You are on page 1of 20

Strategic Buyers and Exclusionary Contracts

Author(s): Robert Innes and Richard J. Sexton


Source: The American Economic Review , Jun., 1994, Vol. 84, No. 3 (Jun., 1994), pp. 566-
584
Published by: American Economic Association

Stable URL: https://www.jstor.org/stable/2118068

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

is collaborating with JSTOR to digitize, preserve and extend access to The American Economic
Review

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
Strategic Buyers and Exclusionary Contracts

By ROBERT INNES AND RICHARD J. SEXTON*

This paper characterizes equilibrium exclusionary contracts between buyers, an


incumbent firm, and a potential entrant when buyers can either vertically
integrate or contract with the outside entrant. In this setting, exclusionary
contracts are generally shown to be efficient and to deter inefficient entry that
would otherwise occur. With multiple unorganized buyers, equilibrium contracts
are shown to take a "divide-and-conquer" form and, in some cases, to deter
some efficient entry. In such cases, efficient contracting can be restored by a
policy that prohibits price discrimination and gives agents free reign to sign
exclusionary agreements. (JEL L10, L40, K12)

In recent papers in this Review, Phillipe that the AB and RRW results hinge upon
Aghion and Patrick Bolton (1987) hence- somewhat limited views of the strategic in-
forth (AB) and Eric Rasmusen, J. teractions that might take place among mar-
Ramseyer, and John Wiley (1991) (hence- ket participants in an entry game. Specifi-
forth RRW) have demonstrated that exclu- cally, we consider market settings where
sionary contracts between an incumbent firm buyers have free reign to form coalitions
and its customers may reduce economic among themselves and to contract with ei-
welfare by deterring otherwise desirable en- ther the incumbent or an entrant .2 In these
try. Both papers show that rational con- settings, buyers have incentives to elicit en-
sumers will sign exclusionary agreements try in order to create competition, even
with incumbent firms even though such when such entry is socially inefficient. How-
agreements are not in society's collective ever, we find that exclusionary contracts ac-
interest. That is, by diminishing the proba- tually prevent inefficient entry from occur-
bility of outside entry, the individual agree- ring and need not deter efficient entry. Thus,
ments impose a negative externality upon results quite opposite from AB and RRW
other consumers and potential entrants. are obtained, and the desirable efficiency
These papers pose a rigorous challenge to properties of exclusionary contracts posited
the view held by "Chicago School" scholars by Posner and by Bork are restored.
such as Richard Posner (1976) and Robert
Bork (1978) that exclusionary contracts must
be efficient or they would not be signed by
buyers.1 However, in this paper we show

Marius Schwartz (1987), who focus on exclusive dealing


in retailing (see also Stanley Ornstein, 1989; Howard
Marvel, 1982).
* Department of Agricultural and Resource Eco- 2Other recent papers have also recognized the im-
nomics, University of Arizona, Tucson, AZ 85721, and portance of strategic buyers in entry games. Michael
Department of Agricultural Economics, University Katz (1987) and David Scheffman and Pablo Spiller
of California, Davis, CA 95616. We are indebted to (1992) study the effects of buyer vertical entry on
two anonymous reviewers, Giacomo Bonanno, Garth incumbent-seller pricing, and Sexton and Terri Sexton
Holloway, and Jeff LaFrance for invaluable comments (1987) study similar phenomena when buyers can jointly
on this paper. In addition, we owe special thanks toenter a market as a cooperative producer. Tore
Tom Hazlett for tireless advice on the cable TV indus-
Ellingsen (1991) examines the effects of buyer partici-
try. The usual disclaimer applies. pation in rent-seeking/lobbying games. We extend
1This view has also been debated in the recent these lines of inquiry by studying the effects of poten-
exchanges of William Comanor and H. E. Frech (1985, tial strategic buyer behavior on the structure and effi-
1987), Frank Mathewson and Ralph Winter (1987), and ciency properties of equilibrium exclusionary contracts.
566

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
VOL. 84 NO. 3 INNES AND SEXTON: STRATEGIC BUYERS 567

Welfare effects of entry and entry-deter- example, the entrant's ex post cost of entry
rence behavior are driven by two related and production is random ex ante, implying
phenomena. First are potential pricing in- that entry is efficient only when the realized
efficiencies due to imperfect competition, value of this cost is sufficiently low. The AB
the extent of which depends on the amount paper shows that an equilibrium exclusion-
of entry. These inefficiencies arise when de- ary contract between an incumbent seller
mand is not perfectly inelastic, seller(s) are and a buyer permits these two players to act
unable either to price-discriminate or offer as a cartel with respect to an entrant, mak-
price-quantity contracts, and price diverges ing commitments that extract some of the
from marginal cost. In the RRW model, for entrant's expected surplus. Unfortunately,
example, efficiency losses from "naked ex- this surplus extraction reduces the entrant's
clusion" are due to these deadweight costs incentive to enter and thereby deters some
of monopoly pricing.3 If price-quantity con- efficient entry.
tracts were permitted in the RRW analysis, A central purpose of our analysis is to
naked exclusion would still occur, but this consider whether this important result is
exclusion would represent an efficient ex- robust to a generalized contracting environ-
traction of surplus by the monopolist. ment wherein buyers can behave strategi-
Because price-quantity contracts are cally and elicit entry in response to poten-
likely to be possible when exclusionary tial monopoly or seller cartel behavior. A
agreements are possible, we focus on a sec- buyer-entrant coalition, achieved through
ond source of entry welfare effect, namely, either vertical buyer integration or contract-
cost differences between an entrant and an ing between a buyer and an outside entrant,
incumbent, both due to fixed setup costs for internalizes any beneficial effects from entry
an entrant and differences in post-entry that accrue to the customer. Thus, the ben-
production costs.4 In the AB model, for efit from entry to a buyer-entrant coalition
will typically exceed the benefit accruing to
an ordinary outside entrant. As a result, a
buyer-entrant coalition will elicit entry
3In RRW, an absence of exclusionary contracts leads when it is socially inefficient, that is, when
to entry and production by two firms who will competi- the net buyer-entrant gains are exceeded
tively price their output at their common minimum by the incumbent seller's profit losses. Our
average cost (and marginal cost) of output. However, analysis shows, however, that equilibrium
when the potential entrant has a minimum efficient
exclusionary incumbent-buyer contracts de-
scale of operation, the incumbent monopolist can offer
a "naked" exclusionary monopoly-price contract which ter this inefficient entry without harming
consumers will sign. Knowing that other consumers will buyers. Thus, buyers' incentives to elicit in-
sign the agreement, each consumer also knows that efficient entry impart an efficiency motiva-
there will be too few "free" (nonsigning) consumers for
tion to exclusionary contracts that has been
entrant production to be profitable. Thus, the monopoly
outcome will be achieved in the final analysis, regard-
absent in prior analyses.
less of the individual consumer's choice, and a con- We develop these points in Section I with
sumer will only require an infinitesimal "signing bonus" a simple entry game among an incumbent,
to elicit his acceptance of the exclusion agreement. entrant, and single customer. One interpre-
4When firms incur fixed setup costs, too much or
tation of the single-buyer model is that con-
too little entry may occur relative to the social opti-
mum (C. C. von Weizsacker, 1980; Gregory Mankiw sumers have merged into a buying coalition
and Michael Whinston, 1986; David Mills, 1991). The or have contracted with a joint purchasing
reason is that an entrant considers only the profit effect agent. However, analyzing the full implica-
from his entry decision, ignoring potential gains to
tions of buyers' coalition-formation possibil-
consumers and losses to incumbent firms. These exter-
nal effects are offsetting, typically leading to inefficient
ities requires a setting where consumers are
entry decisions when demands are elastic and post-en- unorganized initially but have the poten-
try cartel or Cournot behavior is assumed. In the AB tial to form coalitions. We develop this an-
model, which we also follow, these external costs of alysis in Section II, where we show that
entry are not present because demands are inelastic, ex
the incumbent firm will exploit identical
post Bertrand competition is assumed, and thus the
monopolist's profit losses from entry are exactly offset buyers' initial disorganization by offering
by consumers' surplus gains. discriminatory-price contracts that "divide

