In spite of the many successes in the fight against cartels, enforcement is likely to be suboptimal because competition authorities systematically underdeter cartel formation. The reason is simple: convicted cartels are observed, deterred cartels are not. Consequently, many decisions by a competition authority will be driven by a desire to convict cartels – as that is an observable measure upon which to assess performance – which will be to the neglect of deterring cartels. Underdeterrence manifests itself most heavily through underprosecution (some types of collusion cases are not pursued) and underpenalization (penalties are not set to deter future collusion). Some proposed policy changes are offered for enhancing deterrence.
By Joseph E. Harrington, Jr.*
“No modern development in antitrust law is more striking than the global acceptance of a norm that condemns cartels as the market’s most dangerous competitive vice [but] is modern cartel enforcement attaining its deterrence goals?” William Kovacic (OECD Conference, October 2013), former Chair of the U.S. Federal Trade Commission.
I. INTRODUCTION
The President of Bumble Bee Foods is sentenced to 40 months in prison and StarKist is levied a $100 million fine by the U.S. Department of Justice for agreeing to reduce the size of a can of tuna and raise prices. Six major truck manufacturers are fined approximately $4 billion dollars by the European Commission for privately meeting to exchange and agree on list prices. These cases, and many others, illustrate that major corporations continue to collude and even in jurisdictions with a record of aggressive enforcement.
Of course, continued cartel formation does not imply enforcement is suboptimal. Taking into account enforcement and error costs (i.e. direct and indirect costs from wrongful prosecutions), the existence of some cartels is socially optimal and to be expected. Nevertheless, I contend that enforcement is likely to be suboptimal because enforcers – both public and private – systematically underdeter cartel formation. In exploring this issue, this paper addresses three questions: 1) why would there be underdeterrence?; 2) how does underdeterrence manifest itself?; and 3) what can be done to enhance deterrence? While the discussion pertains to any competition authority (“CA”), our analysis will largely be conducted from the perspective of the Antitrust Division of the U.S. Department of Justice (“DOJ”).
II. WHY WOULD THERE BE UNDERDETERRENCE?
There are three facets to cartel enforcement: detection, prosecution, and penalization. Let us consider how the decisions of a CA impacts each of them.
Many of a CA’s cases are either brought to them (such as through a customer complaint), accidently found (such as during a merger review), or sourced from other enforcers (specifically, private litigants and other competition authorities). A CA has two general strategies should it want to be more proactive in building its caseload. The first is to adopt programs that incentivizes others to report suspected cartel episodes, such as leniency programs and whistleblower rewards.[1] The second approach is for the CA itself to actively look for collusion in the marketplace.
Given a suspected cartel has been identified, a CA next decides whether to pursue an investigation and ultimately whether to prosecute. Not all suspected cartels are investigated as evidenced by the many credible cases that the DOJ leaves to private litigants.[2] If a CA does choose to prosecute and obtains a conviction, it then decides the penalty to seek to impose. The associated calculus involves weighing off the benefit from going after a larger penalty against the cost and risk of additional litigation. A CA is often faced with the choice between a penalty that will induce defendants to settle and a harsher penalty that may require taking the case to trial.
In making these decisions that affect detection, prosecution, and penalization, it would be desirable for a CA to do what is best for society; that is, maximize social (or consumer) welfare while taking account of the cost to taxpayers. However, such an expectation is unrealistic and unfair. Those employed at a CA are as much flesh and blood as the rest of us. While they may take account of social goals, they also have personal aspirations. Entering their calculus could be the “warm glow” from putting a price fixer behind bars or the desire to exert less effort and take on less risk from settling a case rather than going to trial. They may pursue cases to contribute to the reputation and resources of the agency. Or a CA official may consider career concerns as they act to improve their performance as viewed by those who might promote them or hire them in the private sector.
Rather than speculate about the incentives of those who work at a CA, let me make two fairly indisputable points. First, the interests of a CA do not fully align with the interests of society or the goal of competition law. To some degree, the personal considerations of a CA’s employees are relevant to their decisions. Second, a CA strives to enhance its measurable performance because, generally, rewards are tied to what is observed. Some observable metrics include the conviction rate (that is, fraction of investigations that lead to a conviction or guilty plea), number of convicted cartels, and amount of penalties as measured by days in jail and fines collected.
