If pricing practices are illegal when implemented offline, they are very likely to be illegal when implemented online as well. Firms involved in illegal pricing practices cannot avoid liability on the grounds that their prices were determined by algorithms. Like an employee or an outside consultant working under a firm's "direction or control", an algorithm remains under the firm's control, and therefore the firm is liable for its actions.
The submission identifies three potential implications for algorithms in vertical cases:
- When algorithm-based price monitoring is used to detect deviations from a fixed or minimum resale price, that does not constitute a resale price maintenance (RPM) offence as such but forms part of the RPM infringement;
- When these are used to enable detection of retailers deviating from a manufacturer's pricing recommendations, allowing manufacturers to retaliate against retailers that do not comply with pricing recommendations, that turns "recommended" prices into a fixed resale price (RPM) which is illegal; and
- When algorithmic price monitoring is relied upon by retailer A to monitor that retailer B adheres to an RPM, and A then follows B in adhering to the price, the RPM effect has spread.
In horizontal cases, the submission also identifies three scenarios where the use of algorithms gives rise to competition concerns:
- If the algorithm-enabled price monitoring is used to monitor prices already agreed between competitors, that simply forms part of the overall infringement, although it might lead to an increased fine;
- If it is used to implement pre-existing explicit collusion, it is no different from setting prices manually to implement a collusive price in the offline world, where competitors are clearly aware of the collusion; and
- Algorithm-enabled price monitoring can also be used as a means of communication to engage in explicit collusion, including through "hub and spoke" collusion and signalling.
The submission presented a future scenario where pricing algorithms could engage in explicit collusion with each other without explicit instructions to do so. For this to become possible, the algorithms would need to develop a sense of communication, for concerted practices, and bargaining and mutual commitment for reaching an 'agreement', as defined in EU competition law. However, if this becomes possible, the firms using such algorithms would remain liable for their behaviour, as it is up to the firms to ensure that their algorithms do not engage in illegal behaviour.
As the OECD Secretariat stressed in the submission, pricing algorithms when combined with 'big data' can also produce important efficiencies, which benefit firms and consumers through new, better products and services. It's therefore important that authorities approach enforcement with care. If algorithms are simply used as a tool for implementing cartelist conduct, it would not come as a surprise to any firm that this is illegal. However, aside from such obvious cases, one can easily see how it could place an additional compliance burden on firms if they have to extend their compliance duties to the effects of using very complex technologies, particularly when having used them in a bona fide way.
Robert Eriksson is a competition law expert at Pinsent Masons, the law firm behind Out-Law.com