Cohort Differences in Wealth and Pension Participation of Near-Retirees

Summary

The approaching retirement of the baby-boom generation has attracted both research and public policy attention. Many social and economic changes occurred during the second half of the twentieth century, changes that are likely to affect the retirement economic security of recent cohorts in many ways. In this article, using data from the Health and Retirement Study (HRS), a longitudinal, nationally representative survey of older Americans, we compare potential retirement economic resources—pension participation and nonpension net worth—of two cohorts of near-retirees. Particularly we look at individuals born from 1933 through 1939, often referred to as depression babies, who were ages 55–61 in 1994 and the more recent cohort consisting of individuals of the same ages (55–61) in 2004, who were born from 1943 through 1949.

Our findings indicate that the more recent cohort of near-retirees has a significantly higher pension participation rate over their working life, and therefore greater opportunity to establish pension income through their working life, compared with the earlier cohort (82 percent versus 64 percent). The increase in pension participation was more pronounced among the recent cohort of women, an expected outcome given the increase in labor force participation of women over the past half century. As a result, although differences by sex in pension participation remained significant, the gap has narrowed for the recent cohort of near-retirees. In addition, we find that the gap in participation rate between those in the highest and the lowest wealth quintiles has widened over time (from 22 percent in 1994 to 26 percent in 2004).

For both cohorts of near-retirees, the evidence indicates that those without a pension have much lower levels of net total worth than those who report having a pension. The pattern that emerges for both cohorts is that about one-fifth of individuals aged 55–61 hold little or no wealth at all, whereas about two-fifths hold a substantial amount of wealth. In addition, housing equity, which rarely is used to finance consumption in retirement, comprises more than one-half of total nonpension net worth for about 60 percent of all households, leaving—on average less than $45,000 jointly in nonhousing wealth and IRA/Keogh assets—a much smaller amount of wealth that is readily accessible if the need arises.

The fact that many near-retirees (about 40 percent) in the lowest-two wealth quintiles have no pension to potentially draw income from, coupled with the very low level of total nonpension wealth raises concern about their income security in retirement; they may be likely to rely heavily on Social Security, rely on welfare programs, or continue work in retirement.

Introduction

In the United States retirement incomes are largely derived from three pillars: Social Security, employer pensions, and personal saving (nonhousing wealth and home equity).¹ In addition, individuals may continue working in retirement to supplement their retirement income, or they can receive income from welfare programs. In this article we focus on two potential sources of income in retirement: (1) employer pension participation and (2) total nonpension wealth. Employer pensions play an important role in assuring a comfortable retirement. Participation in an employer pension plan potentially generates retirement income. Nonhousing wealth is readily available for spending, and some assets such as stocks and bonds generate income flows. Home equity, an important component of total wealth, can also be used to finance retirement through an equity line of credit, a reverse mortgage, or an outright sale (Eschtruth, Sun, and Webb 2006). Only a small proportion of households draw down their housing wealth, however. Fisher and others (2007) using data from the Consumer Expenditure Surveys find that only 1 percent to 4 percent of persons aged 60 or older had a home equity loan from 1998 through 2003. To understand the extent to which families use housing equity to finance consumption in retirement, Venti and Wise (2001) examine data from several household surveys. The authors conclude that, on average, home equity is not liquidated to support general nonhousing consumption needs as households age.

Shift from Defined Benefit to Defined Contribution Plans

Many social and economic changes have occurred since World War II—changes that are likely to affect the retirement income security of baby boomers in many ways. Major changes have occurred in the past few decades in employer-provided pension plans—a shift from defined benefit (DB) plans where the main responsibilities rest with the employer, toward defined contribution (DC) plans where the employee is responsible for his or her economic security in retirement (Munnell and Sunden 2004; Costo 2006). DB plans, usually funded by the employer, provide retirement benefits based on a formula typically involving the final salary, age, and years of service. In contrast, DC pensions are savings accounts where employer and employee contributions into the account are invested and retirement benefits will depend on the account balance at retirement. Using data from the Form 5500, which employers file annually with the Internal Revenue Service (IRS) and the Department of Labor, Buessing and Soto (2006) provide evidence of a dramatic shift since 1981 in participation of private-sector wage and salary workers from DB to DC pensions. In 1981, 27 percent of private-sector workers participated only in a DB plan; 9 percent participated only in a DC plan; and 11 percent had both a DB and a DC plan. Almost two decades later in 1999, about 7 percent participated only in a DB plan; 29 percent participated only in a DC plan; and 14 percent participated in both types of plans.

