Optimal Monetary Policy in a Data-Rich Environment

This paper formalizes and solves the problem of a central bank which has a fully structural model (assumed to be true), observes a large number of noisy indicators, and seeks to (1) extract from these indicators estimates of the relevant shocks and state variables of the economy, and (2) conduct optimal monetary policy, given these estimated shocks and other exogenous variables. We argue that by using the entire data set, the central bank obtains sensibly more accurate estimates of the "true" state of the economy. In addition, depending on the number and quality of indicators considered for the estimation of the latent state of the economy, optimal policy may yield different paths of key macroeconomic variables and important differences in welfare.