Transcript EC 170: Industrial Organization

Entry Deterrence and Predation
Chapter 9: Entry Deterrence and
Predation
1
Introduction
• A firm that can restrict output to raise market price has
market power
• Microsoft (95% of operating systems) and Campbell’s
(70% of tinned soup market) are giants in their
industries
• Have maintained their dominant position for many years
– Why can’t existing rivals compete away the position of such firms?
– Why aren’t new rivals lured by the profits?
• Answer: firms with monopoly power may
– eliminate existing rivals
– prevent entry of new firms
• These actions are predatory conduct if they are profitable
only if rivals, in fact, exit
– e.g., R&D to reduce costs is not predatory
Chapter 9: Entry Deterrence and
Predation
2
Evolution of market structure
• Evolution of markets depends on many factors
– one is relationship between firm size and growth
• Gibrat’s Law
– begin with equal sized firms
– each grows in each period by a rate drawn from a random
distribution
– this distribution has constant mean and variance over time
– result is that firm size distribution approaches a log-normal
distribution
• Very mechanistic
– no strategy for growth
• Including strategic decision making affects distribution but not
conclusion that firm sizes are unequal
– What about the facts in the market place?
Chapter 9: Entry Deterrence and
Predation
3
Monopoly power and market entry
• Several stylized facts about entry
– entry is common
– entry is generally small-scale
• so small-scale entry is relatively easy
– survival rate is low: >60% exit within 5 years
– entry is highly correlated with exit
• not consistent with entry being caused by excess profits
• “revolving door”
• reflects repeated attempts to penetrate markets dominated by large firms
• Not always easy to prove that this reflects predatory
conduct
• But we need to understand predation it if we are to find it
Chapter 9: Entry Deterrence and
Predation
4
Predatory conduct and limit pricing
• Predatory actions come in two broad forms
– Limit pricing: prices so low that entry is deterred
– Predatory pricing: prices so low that existing firms are driven
out
• Outcome of either action is the same—the monopolist
retains control of the market
• Legal action focuses on predatory pricing because this
case has an identifiable victim
– a firm that was in the market but that has left
• Consider first a model of limit pricing
– Stackelberg leader chooses output first
– entrant believes that the leader is committed to this output choice
– entrant has decreasing costs over some initial level of output
Chapter 9: Entry Deterrence and
Predation
5
A limit pricing model
Then the entrant’s
the is
residual demand R1These
,
are the cost curves
$/unit
residual With
demand
entrant
for
potential entrant
Bythe
committing
to output
R1 =the
D(P)
- Q1 can operate profitably.
Entry is not deterred by the d
Q the incumbent deters
incumbent choosing Q1.
d
entry. Market price P
Then the entrant’s
is the limit price
e
e
R1
At price
marginal
P entry
revenue
is The
is MR
entrant’s residual
The
entrant
equates
MC
unprofitable
e
demand is
marginal revenue
Assume
instead that
ACe
Re = D(P) - Qd
with marginal cost
the incumbent
Assume
that the
d
commits
to
output
Q
incumbent
commits
D(P) = Market Demand
to output Q1
Re
Pd
Pe
MRe
qe
Quantity
Qd
Qd
Q1
Chapter 9: Entry Deterrence and
Predation
6
Limit pricing
• Committing to output Qd may be aimed either at
eliminating an existing rival or driving out a potential
entrant.
• Either way, several questions arise:
– Is limit pricing more profitable than other strategies?
– Is the output commitment credible?
– If output is costly to adjust then commitment is possible
• why should this property hold?
– could be claimed to be ad hoc to support the theory
• even if it holds, is monopoly at output Qd better than Cournot?
– may not be if the entrant’s costs are low enough
• Credibility may relate output to capacity
Chapter 9: Entry Deterrence and
Predation
7
Capacity expansion and entry deterrence
• For predation to be successful and rational
– the incumbent must convince the entrant that the market after
the entrant comes in will not be profitable one
• How can the incumbent credibly make this threat?
• One possible mechanism
– install capacity in advance of production
• installed capacity is a commitment to a minimum level of output
• the lead firm can manipulate entrants through capacity choice
• the lead firm may be able to deter entry through its capacity choice
– but is this credible?
