A Theory of Antitrust Limits

Nicolas Petit
Volume 28
,  Issue 4


Like all laws, antitrust laws work within limits.1“Limits” mean conceptual boundaries set in legal constructs. This contrasts with Judge Frank Easterbrook who talked of limits to describe what in reality were “limitations” of antitrust decisionmakers, due to ignorance of facts about markets and the costs of action. See Frank H. Easterbrook, The Limits of Antitrust, 63 Tex. L. Rev. 1, 4 (1984) [hereinafter Limits]. The limits of antitrust laws are defined by statutory legislation, judicial practice, and prosecutorial discretion.

Today, the limits of antitrust face criticism. Loud voices argue that the current limits are too strict. They contend that antitrust must do more even though the evidence supporting their claims is disputable. Modern critics make broad claims about rising economic and political power, particularly in the digital sector. They invoke allegedly lethargic levels of enforcement activity to suggest that antitrust agencies and plaintiffs are discouraged from bringing cases by the strictures of the law.

Modern criticism of antitrust law mainly attacks three limits: (1) the consumer welfare standard, (2) the rule of reason, and (3) a self-imposed neglect of labor markets in the exercise of prosecutorial discretion. The limits would be particularly prevalent in monopolization law and, to a lesser extent, in merger law.

In calling to relax the limits that constrain antitrust power, critics give considerate attention to their foundations. These voices point to a simple reason: ideology. Antitrust limits arguably result from a body of doctrine shaped since the 1970s by opinions favorable to laissez faire, neo-liberal economics, and the Chicago School of antitrust.2Patrice Bougette, Marc Deschamps & Frédéric Marty, When Economics Met Antitrust: The Second Chicago School and the Economization of Antitrust Law, 16 Enter. & Soc’y 313, 330 (2015). Saying that the limits of antitrust are ideological subliminally conveys the message that the limits can be changed by political edict. And it is a way to insinuate that the limits are illegitimate.

Having derided the limits as arbitrary choices, critics proceed as if the sky was the limit. They envision turning antitrust law into a Swiss army knife of social welfare optimization. A broad mandate of goals should be pursued under the antitrust banner, like curbing economic inequality, surveillance capitalism, and global warming.

This Article, however, proposes that (1) ideology is a lazy answer; (2) the present limits in US and EU antitrust law rest mostly on practical considerations; and (3) the question of setting better limits is ultimately empirical and context dependent. The point is that besides ideology, practicality matters. The limits of antitrust stem from practical choices made more or less consciously by institutions tasked with applying statutes designed to curb monopoly power. An understanding of the practical limits of antitrust is necessary to avoid the creation of false expectations about what law reform can achieve, which in turn fuel distrust in majoritarian institutions.3Logjams in law reform nurture demands for authoritarianism or libertarianism. The loser is the rule of law.

Unfortunately, the issue of the foundations of antitrust limits has not been systematically treated.4Scholars like William Kovacic and others have attempted to debunk the view that US antitrust doctrine could be explained predominantly by the influence of Chicago-School ideas. See William E. Kovacic, The Intellectual DNA of Modern U.S. Competition Law for Dominant Firm Conduct: The Chicago/Harvard Double Helix, 2007 Colum. Bus. L. Rev. 1, 12, 14. But Kovacic essentially did this by demonstrating the role and influence of non-Chicago School affiliates, such as by members of the so-called “Harvard School,” on the state of modern US jurisprudence. Kovacic did not consider the possibility that Harvard School and Chicago School experts might just have been disciples of the same ideology, whose members only quibbled about the details. This Article addresses this problem by supplying a theory of the limits of antitrust law. Part I shows that the limits of antitrust are not ideological. Instead, the process of setting limits to antitrust law is an inherent development in any system of enforcement and judicial application. To establish this, a careful review of the case law of the Supreme Court in pre-Chicagoan times is undertaken. Part II then turns to the next, harder question: how are the inherent limits of antitrust law set? The thesis advanced is that the foundations of antitrust limits are diverse, discrediting again the role of ideology but also casting doubt on other mainstream explanations. The study of US and EU antitrust law suggests that multiple institutional, formal, and developmental reasons combined to produce the limits of antitrust. Last, Part III leaves the descriptive world to address a normative question: how to deal with the problem of the obsolescence of the limits of antitrust in contexts of socio-economic change?5Perhaps the issue is only one of low levels of enforcement activity. Complaining that antitrust policing activity is too low seems, however, absurd, because low levels could conceivably denote both optimal deterrence and abysmal performance. If the issue was truly about an activity problem, then increasing funding would assuage all concerns. But opponents seem to go beyond this and advocate for law reform. This Article argues that reforming the limits of antitrust law requires an empirical assessment of the costs of monopoly compared to the costs of judicial errors that goes beyond simplistic presuppositions about the efficiency of market processes and inefficiencies of the legal system. A realistic, context dependent appraisal of legal institutions, technology, and markets should guide the assessment.

I.     The Inherent Limits of Antitrust

A.     Hypothesis

How to explain the current limits of antitrust law? Voices supportive of more aggressive antitrust assert that an ideology committed to limited government intervention known as the “Chicago School” undergirds doctrinal constraints.6See Kovacic, supra note 4, at 12–13. Skepticism about the ability of government agencies and courts to bring Pareto improvements to markets have motivated a limitation of the reach of antitrust proscriptions to horizontal cartels, mergers, and monopoly.

Presenting antitrust doctrine as premised on Chicago-School ideology is the new normal in the antitrust conversation. For example, economist Fiona Scott Morton and law professor Herbert Hovenkamp write that the Chicago School’s influence on the federal courts was essentially ideological.7Herbert Hovenkamp & Fiona Scott Morton, Framing the Chicago School of Antitrust Analysis, 168 U. Pa. L. Rev. 1843, 1871 (2020). Some have attempted to nuance this account by drawing attention to the support of progressive judges to antitrust doctrine.8See James May, Antitrust in the Formative Era: Political and Economic Theory in Constitutional and Antitrust Analysis, 1880–1918, 50 Ohio St. L.J. 257, 343–45 (1989). But the critique of ideological bias stands strong.9See Kovacic, supra note 4, at 27; see also William E. Kovacic, The Chicago Obsession in the Interpretation of US Antitrust History, 87 U. Chi. L. Rev. 459, 464 (2020) (“The greatest omission in Chicago School-centric interpretations of US antitrust policy is the failure to recognize the significant influence of the modern Harvard School and its formative scholars . . . .”). The obvious counter argument is that we might just witness Chicagoan fifth columnism at play. The fact that “Harvard-School” Justice Stephen Breyer has written as many defendant-friendly or small antitrust opinions as “card-carrying” Chicagoans could be interpreted as a sign of successful proselytism by Chicagoan ideology.10Timothy J. Muris & Jonathan E. Nuechterlein, Chicago and Its Discontents, 87 U. Chi. L. Rev. 495, 515 (2020). However, Justice Breyer’s pro-defendant position in many cases might equally be read as traditional application of Harvard-School economics, which insists on serious evidence and consideration of the administrability costs of intervention. Similarly, the fact that current doctrinal strictures also reflect neo-Harvard-School institutional insights does little to undermine the critique that error-cost analysis has been the Chicagoan worm in the fruit that drove subsequent antitrust scholarship and doctrine astray.11Edward D. Cavanagh, A 2020 Agenda for Re-Invigorated Antitrust Enforcement: Four Big Ideas, 105 Cornell L. Rev. Online 31, 61–64 (2020).

As leading antitrust voices (lawyers and economists) now align with the criticisms of current antitrust doctrine, a deeper examination of the Chicagoan ideological attack is warranted.12Daniel Crane’s predicament might become sooner than he anticipated. See Daniel A. Crane, A Premature Postmortem on the Chicago School of Antitrust, 93 Bus. Hist. Rev. 759, 775–76 (2019) (“[T]he prospect of landmark legislative reforms seems low. The populist pressures are almost all external to the antitrust establishment, which, from right to left, has largely circled the wagons around consumer welfare. Without some significant faction of antitrust professionals (lawyers and economists) leading the charge in a Brandeisian direction, it seems unlikely that the inherently conservative courts will abandon the consumer welfare model and economic reasoning in the foreseeable future.”). There are multiple reasons to be skeptical of the narrative that Chicagoan ideology is the DNA of contemporary antitrust limits. Not least is that a similar agenda of criticism of antitrust law’s limits has arisen in the EU, where free market ideology has never been in high currency.

But how to verify the proposition that reasons distinct from Chicagoan ideology explain the limits of antitrust? A logical approach to the problem is to formulate a null hypothesis and assess its validity. One way to go about this is to state the null hypothesis that the contemporary limits of antitrust existed before the Chicago School. If the null hypothesis is true, we can relativize the thesis of a Chicagoan influence on antitrust doctrine.13True, the limits can be shared with another ideology, but then this means that a particular ideology is not determinant, so that the Chicagoan ideological critique is defective. A simple test of the null hypothesis should therefore ask if one can find Chicagoan limits in the pre-Chicagoan antitrust jurisprudence.

A conservative method to implement the test is then twofold. The first step is to collect Supreme Court decisions that followed the adoption of the Sherman Act of 1890 up to and including the first majority opinions of Justice Louis Brandeis. As this period precedes the Chicago School, we would not expect to find Chicagoan influence. The second step is to search for the following markers of Chicagoan ideology: (1) the consumer welfare standard; (2) the rule of reason; (3) a neglect of labor monopolization; and (4) a bias against state regulation.

The presence of Chicagoan markers in the pre-Chicagoan ideological era would then provide tentative support for the null hypothesis. This would imply absence of a significant relationship between ideology and the limits of antitrust. By contrast, absence of Chicagoan markers in this era would provide tentative support against the null hypothesis. This would alternatively support the view that Chicagoan ideological foundations underpin antitrust limits. Of course, the test proposed here is not a strict one. A qualitative analysis of the case law does not allow us to make a claim to evidence, just to derive rough insights.

Let us consider each marker separately. The consumer welfare standard stands as the mother of all evils for antitrust critics.14See Barak Orbach, How Antitrust Lost Its Goal, 81 Fordham L. Rev. 2253, 2275–77 (2013). In Reiter v. Sonotone Corp., the Supreme Court declared the Sherman Act to be a “consumer welfare prescription.” 442 U.S. 330, 343 (1979) (internal quotation marks omitted). The consumer welfare standard conditions antitrust liability to a showing of detriment to consumers. A strict version of the standard requires an “actual” or “short-term” effect on price charged or output served to buyers in the relevant market.15Kevin Caves & Hal Singer, When the Econometrician Shrugged: Identifying and Plugging Gaps in the Consumer-Welfare Standard, 26 Geo. Mason L. Rev. 395, 397 (2018). A soft version of the standard does not make consumer welfare a test of legality in the particular case, but rather a “guide” in the formulation of legal rules of liability and procedure.16Gregory J. Werden, Consumer Welfare and Competition Policy, in Competition Policy and the Economic Approach: Foundations and Limitations 11, 29 (Josef Drexl, Wolfgang Kerber & Rupprecht Podszun eds., 2011); see also Orbach, supra note 14, at 2273–74 (exploring how consumer welfare became the goal of antitrust policy). And an even looser version relies on the consumer welfare standard as a priority setting device.17See Werden, supra note 16, at 15–17. The standard channels prosecutorial discretion towards the protection of specific types of consumers, like those at the end of the distribution chain who cannot pass an increase in prices to downstream sellers.

To add even more complexity, Judge Robert Bork, who popularized Chicago-School economics, talked about consumer welfare in terms of total welfare, that is the overall welfare of both producers and consumers.18Robert H. Bork, The Antitrust Paradox: A Policy at War with Itself 90–91 (1978). Though it is not clear which version of the consumer welfare standard today’s critics attack, a practical fixation on prices by courts and enforcement agencies appears to be their common concern. Attention to price or output effects narrows antitrust decisionmakers’ line of sight, leading them to neglect hard to measure, long-term, and broad harms from competitor exclusion like loss of quality competition, slowing of innovation, and weakening of entrepreneurial dynamism.19See Tim Wu, After Consumer Welfare, Now What? The “Protection of Competition” Standard in Practice, Antitrust Chron., April 2018, 1, 5–6 (2018).

Conduct of antitrust law under the rule of reason is another emblem of Chicagoan ideology that critics oppose.20The rule of reason is the dominant mode of analysis of restraints of competition in US antitrust law. Lina Khan & Sandeep Vaheesan, Market Power and Inequality: The Antitrust Counterrevolution and Its Discontents, 11 Harv. L. & Pol’y Rev. 235, 274 (2017). To be clear, the critics do not dispute the use of the rule of reason as a device per se. There is an acceptance that not all business behavior described literally in the Sherman Act should be considered inherently anticompetitive and that a filter is needed to sort reasonable and unreasonable conduct. But they oppose the Chicagoan re-orientation of the content of the rule of reason.21See id. at 275–76. For the Chicago School, antitrust decisionmakers must determine the reasonableness of business conduct by reference solely to its economic impact on competition. For its critics, the alleged economic rewrite of the rule of reason by Chicagoan ideology makes antitrust a narrow, irrelevant policy unable to address the wealth of socio-political harms arising from business behavior.

Moreover, the focus on economics sophisticates antitrust cases to an extent that confounds lay juries and generalist courts.22See Maurice E. Stucke, Does the Rule of Reason Violate the Rule of Law?, 42 U.C. Davis L. Rev. 1375, 1461–63 (2009). Under the rule of reason, an assessment of the net economic effects of conduct is required. Admittedly, no hard “balancing” of economic harms and benefits is ever carried out in real cases. The rule of reason instead requires a balancing of the strength and weight of evidence of competitive effects.23Andrew I. Gavil, Burden of Proof in U.S. Antitrust Law, in 1 Issues in Competition Law and Policy 147 (Wayne D. Collins ed., 2008). This indirect balancing, however, invites abstract and theoretical claims to the fact-finding inquiry.24Khan & Vaheesan, supra note 20, at 272 (criticizing the “open-ended, fact-intensive” rule of reason). Note that the Chicagoan rule of reason is certainly fact-intensive, but it appears much less open-ended than the pre-Chicagoan one. See Jesse W. Markham Jr., Sailing a Sea of Doubt: A Critique of the Rule of Reason in U.S. Antitrust Law, 17 Fordham J. Corp. & Fin. L. 591, 613 (2012). Moreover, given that an economic proof is required, a showing that competitive injury is caused by the impugned conduct is necessary.25Caves & Singer, supra note 15, at 407. The standard method to prove causation consists of running a counterfactual assessment to see if the competitive injury would arise but for the impugned conduct. Understandably, the level of economic sophistication required under the rule of reason renders antitrust cases costly.26In the past, the heavy costs of antitrust under the rule of reason has led prominent affiliates of the Chicago School like Judge Easterbrook to disfavor the use of the rule of reason in antitrust proceedings and default to per se rules of legality.

The third ideological marker of Chicagoan ideology is a neglect of labor monopolization. The limit is more implied than explicit. Economists Kevin Caves and Hal Singer write “[w]hen it comes to under enforcement of monopsony harms, the problem is one of agency discretion, as opposed to black letter law.”27Caves & Singer, supra note 15, at 415. Chicago-School economics posit a symmetry between harms to sellers in product markets and harms to suppliers in factor markets. If quantities produced go down by exercise of monopoly power, inputs consumed in the manufacturing process should decrease and vice versa. Antitrust can thus limit itself to looking down. If antitrust is good at remedying price increases caused by product market power, it will limit wage decreases caused by labor market power.

A last ideological marker of Chicagoan ideology is a view of the state as primary source of distortions of competition. As Professor Daniel Crane puts it, Chicagoan ideology abhors anticompetitive state or local regulations.28Crane, supra note 12, at 772. Contrary to the three other markers, the bias against the state implies more expansion than limitation of antitrust law.29Timothy J. Muris, Clarifying the State Action and Noerr Exemptions, 27 Harv. J.L. & Pub. Pol’y 443, 444 (2004). That being said, the bias can be envisioned as a limit to the extent that the channeling of resources towards state-induced distortions of competition reduces antitrust activity against anticompetitive business conduct.

This Article’s null hypothesis has limitations. The presence of antitrust limits, which is now associated with the Chicago School in earlier case law, cannot prove that Supreme Court doctrine was ideology-free. But it can prove this: that doctrine was not created by a “system” of beliefs or ideas about markets that appeared only decades later.30George Stigler, a prominent economist associated with the Chicago School, showed that economists had really little theory to bring to the analysis of monopoly and consumer welfare in the formative era. See George J. Stigler, The Economists and the Problem of Monopoly, 72 Am. Econ. Rev. 1, 7–8 (1982). We are interested in seeing if antitrust limits arose in times free from suspicion of pro-Chicagoean ideology. A second difficulty is that we should not expect to find literal evidence of Chicagoan markers in the case law. Our test is thus more semantical than lexical. We look for impressionistic strokes. Subjectivity is inevitable. With these caveats in mind, let us look at the footprint of Chicagoan antitrust limits in the pre-ideological era.

B.     Antitrust Limits in Pre-Chicagoan Times

1.     Introduction

A theory of the inherence of antitrust limits must hold true over time and space. This Section addresses the time dimension, leaving for later the question of antitrust limits across jurisdictions. Were there antitrust limits in pre-Chicagoan times, and how close were they to the Chicagoan’s substantive precepts? The answer to these questions lurks in eighty-five opinions from the Supreme Court rendered between 1891 and 1921.31The list closes with the last case under Chief Justice White. This period is helpful in the sense that it makes the list of cases conservative. Chief Justice Taft was a Republican probably closer in mindset to what would later be the Chicago Schools views than his predecessor, Justice White. See Daniel A. Crane, All I Really Need to Know About Antitrust I Learned in 1912, 100 Iowa L. Rev. 2025, 2031–33 (2015). The list was retrieved from the Supreme Court Database, using both the 2020 release of the modern database (SCDB_2020_01, covering the terms 1946–2019) and the 2019 release of the legacy database (SCDB_Legacy_06, covering the terms 1791–1945). Those cases where the “issue” variable was recorded as “antitrust (except in the context of mergers and union antitrust)” (80010), “mergers” (80020), or “union antitrust: legality of anticompetitive union activity” (70020), were selected. Antitrust scholars call this period the “formative” era to describe the process of judicial construction that followed the enactment of the Sherman Act in 1890.32The formative era is often opposed to the period of judicial “transformation” that took place following the imports of Chicago-School ideas in antitrust doctrine. See William E. Kovacic, The Antitrust Paradox Revisited: Robert Bork and the Transformation of Modern Antitrust Policy, 36 Wayne L. Rev. 1413, 1428, 1462 (1990). The formative era is particularly apt to study our null hypothesis. If the Court had any unconscious bias, it must have been one exactly antagonist to the small government ideology of the Chicago School. In 1915, the Supreme Court declared in D. R. Wilder Manufacturing Co. v. Corn Products Refining Co.33236 U.S. 165 (1915). that the Sherman Act was intended “in the most comprehensive way” to prohibit collusion and monopolization, was founded on “broad conceptions of public policy,” and conferred “amplest discretion in [the] courts.”34Id. at 173–74.

