ARTICLE

A comparative study of the regulation of algorithmic collusion in South Africa and the European Union

Authors: Dylan Abraham Smit & Thalalolwazi Msutu

ISSN: 1996-2177
Affiliations: Chartered Accountant trainee, Nedbank; Lecturer, Department of Mercantile Law, Stellenbosch University
Source: South African Law Journal, Volume 143 Issue 4, p. 799-841
https://doi.org/10.47348/SALJ/v143/i4a8

Abstract

The rapid development of algorithms and artificial intelligence has helped firms to set their pricing in a more efficient way. Among the multiple issues that they bring to competition law, algorithms and artificial intelligence help to facilitate tacit collusion and conscious parallel conduct, leading to anti-competitive harm, a type of collusion that is not always outlawed in competition law. The article investigates algorithmic collusion, assessing whether South African law can regulate it to prevent anti-competitive harm, and conducts a comparative analysis of how the European Union is regulating it. The article finds that competition authorities stand at a crossroads concerning the regulation of algorithms and artificial intelligence. They may rely on current traditional tools of competition law, potentially adapting tools such as the hub-and-spoke conspiracy, minimum resale price maintenance and merger control to restrict market conditions that facilitate algorithmic collusion. Alternatively, they may develop new regulatory tools, including removing the requirement of agreement in collusion cases, reversing the onus of proof once anti-competitive harm is proven and introducing ex ante regulatory measures — such as the European Union’s Digital Markets Act and the Artificial Intelligence Act — or market inquiries to supplement existing competition law. A hybrid approach, combining elements of both traditional and novel tools, may ultimately prove most effective.