TY - JOUR
AU1 - Kovacic, William E
AU2 - Marshall, Robert C
AU3 - Meurer, Michael J
AB - Abstract We provide empirical evidence that many multi-product firms have each participated in several cartels over the past 50 years. Standard analysis of cartel conduct, as well as enforcement policy, is rooted in the presumption that each cartel in which a given firm participates is a singular activity, independent of other cartel conduct by the firm. We argue that this analysis is substantially deficient in many aspects in the face of serial collusion by multi-product firms. We offer policy recommendations to reign in serial collusion, including a mandatory coordinated effects review for any merger involving a serial colluder, regardless of the apparent nature of the merger. I. INTRODUCTION Thus, every agreement both helps to shape, and itself fits into, a logical pattern of cartel control. By defining for each company its area of operation, its duties and its rights, an agreement makes it possible for all companies to plan, individually and in concert, with some assurance of how each will behave. By limiting and regulating the activities of chemical companies, cartel agreements get rid of the uncertainties of a freely competitive market. The cartels provide a relatively stable framework within which conflicting interests may be reconciled by maneuver, bargain, and compromise.1 A rapidly emerging view holds that American markets have grown steadily more concentrated over the last 20 years.2 The evidence shows that a small number of firms dominate sales across most markets, not just high-tech markets.3 This trend is also evident globally, and in most other leading economies.4 Growing concentration is accompanied by growing mark-ups and profits.5 There are many theories about why competitiveness has declined, but no consensus. One theory holds that lax antitrust enforcement over the past two decades is a significant factor.6 Critics of antitrust enforcers focus on lax merger enforcement and tolerance of alleged abuses of dominant positions by high-tech firms. Other economists theorize that rising concentration is best explained by the growing importance of network effects, the growing importance of investments in information technology, and various other factors.7 We take no position in this article on what has caused increased concentration, nor do we take a position on the merits of antitrust enforcement against abuse of dominance. Instead, we hope to shed some light on possible failings in a domain of antitrust enforcement that most commentators believe has been quite successful over the past two decades—anti-cartel enforcement. We analyse enforcement improvements that could better deter cartel formation and ward off the associated suppression of competition. We applaud increased vigor in global anti-cartel enforcement, and we do not dispute there have been remarkably successful prosecutions of global price-fixing in recent years. But certain patterns revealed by these prosecutions make us worry about what more might be out there. We wonder whether aggressive enforcement has deterred most price fixing or if instead, sprawling price fixing agreements remain hidden to enforcers who have become complacent. The evidence of serial collusion by major multi-product firms is readily observable from the public record.8 In the chemical industry, focusing solely on European Commission decisions, the following firms have been caught colluding in at least four products since 1955—Akzo Nobel, BASF, Solvay, Bayer, Degussa, Hoechst, Arkema/Atofina, Shell, Rhone Poulenc/Aventis, ICI, Elf Aquitaine, and Atochem. In the electronics industry, focusing solely on European Commission decisions, the following firms have been caught colluding in at least four products since 1980—Hitachi, Samsung, Mitsubishi, Toshiba, and Panasonic. In the auto parts industry, including European Commission and Department of Justice (DOJ) cases, the following firms have been caught colluding in at least three products since 2000—Hitachi, DENSO, Autoliv, TRW, NGK Spark Plug, Mitsubishi, Mitsuba, Robert Bosch, Toyoda, Panasonic, Valeo, and Yazaki. In the financial industry, focusing solely on European Commission decisions, the following firms have been caught colluding in at least three products since 2000—JP Morgan, Royal Bank of Scotland (RBS), and Union Bank of Switzerland (UBS). In the graphite’s industry, including European Commission and DOJ cases, SGL has been caught colluding in three products since 2000. Disturbingly, many of these firms have been caught colluding in substantially more than four products—for Hitachi the number is 16, for Akzo Nobel the number is nine.9 Note that these are the cartels that have been detected. Effective collusion is clandestine; these firms could be participating in many other undetected cartels. Because of this detection problem credible empirical work that estimates the magnitude of unobserved cartels is difficult.10 At any rate, the collusion in at least three of the industries persists—new investigations have recently been opened regarding collusion in the chemical, auto parts, and financial products markets.11 Evidence that serial collusion predominates over one-shot cartels should help change beliefs about why firms fix prices and violate antitrust law. Many commentators believe that organizational failures within firms are the primary causes of collusion. They argue that lax monitoring and inappropriate incentives allow and encourage rogue division managers to fix prices. Yet, the typical responsible managers who are identified in price fixing investigations do not seem to be rogues, and seem instead to be top officers of the accused firms.12 To be sure, the rogue manager scenario serves the interests of both serial colluding firms and enforcement authorities. Serial colluding firms can declare each episode of the conduct ‘rogue’ and somehow avoid probing inquiries as to whether they have a business model that embraces explicit collusion. Public authorities, whose interests with antitrust enforcement should be at odds with serial colluders, have almost never pointed to the multiple episodes of cartel conduct by these major multi-product firms as a separate and distinct concern because it is in their interest to tout their great success with cartel law enforcement, one case at a time. The operation of serial collusion belies the rogue manager story and suggests instead that collusion is simply a way of doing business for many firms. Consider for example serial collusion in the chemical industry. Firms in this industry relied on the consulting firm, Fides/AC Treuhand, to facilitate, implement, and run seven chemical cartels.13 Every chemical firm listed above was in a cartel that used the services of Fides/AC Treuhand. How would rogue division managers, acting on their own and thus without support from anyone inside their firm, gain access to the cartel consulting services of Fides/AC Treuhand? Would they make an ‘out of the blue’ phone call or knock on the door of Fides/AC Treuhand, without introduction of any kind, and ask for Fides/AC Treuhand to help them form and run a cartel? The regular use of Fides/AC Treuhand by large multi-product chemical firms suggests, as a reasonable conjecture, that (i) each firm has institutional knowledge of the value provided by Fides/AC Treuhand in assisting with suppression of interfirm rivalry, (ii) new division managers are trained to be aware of this value, and (iii) introductions to Fides/AC Treuhand are done by more senior people at the firm who have previous experience with Fides/AC Treuhand. Collusion across a broad range of products by large multi-product firms creates many more degrees of freedom in addressing standard cartel problems than is available to single-product firms engaged in collusion. The central problem of a cartel is secret deviations by members. All cartel members have an incentive to cheat on the collusive agreement, to sell more at elevated prices than what the cartel agreement specifies for them. To address this central problem, cartels implement structures, if they do not already exist, for pricing, allocation of the collusive gain, and enforcement of the cartel agreement. If two large multi-product firms are colluding across many product lines, they may, for example, make side payments to one another across product lines, have one firm exit one product and compensate that firm with the accommodation of entry by their co-conspirator in another, and sacrifice the profits in one cartel to punish deviant behaviour by a single-product co-conspirator as a warning signal to other single-product firms with whom they are colluding in other products. None of these kinds of conducts are available or make sense for single-product cartel members. Again, multi-product serial colluders have a greatly expanded set of actions they can take to enhance cartel profits across their portfolio of cartels. It is important to recognize the possibility that large multi-product serial colluders are managing a portfolio of cartels, not a single cartel. The expanded set of opportunities available to firms that are managing a portfolio of cartels makes such cartels, on average, more profitable, stable, and long lasting. Thus, serial collusion likely causes more social harm and is harder to deter than one-shot collusion. All else equal, socially optimal enforcement efforts should be more rigorous in a world of serial collusion as compared to a world with equally frequent one-shot collusion. Simply asking enforcement agencies and courts to take off their blinders and pay attention to serial collusion could help achieve greater rigor. Courts sometimes dismiss evidence of a history of collusive conduct as irrelevant or as lacking sufficient relevance to offset its potential for prejudice. They should not; this evidence is highly relevant.14 A history of serial collusion should also be used by enforcement agencies when deciding whether to launch price fixing investigations and especially when reviewing mergers that could facilitate future cartel activity. We propose four reforms to improve anti-cartel enforcement. First, we spell out a new cartel reconstruction program. Enforcers should work with colluding firms to learn how each cartel worked, who was responsible, and what other markets might be affected. Second, more extensive monitoring should be imposed on serial colluders in sentencing, and settlement or plea agreements. A better understanding of the nature of serial collusion is needed to guide enforcers to find the red flags that monitors should watch for. Information from monitoring might trigger an investigation, and can also guide merger review where it intersects with anti-cartel enforcement. Third, we discuss problems with current leniency programs and explain why enforcers should complement leniency programs with bounty programs aimed at peeling small firms away from cartels. A successful bounty program would make small firms unappealing members for inclusion in a cartel which could destabilize a cartel or at least limit its ability to raise prices. Finally, we favour a mandated review of mergers from a coordinated effects perspective whenever a serial colluder submits a merger for review. Cartelization might be facilitated even when the merging parties do not compete directly. For example, if firm A proposes a merger with firm B, and firm A makes products 1–50 while B makes products 51–100 there may appear to be, on the surface, no coordinated effects issue. However, if firm A is a serial colluder that has regularly colluded with firm C, and firm C makes many of the products that firm B makes, then a very real concern arises about the potential post-merger expansion of the portfolio of cartels involving A and C. The article is organized as follows. In Section II, we provide more detail regarding the empirical evidence of serial collusion. In Section III, we review the economic and legal analysis of single-product collusion. In Section IV, we extend the standard law and economic analysis of collusion, which is predicated on all collusion being singletons, to multi-product firms engaged in collusion across a broad range of products. In Section V, we flesh out policy recommendations. In Section VI, we offer concluding comments. II. EMPIRICAL EVIDENCE OF SERIAL COLLUSION The label ‘serial collusion’ requires further examination. Multiple cartel episodes by a given firm occurring over the same time interval where this time interval is relatively short would not constitute serial collusion. For example, suppose a firm was found to have participated in six cartels that each lasted 18 months from January 2000 until June 2001. We display this possibility in the Hypothetical Duration Chart below. Collusive behaviour of this sort might be explained by a single top executive who experimented briefly with explicit collusion, and the firm stopped the collusion when it was uncovered by other officers or directors. This pattern does not suggest the firm has adopted collusion as a way of doing business and the label serial collusion does not fit this hypothetical. Figures 1A and B and 2–4 display cartel duration charts in the chemical, electronics, auto parts, and financial products markets. The charts for the chemical, electronics, and financial products markets do not look at all like the Hypothetical Duration Chart. There is considerable variation in the start and end dates of the cartels across different product lines.15 The data in these charts is consistent with the view that firms are serial colluders in the sense that they move into cartels when they have the opportunity and move out of cartels when they collapse because of cheating or detection. The auto parts cartels shows less heterogeneity with respect to duration though we suspect serial collusion is also present in this industry; we will comment further on that industry below. We present additional evidence later in this section that over the span of decades certain serial colluders, especially in chemicals and electronics, have repeatedly restarted cartels after earlier episodes of serial collusion have collapsed. Figure 1. Open in new tabDownload slide Open in new tabDownload slide (A) (See above.) Cartel participation durations for chemical firms participating in four or more cartels since 1955. (B) (See below.) Cartel participation durations for chemical firms, and Fides/AC Treuhand, participating in four or more cartels since 1955. Figure 1. Open in new tabDownload slide Open in new tabDownload slide (A) (See above.) Cartel participation durations for chemical firms participating in four or more cartels since 1955. (B) (See below.) Cartel participation durations for chemical firms, and Fides/AC Treuhand, participating in four or more cartels since 1955. Figure 2. Open in new tabDownload slide Cartel participation durations for electronics firms participating in four or more cartels since 1980. Figure 2. Open in new tabDownload slide Cartel participation durations for electronics firms participating in four or more cartels since 1980. Besides examining cartel duration we believe it is also important to study the pattern of cartel membership across the range of product markets present in an industry. If collusion occurs exclusively between small subsets of the firms in an industry, and these subsets stayed relatively isolated from one another, then this would be less concerning than collusion across a broad range of firms through time. Collusion across a broad range of firms would imply that all participating firms saw the value of collusion across their many product lines and, additionally, cartel issues could be resolved across products rather than solely within a given product. The Hypothetical Cartel Involvement Table below displays a pattern in which the colluding firms have formed collusive subsets that are isolated from each other. This pattern of behaviour provides fewer benefits to cartel members in terms of stabilizing and increasing the profitability of collusion, and thus it is less problematic. Figures 5–9 later in this section display cartel involvement tables for the chemical, electronics, auto parts, and financial products industries. The tables show a complex web of connections between firms that substantially diverges from the Hypothetical Cartel Involvement Table. Now we will take a closer look at the duration charts in Figs 1A and B and 2–4. Figures 1A and 1B are for the chemical firms listed in the introduction. Time in years is plotted on the horizontal axis. A number identifying separate cartels is on the vertical axis. The numbers correspond to the cartels identified in Appendix A. The cartels are grouped by multi-product firms. For Akzo Nobel, the first firm in Fig. 1A, we see a horizontal bar for the first cartel, cartel 19, which is the Organic Peroxides cartel. The solid line shows the duration of the cartel as reported in the EC decision. The Organic Peroxides cartel lasted nearly 30 years, starting in the early 1970s. It would be difficult to credibly attribute a 30-year cartel to a single rogue division manager. Also, it seems difficult to imagine that a cartel could function for 30 years without the knowledge of top management at Akzo. It is clear that there are a range of start dates for the Akzo cartels, from the early 1970s to approximately 2005. The durations are six to eight years for many cartels. The range of start dates as well as the durations leads us to reject the notion that there was a single top executive who tried collusion briefly across many product lines. As we look through the rest of the firms in Fig. 1A and B we see a pattern of cartel conduct similar to the conduct of Akzo.16 Hypothetical Cartel Involvement Table The first firm in Fig. 1B is not a chemical firm. It is Fides/AC Treuhand. According to its webpage, Fides/AC Treuhand is an association management company (AMC), providing management services to trade associations, professional societies, and similar organizations. For many of the European chemical cartels it managed and facilitated a broad range of cartel activities such as coordination of price increases, cartel-related communications, monitoring of compliance with the cartel agreement, and adjustment of members’ share allocations.17 It is evident from Fig. 1B that Fides/AC Treuhand has played the role of cartel facilitator for many years, for prolonged periods of time within each cartel, and across a range of products.18 It is interesting to note that only seven of the nine cartels listed for Fides/AC Treuhand are in the chemical industry. The other two are in the wood products industry. Apparently the value of Fides/AC Treuhand as a cartel facilitator is understood in industries beyond chemicals. Figure 2 provides the duration information for the electronics firms identified in the introduction. Note that the time scale starts in 1980, not 1955. The cartels start at different dates and they last for up to 15 years. Figure 3 presents the information for the auto parts cartels. The durations are coincident for several of the cartels. For example, the first bar for Hitachi pertains to seven cartels, all of which have the same duration. Note that the time range for the figure is 1995–2015. The auto parts cartels, viewed in isolation, leans away from an interpretation of serial collusion since many of the cartels started at the same time. However, the durations of many of the cartels is 10 years, which is difficult to explain as being the action of a single rogue division manager. Figure 3. Open in new tabDownload slide Cartel participation durations for auto part firms participating in three or more cartels since 1995. Figure 3. Open in new tabDownload slide Cartel participation durations for auto part firms participating in three or more cartels since 1995. Note that Hitachi appears in both the electronics and auto parts cartels. Altogether, Hitachi participated in 16 cartels. Similarly, aggregating across electronics and auto parts, Mitsubishi participated in eight cartels and Panasonic in eight. Thus, it is reasonable to conjecture that these three firms brought their prior experience with collusion in electronics to auto parts. Finally, in the financial services industry, there have been fewer cartels detected than in chemicals, electronics, or auto parts. Figure 4 presents information for the financial services cartels.19 Note that the time range in the figure is 2000–15. JP Morgan has participated in five cartels,20 RBS in four, and UBS in three.21 The start dates have substantial variation and the durations are mostly on the order of three years. Figure 4. Open in new tabDownload slide Cartel participation durations for financial firms participating in three or more cartels since 2000. Figure 4. Open in new tabDownload slide Cartel participation durations for financial firms participating in three or more cartels since 2000. We now return to the chemical industry and examine the pattern of firm involvement with other firms across these cartels. Figure 5 presents that information with firm names on the rows and cartel names in the columns. It is clear that the cartel participation is not concentrated within specific pairs of firms but extends broadly across many firms. It is important to note that every chemical firm in Fig. 5 has been involved in at least one cartel where Fides/AC Treuhand was a cartel facilitator. Figure 5. Open in new tabDownload slide Cartel involvement for chemical firms. Figure 5. Open in new tabDownload slide Cartel involvement for chemical firms. Figure 6 presents the pattern of cartel involvement for the firms in the electronics industry. Again, there is broad participation across the firms. In fact, Hitachi has conspired with each of the other four firms in at least one cartel as has Samsung and Toshiba. Figure 6. Open in new tabDownload slide Cartel involvement for electronics firms. Figure 6. Open in new tabDownload slide Cartel involvement for electronics firms. Figures 7 and 8 present the structure of involvement of relevant firms in the auto parts and financial services cartels. There is an important caveat regarding these figures. Unlike chemicals and electronics, the auto parts and financial services cartel cases are sometimes drawn from the DOJ pleadings and admissions. The EC prohibition decisions will enumerate all firms participating in a cartel but the DOJ will typically only identify the firm that is pleading or admitting culpability. Thus the involvement matrices for auto parts and financial services are truncated in contrast to the matrices for chemicals and electronics. Of course, this is true for the duration figures as well. Namely, the firms not identified by the DOJ in the auto parts and financial services cases are not in the duration plots. Figure 7. Open in new tabDownload slide Cartel involvement for auto part firms. Figure 7. Open in new tabDownload slide Cartel involvement for auto part firms. Figure 8. Open in new tabDownload slide Cartel involvement for financial firms. Figure 8. Open in new tabDownload slide Cartel involvement for financial firms. Overall, the frequency of collusion by this collection of firms across chemical, electronics, auto parts, and financial product markets is striking. Despite additional penalties imposed by some competition law systems for