Transcript Predatory Conduct - College of William & Mary

Predatory Conduct
• What is predatory conduct?
– Any strategy designed specifically to deter rival
firms from competing in a market.
– Primary objective of predatory conduct is to
influence the behavior of rivals.
• For an action to be seen as “predatory” it
must only be profitable if it causes rival to
exit market or deters a potential entrant.
Review of Dominant Firm and
Competitive Fringe Model
• One dominant firm in the industry.
– Acts as a price maker.
• Large number of small firms, the “competitive
fringe”.
– Act as price takers.
• Dominants firm moves first and sets the price.
• Fringe firms supply based on the price.
• Similar to Stackelberg, except followers don’t
affect price.
• Dominant firm’s demand is:
DD(P) = D(P) - n*S(P)
Price
DD(P)
N*S(P) =
Fringe Supply
At prices above
this point, fringe
supplies everything
D(P)
Industry Quantity
• Dominant firm maximizes:P*DD(P) - cD(DD(P)).
– Sets MR = MC
Price
N*S(P) =
Fringe Supply
MRD(P)
MCD
DD(P)
qD
D(P)
Industry Quantity
• Fringe supplies based on price set by dominant
firm.
Price
N*S(P) =
Fringe Supply
MRD(P)
MCD
P*
DD(P)
qD qF
D(P)
Industry Quantity
Implications of Dominant Firm
and Competitive Fringe Model
• Dominant firm supplies where MRD = MCD.
– In some cases this is greater than quantity
monopolist would supply, and in some cases less.
– Price will always be less than monopolist’s price
would be.
– Why? Because competitive fringe serves to make
dominant firm’s demand more elastic, so the firm
has less power to price above marginal cost.
• Note that the dominant firm does not drive the
fringe out of the market in this model.
Repeated Version Dominant Firm
and Competitive Fringe Model
• What if there was more than one period?
• Dominant firm could kill competitive fringe by
pricing so low that fringe would not produce.
– In a one-shot game, generally doesn’t maximize
current profits, and therefore not done.
• Once fringe dies, dominant firm can price at
monopoly level.
• Profitability of plan depends on cost of killing
fringe and relative size of monopoly profits.
Summary of Limit Pricing and
Quantity Commitment Model
• Incumbent in the market acts as a Stackelberg
leader and chooses an output level.
• Potential entrant sees incumbent’s quantity and
then decides whether to enter.
• Key assumption: entrant believes that its entry
decision will not affect the leader’s output
choice.
• By picking output level, incumbent can
manipulate potential entrant’s profit from entry.
Price
Residual demand
for PE
D(P)
qL
Industry Quantity
At q*, entrant’s profit is negative
Price
DPE
MRPE
MCPE
ATCPE
D(P)
q*
qL
Industry Quantity
Critiques of Limit Pricing and
Quantity Commitment Model
• Will incumbent really produce at qL once the
entrant is in the market?
• Only if there is someway he can commit to this
level, otherwise the two firms will split the
market as in Cournot.
• If there is no way to commit, entrant will not
believe the incumbent’s threat -- it is not
credible.
Credibility of Threats
• “Threats” are actually just statements about
what players will do in future rounds.
• For a threat to be credible, it must be optimal for
the person making the threat to carry it through.
Example
Entrant
Stay Out
Enter
Low P
Low P
High P
Incumbent
High P
2,2
-1,0
0,5
0,0
•Find optimal strategy for each subgame (prune the tree).
•Find Entrant’s optimal action.
Chain Store Paradox
• Firm A has a store in each of 20 markets.
• In each market, there is a single local
potential entrant. (Different PE in each
market.)
• Currently none of the PE’s has enough
capital to begin operations, but in time they
will.
• How should Firm A price in this situation?
Chain Store Paradox, con’t
• If Firm A accommodates entry, each firm
has a positive profit although A > PE.
– Think Cournot with heterogeneous costs.
• If Firm A fights, he can price low enough so
that PE = 0.
– Think Bertrand with heterogeneous costs.
• However if A fights, profit is less than if A
accommodates.
– Assume A will have to maintain low price to
keep PE out of the market.
Chain Store Paradox, con’t
• Should Firm A price low in the “first”
market (i.e., market where entry occurs
first) and drive the competitor out?
– Will lose money, but this market will serve as
an example for the other PE’s. “Proof” that A
will fight.
• Dynamic game -- must work backwards.
• In the “last” market, Firm A will not price
low because that decreases total profit.
– Dominant strategy is to accommodate entry.
Chain Store Paradox, con’t
• In the “last” market, Firm A accommodates.
• In the next to the last market, no need to
prove threat to PE in last market, since A
will always accommodate. Therefore, Firm
A should also accommodate the PE in the
next to the last market.
• And so on…
• Thus the “paradox”: even in a chain of
markets, predatory threats aren’t credible.
Critiques of Chain Store Paradox
• Requires a fixed number of markets.
• If there are an infinite number of markets,
or even just the possibility of additional
markets, you can find situations under
which predatory action is credible.
• In such a case, a firm may want to develop a
reputation as a tough competitor.
Capacity Expansion to Deter Entry
• aka the Dixit Capacity Expansion Model.
• Same basic setup: one incumbent firm and
one potential entrant.
• Incumbent decides how large to build its
plant (i.e., how much capacity to build).
– With a plant of size K, the incumbent can
produce up to K units at a marginal cost of w.
