The most powerful objection to the theory that income is preferred to free time because of a concern to fund retirement is, however, that one of its crucial empirical entailments—the prediction that workers will save at high rates when at the height of their earning potential in order to finance their postretirement lives—fails to hold true. For the preference for income over more free time during the working years to be convincingly explained as preparation for retirement, we would have to demonstrate that the life-cycle hypothesis, developed by the economist Franco Modigliani (1970) (and Milton Friedman’s [1957] related permanent income hypothesis), is supported by the empirical evidence. The life-cycle hypothesis posits that people aim to generate a constant amount of utility, through their consumption, over the entire course of their lives. According to the theory, because the marginal utility of consumption decreases, individuals, rather than spending all the money they have at any given point in time, will seek to distribute their spending over time such that it maximizes utility over their entire lives, with the result that consumption is “smoothed” across the life course. The theory predicts that people will go into debt when young and poor, save when afluent and middle aged (in order to finance their retirement), and “dissave” when old, spending down their assets. So it goes in theory, but the empirical support for the life-cycle hypothesis is not compelling.35 Economist Angus Deaton (1992), in an overview of the empirical literature on the topic, concludes that the evidence, on balance, does not support the life-cycle hypothesis. According to Deaton, “it is certainly the case that the household data typically show no clear evidence of asset decumulation among the old.” Moreover, he adds, “there are other problems earlier in the life-cycle, and although there is often some hump saving in late middle age, there is also typically a very close articulation of consumption and income over the whole of life. Survey data also show that many households possess little or no wealth, certainly not in financial assets, and many individuals have little beyond social security after they retire.” Deaton concludes that “the fall in the saving rate in the US is not attributable to a redistribution of aggregate income towards the old, but comes from an increase in expenditures, particularly expenditures on durable goods, across all cohorts, particularly prime-age cohorts” (p. 217 [emphasis added]).36 Deaton’s overall assessment accords with that of other economists. Paul Courant, Edward Gamlich, and John Laitner (1986, pp. 278–79), for example, note that “for all its elegance and rationality the life cycle hypothesis has not tested out well.” Moreover, if Modigliani’s life-cycle hypothesis was correct then, as the length of time lived after retirement increased (as it did consistently during the twentieth century), there should have been a commensurate increase in the rate of saving. Yet household savings in the United States did not increase over the course of the twentieth century, despite rising income, increasing life expectancy, and the growing prospect of life after work (Maddison 1992).
64
Time for Things