Micro data studies of household saving often find a significant group in the population with virtually no wealth, raising concerns about heterogeneity in motives for saving. In particular, this heterogeneity has been interpreted as evidence against the life cycle model of saving. This paper argues that a life cycle model can replicate observed patterns in household wealth accumulation after accounting explicitly for precautionary saving and asset-based, means-tested social insurance. We demonstrate theoretically that social insurance programs with means tests based on assets discourage saving by households with low expected lifetime income. In addition, we evaluate the model using a dynamic programming model with four state variables. Assuming common preference parameters across lifetime income groups, we are able to replicate the empirical pattern that low-income households are more likely than high-income households to hold virtually no wealth. Low wealth accumulation can be explained as a utility-maximizing response to asset-based, means-tested welfare programs.

One of the key puzzles in understanding behavior is not so much why people save-the title of this session-but why people don't save. According to the familiar life-cycle model, households should accumulate wealth to provide for their retirement consumption. The surprising result from the data is the sizable fraction of the population who have accumulated so little, even among those nearing retirement. Given that earnings will almost surely decline when households retire, such behavior can imply poor living standards for the elderly. Some might interpret the low wealth accumulation as being evidence of myopia, irrationality, or a failure of households to enforce mental accounting (Richard Thaler, 1994), while others might view the low level of wealth accumulation as evidence of high individual rates of time preference. Determining the underlying causes of low wealth is crucial for public policies that seek to alleviate low aggregate rates in the United States, as well as policies that seek to buttress the adequacy of financial resources for the elderly. In this paper, we outline what we believe to be the causes of why many people do not save. We conclude by speculating about government policies that may be most effective at encouraging saving. Much of the research examining levels of consumption, saving, and wealth, as well as their responsiveness to policy, has been done using a life-cycle model with the simplifying assumption of perfect certainty. Alan Auerbach and Laurence Kotlikoff (1987), for example, developed a model with 55 overlapping generations of individual lifecycle households, each with empirically plausible age-earnings profiles and utility parameters, and used the model to address tax policy and demographic issues in a regime in which all households are identical within a generation, and all generations know future earnings and interest rates. More recently, a line of inquiry has examined the effects of uncertainty on saving, generally in the context of highly stylized models. This research has shown that, in these models, uninsured earnings uncertainty can alter optimal behavior in a variety of important ways. (For a partial review of this precautionary saving literature, see Hubbard et al. [1994].) In two recent papers (Hubbard et al., 1993, 1994), we have combined these two strands of the literature by examining the implications of a life-cycle model of consumption, saving, and wealth accumulation subject to what we think are the three most important sources of uninsured idiosyncratic risk facing households: uncertainty about earnings, medical expenses, and length of life. Our intent has been to create a realistic model in which families live for many periods, working for part of their lives and retiring later in life. To parameterize the uncertainty facing families, we estimate tDiscussants: Christopher Carroll, Federal Reserve Board; John B. Shoven, Stanford University; Laurence Kotlikoff, Boston University.

This paper argues that a life cycle model can replicate observed patterns in household wealth accumulation after counting explicitly for precautionary saving and asset-based, means-tested social insurance. The authors demonstrate that social insurance programs with means tests based on assets discourage saving by households with low expected lifetime income. In addition, they evaluate the model using a dynamic programming model. Assuming common preference parameters across lifetime income groups, the authors are able to replicate the empirical pattern that low-income households are more likely than high-income households to hold virtually no wealth.

Two decades ago, most workers with pensions had a defined benefit (DB) plan. The employer made necessary contributions and investments to meet promised pension benefit payments when the employee retired. By 1993, the tide had turned; more than half of covered employees participated primarily in defined contribution (DC) pensions such as 401(k) plans [Employee Benefit Research Institute (EBRI), 1997]. This dramatic shift was associated as well with growing concerns about the emerging 401(k) plans. Some viewed 401(k) plans as a crisis waiting to happen when current generations, having made minimal (or no) contributions to their DC plan, retired with inadequate pension asset balances (Karen Ferguson and Kate Blackwell, 1995; Anne Willette, 1995). Another concern was that workers switching jobs would use the lump-sum distributions for houses, boats, or other purchases rather than reinvesting them in another retirement account [Ellen E. Schultz, 1995; U.S. Department of Labor (USDOL), 1998a]. In a similar vein, newly retired workers may not roll over the assets into an annuity and therefore risk spending down their retirement wealth too early (e.g., Jeffrey R. Brown and Mark J. Warshawsky, 2001). The most serious charge against 401 (k) plans,

Panel data are used to estimate the effect of marital separation on labor supply. Female labor supply increases substantially and male labor supply declines marginally, lending support to the theory of specialization within the household. The most interesting finding is that women began to increase their labor supply well before the actualsplit occurs, suggesting either that shocks to labor supply change divorce probabilities or that women with a higher likelihood of divorcework more. This argument is pursued by constructing and estimating a simultaneous model of labor supply and divorce risk.

Even among households with similar socioeconomic characteristics, saving and wealth vary considerably. Life-cycle models attribute this variation to differences in time preference rates, risk tolerance, exposure to uncertainty, relative tastes for work and leisure at advanced ages, and income replacement rates. These factors have testable implications concerning the relation between accumulated wealth and the shape of the consumption profile. Using the Panel Study of Income Dynamics and the Consumer Expenditure Survey, we find little support for these implications. The data are instead consistent with “rule of thumb,” “mental accounting,” or hyperbolic discounting theories of wealth accumulation.

The Importance of Bequests and Life-Cycle Saving in Capital Accumulation: A New Answer by Karen E. Dynan, Jonathan Skinner and Stephen P. Zeldes. Published in volume 92, issue 2, pages 274-278 of American Economic Review, May 2002

The question of whether higherlifetime income households save a larger fraction of their income was the subject of much debate in the 1950s and 1960s, and while not resolved, it remains central to the evaluation of tax and macroeconomic policies. We resolve this long-standing question using new empirical methods applied to the Panel Study of Income Dynamics, the Survey of Consumer Finances, and the Consumer Expenditure Survey. We find a strong positive relationship between saving rates and lifetime income and a weaker but still positive relationship between the marginal propensity to save and lifetime income. There is little support for theories that seek to explain these positive correlations by relying solely on time preference rates, nonhomothetic preferences, or variations in Social Security benefits. There is more support for models emphasizing uncertainty with respect to income and health expenses, bequest motives, and asset-based means testing or behavioral factors causing minimal saving rates among low-income households.