The Economics of Social Data

Abstract

A data intermediary pays consumers for information about their preferences and sells the information so acquired to firms that use it to tailor their products and prices.The social dimension of the individual data-whereby an individual's data are predictive of the behavior of others-generates a data externality that reduces the intermediary's cost of acquiring information. We derive the intermediary's optimal data policy and show that it preserves the privacy of the consumers' identities while providing precise information about market demand to the firms. This enables the intermediary to capture the entire value of information as the number of consumers grows large.Individual Data and Data Intermediaries The rise of large Internet platforms-such as Facebook, Google, and Amazon in the US, and of similar large entities in China, such as JD, Tencent and Alibaba-has led to an unprecedented collection and commercial use of individual data. The ever-increasing user bases of these platforms generate massive amounts of data about individual consumers: their preferences, their locations, their friends, their political views, and almost all other facets of their lives. In turn, many of the services provided by large Internet platforms rely critically on these data. The availability of individual-level data allows these companies to offer refined search results, personalized product recommendations, informative ratings, timely traffic data, and targeted advertisements. 1 The recent disclosures on the use and misuse of social data by Internet platforms have prompted regulators to limit the hitherto largely unsupervised use of individual data by these companies. As a result, nearly all proposed and enacted regulation to date is centered on ensuring that consumers retain control over their data. The idea is that, because consumer data must be acquired, aggregated, packaged, and sold, the allocation of control rights impacts the ultimate use of the data. 2 In particular, by assigning ownership and control to individual consumers, regulators hope to enable an efficient use of information or at least to grant individual consumers the appropriate compensation for the data they reveal. 3 Despite the appeal of the Coase theorem implicitly invoked in the regulation, it is far from evident that the resulting allocation of information is efficient due to multiple potential market failures. In this paper, we explore the implications of externalities across consumers.A central feature of the data collected from individuals is its social aspect. Namely, the data captured from an individual user are not only informative about that specific individual but also about appropriately defined nearby individuals. Thus, these individual data are really social data. The social nature of the data generates a data externality, the sign and magnitude of which depend on the ultimate use of the information so gained. In the context of geolocation data, for instance, an individual user conveys information about the traffic conditions for nearby driver...

Data Sharing and Incentives

“…In particular, Acemoglu et al (2019) and Bergemann et al (2019) also consider externalities created by social data. Different from us, these papers study data sharing in environments where consumers may sell their data.…”

“…Different from us, these papers study data sharing in environments where consumers may sell their data. In Bergemann et al (2019), one agent's information improves a firm's ability to price-discriminate against other agents, which can decrease consumer surplus. In Acemoglu et al (2019), agents value privacy, and thus information collected about one agent imposes a direct negative externality on other agents when types are correlated.…”

“…14 A function g > 0 is strictly log-concave if log g is strictly concave. 15 The Gaussian versions of our quality and circumstance linkage models represent special cases of the information environment considered in Meyer and Vickers (1997) and Bergemann et al (2019), both of whom allow for correlation between both types and shocks. The Gaussian version of our quality linkage model also corresponds to a symmetric version of the environment considered in Acemoglu et al (2019).…”