Retirement Plans of Mid die-aged Married Women 1

Transcription

1 Although the majority of middle-aged working women do not plan to retire at the same time as their husbands, having a retired husband does influence women to plan for earlier retirement than they would otherwise. Women's retirement plans are strongly influenced by their own pension and Social Security eligibility. Most women with pensions plan to postpone retirement until their pension eligibility begins even if their husbands will be retired before this time. Implications of these findings for future trends in the age of retirement for women are explored. Key Words: Retirement, Pensions, Middle-Aged, Women Retirement Plans of Mid die-aged Married Women 1 The aging of the population has led to concern about the burdens of supporting a large population of retired persons and to proposals that incentives be provided to induce people to retire later. The factors that influence men to retire have been the subject of considerable research, but much less is known about older women's work and retirement decisions. In the past, the majority of married women were full-time homemakers, but today, when over half of married women aged 45 to 54 are in the labor force, the subject of women's retirement deserves more attention. As the majority of families come to have two wage earners, how are retirement plans being made? Do most husbands and wives want to retire at the same time in order to enjoy leisure activities together, or are they primarily influenced by eligibility for their own individual pensions even if this causes one to retire before the other? This paper examines the retirement plans of middle-aged married women and their husbands in order to provide some preliminary answers to these questions. Factors That Influence the Retirement Decision Among men the availability and size of pensions, the amount of earnings foregone by retiring early, and health are all important influences on actual or planned age of retirement (Burkhauser, 1979; Hall & Johnson, 1980; Parnes & Nestel, 1974; Parsons, 1981; Quinn, 1977). Other influences on retirement found by some researchers include satisfaction or dissatisfaction with the current job, assets, and number of dependents (Burkhauser, 1979; Hall & Johnson, 1980; Morgan, 1980; Parnes & Nestel, 1974). Overall, the most important considera- 'This report was prepared under a contract with the Employment and Training Administration, U.S. Department of Labor, under the authority of the Comprehensive Employment and Training Act. Interpretations or viewpoints stated in this document do not necessarily represent the official position or policy of the Department of Labor. The author thanks Pan-Long Tsai for his able research assistance in preparing this paper. 2 Center for Human Resource Research. The Ohio State University,' 5701 North High Street, Worthington, OH Lois B. Shaw, PhD : tions appear to be the adequacy of pension and other income and health (Clark etal., 1978). Less is known about older women's actual or planned retirement. A recent study (O'Rand & Henretta, 1982) found that, for unmarried women as for men, both health and potential income after retirement were of major importance. For married women, poor health is also known to be an important cause of withdrawal from the labor market (Chirikos & Nestel, 1983; Shaw, 1983). It is less certain whether their own earnings or pension prospects are an important consideration for married women or whether their husbands' retirement is the deciding factor. Evidence can be found for either view, and both factors could be important. Studies of older women's labor force participation (Bowen & Finegan, 1969; Harris & Deming, 1981) suggest that married women do respond to economic incentives in their work decisions; they are more likely to be in the labor force when their own earnings potential is high or their husbands' income is low. Recent research also indicates that married women respond to pension income in much the same way as men and unmarried women: women who are currently eligible for a pension or Social Security are more likely to stop working than are women without these sources of income (Henretta & O'Rand, 1980). On the other hand, the Social Security Administration did not interview married women in its Retirement History Survey (RHS) because preliminary interviews led them to believe that "for most married women of this generation, 'retirement' has little meaning apart from their husbands' stopping work" (Irelan, 1972, p. 6). According to this view of retirement, women might be expected to stop working when their husbands retire. Subsequent research using the RHS offers only partial support for the SSA's opinion. If their husbands are still working, women are more likely to be in the labor force (Anderson et al., 1980; Clark et al., 1980). However, wives' continued employment has also been found to decrease the likelihood that husbands will retire (Hall & Johnson, 1980; Morgan, 1980). In fact, Anderson etal. (1980), using a simultaneous equation model, found that each spouse's retirement influenced the other's, support- 154 The Gerontologist

