Optimal information disclosure

A Sender(internet advertising platform, seller, rating agency, or school) has a probabil-ity distribution over prospects(internet ads, products, bonds, or students). Each prospect is characterized by its pro\u85tability to the Sender and its relevance to a Receiver(internet user, consumer, investor, or employer). The Sender privately observes the pro\u85tability and relevance of the prospect, whereas the receiver observes only a signal provided by the Sender (the prospects rating). The Receiver accepts a given prospect only if his Bayesian infer-ence about its relevance exceeds a private opportunity cost that is uniformly drawn from [0,1]. We characterize the Senders optimal information disclosure rule assuming commit-ment power on her behalf. While the Receivers welfare is maximized by full disclosure, the Senders pro\u85ts are typically maximized by partial disclosure, in which the Receiver is induced to accepting less relevant but more pro\u85table prospects (switches) by pooling them with more relevant but less pro\u85table ones (baits). Extensions of the model include maximizing a weighted sum of Sender pro\u85ts and Receiver welfare, and allowing the Sender to subsidize or tax the Receiver. Acknowledgements to be added. 1