The cases of the formative era featured lots of disputes brought by private plaintiffs, but few by the government. The most common practices complained about concerned: (1) coalition building strategies; (2) sell, combine, or be ruined threats; (3) discriminatory practices; and (4) conflicts of interests created by situations of vertical integration often between railroads and input industries like coal.35See Herbert Hovenkamp, Antitrust Policy, Federalism, and the Theory of the Firm: An Historical Perspective, 59 Antitrust L.J. 75, 82 (1990). These cases featured mostly structural offenses of unlawful combination or monopolization under both sections 1 and 2 of the Sherman Act and were more rarely about the terms of trade that “consummate” the offenses.36See United States v. United Shoe Mach. Co., 247 U.S. 32, 38 (1918).

The picture that emerges from the study of the cases is that some antitrust limits close to Chicagoan ideas existed in rough form. The analogies are imperfect and incomplete. But the idea of evaluating antitrust liability by application of a welfare compass and on the basis of economic arguments was unmistakably there before.

2.     Consumer Welfare

A lawyer trained in reading the bible like the devil would confidently detect consumer welfare wording in the first formative era cases. In 1903, the Court in Northern Securities Co. v. United States37193 U.S. 197 (1904). affirmed the dissolution of a railroad trust spearheaded by JP Morgan on the ground that “the public convenience and the general welfare will be best subserved when the natural laws of competition are left undisturbed.”38Id. at 337–38. The Court adds that “powerful corporations[’s] or combinations[’s]” control over trade and commerce “would be detrimental to the general welfare.” Id. at 339. However, Justice Harlan’s opinion said nothing about the welfare of a specific category of economic agents.39Cf. FTC v. Gratz, 253 U.S. 421, 432–33 (1920) (Brandeis, J., dissenting). Justice Brandeis discussed the objectives of both the Federal Trade Commission Act and of the Sherman Act: The proceeding is thus a novelty. It is a new device in administrative machinery, introduced by Congress in the year 1914, in the hope thereby of remedying conditions in business which a great majority of the American people regarded as menacing the general welfare, and which for more than a generation they had vainly attempted to remedy by the ordinary processes of law. Id. at 432. And it is unclear if the Court had in mind welfare in the economic sense that Robert Bork thought about it when he talked of the “wealth of the nation,” or whether the Court envisioned the social, economic, and political wellbeing of the American people.40See generally Bork, supra note 18.

In 1912, a clearer allusion to a consumer welfare goal appeared under the pen of Justice Oliver Wendell Holmes in Central Lumber Co. v. South Dakota.41226 U.S. 157 (1912). Writing for the majority, Justice Holmes contended that a statute like the Sherman Act aimed “at preventing the creation of a monopoly by means likely to be employed” should be read “as having in view ultimately the benefit of buyers of the goods.”42Id. at 161. Yet, the opinion was devoid of the welfare language of Northern Securities. In 1919, a “general welfare” motivation was again given to the Sherman Act by Justice Brandeis in his dissent in FTC v. Gratz.43253 U.S. 421, 432 (1920) (Brandeis, J., dissenting). But Justice Brandeis’s dissent did not really remove the vagueness found in Northern Securities. Perhaps, the clearest expression of consumer welfare appeared in the 1911 Dr. Miles Medical Co. v. John D. Park & Sons Co.44220 U.S. 373 (1911). opinion, where the Court mentioned in passing “enhanced price[s] to . . . consumer[s].”45Id. at 408.

Short of any clear literal footprint, can we track the spirit of the consumer welfare standard in the case law? The essence of consumer welfare is the principle that the application of antitrust law depends on proof of monopoly power wrongly acquired, protected, or extended. Liability has two elements: bad conduct and the presence of monopoly power. The element of monopoly power is the one that allows an inference of lost consumer welfare. A rule of thumb to detect a situation of consumer welfare detriment is therefore this: a reduction in the total output served to consumers or increased prices to consumers. In modern case law, consumer welfare has been further codified as a hard evidentiary rule. Lost consumer welfare can be proved directly by production of evidence of price increases, or indirectly by showing conditions of monopoly control of output in a relevant market.46See Gavil, supra note 23, at 125, 147–48. In the old case law, we might therefore find signs of consumer welfare by searching for the rule of thumb, not the hard evidentiary rule.

The consumer welfare standard is visible in the big United States v. United States Steel Corp.47251 U.S. 417 (1920). case of 1920. The case involved the “super-combination” of twelve independent iron and steel companies into the U.S. Steel Corporation.48Id. at 420, 437. The claim of the government was that the U.S. Steel Corporation had been formed for the purpose and effect of monopolizing the steel trade in violation of section 1 of the Sherman Act.49Id. at 419. The Supreme Court opinion passed to posterity as a brutal demolition of the government’s case.50See id. at 457. What is interesting for our inquiry, though, is the economic dicta. In what would suggest today’s strong consumer welfare standard, the U.S. Steel Court defined for the first time the problem of monopoly as one of “power over prices,” and proceeded to condition antitrust liability on proof of it.51Id. at 445–46. While recognizing the gigantic size of U.S. Steel, the Court added that it did not matter because U.S. Steel’s “power over prices was not and is not commensurate with its power to produce.”52Id. at 445. The Court’s treatment of the evidence was also highly informed by the consumer welfare standard. Justice McKenna’s opinion spent substantial time lamenting the absence of actual price injury suffered by customers in the record.53U.S. Steel, 251 U.S. at 446, 448. The Court observed that “the only attempt at a fixation of price was . . . through an appeal to and confederation with competitors.” Id. at 446. The Court also gave substantial credit to the testimony of “200 witnesses out of some forty thousand customers” who had no complaints about steel prices. Id. at 448. Unconvinced by the evidence, the majority faulted the government for tilting at windmills. The case, said the opinion, was based on “alleged power for evil, not the exertion of the power in evil.”54Id. at 447, 451 (“[T]he law does not make mere size an offence or the existence of unexerted power an offence.”). The case also contained language suggestive of a concern with “detriment[s]” to “welfare,” like in Northern Securities and Justice Brandeis’s dissent in Gratz.55Id. at 446. The Court identified President Roosevelts approval of the combination as a hint that it could not possibly bring about a detriment to welfare: “[Roosevelt’s] approval, of course, did not make it legal, but it gives assurance of its legality, and we know from his earnestness in the public welfare he would have approved of nothing that had even a tendency to its detriment.” Id.

Another manifestation of the consumer welfare standard appeared the same year in United States v. Reading Co.56253 U.S. 26 (1920). The case had to do with the creation of the Reading Company, a holding combining two large railroad and coal producing companies, designed to acquire stock in competing coal suppliers and carriers. In a statement that hints at a consumer welfare standard, the Court noted that:

[The holding’s] board of directors, obviously, thus acquired power: to increase or decrease the output of coal from very extensive mines, the supply of it in the market, and the cost of it to the consumer; to increase or lower the charge for transporting such coal to market; and to regulate car supply and other shipping conveniences, and thereby to help or hinder the operations of independent miners and shippers of coal.57Id. at 48.

The case, however, displayed substantial differences with the U.S. Steel case. The Reading Court affirmed antitrust liability and ordered dissolution under sections 1 and 2 of the Sherman Act without requiring proof that monopoly power had been exerted.58Id. at 63. Professor Donald Turner later observed that the case was “strong doctrine” because the holding company’s percentage of the anthracite-coal market was well under 50%.59Donald F. Turner, Antitrust Policy and the Cellophane Case, 70 Harv. L. Rev. 281, 289 n.22 (1956). However, the Court may just have established liability on the grounds of incipiency or on account of known practices of anthracite coal price gouging.60Myron W. Watkins, Economic and Legal Concepts in Antitrust Adjudication, 9 Antitrust Bull. 347, 361 n.30 (1964) (“The prices of anthracite coal were reputedly and allegedly, exorbitantly high, though it does not appear that the prosecution proved this circumstance factually. . . . It may be assumed, perhaps, that the Court took judicial cognizance of the notorious gouging of the ‘anthracite monopoly.’”). Under this interpretation, a concern of consumer welfare grounded the test of legality of monopoly conduct. The passage above might even be read as a further elaboration of U.S. Steel, insofar as it stressed that it is not any power over price that denotes monopoly power, but power over price stemming from power over output.

Admittedly, it would be a wild claim to argue that U.S. Steel and Reading endorsed the consumer welfare standard. The decisions remain, however, relevant to a discussion of consumer welfare to the extent that they show the Court’s sympathy for the idea that the test of legality does not inhere only in size, but in the maintenance, and possibly, exertion of monopoly power. Under that half-formulated doctrine, diagnosis of actual or potential power over price, stemming from control of industry output, appeared to be the idea behind the test of legality.61We write half formulated, because no strict requirement of harm to trading partners was required. Under the case, antitrust law prohibits conduct that harms rivalry. The “consumer welfare” issue, that is whether trading partners are exposed to harm in the form of increased quality adjusted prices or reduced innovation, is at best left to conjecture.

An almost literal endorsement of consumer welfare eventually appeared in the 1918 opinion of Justice Brandeis in Board of Trade v. United States (Chicago Board of Trade).62246 U.S. 231 (1918). To support exoneration of a rule restricting the right of grain traders to do business after specific hours, the Court noted that the rule “made it possible for them to pay more to farmers without raising the price to consumers.”63Id. at 240 (emphasis added). Clearly, this case law was not too far from the modern consumer welfare framework. The Court here did not accept to unconditionally clear a restriction of competition with an income or wage increasing effect. The opinion only accepted to credit trading partner welfare efficiencies as long as they did not result in a consumer welfare loss. 64See id. at 239–41. Justice Brandeis here looks surprisingly consumer welfarist. A careful read of the opinion throws doubt on modern critics’ descriptions of Brandeis as the champion of small business interests and trading partner welfare in antitrust doctrine.

3.     Rule of Reason

Why spend time testing for the presence of the rule of reason in the formative era? It is well known that the Court introduced the rule of reason in Standard Oil Co. v. United States,65221 U.S. 1 (1910). grounded it in competition concerns in United States v. American Tobacco Co.,66221 U.S. 106 (1911). and fleshed it out more fully in Chicago Board of Trade.67See Address by R. Hewitt Pate, Assistant Atty Gen., U.S. Dept of Just., Antitrust Div., Antitrust Law in the U.S. Supreme Court (May 11, 2004), https://perma.cc/3AMB-56FD. Nonetheless, the inquiry proposed here offers a distinct contribution. The question is whether analogs of the Chicago-School version of the rule of reason appeared in the formative era. Recall that the rule of reason generally means that restraints of competition are not automatically condemned. Their (wrongful) purpose or (harmful) effect needs to be tested. Judge Frank Easterbrook called the rule of reason a “method of study.”68Limits, supra note 1, at 10.

A first task requires distilling the specific features, or “main themes” of the Chicagoan rule of reason.69Robert H. Bork, The Rule of Reason and the Per Se Concept: Price Fixing and Market Division, 74 Yale L.J. 775, 781 (1965). The Chicago School advocated use of a particular standard of reasonableness under the rule of reason.70See id. at 785–88. The test of legality is solely one of anticompetitive effects. Form, nature, purpose, or intent are irrelevant.71Easterbrook wrote in A.A. Poultry Farms, Inc. v. Rose Acre Farms, Inc., 881 F.2d 1396 (7th Cir. 1989), that “[i]ntent does not help to separate competition from attempted monopolization.” Id. at 1402. Moreover, not every effect is relevant. The quantitative significance of the restraint matters. Last, improvements in economic efficiency can redeem any anticompetitive effects and salvage even the worst restraints on rivalry, like monopoly, from liability. Economic efficiency must, however, be narrowly understood as quality adjusted reduction in prices or costs. This excludes improvements in business morals, the provision of social benefits, redistribution, and public goods. Innovation—a form of public good—occupies an ambiguous place in Chicagoan literature.72See, e.g., Bork, supra note 69, at 831–32.

So much for the reason prong, what about the actual method embodied in the rule of reason? The rule dictates a concrete determination, on a case-by-case basis, upon a market analysis, and on the basis of the totality of the evidence that competition is restrained. The high costs of economic inquiry for lay juries and generalist courts are what led Easterbrook, again, to argue against widespread use of the rule of reason in antitrust litigation, and in favor of simplifying antitrust application by recourse to filters.73See A.A. Poultry, 881 F.2d at 1401–02.

Against this backdrop, what can be seen in the cases? Many opinions of the formative era had started to utter rudimentary versions of the components of the Chicagoan rule of reason. For example, it did not take long for the Court to replace the rough binary inquiry between restraint and non-restraint inherited from United States v. Trans-Missouri Freight Ass’n74166 U.S. 290 (1897). and United States v. Joint Traffic Ass’n75171 U.S. 505 (1898). with a test of “direct and immediate effect” in Hopkins v. United States.76171 U.S. 578, 592 (1898). Bork added to the argument in the preceding case Anderson v. United States., 171 U.S. 604 (1898), which rejected Sherman Act liability for agreements which only “indirectly” and “remotely” affect commerce. Bork, supra note 69, at 795. The analogy with Chicagoan rule of reason, however, stops there. Anticompetitive purpose or absence thereof, not effects, played a key role in the Hopkins Court’s exoneration of the bylaws of the Kansas Livestock Exchange.77Although, the case was about clear restrictions of the freedom to trade of sellers who had been limited in their ability to send interstate telegrams. More to the point is the Chicago Board of Trade opinion. Writing for the majority, Justice Brandeis remarked that the “call rule” in dispute, which limited the right to trade after close of business, “had no appreciable effect on general market prices.”78Bd. of Trade v. United States (Chicago Bd. of Trade), 246 U.S. 231, 240 (1918) (emphasis added). Effects and appreciability, two markers of the Chicagoan rule of reason, were already in clear display.

A threshold inquiry into the quantitative significance of the restraint also appeared in Straus & Straus v. American Publishers’ Ass’n,79231 U.S. 222 (1913). which held that copyright does not shield anticompetitive agreements from section 1 liability.80Id. at 234–35. The case United States v. Union Pac. R.R. Co., 226 U.S. 61 (1912), also led the court to verify on the facts that the amount of suppressed competitive traffic was substantial, id. at 88–89 (noting that “such competing traffic was large in volume, amounting to many millions of dollars,” and as a result “a large and valuable part, of interstate commerce . . . was thus directly affected”). The case concerned the strict resale price maintenance (“RPM”) policy enforced over booksellers by the American Publishers’ Association and the American Booksellers’ Association.81Straus, 231 U.S. at 229. Both associations had refused to sell books to booksellers who would not maintain the retail price for copyrighted books.82Id. In support of its finding of antitrust liability, the Court insisted on the fact that the Publishers’ Association “was composed of probably seventy-five per cent of the publishers of copyrighted and uncopyrighted books in the United States.”83Id. at 235. The opinion is remarkable because in a previous decision in Dr. Miles, the Court had considered RPM as an “obvious” restraint of trade without making liability strictly conditional on proof of substantial market coverage.84Dr. Miles Med. Co. v. John D. Park & Sons Co., 220 U.S. 373, 400 (1911). The Court confirmed that the agreements had been made with most jobbers and druggists, but the language used suggests this was an additional observation, not strictly necessary for its holding of liability. See id.

A primitive version of Chicagoan efficiency again appeared in Justice Holmes’s opinion in United States v. Winslow.85227 U.S. 202 (1913). The case involved the merger of three dominant producers of complementary machines essential in the manufacture of shoes.86Id. at 216–17. The Supreme Court flatly rejected the government charges of unlawful combination under section 1.87Id. at 217. The opinion’s reasoning rested mainly, if not exclusively, on efficiency.88Id. Justice Holmes wrote “the combination was simply an effort after greater efficiency.”89Id. Low on facts, and essentially based on logic—the market power of three corporations each with a dominant position was the same as one corporation with three dominant positions—the opinion did not enter into a case-specific evaluation of the merger’s efficiencies. But compare Justice Holmes’ statement that “[t]he disintegration aimed at by the statute does not extend to reducing all manufacture to isolated units of the lowest degree,” with Judge Richard Posner’s statement whereby “the aesthetic delights of smallness and the yearning to resurrect a nation of sturdy Jeffersonian yeomen will not be permitted to decide antitrust cases.”90Compare id., with Richard A. Posner, The Rule of Reason and the Economic Approach: Reflections on the Sylvania Decision, 45 U. Chi. L. Rev. 1, 13 (1977). Winslow clearly prefigured Chicagoans’ efficiency views towards conglomerate integration.

An even more spectacular endorsement of efficiency can be found in the first United States v. United Shoe Machinery Co.91247 U.S. 32 (1918). case of 1918. The case concerned the formation of United Shoe Machinery Company through merger and acquisition and through tying clauses in lease contracts.92Id. at 33–34. Acknowledging the magnitude of the combination, the majority deemed it to be “the result and cause of efficiency.”93Id. at 56. Discussing the dissolution sought by the government, the Court noted that “the idea is repellent that so complete an instrumentality should be dismantled and its concentration and efficiency lost,” and “the charge that it has been oppressively used is not sustained.”94Id. at 47, 56. And when considering whether the leases were unlawful restraints of competition, the Court spoke full Chicagoan vernacular. The Court observed that the leases had been “the practice of the constituent companies before their union and . . . were substantially the same after.”95Id. at 63. This allowed the Court to draw an inference that market power was not the root cause of the leases, but were instead a sign of efficiency. Years later, Chicagoans like Easterbrook would argue that the survival of long-lived, durable business practices is an indication that they serve some beneficial goal that is not anticompetitive.96See Frank H. Easterbrook, Workable Antitrust Policy, 84 Mich. L. Rev. 1696, 1701 (1986) [hereinafter Workable Antitrust Policy].

