You Say You Want an Antitrust Revolution? The Paradox of Modern Merger Control

A. Narrow Markets, More Merger Candidates

As just described, the HMT often leads to exceedingly narrow antitrust markets, corresponding neither to a business nor a popular interpretation of the locus of firm interactions and competition. It directs attention away from competition in broader product markets and away from competition in a broader sense than with respect to specific products. It suggests a truly reductionist view of antitrust concerns. For example, in the Polypore matter, the Federal Trade Commission (FTC) defined four separate antitrust markets where competitive harms were alleged. ⁷ These included “deep-cycle battery separators used in golf carts,” “motive separators used primarily in forklifts,” and “interruptible power supply separators used in batteries for backup power.” In its challenge to General Electric’s acquisition of Instrumentarium OYJ, Department of Justice (DOJ) set out a market consisting of “C-Arms,” which were described as “fluoroscopic x ray devices that offer real time, continuous images during certain medical and surgical procedures.” ⁸ And the DOJ complaint against the merger of Verso Paper and Newpage Holdings listed as markets “coated freesheet web paper” and “coated groundwork paper.” ⁹ These are not what comes to most observers’ minds when the term “market” is used. Indeed, it is not what most economists would use as examples of markets in common discussion or in tutoring students. But these are what modern merger control policy directs attention to.

Another class of cases involves retail markets where product lines and brand names are important. In such cases, the HMT instructs segmenting a product lineup into multiple “antitrust market” components based on the view that consumers shop and choose only within narrow segments. The significance of this is that when companies within the same segment seek to merge, they are likely to face challenges, but if they merge across these distinct antitrust market segments, the standard approach would likely permit such a merger, despite the fact that it might substantially strengthen product groups and powerhouse brands.

For example, despite the appearance of eight or nine car rental brands, numerous mergers have resulted in only three major groups. Brands have been allowed to consolidate because the antitrust market is not “car rentals” but rather “business car rentals” versus “vacation car rentals” and similar segmentation. This argument has been supported by data showing differences in prices and consumer preferences between these, suggesting that such categories are not viewed as strongly competitive with each other. This narrow view of antitrust markets has justified a series of consolidations of a business-oriented rental car company and a vacation-oriented rental car company, with the result that there are now only three groups—Hertz, Avis, and Enterprise—each with at least one branded rental service from each segment.

This case is by no means unusual. Other examples of products where vast consolidation has occurred based on seemingly noncompetitive segments include beer, eyeglasses, certain clothing brands and stores, hotel chains, writing instruments, and blades for shavers. The result is that increasingly consumers find that apparently different brands are not so different after all. Many prove to be under common ownership and are operated and marketed in a coordinated and generally less competitive fashion. And the rationale for permitting such consolidation is often narrow “antitrust markets.”

B. Narrow Markets and Potential Competition

Narrowly defined antitrust markets exacerbate another difficulty faced by merger enforcement, namely, that involving potential competition. The doctrine of potential competition is directed at mergers between an incumbent and a firm not currently producing in the antitrust market but which has the production capability to enter the market quickly and easily enough to constitute a competitive threat to incumbents. But narrowly defined antitrust markets essentially draw a circle around a small—the “smallest”—number of incumbents, thus leaving a large—the “largest”—number of firms outside the market. That implies some number of firms that might be included in a different and broader view of the market have been effectively defined out of the market and converted into potential as opposed to actual competitors.

This has significant consequences for antitrust enforcement since potential competition mergers are not evaluated by the same standards as those involving two incumbent firms but by a more stringent standard. The Supreme Court has stated that in order for a merger between an incumbent and a potential competitor to be anticompetitive, four facts must be established: The market must be concentrated. The potential competitor must have “the characteristics, capabilities, and economic incentives to render it” a competitive threat. It must be either unique or at most one of very few such well-positioned firms. And lastly, there must be actual evidence that the outside firm has “in fact tempered oligopolistic behavior” by incumbents. ¹⁰

This standard—especially the last criterion—goes well beyond that for mergers between incumbent firms. This standard is a problem in itself since it has handicapped enforcement actions against many mergers that eliminate potential competitors. But in conjunction with narrow antitrust markets that classify more firms as outside the market, it has had the further effect of subjecting more mergers to this standard of proof that is considerably more difficult for the agencies to satisfy. Potential competition mergers are both more numerous and more difficult to challenge.