We have now arrived at the challenge when it comes to deterrence: convicted cartels are observed, deterred cartels are not (or at least not easily). Consequently, we expect many decisions by a CA to be driven by a desire to convict cartels – as that is an observable measure upon which a CA’s employees can be rewarded – which will be to the neglect of deterring of cartels. Private enforcers are even less incentivized to value deterrence. Plaintiffs and law firms are motivated to collect damages from existing cartels (as well as shut them down), and not to deter the formation of future cartels. Indeed, fewer cartels means fewer cases so law firms have little interest in promoting deterrence.
The central claim of this paper is that anti-cartel enforcement is suboptimal because the system is designed for public and private enforcers to insufficiently care about deterrence.
III. HOW DOES UNDERDETERRENCE MANIFEST ITSELF?
Having explained in the abstract why a CA has inadequate incentives to deter collusion, let me be concrete in how it might manifest itself in a CA’s actions. But before doing so, two caveats are warranted. What may appear to be suboptimal conduct from the perspective of deterrence could reflect conduct that is optimal in light of scarce resources. For example, some detection activities may not be performed and some cases not pursued because of a lack of resources, not because of insufficient incentives to deter. However, I don’t think a shortage of resources is the full story for, as I will describe, there are some low-cost actions not being taken. The second caveat is that some apparently suboptimal conduct by public enforcers could be optimal when taking into account the presence of private enforcers. For example, a CA may leave some cases for harmed customers to litigate or seek a less severe penalty because of customer damages. As is explained below, I do not believe this is sufficient to rationalize the conduct I am about to describe.
Starting with a CA’s actions relevant to detection, it is not immediately clear that a CA’s incentives are weak when it comes to discovering cartels. Given the number of convicted cartels is an observable metric, a CA could choose to be very active in finding cases. But a CA may instead be focused on the conviction rate which could lead them to be less active in detection given that fewer cases allow more resources to be brought to bear on each case. Focusing on the DOJ, its leniency program has surely been effective in inducing the self-reporting of cartels though concerns have been expressed of excessive reliance on it for discovering cases.[3] Putting aside leniency applications, the DOJ has generally not engaged in the enterprise of discovering cartels. Notably, the DOJ has chosen not to perform a potentially productive detection activity: cartel screening.[4]
Cartel screening is the examination of market data for collusion. Any investigation begins with some piece of evidence supporting the hypothesis that a market harbors a cartel and screening can deliver that evidence. As some examples, screening has discovered cartels in markets for cement (South Africa), subway construction (Korea), and retail gasoline (Brazil).[5] Studies have shown that bidding rings can be detected from bid data using algorithms based on supervised machine learning.[6] To my knowledge, the DOJ has shown little interest in screening (though the Procurement Strike Force appears to be a recent departure), while many other CAs have engaged in it to varying degrees.[7] One possible reason for why screening may be underutilized is that it may not identify the easiest cases to prosecute for it starts with economic evidence. However, it is worth noting that screening is complementary to a leniency program. An investigation initiated by screening can induce firms to apply for leniency, as occurred with the cement cartel in South Africa.[8]
Of greater concern than underdetection is underprosecution. In their case selection, CAs may be focusing on winning cases and shutting down existing cartels, while avoiding those cases that would do more to deter future collusion. On this issue, three points will be made. First, the DOJ has largely focused on cases involving explicit collusion. Second, the DOJ has not been prosecuting attempts to explicitly collude. Third, the use of consent decrees may be a frugal method for shutting down a cartel but they do not necessarily serve the goal of deterrence.
Perhaps it is due to the leniency program delivering an ample supply of explicit collusion cases but it is rare for the DOJ to take on cases of tacit collusion.[9] For example, there have been numerous episodes in which firms colluded through public announcements such as earnings calls. The DOJ has declined to prosecute, as they are left to private litigants to pursue.[10] Recognizing there is a resource constraint, so not all cases can be pursued, if the goal is solely to shut down cartels then avoiding these challenging tacit collusion cases and focusing on explicit collusion cases may make sense. However, the goal of deterrence can justify prosecuting tacit collusion. If firms know certain practices are not prosecuted by a CA, it gives them license to pursue those practices. This means the conviction of one case of tacit collusion could deter many future episodes involving similar practices. This concern about case selection is magnified by the success of the leniency program in detecting and prosecuting explicit collusion for it may be causing some firms to instead pursue tacit collusion. If tacit collusion is becoming more prevalent in practice, underdeterrence due to underprosecution is then an increasing concern.