Several factors have influenced such a shift. First, because of their portability across jobs, employees find DC plans attractive (Munnell and Sunden 2004). Second, structural changes in the U.S. economy have occurred, such as the shift in the labor force from the manufacturing sector and unionized jobs where DB plans are more often offered, toward the services sector and nonunionized jobs where DC plans tend to be offered (Wiatrowski 2004). Several studies have attributed about 50 percent of the decline in DB plans to such structural changes (Andrews 1992, Gustman and Steinmeier 1992, Ippolito 1995). Third, changes in the law since the 1974 Employee Retirement Income Security Act (ERISA),² with respect to funding requirements for DB plans or the introduction of 401(k) plans, have decreased incentives for employers to offer DB plans. Schieber (1999) documents a shift in the focus of the federal regulation from limiting the loss of federal revenues through excessive deductions associated with employer-sponsored retirement plans prior to ERISA to increasing short-term federal tax collections in the 1980s and 1990s.³ Fourth, pension accounting standards used for calculating long-term pension obligations of DB plans have changed. Schieber (1999) observes that both changes in Financial Accounting and Standards Board (FASB) rules and changes in regulatory measures adopted since the early 1980s have slowed the funding of pension plans for the baby-boom generation during the early part of their career. This contributed to increases in unfunded liabilities that were made more explicit to employers with subsequent changes in FASB rules. Finally, employers' pension liabilities may have increased because of decreases in mortality across all ages and especially among those aged 65 or older. All of these changes have increased employers' costs of providing DB plans and weakened the competitive position of firms with large pension liabilities. Furthermore, such costs have become even more evident in the face of a global economy where U.S. establishments compete with international firms that may not provide occupational pensions. Schieber (1999) concludes that such changes are likely to have significant implications for the retirement security of the baby-boom generation because this is the first generation to have spent its whole career under such a regulated environment of the employer pension system.

In short, while over the past few decades pension coverage rates have remained around 50 percent, all of these factors have contributed to the shift in employer preferences toward DC plans and therefore to a shift in the type of plans these employers offer. According to Munnell and Sunden (2004) there was a "virtual halt" in the formation of new DB pension plans in the 1980s and a surge in the adoption of 401(k)-type pensions by new businesses.

Implications of the Shift in Pension Plans for Retirement Income Security

The shift in pension types that are available to employees has important implications for retirement income security partly because of their different enrollment procedures. In traditional DB plans, employees are automatically included in the plan. In most DC plans, employee participation is not automatic, and employees have to make a decision whether to participate in the plan or not (Munnell and Sunden 2004; Copeland 2006). The employees' responsibilities and risks associated with such plans may discourage them from participating. Research by Madrian and Shea (2001), Choi and others (2002, 2004a, 2004b), and Iyengar, Huberman, and Jiang (2004) have documented delayed participation or lower levels of participation in DC plans than in DB plans, resulting from the complexity of the decision on appropriate contribution rates and investment asset allocations. Madrian (2005) notes that another reason that many employees delay enrolling is that they can put it off. The 2006 Pension Protection Act included clauses permitting employer provision of financial investment advice and automatic enrollment into a default investment fund (American Association of Retired Persons (AARP) 2007; IRS 2007). To the extent that employers will implement such provisions, the participation rate in DC plans is expected to increase in the future. According to Madrian (2005, 11), "the most effective mechanism for increasing savings plan participation is automatic enrollment. Firms with automatic enrollment have participation rates ranging from 85% to 95% among those employees who are impacted." The author cautions, however, that one of the drawbacks of automatic enrollment is the employer-chosen default contribution rate and asset allocation.

Another reason that the shift in the type of pension may affect retirement income security is that DB and DC plans differ with respect to risks associated with them. Traditional DB plans provide protection for longevity risk by paying benefits in the form of a life annuity (that is, a monthly benefit throughout one's life). In addition, since ERISA, DB plans provide spousal and survival benefit rights to the spouse of an eligible employee. The main risks for participants of DB pensions are employee job mobility or job separation, which reduces pension value, and the risk of pension termination from the employer either through bankruptcy or conversion. In recent years, several employers have either terminated or frozen their traditional DB plans, whereas others have converted them to a "cash balance" account that accrues value similar to a DC account (Beller 2005; Cahill and Soto 2003). While the Pension Benefit Guarantee Corporation (PBGC) insures against bankruptcy or termination, benefit payments for DB plans taken over by the PBGC are typically modest relative to the former DB plan.

