A. DOUGLAS MELAMED
The American Bar Association
April 2, 1998
It may be a coincidence, but it is also a fact, that most of the Division's recent contested, civil non-merger cases have involved what are in substance exclusionary vertical agreements. Among the recent or pending cases involving such agreements are Delta Dental(1) -- which concerned most favored nation agreements pursuant to which dental care providers agreed with healthcare payers that they would not charge anyone else a lower price; Stillwell(2) -- a tie-in of utility services; General Electric(3) -- a straightforward horizontal case about competition in equipment servicing that includes, among others, an allegation that the software licensing agreement at issue prevents otherwise willing firms from servicing machines manufactured by third parties and thereby impairs the ability of those parties to compete with GE; and Nielsen(4) -- which concerned tie-in and exclusive distribution arrangements. Several of the Division's pending investigations also concern exclusionary vertical agreements.
These cases, and the different types of exclusionary agreements they involve, raise a variety of different doctrinal questions. But they raise similar conceptual issues.
What I propose to do today is to outline some of my thoughts about how to analyze these issues. This is not a statement of DOJ policy, which is still evolving. I think of this, instead, as part of the kind of "dialogue with the bar and the academic community about antitrust doctrine" that Joel Klein has spoken of.(5)
At the outset, let me put this in context: Antitrust is concerned with private arrangements that reduce competition. They can do so in one of two ways: (i) by effecting collaboration among those that would otherwise be rivals (arrangements, including mergers, joint ventures and cartels -- which we generally call "horizontal" -- that raise the prices or reduce the output of the parties to the arrangement); and (ii) by excluding rivals or potential rivals from the market (or weakening them, reducing their output, and thus diminishing their capacity to discipline the exercise of market power).
My focus today is on practices that have the latter effect. Exclusionary practices can include both (i) unilateral single-firm practices, which are subject to Section 2 of the Sherman Act, and (ii) vertical agreements, i.e., agreements between firms that do not compete with one another or, more precisely, agreements that do not concern the way in which the parties to the agreement will compete against one another. Like all agreements, these agreements are subject to Section 1 of the Sherman Act.
Exclusionary vertical agreements are agreements that tend to exclude competitors of one of the parties to the agreement. Examples include exclusive dealing, tie-in arrangements, and most favored nation agreements.
What these agreements have in common is that they are agreements between a firm -- call it "Firm A" -- and one or more other firms that restrict the ability of those other firms to supply inputs to Firm A's rivals. The restrictions can be explicit and direct, such as those in exclusive dealing arrangements or most favored nations clauses, or implicit and indirect, such as those resulting from tie-in agreements, market share incentives or penalty provisions in license agreements.
For simplicity, I will hereafter refer to simple exclusive dealing agreements between a manufacturer and its distributors, who provide an input in the form of distribution services, that prohibit the distributors from handling products of rivals of the manufacturer. Although I will focus on simple exclusive dealing agreements, I believe that this analysis could apply to all exclusionary vertical agreements.
Vertical Agreements in General
Vertical agreements are generally procompetitive because they usually involve a combining of complements -- for example, the manufactured product and distribution services -- for greater good.(6)
Although vertical agreements are generally procompetitive, they can injure competition, under some circumstances, when they deny (or raise the cost of) a needed or valuable input -- such as distribution services -- to a rival.(7)
A manufacturer seeking to harm a rival by exclusionary vertical agreements must induce the distributors to make an exclusionary commitment or promise, such as the promise not to distribute products produced by the manufacturer's rivals. The essence of an exclusionary agreement is this promise by the distributor (or other input supplier).
This is a key point -- one which I think has generally been ignored in the cases and the commentary and which I think can help us make sense of exclusionary vertical agreements: The distributor's promise imposes a cost on it -- typically, the inability to deal with the manufacturer's rivals -- and the distributor has to be induced to bear that cost; it has to be compensated for it. The critical question is, where does the compensation come from?
