This post has been written by Founding Editor Viraj Ananth.
On September 30, 2018, the Ministry of Corporate Affairs constituted the Competition Law Review Committee. The Ministry explicitly lay out the Terms of Reference of the Committee as follows:
- To review the Competition Act and accompanying rules and regulations and enact necessary changes in line with the changing business environment,
- To incorporate international best practices,
- To study allied regulatory regimes and policies that interact and overlap with the Competition Act,
- To consider any other competition law matter deemed necessary by the Committee.
This move by the Ministry does not come as a surprise, with 2018 playing witness to a wide range of jurisdictions making efforts to reform their anti-trust regulatory landscapes. What does, however, come as a surprise, is the absence of any explicit reference to the role of Big Data and the need to align the Act with the changing circumstances of the digital economy. Nevertheless, it is reasonable to assume that the Committee will consider the implications of Big Data in the digital economy, given the close association of the same with all four Terms of Reference of the Committee.
What do we mean by ‘Digital Economy’ and ‘Big Data’?
‘Digital Economy’ is a broad term that encapsulates markets focusing on digital technologies. The digital economy is characterized by ‘winner take all’ models, due to the strong network effects of digital goods. Incumbents are therefore often found to occupy a strong market position and have a wide range of customers. Essentially, the dynamics of a digital economy market tends to favor market concentration and the creation of structural barriers to entry of newcomers, making such markets an important sphere for anti-trust regulators.
The scope of such data extends to shopping and to-do lists, messages sent via the speaker, and search histories, for example. Such voluminous, heterogeneous data hence allows Amazon to construct a ‘360-degree view’ of its users, and thereby, build a marketing profile, fine-tuned to individual subjectivities. The applications of big data processing are wide-ranging for companies and include carrying out personalized and targeted advertising, improving decision-making capabilities with respect to demand and supply and generating predictive models of market conditions. Therefore, the burden falls upon the Competition Commission of India (CCI) and particularly, the Committee constituted, to decide whether such applications amount to an unfair competitive advantage in terms of competition law.
Anti-Trust Issues Stemming from the Use of Big Data
The collection and processing of data serve as a source of market power for companies, acting as an entry barrier to firms which are unable to engage in a similar magnitude of collection and processing. The problem stems from the fact that a large part of big data’s utility flows from its voluminous nature. An incumbent firm, with a wider consumer base, will not only be able to collect larger quantities of data, but also a wider variety of data, in exchange for the greater number of services it is able to offer. Incumbents are hence commonly left with more users, a greater volume of data and higher revenues, allowing greater investment into processing algorithms and functionalities, thereby further compounding the gap in market share between them and smaller players.
It is noteworthy that mere access to and collection of data does not create tangible commercial value. Instead, expensive processing machinery, equipped with artificial intelligence (AI) technology, is often necessary to meaningfully process such data, only after which does the commercial value, in question, arise. As a result, a newcomer offering free services in exchange for the collection of data, in the absence of the requisite investment for processing, will still likely fail to capture the market. Moreover, incumbents have often been found to build on their primary data-sets through the purchase of third-party data, which may not be practicable for smaller firms which can only afford to exchange free services for data, and not cash per se. Incumbents may even enter into exclusive agreements with such third-party data and analytics providers, depriving smaller firms of the ability to procure similar data and therefore, gaining a competitive advantage.
Big data considerations also call for a reworking of regulatory policies on mergers and acquisitions. For example, ordinarily, a merger between an established incumbent and an innovative newcomer will not attract anti-trust concerns due to the low market share of the newcomer. However, this generalized approach cannot necessarily be transplanted onto the digital economy where such mergers can substantially increase the data concentration in the market (and hence market power), particularly if the newcomer has access to large data sets gained in a different market. Big data mining also has severe implications for cartelization. Use of big data with AI algorithms provides companies unique ways to implement their cartel agreements as well as monitor compliance with the same. One such example is the use of these tools to automatically adjust prices in real-time.
