Prosecuting cartel behavior is the most crucial priority for the U.S. Department of Justice’s Antitrust Division (“the Division”) according to Makan Delrahim, Assistant Attorney General for the Division.[1] To deter cartel activity, the U.S. and numerous international competition authorities offer cartel participants (“cartelists”) an opportunity to apply for leniency in exchange for material information that can be used in an investigation against the cartel.[2] Courtesy of these leniency programs, the Division and its global counterparts have worked to dismantle some of the most sophisticated international cartels and recover billions of dollars in criminal fines.[3] Nevertheless, leniency applications across the globe appear to be declining, sparking concerns that authorities are losing their ability to detect cartels.[4] Countries have already taken measures to overcome these obstacles by harnessing developing technologies to bolster alternative detection mechanisms and restructuring their leniency policies.[5] In combination with these efforts, to ease the process for both potential leniency applicants and enforcement regimes, competition authorities should adopt a one-stop-shop model for their leniency programs.[6]
The Cartel Enforcement Process
A cartel forms when two or more businesses agree to collude with each other, commonly by fixing their prices, allocating market shares, and engaging in bid rigging.[7] Collusion impedes competition, harms consumers, and undermines confidence in the free market system on a global scale.[8] In 1978, the Division introduced a leniency program to aid in the prosecution of cartelists.[9] Though largely ineffective at first, the leniency program experienced major successes after undergoing policy reforms.[10] In 1993, the Division began providing immunity from prosecution to the first leniency applicant, and leniency applications soared as a result.[11] Between 1997 and 2003, the Division successfully prosecuted international cartels engaged in USD 10 billion worth of commercial activity.[12] Observing the triumphs of the U.S.’ leniency policies, international competition authorities followed suit.[13] In 2006, the European Commission (EC) released a Leniency Notice, which established a system for cartelists to apply for a marker to protect their status as the first in line to receive immunity if the cartelists were to provide sufficient material information on the cartel by a certain deadline.[14] Between 2006 and 2017, nearly all of the EC’s cartel investigations were generated from leniency applicants.[15]
Evidence of the Decline
Based on data collected by the Organisation for Economic Co-operation and Development (OECD) between 2015 and 2021, the number of leniency applications submitted in the OECD’s 38 Member countries, which includes the U.S., decreased by 58%.[16] The same pattern can be observed globally, including a 60% decline in Europe, 65% in Asia-Pacific, and 66% in Latin America and the Caribbean region.[17] Although the number of applications is not a perfect way to measure the effectiveness of leniency programs —as it does not take into consideration the quality of the applications or their impact on investigations —scholars consider it a useful tool for gauging whether cartel enforcement is in jeopardy.[18] Two possible reasons for this decline are resource misallocation and a lack of coordination in multi-jurisdictional investigations.[19]
Resource Misallocation
In their early stages, leniency programs tend to work in tandem with alternative non-leniency detection tools.[20] However, once the programs prove to be successful, authorities tend to dedicate their limited time, money, and personnel to them at the expense of alternatives and start to overly rely on leniency applications.[21] Cartelists observing this shift realize that their behavior is unlikely to be detected if they do not self-report, so they decide to risk foregoing leniency.[22] On the other hand, in jurisdictions where authorities have a reputation for robust enforcement practices independent of leniency applications, leniency applications tend to rise and cartel deterrence strengthens as a result.[23]
Cartel Screening & Machine Learning
One way to improve enforcement efforts is through behavioral screening methods that use digital datasets to identify evidence of collusion and the markets in which cartels have formed.[24] Behavioral screens have been particularly effective in detecting collusion markers in bid rigging, such as by analyzing the structure and distribution of the winning and losing bid prices.[25] However, screens run a risk of generating false results by failing to account for certain variables and misconstruing the data, as was the case with a screening tool implemented by the Korean Fair Trade Commission in 2006.[26]