Some cases like United States v. Reading Co.97226 U.S. 324 (1912). show a commitment to a full-blown market inquiry under a Chicagoan “totality of the evidence” standard.98In other cases, the Court stressed the need for a careful analysis of the facts, but its review did not entail a full market inquiry. See Cincinnati, Portsmouth, Big Sandy & Pomeroy Packet Co. v. Bay, 200 U.S. 179, 184 (1906) (“A contract is not to be assumed to contemplate unlawful results unless a fair construction requires it upon the established facts.”); see also Contl Wall Paper Co. v. Louis Voight & Sons Co., 212 U.S. 227, 266 (1909) (“[T]he court will not be restricted to a partial statement of the facts but will consider all the circumstances connected with the transaction so as to ascertain its real nature.”). The case concerned the steps that a large combination of integrated railroads and coal companies had taken to suppress competition from independent coal suppliers.99Reading, 226 U.S. at 337. The combination was said to have engaged in market sharing, prevented the construction of an independent competing railroad, and entered into exclusionary long-term contractual commitments.100Id. at 343–44. The zeal with which the Court reviewed the economic evidence would delight an orthodox Chicagoan (though perhaps not the conclusion). The case contains a description of the economics of coal mining and transportation, a review of market share and price data, and a discussion of the incentive structures of both the defendants and their suppliers.101See id. at 338–40. What is more, the Court adduced evidence of the loss of “actual competition” so dear to Chicagoans, who warn against antitrust cases built on Nirvana fallacies of lost potential competition.102See id. at 369. The Court noted: “Before these contracts there existed not only the power to compete, but actual competition . . . . Such competition was after the contracts impracticable.” Id. Last, even if the finding of liability in the case appeared to rest to a large degree on anticompetitive intent, the Court expressly confined consideration of intent evidence to a subsidiary role in cases where there is only a mere probability of effects.103In a statement of principle, the Court held: “Of course, if the necessary result is materially to restrain trade between the States, the intent with which the thing was done is of no consequence. But when there is only a probability, the intent to produce the consequences may become important.” Id. at 370. For the Reading Court, effects prevailed over intent.

4.     Monopsony

Detecting Chicagoan ideology in the treatment of labor monopsony issues in formative era jurisprudence is harder. A test might consist in verifying whether the Court disregarded harms toward labor or, more generally, harms toward the class of economic agents to which labor belongs (i.e., sellers).

The case law contains striking examples of what modern critics would call antisocial antitrust. The 1908 decision in Loewe v. Lawlor104208 U.S. 274 (1908). is a case in point. Members of the American Federation of Labor involved in the hat business had organized boycotts to force manufacturers of fur hats to unionize their businesses.105Id. at 288 n.15. Unwilling to create an exception in the application of the Sherman Act to anticompetitive conduct by workers, the Supreme Court held that the “act made no distinction between classes,”106Id. at 301 (emphasis added). and that the Sherman Act covered “combinations of labor as well as of capital; in fact, all combinations in restraint of commerce.”107Id. at 302. Does this come close to Chicagoan ideology? The answer is not that simple. If Chicagoan ideology means a bias for businesses and against workers, the answer might be yes.108Some have observed that the case became “a potent weapon in the hands of employers seeking to prevent unionization.” Marshall Steinbaum, Antitrust, the Gig Economy, and Labor Market Power, 82 L. & Contemp. Probs., no. 3, 2019, 45, 57. This sidelines the fact that antitrust remains a potent weapon in the hands of consumers to prevent capital cartelization. But if Chicagoan ideology means a bias for small antitrust, bounded by inherent limits, the answer is a sure no. Loewe is not a case limiting antitrust, just the contrary.

A better question to ask is whether monopoly dominates monopsony in the cases. There is no discussion that cases of supply-side restraints were far more numerous than cases of demand-side restraints during the formative era. But liability decisions related to monopsony harms in input markets were far from absent. In Swift & Co. v. United States,109196 U.S. 375 (1905). the first case involving an attempt to monopolize under the Sherman Act,110For a thorough and rare discussion, see generally David Gordon, Swift & Co. v. United States: The Beef Trust and the Stream of Commerce Doctrine, 28 Am. J. Legal Hist. 244 (1984). the Supreme Court affirmed liability against a conspiracy of buyers of fresh meat.111Swift, 196 U.S. at 402. It remarked that by refraining from bidding against each other for fresh meat, the buyers had been “compelling the owners of such stock to sell at less prices than they would receive if the bidding really was competitive.”112Id. at 391. In American Tobacco Co., the facts that formed the basis of the conspiracy charge was a decision from five tobacco factories to combine into a trust.113United States v. Am. Tobacco Co., 221 U.S. 106, 143 (1911). We might wonder why, but the Court found it necessary at the outset to stress that there was “no doubt that these factories were competitors in the purchase of the raw product which they manufactured,” and that there had been “fierce and abnormal” competition between them.114Id. at 157. One conjecture is that the factories were located in different localities, so that they were not strictly competitors on the selling side. And in Eastern States Retail Lumber Dealers’ Ass’n v. United States,115234 U.S. 600 (1914). the Court declared unlawful a collusive system of reporting that retail lumber dealers had introduced to dissuade wholesalers from selling directly to consumers.116Id. at 611–12, 614. Buyer power played no apparent role in the decision, yet the combination appeared to have the necessary effect of forcing wholesalers to lower prices to retailers.

Occasionally, the Court rejected allegations of antitrust violations in relation to demand-side restraints and selling-side injury. In Anderson v. United States,117171 U.S. 604 (1898). the Supreme Court did not find the evidence sufficient to support a holding of unlawful combination among buyers of cattle.118Id. at 617. But the Court saw the nature of the monopsony power problem eye-to-eye when it said, “there is not the slightest evidence that the market prices of cattle have been lowered by reason of [the combination’s] existence.”119Id. at 614.

The United States v. Patten120226 U.S. 525 (1913). case of 1913 made explicit that no bias towards the demand- or supply-sides of the market characterized the formative era.121See id. at 542. The case had to do with the practice of “cornering the market.”122Id. at 538–39. Traders that corner a market purchase all present and future quantities of a commodity at artificially inflated prices and hoard them to charge monopoly rents on the selling side. Cornering the market harms buyers. But what is important is that the Court held that the direction of harms was irrelevant to an antitrust prohibition of cornering. The Court said that the antitrust concept of “restraints” covered anything that “operates to thwart the usual operation of the laws of supply and demand”123Id. at 542. and that any conspiracy “having as its object the arbitrary increase or depression of prices is in restraint of trade,” regardless of the side or sign of the price effect.124Id. at 528.

There is no evidence that formative era jurisprudence established limits to the application of antitrust with respect to monopsony harms. Our best conjecture is that the limited number of antitrust cases against demand-side restraints on input markets is a byproduct of the antitrust statute itself. Focused on contracts in restraint of “commerce,” the language of the Sherman Act did not invite complaints toward factors of production, like labor, used as inputs in the manufacturing process.125In the famous case of United States v. E. C. Knight Co., 156 U.S. 1 (1895), which was quickly overruled, the Court had initially excluded from the Sherman Act “[c]ontracts, combinations, or conspiracies to control domestic enterprise in manufacture, agriculture, mining, production in all its forms, or to raise or lower prices or wages” as restraints with indirect results, see id. at 16. Similarly, in Blumenstock Brothers Advertising Agency v. Curtis Publishing Co., 252 U.S. 436 (1920), the Supreme Court went as far as to suggest that advertisement agency contracts were not commerce because they were an “incident” to commercial intercourse, see id. at 443.

5.     State Distortions

To close this review of antitrust in the formative era, we ask if a negative disposition toward state regulation characterized the jurisprudence. The evidence from the cases is necessarily indirect. The Sherman Act does not target states but persons.

The case law brings weak support to the idea of a bias against state regulation. In Addyston Pipe & Steel Co. v. United States,126175 U.S. 211 (1899). Justice Peckham wrote that a bid-rigging and price-fixing combination between suppliers of cast iron pipes resulted in a direct and substantial obstruction to, or regulation of, commerce “possibly quite as effectually as if a [s]tate had passed a statute of like tenor as the contract.”127Id. at 230. However, only an incomplete reading of Justice Peckham’s opinion would allow us to discern a per se inclination against government intervention. The opinion made clear that the evil stemming from state regulation was the transaction costs arising from legal and economic fragmentation.128Id. at 232–33 (“It is true that under our system of government there are numerous subjects over which the States have exclusive jurisdiction, resulting in the enactment of different laws upon the same subject in various States, and also in varying and inconsistent judicial judgments in the different States upon the same subject. That condition has never been regarded as an end in itself desirable.”).

The sole case that could be presented as a sign of hostility toward state intervention is International Harvester Co. of America v. Kentucky.129234 U.S. 216 (1914). The state of Kentucky had declared price fixing unlawful only when entered into “for the purpose or with the effect of fixing a price that was greater or less than the real value of the article.”130Id. at 221. The Supreme Court reversed lower court judgments on the basis of the Fourteenth Amendment.131Id. at 222–23. Justice Holmes, emphatically writing for the majority, declared that guessing the real value of a product absent the impugned restraint of competition is “a problem that no human ingenuity [can] solve,” and a “gift[] that mankind does not possess.”132Id. at 223, 224. What is interesting lies however elsewhere. It is that the price fixing exemption written in the law had been the result of lobbying from Kentucky’s domestic tobacco industry and farming sector, which sought to counter the bargaining power of large buyers.

The small and indirect trace of regulatory capture in International Harvester is however insufficient to draw an inference of a case law antipathy towards state intervention in the formative era.133The facts in the 1912 Reading case also point to state intervention as a source of competitive distortions. The Court suggested that the combination charged was “a further direct consequence of the state authorized alliance between coal-producing and coal-transporting companies.” United States v. Reading Co., 226 U.S. 324, 339 (1912).

6.     Conclusion

While Chicagoan antitrust limits cannot be found as an ideological system in the formative era, some were undeniably present. When they appeared, however, it was in a haphazard fashion, lacking in clarity of purpose and specification. Often, the limits arose sporadically only to be washed away in the next case.134Efficiency is a good example. Compare the per se endorsement of efficiency in Winslow in 1913 against the per se rejection of efficiency in Eastern States Retail Lumber Dealers’ Assn in 1914, where the argument that the course of conduct is necessary for “public welfare in providing retail facilities is answered by the fact that Congress . . . has so legislated,” so that “private choice of means must yield to the national authority.” E. States Retail Lumber Dealers’ Ass’n v. United States, 234 U.S. 600, 613 (1914); see also United States v. Winslow, 227 U.S. 202 (1913) (holding that a combination for greater efficiency does not necessarily violate the Sherman Act).

The three limits invented by antitrust judges in the formative era are (1) the concern towards power over price and output; (2) the requirement of thorough empirical inquiry of effect, and to a lesser extent of purpose; and (3) the consideration of efficiencies.

The argument that can be drawn from this is that the Supreme Court consciously or unconsciously came up with the idea of limiting antitrust law. Moreover, the substantive proximity of the limits with those that the Chicago School advanced decades later is unmistakable. However, it is clear that jurisprudence could not be shaped by the corpus of methods, system of normative beliefs, or set of policy prescriptions of the Chicago School. At best, a similar political mindset might have played a role.

The limits that we have highlighted can thus be said to have practical foundations. Compared with ideological foundations, the practical foundations pertain to common sense, logic, and pragmatism. Compared to ideology that mobilizes dogmatic reasoning and yields deterministic results, pragmatism leans toward uncertainty and unpredictability.

The upshot is that ideological criticisms of current antitrust law are unconvincing. Arguably, critics know this. Saying that the limits are ideological could just be a tactic to suggest that it is perfectly fine and legitimate to change the law by political fiat, without caring much about the costs and benefits in the administration of the law.

II.     The Path of Antitrust Law’s Limits

We now want to lay down the basic components of a descriptive theory of the construction of antitrust laws limits by studying where courts draw the line. The theory must describe the concerns that explain the evolution of antitrust doctrine toward what it is today, given its starting conditions and environmental constraints. To account for heterogeneity, Section A comparatively studies the general characteristics of the US and EU antitrust law systems. Section B then shows that several features of each regime correlate with different attitudes of US and EU doctrine towards uncertainty and the treatment of imperfect facts. Finally, Section C observes that the root cause of these differences remains, however, inherently speculative, and the quest for a single explanation is, often, misguided.

A.     General Characteristics of the US and EU Antitrust Law Systems

The US and EU antitrust law systems have specific substantive, institutional, and cultural characteristics, and some corrections have been brought over time.

1.     US Antitrust

The statutory foundation of US antitrust law is the Sherman Act of 1890. The act’s “strict language”135United States v. Trans-Missouri Freight Assn, 166 U.S. 290, 312 (1897). uses “abstractions”136United States v. U.S. Steel Corp., 251 U.S. 417, 452 (1920). like “restraint[s] of trade”137Id. at 436. or “conspiracy.”138Krulewitch v. United States, 336 U.S. 440, 443 (1949). In Krulewitch, Justice Jackson remarked that “[t]he modern crime of conspiracy is so vague that it almost defies definition.” Id. 446 (Jackson, J., concurring). The broad standards found in the Sherman Act give it “a generality and adaptability comparable to that found to be desirable in constitutional provisions.”139Appalachian Coals, Inc. v. United States, 288 U.S. 344, 360 (1933). In the US, antitrust statutes enacted by legislatures are rare.140The main reform of the Sherman Act of 1890 was brought in 1914 with adoption of the Clayton Act and Federal Trade Commission Act. See Clayton Act, ch. 323, 38 Stat. 730 (1914) (codified as amended in scattered sections of 15 and 29 U.S.C.); Federal Trade Commission Act, ch. 311, 38 Stat. 717 (1914) (codified as amended at 15 U.S.C. §§ 41–58). References to US law in this Article refer to federal antitrust law and not the antitrust laws of the fifty states and the District of Columbia. Judicial opinions enacted by courts are the primary source of law. Substantive antitrust law evolves following the incremental, slow, and uncertain process of the common law.141Guido Calabresi, A Common Law for the Age of Statutes 75 (1999). For a description of the common law-like nature of US antitrust, see William F. Baxter, Separation of Powers, Prosecutorial Discretion, and the “Common Law” Nature of Antitrust Law, 60 Tex. L. Rev. 661, 665–66 (1982).

US antitrust law can be mobilized by two federal agencies, fifty states, and any person or entity harmed by a violation of the Sherman Act. Actions can be brought before hundreds of federal courts across the nation.142US federal antitrust laws cannot be enforced in state courts. 15 U.S.C. §§ 15(a), 26. Antitrust cases move from federal courts at the district level to court of appeals at the circuit level, and ultimately to the Supreme Court. Supreme Court antitrust opinions speak for all entities.143Calabresi, supra note 141, at 63; see also Address by R. Hewitt Pate, supra note 67 (“Because there are so few Supreme Court antitrust decisions each year and because each one sets precedent that will govern the application of the antitrust laws in the lower courts for decades to come each decision is an event of major significance for antitrust enforcers and the antitrust bar. Every phrase is studied with care, and every future case is evaluated in terms of the Court’s reasoning process.”).

The courts are generalist. Lay judges and juries cannot be presumed to possess economics training or business expertise.144See Limits, supra note 1, at 39 (explaining that antitrust is “an imperfect tool . . . because we rarely know the right amount of competition there should be, because neither judges nor juries are particularly good at handling complex economic arguments, and because many plaintiffs are interested in restraining rather than promoting competition”). The comprehensive portfolio of cases that courts hear gives them in return an opportunity to gain familiarity with cross-cutting issues.145See id. Judges understand the big picture and they are, or are expected to be, impartial.

All binding antitrust decisions are adopted by an independent court which decides, presumably without bias, for the government or the private plaintiff.146This is true with the possible exception of the FTC. There is a hearing before an administrative law judge, but some suggest that this does not avoid the confirmation bias problem of decision by the Commission. On judicial review, the federal appeals court makes a de novo determination of the law, but reviews the facts on a deferential standard that permits overturning only if there is no substantial evidence in the record to support the Commission. See 15 U.S.C. § 45(c). This aspect is reinforced by the accusatorial nature of the procedure. Judges put no skin in the game of investigation. They must decide in favor of the party that has brought the best evidence before them.

The courts that exercise an antitrust power can hear civil and criminal cases. Criminal remedies consist in sentencing and fines. Civil remedies consist in damages, injunctions, cease and desist orders, disgorgement of wrongful gains, and equitable remedies including supervision by monitoring trustees and dissolution of a company.147Edward Cavanagh, Antitrust Remedies Revisited, 84 Or. L. Rev. 147, 150–52 (2005) (discussing various civil remedies).

Antitrust litigation tends to produce big cases.148For an in-depth discussion on common evidentiary and procedural issues in antitrust cases, see Breck P. McAllister, The Big Case: Procedural Problems in Antitrust Litigation, 64 Harv. L. Rev. 27 (1950). The evidence is often concealed and, in almost all cases, involves combinations or monopolies engaged in interstate commerce.149See generally Cavanagh, supra note 147, at 182–205. The result is that the typical antitrust trial involves broad issues of fact and requires greater proof compared to a conventional case.150Walter L. Rice, Trial Technique in Antitrust Cases, 7 L. & Contemp. Probs. 138, 138 (1940). The proof is predominantly documentary. Litigants can force the production of proof by asking the judge to order discovery or issue subpoenas. “The exhibits selected from these documents furnish[] the [evidentiary] basis” of most antitrust cases.151Id. at 142. Witnesses, often competitors or customers of the defendants but also field experts, bring the rest of the evidence.152See id. at 138–39. In criminal cases, a grand jury can be summoned to supervise the collection of the evidence.153Fed. R. Crim. P. 6(a)(1). Grand juries play a key role to protect corporate citizens from unfounded criminal charges.154See 9-11.010 – Grand Jury: Introduction, U.S. Dep’t of Just., https://perma.cc/J7AS-E84U.

Cases are tried before the judge, sometimes with assistance of a jury for the verdict. Witnesses are cross-examined, often to collect expressions of opinion about the effects of the impugned practices, or to expose contradictions, inconsistencies, or even falsehoods.155This is true even though the case of falsehoods is more rare. See Rice, supra note 150, at 138–39.

