In this paper we consider and illustrate a solution to the inter-related problems of monetary aggregation and estimation of money demand. The problem with the definition of money is that the relative prices of the monetary components fluctuate over time, rendering simple-sum aggregates inefficient. The authors apply Revealed Preference tests to the U.S. monthly data to determine admissible and separable components. These components are then aggregated using the Divisia technique. To deal with the problem of money demand, the dynamic Fourier expenditure system is used to provide estimates of the elasticities of substitution. These, while showing general substitution among the liquid assets studied, are quite variable over time. This finding underscores the inefficiency of both simple-sum aggregation and single-equation, log-linear money-demand estimation. Copyright 1997 by Ohio State University Press.
