COMMENTS OF THE AMERICAN ANTITRUST INSTITUTE
ON HORIZONTAL MERGER ANALYSIS
HORIZONTAL MERGER ANALYSIS
February 10, 2004
2919 ELLICOTT ST, NW o WASHINGTON, DC 20008
Market concentration has often played a controversial role in merger law and policy. In the past, some have argued that "big is bad" and the government must use its antitrust tools to stop all trends toward concentration. On the opposite extreme, others have argued that concentration is never a problem as long as government is not creating or supporting entry barriers, because new competitors are always waiting in the wings, forcing even an apparent monopolist to behave in a competitive manner. Although debates about antitrust have moved beyond these older views, merger enforcement today has been criticized for giving concentration either too little or too much weight. A "post-Chicago" position on the proper role of seller concentration(1) and related issues in horizontal merger analysis that relies heavily on recent advances in empirical economic analysis has not yet clearly crystallized. The American Antitrust Institute(2) offers this Statement, based on extensive conversations and debate within the AAI Advisory Board, as a contribution toward crystallization. While the document attempts to reflect a consensus, it cannot and should not be expected that every member of the Advisory Board necessarily agrees with every word or even with all of the general positions taken.
(a.) Horizontal mergers can raise competitive concerns for a number of reasons. A merger may create opportunities and incentives for unilateral price increases, express collusion or tacit coordination and strategic behavior that artificially disadvantages rivals or suppliers. These effects may lead to higher prices, which are harmful because they transfer income away from consumers and undermine allocative efficiency. They may also lead to higher costs, including the possible creation of so-called x-inefficiency.
(b.) Mergers can also reduce competition along other dimensions, including quality, service, the development of new and better products and other areas that significantly affect consumer choice.In an industry with differentiated products, a horizontal merger may also lead to a reduction in the variety of products, which can also harm consumers.
(c.) Economic research indicates that monopoly slows the pace of innovation. Incumbents may ignore or discourage the development of new products and technologies, particularly radical innovations, and both market and technological uncertainties make it likely that innovations will be forthcoming more rapidly when there are multiple, independent sources of initiative. Enhanced opportunities for express or tacit collusion associated with higher levels of concentration can lead to a reduction in the incentives for innovation and may channel investment by fringe firms or prospective entrants away from projects that would compete against the leading firms.
(a.) Since the Supreme Court's decisions in cases like Philadelphia National Bank, the level of market concentration has played a central role in merger analysis. However, the economics literature of that era that related measures of concentration to profits has been criticized for its over-reliance on questionable measures of profits and its failure to account for factors other than anticompetitive behavior that could explain the correlation between profits and concentration across industries.
(b.) The consensus conclusion from more recent studies using more sophisticated research tools is that increased concentration, at high levels, is associated with higher prices, and is therefore a suitable proxy, at least in the first instance, for an expectation of market power. In particular, as an empirical matter, high seller concentration in a properly defined market with significant barriers to entry is associated with higher prices, all other things being equal, and increases in concentration, particularly substantial ones in markets that are already highly concentrated, may precipitate large price increases.
(c.) Even if one is not persuaded by the economic literature alone, where the literature is inconclusive (as is often the case) it is appropriate to take into account the underlying policies of the antitrust laws, as manifested in legislative history and more than a century of judicial explication, reflecting a preference for open markets and more than a handful of competitors, all other things being equal; and a trust in openness, diversity, and forces of competition.
(a.) Neither economic theory nor empirical economic research supports a single "bright line" level of concentration that separates anticompetitive from benign mergers in all or even most industries. Nonetheless, empirical results are generally consistent with current merger law: namely, that in general a substantial increase in an already high level of seller concentration creates a rebuttable presumption that a merger transaction is likely to have anticompetitive effects. These empirical results also support the appropriateness of a flexible sliding scale approach. That is, the higher the degree of concentration and the larger the magnitude of increase in concentration, the stronger the rebuttal evidence that should be required to overcome the presumption of consumer harm.
(b.) As an empirical matter, small mergers producing a low level of concentration generally are unlikely to be associated with consumer harm. In this regard, cases like Von's Grocery Company obviously no longer reflect appropriate merger policy, despite the statute's incipiency mandate. Even though the Guidelines' statements that low-concentration mergers within their safe harbor are "unlikely" to have anticompetitive effects and "ordinarily" require no further analysis are correct, increased guidance could be provided by specifying those rare circumstances where a challenge might nevertheless be appropriate. These exceptions should be made explicit and transparent, and should be limited to situations involving an industry with a history of collusion, or mergers that involve the elimination of a maverick or a weakening of a maverick's behavioral incentives.
