12 April 2017

Know your algorithms – or pay the price

Companies should be on their toes when using software tools such as pricing or monitoring algorithms that check and adjust prices automatically. If used to implement cartels or monitor resale price maintenance more effectively, this could lead to higher fines. The European Commission recommends companies ensure antitrust compliance by design; blaming the software or limited computer skills for an infringement will not help. If algorithms are designed to lead to collusion, companies will be held responsible by the Commission. Companies should therefore know exactly how their algorithms work and what they do before using them.

Competition Commissioner Vestager recently warned companies to beware of automated systems that monitor and adjust prices automatically, as they could be used for anti-competitive goals. A monitoring algorithm could, for instance, facilitate price-fixing cartels by screening for compliance with the fixed price among cartel members. Similarly, this algorithm could be used to facilitate resale price maintenance by enabling a supplier to easily monitor the prices set by its retailers. The Commission is currently investigating whether the use of this software may have been an aggravating factor in the online sales practices of a number of consumer electronics manufacturers (see our earlier In context for more information).

However, it remains to be seen whether the competition rules will allow competition authorities to tackle all possible unwanted effects of algorithms. As noted in the earlier joint report on competition law and data by the French and German competition authorities, it will be difficult to prosecute if there is no sign of coordination. For instance, if businesses use a similar pricing algorithm built by the same company, this could affect price competition without there being any kind of horizontal coordination between the businesses. In addition, data-based algorithms could have an impact on price competition by integrating in their automated price adjustments the competitors’ reactions to price variations, based on past experiences of price variations, without the need for any contact between competitors. Time will tell how competition authorities deal with these issues.