The Yale Law Journal

VOLUME
130
2020-2021
Forum

Dominant Digital Platforms: Is Antitrust Up to the Task?

18 Jan 2021

abstract. It has been one hundred years since the end of the Progressive Era and twenty years since the Microsoft settlement. This twenty-year period has seen the rise of the Internet and a new set of dominant platforms, as well as increased consolidation in the brick-and-mortar world. Antitrust has become more permissive. This Essay examines both issues: (i) the potential legal difficulties in reining in exclusionary conduct by dominant platforms; and (ii) merger under-enforcement.This Essay argues that it is necessary to strengthen antitrust enforcement. Consolidation through mergers and exclusionary conduct by dominant firms can harm consumers and workers and reduce innovation. Digital networks are a particular concern because barriers to entry, which result from substantial network effects and economies of scale and scope, rise as platforms’ dominance is enhanced. While antitrust law in principle can evolve, new legislation would be a more rapid—and more certain—path to reform.

Introduction

There has been growing public concern over both market concentration generally and the monopoly power of digital platforms and their effects on U.S. competition and consumer and worker welfare.1 These concerns are reinforced by the worsening skew in the distribution of income and wealth.2 Concentration has visibly increased in airlines, beer, agricultural products, and health care, among other sectors, and the dominance of Google, Facebook, Apple, and Amazon has attracted increased attention.

Competitive concerns about digital networks are not new. Payment-card networks—primarily Visa, American Express (AmEx), and MasterCard—have been subject to substantial antitrust attacks.3 Dominant travel-booking platforms like Sabre charge significant fees for airline tickets.4 Control over digital home listings by realtor associations has been used to support high real-estate sales commissions by discouraging discount brokers.5

The growing role of the internet in information dissemination, advertising, social discourse, and commerce raises fears that a small number of firms (most notably, Google, Facebook, Amazon, and Apple) will have control over communication and retail commerce. This control can have political and social effects, as well as more purely economic effects.6 Google controls search advertising and also has gained power in the sale of display and contextual ads.7 It also controls the Android operating system and charges fees for applications that are used by Android device users. Apple has similar power over the applications for iPhones and iPads.8 Facebook’s dominance of social media gives it control over narrowly targeted advertising, including targeted political messaging.9 Amazon controls a substantial share of online retail commerce and charges large fees to sellers who sell through its Marketplace.10 Not only do these platforms have substantial market power, but their conduct also has raised a variety of antitrust concerns. This involves both exclusionary conduct directed at competitors and acquisitions that can enhance or entrench the platforms’ dominance by gaining control over innovative competitors or complementary products that could become competitors or facilitate competition by others.

These concerns harken back to the Gilded Age and the trusts that led to the Sherman Act in 1890. Indeed, some critics refer to themselves as New Brandeisians or New Progressives.11The trusts combined competing firms into integrated enterprises that wielded monopoly power to collusively raise prices and exclude competition. The Sherman Act achieved some substantial market restructuring in response, including the breakup of Standard Oil12 and the Northern Securities railroad trust,13 until the failure of the U.S. Steel case in 1920.14 Since that time, the two most notable attacks on monopoly power have been the 1982 antitrust consent decree that broke up the AT&T monopoly, and the 1998 Microsoft case and 2001 settlement that contributed to the successful entry of Google and other internet services.15

It has now been one hundred years since the end of the Progressive Era and twenty years since the Microsoft settlement. Microsoft’s exclusionary conduct towards Netscape arose from Bill Gates’s fears that the “Internet Tidal Wave” might weaken Microsoft’s power. This twenty-year period has seen the rise of the Internet and a new set of dominant platforms, as well as increased consolidation in the brick-and-mortar world.

Antitrust has become more permissive, which raises the question of whether antitrust laws governing mergers and exclusionary conduct should be strengthened. This Essay will examine both issues: (1) the potential legal difficulties in reining in exclusionary conduct by dominant platforms; and (2) merger under-enforcement, which has permitted this market consolidation and similarly may permit future vertical mergers and acquisitions of new entrants by dominant platforms and other leading firms.

In my view, it is necessary to strengthen antitrust enforcement in these areas. Consolidation through mergers and exclusionary conduct by dominant firms can harm consumers and workers and reduce innovation. Digital networks are a particular concern because barriers to entry, which result from substantial network effects and economies of scale and scope, rise as platforms’ dominance is enhanced. While antitrust law in principle can evolve, new legislation would be a more rapid—and more certain—path to reform.

