This paper builds a dynamic model of the information flow between partially informed financial institutions and a public agency. The financial institutions decide how to allocate their portfolio between a risk-free technology with a known payoff and a risky technology whose payoff is unknown. The public agency learns about the value of the unknown payoff by observing with measurement error the actions of the financial institutions and decides whether to communicate the information at the agency's disposal. The paper characterizes the optimal public communication plan and shows that full transparency (taken as the release of information whenever it is collected) is not always optimal. Instead, optimal plans involve delayed communication, the amount of delay depending on the precision of private information and the size of the agency's measurement error. The explanation of the result lies in the collection process of public information: while releasing information improves the welfare of the agents, it also decreases the informational content of their actions, hampering the agency's learning and reducing the benefits of future public communication.
D80 - Information and Uncertainty. General ; D83 - Search, Learning, Information and Knowledge ; E58 - Central Banks and Their Policies ; E61 - Policy Objectives; Policy Designs and Consistency; Policy Coordination