– capacity must be costly to install and should be irreversible
Chapter 9: Entry Deterrence and
Predation
8
The Dixit model
• Consider a two-stage game
– incumbent in period 1 installs capacity
•
•
•
•
capacity K1 costs r.K1 to install
in second period incumbent can produce up to K1 at unit cost w
capacity can be expanded in period 2 at additional cost r per unit
capacity cannot be reduced in period 2
– potential entrant in period 2 observes incumbent’s capacity
choice
• to enter and produce incumbent needs capacity K2 which costs r.K2
• unit cost of production is w
• note: entrant will never install unused capacity
– if entry takes place firms play a Cournot game in the second
period
• Market demand: P = A – B(q1 + q2)
Chapter 9: Entry Deterrence and
Predation
9
The Dixit model 2
• Costs for the incumbent are:
– C1 = F1 + w.q1 + r.K1 for q1 < K1; marginal cost w
– C1 = F1 + (w + r)q1 for q1 > K1; marginal cost w + r
• Costs for the entrant are:
– C2 = F2 + (w + r)q2 ; marginal cost w + r
• Standard Cournot analysis gives the best response
functions:
– q*1 = (A – w)/2B – q2/2
when q1 < K1
– q*1 = (A – w – r)/2B – q2/2 when q1 > K1
– q*2 = (A – w – r)/2B – q1/2 provided that q*2 > 0
• for the entrant to enter it must expect to cover the sunk costs F2
• this implies a lower limit on the output that the entrant must make
Chapter 9: Entry Deterrence and
Predation
10
The Dixit model 3
• The incumbent’s best
response function has a
break in it at K1
• The entrant’s best
response function has a
break where sunk costs
are not covered
• Equilibrium depends
upon these two breaks
q2
L’
N’
R’
Chapter 9: Entry Deterrence and
Predation
R
N
K1
L
q1
11
The Dixit model 4
q2
• Consider the possibilities
• Suppose that firm 2 enters
• Equilibrium must lie
between T and V
• Where depends upon
location of the break in R’R
• Firm 1’s output is greater
than T1 and smaller than V1
• So capacity choice lies
between T1 and V1
L’
N’
R’
T2
T
V
V2
R
N
T1
Chapter 9: Entry Deterrence and
Predation
V1 L
q1
12
The Dixit model 5
q2
• Now suppose that firm 2
does not enter
• Must be that it cannot break
even at output less than T2
• Then firm 1 would want to
choose capacity M1
L’
N’
– this is the monopoly output
with MC = w + r
• M1 is actually the Stackelberg
output level for firm 1
– firm 1 as market leader will
never choose output and
capacity less than M1
R’
T2
M2
V2
T
S
V
R
N
T1
Chapter 9: Entry Deterrence and
Predation
M1
V1 L
q1
13
The Dixit model 6
• Suppose that the break in the
entrant’s best response
function lies at BL in R’T
• Incumbent chooses capacity
M1 and entry is deterred
• Suppose that the break in the
entrant’s best response
function lies at BS in TS
T2
M
• Incumbent chooses capacity 2
V2
M1 and entry is deterred
q2
L’
N’
R’ BL
TB
S S
V
BL
N
T1
M1
V1 L
• Suppose that the break in the entrant’s
best response function lies at BL in VR
• Incumbent chooses capacity of Stackelberg leader output
M1 and entry is accommodated
Chapter 9: Entry Deterrence and
Predation
R
q1
14
The Dixit model 7
• Now suppose that the break
in the entrant’s best response
function lies at B* in SV
• Incumbent can choose to
install capacity of Stackelberg
leader output and share the
market
• Or install capacity B1 and
maintain monopoly in the T
2
M2
market
V2
• Choice depends upon relative
profitability
q2
L’
N’
R’
T
S
B*V
N
T1
M1 B1 V1 L
R
q1
– If B* is “close to” S then use capacity to deter entry
– If B* is “close to” V then accommodate entry as Stackelberg leader
Chapter 9: Entry Deterrence and
Predation
15
Capacity expansion and entry deterrence 2
• An example:
–
–
–
–
–
–
P = 120 - Q = 120 - (q1 + q2)
marginal cost of production $60 for incumbent and entrant
cost of each unit of capacity is $30
firms also have fixed costs of F
incumbent chooses capacity K1 in stage 1
NOTE: incumbent will always produce at least K1 in
production stage—otherwise it throws away revenue that could
help cover the cost of installed capacity
– entrant chooses capacity and output in stage 2
– firms compete in quantities in stage 2.