To mitigate this issue, authorities can leverage artificial intelligence by using machine learning algorithms to analyze massive data sets, identify patterns of collusion markers within them, and make more accurate predictions.[27] In 2008, the Swiss Competition Commission (COMCO) used machine learning to detect a bid-rigging cartel, compiled sufficient evidence to open an investigation against six firms, and three years later the firms were all fined.[28] In 2019, the Division created the Procurement Collusion Strike Force (PCSF) which utilizes machine learning to detect collusion in government contracts.[29] As of August 2023, the PCSF had 60 open criminal investigations and prosecuted more than 30 companies and individuals, comprising over USD 350 million in government contracts.[30] Furthermore, in 2023, Brazil’s Administrative Council for Economic Defense (CADE) used similar tools to detect bid rigging by firms between 2009 and 2018, resulting in the indictment of 10 companies and 11 individuals.[31]
Multi-Jurisdictional Investigations
As more and more countries have developed their own leniency programs, international cartels have become susceptible to investigations across jurisdictions with divergent leniency policies and enforcement protocols.[32] The proliferation of independent leniency programs means that a cartelist who applies for leniency in one jurisdiction and successfully obtains a marker in only that jurisdiction could still be prosecuted in another jurisdiction if the conduct were to stretch across borders.[33] Likewise, if two countries were to cooperate in an investigation but only one country were to receive a leniency application, the recipient of that application could share information obtained with the other country which could then bring separate charges against the cartelist.[34] While this practice is generally not condoned by enforcement authorities, a situation of this type unfolded involving a leniency application submitted to Ecuadorian authorities, which resulted in a fining decision in Peru in 2021.[35] To avoid a similar fate, cartelists may feel compelled to submit several applications to different competition authorities, which can be very onerous and ineffective if they are not the first in line, so they might be disincentivized from applying for leniency altogether.[36]
ECN+ and the One-Stop-Shop Model
To rectify this coordination challenge, the European Competition Network (ECN) created a policy called ECN+ which allows applicants who have applied for leniency with the EC to also submit summary applications to national competition authorities (NCAs) of the ECN for activities pertaining to the same cartel.[37] If the EC decides not to pursue the case, NCAs under the ECN+ are required to allow the applicant to submit a full application.[38] However, there are notable drawbacks to ECN+.[39] Summary applications must cover the same scope as the full application in terms of product, location, and duration, and many NCAs do not allow applications to be submitted in English.[40]
Going a step further, in 2022, the OECD proposed a one-stop-shop model for leniency applications.[41] Under this proposal, an applicant could obtain a marker in one designated jurisdiction, which would then notify the affected jurisdictions, and the marker would give the applicant priority in all jurisdictions.[42] This model could make the application process more efficient for prospective applicants and create a coordinated system of enforcement.[43 However, the one-stop-shop gives rise to its own enforcement challenges.[44] For instance, an applicant could provide sufficient information concerning its activity in the designated jurisdiction to obtain a marker but not enough for other jurisdictions, and yet still be covered across all jurisdictions; this might deter cartelists from disclosing information in those other jurisdictions given they were not the first in line, leaving the other jurisdictions without fuel for their own investigations.[45] Therefore, competition authorities across jurisdictions would need to have a mechanism for information gathering and withdrawing their markers if necessary.[46] Issues like these would need to be addressed for the one-stop-shop model to be effective.[47] As of now, the one-stop-shop modeled has not been implemented by enforcement authorities in any jurisdiction.[48]
Conclusion
To curb the decline of leniency applications, international competition authorities should consider leveraging machine learning tools to improve their screening capabilities as COMCO, the Division, and CADE have done, and adopt a one-stop-shop model. By doing this, authorities can establish themselves as not only independent enforcers but also a powerful collective, leaving cartelists minimal choice but to cooperate and relinquish their stranglehold over markets worldwide.
*Special thanks to Professor Laurence Sorkin for giving me his permission to publish this post on the JCFL blog, as it will be, in parts, excerpts of my larger paper-in-progress for the International Cartel Enforcement class.
[1] Makan Delrahim, Assistant Att’y Gen., Antitrust Div., U.S. Dep’t of Just., Opening Remarks at Roundtable Discussing the Antitrust Criminal Penalty Enhancement & Reform Act (Apr. 11, 2019) [hereinafter Delrahim’s Opening Remarks 2019], https://www.justice.gov/opa/speech/file/1153656/dl.
[2] Robert B. Bell & Kristin Millay, The Antitrust Division’s Corporate Leniency Program, Learn from the Past or Be Condemned to Repeat It, Hughes Hubbard & Reed LLP (Spring 2019) [hereinafter Bell & Millay 2019], https://files.hugheshubbard.com/files/Antitrust-Division%E2%80%99s-Corporate-Leniency-Program.pdf.
[3] Delrahim Opening Remarks 2019, supra note 1.
