Biennial Report XXII of the Monopolies Commission under § 44(1) ARC, 3 July 2018
Despite the numerous advantages associated with the use of price algorithms, potential disadvantages are likewise increasingly being discussed. Commentators consider anti-competitive effects in connection with collusion possible where price algorithms are used. Collusion is typically understood as a market result in which companies achieve higher profits than in competition where they coordinate, for example in relation to prices or quantities. Collusive behaviour is therefore to the detriment of the demand side of the market and is undesirable from the point of view of society as a whole.
The influence of price algorithms on collusion largely depends on the structural characteristics of the respective market and on other supply and demand factors. Depending on how these factors are shaped, price algorithms as one further element can promote collusion. However, from today’s perspective, no reliable predictions can be made as to whether collusion will occur more frequently in the future. In the end, collusion can be expected primarily in markets that offer favourable conditions for collusion. Such conditions include high barriers to market entry, a relatively small number of companies and a high degree of market transparency.
In data-intensive sectors such as the Internet economy, price algorithms can facilitate collusion by automating collusive behaviour and, thus, accelerating it technically. For example, they can stabilise collusion by collecting information on competitors’ prices and sanctioning deviations from collusive market outcomes more quickly. The use of price algorithms can also make explicit agreements or agreements restricting competition unnecessary. Lastly, in the case of self-learning algorithms, the relevant business decision is already made at the time of the decision regarding the price algorithm and is not made in the price-setting process.
The discovery of collusive behaviour of companies by competition authorities is generally difficult. This holds true for the determination of concerted practices as such. These difficulties tend to increase when price algorithms are used. This also concerns the proof of potentially excessive prices.
Therefore, markets should be monitored for collusive risks. In particular, the Monopolies Commission considers it necessary to strengthen market monitoring through competition sector inquiries. As information on possibly collusive excessive pricing is most likely to emerge through consumer associations, the Monopolies Commission recommends that consumer associations receive a right to initiate competition sector inquiries. Where the competition authorities reject the application, they would have to provide detailed reasons for their decision. If, in the context of market observation, concrete indications were to arise that the use of price algorithms favours collusive market results to a considerable extent and that the enforcement of the competition rules is insufficient, a reversal of the burden of proof with regard to the damage caused by an infringement of competition law could be considered. In that way, the liability for financial losses which the collusive use of price algorithms can entail could, in case of doubt, be assigned to the users of such algorithms.
Finally, price algorithms are often not designed by the companies themselves but are provided by IT service providers with special expertise. Whether such a third party is liable for violations of competition law typically depends on the responsibility of the companies using the price algorithm. This means that IT service providers can either be subject to particularly far-reaching liability or, conversely, benefit from liability gaps, depending on whether the decision on the design of the price algorithm in question lies more with the user or with the respective IT service provider. The liability of such third parties should be generally reviewed.
- Algorithmen und Kollusion (Chapter of the XXII. Biennial Report in German language)
- Algorithms and collusion (Excerpt from Chapter I of the XXII. Biennial Report)