Data and E-commerce: An Economic Relationship

The digital platforms use the Internet to mediate economically advantageous transactions between different user groups. Although the business model of digital platforms is widely discussed and analyzed, a generally accepted definition has so far not been successful. Digital platforms are quasi marketplaces that clearly stand out from pure online shops or online retailers that serve one instead of several different user groups. In summary, the role of e-commerce platforms among online-active companies and in e-commerce as a whole is significant. Online platforms have a function in the digital world that is not perceived by other actors. E.g. Amazon is one of the largest e-commerce platforms worldwide. Due to its importance, Amazon should also have the most extensive collection of customer and transaction data, which makes the company particularly relevant for data sharing. In this context, the following challenges arise with regard to an assessment of the market shares, which are important when analyzing e-commerce platforms: The platform activities are not always clearly differentiated from other activities. The best example represents Amazon, which provides a platform with its marketplace, but also operates as a retailer in the retail sector. Amazon can be compared with eBay. At the same time, it is also conceivable to compare the Amazon Books division with other pure book platforms. Empirically we can separate the market share of the marketplace from that of the retail sector. Ultimately, Amazon Retail acts like another, albeit very large, retailer on the marketplace. This means that the share of sales from the Amazon retail sector can also be attributed to the marketplace. Even if Amazon’s retail sector is so different from the players in the marketplace, the Amazon marketplace has a large market share among e-commerce platforms. In connection with this, the available data provide no evidence that there is an e-commerce platform that has a similarly broad product portfolio and a similar reach to Amazon and could therefore be seen as a substitute to it (LaVecchia, Mitchell, 2016). However, this can be quite different in the individual product categories, which means that this conclusion does not have to apply to all markets in which Amazon is active.

Digital platforms have special properties that also have a direct impact on the dynamics of the markets in which they operate. These characteristics of digital platforms are directly related to their definition, which is by no means uncontroversial. It is particularly important in this context that digital platforms always have at least two different user groups. The platform itself is not a transaction partner, but merely mediates the transaction. Platforms do not have any inventory, but only an infrastructure. An example represents Amazon Marketplace. The two user groups of the platformare retailers and end customers. Amazon Retail itself is not a platform. As a retailer, Amazon itself holds the inventory of goods. In this case, the only user group is the end customer. Ultimately, Amazon Retail is one of many retailers on the Amazon marketplace. Digital platforms are therefore always multi-sided (Šmejkal, 2016) and serve multiple user groups. This means that the definition of the relevant market for digital platforms is not always easy. This ultimately depends on the specific business model. In addition to digital platforms that are only active in one market, there are also platforms whose activities are relevant in several markets.

An essential characteristic of digital platforms is the existence of positive network effects. This means that every additional user of a platform is advantageous for the other users of the platform (Funta, 2020b). Each additional retailer, for example on an e-commerce platform, has potentially positive effects on the (potential) customers on the other hand, because the selection and price competition increase (Šmejkal, 2012). Every additional customer creates positive effect because the potential demand for the product increases. In addition to the classic network effects, so-called data-driven network effects can be identified, which play a major role for many digital platforms. Network effects have the potential to accelerate the rapid growth of digital platforms. A larger number of users make the platform attractive to other users, which in turn leads to further growth. The effect is self-reinforcing (referred to as the “positive feedback effect”). Successful digital platforms are naturally rather large in terms of their number of users. In order for such a business model to achieve a full potential, it is important to reach a critical mass of users (“critical mass frontier”) in the different user groups. Once this critical mass is reached, positive feedback can be used to grow. However, without the critical mass of users, the platform is not stable. Many digital platforms try to reach the critical mass via a corresponding pricing policy. First of all, it should be noted that, in particular, free services on a platform pose challenges for precise market definition, since, for example, conventional methods for determining market power (Karas, Králik, 2012) that are based on pricing options are not applicable. In this case, the question arises, for example, whether the individual market sides need to be examined individually or as a whole with regard to market power. In addition to defining the relevant market, other properties of digital platforms are important for the market power like economies of scale, multi-homing or exchange costs (Funta, Golovko, Juriš, 2020a).

