Regulating Tacit Algorithmic Collusion: Meaning, Scope and its Challenges

PIc Credits-


Development in technology brings with it, both opportunities as well as challenges. With development in the Artificial Intelligence (AI), the use of algorithms in making business decisions is on the rise. While the algorithms are used to predict consumer preferences, it is a challenge when such algorithms are used for tacit collusion to co-ordinate their prices. This article focuses on whether this challenge can be effectively dealt with by the current competition law regime, and if not, what are the potential changes that it should undergo in order to deal with the aforesaid challenge effectively.

Understanding ‘Tacit Algorithmic Collusion’

To understand the meaning of tacit algorithmic collusion, it is imperative to understand the meaning of Tacit Collusion. Tacit collusion is said to occur when two companies collude without any written agreement or formal communication.[1] It is difficult to perceive whether the companies are colluding or just engaging in parallel pricing. There is a thin line of difference between tacit collusion and price parallelism. Price Parallelism is when two firms have the same price or price fluctuations without any intent to collude, whereas tacit collusion is when they consciously have the same price or price fluctuations with the intent to collude.[2] There have been many debates on whether such tacit collusions should be held anti-competitive or not.[3] In India, if it is more or less shown that the intent of the companies is to collude, tacit collusion is established, and the same is hence deemed as anti-competitive, falling under Section 3(3) of the Competition Act, 2002.

Coming to the definition of tacit algorithmic collusion and its regulation, an algorithm is simply a decision-making process that takes into account various factors and inputs given to it. Similarly, a pricing algorithm is decision-making process to arrive at the price of a particular product or service using various decision parameters.[4] Adding on, a pricing algorithm either uses price as an input value to arrive at the output, or, uses a set of values as inputs, to arrive at the output – which is the price. In this regard, the forthcoming sections would focus on the use of pricing algorithm to arrive at price as an output.[5]

In 2018, a complaint was filed before the Competition Commission of India (CCI) alleging that Uber and Ola cab service aggregators are engaging in algorithmic collusion for fixing the price of their service,[6] which was again reiterated by the NCLAT in a recent judgment.[7] The CCI, in this case, held that the prices cannot be said to be fixed as they were being decided by an algorithm, thereby meaning that algorithmically decided prices cannot collude. However, with the development in the Artificial Intelligence (AI), the ability of an algorithm to tacitly or explicitly collude comes into question.[8]

The observation made by the CCI does not seem to be entirely correct. For instance, Ai Deng[9] gave a perfect example of how such tacit algorithmic collusion may occur in any communication; in a case where two online sellers are selling a homogenous product, an algorithm can be designed as follows:

“First, I would raise my price until you also change your price. If you do not change your price in response to my increased price, my algorithm will decrease the price at the cost of the product, or even below cost. This will continue till both the sellers do not come to cooperation…”

In the above example, if the second seller does not raise his price; the first seller would decrease his price below cost to capture the market which will induce the second seller to collude by increasing the price. In such a case, the sellers would come to cooperation and will subsequently collude merely by making changes in their algorithms.

The Ability of an Algorithm to Tacitly Collude

The question that may arise now is whether there is an algorithm which can tacitly collude, and whether the algorithm colludes on its own. There is a mixed opinion for this. Yes, there have been such algorithms developed at an experimental stage which can tacitly collude but no such algorithms have been made functional in real market conditions.

One such algorithm is called the tit-for-tat (TFT) algorithm,[10] which is at an experimental stage. In TFT, both the parties start at the stage of cooperation and the algorithm will copy the same action taken by the other party in any previous interaction. Hence, the algorithm itself depends on the cooperation of both parties. Hence, if Firm A drops its price to Rs. 150, Firm B will do the same. Despite the TFT being appealing, it has a few shortcomings. For instance, for a TFT to work, the information of competitor’s actions is required, so that the algorithm can copy the competitor’s behavior. It is not always possible to have such access to information[11].

Tacit Algorithmic Collusion: Can it be regulated under Indian Competition Law?


Current Framework under the Indian Competition Law

Tacit Collusion was covered under the The Monopolistic and Restrictive Trade Practices Act and now is also covered under the current competition law in India, by way of Section 3(3). An instance for the same is the definition of an ‘agreement’, under Section 2(b) of the Competition Act, 2002 (“Competition Act”), which covers ‘tacit agreement’:

  1. Whether or not, such arrangement, understanding or action is formal or in writing; or
  2. Whether or not, such arrangement, understanding or action is intended to be enforceable by legal proceedings.

Dealing with Tacit Collusion is difficult because tacit agreement cannot be always proved, due to lack of evidence. The question that thus arises is: how have the Courts dealt with cases of Tacit Collusion and how have they established the presence of such Tacit Agreement?

There are two pre-requisites to establish Tacit Collusion; the first of which is that there should be reasonable proof that there was indeed an agreement, and the second being that agreement amounts to collusive behavior[12]. It however becomes very difficult to determine whether there was an agreement. In some of such scenarios, Courts have termed it Tacit Collusion only on the basis of price parallelism, and on the other times, in absence of any solid proof, the Courts have refrained from terming it Tacit Collusion. For instance, in the case of Ghai Enterprise Ltd and Quality Ice creams,[13] MRTPC had linked a tacit agreement with price parallelism. Both the ice cream companies, who held 80% percent of the market share, introduced similar discounts and the kept the prices identical. The MRTPC observed that the ‘preponderance of the probabilities’ in this case, leads to the conclusion that there was a concerted effort between the companies and hence, tacit collusion.