recidivism, many of these firms chose to engage in a series of collusive schemes over a significant period of time.22 The firms active in each cartel change from cartel to cartel because these firms manufactured different portfolios of products. Many firms encountered one another in two or more cartels. Furthermore, for the chemical cartels there is the ongoing use of a cartel facilitator. Thus far our discussion has concentrated on recent cartel activity in four industries for which we have the best data on participation and duration. We conclude this section by highlighting other instances of serial collusion. Our discussion suggests that serial collusion has been a global problem almost since the chemical industry was established in Germany before 1900. Episodes of serial collusion centered in the USA are also well represented in the history of price fixing. We begin by returning to collusion in the electronics industry. In Figs 2 and 6, we presented information about electronics cartels spanning 25 years starting in the late 1980s. Figure 9 presents the pattern of cartel involvement in the electrical industry that was uncovered in the USA in the late 1950s. The products in those earlier cartels tended to relate more to electricity generation rather than components in consumer electronics products featured in the more recent cartels. Figure 9. Open in new tabDownload slide Cartel involvement for 1950s electrical firms. Figure 9. Open in new tabDownload slide Cartel involvement for 1950s electrical firms. The duration of these cartels is not clear, but two of the leading cartel members, General Electric and Westinghouse, were frequently investigated and prosecuted for antitrust violations, including price fixing, in the decades leading up to and following the 1950s electrical equipment conspiracy.23 Twice before the electrical equipment conspiracy General Electric and Westinghouse were sued for using patent licenses to facilitate price fixing in electrical products. In a close and controversial case the companies were acquitted in the first suit, US v General Electric (1924), but they were found liable for price fixing in 1948 in US v Line Material. Southern States Equipment Co, another one of the colluding firms listed in Fig. 9, was also found liable in 1948. Finally, in 1976 the DOJ reached a settlement agreement with GE and Westinghouse that modified the 1962 consent decree.24 The settlement was reached in response to complaints by the electric utility industry that the companies continued to fix prices.25 Steel is another industry that seems to have been plagued by serial collusion.26 A steel cartel was established in Western Europe during the late 1920s to cover exports of various steel products.27 American companies joined European companies to reestablish a steel cartel from 1933–1938. It covered the same products as the 1920s cartel and several additional products.28 ‘[I]n 1938, almost ninety-eight percent of world steel exports were regulated by restrictive production and marketing arrangements.’29 After World War II, US Steel and Bethlehem Steel lead six different multi-market price-fixing schemes.30 ‘VSL construction was similarly involved (including as a leader) in multiple US cartels across several decades with distinct, but overlapping partners.’31 In the post-war US, Hay and Kelley uncovered 10 distinct price fixing agreements related to steel products. Two were regional in scope and the other eight had a national scope. The products were: wrought steel wheels, rolled steel pipe flanges and rings, bed springs, metal library shelving, fabricated structural steel, carbon steel sheets, stainless steel welding electrodes, and wholesale distribution of steel products.32 In the new century, suits have been brought against steel product price fixers in both the Europe and the USA ‘The European Commission has fined 17 producers of prestressing steel a total of €458,410,750 for operating a cartel that lasted 18 years until 2002 and covered all but three of the then European Union Member States.’33 ‘Eight steel producers were accused of price fixing in a private suit filed by US steel purchasers in 2008. Five of the defendants settled for a total of $103.7 million.’34 Finally, ‘German authorities announced an investigation of seven makers of flat steel for price fixing in the fall of 2017.’35 Multi-product collusion has a long history in the chemical industry going back to the 1880s and appearing in most decades since then. A central player is the German chemical company I.G. Farben, its predecessors and its offspring.36 Three of the chemical companies listed in Fig. 5 were at the core of I.G. Farben. Bayer, Hoechst, and BASF merged (along with three other firms) to form it in 1925. And those three firms reemerged during the liquidation of I.G. Farben at the end of World War II. Stocking and Watkins observed that: ‘The breadth of its chemical interests has led I.G. [Farben] into agreements with every major world chemical producer, covering together virtually every branch of chemical manufacturing.’37 Although the system of cartels was based in Europe, it included American firms in global export cartels and reached directly into the domestic American markets. A German–American cartel was formed in magnesium powder and fabricated magnesium that included I.G. Farben, Dow, and Alcoa.38 DuPont found English and Swedish partners to implement global cartels in other markets. In the interwar period duPont and Nobel extended their global cartel from explosives to ‘dyestuffs, artificial leather, celluloid’ and it eventually ‘embraced all products in the sale of which the two companies, save for the agreement, would have been competitors’.39 In the late 1930s, du Pont and ICI updated their chemical cartel agreement with ‘the inclusion of new products, many of which were already being treated as though they were included – notably certain chlorine products, antiknock compounds, synthetic resins and plastics, pharmaceuticals, Neoprene (du Pont’s synthetic rubber), and nylon’.40 III. A FRAMEWORK FOR ANALYSING SINGLE-PRODUCT EXPLICIT COLLUSION Global acceptance of an anti-cartel norm is perhaps the most important development in competition law and policy in the past 20 years.41 For most of the 20th century, the prosecution of cartels was largely the preoccupation of US enforcement agencies. Most jurisdictions regarded supplier collusion as a secondary priority. Some actively resisted US efforts to chase cartels outside its own borders. In the late 1990s, the lysine and vitamins cartel prosecutions changed all of that. The DOJ recovered massive, unprecedented fines against the participants in both collusive schemes, whose audacity, complexity, and effectiveness startled the antitrust world.42 Shamed by their earlier complacency and lured by the prospect of immense financial recoveries, many competition agencies boarded the anti-cartel bandwagon. Today agency leaders regularly call cartels the worst competitive scourge of the market place.43 In the past 20 years, anti-cartel enforcement has been modernized in important ways, especially through diffusion of vigorous enforcement policies in many other countries, and in the widespread adoption of leniency programs.44 By some measures, the resulting international campaign against cartels seems to be working. Enforcement agency bar charts tracing total cases, civil fines, criminal fines, and prison sentences climb impressively upward.45 Enforcement officials in the European Union (EU) and the USA point with evident pride to the greater severity of sanctions imposed on cartel offenders.46 Agency press releases speak in the language of sports: new heights in recoveries are ‘records’ akin to achieving the largest number of goals scored in a season or a career.47 Newer authorities underscore their own progress by pointing to a higher tempo of prosecutions, increased recoveries, and enhanced enforcement tools. The drive to achieve higher recoveries and more convictions derives encouragement from commentators, journalists, and political leaders who regard higher activity levels as good measures of agency performance.48 In addition, huge damage awards in private cases, and the gradual global adoption of private rights of action, reinforce the impression of increasing enforcement effectiveness. The traditional activity-based measures of enforcement performance accurately suggest that the hazards for cartel participants are greater today than they were a few decades ago. The expansion of anti-cartel enforcement is an impressive public policy accomplishment, and the enhancement of modern enforcement tools—notably, the development of more powerful leniency mechanisms49—present admirable progress in policymaking. Despite these accomplishments, reasons for discomfort about the effectiveness of anti-cartel programs remain.50 The volume of fines and prison sentences are badly imperfect proxies for gains in social welfare. They do not indicate whether antitrust enforcement is achieving one of its central aims: to deter firms from engaging in illegal collusion in the first place. Indeed, climbing levels of prosecutions and punishments are ambiguous evidence of effectiveness; they could be taken to suggest that a significant number of firms regard the possibility of detection and prosecution to be acceptable business risks.51 In this section, we sketch the design and enforcement of the legal framework that applies to collusive agreements. Here we also describe the economics regarding the formation and operation of cartels. Our focus in this section is one in which rival firms engage in explicit collusion (ie they resort to an express exchange of assurances) with regard to a single product. This provides valuable background for our consideration of multi-product collusive arrangements in Section IV below. Five major themes emerge from our review of the economics and law related to single-product collusion. First, we wish to underscore the difficulties associated with forming and implementing a successful cartel. These include the three tasks made famous by George Stigler (forming a consensus, detecting cheating on the consensus, and punishing deviations)52 and the likewise important suppression of external threats (most importantly, buyer resistance, new entry, and the incursion of substitute products) illuminated in Michael Porter’s ‘five forces’ model of business strategy.53 As we have said in previous work, nothing about this is easy.54 Rivals who gather furtively in a hotel conference room or a restaurant to declare their commitment to raise prices, yet do not devise plans over the course of the operation of the cartel to surmount all of these obstacles, either by themselves or with the assistance of expert advisors, have taken grave risks with little prospect of success. Our second observation is that the difficulty and complexity of cartel formation and management tasks require considerable technical and managerial skill—for example, in devising and adjusting the assignment of market shares, in tracking sales volumes by individual cartel members, or in constructing methods for executing the side payments that ensure fidelity to the common scheme. As with any other dimension of organization and management, there are benefits from the learning—the repeated performance of the same tasks over time. Cartel members could realize these benefits either by repeated participation in collusive schemes (among themselves or with other firms) or by engaging third parties (eg an accounting firm such as Fides/AC Treuhand) which have broad experience in assisting cartels. The third major point from our research, explored more fully below in Section IV, is that the involvement of senior company officials is typically important for effective cartel coordination. The complexity, durability, and scope of many cartels likely require commitment at the highest level of the company and the participation of key senior officials, as well as middle managers. We can imagine that low level subordinates in some instances devise and execute collusion schemes without the knowledge or acquiescence of top management. This is unlikely to be the main case, especially for the massive, intricate, long-lived arrangements that figure so often in modern cartel prosecutions. Without high level involvement, such schemes would likely not be possible. Our fourth observation is that cartel members, in many cases, cannot perform the collection of complex, difficult cartelization tasks without continuing engagement. The ‘contract’ that represents the shared commitment to raise prices is by necessity highly relational. It should be flexible to adapt to changed circumstances (eg an increase in demand that was not anticipated when the colluding firms first allocated market shares among themselves) and to respond to external threats to the operation of the collusive scheme (eg a move by a major purchaser to vertically integrate into the production of the input currently controlled by the cartel). Adaptation and adjustment require consultations for the group to determine how to proceed. Moreover, there is an inherent degree of incompleteness in all contracts, owing to compromises that inject ambiguity into certain terms or to the inability of the parties to anticipate and address all contingencies. Like legitimate contracts, cartel agreements are incomplete for the same reasons. To address the consequences of incompleteness as performance of the agreement takes place, cartel participants often communicate with each other—to devise solutions to unexpected contingencies or to confront ambiguities or disagreements that the parties sidestepped when they entered the scheme. The importance of continued engagement and communications among participants suggests a fifth theme, developed more extensively in Section IV. Communications can be improved, and the performance of the cartel’s key operational tasks can be enhanced, as the range of the firms’ interaction expands in time and scope. The performance of a single cartel generates considerable knowledge, trust, and understanding that can be applied usefully in additional cartels. Repeated interaction over time in new collusive episodes helps the cartel refine its operational techniques, improve means of communication, increase understanding of each firm’s culture, conduct, and intentions (to improve the ability of each firm to anticipate and interpret the moves of its counterparts), and where suspicion and doubt always lurk in the background, to strengthen trust. Engagement across a broader scope of products provides additional contact points that can enhance coordination—for example, by providing more avenues to make side payments, more possibilities for punishing deviations, and more means to use leniency strategically to channel the attention of enforcement agencies towards a decaying or dead cartel and away from a vibrant or promising collusive arrangement. The phenomena described above place a premium on establishing a framework of legal rules and enforcement mechanisms that accounts for the dynamism of cartel behaviour and makes the fullest possible use of knowledge gained about how cartels operate and how they respond to new advances in enforcement methods. We emphasize that, although the formation and execution of a cartel is no easy task, the rewards for successful cartelization are substantial. The history of antitrust policy since the late 19th century demonstrates that firms have responded to enhancements in the law enforcement regime—the bolstering of substantive legal commands, sanctions, and methods of detection—with a host of countermeasures to avoid detection or blunt the impact of more severe punishments. This history suggests that the arms race between law enforcement and cartelists has not subsided. Competition agencies should be no less adaptive or ingenious than the cartelists they are seeking to apprehend and better, deter. A. Antitrust Law and the treatment of Collusive Horizontal Agreements As suggested above, the prohibition of cartelization is at the heart of antitrust law.55 By statute, by jurisprudence, or by both, the definition of the offense has evolved in stages. In the USA, early cases confronted and condemned the setting of the complete price of the product or service.56 In fits and starts, the US jurisprudence gravitated, by the late 1930s and early 1940s, toward the establishment of a categorical prohibition (a ‘per se’ rule) against horizontal price fixing.57 The principle of strict prohibition was extended to encompass restrictions that affect the final price and quality of the product or service; these include customer and territorial allocations,58 agreements to restrict the use of certain product inputs,59 the fixing of one element of the total price (eg financing terms).60 Most recently, the US DOJ and the Federal Trade Commission have announced that both agencies will prosecute, as per se offenses, ‘no-poaching’ agreements by which firms agree not to make unsolicited calls to recruit employees of their rivals.61 The delineation of agreements subject to summary condemnation has been accompanied by, and sometimes has followed, developments in academic literature and professional commentary about the economic impact of the challenged practices. As a consensus has developed about the economic and social harm associated with cartels, there has been a progressive increase in sanctions and methods of detection, with the most extraordinary enhancements taking place in the past half-century. In the USA, criminal prosecution of cartel agreements is a routine practice. Individual defendants face maximum prison terms of 10 years, and those convicted of cartel offenses, or who plead guilty, are likely to spend at least 2 years in prison. Monetary fines now can run into the hundreds of millions—indeed, billions—of dollars as a result of fining mechanisms that allow the Justice Department to collect double the gain to violators or double the loss to victims. Detection techniques have been enhanced hand in hand with the increase in sanctions. DOJ introduced its modern leniency program in 1993 and since has introduced a series of enhancements. DOJ and the FBI use the full range of electronic surveillance and related information gathering methods available for the investigation of organized crime. Over 60 jurisdictions worldwide have adopted leniency programs, and the use of search warrants to gather information from business premises is a common element of global antitrust law enforcement. Two major issues have accompanied the elaboration of the substantive principles and enforcement methods described above. The first is the definition of what constitutes a collusive scheme worthy of summary condemnation. Courts have recognized that agreements to set prices or other terms of trade are worthy of fuller analysis, and sometimes deserve approval, when they facilitate the development of a new product or service or enable firms to generate efficiencies for consumers that otherwise would not be available in the absence of the restrictions.62 Modern US jurisprudence makes clear that firms always have the opportunity to introduce cognizable, plausible efficiency rationales for challenged restrictions, and that the presence of such rationales precludes per se condemnation. The second pivotal issue is the definition of what constitutes concerted action. If no efficiency rationales are present, the key issue in the case will be whether the firms acted alone or in concert.63 The enhancement of sanctions and means of detection has led firms to take various steps to mask their coordination. Many of the government’s cartel prosecutions and private antitrust cases hinge on decisions about what types of evidence (known as ‘plus factors’), when added to the fact of parallel conduct, permit a judge or a jury to infer agreement. As set out below in Section IV and in our earlier work, we argue that careful examination of past cases and investigations yields valuable insights about which types of behaviour deserve emphasis as plus factors. In the discussion of policy recommendations that concludes this article, we propose that government agencies and researchers undertake larger efforts to collect data on the origins and operations of specific cartels for three purposes: (i) to identify promising focal points for investigations; (ii) to increase our understanding of which types of observed conduct should be treated as plus factors to sustain an inference of agreement; and (iii) to improve the operation of compliance programs through which government agencies and firms monitor adherence to the law against collusion. B. The effects of collusion Competition among vendors has two main advantages. First, it reveals information to a buyer that it might not be able to discover on its own. Second, it helps a buyer get greater quantity or better quality than it could in the absence of competition. To illustrate these advantages, and thus the justification for the prohibition against price fixing, consider an example in which a medical supply firm buys chairs for its new offices. Suppose a firm wants to acquire one wooden office side chair of a given pre-specified size and quality. Suppose the firm can purchase the chair from a local office furniture supplier for $250. In contrast, suppose the firm wants to buy at least 2000 such chairs. If the purchase of a single chair of a given quality costs $250 what should the firm expect to pay for 2000 such chairs? Should the firm expect to pay, per chair, $250 or something less than that, perhaps even quite a bit less than that? A medical supply firm has no expertise to assess what 2000 office chairs should cost. So, the firm uses competition to learn about the market price for such chairs, inviting all qualified vendors of such chairs to submit bids for the provision of at least 2000 chairs. Each invited supplier will know that they are competing with others in their industry for the contract. The competition between the vendors is such that, although winning with a high bid would be quite profitable, each vendor realizes that a high bid has a low probability of winning. To enhance the probability of winning each vendor will consider submitting low bids. The bids of each bidder will be bounded—no bidder will want to earn negative profits by winning. Also, the greater the number of bidders the more aggressive each bidder will need to be to enhance their chance of winning. Suppose there are five vendors invited to bid and the low bid turns out to be $125 per chair. Competition among the vendors has revealed to the purchasing firm information that they would not have been able to acquire on their own—namely, that 2000 chairs can be acquired at half the per chair cost of buying a single chair. But there is an additional benefit of the competition. The bid solicitation had specified the purchase of at least 2000 chairs. With a winning bid of $125 per chair, which may be quite a bit lower than the prior belief that the firm had about what they would pay for the chairs, the firm may decide to purchase an additional 200 chairs. The winning bid of $125 may not leave much of a profit margin for the winning vendor. The five invited bidders each realize that the competition amongst them will negatively impact their profits. If the five