The Conundrum surrounding Price Fixing Algorithms and Cab Aggregators: Should the NCLAT reconsider its stance?
– Palash Moolchandani & Shaivi Shah
Introduction
The recent order passed by the NCLAT in Samir Agrawal v. Competition Commission of India and Ors. has, once again, revealed its woefully underdeveloped understanding of contemporary competition issues. The NCLAT cemented the well-known fact that Indian competition jurisprudence is leagues behind its more developed counterparts and has set a precedent that marks a step back in India’s journey towards recognizing artificial intelligence and algorithm-induced competition violations.
Factual Matrix
On 14th October 2018, Samir Agrawal (“Informant”) filed an information with the CCI alleging that the business model followed by OLA and Uber (“Opposing Parties”) is violative of the Competition Act, 2002 (“the Act”) as it facilitates price-collusion among cab-drivers on the basis of the Hub-and-Spoke principle. It envisions the Opposing Parties as being “Hubs” that algorithmically determine the prices to be charged by the drivers, who, in turn, act as the “Spokes”. Every driver who enters into an agreement to be listed as an independent contractor with the Opposing Parties is aware that it is signing identical terms on pricing as its competitors who are already attached to the aggregators. The Informant alleged that this would amount to a meeting of minds among the drivers and would be violative of section 3(3)(a) read with section 3(1) of the Act.
In November 2018, the Commission ruled in favour of the Opposing Parties on the grounds that a Hub-and-Spoke collusion would require the existence of an explicit “conspiracy” between the cab-drivers to use the Opposing Parties’ algorithms to indulge in price-fixing/co-ordination or exchange of sensitive information. Further, it was stated that the algorithms determined prices on the basis of large sets of big data and numerous variables such as distance, availability of cab, weather, etc. Thus, the prices determined would be different at every instance, making it impossible to indulge in price-fixing.
The order was appealed, and on 29th May 2020, the NCLAT, New Delhi, upheld the decision of the CCI.
Algorithmic Pricing and Collusion
The rapid advancement of technology and the digital economy has led to the emergence of algorithms. Vast collections of data, referred to as ‘Big Data’, have enabled these algorithms to determine the prices in a manner that is quicker and more accurate than humans. Unfortunately, the increased usage of such algorithms enhances the possibility of facilitating cartelization. Prof. Ariel Ezrachi, in his seminal work, has identified four ways through which algorithms can be used to achieve anti-competitive collusion.
1) Messenger – In this case, algorithms are merely used as tools to execute pre-existing price-fixing agreements. They function as messengers to facilitate the exchange of information and price-signalling among cartel members. This was observed for the first time in the case of USA v. David Topkins wherein complex algorithms were used to facilitate a pre-existing agreement to fix the prices of posters sold on Amazon.
2) Hub-and-Spoke – In this scenario, competitors use common platforms or similar algorithms to determine the prices, with a view to sustain a horizontal cartel. The algorithm is used as a central Hub to charge similar prices without actually entering into price fixing agreements. This also enhances price transparency which help firms to adapt to the prices of their competitors, thereby, facilitating tacit collusion. In 2016, the EU Court of Justice prosecuted the administrator of an online travel platform, Eturas, who issued a notice to all its attached travel agents, stating that it was imposing a minimum-discount limit. The court held that such a policy created price parallelism and was in violation of Article 101 of TFEU. Further, it held that a Hub-and-Spoke arrangement does not require an express agreement, and the very fact that the travel agencies knew about such policy and did not oppose it raises a presumption that they were complicit to the same.
3) Predictable Agents – In this case, firms use their own individualized pricing algorithms that constantly monitor and adjust to competitors’ prices.
4) Digital Eye – In this scenario, algorithms collude with each other on the basis of their own self-learning.
Analysing the NCLAT’s Decision: An Error in Judgement?
In this case, it was alleged that Uber and Ola, through their pricing algorithms, had caused their drivers to cartelize themselves in accordance with the Hub-and-Spoke principle. As per Indian competition jurisprudence, in order to establish the existence of an anti-competitive cartel, one would have to establish – a) An agreement suggesting a conspiracy b) A resulting price parallelism in the market and c) An intention to restrict/eliminate competition in the market.
