Estimating Taylor rules in a real time setting

This paper demonstrates how the use of revised data distorts our understanding of past monetary policy decisions Three problems are addressed - the use of (i) contemporaneous rather than lagged data (ii) revised rather than unrevised data; and (iii) leads of data unavailable at the time of policy setting for estimating potential output In order to evaluate each of these distortions separately I have estimated Taylor rules using different sets of estimates of output gap and inflation for three sub-samples corresponding to chairmanship terms of Arthur Burns Paul Volcker and Alan Greenspan Three series of estimates are constructed - series based on revised estimates of data for the whole post-war sample; series based on truncated (excluding leads) subsamples of revised data; and series similar to the previous one but based on unrevised data Although using revised data may produce significantly misleading conclusions the inclusion of leads of the data when estimating the potential level of the economy has a much bigger impact producing coefficients which may have a value less than half of the true one At the same time the use of contemporaneous rather than lagged data does not seem to have a big effect on the final results Among other things I demonstrate that the US monetary policy was less active during Burns' chairmanship and much more anti-inflationary during Greenspan's than traditional analysis would suggest.