When to ask for an update: Timing in strategic communication

Abstract

A principal (receiver) considers whether to accept a project whose total value depends on two aspects. In each period, an agent (sender) privately learns the value of one aspect with positive probability. We compare two common reporting protocols: frequent updating, which requires the sender to report in each period, and infrequent updating, which requires him to report only at the end of the learning process. The sender is biased towards acceptance; he may conceal his signal, but cannot otherwise misrepresent.If the receiver would reject the project under the prior, then the equilibrium outcome is the same regardless of the updating frequency under certain regularity conditions. In contrast, if the project's prior expected value is sufficiently high, then updating frequency matters. Specifically, if it is sufficiently unlikely for the sender to be informed in a later period and the players' interests are sufficiently aligned, then frequent updating is better for the receiver since it encourages the sender to reveal provisionally unfavorable information early on. When it is sufficiently likely for the sender to be informed in a later period, however, infrequent updating is better for the receiver. Conditional delegation captures the benefits of both frequent and infrequent updating under high prior.