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
568 THE AMERICAN ECONOMIC REVIEW JUNE 1994

and conquer" the customers but are capability to contract with a potential en-
nonetheless constrained by buyers' ability to can dramatically increase the con-
trant
form a coalition that can, in turn, contract sumer's payoff. It follows, therefore, that
with a potential entrant or vertically inte- the customers have incentives to seek out an
grate.5 These contracts deter inefficient en- entrant (rather than vice versa as RRW's
try that would occur in their absence; in logic implies) to introduce competition into
some cases, they also deter some efficient the contracting process.
entry. In such cases, efficient contracting Examples of markets with vertical expan-
can be restored by a policy that prohibits sion possibilities are also numerous, particu-
price discrimination and gives agents free larly among intermediate good purchasers.
reign to sign exclusionary agreements. Vertical expansion may occur either unilat-
Are the strategic buyer contracting and erally (by a single firm) or jointly (by a
coalition-formation possibilities consider- cooperative firm).8 The vertical integration
ed here realistic? Recall that entry by a that we study is taken to avoid monopoly
buyer-entrant coalition is generally more power and, therefore, is unlikely to be chal-
profitable than ordinary outside entry and, lenged under U.S. antitrust laws.9 Similarly,
hence, may be more likely to occur. More- a cooperative firm is permitted under U.S.
over, examples of buyer coalitions abound, antitrust laws unless its effect is monopoliz-
including group purchases of insurance, ing (Northwest Wholesale Stationers v. Pa-
buyers' clubs, joint purchasing arrange- cific Stationery&Printing Co., 1985). Coop-
ments by retailers, and common procure- eratives in agriculture are also accorded
ment of stock by independent grocers.6 In specific protection from the antitrust laws
addition, homeowner or neighborhood asso- under the Capper-Volstead Act (1922) and
ciations can play the role of buyer coalitions Section 6 of the Clayton Act (1914).1o
as they do, for example, in the procurement In what follows, we study these poten-
of cable television services. In Section III tial buyer behaviors in the simplest pos-
below, we discuss the cable TV case in sible buyer-incumbent-entrant contracting
some detail. Jonathan Jacobson and Gary games that enable the key interactions and
Dorman (1991) conclude that buyer coali- coalition-formation incentives to be expli-
tions, unless organized for monopsony pur- cated.
poses, have been supported by U.S. courts.
With respect to buyer-entrant agree- I. Efficient Exclusion With Buyer-Entrant
ments, RRW argue that entrants may be Contracting
less able to use exclusionary agreements
than an incumbent firm because, for exam- We consider an Aghion-Bolton-type mod-
ple, an entrant may lack information on el with three players: (i) an incumbent
customers' identities or other characteris-
tics.7 However, we show that a consumer's

8Cooperative firms are most common in agriculture,


5Our results are distinguished from the extant liter-where in the United States they handle 27 percent of
ature on price discrimination (e.g., see Hal Varian, farm input purchases and 31 percent of farm product
1989) in that entry-deterrence considerations motivate sales (Roger Wissman, 1986). Consumer food, housing,
third-degree price discrimination despite the presenceand service cooperatives are also common in the United
of identical buyers. States and elsewhere (Robert Smiley and Riccardo
6See Jonathan Jacobson and Gary Dorman (1991) Veraldo, 1983).
for further discussion. 9See Martin Perry (1989) for discussion of these
7The legal standing of exclusionary contracts be- legal issues.
tween buyers and entrants is unlikely to be in dispute 10Section 6 of the Clayton Act states in reference to
because the charge of anticompetitiveness leveled "agricultural organizations" that "such organizations
against analogous incumbent-buyer agreements is or the members thereof [shall not] be held or construed
based upon the agreements' potential to exclude rivals. to be illegal combinations or conspiracies in restraint
This argument is without merit when the exclusionaryof trade, under the antitrust laws." The Capper-
firm is an entrant attempting to establish a foothold inVolstead Act refers specifically to marketing rather
the market . than buying activities of farmers.

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
VOL. 84 NO. 3 INNES AND SEXTON: STRATEGIC BUYERS 569

monopolist M who has unit production costs time of contracting (e.g., stages 1 and 1') but
C = 2; (ii) a single buyer B who consumes becomes known at the time of the entry
either one unit of M's output good or none decision (by the definition of an opportunity
at all, and who has a reservation price for cost).
this good, v = 1; and (iii) a potential entrant Like AB, we assume in this section that
E who has unit production costs, ce, and an contracts contingent upon entrant costs can-
entry/setup cost, F, which are both random not be signed and that all agents are risk-
ex ante. For simplicity, we assume that ce neutral. Key features distinguishing our
and F are independent uniform random model from prior work are the opportuni-
variables with supports Ce E [0, 1] and F E ties for buyer-entrant contracting and the
distinct role for the entry cost F. The latter
Play occurs in the following sequence. In feature is important in our analysis because
stage 1 (time 0), M offers a product-purchase a buyer may elicit entry in order to reduce
contract to B, and B either accepts or re- the final price that he obtains, even when
jects the contract offered. Next, in stage 1' entrant production does not take place.
(time 1), B can offer a product purchase
contract to E, which E either accepts or A. The AB Setting
rejects. In stage 2 (time 2), ce and F are
realized, and then E either enters by paying Consider the above model when buyer-
F or does not enter. If E does not enter, entrant contracting is ruled out (i.e., there is
then he cannot produce output in the sub- no stage 1'). This is the AB model. Without
sequent stage of the game, and in addition, any contracts from stage 1, stage-2 entry
any B-E contract from stage 1' is automati- then leads to a stage-3 Bertrand price of
cally canceled. Finally, in stage 3 (time 3), P=max(ce, ); hence, entry occurs if and
production and trade take place. We as- only if it is efficient, namely, when Ce + F <
sume that B is free to sign contracts in both C = 2; the corresponding probability of entry
stages 1 and 1', deferring any contract liqui-is
dation decisions to stage 3. For simplicity,
we also adopt the following tie-breaking
conventions: (1) f 1/2 l/2-Fdce 2dF =1/4.
0 0

(i) if B or E is indifferent between signing


With no contracts and
(liquidating) a contract and not signing
the monopolist will ch
(not liquidating), he will sign (liquidate);
P = 1. Expected payoffs in this no-contract
(ii) E enters (in stage 2) if and only if E's
(NC) equilibrium are
gains from entry are strictly positive;
(iii) if B is indifferent between buying from
E and from M, he buys from E. VNC(M)= (1 - )(1-c) =3/8

We denote this structure game I. VNC(B) = O(v - 1/2) = 1/8


Our model captures the following stylized
features of incumbent-entrant interactions VNC(E) = E{max(1/2- Ce - F,O))
in the simplest setting possible; (a) an in-
cumbent has a first-move advantage; (b) an =1/24.
entrant must sink some investment prior to
competing or producing; and (c) an entrant's For this setting, AB show that M and B
costs are uncertain ex ante. AB motivate will sign a stage-1 contract that permits them
feature (c) by pointing to the stochastic ar- to extract some of the entrant's surplus.
rival of investment opportunities to poten- This contract takes the form (P, PO), where
tial entrant(s). In the presence of capital P specifies the price at which B will pur-
constraints, the opportunity cost of an entry chase from M, and PO specifies liquidating
investment is then random. This logic im- damages B must pay M if B elects to pur-
plies that the entry cost is uncertain at the chase from E. Because M moves first and

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
570 THE AMERICAN ECONOMIC REVIEW JUNE 1994

offers a "take-it-or-leave-it" contract to B, age payment above the level that would
M extracts the full surplus generated by the elicit entry when such entry is efficient (i.e.,
{M, B} coalition. M does so by setting Po = P - the incumbent gains the addi-
(P,Po)==, ), a contract which maximizes tional payment in all states of nature in
M's expected profit subject to B obtaining which entry occurs; this gain comes at the
at least his no-contract surplus, P' B ac- cost of inefficient entry-deterrence in some
cepts this offer and the following expected marginal entry state of the world, a cost
payoffs result: which is negligible at the efficient level of
the liquidating damages.
VAB(M) = 25/64
B. Ex Ante B - E Contracting
VAB(B) = 1/8
We now include stage 1' in game I, thus
and permitting buyer-entrant contracts that are
signed ex ante, before uncertainty about
VAB(E) = [(1/24) - (3/128)]. entrant costs is resolved. Two key points
emerge from the analysis: (i) exclusionary
The {M, B) coalition reduces E's profit contracts between M and B no longer deter
from entry, which in turn causes E to enter efficient entry; and (ii) such contracts deter
too infrequently (when ce + F E [0, P - PO)inefficient entry that would occur in their
= [0, ) vs. when ce + F E [0,-) under the absence. In our expanded paper (Innes and
no-contract equilibrium). Thus, the Sexton, 1992), these results are generalized
buyer-incumbent contract induces a socially to models with other buyer-entrant con-
inefficient outcome; {M, B) gain 1/64 over tracting and vertical-entry possibilities.12
the no-contracting equilibrium, but E loses To convey these points, we first consider
3/128 with a social loss of 1/128. The intu- buyer-entrant behavior in the absence of
ition for this inefficient entry-deterrence is any stage-1 M-B agreement. For this case B
quite simple: By raising the liquidating dam- anticipates a price of P = 1 in stage 3 if E
does not enter in stage 2. Therefore, from
the joint perspective of B and E, there are
potential gains from entry whenever
11 Under a (P,Po) contract (and no stage-2 contract-
ing), entry occurs if and only if the maximum price that
E can charge B (and elicit his custom), P - PO, exceeds
E's costs of entry and production, ce + F. That is, the (2) 1-F-min(ce, 1/2) > O.
probability of entry is O(P,PO) = Pr(ce + F < P - Po)
= max(O, [P - PO - 1/4]). Thus, M sets the contract to
solve the following expected-profit maximization prob- Given F < 2 condition (2) is always satis-
lem:
fied, implying that (i) the B-E coalition can
benefit from entry 100 percent of the time,
max4( )PO +[1- ()](P -1/2)
even though entry is only efficient 25 per-
subject to cent of the time and (ii) the potential B-E

(1- P) 2 1/8.

