Antitrust is having a moment. Pundits and policymakers across the political spectrum are calling for increased antitrust enforcement to address all manner of social ills. From technology platforms’ power over speech and encroachments on user privacy, to wage stagnation in more concentrated labor markets, to growing income inequality, reduced innovation, and threats to democracy itself, the list of maladies for which antitrust has been proposed as a remedy goes on and on.

Antitrust enforcers have taken note. Last year, the U.S. Federal Trade Commission hosted 14 public hearings to gather feedback on how antitrust might be revamped to address contemporary concerns. The FTC then joined the Antitrust Division of the U.S. Department of Justice in announcing probes of Google, Apple, Facebook, and Amazon. At the state level, nearly every state attorney general has joined antitrust investigations of Facebook (led by New York’s attorney general, a Democrat) and Google (led by Texas’s attorney general, a Republican).

Given policymakers’ heightened interest in antitrust and the recent flurry of bipartisan enforcement activity, it is worth stepping back to ask a couple of big-picture questions: What are antitrust’s limits in addressing the social harms that are motivating calls for more aggressive enforcement? And how should enforcers and courts proceed in light of those limits?

These questions are not new. In 1984, Frank Easterbrook addressed them in an influential Texas Law Review article entitled “The Limits of Antitrust.” This article revisits his answers in light of the current antitrust moment. It concludes that his overarching prescription remains fundamentally sound but that his view about the relative harms from over- versus under-enforcement, as well as some of the specific screening mechanisms he proposed for optimizing antitrust’s effectiveness, require some adjustment. It then suggests four new screening mechanisms that could assist 21st century courts and enforcers in ensuring that antitrust secures as much social welfare as possible given its inherent limitations.

Easterbrook’s “Limits of Antitrust” Approach

The approach Easterbrook set forth in 1984 included three key components. The first was an overarching goal that courts should pursue in crafting antitrust doctrines. Antitrust is concerned with business behaviors that generate market power: coordinated conduct that leads to collusion and exclusionary actions that create monopoly power. The difficulty is that many acts of coordination between firms enhance market output, and many business practices that usurp sales from the actor’s rivals—and thus “exclude” them from the market—also generate benefits for consumers. Extremely low prices, for example, may drive rivals from the market, but they offer an obvious and immediate benefit to consumers. Similarly, exclusive dealing agreements by manufacturers may raise their rivals’ costs of distribution, but they may also spur manufacturer investment in distributors by reducing free-riding by competing manufacturers.

Regulating these sorts of competitive mixed bags inevitably entails costs. First, there are the costs that result from mistaken judgments. If the law wrongly allows conduct that is, on net, anticompetitive, consumers will face higher prices and/​or reduced quality, and a deadweight loss will occur. But if the law wrongly forbids conduct that is, on balance, procompetitive, market output will be lower than it otherwise would be and, again, consumers will suffer. In addition to these so-called “error costs,” regulating competitive mixed bags entails significant “decision costs”—i.e., costs to business planners and courts from deciding whether contemplated or actual conduct is forbidden or permitted.

False conviction error costs, false acquittal error costs, and decision costs are intertwined. If policymakers try to reduce the risk of false conviction by making it harder for a plaintiff to establish liability or easier for a defendant to make out a defense, they will raise the risk of false acquittal. If they ease a plaintiff’s burden or cut back on available defenses to reduce false acquittals, they will increase false convictions. And if they make the rule more nuanced in an effort to condemn the bad without chilling the good, thereby reducing error costs overall, they enhance decision costs. As in a game of whack-a-mole, driving down costs in one area will cause them to rise elsewhere.

Given this unhappy situation, Easterbrook proposed an overarching goal for antitrust policies: they should be crafted so as to minimize the sum of error and decision costs. Pursuing such an objective, policymakers would not try to prevent every anticompetitive act, allow every procompetitive one, or keep antitrust rules as simple as possible. They would eschew perfection along any single dimension in favor of overall optimization.

