The meta taylor rule

Abstract

We characterize U.S. monetary policy within a generalized Taylor rule framework that accommodates uncertainties about the duration of policy regimes and the specification of the rule, in addition to the standard parameter and stochastic uncertainties inherent in traditional Taylor rule analysis. Our approach involves estimation and inference based on Taylor rules obtained through standard linear regression methods, but combined using Bayesian model averaging techniques. Employing data that were available in real time, the estimated version of the "meta" Taylor rule provides a flexible but compelling characterization of monetary policy in the United States over the last 40 years.