European Union – Data-related Abuse of Dominance
Data has become an increasingly relevant resource – not only for e-commerce but also for brick-and-mortar businesses. Its significance goes far beyond multisided markets, although it seems that issues concerning data-related abuses have been discussed most frequently against the backdrop of the business models of Facebook, Google and the like.However, supermarkets and coffee house chains rely on data for their strategic decision-making just the same.
The access to and subsequent use of data raises several issues under EU competition law.This chapter focuses on the relevance of data in the context of unilateral business behaviour. It boils down to seemingly simple questions: is (data-related) knowledge power and is it possible to misuse such power?
The legal framework of competition law for answering this question is relatively straightforward: Regarding unilateral behaviour, competition law only imposes obligations on undertakings that are dominant pursuant to Article 102 TFEU.Such undertakings are required not to abuse their dominance and must refrain from certain business conduct. Business conduct, it is worth noting, that would be permissible if employed by a non-dominant undertaking.
Translated into competition law terms, the answers to these questions depend on whether data alone or together with other factors might confer a dominant position on an undertaking and whether data can then also be used to abuse that dominant position. Only if both questions are answered in the affirmative can data-related knowledge really be considered power in the context of Article 102 TFEU.
The relevance of data for establishing a dominant position
An undertaking has a dominant position pursuant to Article 102 TFEU if it can behave to an appreciable extent independently of its competitors, its customers and consumers in general, and thus ultimately prevent effective competition from being maintained on the relevant market.The test is twofold. The relevant market needs to be defined and dominance in the market needs to be established.
The theoretical framework
The starting point for the assessment of data-related abuses is the definition of the relevant product or service market and its geographical scope.Under competition law, the concept of the demand-side oriented market is used for this endeavour.
On the demand side, products or services belong to the same market if they are substitutable from the buyer’s point of view.Relevant factors are the product’s characteristics, its intended use and the pricing. Geographically, the coverage of the market depends on the area in which the undertakings are engaged in the supply and demand of the products or services and where the competitive conditions are sufficiently homogenous. The definition of the relevant market, although only the first step for identifying the relevant competitive environment, has an immense – some say even unjustifiably high – influence on the outcome of the assessment. It might make a huge difference for an undertaking whether the market conditions are assessed on a national or global level and whether certain products are deemed interchangeable or not (as this usually influences the market shares of the relevant undertaking).
The demand-side oriented market concept has many nuances. Although intuitive at first sight, its application can become challenging if complex relationships between customers and suppliers, in particular in the context of multisided markets, are involved. Some of these challenges will be addressed later. For now, it is sufficient to note that for applying Article 102 TFEU, the demand-side oriented market concept is the preferred method of the EU Commission (Commission) and the European courts.While, for the reasons stated above, alternative assessments have been tested in particular in the context of merger decisions, conceptually, Article 102 TFEU requires the prior definition of a market before dominance can be assessed.
What is data?
Before the demand-side oriented market concept can be applied in a data context, it is critical to understand what ‘data’ means for the purpose of competition law. Although often used in the current discussion, surprisingly, there is no generally accepted definition.
As a starting point, ‘data’ equals physically manifested information.For example, the Merriam–Webster dictionary defines ‘data’ in its online version as ‘information in digital form that can be transmitted or processed.’ Understood this way, access to and use of information have always been relevant issues under competition law. For example, in the classic 1979 decision Hoffmann-Ra Roche, the Commission and the CJEU based the finding that an undertaking had a dominant position, inter alia, on its ‘technological advantages’, explicitly stating the ‘highly developed customer information’ at its disposal. The IMS Health case, still a landmark decision concerning the ‘essential facility doctrine’, dealt with a right to access to sales information stored in a database of an undertaking offering reports to the pharmaceutical industry.
If data is discussed in a competition law context, this is often done against the background of a very specific kind of data, namely personal data.According to Article 4(1) of the EU Data Protection Regulation, personal data ‘is any information relating to an identified or identifiable natural person.’ This definition captures, for example, a natural person’s domicile, shopping behaviour or internet surfing patterns. The significance of personal data for business dealings is not a phenomenon of the digital era. Undertakings have always been interested in understanding the preferences and needs of their customers and often this understanding is based on what one would today call personal data. Nonetheless, some things are different in the digital economy: certain business models mainly rely on access to and use of personal data. For a long time, personal data was used to boost the sale of goods and services by enabling undertakings to better satisfy the needs of customers. It was a means to an end. Now, having access to personal data is a goal in itself as it can be directly monetised. Further, presumably as a reaction to these new business models, people are becoming increasingly aware of the value of their personal data and expect a reasonable legal framework for its protection. Both developments might be the reason for the special attention that personal data currently receives from competition authorities and scholars alike.