However, in the subsequent case of Re: Domestic Airlines,[14] it was observed that the airlines had started charging exorbitantly , following the strike by Air India’s pilots. The DG submitted that all the airline companies had engaged in parallel behavior to raise prices, thereby tacitly colluding. However, the Commission denied any presence of such agreement as there was no material available. It is pertinent to note, that there was a dissenting opinion by Hon’ble Member, Mr. R. Prasad, who suggested that an inquiry under Section 26(8) should be made to look into the parallel behavior.[15]

Position of EU

The landmark judgment of European Commission in dealing tacit collusion in an online platform is the Eturas Case.[16] Although it did not involve algorithmic collusion, the findings of the Commission in establishing tacit collusion on an online platform can also be applied in the cases involving algorithmic collusion as the primary issue is the same: proving tacit assent.

In this case, there was an online information system known as E-TURAS, which was an online travel booking system used by the travel agencies. The E-TURAS system introduced a limit in the discount rate at 3%, following which no travel agency using the system could offer a discount more than 3%. Before the introduction of this discount rate, two messages were sent to all the travel agencies. In the first message, the notion of capping the discount rate was put forward before the agencies. In the second message, it was informed to the travel agencies that the discount rate has been officially capped between 1-3%.

The Lithuanian Competitive Council filed a case against them under Article 101(1) of the TFEU which prohibits cartels and other anti-competitive agreements. Many travel agencies pleaded that they had no knowledge about both the messages and they did not collude. To solve this issue, the Lithuanian Supreme Court referred two questions to the European Commission:

  1. Should it be presumed that all the travel agencies were aware of the discount rate and gave their tacit approval, because they received a message from the common informational system, i.e., E-TURAS?
  2. If such presumption cannot be made, what factors should be taken into account to arrive that there was tacit collusion?

The European Commission observed two main points: First that it can be presumed that the travel agencies tacitly assented to the message circulated by E-TURAS in cases wherein such travel agencies have not ‘publicly distanced’ themselves, such as, by offering discounts higher than 3% or taken any other steps which rebuts the presumption of a concerted practice, and second that in order to establish a concerted practice (i.e. mutually agreed decision),[17] the subsequent conduct of the agencies must be seen. Mere dispatch of the message will not amount to tacit assent or collusion.

Overcoming the challenges in the existing framework

There has been no unified approach in dealing with tacit collusion. This has amounted to conflicting opinions and judgments, as discussed above. There can be three additions in determining the existence of tacit collusion: (a) To look for economic factors,[18] such as price fluctuations and the firm’s behavior to prove the presence of a collusion through a market study; (b) To look for plus factors[19] – meaning the parallel conduct engaged by oligopolistic firms which forms a part of large coordinated action (as was witnessed in Re: Domestic Airlines); (c) As propounded by the EU in the Eturas case, conduct of the firms in question can be taken into account to see whether the firm is colluding or not.


There are growing concerns surrounding the development of Artificial Intelligence and its effect on collusion. The first step CCI can take is to acknowledge the ability and the use of algorithms to engage in tacit collusion. Largely, the Indian Courts relied on the ‘preponderance of probabilities’[20] while deciding whether there was an agreement. They have not considered the economics of the market while establishing an ‘agreement’. The existing framework cannot be applied in a case of algorithmic collusion, as it is vague and inconclusive. The CCI will have to look for economic factor by assessing the firm’s behavior in the market combined with the study of the functioning of the algorithm to establish whether there was a tacit agreement.

(This post has been authored by Priyanshi Joshi, a Vth Year Student at Institute of Law, Nirma University)



  1. Directorate for Financial and Enterprise Affairs Competition Committee, Algorithms and Collusion – Background Note by the Secretariat, OECD, DAF/COMP(2017)4, (9th June, 2017), available at:
  2. Ibid.
  3. Donald F. Turner, The Definition of Agreement under the Sherman Act: Conscious Parallelism and Refusal to Deal, 75 Harvard Law Review 655 (1962); Richard Posner, Conscious Parallelism and Price Fixing: Defining the Boundary, The University of Chicago Law Review, 52(2):508-535.
  4. Pricing Algorithms, Economic Working Paper on the use of algorithms to facilitate collusion and personalized pricing, Competition and Markets Authority, available at:
  5. Ibid.
  6. Samir Agrawal v. ANI Technologies Pvt Ltd, Case No. 37 of 2018 (CCI), .
  7. Samir Agrawal v. CCI, Competition Appeal No. 11 of 2019 (NCLAT), available at:
  8. See, Basu Chandola, Algorithms and Collusion: Has the CCI got it wrong?, Kluwer Competition Law Blog, (February 28, 2019),
  9. Ai Deng, What Do We Know About Algorithmic Tacit Collusion?, Antitrust, Vol. 33, No. 1, (September 16, 2018).
  10. See Robert M. Axelrod, The Evolution of Cooperation (1984).
  11. Ai Deng, Supra note at ix.
  12. Isha Malhotra, Project Report on Price Parallelism and Tacit Collusion with Respect to Practices Under Indian Competition Law, available at:
  13. RTPE 18 of 1983, order dated 25/4/1986.
  14. Re: Domestic Airlines, Reference Case no. 01/2011,
  15. Isha Malhotra, supra note at xii.
  16. Case C-74/14, 21st January 2016, available at:
  17. Alfonso Lamadrid and Pablo Colomo, ECJ’s Judgment in Case C-74/14, Eturas (on the scope of “concerted practices” and on technological collusion), Chillin’ Competition, January 2016, available at:
  18. Isha Malhotra, supra note at xii.
  19. Builders Association v. CMA: Recognition of Plus Factors in the Indian Competition Regime?, Society of International Trade and Competition Law, available at:
  20. Isha Malhotra, Supra note at xii.


Cite as: Priyanshi Joshi, ‘Regulating Tacit Algorithmic Collusion: Meaning, Scope and its Challenges’ (The Contemporary Law Forum, 24 July 2020) <,-scope-and-its-challenges> date of access.

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.