Antitrust Law 4.0 – The effects of digitisation on antitrust law compliance
In its closing report on the sector inquiry on e-commerce dated 10 May 2017 the EU Commission criticised several business practices which could raise cause for concern under competition law. The use of price adjustment software in particular has been on the watch list of the cartel authorities.
The use of software which enables enterprises inter alia to follow prices on the internet and to (automatically) adjust their own prices on the basis of the observed prices has since become widespread. Cartel authorities all over the world are watching this development closely. Digital transparency on the internet, namely, is raising new questions about anti-competitive behaviour – and thus also increasingly creating new challenges for the antitrust law compliance of enterprises. Enterprises are therefore advised to keep themselves up-to-date on “Antitrust Law 4.0” and to closely follow the dynamic thought process in this field.
Through the use of pricing software, for example, dealers’ deviations from manufacturers’ “recommended” retail prices can be discovered at shortest notice and “cautioned” by manufacturers – in many jurisdictions this behaviour already fulfils the constituent criterion of prohibited vertical price fixing. Due to the availability of price information in real time, however, this can also lead to an automated coordination of prices between competitors. Depending on the market conditions, the use and the distribution of pricing software, cartel authorities therefore fear alarming consequences for competition.
But when is this deemed the case? Several scenarios are conceivable here: besides the classic (digital) cartel, in which price algorithms are used as instruments to implement pricing agreements, a further example are hub-and-spoke constellations. If competitors use the same platform or software programme of the same provider (hub), the interaction between the users (spokes) via the common software or platform can lead to a breach of cartel law. Furthermore, scenarios in the M2M (machine to machine) sector and the use of self-learning algorithms have proven a grey area little noticed to date.
The digital cartel
If the software serves as a means to implement a prohibited agreement, the case is clear: It makes no difference whether competitors agree on prices amongst themselves in the “classic” manner over a shared dinner or whether they deliberately monitor and implement them using a software – both are cases of prohibited cartel agreements.
It was for this reason that the US Department of Justice had no problem in punishing an infringement of antitrust law in the so-called “poster cartel”. Here, several dealers had colluded with each other on the prices of their products to be sold via the trading platform Amazon Marketplace and monitored the observance of the agreed price level over the further course of time through a corresponding programming of price adjustment software.
Back to top
Caution when using the same software/platform (hub-and-spoke scenario)
Less clear, yet no less risky from the compliance perspective, are constellations in which enterprises (spokes) use the same software of a third provider (hub) and during the course of the same are subject to the same price algorithms, or which obtain access to competitively sensitive data of competitors via a commonly used platform. Unlike digital cartels, the software or platform is not necessarily deliberately used for anti-competitive agreements in this case. At the same time, however, it can lead to a (price) transparency or even price adjustment amongst the parties with a negative effect upon the competition.
A pioneering judgement in this area is the “Eturas” judgement passed by the European Court of Justice (ECJ) on 21 January 2016. Several travel agencies were using the same cloud-based booking software. The software administrator had sent all software users an electronic communication to the effect that the system stipulated a maximum rebate of 3%. Higher price rebates were still possible, but needed to be individually provided for in the system by the travel agencies. The majority of the travel agencies using the software considered the rebate limit stipulated by the system to be reasonable and therefore also uniformly subsequently advertised with price rebates of 3%. The ECJ ruled that the unreserved use of the rebate limits adjusted by the system represented an illegal price cartel that had been tacitly entered into. To avoid being held liable for this, the enterprises should have expressly objected to the change vis-à-vis the system administrator or should have provably represented that they were systematically also still granting price rebates above the maximum rebate stipulated by the system.
Against this background, enterprises which use (pricing) software or platforms of third providers should keep a keen eye on possible (unconscious) interactions or collusions with competitors which are initiated by the system. This applies all the more in cases where it is known that a specific software or platform is broadly distributed in the enterprise’s specific branch.
Back to top
“When computers collude” – cartel breaches in the M2M sector?
If price algorithms are programmed with the intention of procuring or facilitating a (price) agreement with other competitors, then this must be appraised as a classic price cartel. Here, namely, it is the person as the initiator of the cartel that clearly stands in the fore.
But how are constellations to be appraised in which the mere use of (pricing) software leads to a (price) collusion between competitors, without this having been intended by or known to the enterprises or programmers? How are cases to be appraised in which a self-learning algorithm – with no human intervention - assumes control and, within the scope of a self-learning process, reaches the conclusion that the best strategy to maximise profit is price coordination with competitors? Can an enterprise be held responsible at all for a computer-controlled, anti-competitive collusion between competitors without human contact, or without any contact initiated by humans?
According to the currently applicable principles, enterprises can only be held liable for the anti-competitive conduct of their (executive) employees. In future, however, it could be that enterprises can also be held responsible pursuant to cartel law for the “behaviour” of the software they use. This was at least suggested in the recent statements of the EU Commissioner for Competition, Magrethe Vestager, at the 18th International Cartel Conference of the German Federal Cartel Office [Bundeskartellamt, BKA]:
“What businesses can – and must – do is to ensure antitrust compliance by design. That means pricing algorithms need to be built in a way that doesn't allow them to collude.”
“And businesses also need to know that when they decide to use an automated system, they will be held responsible for what it does. So they had better know how that system works.”
Enterprises therefore might have to ensure in future that their algorithms are programmed in such a way that an exchange of sensitive competitive information or even an (automatic) coordination of prices with competitors is excluded or at least made difficult – quasi a compliance programme for software applications. Should there nevertheless be a (computer-controlled) collusion between competitors, enterprises might have to bear the consequences of the anticompetitive “behaviour” of their machines. From today’s perspective this may well still seem like a science fiction scenario, but technological development is progressing at a fast and furious pace: twenty years ago no-one would ever have believed that computer-controlled fridges would be able to handle your order for your next shopping trip or that electronic language assistants could tell you the name of the current title playing on the radio.
Brave New World!