An “invitation to collude” case involves a firm proposing to a competitor that they form an agreement to restrain trade but where the agreement is not consummated; typically, because the competitor declined the offer. It has been almost 40 years since the DOJ last prosecuted an “invitation to collude” case under Section 2 of the Sherman Act. In that case against American Airlines, the Fifth Circuit Court held that “an agreement is not an absolute prerequisite for the offense of attempted joint monopolization.”[11] Subsequently, invitation to collude cases have been exclusively handled by the Federal Trade Commission as an unfair practice under Section 5.[12] Some of those FTC cases have been quite egregious. For example, a supplier of barcodes sent a series of emails to two competitors expressing a plan to coordinate price increases.[13]
The problem with leaving the FTC to prosecute these cases is that they do not have an arsenal of penalties to draw upon. The cost of being caught by the FTC for inviting a conspiracy is basically legal fees. The Sherman Act’s Section 2 has greater deterrent potential with its threat of criminal penalties and corporate and individual fines. It would seem appropriate and should be uncontroversial for firms to be severely punished for attempting to form an agreement to restrain trade. That the other firms did not oblige to go along should not get a firm and their executives off the hook.
Finally, underprosecution could be reflected in cases ending with a consent decree rather than a judicial opinion. From the perspective of stopping a particular instance of a harmful practice, a consent decree is eminently sensible when compared to going to trial and possibly losing. However, from the perspective of deterrence, a successful judicial ruling could deter multiple future cartels. As a case in point, Container was important for precedent for it established the rule of reason for a class of information exchanges.[14] That would not have happened if the DOJ had issued a consent decree. In contrast, a consent decree with several airlines in 1994 has left us 28 years later still not knowing when advance price announcements are a violation of Section 1.[15] Returning to the absence of public announcement cases by the DOJ, there has been no jurisprudence providing guidance as to the public communications that a firm should avoid.
As commented by legal scholar and judge Frank Easterbrook: “Deterrence is … the first, and probably the only goal of antitrust penalties.”[16] Accordingly, a competition authority should foremost consider deterrence when deciding on the appropriate penalty. As incarceration must be abhorrent for corporate executives, the DOJ is certainly enhancing deterrence when it pursues criminal investigations and exacts prison sentences. At the same time, corporate fines (even when combined with customer damages) could be insufficient for collusion to be ex post unprofitable much less ex ante unprofitable. The vitamins case is illustrative of this concern. Hoffman LaRoche received a fine of $500 million which at the time was the highest fine in the DOJ’s history.
However, by the DOJ’s own guidelines, the recommended fine was in the range of $1.3 billion to $2.6 billion, which meant the $500 million fine was 40 percent of the minimum recommended fine. Nor did damages sufficiently increase the financial penalty. It is estimated that members of the vitamins cartel incurred U.S. fines and damages that were only 86 percent of U.S. overcharges[17] (which is a reasonable approximation of the incremental profits from collusion). In practice, customer damages have generally proven inadequate to deliver cartel-deterring financial penalties. A recent study found for the median case that customers recovered only 52 percent of damages and in only 20 percent of cases were at least single damages obtained.[18] Given a plausible probability that executives attach to being caught and convicted, the best evidence suggests that it remains highly profitable to form a cartel.
The profitability of collusion also means that boards of directors and senior managers will not be incentivized to encourage their employees to comply with antitrust laws. That a company has adopted a compliance program need not reflect a genuine desire to dissuade employees from colluding. The real test is whether it has in place the same monitoring practices for preventing and detecting collusion as they do with regards to accounting fraud, embezzlement, and other financial crimes.[19] There is no evidence that is the case. The problem of underdeterrence is exacerbated when boards of directors and senior managers do not see collusion as a problem for shareholders and thus do not take adequate action to prevent it.
