THE NEED FOR OBJECTIVE AND PREDICTABLE STANDARDS
IN THE LAW OF PREDATION
Thomas B. Leary, Commissioner
Federal Trade Commission
Speech before the
Steptoe & Johnson and Analysis Group/Economics 2001 Antitrust Conference
May 10, 2001
My topic for today is predation, with specific emphasis on price predation.(1) The topic is timely for a couple of reasons. First, most of the "post-Chicago" commentary seems to have focused on predation issues and, second, there have been some high visibility predation cases in recent years, like American Airlines,(2) Intel(3) and of course, Microsoft.(4) I want to discuss some of the problems that I see, in a way that I hope will clarify and not confuse. But, that is not a promise.
A little history to begin with. As you know, a new view of predation was one of the cornerstones of what was called the "New Learning" when it emerged some 25 years ago. The focus was on price predation. Historical research was done; people went back and looked at the traditional price predation cases like Standard Oil and came to the conclusion that it was not really all that simple. In addition, theorists advanced the idea that price predation is unlikely ever to make sense because the would-be predator -- by definition, the company with the largest market share - is going to lose a lot more in the course of a price war than its competitors. The chance that these losses can be recouped is slight, not only because of uncertainties and the time value of money but also because if the successful predator ever tried to behave like a monopolist and raise prices to a super-competitive level, it would just attract new entry which would defeat the game. It is fair to say that this view of predation as a very rare animal became pervasive and quickly gained acceptance in the courts. The high water mark for judicial recognition of this theory was probably the Matsushita case in 1986,(5) in which the Court expressed great skepticism about whether price predation is likely in the real world.
There was, fortunately, already a handy tool to help courts distinguish between price cuts that were or were not predatory. This was the famous Areeda-Turner test, a cost-based standard for predatory pricing, first proposed in an article published in 1975.(6) Generally speaking, the test proposed that prices below marginal cost be considered predatory, while prices at or above marginal cost be considered nonpredatory. Because marginal cost is generally difficult to measure, average variable cost is used in its place. Nuanced critiques from a technical economic point of view have followed, but one or another version of the Areeda-Turner test has been widely accepted. Acceptance may have been encouraged by a general sense that predation was unlikely to happen much anyway.
Enter now "post-Chicago" theory. There has been a lot of writing on this subject and a large number of theoretical models have been constructed that I personally can only understand in translation.(7) These new models and theories tend to challenge prevailing theoretical assumptions of the "New Learning" in two different ways. One way is to argue that a predator may not incur the high up-front costs that have been assumed. One example of this argument is the theory of raising rivals costs. The proposition is that a predatory strategy does not necessarily involve price cuts; predation may involve, alternatively, various strategies that raise the costs of rivals. One example would be an action that preempts access to a particularly valuable input. This is not a costless thing to do. The would-be predator may have to pay a premium for an exclusive arrangement or may need to forego a profit opportunity otherwise available if the predator happens to be the provider of that input.(8) But, even if preemption imposes costs on the would-be predator, the cost calculation may be very different from the cost calculation in a predatory pricing model. Preemption strategies are probably a great deal less expensive.
The second way that models challenge the New Learning is to address the other side of the equation, and raise the possibility of a higher level of recoupment and/or a more certain recoupment. Some of these models suggest that a predator can send a message to competitors that he is one tough guy. Somehow or other, if that message gets across, it may yield reputational benefits which dissuade competitors from aggressive competition in the business directly involved, and may dissuade other people from aggressive competition in other businesses as well. Existing competitors may be reluctant to expand output or potential entrants may be reluctant to enter. These models suggest a potential for a much larger payoff and a much more certain payoff from predatory conduct than people originally supposed.
There are other theories that rely on market imperfections. Consider the example of railroads. They connect end to end all across the United States; there is no single railroad that goes everywhere. A few people here, who are veterans with me of the rail merger wars, will remember conventional economic learning held that a railroad would not try to keep traffic on its own long haul if a segment of that route could be more efficiently served by another railroad. Yet we observed that when railroads made end-to-end acquisitions, the first thing they were likely to do was cancel or downgrade the interchanges with non-affiliated carriers. Why did they do that? Remember the perhaps apocryphal story of the bumblebee. As a matter of aeronautic theory, it is supposed to be impossible for a bumblebee to fly but the bumblebee does not know it and it goes ahead and flies anyway. That may be the explanation here - a lack of information. If you are a rail employee responsible for selling traffic, and you do not really know whether the competitor's railroad is or is not more efficient, the safest option is to sell the long haul on your own railroad. Nobody is going to criticize you too much if your decision favors the home team. There also may be an agency problem, depending on the railroad's compensation structure. If you are a salesman and you are compensated on the basis of the volume of sales rather than the profitability of sales, you are motivated to keep as much of the traffic haul inside as possible. So, behavior contrary to the predictions of theory may be based on an information problem or an agency problem.
