Reinterpreting money demand regressions

Abstract

It has been two decades since a partial adjustment model was first applied in empirical money demand studies. Since then the partial adjustment specification has become widely used, particularly in quarterly money demand studies. However, the theoretical rationalization for partial adjustment has never been entirely satisfactory. Furthermore, a number of empirical characteristics of the conventional money demand regression have proven difficult to interpret under the partial adjustment specification. This paper offers an alternative interpretation of the conventional money demand regression results, one that does not rely on a partial adjustment rationale. Instead, money demand is assumed to adjust completely each period to current interest rate and transaction variables. But empirical measures of these regressors are contaminated by stochastic measurement error. This interpretation is able to explain many characteristics of the conventional money demand regression that have proven difficult to interpret from a partial adjustment point of view. A conventional money demand regression with a partial adjustment specification is described in Section II. Following this description, a number of specific difficulties in interpreting the conventional money demand regression are detailed. Theoretical weaknesses with the partial adjustment rationalization and difficulties with the partial adjustment interpretation of specific empirical characteristics of the conventional money demand regression are discussed. In Section III, coefficients in a