This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation

Pension Decisions in a Changing Economy

Gender, Structure, and Choice

^a Pepper Institute on Aging and Public Policy and Department of Sociology, Florida State University, Tallahassee

Melissa A. Hardy, Director, Pepper Institute on Aging and Public Policy, Florida State University, Tallahassee, FL 32306-1121 E-mail: mhardy{at}garnet.acns.fsu.edu.

	Abstract

TOP Abstract The Importance of Pensions Gender Implications of the... Lump-Sum Distributions Methods Results Discussion References

 
Objectives. As responsibility for financial security in retirement becomes more individualized, understanding the distribution and determinants of savings behavior grows in importance. Employed men and women often gain access to their pension assets when they change jobs. In this study gender differences in pre-retirement access to and disposition of accumulated pension assets are examined.

Methods. The authors used data from the Health and Retirement Study to model pension participation, disposition of pension assets, and use of cash settlements derived from a pension plan in a previous job. Logit models provided estimates of gender differences in access to pensions and the preservation of pension funds for retirement.

Results. Women were less likely to have participated in employer-sponsored pension plans; more likely to cash out accumulated pension assets when they changed jobs; and, when job changes occurred at relatively young ages, equally likely to spend the settlement. However, by their late 40s, women were more likely to save the settlement, a net gender difference that increased with age at which the settlement was received.

Discussion. The structure of employment compensation continues to place women at a disadvantage. Gender differences in earnings and fringe benefits not only affect current financial status, but also cast a shadow over future financial security. Although the gender gap in pension coverage has been reduced, women with pensions have access to lower benefits and less in accumulated assets. As these continuing deficits are addressed, enhancing women's tendency to save pension assets for retirement can help them build financial security.

FINANCIAL security in retirement requires building claims to multiple sources of retirement income. Women's connection to this structure of compensation places them at a disadvantage. Not only is much of "women's work" performed in the household rather than in the labor force, but employed women continue to be concentrated in occupations and industries characterized by lower levels of compensation. Because access to pensions—the most common savings vehicle for retirement—is connected to the structure of employment compensation, women's retirement security has depended more on the benefit claims they made as spouses than on any additional claims they established as paid employees.

Previous research has emphasized differences in wages and earnings (Baron and Bielby 1980; Bielby and Baron 1984; England 1992; England, Chassie, and McCormack 1982; Reskin and Hartmann 1986; Reskin and Roos 1990); however, all forms of compensation can be relevant to later inequalities in retirement income. Access to employer-sponsored pensions can provide an additional source of retirement income, but pension coverage does not guarantee more income in retirement. Receipt of pension benefits depends upon favorable placement in a multi-tiered structure of eligibility and entitlement. Not all employed persons have the option of participating in employer-sponsored pension plans. Not all employees covered by a plan choose to participate. Not all employees who participate in a plan are vested in that plan; therefore, when employees change jobs, not all participants are entitled to receive (either immediately or in the future) benefits from that plan.

	The Importance of Pensions

TOP Abstract The Importance of Pensions Gender Implications of the... Lump-Sum Distributions Methods Results Discussion References

 
Employer-sponsored pensions provide an example of both collective and individualized approaches to retirement savings. Traditionally, the most common types of employer-based pensions are defined benefit (DB) plans, in which benefit levels are formulaic combinations of employee characteristics such as age, wage, and years of service. Eligibility for benefits from DB plans is regulated by federally mandated vesting rules. Designed as a labor management tool, retirement saving was a collective enterprise funded by employer contributions and available to workers at ages when the firm considered retirement desirable. Because DB plans rewarded seniority and loyalty in large firms, women's entitlement to these benefits was more often achieved through survivors' status than as the primary beneficiaries.

Although DB plans remain the most common type of pension, increases in women's labor force participation have roughly coincided with a shift to defined contribution (DC) plans, which have been rapidly increasing in both absolute and relative terms (Olsen and VanDerhei 1997). In contrast to the wage-indexed annuities promised by DB plans, DC plans resemble tax-exempt savings accounts. Employers and/or employees make contributions to an individualized pension fund, which accumulates resources over time. These funds are invested, and the benefits paid in retirement depend on the level and consistency of contributions and the soundness of the investments. Vesting is often immediate (or within 1 year), and upon leaving the firm, the worker often has the option of shifting the accumulated assets to another pension account or taking a cash settlement. In 1993, 71.5% of participants in employer-sponsored retirement plans reported the availability of a lump-sum distribution from their primary retirement plan, up from 47.8% in 1983 (Scott and Shoven 1996).

