4
Personal Saving Behavior and Retirement Income Modeling: A Research Assessment

James M. Poterba

Two of the central tasks in the design of retirement income policy are forecasting the future financial status of elderly households under the present system of government policies and estimating the likely effects of changes in such policies. Both of these tasks depend critically on the assumptions maintained about the determinants of personal saving, subjects of perennial controversy in both theoretical and empirical economics.

This background paper identifies several aspects of personal saving behavior that bear on projecting the future financial status of elderly households, with a particular focus on the effects of Social Security, pensions, and other government policies that affect personal saving. This paper does not attempt to follow Kotlikoff (1984), Hurd (1990a), or the Organization for Economic Cooperation and development (1994) in presenting a systematic survey of existing research on saving behavior. Rather, it explores why it has been difficult to achieve a research consensus on a number of important aspects of personal saving behavior and how future study and data collection can contribute to developing such a consensus.

The paper is divided into seven sections. The first two are concerned with forecasting future financial status assuming the continuation of current government policies. The first section sketches a simple accounting framework for

I am grateful to Peter Diamond, William Gale, Michael Hurd, Lawrence Thompson, and members of the panel for helpful comments, and to the National Institute on aging, the National Science Foundation, and the Center for Advanced Study in the Behavioral Sciences for research support.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 123
Assessing Knowledge of Retirement Behavior 4 Personal Saving Behavior and Retirement Income Modeling: A Research Assessment James M. Poterba Two of the central tasks in the design of retirement income policy are forecasting the future financial status of elderly households under the present system of government policies and estimating the likely effects of changes in such policies. Both of these tasks depend critically on the assumptions maintained about the determinants of personal saving, subjects of perennial controversy in both theoretical and empirical economics. This background paper identifies several aspects of personal saving behavior that bear on projecting the future financial status of elderly households, with a particular focus on the effects of Social Security, pensions, and other government policies that affect personal saving. This paper does not attempt to follow Kotlikoff (1984), Hurd (1990a), or the Organization for Economic Cooperation and development (1994) in presenting a systematic survey of existing research on saving behavior. Rather, it explores why it has been difficult to achieve a research consensus on a number of important aspects of personal saving behavior and how future study and data collection can contribute to developing such a consensus. The paper is divided into seven sections. The first two are concerned with forecasting future financial status assuming the continuation of current government policies. The first section sketches a simple accounting framework for I am grateful to Peter Diamond, William Gale, Michael Hurd, Lawrence Thompson, and members of the panel for helpful comments, and to the National Institute on aging, the National Science Foundation, and the Center for Advanced Study in the Behavioral Sciences for research support.

OCR for page 123
Assessing Knowledge of Retirement Behavior assessing the financial status of households at various future dates. It also presents a brief overview of the financial status of households currently reaching retirement age in the United States, including the assets these households own and the typical level of such holdings. The subsequent section discusses three prominent models of saving behavior, the life-cycle model, the ''precautionary saving" model, and the bequest model, and examines the degree to which each model appears to be supported by available data. It also considers the central implications of each model for predicting future financial status. The next three sections are concerned with the effect of various policies on personal saving behavior. The third section discusses the interaction between Social Security and private saving, reporting on time series as well as cross-sectional studies of the Social Security offset. The fourth section presents a parallel discussion focusing on private pensions and other personal saving. The fifth section explores the recent increase in the popularity of targeted retirement saving vehicles, such as Individual Retirement Accounts (IRAs) and 401(k) plans, and how they are likely to affect the future financial status of the elderly. The sixth section examines one form of wealth accumulation that is particularly widespread: housing wealth accumulation. It considers patterns of home equity accumulation, the role of housing wealth in financing retirement income needs, and the existing body of research on housing decisions of elderly households. A brief concluding section outlines a number of unresolved research questions. It also describes how some of these questions may be addressed using available data sets and notes why others may be difficult to resolve even with additional data. FORECASTING FINANCIAL STATUS: AN ORGANIZING FRAMEWORK If the average wealth in 1994 of individuals born in year a is Wa, 1994, then the average wealth of individuals in this cohort in year t > 1994 can be forecast by where Sa,k is the forecast net saving in a year k by those born in year a,k > 1994, and r is the projected annual after-tax rate of return on savings. Because cohort averages conceal important differences in the wealth positions of households in a given cohort, and because retirement income policy is often concerned with the financial status of the least well-off groups in the population, a forecasting equation similar to the one above can be applied separately to different segments of a given cohort. Bernheim and Scholz (1993) follow such a strategy in separately forecasting wealth at retirement for those with college degrees, some college, and only high school education. Hurd (1992) describes a number of the more sophisticated microsimulation models that have been used to forecast retirement in-

OCR for page 123
Assessing Knowledge of Retirement Behavior come and also raises a number of other important issues that occur in actual forecasting contexts, such as the treatment of married versus single individuals and the modeling of programs such as Medicare and Medicaid that may alter private consumption expenditures. There are three important inputs to the simple forecasting equation presented above: the current wealth holdings of various cohorts, the prospective rate of return on different assets, and the age-specific pattern of net saving rates. Each of these inputs will be discussed in turn. Patterns of Wealth Holding for Pre-Retirement Cohorts The starting point for any projection of a cohort's future wealth holding is an estimate of its current wealth. For the majority of households over most of their lifetimes, financial asset holdings are relatively small. Principal assets are home equity, the present discounted value of employer-provided pension benefits, and net Social Security wealth. Each of these wealth components is subject to substantial measurement error. For financial assets, there are substantial rates of nonresponse in most surveys, and these problems are most severe among households with higher wealth levels. Juster and Smith (1994) discuss such problems of nonresponse and possible ways of improving survey performance when collecting asset information. Reconciling estimates of total wealth stocks from households surveys and aggregate data sources, such as the Flow of Funds, has also proven difficult with many existing data sets. For pension wealth, the data problem is not measurement error but the rarity of detailed data. Only a few surveys, such as the Survey of Consumer Finances Pension Provider Survey and the Health and Retirement Survey (HRS), provide sufficiently detailed information about the structure of employer-provided defined benefit pension plans and the employee's work history to permit accurate estimates of the present discounted value of pension wealth. Given the complexity of most defined benefit pension plans, another important issue is whether individuals understand the structure of their pension plan and its associated incentives. Some of the same problems arise in estimating Social Security wealth, but they are less severe because the current benefit formulas that apply to individuals are known, and an estimate of future benefits can be made given information on an individual's earnings history. The HRS in particular will provide Social Security earnings histories for many respondents, and this will make it possible to compute precise estimates of prospective Social Security benefits. To illustrate the current level of wealth holding and the relative importance of different assets in household portfolios, Table 4-1 summarizes the asset holdings of households that included individuals between the ages of 55 and 64, and between 65 and 69, in 1991. Mean and median asset holdings differ substantially

