The labor force participation of older women: retired? working? both?

Older Women in the Labor Force
The labor force participation of older
women: retired? working? both?
Noneconomic factors—such as level of education,job flexibility in work hours,and physical stress—appear to influenceolder women’s labor force participationmore strongly than economic ones, resulting inmany “retired” women who are employed Why do older women participate in the Background and literature
labor force? There is some evidencethat on average, women’s incomes at Although a larger proportion of men than women older ages are low; therefore, they may work be- are employed at older ages, the labor force par- cause they have to work. However, more-edu- ticipation rate among older men has fallen, while cated women continue to work till older ages.
that of older women has risen. In 1975, women Thus, to the extent that education and income represented 38 percent of all older workers (ages rise together, some older women apparently work 65 and older), but by 1990, they accounted for 43 because they prefer to work. This article consid- percent.1 Census and Social Security data show ers the question of women working during the that between 1975 and 1990, the labor force par- usual retirement ages: What are the ages of older ticipation rate among 55- to 64-year-olds changed differently by gender. Men’s labor force partici- change as women age? Does the age of those pation rate fell to 68 percent from 76 percent, while who work more weeks per year differ from those that of women rose to 45 percent from 41 percent.
who work more hours per week? Do older women For women older than 65 years, the labor force typically leave the labor force and re-enter later participation rate rose slightly to 8.7 percent from or do they continue working? Do they work pri- 8.2 percent, although among men older than 65 marily because of income needs or do other rea- years, it fell to 16 percent from 22 percent.2 The data are from the Mature Women’s Co- could follow several paths. Workers could either hort of the National Longitudinal Survey of La- remain in the labor force or leave at retirement bor Market Experience, sponsored by the Bureau and reenter later. Joelle R. Weckerle and Kenneth of Labor Statistics. The survey began in 1967 S. Shultz assert that continuing work but chang- with 5,083 women ages 30 to 44, following them ing to a part-time or temporary job at later ages for the past three decades. By 1997, they had (bridge employment) is more common among men reached ages 60 to 74, well into the usual retire- than women partly because nonstandard work is ment ages. The National Longitudinal Survey a more typical situation for women at all ages.3 seeks information about personal and family For women, working part time at the usual retire- characteristics as well as the labor market experi- ment ages is an ongoing labor force participation decision rather than the continuation of work in a different pattern, more typical of men.4 The part-time employ- older ages. Amy M. Pienta and others, studying a group of ment of women workers might mean working fewer weeks per women aged 55 to 64, found that the more strongly family year or fewer hours per week. Studies indicate that older situations caused women to leave the labor force at younger women seek to work fewer hours on average than older men.5 ages, the less they participated in the labor force at older However, Barry T. Hersh and others found that women are pushed into flex-time and part-time jobs so that the lower hours It seems likely that low income provides an incentive for women work may not be their idea.6 Moreover, data from the older women to work. The poverty problem among elderly 1991 Commonwealth Fund Productive Aging Survey indicated women is extensive. Women are 70 percent more likely to that that 19 percent of women older than age 55 who work part spend their retirement in poverty than men.20 For the over-50 time would prefer full-time work.7 Whether they cannot find age group, women make up 60 percent of the lower-income full-time jobs or personal and family commitments keep them quartile.21 In 1989, the income of nearly 20 percent of women from full-time employment is unclear. There is some evidence over age 74 was below the poverty level. Nearly three-fifths that they seek flexibility, if not fewer work hours. Michael C.
of women aged 75 and older had annual incomes below Barth, William McNaught, and Philip Rizzi found that many $10,000.22 And the picture is not likely to improve. Recent older women are employed in the service industry because studies found that some older people, especially women, are more flexible hours are possible in service occupations.8 headed for trouble, having few pension or savings plans.23 Personal characteristics, family situations, and previous labor force attachment, as well as the need for income, may Data and summary statistics
affect the labor force participation of older women. Age itself The NLS Mature Women’s cohort included 5,083 women when is associated with less labor force participation, of course, it began in 1967. By the 1997 wave of the survey—when the whether through personal issues such as health status, or women had reached ages 60 to 74—2,608 of the women re- because of labor force participation factors such as age dis- sponded.24 For most of the tables in this article, the respon- crimination. Nonwhite women seem to fare worse at older dents are classified by age into three classes: younger than ages than white women. If they work, they hold lower-paying age 65, ages 65 to 69, and age 70 or older. These age group- jobs. 9 Moreover, poverty has been shown as more persistent ings are useful in that Social Security may cause work incen- among older black women.10 On the other hand, a number of tives to differ among them. Women under 65 cannot retire studies have concluded that education has a positive effect with full Social Security benefits. In addition, at the time this on the likelihood of working among women at older ages.11 survey was made, Social Security benefits were reduced by Marital status is an important factor in the labor force par- working until age 70 was reached. As table 1 shows, each age ticipation of older women. The retirement of both spouses class represented about a third of the total 1997 respondents.