The remainder of this Essay is organized as follows. Part I focuses on exclusionary conduct by dominant firms, including dominant digital platforms. Part II focuses on vertical mergers and acquisition of potential or nascent competitors. Part III then discusses the need to strengthen antitrust law and enforcement, both generally and specifically with respect to dominant digital platforms.

I. exclusionary conduct by dominant firms

It is important to recognize an inherent limitation of section 2 of the Sherman Act. Section 2 is not a “no fault” statute. It does not prohibit lawful monopolies from charging monopoly prices. It prohibits “monopolization,” which is exclusionary conduct that permits a firm to achieve, enhance, or maintain monopoly power and thereby has anticompetitive effects on consumers.16 These anticompetitive effects include higher prices, but also lower quality products and reduced innovation. For example, Microsoft was found to have engaged in variety of exclusionary conduct (i.e., exclusive contracts; integration of Internet Explorer into Windows; deception and threats) designed to prevent competition in operating systems that would have been facilitated by the success of Netscape and Java.17 The focus of the Microsoft case was innovation competition, not price competition.

A. Hurdles to Successful Section 2 Litigation

Some allegedly anticompetitive conduct by the platforms involves agreements with platform transactors, which can be attacked under section 1 of the Sherman Act. Likewise, unilateral conduct can be attacked under section 2, where potential remedies can be stronger. This Essay will focus on section 2 enforcement, though many of the issues also will apply to section 1 litigation.

Effective section 2 enforcement faces three important hurdles. First, section 2 only applies to firms with monopoly power or a dangerous probability of achieving monopoly power. Second, even if the exclusionary conduct harms competitors, consumers may not be harmed because of offsetting price or quality benefits from the conduct. Third, antitrust liability does not automatically lead to the break-up of the monopoly. A court instead may enjoin the anticompetitive conduct. I will now discuss how these hurdles apply to the tech platforms.

Monopoly Power. While the tech platforms are large, evidence is needed to establish monopoly power. In litigation, the platforms no doubt would argue that competition is “just a click away.” However, the real issue is whether such substitution by users is a sufficient constraint on anticompetitive conduct or monopoly pricing by these platforms—a proposition that seems unlikely in light of their market shares in their core areas and the substantial barriers to entry and expansion facing new entrants and smaller competitors. In addition, if there is “direct evidence” of anticompetitive effects, that evidence also is convincing proof of market power, as discussed below.

Anticompetitive Effects. In litigation, the platforms likely would claim that they have achieved their power “on the merits” by providing innovative products. However, this showing should not be sufficient if the case alleges anticompetitive maintenance of monopoly, rather than achievement of monopoly power. The monopoly maintenance claim would focus on whether the alleged exclusionary conduct raises barriers to entry or competition by smaller or nascent competitors. For example, this was the structure of the Microsoft case.18 Companies may argue that it must be proved that the excluded competitor would have succeeded in weakening the monopoly. However, in Microsoft, the D.C. Circuit rejected Microsoft’s claim that strong causation evidence must be shown. As the court explained, “[t]o require that § 2 liability turn on a plaintiff’s ability or inability to reconstruct the hypothetical marketplace absent a defendant’s anticompetitive conduct would only encourage monopolists to take more and earlier anticompetitive action.”19 This is because “neither plaintiffs nor the court can confidently reconstruct a product’s hypothetical technological development in a world absent the defendant’s exclusionary conduct.”20 After American Express,21 the plaintiff will need to account for that fact that these platforms compete in “two-sided” markets, which makes the analysis more complex. While the effects on both sides of the market can be competently analyzed, they may confuse courts.

Remedy. Obtaining an effective remedy can be a challenge if the government agency or court wishes to minimize intrusion. Standard Oil was successfully broken up into multiple entities. But when the government sought to break Microsoft up into separate operating systems and applications companies, the D.C. Circuit signaled that this remedy would be unwarranted.22 However, conduct remedies face the twin problems of loopholes and market evolution that can affect the firm’s choice of exclusionary instrument, which makes conduct remedies hard for a generalist court to design and enforce. It is difficult for a consent decrees to anticipate all the ways in which the firm might respond in a dynamic industry.For example, the Microsoft consent decree did not anticipate the potential to exclude competing search engines. This is one of the rationales for sectoral regulation, rather than simply relying on antitrust, though regulation also can face similar problems.23

