Estimation of a nonlinear Taylor rule using real-time U.S. data

This paper extends the work in Orphanides (2003) by re-examining the empirical evidence for a Taylor rule in a nonlinear framework. In doing so, it updates the Greenbook dataset used by the afore men- tioned author to the most recent available period. A three-regime threshold regression model is utilized to capture the possibly asymmetric policy reaction function used by the U.S. Federal Reserve. The theoretical foundations for such an approach to monetary policy are discussed in Orphanides and Wilcox (2002). Our results indicate that the estimated Taylor rule for the U.S., based on real-time Greenbook data for the period 1982:3-2003:4, is probably nonlinear.