The second key component of Easterbrook’s approach was his instruction about how antitrust tribunals should weigh the harms from false convictions versus false acquittals. If a procompetitive behavior is wrongly condemned, the adverse effect—squandered efficiencies—will persist until a subsequent judicial decision overrules the erroneous precedent. By contrast, if anticompetitive conduct is wrongly allowed to persist, the result will be the sort of monopoly pricing that invites market entry and thereby self-corrects. Accordingly, Easterbrook reasoned, false convictions are worse than false acquittals, which suggests that liability rules on questionable practices should be calibrated so as to err in the direction of allowing anticompetitive acts rather than banning or discouraging procompetitive ones.

The third key component of Easterbrook’s approach was a set of five screening mechanisms designed to filter out challenges to practices that are likely to be procompetitive:

  • Is there market power? The court should dismiss actions in which the defendant(s) lacked market power because anticompetitive harm is unlikely to occur absent such power.
  • What is the logical relation between the defendant’s profit and reduced competition? Business conduct that injures the perpetrator’s rivals but does not enhance its profits is self-deterring, so antitrust intervention is unnecessary. If the challenged practice could enhance the defendant’s profits even apart from a reduction in competition, condemnation of the practice could deter procompetitive conduct. Courts should thus dismiss antitrust claims that do not show how the defendant’s profits are rising from a reduction in competition.
  • Is there widespread adoption of identical vertical restraints? Easterbrook asserted that for trade-restraining agreements between firms operating at different levels in the same supply chain, anticompetitive harm is likely only if the restraints are widely adopted throughout the relevant market. Challenges to vertical restraints employed by only one or a few competitors in an industry should therefore be dismissed.
  • What is the effect on the defendant’s output? Because an exercise of market power (at least on the seller’s side) involves a constriction of output and consequent increase in price, evidence that the defendant’s output is increasing despite the challenged conduct suggests that the conduct is procompetitive and the challenge without merit.
  • Who is the plaintiff? Because a defendant’s customers benefit from enhanced competition in the defendant’s market but are harmed when competition is reduced—and its rivals, vice-versa—competitor complaints should be viewed with suspicion.

Assessing Easterbrook’s Approach Today

Much has changed in the 36 years since Easterbook published his classic article. The business world has been transformed by the internet, mobile telephony, and digital platforms and social networks. Economic learning has also advanced, with scholars gaining a better understanding of how certain business practices can be anti- and/​or pro-competitive. How does Easterbrook’s late-20th century approach look in light of 21st century market developments and advances in economic learning?

His overarching objective for antitrust policy decisions remains sound. Since 1984, no developments in market structures or economic learning have altered the mixed-bag nature of the behavior antitrust regulates, the consequent inevitability of error and decision costs, or the fact that efforts to reduce one set of costs will drive up another. Scholars have progressed in their understanding of the circumstances under which particular behaviors may occasion anticompetitive harm or create procompetitive benefits, and that new knowledge may allow courts to restructure doctrines so as to reduce costs overall. But antitrust remains an inherently limited enterprise and Easterbrook’s overarching prescription for maximizing welfare in light of those limits—craft policies to minimize the sum of error and decision costs—remains as wise as ever.

His claim that false acquittals self-correct and should generally be favored over false convictions has fared less well. Economic learning has revealed that some forms of exclusionary conduct do not automatically self-correct. For example, some actions by a dominant firm can raise rivals’ costs or prevent them from growing enough to attain scale economies that would enable them to underprice the dominant firm. Indeed, in markets characterized by large economies of scale and network effects (e.g., digital social networking, computer operating systems), entry and underpricing may be particularly unlikely. Easterbrook’s point about incommensurate harms should thus be softened somewhat. In deciding whether to tilt the liability rule in favor of permitting questionable conduct, courts should ask whether any resulting market power would be transitory (as with collusion, which is difficult to maintain and invites entry) or durable (as with some exclusionary practices in some types of markets). Sometimes a pro-defendant bias will be appropriate, but not always.

