The shift in the world’s operation to a burgeoning technopolis, with reliance on data for service optimization, has raised many possibilities and concerns for citizens and policymakers alike. As the world appears to be in the throes of another such revolution, the increased usage of Big Data and its analytics at large has raised the question of algorithmic pricing in the realm of competition law.
Impact on the Market
The advent of Big Data analytics has had a tremendous say in the operation and importantly the prices at which many services operate in the 21st-century economy. From online retail, airline tickets to hotel booking, prices today are set by algorithmic actors (i.e. computers) instead of humans as they are best able to apply and determine the price which matches twin of issues of customer demand as well as prices charged by competing entities prices to increase revenues.
It is arguable to state that these developments might lead to an increase in market transparency by enhancing competitive pressures thus benefiting the consumer. However, this rosy Adam Smith-endorsed picture presents only one side of the narrative. On the other hand, the very same techniques can and have been used to engage in interdependent pricing which may result in an artificial rise of prices and thus raise the question of anti-competitive behaviour.
This issue has so far been examined in a handful of cases in the US & UK. The overarching narrative is that algorithms were used to execute preexisting agreements between competitors to fix prices, thus the criterion of the meeting of minds was fulfilled.
This was found as being the case in the United States. In the Topkins case, a former employee of an online art retailer was prosecuted for conspiring with competitors to fix the prices of certain posters sold through the Amazon Marketplace. Similarly, in the case of Aston, the US Department of Justice indicted a director and his company of a similar offence. The common element in both these decisions was the element of the meeting of minds to collude to adopt specific prices as per their algorithms. Given that in both these cases the algorithms were an extension of an already established agreement, these individuals and entities were convicted.
A unique case, however, presented itself with regard to the claim and subsequent lawsuit against erstwhile Uber CEO Travis Kalanick (with Uber being added as party later on in the suit) over price surcharge activities. According to the complainant, Uber’s vertical agreements with each individual driver gave rise to a horizontal coordination due to parallel use of the same algorithm. In this manner, Uber was alleged to have constituted a “hub and spoke cartel” wherein independent drivers (the spokes) collude with one another through a hub i.e. Uber.
The presiding judges with regard to the allegation of “adverse effect? on competition stated that any such determination would require further investigation in the relevant market. In addition, it was stated that the defendants’ counterclaims of pro-competitive benefits required a fact-finding exercise to be determined conclusively. However, in spite of this, the court dismissed the suit. The question of what constitutes relevant market remains elusive, one which may not be addressed sufficiently with the ongoing arbitration between the two parties.
The Next Wave
The principle of the meeting of minds is well entrenched to establish an offence under competition law. But what happens in cases where the algorithms are able to coordinate prices without being under the aegis of human operators? With greater input of Artificial Intelligence in such algorithms, the likelihood of this happening is a reality rather than a far-fetched thought at this juncture. An action such as this effectively leads to an anticompetitive event happening without the element of the meeting of minds between “human actors” taking place at all.
The European Union which in the past decade has taken the stance of policing technology companies for their alleged anti-competitive behaviour appeared to have developed the answer to this conundrum. EU Competition Law Commissioner, Margrethe Vestager stated that an algorithm’s actions of communicating with competitor’s algorithms will be treated as an extension of human will.
Thus, if necessary safeguards are not undertaken by the host entity (the one using the algorithm) to provide against such systemic collusion, the blame will be imposed on them. However, Vestager consigns the host party to be indicted even if the necessary safeguards have been undertaken as under EU law companies will be held liable for any anti-competitive practices of their employees, even if they can show that they have used their best efforts to prevent these.
The American approach appears to have taken an alternate route as per the opinion of a Senior Official of the US Department of Justice, Barry Nigro. According to him, tacit collusion through such mechanisms is not illegal unless there is an agreement among the participants.
Having stated the European and American outlook on these issues, the language of the law, particularly for the former jurisdiction, appears to provide a loophole for parallel conduct by algorithms in price fixing. Given the evolving nature of algorithms and their eventual autonomous form, price fixing has also come to take place by means of parallel conduct. Herein algorithms would be able to monitor each other’s behaviour and as a result, engage in price fixing which would effectively be covertly done without any agreement.
Notwithstanding Commissioner Vestager’s opinion, parallel conduct as a practice is not explicitly prohibited by law as per Article 101 under the TFEU. The caveat being that independent parallel conduct must be without any direct or indirect contact between the operators. Thus, entities in this manner would be able to engage in price-fixing activities without suffering from any liability from a plain reading of the law.
As a subset of this, tacit collusion works on the premise that the increasing use of pricing algorithms coupled with growing market transparency would result in a scenario where there would be no need to actively collude. While this variant is not considered illegal in most jurisdictions, it is, however, raising concerns amongst many of the antitrust authorities across the world.
An ex-ante merger control approach would be one of the methods to tackle this issue by establishing a system capable of preventing tacit collusion, through the enforcement of merger control rules in markets with algorithmic activities. This may require agencies to consider lowering their threshold of intervention and investigate the risk of coordinated effects not only in cases of 3 to 2 mergers but potentially also in 4 to 3 or even in 5 to 4 mergers. Such an approach would allow the regulators to assess industries where tacit collusion is more sustainable as well as those where it isn’t.
Alternatively, a focus could be based on concentrated markets which involve homogenous products where the algorithms can monitor to a sufficient degree the pricing and other key terms of sale. As well as focusing on those markets which are already susceptible to human coordination, with these algorithms serving as the last key towards achieving coordinated equilibrium.
From the standpoint of businesses, they must ensure that their algorithms are anti-trust compliant by design, meaning that their structure doesn’t allow them to collude.
The relevant question while the EU & US tussle on their respective terrains in regulating the formation of robot cartels is what relevance this has for an emerging and digital-friendly economy like India? The answer lies in understanding the economic scope of these enterprises and how widely they have grown in this decade.
Multiple domestic tech-based companies have flourished, coupled with the integration of pricing algorithms in traditional businesses necessitate the regulatory apparatus to be well prepared. The experience in the EU is particularly essential as it sets the tone for how emerging economies might want to contend with the competition arising out of their digital economy.
In 2018, the Competition Commission of India decided a case pertaining to the very same issue of algorithmic pricing by Uber and Ola. As per existing law, it found their actions as not constituting an anti-competitive act. The application of the hub-spoke principle, in this case, could itself be a subject of a separate discussion. However, this decision highlighted what anti-trust scholar Lina Khan has repeatedly asserted – that the conception of competition law needs a reinterpretation for the digital economy and its functionaries.
With self-learning algorithms such as Libratus becoming increasingly prevalent, the traditional conceptions of agreement and intent as defined in competition law jurisprudence become things of the past. Policymakers now need to ascertain the adverse impact, if any, on the market. If yes, then a determination on how to audit pricing algorithms and where necessary introduce automated checks into their design is needed.
The question then is whether algorithmic pricing regulation is a competition law concern or that of another arena of law and policy?
(The use of Big Data and algorithmic pricing by e-retailers opens up questions of competition law, says Dhruv Shekhar)