Market Manipulation and Collusion by AI in Finance: a Primer
--In the last few decades, thanks to both technological and regulatory developments that have completely reformed global financial markets’ infrastructure and thus financial relationships between market players, algorithmic trading has become a fundamental component of trading activities’ total volumes. In this context, artificial intelligence (AI) applications are considered today as potential game-changer for the financial sector as a whole, leading to several efficiency gains. However, when considering the specificities of some AI applications (i.e. the issue of ‘explainability’ of AI decision-making or the so-called “black-box” phenomenon), robo-traders may also present new technology-specific risks that could ultimately jeopardise the integrity of financial markets as well as undermine their stability.
With this in mind, our current research project studies the emergence of this new market reality dominated by robo-traders through the lens of law & economics and, specifically, by addressing the issues of market manipulation and collusion in financial markets in the context of market failures. We suggest that robo-traders may perform better than humans not only in traditional manipulating techniques but also in discovering new ways for manipulation or even learning to collude with other AI by just observing and populating the market.
Building on these intuitions, we argue that existing market abuse regulations, as well as the current legal framework for the governance of algorithmic trading, might prove inadequate to cope with the risks of market failure posed by robo-traders’ activities. For example, AI algorithms can be programmed or even learn by themselves to game the rules with the risk of being immune from prosecution, as most market abuse regimes at present require the market supervisor to establish an element of ‘intention’ of the manipulator, in order for it to count as a financial crime.
With this and other potential risks in mind, we first aim to open a scientific debate over the challenges posed by robo-traders to regulators and supervisors in fulfilling their mandates of guaranteeing the main pillars of financial markets as a public good – beside efficiency: namely, market safety, trust, and integrity.
In this vein, and emphasizing the need for reform, we explore possible policy recommendations for both regulators and supervisors to achieve effective governance and supervision of Robo-traders. First, we stress the need to rethink existing market abuse regulations also to encompass scenarios of financial crime that can be committed by robo-traders autonomously, i.e. without any human intention or direct negligence. Second, we discuss how to enhance the governance framework of robo-traders, including, on the one hand, the proactive adoption of regulatory technology (RegTech) solutions to ensure optimal compliance at the firm level. On the other, we call for a reinforcement of supervisors’ technological knowledge and capabilities, through the adoption of supervisory technology (SupTech) tools and by strengthening the market conduct supervisory strategy to achieve effective surveillance and response mechanism. Indeed, the current division of competences between national and supranational organizations together with the existing regulatory framework of delegation of supervisory competencies to private and self-regulated organizations do not prove to be optimal for achieving effective supervision of financial market conduct at a global level. Finally, on the enforcement side, this project also examines the case for a new “accountability” framework and liability rules for robo-traders’ misconduct given the ongoing and intense scientific and political debate on the legal implication of AI applications in society.