinvest if its objective is to deter entry, since low prices have negative impli-
cations for the profits of the entrant. However, if the incumbent’s objective is to strategically
accommodate it should underinvest since this will lead to higher prices, good for both it and
the entrant.
5. (a) For q1 k1 , MC1 =6
. For q1 > k1 , MC1 = 12
. MR1 = 60 4q2 8q1.
(b) For q1 k1, the incumbent’s best-response function is q1 = (1=4)(27 2q2). For
q1 > k1, the incumbent’s best- response function is q1 == (1 2)(12 q2). The entrant’s
best- response function is q2 = (1 2)(12
= q1 . )
(c) When k1 = 5, q = 5 and q = 3:5.
1 2
For any k1, the Nash equilibria are:
4 = =4
(i) for k1 < : then q1 q2
5 =5
(ii) for k1 > : then q1 =35
and q2 :
4 5 =
(iii) for k1 : then q1 k1 and q2
= (1=2)(12 k1).
(d) (i) f = 64: k1 = q1 = 6 and q2 = 0
M
(ii) f = 25: k1 = q1V = 5 and q2 = q2V = 3:5
(e) MR1 (4; 4) = 12 > MC1 = 6 if k1 > 4: The investment in capacity converts variable
C
costs into sunk expenditures, making the threat to produce q1 > q1 credible. It is
important that the costs of capacity are sunk, otherwise firm 1 would sell its capacity
(4 4)
and the equilibrium would occur at the simple Cournot outcome ; .
(f) No, since R2(q1 = 7) = 2:5 and MR1 (7; 2:5) = 6 < MC1 = 6:
6. (a) The Cournot free-entry equilibrium number of firms is N c = 3:5. The Cournot free-
entry total surplus is 24.5. The monopoly total surplus is 26.375. In this case, the
monopoly is more efficient because of the duplication of the large fixed cost.
(b) The Cournot free-entry equilibrium number of firms is N c =8 . The Cournot free-entry
total surplus is 32. The monopoly total surplus is 29.375. In this case, the Cournot free-
entry equilibrium is more efficient. The duplication of fixed costs is more than off set by
the benefits of a lower price due to increased competition.
Chapter 15
Strategic Behavior: Principles
1. The second stage competition is between strategic substitutes because the best-response func-
tions are downward sloping.
2. The second-stage equilibrium values are:
x1 = 15 9
2 + 8 k1
x2 = 15
2
3 k1:
8
3. With k1 =2 1
, firm 2’s profits are before paying the fixed costs of entry, but 1 after paying
the entry costs, so entry will be deterred. Profits for firm 1 in this case are monopoly profits
2 8 25
with k1 constrained at its deterrence value of , equal to : . If k1 =0 , there will be a
0 78
symmetric duopoly post-entry, with profits for firm 2 of : and for firm 1 of : . Thus,2 78
the deterrence equilibrium with k1 =2 is the preferred choice for firm 1. This is a Top Dog
equilibrium.
4. (a) The two best-response functions are:
p1 = 15
2
1 1
2 k1 + 4 p2
p2 = 15 +
2 4
1 p1:
(b) The second period equilibrium prices are:
8 k1
p1 = 10 15
2 k1:
p2 = 10 15
(c) This is a Puppy Dog game because the goods are strategic compliments and investment
makes firm 1 tough, so firm 1 will underinvest.
5. (a) The monopoly output in period A is qA = 52 .

71
(b) In the dynamic two period case, the monopoly outputs are qA qB = =5
. Output
increases in period A in order to lower costs in period B. Output increases in period B
because costs are lower. With the linear model, these two effects are equal.
(c) Firm 1’s profits over both periods can be written in the general form:
1(q1A ) = 1A (q1A ) + 1B (q1A ; q1B (q1A ); q2B (q1A )):
so that the first order condition can be written as:
d1 = @1A + @1B + @1B @q1B + @1B @q2B :
dq1A @q1A @q1A @q1B @q1A @q2B @q1A
The first two terms are the same as for the monopoly learning firm. The third term is zero
by the envelope theorem. The last term is the strategic term, and is positive, implying that
the first period output will be higher in the duopoly case, to take advantage of strategic
learning. Given that the goods are strategic substitutes and that investment makes you
tough, this is a Top Dog model.
Chapter 16
Strategic Behavior: Applications
1. m
(a) q11 = 3 and 11m = 9
m
(b) Assume that firm 1 sets q11 =3
in period 1. Then second period Cournot quantities are
C =3 C =3 2
q12 and q22 = : Total undiscounted profits for firm 1 equal 18.
(c) If firm 1 increases period 1 output by one unit, so that q11 =4 , then period 1 profits fall
to 11m =8 . The new Cournot quantities for period 2 are q12 C = 10 3
= and q22C =4 3= .
Period 2 profits increase to 100/9 and thus total profits increase to 19.1.
(d) This is a Top Dog game. By expanding output in period 1 beyond the short term profit
maximizing level, firm 1 gains a strategic advantage in period 2. Investment in first
period output makes firm 1 “tough”, and the second period game is one of strategic
substitutes.
2. Think of the output corresponding to setting marginal revenue equal to first period marginal
cost as the benchmark. In the first period the monopolist will product more than this because
of the effect of increased learning on second period costs, and hence second period profits. In
period two the monopolist will produce more than the benchmark because its costs are now
lower due to first period learning. In general it is possible that output could be declining in
successive periods or increasing.
A simplified algebraic model gives some more insight into this issue. Suppose that inverse
( )
demand is P Q in each period, and that constant marginal costs are c (0) ( )
and c q12 in the
() 0
two periods respectively, where c0 < : Then the monopolist’s undiscounted profits are
given by:
= ( ( ) (0)) + ( ( ) ( ))
P q11 c q11 P q12 c q11 q12 (16.1)
The first order conditions for first and second period quantities are, respectively:
P + P 0q11 c(0) c0 (q11)q12 = 0 (16.2)

P + P 0q12 c(q11) = 0 (16.3)
(0) ( )
so that c ( )
c0 q11 q12 > c q11 implies that q11 < q12 . The expression can be rearranged to
() 0
show that if c00 > (costs decrease with learning but at a decreasing rate), then q11 < q12;
i.e. output is increasing and prices declining .
3. If the spillovers are 100% then the strategic motive for overinvesting in learning will disappear.
73
=
4. Two extreme cases that satisfy this property are C F (all costs are fixed), where the progress
ratio= 100% =
; and C cQ ( Q is cumulative total output, so that all costs are variable with
constant unit costs), with the progress ratio=0 =
: More generally, if costs are given by C
0 1
Q1 where , then the progress ratio is constant and equal to (2 1) 100%

with the two previous cases arising when =1 and =0 respectively.
5. With symmetric Cournot, the equilibrium quantities are q1 = =3
q2 , and the equilibrium
= =9
profits are 1 2 . If firm 1 instructs their manager to maximize sales revenue, the new
equilibrium quantities are q1= 11 3 =8 3
= , q1 = , and firm 1’s profits increase to 1 = 88 9 = .
This is a Top Dog equilibrium because investment makes firm 1 “tough”, and the game is one
of strategic substitutes.
Chapter 17
Advertising and Oligopoly
1. Using post-advertising tastes to measure welfare changes, the change in welfare is given by the
dark shaded area in the figure. The change in profits is the same as given in the text. Since the

difference between the two areas is still x p the equation given in the text W = x p
is still a valid approximation for the welfare change associated with a small increment of
advertising. The reason is that, for small changes, pre- and post- advertising tastes differ by
an amount which is of “second order of smallness”.
2. The equilibrium number of firms are nc = 15; 13; 11 respectively.

3. (a) Advertising is predatory because increments in advertising by firm 1 shift firm 2’s de-
mand curve to the left.
(b) p1 = 4:43 and p2 = 1:74
(c) p1 = 5:83 and p2 = 1:30
4. Yes, advertising can still be used strategically to deter entry. If investment in advertising
makes the incumbent firm “soft” then the entry deterring strategy would involve underinvest-
ment, and would be a lean and hungry look. If there are multiple incumbents then because
of the standard free rider problem associated with public goods, the incumbents will collec-
tively advertise too little from the perspective of their joint profits. If entry deterrence is the
objective of multiple incumbents, then the externality will imply too little investment in entry
deterrence, which implies an inefficiently high level of advertising in this model.
(a) The period 2 game is one of strategic complements, since the price best-response func-
tions have a positive slope.
(b) In a model of linear demand such as the one presented, investment in advertising makes
firm 1 soft because it increases firm 2’s profits in the new equilibrium. Firm 1 will set
a higher price as a result of advertising, and given that the two products are substitutes,
higher prices for product 1 increase profits for product 2.
(c) This is a Fat Cat game which implies strategic overinvestment in advertising.
(d) A Lean and Hungry Look game involving underinvestment in advertising.
0
(e) When w > direct strategic effects are present. These are defined as strategic actions by
firm 1 that directly impact firm 2’s profits i.e. not through the two firms’ choices of price.
75
Price D'
D
p'
D'
Marginal cost
q q' Quantity
Figure 17.1: Problem 1
p2
p1 = 4+ p2/4
Indirect Strategic Effect
of an increase in A to 20
A = 10
equilibrium p2 = 1+ p1/6
Direct Strategic Effect

of an increase in A to 20
1
A = 20
equilibrium
4 p1
Figure 17.2: Problem 3b

Chapter 17. Advertising and Oligopoly 77
p1 = R1(p2, A)
p2
Effect of an increase in A
on firm 1's reaction function
Pre-advertising
equilibrium p2 = R2(p1, A)
Effect of an increase in A
on firm 2's reaction function
Post-advertising
equilibrium
p1
Figure 17.3: Problem 5(f)
In this case, firm 1’s advertising expenditure directly reduces firm 2’s profits. In a best-
response function diagram the direct effects appear as shifts in firm 2’s best-response
function.
(f) Figure 17.4 from the text, reproduced here, describes exactly the analysis of this model.
Chapter 18
Research and Development
1. The diagram shows surplus changes with a drastic innovation. With a drastic innovation
Vs is defined exactly as with a non-drastic innovation - the area between ch and cl and the
demand curve. Areas A and B in the diagram represent monopoly profits, before and after the
innovation, respectively. From the diagram it is clear that
Vc = area A < Vs (18.1)
Vm = area A area B < Vc (18.2)
2. (a) With a non-drastic innovation the price remains constant after innovation, so that S c =
0 : With a drastic innovation, the price falls from ch to the monopoly price corresponding
( ) ( + )2
to costs cl ; pm cl which is a cl = . The gain in consumer surplus from this change
is

