Skip to main content

Internet & media

Pricing algorithms: the risk of (inadvertent) digital collusion

The adequacy of competition in the digital sphere has come under increasing scrutiny over the past years. Regulators around the world, particularly in Europe, are showing a growing interest in the antitrust implications of digitisation and have started bringing test cases, along with a number of consultations, inquiries and reports.

Digitisation raises a number of entirely novel issues, which regulators are going to have to address. In her speech at the 18th German Federal Cartel Office Conference on Competition on 16 March 2017, European Competition Commissioner Margrethe Vestager made this very point in relation to dynamic algorithmic pricing.

It will not come as a surprise in this regard that where pricing algorithms are used to implement and/or operate cartels, companies and their executives will be facing the same penalties as though they had carried out the cartel themselves. But what if the use of pricing algorithms led to companies committing antitrust offenses without anybody at those companies realising? What, if, for instance, algorithms figure out ways to coordinate prices without the developers / users of these algorithms being aware of it? And what if the growing use of dynamic algorithmic pricing creates a level of market transparency which in itself has the effect of dampening competition?

In a world where prices are increasingly set by algorithms, it is unclear to what extent existing competition law concepts are sufficient to deal with possible anticompetitive behaviour or how regulators are going to tackle the likely enforcement challenges that lie ahead. It is therefore important for companies to follow closely authorities’ thinking and keep themselves informed of developments in this area, as it may directly impact on their business strategies. 


Companies make increasing use of algorithms to monitor and then adjust prices automatically. According to the Commission’s e-commerce sector inquiry of last year, out of those online retailers who monitor their rivals’ prices, 67% use automatic pricing software to do so. In almost 80% of cases prices are then adjusted manually or automatically on the basis of the information obtained in this way.

In her speech, Commissioner Vestager warns that certain uses of pricing algorithms could infringe competition law. According to her, ‘the challenges that automated systems create are very real. If they help companies to fix prices, they really could make our economy work less well for everyone else’. Vestager is echoing similar statements made previously by her German, French, UK and US counterparts, for instance, by the French and German competition authorities in their report on ‘Competition Law and Data’.

The Digital Collusion Scenarios

Vestager notes different ways in which computers may collude. These can broadly be grouped into four scenarios, some of which overlap with the categories identified by Ariel Ezrachi and Maurice Stucke in their recent work on ‘Virtual Competition’. 

The four collusion scenarios 

Four collusion scenarios

The first two scenarios (the classic digital cartel and the hub-and-spoke) are relatively straight-forward. The core concepts of what makes a cartel remain relevant to these types of arrangement. However, businesses should be wary of the hub-and-spoke, as, in practice, it may be difficult to spot. The third scenario (M2M & self-learning algorithms) might be more tricky to control and stretches existing antitrust concepts. The final scenario (tacit algorithmic collusion) is an outlier in that it does not, under current rules, result in illegal anticompetitive conduct. However, it raises interesting issues and calls for close monitoring to see how antitrust authorities will address these.

Download our guide to the four scenarios 

What this means for businesses

Given the widespread use of pricing algorithms, antitrust officials are going to have to address the questions that arise with regards to enforcement and the application of traditional antitrust concepts. The policing of such algorithms is likely to become part of the day-to-day work of authorities. The development and use of pricing software will increasingly be subject to thorough antitrust analysis. Ignorance and lack of intent cannot protect against liability. Businesses will therefore need to closely analyse their pricing algorithms from an antitrust perspective and ensure, both at the programming as well as at the application stage, that they have appropriate human and technical safeguards in place, so as not to run the risk of committing competition law offences.