Algorithmic Collusion - An Emerging Antitrust Concern

- Chhavi Jain (Third Year Law Student at National Law University Jodhpur)

Introduction

Technology evolution and the rise of artificial intelligence (“AI”) have given a fillip to innovation and economic development. At the same time, these have also raised numerous antitrust concerns, especially in the emerging digital markets. Big-tech firms are in a position to abuse their dominant position and misuse consumer data, and algorithms can facilitate collusion. The Draft Digital Competition Bill, 2024 (“Draft Bill”) attempts to address some of these concerns by bringing in ex-ante regulations for the provision of core digital services by “Systemically Significant Digital Enterprises.” However, it does not recognise algorithmic collusion. This blog article analyses the potential anti-competitive effects of algorithmic tacit collusion, specifically in oligopolistic market conditions.

Algorithms Facilitate Collusion

Algorithms, found in various forms, are rules or codes which give certain output when given a certain input. They increase supply side and demand side efficiencies by boosting innovation and transparency and supporting consumer choices by making information accessible. Used in different industries for a plethora of purposes, certain algorithms, like pricing algorithms, can risk collusion.

Four models of algorithmic collusion have been identified in leading literature; the focus herein is on the Predictable Agent Model in oligopolistic markets. In concentrated markets, the firms recognise their mutual interdependence and use their collective power for profit maximization. The competitors closely monitor each other and use the same algorithms that facilitate tacit collusion.

Anti-competitive agreements are void and unlawful under Section 3 of The Competition Act, 2002 (“The Act”). Agreements are broadly defined under Section 2 (b) of the Act to include “arrangement, understanding and action in concert.” Since such agreements are usually not based on formal contracts, and can be based on a simple “nod or wink”, tacit agreements have been recognised by the Competition Commission of India (“CCI”). 

In Shikha Roy v. Jet Airways, the CCI recognised this concept for the first time. The information filed alleged certain airlines to have raised airfares of some routes, pursuant to an agreement under Section 3, to exploit passengers during the Jat Agitation. Although the CCI dismissed the case against the Informant, it did acknowledge that algorithms significantly increase the possibility of collusion, “with or without the need for human intervention and coordination”. Forming a prima facie opinion, it investigated the use of algorithms by the Opposite Parties (“OPs”) and held that the OPs were not liable since all airlines were using different software and there was no instance of identical pricing. Moreover, historical data, specific and different for each airline, was the input, thus creating a separate algorithm for each airline.

Conscious Parallelism

Conscious parallelism is when competitors do not act in concert, but make informed business decisions in response to the economic situations, and this results in supra-competitive prices. The United States Supreme Court in Theatre Enterprises, Inc. v. Paramount Film Distributing Corporation did not consider similar prices arrived at, based on independent business decisions, as anti-competitive.

In concentrated markets there is a thin line separating tacit collusion and conscious parallelism. Such actions fall under a grey area, where the acts go beyond conscious parallelism. However, in the absence of clear evidence of concerted action or cooperation and the oligopolistic market conditions, the weight is tilted in favour of conscious parallelism.

The Lithuanian Supreme Administrative Court dealt with a similar issue in the famously called “Eturas” case. The court observed that the travel agencies using the E-TURAS booking system ought to have known that other firms were also using it. Thus, they were obligated to act with “care and diligence.” A presumption of concerted practice is raised against the firm, unless it publicly distances itself from the information exchange or reports the same to the authorities.

Regulating The Use Of Algorithms

The issue that primary arises is rooted in informed business decisions based on prudence. None of the firms would be willing to bear the cost of developing its own algorithm. Moreover, no firm would want to lose out on the advantages that their competitors get by using the industry benchmark algorithms. However, firms should not be allowed to take undue advantage of these defences.

The Draft Bill does not recognise algorithm facilitated collusion. However, the ever-increasing use of algorithms needs to be regulated. Specific regulations for pricing algorithms and algorithmic collusion can be formulated. The authorities may also formulate guidelines for the use of algorithms, covering related concerns, including, but not limited to, antitrust and data privacy. The CCI should tread carefully while regulating such collusion and must balance the economic implications.