Antitrust error.

• Author: Devlin, Alan

ABSTRACT

Fueled by economics, antitrust has evolved into a highly sophisticated body of law. Its malleable doctrine enables courts to tailor optimal standards to a wide variety of economic phenomena. Indeed, economic theory has been so revolutionary that modern U.S. competition law bears little resemblance to that which prevailed fifty years ago. Yet, for all the contributions of economics, its explanatory powers are subject to important limitations. Profound questions remain at the borders of contemporary antitrust enforcement, but answers remain elusive. It is because of the epistemological limitations of economic analysis that antitrust remains unusually vulnerable to error.

The fear of mistakenly ascribing anticompetitive labels to innocuous conduct is now pervasive. The Supreme Court has repeatedly framed its rulings in a manner that shows sensitivity to the unavoidability of error. In doing so, it has adopted the principle of decision theory that Type I errors are generally to be preferred over Type II. It has crafted a pro-defendant body of jurisprudence accordingly. In 2008, the Justice Department picked up the gauntlet and published the first definitive attempt at extrapolating optimal error rules. Yet, in 2009, the new administration promptly withdrew the report, opining that it could "separate the wheat from the chaff" and thus marginalizing the issue of error. Notwithstanding this confident proclamation, error remains as visible as ever. Intel's behavior in offering rebates has been subject to wildly fluctuating analysis by the U.S. and E.U. enforcement agencies. In a marked departure from precedent, the DOJ is again viewing vertical mergers with concern. And the agency has reversed course on the legality of exclusionary payments in the pharmaceutical industry. Antitrust divergence, both within and outside the United States, remains painfully apparent, demonstrable proof that vulnerability to error remains systemic. For this reason, error analysis may be the single most important unresolved issue facing modern competition policy.

This Article seeks to challenge the contemporary mode of error analysis in antitrust law. We explain the causes and consequences of antitrust error and articulate a variety of suggested cures. In doing so, we debunk the current presumption that false positives are necessarily to be preferred over false negatives. We highlight a variety of cases in which the contemporary bias in favor of underenforcement should be revisited.

TABLE OF CONTENTS INTRODUCTION I. THE ROLE OF ERROR IN COMPETITION LAW A. The Contemporary Role of Error Analysis B. Antitrust's Unique Vulnerability to Error C. Questioning Contemporary Error Analysis 1. Basic Error Analysis 2. Debunking Current Error Analysis 3. The Section 2 Report Debacle II. REVISITING ERROR ANALYSIS IN U.S ANTITRUST LAW A. Rules and Standards B. Constructing Behavior-Specific Error Rules 1. Merger Analysis 2. "Pay-for-Delay" Agreements 3. Refusals To Supply 4. Predatory Pricing 5. Vertical Distribution Contracts, Integration, and Product Tying 6. Should Certain Instances of Price-Fixing Be Analyzed Under the Rule of Reason? C. Putting Faith in the Agencies? Prosecutorial Discretion as a Facilitator of a Permissive Rule CONCLUSION INTRODUCTION

In 2006, the Federal Trade Commission (FTC)and U.S. Department of Justice (DOJ or Justice Department) agreed to undertake a review of antitrust enforcement against monopolistic conduct. (1) This review was to culminate in a joint report setting forth their views and proposing, among other things, a general test for assessing arguably anticompetitive, unilateral conduct. (2) The review was undertaken, but the 2008 report that followed was approved by the Justice Department alone. (3) The general test that it proposed--no enforcement unless evidence demonstrated "substantial disproportion[ality]" between the anticompetitive harm and the procompetitive benefit caused by the conduct in question (4)--adopted the view that antitrust enforcement was so prone to administrative and judicial error that a wide margin of safety was necessary to prevent errors that would punish or discourage vigorously competitive conduct. The FTC strenuously disagreed with that proposal, believing that it unfairly biased antitrust inquiry in favor of dominant firms. (5) In 2009, after the change in administration, the newly appointed head of the Antitrust Division of the DOJ withdrew the report, in disagreement with its view of administrative error, piously announcing that there would be no errors on her watch. (6)

These dramatic events underscore the central role of error in antitrust enforcement and adjudication. They also illustrate the crucial importance of identifying optimal modes of error analysis. This Article seeks to engage contemporary error rules in antitrust and in doing so, explain the source of indeterminacy in this area of the law, reveal why the current mode of error treatment is unsatisfactory, and discern the optimal method of approaching rulemaking in the uncertain world of antitrust adjudication.

