The 2023 Merger Guidelines: Law, Fact, and Method

Abstract

The final version of the 2023 Merger Guidelines, which were issued in December 2023, is a vast improvement over an earlier draft—which indicates that the Agencies took the many comments that they received on a draft very seriously. These Guidelines break some new ground that older Guidelines did not address, and make many positive contributions, which this paper spells out. They are also excessively nostalgic for a past era, however, and this may explain their propen sity to treat empirical questions as issues of law: This is one way to insulate these Guidelines from further revision. The excessive reliance on one decision, Brown Shoe, is unfortunate—particularly since that decision has been so often repudiated, even by the Supreme Court itself. This paper pays particular attention to: the Guidelines’ treatment of structural triggers and direct measures of competitive effects; their aggressive position on potential competition mergers; their willingness to weigh a “trend” toward concentration as a factor; and their treatment of serial acquisitions. The Guidelines include a welcome new section on mergers involving multi-sided networks, although their view of networks is too one-sided; and the Guidelines also contain an expanded section on mergers with harmful effects on suppliers—including labor. The Guidelines’ treatment of market definition is likely to lead to underenforcement because they define markets too broadly. Finally, the Guidelines could have made better use of recent retrospective studies—many of which would have provided further support for the substantive positions that the Guidelines take.

1 Introduction

The 2023 Merger Guidelines issued by the Antitrust Division of the U.S. Department of Justice and the Federal Trade Commission are comprehensive, novel in many ways, and revisionist.¹ In understanding them as an enforcement tool, it is important to remember that the Agencies do not have authority to “make” new law through Guidelines, but only to interpret existing law and indicate how they intend to exercise their enforcement discretion. The line between “interpreting” and “making” law is notoriously ambiguous, however, particularly in areas such as antitrust, which are not governed by a detailed code.

In the past the lower federal courts have generally been quite receptive to the Guidelines—largely because the Guidelines were sensibly moderate: They reflected mainstream views, and were not written in the context of particular cases that the Agencies were litigating. In the past, attempts to push the law too far have not only fallen on deaf ears, but have also resulted in quick withdrawals of the document in question.² How these Guidelines will be accepted by the courts remains to be seen.

The 2023 Merger Guidelines provide six specific “Guidelines” that describe the “distinct frameworks” that the Agencies will use to identify competitive concerns, noting that the specific Guidelines are not intended to be mutually exclusive. Five additional Guidelines focus on application. Following that, another section discusses rebuttal evidence, and a final section contains a “non-exhaustive discussion of analytical, economic, and evidentiary tools” that will be used for evaluation, including market definition.

The 2023 Merger Guidelines were originally issued as a draft and made available for comments. They received more than 3000 comments.³ The Agencies appear to have taken the comments seriously.

In retrospect, the initial draft was excessively influenced by a Neo-Brandeisian mindset that is both reactionary and backward looking. That could explain the large number of citations to Supreme Court merger decisions from the1960s and 1970s. It may also explain the tendency to assert issues of fact as if they were conclusions of law, thus insulating them from later revision.

The final version is a vast improvement, much more accommodating of significant new learning in merger policy that has been reflected in lower court decisions since the Supreme Court largely abandoned substantive merger review in the 1970s. It is unlikely that the Supreme Court intended for merger doctrine to be frozen in time as of 1975 or so. Instead, the Court likely intended for the lower courts to continue developing merger policy, and to reflect new economic learning in the process.

Nevertheless, even in final form the new Merger Guidelines are a revisionist document. One issue they must confront is attaining judicial acceptance. The way to do that is to tie stricter merger standards to the best available methodology and evidence. The bulk of recent empirical literature supports the proposition that merger enforcement could be improved by stricter structural standards. This depends in large part on high-quality expert opinion and economic scholarship—including retrospective studies of previous mergers.

Further, this approach would shift the focus away from substantive antitrust rules—which sadly carry a fair amount of ideological baggage—to evidentiary standards where the issues concern methodology rather than conclusions.⁴ A federal judiciary that may be reluctant to embrace more aggressive antitrust generally, might be more receptive to arguments that are based on sound science. That science also places the focus where it should be: on post-merger prices and output, which is consistent with the goals of antitrust generally. The Guidelines should acknowledge that economics can be their friend.

This paper examines the various methodologies that the 2023 Guidelines employ, including their treatment of “law” as opposed to “facts.” That distinction is important in the antitrust context because the antitrust statutes use spare language, which leaves the courts to do a great deal of gap-filling. When they fill those gaps, however, are they making law that is equally applicable to all relevant situations and unchanging until overruled? Or are they making statements of fact, which are typically relevant only to the circumstances of a single or at least closely similar cases?

2 Law and Facts in Merger Policy

The legal system addresses law and facts in different ways. A statement of law is durable until it is overruled by a court of the same or higher level. By contrast, facts are specific to individual or similar cases, and they can change. Further, in antitrust facts are often the subject of expert testimony, where they operate under the constraint that the methodology that the expert employs must be generally accepted in the discipline, testable in a meaningful sense, and with methodologies that are up to date.⁵ Economic or other expert testimony as well as juries address only facts; interpreting the law is the work of judges.

In the 1960s the Supreme Court stated in merger cases that:

• Mergers should be subjected to harsher treatment when there has been a “trend” toward increased concentration in that industry;⁶

• Internal expansion yields greater market output than do mergers;⁷

• Markets should be defined by reference to “reasonable interchangeability,” but with no mention of the margins at which that interchangeability occurs;⁸

• Merger analysis requires a market definition,⁹ including recognition of something called “submarkets”;¹⁰

• One indicator of a submarket is “unique production facilities”;¹¹

• A merger of firms with market shares of 4.7% and 4.2% in an unconcentrated market (HHI of 300 and DHHI of 39), is unlawful;¹²

• A merger creating a post-merger firm whose market share exceeds 30% and results in a significant concentration increase is prima facie unlawful;¹³

• “Possible economies” cannot be a defense to illegality.”¹⁴

None of these statements appears in the text of Clayton Act §7 or any other antitrust statute. A few were discussed, although inconclusively, in the legislative history that led to the statute’s passage. Some of them appear to have no support except that the Supreme Court stated it. The triggering market share and concentration levels appear to be nothing more than stabs in the dark, unsupported by any evidence. Some, such as the “reasonable interchangeability” statement, are known today to be economically incorrect—in that case producing the “cellophane fallacy” in market definition, which tends to define overly broad markets.¹⁵

The same thing is true of “unique production facilities,” which Brown Shoe stated to be an identifier of a relevant submarket. A single production facility is often used to produce noncompeting goods in chemicals, software, agriculture, or other products.¹⁶ Further, sometimes identical products are produced in multiple production facilities, whether by one firm or many firms. That is to say, “unique production facilities” can be both overinclusive and underinclusive. Others, such as the statement that internal expansion yields higher output in the affected market than does a merger were given with no apparent support of any kind.

The market share and market concentration rules that were adopted in Von’s Grocery, as well as those adopted in Brown Shoe, appear to have been pulled out of thin air. They have never been followed by any set of Merger Guidelines—including these. The structural standards that were set in the 2023 Guidelines (HHI > 1800, and DHHI > 100) are not even close—notwithstanding that these Guidelines almost slavishly follow Brown Shoe in other areas. Clearly, in this particular case the drafters of these Guidelines realized that this was not a question of law. Today, determining the market share and concentration levels that trigger merger illegality is the subject of a large economic literature, and views have changed over time.

By contrast, in creating the presumption that a merger that yields a firm with a (pro forma) market share that exceeds 30% is unlawful, the Supreme Court’s Philadelphia Bank decision cited prominent antitrust economists of that period.¹⁷ That presumption was acknowledged to present an issue of fact.

The term “submarkets” appears to have been a Supreme Court fabrication out of practically nothing,¹⁸ convincing some courts that established criteria for market definition need not apply to a smaller grouping of sales if it is characterized as a “submarket.” Finally, the merger statute says nothing one way or the other about merger economies, and the law over time has drifted over a wide range: from condemning mergers precisely because they created economies in the 1960s and 1970s,¹⁹ to exonerating mergers based on largely hypothetical or unproven cost savings.

