- Covid-19 Guidance New
- Editor's Preface
- League Tables New
- Ukrainian Legal Market
Practice Areas and Industries Review
- Aircraft and Airports Finance
- Alternative Dispute Resolution
- Anti-Money Laundering
- Banking Disputes
- Business Crime
- Business Immigration
- Business Protection
- Capital Markets
- Commodities Arbitration
- Competition Investigations
- Contract Law
- Corporate Disputes
- Corporate Governance
- Corporate Intelligence
- Counterfeiting and Piracy
- Criminal Process
- Cross-Border Debt Restructuring
- Currency Regulation
- Due Diligence
- Energy Efficiency
- Enforcement of Foreign Awards
- Family Law
- Financial Restructuring
- Financial Services
- Free Trade Agreements
- Government Relations
- International Arbitration
- International Civil Procedure
- International Tax
- IT Law
- Jurisdiction Issues in Commercial Procedure
- Labor & Employment
- Medicine & Healthcare
- Mergers & Acquisitions
- Non-Performing Loans
- Political Prosecution
- Private Clients
- Procedural Actions
- Project Finance
- Public-Private Partnerships
- Real Estate
- Renewable Energy
- Role of Experts in International Arbitration
- Secured Transactions
- State Aid
- Tax Controversy
- Trade Remedies
- Transfer Pricing
- Unfair Competition
- World Trade Organization
Who Is Who Rankings
- Antitrust and Competition
- Banking & Finance, Debt Restructuring
- Capital Markets
- Corporate and M&A
- Criminal Law/ White-Collar Crime
- Energy & Natural Resources
- Information Technologies, Telecommunications & Media
- Intellectual Property
- International Arbitration
- International Trade: Trade Remedies/WTO, Commodities, Commercial Contracts
- Labor & Employment
- Pharmaceuticals & Healthcare
- Private Clients: Wealth Management, Family Law
- Real Estate, Construction, Land
- Tax and Transfer Pricing
- Transport: Aviation, Maritime & Shipping
- Law Firms Profiles
- Lawyers Profiles
Managing Partner, CLACIS
Algorithmic Pricing Collusion: When does Liability Arise?
The application of pricing algorithms is now widespread in online industries and in traditional businesses. In general terms, a pricing algorithm is a program that monitors and/or sets prices. This can be a program software possessed and applied unilaterally by a business entity, an outsourced software or a simple price comparison website used to monitor the competitive environment, determine pricing policy, plan commercial activities, monitor maximum resale prices, etc.
Though pricing algorithms improve dynamic pricing and the matching of demand & supply, which will result in lower prices and the market functioning more efficiently, such software can significantly increase the risk of a collusive outcome on the market and has already attracted the attention of antitrust authorities around the world.
In this light, after successfully prosecuting cases of collusion facilitated by pricing algorithms, antitrust authorities have paid significant attention to the question of liability of business entities using pricing algorithms, especially in cases of potential collusion involving self-learning autonomous algorithms in the absence of evidence of an agreement.
Antitrust Risks of Use of Pricing Algorithms
Competition authorities believe that pricing algorithms can distort competition by increasing market transparency, simplifying communication between market players and detecting deviations from collusive agreements. There are 4 main competition harm scenarios:
– Messenger: algorithms monitor prices and can facilitate maintenance of a cartel by sending messages to participants;
– Hub-and-spoke: cases where companies agree to use the same pricing algorithm to facilitate collusive agreement;
– Predictable agent: cases where entities unilaterally choose algorithms which monitor the pricing behaviour of rivals and react to it in a certain way, and which can potentially be means of tacit collusion;
– Digital eye: cases where companies unilaterally use self-learning sophisticated algorithms which, being programmed, for example for profit maximisation, can autonomously collude on higher prices.
Liability for “Hub-and-Spoke” and “Messenger” Collusion Scenarios
It is generally recognised that overt collusion is expressly prohibited by competition laws and collusion facilitated by using pricing algorithms is treated like other concerted actions. Considering that price-fixing agreements are restrictions by object, liability for them is presumed per se. Antitrust authorities had earlier successfully prosecuted cases of both horizontal and vertical collusion facilitated by pricing algorithms. For example, in 2015 the US District Court of Northern California held David Topkins, a director of a company selling posters online, liable for agreement with other merchants on levels of prices and specific algorithms to be used. Later, in 2016 the same court found Trod Limited and its director liable for a similar infringement. This case triggered similar issues in the UK where the CMA found liable two merchants selling on Amazon for an agreement not to compete on prices and to adjust the settings of a re-pricing algorithm available on Amazon. A messenger scenario was considered by CJEU in the E-Turas case, where the Court of Justice of the European Union confirmed that though travel agents did not formally respond to the messages, the fact that they were aware of such messages and continued to use the system such entities will be liable for being part of a price-fixing cartel.
Certain exemptions can be made for monitoring of recommended and maximum resale prices, but if monitoring of resale prices results in fixing them, such arrangement shall be considered as prohibited by object. In 2018 the European Commission held Asus, Denon&Marantz, Philips and Pioneer liable for resale price maintenance, which was facilitated by the use of price-comparison websites and special pricing program which helped producers to trace the prices of resellers, detect deviation and maintain the level of retail prices.