Positive indirect network effects often entail so-called “market tipping”. The mutually reinforcing indirect network effects bring about a rapid increase in the market power of the platform. As a result, markets with digital platforms are often characterized by a small number of large companies (often also monopolies). Market tipping leads to so-called “winner takes all markets”. This means that these markets are prone to tipping (Zingales, Lancieri, 2019). In such markets, the market share of the largest company is larger than the share of the other smaller competitors. Competition in digital platforms can therefore take place in part for the market, i.e. for reaching critical mass. Winner-takes-all markets and natural monopolies are usually not clearly differentiated in economic literature. The monopolistic structure of the market is common to both concepts. The difference lies in the reason for the existence of such a market structure: winner-takes-all markets occur primarily in connection with digital platforms. The basic requirement is that the platform is able to attract and retain users. Winner-takes-all markets also depend on demand. Even in winner-takes-all markets, the platform’s market power does not necessarily have to be long-term.

The economic analysis makes it clear that data sharing offers advantages, which are examined in more detail below.

The non-rivalry of the data means, that data are non-rival in consumption and have the ability to be replicated almost free of charge. Especially in connection with the organizational structure of digital platforms, there is the potential for exclusive access to data to secure a dominant position. This can be the case in particular if the data record is not easily replicable, access to the data record is only possible for the dominant company and the quality of the information (Peráček, 2014) derived from the data record is high. The data set then becomes another determining factor for the extent of market power. The question arises whether the market can still be contested in such a case and whether the powerful company is misusing its market power. Sharing the specific data set could be a way to (again) enable competition in the event of abuse of market power (Schweitzer, Haucap, Kerber, Welker, 2018).

Data may be seen as a decisive factor for market power. The background is that every company basically has the opportunity to collect data. At the same time, there is a need to actually convert data into information or valuable products or services (Stehlík, Hamuľák, 2013). A larger amount of data does not automatically lead to better results or greater added value. It remains unclear whether access to data in other constellations such as a platform that holds the data and the platform users would not play a decisive role. It is obvious that in such a constellation, the data generated by the platform could not easily be collected by users in a comparable scope and quality. For example, although a retailer on Amazon Marketplace has access to data related to own sales, Amazon Marketplace (and thus also Amazon Retail) can access data on the sales of thousands of retailers and on (potential) customers (Daňko, Mesarčík, 2015). Depending on the application, the value of data also depends on how current is the respective data record. Even if a platform such as Amazon Marketplace provides data directly to individual retailers, it is conceivable that the smallest technical delays will occur in data transfer. This can lead to a reduction in the usability and thus the value of the data. Even access to the data would not reduce the competitive advantage of the platform operator due to the slightly delayed delivery. This aspect can, however, be technically solved by enabling all actors to have access to the data at the same time. If retailers get fast data access on Amazon Marketplace, but can only adjust their prices manually, for example, pure data access is not sufficient to generate benefits, at least not in the same way as the platform itself.

Data alone can be hardly used. For small companies in particular, this represents an enormous hurdle in terms of financing and know-how, so that access to data per se cannot yet overcome competitive disadvantages. Data is mostly an essential part of the business model, especially if we speak about digital platforms. If the platform cannot make a profit with the data, this reduces the incentive to generate and collect particularly useful data. As a result, the platform’s investment in data innovation could be suboptimal. This in turn would adversely affect the other actors who use the data. The potential of the data that could be used would not be exploited. These disadvantages could be dealt with in the context of the specific design of data sharing. There is also a lack of empirical evidence for the concrete effects of data sharing on investments in digital platforms and on their incentives to innovate.

In order to decide whether Amazon actually has a dominant market position in terms of antitrust law, the relevant criteria for determining such a position must be first analyzed. Whether Amazon meets these criteria will depend in particular on the fact that Amazon is a platform for retailers, offers services for sellers on its marketplace, and acts as a retailer. This means that it has to be determined whether and to what extent these markets have to be viewed separately or analyzed as a whole. When defining the market in e-commerce it should be noted, that Amazon Marketplace interacts with a variety of business partners in a variety of markets. Regarding questions of access to the platform and the obligation to share data, the relationship between Amazon and the retailers appearing on the Amazon Marketplace is thus relevant. Amazon’s services as a platform for retailers differ significantly from the services of other platforms that are primarily not addressed on commercial retailers. Examples of such platforms, which are primarily aimed on consumers with regard to both user sides, are eBay and Facebook Marketplace (Funta, 2018a). In contrast to Amazon, retailers do not have there the same processing services that make commercial use for sale to a considerable extent attractive. The data available to determine the precise market shares of Amazon is limited. Amazon is without a doubt the most important e-commerce platform for retailers. Furthermore, Amazon has particularly high market shares in the product categories like books, toys, electronics & computers.

