The Impact of Artificial Intelligence on Competition Regulation

by Aileen Aditi Sundardas

Abstract

Over the past couple of years, there have been some spectacular developments in the tech world. With emergence of extreme innovation in the field of Artificial Intelligence, businesses all over are trying to find unique and elaborate ways in which they can make commercial use of these modern technological advancements. One of the most interesting sub-fields in artificial intelligence which is now getting popular in businesses is machine-learning. It essentially uses algorithms to process and learn from complex data, by adapting to it and acting on it without human intervention. With the digitisation of the economy and the unexpected commercial use of machine learning, there are some real, pertinent issues that could potentially hamper market competition regulation as it is. This paper aims to critically analyse the impact of artificial intelligence and machine learning on competition regulation. Further, it will also look at issues such as algorithmic pricing and collusion in depth, in order to comprehend the impending threat to competition in markets. The paper will conclude by synthesising an understanding of whether the current competition regime would be adequate to regulate the industry.

Introduction

 The onset of rapid technological innovation has had a cascading effect on the evolution of online marketplaces. Artificial Intelligence [1] has come to play a very crucial role in optimising and regulating the way in which consumers experience the online marketplaces. Majority of businesses are using business models that are increasingly relying on machine learning [2] to produce accurate and efficient results. [3] Companies have begun to rely on self-learning [4] algorithms for logistics, recommendations, planning, price decisions, predictions, trade and much more. [5] These self-learning algorithms compute the large amounts of data fed to it and then they process it, in order to learn and make decisions based on it. The algorithms are not required to follow a set of instructions, rather, they are able to “comprehend” the data to achieve a complex goal. [6] Artificial intelligence is therefore used to help businesses in not only running in a cost benefit and efficient manner, but it also helps in understanding the marketplace in order to achieve competitive advantage. One of the key ways in which artificial intelligence is used in competitive markets is to determine the price point for optimal profit maximisation by using autonomous pricing algorithms. But the issue that arises is whether artificial intelligence is able to achieve these goals without using anti-competitive behaviour and whether artificial intelligence comprehends the kinds of behaviours that are deemed to be anti-competitive.

Algorithmic Pricing and Collusion

One of the aims of competition law is to ensure economic growth and protect consumer welfare.[7] The regulatory [8] concerns that competition law aims to address are issues such as abuse of dominance, merger control, anti-competitive agreements and other such anti-competitive behaviour. Pursuant to that, competition law essentially ensures that all the prices set by sellers in a market are competitive. However, although machine learning algorithms are sophisticated and able to solve complex problems, they are also capable of facilitating price collusion. Price collusion is when certain players in a market, who would be deemed as competitors of a particular market, come together in order to have an influence over the pricing in that market[9]. It is an agreement between market players to behave in an anti-competitive manner so as to distort the market by having an illegal, unfair advantage. This would entail “price-fixing” in order to maximise profits.

Automated pricing systems use pricing data of competitors to determine profit maximising price points for a company. It is a complex machine learning algorithm that does not have any human intervention and is completely automated. Algorithms could potentially facilitate anti-competitive behaviour such as collusion as it could be the easiest way to maximise profits. There are two broad categories of collusion, explicit collusion and tacit collusion[10]. Explicit collusion is when players in a market get together and make explicit agreements to engage in collusive, anti-competitive behaviour by agreeing on a set price level. Tacit collusion occurs when competitors do not explicitly agree to a price level and yet the outcome that accrues is anti-competitive because they implicitly decide to coordinate and maintain such an anti-competitive scenario.

Artificial Intelligence poses a great threat of engaging in anti-competitive behaviour, although it is unlikely that it would engage in explicit collusion, the real threat lies in its potential to engage with practices such as tacit collusion. The fact that this process is entirely automated coupled by the fact that these algorithms are designed to find the most efficient way to maximise the company’s profits, there exists a threat that in pursuance of these goals, machine learning algorithms could decide that the most efficient outcome could be achieved by colluding rather than by competing. The seriousness of this issue is difficult to assess because in tacit collusion there are no agreements, there is no communication and it is an algorithm make these decisions unilaterally without any direction from humans.

The Way Forward

Algorithmic collusion has become a central to the discourse on technology and competition law, however, it is one that has very little evidence that such collusive conduct actually exists in markets.  The discourse is based on anticipated hypothetical scenarios.[11] It is an issue of utmost importance as the potential for this sort of scenario to arise is quite probable, it is an impeding threat to market players and competition regulators. It is important to synthesise an understanding of these issues beforehand in order to make changes and prevent such anti-competitive conduct, because once such conduct is detected it would be difficult to hold a seller liable given that it is the algorithm that chose to engage with such conduct without the seller’s command or awareness.[12]

In the film A Space Odyssey[13] the robot HAL is shown as having the ability to make its own decisions in pursuance of an end goal, however, the robot goes rogue at one point in order to achieve the end goal. This situation could also play out in algorithmic price determination as the algorithm has the ability to make its own decisions to achieve a set end goal of optimal price maximisation. The algorithm may not have even been programmed to behave in an anti-competitive manner but there is no way of knowing the way in which an autonomous algorithm would make decisions with reference to the set end goal.