often occurs within a short time, although men’s health prob- Table 2 summarizes information about their labor force sta- lems do not result in their wives leaving the labor force.12 tus. As expected, age was associated with less labor force Steven Haider and David Loughran, studying men and women, participation. More than 20 percent of all respondents were found that being married was associated with higher labor identified as employed. That percentage fell with age—37 force participation among older people.13 But others have percent of women younger than age 65 participated in the found that the majority of women who work at retirement ages labor force compared with 11 percent of respondents age 70 are unmarried.14 And the results of Franco Peracchi and Finis or older. Nearly three-fifths of all respondents were classified Welch indicate that unmarried women are less likely to leave as retired—rising to 75 percent of women at age 70, up from 37 the labor force and more likely to reenter—but for unmarried percent of women younger than age 65. However, table 3 men, the opposite is true.15 Marital dissolution often reduces reveals that 144 (or 9 percent) of those classified as retired women’s retirement income substantially so that they must were working at the time of the interview.25 often work at older ages.16 Women who have remained singlereact more like men, working a greater amount of time at earlierages and reducing work to a greater extent as they get older.17 Table 1. Age of respondents to the 1997 National
Monetary incentives differ among women by marital status as Longitudinal Survey of Mature Women
well. Although married women respond only to their wages intheir labor force participation decision, unmarried women re- Age group
spond to all financial variables.18 David A. Weaver also found that the presence of children and parents in the household does not affect older women’s working. Labor force attach- ment at younger ages may affect labor force participation at Table 2. Employment status of respondents to the 1997 National Longitudinal Survey of Mature Women
All cases
Under age 65
Aged 65 to 69
Aged 70 or older
Employment status
1 Differs from total in table 1 due to missing information.
Table 3. Time worked by employed or retired respondents in the 1997 National Longitudinal Survey of Mature Women
Mean hours
Mean weeks
Labor force status
Number of
Percent of
Number of
Percent of
respondents
respondents1
respondents
respondents2
1 Percent of those in labor force status who stated work hours greater than 3 Weeks worked since last interview. Range of weeks was 0 to 56. Weeks greater than 56 were set equal to 56.
2 Percent of those in labor force status reporting weeks greater than zero Data are for respondents who reported they were employed or retired at Full-time and part-time work of respondents to the 1997 National Longitudinal Survey of Mature Women
Full-time work
Part-time work
All hours of work
Age group
Full time = 35 or more hours per week; part time = less than 35 hours per week.
Not surprisingly, of the women who work at older ages, the others, who concluded that part-time work was not the proportion that worked full time diminished with age, while the proportion who worked part time increased. Table 4 re- Table 5 displays the industries employing the respondents veals that two-thirds (or 66 percent) of those younger than by age group as well as for all ages. About two-fifths of these age 65 worked 35 or more hours per week, and one-third (or 34 older women are employed in professional services. The per- percent) worked less than 35 hours per week. Among women centage employed in both trade and personal services rose age 65 to 69, 42 percent worked full time; 58 percent, part time.
by 10 percentage points from the youngest to the oldest age Less than 30 percent of women 70 years old or older who group, while employment in manufacturing fell to 3 percent worked were employed full time, more than 70 percent work- from 15 percent. This result supports Barth and others that ing part time. However, further analysis revealed that the older women workers desire the flexibility of employment average hours worked were very similar for all age groups.
available in service industries.27 Not surprisingly, it also sup- Full-time working women for all age groups averaged about ports the findings of Haider and Loughran that older workers 42 hours; part-time workers, 16 hours. If the women were tend not to work in more physically-demanding jobs.28 pushed into part-time jobs when they preferred full-time work, Table 6 reveals that more women of all ages were employed they might be expected to hold more than one job. However, in clerical occupations than in other occupations. Fewer of only 5 percent of the women who worked reported working at the women worked as laborers and operators as they aged, more than one job. This would seem to indicate that if they probably because of the physical demands of such occupa- worked part time, that was their preference. If so, this differs tions. In the older age groups, a larger percentage of those from the assertion of Barth and others as well as Hersh and who worked were employed in household services: 3 percent Table 5. Industries employing respondents to the 1997 National Longitudinal Survey of Mature Women
Under age 65
Age 65 to 69
Age 70 or older
Industry
Finance, insurance, and real estate .