Like his instruction on incommensurate harms, Easterbrook’s screening mechanisms for filtering out procompetitive behaviors require some adjustment. The first two screens—that defendants possess market power and that the challenged conduct enhance their profits by reducing competition—have fared well and continue to enjoy support in the case law. The fifth—weeding out competitor complaints—remains useful in some contexts. In challenges to horizontal mergers, for example, complaints by rivals should raise red flags because competitors benefit from reduced competition and are injured when their rivals become more efficient. Developments in economic learning, though, suggest that the mere fact that the complainant is a competitor does not always signal that the challenged practice is procompetitive. We now understand that many exclusionary practices (e.g., exclusive dealing arrangements involving substantial market foreclosure) may injure competition by raising rivals’ costs. Because such practices hurt both consumers and competitors, the mere fact that a competitor is complaining does not indicate that the challenged practice is procompetitive. A useful revision would be to say that behaviors drawing competitor, but not consumer, complaints are likely procompetitive.

Easterbrook’s third and fourth filters have not stood the test of time. The third, which eliminates challenges to vertical restraints that are not in widespread use throughout the market at issue, rests on a premise that we now understand to be faulty. Contrary to his assertion, a single firm’s vertical restraints can sometimes injure competition, as when a dominant producer’s exclusive dealing contract denies its rivals a critical input or forecloses them from a substantial proportion of sales opportunities and thereby holds them below minimum efficient scale.

Easterbrook’s fourth filter, which screens out actions against any defendant whose output is not falling, is similarly problematic. This screen may be appropriate when the alleged anticompetitive harm is collusion—some kind of agreement to restrain output so as to increase price and enhance profits. But if the defendant has engaged in unreasonably exclusionary conduct to drive rivals from the market or raise their costs, its output may rise, particularly if market demand is increasing.

Four Additional Screens for the Current Era

In addition to softening Easterbrook’s incommensurate harms principle and revising or eliminating some of his particular screening mechanisms, courts attempting to optimize antitrust’s effectiveness in the current antitrust moment should adopt four additional filters:

1) Does the challenged practice entail consumer harm? An initial 21st century filter—no imposition of antitrust liability absent consumer harm—would not have seemed worth mentioning when Easterbrook authored his 1984 article. Only six years before, Robert Bork had published The Antitrust Paradox: A Policy at War with Itself. He purported to show that the purpose of the Sherman Act, as revealed in its legislative history, was to enhance consumer welfare, which he equated with maximizing efficiency. While Bork’s reading of legislative history has been severely questioned, if not discredited, his effort to focus the antitrust laws on consumer welfare met with success. In 1979, the U.S. Supreme Court proclaimed the antitrust laws to be a “consumer welfare prescription” and, ever since, the prevailing view among courts has been that antitrust’s sole end is consumer welfare, a view known as the “consumer welfare standard.” It is thus no surprise that Easterbrook did not, in 1984, propose a screening mechanism to weed out antitrust actions aimed at some other objective.

In recent years, numerous commentators from both the left and right have called for abandoning the consumer welfare standard. They maintain that antitrust’s focus on consumer welfare prevents it from addressing:

  • big companies’ monopsony power over labor and inputs (which tends to drive down prices and therefore may not appear to harm consumers);
  • reduced innovation resulting from “kill zones” around dominant firms (as reduced innovation does not have an immediate effect on price or quantity and is therefore unlikely to register under the consumer welfare standard);
  • harms, such as reduced privacy, in zero-price markets (as defendants can always avoid liability by showing their products are “free” to users); and
  • non-consumer harms that result from companies’ being too big (e.g., job losses and community harms from the failure of small businesses that cannot match their larger rivals’ efficiencies, wealth inequality that is exacerbated as giant businesses distribute their massive profits to managers and large shareholders, and harms to democracy resulting from big businesses having excessive political influence).

Given those concerns, many have called for courts to abandon the consumer welfare standard. Sen. Elizabeth Warren’s “Anti-Monopoly and Competition Restoration Act” legislation, for example, would declare that the “antitrust laws were not created exclusively to enhance the narrowly defined concept of ‘consumer welfare.’ ” The draft act instead provides that the purpose of the antitrust laws is to protect “market structures that … restore and protect competition between rivals” for the benefit of “workers, consumers, entrepreneurs, and citizens.”