Finally, section 2 doctrine has become much more defendant-friendly over time. The expansive approach of Alcoa24 is ancient history. Skeptical conservative commentators successfully advocated for higher procedural and substantive burdens on plaintiffs, even for defendants with substantial market power.25 They argued that exclusionary conduct allegations relied on defective economic reasoning: that monopolists are important innovators and that monopolized markets would rapidly self-correct. However, post-Chicagoans make the point that whatever relevance these assumptions had for 1950s and 1960s antitrust, they clearly are no longer justified.26 A rigorous economic basis for the competitive harms from exclusionary conduct claims has been established. Theoretical and empirical evidence has shown that innovation incentives are stronger when a dominant firm faces competition.27 And it is now clear that the market self-correction claim ignores the fact that the exclusionary conduct itself raises artificial barriers to entry.28

Conservative commentators also overlooked the fact that defendants’ much larger financial stakes give them incentives to spend much more to defend their interests than rivals, who are attempting to enter and create a competitive market, and so have much lower financial stakes and resulting incentives to invest in private litigation. These financial asymmetries tend to skew outcomes in favor of the defendant relative to the underlying merits and thus increase the likelihood of false negatives (i.e., underdeterrence and erroneous convictions).29 Public enforcers similarly are disadvantaged because they face severe budget constraints.

The digital platforms compete in two-sided markets, which can complicate the analysis. This was the case in the recent American Express case.30 While this case involved exclusionary vertical agreements by a platform with market power (not monopoly power) and was alleged to violate section 1 of the Sherman Act (not section 2), it is useful for illustrating the hurdles facing plaintiffs in section 2 litigation involving two-sided digital platforms.

B. Two-Sided Markets

Two-sided digital platforms provide services for multiple users that interact through the platform and comprise an interdependent network ecosystem.31 For example, Google offers search services to users and advertising opportunities to companies who want to reach the searchers. Google and Apple provide operating systems and application stores that permit developers to sell applications to users.32 Facebook and Twitter provide platforms for users to communicate with each other and for advertisers to deliver targeted advertisements to users. Amazon and eBay provide retail marketplaces that connect vendors and consumers. All of these companies also sell data on user characteristics and behavior. Visa and American Express connect banks, merchants and card holders.33 Sabre and Expedia link airlines, travelers and travel agents to make travel bookings.34

These platforms benefit from network effects, whereby an increase in the number of users will increase the attractiveness of the network to the firms on the other side of the platform as well as to other users. In search, for example, more searches generate more user data, which can permit more narrowly targeted advertising and better recommendations. Network effects also can entrench the power of a dominant provider. Network effects are not a unique feature of digital platforms. Farmers’ markets and shopping malls create network effects in that a more desirable set of vendors or stores attract more shoppers and vice versa.35 What makes these digital transaction platforms more worrisome is their geographic and commercial breadth, which exacerbates their influence over commerce and creates higher barriers to entry and expansion for entrants and smaller competitors.

Platforms account for the network effects and interdependencies between users on both sides of the market in setting prices and offering services. For example, Google does not charge users for searches, but instead earns revenue from advertising and data.36 The greater the number of searches, the greater the advertising and data revenues earned by Google. Other platforms act similarly. Credit card networks charge high fees to merchants and offer “rewards” to high volume card users.37

Two-sided competition complicates antitrust analysis because it requires analysis of the competitive interdependence and impact on customers in both affected markets. For example, the Federal Trade Commission (FTC) investigated allegations that Google engaged in exclusionary conduct against rival shopping services by manipulating search results, harming its shopping service rivals, and permitting Google to charge higher prices to the merchants to which it directed traffic.38 However, the FTC did not bring a case, concluding that Google’s conduct may have benefited consumers on the other side of the market by increasing the quality of consumer searches so that consumer welfare was not necessarily harmed on balance.39

The platforms track users and sell data to advertisers that arguably reduce users’ privacy. But Google and others might argue that most users value the targeted ads and users place a higher value on free search (or lower prices in the case of other networks) than any loss in privacy. All these firms also might argue that any loss in privacy does not raise barriers to entry because a new entrant could promise not to share consumer information. However, this argument ignores the other substantial entry barriers. For example, while the DuckDuckGo search engine offers more privacy protections than Google, its share is very small.40 Google might argue that privacy has low value, whereas Google’s opponents might argue that other engines’ low share is the result of insufficient search data.

These platforms also operate in multiple market spaces. Google also runs real-time auctions in which advertisers bid for ad space on the websites of various publishers. Critics allege that Google has used exclusionary conduct in supplying advertising technology services to both publishers and advertisers.41 Google might argue in rebuttal that its conduct is efficient competition on the merits.

