The International Competitive Network (ICN) is the eminent group of competition authorities from over 130 jurisdictions.  The ICN published a scoping paper on 28 April 2020 for its cartels working group sub-group project on “Big data and Cartels”. The purpose of the initiative is for the ICN members to discuss and debate the issues and consider how this might lead to an update of chapters of the ICN Anti-Cartel Manual. 

The paper looks at data and algorithms as a vehicle for collusion, and the second part focuses on data and algorithms as a tool for cartel detection.

The issue of algorithms and collusion, while a recent debate, is not new. G+T first covered competition law issues such as this in our 2016 article “Can Robots Collude?”; which preceded a keynote and panel discussion with ACCC Chairman Rod Sims.  G+T noted that it is relevant for businesses to consider the competition law implications of using algorithms to optimise business operations, and to think about how to design such systems with compliance in mind.  However, the Australian legal position likely accommodated enforcement of such colluding conduct through the concerted practices prohibitions introduced in November 2017 following the Harper Review.

But if there’s one thing for businesses to take away this week it’s that the debate and policy landscape internationally on colluding robots is well and alive – and it is relevant to be aware of their evolution to ensure operationally your business is not caught by ‘surprise’ by new competition laws. The ICN paper reflects an increased and disciplined global competition regulator focus on the use of algorithms and issues of collusion.  Shortly after on 2 June 2020, the European Commission also commenced consulting on the need for new powers to address algorithm-based tacit collusion

A summary of the ICN developments is below.

Is a change to cartel enforcement on the horizon?

Big data (the management of large databases), algorithms and AI (the tools used to process Big data) have become increasingly common features of many industries in our economy, benefitting both businesses and consumers through data driven innovation. Businesses can maximise profits by processing significant amounts of data and responding to markets in real time, and consumers benefit from new and increasingly tailored products.

However, despite these benefits the debate continues around the extent to which Big data, algorithms and AI can be used by competitors to fix prices, allocate markets or rig bids in new ways. For example, in a world where companies use sophisticated algorithms in parallel, competitors in the same market could autonomously decide that a price fixing strategy would be optimal and then ‘agree’ to collude with each other without human intervention. This is known as ‘tacit collusion’, and while it may lead to the same economic detriments as explicit collusion, it does not currently have the same legal implications.

What issues were raised by the ICN?

In its recent scoping paper “The impact of digitalization in cartel enforcement’, the ICN set the stage for potential changes to its Anti-Cartel Enforcement Manual, last updated in 2009.

This scoping paper called for discussion around the following cartel enforcement issues that competition regulators are currently grappling with in this new digital era:

  • How to distinguish explicit collusion from tacit collusion. Under tacit collusion (or conscious parallelism), non-competitive outcomes are achieved by each participant deciding on its own profit-maximising strategy independently of its competitors.  How do you separate this (good) conduct from the bad?
  • How to determine intention and a ‘meeting of the minds’: Given the difficulty of establishing that two companies intended to collude via the use of algorithms, could the consciousness of the collusive outcome be sufficient to fall within the provisions on anticompetitive agreements?
  • Whether market structures should be considered: As algorithms can achieve fast price matching even in markets where traditional methods of price fixing by competitors are unlikely to succeed, should market structures be considered in determining if competition has been restricted? Interestingly, on June 2, the European Commission called for greater market investigation powers on this point, seeking the ability to look beyond abuse by dominant firms and directly address structural competition issues within market sectors.
  • Who is liable: As firms apply an algorithm’s autonomous decisions, when is an algorithm under that firm’s control? Could both a company that developed the algorithm and the benefitting firm be found liable?

Ultimately, the members of the ICN will consider whether competition authorities can tackle algorithmic collusion under existing cartel enforcement measures or whether new measures should be considered.

What’s happening in Australia?

The ACCC has been alive to the potential of algorithmic cartels for a number of years, with Chairman Rod Sims notably stating in 2017 that liability could not be avoided by using the excuse “my robot did it”.

Following the Harper Review of Competition Policy in 2016, the ACCC enacted two new competition laws that take steps towards addressing big data e-collusion:

  • Concerted Practices law, which does not require there to be a “meeting of a minds” when considering whether there has been cooperation between competing businesses that had led to non-competitive outcomes; and
  • New Misuse of Market Power provisions, which looks at the effect or likely effect of conduct rather than the intent of competing businesses.

The digital economy has continued to be a focus for the ACCC, having completed its 18 month inquiry into Digital Platforms, it is now working through two significant ongoing Inquiries into Digital Platform Services more broadly and Ad-Tech.