This paper reports evidence that the relation between interest rates and mo
ney shocks depends on the Federal Reserve's operating procedure. A signific
ant liquidity effect is detected only when a nonborrowed reserve measure is
used to identify policy shocks. When broad monetary aggregates are conside
red, a liquidity effect is not observed. However, where a liquidity effect
is documented, it is primarily due to the 1979-1982 period in which a nonbo
rrowed reserve operating procedure was adopted. If this period is omitted,
then interest rates and money shocks are largely independent regardless of
which monetary aggregate is used to measure policy shocks.