This paper focuses on the demand for money in the United States in the context of five popular locally flexible functional forms—the generalized Leontief, the basic translog, the almost ideal demand system, the Minflex Laurent, and the normalized quadratic reciprocal indirect utility function. We pay explicit attention to the theoretical regularity conditions of positivity, monotonicity, and curvature and argue that much of the older empirical literature ignores economic regularity. We treat the curvature property as a maintained hypothesis and provide a comparison in terms of violations of the regularity conditions and in terms of output in the form of a full set of elasticities. We also provide a policy perspective, in that a strong case can be made for abandoning the simple sum approach to monetary aggregation, on the basis of the low elasticities of substitution among the components of the popular M2 aggregate of money.