2 ing the theory that husbands and wives often plan their retirement jointly. Much of the research on women's retirement has used RHS data, and some of the results have been described as preliminary (Anderson etal., 1980; Clark etal., 1980). Thus further investigation with other data sources is needed to determine the effects of husbands' retirement and wives' own pensions on women's retirement. The present analysis focuses on middle-aged women, a group whose members will have considerably more work experience when they retire than women who are currently reaching retirement age. By comparing the reported retirement plans of wives and husbands, we can determine how many working women plan to retire at the same time as their husbands, Multivariate analysis allows an assessment of the importance of the women's own pensions as well as their husbands' plans for retirement. Although retirement plans may change before retirement actually takes place, the analysis offers insights into anticipated influences on retirement for a new generation of women workers. The Data The analysis uses data from the National Longitudinal Survey of Work Experience of Mature Women (NLS), a nationally representative sample of women who were first interviewed in 1967 and then reinterviewed periodically through In the 1979 interview, when they were years of age, all women who were either employed, seeking work, or intending to seek work in the future were asked questions about their retirement plans. Because some people may associate retirement with the receipt of a pension and many women are not covered by pensions, the question on age of retirement was phrased "At what age do you expect to stop working at a regular job?" All married women were asked a similar question about their husbands' plans. Over 60% of the respondents who were age 45 or over had a planned retirement age or stated that they planned never to retire, and the percentage with plans increased only slightly with age. On the other hand, only half of the women below age 45 had retirement plans. Since it was felt that women who are 20 or more years away from retirement are not only less likely to have thought about retirement but also more likely to give stereotypic answers to the retirement question, the present analysis is limited to respondents in the 45 and older age group. The sample consisted of over 800 married women with retirement plans. Ninety percent were employed at the time of the interview; the other 10% were either seeking employment or planning to do so in the future. About 78% were white, 20% were black, and 2% belonged to other races. Black women were oversampled by the NLS in order to have sufficient numbers to conduct separate analyses by race. Since significant racial differences in the factors affecting retirement plans were not found in the present analysis, all results are presented for the total sample weighted by the inverse of the probability that a given individual would be interviewed. This weighting caused black women to be counted as 8% of the total sample, a figure that is close to their true proportion of the population in this age range. Women who had a planned retirement age differed from those who did not in several ways: they were somewhat better educated and were more likely to be currently employed, to have a pension plan at their place of employment, and to expect to be eligible for Social Security through their own employment. Women with plans about their retirement age were also more likely to have husbands with retirement plans (or at least to have an opinion about their husbands' probable age of retirement). One Iimitation of the data should be noted although the women's plans were self-reported, those of the husbands were not. In most cases, therefore, "husbands' plans" means the wives' perception of those plans. However, about 14% of the women reported that their husbands had already stopped working; in these cases actual retirement rather than expectations was used in the analysis. How accurately the women's reports reflected their husbands' actual plans is uncertain. When the husbands were at home during the interview, their views may have been sought before the question was answered. About 12% of the women who reported a planned age of retirement for themselves said they did not know when their husbands planned to retire. Presumably most of the others based their answers on opinions they had heard their husbands express, but some may have given stereotypic answers such as age. Retirement Plans of Husbands and Wives Of the women with retirement plans, 36% planned to retire before age 62, 22% between ages 62 to 64, and 19% at age. Three percent planned to retire at an age beyond, and the remaining 20% stated that they planned never to stop working, rn general the women planned to retire at younger ages than their husbands. Whereas 36% of the women planned to retire before age 62, their report of their husbands' plans indicates that only one-quarter thought that their husbands would retire before that age. According to the women's reports, equal percentages of husbands and wives expected to retire at ages Over 35% expect their husbands to retire at age, but less than 20% of the women themselves expect to retire at this age. On the other hand, only 15% of the respondents believed that their husbands would never retire while, as noted above, 20% of the women themselves did not plan to stop working One reason the women expected to retire at younger ages than their husbands may be that because they were usually younger, they would have to retire at earlier ages to retire at the same time as their husbands. In order to investigate this possibility, Table 1 shows the relationship between the women's retirement plans and those of their husbands (as perceived by the respondents). Overall, only 30% of the wives planned to retire at the same time as their husbands, nearly half planned to retire after their husbands retired, and 23% planned to retire before their husbands. It appears that retiring at the same time as their husbands was far from being the usual expectation for this group of middle-aged women. Vol. 24, No. 2,