The responses of lower courts have been equally varied. Some have cited these various propositions as if they represented irreversible statements of the law. Others have deviated from them significantly—although typically without observing that the Court was stating only an issue of fact. A case in point is the market share standards in Brown Shoe and Von’s, which no one follows—not even the 2023 Guidelines. One 2023 decision severely but correctly qualified the Brown Shoe “reasonable interchangeability” test, observing that “at a high enough price even poor substitutes look good to the consumer.”²⁰ Numerous courts have qualified the “submarket” idea by stating that a “submarket” definition must satisfy the same criteria as are used to define a market—thus making the idea of a submarket irrelevant.²¹

The Supreme Court all too often fails to tell us whether a particular statement was intended to present a principle of law or a conclusion of fact. Clearly an issue is one of law if the statute states it. For example, §7 of the Clayton Act states that a merger is unlawful if it unreasonably threatens to “lessen competition,” although it gives no meaning to that term. It also makes it unlawful to “acquire” another firm, indicating as a matter of law that the statute does not apply to simple contracts, but only to acquisitions. It also makes clear that both stock and asset acquisitions are covered and that holding companies are permissible.²²

For all other statements, including those in the list above, if the statute does not address them they should presumptively be regarded as statements of fact unless the Supreme Court states that they apply as a matter of law. To conclude otherwise is to freeze the law into a kind of antitrust Dark Ages: intent on the preservation and repetition of existing ideas rather than the acceptance of new ones.

The Brown Shoe opinion itself was helpful, concluding that "Congress neither adopted nor rejected specifically any particular tests for measuring the relevant markets,” that it did not adopt a particular definition of the word “substantially” in the statute, and that a merger must be “functionally viewed” in its particular setting.²³ These statements all indicate that the Court viewed at least some of these issues as ones of fact, subject to market specific analysis or technical change.

The balance of this paper examines a number of issues that are raised by the Guidelines: We seek to determine the appropriate role of law as opposed to facts.

3 Concentration, Performance, and the Philadelphia Bank Presumption

One criticism of the original draft Guidelines was that its large opening section on concentration never associated increased concentration with any measure of performance, such as: higher prices; reduced output; restrained innovation; or the exercise of market power.²⁴ The final version has addressed this issue somewhat; but it remains the case that the Guideline 1 treatment of concentration places little emphasis on these indicia of performance as targets of merger enforcement. That is reserved for Guideline 3; in turn, this separation suggests that the concerns about “concentration” and the concerns about anticompetitive facilitation of coordination are different things. This is in sharp contrast to the 2010 Horizontal Merger Guidelines’ statement of a “unifying theme” that mergers should not increase or enhance market power.²⁵ The 2010 Guidelines’ identification of specific concentration levels was in pursuit of that goal.

The statement in the 2010 Horizontal Merger Guidelines is more consistent with the concerns of current merger case law—although perhaps not those expressed in the 1960s. The exercise of market power is best evidenced by higher prices, although other metrics such as output reduction or diminished innovation are important too.

That concern has broad support today, even among the Justices of the current Supreme Court: Writing for a unanimous Supreme Court in 2013, Justice Sotomayor described merger law as challenging the consolidation of market power.²⁶ That same year, then Judge Gorsuch cited the 2010 Horizontal Merger Guides as illustrating a methodology for proving market power from market share.²⁷ In 2008 then Judge Kavanaugh observed that “merger enforcement, like other areas of antitrust is directed at market power.”²⁸ He was quoting from the Heinz (“baby food”) merger decision, in which then judge Merrick Garland sat on the panel.²⁹ In the 1990 Baker Hughes decision, then Judge Thomas made merger law’s structural presumption a contingent surrogate to measuring market power.³⁰

Judge Kavanaugh’s statement that equated merger concerns with “other areas of antitrust” also reflect hundreds of federal decisions that date to the 1890s that declare that the purpose of the antitrust law to be combatting lower output or higher prices: the indicia of market power.³¹ Further, nothing in this history suggests that merger policy under §7 of the Clayton Act was intended to differ from the other antitrust laws in this regard.

4 The Role of Economists: 1960s and Today

Both in 1950 when the Celler–Kefauver amendments to Clayton Act §7 were passed and also today, economics played an important role in merger analysis. The statute itself states its concerns in unambiguously economic terms that were generally understood by the time the statute was amended in 1950: “substantially lessen competition,” and “tend to create a monopoly.” While participation by economists is well recognized today, they were nearly as important in the 1960s. The Supreme Court’s Philadelphia Bank merger decision a year after Brown Shoe cited no fewer than seven industrial economists, more than any antitrust decision to that time.³²

Antitrust economists in the 1950s and today use different but overlapping methodologies to evaluate competition. Those in the 1950s and 1960s were heavily structuralist: They focused on the relationship between market structure and economic “performance.” Today structuralism is less dominant. Price–cost margins and innovation are more ascendant—mainly because the discipline of economics has developed better tools for assessing them. The prevailing enforcement methodology employs a mixture of structural and performance evidence.

One thing that has not changed, however, is underlying goals. Prominent structuralist economists from mid-century—such as Edward S. Mason,³³ Joe Bain,³⁴ Leonard Weiss,³⁵ Carl Kaysen,³⁶ and Donald Turner (who was the AAG when the 1968 Merger Guidelines were released)³⁷—were uniformly concerned about the threats of oligopoly coordination and lack of price competition.³⁸ Today, the dominant economic literature—including that upon which the Agencies rely—uses different methodologies to address the same concerns.³⁹

On the issue of concentration and performance, the legislative history of the 1950 Celler–Kefauver amendments to §7 is hardly a model of clarity. The Supreme Court’s Brown Shoe decision⁴⁰ compounded the problem by giving a very one-sided view of the issue. Congress was clearly concerned about market “concentration,” and many members expressed the view that it had been increasing. Several members of Congress also expressed concerns that excessive mergers might result in higher prices.⁴¹ No member of Congress spoke in opposition to that view or declared that a concern for higher prices was unimportant. The Brown Shoe opinion amply reflected Congress’ concerns about increasing concentration, but ignored the concern about higher prices.

When not viewed strictly through the lens of Brown Shoe, the legislative history tracked the then reigning “structure–conduct–performance” (S–C–P) paradigm in antitrust economics. Structure was a principal concern, but on the assumption that anticompetitive conduct and poor performance would flow from noncompetitive structures. As structuralist economist Leonard Weiss put it:

The main predictions of the structure–conduct–performance paradigm are: (1) that concentration will facilitate collusion, whether tacit or explicit, and (2) that as barriers to entry rise, the optimal price-cost margin of the leading firm or firms likewise will increase.⁴²

In any event, Brown Shoe’s focus on concentration for its own sake was short-lived. All of the economists that the Philadelphia Bank decision cited 1 year later were members of the reigning structuralist school (the “Harvard School”). For the particular conclusion that a merger should be presumptively unlawful if it creates a firm with a market share of at least 30%, Philadelphia Bank cited four economists—all of whom would have applied a stricter standard than the one that the Supreme Court adopted. These included Carl Kaysen and Donald F. Turner (recommending a 20% trigger for presumptive illegality), George Stigler (20%), and Jesse Markham (25%).⁴³ The Court concluded that it had “no view on the validity of such tests,” but noted that the actual case’s 30% number exceeded all of them.⁴⁴

From that point Supreme Court case law on mergers was increasingly driven by concerns about price or other indicia of performance. One exception was the Court’s 1966 decision in Von’s Grocery, which the 2023 Guidelines do not cite.⁴⁵Brown Shoe and Von’s Grocery should be counted as severe outliers from an antitrust tradition that with few exceptions targets higher prices, lower output, or restrained innovation.

In its 1964 El Paso Natural Gas decision—2 years after Brown Shoe—the Court focused exclusively on price.⁴⁶ El Paso, whose reserves lay to the south and east, was the dominant natural gas supplier to California. Pacific Northwest, with reserves to the north, had repeatedly bid against El Paso for supply contracts into California markets but had always lost the bids. In one case El Paso had to reduce its bid price in order to meet a Pacific Northwest bid.⁴⁷

Justice Douglas concluded that “We would have to wear blinders not to see that the mere efforts of Pacific Northwest to get into the California market, although unsuccessful, had a powerful influence on El Paso’s business attitudes.”⁴⁸ With that, the Court treated the merger as horizonal (“Unsuccessful bidders are no less competitors than the successful ones”) and condemned it entirely on price increasing grounds.⁴⁹

In the Continental Can case that same year⁵⁰ the challenged merger was between a can maker and a bottle producer. At the time competition between the two had been driven mainly by the markets for baby food and beverages. The concern was that the merger limited the ability of large customers to force the can and bottle makers to bid against each other for their trade with the threat of transferring their business. The Court concluded that:

the possibility of such transfers over the long run acts as a deterrent against attempts by the dominant members of either industry to reap the possible benefits of their position by raising prices above the competitive level or engaging in other comparable practices….⁵¹

Then, in 1974 the Court issued a pair of decisions that severely pushed back against Brown Shoe, and rejected the government’s challenges. The problem in the General Dynamics case was that the acquiring firm’s depleted reserves made its current market share an exaggeration of its actual competitive weight.⁵² As a result the Government could not rely exclusively on structural evidence.⁵³ The Court stated the fundamental concern that the defendant’s “power to affect the price of coal was … severely limited and steadily diminishing.”⁵⁴

In Marine Bancorporation⁵⁵—a potential competition merger case—the Court explained that the doctrine applied to concentrated markets in which current participants have “the capacity effectively to determine price and total output of goods or services.”⁵⁶ If the target market were performing competitively, the incumbent firms would “have no occasion to fashion their behavior to take into account the presence of a potential entrant.”⁵⁷ However, the merger precluded entry de novo which would have assisted in “deconcentrating that market over the long run.”⁵⁸ The Court chastised the parties because they never “undertook any significant study of the performance, as compared to the structure of the commercial banking market….”⁵⁹

The Supreme Court’s next word on the issue was in the Cargill case in 1986: a dozen years after the Court had formally stopped reviewing the substance of merger decisions in actions by the government agencies.⁶⁰ In this private merger challenge the plaintiff claimed that after the merger Cargill—a very large beef processor—would reduce its prices, and would thereby injure the plaintiff competitor by forcing it to reduce its margins.