OCR for page 123
Assessing Knowledge of Retirement Behavior TABLE 4-1 Mean and Median Wealth Holdings, 1991, in Thousands of Dollars Wealth Holdings by Age and Type of Asset Mean Median Household heads age 65–69   Total financial assets $52.9 $14.0 Targeted retirement assets 10.9 0.0 Other financial assets 42.0 7.4 Home equity 65.0 50.0 Equity in other property 33.9 6.0 Employer-provided pensions 62.3 16.0 Social Security wealth 99.7 99.2 Net worth 312.3 261.4 Net worth excluding pension and Social Security wealth 150.3 96.6 Household heads age 55–64   Total financial assets $42.0 $8.3 Targeted retirement assets 12.9 0.0 Other financial assets 29.1 3.0 Home equity 57.8 36.0 Equity in other property 44.3 8.2 Net worth excluding pension and Social Security wealth 140.5 74.9   SOURCE: Poterba, Venti, and Wise (1994a). Tabulations are based on the Survey of Income and Program Participation, Wave 4, 1990 Panel. Sample size is 2,799 for the 55-to-64 age group and 1,525 for the 65-to-69 age group. in both age groups, reflecting the concentration of financial assets among a small group of households. For those approaching retirement age, the 55-to-64 age group, home equity is the single largest asset category that can be evaluated; neither pension wealth nor Social Security wealth can be estimated with any precision for this pre-retirement group. Mean (median) holdings of all financial assets for 55- to 64-year-olds total $42,000 ($8,300), with $12,900 (0) of this amount in targeted retirement saving accounts such as IRAs and 401(k)s. The Table 4-1 entries for 65- to 69-year-olds include information on the present value of both publicly provided and private pensions. These data show that Social Security wealth is the single most important asset for households in this age group. The mean value of employer-provided pensions is comparable to the mean value of home equity, but the median for private pensions is much lower, reflecting the fact that more households own homes than are covered by private pensions. Targeted saving accounts are less important for this group than for the 55-to 64-year-olds, in part because they may have been cashed out and in part because this cohort was eligible to accumulate funds in these accounts for a

OCR for page 123
Assessing Knowledge of Retirement Behavior shorter part of their working life than the younger cohort. A conclusion not illustrated in Table 4-1 comes from Auerbach, Kotlikoff, and Weil (1992), who suggest that one important pattern in the postwar period has been the increasing importance of annuitized wealth in the portfolios of elderly Americans. Except for households at the top of the income and wealth distribution, annuitized wealth, which includes private pension wealth and Social Security wealth, has become a much greater fraction of the household portfolio in the last three decades. One of the central statistical patterns illustrated in Table 4-1 is the relatively low level of liquid assets held by the majority of households reaching retirement age. This finding may in part reflect the use of data from the Survey of Income and Program Participation (SIPP). The SIPP tends to yield lower estimates of mean wealth than other data sets, as a result of low response rates among high-income households. This problem should not substantially affect estimated medians, however, and more generally, the SIPP patterns of low wealth holdings at advanced ages are confirmed in other data sets. Only a small fraction of households reaching retirement age have accumulated assets worth more than twice their pre-retirement annual income. The prevalence of households that save very little for retirement is an important consideration in evaluating how various policy interventions may affect the private saving rate. The current asset accumulation profiles of many households provide very limited opportunities to reduce financial wealth in response to initiatives that encourage new forms of wealth accumulation. The wide disparity in patterns of wealth accumulation also suggests the value of analyzing how potential policy shifts affect both those who currently accumulate substantial levels of financial assets in preparation for retirement and those who do not. Why so many households save so little for retirement is an unresolved issue. Some households may be myopic and fail to accumulate assets because they do not recognize the value of providing for their future. Some may be unlucky and experience lower earnings or higher expenses than they expected before reaching retirement. Others may have high discount rates and therefore choose to consume a high fraction of income while working at the expense of lower consumption when retired. Still others may have incorrect expectations about their retirement income from Social Security, private pensions, and other sources, or about life expectancy and post-retirement consumption needs. Further work on the relative importance of these and other explanations would be helpful in guiding public policies that are designed to influence private saving behavior. Rates of Return A second critical input to the forecasting model, and one that is particularly important for forecasting the future wealth of those individuals who actually accumulate assets on their own account, is the long-term average rate of return on these assets. This involves two distinct issues: which assets are individuals likely

OCR for page 123
Assessing Knowledge of Retirement Behavior to invest in, and what rates of returns are different types of assets likely to deliver? There is some evidence suggesting that individuals are very conservative in their portfolio choices. In spite of the well-documented superior performance of equities relative to fixed-income investments over long time horizons, data from the 1989 Survey of Consumer Finances show only 28 percent of households with heads aged 45 to 54 hold any corporate stock either directly or through a stock mutual fund. A similar pattern emerges with respect to investment of self-directed retirement assets. VanDerhei's (1992) analysis of 1989 Form 5500 filings shows that common stock accounts for 21 percent of the asset value in 401(k) plan accounts, while insurance company products such as Guaranteed Investment Contracts account for 41 percent of value. These investment patterns are important because the increasing importance of self-directed defined contribution pension plans makes financial status at retirement substantially dependent on individual investment decisions. In spite of this, the determinants of household portfolio composition are neglected in research on wealth accumulation. A related issue, and one that is largely beyond this paper, concerns the expected rate of return on different types of assets. The usual approach to projecting returns assumes that historical patterns of returns will persist into the future. A number of recent studies, however, have questioned this assumption. Schieber and Shoven (1994) argue that the coincident aging of the populations in most developed economies will raise the demand for assets in the short run and lower it several decades hence, thereby depressing rates of return on bonds and stocks over the financial planning horizon of the baby boom generation. This argument depends critically on the degree of integration of world capital markets and the absence of rapid growth in asset demand from currently less developed nations. Others, including Blanchard (1993) and Siegel (1994), argue that the prospective excess return of stocks over bonds is smaller than its historical value for a variety of reasons related to the financial market conditions of the last half century as well as to prospective conditions. Evaluating these arguments is an important part of long-term financial status forecasting. Age-Specific Patterns of Net Saving The third component of the forecasting equation is projection of the net saving behavior of households as they age. This projection involves several parts, such as the rate of saving out of earnings during the years when the household is working, the date of retirement and associated decline in earnings, and the rate at which the household decumulates or accumulates assets after retirement. Pre-retirement saving behavior is central for predicting the wealth that households will have when they retire, and this will be discussed in the next section. The rising number of years that households spend in retirement and the substantially greater incidence of poverty and other measures of financial hardship among