At the same time, in view of the increasing use of pricing algorithms, certain non-EU jurisdictions consider amending their antitrust legislation to treat the use of pricing algorithms as an aggravating factor in cases of administrative liability of officials for anticompetitive concerted actions and abuse of dominance. This approach can be explained by the fact that the use of pricing algorithms significantly simplifies collusion, which in the absence of algorithms would much more complicated to engage in.
Liability for “Predictable Agent” and “Digital Eye” Cases
The possession and unilateral use of pricing algorithms shall not be considered an antitrust infringement. Competition law of most jurisdictions provides that business entities can be held liable for anti-competitive concerted actions only when there is evidence of an agreement to collude. However, under ’predictable agent’ and ’digital eye’ scenarios pricing algorithms can distort competition without any explicit agreement to do so between their users.
Yet, antitrust authorities emphasise that companies shall be liable if their pricing algorithms autonomously collude on prices. Parallel pricing is not itself prohibited by antitrust laws and is considered illegal only if it cannot be explained by any reason other than collusion. Although prosecuting parallel actions is challenging and is not yet widespread in the EU, it does not mean that it is not possible. In Ukraine the practice of prosecuting parallel actions is already established.
To hold entities liable in the event of ’predictable agent’ and ’digital eye’ scenarios, antitrust agencies will have to link the commercial decisions of entities to algorithmic prices, which will be challenging in the absence of explicit evidence of collusion. To do so antitrust agencies will have to investigate longer periods preceding the alleged collusion to track historical functioning of the algorithm so as to identify whether the price alignment is a result of self-learning or manual adjustments.
German and French antitrust authorities noted that antitrust authorities already have at their disposal the means to efficiently investigate price alignment by self-learning autonomous algorithms, namely, information requests, dawn raids and interviews. For example, they can request business entities to provide a description of implementing principles, explanation of inputs and outputs, usage patterns of the algorithm, frequency of learning, recalibration or manual adjustments, etc. Also, an antitrust authority can ask for internal documents like the specifications for algorithms, user manuals or code used in the development phase.
It is already known that self-learning algorithms must include a code that reveals an intent to collude. The antitrust authorities might investigate the functioning of an algorithm to reveal such code. The authorities can also request a source code of the algorithm to approximate or recreate the algorithm in controlled conditions to understand whether the algorithm was programmed to collude. Taking into account the fact that such a measure would require significant technological resources and competences, it is expected that such measures will be applied as a measure of last resort.
It means that unlike in the ’messenger’ or ’hub-and-spoke’ scenarios, where the liability of business entities arises from the fact of direct collusion, the issue of liability of business entities in case of autonomous collusion by software is still an open question. Though there is still no case of liability of business entities for the autonomous actions of algorithms, antitrust authorities send out warnings to companies that they can be found to have breached competition laws when prices are aligned between competitors by independently using self-learning algorithms, which would be treated as the actions of any human employee.
Needless to say, Ukraine is part of global business and its consumers use both global and local sales platforms. Ukrainian competition law also has all the same basic concepts for prosecution of concerted actions; hence businesses using pricing algorithms shall assess their compliance with such local competition rules, as with time the competition authority may well focus on analysing such matters.
Oxera, ’When Algorithms Set Prices: Winners and Losers’ (2017), Discussion Paper 19 June 2017, 2, <https://www.oxera.com/wp-content/uploads/2018/07/When-algorithms-set-prices-winners-and-losers.pdf.pdf.> assessed 10 February 2020.
Ariel Ezrachi and Maurice E. Stucke, ’Artificial Intelligence & Collusion: When Computers Inhibit Competition’  University of Illinois Law Review 1775,1782-84.
US v Topkins .
U.S. v. Daniel William Aston and Trod Limited .
Trod Ltd/GB Eye Ltd (Case 50223) CMA’s Decision as of 12 August 2016.
C-74/14 “Eturas” UAB and Others v Lietuvos Respublikos konkurencijos taryba  ECLI:EU:C:2016:42.
See for example Asus (Case No 40465) Commission Decision C (2018) 4773 as of 24 August 2018.
Bundeskartellamt and Autorité de la Concurrence, ’Algorithms and Competition’ (2019), 59, < https://www.autoritedelaconcurrence.fr/sites/default/files/algorithms-and-competition.pdf>, assessed 10 February 2020.
A. Ezrachi & M. E. Stucke,’Algorithmic Collusion: Problems and Counter-Measures’ (OECD Roundtable on Algorithms and Collusion 21-23 June 2017) para 71, <https://www.oecd.org/officialdocuments/publicdisplaydocumentpdf/?cote=DAF/COMP/WD%282017%2925&docLanguage=En> , assessed 10 February 2020; see also Article 6 of the Law of Ukraine “On Protection of Economic Competition” No № 2210-III as of 11 January 2001.
Case 48/69 ICI v Commission  ECLI:EU:C:1972:70, para 66.
Monopolkommission, ’Shaping Competition Policy in the Era of Digitisation’ (2018), para 29 < https://ec.europa.eu/competition/information/digitisation_2018/contributions/monopolkomission.pdf> assessed 10 February 2020.
Ibid, para 28.
Bundeskartellamt and Autorité de la Concurrence (no 9), 69.
Charley Connor, ’When robots collude’, 27 September 2019 <https://globalcompetitionreview.com/insight/gcr-q3-2019/1202826/when-robots-collude>, assessed 10 February 2020.
Ibid, see also Bundeskartellamt and Autorité de la Concurrence (no 9), 67.
Charley Connor (no 17)