U.S. Money Demand: Structural Shifts or Heterogeneous Agents?

This paper examines whether observed "structural shifts" in the money demand function could be the result of agent heterogeneity due to different household income levels. Following the methodology of T. M. Stoker (1986), income distribution variables are found to be significant determinants of money demand and, once changes in the distribution of income are accounted for, there is no evidence of parameter instability. A money demand function incorporating income distribution effects is shown to perform as well or better than several standard alternatives on the basis of diagnostic tests, nonnested hypothesis tests, within-sample prediction errors, and out-of-sample forecasting. Copyright 1992 by Oxford University Press.