Since B obtains a payoff of 1- P with or without entry,


this last constraint requires that B obtain at least his
no-contract payoff. The contract (8,Z) solves this prob-12We have considered many variants of the current
lem, yielding an entry probability of 8 and a payoff to model, including those that permit (i) ex post B-E
M of contracts that are signed after uncertainty about en-
trant costs is resolved, (ii) both ex post and ex ante B-E
contracts, (iii) buyer vertical entry, (iv) buyer vertical
VAB(M) = (1- _)(P - C)+ kPo entry and ex post B-E contracts, and (v) alternative
specifications for the B-E contracting subgame (e.g., E
= (7/8)(3/8) + (1/8)(1/2) = 25/64. offers contracts to B, rather than vice versa).

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
VOL. 84 NO. 3 INNES AND SEXTON: STRATEGIC BUYERS 571

joint benefit from a stage-i' contract is:13 therefore

(3) E{max(1 - F -min(ce, 1/2), 0)) E{max(1- F -min(ce,1/2), O)}- TE


- vNC(B) - VNC(E) = 3/8- 1/3 = 1/24 = VNC(E).

3/8- 1/8- 1/24 = 5/24. Without a stage-1 M-B contract, B and E


will thus contract to elicit inefficient entry.
B can capture the gains in (3) by offering The total surplus cost of the inefficient en-
E a modified AB-type contract in stage 1'. try is n
One such B-E contract (among many) is
f1/2f J {min(ce, 1/2) + F - (1/2)} dCe2 dF
0 1/2-F
CE = (pE, POE, TE) = (1, 1/2, 1/3)
= 1/6.
where pE and PE are price and liquidating
damage parameters (now applying to E, These losses are four times as large as the
rather than M) and TE is a buyer signing maximum total expected benefits from ef-
bonus that E pays to B at the outset. ficient entry (1/24).
Under CE, B obtains a payoff of TE= This outcome contrasts sharply with AB,
1/3, which equals the sum of his no-con- wherein an absence of incumbent-buyer
tract surplus, VNC (B) = 1/8, and the maxi- contracts leads to entry if and only if en-
mal joint B-E surplus, 5/24. Clearly, B can- try is efficient. The difference between our
not obtain higher rents and still elicit E's analysis and that of AB is due to the possi-
contract acceptance. bility of buyer-entrant coordination that is
To show that E indeed obtains his re- modeled here. A buyer-entrant coalition
quired no-contract payoff of VNC(E) = 1/24 internalizes the consumer-surplus gains
-and that E will thus accept CE-consider from avoiding monopoly pricing in stage 3,
what happens in stage 3 if CE is signed in whereas this effect is external to an outside
stage 1' and E enters in stage 2. Since entrant who can capture only the efficiency
p =pE - = 2 = c is both the lowestgains pricefrom entry. Thus, entry will often be
that M can afford to offer B and the highest beneficial to the buyer-entrant coalition
price that induces B to liquidate CE, M will even when it is inefficient overall because
offer P = 2 in stage 3. By assumption, B will the buyer's surplus gains from entry will
thus liquidate CE. When c E < 2, B will also often be positive despite larger losses to M.
make a stage-3 price offer of P= 2 in order Of course, if entry costs were zero (F = 0),
this distinction would be unimportant since
to capture the efficiency rents, 2 - Ce. With
entry, E thus obtains an ex post payoff of entry (as distinct from production) would
never be inefficient.
Turning now to stage 1, we will show that
(4) MaX(p E , poE + [1/2 - c,]) - F equilibrium exclusionary incumbent-buyer
contracts restore efficiency. The incumbent's
=1- F-min(ceJ1/2) _!0 problem is to find a contract that, if possi-
ble, has the following properties: (i) it leads
namely, the maximal joint B-E gains from to efficient entry and production choices
entry. E's expected payoff under CE is (maximizing total market surplus, S), (ii) it
extracts all of the entrant's surplus [achiev-
ing V(E) = 0], and (iii) it gives the buyer
exactly his no-contract payoff, V(B) = 1/3.
13If F could take on values above c = ', the B-E
In other words, M will maximize his own
coalition could not benefit from entry 100 percent of
the time but could still benefit from much more fre- payoff, V(M) = S - V(E) - V(B), subject to
quent entry than is efficient. the most minimal constraints on entrant

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
572 THE AMERICAN ECONOMIC REVIEW JUNE 1994

and buyer rationality, V(E) > 0 and V(B) > stage-i' contract) is thus
1/3.
To extract the entrant's surplus, the in- SBE = E{max(P -p - Ce - F, 0)) - R.
cumbent can set a liquidating-damage pa- When this surplus is positive, B can capture
rameter in the incumbent-buyer contract it by offering E the contract
such that, without a stage-i' entrant-buyer
contract, the entrant earns a zero payoff. CE = (PB PO TB)
The entrant will then be willing to sign a
= (P-P P, + SB + R) E> 0
B - E contract that offers up all of his
prospective surplus. For example, with a contract which will elicit B's payment of R
(P, PO) as defined earlier, M can achieve and entry by E whenever Ce + F < P -POL15
this end by setting PO = P, which implies Returning now to the incumbent's prob-
that, without a stage-i' contract, E could lem, we see that efficient-entry choices will
charge at most pE= p - p = O in stage 3. be elicited whenever
However, if such (P, PO) features were also
(Sa) P-PO=c=1/2
to apply in the presence of a stage-i' con-
tract, E would never enter, and the incum-
and
bent's efficient-entry objective would not be
met. To overcome this problem, we implic-
(Sb) 0<R<E{max(P-PO -ce-F,O)}
itly make the liquidating-damage provision
depend upon whether or not an efficient = 1/24.
stage-i' contract is signed.
Specifically, consider incumbent-buyer The second condition in (5) ensures that
contracts of the following form: SBE is positive, so that the incumbent-buyer
contract CM indeed permits some entry.
The first condition ensures that the stage-i'
CM = (P, PO, T, PoL, R) contract elicited by CM leads to entry when
and only when it is efficient (i.e., when
where T is a signing bonus which M pays to Ce + F < c = ). Assuming (5) is satisfied
B at the outset and, in order to extract E's
(and hence, SBE = E{max(1/2- Ce - F, 0)}
surplus, PO = P. The (PoL, R) contract fea-
- R = 1/24 - R 2 0), the buyer's payoff un-
tures give B the option to reduce the der CM is
liquidating-damage amount from PO to PoL
< PO. In order to exercise this option, B (6) V(B) = (1-P) + SBE + T 2 1/3
must pay M the amount R (for "reduce").
(1-P) - R + T ? 7/24
Furthermore, this option is only available to
B before entry has occurred (i.e., before where the inequality represents the con-
stage 2 unfolds). tract-acceptance constraint faced by the in-
To understand how this contract works, cumbent in stage 1. In sum, if the incum-
consider the stage-i' B-E contracting game bent offers a stage-1 contract (P, PO =
that it elicits. In order for entry ever to be F,T, FPLI R) which satisfies condition (5) and
in B and E's collective interest, B must satisfies condition (6) with equality, he
invoke his PO-reduction option by paying M
the amount R. Assuming R is paid, B and
E can gain from entry whenever ce + F <
P - poL. 14 The maximum collective B-E sur- Since pE=p_PO and PE > Po > PoL, E will ob-
tain B's custom if (a) CE is signed and (b) E enters.
plus from a stage-i' contract (vis-'a-vis no Thus, CE gives E an "entry payoff" of P - pL- Ce - F
and thereby elicits entry when Ce + F < P - PoL If B
signs CE and does not pay R, entry by E will require B
to pay Po < PE. Since the expected value of this cost,
14Here we need not distinguish between entryP and
[Pr(ce + F < P - PL)]> E{max(P - PL-Ce-F,O)},
entrant production because contracts will be designed is greater than R (with SBE > 0 by assumption), B will
to elicit entry only when entrant production is efficient. exercise his PO-reduction option after signing CE.

