Artificial Intelligence and EU Competition Policy – The Algorithmic Collusion

Blogs May 4, 2018


Machine learning algorithms power systems from the trivialto the life-changing. AsAIand algorithms have practically invaded all areas of our everyday lives, no wonder they have also reached the antitrust community.In the past few years there has been considerable debate around the potential anti-competitive impacts of pricing algorithms: software that automatically monitors competitors' prices in real time, and adapts prices accordingly. In fact, pricing algorithms are one of thetopissues on the agenda of the European competition authorities,due to two potentially detrimental effects on business conduct: price discrimination and collusive practices.

An increasing number of businesses have adopted pricing algorithms to their business models to obtain better pricing models, customise services and predict market trends. For instance, one study found proof of more than 500 vendors using algorithmic pricing on Amazon.The increasing use of algorithms has also raised concerns of probable anti-competitive behaviour, since they can facilitate the reaching and sustaining of collusion among companies, without any formal agreement or human input.

By constantly monitoring the behavior of competitors, companies are able to stabilise the market resulting in a greater likelihood of parallel conduct. There may be no longer any need for market players to explicitly coordinate their conduct, thanks to algorithms. Algorithmic pricing may therefore create a market environment in which companies engage in tacit collusion and price fixing without running the risk of being held liable under competition laws, more specifically under Article 101 TFEU.However, a tacitly collusive outcome may not be desiredasit deliberates companies the capacity to significantly suppress output or increase prices to the disadvantage of consumers same way an explicit agreement would.

When analysing the current EU competition law framework, one main problem can be easily distinguished: when a company implementing discriminatory pricing isn’t dominant and its practices fall outside the scope of Article 102 TFEU, or companies using pricing algorithms have not entered into any agreements or concerted practices to do so, the use of pricing algorithms falls outside the scope of Article 101 TFEU, even if the use results in higher prices for consumers. Clearly, there is a grey area between the right of businesses to intelligently adapt to the behavior of competitors and anti-competitive collusion. Algorithms falling outside the scope of our competition laws pose a problem, since they facilitate collusion without the need for communication or contact between competitors. Tacit collusion is likely to become a more frequent phenomenon in the digital markets and therefore, competition authorities may wish to adjust their approach to how tacit collusion is legally treated.

However, the EU would currently have difficulties in bringing a case based merely on a purely software driven choice to price fixing, as it lacks an agreement between parties. To cover the issue,competition authorities and courts should be prepared to look beyond traditional forms of restrictive agreements and concerted practices as currently stipulated by Article 101 TFEU.To safeguard effective competition enforcement in the digital era where algorithms have become increasingly prominent, the scope of liability for collusion and current notion of agreementals need re-evaluation. To conclude, although algorithms pose a challenge to competition, careful assessment and cautious approach should be taken in all actions. The complex digital phenomenon is filled with uncertainty and the lack of intervention and over regulation both pose severe costs on society.