At least for the purpose of competition law, it would be wrong to narrow data down to personal information or any specific information for that matter. Personal data is just one (albeit increasingly important) kind of data. There are many others, for example non-personal data, structured and unstructured data, voluntarily provided data and publicly available data.From the outset, there is no reason to assume that one category of data is more valuable than the other. The Federal Cartel Office in Germany recently assessed the establishment of an e-commerce platform for steel producers and steel merchants. In the process, it particularly reviewed whether data about the business behaviour of the steel manufacturers (non-personal data) was secured against access by the other steel manufacturers. Further, even publicly available information can be used in a way relevant for competition law. The technology company Uber Technologies Inc offers an application for smartphones that connects riders with drivers. The prices for fares are set by an algorithm but drivers can raise the standard fare during times of high demand. Correspondingly, Uber informs its drivers upfront about upcoming events that might create such demand (publicly available information). Ultimately, only few business decisions are based on the evaluation of one kind of data alone. More likely, it is a mixture of several input factors.
For competition law, all data can be of relevance. Addressing all potential issues would thus go beyond the scope of this chapter. However, in many cases, access to and use of data does not raise specific competition law issues. Rather, data plays only a subordinate role (like in the Hoffmann-Ra Roche decision). Thus, for the remainder of this chapter, the focus will be drawn to situations in which data is central for the assessment. This is the case for business models which require a great volume and a broad variety of data to be gathered, usually at high speed from multiple sources (also called the ‘3Vs’: volume, velocity and variety).Such business models are often described by the term ‘big data’.
The relevance of data for defining markets
Traditionally, data has most of the time only indirectly influenced the way markets were defined. For example, when the Commission evaluated whether low-end and high-end manufacturers of extruders belong to the same market in the Triton/Logstor merger case, it considered issues such as precision, design and technology used.Data was not expressly mentioned in this context, but access to technology and production methods are usually based on access and use of (non-personal) information. In the Nokia/Navteq merger case, the Commission looked at the scope of a database on which digital maps were based to determine the geographical market, and in Thomsen/Reuters, also a merger case, the relevant market was defined as ‘discrete content sets’ (information services for finance professionals) which consisted of a significant part of data, data-feeds, real-time market data and non real-time data. In all these cases, the relevant market pursuant to competition law was ultimately the market for the goods or service in question, not for data as such.
It also seems conceivable to define ‘pure’ data markets, namely markets in which data is at the centre of supply and demand. Data is a tradable good.There are several undertakings and business models that focus on the gathering and assessing of data, for example, polling institutes. Yet, there is no single market for it. This is a direct consequence of the concept of the demand-side oriented market. For undertakings that sell car parts, personal data about the shopping behaviour of hobby gardeners seems worthless and information about a customer’s location cannot be substituted with information about her or his income.
Such ‘pure’ data markets are, even in the digital economy, the exception. The Commission has neither assumed such a market in the Google Shopping case, nor in any of the recent big data mergers. Even if certain services (and occasionally also goods) are offered in exchange for data, the general consensus seems to be that the actual services are of relevance for defining the market. For example, to use Facebook, users must agree to Facebook’s terms and conditions, which reserve Facebook broad rights with regard to the user’s (mostly) personal data. This information is then the basis for the revenue generated with advertising companies (targeted advertising). One could therefore argue that the social network part of Facebook is actually a pure data market with the user being the supplier of the data and Facebook, as customer, in demand of such data. This approach, however, would ignore that the data exchange is only a by-product of the transaction (which is aimed at allowing access to Facebook) and that qualifying the user as supplier would lead to several further competition law issues (for example, whether the user would then become an ‘undertaking’ pursuant to competition law). Along the same lines, for example, the German Federal Cartel Office has qualified the market on which Facebook is active as the ‘German market for social networks’– and not as a pure data market. If one were to pay for the use of Facebook with cash, one would also not consider that there is a market for the offering of ‘money’ in which the user is the supplier.
At first sight, such ‘free’ business models seem to be at odds with the traditional understanding of how markets work. Even in the recent Google Shopping case, Google therefore insisted that the free offering of its search engine would need to be of relevance for the competition law assessment.However, as indicated above, most services are not truly offered free of charge. The consumers simply do not pay with money. Rather, they provide their personal data. The EU Commissioner for Competition Vestager has referred to data as ‘currency’ and explicitly compared data with cash. More fundamentally, at least under competition law, in order for a service or good to belong to a certain market, a good argument can be made that remuneration is not at all necessary. If a supermarket gives away free products (e.g., because the goods are close to the expiration date) to its customers, this would still be considered a market activity. The principles of competition law would apply in the same manner as they would apply to a regular sale. This is not to say that every time a product is given away for free it always has market relevance. Of relevance is the entrepreneurial conduct behind it (potentially not given in the example above if the products were simply provided to a charity). In any event, in particular the Commission has made it quite clear that even if big data services are offered without a monetary reward, markets pursuant to competition law might very well exist. In Germany, this is now even explicitly laid down in Sec. 18(2a) of the Act against Restraints of Competition.