In cases of explicit algorithmic collusion, the Indian Competition Act [14] can be interpreted in a manner that can cover issues of algorithmic price collusion. Section 3 [15] of the Act details the prohibition on price fixing and the term “agreement” used in this section can be interpreted in a broad sense so as to hold a business engaging in algorithmic price collusion.[16] This section would be much more difficult to apply in situations of tacit collusion because there is no communication nor any explicit agreements between market players. Perhaps one possible way to go forward would be to shift and increase the burden on businesses that use algorithms, to monitor the market to ensure that there is no form of anti-competitive effect on the market by the themselves or other players in the market. Or perhaps the competition regime needs to be expanded to deal with scenarios, like this if they were to arise.

Some pertinent questions that arise from the usage of autonomous algorithms could be;

  • At what level would this anti-competitive behaviour be detected? Would it be at the level of products, product categories, specific markets, specific industries, specific frequencies, specific period of time…?
  • How could a regulator establish that collusion has occurred and what would be the contours of the definition of collusion?
  • What are the actual statistics that indicate high possibility of algorithmic price collusion occurring?

The rapid technological developments in the world are challenging competition regulators and enforcers and as the technology develops, the law must find unique ways to evolve and keep up with these developments.


[1] McCarthy, J. (2007). What Is Artificial Intelligence? Technical report, Stanford University, Available online at: http://jmc.stanford.edu/artificial-intelligence/what-is-ai/index.html.

[2] Mitchell, T. M. (1997). Machine learning. McGraw Hill Series in Computer Science. Maidenhead: McGraw-Hill.

[3] Scharre, P., Horowitz, M., & Work, R. (2018). ARTIFICIAL INTELLIGENCE: What Every Policymaker Needs to Know (pp. 3-4, Rep.). Center for a New American Security. doi:10.2307/resrep20447.4

[4] Gureckis, T., & Markant, D. (2012). Self-Directed Learning: A Cognitive and Computational Perspective. Perspectives on Psychological Science, 7(5), 464-481.

[5] Ezrachi, A., & Stucke, M. E. (2017). Artificial intelligence & collusion: When computers inhibit competition. University of Illinois Law Review, 2017(5), 1775-1810.

[6] Tegmark, M. (2017). Life 3.0: Being human in the age of artificial intelligence.

[7] Singh, P.(2018). Competition policy vis-à-vis consumer welfare. International Journal of Law, 4(5), 18-24. http://www.lawjournals.org/download/382/4-5-14-772.pdf

[8] Ducci, F. (2018). Competition Law and Policy Issues in the Sharing Economy. In Tremblay-Huet S. (Author) & McKee D., Makela F., & Scassa T. (Eds.), Law and the “Sharing Economy”: Regulating Online Market Platforms (pp. 295-318). University of Ottawa Press. Retrieved October 23, 2020, from http://www.jstor.org/stable/j.ctv5vdczv.13

[9] Frass, A., & Greer, D. (1977). Market Structure and Price Collusion: An Empirical Analysis. The Journal of Industrial Economics, 26(1), 21-44. doi:10.2307/2098328

[10] OECD (2017), Algorithms and Collusion: Competition Policy in the Digital Age www.oecd.org/competition/algorithms-collusion-competition-policy-in-the-digital-age.htm

[11] Li, S. , & Xie, C. (2019). Automated Pricing Algorithms and Collusion: A Brave New World or Old Wine in New Bottles?. NERA Economic Consulting. https://www.nera.com/content/dam/nera/publications/2019/Algorithmic%20Collusion%20-%20Li%20and%20Xie.pdf

[12] Ezrachi, A., & Stucke, M. E. (2017). Artificial intelligence & collusion: When computers inhibit competition. University of Illinois Law Review, 2017(5), 1775-1810.

[13] Kubrick, S., & Clarke, A. C. (1968). 2001: A Space Odyssey. United States: Metro-Goldwyn-Mayer Corp.

[14] The Competition Act (2002)

[15] The Competition Act (2002) s.3

[16] Bhaduri, A. (2018, September 18). Pricing Algorithms : How Should India Deal With It?. IndiaCorpLaw. https://indiacorplaw.in/2018/09/pricing-algorithms-india-deal.html#:~:text=In%20India%2C%20section%203%20of,Act%2C%202002%20prohibits%20price%20fixing.&text=The%20Competition%20Act%20is%2C%20thus,to%20deal%20with%20the%20issue.

About the Author

Aileen is a fourth year student at Jindal Global Law School, pursuing B.A. L.L.B (Hons.). She is also an in-house researcher with The Digital Future – Artificial Intelligence team.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s