THE PERMANENT INCOME‐WEALTH APPROACH TO THE DEMAND FOR MONEY: ANALYTICAL AND EMPIRICAL ASPECTS*

Abstract

THIS INVESTIGATION is a critical examination of the permanent income-wealth approach to the demand for money. The study attempts to explore the theoretical models on which such an approach is based, examine the available empirical evidence regarding the validity of the hypotheses, and adduce additional evidence concerning the role played by permanent income and wealth in demand functions for money. Several alternative models are developed in the thesis. Some relatively unrecognized problems concerning the nature of available evidence are treated. Disaggregated quarterly data are used to explore how the hypotheses work on a short-run sectoral basis. Analytically, it is shown that the ideas involved may be treated as special cases within the framework of a Keynesian model. The concepts, in this case, are reflected as the hypothesis that the demand for current output and the demand for money are relatively insensitive to fluctuations in current income. As a subsidiary hypothesis, the permanent income-wealth hypothesis treats the demand for cash as being relatively insensitive to changes in yields on financial assets. In this situation, the conclusion emerges that permanent income-wealth hypotheses imply a heightened importance of monetary versus nonmonetary factors. That is, the change in equilibrium income with respect to changes in the money stock is raised relative to the change in equilibrium income relative to changes in autonomous flows. Hypotheses in a permanent income-wealth approach alternatively may be developed on aggregating a macroeconomic model in which economic units choose optimal portfolios determined by tastes for income and risk. In this context, the hypotheses may be treated as propositions that the expected yields on capital change by relatively small amounts in response to changes in current income. As before, the conclusion emerges that a permanent income-wealth analysis implies a greater relative importance of changes in monetary rather than nonmonetary factors in altering equilibrium income. In examining available empirical work concerning the various approaches, it becomes clear that many of the studies, particularly the earlier studies, are inadequate, if not misleading. The reasons for this are due to problems of autocorrelation, implicit correlation, or specification and estimation of functions in a fashion that conceals such difficulties. When new evidence is adduced in terms of quarterly data for the household and corporate sectors from 1952-63, it appears that, at least in the short run, neither income nor wealth act as important determinants of money holdings. Yields on assets and lagged stock variables seem to be the key variables in explaining cash balances. Moreover, when the lagged stock variable is dropped in the estimates, serial correlation does not appear in residuals so that the distributed lag model does not merely conceal the presence of correlated residuals. The consumer sector seems to be most responsive to fluctuations in yields on mortgages, whereas the corporate nonfinancial sector is sensitive to changes in yields on Treasury bills and on prime commercial paper. In general, a closer fit of the regression estimates is obtained for the consumer data than for the data on nonfinancial corporations. Wide differences in variables and closeness of fit in regression estimates seem to indicate that the disaggregated approach in a study of the demand for money is highly useful.