Competing in the presence of privacy concerns: a model of the market for customer information

Abstract

There is 'no free disposal' (NFD) in the consumption of online personalization services, as this activity inherently involves sharing of personal and preference information that creates disutilities to the consumer. Not only are more services not necessarily better for the consumer, but these services are also provided for free as firms extract value from the usage of consumer information rather than from directly pricing the services. Further, the services themselves may be provided to a consumer as a fixed-length toolbar/deskbar (fixed-services strategy) or with the option of choosing a subset of the portfolio of services offered (variable-services strategy). This paper models a duopoly of firms that are heterogeneous in their marginal value for consumer information (MVI) and interact through a two-stage dynamic game, where the firms choose a fixed- or variable-services strategy in the first stage and their level of services of-fering in the second. After examining a series of subgame equilibria, we arrive at distinct subgame-perfect Nash equilibriua (SPNEs) that allows us to characterize competition between firms of different MVI endowments. Our findings suggest that while there is no SPNE in a duopoly of two small firms, when one firm is small and the large, there is a unique SPNE in pure strategies where both firms offer fixed-services such that they segment the market. As the differences in their valuations increase, the larger firm continues to offer fixed-services while smaller firm enjoys the option of offering variable services. A duopoly of large firms results always results in symmetric SPNE; both firms offer variable services as long as one firm has very large MVI and both offer fixed-services otherwise. Interestingly while the former is consumer welfare maximizing, the latter results in a third of the market (consisting of privacy seekers) not being served. Our results lead to important managerial and policy implications, as well as interesting extensions to the existing location models.