Data has become an indispensable business tool, and, as a result, the collection and use of data by dominant undertakings can give rise to competition law concerns. This article, available here, examines data-related abuses in competition law, and seeks to provide an overview of specific types of abuses arising from the use of data.
Section II looks at the definition of data.
Data is often defined as “information that can be stored and used by a computer program.” Accordingly, “big data” refers to “large amounts of different types of data produced at high speed from multiple sources, requiring new and more powerful processors and algorithms to process and to analyse’. As “data” increases in volume, diversifies in nature and content, and keeps on flowing rapidly through the veins of the global economy, its collection and processing creates increasingly valuable commercial opportunities. Undertakings more and more see data as an indispensable tool for improving business decisions and strategies, and for improving their services and products.
Section III looks at the role of data in market power assessments.
In many industries, undertakings are able to collect and use massive amounts of data. As a result, competition law now needs to focus on and scrutinise the role of data collection in market structure and business practices.
A particular focus is on determining whether data collection raises barriers to entry in the relevant market and contributes to (increasing) market power. A number of criteria are relevant for this analysis, e,g. whether some data is indispensable for operating in the relevant market, or whether other players in the market can also gain access to and collect the relevant data by themselves or through third-party data providers.
Section IV considers the potential for data-related abusive practices.
The gathering and possession of “big data” is not, in and of itself, a competition law violation; however, the way companies yield such data may be detrimental to consumer welfare. This section, which comprises the bulk of the discussion, describes a number of data-related abuses under European competition law.
A first set of practices concerns refusing access to data or providing data on discriminatory terms.
Under the Bronner case law, dominant companies are subject to a special responsibility which means that, in certain circumstances, they may be forced to deal with third parties. A significant number of markets have emerged in which “data” is the most essential means for undertakings’ ability to compete. If a dominant undertaking does not allow current and/or potential competitors or undertakings operating in the downstream market to access essential and non-replicable data which is objectively necessary for them to be able to compete, this may infringe competition law. However, this will only be the case if the ‘data’ is indispensable, i.e. if (i) the data owned by the incumbent undertaking is genuinely unique, and (ii) the competitor does not have any other opportunities or avenues to access the data necessary for the operation of its business. And even if the data is found to be indispensable, one must still demonstrate that the refusal to access is highly likely to eliminate the competition in the secondary market.
These are very demanding conditions, as is apparent from the fact that only a couple of examples of such an infringement were identified in Europe. One such case is the Italian Competition Authority’s finding in Telesystem/Sip. Sip, the Italian provider of voice-telephone services, was found to have abused its dominant position by refusing to allow access to data relating to its subscribers, which was necessary for other companies to be able to compete in downstream markets. Another example comes from France. The French Competition Authority decided that Cegedim, the leading provider of medical information databases in France, had abused its dominant position by refusing to sell its main database (known as “OneKey”) to undertakings that relied on software provided by another company (called Euris). Euris competed with Cegedim in an adjacent market for customer relationship management software in the health services sector. The French Competition Authority concluded that Cegedim’s refusal to supply and discriminatory conduct had impaired Euris’s development and business prospects between 2008 and 2012.
A second set of practices concerns tied sales and cross-usage of data sets.
A dominant undertaking holding a large trove of data on a specific market may adopt tying and bundling strategies. For instance, a dominant undertaking in possession of a valuable dataset may seek to enter the market for data analytics by tying the purchase of its dataset to the use of its analytics services. An example can be seen in a 2010 French Opinion concerning cross-usage of customer databases. This Opinion emphasised that cross-usage of data obtained by an undertaking as a result of a legal monopoly may entail anticompetitive consequences, whereas such practice may increase competition if (i) the cross-used data is obtained through competition on the merits and (ii) the undertaking is a new entrant in the market where relevant data is used. Reflecting this analysis, in the GDF-Suez decision, the French Competition Authority required GDF, a gas supplier, to allow its competitors to access the consumption data that GDF had collected as a provider of regulated offers, in order to provide all gas suppliers with the same level of information regarding consumption data.
A third set of practices concern pricing.
Applying predatory pricing doctrines to data-based businesses is often challenging. For instance, online platforms generally employ skewed pricing structures on different sides of their platforms, and it is not uncommon for them to provide certain services for free. When adopted by dominant companies, zero-pricing would undoubtedly be predatory in a traditional market; however, platforms in multisided markets typically recoup their costs (and then some) on the other side of the market. Therefore, it is crucial to consider the entire pricing structure of online platforms. Another practical difficulty in applying the equally efficient competitor test to platforms is asymmetric competition where competition platforms do not have the same number of sides or do not compete on all sides.
I should point out that the authors also discuss discriminatory and excessive pricing on the pricing section, but do not really do much more than describe the relevant legal tests and describe how it could apply to data at a very high level.
The paper provides a good overview of how unilateral practices concerning ‘data’ might fall within the scope of antitrust. In doing so, they highlight one of the main limitations of most of the literature on this topic (to my eyes): it is unclear what is so special about data. Despite all the claims about big data, most cases identified in this (and other papers) concern access to databases – and, to be more specific, customer databases in the possession of regulated incumbents which authorities require these incumbents to provide to potential competitors downstream in the context of market liberalisation. Further, when discussing “technologically sophisticated” data practices, the literature often conflates practices related to digital platforms, algorithms, AI and data as if they were all and the same.
I acknowledge that there are some issues that are specifically about data – as apparent in a number of merger decisions and the Facebook case, and from the papers I will review this week and next. As a rule, however, such issues either fall squarely within traditional competition frameworks, or are so far outside such frameworks that we are unable to say confidently that the concern raised by a data practice is a competition one.