In DC plans, employees bear all risks involving the adequacy of contributions, investment risk, management of money in retirement, and longevity risk, in contrast to DB plans where the employer is the bearer of such risks. DC plans, in general, offer payments of benefits as a single lump sum or payments that are distributed over a set period of time, or they allow transfers into a tax-sheltered Individual Retirement Account (IRA) from which the retiree withdraws money. Some plans offer monthly payments through an annuity.⁴ Hurd and Panis (2006) using data from the Health and Retirement Study (HRS) find that among workers that separated from a job between 1992 and 2000, about 15 percent rolled over their pension entitlement into IRAs, whereas about 12 percent cashed it out. The cash-out entitlements represented only a small proportion (5.3 percent) of entitlement dollars. Furthermore, evidence suggests that few persons buy annuities, and the main form of distributions from DC accounts is a lump-sum amount that is rolled over into another account (either tax-sheltered or not).⁵ At that point the individual is responsible for managing the process of investing and spending down the account balances, which introduces the risk of "prematurely depleting the account" and outliving one's pension wealth, that is, longevity risk (Society of Actuaries 2006).

DC pensions have less protection for surviving spouses than DB plans. Unless an annuity payment is available, most DC plans do not offer a survivor annuity. There are rules for such plans that protect the surviving spouse as a beneficiary at one's death. However, account balances can be withdrawn in any form at the employee's discretion, without spousal consent when one reaches a distribution date such as retirement or termination of employment.

Despite the drawbacks, DC plans have the potential of generating high account balances because of the compounding effect of long-term retirement saving given the individual made contributions over a substantial period of his or her working life and made sound investment decisions. Simulations indicate that a lifetime DC plan can generate as much or more money than DB plans but usually do not (Munnell and Sunden 2004; Poterba and others 2006). It remains to be seen in years to come whether individuals with such plans will be better off in retirement.

Aside from these developments in the pension arena, dramatic changes have occurred in marriage, family, and women's roles within the family and the workplace (Farley 1996; O'Rand and Henretta 1999; Society of Actuaries 2006; Butrica, Iams, and Smith 2003; Goldin 2006). More specifically, over the past four decades, the age at first marriage increased, the divorce rate increased, and the total fertility rate decreased to the replacement rate level. Multiple marriages over a lifetime also became more common. Furthermore, there has been a "quiet revolution" in perspectives among women about their changing roles, which began in the 1970s and continue today (Goldin 2006), toward increasing labor market experience and earning capacity over their lifetime, and shifting identities from home and family toward economic independence. These changes have fundamentally transformed the occupations and lifetime earnings of many women born after World War II. Moore (2006) observed that as women's labor force participation rates increased over the past half century, succeeding cohorts of women have increased their opportunities for pension coverage. As a result, women's expected retirement incomes are likely to have increased.

Different cohorts, in particular the more recent ones, may be differently affected by such social and economic changes, which in turn are likely to affect pension and nonpension wealth and therefore retirement income. Motivated by all of these developments, in this article, we compare potential retirement economic resources of two cohorts in 1994 and 2004, at ages (55–61) near eligibility for Social Security retired-worker benefits (that is, near-retirees).⁶ Particularly, we look at individuals born from 1933 through 1939, often referred to as "depression babies" who were aged 55–61 in 1994, and the more recent cohort consisting of individuals of the same age (55–61) in 2004 who were born from 1943 through 1949.⁷ Because this age group is 5–10 years away from the Social Security full retirement age, there is time to accumulate additional wealth.⁸ Thus, we believe that information on pension participation and personal saving available at such ages provides a fairly accurate picture of these potential income resources at retirement.

It is important to note that there is a major difference between these two cohorts in the household structure the cohort members established in their twenties and thirties. For the earlier cohort, the norm in the 1950s was to marry and form one-earner households with the husband as the "breadwinner." In contrast, for the later cohort, because of the so-called "quiet revolution," being in a dual-earner household in the 1970s and 1980s was more common. Such a difference is expected to translate into differences in economic resources available in retirement.

