The Velocity of M1 in the 1980s: Evidence from a Multivariate Time Series Model

Abstract

The velocity of M declined in the 1980s after an upward trend over the previous twenty years. This apparent break in the process generating velocity was an important consideration in the Federal Reserve's deemphasis of M as an intermediate target in 1982. Discussion of the behavior of velocity has focused upon whether the process generating velocity underwent a shift, perhaps due to the financial innovation and deregulation in the 1980s, or whether the behavior of velocity merely reflects the underlying variability in its determinants. Tatom [34], Darby et al. [7], Stone and Thornton [33], and Rasche [30] provide summaries of this debate. No consensus has emerged on whether the process generating the velocity of M 1 shifted after 1982. The aim of the present paper is to analyze the behavior of velocity before and after 1982. The approach taken here is quite different from the traditional method of analyzing velocity which relies upon single-equation estimates of velocity or money demand functions. A vector autoregressive (VAR) model that contains the income velocity of M1 and its key economic determinants (income, interest rates, inflation, and money growth variability) is estimated using quarterly data for the period 1961:1-1981:4. The adequacy of this model is evaluated by calculating variance decompositions. A Monte Carlo simulation technique similar to that described in Doan and Litterman [8] is used to compute standard errors for the variance decompositions. This allows a judgement as to the significance of the response of velocity to its economic determinants. The effects of these variables on velocity for the period 1982:1-1988:4, the period of velocity's unusual behavior, are estimated through computation of historical decompositions. Formal stability tests are also performed to determine if there was a change in the process generating velocity after 1982.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 36 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.