	Gender Implications of the Changing Pension Environment

TOP Abstract The Importance of Pensions Gender Implications of the... Lump-Sum Distributions Methods Results Discussion References

 
Employed women are concentrated in the secondary labor market in service jobs characterized by low wages, high turnover, few fringe benefits, and low rates of pension coverage. Women are also more likely to be employed part time, and part-time workers are generally excluded from employer-sponsored coverage. Vesting rules are particularly significant for working women; family responsibilities often have been translated into shorter job tenure and a lower probability of being vested. However, in recent decades, the disparity between rates of coverage for working men and women has narrowed. Unfortunately, the reduced gender gap is due as much to declining coverage rates for working men as to increasing coverage rates for working women (Even and Macpherson 1994; Korczyk 1992).

Given that women's careers have been characterized by both shorter tenure and more frequent job changes, DC plans may offer women increased opportunities for retirement savings. The lump-sum option can potentially increase the proportion of workers entitled to benefits and reduce pension loss resulting from job change. However, aside from tax penalties designed to encourage savings, workers who receive cash settlements from their pension plans can spend or save the money as they choose. Rather than lose pension assets through restrictive vesting provisions or entitlement rules, recipients of cash settlements may gain access to the assets but nevertheless lose pension wealth by spending the settlement on other things.

	Lump-Sum Distributions

TOP Abstract The Importance of Pensions Gender Implications of the... Lump-Sum Distributions Methods Results Discussion References

 
Most previous researchers have used data from the Current Population Survey (CPS) employee benefit supplement to analyze lump-sum disbursements (Andrews 1991; Korczyk 1996; Piacentini 1990; Scott and Shoven 1996; Yakoboski 1994). In the 1993 CPS, around 11% of workers reported ever having received a lump-sum distribution from a retirement plan on a previously held job (Korczyk 1996). But only about 19% of workers who received a lump-sum payment dedicated the entire disbursement to tax-qualified savings; 13% used the entire amount for non-tax-qualified savings or a mix of tax and non-tax savings; and 38% used some or all of the money to fund current consumption (Yakoboski 1994). Estimates from the 1996 Retirement Confidence Survey indicate that half of those receiving a pre-retirement distribution within the past 3 years spent some or all of the money (Employee Benefit Research Institute 1997). Other analyses (Scott and Shoven 1996) suggest that workers have become more likely to save cash settlements over time, a trend largely due to a decline in women's propensity to spend lump-sum distributions. Bivariate tabulations from the 1993 CPS (Korczyk 1996; Yakoboski 1994) indicate that women, younger workers, recipients of small distributions, recipients with less than a college education, and recipients with lower earnings are more likely to use lump-sum distributions for nonretirement purposes. A greater propensity to spend among younger workers, less educated workers, and recipients of smaller sums is also supported by multivariate analysis of Health and Retirement Study (HRS) data (Poterba, Venti, and Wise 1998); however, their analysis assumes no differences by gender, marital status, or characteristics of previous or next job. The fact that many of these characteristics that predict a failure to save are more typical of women than of men raises an interesting question: How do women's and men's behavior compare under similar circumstances?

	Methods

TOP Abstract The Importance of Pensions Gender Implications of the... Lump-Sum Distributions Methods Results Discussion References

 
Our data are from the first wave of the HRS collected in 1992. The HRS is a representative sample of individuals born between 1931 and 1941, aged 51–61 at the time of the interview. Compared with the CPS, the HRS information regarding lump-sum use is not as detailed; however, HRS includes more detailed information on pension histories, allowing us to include as explanatory variables: (a) individual characteristics at job exit, (b) characteristics of the job from which a lump-sum distribution was received, (c) characteristics of the new job, and (d) the reason for the job change.