Overall, the important characteristics of the US antitrust system are the following. First, it is highly judicialized. No remedy can be applied to an antitrust defendant without judicial approval. And judges prioritize facts and consistency over policy concerns. Second, it is extremely decentralized. A multiplicity of forums (courts), and actors (litigants) can apply and mobilize the statutes. Stare decisis and legislative supremacy are not strictly adhered to in the lower courts, especially when Supreme Court and legislative injunctions are perceived as detached from practical or theoretical realities.156For example, the Robinson-Patman Act rule that prohibits price discrimination has been disabled by application of strict requirements. See Herbert Hovenkamp, The Robinson-Patman Act and Competition: Unfinished Business, 68 Antitrust L.J. 125, 125 (2000). Similarly, the judgment of the Supreme Court in Eastman Kodak Co. v. Image Technical Services, Inc. has not been strictly constructed by lower courts, and complaints brought on this basis stand little chance of success. See generally Eastman Kodak Co. v. Image Tech. Servs., Inc., 504 U.S. 451 (1992). For another example, see SMS Sys. Maint. Servs. v. Digit. Equip. Corp., 11 F. Supp. 2d 166 (D. Mass. 1998), aff’d, 188 F.3d 11 (1st Cir. 1999). Third, US antitrust law is litigation friendly. The law embeds powerful incentives to private rights of action, including generous discovery rules,157In his 1984 article, Easterbrook talked of a situation of “ceaseless discovery.” See Limits, supra note 1, at 12. mandatory treble damages, fees for victorious litigants, contingency fees, class actions, etc.158See Cavanagh, supra note 11, at 42, 47–48, 64.

In the course of US antitrust history, corrections were brought.159For example, numerous devices have been introduced to avoid trial, including summary judgment, class certification motions, legal rules like “per se” and “quick look” that simplify the issues, etc. See generally id. at 48, 52–53, 58. To cope with hard, imperfect, and messy facts as well as complex economics, administrative capabilities were raised with the addition of the Federal Trade Commission (“FTC”). Pointed changes were brought to statutory law too. For example, a 1974 reform seeking to limit the number of Supreme Court opinions in antitrust cases reduced incentives for long, drawn out litigation but may have strengthened the centrifugal nature of the US antitrust system.160See generally 15 U.S.C. § 29(b).

2.     EU Competition Law

The textual provisions of EU antitrust law are, to a major degree, “modern restatements” of US law.161James A. Rahl, International Antitrust, 79 Mich. L. Rev. 982, 985 (1981). Similar to US antitrust law, EU law uses broad standards like “concerted practices,” and “abuse of a dominant position” to designate forbidden conduct.162James Calder, Abid Qureshi, Fiona Schaeffer, Shola Ajewole, Christina DeVries, Eric Hochstadt & Richard Julie, A Review of Similarities and Contrasts Between American Antitrust and European Union Competition Law, 2004 Colum. Bus. L. Rev. 380, 382, 395 (internal quotation marks omitted). But the EU statutory provisions embody clearer instructions furnishing examples of bad business conduct. The scope of conduct within the condemnation of EU text law appears also more limited than US antitrust law.163Though, again, section 5 of the FTC Act has tended to broaden the scope of US substantive law. See Federal Trade Commission Act, Pub. L. 75–447, 52 Stat. 111, 112–15 (1938). The offense against combinations contains an exculpatory clause that allows the justification of cartels on grounds of public policy. And the offense against monopolization lies in monopolistic exploitation by an already dominant firm, not in monopoly acquisition, protection, or extension.164EU text law does not prohibit clearly the acquisition of monopoly power. This might be due to the fact that “[t]he European economy . . . has been more monopoli[z]ed . . . and its competitive self-righting mechanisms may be less robust.” John Vickers, Competition Law and Economics: A Mid-Atlantic Viewpoint, 3 Eur. Competition J. 1, 6 (2007). Another possibility is that the size of businesses in Europe remained low, compared to American firms, and the lawmakers did not want to raise obstacles to industrial concentration and deal with the problem of market power when it occurred. This reading is more convincing and confirmed by the absence of a merger control system in the Treaty of Rome. See generally Treaty Establishing the European Economic Community, Mar. 25, 1957, 298 U.N.T.S. 11.

The EU antitrust statutory provisions hint that the European lawmakers contemplated a system closer to economic regulation than to law enforcement.165This was a means to engage in broad public policy choices and/or directly and quickly prescribe market outcomes. The lawmakers expect the system to deal with “value choices”166René Joliet, The Rule of Reason in Antitrust Law: American, German and Common Market Laws in Comparative Perspective 113 (1967). that are normally beyond the limits of judicial competence and assume an ability to regulate and enjoin dominant market power “as soon as exploited.”167René Joliet, Monopolization and Abuse of Dominant Position: A Comparative Study of the American and European Approaches to the Control of Economic Power 241 (1970). In sum, EU antitrust law posits, and requires, enforcement capabilities of benevolent social planning with superior knowledge and perfect information.168See Consolidated Version of the Treaty on the Functioning of the European Union arts. 101(3), 102(a), July 6, 2016, 2016 O.J. (C 202) 88–89, 309 [hereinafter TFEU]. In the same sense, the language of the EU statutory provisions appear to be designed to pursue multiple goals, that is fairness,169See id. arts. 101(3), 102(a), at 88–89. market integration in a large territory,170See id. arts. 101(1)(d), 102(c), at 88–89. and the protection of third parties from disadvantage, without specifying a hierarchy.

The enforcement of EU antitrust corresponds to the system foreseen in statutory law. Competition is deemed better maintained by application of a system of control based on centralized authority with expertise. EU antitrust is mostly enforced by delegation of power to central administrative agencies at the national and pan-European levels (to the European Commission, “EC”).171Within the system, the centralized administrative agency called the European Commission occupies a leading role. See, e.g., id. arts. 17, 32, at 25, 60. Private enforcement of the law before courts is less robust than in the US.172See Anu Bradford, Adam S. Chilton, Katerina Linos & Alex Weaver, The Global Dominance of European Competition Law Over American Antitrust Law, 16 J. Empirical Legal Stud. 731, 734 (2019). And whereas the US agencies have to involve courts when seeking an injunction or another remedy, the EU agencies have “atypically permissive” powers to carry out enforcement without needing to resort to courts.173Stephen Wilks & Lee McGowan, Disarming the Commission: The Debate Over a European Cartel Office, 32 J. Common Mkt. Stud. 259, 259–61 (1995) (“The centrality of the competition policy regime deserves emphasis. In this field the Commission enjoys exceptional autonomy based on Articles 85–94 of the Treaty of Rome and on the atypically permissive Regulation 17 of 1962 which allows it generous powers to investigate, codify, exempt and fine.”). The agencies’ claim to legitimacy is based on the highly qualified experts at their disposal (and the accretion of knowledge and learning by repetition),174The antitrust statutes are administered iteratively, allowing for the accumulation of knowledge. as well as on their substantial autonomy from the executive and legislative branches (though indirect influence can be exerted through appointment and budget oversight).

Compared to the rules in the US, the rules of procedure and evidence in the EU are more bespoke, flexible, and underdeveloped.175The rules are bespoke in the sense that the EU antitrust law system emerged in an environment less shaped by a well-established legal tradition than the common law context in which the Sherman Act took place. Agencies tend to be in charge of both investigation and decision-making (fact finding, liability, and remedy).176However, some countries separate both activities. See European Competition Network, Decision-Making Powers Report 5–10 (2012). Proceedings are ex parte. Agencies can pursue antitrust charges in the absence of a formal plaintiff. There is no right of confrontation and cross-examination of depositions and testimony. There is no right to discovery from third parties. And the agency is not required to take a position on all the facts, legal arguments, and economic evidence brought by a defendant. Moreover, agency decisions tend to be self-executing. Business conduct can be changed, and violators fined, before judicial review. Fines and behavioral remedies are the rule, structural relief the exception. Victims are infrequently awarded damages by national courts.177David Ashton, Competition Damages Actions in the EU: Law and Practice 0.06 (2d ed. 2018) (“[T]he situation in the EU stands in marked contrast to that in the US . . . [and that] the vast majority . . . of antitrust proceedings in the US are initiated by private parties, while in the EU, by contrast, antitrust enforcement remains the quasi exclusive preserve of the public enforcement authorities.”). More generally, on the low levels of damages action, see White Paper on Actions for Breach of the EC Antitrust Rules, at 3, COM (2008) 165 final (Apr. 2, 2008). Note that European private antitrust damages actions have, however, increased in the last decade, with the adoption of Directive 2014/104. See generally European Parliament and Council Directive 2014/104 of Nov. 26, 2014, Certain Rules Governing Actions for Damages under National Law for Infringements of the Competition Law Provisions of the Member States and of the European Union, 2014 O.J. (L 349) 1, 1–19. When this happens, it is often after an agency finding of antitrust liability. Criminal sanctions are rare.178See Wouter P.J. Wils, Is Criminalization of EU Competition Law the Answer?, 28 World Competition 117, 133–37 (2005). There are no criminal sanctions at the European level. See Council Regulation 1/2003, art. 23(5), 2003 O.J. (L 1) 16–17.

Challenges to agency action do not entail a standard appeal. The system of European judicial review is modeled on French administrative law where the administration is subject to its “own” judge.179See generally Jürgen Schwarze, Judicial Review in EC Law—Some Reflections on the Origins and the Actual Legal Situation, 51 Int’l & Compar. L.Q. 17, 17 (2002). The function of scrutinizing administrative action is assigned to a specific judicial body outside of the established system of courts.180See, for example, Guy Carcassonne, The Case of France, 39 Hung. J.L. Stud. 151 (1998). French administrative law is rife with discussions about whether proceedings can be brought against an administrative body before a different, private law judge. See Carole Harlow, “Public” and “Private” Law: Definition Without Distinction, Mod. L. Rev. 241, 251 (1980) (noting “the jurisdictional disputes which are so characteristic of French administrative law”). In line with the continental tradition, the European courts tolerate substantial deference to administrative discretion.181Joana Mendes, Law and Administrative Discretion in the EU: Value of a Comparative Perspective, in Comparative Administrative Law 632, 636 (Susan Rose-Ackerman, Peter L. Lindseth & Blake Emerson eds., 2017). An official in Brussels once said in a nonpublic context: the Commission is the friend of the court. Self-executing antitrust decisions benefit from a presumption of lawfulness. There is no de novo review and no opportunity for litigants to bring new raw evidence. The European courts do not scrutinize the determination of the relevant facts brought in support of antitrust liability, and in cases involving complex economic assessments, the qualification of the facts as an antitrust violation (or not) is subject to marginal screening of gross errors.182Though rarely the court annuls Commission decisions, it can call into question the substance of the Commission’s analysis. E.g., Case T-399/16, CK Telecoms UK Invs. Ltd v. Commn, ECLI:EU:T:2020:217, ¶¶ 174–98 (May 28, 2020). The European courts tend to defer to agency interpretation of statutory antitrust law like under Chevron deference in the US.183See Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 865 (1984). Rulings are binary: they uphold or annul the decision.184In some cases, the court also reduces or increases fines. Note also that the Court might find that the Commission erred in its legal assessment or in its analysis of the facts, but that the error is not substantial enough to trigger an annulment. See Ivo van Bael, Insufficient Control of EC Competition Law Enforcement, in Annual Proceedings of the Fordham Corporate Law Institute 742 (Barry E. Hawk ed., 1993). The European courts judge by consensus without a possibility of dissent.185Josef Azizi, Unveiling the EU Courts’ Internal Decision-Making Process: A Case for Dissenting Opinions?, 12 ERA F. 49, 50 (2011). Vagueness is frequently necessary to obtain approval from all the judges that could block the ruling. There is not a strict culture of stare decisis. The European courts do not practice distinguishing or overruling explicitly. The European courts tend to cite all tangentially related cases, not just the relevant precedents, as if it wanted to suggest that the case law follows an immanent logical course.186Case C-413/14 P, Intel Corp. v. Commn, ECLI:EU:C:2017:632 (Sept. 6, 2017). Intel Corp. is a good example. There, the European Court of Justice went against prior opinions supporting a treatment of loyalty rebates under a per se prohibition rule. However, the Court maintained reference to precedent, adding that the prior “case-law must be further clarified.” See id. at ¶ 138.

The European antitrust system displays great ambitions for, and confidence in, what the law can achieve. The point, put more clearly, is this: the law is based on a strong assumption that decisionmakers have sufficient foreknowledge to make very broad tradeoffs (not just narrow ones), sort multiple objectives, and remedy quickly the manifestations of exploitative market power.187That vision was very much en vogue in the 1950s and 1960s in France, where technical committees of economists and engineers leveraged knowledge to plan the needs and opportunities of the economy. It belongs, moreover, to a long tradition in economics education in France, where good economists are essentially good engineers, trained in the best hard science schools with a heavy emphasis on mathematics. In a way, the drafters of EU antitrust law did not appear worried about problems in the application of the law. In addition, the above ambition, and related assumption, might form the basis for a general agreement about the comparative advantage of administrative agencies to apply laws to the facts of the day and give meaning in rulemaking instruments to the goals of the Treaty on the Functioning of the European Union (“TFEU”). This culture of antitrust law provides grounding to the characterization of European law as more regulatory than its US counterpart.188Larry Bumgardner, Antitrust Law in the European Union, 8 Graziadio Bus. Rev., no. 3, 2005, https://perma.cc/54YF-RNWK.

Another salient point of the EU antitrust system is that the central agency is in charge of challenging the authority of states. Additional provisions in statutory law give the EC a role in policing state-owned enterprises’ conduct and controlling the award of subsidies by governments.189These are the EU State aid rules, which are part of EU competition law and are contained in the TFEU. See TFEU, supra note 168, arts. 106–109, at 90–93. The specific configuration of the judicial review system towards the EC is also partly explained by the impossibility of letting national administrative courts decide cases involving domestic government actors. According to Article 344 TFEU, EU “Member States undertake not to submit a dispute concerning the interpretation or application of the Treaties to any method of settlement other than those provided for therein.” TFEU, supra note 168, art. 344, at 194. In the case law, this Treaty provision has been interpreted as granting exclusive jurisdiction to the European Court of Justice to determine disputes between Member States concerning EU law. E.g., Case C-459/03, Commn v. Ireland, ECLI:EU:C:2006:345 ¶¶ 123, 169 (May 30, 2006). A concentration of trust and power in central agencies is the dominant theme of EU antitrust law.

Over history, law reforms have progressively narrowed down the differences between the EU system and the US model.190This was fueled by increased trade globalization and the development of international policy cooperation. See William E. Kovacic, Competition Policy in the European Union and the United States: Convergence or Divergence?, in Competition Policy in the EU: Fifty Years on from the Treaty of Rome 314, 316 (Xavier Vives ed., 2009) (“Despite differences in philosophy, procedure, analytical technique, and, occasionally, substantive outcomes, the past decade has featured important enhancements in measures by public and non-government bodies in both jurisdictions to improve cooperation in the formulation and implementation of competition-policy standards governing transatlantic commerce.”). National competition laws have been introduced across the various countries of Europe, introducing a process of decentralization of enforcement. In 2014, a statute was adopted by the EU to remove obstacles to private actions for antitrust damages before ordinary courts.191See Council Directive 2014/104, 2014 O.J. (L 349) 1 (EU), for a discussion on certain rules governing actions for damages under national law for infringements of the competition law provisions of the Member States and of the EU. Substantial efforts were made in jurisprudence and administrative practice to raise the degree of protection afforded to due process rights in antitrust proceedings. And there has been increased interest towards an integration of the teachings of economics in enforcement, rulemaking activities, and institutional capabilities since the 2000s.192However, the discipline of economics concerned with the reduction of enforcement errors has had marginal impact.

B.     US and EU Antitrust: Different Attitudes Towards Uncertainty

Assume the above stylized, and oversimplified, descriptions of US and EU law are correct. What are the concerns that might have influenced the judicial construction of substantive limits on antitrust statutes? When compared, the opinions of US and EU courts differ in one key dimension: how to deal with imperfect facts. US doctrine requires the resolution of more factual uncertainty as a condition for antitrust intervention than EU doctrine.193The focus here is on agreements and unilateral conduct law and excludes merger law because the Supreme Court has not decided a merger case since 1974. See United States v. Gen. Dynamics Corp., 415 U.S. 486 (1974).

1.     A Common Issue: Dealing with Imperfect Facts

Antitrust law is about dealing with imperfect facts. The job of an antitrust decisionmaker is to evaluate whether business conduct warrants liability. The facts matter a great deal, even in basic cases where the law adopts rules of strict liability. Take the per se prohibition of “price fixing.”194Adam Weg, Per Se Treatment: An Unnecessary Relic of Antitrust Litigation, 60 Hastings L.J. 1535, 1550 (2009). The facts must be verified to see if defendants have engaged in price fixing. But the facts are ambiguous. Is an agreement among partners in a law firm to set a fee policy akin to price fixing?195This example is taken from Bork, supra note 18, at 264–67. And what about discussions amongst competing banana producers on the weather forecast? Few agreements present themselves to antitrust decisionmakers with a price-fixing label on them.

Sometimes, the facts are unobservable. Consider the concepts of monopoly or dominance. Academic economics suggests that monopoly power shows up in deviations from short-run marginal cost pricing. The problem with this suggestion is that no firm calculates marginal costs. Observable facts, like accounting data, exist. But those data do not measure opportunity costs—a key component of marginal cost—which are unobservable.

Last, some facts are unknowable. The US merger law requirement of a showing of a “substantial lessening of competition” to separate good from bad transactions is a case in point.196See Richard B. Blackwell, Section 7 of the Clayton Act: Its Application to the Conglomerate Merger, 13 Wm. & Mary L. Rev. 623, 626 (1972). How do we predict the rate and direction of future innovation that might dissipate or exacerbate concerns of future competitive harms? Or consider the concept of “attempt to monopolize” in unilateral conduct law.197Katharine Kemp, Misuse of Market Power: Rationale and Reform (Apr. 11, 2016) (P.D. dissertation, University of New South Wales) (on file with University of New South Wales). The voluntary nature of an attempt focuses attention on subjective intent.198Swift & Co. v. United States, 196 U.S. 375, 396 (1905) (“Intent is almost essential to such a combination and is essential to such an attempt. Where acts are not sufficient in themselves to produce a result which the law seeks to prevent—for instance, the monopoly—but require further acts in addition to the mere forces of nature to bring that result to pass, an intent to bring it to pass is necessary in order to produce a dangerous probability that it will happen.”). But how do we divine the mens rea behind business conduct? Much like legislative intent, business intent is murky. Many individuals, groups, and hierarchies are involved in the decision-making processes of the modern business enterprise. And even if intent can be located in a single employee, manager, or director’s decision, the element of a “guilty” mind required for intent is hard to establish (and might not be representative of corporate policy).199See Spectrum Sports, Inc. v. McQuillan, 506 U.S. 447, 459 (1993) (“[T]he necessary intent to monopolize . . . is something more than an intent to compete vigorously . . . .”). Antitrust laws, like business school curriculums, work on the basic assumption that killing (rivals) is firms’ business, and that business is good.

