In Women Working Longer, editors Claudia Goldin and Lawrence F. Katz assemble new research that presents fresh insights on the phenomenon of working longer. Their findings suggest that education and work experience earlier in life are connected to women’s later-in-life work. Other contributors to the volume investigate additional factors that may play a role in late-life labor supply, such as marital disruption, household finances, and access to retirement benefits. A pioneering study of recent trends in older women’s labor force participation, this collection offers insights valuable to a wide array of social scientists, employers, and policy makers.
|Publisher:||University of Chicago Press|
|Series:||National Bureau of Economic Research Conference Report|
|Sold by:||Barnes & Noble|
|File size:||21 MB|
|Note:||This product may take a few minutes to download.|
About the Author
Read an Excerpt
Women Working Longer
Facts and Some Explanations
Claudia Goldin and Lawrence F. Katz
Women have been working longer for a long time in US history. Their labor market participation increased decade after decade during the twentieth century, as more women were drawn into the labor force. But that is an old story. The new story is that a large portion of women are working a lot longer into their sixties and even their seventies. Their increased participation at older ages started in the late 1980s before the turnaround in older men's labor force participation and before the economic downturns of the first decade of the twenty-first century.
Women's increased participation beyond their fifties is a change of real consequence. Rather than being an increase in marginal part-time workers, the higher labor force participation of older women disproportionately consists of those working at full-time jobs. Women are remaining on their jobs as they age rather than scaling down or leaving for positions with shorter hours and fewer days.
Why have women as a group increased their participation at older ages? Increased labor force participation of women in their older ages, we will emphasize, is part of the general increase in cohort labor force participation rates. Successive cohorts, for various reasons, increased their participation at all ages, resulting in an upward shift of participation by birth cohort. As more women graduated from college, held jobs with greater advancement potential, enjoyed their jobs more, were not currently married or were married to men who also extended employment into their senior years, more remained active in the labor force into their sixties and beyond.
Rising cohort effects in labor force participation across successive birth cohorts of US women are clearly visible in the microdata from the Current Population Survey (CPS) Annual Social and Economic Supplement (ASEC) and the Health and Retirement Survey (HRS). But these cohort effects are considerably dampened when education is considered. Higher participation at all ages has been due to greater levels of education, particularly college graduation. The increase in cohort effects in labor force participation for women in their late fifties and early sixties is also lessened by including work experience at younger ages and by adding information on the main prior occupation. We find some (negative) impact on employment at older ages from having been a teacher and discuss why that is the case.
Most important is that we find that those who "enjoyed" their jobs earlier in life remained employed for much longer later in life independent of their hours and earnings on the job six to eight years earlier. The difference between those who agree with the statement about enjoying their job versus those who disagree with the statement is 10 percentage points (on a base of 70) and the effect is twice that between those who strongly disagree with the statement and those who agree. Women who work more hours when fifty-nine to sixty-three years old are far more likely to have worked more hours six years before. But that is in addition to their greater satisfaction in the job earlier and their greater fulfillment contemporaneously. That is clearly not the case for all older workers, but it is the case for most.
Many of the cohorts we consider were those that also experienced greater divorce. Therefore, current marital status is related to employment at older ages. Because couples often coordinate their work and leisure, current employment of the spouse is an additional correlate of whether a woman is working longer.
Most of the factors just mentioned, particularly educational attainment and earlier employment continuity, were determined prior to the employment decision under question. The addition of these factors almost nullifies the cohort effects, except in one important case. For the most recent cohorts of college-graduate women we can study to their sixties (those born from 1949 to 1955), the predetermined, observable factors do not eliminate the cohort effect. Something else, yet undetermined, is keeping them in the labor force at older ages.
Labor force participation rates of women in their early sixties can be observed today for cohorts born up to the mid-1950s. Participation rates of forty- and fifty-year-old women born in the late 1950s and early 1960s have not increased relative to those of prior cohorts. Life cycle cohort labor force functions are no longer the humped functions they once were. They have become flat lines, more like those of men than they had been. These flat lines, moreover, have intersected the humped life cycle participation functions of prior cohorts, showing the decrease in participation relative to previous cohorts. But these new and flatter participation functions appear not to be decreasing at older ages relative to prior cohorts. That may indicate that women will continue to work longer even though their participation rates at middle age had stagnated relative to prior cohorts.
Several factors may operate to offset the stagnation or dip in the participation of US women in middle age. One of the reasons for the dip in women's participation in their late thirties and early forties is that women in these cohorts have had their children later. Therefore, the dip had been accompanied by an increase in their participation in their twenties relative to previous cohorts.
We find in our exploration of the correlates of participation that college-graduate women currently in their early sixties have positive cohort effects that remain substantial even after controlling for their earlier life cycle participation rates. Today's younger women will likely retire later than one would have predicted based on their educational attainment and life cycle participation rates. The finding is particularly noteworthy since female college-graduation rates are continuing to increase by birth cohort.