The NCLAT ruled in favour of the Opposing Parties on the ground that there was an absence of an explicit agreement among the cab-drivers to use the algorithm provided by the cab aggregators to fix prices. However, section 2(b) of the Act provides for the term “agreement” to cover a wide ambit. It not only includes explicit written agreements but also includes non-verbal understandings and agreements entered into in concert. In a Hub-and-Spoke mechanism, an agreement entered into individually by each Spoke with the Hub would not in itself be anti-competitive. However, a multiplicity of such agreements, wherein Spokes have knowledge of the fact that their competitors would be entering into similar agreements, could lead to market-wide collusion. Since it is a well-known fact that such agreements would not be entered into explicitly and all incriminatory evidence would be erased, the Indian competition regime provides for acceptance of evidence that may be “fragmented” or “sparse”, going so far as to include agreements that have been entered into by way of a mere “nod” or a “wink”. Thus, it is safe to conclude that a concerted agreement among the cab-drivers exists.
In the present case, the cab-drivers agreed to use the same algorithm, fuelled by the same big data sets provided by the cab aggregators. Thus, it can be established that they were engaged in price parallelism. As per the Consumer Unity and Trust Society (CUTS), India, “agreements to use a standard formula according to which price will be calculated” would also be considered as price-fixing agreements under the Act. Further, as mentioned in the previous segment of this article, it is largely accepted in well-developed competition jurisdictions that agreeing to use the same algorithm in a market would amount to agreeing to fix prices. The rationale behind this is that if competitors would not be permitted to fix prices in a market by means of an explicit agreement, they shouldn’t be permitted to do through an intermediary either.
In a bid to negate their intention to indulge in cartelization, the Opposing Parties stated that their terms of service permitted cab-drivers to “multi-home” between different aggregators. However, as correctly pointed out by the Informant, such a defence was meaningless as it essentially involved switching from a cartel run by one of the Opposing Parties to another run-in concert with the other Opposing Party.
In view of this, the decision of the NCLAT, under section 26(2) of the Act, stating that a prima facie case was not established was inaccurate. The Informant showcased that the Antitrust Division of the DOJ, USA, an authority that India has frequently relied upon, initiated an investigation into Uber’s business model on the grounds of allegations that were similar to those put forward in the impugned case. The Tribunal stated that an investigation could not be carried out as there was neither any direct evidence and nor could foreign authorities form the basis to determine a prima facie case. However, in the case of Western Coalfields Limited, the Commission expressly opined that in cases of alleged cartelization, the absence of “direct” or “prima facie” evidence could not be used to side-step the validity of circumstantial evidence to infer anti-competitive activity. Further, in the well-known Android Case, the Commission ordered an investigation into allegations of abuse of dominance against Google on the basis of international precedents wherein the said practices by Google were denounced. Additionally, in a stark contrast to the present case, the Commission, on merely receiving an anonymous mail, took suo motto cognizance of the Maruti Suzuki Case and found a prima facie violation of the Act. Thus, the Tribunal seems to have deviated from the standard of determining a prima facie case as has been laid down in these precedents.
Conclusion
In conclusion, the authors believe that the conservative stance of the NCLAT needs to change. The new digital economy calls for the application of progressive and up-to-date regulation on the latest technological modus operandi of firms carrying out an anti-competitive activity. The business models of the Opposing Parties would be in line with the Hub-and-Spoke framework of collusion, and hence, would be violative of section 3(3)(a) of the Act. Further, the allegations put forth to the Tribunal merit the finding of a prima facie case and the subsequent ordering of an investigation.
To sum up, the authors would like to refer to a statement by Maureen Ohlhausen, Federal Trade Commission – “Is it ok for a guy named Bob to collect confidential price strategy information from all the participants in a market, and then tell everybody how they should price? If it isn’t ok for a guy named Bob to do it, then it probably isn’t ok for an algorithm to do it either.”
This article is authored by Palash Moolchandani & Shaivi Shah. They are students of law at National Law University, Odisha.