If a policymaker learns today that there will be a regime change in the future that affects everyone, at what time between now and then should he/she announce it to the public? This paper presents a dynamic model where agents that have a limited amount of attention to allocate between learning about the new regime and everything else. They trade off the benefit of being better informed and making better decisions when the new regime arrives against the cost of paying less attention to current events and making worse decisions now. By choosing when to make the announcement, the policymaker can affect this decision. The policymaker also takes into account that later announcements are more precise, and that agents may inefficiently put too much weight on public signals due to strategic complementarities. I solve for the optimal timing of announcements and the conditions under which it is optimal to keep mum in spite of public interest. As a by-product, I characterize the life-cycle of attention following an announcement, before and after the regime changes.
