A pricing algorithm can be defined as a computer procedure which basing on a set of rules solves certain pricing tasks. Pricing algorithms can be of different sophistication levels: from simple ‘match the lowest price’ to self-learning algorithms with objectives such as profit maximising to be set by user . Such programs allow rapidly response to market changes, adjust prices and therefore optimise business rules . Pricing algorithms are widely used by online merchants such as airlines, hotels, travel agencies as well as within different online market places such as Amazon. However, pricing algorithms are used not only in online markets but also offline. For example, such programs can be used by supermarket chains by marking products with electronic tags allowing such retail networks change prices more quickly and frequently .
Use of pricing algorithms is not itself prohibited. However, for the last several years the widespread use of pricing algorithms has attracted attention of antitrust authorities around the world. The main concern is that use of pricing algorithms can facilitate or create collusion between entities , which is explicitly prohibited by Article 6 of the Law of Ukraine On Protection of Economic Competition, EU and US antitrust laws as well as antitrust laws of other jurisdictions. This article will provide an overview of such issues including explicit and tacit collusion.
Pricing algorithms facilitating explicit agreements
In the recent years antitrust authorities have paid significant attention to the use of pricing algorithms to facilitate overt collusion. This means cases where business entities after having explicitly agreed upon level of their prices also agree the use of certain price algorithms which will monitor and adjust their prices or even when businesses use such algorithms to collude . The use of pricing algorithms can facilitate both horizontal and vertical collusion, for example, by simplifying detection of deviations, reducing chances of errors as well as reducing risks of agency slack .
The first landmark case on pricing algorithms facilitating collusion was Poster Cartel case decided by the US district court of Northern California in 2015 . In this case David Topkins, director of a company selling posters online, was held liable for horizontal price fixing with other merchants on Amazon platform . Having agreed with other merchants on the levels of prices and specific algorithms to be used Mr. Topkins wrote a code for his company’s algorithm to set prices on the posters as they were agreed with other merchants . Later, in 2016 the same court found Trod Limited and its director Daniel Aston liable for a similar infringement . The latter case triggered analogical Trod Ltd/GB Eye Ltd case in the UK where GB Eye Ltd submitted a leniency application to the UK Competition and Markets Authority acknowledging that it had agreed with Trod Ltd its prices in the UK . Both merchants were using the repricing algorithm available on Amazon which is to be adjusted by compete rules determined by each particular merchant . Such rules include, for example, decrease of price for x% for competing goods . The repricer allows to exclude prices of certain merchants from the algorithms by adding such merchants to the ignoring list . Thus, the two merchants had agreed between themselves not to compete on prices and put each other in the ignoring list so that they do not undercut each other prices which resulted in the price-fixing cartel.
On the EU level the issue of pricing algorithms was considered by the Court of Justice of the EU in the E-TURAS case . E-TURAS, an online travel booking system, sent messages to its travel agents through the online system announcing technical restrictions to its pricing algorithm capping discounts at 3% . The Court of Justice confirmed that even though travel agencies did not formally respond to the message the fact that they were aware of such message, did not distance themselves from it and have subsequently continued to use the system, such agencies shall be liable for the price-fixing cartel .
In 2018 the EU Commission also emphasised that pricing algorithms can also facilitate vertical price fixing, in particular, in maintaining resale prices. Thus, Asus, Denon & Marantz, Philips and Pioneer were fined for resale price maintenance which was facilitated by use of price comparison web-sites (which in their turn are based on pricing algorithms) and special pricing program which helped producer to trace prices of online retailers, detect deviation and maintain a certain level of retail prices .
Pricing algorithms as a means of tacit collusion
It is considered that use of pricing algorithms can also lead to tacit collusion between entities i.e. concertation without any agreement to collude or use of pricing algorithms for this purpose . The main reason for such concerns is that extensive use of pricing algorithms can lead to increased market transparency, speed of price changes and calculation of optimal prices, which together create favourable market conditions for collusion . It is considered that tacit collusion through algorithms can happen in the following scenarios :
- Hub and spoke. Use of the same pricing algorithm by multiple market players can lead to similar reaction to market developments and as a result similar pricing pattern. Moreover, if such market players are aware of algorithms used by their competitors such use can lead to indirect exchange of information and coordination of pricing policies. The most sensitive hub and spoke scenario is when market players entrust their pricing policy to the same provider of algorithmic pricing services, which can lead to coordination by such agent without knowledge of market players just through collection of data and application of same pricing algorithm .
- Predictable agent. Use of simple pricing algorithms which react to market conditions in a certain predictable way, for example, ‘lowest price matching’ can lead to very transparent and predictable pricing. This can result into tacit collusion and parallel pricing .