Jettisoning the consumer welfare standard in favor of some sort of multi-goaled, structural approach is both unnecessary and undesirable. It is unnecessary because each of the “blind spots” identified by critics of the consumer welfare standard is either addressed by the standard, more appropriately addressed by a body of law other than antitrust, or best left unaddressed.

Monopsony harms to laborers and input providers, reduced innovation, and harms in zero price markets (e.g., privacy limits on free social media platforms) fall into the first category. Anticompetitive harms occasioned by monopsony power are reachable under the consumer welfare standard because “consumer” refers broadly to a person on the other side of a transaction from the defendant, not necessarily to an end-user consumer. Indeed, the enforcement agencies’ consumer-focused Horizontal Merger Guidelines specifically call for consideration of whether a merger will create monopsony power, which would make no sense if consumers were taken to include only end-user buyers. Those guidelines further explain that the agencies consider potential innovation harms when evaluating proposed mergers, proving that such harms, too, are cognizable under the consumer welfare standard. In fact, of the 164 merger challenges asserted by the FTC between 2004 and 2014, 54 alleged harm to innovation. Non-price harms associated with free services are reachable under the consumer welfare standard because all aspects of the transaction—price, quality, accompanying services, etc.—are relevant to the overall surplus consumers enjoy.

The non-consumer harms stemming from companies’ “bigness”—wealth inequality, harms to democracy, the loss of small businesses and the jobs they provide—fall into the second and third categories: they are better addressed by bodies of law other than antitrust or best left unremedied. Wealth inequality is better handled through tax and redistribution schemes; harms to democracy can be handled by campaign finance rules and restrictions on lobbying (and, most fundamentally, by limiting discretionary government power so that it cannot be used to procure private advantages for politically connected firms). Job losses and harms to communities from the failure of smaller, less efficient businesses may be somewhat mitigated by job-training programs, community investments, and the relocation of government agencies to economically depressed areas. At the end of the day, though, obsolescence is a consequence of economic development; there will always be some losses when new and better displaces old and less good. Using antitrust to protect economic laggards is sure to reduce welfare in the long run.

Not only is it unnecessary to abandon the consumer welfare standard in favor of a multi-goaled public interest standard, but doing so would have adverse consequences for consumers and the rule of law. We know this from experience. During the mid-20th century, courts did embrace multiple goals for antitrust. They would often interpret the law to be aimed at promoting consumer welfare by encouraging competition so as to lower prices, enhance quality, etc. But they would sometimes impose liability in the absence of consumer harm—in the face of obvious consumer benefit, even—simply to protect smaller competitors from larger, more efficient rivals.

In the 1967 Utah Pie case, for example, the Supreme Court upheld a finding of harm to competition when a large, efficient firm entered a market and underpriced a smaller but locally dominant rival. Reinstating a jury verdict in favor of the rival, the Court concluded that “a competitor who is forced to reduce his price to a new all-time low in a market of declining prices will, in time, feel the financial pinch, and will be a less effective competitive force.” Thus, consumer concerns could be paramount in antitrust cases—unless the court decided to eschew consumer benefit to protect a less efficient competitor.

In its 1962 Brown Shoe decision, the Supreme Court essentially admitted that it could pick and choose whether to put consumers or competitors first. Having conceded that the merger under review could enhance the merged firm’s efficiency, the Court wrote:

Of course, some of the results of large integrated or chain operations are beneficial to consumers. Their expansion is not rendered unlawful by the mere fact that small independent stores may be adversely affected. It is competition, not competitors, which the Act protects. But we cannot fail to recognize Congress’ desire to promote competition through the protection of viable, small, locally owned businesses. Congress appreciated that occasional higher costs and prices might result from the maintenance of fragmented industries and markets. It resolved these competing considerations in favor of decentralization.

As Robert Bork aptly observed,

No matter how many times you read it, that passage states: although mergers are not rendered unlawful by the mere fact that small independent stores may be adversely affected, we must recognize that mergers are unlawful when small independent stores may be adversely affected.

