Efficient Timing of Communication in Multiperiod Agencies

This paper examines communication in a two-period principal/agent model in which the agent receives a private signal about the second outcome before the first outcome is realized. No communication is compared with communication at three possible dates: before the first outcome (early), at the first outcome/consumption date (normal), and between the initial consumption date and the second outcome (delayed). Delayed communication is shown to have no value if the agent's information is perfect, but can have value if it is imperfect. Early and normal communication can be used to "smooth" compensation across periods and, hence, generally have incremental value over delayed communication if the agent cannot borrow or save. However, the "smoothing" benefits disappear if he can borrow and save. Early and normal communication are equivalent if the agent has domain-additive exponential preferences and the private signal is uninformative about the first outcome. If the private signal is informative about the first outcome, the incremental value of early compared with normal communication attains its maximum for "medium" informativeness. A unifying example is used throughout.