Pricing Algorithms: How Should India Deal With It?

[Anik Bhaduri is a second-year B.A., LL.B (Hons.) student at NALSAR University of Law, Hyderabad]

The rapid development of the digital world often presents issues which could not have been contemplated before, thus presenting new challenges to the existing legal framework, which has to modify itself to deal with the new problems. The development of pricing algorithms presents one such issue. As self-learning algorithms used by different companies could artificially fix the market price, regulatory agencies across the world are faced with the task of dealing with this new challenge to market competition.

The digital economy depends greatly on the use of ‘big data’, which, according to Prof. Ezrachi, is characterized by the volume of the data, the velocity at which it is collected, the variety of information so collected and the value of the data. This is further aided by the development of ‘Big Analytics’, which refers to the ability to access and analyze the ‘Big Data’. Enterprises operating online often use the ‘Big Data’ to decide their policies.

A pricing algorithm refers to a set of instructions that are fed into a computer programme to enable it to constantly adjust and optimize prices through a trial and error method by finding patterns from the ‘Big Data’. The use of pricing algorithms by a company makes pricing efficient, dynamic and personalized. A number of online platforms like Amazon and Uber use pricing algorithms to automatically adjust prices. The algorithms collect and analyze a variety of personal and market data to optimize and decide the prices for the products available.

Pricing algorithms can, however, be used by enterprises to collude by exchanging market information and fixing high prices. Executives of different firms may agree upon such a plan and carry it out using specifically designed pricing algorithms. The algorithms thus become the tools using which the illegal agreement of the human agents are to be carried out.

The problem, however, is that the use of pricing algorithms by firms for tacit collusion makes it difficult to prove that the executives of the firm engaged in such a practice consciously. The use of pricing algorithms enables a firm to respond almost instantaneously to the price movements of its competitors. The computer analyzes a huge volume of a variety of data at such a speed that humans cannot replicate and it creates a perception that there is no human intervention in the price-setting.  There is no interaction between the executives of the firms, and the firms operate in the market through their own pricing algorithms, which automatically reach an understanding. Thus, although the firms engage in tacit collusion, there is nothing to show that the executives of the firm intended it, or even knew about it. According to the existing laws, the firms cannot be prosecuted unless there is a human element in the form of conscious decision-making. It thus becomes difficult for the regulating agencies to deal with such anticompetitive behavior on a digital platform and to enforce laws which prohibiting price parallelism.

The first successful prosecution of price-fixing using digital algorithms took place in the US in 2015 when David Topkins pleaded guilty to a charge of artificially fixing prices using pricing algorithms. According to the charge, Topkins and his co-conspirators used algorithms to fix the price of posters sold on Amazon and maintained what the prosecutors called “collusive, non-competitive prices”. Topkins was held liable under §1 of the Sherman Act (1890) and had to pay a fine of $20,000.

In this case, however, the pricing algorithms did not start the price parallelism by themselves and were used only as the tools for carrying out a pre-existing agreement between Topkins and other sellers. There was no doubt that the ‘meeting of minds’ had taken place between sellers and that the price parallelism carried out by the pricing algorithms was intended by the entrepreneurs.

The European Commission faced a question of digital collusion in the Eturas case. Thirty travel agents used an online platform named E-TURAS, developed by the company Eturas, to offer tours to their customers. In August 2012, Eturas sent a message to all the travel agents that the maximum discount they could offer was to be capped at 3% and subsequently the online platform underwent a change which limited the maximum discount at 3%. The Lithuanian Competition Council held that this was an instance of concerted practice and held the travel agents liable.