Chapter 9: Entry Deterrence and
Predation
16
Entry deterrence
• Entry may not occur
– entrant’s costs are too high
• blockaded entry
• not predatory
• Entry may be accommodated
– entrant’s costs are low
• incumbent takes advantage of its being first in the market
• but does not deter
• Entry may be strategically deterred
– strategic deterrence profitable for the incumbent
– installs excess capacity as an entry-deterring strategy
– uses a credible commitment
Chapter 9: Entry Deterrence and
Predation
17
Evidence on predatory expansion
• Some anecdotal evidence
• Alcoa
– evidence that consistently expanded capacity in advance of
demand
• Safeway in Edmonton
– evidence that it aggressively expanded store locations in
response to potential entry
• DuPont in titanium oxide
– rapidly expanded capacity in response to to changes in rivals’
costs
– market share grew from 34% to 46%
Chapter 9: Entry Deterrence and
Predation
18
Introduction of Predatory Pricing
• Charges of predatory conduct are not new
– Microsoft is only one of the latest
– goes back to the days of Standard Oil
– more recent examples of predatory pricing
• Wal-Mart
• AT&T
• American Airlines
• But they face problems of credibility
– price low to eliminate rivals
– then raise price
– so why don’t rivals reappear?
Chapter 9: Entry Deterrence and
Predation
19
Predatory pricing: myth or reality?
• Theoretical and empirical doubts
– predation is generally not subgame perfect without
uncertainty regarding the incumbent
• return to this below
– McGee’s argument that predation is dominated by
another strategy
• merger is more profitable than predation
• so predation should not happen
– take an example
•
•
•
•
•
two period market
inverse demand P = A – B(qL + qF)
qF is output of leader and qF is output of follower
leader is a Stackelberg quantity leader
both leader and follower have constant marginal costs of c
Chapter 9: Entry Deterrence and
Predation
20
An example of predation
• At the Stackelberg equilibrium
– leader makes (A – c)2/8B
– follower makes (A – c)2/16B
– if the leader were a monopolist it would make (A – c)2/4B
• Suppose that the leader predates in period 1
– sets output (A – c)/B to drive price to marginal cost
– follower does not enter
– leader reverts to monopoly output in period 2 but the follower
does not enter
– aggregate profit is (A – c)2/4B
Chapter 9: Entry Deterrence and
Predation
21
An example of predation 2
• Suppose instead that the leader offers to merge
with the follower in period 1
– monopoly in both periods
– aggregate profit (A – c)2/2B
– so the leader can make a merger offer that the
follower will accept
• Merger is more profitable than predation but:
– merger may not be allowed by the authorities
• monopoly power
– what if there are additional potential entrants?
• may enter purely in the hope of being bought out
• Main point remains: threat of predation has to
be credible if it is to work
Chapter 9: Entry Deterrence and
Predation
22
Predation and imperfect information
• Suppose that the entrant faces financial
constraints
– must borrow to finance entry
• Entrant also faces uncertainty pre-entry
– faces some probability of “low” returns
• private information that can be concealed from bank
• incentive to misrepresent
• bank must then enforce removal of funding if low returns are
reported
• Incumbent then has incentive to take actions
that increase probability of failure
Chapter 9: Entry Deterrence and
Predation
23
Asymmetric information and limit pricing
• The preemption “games” are ways of resolving
the Chain-store paradox
– indicate that it is rational for incumbents to make
investments that are not profitable unless they deter
entry
• An alternative approach: information structure
– suppose that an entrant does not have perfect
information about the incumbent’s costs
• if the incumbent is low cost do not enter
• if the incumbent is high-cost enter
– does a high-cost incumbent have an incentive to
pretend to be low-cost - to prevent entry?