The law tends to address these questions by requiring further factual inquiries. In the price fixing examples mentioned previously, the law might require answers to the following questions: (1) were the partners of the law firm competing in the same field and for the same clients; and (2) did weather conditions determine prices for the banana farmers?

The factual evidence that antitrust doctrine adds to the demands of statutory law are the limits of antitrust. Doctrinal demands for further factual inquiries come under different names, involve distinct processes, and are more or less demanding. Their common feature is to condition liability on the resolution of some factual uncertainty. Depending on their exact calibration, the limits embody a judgment about the degree of empirical truth necessary to ground antitrust intervention. And the limits reflect antitrust doctrine’s confidence in antitrust decisionmakers’ ability and incentives to make sound decisions under uncertainty.

2.     US Law: Interest in Competitive Realities

US antitrust doctrine requires hard empiricism. Broad statutory language, diverse litigation venues, and other incentives to bring lawsuits correlate with an antitrust doctrine that develops rules ensuring that firms are only condemned on the basis of competitive realities.200In Anderson, the Supreme Court observed, “there is scarcely any agreement among men which has interstate or foreign commerce for its subject that may not remotely be said to, in some obscure way, affect that commerce and to be therefore void.” See Anderson v. United States, 171 U.S. 604, 616 (1898). And, in NCAA v. Board of Regents of the University of Oklahoma, 468 U.S. 85 (1984), the Court noted again that “every contract is a restraint of trade,” id. at 98. Elimination, quasi-elimination, or substantial elimination of factual uncertainty conditions antitrust liability in US law.

US antitrust doctrine’s insistence on the resolution of factual uncertainty is discernible in four broad trends. First, antitrust doctrine has a highly factual component that courts must consider to declare business conduct unlawful under the Sherman Act. The contemporary case law of the Court has shifted conduct away from per se illegality towards the rule of reason end of the spectrum.201Rules of per se illegality allow the fact finder to limit its investigation to the “character of the restraint.” See Jefferson Parish Hosp. Dist. No. 2 v. Hyde, 466 U.S. 2, 9 (1984). In FTC v. Actavis, Inc.,202570 U.S. 136 (2013). the Supreme Court analogized reverse payment settlements to horizontal collusion, yet refused to follow the FTC’s invitation to apply a “quick look” rule, insisting that “complexities,” such as industry specific consideration required a more thorough market investigation under the rule of reason.203Id. at 159. The Court recalled Trinko, saying that horizontal collusion is the supreme evil of antitrust, yet considered that the case of reverse payment settlement had to be dealt with under the rule of reason. See generally Verizon Commcns, Inc. v. L. Offs. of Curtis V. Trinko, LLP, 540 U.S. 398 (2004). The Actavis Court allowed the defendant an opportunity to show “legitimate justifications,” including “traditional settlement considerations, such as avoiding litigation costs.” Actavis, 570 U.S. at 156. Similarly, in State Oil Co. v. Khan204522 U.S. 3 (1997). and Leegin Creative Leather Products, Inc. v. PSKS, Inc.,205551 U.S. 877 (2007). the Court removed vertical price fixing from the per se illegality category, holding that the rule of reason was the required mode of analysis of maximum and minimum RPM.206See id. at 894; State Oil, 522 U.S. at 7.

However, these changes in the case law do not amount to a full conversion to the rule of reason. Far from demanding a plenary market examination in all circumstances, the cases invite more intermediary modes of analysis whose net effect is to increase the set of facts relevant to the resolution of the antitrust case. In California Dental Ass’n v. FTC,207526 U.S. 756 (1999). for example, the Court refused to deal with the case under the “quick look” rule or an abbreviated rule of reason.208Id. at 779. The case had to do with professional advertising restrictions enforced by the California association of dentists.209See id. at 759–61. The Court required a more exacting “enquiry meet for the case, looking to the circumstances, details, and logic of a restraint,” yet suggested avoidance of a “plenary market examination.”210Id. at 779–81. What mattered, according to the majority, was to reach a “confident conclusion about the principal tendency” of the restraint at hand.211Id. at 781. In Actavis, the Court, again, insisted that the rule of reason does not require empirical perfection. It said that a plaintiff need not “empirically demonstrate the virtues . . . of the patent system, present every possible supporting fact or refute every possible pro-defense theory” for the purpose of meeting its burden under the rule of reason.212FTC v. Actavis, Inc., 570 U.S. 136, 159 (2013).

Finally, the Court has resisted every invitation to expand the per se rule.213The rule of per se illegality is limited to horizontal price fixing agreements. See Catalano, Inc. v. Target Sales, Inc., 446 U.S. 643, 647 (1980) (“A horizontal agreement to fix prices is the archetypal example of such a practice. It has long been settled that an agreement to fix prices is unlawful per se. It is no excuse that the prices fixed are themselves reasonable.”). In Nynex Corp. v. Discon, Inc.,214525 U.S. 128 (1998). the Court refused to declare the per se rule against a group boycott applicable in a pure vertical setting.215See id. at 130. And in FTC v. Indiana Federation of Dentists,216476 U.S. 447 (1986). the Court similarly refused to treat a concerted scheme amongst dentists to withhold information from health insurers as a “group boycott” because the setting involved no possibility of harm to a competitor.217Id. at 458. “[T]he per se approach has generally been limited to cases in which firms with market power boycott suppliers or customers in order to discourage them from doing business with a competitor—a situation obviously not present here.” Id.

Another technique the Court has developed is to fashion specific substantive filters, screens, and presumptions that require proof of particular facts to engage liability.218A two-pronged filter or screen says if fact A and B are established, then illegality/legality (e.g., exclusive dealing on more (less) than X% of an input market is unlawful (lawful)). A presumption of liability says if fact A is established, then illegality (legality), unless fact B (e.g., a merger to monopoly is unlawful, unless proof is brought of absence of anticompetitive effects). That trend is clear in Jefferson Parish Hospital District No. 2 v. Hyde.219466 U.S. 2 (1984). The disputed practice involved an exclusive contract between a hospital and a firm of anesthesiologists that prevented competing anesthesiologists from serving patients undergoing surgery at the hospital.220Id. at 4–5. The Court abandoned the per se rule against tying arrangements in favor of a more demanding presumption of illegality that required proof of: (1) defendant market power in the tying product; and (2) an effect on a substantial amount of commerce.221Id. at 8. Less spectacularly, in Spectrum Sports, Inc. v. McQuillan,222506 U.S. 447 (1993). the Court added a market power screen as a proxy for inefficiency and liability.223See id. at 457. The case involved a patent owner trying to terminate licensing and distributorship contracts to gain back market share for patented sports products.224See id. at 450. The Court held that a specific intent to monopolize was not sufficient in attempts to monopolize cases, but that section 2 liability also required “inquiry into the relevant product and geographic market and the defendant’s economic power in that market.”225Id. at 459. Lastly, Brooke Group Ltd. v. Brown & Williamson Tobacco Corp.226509 U.S. 209 (1993). follows similar logic while adding a consumer harm filter in predatory pricing cases.227See id. at 224. A price war with Brown & Williamson had left Liggett, once a successful first mover in generic cigarettes, on the brink of exit. The Supreme Court conditioned liability on a showing that the prey had a “reasonable prospect” of recouping its investment in below-cost prices.228Id.

Observe that the execution of the idea of resolving factual uncertainty by adding tests and filters is not always flawless. For example, Brooke Group appears to be a case that does not resolve factual uncertainty to the extent that it asks parties to debate about future, inherently hypothetical facts, to establish anticompetitive conduct.229See id. at 229, 233, 243. This explains why Brooke Group has remained a disputed opinion. The scholarly criticism of the case focuses as much on the principle of adding a recovery requirement, as it does on the handling of the evidence by the Court.230See Patrick Bolton, Joseph F. Brodley & Michael H. Riordan, Predatory Pricing: Strategic Theory and Legal Policy, 88 Geo. L.J. 2239, 2266 (2000). In its appraisal of the Brown & Williamson’s recovery prospects, the Court turned a blind eye to documented historical patterns of lockstep oligopoly pricing in the industry.231See id. at 2257–58. Instead, the Court erred on the side of theory, which teaches that tacit collusion is a rare phenomenon.232Id. at 2258. The Court noted that “tacit cooperation among oligopolists must be considered the least likely means of recouping predatory losses.” Brooke Grp., 509 U.S. at 228. But the Court did not go as far as to adopt a legal rule that would have made recoupment cases based on oligopoly pricing difficult. It added, “A predatory pricing scheme designed to preserve or create a stable oligopoly, if successful, can injure consumers in the same way, and to the same extent, as one designed to bring about a monopoly.” Id. at 229. And, it further noted “[t]heory will not stand in the way of liability.” Id.

A similar defect is found in the case of Ohio v. American Express Co.233138 S. Ct. 2274 (2018). Like in Brooke Group, the Court added a fact intensive filter, only to devolve back into abstract economic theory to discard the liability finding.234See id. at 2284–85, 2287. The case concerned the lawfulness of American Express’s (“Amex”) so-called “antisteering” rules and practices235Id. at 2280. These essentially prevented merchants that accepted Amex cards from, among other things, charging consumers lower prices (for goods or services) when consumers used other cards that charged merchants a lower fee.236See id. at 2283. Amex enforced the antisteering rules by careful oversight of the merchant’s client manager, random on-site visits, and reports from cardholders.237United States v. Am. Express Co., 88 F. Supp. 3d 143, 166 (E.D.N.Y. 2015), revd, 838 F.3d 179 (2d Cir. 2016), affd sub nom. Ohio v. Am. Express Co., 138 S. Ct. 2274 (2018). The government argued that the antisteering policy restricted competition and entry by reducing the ability of competing card issuers to increase sales by charging lower fees to merchants.238See Am. Express, 138 S. Ct. at 2287–89. Amex countered that the rules were necessary to protect its cardholder goodwill and to enable it to fund its more substantial cardholder rewards programs. These programs allowed Amex to compete by differentiating its products from the Visa and Mastercard alternatives.239See id. at 2282.

The Court held that defining a relevant market was necessary, and that, in what it called a “two-sided transaction platform,” in which a platform facilitates a “simultaneous transaction between participants,” the relevant market had to include both sides.240Id. at 2286. This analysis is largely formalistic. The Court reasoned that, in matters involving a transaction platform, both sides are simultaneously implicated in each transaction and there can be effects on one side from a transaction on the other.241See id. At first blush, the requirement of a single market makes sense. A more encompassing market definition might allow a more complete diagnosis of the totality of the injuries (and benefits) imposed by a restraint. But that question can equally be answered by defining two markets—one on the merchant side and one on the cardholder side—and by taking effects on each side into account as part of the factual context. Including both sides in the same market is thus unnecessary to the extent that it does not resolve any additional uncertainty.242Moreover, it can also be problematic by raising artificially levels of factual uncertainty. The requirement of a single market makes almost meaningless the idea of market shares, which if used properly can help assess a firm’s importance in the market. What is the “single” market share of a firm that has a 60% share on one side and only a 15% share on the other if both are included in the same market? Including both sides in a single market can also introduce needless complexity into antitrust cases by requiring, in effect, that plaintiffs define a single market for each transaction type (cardholders to grocery stores, cardholders to bookshops, cardholders to flight reservation systems, etc.). The Court’s holding multiplies the costs of investigation in markets where platforms address multiple categories of participants.

The second trend reveals that actual effects and direct evidence exert greater weight on the fate of antitrust cases compared to potential effects and indirect evidence.243Outside of cases dealt with under the per se rule, obviously. If there is a clear fact, then the courts cling to it. The presence of a concrete indication of consumer harm is a sufficient condition for liability that trumps the need for further inquiry into market definition and market power. In Indiana Federation of Dentists, the Court held that proof of actual adverse output effects “can obviate the need for an inquiry into market power.”244FTC v. Ind. Fedn of Dentists, 476 U.S. 447, 460–61 (1986); see also id. at 452 (“[T]he Federations policy had had the actual effect of eliminating such competition among dentists and preventing insurers from obtaining access to x[-]rays in the desired manner.”); Jefferson Parish Hosp. Dist. No. 2 v. Hyde, 466 U.S. 2, 29 (1984) (noting that “[t]hat burden necessarily involves an inquiry into the actual effect of the exclusive contract on competition among anesthesiologists”). In the section 2 jurisprudence, the law goes even further, requiring actual harm as a necessary condition of liability. Copperweld Corp. v. Independence Tube Corp.,245467 U.S. 752 (1984). an attempt to monopolize case, held that “[t]he conduct of a single firm is governed by [section] 2 alone and is unlawful only when it threatens actual monopolization.”246Id. at 767; see also Spectrum Sports, Inc. v. McQuillan, 506 U.S. 447, 456 (1993).

That actual facts are legally cognizable as such does not mean that the Court is always satisfied with clear price or output signals. Brooke Group’s neglect of empirical realities drives the point home. In judicial practice, there seems to be a dissymmetry between signs of actual harm—which are not always deemed sufficient to establish liability—and signs of actual efficiency, which appear almost always dispositive of lack of antitrust injury. Consider NCAA v. Board of Regents of the University of Oklahoma,247468 U.S. 85 (1984). a case about collective restrictions on college football teams’ ability to sell telecasting rights to TV channels.248Id. at 94. The Court recognized a price-fixing scheme’s “apparent” anticompetitive consequences but undertook a deeper assessment of the restraint’s impact.249See id. at 106. Now compare this approach with Brooke Group and the more recent American Express case. To reject the plaintiff’s recoupment theory in Brooke Group, the Court remarked that “output expanded at a rapid rate following Brown & Williamson’s alleged predation.”250Brooke Grp. Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 233 (1993). And in American Express, the Court disposed of the plaintiff’s claim of competitive injury towards Discover, Visa, and Mastercard noting that credit card transactions had increased, and that Amex’s merchant fees had been constrained.251See Ohio v. Am. Express Co., 138 S. Ct. 2274, 2288–89 (2018). Admittedly, in both cases, output might have been even higher in the absence of the disputed practices.252To focus on the Brooke Group facts, one could speculate, for example, that the rate of segment growth would have tripled, instead of doubled, without Brown & Williamsons alleged predation. Put in economic terms, a movement upward of the demand curve (industry level increase in output) does not rule out a movement along the demand curve (industry or firm level increase in prices). See Brooke Grp., 509 U.S. at 217. But, as the Brooke Group Court held, “such a counterfactual proposition is difficult to prove in the best of circumstances.”253Id. at 233. The bottom line is that courts display more confidence in uncertain theories of efficiency than in uncertain theories of competitive injury. As Justice Breyer famously opined in Barry Wright Corp. v. ITT Grinnell Corp.,254724 F.2d 227 (1st Cir. 1983). “[t]he antitrust laws very rarely reject such beneficial ‘birds in hand’ for the sake of more speculative (future low-price) ‘birds in the bush.’”255Id. at 234. The bottom line? Actual facts play a key role.

Third, the preference for actual facts not only undergirds the application of antitrust doctrine, but also extends to its formulation. In Eastman Kodak Co. v. Image Technical Services, Inc.,256504 U.S. 451 (1992). the existence of obvious “actual effects” on consumers, and the absence of production of “any actual data on the equipment, service, or parts markets,” gave confidence to the Court in its holding: that a test of monopoly power in primary goods markets—and by implication a requirement of market definition—was unnecessary in aftermarket cases where defendants enjoy market power in secondary markets and can effectuate exclusionary ties in tertiary markets.257Id. at 466. In Kodak, the Supreme Court recalled the appeals court’s finding that discovery had not gone far enough to know if there was market power, but that the fact-finding process was sufficient to trust the evidence of actual events that a reasonable trier of fact could conclude was enough. See id. at 460. More generally, the Kodak Court observed that “[l]egal presumptions that rest on formalistic distinctions rather than actual market realities are generally disfavored.”258Id. at 466. The threshold for the formulation of a legality presumption is that conduct must “always or almost always” enhance competition.259Id. at 479. This holding, to some extent, nuances previous remarks about a predisposition in the case law towards acceptance of uncertain economic theories of efficiency. The majority opinion in Catalano, Inc. v. Target Sales, Inc.260446 U.S. 643 (1980). goes in the same direction.261See id. at 645. The appeals court had entertained a funky theory whereby a cartel intending to raise prices and limit free credit would remove a barrier to sellers willing to enter the market.262See id. at 649. In response, the Supreme Court held:

If that potential justifies horizontal agreements among competitors imposing one kind of voluntary restraint or another on their competitive freedom, it would seem to follow that the more successful an agreement is in raising the price level, the safer it is from antitrust attack. Nothing could be more inconsistent with our cases.263Id.

The Supreme Court drove this point home in Weyerhaeuser Co. v. Ross-Simmons Hardwood Lumber Co.264549 U.S. 312 (2007). in relation to rules of illegality, emphasizing that tests must be based on “economic reality.”265Id. at 318.