OCR for page 123
Assessing Knowledge of Retirement Behavior the oldest old, however, have drawn increased attention to the financial behavior of retired households. The question of whether retired households decumulate assets has been a subject of substantial empirical controversy. The first generation of research on this question, illustrated by Mirer (1979), suggested positive saving rates for retired households. A second wave of research, incorporating more careful definitions of saving and cohort rather than cross-sectional data on saving behavior, has found evidence of decumulation after retirement. Studies in this spirit include Bernheim (1987a), Hurd (1987, 1990b), and Attanasio (1992). Tests of whether households dissave after retirement have featured prominently in discussions of the life-cycle hypothesis. For forecasting the future retirement status of the elderly, however, the knife-edge question of whether retirees decumulate is less important than the rate at which accumulation or decumulation takes place. Forecasting the future saving behavior of currently young cohorts is complicated by the need to separate age and cohort effects in observed saving patterns. Will today's 45-year-old save as much as today's 55-year-old when he or she reaches age 55, or will the change in this person's saving over the next 10 years be the same as the change in saving between ages 45 and 55 was for the currently 55-year-old, or will neither of these modeling assumptions suffice? A number of studies, including Shorrocks (1975) and Attanasio (1993), have shown that the age-saving pattern for a cohort can be very different from the cross-sectional age-saving pattern. This suggests the need for cohort data for studying saving and also indicates the potential value of a formal model of household behavior for predicting the financial status of future elderly households. The difficulty of extrapolating age-specific saving rates across cohorts is illustrated in the Bosworth, Burtless, and Sabelhaus (1991) comparison of personal saving behavior in the 1960s and 1980s. This study shows that most age-specific saving rates fell between the early 1960s and late 1980s. Falling age-specific saving rates and shifting demographics have significantly affected the overall personal saving rate in the United States, which declined from an average of 6.7 percent of disposable income in the 1960s and 7.8 percent in the 1970s to 6.5 percent in the 1980s and only 4.5 percent in the 1990–1993 period. Explaining this decline represents one of the most important challenges to research on personal saving, and one important test of the potential value of a predictive model should be whether it would have tracked this decline. Explaining the decline in age-specific saving rates is also a priority research topic because information on the source of this decline will prove valuable in trying to project the future course of saving rates. The simple wealth-forecasting equation at the beginning of this section illustrates the wide range of ways in which personal saving may respond to a change in government policy. For example, a reduction in the accruing value of Social Security benefits for an individual in cohort a at time t could lead to a change in contemporaneous saving (Sa,t), to a change in one or many future saving decisions

OCR for page 123
Assessing Knowledge of Retirement Behavior (Sa,t+j), and potentially to changes in rates of return if individuals alter their investment behavior or if there are general equilibrium effects associated with the policy reform. MOTIVES FOR AND MODELS OF PRIVATE SAVING: WHAT DO WE KNOW? Projecting the future age-specific saving rates of cohorts that are currently not retired requires an economic model of saving decisions. Simple models of saving behavior, which yield strong predictions about the age-specific structure of saving rates, tend to be rejected by the data. More complex models are highly dependent upon parameter choices and yield results that are not conducive to simple presentation. For the last four decades, the life-cycle/permanent income hypothesis has been the dominant economic model for analyzing saving behavior. The central insight of this model is that individuals save during periods when labor income is high to avoid reductions in consumption when labor income is low. In its simplest form, without uncertainty, bequests, or distinctions between different types of assets, this model yields strong predictions about the age-specific pattern of saving and the link between saving rates and employment. For some simple parameterizations of household utility functions, it is possible to characterize the shape of the lifetime consumption profile, as in Summers (1981). Given estimates of a household's future income path, calculating the optimal, and hence predicted, level of saving in each year until retirement is a straightforward procedure. Researchers have used the stark predictions of the simple life-cycle hypothesis as "straw men" to motivate various empirical research projects on individual and aggregate consumption behavior. Existing household-level data provide very limited support for these predictions. As was noted in the last section, many households reach retirement age with virtually no financial assets, and there is substantial controversy as to whether or not households decumulate assets after retirement. These findings are not particular to the United States; data from six country studies in Poterba (1994) suggest that decumulation of household financial assets after retirement is the exception, rather than the rule, in developed nations. In addition, many individuals leave bequests, a finding that is difficult to reconcile with simple versions of the life-cycle model. The simple life-cycle model also appears incapable of explaining a variety of stylized facts about aggregate consumption fluctuations. Carroll (1992) outlines the difficulties: the strong positive correlation between aggregate consumption growth and income growth, the apparent link between consumption fluctuations and forecastable changes in income, and sketchy evidence suggesting a link between measures of income uncertainty and the level of consumption spending. Some recent studies such as Attanasio and Weber(1995) suggest that the statisti-

OCR for page 123
Assessing Knowledge of Retirement Behavior cal failures of the life-cycle hypothesis may be due to data aggregation. A larger set of studies, however, have accepted the failures, and tried to develop more complex theoretical models that could explain the findings. Recent research has emphasized the role of uncertainty and potential capital market constraints in affecting household consumption choices. A number of studies, notably Hubbard, Skinner, and Zeldes (1994) and Carroll and Samwick (1993), have developed numerical algorithms for finding the optimal consumption rules of households facing different types of uncertainty and various constraints on consumption. These models emphasize the existence of a precautionary motive for asset accumulation and suggest that the comparative statistics of saving behavior may be substantially different from the comparative statistics suggested by the life-cycle model. For example, if part of the stock of household wealth is accumulated to guard against future consumption downturns, wealth holdings may be relatively insensitive to changes in real rates of return. This may explain why empirical work on the intertemporal elasticity of substitution, such as Hall (1988), has suggested generally small effects of expected return fluctuations on the rate of consumption growth, even though simulations of life-cycle models with plausible parameter values suggest that much larger elasticities should be observed. Since expected return fluctuations as well as other exogenous shocks may have different effects on precautionary wealth and on wealth being held for retirement, recognizing precautionary motives for saving makes it more difficult to predict what should be observed in the data. Models that incorporate precautionary motives for asset holding suggest that it may be appropriate to focus on several aspects of the economic environment and the saving process that the standard life-cycle model does not recognize as relevant. Precautionary saving models show that various aspects of the social safety net, such as the availability of unemployment insurance and disability insurance, the nature of health insurance, and other factors that influence the chance that a household will experience a prolonged period of low income relative to expenditure needs, have an important influence on asset accumulation decisions. The Hubbard, Skinner, and Zeldes (1994) computational model can be potentially used to assess how various policy changes affect pre-retirement saving. One of the general research needs on private saving is further exploration of expanded life-cycle models, with realistic parameterizations of uncertainty, to better understand the characteristics of optimal saving behavior in these models. The precautionary saving model is not the only alternative to the life-cycle model. One strand of modeling concerns the role of bequest motives in explaining saving decisions. In the simplest life-cycle model, with a certain date of death and no altruistic links between generations, optimizing individuals would not leave bequests. Yet bequests are not only observed, but appear to account for a substantial share of the stock of household wealth; Kotlikoff (1988) and Modigliani (1988) disagree over the precise importance of bequest flows. Several explanations for bequests have been proposed. Andreoni (1989) considers