Business, repair, entertainment and recreational 1 Includes agriculture; construction; transport, communications, public utilities; and public administration.
Occupations of respondents to the 1997 National Longitudinal Survey of Mature Women
Under age 65
Age 65 to 69
Age 70 or older
Occupation
1 Includes technical, farm workers, and crafts workers.
NOTE: Data include respondents providing occupation.
of women under age 65, but 9 percent at age 70 or older. As tion on working was greater. The percentage of those with observed in Barth and others, service occupations probably the highest educational level who worked was more than double that of the lowest educational level (22 to 49 percent The role of educational level in the labor force participa- for 65- to 69-year-olds, and 15 to 32 percent for those age 70 tion of older women is not clear. Poverty is more likely among women with a low educational level. This might mean that Whether work was measured as hours per week the re- less-educated older women need to work. However, a number spondent was working at the time of the interview, or whether of studies have found that it is more-educated people who the number of weeks she worked since the last interview was tend to work at older ages,30 especially among women.31 Table used as the measure, appears to make a difference. Using 7 provides a breakdown by education level within each age hours of work, there is a small difference by educational level.
group. The results generally agree with the other studies, Those with more education work about the same hours as although the effect differs depending on whether work is mea- less-educated respondents in the younger than 65 age group sured in hours or weeks. The table reports the percentages and for the 65 to 69 age group, there is no clear pattern. How- who were employed any weeks since the last interview, the ever, for the oldest workers, it appears that more educated average hours worked and the average weeks worked since women work somewhat fewer hours (26 if less than high the last interview. For all age groups, the percentage of those school, 17 hours if post college level).
who worked any weeks rose with educational level. For the If weeks worked since the last interview is used to mea- youngest age group (less than age 65), 38 percent of those sure work, the youngest and oldest groups differ from the with less than a high school diploma had worked, while 67 middle-age group. Less-educated women among 65- to 69- percent of those with more than a college degree had been year-olds worked more on average: 49 weeks compared employed. For the two older age groups, while the overall with 45 weeks for the most highly educated. For women percentage of women working was less, the effect of educa- under 65 years of age and women 70 years old or older, the most highly educated worked 5 to 7 more weeks than the emerges as critical. In determining which older women work, controlling for various personal, labor market, and financial The overall pattern seems to be that fewer less-educated factors provides a clearer picture. Age itself is likely to reduce women work as they get older. If they do work, they work employment, either because of a reduction of energy or be- more hours but fewer weeks by age 70. Conversely, more- cause of the custom for older people not to work.33 The num- educated women work more weeks, but they work fewer hours.
ber of household members might increase the woman’s need There appear to be more work opportunities for older, more- to work or decrease the likelihood of her working, depending educated women, and they seem to have more control over on the needs of the other household members. Marital status hours so that they can work more regularly but for shorter may have substantial effects on the labor market behavior of time periods. This is similar to Haider and Loughran’s find- older women because of the decision about how to spend ings for older men.32 The greater number of hours worked time at older ages. If the woman’s husband is retired, she may among women of the lowest educational level at the oldest want to spend her time with him. In addition, marital status ages may be due to the need to work to obtain income.
usually has financial ramifications for women. Because of thetypical higher retirement income of men, older married women Regression results
are less likely to have financial need.
The labor force participation of many of these women has Although many women are employed at older ages, more are been intermittent. In fact, the average years worked by the not. If providing an incentive to engage in market work is respondents as a percentage of their total adult years (since being considered in the development of policy changes in age 18) is 56 percent. Those who have worked most of their programs such as Social Security, then the extent to which adult life may wish to leave the labor force as soon as Social economic factors and personal characteristics or preferences Security benefits begin, as seems the case for men. On the affect the labor force participation decision of older women other hand, women with a greater labor force attachment may Education and work of respondents to the 1997 National Longitudinal Survey of Mature Women
In an educational
In an educational level
level group
group who worked
Usual hours on job
Weeks worked
since last interview
Age and education
Number 2 education
respondents worked 3
respondents worked 4
Under 65 years
All education classes .