The theory was the same one that the Supreme Court had approved in Brown Shoe, which affirmed the district court’s conclusion that small sellers were having an increasingly difficult time competing with larger firms because they had advantages that “result in lower prices or in higher quality for the same price.”⁶¹ In affirming, the Brown Shoe Supreme Court had suggested that Congress was aware that “occasional higher costs and prices might result from the maintenance of fragmented industries and markets.”⁶² The Court did not cite any legislative history that supported that proposition.

In any event, Cargill effectively overruled Brown Shoe on this issue: Justice Brennan, a pro-enforcement liberal, wrote the Court’s opinion and concluded that to condemn a merger because it resulted in lower prices that harmed a rival was not only incorrect but that it would be “inimical to the antitrust laws.”⁶³

A private plaintiff must meet standing and injury requirements that do not apply to the government. However, the Cargill Court expressly rejected a request from the Government as amicus that competitors be denied standing to challenge mergers.⁶⁴ Rather, it decided the case on substantive antitrust policy grounds. It acknowledged that post-merger predatory pricing could be unlawful, but absent that it was not consistent with antitrust goals to condemn a merger simply because competitors suffered lower margins.⁶⁵ That requirement is clearly a requirement of merger policy—not of private plaintiff standing to sue.

Why the 2023 Merger Guidelines fail to discuss Cargill is unclear, because it appears to overrule Brown Shoe on a fairly critical point. Given that the Agencies have the authority to enforce the law, but not to ignore existing law, this absence of discussion creates the impression that the Guidelines are an advocacy document rather than a statement of enforcement policy.

Finally, any interpretation of §7 that uses a reduction in the number of firms rather than the effect on performance as a metric implies that identical language in, different sections of the Clayton Act mean different things. For example, §3 of the Clayton Act reaches tying and exclusive dealing with the same “substantially lessen competition” or “tend to create a monopoly” language as §7, but tying does nothing to reduce the number of firms.⁶⁶ Early on, the Supreme Court interpreted §3 in tying cases to refer to the threat of higher prices in the tied market.⁶⁷

5 Price-Increasing Mergers

When it comes to mergers that actually threaten price increases or output reductions, the 2023 Guidelines reflect an amply supported belief that the 2010 Horizontal Guidelines were too conservative and permitted many mergers that were likely to have adverse price effects. That tendency was then exacerbated by judicial decisions that tilted even more conservatively than the Guidelines stated.⁶⁸ Indeed, one phenomenon that emerged after the 1992 Guidelines is that the risk-averse and underfunded Agencies themselves did not follow them, but generally limited enforcement to mergers that fell in the highest ranges of the articulated standards.⁶⁹

The 2023 Guidelines should be helpful here. Moving the critical target zone from 2500 to 1800 expands the range of mergers that significantly exceed concentration thresholds. That could make judges more willing to act.

In order to have teeth, the more aggressive standards that are articulated in the 2023 Merger Guidelines must be accompanied by more expansive enforcement activity to include the full range of the stated standards, coupled with courts’ willingness to follow them and Congress’ willingness to pay for it. This is one of the greatest improvements in enforcement that these Guidelines could make—but of course they have no control over judges or the Congress.

Under the 2023 Guidelines, if a post-merger HHI exceeds 1800 and the DHHI (the increase in the HHI) exceeds 100 then the merger is presumed to lessen competition substantially.⁷⁰ The Guidelines use the word “presumed,” as Supreme Court precedent requires. The Philadelphia Bank decision made structure at a certain level decisive “in the absence of evidence clearly showing that the merger is not likely to have such anticompetitive effects.”⁷¹ In the Supreme Court’s Marine Bancorp. decision, which involved a potential competition merger, the question was whether the market in question was highly concentrated and its competitive performance impaired as a result. The Court wrote:

The record indicates that neither the Government nor the appellees undertook any significant study of the performance, as compared to the structure, of the commercial banking market in Spokane.⁷²

The Court suggested an absence of parallel behavior as one type of evidence that would defeat the presumption.⁷³ Others include a record of new entry or shifts in market share, as well as other indicia of actual competition among the firms. Yet another, which the Supreme Court identified in its Cargill decision, was price effects: condemning a merger because it reduced prices would be inimical to antitrust goals.⁷⁴ That limitation is important because market concentration is in fact driven by multiple factors—including scale economies and network effects. Higher concentration can also result in lower prices rather than higher ones, but mergers cannot be condemned on that ground.

One thing that is missing from the 2023 Guidelines is a helpful discussion of evidentiary links between post-merger performance and the particular choices of post-merger concentration > 1800 HHI, a 100-point DHHI, or a single firm post-merger share of 30% as enforcement triggers. As is noted below, recent empirical literature has made good contributions in this area,⁷⁵ and they tend to support the 2023 Guidelines approach; but the Guidelines themselves cite only case law. Brown Shoe—the decision that comes closest to condemning concentration for its own sake—would have condemned a merger at much lower levels.⁷⁶

While the Guidelines rely heavily on that decision for other reasons, they do not adopt its structural metrics, but once again without explanation. The most likely one is that the drafters realized that the choice of a metric is a question of fact, not of law, and on this point the Brown Shoe standards are obsolete.

One of the reasons that these numbers have changed over time is that empirical testing continues to progress.⁷⁷ The 2023 Guidelines are correct to restore HHI > 1800 as a presumptive trigger. Indeed, the evidence may justify an even lower one. As an empirical matter, the increase in the HHI is particularly significant.⁷⁸ The 2023 Guidelines requirement of a DHHI > 100 rather than 200 is, if anything, an even bigger change than the change in the triggering overall concentration level. For example, in a market with HHI > 1800 a merger of a firm with an 8% market share and one with a 7% share could be challenged.

Guideline 1 also indicates that a challenge is proper when the post-merger firm’s market share exceeds 30% and the DHHI exceeds 100. The 30% number references the Supreme Court’s Philadelphia Bank conclusion that a merger in that range triggered illegality.⁷⁹ If the resulting market share exceeds 30%, an acquisition of anything other than a very small firm will also increase the HHI by more than 100.⁸⁰ The minimum increase of 100 seems to be an interpretation of Philadelphia Bank’s requirement that in addition to a post-merger firm > 30%, the merger must also result in a significant increase in concentration.

Most mergers of that magnitude are very likely challengeable in any event. The exceptions are cases where the merging firm was the largest in its market and the others were significantly smaller.⁸¹ This is effectively a somewhat-modified revival of the “leading firm proviso” that appeared in the 1982 Guidelines:⁸² It targeted acquisitions by the market leader. As those Guidelines observed, the real concern in such cases was not thought to be facilitation of coordination, but instead single-firm dominance. As a result, those Guidelines made clear that the factors that are relevant to facilitation of collusion would not be applied and the presumption against mergers in that region was strong.

In any event, these conclusions are ones of fact, as is any conclusion about a particular concentration level that threatens to harm competition. The statute itself does not state any minimum concentration level. Nevertheless, with some quibbling about the precise level of the trigger, this particular conclusion remains quite durable.

In contrast to Guideline 1, Guideline 3 does link concentration to performance by addressing mergers that “Increase the Risk of Coordination.”⁸³ Further, this coordination can apply to all dimensions of competition, including: price; product features; customers; wages; benefits; or geography. These dimensions of coordination have always been attached to merger policy since its inception. The metrics present a question of fact for which the data and understanding can change.⁸⁴ In all events, given the historical record of post-merger pricing behavior, this concern could extend to markets where post-merger HHI levels fall below 1800.

