By Kajal Singh (June 29, 2021)
Edited by: N. Raja Singh, Partner
INTRODUCTION
The emergence of online platforms on the global landscape has revolutionised search, review and consumer experiences. Their advent has garnered an increased rate of market participation, widening of markets and strict anti-competitive scrutiny. Interestingly, anti-competitive concerns in platform markets are as unique as the sector itself.
Platform markets refer to two different but connected and interdependent user bases. E.g. Amazon, Google, Uber, Zomato, etc. They give rise to effects and implications for the competition regime that are by no measure similar to traditional restraints. Therefore, antitrust enforcement in these markets begets a crucial question – Whether the Competition Commission of India (CCI) is equipped to effectively curtail, restrict and penalize anti-competitive arrangements and conduct in platform markets?
The scope of the article is confined to discussing the peculiar set of problems that platform markets entail and analyse CCI’s approach and challenges faced in assessing the same. In particular, traverse into the territory of jurisprudence and routine mistakes by CCI in
1) Interpreting platform markets in a traditional value chain model,
2) Examining the interplay between pricing tactics and algorithms, and
3) Antitrust treatment accorded to different pricing strategies employed by platform markets.
Interpreting Platform Markets in a Traditional Value Chain Model
Traditionally, a value chain has three constituents: a wholesaler, retailer, and consumer. Fairtrade regulators across jurisdictions also employ this traditional perception of the value chain in ascertaining whether in a business environment, a particular the restraint is horizontal or vertical. Interestingly, the online platforms function as an intermediary and thus do not always fit into the stated conception of the value chain. The difficulty in placing online platforms at a specific position in the value chain is often the reason that these online platforms elude the clutches of competition law.
The problems associated with aligning platform markets with a traditional value chain model and set of implications arising from the same was posed before CCI in cases pertaining to Resale Price Maintenance (RPM). Therefore, the subsequent discussion will touch upon problems and implications in the backdrop of cases of RPM adjudicated by CCI.
At the outset, it is pertinent to note that for an RPM allegation to succeed, a vertical relationship between the parties must be established. In the vertical relationship, the goods are sold by the seller to the buyer on the precondition that the buyer will not resell the goods at a price lower than the price stipulated by the seller. Therefore, a vertical relationship and resale of goods are two fundamental elements to be proved to sustain an RPM allegation.
To date, CCI has adjudicated multiple RPM cases, but there has been a divided opinion on whether an RPM allegation can succeed against/ by an online platform or not. These divergent conclusions emanate from two varying approaches employed in defining the value chain and in determining whether or not online platforms engage in resale.
Thus, along with reconsidering its understanding of the value chain, CCI will have to reflect upon the relevance of resale in dealing with RPM allegations. But currently, CCI remains divided on the issue of resale being a fundamental element in deciding RPM allegations. While in the case of Jasper Infotech v. KAFF Appliances CCI held that resale need not be strictly interpreted in the conventional sense of a brick-and-mortar store. In the case of Samir Agarwal v. ANI Technologies, CCI outrightly rejected the RPM allegation against cab aggregators like Ola/ Uber on the ground that they do not engage in any sort of resale of services.
Notably, the conclusion that online platforms do not engage in the resale of services is to disregard that those online platforms are essentially parallel distribution channels. Online platforms are intermediaries which adds value to the process concerning both, the buyer and the seller. They link sellers to the consumers and thus can well be included in the value chain. Inevitably, their absence in the value chain will make the communication between the seller and the buyer consume more time, cost and bandwidth.
CCI’s decision in Samir Agarwal manifests the inherent constraints in the Competition Act, 2002 in dealing with the ever-evolving online platforms and constantly revolutionizing commercial relationships. Indubitably, for CCI to regulate the online platforms better, it will have to reconsider its understanding of the value chain. While CCI in the case of Deepak Verma v. Clues Network held that online retail portals are a part of the distribution channel, it remains unclear as to which position in the value chain would they be placed. And the question of whether online platforms engage in resale or not can only be answered after that.
Examining the Interplay Between Prices and Algorithms
Price is a crucial element while assessing competition. From businesses enjoying the privilege of exercising absolute control over deciding prices, the decisional power is now shifting towards intermediary online platforms. Their introduction has transformed access and utilization of price information in determining prices and pricing strategies. However, with technological developments, use of powerful algorithms and AI-powered pricing software has become the most prominent tactic of facilitating and monitoring tacit but actual coordination and collusion among competitors.
Pertinently, price algorithms or AI-enabled algorithmic pricing sets prices without any human intervention thus obviating the need for any overt communication or agreement among competitors. This absence of human involvement and electronically generated prices create difficulty in the detection and prosecution of anti-competitive agreements.
Recently, antitrust authorities in developed jurisdictions like the USA, in the case of US v. Topkins and UK, in the case of Trod Ltd/ GB eye Ltd. business entities have been tried for using algorithmic price-fixing systems in order to collude with one another in the market. However, CCI’s trajectory in assessing technology-aided collusion as opposed to its international counterparts has been regressive and bereft of evolving and developing an understanding of antitrust concerns.
A prominent case involving price-fixing algorithms adjudicated by CCI was Samir Agarwal v. ANI Technologies Pvt. Ltd. cab aggregators were alleged to have fixed prices using pricing algorithms. The informant envisioned cab aggregators as “Hubs” which facilitated collusion among cab drivers as the “spokes”, by algorithmically determining prices to be charged by them.