Critics have alleged that Amazon excludes competition by preventing vendors from charging lower prices on other retail shopping platforms.42 If this case were litigated, Amazon likely would argue that it lacks monopoly power in a relevant retail market that includes brick-and-mortar stores, whereas opponents would allege an e-commerce market. Amazon also might argue that its conduct prevents free riding on Amazon’s website, whereas opponents would allege that Amazon prevents merchants from charging lower prices on other platforms or their own websites, thereby raising entry barriers. For example, a new marketplace entrant might want to charge lower commissions to vendors, allowing them to offer lower prices on the marketplace as a way to attract more customers to the marketplace. If the vendors cannot offer lower prices than Amazon, the marketplace is less likely to attract traffic, which will make entry less likely. Finally, Amazon also might claim that its tracking behavior improves the shopping experience of its users by offering them valuable recommendations, recommendations that they are not forced to accept. However, opponents might counter that Amazon can reduce competition by deceptively skewing recommendations to disadvantage retail competitors.

It is appropriate for antitrust to account for the effects on both sides of a two-sided market. One question is which party should bear the evidentiary burden. Should it be sufficient for the plaintiff to produce evidence of harm on one side to shift the burden to the defendant to produce sufficient evidence of offsetting benefits to then shift the burden back to the plaintiff; or should the defendant have the burden of proof that benefits exceed costs? The other question is the appropriate height of the burdens. In my view, the inquiry should place relatively more weight on avoiding false negatives (i.e., erroneous acquittals and under-deterrence) than false positives (i.e., erroneous convictions and over-deterrence). The competitive concerns are substantial and the high barriers to entry and expansion can allow the conduct to entrench the platforms’ monopoly power for a significant period of time. This increases the cost of false negatives relative to false positives.43 In addition, many of the claimed benefits of the challenged conduct likely can be achieved by less exclusionary alternatives—a fact that reduces the cost of false negatives. As a result, accepting a small reduction in those benefits generally can lead to much larger benefits from more vigorous competition. The legal standards also should take account for the fact that section 2 litigation outcomes likely are distorted by the fact that the defendants have the incentive to greatly outspend private plaintiffs and that public-agency budgets are limited.

C. Ohio v. American Express

The potential complexities from accounting for the effects on both sides of a two-sided market can be illustrated by the recent Supreme Court decision in Ohio v. American Express.44 The Department of Justice (DOJ) and several States filed suit against American Express, Visa, and MasterCard in 2010 for “anti-steering” provisions contained in each of their contracts with merchants.45 These provisions prohibit merchants from responding to high fees charged by a card brand by attempting to steer customers to use another card brand, either with surcharges, discounts, signs or verbal requests. Visa and MasterCard agreed to delete the provisions, but Amex litigated. While the plaintiffs prevailed at trial, the Second Circuit reversed.46 The Supreme Court affirmed the Second Circuit in a 5-4 decision by Justice Thomas.47

Justice Breyer’s dissent explained why the anti-steering provisions are
anticompetitive.48 If steering is prevented, a merchant’s only recourse to a higher network fee is to cease accepting the card altogether. Because each of these networks accounts for a significant fraction of transactions, and many consumers would not simply use another card, that response would be very costly to the merchant, which substantially reduces its countervailing power.49 As a result, the networks’ constraints on fee increases are reduced, which leads to higher fees by all of them. Faced with higher costs of card acceptance, merchants are led to raise their retail prices, which harms both card users and also the generally lower income customers who pay with cash. The card networks spend some of their higher fees on rewarding card usage, particularly by higher income users. Thus, some card holders may gain from higher merchant fees, despite higher merchandise prices. Still, most do not and, of course, cash customers clearly are harmed.50

These rules also create a dysfunctional market dynamic. All customers pay the higher merchandise prices, which creates an incentive to use cards rather than cash, even if the reward is very small. This leads to a vicious cycle in which there is more card usage, which in turn further raises merchants’ costs and retail prices, which further encourages more card usage, and so on. Indeed, card usage likely exceeds the competition level.

Despite this logic and supporting evidence of actual fee increases, the Supreme Court reversed. It characterized the Amex fee increase as competition on the merits because it could support larger user rewards. In doing so, the Court ignored the impact on cash customers and the users of other cards. It also seemed to ignore the fact that the other networks similarly would have incentives to raise their merchant fees in response. The Court erroneously relied on the increasing volume of card transactions over time as evidence of procompetitive output increases, ignoring the fact that card usage rises as the economy grows and ignoring the dysfunctional market dynamic.

Importantly, the Court also found that the plaintiffs did not show that Amex had market power in the relevant two-sided transaction platform market. The plaintiffs had defined the relevant market as services to merchants and concluded that Amex had market power in this market. They also traced through the potential effects on rewards on the other side as a procompetitive efficiency benefit, but found that the rewards did not adequately compensate for the likely higher merchandise prices.