Fourth, US law’s commitment to hard empiricism leads it to practice a strict division of labor. Antitrust doctrine generally prefers institutions with a comparative advantage in fact-finding processes. Simply put, it makes no sense to mobilize antitrust law when a specialized regulatory structure is in place. In Verizon Communications, Inc. v. Law Offices of Curtis V. Trinko, LLP,266540 U.S. 398 (2004). the Court came close to saying this when confronted with the question of whether section 2 liability could be imposed on a telecommunications incumbent for violations of sector-specific sharing obligations.267Kovacic, supra note 4, at 21 (noting that decisions like Trinko “view sectoral oversight more favorably than antitrust decisions”). The Court found no benefits to antitrust intervention. It noted that “[a]n antitrust court is unlikely to be an effective day-to-day enforcer of these detailed sharing obligations.”268Trinko, 540 U.S. at 415. In Pacific Bell Telephone Co. v. LinkLine Communications, Inc.,269555 U.S. 438 (2009). Justice Breyer adopted a close view in a concurring opinion,270Id. at 459 (Breyer, J., concurring) (“When a regulatory structure exists to deter and remedy anticompetitive harm, the costs of antitrust enforcement are likely to be greater than the benefits.”). but proposed to maintain the possibility of antitrust intervention when the government, rather than a private party, is the plaintiff.271Id. at 458 (“A ‘price-squeeze’ claim finds its natural home in a Sherman Act [section] 2 monopolization case where the Government as plaintiff seeks to show that a defendant’s monopoly power rests, not upon ‘skill, foresight and industry,’ but upon exclusionary conduct.” (citations omitted)). The decision in Credit Suisse Securities (USA) LLC v. Billing272551 U.S. 264 (2007). throws more light on the motives that lead antitrust law to defer to sector-specific institutions with informational advantages.273Id. at 283. The case dealt with investor allegations that investment banks underwriting IPOs had systematically gouged share prices and forced investors to buy the shares at escalated prices.274Id. at 267. The decision found that the Securities and Exchange Commission had “considerable power” to regulate investment banks, and had “continuously exercised its legal authority.”275Id. at 276–77. The Court added “[i]t will often be difficult for someone who is not familiar with accepted syndicate practices to determine with confidence whether an underwriter has insisted that an investor buy more shares in the immediate aftermarket.”276Id. at 280.

It would oversimplify the facts to read the above trends as just a defendant friendly evolution of doctrine. In some of the cases that formulated these doctrines, the decisions produced liability findings. In Kodak, NCAA, and Indiana Federation of Dentists, the antitrust courts upheld Sherman Act violations.277Eastman Kodak Co. v. Image Tech. Servs., Inc., 504 U.S. 451, 485–86 (1992); FTC v. Ind. Fed’n of Dentists, 476 U.S. 447, 459 (1986); NCAA v. Bd. of Regents of the Univ. of Okla., 468 U.S. 85, 120 (1984). In Federation of Dentists, the Court remarked that “[a]pplication of the Rule of Reason to these facts is not a matter of any great difficulty.” Fedn of Dentists, 476 U.S. at 459. In the lower courts, plaintiffs have managed to bring probative evidence of anticompetitive effects. A shining example of this is United States v. Microsoft Corp.,27884 F. Supp. 2d 9 (D.D.C. 1999), vacated 253 F.3d 34 (D.C. Cir. 2001). where the district court affirmed liability under sections 1 and 2 of the Sherman Act, in spite of “insufficient evidence” that the middleware layer market would have ignited genuine competition in the operating system (“OS”) market.279Id. at 93. Two facts convinced the district court to accept a lower burden of production with respect to incipient OS competition. First, the district court abstractly singled out the OS computer platform as an “important market.” Id. at 112. Second, the district court insisted that “firms already in the market” like IBM or Be Inc. had sought to develop a viable alternative to Windows, thereby backing its initial conjecture with empirically observed industry facts. Id. at 25. On appeal, the court stressed the “added uncertainty” of this theory. United States v. Microsoft Corp., 253 F.3d 34, 79 (D.C. Cir. 2001). The court of appeals however also supported a lower burden of production, although by somewhat distinct reasoning. Id. at 58–60. The court relied, in part, on theoretically plausible propositions about the purpose of the Sherman Act to protect “nascent . . . competitors” in “industries marked by rapid technological advance and frequent paradigm shifts.” Id. at 79.

None of the above, however, tells us why the courts continually ask for more fact finding. The mainstream hypothesis is that antitrust doctrine protects businesses from courts, non-expert judges, and lay juries. Concerns about the chilling effects of erroneous and costly enforcement in a decentralized system would transcend the history of US antitrust doctrine.280See, e.g., U.S. Dep’t of Just., Competition and Monopoly: Single-Firm Conduct and Section 2 of the Sherman Act 14–15 (2008), https://perma.cc/5Q9L-J5G2. In our view, however, a pro-business rationalization does not explain the entire phenomenon. If this hypothesis was right, we should see some doctrinal movement toward the rules of per se legality that Easterbrook requested in his famous 1984 paper.281Limits, supra note 1, at 9–10. Or, antitrust doctrine should have progressed towards lighter burdens of proof and evidence and a higher role and judicial deference towards public, expert, and centralized antitrust agencies.282See generally Kovacic, supra note 4.

A possible explanation is that other forces are also at play. In the US courts, facts matter more than anything else and increasingly so. As the economy has kept growing, the economics have become more subtle, for better and for worse. Uncertainty has increased. Today, we face an inflection point where scholars believe that computational technologies will be better placed to make sense of the complex facts of antitrust than ever-refined doctrinal rules on evidence, proof, and fact finding.283Thibault Schrepel, Computational Antitrust: An Introduction and Research Agenda, 1 Stan. Computational Antitrust 1, 4 (2021).

3.     EU Law: Interest in Competitive Probabilities

The last Section explained that US law deals with uncertainty by demanding more fact finding in both the application and formulation of the law; by clinging to clear empirics; or by deferring to specialized experts. Now we ask: how does EU law deal with the unknown?

Comparing the US and EU law to answer this question, we are struck by the different attitude towards uncertainty. EU competition doctrine is more probabilistic. Intervention is permitted at higher thresholds of uncertainty. To see this, we can reason by reference to the stylized trends discussed above. Before we start, let us caveat that the differences in statutory language, legal traditions, and doctrinal constructs make comparisons difficult, but not impossible.

First, where the US has expanded the universe of facts relevant to an antitrust case, EU jurisprudence has leaned towards allowing plaintiffs to avoid full inquiry of the market context and focus on the character of the restraint. If it can be said that the per se illegality rule is becoming an empty category in coordinated conduct cases in the US, it can conversely be said that the “restriction by object” category of EU law—its closest equivalent—has operated like a black hole, absorbing within its scope many allegations of antitrust violations.284See Commission Staff Working Document Guidance on Restrictions of Competition “by Object” for the Purpose of Defining Which Agreements May Benefit from the De Minimis Notice, at 5–17, SWD (2014) 198 final (June 25, 2014); see also TFEU, supra note 168, art. 101, at 88. The contrast is clear. In the US, pay for delay agreements have been classified as falling under the rule of reason, while the EU case Generics (UK) Ltd. v. CMA285See Case C-307/18, Generics (UK) Ltd. v. CMA, ECLI:EU:C:2020:52 (Jan. 30, 2020). has assessed them as “by object” restrictions.286See id. at ¶ 21. Where the US courts have been “slow to condemn rules adopted by professional associations as unreasonable per se,”287FTC v. Ind. Fed’n of Dentists, 476 U.S. 447, 458 (1986). EU law has treated rules of sports leagues or payment card systems as “by object” restrictions in cases like Football Ass’n Premier League Ltd. v. QC Leisure (Murphy),288Joined Cases C-403/08 & C-429/08, Football Ass’n Premier League Ltd. v. QC Leisure (Murphy), ECLI:EU:C:2011:631 (Oct. 4, 2011). Groupement des cartes bancaires v. Commission (Cartes Bancaires),289Case C-67/13 P, Groupement des cartes bancaires v. Commn (Cartes Bancaires), ECLI:EU:C:2014:2204 (Sept. 11, 2014). Gazdasági Versenyhivatal v. Budapest Bank Nyrt.,290Case C-228/18, Gazdasági Versenyhivatal v. Budapest Bank Nyrt., ECLI:EU:C:2020:265 (Apr. 2, 2020). and many others.

An objection to the argument is, however, that in European law, the analysis of a restraint as the “by object” category does not mean that facts do not matter.291See Case C-307/18, Generics (UK) Ltd. v. CMA, ECLI:EU:C:2020:52, ¶ 154 (Jan. 30, 2020). A long line of cases require an evaluation of the “economic and legal context” of an agreement, the “nature of the agreement,” the “position and importance” of the parties, “the market of the contract concerned,” and many other variables to declare a violation by object.292Joined Cases 56 & 58-64 Établissements Consten S.A.R.L. v. Comm’n, 1966 E.C.R. 299, 308, 331, 343; Case 56/65, Société Technique Minière v. Maschinenbau Ulm GmbH, 1966 E.C.R. 235, 250. In Generics, the “by object” analysis required by the Court is not so different from the investigation that takes place under the rule of reason in Actavis.293Compare Generics, ECLI:EU:C:2020:52 at ¶ 46, with FTC v. Actavis, Inc., 570 U.S. 136, 159 (2013). And European jurisprudence accepts that successful charges of anticompetitive behavior be brought on counts of “object” and “effect.”294See Budapest Bank, ECLI:EU:C:2020:265 at ¶ 17. That being said, there is still some way to go before the courts deem a “by object” evaluation to require an elaborate market analysis, a multi-factor economic investigation, and the collection of data. In reality, the movement of EU law is one that has iteratively pulled away the “by object” concept from the per se rule, replacing it with a focus on the “inherent nature” and “evident purpose” of a restraint.295In Cartes Bancaires, the Court insisted on a restrictive reading of the concept, informed by experience. See Case C-67/13 P, Groupement des cartes bancaires v. Commn (Cartes Bancaires), ECLI:EU:C:2014:2204 ¶ 51 (Sept. 11, 2014). In Budapest Bank, the Court said that an assessment of object sufficed to find a violation of article 101 TFEU and hinted that an assessment of effects would be redundant. See Budapest Bank, ECLI:EU:C:2020:265 at ¶ 48. It follows that the fact that a finding of a restriction of competition “by object” relieves the competent authority or court having jurisdiction of the need to examine the effects of that restriction in no way means that that authority or court cannot undertake such an examination where it considers it appropriate. Id. at ¶ 40. In his opinion under Budapest Bank, AG Bobek said that what had to be conducted amounted to one an absence of “specific circumstances,” and then no more than a “basic reality check.” See Opinion of AG Bobek, Budapest Bank, ECLI:EU:C:2020:265 at ¶¶ 45, 49. In ING Pensii, Societate de Administrare a unui Fond de Pensii Administrat Privat SA v. Consiliul Concurenței, the Court seemed to make a nominal reference to economic and legal context, stressing the analysis of the “terms” and “objectives” of conduct when it said that [A] finding that an agreement to share clients has an anti-competitive object—in particular a finding that the agreement may have a negative impact on the market—does not depend on the actual number of clients who are in fact shared out but simply on the terms and the objective aims of the agreement, considered in the light of the economic and legal context in which the agreement was concluded. See Case C‑172/14, ING Pensii, Societate de Administrare a unui Fond de Pensii Administrat Privat SA v. Consiliul Concurenței, ECLI:EU:C:2015:484, ¶ 55 (July 16, 2015). At the same time, the scope of the “by object” category has expanded.296See, e.g., C-439/09, Pierre Fabre Dermo-Cosmétique SAS v. Président de l’Autorité de la concurrence, ECLI:EU:C:2011:649 (Oct. 13, 2011) (finding a selective distribution agreement to be restrictive “by object”); see also Case C-32/11, Allianz Hungária Biztosító Zrt. v. Gazdasági Versenyhivatal, ECLI:EU:C:2013:160 (Mar. 14, 2013) (expanding the scope of the “by object” category to a vertical agreement between motor insurers and car repairers). This movement explains the statement contained in Cartes Bancaires, whereby the Court of Justice of the European Union (“CJEU”) held that the concept of a restriction by object must be interpreted strictly.297Cartes Bancaires, ECLI:EU:C:2014:2204 at ¶ 58. Bottom line? A similar direction of travel—towards less per se illegality—but a different final destination characterize EU law compared to US doctrine.

The same evolution characterizes some areas of abuse of dominance law. What do we discern? To start, one must recall that the EU works on a statutory prohibition of abuse of a dominant position but not of monopolization or attempted monopolization.298See TFEU, supra note 168, art. 102, at 89 (“Any abuse by one or more undertakings of a dominant position within the internal market or in a substantial part of it shall be prohibited as incompatible with the internal market in so far as it may affect trade between Member States.”). A showing of existing market power is a preliminary screen in all article 102 TFEU cases, and that requirement makes the fact-finding process intensive by design. A market power screen also applies in section 2 cases in US law in addition to a bad conduct requirement.299Market power in US law must be the outcome of bad conduct. In EU law, market power is the economic context in which bad conduct takes place. In this way, EU law appears necessarily more fact intensive than US law. The fact finder must establish that a market power situation is present before establishing an abuse. In US law, a prediction of future market power is sufficient. Still, this should not obscure that EU and US case law have diverged in relation to market power. While EU law has decreased the amount of fact finding by establishing a rebuttable presumption of dominance in AKZO Chemie BV v. Commission300Case C-62/86, AKZO Chemie BV v. Comm’n, 1991 E.C.R. I-3359. that is triggered once a firm crosses a market share threshold of 50%, US law has raised the burden of production by requiring proof of a high likelihood and sometimes a “dangerous probability” of market power and rejected invitations to establish a structural presumption.301In practice, in the US, the threshold for dominance is usually around 70%. See, e.g., U.S. Dep’t of Justice, supra note 280, at 21–22.

With this in mind, let us turn to the evolution of the European rules, specifically the evolution of the requirement of factual proof in abuse of dominance proceedings.

The evaluation of abuse of dominance analysis by the courts manifests some relaxation of the substantive burdens of proving facts. In some areas, like refusal to deal, the Court has abandoned strict evidentiary requirements imposed in past cases.302See Case C-7/97, Oscar Bronner GmbH & Co. KG v. Mediaprint Zeitungs-und Zeitschiriftenverlag GmbH & Co. KG (Bronner), 1998 E.C.R. I-7791; Joined Cases 241/91 P & 242/91 P, Radio Telefis Eireann (RTE) v. Comm’n (Magill), 1995 E.C.R. I-743. See generally C-418/01, IMS Health GmbH & Co. OHG v. NDC Health GmbH & Co. KG (IMS Health), 2004 E.C.R. I-5039. For example, in Radio Telefis Eireann (RTE) v. Commission (Magill),303See generally Magill, E.C.R. I-743. Oscar Bronner GmbH & Co. KG v. Mediaprint Zeitungs-und Zeitschiriftenverlag GmbH & Co. KG (Bronner),304See generally Bronner, E.C.R. I-7791. and IMS Health GmbH & Co. OHG v. NDC Health GmbH & Co. KG,305See generally IMS Health, E.C.R. I-5039. the CJEU conditioned a finding of abuse on the showing of three facts: the indispensability of the disputed input, the elimination of all competition in an adjacent market, and the absence of an objective justification.306See IMS Health, E.C.R. I-5039 at ¶¶ 44–52; Bronner, E.C.R. I-7791 at ¶¶ 36–47; Magill, E.C.R. I-743 at ¶¶ 51–58. In cases involving intellectual property, the conduct had to also prevent the emergence of a “new product” for which there was potential consumer demand.307See IMS Health, E.C.R. I-5039 at ¶ 32; Bronner, E.C.R. I-7791 at ¶ 40; Magill, E.C.R. I-743 at ¶ 30. More modern judicial statements on refusal to deal like Microsoft Corp. v. Commission,308Case T-201/04, Microsoft Corp. v. Commn, 2007 E.R.C. II-3619. Slovak Telekom v. Commission,309Case C-165/19 P, Slovak Telekom a.s. v. Commn, ELCI:EU:C:2021:239 (Mar. 25, 2021). or Huawei Technologies Co. Ltd v. ZTE Corp.310Case C‑170/13, Huawei Techs. Co. v. ZTE Corp., ELCI:EU:C:2015:477 (July 16, 2015). have reversed this trend. The cases have in common the holding that the structured test or some of its conditions are inapplicable.311See generally Slovak Telekom, ELCI:EU:C:2021:239; Huawei, ELCI:EU:C:2015:477; Microsoft, E.R.C. II-3619. Findings of liability can be reached based on a less intensive, and less exhaustive, study of the facts. For example, in Microsoft, the facts of the case hinted that the interoperability information that was at the heart of the case was not indispensable for a rival to enter the market.312See Pablo Ibáñez Colomo, Indispensability and Abuse of Dominance: From Commercial Solvents to Slovak Telekom and Google Shopping, 10 J. Eur. Competition L. & Prac. 532 (2019).

In regard to predatory pricing cases, a similar evolution can be detected. In France Télécom SA v. Commission,313Case C‑202/07 P, France Télécom SA v. Commn, 2009 E.C.R. I-2369. a telecommunications incumbent had organized a campaign of below-cost pricing in the new market for high-speed internet.314Id. at ¶ 3. The legal issue that confronted the CJEU was whether to add a factual inquiry of recoupment to the simple price-cost test defined in the AKZO case. The CJEU refused, taking an approach diametrically opposed to the Supreme Court in Brooke Group. Economist John Vickers remarked that the CJEU might have considered that the need to show dominance did away with the need to show recoupment.315John Vickers, Competition Law and Economics: A Mid-Atlantic Viewpoint, 3 Eur. Competition J. 1, 6 (2007). But the language of the French Télécom decision points to a distinct rationale. While the Brooke Group court conditioned liability on a factual showing of consumer harm, the French Télécom Court states that the showing of consumer harm could be more or less presumed because

following the withdrawal from the market of one or a number of its competitors, . . . the degree of competition existing on the market, already weakened precisely because of the presence of the undertaking concerned, is further reduced and customers suffer loss as a result of the limitation of the choices available to them.316France Télécom, E.C.R. I-2369 at ¶ 112.

That being said, other areas of abuse of dominance law display a movement toward more fact finding. Tying law is a clear-cut example. Early jurisprudence treated tying practices under a quasi-per se prohibition rule.317Robert O’Donoghue & Jorge Padilla, The Law and Economics of Article 102 TFEU 306 (3d ed. 2020). The case law required application of a three-pronged test.318Case T-30/89, Hilti AG v. Commn, 1991 E.C.R. I-1441, ¶ 46; see also Case T-83/91 Tetra Pak Int’l SA v. Commn (Tetra Pak II), 1994 E.C.R. II-762, ¶ 41 (noting market position and propriety of action). Under influence of the EC, the cases evolved to embed a fourth condition that requires an economic demonstration of foreclosure of competition. The General Court confirmed the validity of this approach in Microsoft, though its judgment is not “a model of clarity.”319O’Donoghue & Padilla, supra note 317; see also Case T-201/04 Microsoft Corp. v. Commn, 2007 E.R.C. II-3619, ¶¶ 862–68.