• for example by pricing as a low-cost firm
Chapter 9: Entry Deterrence and
Predation
24
•
•
•
•
A (simple) example
Incumbent has a monopoly in period 1
Threat of entry in period 2
Market closes at the end of period 2
Entrant observes incumbent’s actions in period
1
• These actions determine whether or not to
enter in period 2
• Incumbent is expected to be high-cost or lowcost
– no direct information on costs
– entrant knows that there is a probability p that the
incumbent is low-cost
• Need to specify pay-offs in different situations
Chapter 9: Entry Deterrence and
Predation
25
The Example 2
• Incumbent profits in period 1 (in $million)
– low-cost firm acting as low-cost monopolist: $100m
– high-cost firm acting as high-cost monopolist: $60m
– high-cost adopting low-cost monopoly price: $40m
• Incumbent profits in period 2
– if no entry, profits according to true type
– if entry occurs:
• low-cost incumbent: $50m
• high-cost incumbent: $20m
• Entrant’s profits in period 2
– competing against a low-cost incumbent: -$20,
– competing against a high-cost incumbent: $20m
Chapter 9: Entry Deterrence and
Predation
26
The Example 3
High Price
E3
Enter
Incumbent: 60 + 20 =
80 Entrant: 20
Stay Out
Incumbent: 60 + 60 = 120
Entrant: 0
Enter
Incumbent: 40 + 20 = 60
Entrant: 20
High-Cost
Nature
I1
Low Price
E4
Stay Out
Low-Cost
I2
Enter
Low Price
E5
Stay Out
Incumbent: 40 + 60 = 100
Entrant: 0
Incumbent: 100 + 50 = 150
Entrant: -20
Incumbent: 100 + 100 = 200
Entrant: 0
Chapter 9: Entry Deterrence and
Predation
27
With uncertainty and
With no uncertainty
The ifexample
a4
low price the entrant
the entrant enters
the
does not
incumbent is high-cost
Incumbent:
60 +know
20 = 80if
Enter
Entrant:
20 at E4 or E5
he is
High Price
E3
Stay Out
Incumbent: 60 + 60 = 120
Entrant: 0
High-Cost
Nature
I1
Low Price
Enter
E4
Stay Out
Low-Cost
I2
Enter
Low Price
E5
Stay Out
Incumbent: 40 + 20 = 60
Entrant: 20
Incumbent: 40 + 60 = 100
Entrant: 0
Incumbent: 100 + 50 = 150
Entrant: -20
Incumbent: 100 + 100 = 200
Entrant: 0
Chapter 9: Entry Deterrence and
Predation
28
The example 3
• Consider a high-cost incumbent
– high price in period 1 - entry happens, profits are 80
– low price in period 1 - if no entry profits are 100
– low price in period 1 - if entry profits are 60
• A high-cost incumbent has an incentive to
pretend to be low-cost
• The entrant knows this
• So a low-price of itself will not deter entry
– it is not a true signal of the incumbent’s type
• Only the probability that low-price means lowcost deters entry
Chapter 9: Entry Deterrence and
Predation
29
The example 4
• Consider the profits of the entrant given that the
incumbent sets a low-price in period 1
– if the incumbent is high-cost - profit is 20 with
probability 1 - p
– if the incumbent is low-cost - profit is -20 with
probability p
– so expected profit is 20(1 - p) - 20p = 20 - 40p
• Will the entrant not enter when it sees a low price?
• Only if p > 1/2
• Only if there is a “sufficiently high” probability
that the incumbent is low cost.
• Provided that pretence is expected to work a highcost incumbent has an incentive to set a limit price
Chapter 9: Entry Deterrence and
Predation
30
Limit pricing and uncertainty
• Monopoly power can persist even if the
incumbent is high-cost
• Entry only takes place if entrants believe that
the incumbent is high-cost
– so entry is more likely when incumbents are
expected to be weak
– entry then consistent with exit: efficient entrants
drive out inefficient incumbents
Chapter 9: Entry Deterrence and
Predation
31
Limit pricing and uncertainty 2
• Note: the model shows how a high-cost firm
can deter entry.
• However, to do this it must set a low price.
– This is how it “fools” the would-be entrant.
• The threat of entry forces the incumbent to
price below the monopoly price it would
otherwise set
• This lower limit price therefore mitigates the
resource misallocation effects of monopoly.
Chapter 9: Entry Deterrence and
Predation
32
Predation and Financial Constraints
• Two period model. Microhard and Newvel both start
in the market.
• Fixed cost are F and monopoly and duopoly profits πM
and πD
• Microhard may try to drive Newvel out in the first
period by driving pricing down so each firm looses
- ε = (πP – F )
• If Newvel is credit constrained, it cannot borrow the
necessary fixed cost for period 2 if it loses money in
period 1
• Microhard will predate if
• πM – F – ε > 2(πD – F) or πM – ε > 2πD – F
• Uncertainty and asymmetric information can generate
credit constraints
Chapter 9: Entry Deterrence and
Predation
33
Predatory Conduct and Public Policy
• The evidentiary requirements for prosecuting
predatory pricing cases are high
– Pricing below cost
– The predator had a reasonable expectation of recouping
the losses
• Very hard to distinguish predation from other procompetitive behavior
– Hard to measure marginal cost
– Learning curves, network effects, and other externalities
Chapter 9: Entry Deterrence and
Predation
34
Empirical Application: Entry Deterrence
In The Pharmaceutical Industry
• It is difficult to test in a statistical sense for systematic
entry deterrence. We need to identify markets:
– where entry was likely; and
– where the incumbent could do something to limit entry
– where deterrent actions can be identified
• Incumbent may not take any action if entry is not likely
or if there is little it can do to stop entry.
