In a discussion with The Regulatory Review, Gina-Gail S. Fletcher explains how financial regulators should approach both the opportunities and the challenges presented by technological advancements. Fletcher argues for a harm-based approach to assigning liability for trading misconduct, rather than an intent-based approach, in an era of sophisticated artificial intelligence in which algorithmic trading is prominent. She notes that although regulators’ technological capabilities tend to lag… Read More »
In a discussion with The Regulatory Review, Gina-Gail S. Fletcher explains how financial regulators should approach both the opportunities and the challenges presented by technological advancements.
Fletcher argues for a harm-based approach to assigning liability for trading misconduct, rather than an intent-based approach, in an era of sophisticated artificial intelligence in which algorithmic trading is prominent. She notes that although regulators’ technological capabilities tend to lag behind industry capabilities, novel technologies such as artificial intelligence have a role to play in identifying and regulating harmful behavior in financial markets. She also argues that the U.S. Securities and Exchange Commission (SEC) and the Commodity Futures Trading Commission (CFTC) should coordinate in establishing regulatory guidelines for algorithmic trading in each agency’s respective market. She concludes that regulation provides a set of guardrails for sustainable innovation in financial markets.
Gina-Gail S. Fletcher is a professor of law at Duke University School of Law. She currently serves as a member of the Financial Industry Regulatory Authority’s National Adjudicatory Council. She previously served on the SEC’s Investor Advisory Committee, the CFTC’s Market Risk Advisory Committee, and the Financial Industry Regulatory Authority’s Investor Issues Committee. Prior to joining Duke Law, she was an associate professor at Indiana University Maurer School of Law and a visiting assistant professor at Cornell Law School.
The Regulatory Review is pleased to share the following interview with Gina-Gail S. Fletcher.
The Regulatory Review: In a recent article, you noted the potential for harm to consumers and financial markets if regulators do not sufficiently regulate sophisticated algorithms and artificial intelligence. What should financial regulators’ approach toward novel and sophisticated algorithmic technology look like?
GSF: Regulation tends to be reactive to market changes, rather than proactive. This is especially true with technological advancements, where industry is likely to be ahead of regulators. In approaching new technology, regulators should consider not just its positive aspects but also how it may have been designed to circumvent existing regulatory tools, how existing regulations apply to it, and what regulations may be needed to address truly novel harms arising from it. Regulators may not have full or even partial answers to all these questions when first encountering new technology, but it ought to be how they first approach new developments in the markets. Even still, at the pace at which AI and trading algorithms are developing, regulators will necessarily be behind the latest technology. Recognizing that many market developments seek out regulatory gray spaces, regulatory focus should be on understanding the pluses of the technology—such as improving pricing and informational efficiency, lowering costs, or improving investor market access—and the minuses of the technology—such as increasing market risks, exploiting systemic weaknesses, or eroding investor protection.
TRR: How should regulators decide when algorithmic trading should qualify for the imposition of liability?
GSF: Under the current regulatory framework in the United States, liability for trading harms, such as market manipulation, depends on the intent of the trader. When human traders dominated the markets, this intent-based framework made sense, especially given that the line between “harmful” and “aggressive” trading can be somewhat blurry. In today’s financial markets, however, in which over 90 percent of traders use AI and algorithms in their trading, basing liability on intent is an ineffective way to impose liability for harmful conduct in the markets. First, non-human actors cannot form intent and, therefore, cannot be liable for intent-based misconduct. Second, even if one focuses on the actors behind the algorithm or AI, deciphering those actors’ intent will be difficult except in the most egregious cases. Intent has always been notoriously difficult to prove in manipulation cases, and with algorithms and AI creating a layer of separation between the trader and the alleged misdeeds, it is likely to be even more difficult to assign liability based on intent.
Given these difficulties, which are only exacerbated by algorithms and AI, liability for trading misconduct should focus on the harm that results rather than the intent of the trader. That is to say, if the algorithm’s trading creates an artificial price or sends false signals to the markets, it should not matter whether the trader intended to manipulate the markets. What should matter is that the conduct did distort the market. A harm-based approach considers whether the trading behavior distorted asset pricing, impaired informational efficiency in the markets, exploited market conditions to heighten volatility, or otherwise undermined market integrity. Such liability should attach regardless of the intentionality of the traders’ actions, with the level of liability or sanctions being determined by whether such conduct was intentional, reckless, or negligent.
TRR: In an era in which sophisticated algorithmic trading in both securities and commodity derivatives remains common, to what extent should the government still regulate these financial products separately
GSF: The question of whether the SEC and CFTC ought to remain separate agencies is a long-running one, the salience of which is only heightened as markets become more interconnected and more complex. Without a larger, more comprehensive overhaul of financial regulation in general, simply combining the two agencies is unlikely to result in improved regulation of algorithmic trading and better enforcement against attendant harms. The agencies have different missions, objectives, and regulatory priorities, and imposing SEC-like regulations over the derivatives markets (and CFTC-like regulations over the equities markets) may cause more confusion and disruption to the markets rather than good. Short of a fundamental overhaul of financial regulation, a better approach would be for each agency to establish regulatory guidelines for algorithmic trading in their respective markets and coordinate their regulations to minimize the gaps and loopholes in which regulatory arbitrage thrives.
TRR: To what extent do resource and technology constraints affect regulators’ ability to police algorithmic behavior?
GSF: It is unsurprising to say that regulators are several technological steps behind the industry, which means that regulators are often playing catch-up in understanding the impact of new technologies in the market. Resource constraints—both monetary and expertise—contribute to regulators lagging behind the industry. But the fact is that even with bigger budgets or greater expertise, regulators will always be behind industry. This only further underscores the importance of having regulations that can develop alongside the markets and technology, which means moving away from human-centric concepts of liability for trading misconduct.
TRR: What role should digital technology, including AI, play in regulatory enforcement?
GSF: AI has an important role to play in regulatory enforcement. It is one of the most potent tools regulators have today to aid them in market surveillance, to identify bad actors in the markets, and to hold those actors accountable. Leveraging AI to assist with monitoring and enforcement is especially necessary given the vast quantities of data that must be reviewed to identify harmful and illegal trading activities in the markets. With the assistance of AI, regulators are better equipped to identify patterns and trading behavior that indicate or strongly imply harmful activities, such as manipulation, front running, or wash trading, faster. As such, regulators must continue to embrace AI as a regulatory tool to safeguard market integrity and efficiency.
TRR: In general, how should regulators and policymakers think about balancing the need for robust, effective regulation that protects consumers and markets with the advantages brought about by technological innovation?
GSF: Too often consumer protection is seen as being in tension with technological innovation, and, as such, the former must suffer in order for the latter to flourish. However, sustainable technological innovation must coexist with consumer protection. Achieving that balance requires regulators to act as arbiters in the marketplace—considering the environment needed for technological advancement that does not require consumer or investor exploitation. It is also important to identify when technological advancements result in new versions of old problems. Disruptive trading done through new technology is still manipulative, regardless of the format used. As such, it is important for regulators not to be blinded by the new technology and, instead, to see the ways in which the misconduct at issue is familiar. Lastly, this balance requires eschewing the belief that all regulation impedes technological advancement. Technological advancements thrive with guardrails that allow for innovators, the markets, and consumers and investors to absorb changes as they occur. Regulation, therefore, should not be viewed as a hinderance to technological innovation but as a set of rules through which sustainable innovations may thrive.
Tagged: algorithmic trading, Artificial Intelligence, Commodity Futures Trading Commission, Securities and Exchange Commission, technology regulation
