Information Disclosure about Booster Efficacy in a Non-Stationary Environment

This paper investigates the dynamic disclosure of information in non-stationary environments. In particular, a planner iteratively discloses information about the efficacy of an immunizing booster shot that stochastically evolves over time amid the long-run spread of an infectious disease whose severity also varies over time. Each time period, a heterogeneous population of agents uses the disclosed information to determine whether they should obtain the booster shot, and then whether to remain isolated or active. The central planner's objective is to ensure that the active population remains above a minimum threshold each period. We characterize a Markov decision process over the state of beliefs and how signalling mechanisms act on them. We highlight the "greedy" disclosure rule which provides the least amount of information possible subject to the planner maximizing the likelihood of achieving the active population threshold in the current period. Our results demonstrate that the greedy disclosure rule becomes optimal in finite time. We show this for settings where the population's belief over the booster's efficacy becomes more pessimistic than the belief required in the long-run.