OCR for page 123
Assessing Knowledge of Retirement Behavior the possibility that donors enjoy giving for its own sake; Barro (1974) focuses on intergenerational altruism; Bernheim, Shleifer, and Summers (1985) stress the potential role of bequeathable wealth in improving the lifestyle of elderly individuals. Another explanation for bequests is simply that with uncertainty about the date of death, some individuals will die unexpectedly, thereby leaving to their heirs assets that they had planned to consume. Because these explanations of bequest behavior are not mutually exclusive, it has proven difficult to design and implement tests of how bequest considerations affect saving decisions. A third alternative to the standard life-cycle model, and one that has attracted substantial attention in the last decade, is the "behavioral approach to saving." Sheffrin and Thaler (1988; also Thaler, 1994) emphasize differences in the way households perceive different types of saving instruments and argue that different assets are held in distinct "mental accounts." A key implication of this view is that changes in the level of one asset may have relatively small substitution effects on holdings of other assets. This stands in contrast to the simple prediction of standard life-cycle models, in which different assets are perfect substitutes and total wealth at a given age is a summary statistic for the household's financial situation. Laibson (1994) presents a particularly intriguing justification for something like mental accounts by suggesting that consumers use hyperbolic rather than exponential discount factors to evaluate future events. While the mental accounts model seems to resonate well with intuitions about how individuals save, there is little empirical evidence that directly supports this model. In part, this reflects the lack of a well-articulated model that can be estimated or tested with data. A natural avenue for further research lies in drawing out the implications of alternative models of saving behavior and exploring how these models would modify life-cycle-based predictions of pre-retirement saving behavior. The possibility that individuals use rules of thumb or other simple heuristics in deciding how much to save is implicit in many behavioral discussions of personal saving. If such rules are relatively insensitive to prospective rates of return, then changes in such returns can have particularly large effects on the wealth of individuals at retirement. To illustrate this possibility, consider an individual who saves 5 percent of annual labor income each year between ages 35 and 65, regardless of the expected return on saving. If this person's real labor income is constant over this period, and the real rate of return is 3 percent per year, he or she will accumulate assets worth 2.43 times his or her annual labor income by age 65. If the annual rate of return is 6 percent, however, wealth at retirement will he 4.21 times annual labor income. Thus, if the flow of saving is insensitive to the expected return, the stock of wealth at retirement will be especially sensitive. The mental accounts framework may be particularly relevant to the current debate on whether households in the baby boom generation are saving enough to provide for themselves in retirement. Bernheim (1993) argues that baby boomers

OCR for page 123
Assessing Knowledge of Retirement Behavior are undersaving, and he bases this conclusion on an analysis of their projected nonhousing net worth at retirement. The Congressional Budget Office (1993) takes issue with this conclusion, pointing out that if housing wealth is included in net worth, then baby boomers are much closer to a wealth accumulation trajectory that will sustain pre-retirement levels of consumption after retirement. A key point of controversy between these studies involves the degree to which elderly households are prepared to decumulate net housing wealth during retirement. Bernheim (1993) implicitly assumes that elderly households are unable or unwilling to draw down their net housing equity to finance consumption during retirement, while the Congressional Budget Office, which focuses on net worth including housing assets, implicitly views housing wealth as fungible and equivalent to holdings of financial assets. Available evidence, discussed in greater detail below, suggests that the elderly may view housing as in a different mental account than other assets, and that they may be reluctant to downsize their homes or borrow against them to finance consumption needs. SOCIAL SECURITY AND PRIVATE SAVING Social Security is the government policy that most directly affects the financial status of elderly households. Benefit payments may not increase the financial status of the elderly by the full amount of the transfer, however, because households may anticipate such benefits and adjust their pre-retirement saving in response. The degree to which such offset occurs depends on many factors, including whether or not households correctly perceive the future value of their benefits, how retirement decisions are affected by Social Security, and whether individuals have operative bequest motives that link their utility to that of their children, who will be called upon to finance Social Security payouts. Although the net saving effect of the Social Security program is a first-order question for evaluating government policies toward retirement saving, the existing empirical literature in this area is weak and there is relatively limited prospect for improvement. The first wave of empirical research on the Social Security offset question involved time series estimates of how aggregate consumption responded to changes in the value of aggregate Social Security wealth. Studies in this vein include Feldstein (1974, 1982), Barro (1978), Barro and MacDonald (1979), and Leimer and Lesnoy (1982). Aaron (1982) surveyed the then-extant literature, but the limited research since the early 1980s makes his survey still current. It is extremely difficult to disentangle the effects of Social Security wealth, a time series with a strong trend, from other trending variables and other shocks in a relatively short time series. Auerbach and Kotlikoff (1983) present simulation evidence showing that reduced form equations relating aggregate consumption to aggregate income, wealth, and Social Security wealth are extremely sensitive to specification changes and to other factors in the economic environment.

OCR for page 123
Assessing Knowledge of Retirement Behavior the early 1980s, they have grown in popularity, and in 1993, the Census Bureau estimated that 39 million employees were eligible to participate in such plans. More than 25 million have participated. Poterba, Venti, and Wise (1994b) present tabulations of age- and income-specific participation rates in these plans and show that eligibility for a 401(k) increases with income, but is not strongly related to age. For those eligible, participation is unrelated to age and is above 60 percent for all income groups. The growing importance of these targeted retirement saving vehicles raises at least three important questions for the analysis of prospective saving behavior. First, how much are individuals who are currently contributing to these accounts likely to accumulate between now and the time they reach retirement? Second, how does the accumulation of balances in these accounts affect other forms of personal saving and wealth accumulation? Finally, how are these account balances likely to affect the evolution of wealth during retirement? Prospective Accumulation in Targeted Saving Accounts The first important issue concerns the asset balances that individuals with targeted retirement saving accounts are likely to accumulate by retirement age. Forecasting future contributions to these accounts raises many of the same problems of separating age effects from cohort effects that arise with respect to saving behavior more generally. Nevertheless, Poterba (1996) and Venti and Wise (1996) present some evidence on this question, using somewhat different approaches to impute future asset growth in targeted saving accounts. Poterba (1996) assumes that the age-specific changes in average 401(k) balances, computed from the 1987 and 1991 SIPP, will apply to the currently youngest cohorts as they age. Venti and Wise (1996) fit a polynomial in age to contributions and then use the fitted values from this equation to predict future contributions. In both cases the results suggest that if currently middle-aged 401(k) contributors behave like the older 401(k) contributors in earlier cohorts, they will accumulate very substantial balances in targeted retirement saving accounts. Forecasts in Poterba (1996) suggest that if a 3 percent annual real return on plan assets is assumed, households between the ages of 35 and 39 in 1991 could expect to reach age 60 to 64 with a pretax value of $26,025 (1991 dollars) in 401(k) plan assets. The after-tax value of the account is somewhat smaller and depends on the marginal federal and state/local income tax rate that the household would face when withdrawing assets. The reported average combines the roughly three-quarters of all households without targeted retirement saving accounts with the one-quarter with such accounts. The account balances for those with these accounts would average roughly $100,000 (1991 dollars). These average balances are an order of magnitude larger than those of households reaching retirement in the early 1990s. They also omit possible saving through other forms of targeted retirement saving accounts, such as IRAs.