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
VOL. 84 NO. 3 INNES AND SEXTON: STRATEGIC BUYERS 573

TABLE 1-EQUILIBRIUM PAYOFFS UNDER ALTERNATIVE CONTRACTING REGIMES

Monopolist Buyer Outside Probability Efficiency


Row Contracting regime (M) (B) entrant (E) of entry loss

A. Nonstrategic Buyer:

1 No contract 3/8 1/8 1/24 1/4 0


2 AB contract 25/64 1/8 1/24-3/128 1/8 1/128

B. Strategic Buyer (Ex Ante Buyer - Entrant Contracting):


3 No M-B
contract 0 1/3 1/24 1 1/6
4 M-B contract 5/24 1/3 0 1/4 0

will achieve the objectives set out at the be- Table 1 summarizes equilibrium payoffs
ginning of this argument, maximizing his and social losses from the alternative con-
expected profit. An example of such a tracting possibilities considered here. The
contract is contained in the following table indicates the buyer's payoff gain from
proposition. B-E contracting opportunities. The buyer's
incentive to create these opportunities-and
PROPOSITION 1: There is a subgame-per- attendant countervailing power-is thus
fect equilibrium to game I that is character- clear.
ized by the following behaviors: (i) In stage 1, Table 1 also indicates possible effects of
M offers and B accepts the contract policy measures that preclude exclusionary
contracts between (i) M and B, (ii) B and E,
CM (P, Po, T, PO, R) or (iii) both. If all exclusionary contracts are
prohibited (row 1), then an efficient equilib-
=(1/2, 1/2, - 1/6, 0, 1/24). rium will prevail, but the buyer will lose
relative to the no-regulation case (row 4).
(ii) in stage 1', B offers and E accepts the
However, suppose instead that M-B exclu-
contract
sionary agreements are precluded, while B-E
contracts are permitted as a way to allow an
CE = (PE, PO, TE) = (1/2,1,1/24).
entrant to establish a foothold in the mar-
At this juncture, B also invokes the option to ket. In this case, inefficient entry will take
reduce the M-B contract-liquidating damage place, leading to a net social loss of 1/6
amount from PO = 1/2 to PO = 0 by paying (row 3). Here, the monopolist's interests
M the amount R = 1/24. (iii) In stage 2, E coincide with society's, both opposing uni-
enters if and only if entry is efficient (i.e., lateral prohibition of exclusionary M-B
ce + F < 1/2). (iv) In stage 3, if no entry has agreements. On the other hand, if only B-E
occurred, then B invokes his stage-1 contract contracting is precluded, with M-B contract-
and purchases one unit of the product from ing permitted, the AB equilibrium is ob-
M at the price 1/2. If entry has occurred, tained, leading to the deterrence of efficient
then B invokes his stage-l' contract and pur- entry, buyer losses, and incumbent-firm
chases from E at the price 1/2. (v) Expected gains (row 2). Clearly, this policy is inimical
payoffs in this equilibrium (and any other to consumer interests but is favorable to the
subgame-perfect equilibrium to game I) are: incumbent, who will thus seek its adoption.
Our application to cable television in Sec-
VI(M) = 5/24 tion III illustrates the judicial and legislative
routes incumbent firms have traveled in
VI(B) = 1/3 seeking to implement this type of policy.
The intuition underlying efficiency of
VI(E) = 0. equilibrium exclusionary contracts is both

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
574 THE AMERICAN ECONOMIC REVIEW JUNE 1994

simple and general. In the AB model, the panded paper (Innes and Sexton, 1992), the
{M, B} coalition can only extract surplus incumbent can gain by giving the buyer such
from the entrant by making commitments an incentive.17
that diminish the entrant's prospective ex
post profits from entry (i.e., the profits an- II. Exclusionary Divide-and-Conquer
ticipated at the time of the entry decision, Contracts
when uncertainty about entrant costs has
been resolved). However, when E and B can The prior section studied implications
sign a contract ex ante, before uncertainty of a buyer ability to create countervailing
about entrant costs is resolved, this contract power for the structure and efficiency prop-
can be used to extract entrant surplus ex erties of equilibrium exclusionary contracts.
ante -and, hence, without distorting ex post Implicitly, this analysis was based on a
entry incentives. Proposition 1 illustrates premise that, if there were multiple buyers,
how M and B can structure their initial they were already organized in a coalition
contract so as to compel the entrant to offer with which the monopolist and potential
up his surplus at the time of B-E contract- entrant would deal as a group. However,
ing. Since the {M,B} coalition can thus in- multiple buyers may be disorganized ini-
ternalize the benefits of efficient entry tially, disorganization which the monopolist
choices, M and B will indeed structure their can exploit to advantage. In this section, we
own agreement so as to elicit such efficient explore the implications of buyers' disorga-
choices. It is important to emphasize that nization for the structure of equilibrium
the efficient-contracting results derived here contracts. However, in contrast to AB and
are robust to virtually any variant of our RRW, we also consider buyers' ability to
model in which ex ante B-E contracting is coordinate and act jointly. We show that the
possible.'6 monopolist can exploit both the initial buyer
We should also stress that ex ante B-E disorganization and economies of scale in
contracting possibilities are likely to be pre- potential buyer coordination by offering
sent in practice. As Mathewson and Winter contracts that discriminate among the buy-
(1987) indicate, offering and consumating a ers. In essence, the monopolist bribes some
contract does not lead, in and of itself, to buyers so that other buyers, who cannot
the incurrence of any sunken costs. Thus, anticipate that the bribed buyers will defect
there are no cost impediments to the from their contracts, can be exploited with-
contracting and competition in contracting out risk of their defection and organization
modeled here. Moreover, our results indi- into a costly small coalition. The monopolist
cate that either (i) the buyer has a strong
incentive to solicit an entrant with whom to
sign the ex ante contracts posited here, or
(ii) for cases which we consider in our ex- 17This second point applies when there are opportu-
nities for both ex ante B-E contracts and ex post B-E
contracts that are signed after uncertainty regarding
entrant costs is resolved. In our expanded paper (Innes
and Sexton, 1992), we find that the buyer does as well
16For example, efficient contracting persists underunder ex post B-E contracting alone as under the
any alternative stage-i' specification that permits E and ex ante B-E contracting modeled here. However, if
B to achieve an allocation in the core of their subgame. only ex post B-E contracting occurs, the payoff attain-
Hence, results are robust to games in which E makes able by the M-B coalition (and, hence, by M) is less
take-it-or-leave-it contract offers to B (rather than vice than under ex ante contracting for two reasons: (i) for
versa) or in which E and B engage in Nash bargaining. the same reasons as in AB, an equilibrium stage-1 M-B
In such alternative games, M can still design the stage-1 agreement will deter not only inefficient entry, but
contract so as to permit efficient stage-i' contracting some efficient entry as well; and (ii) M and B cannot
and to eliminate any further B-E rents; hence; all that reduce E's expected payoff to zero. Therefore, if both
changes (vis-'a-vis the current treatment) is B's no- ex ante and ex post B-E contracting are possible, M
stage-i-contract payoff and the corresponding T sign- can capture efficiency and entrant rents by constructing
ing fee that is required to elicit B's stage-1 contract a stage-1 agreement with B that induces B to sign an
acceptance. ex ante (rather than ex post) agreement with E.