Another issue is still unsolved: For defining markets, the Commission frequently evaluates how customers would react to a small but significant non-transitory price increase (the SSNIP test).If such a price increase makes the customer resort to a different product or service, that product or service would be deemed to belong to the same market (as the customer sees these products or services to be interchangeable in the case of a modest price increase). It seems difficult to apply that test to the services offered by Facebook or Google. Without a monetary remuneration, there is no hypothetical price to be raised. Thus, other benchmarks are needed. Just to name an alternative, one could think about taking the terms and conditions as proxy (‘How would the user react if Google changes its terms and conditions in a particular way?’). But that test would be far more difficult to handle in practice than a price-based approach. How to best deal with such issues is still a source of constant debate.
Overall, data can play a role in defining markets but dealing with that role adequately under competition law does pose some challenges and the process of finding answers to those challenges is still ongoing.
The relevance of data for establishing dominance
Once the relevant market is defined, the question of whether an undertaking is dominant depends on an overall assessment of all relevant facts. Traditionally, market shares have been at the centre of that assessment.However, as the Commission noted, in dynamic environments ‘market shares may turn out to be ephemeral’. To make things even more complicated, it is difficult to calculate market shares for services that are seemingly offered for free, like Facebook or Google. Traditionally, market shares are either based on turnover (set in comparison to the overall volume of the market) or sold volumes. It thus does not come as a surprise that the Commission increasingly turns to other factors for its assessment.
Although the details are disputed, it is generally accepted that access to and use of data can be one factor for determining whether an undertaking is dominant.Customer data was already in the Hoffmann-Ra Roche case one (albeit minor) factor in the assessment. In Germany, the latest revision of the Act against Restraints of Competition included a provision expressly mentioning access to competitively relevant data as a factor in the assessment in multisided markets.
On the outset, it is important to note that data power (i.e. the fact that an undertaking has access to a significant amount of data) does not equal market dominance.A small data set can be far more valuable than a large data set. The assessment depends on the facts of the case at hand, in particular on whether the data poses a relevant barrier for market entrance.
In this context, the special features of data need to be considered. Most data is non-rival.This is in particular true for personal data. It can certainly be valuable – but it can also be duplicated an infinite amount of time. An undertaking might be very interested in knowing the household income of its customers, combined with their location and shopping preferences. But even if this information has already been provided to several undertakings in the past, customers are not prevented from informing additional undertakings tomorrow. This is very common. Many e-commerce shops gather similar data from their customers and many customers use more than one e-commerce website (multi-homing). Some scholars therefore even state that access to personal data cannot be of relevance (or only of very minor relevance) for the assessment of market power.
However, the mere fact that a resource is generally available does not necessarily mean that an undertaking can easily gain access to it and that, in the case of personal data, the customer is willing to duplicate it an infinite amount of times. Awareness about the importance of personal data is rising and there might come a day (and it might have already arrived for some) when people are only willing to share personal data with a limited number of undertakings. Belonging to those ‘chosen few’ could then be one factor for the assessment.Along the same lines, if it becomes increasingly difficult to gather data, this might also emphasise its importance. For example, in the Thomson/Reuters merger proceedings, the Commission noted that it would ‘be costly, time consuming and very resource intensive’ for competitors to offer data services like the merging undertakings. As a consequence, the merging undertakings had to divest copies of some of their databases to win approval from the regulator.
It further needs to be carefully assessed how long data remains valuable.Historic data about a customer’s preference is often irrelevant for advertisers. When a term is entered into an online search engine, what counts is the current preference of the customer in this very moment – not what he or she did days or even a couple of hours ago. This means that for the assessment of dominance, a (potentially outdated) large data archive is of less relevance than the ability to receive new customer data when needed.
The examples discussed previously only dealt with the access to data. This is one side of the coin. It is at least equally important what the relevant undertaking is capable of doing with the collected data.The competitive advantage is often conferred to an undertaking not by an unexploited data set but by, first, the ability to process the data and, second, to make business decisions based on the processed data.
It directly flows from the fact that it is conceivable to define pure data markets that it must, at least theoretically, also be possible to base the assessment of market dominance on access and use of data alone. In the context of big data, no such case has yet emerged.
Data might be a market entrance barrier in particular in cases of scale effects.These occur if access to large data sets (and the potential to use it) confers corresponding benefits to an undertaking and these benefits become bigger the more data becomes available. Thus, with each additional piece of data, the undertaking becomes more powerful – and the competitive gap to its rivals increases. Conversely, if multi-homing is rather common, access to data should usually not be a relevant entrance barrier.
Concerning the issue of evaluating the market position (without turnover-based benchmarks), quantitative factors are currently in the spotlight,for example, user traffic or user numbers. This also poses challenges, as not every user is of the same value for an undertaking in the sense that some might use the relevant service more often than others (and thus might also provide more data than others). Which proxy might replace turnover-based market shares is also a question that can only be answered against the backdrop of the case at hand.