IV. WHAT CAN BE DONE TO PROMOTE DETERRENCE?
A response to the problem of underdeterrence has two layers. First, what actions can be taken to enhance deterrence? Second, how can a CA be incentivized to take those actions? From our preceding discussion, much of the answer to the first (and easier) question is transparent. A CA can increase deterrence by engaging in cartel screening and thereby making it more likely a cartel will be detected. Screening is a relatively low-cost activity; requiring a few economists and some research assistants to collect and analyze data.[20]
In the area of prosecution, the DOJ can take on cases other than explicit collusion. It has long been recognized that a Section 1 violation does not require an express agreement. A judiciously chosen case could aid in deterring cartels and more clearly defining the boundaries of illegal conduct. In particular, a case involving public announcements could aid in deterring such activities and provide guidance to firms as to what they should not say publicly. The DOJ could also aggressively pursue invitation to collude cases under Section 2 and penalize firms for attempting to form an illegal agreement.
While incarceration has effectively punished individuals, more needs to be done to penalize shareholders. Within the limits allowed by the law, fines should be set with the goal of making cartel formation a regretful act which means making it credibly ex ante unprofitable. A necessary condition for that to be so is that fines and damages are several times the incremental profit from collusion. CAs can also be encouraged to consider the use of structural remedies where appropriate. A structural remedy has cartel members sell productive assets, such as capacity, to other firms in order to make future collusion less likely. Compared to existing corporate penalties of government fines and customer damages, divestiture is more of a deterrent under certain conditions, more effective at compensating those consumers harmed, and corrective in reducing the likelihood of recidivism and preventing post-cartel tacit collusion.[21]
Of course, if the problem is one of incentives, identifying deterrence-enhancing actions is pointless if a CA is not motivated to take them. Offering a solution to address that challenge will require an analysis that is beyond the scope of this paper. However, one practical suggestion is for the DOJ to make “deterrence” an explicit consideration in their decisions. When deciding how to allocate resources (do we assign an economist to screening?), whether to prosecute a cartel (should we take on a case involving public announcements?), and the penalty to pursue (how high to make the corporate fine?), it can consider the implications for future cartel formation.
V. CONCLUDING REMARDS
The DOJ is an aggressive and effective actor in the fight against cartels, as are the competition authorities in many other jurisdictions. It has a long list of accomplishments including the leniency program and obtaining multi-year prison sentences. More recently, it has taken on buyer cartels in labor markets and created the Procurement Collusion Strike Force. The DOJ has an exceptional record in prosecuting explicit collusion with the aid of leniency recipients. However, its record is less impressive in detecting cartels (outside of leniency applications), prosecuting tacit collusion, developing precedents to define the boundary of lawful and unlawful conduct, designing new penalties, and, more broadly, deterring cartel formation. My hope is that this paper may offer some constructive suggestions for the DOJ and other competition authorities to develop more effective anti-cartel enforcement with the specific objective of deterring collusion.
* Patrick T. Harker Professor, Department of Business Economics & Public Policy, The Wharton School, University of Pennsylvania, Philadelphia, PA 19104 U.S.A. harrij@wharton.upenn.edu.
[1] Leniency programs provide reduced penalties to cartel participants coming forward, while whistleblower rewards provide monetary payment to non-participants from reporting a cartel.
[2] In a study of 60 large private antitrust suits, 40 percent of them were initiated by plaintiffs. Robert H. Lande & Joshua P. Davis, “Benefits from Private Antitrust Enforcement: An Analysis of Forty Cases,” University of San Francisco Law Review, 42 (2011), 879-918.
[3] “I am worried that the success of the Leniency Program combined with budget constraints that your Division faces will in effect give you incentives to pursue only the companies that come forward. . . . [A]s I know from personal experience, some of the most egregious and harmful of the cartels may have nobody coming forward.” Senator Bill Blumenthal, US Senate, Committee on the Judiciary, Cartel Prosecution: Stopping Price Fixers and Protecting Consumers. Hearing before the Subcommittee on Antitrust, Competition Policy, and Consumer Rights, 113th Cong., November 14, 2013.