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
VOL. 84 NO. 3 INNES AND SEXTON: STRATEGIC BUYERS 575

thus "divides and conquers" the consumers random, with ce still uniformly distributed
with discriminatory-price agreements. on [0,1]. Second, we suppose that there are
As in our single-buyer model, multiple two identical risk-neutral buyers, Bi and
buyers have incentives to create countervail- B2, who have reservation price v = 1 and
ing power. As a result, exclusionary con- who can vertically enter the market in the
tracts here will deter inefficient entry that following way: in stage 2, after ce is re-
would occur in their absence. Furthermore, vealed, the buyers can form either one one-
when the monopolist can either offer con- buyer coalition, two one-buyer coalitions, a
tracts that are contingent upon an entrant's two-buyer coalition, or no coalition. If a
costs or anticipate ex ante contracting coalition is formed, it pays the fixed entry
between buyers and a potential entrant, cost F and is able to produce at the per-unit
contracts can be designed to capture rents cost ce; otherwise, buyers cannot produce
from efficient entry. In these cases, the the product. For simplicity, the stage-2 buyer
monopolist will elicit entry when and only coalition-formation process is assumed to
when it is efficient. On the other hand, be a cooperative game, with a solution that
when entrant-cost-contingent contracts and is an element of the core.19 Thus, any solu-
ex ante B-E contracting are not possible, the tion to the subgame maximizes the sum of
monopolist will exploit buyers' initial disor- the two buyers' payoffs.
ganization to capture some of the buyers' Third, we will assume initially that stage-1
entry rents with contracts that deter some contracts have the following two properties:
efficient entry. In this case, exclusionary
agreements have two offsetting welfare ef- Property (i): The terms of a contract with
fects, deterring both inefficient entry and any one buyer cannot be contingent on the
some efficient entry. However, we show that decision of the other buyer to accept or
if price discrimination is prohibited, the reject his contract offer.20
monopolist will be prevented from pursuing
a divide-and-conquer strategy, and equilib-
rium contracts will be efficient, only deter-
19In our expanded paper (Innes and Sexton, 1992),
ring the inefficient entry that would occur we show that the cooperative stage-2 game structure
without them. This analysis thus finds moti- leads to qualitatively identical outcomes as does a
vations for both anti-discrimination policies plausible noncooperative coalition-formation game. In
this game, a "first-move" buyer, Bi, is chosen by a coin
and a laissez-faire policy toward exclusion-
flip at the start of stage 2. Bi then begins the game by
ary contracts. either entering (i.e., paying F) or not entering. If Bi
enters, he announces a fee for B2's membership in the
A. A Two-Buyer Game with Potential coalition, and B2 subsequently agrees to pay the fee or

Vertical Entry does not join the coalition. Finally, if B2 does not join
the coalition-or if Bi does not enter to begin
with-B2 either enters on his own or does nothing.
Implications of potential multiple-buyer 20Notably, AB allowed decision-contingent con-
coordination are most simply obtained in a tracts in their analysis of the multiple-buyer case. Our
two-buyer model wherein vertical buyer en- view is that decision-contingent contracts are poten-
tially very costly to implement. For example, under the
try is possible.18 Therefore, we consider the
equilibrium contracts in AB's two-buyer model, each
following model. First, for simplicity, we consumer obtains a lower price if the other consumer
assume that the entry cost F E (0,1) is non- has rejected his contract and, hence, will require veri-
fication of the other consumer's acceptance before
acquiescing to the high contract price. Costs of verify-
ing consumers' acceptance decisions by an impartial
18The qualitative implications of this analysis extend outsider thus seem to be an unavoidable consequence
to an environment in which buyers instead contract of their contingent contracts. We believe that these
with an outside entrant. Moreover, such B-E contract- implementation costs may explain their paucity in prac-
ing possibilities have qualitative implications in our tice and thus motivate their preclusion here. Nonethe-
multibuyer model that are no different than those less, we relax property (i) in our expanded paper (Innes
developed above for our single-buyer model. There- and Sexton, 1992) and, in doing so, find an efficiency
fore, we focus here on the simpler case of buyer motivation for a legal prohibition of decision-contin-
vertical-entry capabilities. gent contracts.

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
576 THE AMERICAN ECONOMIC REVIEW JUNE 1994

central theme, this game is designed to stress


Property (ii): Contract terms can be contin-
gent on the realization of ce.21 Thus we the potential strategic role that buyers may
consider contracts of the form, Ci = {Pi(Ce),
play in markets with potential entry.
PO(Ce)} for i = 1,2, where Pi and Po are Given our specification of stage 1, the
now contract price and liquidating damage incumbent's contract offers must satisfy two
functions. This property is assumed to hold sets of constraints in order to elicit buyer
because it permits the simplest derivation of acceptance decisions:
the divide-and-conquer contracting equilib-
rium that is generic to our models. Later in (i) With acceptance, each buyer must ob-
the section, we reverse property (ii). tain an expected payoff that is at least
as high as he would obtain with a unilat-
Finally, we impose the following structure eral rejection (and acceptance by the
on the stage-1 contracting game: M begins other buyer).
the game by posting simultaneous stage-1 (ii) If the first buyer, Bi, can anticipate that
contract offers to both buyers. This assump- B2 will reject his contract offer if Bi
tion is in accord with AB and section III in rejects, then Bi must obtain an ex-
RRW and can be motivated by an inability pected payoff at least as high as he
of the monopolist to specify expiration of would obtain with joint buyer rejection
his contract offers before all buyers have (i.e., no stage-1 contracts).
seen them.22 Once stage-1 offers are posted,
we assume that consumers make their ac- These constraints are the driving force
cept/reject decisions sequentially. Sequen- behind the divide-and-conquer monopoly
tial consumer choice at this stage in our strategy that emerges both here and in n
model is important because, in conjunction buyer generalizations of our model. In par-
with simultaneous seller offers, it gives eco- ticular, the monopolist must give the "first"
nomic force to the notion that buyers may buyer, Bi, a payoff that is as high as could
be able to implicitly coordinate their actions be obtained if all buyers were to reject
to their mutual betterment. If the buyers their stage-1 contract offers and proceed to
instead moved simultaneously in stage 1, stage 2. However, the next buyer only needs
there would be multiple subgame-perfect to be given a payoff that is as high as could
equilibria to the subgame inaugurated by be obtained if all buyers except BJ reject
the incumbent seller's contract offers. their contracts. The latter payoff will be
Among these equilibria, that which is most lower than that required by Bi because Bl's
preferred by both buyers (and, hence, which contract can be designed so that Bi con-
we believe is most plausible) is the unique tributes nothing to the prospective collec-
perfect equilibrium in the subgame with se- tive buyer-coalition payoff.
quential buyer moves. We call this structure We begin the analysis by stating an effi-
game II and note that, per our paper's ciency property of contracts that elicit
"accept" decisions by both buyers and, at
the same time, maximize the incumbent
seller's expected profit. Denoting the in-
cumbent's most-preferred "accept-eliciting"
21AB did not allow ce-contingent contracts. How- contracts by {C , C2 }, we have the following
ever, revelation of ce is particularly troublesome in proposition.
AB's model because the entrant is not a party to
contracts between M and Bi or B2. This problem is
not present when buyers are also the potential
PROPOSITION 2: In game II, the con-
entrant(s). Thus, for example, outside audits to verify tracts {C*, C2 } elicit the entry of a buyer
ce are plausible here. coalition when and only when such entry is
22In their section IV, RRW develop a model in efficient ex post (i.e., when ce < (1 - F)/2).
which contracts are posted and accepted or rejected
sequentially. RRW use sequential consumer choice as
a device to enable the seller to offer contracts contin- If buyer entry is ex post efficient, the
gent on other consumers' responses. incumbent seller can capture the gains

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
VOL. 84 NO. 3 INNES AND SEXTON: STRATEGIC BUYERS 577

under
from such entry (i.e., 2cthe {C,
- C)} contracts,
2ce - F) where
withPi is an
appropriate choice of ce-contingent liq- the average contract price.24
uidating damages. Similarly, if buyer entry To derive the above-mentioned con-
is ex post inefficient, the incumbent can straints on contract terms, suppose that
capture the gains from nonentry (i.e., F buyer j accepts his contract and buyer i
+2min(1/2, ce)-1). Thus, the incumbent rejects. Then the gain to coalition formation
will never leave deadweight losses from in- and buyer entry as of stage 2 is
efficient entry or nonentry "lying on the
table." 23
Since {C, C2*} must deter coalition entry (10) [1 Ce]

when it is inefficient [when ce (1 - F)/2],


we can assume (without loss of generality)
that the liquidating damages in these in- + [max(Pi(ce) - P(Ce), Ce)
stances are set sufficiently high to deter
such entry,
The first bracketed difference in (10) is the
gain to buyer i from coalition entry. The
(7) Po(Ce) = V = 1 VCe 2(1-F)/2 second bracketed difference is the potential
gain from coalition patronage of buyer j;
substituting for PA(Ce) from (7) and (8), this
i=1 ,2. difference is zero when ce > (1- F)/2 and
equals max(gj(F),0) when ce <(1 -F)/2.
Similarly, since {C*,C2*} must elicit coali- Thus, to obtain the minimum unilateral re-
tion entry when it is efficient, we can as- jection payoff to buyer i, R i, we set gj(F)
sume (without loss of generality) that the = 0, substitute (7) and (8) into (10), and
liquidating damages in these instances are then take the expectation of the maximum
set in the following way: of the expression in (10) (Bi's payoff with
stage-2 entry) and zero (Bi's payoff with no
stage-2 entry):
(8) Pi(Ce) = Pi(Ce) -PCe - gi(F)

(11) R min = fmax(1 - Ce - F,O) dce


VCe< (1-F)/2 i = 1,2

where {g1(F), g2(F) = F - g1(F)} is a rule =(1-F)2/2.


to allocate the entry cost F between B1 and
B2. Equation (8) implies that coalition entry
will occur with the indicated ce realizations. With gj(F) = 0, and hence gi(F) = F, the
Together, (7) and (8) reduce the contract- corresponding unilateral rejection payoff for
choice problem to one of choosing the P(ce)
functions and the cost assignment g1(F).
Also note that these liquidating-damage
terms lead to a buyer-i expected payoff of 24It is of interest to note that the seller prices
specified in {C*, C2*), Pi(c_), can be set constant with-
out loss of generality: P(ce) = Pi VCe; with this con-
tract specification, the buyers obtain the same payoff
regardless of the ce realization. If the buyers were
(9) 1- Pi(Ce) dCe =1_ pi
0 risk-averse, such a constant-payoff contract would be
uniquely optimal, thus confronting buyers with a choice
between a riskless contract and risky outcomes under
contract rejection; the buyers' aversion to the risk
inherent in contract rejection would thus permit the
23A proof of Proposition 2 is contained in our ex- incumbent to extract more buyer surplus than is possi-
panded paper. ble here.