3 Table 1. Percentage Comparison of Husbands' and Wives' Retirement Plans by Age Difference Between Spouses Table 2. Percentage Comparison of Husbands' and Wives' Retirement Plans by Age Difference and Husbands' Health Status Timing of Retirement Timing of Retirement Age difference Wife before Same Wife after Husband Time 3 Husband Age difference Wife before Same Wife after Husband Time a Husband Wife 10 or more years younger Wife 2-9 years younger Same age a Wife 2 or more years older "Within 1 year. ' does not add to % due to rounding b The congruence of husbands' and wives' retirement plans was influenced by their relative ages. Nearly half of women who were approximately the same age as their husbands planned to retire at about the same time, whereas only 12% of women who were 10 or more years younger than their husbands planned to retire when their husbands did and the majority of wives who were older than their husbands planned to retire before their husbands. This relative age effect suggests that each spouse's plans were influenced by pensions that become available at certain ages and, perhaps, by some general idea about when it is appropriate to retire. However, some tendency for spouses to plan to retire at the same time may also be seen, especially when the differences in their ages was not too great. Congruence of retirement plans was also influenced by the husband's health. As Table 2 shows, about threequarters of men with health problems planned to retire earlier than their wives in fact, about 60% had already retired at the time of the 1979 interview. When the sample was limited to include only those men who had no health problem, the plans of the spouses became more congruent: 36% of all couples planned to retire at the same time, and over half of the couples who were the same age planned to retire at the same time. However, the pattern of women who were younger than their husbands generally planning to retire later and those who were older retiring earlier remained, again suggesting that each spouse was influenced not only by the other's plans but also by the prospect of receiving a pension or Social Security at a particular age. Multivariate Analysis of Retirement Plans In order to explore further the influences on women's retirement plans, a multivariate analysis of the various factors that may affect these plans was required. As suggested by previous research, the retirement plans of many couples may, in fact, be determined jointly, after taking into account the pension eligibility and adequacy of both spouses. Because information was not available on the age when the husband could receive a pension, it was necessary in the present analysis to assume that the husband first made his retirement plans based on when it would be economically advantageous to retire, Wife 10 or more years younger Wife 2-9 years younger Sameage a Wife 2 or more years older Wife 10 or more years younger Wife 2-9 years younger Same age a Wife 2 or more years older Husband has health problem _ a Within 1 year. b Percentage not shown when n is less than 20. c does not add to % due to rounding Husband has no health problem c 612 the state of his health, and other factors and the wife then made her own retirement plans. This analysis focuses on the importance in her planning of both her own pension prospects and her husband's plans. As previously noted, retirement is defined as not working at a regular job. Since the influences on retirement have been shown to vary over the ages when retirement may occur (Morgan, 1980; O'Rand & Henretta, 1982), separate analyses were made of influences on plans to retire at different ages. To this end, three probit equations were estimated with the three dependent variables taking a value of 1 if retirement was planned before age 62, at ages 62 to 64, and at age, respectively. In each case, planning not to retire by the specified age was given a value of 0. Probit analysis was used instead of ordinary least squares regression analysis because the latter may predict values of the dependent variable that are less than 0 or greater than 1. The equations for retiring at ages 62 to 64 and age estimated the influences on retirement plans for women who planned to continue at work until these ages; women who planned to retire before these ages were excluded. This method of analysis assumes that the possible ages for retirement were considered sequentially; that is, early retirement was considered first, and if the woman decided not to retire at that time retirement at the next age was considered. To measure the effect of pension eligibility, a set of dummy variables was created: these took values of 1 if the woman would first become eligible for a pension (other than Social Security) before age 62, at ages 62 to 64, at age, or if she did not know when or if she would be eligible. The omitted category was having no pension coverage. An additional dummy variable, which took a value of one if the woman could receive a larger pension 156 The Gerontologist