Age 65 to 69
All education classes .
Age 70 or older
3 Those giving educational level and hours on job.
2 Those giving educational level who worked at least 1 week since last interview.
4 Set equal to 56 if weeks since last interview greater than 56.
prefer employment. Race can be considered a labor market tion differ, and if so, how. Therefore, the characteristics asso- factor. Lifelong racial discrimination resulting in less advan- ciated with different work patterns are analyzed with the re- tageous work experience for black women might create their sults reported in tables 8 through 11. Each table reports re- need for employment to bolster income. Nonwage income would directly impact the necessity to work, especially among Table 8 is a probit analysis of whether the woman worked a group such as older women where poverty is such a perva- any weeks since her last interview. Of the personal factors, sive problem. The effect of education could have both eco- age was negative and significant for those under age 65, but nomic and personal preference aspects. While low nonwage was not significant for women older than that. This may re- income, often associated with low educational levels, could flect the availability of a rising Social Security benefit amount indicate the need to work, a higher educational level might from age 62 on.34 Although a larger number of household mean more desirable jobs and working conditions, and thus members was associated with women under 65 working, this provide a desire for continuing employment at older ages.
variable was not significant at age 65 or older. Married women There are several forms a woman’s employment could take.
were less likely to work among those under 65. This result She could work more or fewer hours per week. She could work differs from Haider and Loughran, who studied an older group for a few weeks in temporary jobs. She might continue work- of both men and women and found that being married was ing as she ages, or she might leave the labor force and return associated with working more.35 Other studies have found after several years. One question that arises is whether that most women of retirement age who work are unmarried, women who engage in various forms of labor force participa- primarily through the effect of marriage on income at retire- Probit analysis of respondents to the 1997 National Longitudinal Survey of Mature Women who worked since last interview
Under age 65
Age 65 to 69
Age 70 or older
Independent variable
Coefficient
t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
1 Educational levels compared to high school graduates.
Significant at the 0.10 percent level.
2 (Number of years worked since 18)/(age-18).
** Significant at the 0.05 percent level.
** Significant at the 0.01 percent level.
Table 9. Ordinary Least Squares: usual hours worked by respondents to the 1997 National Longitudinal Survey of Mature Women
Under age 65
Age 65 to 69
Age 70 or older
Independent variable
Coefficient t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
1 Educational levels compared to high school graduates.
** Significant at the 0.05 percent level.
2 (Years worked since age 18)/(Age-18).
*** Significant at the 0.01 percent level.
Data are for hours greater than zero.
* Significant at the 0.10 percent level.
ment ages.36 However, their preference for time use could also ages. Considering a personal characteristic likely to have come into play in that married people often retire near the time labor market effects, race (white) was positive and significant when their spouses retire.37 The signs on the marriage vari- for all age groups. White women in this cohort worked more able were also negative for the older groups but are not sig- than black women.39 Fewer good employment opportunities nificant. Educational level was not strongly associated with may exist for older black women. Barth and others as well as labor force participation if work is measured as working any Mary Bowler found that if older black women worked, their weeks since the last interview. Compared with high school wages were lower than those of older white women.40 The graduates, those with lower educational levels were less likely sign on nonwage income was negative for all age groups but to work under the age of 65, but that variable was not signifi- significant only for those under 65. They worked less if the cant in older age groups. College graduates were somewhat nonwage family income was greater.41 This indicates that al- more likely to work at the oldest ages. The percentage of though poverty is a problem for older women, it is not a strong years the woman worked in her adult life was strongly associ- factor in whether they engage in market work or not.
ated with the labor force participation for all age groups, even Using hours worked at the time of the interview as the those 70 years old and older. This agrees with the findings of measure of work tells a somewhat different story. Table 9 Pienta and others.38 Apparently, women who have a lifetime shows the results of a regression of hours worked. Age was of labor force attachment continue that attachment at older associated with working fewer hours only for the oldest age Table 10. Ordinary Least Squares: usual weeks worked by respondents to the 1997 National Longitudinal Survey of Mature
Women since last interview
Under age 65
Age 65 to 69
Age 70 or older
Independent variable
Coefficient
t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
1 Educational levels compared to high school graduates.
** Significant at the 0.05 percent level.
2 Years worked since age 18)/(age-18).
*** Significant at the 0.01 percent level.
Weeks worked = 56 if greater than 56 weeks since last interview.