For these reasons, narrow markets have facilitated acquisitions by incumbents of numerous firms in the competitive space surrounding them. This has arguably reduced the population of firms overall and certainly the population of firms that might pose indirect threats to incumbents or future threats to invade their competitive strongholds. That in turn may account at least in part for the decline in overall firm numbers that was previously noted and for the increase in entry barriers into many industries. It is also perhaps evidenced in the vast number of acquisitions by the five major tech companies over the past twenty years. Amazon, Apple, Facebook, Google, and Microsoft have collectively been responsible for some 700 acquisitions during this time, some number of which, in retrospect, may have represented possible rivals to their dominance.

C. Narrow Markets and Expansive Remedies

Yet another consequence of narrowly defined antitrust markets concerns merger remedies. Remedies—particularly divestitures—have long been used as a device to resolve some competitive concerns with mergers without having to litigate their complete prohibition. While appropriate in some cases, remedies have recently come to be used in a wider set of circumstances and in more novel forms, to the point that they have nearly displaced challenges to mergers altogether. Economic analysis, data, and case study experience, however, have demonstrated that remedies are often ineffective in preserving competition, a result that has led to reconsideration of their extensive use. ¹¹

That debate will not be recounted here, but it is relevant as a background fact for one aspect of remedy policy that has not received much attention. Narrowly defined antitrust markets can facilitate the use of remedy policy, especially divestitures that require finding a party outside the relevant market to which the overlapping asset or operation between the merging firms must be sold off. A narrowly defined antitrust market implies that the set of such outside firm in order to resolve the competitive problem is correspondingly larger and would include some firms closer in demand or production space to the merging firms than would otherwise be the case. Put differently, drawing a small circle around the “antitrust market” by definition qualifies more outside firms as potential buyers, even if they are still close enough to be more broadly competitively relevant.

This problem reportedly surfaced in efforts to remedy the merger of Bayer and Monsanto in 2017 in the United States and European Union. There were apparently difficulties in finding suitable buyers of Bayer’s Liberty and LibertyLink divisions which owned a type of gene that is tolerant of certain herbicides. The difficulty arose since there were so few remaining companies with suitable complementary capabilities in the underlying “traits” to which the divisions could be divested—the result of prior consolidation in the sector. ¹² It was further reported that identifying a possible divestiture buyer would be easier if the “trait” product was defined more narrowly since that would qualify additional companies as being in seemingly unrelated businesses and therefore potential buyers. Ultimately, the operations were sold to BASF Ag.

Mergers in the pharmaceutical sector raise similar issues. The standard approach to resolving competitive concerns with overlapping product portfolios has been to require divestiture of one of the two overlapping products to another pharmaceutical company that does not have the very same drug in its portfolio or under development. This practice has resulted in a remedy policy that is little more than a constant rearranging of products among the portfolios of the same limited set of drug companies.

A graphic example of this remedy practice is the FTC’s method for resolving the Teva-Allergan merger in 2106. ¹³ These companies were the first and third largest generic drug manufacturers, each with portfolios of about 400 specific drugs. The two companies had about eighty overlapping products—20% of each company’s entire portfolio. The FTC required divestitures, but it proved impossible to find a single buyer, or even two or three, that did not already have one or more of those products in its own portfolio. As a result, the overlapping products were eventually scattered among ten different companies of varying size and significance. It is not at all obvious that this kind of rearrangement of assets and capabilities in the industry in the guise of merger remedies has preserved, much less strengthened, competition in the industry.

A. Contrasting Unilateral and Coordinated Effects

Given this ever higher burden associated with a coordination theory of merger analysis, it seemed unquestionable to represent a major advance when the 1992 revision of the Merger Guidelines introduced a different mechanism through which a merger could result in competitive harm. This mechanism was the so-called unilateral effects, and it was based on formal economic theory of competition between two firms selling products that are partial substitutes for each other. Each firm is restrained in raising its own price by the resulting loss of sales to other firms, one of which is the acquisition target. But by acquiring that other firm, the first firm recaptures some fraction of the otherwise lost sales, thereby justifying a price increase that would otherwise be unprofitable.