Under such an approach, a court could allow a merger that would benefit consumers by enhancing productive efficiency (if the court followed the first three sentences in the Brown Shoe passage above) or it could choose to block the merger (if it followed the last three sentences). Such leeway naturally trickled down to the enforcement agencies, which could articulate grounds for challenging just about any business conduct by emphasizing its adverse effects on either consumers or competitors.

With enforcers and courts free to pick and choose among antitrust’s multiple goals in order to condemn or acquit virtually any business behavior, antitrust became less a body of law and more an exercise of raw political power. Bork compared it to the sheriff of a frontier town: “He did not sift the evidence, distinguish between suspects, and solve crimes, but merely walked the main street and every so often pistol-whipped a few people.” Even a Supreme Court justice admitted that antitrust had become arbitrary and unprincipled. Dissenting in Von’s Grocery, a 1966 decision that condemned a grocery store merger that generated obvious efficiencies and resulted in a merged firm with a paltry 7.5% market share, Justice Potter Stewart confessed, “The sole consistency that I can find is that, in litigation under [Clayton Act] § 7, the Government always wins.”

When government always wins, winning the favor of government officials becomes paramount. For that reason, abandonment of the consumer welfare standard in favor of a multi-goaled public interest standard would promote politicization of the antitrust enforcement agencies. It would also ensure that consumers, widely dispersed and difficult to organize, regularly lose out to firms and organized interest groups even when the total harms to consumers from an enforcement decision exceed the benefits to the organized interests promoting it.

2) Has the defendant extended market power or just exercised it? Two types of antitrust-related business behavior can injure consumers. The first is an exercise of market power, i.e., the power of a firm lacking competitive constraints to enhance its profits by raising its price above its incremental cost. When a firm exercises market power, it extracts for itself more of the surplus, or wealth, created by its transactions with its customers. Firms may also cause consumer harm by extending their market power. When nominal competitors agree to limit their competition so as to raise prices, their collusion creates market power that would not otherwise exist; when a single firm engages in unreasonably exclusionary conduct that drives its rivals from the market or somehow raises their costs so as to render them less formidable competitors, its market power grows.

While both surplus extraction and market power extension can occasion consumer harm, there should be no antitrust liability absent the latter. One reason for this is practical. If surplus extraction involving no extension of market power were illegal, adjudicators and business planners would confront an intractable question: how much extraction is permitted? Every instance of supracompetitive pricing by any firm with any quantum of market power transfers some surplus from consumers to the producer. It would be impracticable for antitrust to forbid all such surplus extraction, so courts would have to draw some sort of line. But how?

A more important reason for refusing to impose antitrust liability for mere surplus extraction is that doing so promotes dynamic efficiency. The prospect of earning supernormal profits from a lack of competition motivates entrepreneurs to develop unique products and services. In addition, the supracompetitive profits gained through surplus extraction often enable innovation by funding research and development efforts. A glance at the top 15 global spenders on research and development reveals that 11 are either technology firms derided by many as monopolistic (#1 Amazon, #2 Alphabet/​Google, #5 Intel, #6 Microsoft, #7 Apple, and #14 Facebook) or pharmaceutical companies whose patent protections insulate them from competition and allow them to charge supracompetitive prices for their products (#8 Roche, #9 Johnson & Johnson, #10 Merck, #12 Novartis, and #15 Pfizer). Because the static inefficiencies (deadweight losses) occasioned by mere surplus extraction may be dwarfed by the dynamic efficiencies that result from rewarding and financing innovation, antitrust should not forbid practices that extract surplus without also extending market power.

This runs counter to a number of recent proposals. New York University law professor Harry First, for example, has argued that “excessive drug pricing” by pharmaceutical companies could be an antitrust violation. The United Kingdom’s Digital Competition Expert Panel suggested that antitrust might restrict algorithmic pricing systems in which digital platforms harness user data to estimate online purchasers’ willingness to pay and craft personalized prices. While both “excessive pricing” and price discrimination schemes extract additional surplus from consumers, they do not extend sellers’ market power.