In conclusion, data can be one factor relevant for determining dominance. Its importance may vary depending on the markets and services concerned. Yet, in an increasingly digitalised economy, it is to be expected that access to and use of data will gradually become more important for this assessment.
Data-related abuses under Article 102 TFEU
Having a dominant position, regardless of whether such a dominant position is founded on access to data alone or based (also) on other factors, is not in itself prohibited by competition law. Rather, certain requirements apply that set limits to the entrepreneurial freedom of such dominant undertakings. In the context of big data, several categories of potentially abusive behaviour are frequently discussed.
Excessive collection of data – ‘looting’ user data
If an undertaking reserves to itself very broad rights to gather data, this could potentially be deemed exploitative. Based on this theory of harm, the German Federal Cartel Office opened its (still ongoing) proceedings against Facebook. The underlying allegation is that Facebook’s terms and conditions allow for the gathering of data not only on the platform itself but also from third-party services (for example independent e-commerce businesses that have no connection to Facebook except that they have incorporated the ‘Like’ button on their homepages). This data is then combined with the already available information about the individual user, allowing for an even deeper insight. The Federal Cartel Office in particular has made the criticism that users are not aware that these third-party services transfer data to Facebook even if the users do not actively engage with the homepage or the ‘Like’ button.
Tackling a potentially excessive thirst for data is only seemingly an unexplored path for competition law. The setup might be different and arguably more complex – but ultimately it might just be a case of dealing with unfair trading conditions.
The Commission stated in the Facebook/WhatsApp merger decision that ‘[a]ny privacy-related concerns flowing from the increased concentration of data . . . do not fall within the scope of the EU competition law rules but within the scope of the EU data protection rules.’This concern has been echoed by scholars who argue that data protection laws should be the sole tool to address privacy concern. It seems doubtful whether each violation of the EU Data Protection Regulation (and more specifically its national implementing provisions), for example, by imposing unlawful terms and conditions, equals a violation of Article 102 TFEU. After all, data protection law has a different aim than competition law. While the first is driven by privacy concerns, the latter is expected to secure market-driven competition. This is, however, not to say that there cannot be a case where both fields of law are concerned.
Some recent statements of Commissioner Vestager reveal a more nuanced position. While she noted that on the one hand competition enforcement should not be expected to fix privacy problems, she also agreed that there might be ‘some truth’ in the allegation that failure to protect personal data might be a sign that markets do not work properly.After all, there is a significant difference between data protection law and competition law. While data protection law in theory allows for the sharing and use of data as long as the concerned person agreed to the conduct in question, such consent is irrelevant for the assessment under Article 102 TFEU.
This raises the question of when the threshold is crossed to make excessive data gathering a competition law issue. Currently, there is no general consensus and it seems that several obstacles would need to be passed to establish an abusive behaviour.It is still up for debate whether there must be causality between the dominant position and the abuse, namely the imposing of the potentially abusive terms and conditions. In any event, it is clear that not every legal violation by a dominant undertaking automatically results in an infringement of Article 102 TFEU, for example in case of a breach of labour or environmental law. In previous decisions, the Commission and the EU courts have assessed whether the terms and conditions in question were necessary for achieving the contract’s objective and also proportionate. This inter alia raises the question: is the compensation (the personal data shared) adequate for the service offered (the social network)? This is difficult to answer as one can neither directly put a price tag on the personal data, nor on the ‘value’ of the service offered by Facebook (which might be very high for some users).
Price or product discrimination
Access to data could also be used to offer different prices for customers or even entirely different products and services to distinct customer groups. An undertaking could offer cleaning services to customers living in areas where people are generally earning above average incomes at a higher price than to customers domiciled in low-income areas.
Such price differentiation is not new. Already today, certain offers are directed only at certain customer groups. For example, Microsoft offers Word to students, pupils and teaching staff under special conditions and has distinct versions for private consumers and businesses. And the different average household incomes in, for example, Norway and Turkey is one of the reasons why certain goods are more expensive in the one country than in the other.
In certain industries, professional customers are presumably very used to being charged differently from their competitors, simply because there are no ‘catalogue prices’ for the goods in question and several factors influence the daily trading price (for example, in the steel industry). In comparison, end consumers (located in the same area) are accustomed to pay the same price at least for standardised goods.It is difficult to foresee how such end customers would react to customer-specific prices for goods and services. Think, for example, about a customer finding out at the supermarket checkout that they pay 20 per cent more for vegetables than her or his next-door neighbour simply because the supermarket analysed the household income. In today’s information society, regardless of how good the respective algorithms are, it seems unlikely that a large-scale ‘hidden’ price differentiation could stay undetected for a long period of time. It can be assumed that such price discrimination would not sit well with most customers because nobody likes the feeling of being overcharged.