Because of our interest in the pre-retirement disposition of pension assets we focused on job shifts and limited our analysis to respondents who reported working full time in a wage-and-salary position in a previous job. HRS includes information on a previous job when the respondent worked in the job for a minimum of 5 years and the employment ended after 1972. We defined previous job as the job prior to the current job for those currently employed and the job prior to that held at the time of retirement for retired respondents (to limit the confounding of disposition at job change versus disposition at retirement). Fifty-seven percent of men and 43% of women reported this type of previous employment. Our analysis sample therefore consisted of 3,324 men and women between the ages of 51 and 61 in 1992.

Descriptive statistics reported in Table 1 show that the average respondent was between the ages of 55 and 56 with more than a high school education. Two in five respondents in this group were women, and more than 8 in 10 were White. Typically, the respondent began working at this job at age 32 and continued to work at this job for more than 12 years. Weekly earnings averaged $753 (adjusted to 1991 dollars with the Consumer Price Index) at the time of departure. Two thirds of these workers were employed either in the public sector (8%) or in monopoly sector industries (60%, as proxied by census industry codes). Almost 6 in 10 workers participated in an employer-based pension plan on their previous jobs. Differences between men and women in this sample occurred primarily in job-related characteristics. Men began their previous jobs at a younger age, held them for a longer time, and averaged higher earnings. Those participating in a pension program were more highly educated, more likely to be White men, and more likely to have worked in either the monopoly or public sector with average job tenure of almost 15 years.

Individual characteristics specified in X include gender (1 = woman), race (1 = non-Hispanic White), age, years of schooling, and weekly earnings at the end of the employment period (in 1991 dollars and expressed in the metric of the natural logarithm). In addition, we included the age at which the job was entered, the duration of employment in that job (tenure), and whether the job was in the public sector or a monopoly versus a competitive sector industry. Job tenure was specified as a quadratic, because we believed that the effect of additional years of tenure on participation was stronger in the lower portion of the distribution. We tested to see whether results were robust relative to adjustments for sample selection, which they were.

Modeling Disposition of Pension Assets
In the second stage of analysis, we limited our sample to those who participated in an employer-sponsored pension plan in the previous job. Our interest was in what happened to the accumulated pension assets when the employee took a new job. We estimated a multinomial logit model that defined J-1 log-odds ratios, where J indicated the number of categories on the dependent variable, as in

The possible pension outcomes included taking a cash settlement, initiating receipt of benefits, deferring benefits and/or allowing pension assets to accumulate, and losing access to the accumulated pension wealth. We normalized the cash settlement outcome to zero (as reflected in the denominator above); therefore, coefficient estimates expressed the log of the odds of a given outcome relative to this "reference" category.

Estimates for the third set of models were based on respondents who reported taking a cash settlement when they changed jobs. Here we estimated logit models predicting whether lump-sums were saved or spent. Because we did not have plan information that indicated whether a lump-sum was offered, we could not distinguish between those who refused the lump-sum option (in favor of initiating or deferring benefits) and those whose plans did not offer this option. At each stage of analysis our interest was in those who failed to retain pension assets for retirement. Those who chose to receive or delay their retirement benefits rather than take a lump sum (assuming availability) were nevertheless using these assets for their intended purpose. Although receipt or deferral of benefits may not have been the optimal financial decision, the benefits were locked in as retirement income. We included adjustments for sample selectivity (Berk 1983; Stolzenberg and Relles 1997) and found results consistent across specifications; therefore, we report only the results from the unadjusted models.

	Results

TOP Abstract The Importance of Pensions Gender Implications of the... Lump-Sum Distributions Methods Results Discussion References

 
Participation
The first model predicted participation in an employer-sponsored pension plan on a previous job (Table 2 ). Consistent with some earlier research, working women were three quarters as likely as men to participate in these plans, a gender difference that persisted when education, tenure, earnings, and employment in the monopoly or public sectors—all of which increase the likelihood of participation—were controlled. Non-Whites were also less likely to participate, although the gender gap was slightly larger than the racial gap. The timing of the job—both in terms of individual career and in terms of calendar year—also correlated with participation. Starting a job at an older age made participation more likely, and the earlier the year in which that job start occurred, the less likely a worker was covered by an employer-sponsored pension plan. The effects of these timing measures reflect both the expansion of employer-based pension plans and the occupational mobility of workers, suggesting that jobs initiated at older ages may well have been better jobs than those occupied at younger ages. Job stability (as referenced by job tenure) also increased the likelihood of participation, although the positive effect of tenure was attenuated at the upper end of the distribution.