S c = 0:5 ch pm (cl ) q(ch ) + qm (cl )
l a c h a cl
= 0:5 c a + c
b + 2b
h
2
which is greater than 0 provided that we have a drastic innovation.
(b) Under monopoly, the price falls from one monopoly price to another whether or not the
innovation is drastic. Hence, the change in surplus can be computed as:

Sm = 0:5 pm (ch ) pm (cl ) qm (ch ) + qm (cl )
h
c c l 2a ch cl
= 0:5 2 2b > 0:
(c) To compute W; we first need to compute V , the private value of the innovation. Under
competition, and a non-drastic innovation, we have
V c (non-drastic) = (c cl )(a ch )
h
(18.3)
b
79
Price
A
ch
B
cl
Quantity
(a + cl )=2 so that
With a drastic innovation, the price falls to
= (a +2 c ) cl a 2bc
l l
V c (drastic)
= (a 4bc )
l2
(d) Under monopoly there is no distinction between drastic and non-drastic, and
V m = (a 4bc ) (a ch )2 :
l2
4b (18.4)
We can now compute the three different expressions for S:

(e) Under competition with a non-drastic innovation
W c (non-drastic) = S c + V c = (c cl )(a ch ) :
h
b (18.5)
(f) Under competition with a drastic innovation
W c (drastic) = S c + V c = 1 ch a + cl a ch + a cl + (a cl )2
2 2 b 2b 4b (18.6)
(g) Under monopoly

h l 2a ch cl (a cl )2 (a ch )2
Wm = Sm + V m = 1 c c + 4b
2 2 2b 4b
= 23 c 2 c 2a 2cb c
h l h l
Chapter 18. Research and Development 81
3. (a) Before the innovation, there are n identical firms.

Using the best-response function for each firm and setting qi = q; 8i, each firm produces
h
q = ba(n +c1) ; (18.7)
Hence, the market price is

p = an++nc1 :
h
(18.8)
So, the profit of each firm before the innovation is
= (p ch )q = b((an +c 1))2 :
h2
(18.9)
Non-drastic innovation:
In this case, the innovating firm (denote by l) competes with the other (n 1) non-innovation
firms (denote by n).
Solving the best-response functions
a cl b(n 1)qn = 2bql (18.10)

a ch b(n 1)qn bql = 2bqn (18.11)
yield
qn = a +b(nc + 1)2c
l h
(18.12)
ql = a + (nb(n 1) ch ncl
+ 1) (18.13)
The market price is then given by
pl = a b (n 1)q + ql = a + (nn +1)1c + c :

h l
(18.14)
So, the profit of the innovating firm after innovation is

a + (n 1)ch ncl 2
l = (pl cl )ql = b(n + 1)2 : (18.15)
The private value of innovation to this firm is thus
V = l
a + (n 1)ch ncl 2 (a ch )2
= b(n + 1)2

a ch + n(ch cl ) 2 (a ch )2
= b(n + 1)2 >0
Drastic innovation:
The innovating firm earns monopoly profits after innovating, hence
V = m (cl )
= (a 4bc ) b(a(n +c1))2 > 0:
l2 h2
4. Demand and costs are given by
P = 1 Q; ch = 0:4; cl = 0:2: (18.16)
Monopoly price and quantity:
pm (c) = 1 +2 c ; qm (c) = 1 2 c : (18.17)
Duopoly price and quantities,
pd = 1 + c3 + c ; qd (ch ) = 1 + c 3 2c ; qd (cl ) = 1 + c 3 2c :
h l l h h l
(18.18)
Let the total duopoly quantity produced be
d = qd (ch ) + qd (cl ): (18.19)
If the incumbent innovates,

The gain in consumer surplus would be
Sm = 21 [pm (ch ) pm (cl )][qm (ch ) + qm (cl )]

= 0:035:
If the entrant innovates,
Sd = 12 [pm (ch ) pd ][qm (ch ) + qd ] = 0:065: (18.20)
Hence, Sd > Sm .
To calculate total welfare, we need to compute the private gain from innovating.
If the incumbent innovates,
Vm = m (cl ) m (ch )
= (1 4c ) (1 4c ) = 0:07:
h2 l2
Chapter 18. Research and Development 83
Therefore, the total surplus gain is
Wm = 0:105: (18.21)
If the entrant innovates, then the total private gain is the difference between total duopoly
profits minus the incumbent’s monopoly profit before the innovation.
Vd = Vdi + Vde
= pdqd cl qd(cl ) ch qd(ch ) (1 4c )
h2
= 0:04
where i denotes the incumbent and e denotes the entrant.
Therefore,
Wd = 0:105: (18.22)
Which is (approximately) the same! The reason is that although innovation by the entrant
leads to a lower price, it also implies less output being produced at the low cost cl : Since total
surplus is the same whether the entrant innovates or the monopolist does, the government has
no incentive to intervene. It is worth noting, however, that the entrant’s profit incentive is
( ) = 0 11
d cl : which is higher than that of the incumbent - so the likely outcome is innovation
by the entrant, and duopoly.
5. The present value of an income stream of $200,000 over 20 years, discounted at 10%, is $1.87
million. In a patent race, each firm would be willing to spend up to this amount, multiplied by
the probability of winning the race.
6. The first order condition for the firm’s problem is
1 rT R c0 e R
2r 1 e 1 c0e 1=0 (18.23)
which defines implicitly the firms optimal choice of research expenditure R (T ). The regula-
tor’s problem is to choose T to maximize
1 1 c0e
R (T ) 2 3+e rT :
8r (18.24)
Chapter 19
The Theory of the Market
1. (a) The first term accounts for the increase in expenditure necessary to purchase the initial
amount of q1. The second term accounts for the fact that the amount demanded for good
1, and hence expenditure, will change as the price changes. The third term accounts for
the change in the demand for good 2, and hence expenditure, when the price of good 1
changes.
(b) If the budget constraint cannot be violated then the change in expenditure from a change
in p1 equals zero.
+ =
(c) Divide through by q1 and use p1q1 p2 q2 M to derive "11 = 1 + "21s2=s1:
(d) In order for good 1 to be inelastic, it must be the case that "21< 0—the two goods are
complements.
2. Using (19.1)
m
"ii (pm ) = pmp c :
Multiply top and bottom of the right-hand side by p0 and substitute in p0 = c to derive
m 0
"ii (pm ) = (pmp pp0)p0 :
Since t = (pm p0 )=p0 and 1 + t = pm =p0,
"ii = "ii (pm ) = 1 + 1=t
as required.
3. (a) Figure 19.1 shows the gain and loss in profits from an increase in price equal to dp. The
= +
gain is area g and the loss is area l. Area g equals dpq1, but q1 q dq where dq < . 0
So
g = dp[q + dq] (19.1)
= qdp[1 + dq=q]
= qpdp=p[1 + (dq=q)(dp=dp)(p=p)]
= qpt[1 "t]
85
dp g
l
P=P(Q)
MC = c
q1 q

Chapter 19. The Theory of the Market 87
as required.
(b) By definition the gross decrease in profits from the increase in price equals the per unit
margin multiplied by price times the decrease in quantity:
l = mpdq (19.2)
= mpdq(q=q)(dp=dp)(p=p)
= mt"pq
as required.
(c) In order for the price increase to be profitable g > l or 1=" t > m.
4. (a) A price increase is profitable if
1="ii > L:
At prevailing prices L = 0.80 which is less than 1=" since " = 1:125.
(b) No:
1="ii = 0:89 < L = 0:98:
(c) For a constant-elasticity demand curve the critical elasticity of demand is defined as
"ii = m1 ++tt :
If c =1then "ii = 1 23 and a profit-maximizing monopolist of Classical would increase
:
price by 5%.
(d) If c =01: then "ii = 1:02 and a profit-maximizing monopolist of Classical would not
increase price by 5%.
(e) If c =4 then "ii = 4:2 and a profit-maximizing monopolist of Classical would increase
prices by 5%.
(f) The price-cost margin determines the profit implications of the reduction in demand. A
large margin means the reduction in demand is very costly, while a small margin means
the reduction in demand is not very costly.
5. (a) pC= 40 and qC = 40.
(b) pM = 50 and qM = 30. The price increase is 25 %.
(c)
" = dq p
dp q
If p = a bq, then q = (a p)=b and dq=dp = b. Substituting in
"= a p p
as required.
(d) "ii = 1 667 > "ii = 1, so this is an antitrust market.
:
(e) (i) p C= 70 and qC =M10.
(ii) pM = 72:5 and q = 7:5, so the price increase is 3.6%.
(iii) "ii= 4:915 < "ii = 7:00 and this is not an antitrust market.
Chapter 20
Exclusionary Strategies I: Raising