vendors met prior to submitting bids they may reach an agreement to suppress their rivalry and share in the incremental profits from doing so. The five bidders may agree that Bidder A will submit the low bid of $187.50 per chair and each of the other four bidders will submit bids ranging from $190 to $230 per chair. Each of the four losing bidders may be compensated by Bidder A for suppressing their rivalry with direct cash payments, subcontracts, or agreements for other bidders to be the designated winners on future procurements. The explicit collusion just described is clandestine, unknown to the purchasing firm. The five bidders are pretending to act competitively in the submission of their bids. The purchasing firm, who does not know there is collusion by the bidders and is relying on competition to reveal information about what should be paid for 2000 chairs, believes that the price of $187.50 per chair has been revealed through legitimate competition to be the lowest price that can be paid for the chairs. At that price, the firm may buy 2000 chairs, but not more. An important incremental loss from the collusion is the reduction in the number of chairs purchased. At the competitive price of $125 the firm bought 2200 chairs, but at a price of $187.50 the firm buys 200 fewer chairs. Mutually beneficial transactions are not occurring due to the collusion by the bidders. This is a social loss. The fundamental points of the chair purchase example above are not special—they apply across all products and markets. Antitrust laws are designed to prevent, or at least mitigate, the loss that occurs from the reduced output sold in the market as well as assure market participants that competition will produce reliable price discovery.64 C. The machinery of collusion Cartels use various techniques to achieve price elevation and/or output reduction. Naturally, cartels sometimes reach a simple agreement to fix an agreed price (or related terms such as discounts, interest rates, credits, rebates, etc). For various reasons cartels sometimes use highly related but alternative pathways to raise prices for example by agreeing to output restrictions, allocation of customers or territories, or to rig bids at an auction or procurement. Antitrust law distinguishes illegal price fixing from merely parallel behaviour and requires that plaintiffs and prosecutors prove that the accused competitors reached an agreement to fix prices. The agreement requirement is justified on the grounds that merely parallel behaviour is less likely to generate the high prices and output restrictions that competitors can achieve via agreements to explicitly collude. Further, the risk of chilling pro-competitive behaviour is likely greater if the law attempts to regulate parallel pricing in the absence of agreement.65 Thus, a key assumption supporting this distinction is that successful collusion is unlikely unless the firms communicate and reach an agreement. Explicit collusion is different from tacit coordination. Tacit coordination in an oligopoly with firms that recognize their mutual interdependence is just rational non-cooperative interaction. On the other hand, an agreement among competitors that relies on interfirm communication and/or transfers to suppress rivalry is explicit collusion, and we refer to the firms involved in the conspiracy as a cartel.66 The communication regarding the agreement can occur at the initiation phase of the explicit collusion and also at the implementation phase.67 Since explicit collusion is illegal the agreement must be self-enforcing. Put differently, during the course of the conspiratorial agreement each co-conspirator will be contemplating if their own profits can be increased by not complying with the agreement. But non-compliance that is observable to fellow co-conspirators will potentially result in punitive actions within the cartel or even the dissolution of the agreement. After the initial collusive agreement is implemented further communication may be required to counter the temptation of cartel members to cheat.68 Therefore, secret non-compliance is the central problem of all cartels. An illustration may be helpful. Suppose that a homogeneous product is made by four firms and some smaller foreign fringe firms, where the four large firms produce 85 per cent of total world output. Initially, the four firms are not in a cartel and the price in the market place is $10 per unit. The four firms meet and agree that they will all be better off selling at $25 per unit.69 So, each cartel firm announces a new price of $25. But, the buyers of such a product are not passive in paying the higher price, especially the larger buyers. Suppose a large buyer that has typically bought from firm A approaches firm B and offers to buy a large quantity at a price of $20. Since it had been profitable to sell at $10 it is even more profitable to sell at $20. Firm B’s only introspective question is whether the sale to the large buyer would be detected by the other cartel firms. Certainly, firm A will know that it lost the business of the large buyer, but to whom? Did the large buyer acquire most of its needs from foreign fringe firms or was there a defection by a cartel firm? If firm B’s defection will go undetected then it will take advantage of the opportunity to sell incremental quantity to a new buyer. But the other cartel firms will be receiving similar offers from buyers. Quickly the price in the market will not be $25 but $20. As the price descends there will more offers by buyers at even lower prices. The price will continue to descend towards $10. This is natural with unconstrained secret deviations. A cartel can be effective in mitigating secret deviations and thus generating a sustainable material profit above non-collusive levels if it implements certain structures, referred to as pricing, allocation, and enforcement structures.70 Pricing structures include explicitly coordinated price announcements, rigged bids, or quantity restrictions. Also, to facilitate the implementation of these prices, the incentives of the sales staff may be altered to ‘price before volume’ from the standard incentive of seeking incremental volume by undercutting rivals. Allocation structures refer to the division of the collusive gain. These may include, but are not limited to, market share allocations, customer allocations, geographic allocations, or combinations of the three. Allocation structures also include transfers between firms to rectify inadvertent deviations from the agreement or address new issues associated with the division of the collusive gain. Enforcement structures include monitoring mechanisms that assure each cartel firm that others in the agreement are keeping their commitments regarding pricing, quantities, and the sharing of the collusive gain. Investment in monitoring can be extensive. Enforcement structures also include the penalties for breaches of the agreement. If good collusive structures are in place then secret deviations can be greatly mitigated. The collusive structures that are effective in one product/industry/market may not be effective in another. It is obvious from the history of cartel agreements that a rich diversity of cartel structures have been implemented. When a cartel implements effective collusive structures there are conducts and outcomes that emerge that are largely inconsistent with unilateral action but consistent with explicitly coordinated action. These conducts and outcomes are above and beyond parallel movements in price that could be occurring for legitimate unilateral reasons, and thus they are referred to as ‘plus factors’. When the conducts and outcomes are such that it is highly unlikely they are the product of unilateral action but, instead, are most likely the result of explicit coordination to reduce interfirm rivalry, then they are referred to as ‘super plus factors’. Kovacic and others (2011) enumerate seven examples of super plus factors.71 We will return to super plus factors in the policy recommendation section of the article. The strength of the incentive to consider explicit collusion depends upon the characteristics of a given product/industry/market. It is useful to organize thinking on this issue through the use of Porter’s Five Forces. The center force is interfirm rivalry. The perimeter forces are the threat of entry, the presence of good substitute products, bargaining power of suppliers, and bargaining power of buyers. For a given product/industry/market, the following perimeter forces lead to low profits Entry is relatively easy Good substitutes are available Suppliers have bargaining power Buyers have bargaining power The converse leads to high profits Entry is difficult There are no good substitutes Suppliers have little bargaining power Buyers have little bargaining power If a given product/industry/market has perimeter forces that lead to high profits, but the few firms in the industry are making a homogenous product and competing largely on the basis of price, we then have a scenario where the suppression of rivalry through explicit collusion potentially generates a large payoff. The perimeter forces are like a protective wall that the firms can take advantage of if they just bring their interfirm rivalry under control. Bringing interfirm rivalry under control does little good if there is a gaping hole in the wall such as the availability of very good substitutes, extremely low entry barriers, or powerful buyer resistance. If the product is differentiated and competition is largely not based on price there can still be plenty of profit opportunities from explicit collusion if the wall is high enough, albeit the suppression of rivalry may be more complicated. Often cartels are not all-inclusive. An outside firm may have been asked to join the cartel but chose not to, or, the cartel may not have invited some firms to join the cartel. The presence of non-cartel firms can be corrosive to a cartel in the long-run. But if a cartel is effective in suppressing interfirm rivalry amongst its members, then the cartel may begin to function as a dominant firm. As a dominant firm it may exert pressure on corrosive non-cartel firms to join the cartel or just try to eliminate them from the market. Also, as cartel profits rise the protective wall of the perimeter forces may become a focus of attention. For example, the cartel may invest in conducts to raise entry barriers or degrade the perception of substitute products in the market. As an extra source of incremental profits, sophisticated cartels sometimes implement policies that discourage entry. These policies include capacity investments, lobbying for favourable laws and regulations, or making some other investment that preserves their cartel profits by discouraging entry into the cartelized market. Coordinating these investments is likely to require significant planning and communication.72 A similar need for communication may arise when a cartel decides to respond to potential or actual entry. For example, in the global alkali cartel of the interwar era a leading cartel member predatorily cut prices in the local market of a South American entrant73 and it then contacted another cartel member that was irritated by the price cut to explain the cut was temporary and was being used to drive out an entrant.74 Extensive communication is often important for a stable and profitable cartel, but it is also a major hazard. Records of cartel communications are used to establish the agreement required to prove a price fixing violation. As the volume of communication grows so does the risk of detection.75 Clever price-fixers think carefully about the trade-off between the profit maximizing cartel structure and the desire to minimize the risk of cartel detection and conviction of cartel members. One of the main contributions of this article, which we develop in Section IV, is an analysis of how this trade-off is improved for cartelists who are active in more than one market. D. Limits to deterrence Antitrust enforcers are pleased when they successfully prosecute price fixing, but the overarching goal is to deter price fixing so that prosecution is unnecessary. Becker observed that rational violation of the law can be deterred if sufficiently large expected penalties are imposed on violators. Expected penalties are determined by the probability of detection and conviction, and the magnitude of the penalty. Thus, antitrust law can deter rational price fixers if the penalties and the probabilities of detection and conviction are sufficiently large.76 In this article, we explore reforms that tailor enforcement policy to better deter serial collusion by large multi-product firms. Obviously, a low probability of detection does not necessarily preclude effective deterrence of price fixing. As long as any cartel can be detected with positive probability then very large penalties can be chosen to make the expected penalty large enough to deter. Block and Sidak addressed this issue and facetiously asked: ‘Why Not Hang a Price Fixer Now and Then?’77 We do not and we will not. Thus, the initial task for antitrust scholars is to explain the limits on the magnitude of penalties. Fairness and efficiency both support limits on maximum penalties. Penalties must be proportional to the culpability of the defendant to stay within the bounds of the Eighth Amendment. Policymakers should be especially concerned about severity of penalties if there is a risk of mistaken conviction. Harsh penalties may be inefficient if they crowd out the possibility of implementing marginal deterrence of additional or aggravated violations of the law. In the cartel setting, an escalating schedule of penalties might be used to marginally deter firms who are colluding in product 1 from extending their collusive agreement to product 2. Also, the incremental deterrent effect of increased financial penalties disappears as defendants near the point of being judgment-proof. Antitrust scholars most often cite concerns about the chilling effects of severe antitrust sanctions as a reason to limit the severity of penalties.78 The specter of severe penalties combined with the possibility that competitive behaviour could mistakenly be labeled collusive by a court might discourage socially desirable behaviour.79 For example, earnings calls by executives are possibly tools used to facilitate cartel behaviour, but generally, earnings calls provide socially valuable information to investors and improve the efficiency of securities markets. Possibly aggressive pursuit of potential price-fixers could chill disclosures in future earnings calls.80 Expected penalties can be calibrated by adjusting the magnitude of penalties or by adjusting the probability of detection and conviction. Adjustment of penalties is straightforward, but it is less clear how to adjust probabilities of detection and conviction. One possibility is to relax the rules characterizing behaviour as price fixing to make it easier to convict an alleged price-fixer. For example, the law could eliminate the requirement that plaintiffs prove an agreement exists and simply provide evidence of parallel behaviour. Concern about over-breadth make this sort of reform unrealistic and we will focus instead on choices by enforcement agencies that increase the probability of discovering signs of collusion and evidence that could be used to establish agreement. Historically, antitrust enforcers relied mostly on disclosures from insiders or complaints from buyers to identify possible collusion. Enforcement agencies are poorly positioned to access information suggestive of collusion. Successful cartels are careful to hide their agreements and as best they can—hide the implementation of the agreement. Economists have labored to identify collusion econometrically through publicly available data, but thus far the research has not translated into significant improvements in agency monitoring. There is one recent policy innovation that has generated many leads for enforcement agencies—the leniency program, we will discuss that program in Section IV.B. Now that we have discussed policy tools that can be used to calibrate cartel deterrence, we move to the question of whether deterrence is adequate today. The short answer is—no one knows, we lack both theory and especially the data to answer that question. We do want to make two crucial observations though. First, many antitrust commentators worry that expected penalties are too low to adequately deter price fixing.81 And second, current analysis of deterrence is guided by analysis of price fixing in the context of an isolated product or market. We conjecture that multi-product price fixing is more difficult to deter than price fixing in an isolated product, and thus join the ranks of commentators who favour more vigorous enforcement against cartels.82 In Section IV, we will explain that when multi-product competitors collude in multiple products or markets they may be able to increase the stability and profitability of their collusive agreement and possibly decrease the probability of detection. We explain how multi-market price fixing provides benefits in terms of economies of scope, better opportunities to hide the implementation of the collusive agreement and improved cartel stability in the face of external shocks.83 To preview the argument that deterrence is more difficult in the multi-product context consider the following example. Suppose firms A, B, and C agree to fix prices of product 1 and they each expect to earn $100 million from the agreement. Also, firms X, Y, and Z agree to fix prices of product 2 and similarly they each expect to earn $100 million from the agreement. Suppose the probability of detecting the collusion in each product is 1/4. Then, a penalty on each firm greater than $400 million if they are caught and convicted would be sufficient to deter a risk neutral decision maker acting on behalf of each firm. How does the story change if the firms merge in a pair-wise fashion so that A and X combine, B and Y combine, and C and Z combine? If they choose to collude in only one of the two products, then the analysis is unchanged. What if they choose to collude in both products? Each merged firm earns $200 million from the collusive agreement. Let us suppose that the penalty would be $800 million if they are caught and convicted of collusion in both products. It is not so clear what the probability of detection and conviction is likely to be in this new environment. Is it possible that competition authorities might detect collusion in one product, but not the other? If yes, how are the probabilities of detection in each product related? Perhaps the probability of detection declines because multi-product collusion is more stable,84 or because tactics to evade detection are more effective in the multi-product setting.85 In Section IV, we suggest that the probability of detection may go down, and is not likely to rise very much. We also conclude that the average duration and per period profit from collusion is likely to grow and thus $200 million understates the gain from price fixing in both products 1 and 2. Increased profit combined with a probability of detection that falls (or does not rise very much) means that deterrence would be more difficult. E. Deterrence of owners, top management, and middle managers Our discussion of collusion in hypothetical products 1 and 2 supposed that the decision-maker who chose whether to collude was a faithful agent of the firm. Faithful in the sense that the decision-maker acted based on the expected costs and benefits to the firm. It may be the case that decisions to collude are made by managers who have different objectives from the owners of the firm. In extreme cases a manager might decide to collude because he or she will be rewarded for the short-term increase in firm profit. The manager might not care much or at all about the expected costs from detection especially if the manager plans to leave the firm soon. When these ‘rogue managers’ are responsible for collusion it might be difficult or impossible to deter collusion with financial penalties on the firm. Optimists believe that antitrust enforcement against cartels is functioning effectively and when a cartel is detected they attribute it to a rogue manager.86 They believe that good corporate governance implies that top managers act on the preferences of owners87 and are adequately deterred from collusion given current enforcement practices. The optimists typically endorse compliance training and criminal punishment of rogue managers as enforcement tools that increase deterrence of rogues and mitigate agency problems within firms that sometimes allow rogues to flourish. We agree that compliance training and criminal punishment of price fixing managers have social value as enforcement tools, but we also are concerned that compliance training is sometimes used to teach managers how to collude, and that penalties against rogue managers may be offset by indemnification or some form of compensation to managers who suffer personal penalties after a collusive agreement is uncovered.88 Furthermore, we are critical of commentators who place too much emphasis on the problem of rogue managers.89 Widespread multi-product price fixing seems to be inconsistent with the notion that rogue managers are responsible for collusion.90 Returning to our earlier observation, we note again that the features of the chemical cartel seem inconsistent with the notion that collusion arises independently at a division level. It seems unlikely that three different rogue managers from a company like BASF independently sought the services for Fides/AC Treuhand in aid of collusion. Further, the ways that Fides/AC Treuhand coordinated the activities of chemical manufacturers across different product lines suggests top management was likely involved in these cartels. If we are right, then enforcement policies that worry very much about addressing the problem of rogues are likely misguided. We fear that inadequate deterrence means that top management and perhaps even shareholders91 sometimes welcome price fixing agreements as a tool for increasing expected profits—even after factoring in the expected costs from detection and punishment.92 Our fear grows out of the evidence we discussed in Section II concerning persistent serial collusion across a variety of industries. If top management is thoroughly implicated in serial collusion then perhaps it would be desirable to impose tougher sanctions on corporate officers responsible for price fixing.93 IV. SERIAL COLLUSION BY MULTI-PRODUCT FIRMS—WHY IS IT DIFFERENT? Suppose a large multi-product firm manufactures and sells well over 50 products and suppose that the same firm is engaged in explicit collusion with rivals at any point in time in 10 of those products. The large multi-product firm can clearly treat each cartel as a separate independent collusive activity, and thus the analysis of the previous section would be directly relevant, 10 times. However, by viewing the 10 cartels together, and solving cartel problems across the 10 cartels, or some subset of those cartels, the large multi-product firm can increase the aggregate profitability of the 10 cartels, decrease detection risk, and resolve the problems of a given cartel more easily. A. Resolving