A Test of the Permanent-Income Hypothesis: Reply

Abstract

Holmes raises three main issues with respect to my (1969) test of the permanent-income hypothesis (PIH): (a) that my definition of permanent consumption is not equivalent to that of Friedman, (b) that the weights used to derive permanent income and permanent consumption do not sum to unity and hence the values of permanent income, permanent consumption and transitory income, and transitory consumption used in my test are not the correct values, (c) that I ignored the averagetrend rate of growth in calculating permanent income. Before I consider these issues, let me state that the conclusions arrived at by Holmes are not inconsistent with mine. He states: This direct test leads to a rejection of the PIH when consumption is defined to include consumer durable expenditure. However, when consumption is defined to exclude such expenditures, this part of Friedman's theory is not easily rejected on the basis of Canadian evidence (1974). Holmes himself points out that his equations (6) and (12) in table 2 are the really relevant equations. The results of these equations lead to a mild rejection of the PIH for Canada. Thus, according to equation (6) of table 2, when consumption is defined to include durable goods, the marginal propensity to consume out of transitory income is equal to .345, and the marginal propensity to consume out of permanent income is equal to 1.024. When consumption excludes durable goods (according to eq. [12] of table 2), the marginal propensity to consume out of transitory income is equal to .272 and out of permanent income is equal to .828. Since consumption, according to the PIH, is defined to include the value of all nondurable goods plus the rental value of durable goods, one might expect that the true marginal propensities to consume transitory and permanent incomes would lie within these limits. It is likely that the true marginal propensities to consume out of transitory and permanent incomes would be approximately .30 and .90, respectively. These results are not very different from those that I reported in my paper and are very similar to those reported in tables 2 and 3 below. Let us now turn to the three issues raised by Holmes.