COMPETITION • EUROPEAN AND FRENCH LAW • Digital age
In Opinion 18-A-03 of 6 March 2018 on the exploitation of data in the internet advertising sector, the Competition Authority defines algorithms as “a sequence of rules to be applied in an exact order to carry out a particular task: it is a logical sequence for obtaining a certain result from a given input. … [They] are now used for the automatic performance of repetitive tasks involving complex calculations and data processing“. More and more businesses are using algorithms to collect large volumes of data in order to monitor the market and adapt their pricing policy in real time to the latest developments in the offer identified by the machine.
Algorithms can therefore be a vehicle for collusion between competing operators, be provided by a third party facilitating collusion or be the driver for the collusion themselves. In the first case, companies agree beforehand on the principle of common pricing before using the algorithms as a tool for implementation, monitoring and concealment. Algorithms can thus constitute the vehicle for collusion between competing operators, or be provided by a third party facilitating collusion or carry out this collusion themselves. In the first case, companies agree beforehand on the principle of common pricing before using the algorithms as a tool for implementation, monitoring and concealment. In its study on algorithms, carried out jointly with the German Bundeskartellamt, the French Competition Authority describes them as “cartel facilitators“. In the second case, competing undertakings use the same pricing algorithm provided by a third party (software developer) which leads to a price alignment without any contact between the competitors themselves. The Guidelines on horizontal cooperation cover information exchanges where data can be shared indirectly through a common agency or a third party (pt. 55). However, for such behavior to fall foul of antitrust law, competitors must know that they are using the same algorithm. Competing undertakings active in oligopolistic markets may, in parallel and without any communication with each other, use algorithms, which, by their ability to self-learn, devise pricing strategies likely to lead to market alignments without any human intervention. The understanding of such a practice with regard to competition law is more delicate as the concurrence of wills criterion is harder to characterize. Algorithms can also be used in the context of a vertical agreement, to allow suppliers to monitor the prices charged by their competitors and to send new instructions to their distributors in real time or monitor the effective application of such instructions. The Commission has sanctioned this type of practice in its decisions in the area of online sales of consumer electronic products.
On the demand side, algorithms make it possible to adopt two types of pricing strategies: i) personalized pricing, determined with regard to the client’s willingness to pay a certain price level, which is itself identified by collecting information on their consumption habits, their place of residence, the use of loyalty schemes, etc.; ii) dynamic pricing, adapted in real time to current market conditions (e.g. yield management). These discriminatory practices nevertheless remain very marginal, as the Commission showed in its final report on electronic commerce.
The use of algorithms can also constitute, in certain circumstances, an abuse of a dominant position insofar as it may, in itself, confer a market power on an undertaking, firstly, because the development of such a tool involves large investment costs and any business using it must have sufficient resources to overcome such a barrier to entry; secondly, because an algorithm makes it possible to collect a very large volume of data, control of the access to it can lead the dominant company to adopt certain abusive behaviors (refusal of access or provision of essential information, threat of de-listing, unfair trading conditions, etc.). Other abusive behaviors that may arise include imposing excessive, predatory or discriminatory prices fixed by algorithm.