The exclusionary promise might of course be more efficient than the alternatives available to the distributor. For example, an exclusive dealing requirement can ensure the distributor's loyalty and prevent free-riding and might thus expand output. If the exclusionary promise is more efficient than the alternatives, the transaction makes sense and is easy to explain. The distributor in effect conducts an auction. If the exclusive dealing arrangement is the most efficient of the alternatives available to the distributor, he can be induced to enter into the agreement by being permitted to share in the value created by the efficient transaction.
But what if the promise is not efficiency-enhancing? What if, for example, the efficiency-enhancing benefits from the prevention of free-riding are less than the benefits that the distributor could realize from distributing products manufactured by others? If the value created by transactions with other manufacturers would exceed the value generated by the efficiency-enhancing benefits of the exclusive dealing agreement, one would expect the market (and thus the distributor) to reject the exclusionary promise and to choose the other, more efficient transactions instead.
Unfortunately, this does not mean that the market will always choose the most efficient vertical arrangement, nor does it mean that anticompetitive exclusionary agreements can never happen. But it does mean that exclusionary agreements that are less efficient than the alternatives can happen only in one of two circumstances. The first is mistake, which the market presumably will correct.
The second, and more important, circumstance is this: If the manufacturer expects to gain or preserve market power by the exclusion of its rivals, it can endeavor to induce the distributors to go along with the exclusionary scheme by sharing with them a portion of the anticipated supracompetitive profits. The sharing of supracompetitive profits could take the form simply of a high price paid for distribution services, or it could be part of the consideration paid to the distributors in more subtle or complex commercial arrangements.
If the supracompetitive profits available to the distributor are large enough, the distributor can be induced to agree to the restraint, even if it is inefficient.(8) In other words, even if the manufacturer cannot induce the distributor to make the exclusionary promise by sharing efficient fruits of the transaction, it can do so by sharing a portion of the supracompetitive profits created or preserved by the restraint.(9)
The foregoing suggests that a necessary condition for an exclusionary vertical agreement to be anticompetitive is that the agreement is likely to enable the manufacturer that would benefit from the exclusion of a rival to gain or preserve market power that it otherwise would not have. The additional market power enables the manufacturer to recoup its investment in the otherwise inefficient restraint, including the consideration it must pay to the distributors.
Articulating this theoretical proposition, however, does not end our work. It is the job of the enforcement agencies to make the theory operational. While the theory would suggest that proof of market power should always be required to condemn an exclusionary vertical agreement, such a requirement might not be optimal.
Proof of market power, and particularly of market power effects, is always difficult. At best, requiring such proof imposes costs and introduces additional uncertainty into the antitrust enforcement process. What is needed are decision-making rules that can enable us to avoid those costs and uncertainties where possible.
To be more precise, the objective should be to articulate decision-making principles that minimize the sum of (i) enforcement costs plus (ii) the costs of enforcement errors. For this purpose, the costs of enforcement errors are the costs incurred when efficient, procompetitive transactions are prevented or when anticompetitive transactions are permitted.(10)
To achieve this objective, I believe that the enforcement agencies might ask the following analytically distinct questions when assessing allegedly exclusionary vertical agreements:(11)
First, are the agreements exclusionary -- that is, do they exclude rivals from the market or materially diminish their competitive efficacy by raising their costs or denying them needed inputs? If not, they are harmless and should be lawful.
This should be a meaningful hurdle for any complainant.(12) The requirement would seem in general to have two components: first, that competitively meaningful rivals cannot readily avoid the cost increases by turning to other input suppliers or alternatives to the foreclosed input; and, second, that the input foreclosure or cost increase has a material impact on the ability of those rivals to do business in the market in which they compete with the manufacturer.(13)
The exclusionary effect might depend, not just on the characteristics of the individual agreements, but on the aggregate impact of all such agreements used by the manufacturer. A set of exclusive agreements with numerous distributors, for example, might be exclusionary even though no one of the agreements by itself would have a material impact on the manufacturer's rivals.