(c.) Another reason why presumptions drawn from high concentration should be rebuttable is the fact that market definition is an imperfect procedure and, as a related point, certain common market definition procedures create the potential for systematic errors in defining markets. Procedures deserving reconsideration include: the use of the prevailing price as the pricing benchmark for the ssnip test ["small but significant and nontransitory increase in price"] for measuring cross-elasticity of demand; the use of critical loss analysis; and the principle that the agencies will adopt the smallest market definition that satisfies the ssnip test. The smallest market principle should be deleted from the Guidelines entirely. The validity of the use of the prevailing price in the ssnip test and critical loss analysis should not be assumed, but rather should be carefully evaluated in every merger investigation.(3)
(d.) Though empirical research admittedly does not support a single "bright line" level of concentration or market share for determining when mergers are anticompetitive, the public identification of rebuttable threshold presumptions has served as a useful policy guide, channeling enforcement discretion and yielding an important degree of predictability for business planning. Recognizing that predictability is limited by the inherent vagaries of market definition and the difficulties of forecasting such factors as future market entry and competitive effects, merger analysis should be not so much a scientific endeavor as an administrable process of applying educated judgment to careful fact-finding within a commonly accepted, albeit ultimately imprecise, methodological framework.
(a.) Merger enforcement, while emphasizing microeconomic analysis, must be carried out in light of the intent behind the antimerger statutes, and it is clear that Congress intended this enforcement to embody an incipiency doctrine. While the Sherman Act blocks mergers likely to lead to monopoly power or the dangerous probability of monopoly power, the Clayton Act is designed to block mergers the effect of which "may be substantially to lessen competition or to tend to create a monopoly." This means that increases in concentration should be prohibited even if the anticompetitive effects might not be quite large enough or certain enough to constitute Sherman Act violations.
(b.) This statutory language and the intent behind the Clayton Act as well as Supreme Court precedent, also require a degree of careful, economically informed prediction on the part of enforcers and the courts. Under the circumstances, errors of both over-enforcement and under-enforcement are inevitable, and the underlying facts and economics will often be inconclusive. The incipiency doctrine means that in close cases decision makers should resolve doubts on the side of blocking mergers that might lead to a reasonable probability of market power.
(c.) Preserving multiple competitors is likely to be an efficient administrative rule in otherwise close cases because mergers, once consummated, are rarely undone. Enforcement policy almost never gets a chance to undo a merger that should not have taken place, but there almost surely will be future opportunities to permit consolidation in the industry in question.
(d.) In the absence of an "incipiency" policy, firms in an industry that might be contemplating consolidation may be induced into merging prematurely. When other large firms in an industry are merging, the firm that waits runs the risk of its later merger becoming the proposed merger that finally triggers agency opposition --even though in principle it is no worse or different from those mergers that preceded and thus got in under the wire. Enforcement decisions ought to take into consideration the likely strategic responses to a consummated merger by rivals and potential rivals.
(a.) At one time, the analysis of coordinated effects in mergers relied too heavily on the level and change in concentration. Expanding the analysis to include other factors has refined the analysis and made it more reliable. Further refinement of this analysis, including analysis of the pre- and post-merger competitive role of mavericks and other merger-induced changes in the likelihood of coordination, would improve predictions of likely merger effects.
(b.) This is not to say that coordinated effects prediction in the merger context should be identical with analysis of cartel incentives in a price-fixing context. The purpose of merger intervention is to prevent a situation that may be conducive to coordination from occurring in the future, not to demonstrate that coordination will inevitably occur.
(c.) With respect to potential coordinated effects, heightened concern has historically arisen around the point at which there will no longer be at least five strong competitors or when a dominant firm may enhance its price leadership role through a merger. We see no reason to revise this general benchmark at this time.
(a.) With respect to unilateral effects, the market shares of the merging firms can sometimes be used as a rough proxy of the closeness of substitution between the brands of the merging firms. However, market shares are at best a rough indicator of substitution and generally are inferior to careful factual and empirical analysis, including estimates of cross-elasticities.
(b.) The apparent minimization of unilateral effects analysis by the current federal enforcement agencies represents a step backwards. Unilateral effects analysis has a substantial history in industrial organization economics and represents a rigorous analytic approach. While there may be some basis for concern about over-reliance on simulation models in their current state of development, as a particular method of demonstrating the magnitude of unilateral effects, there is no good basis for skepticism of unilateral effects analysis itself.
(c.) With respect to unilateral effects, heightened concern has historically arisen around the point at which the leading firm's market share is at least 35%. We see no reason to change this benchmark level at this time.
(a.) When high market shares and concentration resulting from merger create a presumption of consumer harm, the burden should shift onto merger proponents to demonstrate one or more of the following factors(4):
(b.) If one or more of the above is established, the burden should shift to the government or other plaintiff to show that the merger would likely generate a net anticompetitive effect, taking into account all relevant evidence.