Exclusive dealing law has followed a similar direction, though this time not under the impetus of the EC. The 1979 judgment from the CJEU in Hoffmann-La Roche v. Commission320Case 85/76, Hoffmann-La Roche & Co. AG v. Commn, 1979 E.C.R. 461. held that “fidelity rebates” had to be scrutinized under a per se rule of illegality.321Id. at ¶ 90. Fidelity rebates were defined as discounts conditional on buyers acquiring most or all their needs from a dominant seller.322See id. at ¶ 89. In Intel Corp. v. Commission,323Case C-413/14 P, Intel Corp. v. Commn, ELCI:EU:C:2017:632 (Sept. 6, 2017). the CJEU softened considerably the per se rule against fidelity rebates.324Id. at ¶¶ 137–40. The Court granted defendants a procedural right to challenge a presumptive allegation of harm to competition on the basis of supporting evidence and imposed on the EC a procedural duty to review it.325Id. at ¶¶ 140–44. The Intel decision marks a movement away from the view that a dominant firm’s low price policies are inherently suspicious.

The law of exploitative abuses is one area where fact-intensive tests of illegality have remained stable. In unfair pricing and price discrimination cases, the courts have maintained heavy burdens of proof on plaintiffs. In AKKA/LAA,326Case C-177/16, Autortiesību un komunicēšanās konsultāciju aģentūra/Latvijas Autoru apvienība v. Konkurences padome (AKKA/LAA), ELCI:EU:C:2017:689 (Sept. 14, 2017). a case about allegations of excessive rates charged by national copyright management societies, the CJEU directed the referring court to undertake extensive cross-country comparisons and to account for national discrepancies in purchasing power.327Id. at ¶ 46. In MEO,328Case C-525/16, MEO–Serviços de Comunicações e Multimédia SA v. Autoridade da Concorrência (MEO), ELCI:EU:C:2018:270 (Apr. 19, 2018). a case about allegations of discriminatory prices charged by another collecting society towards TV channels, the CJEU declared that the proof of a “competitive disadvantage” written in article 102 TFEU required it to take into account “all the circumstances” relevant to the case.329Id. at ¶¶ 28, 31.

Even if factual investigations of a broad scope remain required in some areas of abuse law, the CJEU has meanwhile adopted evidentiary rules that accept the production of very speculative facts in support of allegations of abuse. In American law, direct evidence of anticompetitive effects (like a price increase) trumps indirect evidence (like a market power showing or a reduction in the number of competitors),330FTC v. Ind. Fedn of Dentists, 476 U.S. 447, 460–61 (1986); see also Brooke Grp. Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 232 (1993) (concluding that the documented existence of tacit collusion was not a circumstance sufficiently convincing to ground a theory of recoupment). and evidence of actual effects (like contemporary exit of a rival) trumps potential harms to competition (like a reduction in levels of innovation).331With the exception of the American Express case, where the Court required a market definition and market power evaluation in spite of observed price effects. Ohio v. Am. Express Co., 138 S. Ct. 2274, 2287 (2018). In EU law, indirect evidence obviates the need to study direct price and output effects, and an abuse can be found even if no actual effect is achieved provided that the practice “likely” harms competition.332See generally supra notes 313–316, 326–329 and accompanying text. This is different from a pure per se treatment. The nature of the restraint is not the source of liability. A conjecture about effects is what triggers antitrust exposition. We are in modified per se territory.

The same impressionistic tendencies characterize other aspects of abuse of dominance law. Whereas American law specifies thresholds of “dangerous probability” of success or “reasonable prospect” of injury in section 2 cases, a higher degree of abstraction is found in European jurisprudence.333Compare supra notes 219–228 and accompanying text, with supra note 302 and accompanying text. EU courts have declined to elaborate a threshold of likelihood of foreclosure effects, and a careful study of the cases provides, at best, imperfect pointers on this question.334See, e.g., Case C-23/14, Post Danmark A/S v. Konkurrencerådet, ELCI:EU:C:2015:651, ¶ 73 (Oct. 6, 2015) (“[F]ixing an appreciability (de minimis) threshold for the purposes of determining whether there is an abuse of a dominant position is not justified.”). The Court based the lack of an appreciability threshold in Hoffmann-La Roche & Co. Case 85/76, Hoffmann-La Roche & Co. AG v. Commn, 1979 E.C.R. 461, ¶ 123 (“[A]ny further weakening of the structure of competition may constitute an abuse of a dominant position.”).

Moreover, in US law many cases apply a “substantiality” screen.335See, e.g., United States v. Dentsply Int’l, Inc., 399 F.3d 181, 191 (3d Cir. 2005); United States v. Microsoft Corp., 253 F.3d 34, 69 (D.C. Cir. 2001). Sometimes, a substantial foreclosure share is required, like in situations where conduct is deployed in a market distinct from the market where monopoly power exists. The obvious example is tying cases. In other cases of strategic exclusion, the share of foreclosure need not be assessed, however, the lower courts have asked for a showing of foreclosure of a “substantial percentage of the available opportunities for [product] distribution”336Microsoft, 253 F.3d at 70–71. of “a substantial number of rivals” or of conduct that “severely restrict[s] the market’s ambit.”337Dentsply, 399 F.3d at 191. In European law, the courts have resisted every invitation to set a threshold of substantiality for abuse of dominance.338See Post Danmark, ELCI:EU:C:2015:651 at ¶ 73. Note, however, that Intel, at paragraph 139, hints at a requirement of appreciability (“share of the market covered by the challenged practice”) in relation to loyalty rebates, though it does not specify a set threshold. Case C-413/14 P, Intel Corp. v. Commn, ELCI:EU:C:2017:632, ¶ 139 (Sept. 6, 2017). Insignificant abuses can be hit. The open arms shown by the courts toward liability for abuses with a minimal foreclosure footprint appears to be based on a form of incipiency reasoning, whereby small effects presage higher exclusionary effects in the future or highly successful entry deterrence in the past.

Lastly, EU courts have consistently refused to construe the terms “abuse of a dominant position” literally.339See Case 6/72, Europemballage Corp. v. Commn, 1973 E.C.R. 215, ¶ 19. This has allowed findings of abuse where the market power conferred by the dominant position was not the instrument used to exclude competitors or exploit customers.340See id. The result has been an expansion of the law to business conduct very close to unfair trading or fraud.341See, e.g., Case C-457/10 P, AstraZeneca AB v. Commn, ELCI:EU:C:2012:770 (Dec. 6, 2012); Case T-814/17, Lietuvos geležinkeliai AB v. Comm’n, ECLI:EU:T:2020:545 (Nov. 18, 2020). More importantly for our purposes, the CJEU’s reticence to require an instrumental relation between dominance and abuse has relieved plaintiffs of the need to prove one key fact in article 102 TFEU cases, namely that of causation.342Europemballage, 1973 E.C.R. at ¶¶ 26–27.

Compared to US law, therefore, EU law requires the resolution of less factual uncertainty. And this trend also guides the criterion for the selection of antitrust rules. In Budapest Bank, the European court held that to be encompassed by a per se prohibition rule, “there must be sufficiently reliable and robust experience for the view to be taken that that agreement is, by its very nature, harmful to the proper functioning of competition.”343Case C-228/18, Gazdasági Versenyhivatal v. Budapest Bank Nyrt., ECLI:EU:C:2020:265 ¶ 76 (Apr. 2, 2020). The difference here between US and EU law might not be as sharp as previous differences, but a slightly more demanding standard of near certainty applies in US law. In Leegin, the Supreme Court required “considerable experience,” and added “it cannot be stated with any degree of confidence that resale price maintenance ‘always or almost always tend[s] to restrict competition and decrease output.’”344Leegin Creative Leather Prods. v. PSKS, Inc., 551 U.S. 877, 886, 894 (2007) (quoting Bus. Elecs. Corp. v. Sharp Elecs. Corp., 485 U.S. 717, 723 (1988)).

To close this comparative analysis, it is important to stress that EU law clearly did not evolve in the American direction of a division of labor favoring sectoral regulators. In EU law, the prior intervention of an expert institution does not exhaust the application of EU competition law. A case like Deutsche Telekom AG v. Commission345Case C-280/08 P, Deutsche Telekom AG v. Commn, 2010 E.C.R. I-9555. is typical of that evolution.346See Nicolas Petit, Circumscribing the Scope of EC Competition Law in Network Industries? A Comparative Approach to the US Supreme Court Ruling in the Trinko Case, 5 J. Network Indus. 347, 356–62 (2004). The case arose from allegations of abusive margin squeeze against the German telecommunications incumbent. The defendant argued that its rates had been approved by the national regulator thereby ruling out a possibility of margin squeeze. Contrary to the position of US antitrust law, the CJEU held that the fact that the sectoral regulator had not objected to the defendant’s rates did not in any way “absolve [it] from responsibility under [article 102 TFEU].”347See Deutsche Telekom, 2010 E.C.R. at ¶ 58. The Court said that the blessing of the sectoral regulator “left scope to adjust [the defendant] retail prices for end-user access service,” thereby leaving a possibility of anticompetitive behavior.348See id. at ¶ 66. In essence, this decision, and more recent ones like Slovak Telekom,349See Case C-165/19 P, Slovak Telekom a.s. v. Comm’n, ECLI:EU:C:2021:239, ¶ 58 (Mar. 25, 2021). The Court found that despite a sector specific regulatory intervention, the defendant retained “decision-making autonomy.” The Court noted, “It was in accordance with that decision-making autonomy that the appellant adopted the practices at issue,” and thus incurred potential liability for abusive access terms. Id. reveal that intervention of expert institutions does not eliminate all possibility of market abuses. This possibility requires falling back on a judicial policy of legal precaution and of possible subsidiary remediation by antitrust agencies.

4.     Summation

US antitrust doctrine stands on limits that condition antitrust enforcement on competitive realities, while modern EU law accepts intervention on the basis of competitive probabilities (see Table 1 below).

Table 1: Comparison of US and EU Antitrust Enforcement

Universe of Facts Extensive More limited
Type of Facts Actual effects and near certainty Likely effects and mere probability
Expert Regulatory Structure Primacy Subsidiarity

That EU competition law is stricter than US law is not disputed. In EU law, the facts matter less, and the limits are looser.

Because EU law allows the fact finders to establish liability without requiring as much evidence to back their allegations, the theories of harm to competition that the law uses to affirm liability can be applied without paying too much attention to confounding factors. Above-cost exclusionary pricing,350Case C-209/10, Post Danmark A/S v. Konkurrencerådet, ECLI:EU:C:2012:172, ¶ 27 (Mar. 27, 2012). tacit coordination,351See Nicolas Petit, The Oligopoly Problem in EU Competition Law, in Handbook on European Competition Law: Substantive Aspects 259, 259–349 (Ioannis Lianos & Damien Geradin eds., 2013). or predatory pricing without a possibility of recovery are good illustrations, even though the authorities have remained cautious to pursue such cases.

An experienced practitioner once half-jokingly told me that there is no such thing as a law of evidence before the EU courts, and that the only burden of production bearing on the agency is that it collects enough facts of bad actions to convince itself. Undeniably, the situation is very different in the US and, perhaps, in the national courts of Europe’s Member States.

C.     Why Different Attitudes?

Transatlantic differences in the legal treatment of uncertainty might reflect institutional, legislative, or developmental reasons.352In addition, educational reasons like differences in legal training may also have an impact in this area, and in particular the often-higher interdisciplinary component of higher education in the US compared to the EU. Let us review each possibility. A majority of US antitrust writers are of the opinion that institutions explain the evolution of doctrine and possible discrepancies with the EU. On this view, the US antitrust system’s decentralized nature allegedly creates a risk of judicial scrutiny of numerous business behaviors by non-expert decisionmakers like generalist judges.353Daniel A. Crane, LinkLine’s Institutional Suspicions, 2008 Cato Sup. Ct. Rev. 111, 124–25 (2008–2009) (“Juries are quirky, unpredictable, and emotional and inherently inferior to technocratic regulators.”). This has, according to some,354See id. motivated judicial elaboration of legal doctrines that take a cautious view towards antitrust enforcement and that impose great burdens on plaintiffs. On the other hand, EU law has been mostly applied by expert institutions. When the decisionmaker is knowledgeable, the law can be more probabilistic.355Or, to put things differently, we can tolerate more “analytical shortcuts” from decisionmakers that hold expert skills and knowledge. For use of the concept of “analytical shortcut” and related elaboration, see Andriani Kalintiri, Analytical Shortcuts in EU Competition Enforcement: Proxies, Premises, and Presumptions, J. Competition L. & Econ. 392 (2020). And centralized administrative institutions not bound by precedent create less risk of replication of errors than decentralized judicial ones governed by stare decisis.

Unfortunately, institutionalism explains too little. Over the years, the US and the EU have seen their antitrust institutions converge, with laws seeking to ease private enforcement in courts in the EU and procedural reforms seeking to protect defendants in the US.356Daniel A. Crane, Toward a Realistic Comparative Assessment of Private Antitrust Enforcement, in Reconciling Efficiency and Equity: A Global Challenge for Competition Policy 341, 342, 353 (Damien Gerard & Ioannis Lianos eds., 2019). Stare decisis has not received strict application in US antitrust law,357Leegin Creative Leather Prods., Inc. v. PSKS, Inc., 551 U.S. 877, 899 (2007) (“Stare decisis is not as significant in this case, however, because the issue before us is the scope of the Sherman Act . . . [and] the Court has treated the Sherman Act as a common-law statute.” (citing State Oil Co. v. Khan, 522 U.S. 3, 20 (1997))); Nw. Airlines, Inc. v. Transp. Workers, 451 U.S. 77, 98 n.42 (1981); Nat’l Soc’y of Pro. Eng’rs v. United States, 435 U.S. 679, 688 (1978); see also Daniel M. Tracer, Stare Decisis in Antitrust: Continuity, Economics, and the Common Law Statute, 12 DePaul Bus. & Com. L.J. 1, 2 (2013). and the CJEU has developed a policy of precedent in its case law manifesting its reticence to overrule past judgments.

The Supreme Court, itself, has cast a skeptical view toward institutionalism. This point was clearly expounded in Actavis, where the Court said that a concern of keeping cases away from the “time consuming, complex, and expensive . . . antitrust game” could not justify a rule of per se legality for patent settlements.358FTC v. Actavis, Inc., 570 U.S. 136, 153 (2013). And in Leegin, the Supreme Court refused the invitation to maintain RPM under a per se illegality rule holding that “[a]ny possible reduction in administrative costs cannot alone justify the Dr. Miles rule.”359Leegin, 551 U.S. at 895.

Another hypothesis is that the differences are rooted in textual provisions.360See, e.g., United States v. U.S. Steel Corp., 251 U.S. 417, 452 (1920). The language of the law might play a determinant role in the formation of antitrust limits. When antitrust statutes formulate broad standards like the Sherman Act, the legal system has a tendency to work towards the adoption of limits, either hard or soft. Recall the dicta in U.S. Steel where the Supreme Court held that the

command is necessarily submissive to the conditions which may exist and the usual powers of a court of equity to adapt its remedies to those conditions. In other words, it is not expected to enforce abstractions and do injury thereby, it may be, to the purpose of the law.361Id.

By contrast, antitrust provisions with narrower language like EU law might lead to softer limits or to a removal of limits against attempts to evade the law. The hypothesis is intriguing. If we follow wording, then EU law appears more prescriptive than US law. EU law embodies a threshold requirement of dominance, specifies “object” or “effect” as sources of competitive injury, focuses the prohibition of abuse on exploitation of market power, and lists examples of anticompetitive restraints.362See Antitrust, Euro. Comm’n, https://perma.cc/9DEE-8AJ6 (providing an overview of EU antitrust law). Such prescriptive legislation sends clear instructions to enforcers, allows economic agents to better understand the conditions for mobilization of the law, and removes discretion from courts about how the language of the statute should be constructed.363Donald J. Black, The Mobilization of Law, 2 J. Legal Stud. 125, 126 (1973) (“Law may be defined as . . . governmental social control.”). With such strong signals of legislative resolve, less limits should be tolerated in the judicial interpretation of the law.

However, the claim that EU law is more specified than US law is superficially attractive. Many of the concepts provided for in the TFEU are as open-ended as those used in the Sherman Act.364Some early cases that reached the Supreme Court were stock transfer schemes or the creation of holdings designed to evade the provisions of the law, implying that they were too formally strict and narrow. See, e.g., N. Sec. Co. v. United States, 193 U.S. 197, 295–97 (1904) (ruling against the railroad company stock holders, who had created a monopoly in violation of the Sherman Act). Moreover, US law has also introduced prescriptive acts that define forbidden practices and that have been completely ignored by the courts. For example, the Robinson-Patman Act of 1938 struck down as unlawful specific practices like discriminatory prices, services, and allowances in dealings between merchants.36515 U.S.C. §§ 13, 13(a), 13(b), 21(a); see also Cong. Rsch. Serv., R40146, Discriminatory Pricing and the Robinson-Patman Act: Brief Background and Analysis 1 (2009). In practice, the prescriptions of the act were de facto read out by judicial inaction.366FTC v. Fred Meyer, Inc., 390 U.S. 341, 349 (1968) (“[The law] by no means represent[s] an exemplar of legislative clarity.”).

One last hypothesis might be that EU law today is where US law was after sixty years of existence. This would place EU law approximately in the 1950s, a period when US antitrust law was rife with structural presumptions, and where a widespread belief was that any reduction in the number of competitors in a market would lead to diminished competition. This is what Americans, who invented antitrust law, have liked to think.367See Rahl, supra note 161, at 985–86 (“Still perhaps laboring under the partially wrong notion that we invented antitrust law and that foreign antitrust is simply our ‘export, Americans tend to look more with the idea of seeing what and how our European friends are doing than what we can learn from them.”). In a word, Europe would still be learning.368Though note that Professor Spencer Webber Waller has argued that the United States lags behind the rest of the world. See Spencer Weber Waller, The Omega Man or the Isolation of U.S. Antitrust Law, 52 Conn. L. Rev. 123, 128 (2020).