6 Unilateral Effects: Price and Innovation

Unilateral effects analysis must be regarded as one of the more important achievements of Agency merger analysis in the last three decades. The core theory is that in product-differentiated markets—where most merger enforcement occurs—not all firms compete with one another to the same degree: The elasticity of substitution among firm pairs is higher for some pairs than for others. Further, when two firms that are close competitors merge, a price increase among the two is more likely to occur than when a firm merges with a more distant competitor.

A price increase is possible, however, even if the two firms are not the closest competitors. Assessing unilateral effects depends on a number of factors, including: the “diversion ratio,” or elasticity of substitution, between the merging firms; the height of margins; the size of the gap between merging and non-merging firms; and the ability of non-merging firms to reposition themselves closer in product space to the merging firms in order to capitalize on the post-merger price increase.

Unilateral effects analysis is only loosely related to overall market concentration. Indeed, the formal methodologies need not even require a market definition.⁸⁵ They are intended to predict the likelihood of merger-driven price increases by the merging firms, with other competing firms either unaffected or not affected as much.⁸⁶

Historically, the methodology is formally price-driven. An “elasticity” measures a firm’s quantity change in response to a change in price, and does so through well-established methodologies. But how are economists going to model a potpourri of effects that includes prices, quality, “attractive features,” wages, etc., as these Guidelines suggest?⁸⁷ One possibility is that these effects will be reduced to their cash value, perhaps by relying on shadow prices.

Unilateral effects theory is also a triumph of scientific modeling, where a stripped down set of assumptions can work better for making predictions than a model that attempts to take every possibly relevant factor into account.⁸⁸

A promising avenue is the development of methodologies for assessing mergers that diminish competitive innovation incentives from particular pairs of firms. At this writing, these techniques show promise, but they will require more development in litigation.⁸⁹

Further, a complex metric is not always necessary. For example, if two firms are uniquely competing to develop a particular product or line of products, their merger could quite readily be found to retard innovation in the affected area. Economists are creative people and can undoubtedly create models that take other forms of nonprice competition into account.⁹⁰

In any event, challenging mergers on innovation-increasing grounds looks much more promising under something that resembles a unilateral effects theory than under one that is based on general market concentration. The links between innovation and overall market structure are simply not robust enough to support a policy with regard to specific mergers.

7 Potential Competition Mergers

The 2023 Merger Guidelines provide a framework for analyzing mergers that involve potential competitors.⁹¹ They largely adopt without significant revision the framework that the Supreme Court and lower courts developed in the 1960s and 1970s.⁹² The “actual potential entrant doctrine” envisions a highly concentrated target market and a firm that is just outside of the market whose new entry could make the market more competitive. However, the outside firm instead acquires one of the firms inside the market (or vice-versa), and thus removes itself as a potential entrant but without increasing competition in the target market.

By contrast, the “perceived potential entrant” doctrine also starts with a highly concentrated oligopolistic market, but one in which an outside firm is perceived by the incumbents as a potential entrant should their prices rise too high. As a result, the insiders constrain their pricing in order to deter the outsider’s entry. Once the outsider acquires an insider (or vice-versa), however, the perception of this threat disappears, and prices rise.

The Supreme Court and lower courts applied the perceived potential entry theory in several decisions in the 1960s and 1970s. The Supreme Court never approved the actual potential entrant theory, although a few lower courts did.⁹³

One problem with the actual potential entrant theory is doubt about statutory coverage. That issue likely looms larger in this day of stricter textualism than it did in the 1970s. Section 7 prevents mergers that “may substantially lessen competition”—not mergers that merely fail to increase competition. A firm’s entry into a market by merger does not reduce the number of firms in that market, and there is no reason for thinking that competition would be otherwise lessened. That is why the Falstaff decision described the doctrine as involving mergers that “would have no influence whatsoever on the present state of competition in the market.”⁹⁴

Indeed, the presence of an aggressive acquirer might even increase competition in the target market. The merger is unlawful, if at all, because it fails to provide the additional competition that would have resulted had the firm entered de novo rather than by acquisition. In that case, of course, there would be one additional firm in the market.

Nevertheless, the courts and the FTC have been divided on coverage.⁹⁵ In 2023 in FTC v. Meta Platforms, Inc., the FTC urged coverage; and the district court assumed that the Ninth Circuit would follow it, but declined to find coverage on the particular facts.⁹⁶

In its Penn-Olin decision in 1964 the Supreme Court took this doctrine to a speculative extreme.⁹⁷ The court condemned the formation of a joint venture to construct a new facility for production of sodium chlorate. The venture itself significantly increased output in that market; but the theory of the complaint was that if the two firms had entered individually the result might have been two plants and output would have increased even more. The Court did not explain why the joint venture transaction threatened to “lessen” competition substantially.

The 2023 Merger Guidelines discussion of the actual potential entrant doctrine states that the Agencies will examine:

(1) whether one or both of the merging firms had a reasonable probability of entering the relevant market other than through an anticompetitive merger, and (2) whether such entry offered “a substantial likelihood of ultimately producing deconcentration of the market or other significant procompetitive effects.”⁹⁸

On questions of new entry, the potential competition theories sit between two extremes: One is the view that potential competition will always discipline monopoly, so there is nothing to worry about. If that is true, then we do not need a potential competition merger doctrine. At the other extreme is the view that potential competition is impotent and that the only competition that counts is that among actual current rivals. But if no one is a potential competitor, then the doctrine is useless as well.

As a result, one element that the potential competition merger cases share is the idea that some firms—but not all—are potential competitors. Antitrust courts must be able to distinguish them.

In the typical potential competition case, there is some firm that is just outside the market, but that might enter if conditions are right. If there were a large number of such firms, then we would have nothing to worry about. As the Supreme Court observed in the Procter & Gamble decision, “the number of potential entrants was not so large that the elimination of one would be insignificant.”⁹⁹

8 Identifying Potential Entrants: Objective or Subjective

Not all firms are equally well placed to enter, and some may have entry advantages over others.¹⁰⁰ Even when the universe of potential entrants is large, it is quite plausible that one or a few are particularly responsive to price or structural changes in the target market. Whether and how many will enter depends on factors such as the size of the price increase. For example, only one or two firms might enter in response to a 3% higher price, while many more might enter in response to a 10% higher price.

How many firms and how large a price increase we should tolerate are questions of fact and policy. The concern does serve as a warning, however, that when a significant number of equally plausible entrants exist, the elimination of one of them is unlikely to have much of a competitive effect.

The discussion of the acceptable universe of potential entrants in the 2023 Merger Guidelines is brief: A footnote states that

Where there are few equivalent potential entrants including one or both of the merging firms, that indicates that the future market, once commercialized, will be concentrated. The Agencies will consider other potential entrants’ capabilities and incentives in comparison to the merging potential entrant to assess equivalence.¹⁰¹

The Merger Guidelines also state that in identifying potential entrants the Agencies will rely mainly or perhaps even exclusively on objective evidence.¹⁰² During the era of Supreme Court potential competition cases this was a contentious issue, which led then Associate Justice Rehnquist to protest the majority’s reliance on objective evidence.¹⁰³

The FTC itself took the position that the “best evidence concerning the incentives of the acquiring firm to enter independently… is likely to be subjective.”¹⁰⁴ While we presume that a firm wishes to maximize its profit, we would hardly expect a firm to enter all or even a small portion of the markets into which an expert observer concluded that entry would be profitable. New entry is an investment—often a costly and risky one—and every firm is faced with a large number of investment possibilities.

Justice Rehnquist’s position would have reduced the range of potential competition merger decisions by limiting them to situations in which the acquired firm had actually contemplated or “intended” to enter a market de novo as an alternative. The distinction is relevant to the probabilistic standard that is incorporated into the Clayton Act: “where the effect may be.”

Considered purely objectively, the range of potential competitors is very large. For example, weighed objectively Ford—an automobile manufacturer—might have technical or business advantages that make it a plausible entrant into the markets for auto repair, bicycles, tires, gasoline, roadside motels, or digital maps. Does that mean that it should be regarded as a potential entrant based on those considerations alone? If the answer is yes, then GM, Chrysler, Toyota, Subaru, and several others would very likely also be potential entrants. But if the query is limited to markets where Ford has seriously contemplated entry, the range could be much narrower.

To summarize: The use of purely objective evidence makes it easier to “show,” or at least speculate, that an outside firm is a potential entrant based on capabilities and predictions of profitability. In the process, however, this also tends to make the universe of potential entrants larger. In Meta Platforms the district court did not reject an objective approach, although it did find the evidence in that case to be insufficient.¹⁰⁵

9 Relevance of Post-Entry Prices

Entry de novo at viable scale into a highly concentrated market typically results in lower prices in that market. Entry by acquisition does not typically have that effect. As the 2023 Guidelines note in their discussion of entry barriers, “Firms make entry decisions based on the market conditions they expect once they participate in the market.”¹⁰⁶ While new entry in such a situation is desirable, the fact of anticipated lower prices makes it less likely that such entry will occur.