Rivals’ Costs
1. (a) The equilibrium prices are p1 =

p2 = 13
, and the equilibrium profits are 1 = 2 =
100: The best-response functions are illustrated in the figure.
(b) The new equilibrium prices are p1 = 14
, p2 = 15
; and the new equilibrium profits are
1 = 121 ; 2= 81 : The new best-response function for firm 2 is shown as the colored
line in the figure.
(c) The equilibrium prices are p1 = 14 7
: , p2 = 15:3; and the equilibrium profits are 1 =
113 8
: ; 2 = 87 1
: : Firm 1 is still better off.
(d) Since the rival’s costs are raised to c2 =6 , by examination of the equilibrium profit
6
functions, it is easy to see that firm 1’s profits will be higher provided that c1 .
2. (a) The dominant firm will produce at capacity qd =5
and the price is set by the fringe
marginal cost, so P =3
: Dominant firm profits are thus d = (3 2) 5 = 5
:
(b) The price will increase by $1 but the dominant firm output will remain unchanged.
Hence d =1 5=5 :
(c) The net effect on profits is (increase in fringe marginal costs minus increase in dominant
firm average costs) dominant firm output. Or (1 0 75) 5 = 1 25
: : : Raising rivals’
costs is still a profitable strategy.
3. (a) The price will still be given by the fringe marginal cost =3
. The dominant firm’s profits
are given by the area between the dominant firm marginal cost curve and the price, or
d = 0:5 3 3 = 4:5:
(b) The new price will increase by $1. Dominant firm output will rise to qd = 4 and new
profits are d =05 4 4=8
: : Hence the change in profits is d : = 3 5:
(c) part (c) of question 2 (first printing only) is not applicable.
4. Yes, it is generally in the interest of D2 and U2 to retaliate with a vertical merger. D2 will
have to pay more to acquire U2 than did D1 to acquire U1 . The upstream firms will have an
incentive to hold out so as to be the last one offered a merger deal.
89
p2 p1 = 6.5 + .5p2
p2 = 8 + .5p1
p2 = 6.5 + .5p1
8
6.5
6.5
p1
Figure 20.1: Problem 1a

Chapter 21
Exclusionary Strategies II:

Predatory Pricing
1. The subgame originating from node 4, with modified payoffs, is shown in the figure. The
equilibrium of this subgame now has the first period entrant indifferent between staying in
in response to predation and exiting (with no discounting). The subgame perfect Nash equi-
librium of the complete game has the Incumbent choosing to Share in both periods with the
first victim (E1) choosing to Stay in the market i.e. the same as the original game without the
altered payoffs.
2. (a) The figure shows the modified extensive form game, where it is assumed that, if the
entrant exits immediately without cost, the predator incurs no costs either. In effect, the
new assumption simply adds a period at the beginning of game tree, so that four “deci-
sion periods” of the price war are required to force the exit of the prey. The condition
required for predation to be the SPNE is still M 200> D .
(b) Figure ? shows the extensive form with discounting. The condition for predation and
exit to be the unique SPNE is now just M > D which is always satisfied.
3. (a) Assuming that that the dominant firm serves the entire market initially, then since P =
74and MC = 3 126 = 378
the price is certainly predatory under the Areeda-Turner
marginal cost criterion.
(b) Assuming that that the dominant firm serves the entire market initially, then since P =
74 and AV C = 1 5 = 189
:q the price is predatory under the Areeda-Turner marginal
cost criterion.
(c) To find prices that are both “not predatory” and imply negative profits for the victim we
need to look at each condition separately.
12
i. Taking the victim’s profits first, if the victim gets = the market, we can write their
profits as e =( P 110)(0 5(200
: )) 100
P . The values of P that satisfy
0
the inequality e are P 113 ?
(ignoring the high pricing strategy of P as
unrealistic, because the victim would undercut).
ii. The prices that are not below the dominant firm’s AVC must satisfy P : 15
0 5(200
: ) 86
P or approximately P > . So the full set of prices that satisfy both
conditions are 86
P . 113
91
Stay Out {-20,-10;0,0;0,0}

ate E2 Enter
Pred
{-20,-20;0,0;0,-10}
Stay Out
DF {-20,60;0,0;0,0} 1
Share E2
E xit {-20,30;0,0;0,10}
4 Enter 3
E1
Sta
yI Exit {-20,-10;-10,0}
n
E1 Stay In
ate
Pred {-20,-10;-10,-10}
DF Exit {-20,60;-10,0}
Share E1
{-20,30;-10,10}
Stay In
(πD , πD)
( πD – 200, πD– 100) (πD – 600, πD– 300)

are
Sh
(πD – 400, πD– 200)
re are are
S ha Sh Sh
Stay Stay Stay
I E I E I E I E
Fight Fight Fight Fight
Exit Exit Exit Exit
(πM , 0) (πM – 200, – 100) (πM – 600, – 300)

M
(π – 400, – 200)
Figure 21.2: Problem 2(a)
(πD , πD)
are ( δπD – 200, δπD– 100) (δ3πD – 200(1+ δ + δ2), δ3πD– 100(1+ δ + δ2))
Sh
(δ2πD – 200(1+ δ), δ2πD– 100(1+ δ))
are are are
Sh Sh Sh
Stay Stay Stay
I E I E I E I E
Fight Fight Fight Fight
Exit Exit Exit Exit
(πM , 0) (δπM – 200, – 100) (δ3πM – 200(1+ δ + δ2), – 100(1+ δ + δ2))

(δ π – 200(1+ δ), – 100(1+ δ))
2 M
Figure 21.3: Problem 2(b)

Chapter 21. Exclusionary Strategies II: Predatory Pricing 93
(d) The incumbent has the lowest value of min(ATC) but only when producing at a small
scale.
4. The entrant’s expected payoff in a one-shot game is = 14 = 1 2+3 4 1 2 = 1 4 0

= = = > .
Let Pr( )
sjf be the posterior probability that the incumbent is “strong” given that “fight” is
observed in the first period. Using Bayes’ Rule, we can obtain
Pr(sjf ) = p + y(1p p) (21.1)
Then the second entrant must be indifferent between entering and staying out, so that
Pr(sjf )E (P ) + (1 Pr(sjf ))E (A) = 0 (21.2)
Combining the two equations, substituting in the parameter values and solving for y we obtain
y = 1=3: For the weak incumbent to randomize in the first period, it must be indifferent
between fighting and accommodating. This requires
I (P ) + xI (A) + (1 x)I (M ) = 2I (A) (21.3)
Substituting for the parameters, we obtain x = 1=2:
5. Solve
e
q2 = A c22 q1 (21.4)
where q1e = 12 q1 + 12 q1,

q1 = A c21 q2 (21.5)
and
(A q1 q2 c1)q1 + 1C (c1; c2) (A q1 q2 c1)q1 + 1M (c1) (21.6)
simultaneously for the required quantities (q1; q1; q2) .

The last equality can be re-written
q21 q21 (A c1 q2)(q 1 q1 ) = 1M (c1 ) 1C (c1; c2) = K: (21.7)
Using the best-response function for the high-cost firm 1,
q12 q21 2q1(q1 q1) = K (21.8)

() (q 1 q1 )2 = K (21.9)
p
() q1 q1 = + K = k (21.10)
Solving,
q1 = A 2c31 + c2 + 76k (21.11)
q1 = A 2c31 + c2 + k6 > qC1 (21.12)
q2 = A 23c2 + c1 k3 < q2C (21.13)

p
where k =+ K: The low-cost firm 1 expands production in order to signal its type. How-
ever, the high-cost firm 1 produces in excess of its perfect-information Cournot quantity even
though it is not signaling.
6. (a) If firm 2 were faced with a high-cost rival, the Cournot quantities are
q2C = A + c13 2c2 (21.14)
qC1 = A + c23 2c1 : (21.15)
Firm 2’s profit would then be
2C (c1; c2) = (A + c19 2c2) :

2
(21.16)
If it were facing a low-cost rival, the Cournot quantities are
q2C = A + c13 2c2 (21.17)
qC1 = A + c23 2c1 : (21.18)
Firm 2’s profit would then be
2C (c1; c2) = (A + c19 2c2) :