The economic theory underlying this model is straightforward and unimpeachable. Even better, the theory underscores the crucial importance of two potentially measurable parameters—the degree of substitution between the products and the relevant profit margins. For mergers involving partial overlap of products or markets, this theory represented a major advance over the empirically less precise concern over coordination. Unilateral effects mergers do not require coordination. There is no “magic number” of firms to suggest when the concern becomes substantial. The incentives to raise price arise directly out of the characteristics of the products and consumer behavior.

Given these advantages, the agencies have sought to represent as many mergers as possible in terms of unilateral effects rather than the less formal and less precise coordinated effects theory. Moreover, at a time when the judiciary was becoming increasingly familiar and comfortable with, and sometimes even demanding of, empirical validation to competitive concerns, unilateral effects theory seemed to satisfy those expectations. By contrast, coordinated effects allegations cannot match this precision. That theory lacks a clear definition of when coordination began, how it is implemented and preserved, and what the magnitude of any effect might be. Already problematic before the courts, coordinated effects concerns increasingly face the added burden of an unfavorable comparison with a rival theory that offered all of these features.

The sharp contrast with unilateral effects analysis has served as a further obstacle to the agencies’ ability to challenge mergers based on coordinated effects concerns—this despite the fact that the latter are equally meritorious in principal and often more applicable in specific cases. That in turn has led to extreme caution by the agencies in considering bringing such cases, and indeed, there have been very few recent challenges to mergers alleging coordinated effects as the principal competitive concern. ¹⁴ Rapid development of the more rigorous unilateral effects theory has had the unintended consequences of handicapping other concerns.

B. Unilateral Change in Policy Toward Efficiencies

The formal incorporation of unilateral effects analysis in the 2010 merger guidelines has had another presumably unintended effect that has made merger enforcement more difficult. By way of background, it should be noted that the antitrust laws do not provide an explicit role for cost savings from a merger. The Clayton Act explicitly prohibits mergers whose effects “may substantially lessen competition,” without mention of the possibility of offsetting benefits. Consistent with that, the Supreme Court in 1967 declared that “despite widespread acceptance of the potential benefits of [merger] efficiencies,” they “cannot be used as a defense to illegality” because Congress “struck the balance in favor of protecting competition.” ¹⁵

That said, economics has long viewed efficiencies as integral to the competitive analysis of mergers, since from a strict economic efficiency standpoint, cost savings can offset the deadweight loss and consumer harm from a price increase. ¹⁶ Past Merger Guidelines have walked a fine line with respect to the treatment of efficiencies, seeking both to recognize their potential importance but to avoid an explicit statement of their precise role or weight in the overall merger analysis. That line was expunged in the 2010 merger guidelines and not as a result of a deliberate process. Rather, it incorporated a version of the unilateral effects model that explicitly assumed a trade-off between anticompetitive effects and cost savings from a merger.

The rationale for including a role for efficiencies is that the simple and basic version of unilateral effects theory is based on the economic model of differentiated product Bertrand pricing that implies that any merger between substitute products results in a price rise. That price rise might be quite small for products that are weak substitutes, but the implication of a price rise is inescapable under its assumptions. As an analytical tool for economics, that may be perfectly acceptable, but for antitrust litigation purposes, a model that always shows a price increase can be—and has been—criticized for prejudging the answer to the critical question of whether and how price might change as a result of a merger.

Accordingly, starting with the 2010 Merger Guidelines, the earlier unilateral effects model was modified to allow for the possibility of no price change or even a price decrease. The theoretical modification required was to integrate cost savings directly into the model and thereby create a “generalized” measure of pricing pressure in which the inevitable upward pressure on price from product substitution could be offset by the price-reducing effects of cost-saving efficiencies. While again perfectly sensible from an economic perspective, this modification of the Merger Guidelines was no minor change: It represented a full incorporation of efficiencies into merger analysis, the very thing that the courts had declined to do, and that previous Merger Guidelines had carefully avoided doing in order to preserve discretion over the treatment of efficiencies. That discretion has been effectively sacrificed as a by-product of this generalized unilateral effects modeling.

The result of these changes is that cost-saving efficiencies and other merger-related benefits are now fully incorporated in the process of reviewing mergers. ¹⁷ This is a policy change of no small significance but one that crept into the merger guidelines in the guise of a better model of unilateral effects, without much debate or recognition of its import, and seemingly in contrast to established judicial views.