4 by retiring later, was created for the first two equations. Because very few women who first became eligible for a pension at age could receive a larger pension later, this variable was not included in the age regression. Eligibility for Social Security benefits through the woman's own work record was also included. The influence of husband's retirement plans was measured by a dummy variable that took a value of 1 if the woman believed that her husband would be retired by the time she reached the age under consideration. Another dummy variable was included for cases in which the husband's plans were not known, and the omitted category was the case in which the husband planned to retire later than the time under consideration. Husbands' health status was also included. Husbands in poor health may be forced to retire early, and their retirement income might therefore be less adequate than that of husbands whose health was good. Other explanatory variables included were the woman's current health status; her earnings potential as indicated by her education; her age, which may affect the way in which retirement was perceived; and whether she expected income from assets to be an important source of retirement income. Self-employment was controlled for because it may offer opportunities for continued work with flexible hours that can accommodate partial retirement more easily than working for someone else(quinn, 1981). A race variable, taking the value of 1 if the respondent was black, was included to capture any racial differences in the probability of retirement after other factors were taken into account. The analysis was initially performed separately for black and white women, but since few differences were found in the importance of the explanatory variables for the two groups, only the combined results are shown. Table 3 presents the regression results for the probability of retiring before age 62, at ages 62 to 64, and at age. The strong influence of women's pension eligibility for their retirement plans is evident. At each age women who would become eligible for a pension at that age were much more likely to plan to retire than were women who had no pension (the omitted category). However, women who would become eligible for a partial pension but receive a full pension later were deterred from retirement and had about the same probability of planning to retire as women without a pension. Women who would become eligible for a pension after age 62 as well as women who would become eligible for their own Social Security benefits at age 62 were less likely to plan to retire before age 62 than women who had no coverage. Eligibility for their own Social Security benefits increased the probability that women would plan to stop working after they reached age 62. The retirement plans of husbands were also extremely important. At all ages a woman was much more likely to plan to retire if she expected her husband to be retired by the time she reached the end of the given age range (i.e., by ages 61, 64, and, respectively). Evidently the desire to be retired with their husbands exerted a strong influence on wives. Women whose husbands currently had a health limitation were less likely than other women to plan to retire Vol. 24, No. 2, before age. These women may have felt financial pressure to work longer because their husbands' poor health led to premature retirement or reduced earning capacity. Increased medical expenditures might also lead to financial burdens on these families. In contrast, the women's own health status was only marginally significant for their retirement plans at ages 62 and beyond. It is probable that most women with severe health problems had already left the labor force in 1979, and the health problems of women who continued to work were generally minor ones. Self-employed women were less likely to plan to retire before age 62 or at age than other women without pensions and less likely to retire at all ages than women with pensions. Self-employed women were not asked whether they had pensions. Flexibility in working hours, enjoyment of their work, and lack of pensions among the self-employed may all contribute to their plans for working longer than wage and salary workers. Education, which indicates higher earning capacity but probably also higher pension benefits, had no effect on retirement plans before age and a positive effect on planning to retire at age. These two educational effects work in opposite directions, but at age the effect of higher potential pension benefits appear to be predominant. The effect of age on retirement plans at ages was different than its effect on both earlier and later retirement plans. Women who were older and therefore nearer to the actual time of retirement were more likely to plan to retire at ages 62-64, but younger women were more likely to plan to retire either before age 62 or at age. It is likely that older women were more aware of the actual financial situation that would face them upon retirement and were therefore less likely to plan to retire Table 3. Probit Equations for Retirement Plans of Married Women Aged Age eligible for pension Before Don't know Full pension later Own Social Security Self-employed Highest grade completed Husband retired Don't know husband's plans Own health limitation Husband has health limitation Asset income important Age Race Constant (-2.0) x log likelihood ratio N Mean of dependent variable Planned retirement age Before (6.72) a (3.66) -.171(0.70).142(0.88) (4.84) -.4(5.29) -.592(3.02).009(0.43).680(5.89).216(1.29) (0.02) -.492(3.83).291 (2.49) -.024(1.60) -.307(2.28) (3.03).849(4.22) -.213(0.76).216(0.98) (3.91).643(3.21) -.170(0.77) (0.88).951 (5.89).563(2.61).296(1.58) (2.82).408(1.69).861 (2.38).445(1.63) _b.402(1.93).596(2.35).079(2.53).905 (5.09).040(0.16).370(1.57) -.414(2.88) -.209(1.21).347(2.33).074(4.09).512(2.57) -.026(1.20).062(0.41) -.123(0.) a Asymptotic t-ratios in parentheses. b Not used in this equation due to small number of cases (3)