Relying on this sentiment alone (and thereby the hope that the market will regulate itself) seems naive.If an undertaking is dominant (and only then would a price or product discrimination become a problem under competition law), the customer would not have many options to switch to competitors. Arguably, it might make a difference for the assessment whether the undertaking has a market share of just above 50 per cent or whether it is the only supplier in the market. Nonetheless, the more relevant the goods or services are for the individual customer, the more difficult it would become to evade such price discrimination.
Exclusionary conduct and data as ‘essential facility’
Exclusionary behaviour could potentially come up in two different scenarios. Either an undertaking is hindered in collecting the relevant data itself directly from the source (first party data) or it cannot access data that has already been collected (third party data).Access to first party data could be restricted if an undertaking concluded exclusivity arrangements with its suppliers or customers. If, for example, a dominant social network required its users not to sign up with any other social network, competitors could not collect the relevant personal data on their own. Practically, exclusivity could become an issue if data can only be gathered from limited sources (for example very specific product-related data). For example, the IMS Health case dealt with access to third party data, namely a database necessary to offer reports to the pharmaceutical industry.
The question whether a behaviour has exclusionary effect would need to be assessed on a case-by-case basis. Moreover, particularly in the digital economy, one would need to assess carefully whether the data is necessary to be active on the market. In the past, several new entrants have attacked and ultimately overturned seemingly untouchable market leaders. Before Google, there was Yahoo!; before Facebook, there was MySpace.While both services still exist (partly with a different offering than before), their market relevance has significantly declined. This was done without Google or Facebook having access to the data of their rivals.
This is not to say that big data is always contestable and that market leaders are constantly exposed to the danger of being replaced in a matter of seconds. Google+ entered the market for social networks with the support of one of the most proficient big data undertakings ever – and failed to threaten even remotely the position of Facebook.Along the same lines, during the past decade we have not seen an operating system for personal computers emerge that even comes close to the popularity of Microsoft’s Windows. In a joint paper, the German and French competition authorities stated that it might very well be that some undertakings have such a large base of information and customers (and thus such a significant advantage over potential competitors) that new entrants would be unable to challenge their market position on their own. Thus, while the existence of potential competitors might be increasingly relevant for data-driven innovation markets, it is by far not certain that such potential competitors exist simply because the market is deemed to be ‘dynamic’.
A related question in this context is whether a dominant undertaking could be forced to grant access to its data. This would be the case if the data were deemed to be an ‘essential facility’. Dominant undertakings must grant access to competitors if this is indispensable for entering a downstream market. The Commission has already found a database to be an ‘essential facility’, namely in the IMS Health case,although arguably the facts were special. Considering that most data is non-rival and often not strictly necessary to enter a certain market, one would expect the threshold for a successful access claim to be rather high. If personal data is involved, one would also need to align the requirements of data protection and competition law.
Cross-usage of data
When undertakings offer more than one service, combining the data gathered through these services allows an even more detailed analysis of the customers.One example is Google that, next to its popular search engine, offers a wide variety of services touching very different areas of daily life (from the mobile phone operating system Android, cloud migration services, virtual reality studios and voice communication providers). For the purposes of merger control law, it seems possible that the fusion of different data sets (by external growth) might give rise to competitive concerns, although, as mentioned, the Commission has been rather cautious in this regard. Under Article 102 TFEU, the thresholds for tackling the cross-usage of data are rather high. Traditionally, competition law does not prohibit the cross-subsidising of different business units within one group of undertakings. Along the same lines, cross-usage of data (which is ultimately a cross-subsidising) should in itself not be an issue under competition law.
Tying and bundling
Like with any physical product or service, data could potentially be used for tying and bundling practices. The French Competition Authority dealt with a case in which an undertaking operated a database (with the names and addresses of medical doctors) for medical sales representatives.The undertaking refused to sell the database to laboratories that used customer management software from a specific competitor (although it did offer the database to laboratories that used software from other competitors). As the undertaking treated customers differently, depending on the software used, the French Competition Authority ultimately found an abusively discriminatory treatment. However, the case also had aspects of a tying and bundling situation, showing that such abuses are also possible in the digital world.
Data and consumer welfare – an outlook
Much of this chapter has focused on the interplay between data, dominance and abusive behaviours. It is, therefore, noteworthy to also remember the immense benefit that the (advanced)access to and use of data has provided. There is a wide variety of data-driven innovations.One should not take for granted that complex services such as Facebook or even certain mobile games are offered without monetary compensation. Google’s ability to process data has presumably led to significantly more precise search results – the same benefit has been achieved by certain dating platforms. Targeted advertising has helped to bring customers and suppliers together. Price comparison services (and also marketplaces such as Amazon) created an unprecedented transparency with regard to product pricing and quality that has not only helped consumers in their decision-making process but has also enabled the market entrances of suppliers and manufacturers (by using some of these marketplaces, for example). Price discrimination might enable certain customer groups to enjoy products or services that would have otherwise been unaffordable for them.
