In Internet of Things (IoT) ecosystems, one firm often has exclusive control over the data produced by a smart device, as well as of the technical means of access to this device. Such a gatekeeper position can empower firms to eliminate competition for aftermarket and other complementary services in these ecosystems.
This paper, available here, analyses whether competition law can help address problems concerning access to data and interoperability in this context, by reference to connected vehicles. In short, it argues that, while competition offers some solutions to these data access problems, on its own it is insufficient to fully address these problems. As such, additional solutions such as data portability requirements, data access rights or sector-specific regulation might also be needed.
Section II provides a brief overview of the economics of digital ecosystems and of data interoperability.
Data tends to be non-rivalrous in use. It follows that data should be used as much as possible to maximise its value. In many IoT contexts, however, one firm often has exclusive control over the data produced by a device. This gives rise to holdup with regard to other stakeholders, to data access being subject to excessively high prices, and to data being inefficiently underused.
In addition, the same firms that control device data usually also decide on the degree of interoperability of their devices. Important results of the economics of interoperability are that: (1) increased interoperability leads to benefits in terms of more competition and innovation in markets for complementary products and services, and to more choice for users; but (2) closed proprietary systems also have advantages, especially as regards quality and product differentiation.
It follows that two main market failures can emerge in IoT ecosystems as regards data: (a) exclusive control over data, and over technical access to this data and to the device producing it, can lead to reduced competition and innovation in aftermarkets and markets for complementary products; (b) companies may adopt a level of interoperability which is too low from a societal welfare standpoint. Both market failures can produce negative effects for consumers in the form of higher prices, less innovation, reduced consumer choice and underuse of available data.
Companies that control data and how to access it can act as monopolistic gatekeepers to entire ecosystems of services and products. Inasmuch as access to data is necessary for entering and competing in a market, monopolistic gatekeepers can leverage their position across all markets that depend on this access, and can therefore control such markets by means of bundling their offerings across them. However, it is not clear whether closed ecosystems and bundling strategies harm consumers, even if some companies are foreclosed in these secondary markets. Whether such conduct is anticompetitive ultimately depends on whether systems competition among the firms controlling the relevant ecosystems works well enough to act as a substitute for the absence of competition in aftermarkets and markets for complementary services. Therefore, a careful case-specific analysis is required to determine whether control over data and how to access it are anticompetitive.
Section III looks at competition and market failures in connected driving ecosystems.
Connected cars collect, produce and process huge amount of different kinds of data. This data can be transmitted in real time to and from the car, allowing for the provision of services such as repair and maintenance, navigation, parking, hotel and restaurant search, and insurance, among others. The connected driving ecosystem therefore encompasses a large number of complementary services and aftermarkets. This ecosystem can be contrasted with the “extended vehicle concept”, under which manufacturers enjoy exclusive control over connected vehicles’ data and are able to restrict the provision of services by third-parties in complementary markets. This concept endows manufacturers with a gatekeeper position with respect to complementary services and aftermarkets, and locks-in car owners to the manufacturers’ service offerings.
Each of these approaches to connected cars has staunch defenders, who have engaged in an intense policy debate in Europe regarding the regulation of access to connected vehicles and their data. The author describes this debate – and it is important to point out that the author, here and in previous work, has come out for the opening of such data and complementary markets to third parties via interoperability schemes or standards.
Section IV discusses to what extent European (and German) competition law can be to address data access constraints in digital ecosystems.
The literature has shown marked scepticism about whether competition law is suitable for dealing with data access problems, particularly in light of the stringent requirements which a successful refusal to supply (data) case would have to meet. The author, on the other hand, is an optimist.
He considers that, under EU and German law, access to exclusively owned data by third parties is a natural remedy against dominant companies not only under the “essential facility doctrine”, but also in situations where the practical refusal to grant access to data can foreclose independent competitors on aftermarkets and other complementary markets within an ecosystem. This can encompass well-known strategies such as bundling, exclusivity agreements, or other predatory strategies that make it harder or even impossible for others to get access to data. Another set of strategies might consist of designing technically closed ecosystems that grant a firm exclusive technical control over a device and makes interoperability impossible.
Furthermore, recent policy and legislative proposals in Germany suggest that data access issues could also be addressed by controlling abusive behaviour even when a company is not dominant. Under German competition law, a firm with “relative market power” is not allowed to abuse its power vis-à-vis small- or medium-sized companies, particularly when such companies do not have sufficient and reasonable possibilities of switching to other firms. More recently, a policy-making report recommended that such provisions should be used to help firms that are dependent on specific platforms or have problems getting access to data from other firms (including in IoT and aftermarket contexts), particularly when these firms are faced with ‘unfair bargaining positions’. The author considers that economic theory supports such proposals. While economic analysis is sceptical of approaches to market power that focus on individual relationships between firms, there are two exceptions to this. The first is buyer power (e.g. by large retailers), where it is broadly accepted that bilateral bargaining power plays a crucial role for competition. The second is holdup, where a firm can become dependent on another due to transaction-specific investments (e.g. sunk costs) leading to lock in. These are typically the situations to which the provision on ‘relative market power’ has been applied.