The plaintiffs had evidence that merchant fees were raised as a result of Amex’s conduct and the rewards were insufficient to offset the harms from the higher fees. As Justice Breyer correctly explained, “proof of actual adverse effects on competition is, a fortiori, proof of market power. Without such power, the restraints could not have brought about the anticompetitive effects that the plaintiff proved.”51 But, the Court was unwilling to rely on this direct evidence of anticompetitive price effects to infer Amex’s market power.52 Moreover, the Court’s rejection of direct evidence was in contrast to the modern trend in antitrust law adopted for horizontal restraints in NCAA53 and Indiana Federation of Dentists.54 Similarly, in Microsoft,the D.C. Circuit relied on direct proof of monopoly power as well as circumstantial evidence.55

However, the Court here took the position that direct evidence was insufficient for vertical agreements such as Amex’s merchant agreement. Instead, it was necessary to prove market power with circumstantial evidence, evenbefore reaching the issue of the likelihood of anticompetitive effects.56 This is an inferior approach. Using circumstantial evidence to support a two-sided platform market and market power is complicated because the parties on the two sides are very different, as are the prices charged to each of them. Antitrust markets are normally defined to include reasonable substitutes. But the parties on the two sides of a platform purchase complementary products from the platform, not substitutes. Moreover, the requirement adds nothing to the substantive analysis when there is direct evidence.

Perhaps most surprising, the Court also did not remand the case to analyze the unitary two-sided platform market but instead concluded that the conduct did not violate the Sherman Act, thereby ending the case. Potential cases against the dominant digital platforms will need to contend with this precedent and the noninterventionist antitrust approach it supports.

II. anticompetitive mergers

The historical trusts amounted to mergers among competing firms. Mergers are governed mainly by section 7 of the Clayton Act, though the Sherman Act also may be applied. Current merger law has features that might be expected to make it effective in preventing anticompetitive consolidation.

First, most significant mergers must be reported in advance, and trigger waiting periods and a “second request” process in which the parties are required to submit documents and data in response to a Civil Investigative Demand from the reviewing enforcement agency. This provides the agency with time and information to carry out a detailed competitive effects analysis. If the reviewing agency concludes that the merger raises significant competitive concerns, the merging parties will often accept a corrective settlement—generally a divestiture or behavioral remedy.

Second, if the parties refuse the settlement, the agency can seek an injunction to stop the merger (unlike in Europe where the European Commission’s Competition Directorate can stop a merger on its own, subject to appeal to the courts).57 However, U.S. merger law is tilted somewhat in favor of the agencies. Section 7 of the Clayton Act condemns mergers that “may be substantially to lessen competition” or “tend to create a monopoly.”58 This language is not interpreted to require evidence that harm “is more likely than not.” Instead, the standard requires showing only a “reasonable probability” that competition will be reduced.59

A. Horizontal Mergers

For horizontal mergers, there is another agency advantage. The law has developed an anticompetitive presumption for mergers between competing firms with substantial market shares in concentrated markets.60 While the strength of this “structural presumption” has declined over time, it importantly contains a “sliding scale,” whereby stronger structural evidence increases the parties’ rebuttal burden.61

The agencies nonetheless have permitted substantial consolidation. One possible explanation is that the agencies are resource constrained and risk averse to litigation, which leads them to approve problematic mergers or accept weak consent decrees rather than risk losing at trial. The stakes for the parties are huge, incentivizing large expenditures to defend their deals. By contrast, agency budgets have declined relative to the number, size, and complexity of mergers.62 The very few matters that are not cleared or settled simultaneously with consent decrees typically involve markets with extremely high concentration, barriers to entry, and substantial potential concerns, and the agencies almost always prevail. Between 2012 and 2018, the merging parties either abandoned challenged mergers, subsequently settled before trial, or lost in court in more than thirty-five cases, whereas the parties prevailed in only three.63 Only one of those three litigated cases involved a plain-vanilla horizontal merger.64 Two others were nonhorizontal mergers where the structural presumption does not apply—one was a vertical merger65 and the other was a potential entry merger.66 Another notable recent loss was a vertical-plus-horizontal acquisition by a two-sided platform decided in 2020.67

The role of empirical evidence quantifying anticompetitive harms has also increased, and the agency’s litigation advantage has declined as a result. Econometric studies and other data analyses can be challenging for judges, who are time and resource constrained. If the agency presents its empirical evidence in its case in chief, it may tend to be treated as part of the prima facie case rather than being seen as countering the parties’ empirical evidence in rebuttal, so that the agency may be seen as having the burden, notwithstanding the structural presumption. This is important because data sets are invariably imperfect, and it is always possible to poke little holes in even highly rigorous econometric studies. Econometric evidence can be probative, so it is important for courts to spend the necessary time to properly evaluate the evidence and not permit nitpicks to obtain disproportionate importance. Courts also should be cognizant that standard tests of statistical significance are focused on avoiding false positives, not false negatives.68 In this regard, courts might consider retention of court-appointed independent experts or advisors.