One apparent sign of higher maturity of US law might be observed in the output of the antitrust system.369Lawrence M. Friedman, The Law and Society Movement, 38 Stan. L. Rev. 763, 777 (1986) (discussing the role of economics in the law and society movement). As a matter of fact, US antitrust law tends to produce a lower number of decisions, orders, and remedies than the EU, yet the statements of the law of US courts tend to be of higher formal quality. Simply put, the case law tends to speak more intelligibly.370This might actually explain the strange situation in which the EC reportedly talks about fears of “losing” cases in court, while the EU courts’ case law is more plaintiff- than defendant-friendly. Doctrine also embodies more economic references. For right or wrong, Supreme Court decisions have made extensive references to the economics of information asymmetries,371Cal. Dental Ass’n v. FTC, 526 U.S. 756, 772 (1999). error costs,372Verizon Commc’ns, Inc. v. L. Offs. of Curtis V. Trinko, LLP, 540 U.S. 398, 414 (2004). free riding,373Leegin Creative Leather Prods., Inc. v. PSKS, Inc., 551 U.S. 887, 890 (2007). lock-in,374Eastman Kodak Co. v. Image Tech. Servs., Inc., 504 U.S. 451, 472–73 (1992). and multisided platforms.375Ohio v. Am. Express Co., 138 S. Ct. 2274, 2280–81 (2018). In some cases, like Brooke Group, economic theory (suggesting a difficulty of collusion) might have been used to weaken documented facts (patterns of lockstep oligopoly pricing).376David F. Shores, Law, Facts and Market Realities in Antitrust Cases After Brooke and Kodak, 48 SMU L. Rev. 1835, 1852 (1995). In other cases, like Kodak, economic theory (suggesting impossible aftermarket monopoly in presence of primary market competition) was rejected by confrontation to empirical facts (consumer lock-in).377Kodak, 504 U.S. at 472–73.

There is, however, a key problem with the idea that the lower limitations imposed on European antitrust might reflect less maturity, specifically economic maturity. EU law started in a historical context in which the economic problems of monopoly power (and collusion) were far better understood than when the Sherman Act was adopted. As economist George Stigler once observed, the passage of the Sherman Act “did not attract the attention of any economists,” and a rise in popularity with economists only occurred in the 1950s and 1960s.378Stigler, supra note 30, at 2, 4. EU competition policy thus benefited from a far richer intellectual context and was almost on equal footing with the Sherman Act when it was adopted. Moreover, the 1950s were marked by an intense transfer of antitrust know-how from the US to the EU in the context of the Marshall Plan, the decartelization of Germany, and of concerns about national discrimination by Member States against domestic firms. Against this backdrop, if EU law relies today less on economics, or tends to produce more ambiguous statements of the law, this might just have to do with a different tradition of accountability in judicial decision-making. This reliance is likely deepened all the more when proceedings are discretionary and judges tend to respect, and protect, what is perceived as agencies’ superior analytical capabilities (hence, the well-known policy of deference toward agencies’ complex economic analyses).379The many languages used in the practice of European law might also interfere.

None of the reasons examined appear to, alone, explain the fundamental differences between the limits of US and EU antitrust laws. A more realistic analysis is that multiple causes have shaped the evolution of the case law in both regimes. If this is correct, then this is another element to add to a theory of the non-ideological limits of antitrust.

III.     The Optimal Limits of Antitrust

A.     Legal v. Market Processes

This Article has shown that pragmatic, not ideological, considerations have shaped the design of substantive antitrust doctrine, yet a normative question remains: what are the optimal limits of antitrust? Or more precisely, should substantive antitrust law lean toward intervention or non-intervention?

Antitrust’s limits reflect an attitude towards uncertainty. The limits are set by more or less conscious and informed opinions about institutions, markets, and technology’s relative ability to improve the public interest in a context of uncertainty.

In three papers in the 1980s, Easterbrook argued that the calibration of antitrust’s limits ought to be based on an empirical comparison of the efficiency of markets as opposed to institutions at correcting market power failures.380See Limits, supra note 1, at 24–25; Workable Antitrust Policy, supra note 96, at 1700–01; Frank H. Easterbrook, Allocating Antitrust Decisionmaking Tasks, 76 Geo. L.J. 305, 320 (1987) [hereinafter Allocating Antitrust Decisionmaking]. His intuition has remained strong, but his application was wrong. Easterbrook declared that markets systematically outcompete institutions at dissolving monopoly power unlawfully acquired, maintained, or strengthened.381See Limits, supra note 1, at 6–7. And if, in some cases, both institutions and markets fail to timely correct market power, a situation known as a Type 2 error, institutions have an additional, fatal, deficiency.382See id. at 2–3. Institutions occasionally make Type 1 errors by condemning procompetitive conduct, and, unlike Type 2 errors, the costs of these Type 1 errors are irreversible and reproduced on third parties—who may forego procompetitive conduct as a result of the error.383See id. at 5–6. Contrast this with market processes. If a business practice that injures welfare persists in markets, then it probably has a good, procompetitive explanation otherwise it would disappear under attack from enhanced competition.384George L. Priest, The Limits of Antitrust and the Chicago School Tradition, 6 J. Competition L. & Econ. 1, 8 (2010). This difference led Easterbrook to contend without nuance that “markets work better than judges at penalizing inappropriate conduct.”385Limits, supra note 1, at 24.

Easterbrook’s framework has fallen into disrepute. But it directs attention towards a good empirical procedure to set the limits of antitrust: if both processes commit Type 2 errors, but institutions distinctively substitute some Type 1 errors for Type 2 errors, then the relevant question for society is whether we prefer a mix of Type 1 and Type 2 errors by institutions (and if so, what mix) to a homogeneous set of Type 2 errors created by courts and markets. Put more simply, which error is more costly: the truly distinctive judicial error of forbidden competition (Type 1 error) or the undistinctive judicial and market error of excused monopoly (Type 2 error)? A close study of Easterbrook’s framework suggests that the answer to the question of the costs of judicial versus market errors is essentially a function of the duration, diffusion, and frequency of error under both processes.

And because business practices, institutions, and markets change, the weight of these variables might evolve over time. A question that naturally arises is thus the following: given changes in the economic environment, should we reconsider antitrust’s current limits? Easterbrook himself conceded that “novel” practices required a fresh look, not a mechanical application of precedent.386Allocating Antitrust Decisionmaking, supra note 380, at 307 (“[N]ovel practices . . . require substantial independent analysis rather than simple application of a precedent ready-made for the case.”). Remembering that Easterbrook’s framework was, at heart, an empirical compass, not just an ideological one, allows the antitrust policy conversation to operate on a dispassionate analytical level, insulated from political polarization.

To that end, this Article now describes the three determinants of error costs and looks at them empirically in the context of the modern economy.

B.     What Makes Errors Costly?

There seem to be three mechanics of harm that make errors costly in Easterbrook’s framework. First, there is the duration of harm. Harm is durable when correction of error is limited.387Harm can be perpetual if it is irreversible. Second, there is the diffusion of harm. Harm spreads when error replicates. Third, there is the frequency of harm. Harm is frequent when errors are likely. So far, so good.388Note that together, these three mechanics add up to give the magnitude of harm. Note also that Easterbrook never explicitly established these three mechanics or determinants of harms, but that we derive these three categories from our own reading of Easterbrook’s framework. See infra notes 389–397 and accompanying text.

The next step is more problematic. For each determinant, Easterbrook established an inequality that disadvantaged judicial over market errors. Judicial errors are durable, because of the “slow and costly” judicial process.389See Limits, supra note 1, at 29. There is no correction of judicial error other than occasional overruling by future courts. This is slower than markets which self-correct even if “this long run may be a long time coming.”390Id. at 2. Note that in another paper, Easterbrook said that the law moved too fast, and complained that “courts act in advance of the arrival of explanation.” See Allocating Antitrust Decisionmaking, supra note 380, at 308. What he meant, though, is that judges “move slower than markets but faster than the economics profession,” which takes time to produce explanations that require long-term scientific study. Id.

The diffusion of judicial errors is high due to “stare decisis.”391Limits, supra note 1, at 2. Harms spread to other firms who are either sanctioned when they adopt wrongly condemned practices, nudged to adopt less efficient conduct, or deterred from market activity.392Id. (“Any other firm that uses the condemned practice faces sanctions in the name of stare decisis . . . .”). By contrast, market errors are not compounded by replication but are corrected by competition. Easterbrook explains that a “firm collecting an overcharge ultimately loses sales to firms” doing something different, that is “charging the competitive price.”393Id. at 11.

Further, he argues that the likelihood of harm is compounded in virtue of courts’ “ignorance” and insufficient “wisdom.”394Id. at 7, 29. Confronted with a novel business practice not covered by stare decisis, the probability of erroneous condemnation is substantial. Courts are typically more blind to efficiency than monopoly, because the legal system is trial friendly and is marked by “ceaseless discovery” and other “deadly” traps for defendants.395Id. at 12. Easterbrook spoke of the “inhospitality” of the judicial system.396Id. at 4–9. By contrast, markets presumably select neutrally, and the survival of durable and complex practices of which we know little “indicates that they serve some function.”397Workable Antitrust Policy, supra note 96, at 1701.

When comparing these pitfalls with market processes, Easterbrook found that the duration, diffusion, and frequency of error was systematically higher in judicial processes (see Table 2).398Allocating Antitrust Decisionmaking, supra note 380, at 309 (“Law has a comparative advantage, then, when legal processes are rapid, when false positives are few or quickly corrected, and when markets are sluggish about correct false negatives. These criteria . . . are rarely met for novel business practices, those courts are encountering for the first time. The rate of false positives may be particularly high because cases arrive in court ahead of explanations for the practices.”). A good dose of rhetoric was used as subterfuge for passing off concealed simplifications to readers. On close study, three tricks undergirded Easterbrook’s finding of inequality. First, a view that markets do not differ in fundamental ways, so that monopoly power tends to attract entry, cartels are in general unstable, and mergers are almost always efficiency enhancing.399Hovenkamp & Scott Morton, supra note 7, at 1848. Second, an assumption that technology is mature, knowledge is widely available, and innovation spillovers are perfectly appropriable. Third, a description of antitrust institutions as mostly judicial, homogeneous, and immutable.

Table 2: Magnitude of Harm in Easterbrook’s Model Judicial vs. Market Processes

Mechanics of Harm Judicial Processes Market Processes
Duration of Harm High (slow and costly) Low (entry)
Diffusion of Harm High (stare decisis) Low (price competition)
Likelihood of Harm High (ignorance) Low (neutrality)

Now any objective observer can understand that these are not average properties, at least not today in the twenty-first century. The idea that judicial errors cannot be corrected easily, and that markets perpetually self-correct by eroding monopoly was not based on facts but on faith. And even if the world of the 1980s still existed today, antitrust law considers the marginal case. It makes sense to bring cases to courts, even wrong ones, because this helps reveal the contours of the law. Moreover, as economist and Professor Michael Scherer once observed, a policy of antitrust limits that would perfectly eliminate Type 1 errors by “er[ring] on the side of questionable practices” is weird, because these errors are at least visible compared to Type 2 errors.400F. M. Scherer, Conservative Economics and Antitrust: A Variety of Influences, in How the Chicago School Overshot the Mark: The Effect of Conservative Economic Analysis on U.S. Antitrust 30 (Robert Pitofsky ed., 2008). Those concerned with the inefficiency of the legal system should be willing to see Type 1 errors if only to correct them and allow some degree of intervention when institutions fail to correct Type 2 errors.

The inevitable question that arises again is where to put the cursor between expansion or limitation of antitrust law.

C.     Mechanics of Harm in the Modern Era

The issue here is not about whether there must be a conservative or progressive bias in our antitrust system. The issue here is about whether the doctrinal limits should be softened or hardened. And this question requires a reconsideration of the mechanics of harms caused by errors inherent in judicial versus market processes. The issue ought to be addressed by reference to empirical realities, not beliefs based on theories. Put simply, adjustments to antitrust doctrine should be discussed in light of the properties of monopoly power in the global and digital economy.

Consider the mechanics of harm separately. The duration of judicial harms is certainly not a given. Observing in Microsoft that “a mere three years” separated the first complaints from the appellate decision, the Court of Appeals for the DC Circuit said that the litigation timeline was “hardly problematic” for “a case of this magnitude and complexity.”401United States v. Microsoft Corp., 253 F.3d 34, 48 (D.C. Cir. 2001). By conservative standards, three years remains the short term, even in the modern economy.402Jessica Guynn, Google Justice Department Antitrust Lawsuit Explained: This is What it Means for You, USA Today (Oct. 20, 2020, 8:31 PM), https://perma.cc/TPX4-U7YK (explaining that the antitrust lawsuit against Google could take years, with a trial lasting around twelve to eighteen months and appeals adding years). Easterbrook, himself, conceded that “entry itself may take a while”; that it is rare for firms to take less than two years to enter an industry; and that five years was more realistic.403Limits, supra note 1, at 33. Against this backdrop, an opportunity of legal correction on appeal within three years appears to beat the market process by a margin.404Recalling the role of appeals in the correction of wrong legal rules, see United States v. Forness, 125 F.2d 928, 942 (2d Cir. 1942) (“An impeccably ‘right’ legal rule applied to the ‘wrong’ facts yields a decision which is as faulty as one which results from the application of the ‘wrong’ legal rule to the ‘right’ facts. The latter type of error, indeed, can be corrected on appeal.”).

The same doubts arise about the diffusion of harms. Antitrust courts do not follow stare decisis zealously.405See Tracer, supra note 357, at 3 (“[S]cholars now take it for granted that stare decisis has a somewhat modified application in the area of antitrust. Though certain notable exceptions have persisted in the case law, scholars and practitioners are no longer sure that any particular rule or doctrine will survive the next grant of certiorari.”). Though, note that in American Express, the court embraced lots of legal principles that had been applied in very different circumstances to dismiss the case. See Ohio v. Am. Express Co., 138 S. Ct. 2274, 2283–90 (2018). For example, lower courts have drastically narrowed down Supreme Court precedents like Kodak.406See David A.J. Goldfine & Kenneth M. Vorrasi, The Fall of the Kodak Aftermarket Doctrine: Dying a Slow Death in the Lower Courts, 72 Antitrust L.J. 209, 209 (2004) (“[F]ederal district courts and federal courts of appeal have bent over backwards to construe Kodak as narrowly as possible . . . to the point where it is simply no longer an effective weapon for antitrust plaintiffs.”). And sophisticated findings of lower courts in cases of anticompetitive strategic interaction like United States v. Dentsply International Inc.,407399 F.3d 181 (3d Cir. 2005). Conwood Co. v. United States Tobacco Corp.,408290 F.3d 768 (6th Cir. 2002). or LePage’s Inc. v. 3M409324 F.3d 141 (3d Cir. 2003) (en banc). show that liability can be established on specific grounds, without creating an identifiable category of cases that would disincentivize too much market competition.410This is why a concern of “diffusion” by the deterrent impact of the decision that embodied the error is not as material as it may look. Firms who know that courts do not stick automatically to precedent will not necessarily avoid new litigation and might take their chances at testing the boundaries of the erroneous decision. Courts are not automatons that follow any judicial or congressional command.411Note that section 5 of the FTC Act has historically been read narrowly for reasons of clarity, predictability, and, perhaps, distrust of government. And agency rulemaking and congressional legislation can override bad case law. These lower diffusion risks in the legal system must be compared with the nature of competition in the digital economy. Errors of unchecked monopoly power might actually diffuse more quickly, at least for a subset of large digital firms that expand, grow, and diversify by virtue or economies of scale, economies of scope, and network effects.412See generally Nicolas Petit, Big Tech and the Digital Economy: The Moligopoly Scenario (2020). At the same time, a credible claim might be made that monopoly concentration is efficient in digital markets (see Table 3).

Table 3: Correction of Harm in Easterbrook’s Model Judicial vs. Market Processes

Mechanics of Harm Judicial Processes Market Processes
Duration of Harm 3 years (duration to appeal) 2–5 years (entry)
Diffusion of Harm Low (weak stare decisis) High (growth, diversification)
Likelihood of Harm Mixed (learning) Low (neutrality)

Compared to the two others, the likelihood of errors might be higher in the digital economy. New, complex, and dynamic business conduct creates substantial challenges for antitrust courts. Innovative value capture strategies do not follow conventional explanations. And competition in digital markets involves business model experimentation. Easterbrook’s statement that “[i]t is useful for many purposes to think of market behavior as random. Firms try dozens of practices. Most of them are flops, and the firms must try something else or disappear” appears as true today in the modern economy as it was in the eighties.413Limits, supra note 1, at 5. Moreover, business conduct might not be as easily justifiable. Surrounded by noisy information, and unable to proffer a clean justification, defendants will face even more “hostile reactions.”414Id. at 8. This, however, might not generally justify falling back on market processes for the reasons that were just outlined. Recall that duration of harm is more limited, and diffusion of harm can be moderated, compared to the world sketched by Easterbrook. If this is the case, then a legal process of incremental and experimental common law development might create errors with tolerable harms, especially in light of the learning benefits resulting from the succession of cases, both right and wrong.

The above remarks set aside the role of administrative institutions. The expert bodies that operate outside the courts can assist in the educative process and development of antitrust law by facilitating the implementation of the latest knowledge in law and economics.

D.     Summation

The questions that we should ask ourselves in deciding whether to select a more realistic model of antitrust limits like the US or a more probabilistic system like the EU force us to reconsider the mechanics of harm that underpinned Easterbrook’s framework.

The legal and market processes that characterize the modern era do not suggest that markets outcompete courts at the game of monopoly power dissolution. US antitrust doctrine may thus benefit from a slight relaxation. At the same time, there is no point in a full convergence on the EU model. In Europe the law is still enforced very centrally, in spite of growing private enforcement. When antitrust uses legal processes that are centralized, expert, and discretionary, the law can be more probabilistic. Errors impose low short-term competitive harms because the law sends ad hoc signals without precedential value.415And quite unreadable statements. However, the law’s “ad hocism” creates opportunities for gaming, moral hazards, and politicization which undermine long-term trust in the economic system.


This Article has shown that the limits of antitrust are inherently practical, and that they correlate with institutional, formal, and cultural features of antitrust systems. The modern critics thus advance a wrong diagnosis with the wrong remedies. An understanding of the inherent practical demands placed on decisionmakers is a necessary condition for antitrust law reform.

Courts and agencies must apply abstract antitrust provisions to imperfect facts. Depending on each antitrust regime’s preferences in situations of uncertainty, the courts have set limits at different ends of the spectrum. In the US, the case law has preferred to require courts and agencies to intervene on the basis of an evaluation as close as possible to reality. At the other end of the spectrum, the EU has adopted rules and standards that allow intervention on the basis of more probabilistic evaluations of the facts.

All legal arrangements obsolesce. Antitrust is no exception. Economic, legal, and technological change require reconsideration of established doctrine. Allowing marginally more intervention by decisionmakers might be the price to pay to derive empirical learning from the cases, increase our collective stock of knowledge, and answer if, and how, antitrust limits deserve renovation.

Share this article