Commentators have also raised antitrust concerns about sharp business practices that, while perhaps unsavory (or even tortious), do not extend market power. University of Miami law professor John Newman, for example, points to the “digital blackmail” that may occur when an internet platform manipulates the publication of information in order to extract value from users. Real estate comparison site Zillow, for example, has been accused of suppressing market value estimates that are below a listed property’s sale price (and thus have a depressive effect) in exchange for payments from the listing agent. Restaurant review site Yelp has purportedly threatened to remove or demote favorable reviews of restaurants that decline to purchase advertisements on its site. As unseemly (and perhaps deceptive) as these surplus extractive practices are, they do not extend the perpetrators’ market power; if anything, they threaten it by inviting competition from truly neutral rivals.

3) Does private ordering or another body of law address the potential anticompetitive problem? A third screening mechanism for 21st century antitrust attempts to account for the law’s unique enforcement structure. Because many antitrust violations occur in secret, antitrust seeks to secure optimal deterrence by requiring defendants to pay three times the amount of proven damages. Damage-trebling may lead to overdeterrence, though, when the challenged behavior is (1) “mixed bag” (i.e., sometimes efficient and sometimes inefficient), so that it should not be universally deterred; and (2) not hidden, so that the likelihood that the conduct will be detected and proven is greater than one-in-three. To account for potential overdeterrence resulting from automatic damage-trebling, antitrust should stay its hand when a potentially anticompetitive behavior occurs in the open and another body of law or some sort of contract is capable of preventing any anticompetitive harm the behavior may produce.

This filter would reject current proposals to treat employee non-compete agreements as antitrust violations. Covenants not to compete are not entered in secret and are mixed-bag practices in that they are sometimes anticompetitive (because they unnecessarily limit competition for workers) and sometimes procompetitive (because they create incentives for employers to invest in employee training and give employees access to valuable firm secrets). Contract law polices potential anticompetitive effects, denying enforcement to non-competes that are broader than necessary to secure their purported ends. There is thus little need to invoke antitrust with its distortive enforcement features.

The filter would also have prevented several recent enforcement actions against holders of standard essential patents (SEPs). When a patented technology is incorporated into a technology standard (so that the patent becomes “standard essential”), there is a risk that producers utilizing the standard—implementers—will invest extensively and then face unreasonable royalty demands from SEP-holders, who will know that the implementers cannot utilize a different technology without incurring exorbitant switching costs. To avert the risk of such “patent holdup,” standard-setting organizations typically procure up-front commitments from potential SEP-holders that if their technology is included in the standard, they will license it on fair, reasonable, and non-discriminatory (FRAND) terms. In recent years, the enforcement agencies have concluded that antitrust should also be used to police holdup by SEP-holders. They have sought to impose liability when SEP-holders have (1) pursued injunctive relief (rather than damages) against infringers, (2) sought to renegotiate royalty rates on their SEPs, and (3) refused to license their SEPs to rivals.

In each of these cases, the allegedly anticompetitive behavior was not conducted in secret. Moreover, each challenged behavior can be efficient: a holder of a SEP might seek injunctive relief because the infringer is judgment-proof or has rejected (or expressed the intent to reject) a FRAND royalty; a SEP-holder might legitimately renegotiate royalties in light of some market shift that undermines the original royalty rate; a SEP-holder could refuse to license to its direct rivals to prevent the sort of free-riding that diminishes incentives to innovate.

Finally, in each case, the alleged anticompetitive harm could have been addressed—with less distortion from potential treble damages actions—by private ordering or another body of law. Patent law requires holders seeking injunctions to show that their requested relief is in the public interest, something a SEP-holder engaged in hold-up could not do. Contract law polices hold-up in renegotiations, denying enforcement to contract modifications procured via duress while enforcing those that are commercially reasonable (e.g., because of a market shift) and in good faith. A SEP-holder’s obligation to license to its rivals can be—and routinely is—imposed by the FRAND commitment it makes to the standard-setting organization responsible for the technology standard. As intended third-party beneficiaries of FRAND agreements, rivals may enforce them.