- Autonomous algorithms. Where algorithms are sophisticated enough to learn by themselves, they can tacitly coordinate prices. Thus, if person instructs the algorithm with a profit maximizing objective, such algorithm can through ‘trial and error’ rule on its own figure out that the most profit mixing pricing is to align prices with those of competitors . For such situations, big data is an important factor. The more data is available to an algorithm, the better results can be obtained. Data used by pricing algorithms can include, in particular, prices of competitors, past pricing/profit/revenue data; individual customer information; market information such as competitors’ stock; external information such as weather patterns; or firms’ costs, such as production, storage and fulfilment .
However, the mere fact that market players use the same algorithms is not enough to establish a fact of collusion and so far, there is no examples of such collusion. In order to prove tacit collusion, there shall be an intention to collude, which is highly unlikely to be proved without direct communication . Moreover, most algorithms can be changed by users and its mostly impossible for market players to have the same algorithm settings without agreeing them . Therefore, any announcements or communication of name and particularities of can lead to establishing the fact of collusion .
In the meantime, EU Commissioner Vestager noted that the fact that entities were not aware of collusive consequences of algorithms they used would not let businesses get away from sanctions . Instead, she noted that businesses have an obligation to ensure that algorithms function properly and do not anyhow violate antitrust laws by design . It appears that where a company applies an algorithm, it shall be liable for any antitrust consequences of use of such algorithm . Moreover, in other jurisdictions outside the EU legislative bodies are already working on legislative definition of a pricing algorithm and direct prohibition of collusion through use pricing algorithms .
It is worth mentioning here that use of pricing algorithms can also lead to issues of abuse of dominance and personalised pricing.
Personalised pricing and pricing algorithms
Pricing algorithms can also be instruments of personalised pricing – practice when businesses can use available data on customers to charge them different prices . As of now there is no evidence of personalised pricing, but big data and sophisticated pricing algorithms can contribute to such practice . Personalised pricing can be an issue in case of explicit coordination between entities to engage in personalised pricing and even lead to exploitation of joint dominant position by means of discrimination between customers .
Dominance of online market platforms
In some jurisdictions antitrust authorities are also concerned with the role of online market platforms and their possession of big data and algorithms used by merchants. It is considered that online market platforms can be dominant if network effects that arise from the functioning of such platforms and respectively data possessed by such platforms can significantly affect conditions of marketing of products sold on such platforms or squeeze out merchants from the market .
 Competition and Markets Authority, ‘Pricing Algorithms: Economic working paper on the use of algorithms to facilitate collusion and personalised pricing’ CMA94 8 October 2018 <https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/746353/Algorithms_econ_report.pdf> assessed 22 January 2019, paras 2.7-2.9.
 CMA (n 1) para 2.7.
 CMA (n 1) para 3.11.
 CMA (n 1) paras 5.1-5.2.
 CMA (n 1) para 5.4.
 CMA (n 1) para 5.6.
 US v Topkins  <www.justice.gov/atr/case-document/file/628891/download> assessed 22 July 2019.
 U.S. v. Daniel William Aston and Trod Limited  <www.justice.gov/atr/file/840016/download> assessed 22 January 2019.
 Trod Ltd/GB Eye Ltd (Case 50223)  <https://assets.publishing.service.gov.uk/media/57ee7c2740f0b606dc000018/case-50223-final-non-confidential-infringement-decision.pdf> assessed 22 January 2019.
 Ibid.para 3.85 – 3.87.
 C-74/14 "Eturas" UAB and Others v Lietuvos Respublikos konkurencijos taryba  ECLI:EU:C:2016:42 < http://curia.europa.eu/juris/liste.jsf?&num=C-74/14> assessed 22 January 2019.
 Ibid paras 44-47.
 See for example Asus (Case No 40465)  <http://ec.europa.eu/competition/antitrust/cases/dec_docs/40465/40465_337_3.pdf> assessed 22 January 2019.
 CMA (n 1) para 5.17
 CMA (n 1) paras 5.25-5.29.
 CMA (n 1) para 5.15; Ariel Ezrachi and Maurice Stucke, ‘Artificial Intelligence & Collusion: When Computers Inhibit Collusion’ (2015).
 CMA (n 1) para 5.17.
 ibid para 5.22.
 ibid para 5.24.
 Ibid para 2.23.
 Ibid para 5.18; Arnold & Porter, ‘Pricing algorithms: Antitrust implications’ < https://www.arnoldporter.com/en/perspectives/publications/2018/04/pricing-algorithms-the-antitrust-implications> assess 22 January 2019
 CMA (n 1) para 5.18.
 Freshfields Bruckhaus Deringer, ‘Pricing algorithms: the digital collusion scenarios’ < https://www.freshfields.com/globalassets/our-thinking/campaigns/digital/mediainternet/pdf/freshfields-digital---pricing-algorithms---the-digital-collusion-scenarios.pdf> assessed 22 January 2019.
 Draft amendment to the Law on Protection of Competition < https://www.eg-online.ru/document/law/370205/> assessed 22 January 2019.
 СMA (no 1) para 7.2.
 СMA (no 1) para 7.12.
 СMA (no 1) paras 7.31-7.34.
 Draft (no 32)