• Incumbent may take action if entry is fairly likely in an
effort to limit the number coming in
• Efforts by incumbent to build brand loyalty may seem
like entry deterrence but it may instead result in
incumbent not needing to price aggressively when rival
enters, which makes entry easier
Chapter 9: Entry Deterrence and
Predation
35
Empirical Application: Entry Deterrence
In The Pharmaceutical Industry 2
• Ellison and Ellison (2006) try to overcome these issues
in a recent study of the pharmaceutical industry
• They look at the advertising behavior of companies in
the case of 64 drugs about to lose their patents
• Their first step is to identify those markets where entry
following patent expiration was likely
Chapter 9: Entry Deterrence and
Predation
36
Empirical Application: Entry Deterrence
In The Pharmaceutical Industry 3
• For this purpose, they estimate the equation:
Entryi =constant + β1Revi + β2Hospi + β3Chronici + εi
– Entryi is a 1,0 dummy variable equal to 1 if there was entry
within three years after patent expiration in market i
– Revi is the average revenue earned by the incumbent in market i
for the three years before expiration
– Hospi is the fraction of drug revenues paid by hospitals
– Chronic is 1 if the drug treats a chronic/acute condition and 0
otherwise
• Because they estimate a Probit equation, the results give the
probability of entry as a function of market features
• They identify three market types where the probability of entry is: 1)
low; 2) medium; and 3) high
Chapter 9: Entry Deterrence and
Predation
37
Empirical Application: Entry Deterrence
In The Pharmaceutical Industry 4
• What is predatory or entry-deterring behavior here? Ellison
and Ellison focus on advertising.
• They note that when one firm advertises a prescription
drug, the benefits of that advertising spill over to rivals
• The decision-makers with respect to pharmaceuticals are
doctors. They will have a keen sense of the chemical identity
of generic competitors.
• If an advertisement trumpets the ability of a a particular
hydrochloride to alleviate depression, doctors understand
that near identical hydrochlorides will have the same effect
• In this environment, an incumbent wishing to deter entry
may advertise LESS
Chapter 9: Entry Deterrence and
Predation
38
Empirical Application: Entry Deterrence
In The Pharmaceutical Industry 5
• Ellison & Ellison (2006) estimate the following equation:
Advertisingit
– 1 = [1Low + 2Medium + 3High]Time + i
Advertisingi
• The dependent variable measures advertising for each
firm i in each of the 12 months just prior to the patent
expiration relative to the average level of advertising in
each of the 24 months before that
• The dependent variable measures advertising for each
firm i in each of the 12 months just prior to the patent
expiration relative to the average level of advertising in
each of the 24 months before that
Chapter 9: Entry Deterrence and
Predation
39
Empirical Application: Entry Deterrence
In The Pharmaceutical Industry 6
• On the right-hand-side, the coefficient on the time to patent
expiration variable depends on what probability of entry
category for that market
• Ellison and Ellison argue any entry-deterring efforts will
only occur in the middle group
– No need to deter entry when probability is Low
– Impossible to deter entry when probability is High
• So, given our understanding of entry deterrence, advertising
should be low in Medium profanity markets
• That is, 2 should be negative
Chapter 9: Entry Deterrence and
Predation
40
Empirical Application: Entry Deterrence
In The Pharmaceutical Industry 7
• Ellison & Ellison (2006) estimates are shown below
Advertising Intensity Time Trend By Category of
Entry Probability, 64 Pharmaceutical Markets
Coefficient
1
2
3
Estimated
Value
Standard
Error
-0.007
0.013
-0.032
0.009
0.009
0.007
• 2 is significantly negative. Firms facing a medium
probability of entry after patent expiration do reduce their
advertising in the months prior to that date
• Some mild support for entry deterring behavior
Chapter 9: Entry Deterrence and
Predation
41