OCR for page 123
Assessing Knowledge of Retirement Behavior One of the particularly difficult issues in modeling the long-run evolution of balances in targeted retirement saving accounts is that a substantial number of such accounts are terminated each year, and the assets are withdrawn as lump sums. In the March 1993 Employee Benefit Supplement to the Current Population Survey, 12.4 million individuals reported that they had received a lump-sum distribution from a retirement plan at some point. The median (mean) distribution was $3,500 ($10,800). Less than a quarter of these respondents indicated that they had rolled over all of their distribution into another tax-deferred retirement account, while 42 percent reported some rollover activity. Salisbury (1993) notes that recent changes in the tax treatment of lump-sum distributions are likely to affect the disposition of these payments. Further research on the link between individual characteristics, the incidence of 401(k) lump-sum distributions, and the disposition of these distributions is one of the top priorities for research on this aspect of the retirement saving system. Net Saving Effects A second critical issue related to the growing importance of assets in targeted retirement saving accounts concerns the interaction between rising balances in these accounts and other forms of personal saving. This is a topic of active research and controversy. A number of studies, prominently Venti and Wise (1987, 1990b) and Gale and Scholz (1994), have examined the net saving effect of contributions to IRAs. They reach very different conclusions: the former study suggests that most IRA saving offsets other saving, while the latter study suggests most IRA contributions are "net saving." Several more recent studies, including Engen, Gale, and Scholz (1994), Gale and Engen (1995), Poterba, Venti, and Wise (1995a, 1996), and Venti and Wise (1996), have focused on 401(k) accounts. Analyzing the net saving effects of targeted saving vehicles is complicated by the absence of any controlled experiments and the presence of substantial heterogeneity in household saving propensities. Since households with higher saving propensities may contribute to targeted retirement saving plans and also accumulate other assets more quickly, cross-sectional comparisons of the net worth of targeted retirement saving plan participants and nonparticipants can be an unreliable way to assess the net saving effects of these plans. This argument applies with particular force to comparisons of IRA savers and nonsavers, since there are no exogenous factors that determine eligibility for these plans. It is weaker as an objection to studies of 401(k) saving that use employer-determined eligibility for such plans as the source of variation in household saving opportunities although it is, at least in principle, possible for such variation to be related to worker attributes and therefore to underlying saving propensities. Poterba, Venti, and Wise (1996) and Gale and Engen (1995) discuss the saver heterogeneity problem at some length. Because most households hold relatively low bal-

OCR for page 123
Assessing Knowledge of Retirement Behavior ances of other financial assets, there is limited prospect that contributions to targeted retirement saving accounts will be offset on an ongoing basis by reductions in other asset balances. For some high-income households, however, the prospects are likely to be much greater. One margin on which targeted retirement saving plans may substitute for other saving is by replacing traditional pension plans. The 401(k) plans that were started in the early and mid-1980s do not appear to have replaced more traditional pension arrangements; in many cases they were converted thrift plans. Papke, Petersen, and Poterba (1996) present some evidence on these transitions. Venti and Wise (1996) find virtually no offset between private pension wealth and balances in 401(k) and similar targeted retirement saving accounts. The more recent growth of 401(k) plans may involve some displacement of alternative pension assets. Yakoboski and Reilly (1994) report that nearly three-quarters of current participants in salary reduction plans view these plans as their primary pension plan, a substantial increase from 1988 when 49.1 percent of participants viewed these as primary plans. Whether this change is due to a changing perception of a given set of retirement plans by the participants, to changes in the nature of the plans, or to changes in the set of workers who are covered by these plans, should be explored. Another unresolved aspect of the net saving debate is the degree to which IRA and 401(k) plan contributions are financed by increased borrowing. Poterba, Venti, and Wise (1995a) report median nonhousing household indebtedness for households without IRAs, but with 401(k) plan accounts, of $1,240 (1987 dollars) in 1991. Moreover, the median (mean) household debt levels for 401(k) participants without IRAs, measured in 1987 dollars, were $1,153 ($3,261) in 1984, $1,247 ($3,071) in 1987, and $1,240 ($3,223) in 1991. For those with both IRAs and 401(k) plan accounts, the pattern is similar. The low level of household indebtedness suggests that debt-for-401(k) swaps are unlikely to be a substantial factor in the run-up of 401(k) balances. The statistics presented above do not include mortgage debt, which could provide one channel for household borrowing against rising 401(k) and IRA wealth. This issue is not yet resolved. Gale and Engen (1995) present evidence suggesting substantial offset, while Poterba, Venti and Wise (1996) present data that support smaller offset effects. The phase out of tax deductibility for interest on unsecured consumer debt during the years since the Tax Reform Act of 1986 has made home equity loans a particularly attractive way for households to borrow. Mortgage borrowing has risen, hut the interaction with 401(k) contributions remains unclear. A further question related to the net saving effects of targeted retirement saving accounts involves the relationship between contributions to these accounts and aggregate statistics on personal saving. At precisely the same time that individual contributions to various targeted retirement saving became substantial, the National Income and Product Accounts measure of personal saving fell. This