Another possible reason for weak enforcement is a belief by agency leadership that a market with at least four major competitors is normally enough to ensure sufficient competition.69 This belief, which goes back to the writings of Robert Bork,70 stems from a presumption that cartels and tacit coordination are unstable, particularly if there are more than a few firms. It also relies on a presumption that most mergers lead to procompetitive efficiencies. However, these views are unsupportable in light of the evidence that there have been many durable cartels with significant numbers of firms, and numerous merger retrospective studies have shown price increases from mergers that have been cleared, sometimes despite consent decrees intended to remedy concerns.71

B. Vertical Mergers and Acquisitions of Potential Competitors

Acquisitions of vertically related firms and potential (including nascent) competitors may have helped dominant digital platforms to maintain their power. Possible entrenching acquisitions include Facebook’s acquisitions of Instagram, WhatsApp, and now Giphy; Google’s acquisitions of YouTube, AdMob and Waze; and Amazon’s acquisitions of Quidsi (the parent of Diapers.com), Zappos, Wholefoods and Kiva (robotics).72

These cases are harder for the agencies to win. Current legal doctrine does not recognize any structural or other anticompetitive presumptions applied to vertical mergers or potential competition acquisitions. In fact, the agencies lost the only potential competition case litigated in some time and the two cases involving vertical merger issues.73 We turn now to a discussion of these three cases.

1. FTC v. Steris

There have been few potential-competition mergers litigated in the last thirty years. One reason is the high burden placed on merger challengers to show that the acquired firm is a likely entrant. For example, in its 1984 B.A.T. decision, the FTC required “clear proof” that the entry would have occurred.74 In 2015, the FTC failed in its case to block Steris’s acquisition of Synergy. Steris is the leading U.S. competitor in contract medical sterilization, and the FTC alleged that Synergy was a likely potential entrant into the concentrated U.S. market.75 The court rejected the claim, holding that the FTC did not show that Synergy “probably” would have entered the market absent the merger.76

If the agency passes this first hurdle, it also must show that the independent entry likely would have significant procompetitive effects, relative to the acquisition. There is no anticompetitive presumption for potential competition mergers. This second hurdle could be significant if the agencies attack similar proposed or consummated acquisitions by digital platforms. The acquired firm may not have grown if it entered independently. For example, at the time of Facebook’s acquisition of Instagram in 2012, Instagram was a very successful startup but with only thirteen employees and around thirty million users.77

If the FTC attacks this consummated merger, the following issues would arise. The FTC likely will produce documents and testimony that Facebook saw Instagram as a major threat to its market power, and that fear motivated the billion-dollar
acquisition.78 By contrast, Facebook likely will argue that Instagram’s standalone prospects were low, and it was Facebook’s investment and expertise that made Instagram highly successful. Like others, I would be interested to see the evidence on both sides of this case and the others discussed in this Essay.

My policy concern is that the courts may be overly concerned with false positives and hold the agencies to an overly high burden that will allow dominant platforms to entrench their monopoly power by systematically purchasing the most threatening potential and nascent entrants. For example, suppose that Facebook’s acquisition did grow Instagram somewhat faster and added some useful features. These product benefits likely would be swamped by the competitive benefits of Facebook facing a powerful new competitor, benefits which might have included placing more pressure on Facebook to innovate faster and better. As Judge Learned Hand suggested in Alcoa, “Many people believe that immunity from competition is a narcotic, and rivalry is a stimulant, to industrial progress.”79

2. United States v. AT&T80

In 2017, the Department of Justice sued to block the acquisition of Time Warner by AT&T. This was a vertical merger; Time Warner provided video content to AT&T’s DirecTV and video distribution competitors like Dish, Comcast, Charter and
Verizon.81 Because the merging firms are not direct competitors, vertical mergers do not directly increase concentration and the structural presumption does not apply. Nor have courts devised alternative anticompetitive presumptions.82

The most common competitive concern in vertical mergers is “foreclosure” of the input sold by the merging firm, which can raise the costs or otherwise disadvantage rivals and thereby increase or preserve the market power of the merging firm.83 In modern parlance, foreclosure is gauged by the effect on the rivals’ ability to compete, not simply by the fraction of customers or input supplies tied up by the defendant’s exclusive contracts or ownership.84 The DOJ’s expert economist testified that the foreclosure effect was Time Warner’s ability to gain additional bargaining leverage over DirecTV’s competitors, which in turn would give Time Warner the ability to negotiate higher license fees by threatening not to renew the licenses. The higher fees would lead to consumer harm in the form of higher monthly subscription fees charged by DirecTV and its distribution market competitors. The court disagreed, relying on testimony by the parties’ executives and criticism of the model and empirical evidence presented by the DOJ’s expert.