The remainder of the article is organized as follows. We describe the data and then present and discuss results of lifetime access to pensions and pension types for the two cohorts, by selected characteristics and by household type. Among couple households for each of our two cohorts, we compare husbands' pension participation and pension types (based on their own employment), wives' pension participation and pension types (based on their own employment), and couples as a unit (based on either spouse's employment). Next, we examine wealth holdings across cohorts by pension type and household composition (couples, single women, and single men). Our conclusions are presented in the last section.

Data Issues

In this analysis we use data from the Health and Retirement Study (HRS), a longitudinal, nationally representative survey of older Americans aged 51 or older and their spouses of any age. The first wave of interviews was conducted in 1992 and follow-up interviews were conducted every other year since then (see Table A-1 for an illustration of different birth cohorts as they enter the survey and as they age throughout the survey). Because of our interest in changes over a decade, for this analysis we use the 1994 and the 2004 waves. More specifically, we restrict our samples to those individuals aged 55–61 in 1994 (born in the 1933–1939 period) and those of the same age in 2004 (born in the 1943–1949 period).⁹

We focus in particular on two potential income resources for retirement: (1) pension participation as a measure of potential income from an employer pension, and (2) total nonpension net worth.¹⁰ Of course, a more complete picture would include pension and Social Security wealth, but calculating such wealth at retirement age is outside the scope of this article, however.¹¹ Furthermore, pension participation and pension types provide information only on the opportunity to establish pension income, but do not tell us whether increased pension participation and shifts in pension type translate into higher or lower levels of pension wealth for the more recent cohort of near-retirees relative to the earlier one.

It is common in previous research to look at pension coverage of workers in the current job at a point in time. However, a worker's access to and decision to participate in a pension plan will vary across jobs and at different stages of his or her working life. Moreover, some people in this age group (55–61), in particular, may have retired from a career job with a DB plan, for example, and may have taken another job that offers a DC plan (or no plan at all). Focusing on pension coverage and type of pension in the current job will classify individuals as having a "DB-only" plan, a "DC-only" plan, or "no pension" for that job. Looking only at pension coverage in the current job is likely to underestimate lifetime access to pensions to the extent that individuals who do not have a pension in their current job might have had one in a previous job(s).¹²

In contrast to previous research that focuses on pension coverage of workers in the current job, we focus on the broader measure—access to pensions over one's working life (to the extent it is retrospectively reported). This broader measure provides a better indication of the opportunity to establish pension income. The HRS collects information on all pension plans on the current job for respondents currently working and on the most recent job for respondents not currently working.¹³ In addition, it collects information on all pension plans for up to three jobs previously held (for at least 5 years) by either working or nonworking respondents. Our lifetime measure of pension participation is defined as ever having had a pension in a job (whether current, last, or previous jobs) as reported in the current wave or in any of the previous waves in which we observe the individual.¹⁴ We define variables for pension types in the same way.¹⁵ In addition, focusing on pensions on an individual basis or on a household basis will provide different estimates. In married households, spouses may have access to pension income through their spouse's pension. Therefore, we construct a lifetime measure of pension participation for couples as a unit, defined as at least one of the spouses having ever participated in a pension; we do the same for pension types.

With respect to wealth, our variables of interest, which come from the RAND Corporation's HRS data file,¹⁶ are: total net worth, total nonhousing wealth, home equity, assets in individual retirement accounts (IRA/Keogh), homeownership rate, and IRA/Keogh ownership rate. Total net worth is the sum of nonhousing wealth, home equity, and IRA/Keogh assets; it does not include employer pension and Social Security wealth. Total nonhousing wealth includes financial assets, business, vehicles, and other properties or assets, net of debt.¹⁷

Cohort Differences in Lifetime Access to Pensions and Pension Types

In this section we provide evidence on differences in lifetime pension participation and pension types for the cohort of near-retirees in 1994 and in 2004.¹⁸ In 2004 about 72 percent of near-retirees reported participating in a pension over their working life (Table 1).¹⁹ As expected, pension participation is strongly associated with education level and household income. Near-retirees with less than a high school degree are significantly less likely to have participated in a pension over their working life than those with a college degree (39 percent versus 84 percent, respectively).²⁰ In addition, only 46 percent of those in the lowest household income quintile have participated in a pension, compared with 83 percent of those in the highest quintile. Table 1 indicates that about 64 percent of near-retirees in 1994 report being covered by a pension over their lifetime, significantly less than their counterparts in 2004. Over the study period, the increase in the proportion of near-retirees with lifetime pension participation was more pronounced in particular among women, Hispanics, widow(er)s, and part-time employees. It is plausible to attribute the increase in pension participation among women of the recent cohort to their higher level of education and increased labor force attachment. As a result, although gender differences in pension participation remained significant, the gap has narrowed for the recent cohort of near-retirees.

