THE DEMAND FUNCTIONS OF THE HOUSEHOLD SECTOR FOR FINANCIAL ASSETS: AN ECONOMETRIC STUDY*

Abstract

THIS STUDY investigates the demand functions of the household sector for financial assets, on both theoretical and empirical levels. A model of asset selection is presented for an individual household, then is aggregated to represent the entire household sector. This model is an outgrowth of work performed by Farrar and de Leeuw and is based on the efficient portfolio theory of Markowitz and Tobin. The hypothetical household of the model faces an array of possible financial assets and liabilities it can choose among, thereby receiving an expected return and incurring a level of risk exposure. The household's utility function is given a quadratic approximation and is maximized subject to a net wealth constraint. From this maximization is derived a generalized demand function linear in net wealth and rates of return. The model is then altered to produce a submodel in which the resultant demand functions are homogeneous of degree one in net wealth. Demand functions are estimated for seven financial assets in addition to gross financial assets. Two different specifications of the generalized demand function are employed. The first uses aggregate net wealth as the constraint on the household sector. In the second the aggregate net wealth constraint is divided into two parts, deflated net wealth and capital gains on financial assets. The latter is designed to capture the differential effects of various subclasses of net wealth upon the financial asset demand functions of the household sector. Estimation is based on a sample of 57 quarterly observations of the household sector, first quarter, 1952, through first quarter, 1966. The major conclusions of this study are as follows: (1) Aggregate net wealth is an important determinant of households' financial asset demands. While there is some evidence of a lagged response in the adjustment of households' financial asset stocks to changes in net wealth, the same evidence indicates that this lag seldom has any importance when extended beyond four quarters. (2) Disaggregation of the net wealth argument into two separate arguments, consisting of deflated wealth and capital gains on financial assets, results in the deflated wealth argument behaving in a fashion similar to net wealth. Deflated wealth proves to be an important influence upon the financial asset demands of households. (3) The capital gains argument used when net wealth is disaggregated does not possess great explanatory power and as a result only short lags in adjustment are discovered with reference to changes in the capital gains type of wealth. The signs of the coefficients of the capital gains argument are frequently negative in the current period and/or when the lag is short. This evidence is consistent with, but fails to demonstrate firmly, the contention that the receipt of a capital gain stimulates households to increase their consumption in the very short-run and perhaps temporarily to draw down on certain financial asset stocks.