An Informational Theory of Instrument Policy

Abstract

I study how to optimally set instrument and disclosure policy in a dispersed information model in which people learn about an unobserved state from economy-wide outcomes. A policy maker can use its information about the state either directly by disclosing it or indirectly through its choice of the policy instrument. I show how exclusive use of instrument policy, in this setting, is always a more efficient way to use the information that the policy maker has than directly disclosing it. Instrument policy achieves a better balance between the direct benefit of using the policy maker’s information and the indirect adverse cost of reducing the informational efficiency of economy-wide outcomes. I detail how instrument policy achieves this superiority by avoiding distorting the policy maker information used in equilibrium. The analysis further considers how this superiority depends on direct pay-off externalities, “tremble-hand” instruments and the presence of multiple fundamentals. I conclude with two business cycle applications.