To illustrate: If the minimum viable scale of entry into a concentrated market is a 15% market share and the price elasticity of demand is 1, then entry that increases the market’s output by 15% will reduce the post-entry price by 15% (on the assumption that the elasticity remains constant). To the extent that elasticity declines at lower prices, as it would with a linear demand curve, the price decrease would be less. To the extent that rivals reduce their own prices in response to new entry, the price decrease might be greater: Predicting the post-entry price decline involves some behavioral assumptions as well as arithmetic; but we can assume that a price decrease will occur in most markets where potential competition doctrines have relevance.

That price decrease is unlikely to occur, however, in response to a simple change of ownership of a firm in the target market: When the outside firm proceeds by merger with an insider. Indeed, an important point of a policy that encourages independent entry is to reduce target market prices. Nonetheless, the fact remains that the same lowered prices that make entry de novo desirable as a matter of policy, also tends to make it unprofitable to the entering firm.¹⁰⁷ Before it will enter, the potential entrant must anticipate profitability at post-entry prices.¹⁰⁸

The Guidelines’ description of incumbent firms’ exclusion strategies under a perceived potential entrant doctrine is not limited to pricing. The Guidelines mention other strategic decisions—although without specifying what they are. Descriptively, that may be a better way to describe strategic entry deterrence, for entry need not always be about price. But the entry description also complicates the query: What exactly is it that the incumbent firms do in order to stave off a perceived entry threat? Further, is there some reason that their response cannot be translated into price as a metric? For example, a firm might offer free delivery, longer warranties, or design changes in order to make entry less attractive to an outsider; but in order to assess the effect we would likely have to reduce them to a cost metric.

In any event, the choice between entry de novo and by merger is complex. The failure rate for mergers is high;¹⁰⁹ but for new entry the failure rate is very likely even higher.¹¹⁰ Successful new entry involves the displacement of existing firms and their possible bankruptcy, so it can harm smaller rivals more and produce more waste in the form of prematurely retired productive assets. One well known story is Walmart, which enters new markets mainly de novo, and in the process harms competing grocers both large and small.¹¹¹ This is particularly likely to be true in more concentrated markets where scale economies are significant.

10 Potential Entry and Market Definition

One thing that the 2023 Merger Guidelines do not address is whether changes in the methodology for defining markets affects the scope or even the continuing need for potential competition doctrines. The hypothetical monopolist test for a relevant market considers not only who is in a market at a particular instant, but also who would be in the market in response to a small but significant increase in price. The 2023 Guidelines revise this well-known SSNIP test to “SSNIPT,” to account for the view that not only price increases but other changes in terms must be considered.¹¹² The test is inherently dynamic in that it predicts firm (or sometimes consumer) movement in response to price changes. But that is also what the potential competition doctrines do.

A viable potential competition doctrine targets firms who are potential entrants but nevertheless not “in the market” in response to a small but significant price increase. If such a firm were in the market, then it should be treated as a competitor and the merger would be horizontal. That was the route that the Supreme Court took in the El Paso case,¹¹³ where the acquired firm had previously bid to come into the market but had not ever made any sales. The Court responded that “[u]nsuccessful bidders are no less competitors than the successful one,”¹¹⁴ and treated the merger as horizontal.

The interesting policy question that leaves is this: Are there firms that would not be considered as “in the market” under a hypothetical monopolist market definition, but that nevertheless should be considered as potential entrants? These would necessarily be firms that would incur significant sunk costs from entering.

11 “Trend” Toward Concentration or Vertical Integration

The 2023 Merger Guidelines express concerns about “trends” toward concentration as well as trends toward vertical integration.¹¹⁵ The Brown Shoe decision had emphasized that concern, and it was repeated in some other decisions from that era.¹¹⁶ It also appeared in the 1968 Guidelines, which applied a stricter market share standard to mergers in markets that exhibit such a trend.¹¹⁷ Prior to that, it was reflected in several reports from New Deal Temporary National Economic Committee (TNEC) Reports, which Justice Douglas discussed in his dissent in the Standard Oil exclusive dealing decision.¹¹⁸

Concentration trends continue to be articulated in lower court merger decisions to this day.¹¹⁹ The 1968 Guidelines also indicated that the government would apply a stricter standard to vertical mergers if there was a significant trend toward vertical integration.¹²⁰ The concern was dropped in the 1982 Guidelines and did not reappear until 2023. In the 2023 Guidelines the concern is not stated particularly strongly. On horizontal mergers the Guidelines state only that the trend “may suggest greater risk of harm.” For vertical integration, they state that a trend could “magnify the concerns….”¹²¹ The Guidelines say little about the causes or the competitive consequences of such trends.

Why do markets exhibit a “trend” toward concentration? One historically important reason is changes in technology, which often involve investment in larger plants with greater fixed costs, thus leaving room for fewer firms.

For example, the migration of transportation from horse-drawn to gasoline vehicles led to many fewer manufacturers. Many of the early “trust” cases—such as steel, cans, wire nails, and tobacco—arose out of a movement from hand-made products made by very small producers to machine made-ones that required much larger firms.¹²² In a few situations the trend has worked in reverse: For example, the digital computer revolution moved from the (comparatively few) large mainframes that dominated the 1960s and 1970s to the era of a vastly larger number of smaller digital units, which permitted many more firms to make them.

Closely related is improved modes of transportation or transmittal, which can make markets bigger and typically less concentrated. For example, Amazon.com and other internet sellers have undoubtedly been a significant factor in reducing market concentration to the extent that they provide online availability for products in addition to local offline suppliers. A town with two hardware stores can come to have at least three when Amazon and other firms are able to deliver hardware items efficiently to that location.

The most important point is that competition itself creates trends toward concentration or vertical integration if the consequences of the trend are to reduce a firm’s costs or improve its products. Rivals will be forced to copy the first mover or else go out of business. If there are no available scale economies, or if vertical integration does not produce cost savings, then there is no reason to expect the trend.

In sum, a trend toward concentration or vertical integration is presumptive (perhaps not conclusive) evidence that firms are competing to lower their costs.

Another factor that increases concentration is differential rates of growth. Some firms have lower costs or are more innovative than others.¹²³

For example, imagine a market that starts out with 10 equal size firms, each with 10% of the market and thus an HHI of 1000. Suppose that subsequently two firms grow because they have lower costs or superior technology. Others decline. So later the array of firm sizes becomes 25, 25, 10, 10, 5, 5, 5, 5, 5, 5. This new market has an HHI of 1600 (25², 25², 10², 10², 5², 5², 5², 5², 5², 5²). In this case the lagging firms lost market share but did not go out of business. If some had shut down, the resulting HHI would be even higher.

Further, none of the increase in concentration resulted from a merger. Relatedly, product differentiation, branding and advertising strategies tend to favor more concentrated markets—although without diminishing the amount of competition and often increasing it.¹²⁴ These large firms often tend to be both more productive and more innovative.¹²⁵

An analogous phenomenon occurs with respect to vertical mergers or other mergers of complements. Here, the case for production or transaction cost savings from mergers is particularly high, and even Brown Shoe recognized that vertically integrated shoe manufacturers had lower costs.¹²⁶ When one or more firms attain these lower costs by integrating, however, rivals will be pressured to integrate themselves or else lose market share or even be forced out of business by lower-cost competitors. The result will be a trend toward integration.¹²⁷

The empirical evidence indicates that markets experience increasing concentration for a variety of reasons—most of them competitively benign. As a result, there is no presumptive reason to apply differential scrutiny toward mergers that are part of such a trend.

The Guidelines do not appear to contradict that proposition; they state only that trends toward concentration or vertical integration should be considered. Unlike the 1968 Merger Guidelines,¹²⁸ they do not suggest that a merger should be condemned at lower concentration numbers when such a trend exists—although they do indicate that trends can indicate an “arms race for bargaining leverage.”¹²⁹ They do not indicate how to distinguish such an arms race from simple competition to achieve lower costs.

12 Mergers and Networks

Guideline 10 of the 2023 Merger Guidelines states that “When a Merger Involves a Multi-Sided Platform, the Agencies Examine Competition Between Platforms, on a Platform, or to Displace a Platform.”¹³⁰ Adding coverage of networks is welcome, because networks represent a growing portion of the economy, and assessing power and their structural effects poses unique problems.