In retrospect, the DOJ may have erred in choosing to litigate. DOJ previously had accepted a consent decree in the parallel vertical merger between Comcast and NBCU, but rejected a similar voluntary consent decree offered by AT&T. Worse for the DOJ, the judge assigned to hear the case was the same judge who oversaw the Tunney Act judicial review of the Comcast/NBCU consent decree, a matter in which DOJ claimed that the consent decree would be effective.85

The DOJ also accepted a greater litigation burden than necessary. It deviated from the standard three-step burden-shifting framework established in the Baker Hughes merger case.86 As applied to a vertical merger, it would have been sufficient in the first step for the DOJ to produce evidence of likely competitive harm, under the assumption that there are no efficiency benefits. The burden then would have shifted to the merging parties to rebut this evidence or to produce sufficient evidence that there would be procompetitive efficiency benefits. Only if the parties were able meet this burden would the burden have shifted back to the agency to show that there was a reasonable probability that the merger would be harmful overall.

However, the DOJ did not follow this approach.87 Rather than simply relying on the fundamental economic drivers of increased bargaining leverage in the first step (the prima facie case), they also chose to introduce substantial quantitative evidence of “upward pricing pressure” from the leverage as the centerpiece of their case.88 The DOJ also conceded that the merger would lead to an “elimination of double marginalization” (EDM) efficiency benefit that would create “downward pricing pressure” on DirecTV to reduce the prices it charged subscribers after the merger. This is because DirecTV would obtain Time Warner content for free after the merger.89

Thus, the DOJ essentially embraced from the outset the added burden to show in their affirmative case that consumers would be harmed on balance. The DOJ did not attempt to place the burden of showing these procompetitive efficiencies on the merging parties. Had they done so, the parties would have to explain why they could not have achieved EDM by contract absent the merger.90 After all, if the efficiencies were not “merger-specific” but could be achieved absent the merger, then there would be no policy rationale for consumers to bear the risk of higher prices from the
merger.91

If the DOJ instead took the approach of presenting a simpler prima facie case that focused on qualitative and less complex quantitative evidence to shift the burden to the parties, their chances would have been higher. The parties may have had a hard time presenting credible testimony for why they failed to solve the efficient contracting problem before the merger, yet would pass-on the EDM as lower prices to subscribers after the merger. For example, if premerger EDM would have reduced the likelihood of anticompetitive coordination, then they would not have an incentive to pass-on the savings after the merger. However, a DOJ victory certainly would not have been assured. The trial judge was very skeptical (overly skeptical in my view) of the credibility of the testimony of Time Warner’s cable TV customers, who also were competitors of DirecTV, and overly accepting of the credibility of the testimony of the AT&T and Time Warner executives.92

3. United States v. Sabre Corp.93

This most recent agency loss involved an acquisition by a dominant digital platform. Sabre is a digital platform that permits airlines to post schedules, fares and seat availability and allows travel agents to access this information, make travel bookings and pay for them. Sabre proposed to acquire Farelogix, which provides technology to airlines. This technology allows an airline to disintermediate Sabre by allowing the airline to connect directly to travel agencies and provide travel agencies with information and ticket-booking services itself. Thus, this acquisition was analytically like a vertical merger, where Farelogix sells a critical input (i.e., its technology) to airlines, which they use to compete with Sabre for the business of travel agents. The competitive concern is that Sabre would foreclose airlines’ ability to acquire the Farelogix technology input.

Perhaps attempting to exploit the horizontal-merger structural presumption and avoid the difficulties they faced in AT&T/Time Warner, the DOJ did not litigate the case as a vertical merger. Instead, the complaint alleged that Sabre and Farelogix competed in the provision of booking services for airline tickets sold through travel agencies. This competition is indirect, resulting from Farelogix working with the individual airlines to disintermediate Sabre. However, the trial court did not miss the point. It observed that “Sabre and Farelogix view each other as competitors” and found that “the record reflects competition between Sabre’s and Farelogix’s direct connection solutions for airlines.”94

Having concluded that competition was reduced by the merger, the trial court nonetheless rejected the DOJ’s complaint on the grounds that Farelogix and Sabre do not compete in the two-sided platform market.95 While Sabre provides services to customers on both sides (i.e., to both airlines and travel agencies), Farelogix provides services to only one side (i.e., to airlines, but not to travel agencies). The travel agency services are provided by the airlines themselves, using the Farelogix technology.