OCR for page 123
Assessing Knowledge of Retirement Behavior negative relationship, while sometimes cited by opponents of such plans as evidence of their failure, is not well understood. None of the careful studies of the offset between contributions to targeted retirement saving accounts and other saving suggest a negative net effect on personal saving, and many studies suggest substantial positive effects. Understanding the relationship between household-level data on saving trends and aggregate information is therefore a research priority. Withdrawal Behavior A third important issue concerning the saving effects of targeted retirement saving accounts, but one that has received little research attention to date, concerns the pattern of withdrawals from these accounts after households become eligible for penalty-free withdrawals at age 59-1/2. Preliminary evidence on the flow of withdrawals and the age and income pattern of their receipt is presented in Andrews (1991), Fernandez (1992), Yakoboski (1994), and Poterba, Venti, and Wise (1995b), but none of these studies develops a formal model of what determines withdrawals. In analyzing withdrawals, it is important to consider the disposition of assets in targeted retirement saving accounts when households reach retirement, as well as pre-retirement withdrawals. A married couple in which the husband is 65 and the wife is 62 can expect to need retirement income for nearly 25 years. Yet we do not know whether households tend to draw down these accounts relatively rapidly, in which case these assets may not have a substantial effect on the wealth position of the oldest old, or whether such assets remain in targeted retirement saving accounts until age 70-1/2, the age at which withdrawals must begin. In the latter case, the growing accumulation of targeted retirement saving account assets may affect the financial status of the oldest old. HOUSING WEALTH AND OTHER PRIVATE SAVING For the typical household reaching retirement age in the early 1990s, home equity is the second most important component of net worth, after Social Security wealth. Housing wealth is typically much greater than net holdings of financial assets. Prospective patterns of both the accumulation of housing wealth before retirement and the decumulation after retirement therefore play an important role in defining the financial status of future retirees. Decisions made in youth and middle age, such as whether and when to buy a home, how rapidly to repay its mortgage, and whether to borrow against accumulating home equity, are critical determinants of the net housing wealth of prospective retirees. In addition, the rate of house price appreciation is a key factor in determining net housing wealth at retirement, just as the rate of return on financial assets plays a central role in determining financial net worth at retirement.

OCR for page 123
Assessing Knowledge of Retirement Behavior Although one widely cited study, Mankiw and Weil (1989), suggests that real house prices may decline substantially over the next several decades, a number of other studies, cited, for example, in Poterba (1991), call this conclusion into question. Further work both on housing consumption decisions in middle age and on the link between population aging and real house prices is therefore likely to yield important information that bears on the future financial status of elderly households. The popular perception that many elderly households are "house rich but cash poor" has stimulated research on two aspects of housing behavior: the extent to which elderly homeowners downsize their homes, thereby enabling them to use some of their wealth for other purposes, and the potential demand for reverse-annuity mortgages among elderly homeowners who for various reasons are not able to downsize. A number of research studies have analyzed housing consumption and mobility decisions of elderly households, examining in part the interaction between housing wealth and other portfolio components. For elderly homeowners in their sixties and early seventies, Feinstein and McFadden (1989) and Venti and Wise (1989, 1990a) estimate annual mobility rates of approximately 4 percent, with relatively little reduction of housing wealth conditional on moving. This is the result of two types of behavior whose relative importance is not known. First, some elderly homeowners sell their homes and reinvest the proceeds to buy new homes that are not substantially less expensive than their original homes. Second, some elderly homeowners sell their homes, buy less expensive replacement homes, and either spend part of the proceeds or transfer some of the proceeds off their balance sheets, for example, with gifts to children. The pattern of housing wealth accumulation among elderly households may be very sensitive to overall patterns of house price movements. For example, Merrill (1984) found evidence that net housing wealth increased with age among respondents to the Retirement History Survey, but this may reflect the secular rise in house prices during the data sample. The pattern of housing decumulation among the older old appears to be different from that for the younger old. Sheiner and Weil (1992) analyze data from the Current Population Survey, which includes a larger sample of older old households than the data sets used in previous studies. They find substantially greater rates of homeowner mobility than in earlier studies, particularly surrounding times of other shocks such as the death of a spouse, and they find a much greater incidence of homeowner-to-renter transitions among this group than among the younger elderly. The second issue that has attracted research attention concerns the demand for reverse annuity mortgages among elderly households. Capozza and Megbolugbe (1994) survey a number of recent studies on this issue and outline the key economic considerations in the demand for reverse annuity mortgages. Venti and Wise (1989) cast doubt on the stylized view of cash-poor, house-rich

OCR for page 123
Assessing Knowledge of Retirement Behavior elderly by showing that there is a positive correlation between cash income and net housing wealth. The elderly households with substantial housing assets to borrow against are much less likely to be cash constrained than elderly households with limited housing assets. Mayer and Simons (1994) present more recent evidence on the potential demand for reverse annuity mortgages, based on information in the 1984 and 1990 SIPPs. Their results suggest that at most one-quarter of households aged 65 or above would be strong candidates for reverse annuity mortgages, but they do not disaggregate the elderly to compare the young and old old. There is some evidence that elderly households in California and some parts of the Northeast, regions that have experienced rapid house price appreciation over the last two decades, would be particularly likely to raise their income if they participated in reverse annuity mortgage programs. The households that retired during the last two decades lived through a period of substantial increase in real house prices. Between 1974 and 1980, for example, real house prices rose nearly 30 percent (see Poterba, 1991). The associated capital gains on housing were probably unanticipated, so the value of net housing equity at retirement was probably greater than what these retirees would have predicted 10 or 20 years before retirement. Even if real house prices do not decline during the next 30 years, there is no strong basis for projecting a repetition of the real house price growth of the recent past. This suggests that future elderly households are likely to have a smaller relative stock of net housing wealth than their predecessors. Analyzing how this will affect saving behavior as households approach retirement, and asset profiles after retirement, is an important research issue. CONCLUSION AND FUTURE RESEARCH NEEDS This background survey suggests that for many of the important questions of interest with respect to private saving and the future financial status of the elderly, the existing research base does not provide detailed and convincing information on crucial parameter values and behavioral elasticities. In some cases, this is the result of data limitations that will be partly remedied by the HRS. The HRS will combine detailed projections of Social Security and private pension benefits once an individual retires with data on wealth holdings before and after retirement, and will permit new estimates of the effect of retirement income streams on other asset accumulation. It will also replace a number of data sets such as the Retirement History Survey that provide a dated indication of the financial status and saving behavior of elderly households. With respect to a number of issues, however, even the HRS will not resolve the research controversy. This is because existing empirical work on saving is hamstrung by the absence of exogenous shocks to saving opportunities. Past and potential future work is therefore limited to focusing on potentially contaminated sources of information on the relationship between Social Security wealth, pen-