Competition law can also intervene against the use of horizontal or vertical agreements whereby firms either gain exclusive control over data or defend such an exclusive position, such as agreements between manufacturers of connected devices on technological parameters that restrict competition by creating or protecting exclusive gatekeeper positions as regard ecosystems and related data. Such agreements can lead not only to technological collusion (closed systems/lacking interoperability), but also to collusion regarding data governance. The author provides a detailed example of this by reference to an agreement by car manufacturers to establish the ownership of connected car data by each manufacturer.
Yet another way in which competition law can address data access concerns is by preventing a firm from getting into a position of exclusive control over certain data sets. This can be achieved by prohibiting mergers that may create monopolistic positions with regard to certain non-replicable or non-substitutable data sets. Merger control can be (and in fact already is) used for preventing the emergence of exclusive monopolistic control over data sets that can impede competition in other (up/downstream or adjacent) markets.
Section V considers other possible solutions for addressing issues related to data-sharing and data access in digital ecosystems.
Despite competition law offering a number of options for solving data access and data-sharing problems in IoT ecosystems, one should keep in mind that other legal instruments can also be used to this end. One solution can be found in proposals to introduce data rights, either as property-like exclusive rights (“data producer right”) or as access to data rights. Another solution builds on data portability rights. A third group of solutions asks whether contract law solutions (and unfair trading law) can be used to deal with these data governance problems. Another possibility is to adopt sector-specific regulation. Policy efforts to promote standardisation (for solving interoperability problems) and, in some contexts, sector specific regulations which allow for more tailor-made solutions are particularly interesting – and apposite for connected cars.
All these solutions have their own problems. From an economic perspective, however, the necessary balancing between the benefits and costs of different data governance architectures, and of closed or open (and interoperable) systems, will be similar whatever policy solutions and legal instruments are used. Given the different economic and technological conditions that prevail in different economic sectors, it can be expected that different legal and regulatory solutions will be most appropriate for different IoT ecosystems.
This is an insightful and relevant paper, even if I believe it could have been shorter and more to the point. I particularly enjoyed its discussion of how EU and German competition law can apply to data access issues, and the discussion of regulatory alternatives.
On the other hand, I have the impression that the main objective of the paper is to develop a general theory of when access to data can legitimately be imposed in digital ecosystems. From this perspective, the references to connected vehicles should have been incidental, and served mainly to provide examples (as I do in my review). Instead, the paper seems to want to discuss both topics comprehensively. This makes the paper longer, and the argument less focused, than would otherwise have been the case.
The focus on connected carts seems to have led the author not to engage with the fact that many of the issues he identifies have long been present in discussions of complementary product markets and aftermarkets. Looking at such debates could have proved more conducive to the development of a general approach to data access issues. A failure to engage with such broader questions also leads the author to arrive at conclusions (e.g. what the relevant market for complementary products should be, even when there is competition) that, to my mind, require a case-specific empirical analysis.
The focus on a specific industry application of IoT also leads to some odd dead-ends, since it is clear that the principles governing data access for connected cars do not fully align with the principles that should govern data access in other sectors – or even that should apply to all connected cars. For example, the author notes that, despite what he describes as widespread support for his preference for interoperability and opening of different car brand ecosystems to third parties, one of the reasons why no regulation to this effect has yet been adopted is that in-vehicle data is heterogeneous. Data is not only personal and non-personal, but there are also distinctions between raw and processed/aggregated data, data about technical functions of the (components of the) car and about traffic, road and weather conditions, and so forth. This heterogeneity means that the optimal data governance solutions might be different for different types of data.
This reflects the main issue with recent proposals to lower the standards governing refusal of access to data under competition law (as discussed in both of today’s papers, but also in a number of recent reports) or to adopt ex ante interoperability/data access rules: how are we to do this in practice? It is unclear what data we are talking about (rough, structured, unstructured, at which level of structure or analysis, whether it contains business secrets or incorporates analysis by the data holder, etc.), and what standards should govern data interoperability schemes. Just to provide an example, the author (and many others) assume that data is not protected by IP rights. However, if the data to which access will be granted is structured, and actually amounts to a database, it may well be protected by IP rights, which may change the whole calculus of whether to grant access, let alone how. Further, it is likely that generic interoperability duties will require data sharing standards that will have to be developed on a case-by-case (or, at best, sector-by-sector) basis. This is likely to prove difficult, onerous and fertile with possibilities for conflict and delay.
To be clear, this is not to say such measures are unrealistic or impractical – I genuinely do not know enough about the topic to have an opinion one way or the other. However, I have not seen this dimension being discussed, and the work I have read in this regard – including this OECD report on Enhancing Access to and Sharing of Data – makes it clear that it is an important one.