the issues of a given cartel across cartels Serial multi-product cartels are likely to be more resilient to economic shocks and defection by cartel members who may be tempted to depart from the cartel or cheat other cartel members. One reason is that multi-product cartel members that defect face stronger punishment—the potential loss of cartel profit in multiple products, not just one.94 Alternatively, cartels may find it easier to mollify potential defectors by making side-payments in one product to offset losses they might suffer in another product. Multi-market contact provides greater opportunities to hide these side payments from antitrust enforcers; they may be disguised in subcontracts, joint marketing arrangements or patent licenses. The opportunities for such hidden transfers may exist in single-product cartels but they are likely to be more available and can take more complex forms in multi-product cartels. The European Commission noted in the vitamins conspiracy decision that BASF exited the production of thiamin (vitamin B1) in 1989.95 At the same time, Roche, who was a duopolist in the production of beta carotene, accommodated the expansion of BASF’s market share of beta carotene from 12 per cent of worldwide market in 1988 to 17 per cent in 1989, and an average share of 22 per cent for 1990 to 1998.96 The Commission decision does not connect these two events. However, a unilateral decision by BASF to exit thiamin production in the midst of a conspiracy that would have increased thiamin profits does not make sense.97 Furthermore, it makes little more sense for Roche to unilaterally accommodate the expansion of beta carotene production by BASF when Roche was dominant in that product, and thereby erode near-monopoly profits. More plausibly there was a mutual agreement among the two leading vitamin cartel firms to have one exit the thiamine market and get compensated with accommodation of their share expansion in the beta carotene market.98 Coordinated product exit/contraction and entry/expansion is difficult to observe. For large multi-product serial colluders, exit from a single product line and entry into a different product line would likely be viewed as immaterial, and thus not included in filings to the SEC or other regulatory bodies. But we have some window into this potential tactic through European Commission cartel decisions. The four chemical cartel cases listed below featured exit by a large multi-product cartel participant from one of the cartelized markets in the midst of the explicit collusion. Given a well-functioning cartel, it is unlikely that the cartel price was so low that continued sales were unprofitable for the exiting firm.99 It also seems unlikely that the exit was rooted in a fear of being caught, especially since the exiting firm joined the cartel at or near its inception. Given that each of the exiting firms is a potential serial colluder, it seems unlikely that collusion was discovered internally in the midst of the collusion and upper management insisted on exit from the product in question. In this light, we conjecture a cartel portfolio explanation—namely, there is some hidden countervailing transaction in another product, perhaps accommodated entry, which compensates these firms for exiting the products listed below. For the Butadiene Rubber and ESBR cartel, the duration of the infringement was 20 May 1996 to 28 November 2002.100 Shell and Dow were participants in the cartel at the start. Shell’s market share was larger than Dow’s. Nevertheless, Shell sold its entire synthetic rubber business to Dow in the middle of 1999.101 For the Methacrylates cartel, the duration of the infringement was 23 January 1997 to 12 September 2002.102 ICI, Degussa, and Elf Aquitaine were participants from the start. Nevertheless, ICI sold the entirety of its business to Lucite on 11/1/1999, and from that date forward Lucite was a participant in the cartel.103 For the MCAA cartel, the duration of the infringement was 1 January 1984 to 7 May 1999.104 Akzo, Arkema, Elf Aquitaine, and Hoechst were participants from the start. Nevertheless, Hoechst sold its MCAA business to Clariant on 1 July1997 and then Clariant participated in the cartel until 7 May 1999.105 For the Polypropylene cartel, the duration of infringement was 1977–83.106 Rhone Poulenc was a member of the cartel from the start, however, Rhone Poulenc sold its polypropylene business to BP in 1980.107 Besides using coordinated entry and exit decisions to reallocate cartel profit across cartel members, multi-product cartels can use joint ventures, subcontracts, and patent licenses to reallocate profits internally. Joint ventures were common among members of recent electronics cartel. For example, LG began participation in the cathode ray tube cartel in December of 1997. In April of 2001 it formed a production joint venture with another cartel member, Phillips, and the joint venture became a cartel member. When cartel members form production and marketing joint ventures that displace their own production and marketing of a cartel product it provides a great opportunity to adjust output and reallocate cartel profit. By comparing European Commission cartel decisions to joint venture approvals we were able to identify six other joint ventures between electronics cartel members covering a cartelized product that started before or during the cartel: NEC and Hitachi joint venture in DRAMs, LG and Phillips joint venture in LCDs, Hitachi and Mitsubishi joint venture in smart card chips, Hitachi and LG joint venture in optical disk drives, Toshiba and Samsung joint venture in optical disk drives, and Sony and NEC joint venture in optical disk drives.108 Joint ventures may be subject to regulatory scrutiny and we would hope that a vigilant competition authority would not approve joint ventures that facilitate cartel activity,109 but the other tactics for reallocating cartel profits are not likely to face regulatory scrutiny. Exit/contraction properly crafted, is not reviewable by any regulatory authorities.110 Entry/expansion is invariably viewed as a unilateral decision of a firm that is considered to be part of the normal workings of the marketplace. Thus, coordinated exit/contraction and entry/expansion decisions are almost invisible to enforcement authorities. Also, we are unaware of any inquiry regarding such conduct in civil litigation.111 Patent licenses have been used to implement cartels and cross-licensing agreements can certainly be used to reallocate profit within a cartel.112 Subcontracts are sometimes used to reallocate cartel profit,113 but really any sort of contract relationship between cartel members might serve that purpose. When multi-product serial colluders enjoy a dense set of contractual relationships (joint ventures, patent licenses, subcontracts, or other) it makes it easier to hide cartel side-payments, and, as a bonus to the cartel, the threat of losing these other contractual relationships in response to cheating helps stabilize cartels. A different sort of benefit that arises from serial multi-product collusion concerns strategies that reduce the risk of cartel detection. Multi-product cartels can better hide price increases from buyers compared to single-product collusion. Buyer resistance to price increases may diminish because multi-product collusion can remove the competitive benchmarks that buyers use to make judgments about the presence of a cartel. They have more trouble judging whether a price is ‘too high’ and whether price increases are the result of cartelization or some other cause. Coordinating price increases across multiple products enhances the credibility of sellers’ assurances that price increases are grounded in genuine overall cost and demand pressures. A procurement division of a large buyer that gets a common justification for price increases across multiple products lines may be more willing to accept the price increases than resist them. B. Strategic use of amnesty to punish deviators—the ‘Chain Store Paradox’ for serial multi-product colluders114 A different version of our argument about more vigorous punishment in multi-product cartels is based on antitrust leniency programs. The leniency programme allows a cartel member to avoid criminal liability for price fixing by making the first disclosure of the cartel to enforcement officials. In a single-product setting this policy works well to destabilize cartels.115 In the case of serial multi-product collusion, it might be used by a prominent cartel member to punish defectors by sacrificing collusive profit in one product, with the goal of strengthening the punishment threat and the stability of the cartel in other products. Amnesty is typically characterized as a race to the antitrust authority, where the payoff for being first to admit to a conspiracy is the absence of a fine, provided that compelling evidence of a conspiracy is put forward. In other words, it is typically seen as a unilateral decision with no benefit to the amnesty awardee beyond the current cartel in question. However, if the amnesty applicant is a large multi-product firm that is running tens of cartels in diverse products, is there a benefit to amnesty beyond the one cartel for which it sought amnesty? Often large multi-product firms engage in explicit collusion with smaller firms, perhaps firms that make a single product. A large multi-product firm may be involved in a cartel in vitamins, a cartel in concrete additives, and a cartel in synthetic rubber. Small single product firms in each of the cartels may make only a certain range of vitamin products, or only concrete additives, or only synthetic rubber. There is no reason to believe that these smaller firms would have knowledge of cartels in other products, and these firms may not even be aware that the large multi-product firms in their cartel makes these other products. Recall that the central problem of explicit collusion is cheating by members. If a small cartel firm with an 8 per cent market share within the cartel cheats on the cartel agreement, expanding its market share to as much as 10 per cent, it may not be worthwhile for a large multi-product firm with a within-cartel share of 40 per cent to punish the deviating smaller firm. Across tens of cartels this kind of conduct by small single product firms can be, in aggregate, very costly in terms of foregone collusive profits to the large multi-product cartel firm. Prior to amnesty, how would a large multi-product serial colluding firm mitigate this kind of conduct by small single product firms? A punishment within one cartel would be almost surely invisible to any other cartel since the small single product firms operate in different product markets. Even intentionally revealing a cartel to authorities would not get much attention since prior to amnesty such an action would likely be viewed as an accidental detection or a one-off incident. However, amnesty can be used by a large multi-product firm to greatly mitigate cheating by small, single-product firms across its portfolio of cartels. The European Commission makes it clear in their decisions who is receiving full amnesty and their decisions are in the public domain. It is reasonable to presume that any small firm that is engaged in a conspiracy with a large multi-product firm will be aware that their large multi-product co-conspirator has sought amnesty in a different cartelized product. In light of this information, the small single product firm will likely inquire of their large multi-product co-conspirator why they took such an action. The large multi-product firm will explain that they are not willing to tolerate deviations by small firms in their cartels and when these deviations occur they will seek amnesty. This reputation for toughness by the large multi-product serial colluders can have two benefits for cartel payoffs.116 First, cheating by small co-conspirators will be substantially deterred across their entire portfolio of cartels. Second, if there are products that the large serial colluding firm wants to cartelize, but smaller firms in that industry have been reluctant because they believe cheating by others will greatly reduce cartel payoffs making the prospective venture not worth the risk, then for those products, cartels will be more likely to form. One would expect a large multi-product serial colluder to account for the external benefits to other existing and potential cartels when seeking amnesty in a single cartel. This line of argument might explain why two multi-product chemical firms, Akzo Nobel and Degussa, applied for and received full amnesty in particular chemical product cartels. First, Akzo sought amnesty for a cartel in which Degussa was a member.117 But Akzo and Degussa were participating in more than that one cartel together at the time of the Akzo amnesty application. If Akzo was motivated by the standard unilateral logic for seeking amensty, why did not Akzo report the other cartels at that time? Shortly thereafter Degussa sought and received full amnesty for their participation in a second cartel in which Akzo was a participant.118 Note that the cartels in which Akzo and Degussa were participating at the time did not end when Akzo first applied for amnesty. Soon after Degussa’s application for amnesty in the second product Akzo applied for and received amnesty for yet a third cartel, which begs the question why did not Degussa report the cartel in the third product when it applied for amnesty in the second product?119 It is a reasonable conjecture that Akzo and Degussa were working together to send warning signals to deviant firms. It is also reasonable to conjecture that Akzo and Degussa applied for leniency in the markets sequentially to escalate the punishment, alternating applications to share the benefit from leniency. C. Monitoring compliance with cartel rules is likely an increasing Returns Technology Economies of scale and scope are an easy to understand advantage of serial multi-product collusion. Likely, there are efficiencies associated with monitoring the actions of cartel members across multiple related product lines. In other words, checking firm A’s prices for products 1 and 2 is probably less than twice as costly as simply checking firm A’s price for product 1. Next, colluders and their allies (like Fides/AC Treuhand) likely get better at managing collusion with experience. Some aspects of corporate culture, corporate organization, and corporate incentives that work to stabilize collusion in product 1 may be transferred to product 2.120 Finally, developing and using outside cartel consultants, like Fides/AC Treuhand, should be easier and less costly on a per cartel basis.121 In the recently prosecuted chemical cartels Fides/AC Treuhand helped the colluding firms capture these benefits from economies of scale and scope. Fides/AC Treuhand gathered monthly reports on production, sales, and pricing with the right to verify these reports via audit if necessary. It then disseminated the aggregate data to cartel members and advised the members about the health of the cartel.122 On occasion, Fides/AC Treuhand served as a sort of a court to resolve disputes between cartel members. We conjecture that Fides/AC Treuhand was especially helpful to the cartel in dealing with disputes that arose when a cartelized chemical was used as an input in the manufacture of another chemical by a cartel member. When one cartel member buys a product from another cartel member it is likely the seller will discount the price to the buyer, but the buyer would be expected to provide some form of compensation to the seller.123 The involvement of Fides/AC Treuhand124 in the implementation and operation of cartels is well documented in European Commission decisions, and this involvement can be characterized as substantial. The Organic Peroxide decision at paragraph 92 makes it abundantly clear that Fides/AC Treuhand had a thorough understanding of the pricing, allocation, and enforcement structures needed to avert secret deviations within the Organic Peroxide cartel and knew how to translate that understanding into actionable steps for the members of the cartel.125 Fides/AC Treuhand is the predominant firm in the European Commission decisions that we can identify that is in the business of assisting cartels, but it is not the only one.126 Fides/AC Treuhand is an AMC. Many AMCs are primarily involved in the innocuous activity of helping professional and industry/product groups set up and run conferences and other meetings. These AMCs have the expertise and leverage to secure venues and the associated services for conferences and meetings that smaller professional organizations (say, a highly specialized group of surgeons) do not have the infrastructure to put together in a cost-effective manner. However, there are several AMCs that list services or market themselves on their websites in ways that go beyond organizing conferences and into the realm of potentially assisting cartels.127 Examples of some of these include:128 benchmarking services within a trade association and across members of an industry, facilitating collaboration among members of industry, promotion of information sharing, and finding joint solutions to problems within an industry. Some of the Fides/AC Treuhand cartels overlap in terms of the cartel duration as determined by the European Commission.129 Apparently, Fides/AC Treuhand is either not asked by the European Commission about any other cartels in which they are involved or they refuse to answer such queries by authorities. Large multi-product firms will often cite their vast size as being the reason they cannot know of every potential product division involved with collusion. Such a rationale is almost surely absent for a cartel consulting firm such as Fides/AC Treuhand—their cartel advising business is likely self-contained and senior management of that activity is likely aware of all cartels in which Fides/AC Treuhand is involved. The lack of revelation by Fides/AC Treuhand of their involvement with other cartels suggests the possibility that the extent of their historical involvement in assisting cartels may be much greater than what is observed in the EC Decisions. V. SUGGESTED POLICY REFORMS With respect to policy reforms, it is worth keeping in mind that experienced cartel firms solve remarkably difficult problems in initiating and implementing a self-enforcing agreement to suppress rivalry. They use cartel structures that are tailored to the product/industry/market in question, and as conditions change they mutate the cartel structures to adapt. Any policy reform must account for the clever, malleable adaptability of experienced cartel firms. This being said, there are some common features of cartel structures that can form the basis for meaningful policy reforms. A policy reform that will be difficult for cartel firms to use to their advantage should substantially encumber (i) the ability of cartel firms to monitor one another, (ii) the ability of cartel firms to rectify unintentional deviations, and (iii) communication between cartel firms. A policy reform that will additionally be difficult for serial colluders to use to their advantage should empower cartel members that make a limited range of products, have smaller market shares, and have not been detected in previous cartels to be rewarded for revealing cartel conduct. Competition law and policy could take several steps to address the phenomenon of serial collusion and create stronger incentives for firms to comply with the prohibition against agreements to set prices, output levels, or distort other dimensions of rivalry. We suggest that firms and enforcement agencies, as a condition of resolving cases, examine the causes of collusion. We also propose that convicted colluders be subject to heightened scrutiny and reporting obligations designed to deter future collusion. Finally, enforcement agencies should give heavier weight to collusion as a factor in evaluating mergers and treat repeated prior collusive behaviour as an indicator that a merger may facilitate future collusive conduct. A common thread that runs through these proposals is that competition authorities should collect and make better application of what might be called ‘antitrust big data’. A great deal of contemporary discussion about the information services sector focuses on how firms use data analytics to improve their capacity to understand markets and target potential consumers. Over the course of decades (or, in the case of the USA, over more than a century), the operation of the competition system has generated an immense body of information about how firms form cartels and operate them. Before 1993, enforcement and litigation provided a crucial source of empirical knowledge about collusion. Scholarly studies such as Myron Watkins’s and Simon Whitney’s detailed industry case studies and popular accounts of cartels (for example, John Fuller’s study in the early 1960s of the electrical equipment cartel in the USA) drew chiefly upon the records of litigated cases and reports of government agencies, such as the Federal Trade Commission’s study from the late 1940s and early 1950s of the international petroleum cartel. The enforcement programs that developed after the early 1990s greatly expand the volume of publicly available information about how cartels take shape and operate. The impact of this modern outpouring of information about cartels is evident in the new scholarship and popular literature about cartels. The revelations of the lysine cartel provided the foundation for a great book (Kurt Eichenwald’s The Informant in 2000), as well as a less successful movie of the same name, where Matt Damon’s depiction of Mark Whitacre awkwardly straddles the boundary between drama and comedy. The landmark contributions of the modern scholarly literature about the operation of cartels, bid-rigging conspiracies, and the injury caused by illegal collusion frequently build on recent case records or enforcement agency decisions. Thus, in the current environment, government agencies can make use of four significant information streams: (i) judicial decisions, and the publicly available record of litigated cases, involving prosecutions by government agencies or private parties; (ii) nonpublic files that public agencies compile in the course of law enforcement investigations; (iii) government reports on individual industries; and (iv) academic research (such as individual company histories). The research presented in this article draws on all of these sources, mainly the first, third, and fourth. We have identified patterns of firm behaviour from our study of reported decisions involving cartel prosecutions, augmented by our own review of publicly available information about the relevant firms and government enforcement policy. We also use our experience in private litigation and in government agencies to interpret this information. From this work, we are convinced that fuller study of these and related information sources would enable the competition community as a whole, and government agencies in particular, to improve their understanding of how cartels function and how public policy might be adjusted to deter them more effectively. Presented below are our proposals for measures that would expand the policy of