OCR for page 123
Assessing Knowledge of Retirement Behavior sion wealth, and other personal financial saving. A key direction for future work should be modeling and evaluating the potential biases that result from individual heterogeneity in saving behavior and from the possibly endogenous choice by individuals of what types of saving to do (IRAs, 401(k)s, etc.) and even what firms to work for. The difficulty of resolving some perennial questions about saving behavior should not deflect research attention from assessing a number of the emerging trends in the saving and financial behavior of future retirees. Among the central issues that warrant study are the rise of defined contribution as opposed to defined benefit pension plans, the growing popularity of targeted retirement saving accounts as vehicles for personal saving, and the impact of growing numbers of two-earner couples who will reach retirement with multiple sources of income. REFERENCES Aaron, H.J. 1982. Economic Effects of Social Security. Washington, D.C.: Brookings Institution. Andreoni, J. 1989. Giving with impure altruism: Applications to charity and Ricardian equivalence . Journal of Political Economy 97:1447–1458. Andrews, E.S. 1991. Retirement savings and lump sum distributions. Benefits Quarterly 2:47–58. Attanasio, O. 1992. An Analysis of Life-Cycle Accumulation of Financial Assets. Unpublished manuscript. Stanford University. 1993. A Cohort Analysis of Saving Behavior by U.S. Households. NBER Working Paper #4454. Cambridge, Mass.: National Bureau of Economic Research. Attanasio, O., and G. Weber 1995. Is consumption growth consistent with intertemporal optimization? Evidence from the Consumer Expenditure Survey. Journal of Political Economy 103:1121–1157. Auerbach, A.J., and L.J. Kotlikoff 1983. An examination of empirical tests of Social Security and savings. Pp. 161–174 in E. Helpman, ed., Social Policy Evaluation: An Economic Perspective. New York: Academic Press. Auerbach, A.J., L.J. Kotlikoff, and D.N. Weil 1992. The Increasing Annuitization of the Elderly: Estimates and Implications. Unpublished manuscript. University of Pennsylvania. Barro, R.J. 1974. Are government bonds net wealth? Journal of Political Economy 82(6):1095–1117. 1978. The Impact of Social Security on Private Saving: Evidence from the United States Time Series. Washington, D.C.: American Enterprise Institute. Barro, R.J., and G.M. MacDonald 1979. Social Security and consumer spending in an international cross-section. Journal of Public Economics 11:275–289. Beller, D.J., and H.H. Lawrence 1992. Trends in private pension plan coverage. Pp. 59–96 in J.A. Turner and D.J. Beller, eds., Trends in Pensions 1992. Washington, D.C.: U.S. Department of Labor, Pension and Welfare Benefits Administration.

OCR for page 123
Assessing Knowledge of Retirement Behavior Bernheim, B.D. 1987a. Dissaving after retirement: Testing the pure life-cycle hypothesis. Pp. 237–274 in Z. Bodie, J. Shoven, and D. Wise, eds., Issues in Pension Economics. Chicago, Ill.: University of Chicago Press. 1987b. The economic effects of Social Security: Toward a reconciliation of theory and measurement. Journal of Public Economics 33:273–304. 1993. Is the Baby Boom Generation Preparing Adequately for Retirement? Summary Report. Princeton, N.J.: Merrill Lynch. 1994. Do Households Appreciate Their Financial Vulnerabilities? An Analysis of Actions, Perceptions, and Public Policy. Unpublished manuscript . Department of Economics, Stanford University. Bernheim, B.D., and J.K. Scholz 1993. Private saving and public policy. Pp. 73–110 in J. Poterba, ed., Tax Policy and the Economy Vol. 7. Cambridge, Mass.: MIT Press. Bernheim, B.D., A. Shleifer, and L.H. Summers 1985. The strategic bequest motive. Journal of Political Economy 93:1045–1076. Blanchard, O.J. 1993. Movements in the equity premium. Brookings Papers on Economic Activity 2:75–138. Blinder, A.S., R.H. Gordon, and D.E. Wise 1983. Social Security, bequests, and the life cycle theory of saving: Cross-sectional tests. Pp. 89–122 in F. Modigliani and R. Hemming, eds., The Determinants of National Saving and Wealth. New York: St. Martin's Press. Bosworth, B., G. Burtless, and J. Sabelhaus 1991. The decline in saving: Evidence from household surveys. Brookings Papers on Economic Activity 1:183–256. Cagan, P. 1965. The Effect of Pension Plans on Aggregate Saving: Evidence from a Sample Survey. National Bureau of Economic Research Occasional Paper 95. New York: Columbia University Press. Capozza, D., and I. Megbolugbe 1994. Introduction [to special issue on housing finance for the elderly]. Journal of the American Real Estate and Urban Economics Association 22:197–203. Carroll, C.D. 1992. The buffer-stock theory of saving: Some macroeconomic evidence. Brookings Papers on Economic Activity 2:61–135. Carroll, C.D., and A.A. Samwick 1993. The Nature and Magnitude of Precautionary Wealth. Unpublished manuscript. Federal Reserve Board of Governors, Washington, D.C. Congressional Budget Office 1993. Baby Boomers in Retirement: An Early Perspective. Washington, D.C.: U.S. Government Printing Office. Diamond, P.A., and J.A. Hausman 1984. Individual retirement and saving behavior. Journal of Public Economics 23:81–114. Edlin, A.S. 1993. Is college financial aid equitable and efficient? Journal of Economic Perspectives 7:143–158. Engen, E.M., W.G. Gale, and J.K. Scholz 1994. Do savings incentives work. Brookings Papers on Economic Activity 1994(1):85–180. Feinstein, J., and D. McFadden 1989. The dynamics of housing demand by the elderly: Wealth, cash flow and demographic effects. Pp. 55–86 in D.A. Wise, ed., The Economics of Aging. Chicago, Ill.: University of Chicago Press.

OCR for page 123
Assessing Knowledge of Retirement Behavior Feldstein, M.S. 1974. Social Security, induced retirement, and aggregate capital accumulation. Journal of Political Economy 82(5):905–926. 1982. Social Security and private saving: Reply. Journal of Political Economy 90:630–641. 1995. College scholarship rules and private saving. American Economic Review 85(June):552–566. Feldstein, M.S., and A. Pellechio 1979. Social Security and household wealth accumulation: New microeconometric evidence. Review of Economics and Statistics 61(3):361–368. Fernandez, P.A. 1992. Preretirement lump sum distributions. Pp. 285–318 in J.A. Turner and D.J. Beller, eds., Trends in Pensions 1992. Washington, D.C., U.S. Department of Labor, Pension and Welfare Benefits Administration. Gale, W.G. 1995. The Effects of Pensions on Wealth: A Re-evaluation of Theory and Evidence . Unpublished paper. Brookings Institution, Washington, D.C. Gale, W.G., and E. Engen 1995. Debt, Taxes, and the Effects of 401(k) Plans on Household Wealth Accumulation. Unpublished manuscript. Brookings Institution, Washington, D.C. Gale, W.G., and J.K. Scholz 1994. IRAs and household saving. American Economic Review 84(5):1233–1260. Hall, R.E. 1988. Intertemporal substitution in consumption. Journal of Political Economy 96:339–357. Hubbard, R.G. 1986. Pension wealth and individual saving. Journal of Money, Credit, and Banking 18:167–178. Hubbard, R.G., J. Skinner, and S.P. Zeldes 1994. The importance of precautionary motives in explaining individual and aggregate saving. Carnegie-Rochester Conference Series on Public Policy 40(June):59–126. Hurd, M. 1987. Savings of the elderly and desired bequests. American Economic Review 77(3):298–312. 1990a. Research on the elderly: Economic status, retirement and consumption and saving . Journal of Economic Literature 28(2):565–637. 1990b. Wealth Depletion and Life Cycle Consumption by the Elderly. NBER Working Paper #3472. Cambridge, Mass.: National Bureau of Economic Research. 1992. Forecasting the Consumption, Income, and Wealth of the Elderly. Unpublished manuscript. Department of Economics, The State University of New York, Stoney Brook. Juster, F.T., and J.P. Smith 1994. Improving the Quality of Economic Data: Lessons From the HRS. Unpublished manuscript. Institute for Social Research, University of Michigan. King, M.A., and L. Dicks-Mireaux 1982. Asset holdings and the life cycle. Economic Journal 92:247–267. Kotlikoff, L.J. 1984. Taxation and savings: A neoclassical perspective. Journal of Economic Literature 22:1576–1629. 1988. Intergenerational transfers and savings. The Journal of Economic Perspectives 2(2):41–58. Krueger, A., and J.S. Pischke 1992. The effect of Social Security on labor supply: A cohort analysis of the notch generation. Journal of Labor Economics 10:412–437.