In connection with the digitization of the economy and in view of the emergence of platforms, the question about collection of large scale of user data (and in particular what data-power means for a company’s dominant position) has become a focus of the antitrust debate (Bundeskartellamt, B2-88/18-Amazon; Bundeskartellamt, Big Data und Wettbewerb, October 2017; Autorité de la concurrence/Bundeskartellamt, May 2016). The statements made by the European Commission in the Facebook/WhatsApp (COMP/M.7217 – Facebook/WhatsApp) and Google/DoubleClick (COMP/M.4731 – Google/DoubleClick) mergers also show that large databases can be of competitive relevance. It has been stated, that an existence of large database does not necessarily indicate a dominant position. It is still necessary to examine all the circumstances of the individual case in order to determine the significance of the specific data for the market position. The assessment criterion is therefore whether Amazon can use data in general or specifically in its own retail area in order to behave independently of the competition. As a result, however, the combination of market shares and particular strong network effects and the potential resulting from data, result in a dominant position of Amazon Marketplace on the market for the commercial use of e-commerce platforms.

Access to data offers various advantages, in particular the detailed overview of specific clicks or fluctuations in demand and the possibility of precise adjustments to the pricing. The data collection could in itself represent an essential facility (Železná, 1999). Amazon business model suggests that the proportion of data relevant to e-commerce in Amazon’s database is significantly higher in relation to other companies that collect more heterogeneous data. Amazon’s market share provides an indication of a dominant position. The question of whether market shares and the usual threshold values are suitable criteria for determining market power in markets for the use of data also requires further clarification. The fact that other companies collect large amounts of data does not put the data advantage into perspective as long as they are not active in the same market. The same applies to the option of purchasing data from third parties. The corresponding data streams and data collections could for this reason be regarded as unique and consequently not duplicable. For a more precise determination of the types of data that cannot be duplicated, the following limitations can be made, e.g. the entire raw database of Amazon transactions, the data set cannot be created to the same extent by other (non-dominant) platforms. To that extent there is a lack of duplicability (the results of the data evaluation of the raw data by Amazon cannot be duplicated). Thus, due to better data access that Amazon has about its competitors and their customers, the platform could gain a unique, uncatchable advantage that makes it impossible for marketplace retailer to compete successfully (Furman, 2019). As long as the retailers are generally successful in the market, it therefore seems questionable whether the relevant authorities and courts would see Amazon database (due to lack of substitutability) as an essential facility under the current legal regime. This does not mean that the independent retailers would have an equal chance in competition. Overall, it can be noted that the assumption of an essential facility in the form of data collection by Amazon appears theoretically possible in accordance with European competition law, but in practice it is likely to be unenforceable, primarily because of the strict requirements for non-substitutability. The essential facilities doctrine in European antitrust law in its current form would not be a sufficient basis to enable third-party dealers to participate fairly in competition.

Assuming that the Amazon Marketplace platform already embodies an essential facility, this would in practice mean that Amazon would have to enable every free retailer to sell products on the platform at fair conditions and would not deny them to use the platform without justifying reason. A refusal of access to parts of the essential facility would suffice for abuse. However, this does not necessarily result in Amazon’s obligation to make all data or even parts of this data accessible, since only access to the platform as such is affected as such. The database is part of the facility and the basis for its functioning. The essential facilities doctrine under European law can also relate to access to platforms. The criterion of non-duplicability seems to be problematic. A duplication would have to be economically impossible or unreasonable even for an economically strong company comparable to the dominant company. It has to be stated, that it is neither completely impossible nor necessarily unprofitable to set up an alternative online platform. The existence of other, smaller platforms such as eBay underpins this assessment. Online platforms act as important intermediaries for transactions. At the same time, strong indirect network effects have an effect on platform markets, so that companies are dependent on few or even one online platform when it comes to access to markets and consumers (a so-called gatekeeper function, in terms of “access to consumer data and capacity to deliver content and services”, Crémer, de Montjoye, Schweitzer, 2019)). In this respect, a look on Google as a search engine is worthwhile as a benchmark for comparison. Although there are various search engines available, these alternatives are hardly relevant in practice. In any case, it is currently not to be expected that Amazon Marketplace as a whole will be classified as an essential facility.

Instead of focusing on the data set (Polčák et al., 2018) or access to marketplace, Amazon’s marketing services could be seen as an essential facility, i.e. the services that Amazon uses to better market products on the platform. This can include pricing algorithms, website placements, or purchase suggestions. In practice, Amazon conducts merchandising and promotions even for external products. Amazon also determines unilaterally about advertisements that are placed as part of the sponsored products program. In particular a special positioning or ranking position is not guaranteed, and advertisements can be changed at its own discretion or removed without notice. As a result, Amazon is not a neutral provider of marketing services over which the retailer has a determining influence.