[4] The Future of Effective Leniency Programmes: Advancing Detection and Deterrence of Cartels, OECD Roundtables on Competition Policy Papers 299 (June 9, 2023) [hereinafter OECD Roundtables 2023], at https://doi.org/10.1787/9bc9dd57-en.
[5] Id.; Joseph E. Harrington, Jr. & David Imhof, Cartel Screening and Machine Learning, 2 Stanford Computational Antitrust 133 (Aug. 2022) [hereinafter Harrington & Imhof 2022], https://law.stanford.edu/wp-content/uploads/2022/08/harrington-imhof-2022.pdf.
[6] OECD Roundtables 2023, supra note 4.
[7] David Harper, Business cartels: understanding competition law, Companies House, Gov.UK (Oct. 19, 2018), https://companieshouse.blog.gov.uk/2018/10/19/business-cartels-understanding-competition-law/.
[8] Delrahim’s Opening Remarks 2019.
[9] Bell & Millay 2019.
[10] Id.
[11] Id.
[12] Id.
[13] OECD Roundtables 2023.
[14] Siobhan Kahmann & Johan Ysewyn, The decline and fall of the leniency programme in Europe, Concurrences 54 (2018), https://www.cov.com/-/media/files/corporate/publications/2018/02/the_decline_and_fall_of_the_leniency_programme_in_europe.pdf.
[15] Id. at 46.
[16] OECD Roundtables 2023.
[17] Id.
[18] Id.
[19] Id.
[20] Id.
[21] Id.
[22] Id.
[23] Id.
[24] Harrington & Imhof 2022.
[25] Id.
[26] Data Screening Tools in Competition Investigations – Background note by the Secretariat, OECD (Nov. 28, 2022), https://one.oecd.org/document/DAF/COMP/WP3(2022)5/en/pdf.
[27] Sylwester Bejger, Machine Learning in Cartel Screening – The Case of Parallel Pricing in a Fuel Wholesale Market, 17 Energies, MDPI (Aug. 22, 2024), https://www.mdpi.com/1996-1073/17/16/4184.
[28] Harrington & Imhof 2022, supra note 24.
[29] David A. Higbee et al., US DOJ tests new approaches to boost cartel enforcement revival efforts, A&O Shearman (Aug. 25, 2003), https://globalcompetitionreview.com/review/the-antitrust-review-of-the-americas/2024/article/us-doj-tests-new-approaches-boost-cartel-enforcement-revival-efforts.
[30] Id.
[31] Press Release, Administrative Council for Economic Defense, Office of the Superintendent-General of CADE recommends conviction of cartels in public procurements, Cartel, gov.br (Aug. 4, 2023), https://www.gov.br/cade/en/matters/news/office-of-the-superintendent-general-of-cade-recommends-conviction-of-cartels-in-public-procurements.
[32] OECD Roundtables 2023.
[33] Roundtable on challenges and co-ordination of leniency programmes – Note by the United States, OECD (June 5, 2018), https://www.ftc.gov/system/files/attachments/us-submissions-oecd-2010-present-other-international-competition-fora/leniency_united_states.pdf.
[34] OECD Roundtables 2023, supra note 32.
[35] Pedro Anitelle & Eduardo Frade, The Andean Community and Leniency in Latin America, CPI Columns Latin America, Competition Policy International (July 1, 2024), https://www.pymnts.com/cpi-posts/the-andean-community-and-leniency-in-latin-america/.
[36] OECD Roundtables 2023, supra note 32.
[37] Id.
[38] Id.
[39] Helen Gornall et al., ECN+ Directive misses the boat on compliance and one-stop-shop leniency programmes (Mar. 14, 2019), https://www.debrauw.com/articles/ecn-directive-misses-the-boat-on-compliance-and-one-stop-shop-leniency-programmes; Jennifer Boudet & Johan Ysewyn, Leniency and competition law: An overview of EU and national case law, e-Competitions Antitrust Case Laws e-Bulletin (Aug. 2, 2018), https://www.cov.com/-/media/files/corporate/publications/2018/08/leniency_and_competition_law_an_overview_of_eu_and_national_case_law.pdf.
[40] Id.
[41] OECD Round Tables 2023.
[42] Id.
[43] Id.
[44] Challenges and Co-Ordination of Leniency Programmes – Background Note by the Secretariat, OECD (June 1, 2018), https://one.oecd.org/document/DAF/COMP/WP3(2018)1/en/pdf.
[45] Id.
[46] Id.
[47] Id.
[48] OECD Round Tables 2023.