With respect to pension type, the recent cohort of near-retirees in 2004 was almost half as likely as their counterparts in 1994 to have a DB-only plan over their working life. Furthermore, about 38 percent of the recent cohort of near-retirees had the opportunity to establish pension income from both a DB and a DC plan over their working life, a significantly higher proportion, compared with the earlier cohort (about 25 percent) in 1994. The cohort differences in the overall figures of pension participants having at least a DB plan (either as DB only or both DB and DC) and at least a DC plan (either as DC only or both DC and DB) over their working life are noteworthy. Although the prevalence of people with at least a DB plan is almost the same for the two cohorts of near-retirees (about 52 percent), the proportion that has had at least a DC plan is substantially higher for the more recent cohort of near-retirees (37 percent in 1994 versus 56 percent in 2004). To corroborate the prevalence of lifetime DB plans (that is, at least a DB plan) for the earlier cohort of near-retirees, we use information on whether the respondent reported receiving any pension or annuity income from an employer pension (which we assumed to be a DB plan) in any of the survey waves from 1992 through 2004.²¹ Interestingly, we find that overall about 44 percent of the earlier cohort report receiving income from a pension or annuity at some point during the survey, compared with 52 percent who reported having at least a DB plan (Table 1). It is possible that such a difference could be due to cash-out of DB balances at job separation, given the increase in the lump-sum distribution option at job separation over the past decade. Hurd and Panis (2006) using 1992–2000 HRS data found that among those who reported having a DB plan and who had a job separation between 1992 and 2000, about 11 percent cashed-out their pension balances, a finding that supports our results.²²

In married households, each spouse may have access to pension income not only through his or her own pension(s) but also through a spouse's pension(s).²³ Table 2 shows the joint distribution of pension participation by wealth quintiles and marital status.²⁴ The evidence indicates that there is a strong positive relationship between pension participation and total net worth. In 2004, about 52 percent of people aged 55–61 in the lowest total net worth quintile have had a pension over their working life, compared with 78 percent of those in the highest wealth quintile. The pattern is similar if we look at single or married people or at couples as a unit. Overall, single people (either men or women) are less likely than their married counterparts to have a pension. Married women are less likely to have a pension through their own employment than are single women (70 percent versus 83 percent, respectfully, in the middle wealth quintile in 2004). However, they are more likely to have a pension when we look at couples as a unit (93 percent of women have a pension through either their own or their husbands' employment). Across all wealth quintiles, less than a quarter of couple households have never had a pension. Similar patterns existed in 1994, by marital status within wealth quintiles.

Over the decade, lifetime pension participation through one's own employment increased. Within each wealth quintile, married women as a group experienced the largest increase in access to a pension through their own employment, compared with other marital/sex subgroups (Table 2). This is not surprising given the increasing levels of education and labor market attachment of married women of the recent cohort.

With respect to the type of pension, the pattern of shifting away from DB plans is evident across all household types and wealth quintiles. The prevalence of near-retirees with both types of plans increased dramatically over the decade particularly for couples as a unit and for single women. For example, in the highest wealth quintile, the prevalence of both plans increased for couples from 49 percent in 1994 to about 69 percent in 2004. There is no clear pattern of the prevalence of DB-only or DC-only plans by wealth quintiles.

To summarize, the recent cohort of near-retirees, particularly married women, is more likely than the earlier cohort to have a pension over their working life. Still, a wide gap in pension participation exists across wealth quintiles. Overall, about 75 percent of the recent cohort of near-retirees in the highest-three wealth quintiles report having a pension, compared with about 60 percent of those in the lowest-two wealth quintiles. The fact that many near-retirees (about 40 percent) in the lowest-two wealth quintiles have no pension from which to potentially draw income raises concern about their retirement income security; they may be more likely to rely heavily on Social Security, welfare programs, or continued work in retirement.

