Algorithmic collusion raises several challenges in the (online) retail sector. Are surveillance and pricing algorithms compatible with competition law or should they go to law school?
As in almost every field of society, competition law is not immune to the challenges posed by the digital economy. The rise of powerful digital platforms (e.g., Alphabet, Apple, Amazon, ByteDance, Meta and Microsoft) and the increasingly widespread use of monitoring and pricing algorithms have stretched the traditional boundaries of competition law and forced competition watchdogs to think beyond traditional models and adapt in the face of these challenging phenomena.
But what are algorithms and why are they increasingly attracting attention from a competition law viewpoint?
In broad terms, an algorithm is an unambiguous, precise list of simple operations that transform an input into an output. Algorithms can be classified according to the specific function they perform (allocation, search, recommendation, surveillance/monitoring, pricing). Pricing algorithms, for instance, essentially set or recommend prices using data on observable customer characteristics or market conditions, whereas surveillance algorithms monitor market conditions to track the behaviour and strategic decisions of competitors, such as prices (e.g., Wiser Solutions).
The use of algorithms is not problematic in and of itself. It is generally recognised that they may be pro-competitive and efficiency-enhancing as they can contribute to the creation of new and better products, substantially lower production costs, lower search costs and achieve a better balance between supply and demand.
However, the ever-increasing number of commercial decisions (in particular, pricing decisions), that are delegated to algorithms may raise serious concerns from a competition law perspective. Several studies suggest that the use of self-learning algorithms (namely, reinforcement learning algorithms) to monitor and set prices substantially increases the likelihood of collusion. Notably, these algorithms may autonomously learn to reach collusive outcomes without being instructed or programmed to do so and in the absence of explicit communication (autonomous algorithmic collusion).
In fact, as the European Union (EU) Commissioner for Competition Margrethe Vestager stressed, “it is a hypothesis that not all algorithms will have been to law school. So maybe there is a few out there who may get the idea that they should collude with another algorithm who haven’t been to law school either”.
Aware of this dual nature of algorithms, the European Commission (EC) and national competition authorities have conducted several market surveys to understand how prevalent pricing algorithms are in the internal market and published numerous policy papers regarding the relationship between algorithms and competition. Some noteworthy examples include the United Kingdom (Competition and Markets Authority, 2021), the Netherlands (Authority for Consumers and Markets, 2020) and a joint paper by France and Germany (Autorité de la concurrence & Bundeskartellamt, 2019).
Algorithms have been increasingly used in Portugal in several sectors of the economy and the COVID-19 pandemic has been an important catalyser for the digitalisation of the national economy and the usage of algorithms has probably increased since then.
The Portuguese Competition Authority (PCA) has published two policy papers addressing the use of monitoring and pricing algorithms in Portugal. In 2019, the PCA published its Digital Ecosystems, Big Data and Algorithms Issues Paper(Issues Paper), which addressed the main challenges that the digital transition raises for competition policy. In this document, the PCA emphasised that firms are responsible for the algorithms they use and that the use of algorithms as a conscious and deliberate strategy to fix prices is incompatible with EU and national competition law. In the same vein, the EU reinforced that firms cannot avoid liability for illegal pricing practices on the grounds that the prices were determined by algorithms (Note to the OECD regarding algorithms and collusion).
Following the adoption of its Issues Paper, the PCA set up a special Task Force for the digital sector, focused in investigating complaints and detecting potential anticompetitive behaviour in digital markets, and included in its Competition Policy Priorities the investigation of collusion and abuse on digital sectors. In fact, the PCA’s digital task force has been actively investigating complaints and indicia of anticompetitive behaviour in digital markets. For example, in May 2022, the PCA opened an investigation into Google´s alleged abuse of dominance in online advertising, in Portugal, which was, subsequently, called back by the EC in the context of an ongoing investigation at EU-level.
Furthermore, in 2022 the PCA issued its Defence of Competition in the Digital Sector in Portugalpolicy brief, providing a policy update for digital markets, as well as a brief overview of the PCA’s recent enforcement initiatives. According to the survey conducted by the PCA in the online retail of electronic products and household appliances, 21% of the inquired firms acknowledged using price surveillance algorithms and only 12% confirmed using price definition algorithms for some of its products. In addition, some stakeholders (e.g., ANACOM) emphasised that they expect the use of pricing algorithms to increase in the future.
The PCA has hinted that it will pay particular attention to cases where competing firms use common algorithms or data pools to coordinate their pricing strategies (hub-and-spoke scenarios). Beyond the challenges posed by autonomous algorithmic collusion and hub-and-spoke-like scenarios, price surveillance algorithms may play an instrumental role in the implementation of existing anticompetitive agreements. For example, the EC has recently fined four electronic products manufacturers (Asus, Denon & Marantz, Philips and Pioneer) for allegedly fixing online resale prices. According to the Commission, the manufacturers used sophisticated monitoring tools to track the resale prices in their distribution networks, which allowed them to swiftly intervene in cases of price changes.
In sum, firms must be aware of the several challenges that algorithmic pricing raises from a competition law standpoint. On the one hand, pricing algorithms may create appreciable efficiencies and, thus, be pro-competitive. On the other hand, the use of surveillance and pricing algorithms may constitute an infringement of competition rules and lead to the imposition of hefty fines. All in all, the national competition authorities (including the PCA), as well as the EC, are closely monitoring the use of surveillance and pricing algorithms in the retail sector. Hence, companies should not discard a case-by-case analysis of the use of algorithms when designing and implementing algorithms in their commercial and pricing strategies.
#Tech_Hub | Tech Hub by VdA - A Global Digital Impact