– To produce more than K units, he faces an
additional MC of r for each unit above K.
Incumbent’s Marginal Cost
w+r
w
K
Quantity
Capacity Expansion, con’t
• It costs the potential entrant F to enter the
market.
• If the PE enters, the firms choose quantity
as in a Cournot game.
• Since the PE must build his capacity and
produce simultaneously, he faces a MC of w
+ r.
• If the PE doesn’t enter, the incumbent acts
as a monopolist.
Capacity Expansion, con’t
• In a Cournot game with two firms, quantity
produced is a function of the firms, MC.
– BR for firm i: qi = (A+cj-2ci)/3b.
• As long as the incumbent produces less than
K, he has a lower MC, and thus will
produce more than the entrant and make a
larger profit.
Best Responses of the Two Firms
qPE
For output less than K, incumbent
has lower MC and is on this BR
curve
For output greater than K,
incumbent has higher MC and is
on this lower BR curve
q*PE
K
= q*I
qI
Capacity Expansion, con’t
• By increasing K, the PE’s optimal quantity
(and profit) is decreased, which makes entry
less profitable.
• In some cases, it may not be profitable for
the PE to enter at all (if he can’t cover F).
• Is the threat of the incumbent producing a
high quantity of output credible?
– Yes. It is his Best Response.
• How does the incumbent pick K?
Finding the Optimal K
qPE
Max. PE
will produce
Min. PE will
produce
Minimum that
Incumbent will
produce if PE enters
Maximum that
Incumbent will
produce if PE enters
qI
Monopolist’s
optimal quantity
Can the incumbent keep the entrant out?
qPE
Max. PE
will produce
Depends on the PE’s “break even” quantity.
If break even q above this
quantity, PE will never enter
If break even q below this
quantity, PE will always enter
If break even q in this range,
choice of K is critical
Min. PE will
produce
qI
Capacity Expansion, con’t
• If the incumbent picks K* so that the BR for
the PE would be just below the break even
quantity, the PE will not enter the market.
• If K* > M* (the monopolist’s optimal
quantity) the strategy is predatory.
• If the K* < M*, the incumbent will build
capacity equal to M*, as this is the level at
which he will produce. This is not
predatory, but is termed “blockaded entry”.
Extensive Form Capacity Expansion Game
Incumbent
Low K
Potential Entrant
High K
DNE
DNE
Enter
Enter
6,0
L
3,1
H
1,0
H 1,-1
0,-2
L
I
5,0
L
1,1
H
0,0
H 2,-1
1,-2
L
PE
Version 2
Incumbent
Low K
Potential Entrant
High K
DNE
DNE
Enter
Enter
35,0
6,0
L
4
L 3,1
H
1,0
H 1,-1
0,-2
I
L
1,1
H
0,0
H 2,-1
1,-2
L
PE
Final Comments on the Capacity
Expansion Model
• If the capacity cost is not sunk, if it can be
recovered, then the threat is not credible.
• Model is consistent with evidence that early
firms maintain market share -- early firms
are able to make capacity commitments that
give them Stackelberg leadership role.
• In several antitrust cases, firms have been
found guilty of attempting to monopolize a
market by expanding capacity.
Limit Pricing and Imperfect
Information
• Assume there is imperfect information, that
is the potential entrant does not know about
the incumbent’s true cost and efficiency.
• It may be possible for the incumbent to
“fool” the potential entrant with his pricing
and discourage the entrant from entering.
Limit Pricing con’t
• Incumbent is a low cost firm with
probability  and is a high cost firm with
probability (1-).
• PE knows the probabilities, but not what the
incumbent’s cost actually is. PE is a high
cost firm for sure.
• In first period, incumbent prices. After
seeing price, PE decides whether to enter.
• Once PE makes entry decision, incumbent
prices based on actual cost.
Limit Pricing Game
Low cost, l
Nature
High cost, h
P=
Incumbent
P*(l)
P = 1-
P*(h)
P*(l)
P*(h)
PE
E
DNE
E
DNE
E
DNE
10+5,-2 10+10,0
E
6+2,2
5+5,-2
5+10,0
4+2,2
4+6,0
DNE
6+6,0
Limit Pricing con’t
• If incumbent is a low cost firm, pricing low will
always provide as much profit as pricing high,
so he will always price low if he is low cost.
• Since the incumbent will only price high if he is
a high cost firm, if PE sees high price, he
assumes high cost and enters.
• However, incumbent may try to masquerade as
a low cost firm, so if PE sees a low price, he
knows the incumbent could be bluffing.
Limit Pricing Game
Low cost, l
Nature
P=
Incumbent
P*(l)
High cost, h
P = 1-
P*(l)
P*(h)
PE
E
DNE
E
10+5,-2 10+10,0
DNE
E
6+2,2
4+2,2
4+6,0
DNE
6+6,0
Limit Pricing con’t
• When PE sees a low price, he doesn’t know
what costs are.
• Expected value from entry given a low price
depends on the probabilty of each state:
– (-2) + (1-)(2).
• If  > 0.5, entrants stays out, otherwise enters
when he sees a low price.
Limit Pricing Game
Low cost, l
Nature
P=
Incumbent
P*(l)
High cost, h
P = 1-
P*(l)
P*(h)
PE
E
DNE
E
10+5,-2 10+10,0
DNE
E
6+2,2
4+2,2
4+6,0
DNE
6+6,0