The Guidelines note that digital multi-sided platforms “have characteristics that can exacerbate or accelerate competition problems.”¹³¹ Principally, network effects can create a “tendency toward concentration in platform industries,” and also that a “conflict of interest” may result when a platform operator is also a platform participant.¹³² In that case the platform operator may have “an incentive to give its own products and services an advantage over other participants competing on the platform.”¹³³

The Guidelines are not using the term “conflict of interest” in the traditional legal sense where it is applied to fiduciaries. Rather, the term as it is used here means something more akin to the “conflict” that might occur when Walmart sells both the active wear of third parties such as Nike and its own “Athletic Works” house brand. Then it might be tempted to favor its own brand in display, pricing, or other customer convenience. Such preferencing is universal among multi-brand stores that have their own house brands—both on- and offline, and such preferencing rarely has any antitrust implications.

In any event it should not, except for a firm that is dominant in that product, and only if the preferencing provably leads to higher prices or reduced output. To the best of my knowledge, it has never been theorized to do that as a general matter—although there may be exceptions.

Why the Guidelines target this phenomenon for digital platforms but not for old economy stores is not entirely clear—particularly since customer switching costs are typically lower online than they are in physical stores. In general it is easier for a customer to switch from an Amazon house brand than from one that is offered at Walmart—although switching from either may not be all that difficult.¹³⁴

Here, the Guidelines do not state a dominance requirement, but they do state that when a platform owner is dominant the Agencies will “seek to prevent even relatively small accretions of power from inhibiting the prospects for displacing the platform or for decreasing dependency on the platform.” That sentence suggests that the dominance requirement applies to the platform, rather than the particular product or service in which the self-preferencing is threatened.

For example, Amazon is a dominant online platform retailer, but it has less than 3% of grocery sales, and faces substantial competition from offline stores. So would the Amazon/Whole Foods acquisition be challenged on the theory that it would encourage Amazon to display its own Whole Foods items over competing items that are offered by third-party sellers? The proposition that this is merely an “interpretation” of §7 rather than an attempt to make new law is more than dubious unless the challenger can show a reasonable probability of reduced output, higher prices, lower quality, or other indicia of competitive harm.

Suppose that Amazon sells multiple brands of toasters of various designs and price points. Suppose now that it acquires “Toastrite” (a hypothetical toaster manufacturer). Amazon may wish to arrange its display or search results in such a way as to favor Toastrite over other brands that it carries, such as Cuisinart or Hamilton-Beach. Indeed, it may even terminate its relationship with other brands and sell Toastrite exclusively. In any event this is a vertical merger. Absent measurable foreclosure in the toaster market, it is difficult to see any impact at all—incipient or otherwise—on the price or quantity of toaster sales.

The issues in these cases are similar to those in a vertical merger case, where the concern would be foreclosure of rivals; and the Guidelines recommend a 50% foreclosure trigger. However, merely preferential treatment—such as a higher position in a search result or a default—would have to be counted as less severe than outright exclusion.¹³⁵

In the ongoing Google search monopolization case the court measured foreclosure by examining the percentage of sales that are covered by Google’s default search engine contracts.¹³⁶ That may be a good place to start, but a default is not the same thing as an exclusive deal.¹³⁷ Pointing the other way is the fact that the §7 “where the effect may be” standard is more aggressive than the Sherman Act standard that is being applied in Google. Nevertheless, if fear of this “conflict of interest” is to be a rationale for condemning a merger, there must be at least a probabilistic inference of injury to competition—manifested by lower market output or higher prices.

Also important here would be proper market definition. For example, in the Amazon toaster example above, toasters are retailed competitively across a wide variety of sellers. Both Amazon and Walmart have significant national shares (very likely in the low 20% range), but other large retailers sell them as well. Online and traditional brick-and-mortar sales are presumptively competitive with one another, or else one of them would have to be proven to be sufficiently insulated from the other so as to justify supracompetitive prices. That question is market-specific.

The real question is whether competition from internet sales can hold traditional sales close to cost, or vice-versa. For example, video streaming is a technology that is largely unavailable in offline stores, so that market would be limited to internet sellers. Groceries and try-on clothing are entirely different, and online sellers face intense competition from traditional sellers and very likely operate at a competitive disadvantage.

With respect to mergers between platforms, this particular Guideline also does not mention potential offsetting competitive benefits, which can be very substantial—stronger than in the general run of merger cases. One strong network effect is that value can increase dramatically as a network expands: The economies operate in consumption rather than production, and can produce gains that resemble those from mergers of complements.

For example: A hypothetical merger of Uber and Lyft might eliminate competition between the two largest, or even the only two, ride-hailing apps in a particular city. The effects of such a merger would have to be tested against the market for competing products, whatever those might be.

The other effect of this merger, however, is to consolidate the two networks, which would produce more drivers on one side and more riders on the other. This may increase availability, shorten wait times, and yield lower costs to the extent that hailed drivers would be more likely to be in close proximity to passengers. It might also induce more entry into driving by those who found the smaller networks less desirable. Of course, here the gains depend on whether the two networks are consolidated into one, and that might not always be the case.

Here, the Guidelines’ statement about “small accretions” seems overbroad—particularly if the acquired network is a competitor and the principal effect of the acquisition is to make the network larger. Such a merger is at least as likely to result in an improvement in performance as to result in competitive harm: Efficiency defenses can be more robust when mergers occur in networked markets where a likely impact of the merger is to increase direct or indirect network benefits.

One particular problem of the Guidelines’ provision on networks is that it does not give much guidance. It has far too little discussion of: the applicable markets (whether upstream or downstream); minimum market shares; offsetting effects; or how harm is to be assessed or offsetting gains proven. As such it is likely to invite additional litigation and produce poor results.

I acknowledgements That this is largely new territory, and the economic literature not been especially clarifying; but the Guidelines could be clearer about how a market power requirement should be applied, what are the minimum standards for competitive harm, offsetting economies or other savings, and other particulars. They should state clearly that mergers will be challenged in cases where output or quality is predictably lower, prices higher, or innovation restrained.

13 Roll-Ups: Serial Acquisitions

2023 Merger Guidelines, supra note 1, § 2.8.

One lacuna in previous Guidelines is that they tended to address each acquisition in isolation: involving a single pair of firms. A merger that increased the HHI by less than 100 would be treated as presumptively lawful under both the 2010 Horizontal Merger Guidelines and the 2023 Merger Guidelines. To illustrate, if a 20% firm should acquire a 2% firm the increase in the HHI (DHHI) would be 20 X 2 X 2, or 80: a presumptively lawful merger.¹³⁹

But suppose that this firm acquired three different 2% firms with the acquisitions spaced 6 months apart. Examining each merger in this series individually, they would look like this:

#1: 20 X 2 X 2: DHHI = 80.

#2: 22 X 2 X 2: DHHI = 88.

#3: 24 X 2 X 2: DHHI = 96.

For each successive acquisition the size of the acquiring firm has grown by the amount of the previous acquisition, but even the final acquisition is under the 100 DHHI threshold.

Alternatively, however, suppose that the firm acquired all three of the firms, with an aggregate 6% market share, in a single transaction. In that case:

20 × 6 × 2: DHHI = 240: This would be a presumptively challengeable merger—certainly if the post-merger HHI exceeds 1800.

The arithmetic is simple, and certainly overly simplistic. The policy question is both important and more complex: Should this series of acquisitions be treated as three discrete events or as an aggregated acquisition of the 6% firm? In both cases the firm has gone from a 20% share to a 26% share. The difference is that in the first circumstance this growth by acquisition was spaced over a year, while in the second it occurred in a single transaction.

The Guideline quotes the House Report on the Celler–Kefauver amendment as expressing concern about

a cumulative process when the effect of an acquisition may be a significant reduction in the vigor of competition, even though this effect may not be so far-reaching as to amount to a combination in restraint of trade, create a monopoly, or constitute an attempt to monopolize.”¹⁴⁰

The point would have been clearer had it also quoted the previous sentence, which stated that competitive harm “may be achieved not in a single acquisition but as the result of a series of acquisitions….”¹⁴¹ The Guidelines statement refers to discovering “a pattern or strategy of growth through acquisition by examining both the firm’s history and current or future strategic incentives.”¹⁴² That statement seems unnecessary and calculated to produce unnecessary investigative expense.

Section 7 provides that a merger is to be evaluated by its probabilistic effects (“where the effect may be”)—not by its purpose. The only real question about such a series of transactions is whether its effect on price, output, innovation, wages, or some other indicator of competitive effects is similar to that of an aggregated single transaction. The rule needs to be simpler than one that queries the firm’s overall motive and strategy.

Merely as a working presumption, I suggest that the occurrence of multiple acquisitions within a 3-year period should presumptively entitle the Agency to aggregate them. The 3-year period is presumptive in both directions; either side should be permitted to make a case for a longer (government) or shorter (defendant) period. It is also arbitrary, and further study might suggest a longer or shorter presumptive period.

In addition, aggregation is generally appropriate only when the firms that are consecutively acquired operate in the same market. For example, suppose that a large retailer with stores that are scattered across the country should acquire individual stores that are located in Montana, Texas, and Delaware. At least at the level of in-person retail output, these stores do not compete with one another. The upstream situation may be different—depending on the facts.