5 early. On the other hand, they might also be more aware that Social Security eligibility begins at age 62. Some younger women may have cited age for retirement because it is still considered to be the "standard" retirement age. Some women who were too far from retirement to have made definite plans might nevertheless have had a general idea that they would retire at about the age they thought to be most common. In order to illustrate the magnitude of some of the effects just described, Table 4 lists probabilities of planning to be retired by different ages depending on whether a woman was eligible for a pension and whether her husband would be retired. These probabilities were calculated for women who were 50 years old, had completed high school, had no health problem themselves, had husbands without health problems and did not anticipate that income from assets will be an important source of retirement income. Except for the final entry in each group the women expected to be covered by Social Security on their own accounts. The figures shown are cumulative probabilities of planning to be retired by a given age that is, they show the probability of planning to retire at the given age or earlier (see note, to Table 4 for details of the calculation method). The great majority of women who had worked long enough to be eligible for a pension planned to retire before age 62 if their husbands would be retired, and over half planned to retire even if their husbands continued to work. If their husbands planned to retire, women who had little work experience and therefore neither pension nor Social Security coverage also expected to retire before age 62. Women who could later receive a pension or Social Security were likely to plan to wait until eligibility began even if their husbands would already be retired; for women with pensions the probability of early retirement was very low if their husbands would still be working. The majority of women whose husbands planned to retire early expected to be retired themselves by ages 62-64, but women who could receive a pension by working until age were least likely to plan to be retired. Not surprisingly, a great deal of retirement was planned for ages by women who would become eligible for a pension during these years; their probability of retirement increased from 16% before age 62 to 77% by age 64. Except for those who expected to be eligible for a pension before age 62, most women whose husbands would not be retired did not plan to be retired themselves at this age. Almost all women who would have retired husbands and pensions of their own by age expected to be retired then, and over 60% of women with pensions expected to retire by age even if their husbands did not. Only among women who had no pensions and whose husbands would not be retired was the probability of planning to retire by age less than 50%. Thus, both a woman's own pension and her husband's retirement provide strong incentives for retiring. When these effects worked in the same direction, most women planned to retire. When women could obtain a pension by working longer, they generally planned to wait for their eligibility to begin even if this meant that they would retire later than their husbands. It is also interesting to note that Wife's pension Table 4. Cumulative Probabilities of Planning to Retire Before age None None and no Social Security Before age None None and no Social Security Before age 62 Age 64 or earlier Age or earlier Husband retired when wife reaches age Husband not retired when wife reaches age Note: The cumulative probabilities of planning to retire are calculated as follows: CP, = P,;CP 2 = M1-P,)(1-P 2 ); CP 3 = 1-0-P,) (1 P2) (1 P3) where P,, P 2, and P 3 = the probability of planning to retire before age 62, at ages 62-64, and at age, respectively, and CP 1( CP 2; andcp 3, = the cumulative probability of planning to retire before age 62, at age 64 or earlier, and at age or earlier, respectively. Pi, P 2, and P 3 are calculated from the probit equations in Table 3 evaluated for the average woman who was 50 years of age and a high school graduate, had no health problems, had a husband without health problems, did not expect to have substantial income from assets, and with the exception of the final row was covered by Social Security on her own account. Social Security eligibility actually had the net effect of slowing the pace of planned retirement before age. This effect is illustrated in Table 4 for women without other pensions, but the same result can be shown for those with other pensions as well. Summary and Conclusions The majority of middle-aged married women in the NLS did not plan to retire at the same time as their husbands. However, husbands' plans did have a strong impact on wives' plans: having a husband who would be retired made it much more likely that a woman would plan to retire. Women's own pension eligibility also exerted a strong influence on their retirement plans. Women's chances of retiring were increased at the point of pension eligibility and decreased for the ages before they became eligible. Eligibility for Social Security benefits had a similar, though