4) Does the contemplated remedy require an excess of particularized knowledge or endow government officials with a great deal of discretionary authority? Just as markets may systematically fail under certain conditions (externalities, etc.), so may government interventions. First, as Friedrich Hayek observed, when the contemplated intervention requires central planners to acquire and process troves of information that is widely dispersed among economic actors, losses are likely to occur as the planners, who cannot gather and process such information, misallocate productive resources away from their highest and best ends.

Second, losses are particularly likely when interventions endow government officials with great discretion over the allocation of productive resources. As public choice scholars have demonstrated, discretionary authority invites special interest manipulation of governmental power for private ends. Rather than using their authority to maximize social welfare, government officials—who retain their rational, self-interested natures when acting in their official capacities—will frequently exercise state power in a manner that benefits them personally. Organized groups—often incumbent firms—will find ways to exploit this tendency in their favor. And because the costs of special interest manipulation are widely dispersed, individual members of the public do not have an adequate incentive to mount a response even if their losses, in the aggregate, exceed the benefits that are concentrated on the organized groups.

In light of the Hayekian knowledge problem and public choice concerns, courts and enforcers should typically avoid antitrust interventions that either require a great deal of particularized knowledge or endow government officials with a large store of discretionary authority. This general guideline calls into question a number of recent antitrust proposals.

One such proposal is to treat the user data collected by digital platforms like an “essential facility” that must be made available to rivals. A court imposing a duty to share data with rivals would have to create an elaborate price schedule that takes into account such information as the cost of collecting and organizing different sorts of data and the value each sort provides—information that is largely inaccessible and likely to change over time. Courts are ill-equipped to gather and process all that information.

The knowledge problem also bedevils recent calls to break up the largest digital platforms: Google, Facebook, and Amazon. As the American Action Forum’s Will Rinehart has observed, the leading digital platform firms utilize complex business models, teams, and technologies, which makes breaking them up difficult. With respect to business models, the firms operate multi-sided platforms where the value to users on one side (e.g., advertisers) is largely dependent on the number and intensity of users on the other side (e.g., individuals engaged in search or social networking). Moreover, the firms tend to engage in internal cross-subsidization, using revenues from one line of business (e.g., Google search) to support less profitable services (e.g., Google’s YouTube, which is widely assumed not to be profitable on its own). Given that an adverse effect on one part of the business can wreak havoc on seemingly unrelated operations, any break-up plan would have to accurately account for a highly complex set of interrelationships.

Breaking up big technology companies is also complicated by the fact that they employ teams and technologies that work across the entire enterprise. Facebook’s software engineers, for example, support Facebook, Messenger, Instagram, and WhatsApp. Its technology stack includes a number of proprietary technologies designed to assist with common tasks engaged in by all its various services: “BigPipe” serves pages faster, “Haystack” stores billions of photos efficiently, “Unicorn” searches the social graph, “TAO” stores graph information, “Peregrine” assists with querying, and “MysteryMachine” helps with performance analysis. Mistakes in disintegrating teams and technologies are likely to occasion a massive reduction in productive efficiency.

Whereas proposals to treat user data as an essential facility and to break up major digital platforms involve significant knowledge problems, other recent antitrust proposals would endow government officials with significant discretionary authority and thus raise public choice concerns. One such proposal, discussed above, is to jettison the relatively cabined consumer welfare standard in favor of a more amorphous public interest standard. Another is to create a federal agency with broad powers to regulate digital platforms. The history of sector regulation suggests that such an approach would reduce, rather than enhance, competition by entrenching incumbents and stifling innovation.

Conclusion

As courts and enforcers confront an ever-growing chorus calling for bigger and bolder antitrust, they would do well to embrace Easterbrook’s general model, revise some specifics, and supplement it with four additional filters that limit antitrust’s reach. In particular, they should restrict interventions to instances of consumer harm arising from behavior that extends market power, where no other body of law or instance of private ordering is likely to prevent the harms at issue with less distortive effect, and the remedy imposed does not entail excessive knowledge requirements or conferral of discretionary governmental authority.

Such an approach may disappoint those who imagine that antitrust can solve a host of social problems, but it alone will ensure that 21st-century antitrust succeeds at the things it does well.