The travel agents appealed to the Court of Justice of the European Union (CJEU) claiming that they had not consented to capping the discount at 3% and some claimed that they had not even read the message. They contended that limiting the discount was a unilateral act by Eturas and the travel agents cannot be held liable for engaging in price parallelism. In its decision, the CJEU held that all agents except those who had explicitly refused to limit the discount were liable. The Court held that in such cases there shall be a rebuttable presumption that the parties knew of the price parallelism being carried out through the algorithms, unless there is sufficient evidence to prove the contrary. The Court clarified that such a presumption does not entirely shift the burden of proof on to the defendants and thus does not go against the concept of a fair trial.

In 2016, the UK prosecuted Trod and GB Eye for colluding in a manner similar to what David Topkins had done. These two firms had entered into an agreement not to undercut each other’s prices for certain products sold through Amazon and used a pricing algorithm to carry out this agreement.

In India, section 3 of the Competition Act, 2002 prohibits price fixing. The section prohibits any agreement between or “practice carried on, or decision taken by, any association of enterprises or association of persons, including cartels, engaged in identical or similar trade of goods or provision of services,” which determines the market price.

The definition of “agreement” in the statute is wide and does not require an explicit agreement between the firms to hold them liable for price parallelism. The use of pricing algorithms by two distinct enterprises with the knowledge that the algorithms would exchange information and fix prices is sufficient to constitute an “agreement” under the Act. The Competition Act is, thus, sufficient to deal with collusion by pricing algorithms, and India does not need new laws to deal with the issue.

The section, as interpreted by the courts, has led to the development of a body of law which makes it essential for the regulatory agency to prove the existence of an agreement between the enterprises in order to hold the enterprises liable for price parallelism, and mere similarity in pricing trends over a period of time is not sufficient to hold the enterprises liable for concerted practice. The existence such an agreement, however, need not be proven using direct evidence and may be deduced from circumstantial evidence, which may be “fragmentary and sparse.”

In case of algorithms automatically reaching an argument to optimize prices, however, there is no communication between the enterprises and even circumstantial evidence is difficult to obtain. Thus, it becomes almost impossible to prove the existence of a price-fixing agreement between the enterprises and the law turns out to be inadequate. The solution to this problem, at least for the time being, is to follow the path laid down by the CJEU in the Eturas – and to create a presumption that a similarity in the pricing patterns of two or more firms that use pricing algorithms would indicate that the firms knew of price fixing being carried out by the algorithms, and were thus engaged in concerted practice. Such a presumption by the Indian courts shall not be unjust to the defendants and may be rebutted by providing evidence to the contrary. Without this presumption, it is impossible for the regulating agency to prosecute firms engaging in price parallelism unless there is an improvement in technology and the regulators are able to keep a check on the working of the algorithms. In the long run, however, regulating agencies should aim at devising a way of monitoring the operation of pricing algorithms while states should introduce new data protection laws to ensure that the personal data of individuals is not available online.

A presumption that firms are engaging in concerted practice if their pricing algorithms show a similarity in pricing patterns increases the responsibility of the firms, who now have to keep a check on their self-learning pricing algorithms and ensure that the algorithm does not share information with the other firms online or reach an understanding with the algorithms of other firms to not undercut each other’s prices. This additional burden on the companies would mean that all algorithms are to be supervised and even artificial intelligence is to put under the control of individuals. While this reduces the efficiency with which the pricing of various products by a firm, it is essential in order to ensure that anti-competitive behavior is curbed. 

Thus, although the development of pricing algorithms has created a major challenge to competition law regimes across the world, leading to “the end of competition as we know it”, the Competition Act, 2002 seems sufficient to deal with a case of price fixing using algorithms. However, noting that it is almost impossible for the regulating agency to find the evidence of an agreement between firms when self-learning are used to fix the market price, it is imperative that the Indian courts follow the method adopted by the CJEU in dealing with such issues and presume that the executives of firms were aware of such collusion being carried out by firms. At the moment, such a presumption is the only way to deal with the problem and other ways of dealing with digital collusion are yet to be ascertained.

– Anik Bhaduri

About the author

Add comment

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Recent Posts

Topics

Recent Comments

Archives

web analytics