A Stepwise Approach to Antitrust Review
of Horizontal Agreements
JOEL I. KLEIN
The Westin Hotel
November 7, 1996
During the last year and a half, while I've been at the Antitrust Division of the Department of Justice, I've given a fair amount of thought to the question of antitrust doctrine. For several decades, the Chicago School has largely dominated that discussion, in the courts as well as in the journals. And while their basic focus on consumer welfare as the key to sound antitrust analysis is now widely accepted, the application of that view to a new competitive paradigm -- one dominated by extraordinarily rapid changes in technology coupled with an increasingly globalized economy -- raises some potentially important and challenging doctrinal issues for those of us in the antitrust field. This evolution led Bob Pitofsky and the Federal Trade Commission to conduct extensive hearings leading to a report which, among other things, calls for a joint task force with the Antitrust Division to review the question of efficiencies in the merger area. You will hear a good deal about the fine work that the Commission has done during this Conference.
For my part, I've been thinking about these issues largely in the civil non-merger area during the time that I had responsibility over those matters while serving as Anne Bingaman's Principal Deputy. And when I was named as acting head of the Division, one of the first things on my agenda was to open up a dialogue with the bar and the academic community about antitrust doctrine. I plan to make several speeches, raising issues of doctrinal concern and attempting both to set forth how the Division is addressing those issues and to elicit responsive comments from others. I hope this project will get people thinking and talking more about such matters.
This is the first of those speeches and, as such, I decided to begin by discussing different methods of antitrust analysis -- in short, what questions do we at the Division ask, and what questions do we think the courts should ask, in deciding whether a practice violates the antitrust laws. Specifically, I will focus on horizontal agreements and revisit some perplexing questions about the uses of per se and rule of reason analysis. In the future, I would hope to open up similar discussions with respect to other areas that I think may be (or, perhaps, should be) in flux, such as the relationship between antitrust and intellectual property, the significance of phenomena like network effects and tipping points for antitrust analysis, the impact of the differentiated-products theory on enforcement, the role of potential competition and innovation markets, and the appropriate criteria for deciding whether the "agreement" requirement of Section 1 has been satisfied.
Before I turn to today's topic, let me emphasize precisely what I'm trying to do here. While I want to discuss doctrine as it affects enforcement policy, I'm not seeking to answer all the questions. I hope to stimulate discussion about doctrine -- something I think is needed in our field -- while also providing the best guidance I can as to how the Division is thinking about a particular problem or area. I strongly believe that it's in the mutual interest of the Division and the businesses over which we have enforcement authority for us to be as clear as we possibly can about what we're doing and why. That not only enables businesses to intelligently arrange their affairs and you to counsel your clients effectively, but it also helps ensure public accountability for the Division. Despite those important concerns, however, I would hope you don't take my comments on any of these matters as necessarily being the final word on where the Division is going. It's where we are now, but the nature of a discussion is to be able to listen as well as to talk and so, if I stimulate a persuasive response, our position might evolve.
With that introduction in mind, I want to turn to today's topic -- the appropriate method for analyzing horizontal agreements under Section 1 of the Sherman Act. This much seems clear and well-established: the courts and the enforcement agencies all recognize at least two basic methods of analysis. The first is called per se analysis, which applies to a relatively small group of practices that, as the Supreme Court has put it, "because of their pernicious effect on competition and lack of any redeeming virtue are conclusively presumed to be unreasonable."1 And the second method of analysis is called the rule of reason, the potential breadth of which was set out by Justice Brandeis in 1918 in his Chicago Board of Trade opinion, where he instructed courts to consider "the facts peculiar to the business to which the restraint is applied; its condition before and after the restraint was imposed; the nature of the restraint and its effect, actual or probable. The history of the restraint, the evil believed to exist, the reason for adopting the particular remedy, [and] the purpose or end sought to be attained."2 In short, a full-blown rule of reason analysis requires a comprehensive market analysis of pro- and anti-competitive effects and allows for any evidence that might bear on an assessment of those effects. Based on this analysis, the validity of an agreement then turns on whether its anticompetitive effects outweigh its beneficial effects.