Error is uniquely prevalent in this field because antitrust is routinely called upon to deliver answers to unsolvable problems. (7) The intertemporal impact of commercial conduct denies policymakers crucial information about future effects, which, combined with the epistemological limitations of contemporary economic theory, necessitates decision making under uncertainty. The chronic degree of indeterminism that pervades this area of law makes mistaken conclusions understandable--even inevitable--but it does not render them any less costly. For these reasons, it is hardly surprising that competition law displays a unique fixation with error. (8)

Courts, agencies, and academics have reacted to antitrust's unusual vulnerability to error by adopting a bias in favor of false negatives (Type II errors). (9) Believing that procompetitive behavior erroneously condemned will result in a permanent loss of the behavior's benefit, and reasoning that anticompetitive conduct wrongly permitted will be ephemeral due to the market's self-correcting nature, the law seeks to err on the side of underenforcement. (10) This principle has proven to be remarkably influential with respect to antitrust scrutiny of dominant-firm behavior, in particular. Meanwhile, the various premises upon which the modern U.S. error rule has been built have remained unquestioned.

Unfortunately, the role of error analysis in U.S. antitrust law is imperfect. First, the current approach assumes that the risk of mistake is uniform. The law thus posits that all Type I errors are equal in magnitude and probability. Similarly, Type II errors are assumed to be constant. Second, the law acts as if error is apt to arise to the same extent for all offenses, other than for those condemned as per se illegal. This is particularly true of violations allegedly committed by a monopolist or within a vertical market structure. Yet, there is limited theoretical or empirical basis for such an assumption. Third, the law assumes that the same error rule--Type II mistakes are to be preferred over Type I--should apply to all instances in which the risk of error is present. This uncritical conclusion results in an overbroad heuristic that departs from the optimal rule with unacceptable regularity.

The prevailing response to the systemic problem of indeterminism is therefore unsatisfactory. We suggest in particular that antitrust law's unwavering promotion of false negatives over false positives is nonoptimal, and reflects inadequate engagement with the substantive problem. To construct an optimal heuristic, one must first determine the frequency with which the two forms of error actually occur in administration and litigation. Next, one must determine the severity of the errors in each category and the likely temporal range of this severity. In particular, one must determine the length of time necessary for Type II errors to be washed from the system--that is, for the market to return to its predistortion competitive state.

The law assumes that Type I errors result in the perpetual loss of efficiencies, but what about the second-best solutions that markets will devise in response to Type I errors? Surely some of these will be almost as desirable as the impugned behavior. If so, this suggests that the true cost of Type I errors might be lower than claimed. Moreover, although currently assumed, it is not the case that erroneous rules are perpetual. Nor can the market always be trusted to correct anticompetitive conditions mistakenly condoned. Antitrust history is replete with examples of improper rules quickly being circumvented long before they are formally overruled. Furthermore, recent experience suggests that monopolistic behavior may not always be eliminated by the market in a timely fashion, especially where powerful network effects are present.

These nuanced points have largely escaped the attention of enforcement agencies, courts, practitioners, and academics. The current default rule in favor of underenforcement thus lacks a solid intellectual foundation and needs revision. We find that one-size-fits-all error rules are unlikely to yield optimal enforcement and liability determinations. Instead, behavior-specific heuristics that depart from a dogmatic aversion to Type I errors in appropriate circumstances can and should be developed. This Article explores a range of commercial conduct and suggests that scenarios exist in which an unqualified predilection in favor of false negatives is unsuitable.

Although we conclude that antitrust should depart from its predisposition to avoiding Type I errors in certain settings, even then it should approach lawsuits initiated by private parties with a high degree of suspicion. (11) Private plaintiffs' assertions of anticompetitive conduct are paradigmatically noncredible, which ought to color courts' reception of...