[By Srishti Suresh]
The author is a student at NALSAR University of Law, Hyderabad.
Introduction
The recent National Company Law Appellate Tribunal (“NCLAT”) ruling in the case of Samir Agrawal v. Competition Commission of India & Ors.[i], effectively upheld the previous CCI Order[ii], in holding that the algorithms used by cab aggregators such as Uber and Ola, do not result in the creation of a hub-and-spoke cartel. In its decision, the NCLAT laid emphasis on the form of arrangement between the cab aggregator and the independent drivers, as against the substantial effect such a scheme could have on free-market competition.
In the appeal, the contention raised by the informant was in relation to the centralized power of the aggregator in fixing prices for the drivers though the App, thereby effectively barring them from competing with the prices in the market. Whilst possessing asymmetric information related to personalized rider data and other crucial information such as surge demand, traffic etc. gathered from AI-fueled algorithms, drivers engaged by the aggregators were required to charge their customers, a price determined by the App (which is based on preset factors). Scope for potential negotiation and bargain is stripped away, and the market forces of demand and supply are artificially distorted. This, it was alleged, leads to a potential increase of the fares and leads to price-fixing by the aggregator. At the helm, cab aggregators operating as intermediaries between the drivers and the riders, have the leverage to fix prices, distort free competition between similarly situated service providers, and effectively violate Section 3 of the Competition Act, 2002 (“the Act”).
A striking aspect of the ruling, which assumes importance for the purpose of this article, is the “connectivity” between drivers. The scheme of arrangement between the drivers and the aggregator is as follows: each driver enters into a vertical agreement with the aggregator, and each driver is well aware of the fact that multiple competing drivers are simultaneously entering into an analogous agreement with the same aggregator. No driver enters into a direct agreement with the other. The process works on an implicit acknowledgement of the role played by each driver, as well as the aggregator. NCLAT in its order, focused on the absence of connectivity between drivers inter se. For a collusion inhibiting healthy market competition to exist, an understanding or agreement between each party, with the other, is a vital necessity.[iii] However, owing to the vertical arrangement between the parties, the possibility of an anti-competitive collusion was rejected.
Understanding the Hub-and-Spoke Model of Cartels
Cartels conventionally involve communication between cartel participants, agreeing to engage in illicit conduct. However, in the wake a more stringent antitrust regime across jurisdictions, the hub-and-spoke model has assumed great importance[iv]. In this model, the hub is either an upstream supplier or a downstream customer, and the spokes are the various colluding competitors. Each spoke enters into a separate vertical agreement with the hub and offers sensitive information.[v] The same information is disseminated by the hub to the other spokes, while engaging with them vertically. In essence, the information that is stored centrally with the hub is schematically disbursed to the spokes, without them having to communicate with each other directly.
The Shortcoming of the NCLAT ruling
In analyzing the narrow approach taken by the CCI and the NCLAT, one needs to steer clear of any ambiguity that might exist in understanding Section 3 of the Act, in the context of algorithmic collusion.
First, the deemed provision prohibits any agreement, which among other aspects, relates to the provision of services causing an adverse appreciable effect on competition (“AAEC”) in India.[vi] A textual interpretation of the provision seen through a human prism[vii] (conventionally), prohibits agreements that cause a direct or indirect effect on competition. But AAEC is not restricted to an exhaustive list of agreements, the form of which cannot be delineated with clarity and certainty. Any arrangement with the potential to cause economic consequences, operating to the prejudice of public interests and unduly restricting competition, can be construed to pose AAEC on the market.
In the present case, the information that is curated by the algorithm creates a resource and data pool, that is commonly accessed by all the drivers. By analyzing rider data, their frequency in hailing services, and any surge in a given locality, the algorithm has the potential to reduce its output in terms of services, while escalating prices for the same. By offering drivers a lucrative opportunity to earn more, by the way of increased fare charges, the aggregator’s app is at the helm of unilateral “price fixing”. While no horizontal agreement between the drivers exists, the multiplicity of analogous arrangements with the aggregator (the “hub”), coupled with a common pool of information and resource sharing among drivers (the “spokes”), increases the possibility of collusion vis a vis a hub-and-spoke agreement. In addition, given that all the drivers are aware of other competing drivers entering into similar agreements with the aggregator, for the same purpose, makes the agreement a perfect candidate for a hub-and-spoke model.
