By: Christopher J. Allen
The National Association of Attorneys General (NAAG)’s Eastern Region Meeting, held August 1-3, 2023 in Connecticut in conjunction with the Attorney General Alliance, included an “Antitrust Bootcamp” that surveyed the history and current state of and future challenges for state and federal antitrust enforcement. AGs from 9 states (CT, DE, ME, NH, NY, OK, RI, TN, and VT) and Puerto Rico were in attendance, plus senior staff and antitrust AAGs from at least 10 additional states (CA, CO, DC, MA, MS, NV, NJ, NC, PA, WI).
This third in a series of three articles builds on the previous articles in the series (article 1 + article 2) and summarizes the panel discussions on non-competes, AI and healthcare as each of these topics relates to antitrust.
“Non-Competes” Panel Looks At Competitive Impacts
The non-compete panel comprised of Maine AG Aaron Frey, University of Wisconsin Professor Peter Carstensen, US DOJ Principal Economist Ioana Marinescu, and FTC Bureau of Competition Deputy Director Rahul Rao, first discussed the alleged overuse of non-compete agreements across positions with lower wages and responsibilities. The result, according to the panelists, is to limit the availability of jobs and to suppress employee pay and mobility, resulting in harm to employees both during and after their employment, as well as hampering innovation and the economy as a whole. The panelists recognized there may be legitimate use of non-competes to protect trade secrets and proprietary information, and in some industries they may actually serve the public interest by preventing management and executives from moving freely between competitors and creating opportunities for coordination. At the same time, non-competes must be narrowly drafted and fine-tuned, and current agreements are far too broad, pervasive, and have been deployed by employers even in states like California where they cannot legally be enforced. As such, this will continue to be a major area for state and federal enforcement, as well as an FTC rulemaking on the issue.
Inter-Industry Information Exchange Can Limit Competition
The panel then shifted to discuss the related topic of inter-industry information exchange as an anticompetitive strategy, specifically identifying competitors publicly sharing data like wages, list prices, and inventory. The panelists argued that industry may justify such practices as allowing companies to improve their practices, but it also may create opportunities for rivals to coordinate and collude. This has resulted in the withdrawal of federal guidance that had said that sharing information was permissible and potentially beneficial under rule of reason. The panelists argued that rivals sharing information, even by means of a third-party aggregator service, should bear the burden of showing that sharing is reasonable. Otherwise, the practice can go hand-in-hand with other measures like anti-competes to limit worker pay and mobility.
AI Tools and Data Concentrated in the Hands of A Few Large Firms
New York AG Leticia James moderated the next panel, which included US DOJ Policy Advisor and Economist Jerry Ma, Boston University School of Law Technology & Policy Research Institute Executive Director James Bessen, Penn State Dickstein Law Professor Daryl Lim, and Wilson Sonsini privacy partner Maneesha Mithal. A key concern of the panelists was the control of AI being consolidated in the hands of a few large firms, which already have enormous reserves of data that they already have used to obtain significant competitive advantages by investing in traditional analytical software. The collection of vast amounts of data collected for use by AI raises intellectual property, antitrust, consumer protection, and data privacy and security concerns, as well as the risk that AI analytics will reflect and perpetuate real world biases in the underlying input materials. The panel stated that, after the initial period in which industry disruption peaked in the 1990s, traditional software utilization actually now has led to decreased disruption by increasing incumbents’ competitive advantage, as only very large firms can bear the very large costs of collecting and leveraging huge amounts of data. AI has the potential to exacerbate this trend, and while consumers potentially can benefit from the use of AI, there is also a risk of greater consolidation and undermining the remedies that can be used to address competitive concerns.
Advanced software also can help companies evade regulation, as the significant complexity provides opportunities to deceive and obfuscate when regulators lack the tools, resources, and sophistication to keep up. Algorithms can be used to coordinate among competitors indirectly, (such as AI-stabilized coordination on prices) without human conspirators. Also, the need for massive data sets to train AI creates issues for companies seeking to remain compliant with state data privacy/security laws, especially when sensitive and personal identifying information is involved.
The panelists identified measures that could be taken to address these concerns. They include mandating transparency about software from the companies that develop and use the technology, as well as new remedies for competition concerns like requiring companies to provide competitors with access to data sets. To address AI facilitated cooperation, regulators may have to reevaluate how they define markets and potential practices’ impacts on competition. Greater consumer disclosures and opt-outs may be required to explain the collection and use of data to train AI, as well as requirements that companies undertake bias evaluations in algorithmic systems and be able to explain those systems to consumers.
AI Also Brings Advantages for Regulators
The panelists also noted that AI also can be a boon for regulators, such as the growing practice of “computational antitrust” in which AI can be used to better explore and understand market dynamics, as well as public perception and consumer sentiment (e.g., monitoring and analyzing social media). The concluding sentiment was that AI is novel, but the issues it raises are not and our current antitrust statutes are pretty flexible and adaptable to allow regulators and enforcers to address AI just as any other issue. If an enforcement authority evaluating suspected anticompetitive conduct substituted a person for an AI and the concerns remain the same, then the approach should also be the same.
Consolidation in Healthcare Markets Presents Unique Challenges
The final panel examined the unique challenges posed by consolidation in the healthcare market and was co-moderated by New Hampshire AG John Formella and Rhode Island AG Peter Neronha, who were joined by FTC Bureau of Competition Deputy Director Rahul Rao, California Supervising Deputy Attorney General for Healthcare Rights and Access Emilio Varanini, and US DOJ Chief of Healthcare & Consumer Products Section Eric Welsh. The panelists noted that healthcare presents unique competition issues, including a trend towards consolidation at all levels of healthcare, dealing with post-COVID effects such as the lack of personnel, the significant role played by charitable institutions, and dichotomy between big broken urban centers and undermanned rural facilities.
Overall, there is a tension between healthcare systems that are weak and unsustainable as they currently are, but at the same time that cannot be allowed to merge/consolidate because of the impacts on competition. The panel noted that AGs can play a key role in keeping federal regulators informed on the potential for a competition crisis. The panel also noted that healthcare mergers tend to lead to higher prices, which is causing the FTC to scrutinize potential effects even when under traditional market definitions they might not do so. Finally, the panelists expressed concern regarding the role of private equity in healthcare, noting that providing adequate healthcare requires adequate financial and resource support, yet private equity tends to focus on cost-cutting, which creates conditions where further consolidation becomes inevitable.