The predictive power of alternative indicators of monetary policy

Nonborrowed reserve mix as a measure of monetary policy is found to dominate the federal funds rate and nonborrowed reserves in terms of forecasting power for real activity in samples starting in 1959 with a moving endpoint from 1975:8 to 1997:8. In contrast, variance-decomposition results from moving samples of fixed length, that is from a rolling VAR, do not lead to the same conclusion. When data is excluded from the early and mid 1960s, the federal funds rate starts to provide more information concerning future movement in real activity relative to that provided by nonborrowed reserve mix. Results are determined from well established VAR models in which each variable in turn is identified as monetary policy indicator.