A CROSS SECTION STUDY OF THE DEMAND FOR MONEY: THE UNITED STATES, 1960–62

Abstract

A BRIEF SURVEY is made of eight previous cross section studies of the demand for money. This is followed by a replication with U.S. data of a 1945 Czechoslovakian cross section study which concluded that wealth and not income was the determining variable in the money demand function. The U.S. data, which consists of the 1960-62 reinterview Survey of Consumer Finances (SCF), indicates that income is more important than nonhuman wealth. A regional interest rate variable and indicators of human wealth (age, family size, education, race, occupation, and place of residence) are added to the regression with a resulting increase in explanatory power. The extent of various biases-understatement of balances bias, omission of currency bias, and panel bias-is examined and the biases appear to offset each other. Since four years of income data are available for each of the 1057 spending units interviewed in the SCF panel study, a distributed lag approximation of permanent income is formed and is found to be more significant than current measured income. The final set of regressions includes income, regional interest rate, and indicators of both human and nonhuman wealth. The two rival models are the absolute income-wealth hypothesis and the permanent income hypothesis according to which money is a consumer durable which is uncorrelated with transitory income and which depends on permanent rather than current income. The findings of the present study suggest a confirmation of the permanent income hypothesis.