Explanatory variables in this model included gender, race, marital status (at the time the previous job was occupied), tenure (again as a quadratic), the natural log of weekly earnings at the time of departure (in 1991 dollars), and the age at which the respondent left the job (centered at its mean value). Because we did not have estimates of the amount of pension wealth accumulated for persons other than those who chose a cash settlement, we included these last three variables—age, tenure, and earnings—as a proxy for the accumulated assets in (particularly) DB plans, because benefits frequently are determined on the basis of these three characteristics.

We also included in the model features of the job change, such as the time interval between jobs, whether the new job offered a pension plan, and whether the exit from the previous job was voluntary or forced by a plant shutdown or dismissal. We distinguished among DB, DC, and combination pension plans (with DB plans serving as the reference category). Finally, we investigated whether the effects of marital status, age of departure, the time interval between jobs, and the voluntarism of the exit differed for men and women by including interaction terms.

In our sample of pension participants, one third of participants were women, and almost 9 in 10 were White. Slightly more than one quarter were in DC plans, only 6% were in combination plans, and the remainder were in DB plans. Although the majority of both men and women were in DB plans, more than one third of the women were in DC plans, compared with only one quarter of the men. On average, respondents changed jobs at age 46. Women were underrepresented among those who began receiving their benefits and overrepresented among those taking cash settlements. In fact, almost half the women (45%), compared with 1 in 4 men, received a cash settlement. Also, those who lost their pension assets tended to leave their jobs at younger ages after shorter durations of employment. Participants in DB plans were more likely to initiate (perhaps because lump sums were not an option) or to lose their benefits, whereas participants in DC plans were more likely to defer benefits (allowing the assets to continue to build) or to take cash settlements.

Table 3 shows that longer job tenure was linked to receipt or deferral of pension benefits rather than to cash settlements, but longer term workers were more likely to take the cash settlement than to lose the benefits. In general, then, job tenure acted as a buffer against loss. The fact that the effect leveled is likely a function of vesting provisions; once satisfied, the pension benefits of respondents are secured. Higher wage workers were also more likely to take or to defer benefits rather than to lose them or to take a cash settlement. Those whose next job also offered a pension were more likely to defer the benefits. Participating in a DC or combination plan made a cash settlement more likely. Our results show that the odds of initiating benefits were almost 40 times higher for DB participants compared with DC participants; in addition, the odds of losing the accumulated pension wealth among DB participants were almost 9 times those of DC participants. These effects highlight the impact of the incentive structure within DB plans: The best option in a DB plan is to leave the job only when benefits can be initiated, because the real value of deferred benefits erodes with time and cash settlements are not always allowed.

"Leaky" Pensions
We then examined the question of pension "leakage." If funds accumulated in pension accounts are used for nonretirement expenses, then the pension plan loses efficacy as a vehicle for retirement savings. The fact that an increasing proportion of both DB and DC plans allow cash settlements has raised concerns that too many pension dollars are being spent before retirement. In this analysis we addressed the question: What happened to the cash?

We estimated a binary logit model that described whether respondents spent or saved their cash settlements, with spending being normalized to zero. In addition to explanatory variables previously mentioned (gender, race, plan type, marital status, age of departure), we included the amount of the cash settlement. As was the case with earnings, the amount of the cash settlement was expressed in 1991 dollars and transformed to natural logarithms. We again included whether the exit was involuntary, because an unplanned exit may affect how a settlement is used. We also included interaction terms to test whether the impacts of age at receipt and the voluntarism of the exit differed for men and women. Women were almost the same proportion of savers as they were lump-sum recipients. Comparison of men and women recipients, however, indicated persistent differences in tenure (women were at their jobs 1.5 fewer years than men, on average), earnings (women earned 57% of what men earned), pension coverage (47% of women had pension coverage on the new job, compared with 54% of men), and size of lump sum (the average disbursement to women was less than half the average disbursement to men).