Headway in aligning anti-trust policies with the contemporary needs of the digital economy has largely been made by European jurisdictions, particularly Germany, France and Austria. In June 2017, Germany introduced amendments into its national anti-trust regime, one of the most significant of which was the introduction of novel criteria to determine the market power of a firm. These criteria included: the existence of direct and indirect network effects, economies of scale arising in relation to such network effects and access to competition-relevant data. The criteria of access to competition-relevant data falls perfectly in line with the two broad questions highlighted in the joint report submitted by the French and German competition regulators, a year prior to the amendment:
- Whether the data in question is easily obtainable by competitors or is it scarce/not easily replaceable?
- Whether the data in question has a bearing on the market power of the firm?
The German and Austrian regulators also incorporated additional thresholds focusing on the value of the transaction, in addition to the traditional thresholds of turnover of the companies, for example. This is of immense significance in digital economy markets where newcomer firms often lack ‘traditional’ corporate value due to an inability to generate high turnover initially, but which nonetheless may possess corporate value in terms of their access to data sets.
Jurisdictions across the globe are gradually beginning to acknowledge the benefits flowing from a model of data openness, with the United Kingdom (UK) in particular, recognizing the significance of the same in the context of antitrust in the digital economy. The UK’s Competition and Market’s Authority, while conducting market investigations in the banking and energy sector, directed mandatory data-sharing between incumbents and new operators, as a part of the remedy package. While the remedy package in the former case was restricted to the sharing of customer lists, there certainly exists the possibility of extending the scope of such sharing to aggregated, commercially valuable data sets, in cases where an incumbent has overreaching market power, for example.
Principles of data openness and data sharing can be well incorporated into S. 31 of the Competition Act, 2002. Under. S. 31(3), the Commission may propose a “suitable modification” to a combination, if it believes the same is capable of eliminating the adverse effect on competition in question. The Commission has frequently proceeded under this provision, as was seen as recently as September 2018, in the Linde-Praxair deal. It is possible for such ‘modifications’ to extend to data sharing agreements between incumbents and newcomers, provided the same is accepted by the parties, under S. 31(4) of the Act. Furthermore, the jurisprudential shift towards policies of data openness and data sharing has already begun in the Indian context, as can be seen in Cl. 2.4 of the (now withdrawn) National Draft Electronic Commerce Policy, for example, which prescribed data sharing with start-ups below a certain turnover threshold.
In pursuance of the third Term of Reference, i.e. studying overlapping regulatory regimes, it is essential for the Committee to consider the implications of S. 40 of the Personal Data Protection Bill. S. 40(1) necessitates that any company seeking to transfer data beyond Indian borders, maintain a ‘live serving copy’ of the same within India. This requirement has, however, been met with opposition by start-ups due to concerns over the significant increase in compliance costs, and consequently, the relative ease with which incumbents can meet the same and isolate smaller firms.
With respect to the issue of cartelization being made easier in the circumstances of digital economy markets, regard may be had to ‘anti-trust compliance by design’, a phrase first used by the European Commissioner for Competition, Margrethe Vestager. According to this approach, companies cannot rely on the defence of lack of agency and involvement in the decision making processes of their algorithms. Instead, they must ensure that all algorithms employed are built in a manner that does not allow for collusion in the first place, essentially amounting to a rule of strict liability.
The Indian economy is witnessing a huge boom in respect of technology driven markets, such as Online Payments, e-Commerce and Aggregation Services, for example. These markets, being subject to strong network effects and a heavy reliance on access to commercially valuable data sets, make mergers and acquisitions a compelling prospect for numerous players. It is essential that the CCI keep pace with changing needs and circumstances, and incorporate an understanding of the wide-ranging implications of such ‘big data driven’ mergers and acquisitions, into the regulatory landscape of the country. In doing so, two overarching factors must be considered: first, the costing structure of the numerous stages of data processing, which renders it near impossible for a newcomer to extract commercial value on par with an incumbent. Second, the lack of a reasonable substitute, in digital economy markets, for the data accessed and processed by incumbents.