available data and apply what we already know more effectively in law enforcement and other areas of policymaking. A. Cartel reconstructions A cartel infringement reflects a serious failure of corporate governance, management, or operations. So long as human beings run companies, we cannot expect that firms will comply perfectly with all legal commands all the time. A realistic view of compliance is that firms can strive to minimize the severity of individual lapses and take measures to avoid the repetition of past infringements. Firms periodically experience catastrophes—for example, transportation disasters that result in a large loss of life, or operational calamities (eg Deepwater Horizon) that cause death and massive injury to property. In most instances, companies and government agencies (such as the National Transportation Safety Board) respond to these events by carrying out intensive ex post investigations that reconstruct the circumstances that led to the failure and identify improvements for the future. Commercial airlines, for example, devote great effort to incorporate lessons from operational failures into the training of flight crews. Cartel episodes should elicit the same effort by enforcement agencies and cartel participants to reconstruct the forces that led to the illicit collusion.130 As a condition for participation in a leniency program or for entering a plea agreement, or as an element of sentencing in a contested case, we would require the culpable firms to engage in an externally audited reconstruction of the cartel and convey the results of the review to the DOJ. The inquiry would be supervised by a monitor appointed from within DOJ or by an expert outsider approved by the Department. The ex post review would identify who conceived the collusive scheme, who approved its implementation, who cooperated in its execution, what incentives inspired the cartel’s formation, and which features of corporate management, governance, or culture facilitated participation in the cartel. Additionally, the review should reveal all individuals who provided advice and input regarding the initiation and implementation of the cartel. Importantly, the review must, when applicable, identify links between collusive activities related to the product under investigation, and activities that did in the past, or might in the future, facilitate collusion in other products that the defendant manufactures or otherwise controls.131 We see the cartel reconstructions as serving three ends. The first is to provide greater insights for public enforcement agencies about how cartels take shape and how they operate. A better understanding about the formation and functioning of cartels would put the public agencies in a better position: to detect cartels by identifying behaviour consistent with the coordination among rivals, and to devise superior remedies. We advocate that agencies augment stovepipe price-fixing investigations, expanding investigations across multiple product markets simultaneously. The second result would be the opportunity for firms to develop more effective compliance strategies. A necessary starting point for responding to a serious failure is to understand its causes. As noted above, these include incentives that may have induced the firm’s employees to engage in improper conduct, and weaknesses in internal monitoring and supervision that enabled wrongful agreements to form and prosper. An enforcement agency might decide, in determining punishment, to give more credit to a compliance program that emerged from a determined firm effort to undertake a careful self-assessment and retool its operations to cure existing deficiencies.132 Needless to say, certain firms may need stronger motivation to craft effective compliance programs and renounce price-fixing as a business model. For these firms, stronger deterrence via stiffer penalties, better detection, and an intensified threat of successful prosecution may be a desirable complement to the higher quality information disclosure that would come from cartel reconstruction. The third result is to enrich scholarship about cartels by giving scholars better insight into how cartels are created and managed. The modern renaissance of cartel-related scholarship has built significantly upon reported records of cartel cases, especially decisions of the European Commission. Fuller disclosure of information about the architecture and machinery of cartels would propel further advances in this literature, which in turn would illuminate paths for enhancements of public policy and legal doctrine.133 B. Settlement and monitoring Serial cartel participants should be subject to progressively more stringent monitoring by law enforcement institutions. Agreements reached in settlements or plea agreements or conditions imposed in sentencing should expand the ability of government agencies to monitor the behaviour of cartelists. For a serial cartelist, either a DOJ official or a monitor approved by DOJ and/or the court could insist on regular reporting involving communications and transactions with competitors; adjustments in the incentives of sales personnel; and changes in capacity related to previously cartelized products.134 If a multi-product firm has been found to have participated in multiple cartels, then it should be subject to additional requirements beyond standard penalties and requirements for antitrust compliance training as part of any plea agreement. It should be required to report to the General Counsel’s office: Any and all communications to or from competitors, directly or indirectly, both within a product area and across product areas. Any and all transactions with competitors, directly or indirectly, both within a product area and across product areas. We define transactions broadly to include settlement of contract or other business disputes, intellectual property license, or in-kind transfer. It should also be required to submit the above information for review by outside counsel to determine if there are any indications of anti-competitive conduct. If there are any indications of anti-competitive conduct then a report must be made to the relevant antitrust enforcement agency. Such a report would immediately trigger the following actions with regard to all products, directly or indirectly, potentially impacted by the anticompetitive actions. The firms under investigation will provide the relevant information and sufficient data that would allow for a rigorous econometric evaluation of product prices so that a but-for price can be calculated for the period of the potentially inappropriate actions. This but-for price will then be used to assess whether the actual prices during the period in question were elevated above non-collusive levels. The firms under investigation will provide all information regarding the involvement of third parties in assisting with potential anti-competitive conduct for all products under consideration. The firms under investigation will provide all information regarding changes to the incentives of their sales force that involve shifts away from the pursuit of market share and toward the maintenance of elevated prices. The firms under investigation will provide all information regarding capacity changes for all products potentially impacted by the possible anticompetitive conduct. Changes in capacity include exit from and entry to all products potentially impacted by the possible anticompetitive conduct. The firms under investigation will provide all information regarding any conduct that would be considered a dominant firm conduct for all products potentially impacted by the possible anticompetitive conduct. C. Merger review Effective merger policy interacts with anti-cartel policy in two ways. First, absent compelling efficiencies, it should block horizontal mergers by firms attempting to regain the pricing discipline they had previously enjoyed until their cartel was uncovered and prosecuted. There has been a discouraging pattern of mergers in Europe following recent success prosecuting cartels.135 These mergers may have compromised much of the competitive gain that consumers expected to see from vigorous anti-cartel enforcement by the European Commission. Second, and perhaps more importantly, effective merger policy should block mergers that stabilize or encourage cartels. This second concern is subsumed within a broader concern known as coordinated effects.136 A coordinated effects problem arises if a merger ‘change[s] the competitive environment so that the remaining firms could or could more easily coordinate on price, output, capacity, or other dimension of competition’.137 Enforcement agencies should make use of cartel reconstructions and the results of expanded monitoring when reviewing mergers involving a serial cartelist. This is most important in building coordinated effects cases, where the propensity of firms to explore coordination with rivals, post-merger, should be a major factor in predicting whether a merger will have an adverse competitive effect. In the ordinary merger review, we propose that government agencies treat past collusive behaviour as a serious indication of likely future anticompetitive effects. The US Horizontal Merger Guidelines already recognize that collusion may be relevant to a coordinated effects analysis.138 But ‘[m]erger policies in the EU, the US, and many other jurisdictions may not adequately take into account the collusive history of merging firms.’139 Building analysis of collusive activity into coordinated effects review may be impeded if enforcers lack the sort of information that could be gathered in a cartel reconstruction or by monitoring the behaviour of serial colluders. And such analysis is surely impeded by ‘stove-piping’ which occurs for example when price fixing and mergers are handled by separate units in the Antitrust Division of the DOJ and those units do not communicate adequately.140 Remedies such as disclosure, divestiture of assets, selling minority shares in competitors, or licensure of intellectual property to competitors may change the nature of competition in the market and make collusion more difficult (see Marx & Zhou, 2015 regarding post-cartel mergers). This is particularly relevant if recidivism is industry-driven.141 D. Beyond leniency In Section IV.B, we explained that leniency programs can be manipulated by serial colluders to intimidate small firms. Perhaps beefed up monitoring of serial colluders will minimize abuse of leniency, and antitrust authorities can implement a leniency program that destabilizes cartels without unintended consequences. Regardless of how well manipulation of leniency programs can be controlled, we favour a complementary reform patterned after familiar bounty programs. We envision a bounty program that rewards certain small firms when they disclose they are participating with a serial colluder in a cartel. Specifically, the disclosing firm must have a relatively small market share, must make a limited range of products, must not be a cartel instigator, and must not have participated in any other cartels.142 The disclosing firm should be treated as a qui tam relator, exempt from all criminal penalties and paid a percentage of the fines recovered by enforcement authorities from the other cartel members. Of course, creating qui tam incentives for small cartel members will create a disincentive for large multi-product firms to include these smaller firms in cartels, which is socially beneficial. In the vitamins cartel it can be argued that Roche and BASF were colluding from 1985 to 1989 but then expanded the collusion to many other producers in the industry resulting in substantial price increases in vitamins from late 1989 to the end of the cartel. We conjecture that if Roche and BASF faced substantial disincentives to include smaller firms in the cartel then the social harm from the collusion would have been greatly reduced. Given the disincentive to have smaller firms join the cartel, a natural response by serial colluders will be to attempt to force smaller firms from the market at the inception of the collusive conduct. But, as noted in Kovacic and others,143 predatory conduct against a small firm when no firm in the industry is a dominant firm is a super plus factor—it almost surely cannot occur without explicit collusion by the large firms in the industry. In fact, as a more general point, antitrust enforcement authorities should conduct an extensive review when a small firm in an industry is claiming predatory conduct by larger firms that are known serial colluders. E. Cartel facilitators In Section IV.C, we documented the many ways that Fides/AC Treuhand aided cartels by gathering information, analysing cartel problems, and guiding cartels members to adopting practices that increased cartel stability and profitability. Yet, European cartel enforcers initially made little effort to punish or deter repeated cartel involvement by this cartel facilitator. Enforcers have noted in passing the presence of cartel facilitators in at least two other settings,144 but they do not seem to be a high priority. We believe anti-cartel policy has underestimated the significance of intermediaries such as Fides/AC Treuhand in the execution of successful collusive schemes. These advisors apply valuable know how accumulated in the course of managing a number of cartels over time. Consequently, we would elevate these enterprises as focal points of an anti-cartel programme. As a starting point, we would urge the application of penalties that recognize their significant contributions to illegal collusion, including injunctions that bar the consultancy from providing cartel management advice in the future. We also would insist that third party advisors reveal the full breadth of their cartel management experience as a condition of obtaining leniency in an individual case. Furthermore, we would demand that such advisors assist enforcement agencies in the reconstruction of past cartel episodes. We anticipate that cartel reconstructions and monitoring of serial colluders will improve the detection of cartel facilitators in the future. VI. CONCLUSIONS We documented in this article that many large multi-product firms have engaged in explicit collusion for extended periods of time across a variety of products with inclusive participation. Few of these cartels have short duration. The time periods of many of the cartels overlap but they are not coincident. We argued that this evidence points away from a rogue division manager explanation of price fixing. It fits better with our conjecture that many firms have embraced explicit collusion as a business model. The frequent use of an external cartel facilitator in the chemical industry provides further evidence for this conjecture. We explained that large multi-product firms enjoy multiple advantages when embracing explicit collusion as part of their business model. First, as noted in Section IV, they can address cartel issues across their portfolio of cartels rather than within each cartel in isolation. Second, they can use their experience in previous collusive conduct to enhance the profits of current and future cartels in which they are involved. Third, serial colluders can take advantage of some antitrust enforcement tools, such as leniency programs. Our policy recommendations begin with two premises. First, given the obvious evidence of serial collusion, antitrust enforcement should expand beyond stovepipe investigations of collusive conduct.145 Second, given that serial colluders are enormously clever and flexible, policy reforms should attack the underpinnings of cartel structures. In this light, we recommend that large multi-product firms classified as serial colluders by enforcement authorities be required, as part of any antitrust plea or settlement, to provide a full reconstruction of the history of the cartel as well as detailed information going forward of any and all (i) communications and information exchanges with competitors, (ii) transactions and/or transfers, including settlements from litigation, with competitors, (iii) sufficient data required by authorities to construct a reliable econometric model to construct a but-for price to compare to actual prices, (iv) communications with third parties that provided advice regarding anti-competitive conduct, (v) changes to the incentives of their sales force, (vi) capacity changes, including entry and exit, in any and all products, and (vii) exit and entry by rival firms. We also recommend that any serial colluder applying for a merger be subject to a mandatory coordinated effects review, even for products that are only produced by one of the merging parties. Finally, we recommend that small firms with no record of serial collusion be given not just leniency for reporting a cartel to antitrust authorities but be actually rewarded with a share of the fines collected by enforcement authorities. We also recommend that cartel enforcers give higher priority to the goal of detecting, punishing, and deterring firms like Fides/AC Treuhand that facilitate price-fixing agreements. Acknowledgement We are grateful to Ales Filipi, Erik Hovenkamp, and BU Law faculty workshop participants, and participants at the Competition Law Conference at BU. We are grateful to Alexandra Kaminsky, Luc Lallement, Arazi Lubis, Margaret Sabelhaus, Jinzi Sheng, and especially Charles Horgan for skillful research assistance. Of course, all errors are our own. Appendix A EC Decisions and Cartel Firms listed in Fig. 1A, 1B, and5 1. Bitumen: Case COMP/38.456—Bitumen—NL, 13 September 2006 a. Shell 2. Butadiene Rubber, Case COMP/F/38.638—Butadiene Rubber and Emulsion Styrene Butadiene Rubber, 29 November 2006 a. Bayer, Shell 3. Calcium Carbide, COMP/39.396—Calcium carbide and magnesium-based reagents for the steel and gas industries, 22 July 2009 a. Akzo Nobel, Degussa 4. Candle Waxes, Case COMP/39181—Candle Waxes, 1 October 2008 a. Shell 5. *Cartonboard, IV/C/33.833—Cartonboard, 13 July 1994 a. Fides/AC Treuhand 6. Chloroprene Rubber, COMP/38629—Chloroprene Rubber, 5 December 2007 a. Bayer 7. Choline Chloride: Case COMP/E-2/37.533—Choline Chloride, Comm’n Decision, 9 December 2004 a. Akzo Nobel, BASF 8. Citric Acid: Case COMP/E-1/36.604—Citric Acid, Comm’n Decision, 2002 OJ(L239) 18. 5 December 2001. a. Bayer 9. *Fatty Acids: IV/31.128—Fatty Acids, Comm'n Decision, 2 December 1986 a. Fides/AC Treuhand 10. Food Flavor Enhancers: Case COMP/C.37.671—Flood Flavour Enhancers, Comm’n Decision 2004 (L 75) 17 December 2002 a. <None from those listed in Fig. 5> 11. Heat Stabilizers, COMP/38589—Heat Stablisers, 11 November 2009 a. Akzo Nobel, Arkema/Atofina, Elf Aquitaine, Fides/AC Treuhand 12. *Hydrogen Peroxide, IV/30.907—Peroxygen products, 23 November 1984 a. Atochem, Solvay, Degussa 13. Hydrogen Peroxide, Case COMP/F/38.620—Hydrogen Peroxide and Perborate, 3 May 2006 a. Akzo Nobel, Arkema/Atofina, Degussa, Elf Aquitaine, Solvay 14. Lysine, Case COMP/36.545/F3. Amino Acids, 7 June 2000 a. <None from those listed in Fig. 5> 15. Methacrylates, Case No COMP/F/38.645—Methacrylates, 31 May 2006 a. Arkema/Atofina, Degussa, ICI, Elf Aquitaine 16. Methionine: Case C.37.519—Methionine, Comm’n Decision, 2002 (L 255) 1. 2 July 2002 a. Degussa, Rhone Poulenc/Aventis 17. Methyglucamine: Case COMP/E-2/37.978—Methylglucamine, Comm’n Decision, 27 November 2002 a. Rhone Poulenc/Aventis 18. Monochloroacetic Acid: Case COMP/E-1/.37.773—MCAA, Comm’n Decision, 19 January 2005 a. Akzo Nobel, Arkema/Atofina, Elf Aquitaine, Fides/AC Treuhand, Hoechst 19. Organic Peroxides: Case COMP/E-2/37.857—Organic Peroxyde, Comm’n Decision, 10 December 2003 a. Akzo Nobel, Arkema/Atofina, Degussa, Fides/AC Treuhand, 20. *Polyethylene, IV/31.866, LdPE, 21 December 1988 a. Atochem, BASF, Bayer, Dow, Enichem, Fides/AC Treuhand, Hoechst, ICI, Repsol, Shell 21. *Polypropylene, IV/31.149—Polypropylene, 23 April 1986 a. Atochem, BASF, Fides/AC Treuhand, Hoechst, ICI, Rhone Poulenc/Aventis, Shell, Solvay 22. *Potash, IV/795—Kaliand Salz/Kali Chemie, 21 December 1973 a. BASF, Solvay 23. *PVC, IV/31.865, PVC, 21 December 1988 a. Atochem, BASF, Enichem, Fides/AC Treuhand, Hoechst, ICI, Shell, Solvay 24. Rubber Chemicals: Case COMP/F/38.443—Rubber Chemicals, Comm’n Decision 21 December 2005 (summary at 2006 (L 353) 50) a. Akzo Nobel (through Flexsys),146Bayer 25. *Soda Ash, COMP/33.133-B: Soda-ash, 19 December 1990 a. BASF, Solvay 26. Sodium Chlorate, Case COMP/38.695—Sodium Chlorate, 11 June 2008  a. Akzo Nobel, Arkema/Atofina, Elf Aquitaine 27. Sodium Gluconate, http://europa.eu/rapid/press-release_IP-01-1355_en.htm?locale=en#file.tmp_Foot_1, 19 March 2002 a. Akzo Nobel 28. Sorbates: Case COMP/E-1/37.370—Sorbates, Comm’n Decision 1 October 2003 a. Hoechst 29. *Synthetic Fibers, IV/30.810 - Synthetic fibres, 4 July 1984 a. Bayer, Hoechst, ICI, Rhone Poulenc/Aventis 30. Vitamins: Case COMP/E-1/37.512– Vitamins, Comm’n Decision, 2001 O.J. (L6) 21 November 2001 a. BASF, Rhone Poulenc/Aventis,Solvay 31. *Woodpulp, IV/29.725 - Wood pulp, 19 December 1984 b. Fides/AC Treuhand 32. Nitrile Butadiene Rubber, COMP/38.628 - Nitrile Butadiene Rubber, 23 January 2008 a. Bayer EC Decisions and Cartel Firms listed inFig. 2and6 1. DRAMs, COMP/38511—DRAMs, 19 May 2010 a. Hitachi, Mitsubishi, Samsung, Toshiba 2. LCD, COMP/39.309—LCD—Liquid Crystal Displays, 8 December 2010 a. Samsung 3. TV and computer monitor tubes, AT.39437—TV and computer monitor tubes, 5 December 2012 a. Panasonic, Samsung, Toshiba 4. Smart Card Chips, AT.39574—Smart Card Chips, 3 September 2014 a. Hitachi, Mitsubishi, Samsung 5. Optical Disk Drives, Case AT.39639—Optical Disk Drives, 21 October 2015 a. Hitachi, Samsung, Sony 6. Power Transformers, COMP/39.129—Power Transformers, 7 October 2009 a. Hitachi, Toshiba 7. Gas Insulated Switchgear, COMP/F/38.899—Gas Insulated Switchgear, 24 January 2007 a. Hitachi, Mitsubishi, Toshiba 8. Power Cables, AT.39610—Power Cables, 2 April 2014 a. Hitachi, Mitsubishi 9. Alternators and Starters, AT.40028—Alternators and Starters, 27 January 2016 a. Hitachi, Mitsubishi 10. Professional Videotape, COMP/38.432—Professional Videotape, 20 November 2007 a. <None from those listed in Fig. 2 and 6 11. Elevators and Escalators, COMP/E-1/38.823—PO/Elevators and Escalators, 21 February 2007 a. Mitsubishi 12. CRT Glass, COMP/39605 CRT Glass, 19 October 2011 a. Samsung 13. Rechargeable Batteries, AT.39904—Rechargeable batteries, 12 December 2016 a. Panasonic,Samsung, Sony 14. Refrigeration Compressors, COMP/39600—Refrigeration Compressors, 7 December 2011 a. Panasonic 15. Capacitors, AT.40136—Capacitors, 21 March 2018 a. Panasonic,Hitachi DoJ Decisions and Cartel Firms listed inFig. 