This approach was both defective and unnecessary because Sabre competed with the combination of Farelogix and the airlines.96 Yet the court thought that American Express compelled the opposite result, despite its own fact-finding and the vertical nature of the transaction. If other U.S. courts similarly follow this same defective approach, the result will be underdeterrence of anticompetitive acquisitions by digital platforms.97 Indeed, this approach would lead to ludicrous results. Under this reasoning, Microsoft could have legally ended the competitive threat from Netscape and Java simply by acquiring them instead of trying to destroy them.

III. strengthening antitrust

This Essay has indicated some difficulties that antitrust enforcers will face under current law if they conclude that a more aggressive approach is needed to rein in dominant digital platforms. Antitrust is a common-law process, so doctrine in principle could accommodate this new set of circumstances.

First, merger law could evolve by adopting a presumption that all acquisitions by dominant digital platforms protected by network effects are presumptively anticompetitive.98 More interventionist merger guidelines governing vertical mergers and potential entry mergers could influence courts in this direction. However, while the recently promulgated Vertical Merger Guidelines had an opportunity to do so, the agencies instead adopted a nonenforcement bias. This approach was criticized by the two dissenting FTC Commissioners as well as my coauthors and me.99

Second, exclusionary conduct by dominant firms could be subjected to a sliding scale with a reduced burden of proof on the plaintiff and a more skeptical view of their efficiency claims. These doctrinal changes could lead to more antitrust enforcement and greater deterrence of exclusionary conduct by dominant digital platforms. This change would be justified by the recognition that such agreements and conduct by dominant networks involve greater harms from false negatives than false positives.100 Greater emphasis on preventing false negatives is particularly appropriate in that antitrust condones monopoly pricing not found to arise from anticompetitive conduct.101 Thus, it is more important to weaken the monopoly instead. In addition, as noted earlier, excluded rivals who bring private cases are disadvantaged in litigation by the asymmetry in the financial stakes.

Third, another possibility to improve merger enforcement could be legislation that increases agency budgets and mandates both a lower burden on the plaintiff and a higher rebuttal burden on the defendant. For example, Senator Amy Klobuchar introduced a bill to amend section 7 of the Clayton Act to reduce the agencies’ evidentiary burden by changing the standard from “substantially” lessens competition to “materially” lessens competition, where “materially” is defined to mean “more than a de minimis amount” and requires the firms in rebuttal to show that their deal would not materially harm competition.102

Fourth, a similar strengthening also could be applied to section 2 of the Sherman Act as a complement to the sliding scale approach. Some commentators also have proposed regulation of exclusionary conduct by a new Digital Authority.103 In light of the potential need to closely monitor the changing conduct of dominant digital networks over time, regulation may be necessary to supplement antitrust.

This proposed strengthening also harkens back to history. Shortcomings in the Sherman Act led to the creation of the FTC and the adoption of the Clayton Act and FTC Act in 1914. The Clayton Act was amended in 1936 with the Robinson-Patman Act to better deter discriminatory prices that reduced competition. Congress further amended section 7 of the Clayton Act to close loopholes and strengthen merger enforcement in 1950. Finally, the Hart Scott Rodino (HSR) Act in 1976 improved merger enforcement by requiring premerger notification of large mergers and the “second request” process. Prior to the HSR Act, mergers could be consummated before the agencies had a chance to act, which caused enforcement delays. It also reduced the ability to obtain meaningful relief because it often was difficult to “unscramble the eggs” after the merging firms integrated their operations.

However, economic deregulation and the loosening of antitrust standards that began in the late-1970s have weakened constraints on mergers and dominant firms. The time has come today to move in the other direction. Because the common law can be slow to adjust, and in light of the conservative antitrust views of the Supreme Court majority, new legislation would be a faster and more certain route to reform.

Professor of Economics & Law, Georgetown University Law Center and Senior Consultant, Charles River Associates. I have benefited from helpful comments and discussions with Jonathan Baker, Dale Collins, Carl Shapiro, and John Woodbury. I regularly consult for law firms, corporations and government agencies on antitrust matters. All opinions expressed here are my own.