C. The Ever-Rising Burden of Proof

As noted, one of the distinctive features of the unilateral effects approach is its potential to simulate and quantify the impact of any merger to which the model applies and for which a few simple parameters can be measured and applied. Demonstrations of the potential of these models in some merger cases have become nearly legendary. For example, the FTC’s challenge to the proposed merger of Staples and Office Depot rested on a showing that these were two of only three companies that shared a customer base, so that a merger between them would serve to recapture a significant fraction of otherwise lost sales and thereby justify a price rise that would otherwise not occur. ¹⁸ While less formal than later demonstrations of unilateral effects theory, this conclusion was buttressed by documentary evidence from the parties as well as evidence from localized prices reflecting geographic market competition. These factors combined to produce a dramatic—and certainly convincing—case against the merger.

But this case also contributed to another unfortunate side effect of modern antitrust economics, namely, the expectation that challenges to mergers generally could and should include convincing demonstrations along those lines. ¹⁹ This is incorrect and unfortunate. It is incorrect since even when such models are used and simulations run, their predictions are often based on unrealistic promises about what can be delivered. Demonstrations may not fully explain their limitations, including sensitivity to assumptions about the shape of demand curves, the extent of any efficiencies, and constancy of firm behavior. Regardless, where such demonstrations are not provided by the antitrust agency, there is an increasing tendency for courts to be dissatisfied with other evidence.

Moreover, as such models have gained prominence, reliance on them has prompted counterdemonstrations by defendants. They offer alternative models, data, parameter values, and estimation techniques, inevitably with different conclusions. This has shifted the focus of debate but not necessarily narrowed differences between parties and rarely brought clarity to the court tasked with understanding the economics of the merger before it. Rather, some cases have devolved into arcane debates between economists about facts and factors that may well be significant but may also appear to miss important issues of competition.

A recent and telling example is the Justice Department’s challenge to AT&T’s proposed vertical acquisition of Time Warner. The government’s case rested in no small part on a demonstration of the gain to AT&T’s DirecTV division from the merged company’s withholding Time Warner programming from rivals to DirecTV. Considerable court time was devoted to witnesses explaining the basis for estimating the crucial parameters, namely, the loss of customers by rivals and the fraction of those that DirecTV would gain. ²⁰ This demonstration quickly devolved into a debate as to the quality of data being used for those estimates, the similarity of the scenarios used for comparison, the statistical validity of the surveys of customer shifts, the appropriateness of the time frames over which shifts were being measured, and so forth.

In the court’s final opinion, the court ruled that the Justice Department’s demonstration was not convincing and for that and other reasons the DOJ had not met its burden of proof. What was lost in focus on this issue were the broader and arguably more important issues such as the standards by which vertical mergers should be evaluated, what the major competitive concerns might be, and whether that particular merger raised those concerns in a substantial way. If so, the merger arguably could and should have been enjoined. Instead, much of the debate and clearly one of the major stumbling blocks for the court were such things as whether or not certain survey results were fully up to date.

This example illustrates the ambiguous effect of new theory: While it provides clarity and specificity, it can also draw undue attention and in any event it is less likely to resolve issues than simply to shift the contentious debate between the antitrust agencies and the merging companies to new issues. Moreover, as these techniques become more familiar, they can have the unfortunate side effect of raising the bar for a successful challenge to a merger. Rather than the statutory standard prohibiting a merger that “may substantially lessen competition,” the courts’ expectations seem increasingly to be on a demonstration, if not actual proof, that it will in fact do so. This is not generally a realistic expectation of what economics can provide.

Most recently, this emphasis on a convincing demonstration of the adverse effects of a merger has taken the form of calls for “evidence-based effects” analysis. No one, of course, disputes the enormous and potentially conclusive value of analysis of actual evidence of effects. Where that exists, as in the Staples case, it should be the centerpiece of analysis and demonstration. But the implication of the emphasis on evidence-based effects is that mergers lacking direct and convincing evidence of effects should undergo less if any scrutiny since concerns over their effects are not fully supported. This policy would effectively exempt a wide range of mergers, including most where the concern is with coordinated effects and others that rely on predictions that are inevitably subject to questions and criticism.