smaller, impact. Women whose husbands suffered from health limitations were deterred from planning to retire before age, but their own health limitations had only small effects, probably because women with severe limitations were not included in the sample. Women who had no pensions were much more likely to plan to continue working until age and beyond than were women who expected to receive pensions. Thus, in the future, as women have more continuous employment records and are more likely to be employed in jobs with pension plans, they may retire even earlier than they do at present. Any continuation of the trend toward earlier retirement among men would also reinforce the trend for women, who are influenced by their husbands' retirement plans. Because of the financial burden that an increasing population of retired people will place on the actively The Gerontologist

6 working population, methods of reversing trends toward earlier retirement are increasingly sought. Recent legislation provides for future increases in the age of eligibility for full Social Security benefits, while allowing reduced benefits to continue to begin at age 62. The present analysis suggests that increasing the age for full eligibility would indeed have some effect on women's planned retirement age, both directly, through the women's own eligibility, and indirectly, through their husbands' retirement plans. However, raising the age of initial eligibility would probably be necessary to achieve a large impact since Social Security effects on retirement are greatest at the initial point of eligibility. The results also suggest that the effects of other government and private pension plans cannot be ignored. Among women in the present sample who had pensions and who answered the question on the age at which they would first become eligible, over half of those employed in the private sector and about two-thirds with government pensions believed that they would first become eligible before age 62; only 40% of these believed that they would receive larger pensions by retiring later. The prevalence of pension plans allowing for retirement before age 62 could dilute the effect of any increase in the age of eligibility for Social Security, especially if the percentage of women eligible for these pensions increases. Whether private and public employers can be induced to raise the age of pension eligibility or offer larger benefits as incentives for later retirement seems problematic, especially during the coming years when the population of younger workers will remain very large. Employers may often use retirement incentives to reduce labor costs by replacing better paid older workers with younger workers. Only when the baby boom generation reaches retirement age and the pool of younger workers becomes smaller may employers be willing to provide incentives for older workers to remain employed. Thus, it may be some time before the trend toward earlier retirement is reversed. References Anderson, K., Clark, R. L, & Johnson, T. Retirement in dual career families. In R. L. Clark (Ed.), Retirement policy in an aging society. Duke University Press, Durham, NC, Bowen, W. C, & Finegan, T. The economics of labor force participation. Princeton University Press, Princeton, NJ, Burkhauser, R. V. The pension acceptance decision of older workers, journal of Human Resources, 1979, 14, Chirikos, T. N., & Nestel, C. Economic consequences of poor health in mature women. In L. B. Shaw (Ed.), Unplanned careers: The working lives of middleaged women. D. C. Heath, Lexington, MA, Clark, R. L, Johnson, T., & McDermed, A. A. Allocation of time and resources by married couples approaching retirement. Social Security Bulletin, 1980, 43, 3-6. Clark, R. L, Kreps, J., & Spengler, J. Economics of aging: A survey. American Economic Review, 1978, 16, Hall, A., & Johnson, T. R. The determinants of planned retirement age. Industrial and Labor Relations Review, 1980, 33, Harris, R. J., & Deming, M. B. Changes in the employment status of the older population: Paper presented at the annual meeting of the Population Association of America, Washington, DC, March Henretta, J. C, & O'Rand, A. M. Labor force participation of older married women. Social Security Bulletin, 1980, 43, Irelan, L. M. Retirement history study: Introduction. Social Security Bulletin, 1972,35,3-8. Morgan, J. N. Retirement in prospect and retrospect. In C. T. Duncan & J. N. Morgan (Eds.), Five thousand American families: Patterns of economic progress (Vol. 8). Institute for Social Research, University of Michigan, Ann Arbor, O'Rand, A. M., & Henretta, J. C. Delayed career entry, industrial pension structure, and early retirement in a cohort of unmarried women, American Sociological Review, 1982, 47, Parnes, H. S., & Nestel, C. Early retirement. In H. S. Pamesetal., The preretirement years (Vol. 4). Center for Human Resource Research, The Ohio State University, Columbus, Parsons, D. O. Black-white differences in labor force participation of older males. In H. S. Parnes (Ed.), Work and retirement: A longitudinal study of men. MIT Press, Cambridge, MA, Quinn, J. F. Microeconomic determinants of early retirement: A cross-sectional view of white married men. lournal of Human Resources, 1977, 12, Quinn, J. F. The extent and correlates of partial retirement. The Cerontologist, 1981, 21, Shaw, L. B. Causes of irregular employment patterns. In L. B. Shaw (Ed.), Unplanned careers: The working lives of middle-aged women. D. C. Heath, Lexington, MA, Vol.24, No. 2,