3 and 7 1. Airbags: Case 5:12-cr-20383-JCO-LJM, 6 June 2012 a. Autoliv 2. Airbags: Case 3:14-cr-00349-JZ, 29 September 2014 a. Toyoda 3. Airbags: Case 2:12-cr-20491-BAF-LJM, 30 July 2012 a. TRW 4. Air Conditioning Systems: Case 2:13-cr-20713-GCS-PJK, 26 September 2013 a. Valeo 5. Air Flow Meters: Case 2:13-cr-20707-MOB-LJM, 26 September 2013 a. Hitachi 6. Alternators/Starters: Case 2:13-cr-20707-MOB-LJM, 26 September 2013 a. Hitachi 7. Alternators/Starters: Case 2:13-cr-20712-PJD-DRG, 26 September 2013 a. Mitsuba 8. Alternators/Starters: Case 2:13-cr-20710-RHC-MAR, 26 September 2013 a. Mitsubishi 9. Alternators/Starters: Case 5:15-cr-20197-JCO-APP, 31 March 2015 a. Robert Bosch 10. Anti-Vibration Rubber Products: Case 3:14-cr-0068-JZ, 13 February 2014 a. <None from those listed in Fig. 7> 11. Anti-Vibration Rubber Products: Case 3:13-cr-0059-JZ, 6 February 2014 a. <None from those listed in Fig. 7> 12. Anti-Vibration Rubber Products: Case 3:13-cr-00439-JJH, 26 September 2013 a. <None from those listed in Fig. 7> 13. Automotive Brake Hoses: Case 3:14-cr-00394-JJH, 31 October 2014 a. Hitachi 14. Automotive Brake Hoses: Case 3:14-cr-00349-JZ, 29 September 2014 a. Toyoda 15. Automotive Wire Harnesses: Case 2:12-cr-20064-DML-MKM, 30 January 2012 a. Yazaki 16. Bearings: Case 1:13-cr-00104-HJW, 26 September 2013 a. <None from those listed in Fig. 7> 17. Bearings: Case 1:13-cr-00103-HJW, 26 September 2013 a. <None from those listed in Fig. 7> 18. Compressors/Condensers: Case 2:13-cr-20711-AJT-LJM, 26 September 2013 a. Mitsubishi 19. Electronic Control Units: Case 2:12-cr-20063-DML-DRG, 30 January 2012 a. Denso 20. Electronic Throttle Bodies: Case 2:13-cr-20707-MOB-LJM, 26 September 2013 a. Hitachi 21. Fan Motors: Case 2:13-cr-20712-PJD-DRG, 26 September 2013 a. Mitsuba 22. Fuel Injection Systems: Case 2:13-cr-20707-MOB-LJM, 26 September 2013 a. Hitachi 23. Fuel Ratio Sensors: Case 2:14-cr-20494-MOB-DRG, 19 August 2014 a. NGK Spark Plug 24. Fuel Senders: Case 2:12-cr-20064-DML-MKM, 30 January 2012 a. Yazaki 25. Heater Control Panels: Case 2:12-cr-20063-DML-DRG, 30 January 2012 a. Denso 26. Ignition Coils: Case 2:13-cr-20707-MOB-LJM, 26 September 2013 a. Hitachi 27. Ignition Coils: Case 2:13-cr-20710-RHC-MAR, 26 September 2013 a. Mitsubishi 28. Instrument Panel Clusters: Case 2:12-cr-20064-DML-MKM, January 30, 2012 a. Yazaki 29. Inverters: Case 2:13-cr-20707-MOB-LJM, 26 September 2013 a. Hitachi 30. Motor Generators: Case 2:13-cr-20707-MOB-LJM, 26 September 2013 a. Hitachi 31. Oxygen Sensors: Case 5:15-cr-20197-JCO-APP, 31 March 2015 a. Robert Bosch 32. Oxygen Sensors: Case 2:14-cr-20494-MOB-DRG, 19 August 2014 a. NGK Spark Plug 33. Power Window Motors: Case 2:13-cr-20712-PJD-DRG, 26 September 2013 a. Mitsuba 34. Seatbelts: Case 5:12-cr-20383-JCO-LJM, 6 June 2012 a. Autoliv 35. Seatbelts: Case 2:12-cr-20491-BAF-LJM, 30 July 2012 a. TRW 36. Spark Plugs: Case 5:15-cr-20197-JCO-APP, 31 March 2015 a. Robert Bosch 37. Spark Plugs: Case 2:14-cr-20494-MOB-DRG, 19 August 2014 a. NGK Spark Plug 38. Speed Sensor Wire Assemblies: Case 2:12-cr-20215-MOB-LJM, 3 April 2012 a. <None from those listed in Fig. 7> 39. Steering Angle Sensors and Switches: Case 2:13-cr-20540-SJM-MAR, 18 July 2013 a. Panasonic, Toyoda 40. Steering Wheels: Case 2:12-cr-20491-BAF-LJM, 30 July 2012 a. TRW 41. Steering Wheels: Case 5:12-cr-20383-JCO-LJM, 6 June 2012 a. Autoliv 42. Steering Wheels: Case 3:14-cr-00349-JZ, 29 September 2014 a. Toyoda 43. Variable Valve Timing Devices: Case 2:13-cr-20707-MOB-LJM, 26 September 2013 a. Hitachi 44. Air Conditioning Systems: Case AT.39960—Thermal Systems, 3 August 2017 a. Denso, Valeo 45. Alternators/Starters: AT.40028—Alternators and Starters, 27 January 2016 a. Hitachi, Mitsubishi, Denso 46. Automotive Wire Harnesses: Case AT.39748—Automotive Wire Harnesses, 7 October 2013 a. Yazaki 47. Bearings: Case AT.39922-Bearings, 19 March 2014 a. <None from those listed in Fig. 7> 48. Compressors/Condensers: Case AT.39960—Thermal Systems, 3 August 2017 a. Denso, Valeo,Panasonic (e-compressors) 49. Radiators/Fans: Case AT.39960—Thermal Systems, 3 August 2017 a. Denso, Valeo 50. Spark Plugs: Case AT.40113—Spark Plugs, 21 February 2018 a. Bosch, Denso, NGK Spark Plug EC Decisions and Cartel Firms listed inFig. 4 and8 1. CEEMEA: Case 1:18-cr-00061-JSR—United States of America v BNP Paribus USA, Inc., 25 January 2018 a. <None from those listed in Fig. 8> 2. Euro Interest Rate Derivatives a. JPMorgan <http://europa.eu/rapid/press-release_STATEMENT-16-4307_en.htm> 7 December 2016 b. RBS—Case AT.39914, 4 December 2013 3. EUR—USD Forex a. JPMorgan—Case 3:15-cr-00079-SRU, 20 May 2015 b. RBS—Case 3:15-cr-00080-SRU, 20 May 2015 4. Swiss Franc a. JPMorgan, RBS,UBS Case AT.39924, 21 October 2014 Case AT.39924 (CHF LIBOR), 21 October 2014 5. Yen a. JPMorgan, RBS, UBS Case AT.39861, 4 February 2015 6. Municipal Bonds a. JPMorgan Hertz, James L.—Case 1:10-cr-01178, 30 November 2010 Wright, Alexander—Case 1:12-cr-00551-SAS, 18 July 2012 b. UBS Heinz, Gary—Case No 15-432—Gary Heinz and others, Petitioners v United States of America Ghavami, Peter—Case No 15-432—Gary Heinz and others, Petitioners v United States of America Welty, Michael—Case No 15-432—Gary Heinz and others, Petitioners v United States of America Zaino, Mark—Case 1:10-cr-00434-TPG, 19 May 2010 Footnotes 1 George W Stocking and Myron W Watkins, Cartels in Action: Case Studies in International Business Diplomacy (William S. Hein and Co., Inc., Buffalo, NY 1991) 516. 2 See eg David Autor and others, ‘Concentrating on the Fall of the Labor Share’ (2017) 107 Amer Econ Rev 180; Jay Shambaugh and others, The State of Competition and Dynamism: Facts about Concentration, Start-Ups, and Related Policies (June 2018) <https://www.brookings.edu/research/the-state-of-competition-and-dynamism-facts-about-concentration-start-ups-and-related-policies/> accessed 27 September 2018. 3 See Autor and others (n 2) 4 ibid. 5 See Shambaugh and others (n 2) 2; Jan de Loecker and Jan Eeckhout, The Rise of Market Power and the Macroeconomic Implications (2017); Sara Calligaris, Chiara Crisculolo, and Luca Marcolin, Mark-ups in the Digital Era <https://www.oecd-ilibrary.org/industry-and-services/mark-ups-in-the-digital-era_4efe2d25-en>. 6 See generally, Gustavo Grullon and others, Are US Industries Becoming More Concentrated? (2016) <https://finance.eller.arizona.edu/sites/finance/files/grullon_11.4.16.pdf>. 7 See Autor and others (n 2) 185; James Bessen, Information Technology and Industry Concentration (2017), <https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3044730> accessed 27 September 2018. 8 It is important to note that a large multi-product serial colluder often has co-conspirators that are single product firms. When we use the terminology ‘serial multi-product cartels’ we mean that there is at least one firm in the cartel that is a large multi-product firm and that has been involved in cartels in different product areas through time. Connor examined cartel convictions from 1990 to 2009 and found that ‘52 firms were members of seven or more cartels; 26 were in ten or more cartels; and six companies engaged in 20 or more cartels … . These top recidivists are primarily headquartered in the EU.’ John M. Connor, Recidivism Revealed: Private International Cartels John M Connor, ‘Recidivism Revealed: Private International Cartels 1990-2009’, 16. See Section II for more detailed empirical evidence regarding ‘serial multi-product cartels’. 9 More details will follow in Section II. In some cases we include cartel participation by predecessor firms that were acquired by the named firm. Connor reports that Hitachi participated in 20 international cartels with 18 of those ending in the current millennium. ibid 29. 10 A literature review by Connor indicates that ‘[c]artel studies generally conclude that only about 10 percent to 30 percent of all such conspiracies are discovered and punished.’ ibid 14, but we read these studies with caution! 11 See Allissa Wickham, EU Fines Ball Bearing Makers $1.3B For Price-Fixing,Law360, March 19, 2014, <https://www.law360.com/articles/519825/eu-fines-ball-bearing-makers-1-3b-for-price-fixing> accessed 27 September 2018 (‘To date, more than 100 products and 70 companies have been investigated in the commission’s car parts probe, according to Almunia. He noted that the EC is still examining alleged cartels in seat belts, air bags, steering wheels, air conditioning parts and lighting systems, among other car products.’); Matt Bernardini, Bayer, Dow Hit with Polyurerthane Price-Fixing Suit,Law360, 2 August 2018 <https://www.law360.com/articles/1068955> accessed 27 September 2018 (class action price-fixing lawsuit against Bayer, Dow and other alleging price fixing in two chemical markets); David Keohane and Jessica Dye, ‘BNP Paribas’ US Unit Pleads Guilty to FX Price-fixing Conspiracy’ Financial Times (26 January 2018) (BNP is ‘the sixth major bank to face criminal charges as a result of the DoJ’s investigation into antitrust and fraud crimes’ in the foreign exchange market) <https://www.ft.com/content/76dda416-02af-11e8-9650-9c0ad2d7c5b5> accessed 23 July 2018; Joe Rennison and Alexandra Scaggs, ‘US Treasury Dealers Accused of Collusion by Big Investors Paribas’ US Unit Pleads Guilty to FX Price-fixing Conspiracy’ Financial Times (16 November 2017) (10 banks including JP Morgan, Barclays, Goldman Sachs, UBS, and Morgan Stanley have been accused of fixing prices for US government securities) <https://www.ft.com/content/45bc40da-cab1-11e7-ab18-7a9fb7d6163e> accessed 23 July 2018. 12 See Section III.D. 13 European Commission cartel decisions document Fides/AC Treuhand’s involvement in seven chemical cartels (and another two in the wood products industry). We discuss this in detail in Sections II and IV.B. 14 See Christopher R Leslie, ‘Foreign Price-Fixing Conspiracies’ (2017) 67 Duke L J 557. 15 Jun Zhou, ‘The Rise and Fall of Cartels with Multi-market Colluders’ (2016) 48 Rev Indus Org 381: [C]artels with different markets but a common colluder usually start sequentially, with, on average, approximately 3 years between each starting date. A similar pattern exists for cartel dissolution where there is an overlap in membership between different cartels: there are, on average, approximately two-and-a-half years between each dissolution date. 16 The European Commission recently carried out unannounced inspections of chemical firms that purchase styrene monomers as part of an investigation of potential cartel conduct. The names of the raided firms were not released by the European Commission. See ‘European Commission Inspects Styrene Monomer Purchasing Firms Suspected of Antitrust Breaches’, ICISNews (8 June 2018). <https://www.icis.com/resources/news/2018/06/08/10229758/european-commission-inspects-styrene-monomer-purchasing-firms-suspected-of-antitrust-breaches/> accessed 27 September 2018. 17 See, Eric Janssen, ‘AC Treuhand and the Facilitating of Cartels’ Leg Know Port (6 November 2015) <https://legalknowledgeportal.com/2015/11/06/ac-treuhand-and-the-facilitating-of-cartels/> accessed 27 September 2018. AC Treuhand had detailed knowledge of the content of the cartel agreements and in fact had formulated and distributed all the information regarding prices, quota and customers. It was allowed to carry out audits in the offices of the cartel members. Only the data which had ultimately been approved by AC Treuhand could serve as the basis for the negotiations and the agreements. AC Treuhand had made its business premises available to keep the cartels hidden. According to the European Commission, it was the task of AC Treuhand to prevent the discovery of the two cartel infringements. 18 See, Rick Busscher, Martin Herz and Hans Vedder, ‘The Shortest Competition Judgment Ever; AC Treuhand II’ European Law Blog (9 November 2015) <http://europeanlawblog.eu/2015/11/09/the-shortest-competition-judgment-ever-ac-treuhand-ii/> accessed 27 September 2018. … as with any cartel, there needs to be supervision and enforcement. This is tricky for cartelists themselves, as the supervisor is exposed to the risk of being qualified as a ringleader, negating the possibility of immunity under the Leniency Guidelines. Hence, the heat stabilisers cartel thought of a better way, namely hiring an independent firm to ensure that parties stick to the cartel. 19 Euro Interest Rate Derivatives Cartel involved seven banks: Barclays, Deutsche Bank, The Royal Bank of Scotland, Societe Generale, Credit Agricole, HSBC, JPMorgan Chase. See Euro Interest Rate Derivatives (2017) 60 Off J Europ Union 17–20 <https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52017XC0630(02)#document1> accessed 27 September 2018. The first four banks settled with the EC in December 2013 and paid over €820 million in total. The last three paid €484 in total in December 2016 after refusing to settle. Swiss Franc Interest Rate Derivatives two price fixing agreements were prosecuted. The Royal Bank of Scotland and JPMorgan Chase were identified when RBS applied for immunity on 9 August 2011. Swiss Franc Interest Rate Derivatives, Chf Libor (2015) 58 Off J Europ Union 9–11 <https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=uriserv:OJ.C_.2015.072.01.0009.01.ENG> accessed 27 September 2018. ENG RBS received full immunity and JPMorgan Chase was fined over €61 million. Swiss Franc Interest Rate Derivatives, Bid Ask Spread Infringement (2015) 58 Off J Europ Union 14–16 <https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=uriserv:OJ.C_.2015.072.01.0009.01.ENG> accessed 27 September 2018. ENG The Yen Interest Rate Derivatives Cartel involved the United Bank of Switzerland, the Royal Bank of Scotland, Deutsche Bank, Citigroup, JPMorgan, RP Martin, and ICAP. Yen Interest Rate Derivatives (2015) 60 Off J Europ Union 10–14 <https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52017XC0915(02)> accessed 27 September 2018. 20 Evidence of collusion within municipal bonds from JP Morgan was obtained via DOJ cases against two JP Morgan employees: James L Hertz and Alexander Wright. See JPMorgan Chase Admits to Anticompetitive Conduct by Former Employees in the Municipal Bond Investments Market and Agrees to Pay $228 Million to Federal and State Agencies <https://www.justice.gov/opa/pr/jpmorgan-chase-admits-anticompetitive-conduct-former-employees-municipal-bond-investments> accessed 27 September 2018; In the Matter of Alexander W. Wright, <https://www.sec.gov/litigation/admin/2014/34-73074.pdf> accessed 27 September 2018. 21 See ‘Three Former UBS Executives Convicted for Frauds Involving Contracts Related to the Investment of Municipal Bond Proceeds’ Department of Justice, Office of Public Affairs (31 August 2012), <https://www.justice.gov/opa/pr/three-former-ubs-executives-convicted-frauds-involving-contracts-related-investment-municipal> accessed 27 September 2018; See ‘Former Employee of Financial Services Company Pleads Guilty for Role in Bid-rigging and Fraud Conspiracies Involving Proceeds of Municipal Bonds’ Department of Justice, Office of Public Affairs (19 May 2010) <https://www.justice.gov/opa/pr/former-employee-financial-services-company-pleads-guilty-role-bid-rigging-and-fraud> accessed 27 September 2018. 22 Other researchers have documented repeat cartel offenses by many firms. Zhou observes: ‘Cartels often involve firms that collude in more than one market … European Commission … cartel decisions for 1985-2014 shows that 112 (71%) of the prosecuted cartels involved multi-market colluders.’ Zhou (n 15) 381. Connor notes that: [d]uring 1990-2009 the French petrochemical company Total SA participated in 27 cartels with 24 of those ending in the current millennium. The French pharmaceutical firm Sanofi-Aventis participated in 22 with 8 of those ending in the current millennium. The German chemical company BASF participated in 21 with 17 of those ending in the current millennium. The French cement company Lafarge SA participated in 21 with 16 of those ending in the current millennium. The German pharmaceutical company Bayer AG participated in 20 with 5 of those ending in the current millennium. The Japanese conglomerate Hitachi Ltd. participated in 20 with 18 of those ending in the current millennium. Connor (n 8) 29. ‘In the set of 510 cartel members fined by the EC in 1998-2014,’ Marvão identifies 89 ‘multiple offenders’ (firms fined for collusion more than once), 10 ‘repeat offenders’ (firms which initiate a cartel after being investigated for another cartel), and 5 recidivists. Marvão, The Issue of Repeat Cartel Offences, Stockholm School of Economics (2015) <http://fpb.dev.kebbeit.lv/wp-content/uploads/2016/01/freepolicybriefs_feb1.pdf> accessed 27 September 2018. Marvão’s definition of recidivist follows that used by Gregory J Werden, Scott D Hammond and Belinda A Barnett, ‘Recidivism Eliminated: Cartel Enforcement in the United States Since 1999’ [2011] SSRN Electronic Journal: firms which initiate a cartel after being fined for another cartel. Marvao describes her work with Levenstein and Suslow and concludes: The authors show that there is a large share of repeat and multiple offenders in chemicals and a surprisingly high proportion of repeat offenders in the manufacture of transport and electrical equipment. The highest proportion of multiple offenders is found in pharmaceuticals and refined petroleum products. The transportation and storage market is a sector with a high incidence of collusion (83 convicted cartel members), but no repeat offenders. ibid discussing Margaret C Levenstein and others, ‘Serial Collusion in Context: Repeat Offenses by Firm or by Industry?’ OECD Global Forum on Competition DAF/COMP/GF (2015). Finally Levenstein and others review cartel convictions from 1990 to 2009 and note that ‘52 firms were members of seven or more cartels; 26 were in ten or more cartels; and six companies engaged in 20 or more cartels … .’ Margaret C Levenstein and others, ‘The Effect Of Competition On Trade: Evidence From The Collapse Of International Cartels’ (2015) 39 Int J Indus Org 56. 23 Westinghouse had 20 and General Electric had 19 antitrust convictions from 1955–1980. James M Clabault and Michael K Block, Sherman Act Indictments James M. Clabault & Michael K. Block, Sherman Act Indictments: 1955–1980, 905–11 (1981). Cited in Connor (n 8) 29. 24 The 1962 consent decree resolved the federal case against GE and Westinghouse related to the price fixing activities covered in Fig. 9. See Luis Cabral, GE and Westinghouse, NYU Stern (2010). <http://luiscabral.net/economics/teaching/gewestinghouse.pdf> accessed 27 September 2018. 25 ibid. 26 US Steel had 11 and Bethlehem Steel had 7 antitrust convictions from 1955–1980. Clabault and Block (n 23). Cited in Connor (n 8) 29. 27 Stocking and Watkins (n 1) 186: ‘Syndicates were set up to sell semifinished steel, structural steel, merchant bars, hoops and strips, and plates in export markets for all cartel members at uniform prices.’ 28 Ibid 187–89. Second international steel cartel included American and European companies from 1933 through 1938. It controlled semi-finished steel, structural steel, merchant bars, cold rolled bands and strips, hot rolled bands and strips, and medium plates, thick plates, galvanized plates, tin plates, wired rods, tubes and other products. ibid. 29 See Martin J Rosent, ‘The Brussels Entente: Export Combination in the World Steel Market’ (1958) 106 U Penn L R 1079. 30 See Margaret C Levenstein and Valerie Y Suslow, Price-Fixing Hits Home: An Empirical Study of US Price Fixing Conspiracies (Ross School of Business Paper No 1290) 7 (2015) <https://ssrn.com/abstract=2691579> or <http://dx.doi.org/10.2139/ssrn.2691579> accessed 27 September 2018. the firm U.S. Steel was involved in six different conspiracies between 1948 and 1969, some of which involved traditional steel products and producers, as in a 1963 case in which U.S. Steel, Bethlehem Steel and others were convicted of conspiring to eliminate competition in the production of wheels, others were less finished steel products, and yet others were in plastics. Some involved the same co-conspirators, but others were an entirely new set of firms. 31 See Catarina Marvão, The Issue of Repeat Cartel Offences, Free Policy Network, 1 February 2016, <http://freepolicybriefs.org/2016/02/01/the-issue-of-repeat-cartel-offences/> accessed 27 September 2018. 32 See George A Hay and Daniel Kelley, ‘An Empirical Survey of Price Fixing Conspiracies’ (1974) 17 J L Econ 13. 33 ‘Antitrust: Commission Fines Prestressing Steel Producers €458 Million For Two Decades Long Price-Fixing and Market-Sharing Cartel’ European Commission, 6 October 2010 <europa.eu/rapid/press-release_IP-10-1297_en.pdf> accessed 27 September 2018. 34 See Nick St. Denis, Three Steel Companies Hold their Ground in Price-Fixing Lawsuit, 30 November 2015 <https://www.usglassmag.com/2015/11/three-steel-companies-hold-their-ground-in-price-fixing-lawsuit/> accessed 27 September 2018. 35 See Reuters, German Cartel Office Deepens Steel Investigation, Searches Firms, 4 September 2017 <https://www.reuters.com/article/us-steel-antitrust/german-cartel-office-deepens-steel-investigation-searches-firms-idUSKCN1BF1E1> accessed 27 September 2018. 36 See Stocking and Watkins (n 1) x. 37 ibid 466. 38 ibid 274–303. 39 ibid 448. 40 ibid 454. 41 This trend is reflected in the proceedings, recommendations, and reports of international organizations that seek to promote the development of global competition policy norms. See International Competition Network, Cartel Working Group, Setting of Fines for Cartels in ICN Jurisdictions (2017); International Competition Network, Cartel Working Group, Compilation of Good Practices from the Anti-Cartel Enforcement Manual (25 November 2016); Organization for Economic Cooperation and Development, Fighting bid rigging in public procurement: report on implementing the OECD recommendation (2016); Organization for Economic Cooperation and Development, Hard Core Cartels: third report on the implementation of the 1998 OECD recommendation (2005); United Nations Conference on Trade and Development, Model Law on Competition (2010). 42 Andrew I Gavil and others, Antitrust Law In Perspective: Cases, Concepts And Problems In Competition Policy (3rd edn, West Publishing 2017) 6–17. 43 William E Kovacic, Chairman, US Federal Trade Commission, ‘Competition Policy in the European Union and in the United States: Convergence or Divergence?’ Bates White 5th Annual Antitrust Conference, 2 June 2008, Washington, DC. 44 Caron Beaton-Wells and Christopher Tran (eds) Anti-Cartel Enforcement In A Contemporary Age: Leniency Religion (Hart Publishing 2015). 45 Scott Hammond and Ann O’Brien, ‘The Evolution of Cartel Enforcement over the Last Two Decades: The U.S. Perspective’ Malgorzata Krasnodebska-Tomkiel (ed), Polish Office Of Competition And Consumer Protection, Changes In Competition Policy Over The Last Two Decades (Office of Competition and Consumer Protection, Warsaw, Poland 2010) 11. 46 See, eg European Commission, Press Release, Statement by Commissioner Vestager on Commission decision to fine seven companies for participation in four cartels (21 February 2018) <europa.eu/rapid/press-release_STATEMENT-18-1022_en.htm>. 47 See European Commission, Statement by Commissioner Vestager on fining Scania for participating in trucks cartel (27 September 2017) <europa.eu/rapid/press-release_STATEMENT-17-3509_en.htm> accessed 27 September 2018. In describing developments in the prosecution of a European truck manufacturers cartel, Margrethe Vestager, the Commissioner for Competition, said ‘We have fined the six companies of Euros 3.8 billion – a record fine for a cartel in the EU’s 60 year history.’ 48 On the popularity of these measures of effectiveness, See William E Kovacic, ‘Rating the Competition Agencies: What Constitutes Good Performance?’ (2009)16 Geo Mason L R 903. 49 The establishment of the modern leniency mechanism by the US Department of Justice in 1993 is an impressive example of what Philip Weiser has called ‘entrepreneurial administration’. Philip J Weiser, ‘Entrepreneurial Administration’ 97 (2017) B U L Rev 2011. 