2
(21.19)
(
Since c < c, we have 2C c1 ; c2 ) < 2C (c1; c2).
(b) 2e = 1=22C (c1 ; c2) + 1=22C (c1 ; c2) which obviously lies between the two values.
(c) Since the signaling quantity is higher than the Cournot quantity of the low-cost type firm
1, whenever firm 1 deviates, it will prefer to produce its Cournot quantity ( q1C ). When
it does, the high-cost firm 1 will prefer to mimic the low-cost type and hence a pooling
equilibrium prevails.
The incentive compatibility constraint is given by
(A qS1 q2 c1)qS1 2C (c1; c2)+ 1M (c1) (A qC1 q2 c1)qC1 2e + 1M (c1) (21.20)
or,
R = 2e 2C (c1; c2) (qS1 )2 (qC1 )2 (A q2 c1 )(qS1 qC1 ) (21.21)
1 = A q2 c 1 .
Note that the best-response function for the low-cost firm 1 is given by 2qC
Hence, the constraint can be written
R (qS1 qC1 )2 (21.22)
7. (a) Subgame perfection requires that a dishonest officer accept a bribe (if offered).
Payoff to motorist from not offering a bribe: 100
Payoff to motorist from offering a bribe: ( 200) + (1 )( 20) = 20 180
Thus, motorist offers a bribe if
20 180 > 100 (21.23)
80 > 180 (21.24)
< 94 (21.25)
So PBE is where the motorist offers a bribe if < 94 , does not offer a bribe if > 49 ;
= ;
and is indifferent if 49 the police officer accepts a bribe if it is offered.
(b) i. Subgame perfection requires that a dishonest police officer accept a bribe in the 2nd
period.
ii. Since an honest police officer would never accept a bribe, the police officer must be
dishonest. Thus, the second period motorist offers a bribe, knowing that it will be
accepted by the dishonest officer.
iii. Let p2= Pr fhonestjrejectg: Possible first-period strategies for the dishonest officer:
A. Always reject a bribe; then p2 = : By assumption < 94 (since 1st-period
motorist did offer a bribe), so the second period motorist will also offer a bribe.
Payoff to dishonest officer: 0+20 = 20 : Deviating by accepting the first period
bribe would increase payoff to 20 + 20 = 40 ; so this cannot be a PBE strategy.
B. Always accept a bribe; then p2 =1 : Payoff to officer:20 + 20 = 40 : Deviating
by rejecting the first-period bribe would raise payoff to 0 + 100 40 > ; so this
cannot be a PBE strategy.
C. Thus, PBE must be to reject the bribe with probability q: Let r be the proba-
bility that the second motorist offers a bribe after seeing the first bribe rejected.
Expected payoff from rejecting bribe:
0 + r(20) + (1 r)(100) = 100 80r (21.26)
Expected payoff from accepting bribe: 20+20 = 40: Thus, officer is indifferent
between the two strategies if 100 80r = 40 ) r = 34 :
D. For the motorist to be indifferent between offering a bribe and just paying the
=
ticket, it must be that p2 49 : Thus,
p2 = 49 = Prfrejectjhonestg Pr
Prfrejectjhonestg Prfhonestg
fhonestg + Prfrejectjdishonestg Prfdishonestg
) 94 = + q(1

) ; or q =
5
4(1 ) (21.27)
Thus, a dishonest officer rejects a bribe in the first period with probability
5=4(1 3): So in the PBE, the second period motorist offers a bribe with
probability 4 if he knows that the first bribe was rejected.
{0,10}
Out
Entrant
{3,3}
Enter
Accommodate
Incumbent
Prey
{−1,−1}
E. Payoff to first period motorist from paying the ticket: 100

: If she offers a
Pr
bribe, then fbribe rejectedg = + (1 ) Pr
q and fbribe acceptedg =
1 [ + (1 ) ]
q : Thus, expected payoff from offering a bribe is
[ + (1 )q] ( 200) + [1 ( + (1 )q)] ( 20) (21.28)
So she should offer a bribe if
200 200(1 )q 20 + 20 + 20(1 )q > 100 (21.29)
80 180 > 180(1 )q (21.30)

80 180 = 4 9
q < 180(1 ) 9(1 ) (21.31)
5 < 4 9
4(1 ) 9(1 ) (21.32)
45 < 16 36 (21.33)

16
< 81 (21.34)
Thus, first period motorist offers a bribe if 16 and pays the ticket otherwise.
< 81
8. (a) The two Nash equilibria to this game are (Enter, Accommodate) and (Out, Prey). The
SPNE is (Enter, Accommodate). The extensive form is shown in the figure.
(b) The incumbent will not Prey in order to establish a reputation for toughness and the
SPNE has all 10 entrants playing Enter , with the incumbent playing Accommodate, in
all markets.
{0,11}
Out
Entrant
{3,3}
Enter
Accommodate
Incumbent
Prey
{−1,−1}
(c) With these strategies (like a tit-for-tat supergame), if the incumbent preys against the
first entrant, entry will not occur again. This will pay for the incumbent if
1 + 1 10 1 3 (21.35)
or
4
11 (21.36)
(d) A sane incumbent may now Prey in early periods, so as to increase later entrants’ pos-
terior probability that they are in fact crazy. Even though the sane incumbent would not
prey in a finite game without uncertainty, once there is some doubt as to whether he is
crazy, he can exploit this profitably by mimicking the behaviour of the crazy incumbent.
Doing so will reduce the probability of entry in later periods, because of the incumbent’s
reputation (see the development of this model in the chapter).
9. (a) The two Nash equilibria to this game are (Enter, Accommodate) and (Out, Prey). The
SPNE is (Enter, Accommodate). The extensive form is shown in the figure.
(b) The incumbent will not prey in order to establish a reputation for toughness. The SPNE
is that all N entrants play Enter, and the incumbent plays Accommodate in all markets.
The paradox is that we might expect the incumbent to have an incentive to prey early to
establish a reputation for toughness, which then deters later entrants.
i. The Sane incumbent accommodates, independent of p .
ii. The Crazy incumbent preys, independent of p:
34 34
iii. The Entrant enters if p < = ; stays out if p > = and is indifferent if p = 3=4 .
(c) i. Prey with probability 1 means that the second entrant learns nothing from period 1
and will play Enter, and the expected payoff for the Sane incumbent from playing
Prey is 2. The Sane incumbent can do better by playing Accommodate in period 1,
since the expected payoff is 6. Thus, Prey with probability 1 is not an equilibrium.
U D
s 2, 6 2, 6
u 0, 4 0, 2
d 4, 2 0, 1
ii. Accommodate with probability 1 means that both Entrants play Enter and the ex-
pected payoff for the incumbent is 6. But then if the 2nd entrant observes Prey in
period 1 they will be certain that the incumbent is crazy and stay out. Thus, the
incumbent can increase payoffs to 10 by playing Prey. Thus Accommodate with
probability 1 is not an equilibrium.
(d) i. If the first entrant enters, then the probability they play Prey is
q = 3(1 p p) = 12 : (21.37)
ii. The first Entrant plays Enter if

3 4p > q
4(1 p) (21.38)
which is violated, so the first entrant does not enter.

iii. If the first Entrant Does Not Enter then the second Entrant does not change their
prior beliefs and the second Entrant will play Enter. If the incumbent plays Accom-
modate then the second Entrant knows that the Incumbent is Sane and the second
Entrant will Enter. If the Incumbent plays Prey in the first period then the second
Entrant enters 1/2 of the time.
iv. There will exist a market N0 ; such that in any market before N0 the Sane Incumbent
will Prey with certainty in order to establish a reputation for preying. This then
implies that the entrant will play Out with certainty in all markets before N0 : In
markets after N0 ; the Sane Incumbent will mix. In order for the incumbent to
mix the Entrant must also be mixing . As the market nears the final market, the
incumbent is more likely to Accommodate, which means p has to increase as N
increases in order to get the Entrant to mix. Thus, for N > N0 the Incumbent
and the Entrant both mix, but as N increases, p has to increase in order to keep the
Entrant mixing because the probability that the incumbent Accommodates increases.
10. Figure 21.6 is the normal form for this game.
( ) ( )
There are two Nash equilibria, d; U and s; D . These are also the two SPNE since the
( )
only subgame is the game itself. The equilibrium s; D involves a non-credible threat. The
strategy U strictly dominates D for player 2 at its information set. Thus player 1 should expect
player 2 to play U regardless of whether it plays u or d. Therefore player 1 should play d and
based on 1’s strategy player 2’s believes should be that with probability 1 it is at the lower
node in its information set and probability 0 that it is at the upper node. Player 2 of course
plays U .
Chapter 22
Vertical Integration and Vertical

Restraints
1. ( + )
(a) The publisher’s profit is w c e Q .The publisher will choose P to replicate the
vertically integrated monopoly price, namely
P = (a + c 2 e + d) (22.1)
The wholesale price is chosen so that the carrier just covers costs, or
w = (a + c 2 e + d) d or (a + c 2 e d) (22.2)
The figure illustrates the price that the carrier would choose, given w , which is labeled
P DM . This price, which is the product of double marginalization, is obviously higher
than P .
(b) For any given w, P will set by the carrier at a monopoly level, namely P = (a+w +d)=2.
We can find the derived demand for the publisher by setting this equal to inverse demand,
or
=
P a Q a w d= =( + + ) 2 (22.3)
and solving for Q to yield
Q = (a w2 d) (22.4)
Substituting this into the publisher’s profit function, given above, we have
= (w c + e) (a w2 d) (22.5)
Maximizing this expression w.r.t. w we obtain the optimal wholesale price

w0 = (a + c e d)
2 (22.6)
and the corresponding retail price is
P 0 = (3a + c 4 e + d) (22.7)
99
Price
PDM
P*
d
w*
MR(Q) P(Q)
Quantity
It follows that P 0 > P (double marginalization). Note that w0 = w so that the figure
for part (a) illustrates the double marginalization solution exactly.
(c) Since P 0 > P consumers’ surplus is lower where RPM is illegal.
(d) No, because in the “complete contracting” solution given in part (a) the retail margin
is zero. Thus, the incentive for sales effort is also zero. In order to achieve optimal s
we would need either two part pricing (“selling the business to the carrier”) or quantity
forcing, which would force the carrier to sell the optimal quantity.
2. (a) The integrated monopoly price of skates is P = (a + c)=2; and the profits of the inte-
grated monopolist are =(
a c 2= b . ) 4
(b) The derived demand for wheels is Q = (a w)=2b . The wheel monopolist’s profit is
= (w c)(a w)=2b and the skate manufacturers profit is (P w)(a w)=2b. Thus,
without vertical integration, total profit can be written as = (P c)(a w)=2b: If
w = c this is the same as the vertically integrated problem; if w > c then total profits
will be lower, in equilibrium equal to 3(a c)2 =16b .
(c) There is no Nash equilibrium in pure strategies. Consider the skate manufacturer’s best
response to a given w - it is to set P =( + ) 2
a w = : But the best response of the wheel
maker to any given P is to increase w so as to reduce downstream profits to zero i.e.
=
w P . Clearly these have no intersection.
3. (a) In the competitive case, we can set the input prices equal to marginal cost, to obtain the
optimal input mix of plastic and steel as:
S = 1

cs 1
(22.8)
P cp
Chapter 22. Vertical Integration and Vertical Restraints 101
(b) If the plastic input is monopolized, its price is given by p = pm . The new profit
maximizing input mix is:

S m = 1 cs 1 > S
P pm P (22.9)
so that too little plastic relative to steel will be used in production because the watch
manufacturer will substitute away from the monopoly priced plastic.
4. For a retailer to have efficient incentives, we require @i=@Pi =
@ =@Pi and @i =@Si =

@ =@Si in equations (22.6) and (22.7) respectively in the Appendix. Substituting these con-
ditions into the equations, we have
@qi = (P c) @qj :
(w c) @P j @Pi (22.10)
i
@q
(w c) @Sii = (Pj c) @S @qj : (22.11)
i
Dividing both equations by (w c) and rearranging, we obtain
@q @q
(Pj c) = @P = @S
i i
(w c) @P
i i
@q @q j j
(22.12)
@S i i
Since qi + qj = Q or market demand, we can rewrite the above as

@qi @qi
@Q
@Pi
@qi = @Q
@Si
@qi (22.13)
@Pi @Pi @Si @Si
Divide both sides by qi=Q and muliply the numerator and denominator of the LHS by Pi; and
the numerator and denominator of the RHS by Si to obtain
"rP = "M"S "r
r
"M
P "rP S S
(22.14)
Dividing the LHS numerator and denominator by "M

P and the RHS numerator and denomina-
tor by "M
S gives the desired result.
5. (a) In order to prevent consumers from switching to new equipment in period 2, the price of
service must be set such that v p u E or p E ( )
u v with equality for the
profit maximizing price p .
(b) Profits from servicing locked in customers are p c which from (a) equals E u v c: ( )
Competition in the primary market will force total profits, on the sales of the primary
good plus the discounted value of the profits from service, equal to zero. Thus
E K + (E (u v) c) = 0 (22.15)
Solving this equation for E; we have
E = K + 1(u+ v + c) (22.16)
which is less than K since v c u K implies u v + c < K .

Chapter 23
Horizontal Mergers
1. (a) For firms of equal size, HHI =1 10 000

=N ; so pre-merger HHI = 2000; post-
merger HHI = 2800 so merger does qualify for further scrutiny.
(b) Now post-merger HHI =1
=N ; 10 000 = 1 4 10 000 = 2 500
= ; ; : Once again the
change in HHI exceeds 200 and the merger does qualify for further scrutiny.
2. Using the best-response function given, the pre-merger quantity produced by each firm i is
q0 = a 4b c (23.1)
which yields the price

P 0 = a nbq0 = a +4 3c (23.2)
and the profit to each firm

0 = (a 16bc) :
2
(23.3)
When firms 1 and 2 merge, the two firms in the industry produce
qm = q3 = a 3b c (23.4)
The profit to the merged firm is

m = (a 9bc)
2
(23.5)
Since
m = (a 9bc) < (a 8bc) = 20 ;
2 2
(23.6)
the merger is not profitable to the two merging firms.

Clearly, when all three firms merge, we have a monopoly and
m = (a 4bc) > 3(a16bc) = 30 :

2 2
(23.7)
Thus, the merger is profitable.
103
For the general case of n firms. The best-response function given yields
q0 = (na+ 1)c b (23.8)
and each firm earns a profit

0 = ((na+ 1)c)2 b :
2
(23.9)
When m firms merge (i.e. n m firms do not participate in the merger), there are now
n m + 1 firms in the industry.
The post-merger quantities and profits are
qm = (n am +c 2)b (23.10)
m = (n (am +c)2)2 b :
2
(23.11)
Hence, for the merger to be profitable to the participating firms, the condition required is
m m0 or (n + 1)2 m(n m + 2)2: (23.12)
3. Post merger, Cournot quantities are given by
q1 = 1 2c31 + cm
qm = 1 2c3m + c1
The post-merger price can be computed as
P m = 1 + c13 + cm (23.13)
For P m to fall below the pre-merger level of 0.46 we set

1 + c1 + cm 0:46
3 (23.14)
and substituting for c1 = 0:1; we find that cm 0:28:

4. Firms 1 and 2 merge in this question so cm = 0:1; c3 = 0:4. The best-response functions are
qm = 1 0:21 q3 ; q3 = 1 0:24 qm : (23.15)
The post-merger quantities and price are

qm = 0:4; q3 = 0:1; Q = 0:5; P = 0:5: (23.16)
Welfare is given by
W = 0:5(1 P )Q + qm (P 0:1) + q3(P 0:4) = 0:295: (23.17)
The net gain in welfare is 0:005 which is smaller than the net welfare gain resulting from a
merger between firms 2 and 3. The efficiency effect is smaller because the non-merging firm
is relatively inefficient compared to the other case where the non-merging firm (firm 1) was
the most efficient of the three.
Chapter 23. Horizontal Mergers 105
5. In the pre-merger Cournot duopoly, we have
qi = a 3 c = 0:2
Q = 0:4
P = 0:6
and hence
CS = 0:5 0:42 = 0:08

PS = (0:6 0:4) 0:4 = 0:08
TS = CS + PS = 0:16
Post merger, we have a monopoly with costs equal to 0:4(1 X )q: So
P = 1 + 0:4(1 X ) = 0:7 0:2X
2
Q = 0:3 + 0:2X
CS = 0:5Q2
PS = Q2
TS = 3=2 (0:3 + 0:2X )2
Setting Total Surplus equal in the two cases and solving for X we obtain X = 0:13; or at least
a 13% cost reduction is required if total surplus is to increase post-merger.
6. (a) Pre-merger p1 = = 5 16; = 9 32
p2 = p3 = . Post-merger p1 p2 = p3 = . = = 5 4; = 3 4
Because the merging products are now closer substitutes, prices (of merging and non-
merging products) increase by more than in the benchmark case.
(b) Pre-merger p1 = p2 = 11 30; = 2 5
= p3 = . Post-merger p1 p2 = = 11 14; = 5 7
= p3 = :
Note the price of the merging products increases but not by as much when these products
are not close substitutes.
Chapter 24
Rationale for Regulation
1. The industry is a natural triopoly if the cost of aggregate output is minimized when there
are three firms (plants). Because marginal cost is increasing costs are minimized for a given
number of firms if the output is divided equally. This means that
TC (q; N ) = N [c(q=N )2 + f ]:
Substituting in N = 2, 3, and 4 and comparing, the industry is a natural triopoly if
r r
6f < q < 12f ;
c c
where the lower bound is determined by TC(q; 2) = TC(q; 3) and the upper bound is deter-
mined by TC(q; 3) = TC(q; 4).
2. See Figure 24.1. The minimum industry average cost is the same when there are n firms as
when there is a single firm, but it is only reached when all firms produce at minimum efficient
scale. For demand Dn when P = c (minimum average cost) average cost for the industry
(and hence aggregate cost) is minimized when there are n firms each producing at qmes . Of
course this is the long-run competitive equilibrium.
P P
3. (a) The cost function is subadditive: C (q) < C (qi) where qi = q since f < Nf for
N . 2
(b) Using (8.28) in Church and Ware:
NC =0 if
f (1=4)(A c)2
N C = 1 if
(1=4)(A c)2 > f (1=9)(A c)2
N C = 2 if
(1=9)(A c)2 > f (1=16)(A c)2
N C = 3 if
(1=16)(A c)2 > f (1=25)(A c)2
107
AC1 ACn
Lower Envelope
Dn
Q
mes
nq
Chapter 24. Rationale for Regulation 109
4. (a) The minimum efficient scale is found by setting AC (q) equal to MC(q) and solving for
q: qmes . =3
(b) The technology is a natural monopoly for q 4:24.
(c) To find the Ramsey price and output set P (10 Q) equal to AC (Q + 9=Q) and solve
for Q: QR = 3:82. At this output both price and average cost are P R = 6:18.
(d) Yes, an entrant can profitably enter by charging a price P = 6:18 t, (where t is small)
and sell qE = qmes = 3. By doing so its profits would be 0:54 3t > 0 for t < 0:18.
(e) If the incumbent’s fixed costs are sunk then the profit-maximizing level of output for the
incumbent is qI = (7 4)
= and the quasi-rents of the incumbent are 6.125. Since these
are less than its fixed costs, the incumbent would exit the market if its fixed costs are
avoidable.
(f) The change in total surplus from entry isTS = 2:2562.
(g) The change in total surplus in the short run is TS = 11:8376. The change in total
surplus in the long run is TS = 2:2562.
(h) No, entry is not socially beneficial in this case since the incumbent (eventually) exits if
the entrant produces at qmes . The natural monopoly is not sustainable since there does
not exist a price that is both feasible and does not attract entry.
5. (a) QR = 3 and P R = 6.
(b) No. There does not exist a P and q such that 6 > P > AC (q).
(c) Yes if qI =6
post-entry is committed. If competition is Cournot post-entry then the
natural monopoly is still sustainable because the entrant would earn negative profits.
6. (a) The Ramsey quantities are found by setting P equal to AC and solving for Q:
40 + p1600 4f
QR = 2 :
Total surplus equals consumer surplus which is (QR )2 =2.
(i) If f = 1 then TS = 799:00.