Wealth Distribution by Pension and Household Type

In addition to employer pensions, accumulated wealth is another source of income security in retirement. We now turn to the joint distribution of wealth holdings and lifetime pensions of near-retirees in 1994 and 2004, by household type (Table 3).²⁵ Because the wealth distribution is highly skewed, looking at the mean may be misleading; such estimates are affected by a few observations in the upper end of the distribution. Therefore, we focus on the median, which represents the midpoint of all households. Table 3 shows that for both cohorts median wealth holdings (total net worth and its components—nonhousing wealth, home equity, and assets in IRA/Keogh accounts)—vary by access to pension and pension type. In 2004, median net worth was substantially higher among those who had a pension (the highest was about $129,000 for those with both plans), compared with those without a pension (about $51,500). Median wealth for those with a DC-only plan was twice the level of wealth ($107,000) of those without a pension. Across all pension categories, the median home equity is higher than nonhousing wealth. The higher median home equity among pension holders may reflect their higher homeownership rate (about 85 percent to 89 percent depending on type of pension) relative to that of nonpension holders (about 75 percent). Among pension holders, the level of total net worth and its components did not consistently vary by marital status. Although married couples with a DC-only plan have higher levels of total net worth than those with a DB-only plan, the opposite is true among single men and women. Strikingly, among nonpension holders, single men and women have very little or no wealth at all.

The main difference between the two cohorts of near-retirees is that the gap in total net worth between those without a pension and those with both types of pension has increased, mainly because of a decrease in the wealth of nonpension holders. In addition, among married couples, the total net worth of those with a DC-only plan in 2004 is higher than that of their counterparts in 1994. The opposite is true for single men and women. Across all pension types, total net worth of single men in 2004 is substantially lower than that of their counterparts in 1994.²⁶ Furthermore, from 1994 through 2004, the median net worth of those with a DB-only pension or both pension types remained stable (while increasing for single women but decreasing for single men). Also, the median net worth increased by 15 percent for those with a DC-only pension (increasing for married couples but decreasing for single people, especially men) and decreased by about 19 percent for those without a pension. In sum, as expected, our findings indicate a positive association between total net worth and lifetime access to pensions.

We now turn to the level and composition of wealth holdings at selected points in the wealth distribution. More specifically, we rank households, separately for each cohort, by total net worth and classify them into wealth quintiles. Table 4 reports the mean of wealth holdings in each of the wealth quintiles for all households and separately for each household type (married couples, single women, and single men).²⁷ The figures indicate that the wealth distribution is markedly skewed across all household types. The pattern that emerges for both cohorts is that about one-fifth of people aged 55–61 hold little or no wealth at all, whereas about two-fifths hold a substantial amount of wealth ($179,400 or more). Furthermore, Table 4 confirms the well-known fact that the degree of wealth inequality has increased over time, with those at the top of the distribution becoming even wealthier. In 2004, for example, the mean total net worth in the highest quintile was $845,700, almost 4 times the level in the fourth quintile; over 8 times the level in the middle quintile; and about 20 times the level in the second quintile. The ratios in 1994 are about 4, 7, and 15, respectively. The quintile patterns are similar for married couples, single women, and single men. Between 1994 and 2004, mean net worth increased by 32 percent in the highest quintile and 21 percent in the fourth quintile, whereas it remained fairly stable in the middle and second quintile. For the most part, the increases over time were greatest among married couples.

Regarding components of total net worth, for the recent cohort of near-retirees in 2004, home equity comprises the largest share of total wealth (around 50 percent) in all but the highest quintile. In the later quintile, nonhousing wealth comprises more than 50 percent of total wealth, followed by home equity (about 27 percent). It is worth noting that in the lowest-three quintiles, the amount of nonhousing wealth is below $35,000 and the amount of assets in IRA/Keogh accounts is less than $12,000.²⁸ Home and IRA/Keogh ownership rates are directly related to greater wealth holdings. Only 40 percent of households in the lowest quintile actually own a home, compared with more than 90 percent of those in the other four quintiles. Furthermore, the IRA/Keogh ownership rate sharply increases from 10 percent in the lowest quintile to almost 50 percent in the middle quintile and to about 79 percent in the highest quintile. Within each wealth quintile, ownership rates of married couples are higher than those of single men and women.