Second, if the agreements are exclusionary, are there plausible efficiencies? If not, it might be appropriate, at least in some cases, simply to find them to be unlawful without more.
I suspect that this suggestion will be greeted with skepticism in some quarters by those who fear that, if it is adopted by the courts, it would subject all types of business conduct to antitrust second-guessing and that it would invite meritless lawsuits by strike-suiters and by ineffective competitors seeking to insulate themselves from competition by more efficient or innovative rivals.(14)
These concerns are by no means insubstantial. I find them convincing in many circumstances. But they are less convincing in this context.
In the first place, the inquiry I am suggesting would apply only to agreements that actually exclude rivals. Such agreements impose real costs on rivals and their customers. If that is all that they do, they are harmful to the economy.
Moreover, the efficiency screen that I am suggesting is rather modest. While it would require more than just lawyers' arguments -- presumably, some factual demonstration that efficiencies are reasonably to be expected under the circumstances -- it would not require proof of those efficiencies or of their magnitude. In most cases, it should be enough to articulate and begin to document a coherent story of nontrivial efficiencies that appears consistent with the readily available facts.
Those who would prefer a market power screen observe that exclusionary vertical agreements will often have efficiencies and, indeed, that exclusion and efficiency are often two sides of the same coin. To the extent that exclusionary agreements in one form or another give the manufacturer exclusive or preferential access to input suppliers, they are, for example, likely to generate efficiencies from supplier loyalty, elimination of free-riding or economies of scale. But it is precisely because efficiencies are so common and easy to identify that it might be appropriate to condemn naked exclusionary agreements for which no plausible efficiency can be demonstrated.
The alternative would be to jump immediately to an examination of market power. While a market power requirement would make theoretical sense, it might not make practical sense. A market power screen would not only impose costs and uncertainty, but also would create a powerful default rule. Difficulties of proof mean that many instances of market power would go unproven and, thus, that anticompetitive agreements would escape antitrust condemnation. There is no need to bear the cost of such false negatives with respect to exclusionary agreements for which there is no plausible efficiency justification.
As a practical matter, the inquiry into efficiencies by the enforcement agencies does not proceed in isolation. The agencies will and ought to review information that will enable them to make at least educated guesses as to whether the beneficiary of the exclusionary agreement is likely to gain market power. If that is very unlikely, the agencies are unlikely to challenge the agreement, at least without further investigation, both because they will not expect the agreement to have any enduring adverse impact and because the unlikelihood of market power effects might cast doubt on the conclusion that there are no efficiencies.
But if market power consequences appear likely, although unproven, it might make good sense for the agencies to challenge the exclusionary agreements. It should be enough to prove that the agreements are exclusionary and are justified by no plausible efficiencies, without the costly and uncertain task of actually proving market power effects. While the law is unsettled, I think there is room in the cases for such an approach.(15)
Third, if there are plausible efficiencies, are the exclusionary agreements likely to create or preserve market power for the manufacturer? If not, the agreements should be lawful because, absent such market power, the compensation needed to induce the distributors to agree obviously would not consist of a share of supracompetitive profits created by the agreements. The distributor's participation, and the existence of the agreement, could thus be presumed to be the result of the efficiencies created by the agreement. In that event, we are willing to tolerate the exclusionary effects because it has long been clear that antitrust "protects competition, not competitors."(16)
The focus here is not on whether the manufacturer has market power, but rather on whether the exclusionary agreement is likely to create or preserve market power for the manufacturer. It is only such additional or incremental market power that could explain the willingness of the manufacturer and the distributors to enter into an inefficient agreement. This distinction is likely to matter, however, only with respect to manufacturers that do not already have market power; exclusionary agreements that benefit a manufacturer that already has market power will almost always increase its market power.(17)
Fourth, if market power is implicated, are some or all of the exclusionary aspects of the agreements not reasonably necessary to achieve the efficiencies? For example, has the monopoly manufacturer obtained broader exclusivity agreements, or entered into exclusive agreements with more distributors, than necessary to realize the efficiencies? If so, the unnecessary agreements (or the unnecessary part of the agreements) should be illegal.(18)
The Hardest Cases
If you've been listening carefully, you will note that these four questions do not resolve all cases. What's left is the hardest case -- vertical agreements that simultaneously (i) exclude or significantly harm rivals, (ii) are likely to create or preserve market power for the manufacturer, and (iii) create genuine efficiencies.