50 We share the concern of other commentators who have questioned how successful enforcement has been and whether deterrence is currently adequate. See eg Louis Kaplow, ‘An Economic Approach to Price Fixing’ (2011) 77 Antitrust L J 343, 424–25: Although subject to reservations and some ambiguity, it appears that both law and practice base the fine on something closer to a fixed fraction of firms’ revenues in the affected markets, with little or no adjustment to reflect the actual price elevation. … Accordingly, setting to the side other penalties (which may be substantial in the United States because of private suits, but less so elsewhere), deterrence is likely to be highly inadequate when large overcharges occur since fines will be less than firms’ profits, even ignoring any probability discount.‘ 51 In this respect, we are intrigued by the European Commission’s recent prosecutions of manufacturers of heavy trucks for colluding to coordinate pricing and to time the introduction of emissions abatement technology. European Commission, Press Release, Antitrust: Commission fines Scania for participating in trucks cartel (27 September 2017) <europa.eu/rapid/press-release_IP-17-3502_en.htm> accessed 27 September 2018. Through September 2017, the Commission had reached settlements totaling nearly 3 billion euros with MAN, DAF, Daimler, Ivaco, and Volvo/Renault and had imposed a fine of 880 million euros on Scania. The alleged cartel participants are substantial enterprises. By any measure, the firms involved are well versed in the rules of competition law by reason of having substantial in-house legal departments and access to law firms with the best competition law practices. The cartel is alleged to have begun in 1997 and run through at least early January 2011, a period in which the European Commission’s Directorate for Competition was visibly ramping up its cartel enforcement program and bringing high profile cases. In the trucks matter the Commission carried out a dawn raid in January 2011 after receiving a leniency application from MAN. Why did these large companies execute an illicit scheme in the face of easily recognized and progressively more aggressive efforts by the European Commission to prosecute cartels? Two explanations come to mind. First, the theory of the case may have involved a Commission effort to expand the zone of behaviour deemed to be improper collusion—perhaps, a bolder effort to treat information exchanges as proof of improper coordination. If the prosecution indeed represented an outward extension of the existing law enforcement perimeter, one can understand that the firms may have thought that their behaviour, though risky from an antitrust point of view, would not be treated as a hard core infringement. If the Commission was using a novel theory of illegality, why would not all six firms resist instead of settling? Five firms may have thought the settlement amount was better than the risk of a still higher fine (the EU maximum is 10 per cent of global turnover). Scania alone is contesting the charges. A second explanation is that the firms are accused of forms of conduct that the Commission previously and clearly had treated as illegal collusion. If this is the case, one should ask again why the companies would persist in the behaviour through the 2000s in the face of increasingly stringent EC anti-cartel enforcement? In this scenario, there is the sobering possibility that the massive settlement amounts are not a sign of policy success but instead suggest a fundamental failure of deterrence. 52 George J Stigler, ‘A Theory of Oligopoly’ (1964) 72 J Pol Econ 44. For a vivid illustration of the difficulties that a cartel can face in setting output limits and allocating market shares among the participants, see United States v Andreas, 216 F.3d. 645 (7th Cir 2000). 53 See Michael E Porter, Competitive Strategy: Techniques For Analyzing Industries And Competitors (1980) 4 . 54 See Robert C Marshall and Leslie Marx, The Economics Of Collusion: Cartels and Bidding Rings (Free Press (A Division of Simon and Schuster Inc.), New York 2012); William E Kovacic and others, ‘Bidding Rings And The Design Of Anti-Collusion Measures For Auctions And Procurements’ in Nicolas Dimitri and others (eds), Handbook Of Procurement 149 (Cambridge University Press 2006); Randal Heeb and others, ‘Cartels as Two-Stage Mechanisms: Implications for the Analysis of Dominant-Firm Conduct’ (2009) 10 Chicago J Int L 213. 55 The developments covered here are examined in more detail in Alison Jones and William E Kovacic, ‘Identifying Anticompetitive Agreements in the United States and the European Union: Developing a Coherent Analytical Framework’ (2017) 62 Antitrust Bull 254; Alison Jones, ‘Analysis of Agreements under U.S. And E.U. Antitrust Law – Convergence Or Divergence?’ (2006) 57 Antitrust Bul 691. 56 United States v Trans-Missouri Freight Ass’n, 166 U.S. 290 (1897). 57 United States v Socony-Vacuum Oil Co., 310 U.S. 150 (1940). 58 Timken Roller Bearing Co. v United States, 341 U.S. 593 (1951). 59 National Macaroni Manufacturers Ass’n v. Federal Trade Commission, 345 F.2d 421 (7th Cir. 1965). 60 Catalano , Inc. v Target Sales, Inc., 446 U.S. 643 (1980). 61 US Department of Justice, Press Release, Justice Department and Federal Trade Commission Release Guidance for Human Resource Professionals on How Antitrust Law Applies to Employee Hiring and Compensation (20 October 2016). In the announcement, the Justice Department stated its intention to prosecute infringements of this policy as crimes. 62 See Jones and Kovacic (n 55). 63 See, William E Kovacic and others, ‘Plus Factors and Agreement in Antitrust Law’ (2011) 110 Mich. L. Rev. Mich L Rev 393. 64 Antitrust law assumes there are not likely to be offsetting social gains that could excuse price-fixing. There are exceptional cases—collective rights organizations like ASCAP and BMI, or restrictive agreements in patent licenses. See, BMI v CBS, 441 U.S. 1 (1979); Kaplow (n 50) 357; Department of Justice and Federal Trade Commission, Antitrust Guidelines for the Licensing of Intellectual Property, 30–32 (2017) <https://www.ftc.gov/system/files/documents/public_statements/1049793/ip_guidelines_2017.pdf> accessed 27 September 2018. 65 In theory monopoly prices could be achieved by tacit coordination and such parallel behaviour is outside the scope of the price fixing prohibition. The pragmatic judgment made by antitrust law is that the expected costs from chilling desirable behaviour exceed the expected gains from deterring tacit collusion. 66 See Marshall and Marx (n 54) ix. Recent research illustrates how communication can be used to support output restriction and price increases. See Gaurab Aryal, Federico Ciliberto and Benjamin T Leyden, Public Communication and Collusion in the Airline Industry (12 February 2018) <https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3123972> accessed 27 September 2018. The authors study communications used in the airline industry to communicate plans to reduce capacity on competitive routes. Specifically, they find that earnings calls from top managers to the investing public contained comments about capacity that served as a communications channel between competing airlines. The capacity comments apparently convey information about demand conditions that is hard for other airlines to observe directly. Their econometric analysis indicates, ‘that when all legacy carriers operating in an airport-pair market communicated about capacity discipline in a given quarter, the average number of seats offered in those markets decreased by 1.45% in the next quarter.’ ibid 4. 67 For a discussion of this distinction See Edward J Green, Robert C Marshall and Leslie M Marx, ‘Tacit Collusion in Oligopoly’ Oxford Handbooks Online(2014). 68 Cartel stability is promoted by frequent exchange of price, quantity and other information. When information is not exchanged or exchanged infrequently, then cartel members are less certain of whether a decline in the market price was caused by defection or unfavourable shift in demand. If cartel members mistakenly fear others are cheating this may cause cooperation to unravel. 69 A cartel does not always need to seek a price increase. In industries where prices naturally fall under competitive conditions, a cartel may instead attempt to slow the rate of decline. If the cartel is effective, the price descent will be less severe than in the absence of explicit collusion as was the case for DRAM products during that conspiracy. See European Commission Prohibition Decision 19 May 2010, Comp 38511/DRAM <http://ec.europa.eu/competition/antitrust/cases/dec_docs/38511/38511_1813_8.pdf> accessed 27 September 2018. 70 See Marshall and Marx (n 54) Ch 6. There are product/industry/markets where some structures are in place prior to cartel formation so the cartel need not be concerned with the implementation of those. For example, the requirements of the Federal Acquisition Regulations that all bids be revealed at the end of a competitive bidding process eliminates the need for a cartel to invest in ex post monitoring of bids for compliance with the cartel agreement. 71 See Kovacic and others (n 63) 393–436. The seven super plus factors identified by the authors are as follows. A subset of firms restricts production when prices and profits are relatively high or increasing. Among a subset of producers, market shares, customer incumbency, or geographic dominance is stable when the firms have excess capacity, and prices and profits are relatively high or increasing. A reliable predictive econometric model that accounts for all material noncollusive effects on price, estimated using benchmark data where conduct was presumed noncollusive, produces predictions of prices that do not explain the path of actual prices in the period or region of potential collusion, at a specified high confidence level. A firm or subset of firms has knowledge of the details of another firm’s transactions, production, sales, and/or inventories where the latter firm would be competitively disadvantaged by conveying that information unilaterally. Firms engage in interfirm transactions that are transfers of resources and are largely void of productive noncollusive motivations. In an industry where the product made by different firms is largely homogeneous, there is a discrete change in the intrafirm incentives of sales forces, across a subset of firms during a given period, that shifts from the pursuit of market share to maintenance of elevated prices (such as a shift to ‘price before volume’). A subset of firms with an aggregate market share large enough to have dominant-firm market power jointly engage in a dominant-firm conduct when no single firm has the market power to act unilaterally as a dominant firm by engaging in that dominant-firm conduct. 72 As noted in the introduction to this section, the difficulties associated with successful cartel formation and operations can be significant. 73 A cartel member cut its price in Argentina in response to a threat of entry by an Argentine firm. See Stocking and Watkins (n 1) 437. 74 ibid. 75 It is easy to understand why the risk of detection and conviction increases with the volume of cartel communication. More communication likely means more people become aware of the cartel and more incriminating documents or recordings are produced and stored. This increases the risk that an innocent insider will discover and expose the cartel and the risk that a cartel investigation started for some other reason will uncover damming evidence of the agreement. 76 See William M Landes, ‘Optimal Sanctions for Antitrust Violations’ (1983) 50 U Chi L Rev 652; Connor (n 8) x: In the case of collusive group crimes like price-fixing, deterrence means that companies or individuals, after weighing the probable gains versus expected losses associated with overt collusion, decide that it would be less profitable to form a cartel (or join an existing cartel) than to adopt a form of business conduct that does not involve illegal manipulation of markets. 77 Michael K Block and Joseph Gregory Sidak, ‘The Cost of Antitrust Deterrence: Why Not Hang a Price Fixer Now and Then?’ (1980) 68 Geo L J 131, 138–39. 78 ibid. 79 Chilling effects also shape the types of proof relied on by antitrust courts. See, Kaplow (n 50) 366 (‘Whether and what sorts of chilling effects may arise depend on what forms of proof are employed. Suppose, for example, that sudden, substantial price increases are considered as possible evidence of coordinated oligopolistic price elevation.’) 80 Chilling effects are easy to describe but difficult to quantify. See ibid 366 (‘The chilling of desirable behavior is a concern that strongly motivates past discussions of price-fixing rules but almost always remains implicit, which makes it difficult to assess its importance and how the magnitude of the problem is influenced by the nature of the rules that are adopted.’) 81 See eg William H Page, ‘Optimal Antitrust Remedies: A Synthesis’ in Roger D. Blair and D. Daniel Sokol (eds), The Oxford Handbook Of International Antitrust Economics, vol 1 Roger D. Blair and D. Daniel Sokol (Oxford University Press, New York 2014) (‘the formula used to calculate fines is only weakly related to the optimal penalty.’) Kaplow (n 50) 370 (‘Limited evidence to date suggests that the probability of detection is quite low.’) 82 ibid (‘… even with substantial modern enforcement and penalties much stiffer than had existed in the past, the existing level of deterrence may be insufficient. Indeed, even detected and punished cartels may suffer little or realize net gains.’); See note 81, x (compliance ‘programs may be less effective in controlling price fixing than exclusionary practices, because price fixing is necessarily clandestine … A manager may get the idea that he or she can surreptitiously fix prices and enhance the profitability of his or her division (and compensation) without drawing the notice of supervisors and antitrust counsel.’) 83 Much of our thinking runs parallel to the analysis by Bernheim and Whinston of multi-market contact in the context of oligopoly pricing. See generally B Douglas Bernheim and Michael D Whinston, ‘Multimarket Contact and Collusive Behavior’ (1990) 21 Rand J Econ1. 84 Cartel instability likely generates communications and defections that increase the probability of detection. 85 For example, buyer resistance and complaints may be softer when buyers face rising prices in both products 1 and 2. They may be more likely to attribute the general price rise to exogenous factors and less likely to attribute it to collusion. On the other hand, multi-product price fixing may make detection more likely because more parties are privy to cartel-related communications. 86 See eg Kiran Desai, ‘Antitrust Compliance: The Effects of Perceived Regulatory Failure’ [2010] European Anti Rev 15, 17 (‘In creating an antitrust programme it should be recognised that there may be personnel who act in bad faith, for which no amount of education and admonition will act as a deterrent. Sales targets and bonuses can be too much of an incentive to break the law. Indeed, some may even go to great lengths to hide their activities from in-house counsel.’); Theodore Banks, The Trouble With Antitrust Compliance and 10 Ways to Fix It, Corporate Counsel 30 April 2012 (expressing concern the price-fixing will arise through the actions of a rogue manager). 87 The threat of shareholder derivative suits could be used to deter directors and top managers from embracing price fixing agreements. Derivative suits might be effective tools of corporate governance in this context. See John M Connor, ‘The Global Lysine Price-Fixing Conspiracy of 1992-1995’ (1997) 19 Rev Agri Econ 412, 420 (‘ADM settled shareholders’ suits for $30 million’); James D Cox, ‘The Social Meaning of Shareholder Suits’ (1999) 65 Brook. L. Rev. Brook L Rev 3, 44: The need for such an inhibition was the basis raised in the petition for certiorari in California Public Employees' Retirement System v. Felzen. The derivative suit grew out of more than $190 million Archer Daniels Midland (ADM) had paid to settle various civil and criminal antitrust cases for its alleged price fixing activities. The derivative suits alleged ADM's officers and directors had breached their fiduciary obligations in connection with the antitrust violations and sought to recover $190 million. Unfortunately, empirical studies find that cartel member firms have less vigilant directors than other firms and cartel employees exercise stock options more rapidly than employees in other firms. See Paolo Buccirossi and Giancarlo Spagnolo, Corporate Governance and Collusive Behaviour (2007) <https://ssrn.com/abstract=1136675> accessed 27 September 2018. Tolerance of collusion may affect the nature of executive compensation. Spagnolo shows that capped bonus plans and incumbency rents for managers encourage collusion by relaxing the incentive compatibility constraint. Giancarlo Spagnolo, ‘Managerial Incentives and Collusive Behavior’ (2005) 49 European Econ Rev 1501–23. 88 ‘Senior management in EU cartels does not seem to suffer from their participation in cartels. For example, Robert Koehler became CEO of SGL Carbon in 2012, after being convicted in 1999 of price-fixing in the graphite electrodes cartel.’ Marvão (n 31). 89 Harrington reviewed recent price-fixing investigations at the EU and found that in most cases top management was involved in the collusion. See Joseph E Harrington Jr., A Proposal for a Structural Remedy for Illegal Collusion 2 (31 July 2017) <https://ssrn.com/abstract=3038174> accessed 27 September 2018. Connor notes: ‘The evidence on convicted cartelists suggests that rogue managers are in a minority. Cartel executives tend to have titles that place them in a company’s top layer of management.’ Connor (n 8) fn 53. Harry First reviewed DOJ Antitrust Division press releases issued between March 2014 and March 2016 and found 47 named individuals involved in ‘major international cartels.’ The industry distribution was: ‘twenty-seven in auto parts, seven in Libor, seven in ocean shipping, five in cathode display tubes, one in marine hose … .’ ‘[N]ot one person mentioned during this time period was in a low-level corporate position. Their described positions varied—executive, general manager, group and department manager, high-level manager, director of sales and marketing—and thirteen of them were identified either as president, CEO, owner, or chairman. These are not line-level employees.’ Harry First, Are Cartel Participants Rogues? June 2016 <http://www.nortonrosefulbright.com/knowledge/publications/140888/are-cartel-participants-rogues> accessed 27 September 2018. 90 Our case study of price fixing in the chemical industry highlights three facts about price-fixing in the chemical industry that are hard to reconcile with the assumption that collusion is accomplished strictly (or mostly) by middle managers: (1) repeated price-fixing, (2) coordination of price-fixing across different product lines, and (3) the use of outside consultants for addressing the double marginalization problem by coordinating cartel conduct between upstream and downstream cartels and organizing other kinds of coordination. Levenstein, Marvão, and Suslow come to a similar conclusion: ‘There are cases where a firm’s corporate culture encourages participation in cartels. In such a case, the leadership of the organization expects managers to collude, and we observe collusion in many markets in which the firm operates. Managers may learn to collude in one division and then take those practices to another.’ Margaret C Levenstein, Catarina Marvão and Valerie Y Suslow, Serial Collusion In Context: Repeat Offenses By Firm Or By Industry?, 18 December 2015 <http://www.oecd.org/officialdocuments/publicdisplaydocumentpdf/?cote=DAF/COMP/GF(2015)12&docLanguage=En> accessed 27 September 2018. 91 If deterrence is sufficient to make price-fixing unattractive to shareholders then good corporate governance would be the key to inducing top management to take the steps necessary to stop price fixing. There is evidence that good governance, in the form of a significant presence of outside directors on the Board leads to better cooperation with cartel enforcers once collusion is found. See Murillo Campello, Daniel Ferrés and Gaizka Ormazabal, ‘Whistle-Blowers on the Board? The Role of Independent Directors in Cartel Prosecutions’ (2017) 60 J Law Econ241, 244. 92 ‘In such a case, the leadership of the organization expects managers to collude, and collusion occurs in many markets in which the firm operates. Firm norms and expectations of managerial behavior can repeatedly encourage collusion and ‘disregard’ previous fines, as illustrated in the ADM case (Eichenwald, 2000).’ Marvão (n 31). 93 ‘Imposing tougher sanctions, such as individual prison sentences or disqualification of senior executives from employment in their sector or role, may prevent repeated collusive behaviors (in new firms) and thus, increase deterrence levels.’ ibid. 94 Bernheim and Whinston explain how multimarket contact facilitates price fixing by strengthening the threat of punishment faced by potential defectors—if a firms defects in one market then it will suffer in the future when it gets punished in the market in which it defected but also in other markets in which it is participating in a cartel. See Bernheim and Whinston (n 83) 6–11. Econometric research largely confirms the role of multimarket contact in facilitating higher prices, but the research has looked at firms competing across a number of geographic markets rather across different product lines. Bernheim and Whinston’s model suggests the effects are likely to be stronger across product lines assuming that various product lines are more heterogeneous than various geographic markets. Hay and Kelley examine all US price fixing cases from 1963 to 1972 and find that collusion in one geographic market is associated with collusion in other geographic markets. George A Hay and Daniel Kelley, ‘An Empirical Survey of Price Fixing Conspiracies’ (1974) 17 J L Econ 13, 27–28. Work finding that multimarket contact across separate geographic markets leads to higher prices includes: Ivette Jans and David I Rosenbaum, ‘Multimarket Contact and Pricing: Evidence from the US Cement Industry’ (1994) 15 Int’l J. Industrial Org. Int J Indus Org 391 (multimarket contact associated with higher cement prices); William N Evans and Ioannis N Kessides, ‘Living by the ‘Golden Rule’: Multimarket Contact in the U. S. Airline Industry’ (1994) 109 Q. J. Econ. Q J Econ 341 (multimarket contact tends to increases airline ticket prices); Philip M Parker and Lars-Hendrik Roller, ‘Collusive Conduct in Duopolies: Multimarket Contact and Cross-Ownership in the Mobile Telephone Industry’ (1997) 46 RAND. J. Econ. RAND J Econ 301 (multimarket contact associated with higher prices in the mobile phone industry); Nerea Fernandez and Pedro L Marin, ‘Market Power and Multimarket Contact: Some Evidence from the Spanish Hotel Industry’ (1998) 28 JIndus Econ 304 (multimarket contact associated with higher prices in the Spanish hotel market); Meghan R Busse, ‘Multimarket Contact and Price Coordination in the Cellular Telephone Industry’ (2000) 9 J Econ Manag Strat 287 (multimarket contact associated with a pattern of setting higher identical prices across markets); Federico Ciliberto and Jonathan W Williams, ‘Does multimarket contact facilitate tacit collusion? Inference on conduct parameters in the airline industry’ (2014) 45 RAND J. Econ. RAND J Econ 764 (carriers with a significant amount of multimarket contact can sustain ‘near-perfect’ cooperation in setting fares); Haizhen Lin and Ian M McCarthy, ‘Multimarket Contact in Health Insurance: Evidence from Medicare Advantage’ NBER Working Papers 24486 (2018) (multimarket contact responsible for higher health insurance premia). But Waldfogel and Wulf find only a small effect on radio advertising revenue. See, Joel Waldfogel and Julie Wulf, ‘Measuring the Effect of Multimarket Contact on Competition: Evidence from Mergers Following Radio Broadcast Ownership Deregulation’ (2006) 5 BE J Econ Anal Pol, art 17. 