One common observation is that very large firms—such as Alphabet (Google) or Amazon—have acquired numerous smaller companies; many of them are extremely small.¹⁴³ However, the acquired firms operate each in their own market—which is often different from the market of other firms that were acquired in that same year. Should these be aggregated in some way?

To illustrate: In 2017 Amazon acquired: Whole Foods, a high-end grocery chain; Graphiq, which operates as an input into Amazon Echo digital sound systems; and Body Labs, a software producer with artificial intelligence capabilities for analyzing human body shape and motion. There does not seem to be a compelling case for “aggregating” the output of these firms. Even on the upstream side, it is not clear that they operate in the aggregate to increase Amazon’s market share in any product beyond the forbidden limits—or more importantly, to facilitate reduced output, higher prices, less innovation, or some other evidence of competitive harm.

This situation is much different from the owner of a health care network that sequentially acquires pediatrics practices in the same city. Then a stronger case can be made that the acquisitions should be aggregated.

14 Mergers Affecting Supplier Competition, Including Labor

Guideline 10 of the 2023 Guidelines provide that “When a Merger Involves Competing Buyers, the Agencies Examine Whether It May Substantially Lessen Competition for Workers, Creators, Suppliers, or Other Providers.”¹⁴⁴ The concerns that are stated in this Guideline mirror somewhat those that are stated in earlier Guidelines for seller-side harms—just as monopsony is the economic mirror image of monopoly. They reflect the fact that §7 of the Clayton Act applies equally to sell-side and buy-side harms. Notwithstanding that, historically there has been considerably less enforcement of the latter.¹⁴⁵

Measuring concentration in supplier markets—particularly labor markets—poses some challenges: In some cases labor markets may be less competitive than product markets, exhibiting higher switching costs or sunk investments. Labor markets may often be smaller geographically than product markets; but the issue is complex. For example, while the geographic market for low-skill workers may be small, limited to a commuting area, to the extent that these workers are less specialized and can be used across multiple output lines the product market could be larger.

To illustrate: The market for minimum-wage window washers may be geographically limited to reasonable commuting distances, but business firms in many markets need their windows washed, as do some residences. Further, people who are employed as window washers may view alternative occupations at the same skill level as viable options. Nurses are a different matter, for their skills are much more specialized to specific employers.

The same thing is largely true for non-labor input suppliers. Some are highly specialized, perhaps even for a single customer.¹⁴⁶ Others, such as providers of cleaning supplies or common building components, can supply a wide range of product producers. So here it is critical that concentration be measured correctly. For example, the nurses who are employed by a hospital almost certainly work in a much more concentrated market than do the firms that supply the hospital with cafeteria food or cleaning agents.

In a monopsony market, a firm exercises its market power by purchasing less, with the effect that its purchase prices (or wages) are suppressed. Thus one important difference is that the concern in buying markets is with lower prices, not higher ones. Whether this occurs depends on the extent of the market power that is held by the actor(s). This typically requires a market definition. The theory of “unilateral effects” could also be applied to close substitutes on buy-side markets, but at this writing it is relatively undeveloped.¹⁴⁷

The Guidelines appreciate that not every merger that reduces demand for a certain input—including labor—is anticompetitive. Consolidation often leads to reduced demand, including a lower demand for workers.¹⁴⁸ This can result from the elimination of duplication. For example: If two stores merge, they may require only one sales manager rather than two, or they may require the services of only one accountant. While these consolidations eliminate suppliers or jobs, the reductions are not a consequence of monopsony output suppression but rather through resource savings. The antitrust laws do not incorporate a preference for featherbedding.

Evaluation of market shares will address many of these problems; but even firms that operate in concentrated markets can save resources by eliminating duplication. Analyzing that claim would be the same as any other efficiency defense. In any event, acknowledgement should be made more explicit.

15 The Single-Market Rule for Input Markets

The Guidelines do note that competitive harm in labor markets cannot be offset by purported benefits in product markets: “If the merger may substantially lessen competition or tend to create a monopoly in upstream markets, that loss of competition is not offset by purported benefits in a separate downstream product market.”¹⁴⁹ That follows from the “single market” rule that was embraced by the Philadelphia Bank decision,¹⁵⁰ which interpreted the “any line of commerce” and “any section of the country” language of §7 of the Clayton Act. True competitive harm in one market cannot generally be offset by claimed benefits in another—no matter which side is the source of harm.

16 Labor Harms and Product Markets

The Guidelines do not mention the important relationship between the health of labor markets and the robustness of product markets. Here, economists Jan De Loecker, Jan Eeckhout, & Gabriel Unger make this observation:

An increase in markups implies a decrease in aggregate output produced, whenever demand is not perfectly inelastic. Lower output produced then implies lower demand for labor. This results in both lower labor force participation and lower wages. Even if supply is perfectly elastic, real wages decrease with market power because the price of the output good has increased….”¹⁵¹

This often-overlooked principle is critical to any antitrust policy that must simultaneously manage input and output markets. Labor is largely a variable cost—particularly with respect to lower-wage hourly workers. As a result, the number of jobs available is critically dependent on output in the corresponding product market. This creates a problem if product market antitrust rules protect higher-cost businesses, which results in higher product prices and lower output. As was noted above, the Supreme Court would very likely not permit condemnation of a merger on the grounds that it reduced prices.

Further, enforcement that is excessive in the sense that it perpetuates higher costs and lower output could harm labor—depending on the amount of market power that is present in labor markets. On this issue Guideline #1 is troublesome. Its suggestion of condemnation based on concentration without regard to performance suggests at least the possibilities that the Agencies will challenge some output-increasing mergers. If that happens, consumers and labor will both suffer. Whether the Supreme Court would allow this to happen is another matter.

One implication is that true efficiencies that tend toward higher output in product markets should be taken seriously. They should be difficult to prove, and the efficiencies concept has been overused. But that is no reason to impose a policy of using merger law as an output limitation device. For example, a recent empirical paper relates active antitrust enforcement to higher output and lower prices in product markets, with corresponding improvement in the availability of jobs, increased wages, and higher worker participation rates.¹⁵²

It is plausible that a policy of producing robust competitive levels of output in product markets would benefit labor more than would a more aggressive policy toward a practice such as noncompete agreements. These cover perhaps 20% of the workforce, while all workers can benefit from healthy product output. It is not unreasonable to assume that a 15% reduction in product output—whether caused by a cartel or by an overly aggressive antitrust rule—will be followed by a comparable decline in jobs. This decline would appear quickly for lower-wage hourly workers, whose jobs are typically the most sensitive to changes in product demand.

17 Efficiencies and Double Marginalization

Efficiencies are often claimed but much less often proven. The 2023 Merger Guidelines follow a tradition that was established in previous Guidelines of being skeptical and requiring strict proof of efficiencies. They note that firms can often use “contracts short of a merger to combine complementary assets without the full anticompetitive consequences of a merger.”¹⁵³ Of course, the contract itself would have to be legal. For example, for a firm that is involved in retail distribution, exclusive dealing might operate as an alternative to a vertical merger. But the exclusive dealing agreement could also be anticompetitive to the extent that it forecloses rivals unreasonably.

The 2023 Guidelines also adhere to the well-accepted “single market” rule and refuse to “credit benefits outside the relevant market.”¹⁵⁴ The Guidelines also generally adhere to a practice of assuming ordinary efficiencies without specific proof, but requiring that larger claimed efficiencies be verified. Previous Guidelines—such as those that were issued in 2010—required the passing-on of cost savings sufficient to hold consumers harmless.¹⁵⁵ The 2023 Guidelines simply state that proffered efficiencies must be such “that no substantial lessening of competition is threatened by the merger,” which presumably means about the same thing.¹⁵⁶

Notwithstanding the dominance of efficiencies in discussions of merger policy since the 1970s, the fact is that empirical evidence on the subject is underdeveloped. Several recent studies on post-merger pricing are helpful. They are largely limited to examining post-merger prices rather than costs: for the simple reason that price data are more readily available for the purposes of comparing large number of firms.

Post-merger price changes are generally a useful surrogate for changes in costs. They are not perfect. Some mergers may create substantial efficiencies but have market power effects that offset any cost reduction. Mergers in competitive, undifferentiated markets may reduce costs without having any measurable impact on prices. Nevertheless, equilibrium lower prices in a concentrated market following a merger are most likely explained by some cost savings, although these could also result from technological changes unrelated to the merger.

The difficulty of specification and proof justifies both the selection of prima facie concentration levels and the size of any built-in “efficiency credit” provided without specific proof. Further, efficiency gains are not driven by market concentration. Rather, they require firm- or at least technology-specific assessment of the potential for cost reductions by or within a single firm.