1.1 Labor Force Participation Rates
1.1.1 By Age, Sex, and Education Level
The central facts concerning the labor force participation of women by age are shown in figure 1.1, which uses the March CPS-ASEC microdata samples and gives contemporaneous labor force participation rates during the survey reference week for women by five-year age groups since 1962. Throughout much of the period shown, participation rates increased for women in the thirty-five- to fifty-four-year-old group. The thirty-five- to forty-nine-year-old group flattens out in the early 1990s. In contrast, rates for women fifty-five years and older were flat until the 1980s, when an almost continuous increase ensued, even for the seventy- to seventy-four-year-old group.
The labor force participation data are also given in figure 1.2 for college-graduate women, since school attainment increases by birth cohort. The series is restricted to currently married women because a large fraction of the earlier cohorts of college-graduate women-those born from the 1890s to the 1910s-never married or married late. In consequence, a large fraction of college-graduate women, even those who eventually married, never had children and had higher labor force participation rates (Goldin 1997). Considering only the group who were currently married dampens the selection problem but does not eliminate it because of later marriage ages. Participation rates for college-graduate women, therefore, decline somewhat over time as their marriage and childbearing rates become more like others in their cohort.
If one ignores cohorts born before around 1920, the series for all women and that for college-graduate women fifty-five years or older are first relatively constant and then increase, particularly after the mid- to late 1980s. The percentage point increase during the past twenty-five years, shown in table 1.1, is not much different between the aggregate group of women and the college graduates. But because college-graduate women have had considerably higher participation rates than less educated women, the shift toward college has increased participation rates for older women and the growth of women's employment at older ages.
Also clear in table 1.1 is that the increased participation of older women exceeds that of older men in the last twenty-five years, both absolutely and relative to the base levels. Among sixty- to sixty-four-year-old women, for example, participation increased by 17 percentage points on a base of 34 percent, but for males the increase is just 6 percentage points on a base of 55 percent. The percentage point increase for sixty-five- to sixty-nine-year-old males and females is similar in absolute magnitude, but the initial base for women is far lower (15 versus 26 percent).
The relative increase for older women has meant that the gender gap in participation at older ages has greatly decreased, as can be seen in figure 1.3. Differences in participation by sex have, of course, decreased more generally. But the absolute percentage point difference at some of the older ages is now smaller than for the younger age groups. For sixty- to sixty-four-year-olds, for example, the difference in participation rates between men and women was about 50 percentage points in 1962. In 2014, the difference was just 9 percentage points, when that for males and females in their thirties to mid-forties was around 16 percentage points.
Men and women are doing more of the same things throughout their lives, and this is even truer at older ages. But is that also true within couples? The answer is that, for women fifty-nine to sixty-three years old and presently married, far more of these couples are both currently working than currently retired. In addition, in 2014 about as many of these couples had a wife who was working and a husband who was not than the reverse. More women are working into their sixties and more are coupled with men who are also working. But there are also substantial numbers of women who are working into their sixties even though their husbands are retired. We return to the issue of joint employment and leisure below.
1.1.2 Full-Time versus Part-Time Employment of Women at Older Ages
The labor force participation rate for older women increased largely because of an increase in those working full time and full year. The expansion of full-time employment among participants has been especially evident for the sixty-five years and older group.
As seen in figure 1.4, the fraction of sixty-five- to sixty-nine-year-old women in the labor force who worked full time and full year increased from around 30 percent to almost 50 percent, with much of the increase occurring after 2000. The fraction of seventy-to seventy-four-year-old labor force participants working full time and full year increased from 20 percent to almost 40 percent. We emphasize that figure 1.4 gives the fraction working full time, full year among those in the labor force rather than among the population in that age group. Although the timing could indicate the impact of changes in the Social Security earnings test, the increase began before 2000 for both younger and older age groups of women.
1.1.3 Cohort Trends
Increased employment among older women would appear to be related to their increased participation earlier in their lives. The conclusion can be deduced from the fact that all cohorts in figure 1.5, panel A, that have had increased participation in their sixties, relative to earlier cohorts, also had increased participation relative to the same cohorts when they were younger. That is, the cohorts that have begun to "work longer" had higher participation rates throughout their life cycles than did previous cohorts.
Figure 1.5 begins with the cohort born in 1930, but the pattern just mentioned is evident as well for some of the earlier birth cohorts not shown. However, cohorts born in the early 1920s show no discernible increase in participation among women in their sixties despite modest increases earlier in their lives. The data for college graduates given in figure 1.5, panel B, reveal similar findings, but participation levels are higher.