95 ‘BASF ceased its own production of vitamin B1 in 1989 but continued to act as a major supplier selling product bought in from other producers.’ (EC Decision in Vitamins, para 84) BASF continued to bid for Thiamin business. When it won a contract the order would be fulfilled by Roche. 96 Expert Report of B Douglas Bernheim, 93, M.D.L. No 1285, In Re: Vitamins Antitrust Litigation, Misc. No 99-0197 (TFH), 24 May 2002. This report was submitted as exhibit number and filed in the District Court of the District of Columbia. We obtained the document through a request to the law clerk to Chief Judge Thomas F Hogan. The document was made available based on D.C. Local Civil Rule 79.2 and the US District Court for the District of Columbia's policy of not retaining exhibits that are admitted into evidence at trial in civil cases. 97 See Robert C Marshall, Leslie M Marx and Matthew E Raiff, ‘Cartel Price Announcements: The Vitamins Industry’ (2008) 26 Int’l J. Industrial Org. Int J Indus Org 762 (discussing circumstantial economic evidence that the vitamins conspiracy started in 1985). 98 A similar finding can be found in Zhou (n 15) 382 (EC investigation of one market decreases entry and increases exit from other cartelized markets). 99 We do not observe the price at which the transactions occurred. 100 EC Decision in Butadiene Rubber and Emulsion Styrene Butadiene Rubber, art 1 in Decision, 62. 101 EC Decision in Butadiene Rubber and Emulsion Styrene Butadiene, para 286. 102 EC Decision in Methacrylates, art 1 in Decision, 74. 103 ibid para 30. 104 EC Decision in MCAA, art 1 in Decision, 63. 105 ibid para 16. 106 EC Decision in Polypropylene, art 1 in Decision, 36. 107 ibid para 19. 108 We counted more than 15 other joint ventures involving firms that were members of the electronics cartel but the joint ventures did not consolidate production or sales between the cartel members. These joint ventures can also be used to reallocate cartels profits assuming the joint ventures generated a profit, the parties can always agree to any profit share or cost share they desire. 109 The Europe Commission does not have a good track record on this front. See, Leslie M Marx and Jun Zhou, The Dynamics of Mergers Among (Ex)Co-Conspirators in the Shadow of Cartel Enforcement (March 2014), TILEC Discussion Paper No 2014-013 <https://ssrn.com/abstract=2416893> accessed 27 September 2018 (‘[I]n over one-third of the cartels discovered by the European Commission … between 1985 and 2012, conspirators sold majority shares or relocated major production and sales capacities to their co-conspirators.’) 110 If exit occurs through a merger or takeover then it is reviewable. But a multi-product firm can often convert a production process for one product into that for another. Or, alternatively, sell off piecemeal the capital stock. 111 And even if such an inquiry were made it would likely remain unanswered on the grounds of being beyond the scope of the current cartel inquiry. 112 See US v Line Material 333 U.S. 287, (1948); US v Harford Empire, 323 U.S. 386 (1945). 113 See Marshall And Marx (n 54) 16–17. 114 See generally, Richard Selten, ‘The Chain Store Paradox’ (1978) 9 Theor Decis 127. 115 Theoretical work shows that well designed leniency programs facilitate cartel detection in the single-product setting. For a survey, See Giancarlo Spagnolo, Leniency and Whistleblowers in Antitrust, CEPR Discussion Paper No 5794 (August 2006) <https://ssrn.com/abstract=936400> accessed 27 September 2018. Poorly designed leniency programs that are too generous to cartel members undercut deterrence. There is reason to be concerned that the EU may be too generous with its grants of leniency. ‘[A] firm received a 100 percent reduction in the fine through the leniency program in 55 (54 percent) of the 101 products in which firms were prosecuted’ in the EU between 2001 and 2012. Leslie M Marx, Claudio Mezzetti and Robert C Marshall, ‘Antitrust Leniency with Multiproduct Colluders’ (2015) 7 Amer Econ J: Microecon 205. 116 See Shigeki Isogai and Chaohai Shen, Multiple Cartels and Strategic Use of the Leniency Policy, Working Paper, 2017 for a model where a multi-product firm uses amnesty to signal intolerance for deviant conduct to members of its many other cartels. ‘Marvão (2015) shows that repeat offenders appear to receive larger EC leniency reductions, which suggests that firms can learn the ‘rules of the game’, colluding repeatedly and reporting the cartel to reduce their penalties. As such, fines need to be tougher and recidivism needs to be dealt with differently.’ Marvão (n 31). See also, Robert C Marshall and Leslie M Marx, Participation in Multiple Cartels through Time and the Potentially Strategic Use of Leniency, Working Paper, October 2015. 117 Akzo received full amnesty in Organic Peroxides (EC Decision in Organic Peroxides, paras 505–06). 118 Degussa received full amnesty in Hydrogen Peroxide and Perborate (EC Decision in Hydrogen Peroxide and Perborate, para 501). 119 Akzo received full amnesty in Calcium Carbide (EC Decision in Calcium Carbide, paras 335–36). 120 See Levenstein and Suslow (n 30) 8: There are cases where a firm’s corporate culture encourages participation in cartels. In such a case, the leadership of the organization expects managers to collude, and we observe collusion in many markets in which the firm operates. Managers may learn to collude in one division and then take those practices to another. Firm norms and expectations of managerial behavior can encourage collusion, as the well-known ADM case illustrated (Eichenwald 2000). Van Driel (2000) analyzes the role that cultural and organizational factors played in a study of four European transportation cartels. See also Hugo Van Driel, ‘Collusion In Transport: Group Effects In A Historical Perspective’ (2000) 41 J Econ Behav Org385; Kurt Eichenwald, The Informant: A True Story (Broadway Books (division of Random House), New York 2000). 121 See, Rick Busscher, Martin Herz and Hans Vedder, ‘The Shortest Competition Judgment Ever; AC Treuhand II’, European Law Blog, 9 November 2015 <http://europeanlawblog.eu/2015/11/09/the-shortest-competition-judgment-ever-ac-treuhand-ii/> accessed 27 September 2018 … as with any cartel, there needs to be supervision and enforcement. This is tricky for cartelists themselves, as the supervisor is exposed to the risk of being qualified as a ringleader, negating the possibility of immunity under the Leniency Guidelines. Hence, the heat stabilisers cartel thought of a better way, namely hiring an independent firm to ensure that parties stick to the cartel. 122 We expect that the information sharing fostered by Fides/AC Treuhand helped chemical firms coordinate on the initiation of cooperative pricing and to monitor for deviations. Shared information reduces the possibility of the misinterpretation of market-wide data. Recent papers by Awaya and Krishna and Spector show that information sharing among firms improves their ability to support elevated prices. Awaya and Krishna emphasize that information sharing improves monitoring, which reduces uncertainty and allows greater profits from coordination. See Yu Awaya and Vijay Krishna, ‘On Communication and Collusion’ (2016) 106 Amer. Econ. Rev.Amer Econ Rev 285. Spector emphasizes that information sharing allows firms to detect deviations more quickly, making deviations less profitable and so reducing the incentive for firms to deviate in the first place. See David Spector, Facilitating Collusion By Exchanging Non-Verifiable Sales Reports. PSE Working Papers No 2015-07 (2015) <https://halshs.archives-ouvertes.fr/halshs-01119959/document> accessed 27 September 2018. 123 Economists have identified an inefficiency called double marginalization that arises from successive monopolies. A cartel can increase overall profit by responding to this inefficiency. 124 We refer to the firm as Fides/AC Treuhand because, ‘AC Treuhand is the result of a 1993 management-buyout of the division offering association-management within a company called Fides Trust AG.’ (EC Organic Peroxide decision, para 20). 125 Role in explicit collusion (from EC OP Decision, para 92): organized meetings of the members of the agreement, often in Zurich; produced, distributed, and recollected the so called ‘pink’ and ‘red’ papers with the agreed market shares which were, because of their colour, easily distinguishable from other meeting documents and were not allowed to be taken outside the AC Treuhand premises; calculated the ‘pluses and minuses’, ie the deviations from the agreed market shares, which were used for compensations; reimbursed the travel expenses of the participants, in order to avoid traces of these meetings in the companies’ accounts; collected data on OP sales and provided the participants with the relevant statistics; stored the original agreement from 1971 and other relevant documents concerning the agreement in its safe and handed them over to PC; acted as a moderator in case of tensions between the members of the agreement and encouraged the parties to find compromises. AC Treuhand would try to stimulate the parties to work together and reach an agreement. ‘The message from AC Treuhand was that it would get worse for the participants if they discontinued the discussions.’ was actively involved in reshaping the arrangement among producers in 1998 during a bilateral meeting in Amersfoort between Akzo representatives and […] of AC Treuhand. During this bilateral meeting a solution aimed at meeting Atochem’s demand was developed. The solution consisted of a proposal of AC Treuhand for the new quotas; AC Treuhand advised the parties whether or not to allow other participants into the agreement; instructed all participants on the legal dangers of parts of these meetings and on what measures to take to avoid detection of these arrangements bearing on Europe; participated mainly the ‘summit’ meetings but at least at once instance in the nineties attended also a working group meeting; according to Akzo chaired at least some of the meetings, (AC Treuhand sees itself his in its reply to the SO not as chairman but as moderator); was aware of the Spanish sub arrangement and was asked to calculate the deviation between agreed quotas and effective sales in Spain; organized the auditing of the data submitted by the parties; calculated the new quotas after the acquisition and integration of competitors in the agreement. 126 The EC decision in Marine Hoses as well as Yen Interest Rate Derivative make explicit note of a cartel facilitator. For Marine Hose the cartel consultant was Peter Whittle of PW Consulting. See Department of Justice, Antitrust Division (2017) <https://www.justice.gov/atr/case-document/plea-agreement-peter-whittle> accessed 27 September 2018. For Yen Interest Rate Derivatives the cartel consultant was the brokerage firm ICAP. 127 See John M Kuhlman, ‘The Nature and the Significance of Price Fixing Rings’ (1969) 2 Anti L Econ Rev 69 (examining 50 price fixing complaints and indictments in the 1960s and finding that 23 trade associations were named (76)); George A Hay and Daniel Kelley, ‘An Empirical Survey of Price Fixing Conspiracies’ (1974) 17 J L and Econ 13, 28 (‘The current study demonstrated that trade associations are often used as a coordinating device, especially when more than a given number of firms are involved.’) 128 See <https://www.2mpact.be/weblog-detail-pagina/2013/11/20/From-AMC-tot-AMS-and-back-an-eye-opener-for-association-management-professionals?originNode=277> accessed 27 September 2018, for a description of AMCs as well as a list of AMCs throughout Europe. 129 See Fig. 1B. 130 ‘After securing criminal convictions, the Division should also inquire, and publicly report details on, how cartels were able to collude and sustain their collusion.’ American Antitrust Institute, 26 <https://www.antitrustinstitute.org/wp-content/uploads/2018/08/Cartels-Chapter-from-AAI-Transition-Report_100520082109.pdf> accessed 27 September 2018. 131 We have identified eight firms in the chemical or electrical cartel that applied for leniency for one (or more) cartel while still participating in other cartels: Akzo Nobel applied for leniency in the sodium gluconate cartel while still active in seven other cartels; Arkema/Atofina applied for leniency in the monochloroacetic acid and organic peroxides cartels while still active in three other cartels; Bayer applied for leniency in the citric acid cartel while still active in three other cartels; Bayer applied for leniency in the rubber chemicals cartel while still active in two other cartels; Degussa applied for leniency in the methionine cartel while still active in three other cartels; Degussa applied for leniency in the organic peroxides cartel while still active in one other cartel; Rhone Poulenc/Aventis applied for leniency in the methionine and vitamins cartels while still active in another cartel; Solvay applied for leniency in the vitamins cartel while still active in one other cartel; Hitachi applied for leniency in the power cables cartel while still active in two other cartels; Hitachi applied for leniency in the alternators and starters cartel while still active in one other cartel; and Samsung applied for leniency in the monitor tubes and smart card chips cartels while still active in another cartel. 132 Compare Anne Riley and D Daniel Sokol, “Rethinking Compliance,” (2015) 3 J Anti Enforce 31 (discussing and supporting conditions in which a firm's maintenance of a robust compliance program should dictate a mitigation of punishment). On the design and operation of antitrust compliance programs generally see Roger D Blair and D Daniel Sokol, The Oxford Handbook of International Antitrust Economics (Oxford University Press 2014), 265. 133 American Antitrust Institute (2008), 26: EU, U.K., Korean, and Canadian enforcers release far more details about the conduct and harm caused by cartels than does the Division. The information released by the Division rarely, if ever, includes data about unindicted coconspirators, affected sales, conduct, and injuries caused by cartels. The Division should reveal more of what it knows about these matters, either in plea agreements, informations, sentencing agreements, or in follow-up studies using anonymous data. It should publish all sentencing agreements, whether submitted to courts or not, on its Web page. This could be done in a manner that would not interfere with the Division’s law enforcement efforts. <https://www.antitrustinstitute.org/wp-content/uploads/2018/08/Cartels-Chapter-from-AAI-Transition-Report_100520082109.pdf> accessed 27 September 2018. 134 American Antitrust Institute (2008) 25: ‘The Division could require in sentencing agreements that defendants turn over simple post-conviction reports for five years on their production costs, sales, and prices in the affected market. For a representative sample of successful cartel prosecutions, the Division should report on the state of competition in the affected industries.’ <https://www.antitrustinstitute.org/wp-content/uploads/2018/08/Cartels-Chapter-from-AAI-Transition-Report_100520082109.pdf> accessed 27 September 2018. 135 See Marx and Zhou (n 109) 1 (Introduction of leniency in the EU in 1996 increased the pace of horizontal mergers involving cartelists. But the cartel settlement procedure slowed the pace.) ‘Competition authorities have repeatedly permitted mergers among former cartel members, often without review, let alone structural intervention. Davies and others (2014) examine mergers among former cartel conspirators and conclude that only 29% of the mergers were investigated by the EC.’ Marvão (n 31). 136 Both the USA and the EU examine coordinated effects when deciding whether to approve horizontal merges. The FTC advises that a horizontal merger may ‘lessen competition … by creating or enhancing the ability of the remaining firms to act in a coordinated way on some competitive dimension.’ <https://www.ftc.gov/tips-advice/competition-guidance/guide-antitrust-laws/mergers/competitive-effects> accessed 27 September 2018. And the EC Horizontal Merger Guidelines—s 22 state: … horizontal mergers may significantly impede effective competition … by changing the nature of competition in such a way that firms that previously were not coordinating their behaviour, are now significantly more likely to coordinate and raise prices or otherwise harm effective competition. A merger may also make coordination easier, more stable or more effective for firms which were coordinating prior to the merger (coordinated effects). 137 FTC Competition Guidance <https://www.ftc.gov/tips-advice/competition-guidance/guide-antitrust-laws/mergers/competitive-effects> accessed 27 September 2018. 138 The Horizontal Merger Guidelines of 2010 recognize the significance of past collusion for a proper coordinated effects analysis: The Agencies presume that market conditions are conducive to coordinated interaction if firms representing a substantial share in the relevant market appear to have previously engaged in express collusion affecting the relevant market, unless competitive conditions in the market have since changed significantly. Previous express collusion in another geographic market will have the same weight if the salient characteristics of that other market at the time of the collusion are comparable to those in the relevant market. Failed previous attempts at collusion in the relevant market suggest that successful collusion was difficult pre-merger but not so difficult as to deter attempts, and a merger may tend to make success more likely. Previous collusion or attempted collusion in another product market may also be given substantial weight if the salient characteristics of that other market at the time of the collusion are closely comparable to those in the relevant market. Horizontal Merger Guidelines, US Department of Justice and Federal Trade Commission, 25, 19 August 2010 <https://www.ftc.gov/sites/default/files/attachments/merger-review/100819hmg.pdf> accessed 27 September 2018. 139 Marx and Zhou (n 109) 24. 140 American Antitrust Institute (2008) 25–26: Price fixing and mergers are handled by separate units in the Division, yet the two may be related. A single horizontal merger in the United States or abroad can make formation of a cartel feasible. It is frequently the case that cartel convictions are followed by spin-offs and other industry restructuring. A history of collusion in an industry may signal that a rise in coordinated effects is likely after a proposed merger is consummated. The Division should study whether there is a pattern of cartel members’ acquiring rivals, large customers, or suppliers in the affected industry anywhere in the world before, during, or immediately after, the violation. Any negative findings should be incorporated into the Division’s enforcement decisions. <https://www.antitrustinstitute.org/wp-content/uploads/2018/08/Cartels-Chapter-from-AAI-Transition-Report_100520082109.pdf> accessed 27 September 2018. 141 Marvão (n 31). 142 Note that we intend to limit these programs to disclosures that expose at least one serial colluder as a cartel member. Thus, the effectiveness of the program depends on transparent enforcement of anti-cartel laws that clearly identify serial colluders. We leave open the question of exactly how the program would define serial colluders, and leave open the question of whether this sort of bounty program should be available more broadly. Counseling against broad availability of such a program is the risk of chilling desirable competitive behaviour. 143 See Kovacic and others (n 63) 425–26. 144 Note 126 states that cartel facilitators were present in the Marine Hoses as well as Yen Interest Rate Derivative cases. 145 To get a rough sense of the inattention shown by antitrust enforcers and antitrust law scholars to problems created by multi-market contact we compared citation patterns to two prominent articles from the industrial organization literature: the 1990 Multi-market Contact and Collusive Behavior paper by Bernheim and Whinston to Steven C Salop and David T Scheffman, ‘Raising Rivals’ Costs’ (1983) 73 Amer. Econ. Rev. Amer Econ Rev 267. Even though the Bernheim and Whinston article was published 7 years after the Salop and Scheffman article they have a roughly comparable number of citations in Google Scholar—1536 for Salop and Scheffman compared to 1496 for Bernheim and Whinston. We interpret this to mean both articles are highly influential among economics and management scholars. Next we compared the citations to the two articles in the Hein Law Journal Library. Salop and Scheffman got 165 citations compared to only 25 citations for Bernheim and Whinston. Next we looked at Herbert J Hovenkamp and Phillip E Areeda, ‘Antitrust Law: An Analysis of Antitrust Principles and their Applications’ (2012) and found nine cites to Salop and Scheffman but no cites to Bernheim and Whinston—though we did find one cite to Bernheim and Whinston in Louis Kaplow, Competition Policy and Price Fixing (Princeton University Press, Princeton, New Jersey 2013). Finally, we examined documents posted on the websites of the FTC and the Antitrust Division of the DOJ and we found 35 citations to Salop and Scheffman and only five citations to Bernheim and Whinston. We take this citation pattern as evidence that antitrust enforcers and antitrust law scholars cite and discuss significant economic theory, as represented by Raising Rivals’ Costs, when they think the theory may have antitrust applications. Apparently they think of price fixing as a one-shot phenomenon and thus rarely cite Bernheim and Whinston. We think that if they better understood the multi-market contact associated with serial collusion they would show more interest in applications of the multi-market contact theory to anti-cartel enforcement. 146 EC Decision, para 13: ‘The holding company for Flexsys is Flexsys Holding B.V. of which Akzo Nobel Chemicals International B.V. holds 50%, the remaining 50% being held by Solutia Inc and Solutia Europe N.V. together.’ © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
TI - Serial collusion by multi-product firms
JF - Journal of Antitrust Enforcement
DO - 10.1093/jaenfo/jny014
DA - 2018-10-01
UR - https://www.deepdyve.com/lp/oxford-university-press/serial-collusion-by-multi-product-firms-JSOASIH4pe
SP - 296
EP - 354
VL - 6
IS - 3
DP - DeepDyve
ER -