Market power, while a necessary condition of illegality, should not be sufficient to condemn such exclusionary agreements because they might be the most efficient form of distribution, even if used by a monopolist. In that event, the market power gained by the manufacturer can be said to be the result of its "superior skill, foresight and industry"(19) and would not provide a sufficient basis for condemning the agreement.
By the same token, we should not uphold the arrangements simply because they have some efficiency properties. To be sure, if we were dealing with single-firm conduct by the manufacturer, we might be more likely to uphold the conduct, in part because, absent predation, certain types of allegedly anticompetitive single-firm conduct will succeed in the marketplace only if they are more efficient than the alternatives.
But the analysis is different with an exclusionary agreement. There is another party to the allegedly unlawful scheme -- the input supplier or distributor -- and it helps to know why it agreed to the arrangement.
To illustrate this point, let me return again to the metaphor of the distributor conducting an auction. The distributor chooses between the exclusionary agreement and all the other alternatives available from the manufacturer or its competitors. We cannot infer, however, from the facts that the exclusionary arrangement has some efficiency benefits and that it won the auction, that the distributor agreed to the arrangement because it was more efficient than the alternatives. Other arrangements may on balance have been even more efficient and might thus have won the auction but for the inducements offered by the prospect of sharing the manufacturer's supracompetitive profits.
Unfortunately, we usually cannot directly observe the payment of supracompetitive profits to distributors. Sometimes, the payments are disguised in complicated arrangements; in other instances, the distributors might receive a simple form of consideration, such as a high price, that could reflect either efficiency or market power.
Because we cannot observe the payment of supracompetitive profits, we need to evaluate the exclusionary agreements ourselves, without relying on the market to tell us whether they are on balance procompetitive or anticompetitive. Thus, where (i) exclusion, (ii) market power and (iii) efficiency are inextricably intertwined, the proper resolution is a more comprehensive rule of reason analysis -- one that weighs the anticompetitive consequences of the agreements against their procompetitive or efficiency-enhancing implications.
It is important to be careful about what the term "rule of reason" means in this context. Often -- perhaps usually -- the term "rule of reason" is used to connote the evaluation of a collaboration among competitors, such as a joint venture, an information exchange, a joint buying arrangement, and the like. In those cases, the issue is whether the arrangement reduces or increases output of the parties to the arrangement.(20)
In the case of exclusionary vertical agreements, such as exclusive dealing arrangements between a manufacturer and its distributors, the question is very different. In such a case, the question is whether overall output is increased or decreased in the market in which the manufacturer is likely to gain market power, taking into account both the exclusion of rival manufacturers -- which reduces their output -- and the efficiencies -- which could increase the output of the manufacturer that is a party to and benefits from the agreements.
Therefore, in applying the rule of reason to exclusionary vertical agreements, we should undertake an analysis much like that which the agencies use in considering efficiencies in merger cases. We should first estimate the impact of the exclusionary agreements on rivals or, to be more precise, on their output in the market. We should then ask, assuming that impact, whether the efficiencies created by the agreements will increase the beneficiary's output enough so that, on balance, marketwide output will be higher, in the market in which the agreements will create or preserve market power, than if the exclusionary agreements were prohibited.(21) If not, the agreements are illegal.
I am not sure that we will have sufficient information to undertake an analysis like this in many or most cases. But this, in principle, is the question we should ask, and we should do so where we can.