As will be emphasized later, regressions of the labor force rate at older ages on birth cohort dummies indicate that cohort effects are greatly muted by the addition of various predetermined factors such as education, earlier employment continuity, and women's past occupations. That is, cohort differences in labor force participation later in life are largely, but not entirely, a function of earlier changes in human capital accumulation. These human capital advances occurred because women perceived that their investments would pay off in the labor market and that their employment would be higher and more continuous than for previous cohorts.
We noted before that the function tracing out life cycle labor force participation was transformed from being hump-shaped to being almost a flat line after the mid-1950s birth cohorts. Participation rates around age twenty-five to the early thirties greatly increased from the 1930s to the 1950s birth cohorts because women with infants had much higher labor force participation and because the birth rate decreased.
The new flatter cohort life cycle functions have begun to cross each other. The crossing creates an interesting "twist" in participation for the most recent cohorts in figure 1.5, panel A, and more so for college-graduate women in figure 1.5, panel B. The twist is the cohort analog of the oft-mentioned decrease in the participation of women in their thirties and forties. One clear way to see the change is to observe that slicing the cohort graphs at ages thirty and fifty yields the usual cohort progression. Younger cohorts have higher participation rates than older cohorts. But slicing the cohort graphs in between, say at age forty, does not yield higher rates for the most recent cohorts, such as those born from 1959 to 1973. The cohort lines appear to have twisted.
Does this mean that participation rates for these women in their fifties, sixties, and beyond will also be lower? Their increased education and labor force participation in their younger years would argue the opposite. Why they have decreased participation is still an ongoing research question, although some of the answers concern the delay of births, on the one hand, and an absence of mandated leave policy of more than twelve weeks, on the other. The decrease in participation is not large, but the disruption of the increasing trend is clear and could argue for a break in the increase of women working longer.
The bottom line for cohort change is that increased participation at older ages has occurred for cohorts that had greater attachment to the labor force throughout their lives. The upshot is that greater attachment to the labor force earlier in the work life means longer employment at older ages. We now turn to using longitudinal information from the HRS matched to Social Security earnings records to understand the role of cohort effects.
Because we rely on the CPS for the general trends and the HRS for analysis, we provide evidence that the HRS reasonably tracks general trends in the CPS for these cohorts and age groups. Appendix tables and figures show the close relationship between CPS and HRS participation rates (figure 1A.1), marital status (table 1A.2), education (table 1A.3), and number of children (figure 1A.2). Labor force participation rates in the HRS and the CPS are almost identical for women in their fifties and sixties; however, the HRS has higher participation rates than the CPS for women in their seventies.
1.2 Exploring the Role of Cohort Effects Using the HRS
Cohorts born later have higher labor force participation rates at older ages than do those born earlier. We explore whether these cohort effects are primarily due to changes in factors determined largely prior to the retirement option. These variables can include educational attainment, number and ages of children, and earlier life cycle labor force participation. We will also consider the degree to which the individual had relatively high earnings when employed, which we term the "career condition." These largely predetermined characteristics will be measured in our empirical work prior to around age fifty-five, whereas the retirement option is considered from ages fifty-nine to sixty-three.
The retirement decision may instead be determined primarily by factors that are contemporaneous, such as a set of shocks or transitory factors. These factors may have served to increase participation at older years in the post-1980s period and may include marital status change, fluctuations in the value of real estate or financial assets, pension losses, reductions in Social Security payments, and deteriorating health status.
Excerpted from "Women Working Longer"
Copyright © 2018 National Bureau of Economic Research.
Excerpted by permission of The University of Chicago Press.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
Table of Contents
Claudia Goldin and Lawrence F. Katz
I. Transitions over the Life Cycle
1. Women Working Longer: Facts and Some Explanations
Claudia Goldin and Lawrence F. Katz
2. The Return to Work and Women’s Employment Decisions
3. Understanding Why Black Women Are Not Working Longer
Joanna N. Lahey
II. Family Matters: Caregiving, Marriage, and Divorce
4. Changes in Marriage and Divorce as Drivers of Employment and Retirement of Older Women
Claudia Olivetti and Dana Rotz
5. Women Working Longer: Labor Market Implications of Providing Family Care
Sean Fahle and Kathleen McGarry
III. Financial Considerations: Resources, Pensions, and Social Security
6. Older Women’s Labor Market Attachment, Retirement Planning, and Household Debt
Annamaria Lusardi and Olivia S. Mitchell
7. Teaching, Teachers’ Pensions, and Retirement across Recent Cohorts of College-Graduate Women
Maria D. Fitzpatrick
8. The Role of Social Security Benefits in the Initial Increase of Older Women’s Employment: Evidence from the Social Security Notch
Alexander Gelber, Adam Isen, and Jae Song
9. The Hidden Resources of Women Working Longer: Evidence from Linked Survey-Administrative Data
C. Adam Bee and Joshua Mitchell
Appendix: The Health and Retirement Study (HRS)