When it comes down to the assessment of whether certain data-related business behaviour is abusive, there is a lot to consider. When big data is discussed in the public domain, these consumer welfare aspects are often forgotten. But these benefits need to be reflected in a thorough assessment under Article 102 TFEU – even if they are occasionally difficult to grasp and to value economically. On the one hand, competition law should not be scared to tackle data-related abuses – on the other hand, competition law must make sure that the overall picture is taken into account to warrant its application.
The initial questions can be answered in the affirmative. Knowledge is indeed power and, as is often the case, with that power comes the risk of an abuse. As far as data is concerned, this is not a novel finding. Yet, big data has given this topic significantly more relevance. Some issues that come up in this context have been discussed above. Only a few principles are settled and scholars and competition authorities are well advised to closely monitor the developments to come. It is certainly possible that data-related abuses of dominance may play a more important role in the future. In general, Article 102 TFEU should adequately deal with such potentially anticompetitive behaviour. The actual assessment can only be done on a case-by-case basis, recognising not only the special features of data but also balancing the immense benefits that the digital economy has brought against any countervailing effects.
 Dr Thorsten Mäger is a partner and Dr Philipp Otto Neideck is an associate at the German law firm Hengeler Mueller. Both are based in in the firm’s Duesseldorf office. All hyperlinks were last accessed 9 October 2018.
 See for example the German Monopolies Commission, Biennial Report XX (2014), p. 58 et seqq. (Current Problems of Competition Policy)). The full report (in German) can be downloaded at https://www.monopolkommission.de/images/PDF/HG/HG20/HG_XX_gesamt.pdf.
 Davilla, Journal of European Competition Law & Practice 2017, 370 (371).
 For the remainder of this chapter, the term ‘competition law’ is used.
 A similar question has been raised, for example, by Körber, NZKart 2016, 303 (305).
 Treaty on the Functioning of the European Union, OJ C 326, 26 October 2012, p. 47 et seqq. For the sake of completeness, it has to be noted that certain national competition laws, like, for example, in Germany, have provisions that also impose obligations on undertakings that do not fulfil the dominance requirement but have otherwise obtained a significant market position.
 Commission, Decision of 27 June 2017, AT.39740, para. 264 – Google Shopping; European Court of Justice (CJEU), Decision of 13 February 1979, Case 85/76, ECLI:EU:C:1979:36, Court Reports 1979, 461, para. 38 – Hoffmann-La Roche.
 CJEU, Decision of 13 February 1979, Case 85/76, ECLI:EU:C:1979:36, Court Reports 1979, 461, para. 21 – Hoffmann-La Roche.
 Commission, Decision of 27 June 2017, AT.39740, para. 147 – Google Shopping; see already CJEU, Decision of 21 February 1973, Case 6/72, ECLI:EU:C:1973:22, Court Reports 1973, 215 para. 32 – Continental Can.
 Commission, Decision of 27 June 2017, AT.39740, para. 148 – Google Shopping; see already CJEU, Decision of 14 February 1978, Case 27/76, ECLI:EU:C:1978:22, Court Reports 1978, 207, para. 44 – United Brands.
 Podszun, Die 9. GWB Novelle, Kersting/Podszun (Eds.), 1 (12).
 See for example the explanations at http://ec.europa.eu/competition/antitrust/procedures_102_en.html.
 One such example is the assessment of upward pricing pressure (UPP test), Commission, Decision of 2 July 2014, Comp/M. 7018, Non-technical summary, para. 28 et seqq. – Telefónica Deutschland/E-Plus.
 Davilla, Journal of European Competition Law & Practice 2017, 370 (371).
 French Competition Authority and German Federal Cartel Office, Competition Law and Data, p. 4; available at www.bundeskartellamt.de/SharedDocs/Publikation/DE/Berichte/Big%20Data%20Papier.pdf?__blob=publicationFile&v=2; Nuys, WuW 2016, 512 (513); Zech, Daten und Wettbewerb in der digitalen Ökonomie, Körber/Immenga (Eds.), 31 (39).
 Tamke, ZWeR 2017, 358 (359).
 CJEU, Decision of 13 February 1979, Case 85/76, ECLI:EU:C:1979:36, Court Reports 1979, 461, 467, 522 and para. 48 – Hoffmann-La Roche.
 CJEU, Decision of 29 April 2004, Case C-418/01, ECLI:EU:C:2004:257, Court Reports 2004, 5039 et seqq. – IMS Health GmbH & Co OHG.
 See, for example Colangelo/Maggiolino, Journal of European Competition Law & Practice 2017, 363 et seqq.
 Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation).
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 8 et seqq.