So, those are some of my thoughts about an analytical framework that might help us evaluate exclusionary vertical agreements. While the approach might seem complex, as a practical matter it boils down to something like this:
-- If your agreements are likely, on balance, to reduce price or increase marketwide output -- even taking into account their effect on rivals -- you probably do not have to worry.
-- But, if neither of those is the case, you probably ought to take another look at your agreements.
1. United States v. Delta Dental, 943 F. Supp. 172, 179-80 (D.R.I. 1996) (holding that challenge to Most Favored Nations clause set forth facts sufficient to state a Section 1 claim); United States v. Delta Dental, 1997 WL 527669 (D.R.I. March 4, 1997) (consent judgment and competitive impact statement).
2. United States v. City of Stillwell, Oklahoma, Civ. 96-196-B (filed Apr. 25, 1996).
3. See United States v. General Electric, 1997 WL 269491 (D.Mont. March 18, 1997) (denying motion to dismiss Section 1 claim and dismissing Section 2 claim).
4. See Antitrust Division press release, December 3, 1996.
5. Joel Klein, "A Stepwise Approach to Antitrust Review of Horizontal Agreements" (November 7, 1996); see generally, Joel Klein, "Making the Transition from Regulation to Competition: Thinking About Merger Policy During the Process of Electric Power Restructuring" (January 21, 1998). Tom Krattenmaker, Steve Salop, David Seidman and Greg Werden made especially provocative and valuable comments on an earlier outline of these thoughts.
6. They should be distinguished from horizontal agreements, which are agreements among rivals that constrain their rivalry regarding the purchase or sale of substitutes. Agreements among rivals not to deal with other rivals or to deal with them only on unfavorable terms raise issues in addition to those addressed here. Such agreements are sometimes condemned as illegal horizontal agreements without detailed analysis of their exclusionary effects. See, e.g., Fashion Originators' Guild v. FTC, 312 U.S. 457 (1941).
7. Other types of vertical agreements (e.g., resale restrictions) can, under some circumstances, injure competition by facilitating a dealer or manufacturer cartel; but vertical agreements that injure competition by facilitating such horizontal arrangements are not the type of vertical restraints with which I am presently concerned. I also do not address here the use of exclusionary agreements as a device to evade regulation.
8. Some courts have suggested that exclusionary contracts of short duration cannot be unlawful because the excluded rival(s) are able to bid frequently for the patronage of the input supplier and can thus prevent anticompetitive harm. But if the manufacturer that is the beneficiary of the agreement is more likely to gain market power and thus to be able to share supracompetitive profits with the distributor, it should be able to retain the exclusive arrangement regardless of the duration of its contract with the distributor.
9. I assume here, of course, that transaction costs prevent consumers from acting to protect their interests.
10. The two types of costs are related, but it is unlikely that they always correlate either directly or inversely. As a general matter, an increase in enforcement costs should tend to reduce enforcement errors in the particular case, although there are likely to be diminishing returns to that benefit. But an increase in enforcement costs could be correlated with increased enforcement errors if the enforcement costs reflect investigation of, or a need to resolve, issues that (i) do not significantly enhance the likelihood of determining whether the conduct at issue is anticompetitive and (ii) are difficult to resolve reliably. Moreover, legal rules that require substantial enforcement costs for their application will lead to an increase in enforcement errors to the extent that the prospect of such costs deters efficient transactions or legal challenges to anticompetitive transactions.
11. The approach set forth here is consistent with the joint DOJ/FTC Antitrust Guidelines for the Licensing of Intellectual Property (April 6, 1995) and the approach to horizontal restraints suggested by Joel Klein in his "Stepwise" speech several months ago. See note 5, above.