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
578 THE AMERICAN ECONOMIC REVIEW JUNE 1994

yields greater per-buyer payoffs than unilat-


eral rejection). Hence B1 will also prefer
acceptance to unilateral rejection, even
(12) R max J(1 -F)12
when (1- e)dc
the other "unlucky" buyer, B2, ob-
tains liquidating-damage terms that maxi-
mize his contribution to the coalition fixed
+11
(1-F)/2
max(l-ce-F,
costs, g2(F) O)dce
= F. In sum, when Bi is offered
the R2-payoff contract, the only constraint
imposed on B2's contract is (13).
=(1-F)/2.

PROPOSITION 3: In game II, the incum-


At a minimum, the first set of constraints on bent's most-preferred "accept-eliciting" con-
incumbent contract offers requires that both tracts, {C*, C2j}, have the following proper-
buyers earn a payoff of at least R rni on ties:
their contracts:

(15a) B] ("lucky"buyer):
(13) 1_ pi PI2 R min i = 1,2.
pl = 1-R2 g1(F) = O
In addition, if constraint (13) binds for one
buyer, then the other buyer must obtain a
payoff of at lest R max on his contract.
(15b) B2 ("unlucky" buyer):
The final constraint facing M is deter-
mined by the expected payoff to a single p2 = 1-R7in g2(F) = F
buyer when both buyers reject their respec-
tive contract offers. Assuming a symmetric with liquidating damages set according to
division of the entry cost F, this per-buyer equations (7) and (8).
joint-contract-rejection payoff is:25
The incumbent gives Bi a payoff of R2 in
(14) R2 = E{max(1 - Ce- [F/2], O)} order to deter a two-buyer coalition and is
then free to give B2 a payoff that is just
= [1-(F/2)]22/2. sufficient to deter unilateral contract rejec-
tion. Further, the incumbent minimizes the
If the incumbent offers buyers contracts cost of deterring unilateral rejection by
that yield expected buyer payoffs that are specifying such large liquidating damages
less than R2, then B1 can anticipate that B2 for Bi that Bi never contributes to a
will reject if he (B1) rejects; hence, in this prospective one-buyer (B2) coalition.
case, B1 will anticipate a payoff of R2 with Since the {C*,C*j} contracts maximize to-
rejection and will himself reject. However, tal expected surplus (by Proposition 2) and
suppose that the incumbent offers one of just satisfy the minimal buyer-payoff re-
the buyers a contract with an expected pay- quirements, the incumbent cannot obtain a
off of R2. This "lucky" buyer, say Bi, will higher expected profit with any other con-
weakly prefer acceptance to the joint rejec- tract offers in this game. Thus, as claimed
tion outcome. Furthermore, note that R2 is above, there is no loss in generality from
greater than Rm' (i.e., two-buyer rejection considering the two-part contracts de-
scribed here, and (C> ,C} are indeed equi-
librium agreements in game II.
Despite the presence of identical buyers,
25The symmetric treatment of F in equation (14) we find that the equilibrium contracts here
follows from a very weak restriction on buyer beliefs, entail price discrimination (P l < p 2). More-
namely, that ex ante (as of stage 1), each identical
over, these contracts can take a form that
buyer believes he is equally likely to receive the high
and the low payoffs in any nonsymmetric stage-2 core is common in practice, namely, "tying"
allocation that can emerge with joint contract rejection. contracts with liquidating damage pro-

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
VOL. 84 NO. 3 INNES AND SEXTON: STRATEGIC BUYERS 579

visions. For the plausible case of constant a subset of market participants (e.g., M and
contract prices in our model (i.e., P(Ce) = B1) extracts rents from the outsider (e.g.,
P Vce, i = 1, 2), the liquidating damages for B2) using commitments that reduce the out-
our "lucky," low-price buyer are strictly sider's ex post surplus and thereby deter
higher than those for the "unlucky," high- some efficient entry.
price buyer whenever these damages are In our model, contracts also prevent in-
actually paid in the equilibrium (i.e., when efficient entry, which offsets the inefficient
Ce < [1 -F12). deterrence impact. Which of these effects
Without exclusionary contracts, buyer en- will be larger is unclear in general. How-
try occurs whenever the collective buyer ever, in the model analyzed here, with ce-
gains from entry, 2(1 - ce)- F, are positive contingent contracts precluded, the benefi-
(i.e., when ce < 1 - [F/2]). With the equilib- cial entry-deterrence effects of exclusive
rium exclusionary contracts, entry occurs contracts dominate the inefficient deter-
when and only when it is efficient, ce < rence effects.27 Thus, contracts in our model
(1/2)-(F/2). Thus, as in our single-buyer lead to gains in total net surplus even when
model, incumbent-buyer contracts deter in- ce is not observable.
efficient entry and thereby lead to gains in The argument for exclusive contracts is
social welfare. However, these gains are even more compelling when they are com-
achieved at the expense of one buyer (B2), bined with a statute to preclude price dis-
who obtains a lower payoff under his con- crimination. In this case, the monopolist
tract, R mn, then he would obtain if no con- must give both buyers the same contract
tracts were allowed, R2. The incumbent terms, and both buyers must thus obtain a
seller's expected profit increases by more payoff of R2 in order for either buyer to
than this loss AV(M) > R2 - R in > 0. accept the monopolist's offer. By the logic
of Section I, the no-discrimination contract
B. Non-Cost-Contingent Contracts which maximizes M's expected profit subject
to this buyer-acceptance constraint is
When the incumbent's contract offers
cannot be ce-contingent, they do not have
the efficiency property described in Proposi- (P, Po, T)
tion 2 for two reasons. First, the monopolist
can no longer capture any ex post rents = (1,1/2, R2 -[E{max(l-2ce -F,0)}/2]).
from efficient entry without altering the liq-
uidating damages in all states of nature for
which entry occurs; hence, the logic of This contract leads to entry when and only
Proposition 2 no longer applies. Second, in when it is efficient, thus maximizing total
stage 1, the monopolist's optimal "divide- market surplus. Furthermore, it gives the
and-conquer" strategy is, in essence, to form two buyers the minimum payoff required to
a coalition with the lucky buyer B1, with elicit acceptance, namely, R2.
whom he extracts rent from the unlucky Two policy implications are evident. First,
buyer B2. Thus, in the present model, M a no-discrimination statute eliminates inef-
and Bi act as a monopolist with respect to ficient entry deterrence, raising both social
B2, just as the incumbent and single buyer welfare and buyer surplus. Second, in the
acted as a monopolist with respect to the presence of a no-discrimination statute, the
entrant in the basic AB model discussed in only welfare effect of exclusionary contracts
Section I. In both cases, a coalition between is to deter inefficient entry, thereby increas-
ing social welfare without harming buyers.

26From equations (8) and (15), this statement can be


written formally as follows: P2 - F < Pl. Substituting
for Pl = (1- R2) and P2 = (1- R -in), and using equa- 27Details are available in our expanded paper
tions (14) and (11), this inequality can be established. (Innes and Sexton, 1992).