 Capobianco/Nyeso, Journal of Competition Law & Practice 2018, 19 (22).
 A different view is expressed by Colangelo/Maggiolino, Journal of European Competition Law & Practice 2017, 363.
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 5 et seqq; Körber, NZKart 2016, 303 (304); Tamke, ZWeR 2017, 358 (363).
 Federal Cartel Office, case report dated 27 March 2018, ‘Establishing an e-commerce platform for steel products’, available at www.bundeskartellamt.de/SharedDocs/Entscheidung/DE/Fallberichte/Kartellverbot/2018/B5-1-18-01.pdf?__blob=publicationFile&v=4.
 This was discussed in the opinion and order dated 31 March 2016 by Judge Jed S. Rakoff in Case 15 Civ. 9796, Court of the Southern District of New York, Meyer/Kalanick (2016).
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 4; Schepp/Wambach, Journal of European Competition Law & Practice 2016, 120.
 Schepp/Wambach, Journal of European Competition Law & Practice 2016, 120; Nuys, WuW 2016, 512, Tamke, ZWeR 2017, 358 (359, 362).
 Commission, Decision of 23 August 2013, Comp/M. 6922, para.13 et seqq. – Triton/Logstor.
 Commission, Decision of 2 August 2008, Comp/M. 4942, para. 56 – Nokia/Navteq.
 Commission, Decision of 19 February 2009, Comp/M. 4726, para. 43 – Thomson/Reuters.
 Tamke, ZWeR 2017, 358 (364, fn. 38).
 Telle, InTeR 2018, 3 (6).
 Körber, NZKart 2016, 303 (304).
 Körber, NZKart 2016, 303 (304); Schepp/Wambach, Journal of European Competition Law & Practice 2016, 120 (121).
 Last example taken from Kaben, Daten und Wettbewerb in der digitalen Ökonomie, Körber/Immenga (Eds.), 123 (128).
 Federal Cartel Office, Press statement from 19 December 2017, ‘Preliminary assessment in Facebook proceeding’, available at www.bundeskartellamt.de/SharedDocs/Meldung/EN/Pressemitteilungen/2017/19_12_2017_Facebook.html;jsessionid=6A4A488BEE8FC96AD7BE65BA646EC20E.2_cid362?nn=3591568.
 Commission, Decision of 27 June 2017, AT.39740, para. 319 et seqq. – Google Shopping.
 Dewenter/Lüth, Daten und Wettbewerb in der digitalen Ökonomie, Körber/Immenga (Eds.), 9 (25).
 Vestager, ‘Competition in a big data world’, available at https://ec.europa.eu/commission/commissioners/2014-2019/vestager/announcements/competition-big-data-world_en.
 A different question then obviously is whether the behaviour does indeed infringe competition law.
 Commission, Decision of 27 June 2017, AT.39740, para. 157 – Google Shopping; also Court of the First Instance, Decision of 17 September 2007, Case T-201/04, EU:T:2007:289, Court Reports 2007 II, 3601 para. 966, et seqq. – Microsoft.
 See, for example, Commission, Decision of 26 October 2004, para. 86, para. 102 – Oracle/Peoplesoft.
 Capobianco/Nyeso, Journal of Competition Law & Practice 2018, 19 (23); Dewenter/Rösch/Terschüren, NZKart 2014, 387 (390); Schepp/Wambach, Journal of European Competition Law & Practice 2016, 120 (124).
 Podszun, Die 9. GWB Novelle, Kersting/Podszun (Eds.), 1 (10).
 See, for example, Commission, Decision of 27 June 2017, AT.39740, para. 266 – Google Shopping.
 Commission, Decision of 3 October 2014, Case Comp/M. 7217, para. 99 – Facebook/WhatsApp.
 Podszun, Die 9. GWB Novelle, Kersting/Podszun (Eds.), 1 (13).
 See also Tamke, ZWeR 2017, 358 (364).
 Nuys, WuW 2016, 512 (515).
 Sec. 18(3a).
 Körber, NZKart 2016, 303 (305); Louven, NZKart 2018, 217 (220); Nuys, WuW 2016, 512 (515); Telle, InTeR 2018, 3 (5).
 ‘Non-rival goods’ may used by one user without preventing others from using it, e.g. movies, designs and algorithms.
 Telle, InTeR 2018, 3 (6).
 Nuys, WuW 2016, 512 (513); Kaben, Daten und Wettbewerb in der digitalen Ökonomie, Körber/Immenga (Eds.), 123 (125); Telle, InTeR 2018, 3 (5).
 Nuys, WuW 2016, 512 (516); Schepp/Wambach, Journal of European Competition Law & Practice 2016, 120 (121).
 Kaben, Daten und Wettbewerb in der digitalen Ökonomie, Körber/Immenga (Eds.), 123 (125).
 Louven, NZKart 2018, 217 (220); Davilla, Journal of European Competition Law & Practice 2017, 370 (378).