(ii) If f = 10 then TS = 789:97.
(iii) If f = 100 then TS = 696:39.
Cournot free-entry equilibrium:
(i) If f = 1 then TS = 760:50.

(ii) If f = 10 then TS = 672:34.
(iii) If f = 100 then TS = 450.
Unregulated Monopoly:
Fixed Cost Monopoly Cournot

1 0.7497 0.9518
10 0.7469 0.8511
100 0.7180 0.6462
Table 24.1: Relative Efficiency
Fixed Cost Regulation Monopoly Cournot

1 719.10 599 760.5
10 710.97 590 672.34
100 626.75 500 450
Table 24.2: Efficiency of Alternative Governance Mechanisms
(i) If f = 1 then TS = 599.

(ii) If f = 10 then TS = 590.
(iii) If f = 100 then TS = 500.
(b) See Table 24.1.
(c) Table 24.2 shows the net surplus for the three alternatives—including the cost of regu-
lation. If f is small—so economies of scale are not extensive—the free-entry Cournot
equilibrium is the preferred governance mechanism. For more extensive economies of
scale—larger values of f—regulation is preferable.
Chapter 25
Optimal Pricing for Natural

Monopoly
1. (a) See Figure 25.1. Quantity demand equals long-run marginal cost at price P1.
(b) To maximize total surplus in the short run, price should equal short-run marginal cost.
It might differ from long-run marginal cost if demand has changed so that for the level
of output where price equals short-run marginal cost, the level of capital is no longer
optimal in the long-run. In Figure 25.1 when capital stock is chosen optimally in the
long-run, the efficient price is P1 when demand is D. However if demand increases to
D1 then in the short-run the efficient price is P2, but in the long-run it is P3.
(c) In the usual treatment avoidable costs in the short-run are called variable costs and short-
run average costs do include unavoidable fixed costs—as in Figure 25.1. Following
the usual convention, some of the firm’s capital expenditures are recovered if the price
exceeds minimum short-run average variable cost. Alternatively, if the price exceeds
average avoidable cost in the short-run, then some of its capital expenditures will be
recovered. Not all capital expenditures will be recovered since that requires price to ex-
ceed long-run average cost, but for a strong natural monopoly the efficient price (equals
long-run marginal cost) is always less than long-run average cost.
(d) Yes, since if price did not exceed minimum short-run average variable cost—conventional
terminology—or average avoidable cost in the short-run—our treatment—the firm would
shut down.
=
(e) Yes, if there was an increase in demand such that P SRMC > SRAC. No, since in the
long run the efficient response to the increase in demand is to increase capital and reduce
price to the new (lower) marginal cost. The result is lower costs, increased output, and
negative profits. An example is shown in Figure 25.1. Demand increases from D to D2 .
The efficient short-run price is P4 > SRAC. However, in the long-run the efficient price
is P5 and the firm once again is in a loss position.
2. (a) See Figure 25.2. The inefficiency associated with P1S is area A. The inefficiency as-
sociated with P2S is the sum of areas A, B, and C. The greater the difference between
marginal cost and average cost, the greater the inefficiency— area B C .+
(b) See Figure 25.3. The inefficiency associated with D1 (inelastic) is the sum of areas B
and C. The inefficiency associated with D2 (elastic) is the sum of areas A and C. The
111
$/Q SRMC(Q)
LRAC(Q)
LRMC(Q)
SRAC(Q)
P4
P2
P5
P3
P1
D1
D D2
Q Q

Chapter 25. Optimal Pricing for Natural Monopoly 113
PR
A
S AC MC
P1 1
B C
MC2
P2S
Q
QR Q1S Q2S

PR
AC
B
A
C
P1S=P2S MC
D2
D1
Q
Q2R Q1R Q2
more elastic D2 the larger A and the less elastic D1 the smaller B.
3. If is the common value for the markup then for each product i pi = MCi where =
1=(1 )
. It is inefficient because it does not account for the elasticity of demand. The prices
of relatively elastic goods will be too high, while the prices for relatively inelastic goods will
be too low.
4. (a) If the demand function is Qi = 1 biPi then to find the inverse demand function, solve
for Pi as a function of Qi:
Pi = 1 Qi ; bi
1
as required. This is linear with intercept =bi . Consumer surplus is the area of a triangle
1
with height =bi Pi and width Qi . As a function of Qi,
2
CS (Qi) = 2Qbii :
(b) Denote aggregate welfare by W. Then
W = CS1(Q1 ) + CS2(Q2 ):
Q2
Q2R
Q1
Q1R
Iso-welfare contours are combinations of Q1 and Q2 such that W is the same. Substi-
tuting in for consumer surplus and bi ,
2
W = Q21 + Q22:
Solving for Q2 the equation of an iso-welfare contour is
Q2 = (W Q21=2)1=2:
Three iso-welfare contours (corresponding to different values for W) are shown in Figure
25.4.
= +
(c) The profits of the firm are P1Q1 P2 Q2 f . Substituting in for Pi and f the profits
of the firm as a function of Q1 and Q2 are
(Q1 ; Q2) = (1 Q1 )Q1 + (2 2Q2 )Q2 0:72:
Setting this equal to zero, the relationship between Q2 and Q1 such that profits are zero
is q
2+ 4 8(Q12 Q1 + 0:72)
Q2 = 4 :
The zero-profit contour is also shown in Figure 25.4.

(d) See Figure 25.4. The welfare-maximizing quantities are found by finding the point on
the zero-profit constraint that is on the highest iso-welfare contour. This is the point
( )
QR1 ; QR2 in Figure 25.4 which is (i) on the zero-profit constraint; and (ii) is tangent to
an iso-welfare contour.
(e) The tangency condition requires that the slopes of the iso-welfare contour and the zero-
=
profit constraint be equal. Doing so shows that tangency requires that Q1 Q2 . Substi-
= =
tuting Q Q1 Q2 into the zero profit constraint:
Q2 Q 0:24 = 0:
The relevant root is the largest, so QR1 = QR2 = 0:60; P1R = 0:40, P2R = 0:80; and
= 0 54
W : .
( )
(f) The regulated prices are found by (i) finding the revenue Ri that must be recovered for
= 0 72
each good and then (ii) setting prices appropriately. R1 : and R2 = (1 )0 72
: .
(1 )
The revenue required for good i is raised if =
Qi Qi =bi Ri .
When = 0:325, then P1 = 0:374 and P2 = 0:832.
When = 0:333, then P1 = 0:40 and P2 = 0:80.
When = 0:345, then P1 = 0:455 and P2 = 0:763.
(g) The welfare associated with each of the three different weights is (i) 0.537, (ii) 0.540,
= (
and (iii) 0.530. The elasticity at the FDC prices equals "i bi Pi =Qi . )
(i) "1 = 0 597 = 0 712
: and "2 : and as expected welfare is increased by increasing P1
(the price of the inelastic good) and decreasing P2 (the price of the elastic good).
(ii) =1 3
= corresponds to a common cost allocation such that FDC prices are Ramsey.
(iii) "1 = 0 835 = 0 618
: and "2 : and as expected welfare is increased by decreasing P1
(the price of the elastic good) and decreasing P2 (the price of the inelastic good).
(h) At the Ramsey prices, revenues paid by consumers of good 1 are 0.24 and 0.48 for good
2. The stand-alone costs for good 2 are 0.44, so good 2 fails the stand-alone cost test.
An entrant could profitably enter by undercutting the Ramsey price in market two.
(i) The Ramsey price for a firm that only produces good 2 is the price that maximizes
consumer surplus in that market subject to the entrant breaking even. To find it set
=
P AC and solve for QR2 , the Ramsey quantity: QR2 = 0 673
: . Substitute this into the
demand curve for good 2 to solve for the Ramsey price: P2R QR ( ) = 2 2(0 673) =
2 :
0 654 = 0 80
: : This is less than firm 1’s Ramsey price in the multiproduct case of P2R : .
If firm 1 were to charge the monopoly price in good 1, its quasi-rents would only be
0.25, less than good 1’s stand-alone fixed costs. In the long-run, firm 1 would exit from
market 1 and surplus would only be produced in market 2. Total surplus from Ramsey
pricing in market 2 is 0.453 which is less than the 0.54 when Ramsey prices are charged
by a multiproduct firm that is the only producer in both markets.
(j) At the Ramsey prices, revenues paid by consumers of good 1 are 0.24 and 0.48 for good
2. The stand-alone costs for both goods are 0.60 and the incremental cost for both goods
is 0.12. Hence both goods pass both the incremental cost test and the stand-alone cost
test.
5. (a) Since its demand is perfectly inelastic, Ramsey pricing requires raising the price for
good 1 high enough to cover its marginal costs and all fixed costs, and setting the price
of good 2 equal to its marginal cost: pR
1 = 14and pR
2 = 20
:
(b) At the Ramsey prices, revenues paid by consumers of good 1 are $14,000 and $1600
for good 2. The stand-alone costs for good 1 are $12,000 and the stand-alone cost test
for good 1 is not satisfied at the Ramsey prices, indicating that consumers of good 1
subsidize consumption of good 2. The incremental costs for good 2 are its variable costs
of $1600 plus the increase in fixed costs from producing both goods. Efficient joint
production entails a fixed cost of $4000 while efficient production of just good 1 entails
a fixed cost of $2000, so the increase in fixed costs from adding production of good 2
is $2000 and its incremental costs in total are $3600 and product 2 fails the incremental
cost test.
6. Constant average cost implies Ct =
cQt where c equals average cost. If the firm charges
monopoly prices in the first period, then in the second period the firm has to satisfy the con-
straint P2QM M
1 cQ1 or dividing through by Q1 P2 c
M =
P R . When average cost is
constant the Ramsey price is reached in two periods.
7. Let c be the unit cost of the firm. If the firm does not waste then its stream of profits under the
VF mechanism is:
1NW = p1 q1 cq1 .
tNW = 0 for t 2:
If it does waste, its stream of profits under the VF mechanism is:
1W = p1q1 (c + w)q1.
2W = (c + w)q2 wq2.
3W = 0 for t 3:
If is the discount factor then the present value of W exceeds the present value of NW if
[p1q1 (c + w)q1 + ((c + w)q2 wq2)] p1q1 cq1 > 0:
Letting q2 = q1 + q and cancelling terms, this equals
(q1 + q) q1 > 0:
This is true provided is sufficiently close to 1 since q is positive. It is the increase in output
in the second period from the decrease in price: p2 = c + w < p1 .
8. (a) If the firm wastes w per period in equilibrium, then it earns zero profit and the price in
= +
period t is pt c w were c is actual average cost. However it could increase its profits
by eliminating waste in period t +1
: t+1 = [( + ) ] 0
c w c qt > . Of course its profits
in subsequent periods would be zero since pt+2 c. =
(b) Again assume that actual unit costs are c and per unit waste is w. Suppose the firm
=1
charges p1 at t . Under the VF mechanism the constraint on prices in the second
period is p2 c. If the firm engages in waste at t =2 , then its profits will be 2 =
[ ( + )] = =3
c c w q2 wq2 and the pricing constraint at t is p3 c w. Its profits if +
=3
it stops wasting at t are 3 =[ + ] =
c w c q3 wq3. However, this is less than its
=2
“investment” at t to relax the VF constraint at t=3 since q2 > q3 . Why? Because
demand at price c (q2) is greater than demand at the higher price c w (q3).+
9. If product i fails the incremental cost test then
pi qi < C (qi ; qj ) C (0; qj );