CCI discrediting the allegation of Hub and Spoke conspiracy ruled in favour of cab aggregators. It held that there was no express agreement among the cab drivers to use price-fixing algorithms. Further, prices determined algorithmically were different from the conventional notion of the Hub and Spoke arrangement. The order was subsequently upheld by NCLAT in Samir Agrawal v CCI.
In the author’s mind, CCI’s and NCLAT’s conclusion and reasoning are misplaced and premised on a rather restricted and parochial interpretation of the term “agreement”. The reasoning employed fails to factor in the broader nature of the term. The fact that cab drivers agreed to use a price determined by a standard formula is sufficient proof of the agreement between the same.
Similarly, when different entities agree to use the same price-determining algorithm they are agreeing to fix prices. Since they are prohibited to fix prices directly, they attempt and achieve the same through intermediaries by the means of algorithms. This significantly reduces the chances of getting caught, human errors and instances of deviation.
The foregoing analysis was centred on entities using the same price algorithm, however, there will be instances wherein the competitors will use algorithms that though designed unilaterally are designed to react the same way to market stimulants (Predictable Agent). Further, competitors may also deploy self-learning algorithms that have the potential to collude on their own (Digital eye). These are collectively referred to as self-learning algorithms.
Collusion through self-regulating algorithms is extremely difficult to establish primarily on account of competitors arguing that they did not design their algorithms to behave or react in a certain way. They could certainly be imputed with prior notice, however, would the same be sufficient for CCI to indict them under the competition regime remains unsettled and unanswered.
Another concern for CCI in platform markets is to ascertain whether parity in prices across platforms is achieved through price-fixing agreements and self-learning algorithms or is a natural outcome of increased price transparency. The effects of price transparency across platform markets have been positive for consumers. They have better access to information and could make an informed choice. However, with the increased consumer protection, price transparency makes it easier for competitors to monitor their rival’s behaviour and respond in an anti-competitive manner.
Thus, in assessing the anti-competitive effect of collusion, CCI will have to undertake extensive research programmes to first understand the modus operandi and the scope of algorithmic collusion better. While it has looked into self-learning algorithms being used in the airline industry it is yet to scrutinize a case of self-learning algorithms being used in the platform markets.
It will have to keep a constant check on the use of pricing algorithms. Another area of research should focus on whether there can be algorithms that can per se be treated as anti-competitive. Further as suggested by Mr DK Sikri, Former Chairperson, CCI, , the Commission should develop its own algorithm(s) to assess the issues better. It will be interesting to see if CCI itself can harness the power of AI in accurately analysing and evaluating what could be the competitive benchmark for prices and thus decide upon how much the market price has deviated from the same.
Cross-Subsidization Depending Upon the Elasticity of Demand
Pricing structures on platform markets vary as per the way consumers on both sides of the platform would respond to change in the prices of the product. The implications of the pricing structure are such that losses incurred by offering lower priced services to the side of the platform which has more elastic consumers are compensated by excessively pricing that side of the platform which has an inelastic consumer base.
The effect of the pricing structure is more pronounced in a scenario wherein the platform on one side of the market enjoys a virtual monopoly over access to the consumer base. For instance, Google is the most preferred search engine which has nonpareil access to users’ data and a monopoly in providing search services. It provides majority of its services for free to consumers on the side of the platform which avails its search services and recover the cost incurred by excessively pricing publishers on the other side of the product availing its advertisement services.
The same pricing strategy is employed in instances wherein a particular product sold on the platform experiences high elastic demand. The platform has no choice but to offer the product at low prices otherwise the consumer will shift to other alternatives. The platform will make up for the lost profits by pricing another product that has an inelastic demand. For instance, if Amazon incurs losses by pricing non bestseller books really low it will compensate for the same by pricing bestseller books high.
CCI had the opportunity of addressing cross-subsidization in the case of MCX Stock Exchange v. NSE. In this case, one of the subsidiaries of NSE in the currency derivative sector engaged in zero pricing. It was a standard example of cross-subsidization, however, CCI held NSE liable for unfair pricing and not for cross-subsidizing by leveraging its strength. Thus, in the author’s mind, the judgement and reasoning failed to firstly, appreciate the existing jurisprudence of Competition laws and secondly, furthering it for assessing cross-subsidization.
Undeniably, the practice of cross-subsidization is gaining traction and has become extremely common. Its effects are more pronounced on platform markets because of their ability to perfectly price discriminate. Moreover, owing to multiple price changes on the platform market in a day makes the activity untraceable.
These instances of cross subsidization, as the one in MCX stock exchange v. NSE or cross-subsidization across platforms depending upon the elasticity of consumer demand could be dealt better by applying the “Incremental Cost Test” and the “Stand-Alone Test”.
By applying the same, the Incremental Cost Test, it can be ensured that prices charged for the service rendered are equivalent to the additional cost incurred by the company in offering that service. Further, the Stand-Alone Test will compare the rate that would prevail in a competitive market and the rate so challenged, and thus will judge the reasonability of the challenged rate. Whereas the cases of cross-subsidization as illustrated through Amazon demands a constant surveillance over prices of elastic and inelastic demand goods sold on the platform.
Conclusion
Thus, CCI will have to keep itself abreast of advanced jurisprudence, regulation, and reconsider traditional conceptions and tools in assessing platform market models. Another area of focus would be amendments to the Competition Act to make it more suited to the changing trend of the platform markets.
_____________________
1 Francisco Beneke et al, Artificial Intelligence and Collusion (Dec. 20, 2018).
2 Anupam Manur, Regulating Multi-Sided Platforms: Challenges in the Indian Context TAKSHILA REVIEW PAPER (July 31, 2018) pg 18.