The empirical evidence from pricing suggests that a significant minority of approved mergers—even in more highly concentrated markets—result in lower prices.¹⁵⁷ Most such studies are skewed in the overall universe of mergers because they focus on mergers that are reported and reviewed. The majority of mergers fall below reporting thresholds; and, unless the screening process is worthless, these are more likely to yield cost savings and lower prices.¹⁵⁸

The reported mergers in one recent high-quality study show somewhat more price-increasing than price-reducing mergers; but most of the mergers in that study had post-merger HHIs in the 2000–4000 range. Some reached as high as 6000.¹⁵⁹ Most of these mergers were challengeable under the 2023 Guidelines. Even within this group, however, 25% of mergers reduced prices by an average of 5.2%, while another 25% increased prices by an average of 5.9%¹⁶⁰ The study also found that price increases were correlated, although not perfectly, with quantity decreases.¹⁶¹

Most importantly, price increases were also correlated with increases in the HHI. Further, the distribution of price changes (some up and some down) for mergers with similar HHI numbers indicate that efficiencies can vary significantly. That is why any rule that is based on concentration screens can be no more than presumptive.

The literature also indicates that the HHI is a useful tool for analysis but that the standards that are stated in the 2010 Horizontal Merger Guidelines were underdeterrent.¹⁶² HHI increases above 200 were correlated with price increases in nearly all cases—even in less concentrated markets.¹⁶³ Further, while lowering the thresholds would envelope many more mergers, enforcement costs would also rise considerably. That is the eternal political problem that the enforcement agencies face: While the need for more enforcement is widely acknowledged, it costs money.

Overall, studies such as this provide ammunition for stricter examination, provided that: (1) the Agencies actually enforce to the full extent of the stated threshold; and (2) the courts go along. With reference to the 2010 Guidelines, the authors conclude:

We find that current levels of antitrust enforcement are such that the probability of blocking a pro-competitive merger is very low, while the probability of allowing anti-competitive mergers is substantial. However, tightening standards would lead to a drastically higher burden on the agencies.¹⁶⁴

The 2023 Guidelines require a particularized showing of “merger specificity”: The claimed efficiency could not be attained otherwise than by a merger.¹⁶⁵ That is consistent with the treatment in previous Guidelines.

In making this assessment the Agencies may examine the potential for “organic growth,” or contractual alternatives that are less anticompetitive than mergers, or a partial merger that involves only some assets that give rise to the claimed efficiencies. While the Guidelines are quite strict about proof that claimed efficiencies must be verifiable, they do not indicate that they will apply the same strictness in assessing these alternatives, many of which are likely to be hypothetical or anticompetitive themselves.

A problem that remains in the background is the willingness of at least some courts to recognize efficiency claims that do not meet the Guidelines requirements.¹⁶⁶ It is not obvious that the presentation in these Guidelines will improve that situation.

18 Market Definition

Use of the HHI (or any concentration measure) to evaluate the competitive effects of mergers is critically dependent on a correct market definition. The fact that the HHI numbers are squared tends to exaggerate errors—more than do older indexes such as the CR4.¹⁶⁷

Further, in order to relate to a measure of concentration, the market definition should be crafted so as to identify the threat of monopoly or collusion. The hypothetical monopolist test (HMT), which seeks to identify the smallest grouping of sales that is capable of being monopolized or cartelized, is the theoretically correct market definition for applying the HHI.¹⁶⁸ However, the 2023 Guidelines embrace multiple approaches to market definition, some of which do a poor job of assessing concerns for competition.

The 2023 Guidelines statement on market definition begins with boilerplate—largely taken from the Brown Shoe case—that a relevant market is measured by “reasonable interchangeability or use or the cross-elasticity of demand between the product itself and substitutes for it.” They also note that markets may include narrower groupings of sales that are fully encompassed in broader markets.

As noted above, on this particular point Brown Shoe was incorrect, because the reference to observed interchangeability said nothing about the margins at which substitution was occurring. Interchangeability at any price is rarely sufficient to establish a market. There must also be some warrant for thinking that each of the goods in a market is able to force the price of other firms’ goods to cost. As a product’s price rises above the competitive level it becomes increasingly “interchangeable” with other products, as more and more customers substitute away. As a result, high observed interchangeability may reflect no more than that one of the products is already being sold at a monopoly price.

The 2023 Guidelines list additional factors: Some are relevant to the determination of a smallest grouping of sales that is capable of being monopolized; and others seem to reflect no more than casual observation of substitution at any price.

To illustrate: Consider the fact that movie goers are observed to attend theaters, purchase DVDs, or stream them from a website such as Netflix. This would seem to show reasonable interchangeability, or the “substantial competition between the merging parties,” that the Guidelines call for. But does it really show that these three highly diverse technologies belong in the same relevant market for antitrust purposes? This is a version of the “Cellophane fallacy,” named after a Supreme Court decision (in a monopolization case) that reasoned too quickly from observed substitution among diverse wrapping materials to the conclusion that all belonged in the same market.¹⁶⁹

As the Cellophane decision also reveals, casual market definitions that ignore the relationship of market definition to economic monopoly can result in overly broad markets, as it did in Du Pont—although occasionally the error works the other way as well. The result, which does not seem to be consistent with the Guidelines’ general goals, is likely to be underenforcement.

The hypothetical monopolist test (HMT) is a more discerning methodology that has been developed to address such questions for horizontal mergers. Under that methodology the fact finder usually tries to identify the smallest grouping of sales that could yield relatively durable monopoly prices if the sellers of that grouping were united as either a single firm (a “hypothetical monopolist”) or a cartel.

In all events, the 2023 Merger Guidelines statement on market definition is dominated much more by the hypothetical monopolist test than was the earlier draft, which indicates that the Agencies got the message that there must be a sensible correlation between merger policy’s pursuit of undesirable monopoly and the means of measuring it.

I acknowledge that there are situations in which the HMT will not work, such as when adequate data about prices and diversion are not available, or when the good is not yet on the market or sold at a price of zero. For example, in its Illumina decision the Fifth Circuit declined to apply the HMT because the products in question were in the research-and-development stage and had yet to reach any market.¹⁷⁰ In essence, the HMT is clearly preferable when the data and conditions for applying it are adequate; but there are some cases that require alternative approaches.

In sum, a better position on market definition would be that the Agencies will use the HMT whenever adequate data for measurement are available. If not, as in Illumina, they will look to alternatives.

The Guidelines also attempt to modify the HMT so as to consider nonprice terms. As was noted above, whether this effort at descriptive realism will have a net positive value remains to be seen. Market definition consists of two queries: On the demand side, to what extent will customers substitute away from a price increase? On the supply side, to what extent will potential rivals enter or reconfigure their products so as to place themselves within the market?

On the first, merger analysis has used economic techniques to consider the extent to which sales would be lost in response to a SSNIP (or a “SSNIPT,” under the 2023 Merger Guidelines), which is usually considered to be about a 5% increase from prevailing prices.¹⁷¹ The 2023 Guidelines indicates that the Agencies will stick with this 5% price increase in order to make estimates. That already suggests that the “terms” portion of the SSNIPT test will be reduced to a dollar value, making it effectively little more than a price response. In economic practice SSNIPT is likely in most cases to revert to SSNIP.

On the supply side, the Guidelines count as “in the market” any firm that could readily shift production into the market in response to a SSNIPT. Both the 2010 and 2023 Guidelines term these firms “rapid” entrants. The principal difference is that the 2010 Guidelines describe firms that could easily shift production in response to a price change, while the 2023 Guidelines discuss a “small but significant change in competitive conditions.” Once again, if that metric is taken seriously it will complicate economic analysis unless these unspecified economic conditions are simply converted to their cash value.

19 Conclusion

The 2023 Merger Guidelines make many important contributions to merger enforcement policy. In some areas, however, they “take their eye off the ball” by pursuing theories that are not reasonably calculated to identify mergers that truly harm competition by reducing output or quality, increasing price, or restraining innovation. Part of the problem is their tendency to prefer old Supreme Court decisions over a 60-year record in which the lower federal courts wrestled with numerous issues, adopted new methodologies, or availed themselves of new information since the Supreme Court largely abandoned substantive merger review in the 1970s.

One important question, therefore, is the appropriate role of “new learning” that has been expressed in lower court decisions and economic analysis over the last 60 years since the Supreme Court largely withdrew from the substantive analysis of mergers. If doctrine from the 1960s and 1970s is read as declaring rules of law, they will remain petrified in time until they are overruled by a competent tribunal. Precisely the opposite constraint applies to issues of fact: They must incorporate up-to-date and acceptable methodologies in an area where ongoing revision is the norm.