but we know from the zero-profit constraint that
C (qi; qj ) = pi qi + pj qj :
Substituting in, this means
piqi < pi qi + pj qj C (0; qj )

or
C (0; qj ) < pj qj :
To prove that if product i passes the incremental-cost test, product j passes the stand-alone
cost test reverse the inequalities.
10. (a) The average willingness to pay for capacity is
100 k 2 + 50 k 2
2
and the average cost of capacity is 4. Equating and solving k = 69
which is greater than
demand in the off-peak period. At the optimum the willingness to pay of the off-peak
consumers will be zero and the true willingness to pay for capacity is
100 k 2 :
2
Setting this equal to the average cost of capacity (4) and solving, k = 90, so P p = 10
and P o =2 .
(b) The optimal capacity is k = 85. The peak price is P p = 15 and the off-peak price is
Po . =5
11. (a) Capacity imposes costs only in the period with the most consumption. Assume that this
is during the day. Willingness to pay for capacity in the day is
96 10k :
Setting this equal to the cost of capacity (6) and solving k = 900
: We need to check and
(
= 4)
make sure that at marginal-cost pricing P n nighttime consumers do not want to
consume more than 900: Qn (4) = 500 30(4) = 380 :
(b) P d = 10 and P n = 4.
(c) The technology allows the excess capacity at night to be used to produce for the day at
marginal cost of 4. Excess capacity at night is 520. Demand in the day when P d =4
is Qd (4) = 1000 10(4) = 960 . So marginal cost pricing in the day is possible since
total available capacity (day+night) is 1420 and demand is only 960. The optimal prices
are P d =4 =4
and P n .
(d) In the long-run, the marginal cost of capacity of 6 must be recovered for efficiency. So
= 10
the efficient prices are again P d and P n =4 , but now capacity at night can be
used to create supply for the next day. Demand at night when there is marginal cost
pricing is 380, so the minimum capacity to provide 900 units (demand during the day at
its marginal cost) is determined by k 380 + = 900
k , or k= 640 : Capacity of 640
means that 1280 units can be produced during both periods, 380 of which is consumed
at night and 900 during the day. The savings in cost from reducing capacity from 900 to
640 is 1560.
12. Consider Figure 25.5 which shows the demand for the largest consumer under the two-block
tariff
P1Q if Q Q1
E = + (
P1Q1 P2 Q Q1 if Q > Q1 )
Then define the three-block tariff
8
<P1 Q if Q Q1
E = : P1Q1 + P2(Q Q1) if Q2 Q > Q1
P1Q1 + P2(Q2 Q1) + P3(Q Q2) if Q > Q2
where Q2 is the consumption of the largest consumer under the original two-block tariff.
This consumer cannot be made worse off since they can still purchase Q2 for the same total
outlay. However, they will not choose to consume Q2 since the marginal price they face has
fallen. Instead they will increase consumption to Q3 , increasing their surplus by area A and
the profits of the firm by area B. Similarly, all other consumers can be no worse off since the
options available under the two- block tariff remain available under the three-block tariff.
P1
P2
P3 A
B MC
Q
Q1 Q2 Q3

Chapter 26
Issues in Regulation
1. (a) The price of a system equals the sum of the prices for its two components: P R P U + =
4 + 6 = 10 . At this price when demand is Q = 64 P , the quantity demanded is
Q = 54
.
(b) Unregulated and undiversified monopolist: The competitive price of U is 6, so the
demand for component R is QR = 64 6 PR = 58 P R since Q QR =
due to fixed proportions. This means the marginal revenue for the monopolist is
MR = 58 2 QR . Setting it equal to marginal cost (4) and solving, the profit-
maximizing quantity is 27 and P R = 27. At this price and quantity the monopolist
makes profits M = 729 .
Regulated and diversified monopolist: By tying the monopolist monopolizes good
U. Since P R =4 , the demand for good U is QU = 60 P U , where again Q QU=
due to the assumption that systems have fixed proportions. This means the marginal
revenue for the monopolist is MR = 60 2 QU . Setting it equal to marginal cost (6)
and solving, the profit-maximizing quantity is 27 and P U = 33 . At this price and
quantity the monopolist makes profits M = 729 . By tying the goods, the firm is
able to earn the same profits as if were an unregulated monopolist.
2. (a) The efficient price of B is 28. The marginal cost function, efficient level of B output and
quasi-rents from sales of B for the two technologies are:
Technology 1: MCB
1 = QB , QB1 = 28, 1B = 392:
Technology 2: MCB
2 = 28QB , QB2 = 1, 2B = 14:
The efficient price of good A is found by maximizing surplus in market A subject to the
firm breaking even. For
Technology 1 this requires:
QA1 (64 QA1 ) = (720 392) + 16QA1 ;

1 = 39:75 and P1 = 24:25:
or QA A
Technology 2 this requires:
QA2 (64 QA2 ) = (120 14) + 20QA2 ;

2 = 41:44 and P1 = 22:56:
or QA A
121
Total surplus in B is identical, but surplus in market A is greater with technology 2 since
QA2 > QA1 :
(b) From (a) we know that the quasi-rents earned from the sale of B for the two technologies
are 1B = 392
and 2B = 14
: The quasi- rents in the market for A following the indicated
FDC pricing methodology are:
For technology 1, the regulated quantity is such that
QA1 (64 QA1 ) = 360 + 16QA1 ;
1 = 38:70, P1 = 25:30; and 1 = 359:91:
or QA A A
For technology 2, the regulated quantity is such that
QA2 (64 QA2 ) = 60 + 20QA2 ;
2 = 42:59, P2 = 21:41; and 2 = 60:05:
or QA A A
The profits of the regulated firm are the sum of its quasi-rents in each market less fixed
costs:
1 = 392 + 359:91 720 = 31:91:
2 = 14 + 60:05 120 = 45:95:
The firm chooses technology 1, which is inefficient since QA A
1 > Q2 . The choice is still
inefficient if the regulator raises the price of A such that the firm just breaks even when
it chooses technology 2. From (a) we know that in this case QA = 41 44
2 : :
3. (a) The marginal cost of an integrated firm is 8 and its marginal revenue is 64 2Q, so its
profit-maximizing output is 28, its profit-maximizing price 36 and its profits 784.
(b) The profits of the integrated firm downstream are
m = (64 qm qe)qm (A + 4)qm + (A 4)qm ;
or
m = (64 qm qe )qm 8qm ;
where A is the access price, qe the quantity supplied by its competitor, and the integrated
firm recognizes that the actual marginal cost of downstream output is 8 regardless of the
access price it actually charges. Solving for the Cournot equilibrium as a function of
the access price:
qe = 64 3 2A ;
qm = 52 +
and
A
3 :
The downstream profits of the integrated firm are increasing in the access price.
The first-stage problem of the integrated firm is to set the profit-maximizing access price.
The profits of the integrated firm at the first stage are
m = (A 4)qe + (64 qm qe )qm 8qm ;
where the integrated firm knows that
qe = 64 3 2A ;
Chapter 26. Issues in Regulation 123
qm = 52 +
and
A
3 :
Substituting in and maximizing m with respect to A, the profit-maximizing access price
is 32, and the equilibrium price and output downstream are P = 36 and Q = 28 . At this
=0
access price production by the entrant is not profitable, qe and the integrated firm is
able to implement the monopoly outcome by raising the price of access.
(c) A = 52, which is less than the marginal cost of access. If A = 52, then P = 8
and Q = 56, the first best. However, at this price the profits of the access monopolist
are negative. The second-best price of access is the smallest access price such that the
= 5 56
profits of the vertically integrated firm are zero. This price is A : .
The welfare maximizing price is found by maximizing total surplus,
2
TS = Q2 + m + e
given that Q = qe + qm ,
qe = 64 3 2A ;
qm = 52 +
and
A:
3
4. (a) The profits of the integrated firm are:
= (w 4)Q + (64 qI qe )qe (A + 4)qI ;

but Q = qI + qe, so
= (w 4)qe + (64 qI qe)qe 8qI
and as required the coefficient on qI is 8.
(b) The profits of the incumbent are
I = (64 qI qe )qI 8qI

and the profits of the entrant are
e = (64 qI qe r)qe (w + 4)qe :