* Significant at the 0.10 percent level.
Table 11. Ordinary Least Squares: years worked 1987–97 by respondents to the 1997 National Longitudinal Survey of Mature Women
Under age 65
Age 65 to 69
Age 70 or older
Independent variable
Coefficient
t-statistic Coefficient
t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
Percent of adult years worked before 19862 .
1 Educational levels compared to high school graduates.
* Significant at the 0.10 percent level.
2 (Years worked before 1986)/(age 86-18).
** Significant at the 0.05 percent level.
*** Significant at the 0.01 percent level.
group. Women younger than 65 worked more hours if there Table 11 looks at the labor force participation of the re- were more household members, while women at older ages spondents over the 10 years preceding the survey. The re- did not. Marital status and educational level were not related sults are similar to Table 10 where the number of weeks worked to hours worked except that less-educated women under 65 since the last interview is measured. Not surprisingly, the worked fewer hours. The variable for the proportion of years older the women were, the fewer years out of the last 10 that worked as an adult was not statistically significant with re- they worked. Women younger than 65 tended to work if there gard to hours worked at any age. Race had no clear pattern in were more members in their households, although the number that while white women less than age 65 worked marginally of household members was not a significant variable for those more hours, the other age groups showed no significant dif- aged 65 and older. Being married was associated with work- ferences according to race. However, lower nonwage income ing fewer years only for the oldest age group. This may be among women under 65 years old was associated with work- because the women’s husbands were more likely retired over ing more hours. Although Haider and Loughran found that those years. To some extent, education was associated with older women worked fewer hours,42 the results of this table working more of the last 10 years for the age groups under 70 seem to indicate that the hours worked are probably driven years old. Those with some college worked more at ages 65 more by the requirements of the job than by the characteris- to 69, and those with less than a high school diploma worked less at ages less than 65. Once again, the previous labor Table 10 examines which women worked more than others force attachment (proportion of adult years worked before when the number of weeks worked since the last interview is 1986 in this table) was strongly associated with working from used as a work measure. With the under 65 age group, age 1987 to 1997. Considering the labor market and financial is- may be an economic issue. Only women in the under 65 age sues, white women worked more years between 1987 and group tended to work fewer weeks as their age rose. This 1997 than black women. This was true for all of the age agrees with Haider and Loughran who studied that age groups. Nonwage income was negative and significant only group.43 This may be due to the fact that although some for women under 65 years of age. There seems to be little Social Security benefits are available at age 62, full benefits evidence of these women leaving the labor force and return- are available only at age 65. Women under 65 and those 70 ing to it, a path often followed by men. If women left and and older tended to work more weeks if there were more mem- eventually returned to the labor force, the proportion of adult bers in their households, perhaps because they had more help years worked would likely show a weaker association with with tasks at home or because additional income was needed working more out of the last 10 years. This result supports for larger households. Married women under 65 worked less, the finding of Honig and Weckerle and Shultz that women and the sign was also negative for the older age groups, al- engage in bridge employment less often then men.45 Rather,
though the marital status variable was not significant at ages they continue to choose to participate in the labor force, 65 and older. Generally, working more weeks tended to rise along with the woman’s educational level, supporting thestudies linking education and work at older ages among Does “retired” mean “not working?”
women.44 For women younger than 65 years old, those withless than high school educations worked less compared with The assumption that being retired means not working is not high school graduates while college graduates tended to work necessarily correct. Moreover, the descriptive statistics of more. In the oldest group, women with college level educa- the 1997 wave of the NLS for this cohort of older women show tions or more tended to work more weeks.
that if women classified as retired work, they work nearly as The proportion of adult years worked had a strong asso- many hours as those classified as employed. However, among ciation with weeks worked, significant at the 1-percent level.
the women who work at older ages, more work part time as The more a woman was attached to the labor force during her they get older. Not surprisingly, the industries and occupa- life, the more weeks she tended to work at older ages. This tions employing the women reveal the move toward perform- was the case for all of the age groups. Of the labor force ing less physical work as they age, along with the need or conditions, white women in all age groups worked more weeks, desire for jobs that include more flexible hours. Tables 8 likely having better employment opportunities than black through 11 compare the personal and preference factors with women. The higher the nonwage income, the less women the labor market and economic factors. Results lead to the under age 65 tended to work. The sign on the variable for conclusion that even though poverty may be a real possibil- nonwage income was negative but not significant with regard ity, the personal or noneconomic aspects of the women’s lives to weeks worked by those in older age groups. These results appear generally more influential than economic factors on seem to indicate that these women worked less because of whether she works at older ages. This result differs some- financial factors than personal characteristics or preferences.