OCR for page 123
Assessing Knowledge of Retirement Behavior Laibson, D. 1994. Golden Eggs and Hyperbolic Discounting. Unpublished manuscript. Harvard University. Leimer, D.R., and S. Lesnoy 1982. Social Security and private saving: New time series evidence. Journal of Political Economy 90:606–629. Leimer, D.R., and D.H. Richardson 1992. Social Security, uncertainty adjustments, and the consumption decision. Economica 59(235):311–335. Mankiw, N.G., and D. Weil 1989. The baby boom, the baby bust, and the housing market. Regional Science and Urban Economics 19:235–258. Mayer, C.J., and K.V. Simons 1994. A new look at reverse mortgages: Potential market and institutional constraints. New England Economic Review March/April:15–26 Merrill, S. 1984. Home equity and the elderly. Pp. 197–227 in H. Aaron and G. Burtless, eds., Retirement and Economic Behavior. Washington, D.C.: Brookings Institution. Mirer, T.W. 1979. The wealth-age relationship among the aged. American Economic Review 69:435–443. Modigliani, F. 1988. The role of intergenerational transfers and life cycle saving in the accumulation of wealth . The Journal of Economic Perspectives 2(2):15–40. Munnell, A. 1982. The Effect of Social Security on Private Saving. Washington, D.C.: Brookings Institution. Organization for Economic Cooperation and Development 1994. Taxation and Household Saving. Working Paper No. 2 on Tax Analysis and Tax Statistics. Paris. Papke, L., M. Petersen, and J.M. Poterba 1996. Did 401(k) replace other employer-provided pensions? In D. Wise, ed., Further Studies in the Economics of Aging. Chicago, Ill.: University of Chicago Press. Poterba, J.M. 1991. House price dynamics. Brookings Papers on Economic Activity 2:143–203. 1996. 401(k) plans and personal saving in the United States. In S. Schieber and J. Shoven, eds., Public Policy Toward Pensions. Washington, D.C.: Twentieth Century Fund. Poterba, J.M., ed. 1994. International Comparison of Household Saving. Chicago, Ill.: University of Chicago Press. Poterba, J.M., S.F. Venti, and D.A. Wise 1994a. Targeted retirement saving and the net worth of elderly Americans. American Economic Review 84(2):180–185. 1994b. 401(k) plans and tax-deferred saving. Pp. 105–138 in D. Wise, ed., Studies in the Economics of Aging. Chicago, Ill.: University of Chicago Press. 1995a. Do 401(k) contributions crowd out other personal saving? Journal of Public Economics 58:1–32. 1995b. Lump Sum Distribution from Retirement Savings Plans: Receipt and Utilization. NBER Working Paper #5298. Cambridge, Mass.: National Bureau of Economic Research. 1996. Do retirement saving programs increase saving? Journal of Economic Perspectives (forthcoming).

OCR for page 123
Assessing Knowledge of Retirement Behavior Salisbury, D. 1993. Policy implications of changes in employer pension protection. In Employee Benefit Research Institute, Pensions in a Changing Economy . Washington, D.C.: Employee Benefit Research Institute. Samwick, A.A. 1994. The Limited Offset Between Pension Wealth and Other Private Wealth: Implications of Buffer Stock Saving. Unpublished manuscript. Dartmouth College. Schieber, S., and J.B. Shoven 1994. The Consequences of Population Aging on Private Pension Fund Saving and Asset Markets. NBER Working Paper #4665. Cambridge, Mass.: National Bureau of Economic Research. Sheffrin, H.M., and R. Thaler 1988. The behavioral life-cycle hypothesis. Economic Inquiry 26:609–643. Sheiner, L., and D.N. Weil 1992. The Housing Wealth of the Aged. NBER Working Paper #4115. Cambridge, Mass.: National Bureau of Economic Research. Shorrocks, A.F. 1975. The age-wealth relationship: A cross-section and cohort analysis. Review of Economics and Statistics 57:155–163. Siegel, J.J. 1994. Stocks for the Long Run: A Guide to Selecting Markets for Long Term Growth. Homewood, Ill.: Irwin. Summers, L.H. 1981. Capital taxation and capital accumulation in a life cycle growth model. American Economic Review 71:533–544. Thaler, R.H. 1994. Psychology and savings policies. American Economic Review 84(2):186–192. VanDerhei, J. 1992. New evidence that employees choose conservative investments for their retirement funds. Employee Benefit Notes 13(February):1–3. Venti, S.F., and D.A. Wise 1987. IRAs and saving. Pp. 7–52 in M. Feldstein, ed., The Effects of Taxation on Capital Accumulation. Chicago. Ill.: University of Chicago Press . 1989. Aging, moving and housing wealth. Pp. 9–48 in D.A. Wise, ed., The Economics of Aging. Chicago, Ill.: University of Chicago Press. 1990a. But they don't want to reduce housing equity. Pp. 13–29 in D.A. Wise, ed., Issues in the Economics of Aging. Chicago, Ill.: University of Chicago Press. 1990b. Have IRAs increased U.S. saving? Quarterly Journal of Economics 105:661–698. 1996. The wealth or cohorts: Retirement saving and the changing assets of older Americans. In S. Schieber and J. Shoven, eds., Public Policy Toward Pensions. Washington, D.C.: Twentieth Century Fund. Yakoboski, P. 1994. Retirement program lump-sum distributions: Hundreds of billions in hidden pension income. EBRI Issue Brief Number 146(February). Washington, D.C.: Employee Benefit Research Institute. Yakoboski, P., and A. Reilly 1994. Salary reduction plans and individual saving for retirement. EBRI Issue Brief Number 155(November). Washington, D.C.: Employee Benefit Research Institute.