The Cournot equilibrium quantities as a function of r are therefore:
qI = 52 + 3r + w
qe = 64 23r 2w :
and
In order for the incumbent to be a monopolist in the final-goods market, it must be the
case that the entrant’s profit-maximizing choice in equilibrium is to produce qe =0 .
+ = 32
This is true if r w .
(c) The monopolist would be willing to pay the difference between the profits it earns as a
monopolist and the profits it earns as a duopolist. This difference is 435.56.
5. The incentive compatibility constraint for the low-cost firm is:
(p2 ; T2; 2 ) (p1 ; T1; 2 );

and the individual rationality constraint for the high-cost firm is
(p1 ; T1; 1 ) 0:
The cost function for the high-cost firm is
C1 (q1; e1) = (1 e1 )q1 + (e1 );

and for the low-cost firm
C1 (q2; e2) = (2 e2 )q2 + (e2 ):

The ICC for the low-cost firm equals
q2p2 + T2 C2(q2; e2 ) q1p1 + T1 C2 (q1; e~2):

The IRR for the high-cost firm when satisfied with equality is
p1q1 + T1 C1(q1; e1 ) = 0
or
p1q1 + T1 = C1 (q1; e1):
Substituting this into the ICC for the low-cost firm,
q2p2 + T2 C2(q2; e2 ) C1 (q1; e1) C2(q1 ; e~2):

But C2(q1 ; e~2) (~e2 ) = 1 e1 if the low-cost firm is to have the same costs as the high-cost
firm. This means that e~2 is defined by 1 e1 = 2 e~2 or e~2 = 2 1 + e1 . So the ICC
for the low-cost firm is
q2p2 + T2 C2(q2 ; e2) (1 e1 )q1 + (e1 ) (1 e1 )q1 (~e2 ) (26.1)
(e1 ) (~e2 )
(e1 ) (2 1 + e1 )
as required.
6. (a) The total benefits of the investment are
TB = 1 1 :
The total costs of the investment are TC = c. The investment is efficient if total benefits
are greater than total costs, or
1 > c:
1
Chapter 26. Issues in Regulation 125
1
ε = 1/4
ε = 1/2 t
Figure 26.1: Fig26.6(d)
(b) The firm will not choose to invest immediately in the efficient level of capacity because
it foresees that the regulator will then choose Pt =0
for all time and would then make a
loss of c.
(c) If the level of capacity ever reached the efficient level, the regulator would then set price
equal to zero and the firm would make a loss on its final incremental investment. There
is no way that the efficient level of capacity will be attained when there is no regulatory
commitment.
(d) See Figure 26.1. As t goes to infinity, the level of capacity approaches 1.
(e) Pt Kt = c(Kt Kt 1), but Kt = 1 "t . So
Pt (1 "t ) = c(1 "t (1 "t 1)) (26.2)
Pt = c("1 "t " )

t 1 t
Pt = c(11 ")""t
t 1
as required. As t goes to infinity, "t , and "t 1 both go to zero and so does the price.
(f) This pricing policy makes sure that the firm is able to recover all of its costs.
(g) Renege. If the regulator reneges at t=0and sets the P =0forever, then consumer
surplus every period is (1 ) " which has a present value of
V R = 11 " :
Follow. If the regulator prices according to (26.39) then consumer surplus at t (CSt )
is
CSt = Kt(1 Pt) (26.3)

= 1 "t 1[" + c(1 ")]:
The present value of this stream of consumer surplus is
VF = 1 1 " + c(1 ") ;

1 "
as required.
VF > V R if
" > c(1 ) :
(h) As increases the critical value of " decreases. An increase in means that the regulator
values the future more and as a result the rate at which the efficient level of capacity is
approached can be increased.
(i) The incentive to renege is mitigated by the promise of more investment (and consumer
surplus) in the future. Provided the discount factor is large enough or the future gains
large enough, the regulator will not renege.

IO+solutions

Uploaded by

Copyright:

Available Formats

IO+solutions

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

IO+solutions

Uploaded by

Copyright:

Available Formats

i

September 18, 2000

September 18, 2000

2 The Welfare Economics of Market Power 3

3 Theory of the Firm 17

4 Market Power and Dominant Firms 21

5 Non-Linear Pricing and Price Discrimination 25

6 Market Power and Product Quality 27

7 Game Theory I: Static Games of Complete Information 29

8 Classic Models of Oligopoly 31

9 Game Theory II: Dynamic Games of Complete Information 39

10 Dynamic Models of Oligopoly 43

12 Identifying and Measuring Market Power 55

13 An Introduction to Strategic Behavior 59

15 Strategic Behavior: Principles 71

16 Strategic Behavior: Applications 73

17 Advertising and Oligopoly 75

18 Research and Development 79

19 The Theory of the Market 85

20 Exclusionary Strategies I: Raising Rivals’ Costs 89

21 Exclusionary Strategies II: Predatory Pricing 91

22 Vertical Integration and Vertical Restraints 99

23 Horizontal Mergers 103

24 Rationale for Regulation 107

25 Optimal Pricing for Natural Monopoly 111

26 Issues in Regulation 121

The Welfare Economics of Market

1. Profit maximization by a monopolist requires producing where

Figure 2.1: Problem 2.3

q = 1400, p = 5, and  = 0. It is reasonable to assume that each cab will operate at

ny(p) = S (p) = 2npw

can be jointly maximized with respect to y and n.

Figure 2.7: Problem 2.8(a)

Figure 2.8: Problem 2.8(b)

Figure 2.9: Problem 2.8(c)

10. (a) MR(Q) = 100 2Q. See Figure 2.10.

Theory of the Firm

2. Economies of scale are global if S (q) = MC(q)

5. (a) The full information contract is:

 = 16(e1 + e2 ) 2e12 2e22:

Market Power and Dominant Firms

WTP = P1 + P2 (4.1)

Setting MR equal to MC, Q = 500

1 = (1000 Q12)(2 +  )Q1 +  (1000 4 Q1) :

So the profit-maximizing production profile and prices are

This is true if c  : .51 32

Non-Linear Pricing and Price

Market Power and Product Quality

(b) At e= 0,  = l, expected consumer utility is given by (1 l)V + l(1 a)V = V laV .

Game Theory I: Static Games of

1. The Nash equilibrium strategy profile is s = (M; M ).

(H )J = pBH 3 + (1 pBH )0

Figure 7.1: Normal Form for Problem 7.4

Classic Models of Oligopoly

qiD = 180 (12 )90 :

Firm 2's Best Reponse

iC = (30 218ci + cj )2 :

(b) Demand in Country A is QA = 40 P and demand in Country B is QB = 40 P , so

40 qBj =2 qA=2 qBi cB = 0:

Game Theory II: Dynamic Games of

q = 1400, p = 5, and = 0. It is reasonable to assume that each cab will operate at

= 16(e1 + e2 ) 2e12 2e22:

WTP = P1 + P2 (4.1)

1 = (1000 Q12)(2 + )Q1 + (1000 4 Q1) :

This is true if c : .51 32

(b) At e= 0, = l, expected consumer utility is given by (1 l)V + l(1 a)V = V laV .

(H )J = pBH 3 + (1 pBH )0

qiD = 180 (12 )90 :

iC = (30 218ci + cj )2 :

V r = 2025 + 2025 + 11600

3. (a) If competition is Bertrand, then i = 1800

i = (a 2q )q ;

since qr = (1=2)(A q ): Use (10.7) and solve for q as a function of :

qi = A3 (5( 9)9) :

At t play q if all firms played q at t 1.

V = (q ) + (q ) + 2(q ) + :::

V r = r (q) + (qp ) + 2(q ) + :::

V p = (qp ) + (q ) + 2 (q ) + :::

V rp = r (qp ) + (qp ) + 2 (q ) + :::

i = (pi 1)(50(pj pi + N2 )) + (pi 1)(50(ph pi + N2 ));

@ i = (50(p p + 2 )) + (50(p p + 2 )) 2(p 1) = 0:

Behaviour v Market Lerner Index

q1S = 9; q2S = 4:5; P S = 29; 1S = 81; 2S = 40:5:

q1C = 6; q2C = 6; P S = 32; 1S = 72; 2S = 72:

If f = 50, then q1l = 8 and 1 = 1M = 162 f .

If f = 18, then q1l = 12 and 1l = 144 : f

If f = 2, then q1l = 16 and 1l = 64 :f