Table 4 shows some evidence that married workers and stronger evidence that those whose cash settlements were larger were more likely to save their lump-sum distributions than to spend them. As age of departure from the previous job increased, both men and women appeared more likely to save than to spend the cash, but the effect was twice as strong for women as for men. When we compared men and women at average age of departure, we found no difference in their propensity to save, other factors equal. In addition, men whose exits were involuntary were more likely to spend than to save the settlement, but we found no such tendency for women.

	Discussion

TOP Abstract The Importance of Pensions Gender Implications of the... Lump-Sum Distributions Methods Results Discussion References

Our analyses demonstrate persistent structural differences between men and women in their participation in employer-sponsored pension plans, in the types of plans they join, and the amount of assets they accumulate before they change jobs. The gender differences in pension participation appear to be partly compositional. Tenure, earnings, and industrial sector are all characteristics on which the distributions of men and women are quite different, with men being longer tenured, with higher earnings, and more likely to work in occupations and industries that provide pension coverage. Nevertheless, when these compositional differences are controlled, women remain less likely to be in pension plans. Success in building economic security for women in old age, therefore, continues to be hindered by limited access to this important form of occupational welfare. Among men and women who gain access, we also find gender differences in resource preservation. Married women are more likely to take cash settlements than married men or unmarried men or women. But at older ages men and women prefer to initiate benefits, a preference that increases more rapidly with age among women than among men. Further, the longer the wait to start the next job, the less likely the departing worker is to take a cash settlement and the more likely he or she is to begin benefit receipt. Again, this tendency is more pronounced among women than men. And whereas an involuntary job exit is associated with men's taking cash settlements rather than initiating benefits, it makes little difference in sorting women on those options. However, it does make a difference in whether she loses her benefits; women whose job exits are not voluntary are more likely to receive a cash settlement than to lose the assets. Similarly, among those receiving cash settlements, women are more likely to save the cash at older ages of receipt, and men whose job changes are involuntary are more likely to spend the cash, suggesting that women and men may be responding to somewhat different timetables for retirement preparation.

Although we recognize that the pension decisions of married workers may take into account the behavior of the spouse, our models address individual differences in access to and use of pension assets and include marital status as a rough proxy to determine whether gender differences are contingent on marital status. Only in the models of pension disposition do we find evidence of this contingency. Women are more likely to cash out their pension assets, and this tendency is stronger for married women than for unmarried women. Although it is tempting to conclude that this gender difference is a function of plan type, it reflects gender differences net of the type of pension plan. Given that we also controlled for age at departure (among a variety of other characteristics), we cannot conclude that women are taking the cash because they change jobs at younger ages.

In addition, although our evidence suggests that older women are more likely to save their lump sums, we cannot interpret this pattern as an aging effect. Because we relied on retrospective data collected in the first interview rather than on following workers prospectively through their careers, this age difference could be serving as a proxy for cohort differences in either behaviors or opportunities. We can say that workers who leave their previous jobs at older ages are more likely to save the lump sums than workers who leave at a younger age and that this likelihood increases more sharply among women than among men. As is the case with all cross-sectional studies, we did not directly observe process, only outcomes. Further, this gender difference may be confounded with gender differences in opportunities to take lump sums. Our second set of estimations did not control for the choice set (i.e., the options from which the worker could choose) because that information was not available for previous jobs. However, as previously noted, any worker entitled to a lump sum would also have been able to defer or initiate benefits, but a worker entitled to receive or defer benefits need not have been offered the option of a lump sum. Given our interest in identifying gender differences in pension leakage, those who chose benefits over lump sums were retaining pension assets for retirement, although their method may not have been optimal. Therefore, those who had the lump-sum option and chose otherwise made the preservation decision by not cashing out the assets; those who had the option and took the lump sum made the preservation decision in saving rather than spending the cash; and those who did not have the option of the lump sum had the decision made for them by the structure of the program.

Changes in the structure of pensions indicate that the process of accumulation—and the responsibility to ensure growth—will be increasingly located with the individual rather than with the employer. Analyses of net worth suggest that the financial security of both men and women will continue to be enhanced by partnerships that allow individuals to pool resources and share costs. Future research should address these issues within the household context, because the pension decisions of husbands and wives, similar to the retirement decisions of husbands and wives, are likely to be intertwined. Nevertheless, concerns about the permanency of contractual arrangements—whether they exist as part of employment or marriage—suggest that men and women must continue to build their own individual nest eggs. Their successes will hinge on their opportunities and how well these opportunities are managed.