[4] The DOJ has also chosen not to offer whistleblower rewards though space prevents me from critically examining the purported rationale.
[5] These cases can be found in Ulrich Laitenberger & Kai Hüschelrath, “The Adoption of Screening Tools by Competition Authorities,” CPI Antitrust Chronicle, 2 (September 2011).
[6] Martin Huber & David Imhof, “Machine Learning with Screens for Detecting Bid-Rigging Cartels,” International Journal of Industrial Organization, 65 (2019), 277-301; and Manuel J. García Rodríguez et al, “Collusion Detection in Public Procurement Auctions with Machine Learning Algorithms,” Automation in Construction, 133 (2022), 10407.
[7] At the 2016 ICN Chief/Senior Economist Workshop, 27 competition authorities in attendance were surveyed and 15 reported they were doing some screening. Report on 2016 ICN Chief/Senior Economists Workshop, prepared by Nigel Caesar, Renée Duplantis, & Thomas Ross, April 25, 2017.
[8] The case for private actors engaging in cartel screening is put forth in Joseph E. Harrington, Jr. “Cartel Screening is for Companies, Law Firms, and Economic Consultancies, not just Competition Authorities,” Centro Competencia, 3 November 2021 https://centrocompetencia.com/harrington-cartel-screening-is-for-companies-law-firms-and-economic-consultancies/.
[9] By “tacit collusion,” I mean firms have an agreement to restrain competition which was achieved through means that, while indirect, involve distinct identifiable acts of communication. The distinction being made is between cases that “involve direct, readily observable proof that the defendants have exchanged assurances that they will pursue a common course of action” (explicit collusion) and those “in which the plaintiff invokes ‘indirect’ or ‘circumstantial’ evidence to establish the fact of the agreement.” (tacit collusion). William E. Kovacic, “The Identification and Proof of Horizontal Agreements under the Antitrust Laws,” Antitrust Bulletin, 38 (1993), 5-81; pp. 19-20.
[10] These cases can be found in Joseph E. Harrington, Jr., “Collusion in Plain Sight: Firms’ Use of Public Announcements to Restrain Competition,” working paper, November 2021 (Antitrust Law Journal, forthcoming).
[11] United States v. American Airlines, Inc., 743 F.2d 1114, 1122 (5th Cir. 1984).
[12] For some cases, see Larry Fullerton, “FTC Challenges to ‘Invitations to Collude’,” Antitrust, 25 (Spring 2011), 30-35.
[13] One email sent by Jacob Alifraghis of InstantUPCCodes.com stated: “All 3 of us – US, YOU and [Competitor A] need to match the price that [Competitor B] has. … I’d say that 48 hours would be an acceptable amount of time to get these price [increases] completed for all 3 of us. The thing is though, we all need to agree to do this or it won’t work.” In the Matter of Jacob J. Alifraghis, Complaint, FTC Docket Number C-448, August 29, 2014; p. 3.
[14] United States v. Container Corp. of Am., 393 U.S. 333 (1969)
[15] For a discussion of the case, see Severin Borenstein, “Rapid Price Communication and Coordination: The Airline Tariff Publishing Case (1994),” in The Antitrust Revolution, John E. Kwoka, Jr. & Lawrence J. White, eds., 2004.
[16] Frank H. Easterbrook, “Predatory Strategies and Counterstrategies,” University of Chicago Law Review, 48 (1981), 263-337; p. 318.
[17] John M. Connor, “The Great Global Vitamins Conspiracy: Sanctions and Deterrence,” working paper, Purdue University, February 2006.
[18] John M. Connor & Robert H. Lande, “Not Treble Damages: Cartel Recoveries Are Mostly Less Than Single Damages,” Iowa Law Review, 100 (2015), 1997-2023.
[19] For example, a company could use screening to look for collusion by its employees.
[20] For an overview of cartel screening, see Joseph E. Harrington, Jr. & David Imhof, “Cartel Screening and Machine Learning,” working paper, February 2022.
[21] These points are argued in Joseph E. Harrington, Jr., “A Proposal for a Structural Remedy for Illegal Collusion,” Antitrust Law Journal, 82 (2018), 335-359.
