Algorithmic collusion and competition law

Guest CGB-CLARS seminar by A/Prof Rob Nicholls, UNSW Business School

Inter-faculty Competition Law and Economics Seminar of 13 May 2022

Monash University, Caulfield campus

Full details: Visit the event page

To paraphrase a prominent warning from the 19th century, a ‘spectre’ is haunting our digital future – the spectre of Algorithmic Coordination. More prosaically: it seems likely that the expanding use of algorithms to set and adjust prices, and the potential of such algorithms to facilitate a variety of outcomes—classic price-fixing, enhanced conscious parallelism, or unconscious parallelism—will increasingly pose significant challenges for the application of competition law principles that developed in an analogue world.

This was a principal theme discussed on 13 May 2022 by Professor Rob Nicholls as part of an inter-faculty Competition Law and Economics seminar series co-organised by the Centre for Commercial Law and Regulatory Studies (CLARS) and the Centre for Global Business (CGB) of the Monash Business and Economics Faculty. This special video-recorded seminar can be watched below.

As the global conversation around the subject of ‘algorithmic coordination’ has taken shape, particularly following the early work of Ariel Ezrachi (Oxford University) and Maurice Stucke (University of Tennessee) in 2016, Professor Nicholls has been a leading contributor to that discussion.

As explained in the literature, price algorithms can contribute to welfare-detrimental outcomes in a number of ways. At a basic level, competing firms can use algorithms to implement or enhance their already-agreed upon scheme to fix, maintain or control prices. An algorithm could also be implemented or employed by a ‘hub’, possibly but not necessarily operating on the same market, thus functioning as a conduit for a cooperative price coordination mechanism involving various ‘spokes’. In other cases, which until now have been more hypothetical compared to the first two, conventional legal principles seem unlikely to prevent market equilibria that could transfer vast amounts of wealth from consumers to producers. In particular, in a third scenario known as ‘predictable agents’, the algorithms of competing firms practise mutual monitoring and continual adjustment to prices and market data with a view to maximising profits. This amounts to lightning-quick, highly stable conscious parallelism, also sometimes called ‘tacit collusion on steroids’. Finally, a fourth scenario known as the ‘digital eye’ involves autonomous machine learning by algorithms leading to independently derived, evolutionary profit-maximising commercial decisions. Some of these scenarios may raise particular concerns, especially given the possibility that discerning genuine collusive pricing from mere independent adaptation to market conditions may be extremely difficult (thus likely raising the risks and costs of enforcement errors).

Competition law enforcers are not wholly without means to develop techniques to counter the expected rise of algorithmic coordination. For example, Ezrachi and Stucke (2017), drawing on the work of economists Jin Li (HKU and Northwestern) and Charles Plott (California Institute of Technology), note that agencies can use modelling and simulation tools such as so-called incubators to enhance their capacity to detect collusion. In particular, the agency’s own digital tool ‘could shadow the industry’s algorithms, until it was mirroring the industry responses. The agency would then test what conditions added to (or removed from) the incubator would make tacit collusion likelier and more durable. […] Here, the agency can see how the algorithms respond, and what factors help promote, stabilize and destabilize algorithmic tacit collusion.’

Nevertheless, as Professor Nicholls discussed in this seminar, it is doubtful that the Australian rules prohibiting cartels, or indeed, the cartel rules in any jurisdiction, are well-suited to address the third and fourth scenarios above. He further queries whether even the relatively new legal provision prohibiting concerted practices that substantially lessen competition, which may enable the ACCC or a plaintiff to bypass the usual ‘meeting of the minds’ and ‘commitment’ criteria, is supple enough to cope with these challenges that are still hypothetical but fast approaching.