Now, it is important to remember the limitations of these various theories and models. They do suggest that predation is not necessarily an irrational strategy from the standpoint of the actor. Predation could be more rational than we once thought it was, and hence more common. But we do not really know because the models do not provide an operational basis for deciding when there is predation and when there is not. Therein lies the problem, and that is what this talk is about.
The Existing Law
What does the law say?(9) The stated tests for finding illegal monopolization or attempted monopolization are nearly parallel. In a monopolization case, a key test is whether or not a company has monopoly power. In an attempt case, the test is whether or not the company is likely to get monopoly power. As a second element, general intent is supposed to be enough in a monopolization case, while it supposedly is necessary to prove specific intent in an attempt case. Finally, both kinds of cases supposedly require proof of conduct that can be called "anticompetitive." These are supposed to be independent elements of proof.
How do these tests work out in practice? If you look at the issue of monopoly power or the likelihood of obtaining it, there initially tend to be safe harbors based on market share. In a monopolization case, if the market share is under 70 percent, counsel probably do not have to worry too much; in an attempt case, the dividing line may be around 50 percent. These safe harbors are useful, of course, only to the extent that you can figure out what the relevant market is. I talked here last year about how hard it may be to do that.(10) Although the market-power element may be less controversial in the literature up to now, it may become more controversial down the road, as courts realize that market definition is an increasingly complicated subject.
Intent is next. It is very hard to determine the intent of a large organization with any degree of accuracy. In one of the rail merger cases referred to earlier, there actually was a memo by the chairman of the board that said the proposed deal would permit the merged company to charge higher prices. (Needless to say, the memo was prominently highlighted in the opposition briefs.) But, that does not happen very often.
Perhaps because "intent" evidence is likely to be ambiguous, courts tend to say intent is probably irrelevant if there is "anticompetitive conduct." Other courts will purport to apply an intent test, but say that anticompetitive conduct is evidence not only of general intent in a monopolization case, but also specific intent in an attempted monopolization case. In this group of cases, the decisive factor (once past the market-share screen) becomes the definition of anticompetitive conduct. What is anticompetitive conduct?
Something that looks like an intentional tort to a lawyer is anticompetitive conduct. If you burn down your rival's plant or gratuitously slander his reputation, it should qualify, but that kind of thing is very rare. More common strategies, like price cuts or exclusive deals, are competitively ambiguous. Absent an Areeda-Turner cost-based presumption, the issue of anticompetitive conduct may be determined by evidence of anticompetitive intent. Bad intent may support an inference of bad conduct just as bad conduct may support an inference of bad intent.(11) Supposedly separate elements are related to each other in a circular way.
The point here is that there may not really be three independent elements of a predation claim, as conventional wisdom suggests. The tests all tend to implode inward to an inquiry about the likely competitive effects. (There may be evidence of actual competitive effects if there is a lag between the conduct in issue and the legal challenge, as in American Airlines - but liability is still dependent on subsequent rather than immediate effects of the conduct.(12)) Any company with a substantial enough position to pass through the market-share screen up front, will be vulnerable to retroactive monopolization claims if it does something aggressive in the marketplace and there are not some objective standards that can be applied prospectively.
The Importance of Predictable Standards
The importance of this issue will be evident to anyone who has had experience counseling a corporation that may be vulnerable in this way. That counselor's perspective is very different than the perspective of a scholar or a government decision maker.(13) Someone who works inside a company or advises a company on a day-to-day basis may have to decide very quickly whether the company can legally cut a price, and if so, how much. There is not enough time to do complicated economic studies. There is not enough time to do a massive file search to see whether there are bloodthirsty sentiments expressed somewhere in the bowels of the company. There is not enough time to recast the company's accounting system in order to accommodate the most sophisticated and nuanced expressions of an Areeda-Turner test.
When I was working for a company that was then popularly supposed to have market power, I found that one practical approach was to ask the business people whether they expected "to make any money" under the newly proposed prices. And if they asked "by what standard," I would say by "your standard." The question is "do you anticipate that you are going to make any money, as you understand it, at this price?" If the answer was yes, then that may be enough. If the answer was no, then the next question is "why are you doing it?" That might yield some interesting answers. Based on this experience, I tend to have a predilection in favor of presumptive standards like this that can be applied prospectively.
If you go beyond that and abandon a cost-based standard, what are some of the problems that arise? A number of alternative approaches have been, and still are, being proposed. This talk cannot deal with each of them. To illustrate the issues, I will focus on a conspicuous recent example: the proposed test that the court addressed in American Airlines. (I do not know whether the court's opinion precisely captures the government's views, but that does not affect the value of the illustration.)(14) I see six separate problems. Others can probably identify more.
First, the court characterized one of the government's proposed tests as an inquiry into whether the defendant failed to maximize its profits.(15) According to the report of the government's expert, this proposed test would require a determination of "whether the incumbent had clear alternatives that the incumbent knew or could reasonably be expected to have known would have made it more money absent any predation profits."(16) What does that mean? As soon as there is new entry or some kind of price action by a competitor, the profit maximizing price of the dominant firm or any firm will change. And the profit-maximizing price continues to change depending upon what that competitor or other competitors do with their prices, and depending upon how much business they capture. As a practical matter, the market's adjustment to a new equilibrium is likely to be an iterative process. If a plaintiff or a prosecutor were to claim that any price below this constantly changing profit maximizing price is presumptively suspect, the test is not only impractical but also anticompetitive.
A second issue in one of these cases is whether the dominant firm can, at the least, meet competition. This can be a problem in either the monopolization or the attempt context. Leaving aside the government's argument in American Airlines that meeting was actually beating because of service differentials,(17) is it necessary for the defendant firm to hold an umbrella at some level above the competitor's price? This is a very strange concept.
What is the established firm supposed to do if there is a new entrant building its market share? Is it supposed to continue charging its profit-maximizing price on the assumption that the old competitive dynamics still exist? Is there some kind of a market position that it has to cede to a new competitor? I doubt very much that any company could fine tune its market responses on the assumption that it is legally required to cede the new competitor a certain amount of market share, and I doubt that we would want it to.
On the other hand, recognition of a "meeting competition " defense necessarily implicates a third issue. Should it make a difference whether the defendant's price responses are offensive or defensive? This is likely to be the difference between attempted monopolization and monopolization. In attempted monopolization, the arguably predatory price action is often offensive; in monopolization, the arguably predatory price action is often defensive. Generally in life, we tend to be more accepting of defensive conduct; we seem to have some primordial sympathy for retaliation, as opposed to aggression. It is one of the underlying principles of the United Nations, but does it make any sense as antitrust policy? If you base antitrust policy on that primordial feeling you are going to treat existing monopolists more benignly than would-be monopolists, which may be a backwards way to look at things. On the other hand, for the reasons expressed above, it is unappealing to argue that competitors should not respond.
A fourth issue is whether it is relevant that the entrenched firm may want to cultivate a reputation for aggressiveness. This is an explicit part of some economic models, but it is difficult to apply in practice. The reputation argument was rejected by the court in American Airlines for lack of empirical evidence.(18) The interesting question here is how do you fine tune your reputation? How can public policies be based on the notion that it is alright -- even desirable -- for a company to seek a reputation as an aggressive competitor, but an offense for the company to seek a reputation as a punitive competitor? How does a company control the subjective response to its actions? Think about it.
Fifth, what about a test that would take into account a history of "predation?" Is this the first time a company responded and, say, squelched the would-be competitor and caused it to exit, or has it happened before? Is that material? (If people keep trying to enter over and over and over again, it might suggest that the company does not have much of a reputation for predation, or it could suggest that its prices are well above competitive levels and the newcomers are just inept.) Maybe there is something else going on that we don't really understand. By the way, consider how this would work in a private antitrust action. The first would-be entrant who claims it was driven out does not have a cause of action but the second or the third entrant does. And if the second or the third would-be entrant does recover, can this retroactively create a cause of action for the first entrant that it would not have initially? There are some interesting questions about the impact of history.
Sixth and finally, consider another intriguing argument that apparently was made in the American Airlines case. According to the court, the claim apparently was made that, even if prices were non-predatory under a marginal cost test, it was overly aggressive for American Airlines to expand its capacity. The court said this test does not make any sense because it would suggest that it is alright for the defendant company to cut price, but it is wrong to gear up to serve the new customers that are attracted by the lower prices.(19) Surely, the court's criticism is correct. We cannot have a rule of law that requires the defendant company to engage in some kind of bait-and-switch operation.
There was not time in this talk to discuss each of the innovative proposals that have been proposed as alternatives or supplements to the traditional cost-based tests for predation. The tests discussed by the court in the recent American Airlines case were selected to illustrate the difficulties involved when the traditional tests are abandoned -- difficulties that are particularly apparent when you consider the practical implications for business decisions. I believe that policy makers should be very wary of innovation in this area.(20)
There are very serious theoretical and practical problems that will arise if we abandon objective cost-price tests, that companies can apply up front. I recognize that these tests may have to be modified or refined when applied to industries with virtually no marginal costs, like some in the high-tech arena, but it is worth the effort to do so.(21) This is a classic illustration of the tension between the goals of accuracy and predictability. I suggest that predictability is uniquely important here, because pricing decisions typically have to be made very quickly. Companies should be encouraged to price aggressively and promptly.
Mergers seem different. Merger standards can tilt more in the direction of accuracy, because people typically have a longer time to consider the competitive implications. In addition, I have a sense that companies do not have the same "right" to merge as they have to unilaterally price their products as they please. This may be a visceral response, like the offensive/defensive distinctions discussed above, but there also are some rational reasons for it. The Hart-Scott-Rodino law reflects a policy decision that mergers are special; above a certain size, they must be notified in advance and everything slows down. Pricing decisions normally do not have to be notified in advance, even in some of the regulated industries where it was once required. Pricing flexibility is now encouraged. In addition, unlike mergers, pricing decisions can be readily reversed. Merger policy can, therefore, afford to be more holistic and open-ended.
Relatively objective standards that can be applied up front are not only essential for business decisions, they also facilitate summary disposition by courts. It is important to remember that if we were to adopt the more holistic tests that have been proposed, the scholars who propose them are not going to be the ones who apply them. The decisions are likely to be made by twelve citizens selected at random from a jury pool. The potentially chilling effect on pro-consumer competition is obvious.
1. I want to acknowledge the assistance of my advisor Thomas J. Klotz in the preparation of this speech, but take full responsibility for the conclusions and do not purport to speak for any of my colleagues on the Commission.
2. United States v. AMR Corp., 2001-1 Trade Cas. (CCH) ¶ 73,251 (D. Kan. 2001).
3. In re Intel Corp., No. 9288 (FTC Aug. 3, 1999), </os/1999/9908/intel.do.htm>.
4. United States v. Microsoft Corp., 97 F. Supp.2d 30 (D.D.C.), judgment entered, 97 F. Supp.2d 59 (D.D.C. 2000), appeal docketed, No. 00-5212 (D.C. Cir. June 13, 2000).
5. Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574 (1986).
6. Philip Areeda & Donald F. Turner, Predatory Pricing and Related Practices Under Section 2 of the Sherman Act, 88 Harv. L. Rev. 697 (1975).
7. I am fortunate to have an advisor who is competent in the field and can translate the economic literature for a lawyer.
8. General Motors, my old employer, used to provide transmissions for some of its competitors. The story is told that Alfred Sloan, who established the company's guiding business philosophy, was asked by one of his directors why he continued to sell these parts to the Ford Motor Company. He is supposed to have replied that he liked making money, not only when General Motors sells cars but also when Mr. Ford sells cars.
9. See generally ABA Section of Antitrust Law, Antitrust Law Developments 229-302 (4th ed. 1997).
10. See Thomas B. Leary, The Significance of Variety in Antitrust Analysis, 68 Antitrust L.J. 1007 (2001).
11. See, e.g., Aspen Skiing co. v. Aspen Highlands Skiing Corp., 472 U.S. 585, 602 (1985) ("evidence of intent is . . . relevant to the question whether the challenged conduct is fairly characterized as 'exclusionary' or 'anticompetitive' . . . or 'predatory'"); A.A. Poultry Farms v. Rose Acre Farms, 881 F.2d 1396, 1400-01 (7th Cir. 1989) ("Frequently, courts use intent to resolve ambiguities in interpreting price-cost data; sometimes, though courts assume that bad intent is unlawful and use price-cost data to infer it").
12. Contrary to the general impression, monopolization and attempted monopolization are both incipiency statutes. The harm we are concerned about is future harm in both types of cases. In a monopolization case, as in an attempt case, the immediate impact on consumers of the supposedly predatory conduct may be benign or favorable.
13. Thomas Leary, Do the Proposals Make Any Sense From a Business Standpoint? Con, 49 Antitrust L.J. 1281, 1285-88 (1980).
14. I am not personally familiar with the record in American Airlines, and these remarks should not be interpreted as a view on the ultimate merits of the case. There was, for example, also a claim that the defendant's prices were below an appropriate measure of cost, and there may have been unusually compelling evidence of specific intent.
15. United States v. AMR Corp., 2001-1 Trade Cas. (CCH) ¶ 73,251 at 90,195.
16. Id. at 90,181. The test, as stated, apparently would not require a defendant to choose the most profitable alternative that it could be expected to know about, but the distinction between a more profitable strategy and the most profitable strategy may be more semantic than real.
17. Id. at 90,200.
18. Id. at 90,205 - 06.
19. Id. at 90,190.
20. There are many other aspects of predation that could be the subject of a separate talk, like refusals to deal or so-called "predatory innovation," which are lively issues in recent cases like Kodak, Intel, and Microsoft. These predation issues are difficult today and they will get progressively more difficult as market definition issues, and market-share screens, get harder to apply.
21. Some of the Justice Department's arguments in American Airlines attempt to establish appropriate cost standards in the context of air passenger service, which also has very low marginal costs - at least, in the short term. I do not know enough about the facts of the case to have a view on the precise standard proposed, but I believe this is the right path to follow.