[V. Undersaving] It is common wisdom that people save too little. To compensate for this failure, most developed country governments heavily support the elderly in retirement. In addition, a very large number of employers require and subsidize pension contributions of their employees. Many forms of saving receive tax advantage. Even with these legs up, the common wisdom is that financial a**ets of most households still fall considerably short of what they need to maintain their consumption in retirement. For New Cla**ical economics, saving too little or too much, like involuntary unemployment, is an impossibility, a straightforward contradiction of the a**umptions of the model. Since saving is the result of individual utility maximization, it must, absent externalities, be just right. Behavioral macroeconomics, in contrast, has developed theoretical tools and empirical strategies to advance understanding of such time-inconsistent behavior. A key theoretical innovation permitting systematic an*lysis of time-inconsistent behavior is the recognition that individuals may maximize a utility function that is divorced from that representing “true welfare.” Once this distinction is accepted, “saving too little” becomes a meaningful concept. The idea can be illustrated by the ancient myth of the lemmings, who every few years are said to converge in a d**h march, which ends with their final plunge into the sea. The alleged behavior of those lemmings reveals a distinction common among psychologists, but rare for economists. Unless the lemmings experience an unusual epiphany in that final plunge, their utility or welfare is given by one function; yet they maximize another. Think about it: the popular view of saving, that people undersave, is similarly described. Determining whether people save too much or too little involves asking whether people, like the lemmings, have one (intertemporal) utility function which describes their welfare, but maximize another. Such evidence as there is suggests potentially large difference between the two concepts. High negative rates of time discount are necessary to explain actual wealthearnings ratios. Yet, questionnaire responses on the consumption-saving trade-offs that people think they ought to make reveals an intertemporal discount rate that is on average slightly positive. The hyperbolic discount function, which has been used to study intertemporal savings choices, can be used to formalize the distinction between the utility function that describes actual saving behavior and the utility function that measures the welfare resulting from that behavior. The hyperbolic function captures the difficulty people have in exercising self-control. In contrast to the constant discount rates that are standard in neocla**ical theory, the hyperbolic function a**umes that the discount rates used to evaluate trade-offs between adjacent periods decline as the time horizon lengthens: individuals use high discount rates to evaluate options that require an immediate sacrifice for a future reward and lower discount rates when the same sacrifice is deferred into the future. Thus, they are patient in making choices requiring gratifi- cation delays when those sacrifices are deferred; but impatient in delaying gratification in the short run. Because present consumption is more salient than future consumption, individuals procrastinate about saving. The hyperbolic
function accords closely with experimental findings: Human and animal subjects are far less willing to delay gratification immediately than to commit to such delays in the future.66 Two forms of procrastination may result from hyperbolic discounting. “Naive procrastination” occurs when an individual a**umes incorrectly that her utility function will be different in the future. She mistakenly projects that, although today is salient, tomorrow will be different. She fails to see that tomorrow's self will be different from today's self, so that tomorrow will be just as salient as today once it has moved one step closer. The naive procrastinator mistakenly believes that she will save (diet, exercise, quit smoking, etc.) tomorrow, although she has not done so today, and is surprised that the sacrifices deferred today are also deferred again tomorrow. More sophisticated procrastination takes the form of preproperation, according to the terminology of Ted O'Donoghue and Rabin (1999). The preproperator has fully rational expectations about who her future self will be. She says to herself: there is no reason to save today if tomorrow is going to be especially salient. If tomorrow is especially salient then I will spend whatever savings I have laid aside today when it was also especially salient. So I should not make the sacrifice today. Laibson has used hyperbolic discounting as the basis of a research program on saving behavior and policy. With coauthors Andrea Repetto and Jeremy Tobacman (1998) he has simulated the effects of different tax incentive programs in a world in which consumers preproperate. They estimate that large positive welfare effects result from small changes in incentives to save which reduce the amount of preproperation. Because of this work the regulations regarding tax-advantaged 401(k) savings plans have been changed. If firms so choose, workers may now be automatically enrolled with an automatic default contribution. Adoption of such plans significantly increases plan participation and many workers maintain their contributions at the level of the default. Besides the popularity of social security and other programs that “force” consumers to save, the best evidence of undersaving is probably the observation that, upon retirement, individuals, on average, reduce consumption substantially. In fact, consumption at retirement declines discontinuously. Those with more wealth and higher income replacement reduce their consumption by much less. This finding is difficult to explain with the standard life cycle, exponential discounting model. Thaler and Shlomo Benartzi (2000) have devised a savings plan to overcome workers' tendency to procrastinate and have tested it on an experimental basis at a mid-size manufacturing firm: employees were invited to join a savings plan allowing them to elect, in advance, the fraction of wage or salary increases to be set aside for savings. Consistent with hyperbolic discounting, but not with the standard exponential model, workers chose relatively modest saving out of current income but committed to save large fractions of future wage and salary increases. Within a short period of time, the average savings rate had doubled.