The need for a proactive approach
Technological advancements in service sectors have transformed the competitive landscape. The rate at which information is accessed, analyzed and processed, has increased manifold[viii]. Algorithms have a peculiar characteristic of allowing sellers to shadow their customers, by harvesting data on their consumption behaviour.[ix] While the traditional route taken by most competition regulators mandates a concurrence of understanding through a tacit or explicit agreement between participant themselves, the new age of algorithms might not follow the traditional route of concurrence.[x] OECD in its Report has acknowledged the highly uncertain and complex nature of algorithms, but it has also issued a caveat for regulators. A lack of an interventionist approach and over-regulation, could impose a high cost on the society[xi]. Algorithms have proven to be excellent automotive tools in increasing market efficiency. In reviewing the current framework, the Competition Law Review Committee (“CLRC”) Report has highlighted the broad ambit of Section 3. ‘Algorithmic collusions’ could very well be brought into the ambit of anti-competitive agreements. However, the investigative tools in the current framework do not facilitate an effective investigation into algorithmic collusions. The ‘Rule of Reason’, allows regulators to analyze hub-and-spoke agreements that are not per se illegal.
Under Rule of Reason, the overall competitive effects of an agreement are analyzed, while according great weightage to the nature of the agreement and its business purpose. Only facts that are perceived to be necessary in understanding is the intent behind such vertical arrangements with drivers, the common resource pool generated and shared by a centralized algorithm, and its overall effect on the market, are analyzed.[xii] The aforementioned case of cab aggregators, akin to a hub-and-spoke model, effectively transfers decision making power and financial control in the hands of the “hub”. The resultant distortion of market demand and supply, strikes at the root of healthy competition between similarly situated competitors. The increased prices harm end consumers to a considerable extent.
The NCLAT in its ruling has questioned the locus standi of the informant, as he was unaffected by the services offered by the aggregators, and has discounted the plausible anti-competitive nature of a hub-and-spoke arrangement for collusion. A more detailed investigative and proactive approach in understanding the operations of algorithms, the deployment of the same by service providers, and a subjective assessment of the intent of individual participants[xiii] (the spokes), may shed light on the true effects of the arrangement. With the growing adoption of AI and algorithms in different service sectors, the challenges in analyzing and detecting collusive behaviour is only set to increase.
Endnotes:
[i] Samir Agrawal v. Competition Commission of India & Ors., CA No. 11/2019.
[ii] Samir Agrawal v. ANI Technologies Pvt. Ltd. & Ors., LNINDORX 2018 CCI 20 (Case No. 37/2018).
[iii] Supra note 1, ¶ 17.
[iv] Ariel Ezrachi and Maurice E. Strucke, ‘Artificial Intelligence & Collusion: When Computers Inhibit Competition’, 2017 U. III. L. rev. 1775 (2017).
[v] A McCabe, ‘The English Court of Appeal’s Legal Test for “hub and spoke” Cartels- is it Compatible with EU Jurisprudence?’, Eur. Comp. L. Rev. (2012) 33(10) 452-457; A Bolecki, ‘Polish Antitrust Experience with Hub-and-Spoke Conspiracies’, Yearbook of Antitrust and Regulatory Studies (2011) 4(5) 26-27.
[vi] The Competition Act, 2002, §3.
[vii] The Antitrust Laws, Fed. Trade Comm’n, https://www.ftc.gov/tips-advice/competition- guidance/guide-antitrust-laws/antitrust-laws.
[viii] McKinsey & Co., Big Data: The Next Frontier For Innovation, Competition, and Productivity 98 (2011).
[ix] David J. Lynch, Policing the Digital Cartel, Fin. Times (Jan. 8, 2017), <https://www.ft.com/content/9de9fb80-cd23-11e6-864f-20dcb35cede2>.
[x] In re Text Messaging Antitrust Litigation, 630 F.3d 622, 628-29 (7th Cir. 2010).
[xi] oecd, ‘big data: bringing competition policy to the digital era’, (27 October 2016), <https://one.oecd.org/document/DAF/COMP(2016)14/en/pdf>.
[xii] Board of Trade the City of Chicago v. United States, 246 U.S. 231, 238 (1918).
[xiii] Id.