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
580 THE AMERICAN ECONOMIC REVIEW JUNE 1994

III. An Application In the United States, this industry has a


history of franchise monopolies awarded by
What industries may exhibit the types local governments and concomitant at-
of behavior described here? Key conditions tempts to regulate the price and services
for application are (i) structural concentra- provided by cable operators. Most regula-
tion and sunk entry costs on the selling tions for basic and premium service rates,
side so that seller market power is an issue however, were eliminated by Congress
and (ii) relatively few buyers with sufficient (effective at the end of 1986) under the
financial interest in the seller's product Cable Communications Policy Act of 1984
to make coalition organization and vertical (henceforth termed the "Cable Act"). Ana-
expansion realistic possibilities. Many lysts of the industry such as Hazlett
intermediate-goods industries satisfy these (1985,1986) and Adam Jaffe and David
criteria. Notably, these industries are the Kanter (1990) have argued that competition
focus of both antidiscrimination provisions provides an effective substitute for regula-
in U.S. law and most litigation concerning tion, but evidence of scale economies in
exclusionary deals. Consider, for example, cable (Eli Noam, 1985; Bruce Owen and
the case of Sewell Plastics v. Coca Cola et Peter Greenhalgh, 1986) suggest that direct
al. (1989). The plantiff in this case manufac- competition ("overbuilds," in industry par-
tured plastic bottles for a number of Coca lance) may be inefficient, an example of too
Cola bottlers in the southeastern United much entry.
States. The bottlers approached Sewell for Our analysis suggests that the trade-off
price concessions, armed with the threat to between competitive pricing and inefficient
form their own joint bottle manufacturing entry need not occur if participants are given
firm. Court records cite evidence that Sewell free reign to form coalitions and sign exclu-
"considered and pursued a strategy of of- sive contracts. What is the history of the
fering price concessions to certain of the cable TV industry concerning these prac-
defendent bottlers to make the 'investment tices? Buyer coalitions or their agents have
of self manufacture unattractive"' (p. 1210). often solicited an outside entrant to provide
The records further show that these "con- cable TV services (see Hazlett, 1985,1986).
cessions" were offered as part of a long-term A principal form for potential entry has
exclusive deal. been satellite master-antenna television
Applications are not limited to industrial (SMATV). SMATV operators often adver-
products. Consider cable TV. It is charac- tise their offerings in real-estate trade publi-
terized by significant sunk entry costs for cations, making both their availability as
wiring and transmission equipment; and potential entrants and the ex ante expected
when consumers are densely located in costs of SMATV service rather widely
apartment complexes, housing cooperatives, known. When entry has occured in cable
or planned developments, they may form TV markets, it has often involved exclusive
buyer coalitions at modest cost through their agreements between SMATV operators and
cooperative, homeowner, or condominium a buyer coalition. Exclusivity is a key tool
association, or through the initiative of pri- for buyer coalitions to solicit entry in this
vate developers or landlords acting as their type of market because it is a way to guar-
agents.28 antee the entrant's profitability. Without ex-
clusivity, buyers cannot credibly protect the
entrant from an aggressive response by the

28Developers and landlords can be expected to act


as effective agents for consumers in bargaining for the
provision of services such as cable TV since attempts to
extract monopoly prices for such services can lead
to lower values of the agent's housing unit (Thomas however, that franchising has not played this role for
Hazlett, 1985). City governments may potentially also two reasons: (i) regulation has not been undertaken
act as agents for their populace, in which case a fran- with the public interest in mind (Hazlett, 1991), and (ii)
chising process of the type proposed by Harold Dem- franchise bidding processes are vulnerable to oppor-
setz (1968) may represent an effective substitute for tunistic behavior by franchise winners who renege on
direct competition. Experience with cable TV suggests, contract provisions (Robin Prager, 1990).

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
VOL. 84 NO. 3 INNES AND SEXTON: STRATEGIC BUYERS 581

incumbent operator. This observation sug- monopolizing and, therefore, not in viola-
gests an important principle: the opportu- tion of the antitrust laws.
nity of buyers, via exclusive contracting, to Second, incumbent franchisees have at-
guarantee an entrant's profitability is an av- tempted to overturn exclusive arrangements
enue to achieving contestable market out- by promoting mandatory-access laws. Litiga-
comes in markets characterized by signifi- tion has evolved around interpretation of
cant sunk costs of entry. section 621(aX2) of the Cable Act, which
Active competition, however, is the ex- grants to cable systems access to certain
ception in the cable industry; only 2 percent public rights-of-way and easements. Cable
of U.S. households currently have access to incumbents have attempted to use this
multiple cable providers (Hazlett, 1994). provision to gain access despite exclusive
This outcome is not surprising given the deals between SMATV operators and
cost structure of the industry and our analy-buyer coalitions or their agents.30 Franchise
sis in Section II, which shows that an incum- operators have also promoted mandatory-
bent can deter entry when it is not socially access legislation at the state and municipal
efficient through a strategy of offering dis- level, achieving success in several states in-
criminatory divide-and-conquer contracts. cluding Delaware, Florida, Illinois, Massa-
Indeed, incumbent cable operators pursue chusetts, New Jersey, New York, and Rhode
this strategy through a practice known as Island.31 Notably, judges upholding rights to
bulk billing whereby members of apartment access under section 621(aX2) have relied
complexes, housing cooperatives, or condo- on Congressional intent to promote compe-
minium associations receive special "bulk" tition in cable TV through passage of the
rates.29 Cable Act. Our analysis suggests, however,
Entry deterrence in these cases may be that mandatory access plays an anticompeti-
achieved through establishing and maintain- tive role in this type of market. Quite sim-
ing prices that dissuade consumers from ply, under mandatory access, competitive
forming coalitions and soliciting entry. Sinceentry is unlikely to take place. By granting
such pricing strategies can impose a consid- incumbents access to any individual cus-
erable profit cost upon the incumbent oper- tomer, these laws prevent the emergence of
ators, cable providers have appealed to the effective customer coalitions. Hence, cus-
political and judicial process for less costly tomers cannot guarantee an entrant's prof-
means (from their perspective, but not soci- itability, making any entrant vulnerable to
ety's) to achieve entry deterrence. These aggressive post-entry behavior by the in-
appeals have taken the form of three lines cumbent in this type of high-sunk-cost in-
of attack upon consumers' opportunity to dustry.
form coalitions and sign exclusive agree-
ments.
First, exclusive dealings in cable TV have
been challenged under the federal antitrust
laws. However, in Satellite Television and
Associated Resources v. Continental Cable- 30To date, the courts have issued conflicting inter-
vision of Virginia (1983) the U.S. 4th Circuitpretations on this provision of the Cable Act. The 11th
Circuit Court in Centel Cable Television Co. of Fla. v
Court used a rule-of-reason criterion and a
Thos. J. White Development Corp. (1990) affirmed the
rather broad definition of the relevant mar- franchisee's right of access, but the 3rd Circuit Court
ket to rule that such an agreement was not in Cable Investments Inc. v. Woolley (1989) held that
section 621(aX2) cannot be used to gain access to
apartments or multiunit dwellings (MDU's) when the
owner wants to substitute a SMATV operator. Recent
district-court decisions in California, Georgia, and
Maryland, however, have not followed the 3rd Circuit
Court's precedent in Woolley and have held that cable
29These arrangements involve either direct billing to operators can use section 621(aX2) to gain access to
the landlord or association based on the total number single-family developers or even MDU's.
of units or subscribers or direct billing to the customers 31In general, these laws have been upheld by the
and remission of a fee to the landlord or association by courts when they provide nominal compensation to the
the cable operator. owner for the "taking" of property.

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
582 THE AMERICAN ECONOMIC REVIEW JUNE 1994