12. Much of the case law dealing with exclusionary vertical agreements has concerned the issue of exclusionary effects, but the cases often deal with the issue only indirectly. Thus, for example, tie-in cases often turn on whether the defendant has market power in the tying product, and exclusive dealing cases often turn on the percentage of the input suppliers that are subject to the exclusive dealing restrictions. Tying product market power and percentage foreclosure of input suppliers are, to be sure, factors relevant to predicting exclusionary effects, but they do not measure those effects directly. It would be preferable, I think, for the enforcement agencies to focus explicitly on the exclusion of rivals, both in order to keep in mind the purpose of the kinds of proxies that are often addressed in the cases and so that direct evidence of exclusion will not be overlooked. Cf., Jefferson Parish Hosp. Dist. No. 2 v. Hyde, 466 U.S. 2, 16 and 21 n.34 (1984).
13. The issue here is exclusion of rivals, regardless whether the manufacturer itself has or is likely to gain market power. The two are likely to coincide in most cases. But (i) the exclusion test is less rigorous, and easier to apply, than the market power test because it focuses on rivals (competitors) rather than competition in the market as a whole, and (ii) the beneficiary of an exclusionary agreement might not have market power in the same market, especially if it uses the inputs in multiple markets.
14. See T. Krattenmaker & S. Salop, "Anticompetitive Exclusion: Raising Rivals' Costs to Achieve Power over Price," 96 Yale L. J. 209, 247-48 (1986) (imposing antitrust liability without proof of market power "would permit rivals routinely to complain of efficient exclusionary practices"); F. Easterbrook, "The Limits of Antitrust," 63 Texas L. Rev. 1, 14-23 (1984).
15. See FTC v. Indiana Fed'n of Dentists, 476 U.S. 447, 457 (1986) ("[a]s a matter of law, the absence of proof of market power does not justify a naked restriction on price or output" and such a restriction "requires some competitive justifications") (quoting Board of Regents of the Univ. of Okla. v. NCAA, 468 U.S. 85, 109-10 (1984)). While the cited cases concerned horizontal restraints, the language and analysis could encompass exclusionary vertical agreements as well. See generally, P. Areeda & H. Hovencamp, Antitrust Law § 1503 at 929 (1996 Supplement); cf., Premier Electrical Construction Co. v. NECA, 814 F.2d 358 (7th Cir. 1987) (Easterbrook, J.).
16. See Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 224 (1993) ("It is axiomatic that the antitrust laws were passed for 'the protection of competition, not competitors.'") (quoting Brown Shoe Co. v. United States, 370 U.S. 294, 320 (1962)).
17. The simple term "market power" obviously glosses over a number of issues that I do not address in this speech. These include the distinction between the traditional notion of market power as the power to raise price above marginal cost and notions of exclusionary market power as the power to exclude rivals, and the circumstances under which a firm may be unable to increase its profits by excluding rival suppliers of complements to a product over which it already has a monopoly.
18. This is simply the overbreadth or more-restrictive-than-necessary principle that has been a fundamental part of antitrust law since Justice Taft's opinion in United States v. Addyston Pipe & Steel Co., 85 F. 271 (6th Cir. 1898), aff'd, 175 U.S. 211 (1899). See also Rothery Storage & Van Co. v. Atlas Van Lines, Inc., 792 F.2d 210, 224 (D.C. Cir. 1986) (Bork, J.), cert. denied, 479 U.S. 1033 (1987) ("If it is so broad that part of the restraint suppresses competition without creating efficiency, the restraint is, to that extent, not ancillary.") (citing Addyston Pipe).
19. United States v. Aluminum Co. of America, 148 F.2d 416, 430 (2d Cir. 1945).
20. The ultimate issue is whether the arrangement reduces output in the market as a whole. A purely horizontal collaboration, however, will reduce total market output if, but only if, it reduces the output of the parties, so we can use the simpler latter issue as a proxy for the former.
21. There may be circumstances, which I do not address here, in which it is appropriate to consider efficiencies in other markets or efficiencies that do not result in increased output in the short run. See DOJ/FTC Revision to the Horizontal Merger Guidelines, section 4 and nn. 2 & 3 (April 8, 1997).