what, depending on the measure used to determine the extent of labor force participation. It is weaker when hours worked is (These may be household members who help with chores at the measure, seeming to indicate that hours are more often home rather than needing the woman’s care.) determined by the needs of the job. When weeks worked, It appears that older women who work do so because they working at all recently, or working more of the last 10 years are prefer to engage in market work. Those with more education used as measures, the strength of personal factors appears are likely to work more weeks but fewer hours. They probably hold jobs they enjoy and have more flexibility about their Economic issues do have some effect. White women work work schedules. Market work for this cohort seems to have more. If there is no systematic difference in attitudes about been a part of their lives but not an overriding part. On aver- market work between older white and older black women, ra- age, the proportion of their adult years they worked is not cial discrimination may be strong with regard to the black much over half. However, if market work has been a part of women in this cohort. Less family income (family income with- their lives, it is strongly associated with their labor force par- out the woman’s wage) is somewhat associated with her work- ing more weeks and more years out of the last 10 years. How- Compared to this cohort, more recent cohorts of women ever, the assertion of Honig that labor force participation is an have experienced a stronger lifetime labor force attachment.
ongoing personal decision among women—and that of Whether the factors associated with working at older ages Haider and Loughran that non-economic factors are more im- remain similar, or change with future cohorts of women who portant among older people in work decisions—seems borne have more continuous work histories, should be addressed in out by the results of this study.46 Even controlling for income, future research in order to clarify the effect of policy recom- married women are less likely to work, perhaps because of mendations. If these results hold for future cohorts of women, preferences for the use of time. Although David A. Weaver changing the Social Security program or other programs to found no effect from parents or children in the household, provide work incentives would probably be more successful these results show that living with more household members with older women if they focused on providing flexibility in is generally associated with greater labor force participation.47 work situations rather than financial incentives.
1 John R. Besl and Balkrishna D. Kale, “Older workers in the 21st people,” in Scott A. Bass, ed., Older and Active (New Haven, CT, and century: active and educated, a case study,” Monthly Labor Review, June London, Yale University Press, 1995), pp. 71–96; Franco Peracchi and Finis Welsh, “Labor force transition of older workers,” Journal of 2 Patrick J. Purcell, “Older workers: employment and retirement Labor Economics, April 1994, pp. 210–242; Giora Hanoch and Marjorie trends,” Monthly Labor Review, October 2000, pp. 19–30.
Honig, “Retirement, wages, and labor supply of the elderly,” Journal ofLabor Economics, April 1983, pp. 131–152.
3 Joelle R. Weckerle and Kenneth S. Shultz, “Influences on the bridge Richard W. Johnson and Melissa M. Favreault, “Retiring together or USA workers,” Journal of Occupa- tional and Organizational Psychology, September 1999, pp. 317–329.
retiring alone,” Working paper CRR WP 2001–01 (Chestnut Hill, MA,Center for Retirement Research at Boston College, March 2001); 4 Marjorie Honig, “Partial retirement among women,” Journal of Hu- Seongsu Kim and Daniel C. Feldman, “Working in retirement: the man Resources, Vol. 20, fall 1985, pp. 613–621.
antecedents of bridge employment and its consequences for the quality 5 Steven Haider and David Loughran, “Elderly labor supply: work or of life in retirement,” Academic of Management Journal, December play,” Working paper CRR WP 2001–04 (Chestnut Hill, MA, Center for Retirement Research at Boston College, 2001); Michael C. Barth, Wil- 13 Steven Haider and David Loughran, “Elderly labor supply….” liam McNaught, and Philip Rizzi, “Older Americans as workers,” in Scott A. Bass, ed., Older and Active (New Haven, John R. Besl and Balkrishna D. Kale, “Older workers in the 21st University Press, 1995), pp. 35–70.
century…”; Marjorie Honig, “Partial retirement….” 6 Barry T. Hersch, David A. MacPherson, and Melissa A. Hardy, “Occu- Franco Peracchi and Finis Welsh, “Labor force transition….” pational age structure and access for older workers,” Industrial and 16 Donald O. Parsons, Poverty Dynamics among mature women; Labor Relations Review, April 2000, pp. 401–418.