In my view, even in non-per se cases, a reasonable mode of antitrust analysis needn't always require the kind of full-blown assessment of market effects that's called for under Justice Brandeis' formulation. Rather, some middle ground between the two traditional approaches often makes sense. A number of courts and commentators have made a similar point, although their proposals have differed both in substance as well as in description, and have been called everything from presumptive illegality, to strict antitrust scrutiny, to hard-boiled rule of reason, to truncated rule of reason, to quick look.3
Let me spell out in the specific approach that the Antitrust Division currently uses -- and the approach we think the courts should also use -- to analyze at least certain types of horizontal restraints. To begin with, as I just said, we reject the notion that there should only be two methods of analysis -- per se or full-blown market analysis. As a matter of both sound and efficient antitrust analysis, we think this dichotomy is too stark and, frankly, that it leads to far too much of a front-end emphasis on which approach to apply, a choice that can sometimes be outcome determinative. More to the point, adhering to such a dichotomy runs the risk of submerging thoughtful analysis in a battle over the selection of the proper mode of inquiry; and, consequently, either matters that are too complex for per se condemnation sometimes get shoe horned into that category, or some facially anticompetitive practices that have no demonstrable virtue escape condemnation because of insufficient evidence to satisfy the demanding requirements of a full-blown market analysis.
Rather than restrict ourselves to these polar alternatives, the Division employs a three-step analysis when reviewing any horizontal agreement directly limiting competition on price or output that would have occurred but for the agreement.4 When dealing with such an agreement, we first ask whether it is the type of restraint that is currently recognized by the courts as being a per se violation, such as an unadorned agreement to fix prices, curtail output, or divide markets. This category appears to be reasonably well-defined and usually the sole question we face in deciding if a particular agreement fits within it is whether, despite the effort to make it seem like the agreement is ancillary to some other arrangement, it in fact isn't.
We recently confronted this issue in a case that we are currently litigating against General Electric based on that company's licensing of diagnostic software to hospitals that both purchased GE's medical equipment (and wanted the software to service that equipment) and also competed with GE in servicing other hospitals' medical equipment. As a condition of the licensing agreement, GE insisted that the hospitals could not compete with it in servicing any one else's medical equipment, regardless of type or brand. We challenged this restraint as a per se violation because, apart from the fact that it was contained in the licensing agreement, it was facially unrelated to the use of the licensed software and thus should properly be seen as a naked agreement not to compete in the medical equipment servicing market.5
Moving back to our general approach, I should also note that, while I suspect the per se category is by now pretty well defined, I don't want to rule out the possibility that experience will teach us that other forms of horizontal agreements also merit such treatment. That's what the FTC recently concluded in the California Dental case with respect to agreements restricting advertisements of the price, quality, and availability of dental services. It is also conceivable, of course, that a practice currently considered per se should not remain in that category. And in fact, that happened not so long ago with respect to certain vertical agreements when the Supreme Court overruled Schwinn in the GTE Sylvania case.6
Now, to move on to the next step of our analysis, if we conclude that a horizontal agreement that directly limits competition on price or output between or among competitors is not per se illegal, we then inquire whether there's a procompetitive justification for the agreement. We put that question to the party defending the agreement, and we expect a response that doesn't merely speculate about the existence of efficiencies, but rather comes forward with real-world evidence - - factual evidence, expert economic evidence, and preferably both -- to support the claim. In this regard, I'd also like to note that, while contemporaneous evidence of intent may have some probative value on whether an agreement creates real efficiencies, our ultimate focus is on actual or likely effects, not intent.7 And, if we find that the proffered procompetitive justifications are unsubstantiated, we conclude that the agreement should be struck down.
On the other hand, if we find there are significant procompetitive benefits to the agreement, we then move on to the third step in our analysis and seek to determine whether its likely anticompetitive effects outweigh its procompetitive benefits. This weighing and balancing of the sort described by Justice Brandeis often requires an elaborate market analysis, unless, of course, there is convincing evidence of a direct market effect on price or output. But in either event, the key point I want to stress here is that only if there are real procompetitive benefits should there be any need to show actual anticompetitive effects.
Again, let me give you a specific example of how this process works by discussing a very recent case involving a joint selling agreement, or JSA, between two radio stations in Rochester, N.Y., a case in which we secured a consent judgment that is currently going through Tunney-Act proceedings. In essence, that case involved two radio stations, which were direct competitors, that entered into an agreement providing that station A would pay station B a fixed price, in return for which station A could then sell all of station B's advertising and keep the revenue. The agreement obviously eliminated price competition between the two stations by repositing all pricing decisions in one of them. Even so, because we were unable to conclude that joint selling arrangements as a group could categorically be considered per se illegal -- or even that this particular arrangement was merely a naked price-fixing agreement -- we proceeded to ask whether the arrangement produced any procompetitive benefits. The parties offered none and the evidence we examined didn't suggest any so we concluded that, given the elimination of price competition between two direct competitors, the agreement violated Section 1. We didn't require a showing of anticompetitive effects in those circumstances (although we did note that the evidence indicated that there probably was a direct effect on price). We also acknowledged in our Competitive Impact Statement that there could be other JSAs out there that might provide procompetitive benefits which would require a more thorough inquiry under the rule of reason. For example, there could be a JSA involving two stations that reached an agreement to package their sales in order to compete effectively with a company that owned a number of stations in the area. On those facts, we'd almost certainly want to look for evidence of actual anticompetitive effects before invalidating the arrangement.
Now that I've set out our basic method of analysis, there are two other things I'd like to discuss in the time remaining. First, I want to highlight and defend the policy choices reflected by this approach and second, I'd like to explain why I think it is fairly grounded in, or at least consistent with, existing Supreme Court precedent.
As for policy, there are a couple of choices implicated by this step-wise approach. First, it should open the door to at least some additional consideration of procompetitive justifications for horizontal agreements. For example, if we were to have a case presenting facts like those at issue in the Supreme Court's decision in Topco, which involved a cooperative buying arrangement with product-name identity and a territorial division on the sales side, we would most likely move to step 2 of our analysis rather than condemn the challenged restraint as a per se violation, as the Supreme Court did.8 In condemning the restraint as per se illegal, the Court employed a formalistic approach that ignored the crucial point that the territorial restriction might have been legitimately ancillary to a procompetitive arrangement.
I believe that our more flexible approach is preferable to such a formalistic view of per se analysis, especially given how much of our civil enforcement is done by consent decree, which heightens the need to make sure that we're not eliminating efficient, procompetitive practices from the market. Without at least some analysis, I don't see how you can be confident you're not doing just that in a case like Topco or even in the JSA case that I just mentioned.
Second, our approach ensures that facially anticompetitive, but potentially justifiable, agreements do not end up being upheld simply because they cannot be condemned as per se illegal. That is, by adopting this middle-step and requiring proof of procompetitive benefits where the parties have entered an agreement that directly limits competition on price or output before we have to prove actual anticompetitive effects or define markets and show market power, we are lessening the burden of proving a violation compared to what we'd have to show under a full-blown rule of reason analysis. I believe this is justifiable because it reflects the same basic policy choice that supports the per se doctrine in the first place. In particular, the rationale for per se analysis is that, once we conclude that a category of agreements directly limiting competition on price or output "lack[s] any redeeming virtue,"9 we strike down all agreements falling in that category without inquiring whether the particular agreement in question actually has any adverse market effects.10 And we do this even though everyone knows not every agreement so condemned has demonstrable anticompetitive effects - - for example, a couple of manufacturers with a very small market share violate Section 1 when they agree on the price they will charge, even if they can show that the price is the same that they would have charged in the absence of their agreement.
Now, you might wonder why it is that we condemn such agreements? And I think there are two reasons for this result. First, we recognize that there are market effects -- and there are market effects. What I mean is that, usually when we are concerned about such effects in a full-blown market analysis under the rule of reason, we require some form of market power to satisfy ourselves that the anticompetitive impact will be enduring rather than transitory. But it is also true that transitory effects often occur and they aren't a good thing either. As the Supreme Court explained in the Trial Lawyers case, "[f]or reasons including market inertia and information failures, . . . a small conspirator may be able to impede competition over some period."11 Given that recognition, and also given that we're starting from the premise that experience has taught us that the class of agreements in question has no potential for procompetitive benefits, the second reason to support a per se approach is simple efficiency: enough is enough in terms of the use of judicial and enforcement resources.12
The same reasoning supports the approach to rule of reason analysis that I've discussed here today. Since we're starting with an agreement that directly eliminates some competition on price or output, our only hesitation about striking it down should be our concern that, unlike what we've come to be confident about in per se cases, there may be something good going on in these other cases. And we should satisfy ourselves whether there is or there isn't. But where there isn't some procompetitive benefit at stake, it follows that precisely the same justifications for applying the per se rule without any case-specific showing of effects come into play and the inquiry can legitimately be ended.
In supporting the wisdom of this approach, I think it's important to underscore how critical it is to take efficiency analysis seriously. As I mentioned, this is something that is at the heart of the FTC's recent work concerning merger enforcement and I know Bob Pitofsky is going to talk to you about it later today. I'd just like to say that I'm especially glad he's focusing on this subject right now, because I think some of the points he'll make -- and that will emerge from the joint work that we and the FTC will do in the efficiency area -- should be very helpful in enriching the approach that I've been discussing today. For example, the Commission Report has raised questions about how to evaluate and weigh different kinds of efficiencies and the significance of whether efficiencies will be passed on to consumers. These questions are certain to be relevant to efficiency analysis in the non-merger area as well.
I should also add that, although I don't suspect this will happen often, if after doing our analysis we are uncertain about the nature or significance of the proffered efficiencies, my view is that we should go on to consider the anticompetitive effects of the agreement and, if they're aren't likely to be any, then I don't believe a violation has been made out. In the language of the law, what I've just said means that the burden of coming forward with significant evidence of procompetitive benefits is on the defendant, but if the defendant satisfies that burden, we must then prove the absence of such benefits or move on to the third step of the analysis and consider anticompetitive effects. Of course, if there is true uncertainty about the procompetitive aspects of an agreement, it shouldn't require a lot of evidence of anticompetitive effects to tilt the balance in favor of finding a violation.
Lastly, I'd like to take a moment to explain why this approach is at the very least consistent with, and most likely the best reading of, the Supreme Court's recent decisions in this area, namely NCAA13 and Indiana Dentists.14 In both cases, the Court examined seemingly per se-type restraints that inhered in or were ancillary to some larger horizontal arrangement -- an agreement to limit the sale of television rights in NCAA and an agreement not to provide X-rays to insurers in Indiana Dentists. And in both cases, the Court rejected a per se approach, indicating that it was concerned about possible procompetitive benefits that might be at issue. But, at the same time, the Court made clear that there is "often no bright line separat[ing] per se from Rule of Reason analysis,"15 and condemned the restraint in both cases after finding such procompetitive benefits.
It's true that the Court never expressly said that, in the absence of any significant procompetitive benefits, these horizontal agreements would fall without considering anticompetitive effects. But I think that such a reading can be supported not just by the Court's actual method of analysis, but also from the fact that, in both cases, the Court was dealing with traditional per se-type restraints that it surely would have struck down on their face had it not been concerned about their possible benefits. Indeed, the Court said as much in both cases, explaining that "'[a]s a matter of law, the absence of proof of market power does not justify a naked restriction on price or output,' and that such a restriction requires some competitive justification even in the absence of a detailed market analysis.'"16 Finally, I would also point out that, while there were demonstrable anticompetitive output effects in NCAA, as the opinion observed, that wasn't necessarily the case in Indiana Dentists.
All of this leads me to conclude that the approach I've been talking about is soundly supported by the case law,17 although unfortunately not everyone agrees. And so I suppose that, at least until the Supreme Court explicitly endorses our approach, this debate will go forward. In the meantime the Antitrust Division will continue to follow this three-step mode of analysis in our own deliberations, and we will also urge the courts to apply it as well.
Well, even though I'm sure I haven't exhausted the subject that I set out to discuss today, I suspect I've exhausted you and I know I've exhausted myself. So let me thank you for your patience and leave with the hope that I've stimulated some thought about the appropriate method of analysis in cases involving horizontal agreements. Thank you.
3. See, e.g., Richard Steuer, Indiana Federation of Dentists: The Per Se--Rule of Reason Continuum (And A Comment on State Action), 8 Cardozo L. Rev. 1101 (1987); Donald L. Beschle, "What Never? Well, Hardly Ever": Strict Antitrust Scrutiny As An Alternative to Per Se Illegality, 38 Hastings L.J. 471 (1987); Laurence Popofsky, The "Hard-Boiled" Rule of Reason Revisited, 56 Antitrust L.J. 195 (1987); William J. Sims, Note, NCAA v. Board of Regents And A Truncated Rule of Reason: Retaining Flexibility Without Sacrificing Efficiency, 27 Ariz. L. Rev. 193 (1985).
