Algorithmic Price-Setting by Platforms

A recent criminal cartel case where sellers of posters used an algorithm to coordinate their prices on the Amazon platform created buzz in the competition community. While interesting with regard to the use of technology, in the end it was a “normal” horizontal case: competitors agreed on the price and the US and the UK punished them.

Based on a recent paper, this post explores questions arising if prices are not set by the sellers on a platform but by a platform’s algorithm. In other words: what if Amazon would use an algorithm to set uniform prices for its sellers? On a more practical level: how do we treat the setting of the prices/fares by ride sharing platforms (e.g. the Uber case, in the end not decided).

Challenging the established competitive order

Platforms directly connect consumers with suppliers of goods or services e.g. drivers. For this connection service, platforms receive a commission (often a % of the transaction value). These platforms have been highly innovative using the internet and mobile devices to disrupt traditional industries such as the taxi business. The price, as in every market, serves as an important signal and helps in balancing demand and supply (e.g. higher prices on the platform increase the number of suppliers when consumer demand is high). Thus, prices help platforms to balance supply and demand and they might set them algorithmically. Yet, some economic research suggests that platforms have the same interest in supra-competitive prices as suppliers, at least if the revenue model is a commission based.

Challenging the establish legal frameworks

Algorithmic price-setting by platforms raises a number of questions for competition law which center on the seller-platform relationship.

Competition law would not apply to the seller-platform relationship, if suppliers were the platform’s employees or the platform and their suppliers are one single economic entity. Platforms often argue that their suppliers are independent contractors/suppliers and not employees. If that argument were accepted, competition law would likely apply, because (at least) tworather than oneeconomic entities exist: independent suppliers and the platform do not perform the same economic activity (provision of goods/services vs. connecting buyers-sellers) and typically do not share each other’s business risks. 

Amongst the traditional legal frameworks, resale price maintenance (RPM) in a vertical relationship or a brokerage/hub-n-spoke situation seem most similar to situation of platforms setting their seller’s price. While some traditional justifications for RPM - preventing free-riding, ensuring retailers stock sufficient quantities- are absent, another -facilitating entry- seems to apply during the market entry phase of platforms. Yet, platforms are like marketplaces or real-estate brokers who are not producing or reselling any good/services but rather connect buyers and sellers. Thus, a comparison to brokerage/hub-and-spoke situations seems sensible. In such situations, brokers (the hub) coordinate a horizontal cartel amongst the sellers (the spokes) to increase prices charged to consumers. Yet, hub-and-spoke structures are typically initiated by the sellers and not, as here, by the brokers setting the price independentlyand via algorithms.

Challenging Enforcement Questions

Algorithmic price-setting by platforms challenges traditional antitrust thinking and thus gives rise to new policy and legal questions. An enforcement agency or court would have to think hard about whether the seller-platform relationship is classified as vertical, horizontal or some third, altogether different, kind of relationship. What standards of proof should apply in this context: should effects on competition be presumed or is it for the claimant to prove those? How closely should claims that price-setting by the platform is necessary be scrutinised? How should the two-sided-nature of the market and the broader benefits (often producing substantial competitive pressure in the market) be included in the analysis? Moreover, strategically one might ask: should competition agencies engage with these problems and take the lead, or should they let them play out in the courts first?

In the end, the fundamental questions are: do we trust the invisible digital hand, an algorithm programmed by the platform? Do we except a second best outcome (the loss of competition between the sellers) because it provides cheaper prices than the third best outcome (more competition exists with the platform than without it)? And, is it really a black or white choice between second and third best? 

Julian Nowag is Associate Professor at the Faculty of Law, Lund University, and an Associate at the CCLP Oxford.