Micro Theory Work in Progress

Title: Persuasion via Sequentially Acquired Evidence

Abstract: I study a sender who can privately acquire and partially disclose hard evidence to persuade a receiver about a binary state of the world. The sender sequentially acquires noisy binary signals. Signals are time-stamped and costly to acquire. When she stops, the sender discloses a left truncation of the signals. That is, it is possible to omit most dated signals. The receiver, uncertain of how many signals the sender acquires, takes an action based on the difference between the number of good and bad signals in the disclosure. If the cost of acquiring each signal is not too high, there are multiple persuasion equilibria. In every equilibrium, the receiver's posterior belief is supported on two points. This is akin to Bayesian persuasion. If full disclosure of signals is mandatory, the game is equivalent to costly Bayesian persuasion. Mandating full disclosure benefits the sender and hurts the receiver.