John R. Besl and Balkrishna D. Kale, “Older workers in the 21st 7 Michael C. Barth, William McNaught, and Philip Rizzi, “Older Ameri- 17 “Work and family: work patterns of women near retirement,” Report 830 (Bureau of Labor Statistics, October, 1992), pp. 1–6.
9 Ibid ; Mary Bowler, “Women’s earnings: an overview,” Monthly Labor 18 David A. Weaver, “The work and retirement decisions of older women: Review, December 1999, pp. 13–21.
A literature review.” ORS Working Paper Series No. 61. (Social Security 10 Donald O. Parsons, Poverty Dynamics among mature women: evi- Administration, Office of Economic Research, May 1994.
dence from the National Longitudinal Surveys 1967–1989, NLS Report 19 Amy M. Pienta, Jeffrey A. Burr, and Jan E. Mutchler, “Women’s 95–95 (U.S. Department of Labor, Bureau of Labor Statistics, 1995).
labor force participation in later life: the effects of early work and 11 Steven Haider and David Loughran, “Elderly labor supply…”; Francis family experience,” Journal of Gerontology, Vol. 49, no. 5, 1994, G. Caro and Scott A. Bass, “Increasing volunteering among older 20 Donald O. Parsons, Poverty Dynamics among mature women. 34 Age may also reflect health problems. When a variable for health- 21 J.S. Chatzky, “How’s the nest egg,” limiting work was included, the sign was negative but the variable was not statistically significant for any age group.
22 Michael C. Barth, William McNaught, and Philip Rizzi, “Older Ameri- Steven Haider and David Loughran, “Elderly labor supply….” 36 John R. Besl and Balkrishna D. Kale, “Older workers in the 21st 23 Jon E. Hilsinroth, “Retirees becoming wealthier, healthier,” Wall Street century…”; Marjorie Honig, “Partial retirement….” Journal, May 23, 2001, pp. A2 and A6; Patrick J. Purcell, “Older 37 Richard W. Johnson and Melissa M. Favreault, “Retiring together…”; Seongsu Kim and Daniel C. Feldman, “Working in retirement….” 24Although the women were selected originally from age range 30 to 44, 38 Amy M. Pienta, Jeffrey A. Burr, and Jan E. Mutchler, “Women’s labor by 1997 there were a few whose ages were outside the expected range 39 All except 34 women were either white or black. The 34 whose race 25 The questionnaire asked “are you retired from a job or business?” If a was ‘other’ were left out of this analysis. Removing race from the woman responded affirmatively, she was classified as retired.
regression did not substantially change the results.
26 Michael C. Barth, William McNaught, and Philip Rizzi, “Older Ameri- 40 Michael C. Barth, William McNaught, and Philip Rizzi, “Older Ameri- cans…”; Barry T. Hersch, David A. MacPherson, and Melissa A. Hardy, cans…”; Mary Bowler, “Women’s earnings….” 41 Nonwage income is total family income without the wage income of 27 Michael C. Barth, William McNaught, and Philip Rizzi, “Older Ameri- the respondent. When nonwage income was replaced with Social Secu- rity amounts, the sign was negative but it was not significant for any agegroup. Using pension income reduced the sample size substantially.
28 Steven Haider and David Loughran, “Elderly labor supply….” 42 Steven Haider and David Loughran, “Elderly labor supply….” 29 Michael C. Barth, William McNaught, and Philip Rizzi, “Older Ameri- 44 Francis G. Caro and Scott A. Bass, “Increasing volunteering…; Franco 30 Francis G. Caro and Scott A. Bass, “Increasing volunteering…; Franco Peracchi and Finis Welsh, “Labor force transition…”; Giora Hanoch Peracchi and Finis Welsh, “Labor force transition…”; Giora Hanoch and Marjorie Honig, “Retirement, wages….” and Marjorie Honig, “Retirement, wages….” 45 Marjorie Honig, “Partial retirement…”; Joelle R. Weckerle and Ken- 31 John R. Besl and Balkrishna D. Kale, “Older workers in the 21st neth S. Shultz, “Influences on the bridge employment decision….” 46 Marjorie Honig, “Partial retirement…”; Steven Haider and David 32 Steven Haider and David Loughran, “Elderly labor supply….” 33 It is likely that age discrimination would already have had its effect by 47 David A. Weaver, “The work and retirement decisions of older the time the women reached the ages in this wave of the survey.

Source: http://wall.oise.utoronto.ca/inequity/3hill_e.pdf