Third, at the local government level, the structure and efficiency properties of exclu-
cable industry has sought to forestall sionary contracts between an incumbent
buyer-entrant agreements by imposing upon seller and consumers. We find that entry is
an entrant the requirement of universal ser- a better investment for either a potential
vice (e.g., service throughout a metropolitan vertical entrant or a buyer-entrant coalition
area) at uniform rates. This type of require- than it is for a potential outside entrant.
ment also prevents the emergence of effi- Unifying buyer and entrant interests inter-
cent customer coalitions, this time by man- nalizes the beneficial effects to buyers from
dating that entrants deal only with the grand entry which leads, in turn, to excessive en-
coalition of customers. This type of regula- try. However, exclusive contracts between
tion has been recently upheld at the district the incumbent and buyers can be designed
court level in Telesat Cablevision, Inc. v. The to deter such inefficient entry and thereby
City of Riviera Beach (1991), a case involving increase social welfare. We thus find effi-
a Florida SMATV operator's attempt to ciency benefits of exclusionary agreements
contract with several condominium associa- that are absent in earlier analyses of such
tions. agreements wherein customers' opportuni-
The cable industry's success in fostering ties to form coalitions, sign contracts, and
market conditions inimical to entry has, fromgenerally play a proactive role in the market
all available evidence, enabled the industry are limited.
to exercise considerable market power. For With multiple buyers, a monopolist may
example, post-regulation prices have risen be able to exploit buyers' disorganization,
at three times the rate of inflation.32 These while still constrained by buyer incentives to
market conditions prompted calls for rereg- coordinate and create countervailing power.
ulation of cable prices, culminating in the Equilibrium contracts in this setting are
enactment of the Cable Television Con- shown to take a divide-and-conquer form,
sumer Protection and Competition Act of where the incumbent selectively offers high
1992. Unfortunately, this act is silent on payoffs to some consumers so as to elimi-
rights to access, meaning that the status quonate their incentive to join buyer coalitions.
prevails. Our analysis indicates that, by The incumbent is then free to offer poorer
eliminating such impediments to the estab- contract terms to the remaining customers.
lishment of customer coalitions and the en- These contracts deter the inefficient entry
actment of exclusive contracts, monopoly that would occur in their absence. However,
power in the cable TV industry could be exploitation of unlucky, "conquered" cus-
brought in check without incurring the bu- tomers may take the form of commitments
reaucratic and possible efficiency costs of that deter some efficient entry, thereby cre-
reregulation. ating offsetting welfare effects. In this case,
a combination of exclusive contracts and a
IV. Conclusion policy to prohibit price discrimination gives
rise to equilibrium contracts that deter only
This paper studies the implications of po- inefficient entry.
tential strategic buyer behavior for the The implications of this paper differ
rather sharply from the Aghion and Bolton
(1987) and Rasmusen et al. (1991) analyses,
in which no-contract outcomes are efficient
32Robert Rabinovitz (1993) has estimated that and exclusionary contracts between con-
43 percent of this increase was due to the increa-
sumers and the incumbent seller promote
sed exercise of market power after passage of the
Cable Act. Unpublished statistics compiled by Thomas inefficiency. These analyses remain impor-
Hazlett (pers. comm.) also reflect the post-Cable Act tant counterexamples to the general conclu-
exercise of market power by monopoly cable providers; sions of Posner (1976) and Bork (1978) con-
between 1988 and 1991, Hazlett has found that the
cerning the efficiency properties of exclu-
price per channel markup for monopoly systems over
duopoly systems ranged from 25 percent to 58.5 per-
sionary contracts. However, this analysis
cent, while monopoly systems sold at a per-subscriber supports the Posner and Bork conclusions
premium of about 25 percent over duopoly systems. for the rather broad class of markets

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
VOL. 84 NO. 3 INNES AND SEXTON: STRATEGIC BUYERS 583

wherein consumers can behave strategically titrust." Antitrust Bulletin, Spring 1991,
to elicit entry either through vertical inte- 36(1), pp. 1-79.
gration or buyer-entrant contracting. Jaffe, Adam B. and Kanter, David M. i'Market
Power of Local Cable Television Fran-
chises: Evidence from the Effects of
REFERENCES Deregulation." Rand Joumal of Eco-
nomics, Summer 1990, 21(2), pp. 226-34.
Aghion, Phillipe and Bolton, PatricL "Con- Katz, Michael. "The Welfare Effects of Third
tracts as a Barrier to Entry." American Degree Price Discrimination in Interme-
Economic Review, June 1987, 77(3), diate Goods Markets." American Eco-
pp. 388-401. nomic Review, March 1987, 77(1), pp.
Bork, Robert. The antitrust paradox: A policy 154-67.
at war with itself. New York: Basic Books, Mankiw, N. Gregory and Whinston, Michael D.
1978. "Free Entry and Social Efficiency." Rand
Cable Investments, Inc. v. Woolley. (3rd Cir., 867 Joumal of Economics, Spring 1986, 17(1),
F.2d 151, 1989). pp. 48-58.
Centel Cable Television Co. of Fla. v. Thos. J. WhiteMarvel, Howard. "Exclusive Dealing." Jour-
Development Corp. (llth Cir., 902 F.2d 905, nal of Law and Economics, April 1982,
1990). 25(1), pp. 1-25.
Comanor, William S. and Frech, H. E., III. Mathewson, G. Frank and Winter, Ralph A.
"The Competitive Effects of Vertical "The Competitive Effects of Vertical
Agreements ?" American Economic Re- Agreements: Comment." American Eco-
view, June 1985, 75(3), pp. 539-46. nomic Review, December 1987, 77(5), pp.
Demsetz, Harold. "Why Regulate Utilities?" 1057-62.
Joumal of Law and Economics, April 1968,Mills, David E. "Untimely Entry." Journal of
11(1), pp. 55-65. Industrial Economics, December 1991,
Ellingsen, Tore. "Strategic Buyers and the 39(6), pp. 659-70.
Social Cost of Monopoly." American Eco- Noam, Eli M. "Economies of Scale in Cable
nomic Review, June 1991, 81(3), pp. 648- Television: A Multiproduct Analysis," in
657. E. M. Noam, ed., Video media competi-
Hazlett, Thomas W. "Private Contracting ver- tion: Regulation, economics, and technol-
sus Public Regulation as a Solution to the ogy. New York: Columbia University
Natural Monopoly Problem," in William Press, 1985, pp. 93-120.
Poole, ed., Unnatural monopolies. Lexing- Northwest Wholesale Stationers, Inc. v. Pacific Sta-
ton, MA: Lexington Books, 1985 pp. tionary & Printing Co. (472 U.S. 284, 1985).
71-114. Ornstein, Stanley. "Exclusive Dealing and
" Competitive vs. Franchise Antitrust." Antitrust Bulletin, Spring 1989,
Monopoly in Cable Television." Contem- 34(1), pp. 65-98.
porary Policy Issues, April 1986, 4(2), pp. Owen, Bruce and Greenhalgh, Peter. "Competi-
80-97. tive Policy Considerations in Cable Tele-
_ "The Demand To Regulate Fran- vision Franchising." Contemporary Policy
chise Monopoly: Evidence from CATV Issues, April 1986, 4(2), pp. 69-79.
Rate Deregulation in California." Eco- Perry, Martin K. "Vertical Integration:
nomic Inquiry, April 1991 29(2), pp. Determinants and Effects," in R.
275-96. Schmalensee and R. Willig, eds., Hand-
. "Cable TV Re-regulation." Regula- book of industrial organization. Amster-
tion, 1994 (forthcoming). dam: North-Holland, 1989, pp. 183-255.
Innes, Robert and Sexton, Richard J. "Strategic Posner, Richard A. Antitrust law: An eco-
Buyers and Exclusionary Contracts." nomic perspective. Chicago: University of
Working paper, University of Arizona, Chicago Press, 1976.
December 1992. Prager, Robin A. "Firm Behavior in Fran-
Jacobson, Jonathan M. and Dorman, Gary J. chise Monopoly Markets." Rand Joumal
"Joint Purchasing, Monopsony, and An- of Economics, Summer 1990, 21(2), pp.

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms
584 THE AMERICAN ECONOMIC REVIEW JUNE 1994

211-25. Sexton, Richard J. and Sexton, Terri A. "Coop-


Rabinovitz, Robert N. "Market Power and eratives as Entrants." Rand Journal of
Price Increases for Basic Cable Service Economics, Winter 1987, 18(4), pp.
Since Deregulation." Rand Journal of 581-95.
Economics, Spring 1993, 24(1), pp. 1-28. Smiley, Robert and Veraldo, Riccardo. "The
Rasmusen, Eric B; Ramseyer, J. Mark and Development of Consumer Cooperatives
Wiley, John S. "Naked Exclusion." Ameri- in Western Europe and in the United
can Economic Review, December 1991, States." Economia Azindale, August 1983,
81(5), pp. 1137-45. 2(3), pp. 405-23.
Satellite Television and Associated Resources v. Telesat Cablevision, Inc. v. The City of Riviera
Continental Cablevision of Virginia. (4th Cir., Beach. (773 F. Suppl. 383, S.D.F., 1991).
714 F.2d 351, 1983), cert. denied (465 Varian Hal. "Price Discrimination," in R.
U.S. 1027, 1984). Schmalansee and R. Willig, eds. Hand-
Scheffman, David T. and Spiller, Pablo T. book of industrial organization, Vol. 1.
"Buyers' Strategies, Entry Barriers, and Amsterdam: North-Holland, 1989, pp.
Competition." Economic Inquiry, July 597-654.
1992, 30(3), pp. 418-36. von Weizsacker, C. C. "A Welfare Analysis of
Schwartz, Marius. "The Competitive Effects Barriers to Entry." Bell Journal of Eco-
of Vertical Agreements: Comment." nomics, Autumn 1980, 11(2), pp. 399-429.
American Economic Review, December Wissman, Roger A. "After Decade of Growth,
1987, 77(5), pp. 1063-68. Co-ops' Share in Purchasing, Marketing
Sewell Plastics, Inc. v. Coca Cola et al. (720 F. Stabilizes in 80s." Farmer Cooperatives,
Suppl. 1196, W.D.N.C., 1989). March 1986, 52(3), pp. 12-13.

This content downloaded from


189.203.193.200 on Mon, 25 Jul 2022 12:52:22 UTC
All use subject to https://about.jstor.org/terms

You might also like