There is reason to doubt the empirical significance of the strongest version of the 'contestable market' theory, which holds that potential entry can cause even a monopolist benefiting from significant economies of scale to price competitively. This theory wrongly assumes both that entry requires no significant sunk costs (i.e., the entrant's expenditures on inputs can be fully recovered if entry fails) and that the monopolist's price response to entry is delayed.
(a.) Despite the very limited applicability of the pure contestable market model to real world settings, the more general potential entry concept nonetheless is an important element in the analysis of the likely competitive effects of a merger. On the one hand, potential entrants can reduce the likelihood of anticompetitive effects from a merger, particularly where efficient small scale entry by multiple firms is possible and where entry can be secret or sponsored by large buyers. On the other hand, mergers between an incumbent and a potential entrant can cause anticompetitive harm. Accordingly, competitive concerns may arise from mergers that remove significant potential entry, both perceived and likely actual potential entrants. This is a particular concern in high technology markets, where significant competition may occur well before products are sold to consumers.
(b.) Because of the competitive importance of potential entry in many industries, merger policy should place more emphasis on preventing mergers that reduce potential competition. This is an area where the case law has moved too far in the direction of laissez-faire. Federal and state enforcement agencies should undertake greater efforts to bring appropriate enforcement actions, refine the analysis and educate the courts.
(a.) National merger policy since 1981 has rested on the assumption that most mergers generate important efficiencies and therefore significantly contribute to consumer welfare. This is reflected in the fact that, typically, only 2-3% of mergers large enough to require federal pre-notification are pursued to the second request level of investigation. Yet, respected economic research has found that many, perhaps most, mergers do not lead to significant reductions in cost, although a small proportion of horizontal mergers have led to very significant efficiencies. Many of the predicted efficiencies of mergers have failed to materialize.
(b.) The practical importance of this research is that it is time to re-examine the underlying assumption that allows such a high proportion of significant horizontal mergers to be consummated.
(c.) In the meantime, in specific investigations, claims of efficiency benefits arising from a merger should be viewed skeptically. This is particularly true of theoretical arguments for gains arising from consolidated management and marketing. Moreover, the empirical evidence supporting claims of efficiency gains should be based on the specific cost structure and technology of the firms, and should be accompanied by further evidence that demonstrates how these cost reductions will benefit consumers.
(d.) To be cognizable, efficiencies must be non-speculative, merger-specific, and provide substantial direct benefit to customers. Only efficiencies net of any higher costs caused by the merger represent potential consumer benefits. Claimed benefits that will only arise in the long run are often more uncertain and for that reason should be given less weight.
(e.) Because a high proportion of mergers fail to provide the benefits that were predicted by their proponents and because there are large costs for society when anticompetitive mergers occur, Congress should provide federal antitrust enforcers additional resources to permit more detailed scrutiny of more proposed mergers than is possible today. Enforcers should be encouraged to scrutinize more mergers that might currently be deemed marginal.
Recognition of the failure of so many mergers to produce their predicted benefits suggests that more research be devoted to examination of:
(a.) consummated mergers to evaluate whether or not they led to significant savings and/or price increases;
(b.) proposed mergers that were stopped or restructured as a result of antitrust intervention in order to evaluate the effectiveness of the government's intervention, including the sufficiency of remedies utilized;
(c.) the effects of merger enforcement on innovation, including both the extent to which innovation concerns played a role in past enforcement decisions, and the extent to which merger enforcement and non-enforcement has affected various types of innovation; and
(d.) merger dynamics in network industries, where predictions of merger-enhanced tipping effects may deter entry by potential competitors.
Greater transparency on the part of the government is a necessary foundation for the beneficial evolution of antitrust policy. Although the government has recently made positive strides toward increased transparency, there remains a need for more detailed explanations of the agencies' reasoning with respect to actions taken (and, in certain instances, not taken); for projects like the joint FTC/DOJ compilation of data on completed investigations; and for other initiatives that will facilitate research by the government and by academics. The history of antitrust should not be characterized as pendulum-like, but rather as an on-going dialogue, continually evolving toward a better understanding of markets and competition within the context of a politically-determined legal framework.
1. We have not focused in this statement on concentration at the buyer level, which will be the subject of an AAI conference on June 22, 2004, exploring ways in which buyer power may differ from seller power.
2. The AAI is an independent education, research, and advocacy organization, described on the Internet at www.antitrustinstitute.org. The drafting of this document has been a nine-month iterative process featuring very heavy input by a drafting committee of seven and repeated circulations to the full Advisory Board for comment. The author and the Board of Directors bear responsibility for the final version.
3. The AAI will conduct a symposium on "Combining Horizontal and Vertical Analysis in Antitrust: Implications of the Work of Robert L. Steiner" on June 21, 2004. This will explore whether the role of retailers gross margin is given adequate consideration in market definition and other merger-related issues that arise in the consumer goods sector.
4. Merger case law and the Horizontal Merger Guidelines also properly recognize a narrow failing firm defense.