	Acknowledgments

 
We thank Lawrence E. Hazelrigg, Jill Quadagno, the journal editor, and anonymous reviewers for their helpful comments on earlier drafts of this article.

Received for publication June 8, 1999. Accepted for publication April 7, 2000.

	References

TOP Abstract The Importance of Pensions Gender Implications of the... Lump-Sum Distributions Methods Results Discussion References

 

• Andrews E. S., 1991. Retirement savings and lump-sum distributions. Benefits Quarterly 2:47-58.
• Baron J. N., Bielby W. T., 1980. Bringing the firms back in: Stratification, segmentation, and the organization of work. American Sociological Review 45:737-765.
• Berk R., 1983. An introduction to sample selection bias in sociological data. American Sociological Review 48:386-398.
• Bielby W. T., Baron J. N., 1984. A woman's place is with other women: Sex segregation within organizations. Reskin B. F., , ed.Sex segregation in the workplace 27-55. National Academy Press, Washington, DC.
• Employee Benefit Research Institute1997. Lump-sum distributions and rollovers. EBRI databook on employee benefits 4th ed. 135-140. Author, Washington, DC.
• England P., 1992. Comparable worth: Theories and evidence Aldine de Gruyter, New York.
• England P., Chassie M., McCormack L., 1982. Skill demands and earnings in female and male occupations. Sociology and Social Research 66:147-168.
• Even W. E., Macpherson D. A., 1994. Gender differences in pensions. The Journal of Human Resources 29:555-587.
• Korczyk S. M., 1992. Gender and pension coverage. Turner J. A., Beller D. J., , ed.Trends in pensions 1992 119-133. US Government Printing Office, Washington, DC.
• Korczyk S. M., 1996. Pre-retirement pension distributions in a lifetime perspective American Association of Retired Persons, Washington, DC.
• Olsen K., VanDerhei J., 1997. Defined contribution plan dominance grows across sectors and employer sizes, while mega defined benefit plans remain strong: Where we are and where we are going (EBRI Issue Brief No. 190) Employee Benefit Research Institute, Washington, DC.
• Piacentini J. S., 1990. Preservation of pension benefits (EBRI Issue Brief No. 98) Employee Benefit Research Institute, Washington, DC.
• Poterba J. M., Venti S. F., Wise D. A., 1998. Lump-sum distributions from retirement savings plans: Receipt and utilization. Wise D. A., , ed.Inquiries in the economics of aging 85-105. University of Chicago Press, Chicago.
• Reskin B. F., Hartmann H., 1986. Women's work, men's work: Sex segregation on the job National Academy Press, Washington, DC.
• Reskin B. F., Roos P., 1990. Job queues, gender queues: Explaining women's inroads into male occupations Temple University Press, Philadelphia.
• Scott J. S., Shoven J. B., 1996. Lump-sum distributions: Fulfilling the portability promise or eroding retirement security? (EBRI Issue Brief No. 178) Employee Benefit Research Institute, Washington, DC.
• Stolzenberg R. M., Relles D. A., 1997. Tools for intuition about sample selection bias and its correction. American Sociological Review 62:494-507.
• Yakoboski P., 1994. Employment-based retirement income benefits: Analysis of the April 1993 Current Population Survey. (EBRI Issue Brief No. 153) Employee Benefit Research Institute, Washington, DC.

This article has been cited by other articles:

				P. Moen and K. Chermack Gender Disparities in Health: Strategic Selection, Careers, and Cycles of Control J. Gerontol. B. Psychol. Sci. Soc. Sci., October 1, 2005; 60(suppl_Special_Issue_2): S99 - S108. [Abstract] [Full Text] [PDF]

				D. A. Cotter, J. M. Hermsen, and R. Vanneman Gendered Opportunities for Work: Effects on Employment in Later Life Research on Aging, November 1, 2002; 24(6): 600 - 629. [Abstract] [PDF]

				W. H. Haas III and W. J. Serow The Baby Boom, Amenity Retirement Migration, and Retirement Communities: Will the Golden Age of Retirement Continue? Research on Aging, January 1, 2002; 24(1): 150 - 164. [Abstract] [PDF]

This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation