I study the welfare and price implications of consumer privacy. A consumer discloses information to a multiproduct seller, which learns about his preferences, sets prices, and makes product recommendations. Although the consumer benefits from accurate recommendations, the seller may use the information to price discriminate. I show that the seller prefers to commit to not use information for pricing in order to encourage information disclosure. However, this commitment hurts the consumer, who could be better off by precommitting to withhold some information. In contrast to single-product models, total surplus may be lower if the seller can base prices on information. (JEL D11, D83, L81, M31)

I study a model of competition between data intermediaries, which collect personal data from consumers and sell them to downstream firms. Competition has a limited impact on benefiting consumers: If intermediaries offer high compensation for data, consumers share data with multiple intermediaries, which lowers the downstream price of data and hurts intermediaries. Anticipating this, intermediaries offer low compensation for data. Although consumers are exclusive suppliers of data, the nonrivalry of data can lead to concentration and high intermediary profits in data markets. In particular, if downstream firms use data to extract surplus from consumers, competing intermediaries sustain a monopoly outcome.

No abstract

, and participants in various seminars and conferences. I am grateful for the financial support from the Yoshida Scholarship Foundation and the Stanford Graduate Fellowship. The opinions expressed in this article are the author's own and do not reflect the views of the Bank of Canada. The author declares that he has no relevant material or financial interests that relate to the research described in this paper.

This paper studies how the outcome of Bayesian persuasion depends on a sender's information. I study a game in which, prior to the sender's information disclosure, the designer can restrict the most informative signal that the sender can generate. In the binary action case, I consider arbitrary preferences of the designer and characterize all equilibrium outcomes. As a corollary, I solve a problem of how to maximize a receiver's payoffs by restricting the sender's information: Whenever the designer can increase the receiver's payoffs by restricting the sender's information, the receiver-optimal way coincides with an equilibrium of the game in which the receiver persuades the sender.