Surprisingly, although assets in IRA/Keogh accounts increased across all quintiles, the ownership rate has not increased. Two factors may have contributed to such an outcome. First, as we observed above, the recent cohort of near-retirees is more likely to have had a DC plan over their working life. As a result, it is plausible that they may be more likely to have saved through such accounts with their employer, and therefore, less likely to save through IRA/Keogh accounts.²⁹ Second, because by their nature, assets in DC accounts are more portable than accrued wealth in DB plans; the observed increase in the amount of assets in IRA/Keogh accounts could be a result of an increased inflow (or rollover) of funds from DC accounts at or after job separation. However, over the past 10 years, employers with DB plans have also allowed employees to take a lump-sum distribution of their accrued DB wealth upon job separation. Different cohorts of near-retirees may have been differently affected by the types of plans they participated in and especially the availability of options for the disposition of their pension rights. Thus, for the more recent cohort of near-retirees, it is likely that the majority of funds in IRA/Keogh accounts represent employer pension wealth rather than personal saving aside from employer pensions. For the earlier cohort of near-retirees, however, the majority of funds in IRAs may constitute personal retirement saving.

Evidence suggests that the sharp growth of assets in IRAs since the mid-1990s was mainly a result of rollovers from employment-based retirement plans and asset returns and not from new contributions (Copeland 2007). Furthermore, Copeland (2006) using data from the 2001 Survey of Income and Program Participation finds that workers who participated in an employment-based pension plan had a higher probability of owning an IRA; by 2003, about 70 percent of most recent lump-sum distributions were rolled over into an IRA.³⁰

To summarize, for both cohorts of near-retirees, the evidence indicates that those without a pension have much lower levels of net worth than those who report having a pension. In addition, housing equity comprises more than half of households' total net worth for all but those households in the highest net worth quintile; whereas three-fifths of all households have on average less than $45,000 jointly in nonhousing wealth and IRA/Keogh assets. The very low level of wealth among those without a pension coupled with the very low amount of IRA/Keogh and nonhousing wealth (the most liquid assets) are indications that a considerable proportion of the recent cohort of near-retirees are not well prepared for retirement and therefore may be more likely to depend heavily on the social safety net at some point in retirement.

Conclusions

As baby boomers approach retirement, many are concerned about their economic security during retirement. Based on a comparison of the retirement economic resources of near-retirees (aged 55–61) in 2004 with those of the same age in 1994, we find that in both cohorts about 40 percent of near-retirees hold little or no wealth at all, whereas another 40 percent hold a substantial amount of wealth. Moreover, the degree of wealth inequality has increased among the more recent cohort of near-retirees compared with the earlier cohort as the wealth holdings of those at the lower end of the wealth distribution remained low, but the holdings of those in the highest wealth quintile increased substantially. In addition, housing equity, which rarely is used to finance consumption in retirement, comprises more than half of total nonpension net worth for about 60 percent of all households, leaving a much smaller amount of wealth readily accessible if the need arises. Furthermore, we find that the median total net worth among those without a pension is about half of the median total net worth of those with a pension.

We also find that the recent cohort of near-retirees has had a greater opportunity to establish pension income throughout their working life. Overall figures hide differences that exist by demographic groups and wealth quintiles, however. Thus, about 52 percent of those in the lowest wealth quintile have participated in at least some type of a pension plan over their working life, compared with 78 percent of those in the highest wealth quintile.

Even though recent near-retirees are more likely than their earlier counterparts to have had a pension during their working life, whether that will translate into higher pension wealth remains to be seen. Therefore, we cannot infer whether overall they will be better off at retirement than earlier cohorts. This is especially true because the type of pensions available to them has shifted toward DC plans and also because of the increasingly lower level of nonhousing wealth. If such increases in pension participation turn out to be associated with an increase in pension wealth that offsets the decrease in nonpension wealth, then the very low levels of nonpension wealth would be less of a concern.

Finally, looking at the joint distribution of wealth and pensions has revealed important information, with some important policy implications, that would otherwise have been obscured in aggregated samples The very low level of total net worth, for a substantial proportion of recent near-retirees, coupled with lack of pension access, raises concerns about their income security in retirement. Future research, as the recent cohort of near-retirees approaches retirement, may extend this analysis by including employer pension wealth and Social Security wealth.

Appendix