4. See Statements of Antitrust Enforcement Policy in Health Care at 75-76 (1996) (calling for an abbreviated inquiry of any restraint that "facially appears to be of a kind that would always or almost always tend to reduce output or increase prices, but has not been considered per se unlawful"); Antitrust Guidelines for the Licensing of Intellectual Property at 16-17 (1995)(applying abbreviated inquiry to any restraint that "facially appears to be of a kind that would always or almost always tend to reduce output or increase prices"); Brief of United States as Amicus Curiae in NCAA v. Board of Regents at 12-13 (calling for abbreviated inquiry where a restraint "pose[s] high anticompetitive risks because, on its face, it restricts output or restrains price competition.").
5. Thus, we viewed the restriction contained in GE's licensing agreement as comparable to that invalidated by the Supreme Court in Palmer v. BRG of Georgia, 498 U.S. 46, 49 (1990) (per curiam). In that case, the Court held per se illegal a territorial restriction, contained in a licensing agreement, that required that two former competitors not compete in each other's territories. Ibid.; see also Brief for United States as Amicus Curiae in Palmer v. BRG of Georgia at 14 n.14 (explaining why this licensing arrangement was not ancillary to "any efficiency-producing integration of functions or assets by competitors.").
7. See Board of Trade of City of Chicago v. United States, 246 U.S. 231, 238 (1918) ("a good intention will not save an otherwise objectionable regulation or the reverse[, but] knowledge of intent may help the court to interpret facts and to predict consequences"); General Leaseways, Inc. v. National Truck Leasing Association, 744 F.2d 588, 595-96 (7th Cir. 1984) (Posner, J.) (evidence of intent is entitled to little weight as it "cast[s] only a dim light on what ought to be the central question in an antitrust case: actual or probable anticompetitive effect.").
8. United States v. Topco Associates, 405 U.S. 596 (1972), concerned a cooperative buying association formed by several small grocery stores. After examining the buying association's territorial restriction on where its members could sell, the Court held the restraint per se illegal on the ground that, considered as a purely formal matter, the restriction had the characteristics of the type of restraint previously judged per se unlawful. Id. at 608. Although this formalistic reasoning is not sustainable, some commentators have suggested that the Court may have happened upon the right result in that case. See, e.g., Robert Pitofsky, A Framework For Antitrust Analysis of Joint Ventures, 74 Geo. L. J. 1605, 1621 (1986).
10. See Robert Bork, The Antitrust Paradox 269 (1978) ("But considerations of law enforcement efficiency support the invocation of the per se rule against the naked commission-fixing agreement. There being no possibility of efficiency, nothing is lost to society by outlawing the agreement.").
12. See Continental T.V., Inc. v. GTE Sylvania, Inc., 433 U.S. 36, 50 n.16 (1977) ("Per se rules thus require the Court to make broad generalizations about the social utility of particular commercial practices. . . . Cases that do not fit the generalization may arise, but a per se rule reflects the judgment that such cases are not sufficiently common or important to justify the time and expense to identify them.").
17. See, e.g., United States v. Brown University, 5 F.3d 658, 669 (3d Cir. 1993) (court may presume competitive harm and require the defendant to advance a procompetitive justification when dealing with an anticompetitive, albeit not per se illegal, restraint); see also Richard Steuer, Indiana Federation of Dentists: The Per Se--Rule of Reason Continuum (And A Comment on State Action), 8 Cardozo L. Rev. 1101, 1107 (1987) ("[I]f a restraint on price or output is naked, but not illegal per se, it is unnecessary for the plaintiff to prove the relevant market with precision or to demonstrate an actual effect on prices in order to make out a violation. Instead, the defendant bears the burden of establishing some competitive justification."); Diane Wood Hutchinson, Antitrust 1984: Five Decisions in Search of a Theory, 1984 Sup. Ct. Rev. 69, 111 (NCAA "rejected the argument that proof of market power was necessary" and placed "the burden on defendants to show the necessity of adopting a particular [restraint].").