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 38.
 Commission, Decision of 19 February 2009, Comp/M. 4726, para. 261 – Thomson/Reuters.
 Nuys, WuW 2016, 512 (513); Dewenter/Lüth, Daten und Wettbewerb in der digitalen Ökonomie, Körber/Immenga (Eds.), 9 (20).
 Dewenter/Lüth, Daten und Wettbewerb in der digitalen Ökonomie, Körber/Immenga (Eds.), 9 (20).
 Nuys, WuW 2016, 512 (515).
 Kaben, Daten und Wettbewerb in der digitalen Ökonomie, Körber/Immenga (Eds.), 123 (130); similar also Schepp/Wambach, Journal of European Competition Law & Practice 2016, 120 (122).
 Dewenter/Lüth, Daten und Wettbewerb in der digitalen Ökonomie, Körber/Immenga (Eds.), 9 (21); also French Competition Authority and German Federal Cartel Office (fn. 15) p. 13.
 Tamke, ZWeR 2017, 358 (368).
 Grave/Nyberg, WuW 2017, 363 (364).
 Telle, InTeR 2018, 3 (5).
 German Federal Cartel Office, Press statement of 19 December 2017 (fn. 38).
 See Schneider, Journal of European Competition Law & Practice 2018, 213 (215, 221); Tamke, ZWeR 2017, 358 (375).
 Commission, Decision of 3 October 2014, Case Comp/M. 7217, para. 164 – Facebook/WhatsApp.
 Davilla, Journal of European Competition Law & Practice 2017, 370 (381).
 Similar Schneider, Journal of European Competition Law & Practice 2018, 213 (222); Körber, NZKart 2016, 348 (356).
 Colangelo/Maggiolino, Journal of European Competition Law & Practice 2017, 363 (367).
 Schepp/Wambach, Journal of European Competition Law & Practice 2016, 120 (123).
 Vestager (fn. 41).
 Nothdurft, Kartellrecht, 13th Edition, Langen/Bunte (Eds.) § 19 GWB, para. 192.
 A summary is provided by Schneider, Journal of European Competition Law & Practice 2018, 213 (222 et seqq.).
 Nuys, WuW 2016, 512 (519).
 CJEU, Decision of 27 March 1974, Case 127/73, ECLI:EU:C:1974:25, Court Reports 1974, 313, para. 15 – BRT II (‘not absolutely necessary for the attainment of its object’); Commission, Decision of 20 April 2001, D3/34493, para. 111 – DSD (‘the price charged for a service is clearly disproportionate to the cost of supplying’). The decision was later confirmed both by the Court of First Instance (T-151/01, see para. 121) as well as the CJEU (Case C-385/07).
 Telle, InTeR 2018, 3 (7, 10).
 See also French Competition Authority and German Federal Cartel Office (fn. 15) p. 10 et seqq.
 Körber, NZKart 2016, 303 (308).
 Körber, NZKart 2016, 303 (308).
 This is the approach suggested by Körber, NZKart 2016, 303 (308).
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 21, fn. 46; Dewenter/Lüth, Daten und Wettbewerb in der digitalen Ökonomie, Körber/Immenga (Eds.), 9 (17 et seqq.).
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 12; Tamke, ZWeR 2017, 358 (366).
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 19 et seqq; Körber, NZKart 2016, 303 (309).
 CJEU, Decision of 29 April 2004, Case C-418/01, ECLI:EU:C:2004:257, Court Reports 2004, 5039 et seqq. – IMS Health GmbH & Co. OHG.
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 29.
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 30.
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 12, 30.
 Tamke, ZWeR 2017, 358 (370); Klotz, WuW 2016, 58 (63).
 CJEU, Decision of 29 April 2004, Case C-418/01, ECLI:EU:C:2004:257, Court Reports 2004, 5039 et seqq. – IMS Health GmbH & Co. OHG.
 Nuys, WuW 2016, 512 (517); Schepp/Wambach, Journal of European Competition Law & Practice 2016, 120 (123); Tamke, ZWeR 2017, 358 (371).
 Nuys, WuW 2016, 512 (517).
 Nuys, WuW 2016, 512 (514); Schepp/Wambach, Journal of European Competition Law & Practice 2016, 120 (121).
 French Competition Authority, Decision of 8 July 2014, 14-D-06 – Cegedim. The English press release it available at www.autoritedelaconcurrence.fr/user/standard.php?lang=en&id_rub=592&id_article=2403.
 Capobianco/Nyeso, Journal of Competition Law & Practice 2018, 21.
 Körber, NZKart 2016, 303 (307).
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 9 et seqq.
 Commission, Decision of 11 March 2008, Comp/M.4731, para. 360 – Google/DoubleClick.
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 14.
 French Competition Authority and German Federal Cartel Office (fn. 15) p. 22.