Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2014 Sep 2.
Published in final edited form as: J Econ Perspect. 2010 Winter;24(1):139–160. doi: 10.1257/jep.24.1.139

How Longer Work Lives Ease the Crunch of Population Aging

Nicole Maestas 1, Julie Zissimopoulos 1
PMCID: PMC4151111  NIHMSID: NIHMS403613  PMID: 25190906

Population aging is not a looming crisis of the future—it is already here. Populations age when life expectancy rises and fertility declines. Economic challenges arise when the increase in people surviving to old age and the decline in the number of young people alive to support them cause the growth in society’s consumption needs to outpace growth in its productive capacity.

The ultimate impact of population aging on our standard of living in the future depends a great deal on how long people choose to work before they retire from the labor force. Here, there is reason for optimism. The end of the twentieth century witnessed a profound change in retirement behavior. For over a century, the labor force participation rate of men over age 65 declined—falling steadily from 75 percent in the late 1800s to just 16 percent in 1990 (Moen, 1994; Costa, 1998). At the end of the twentieth century, however, the labor force participation rate of older men began to rise (Quinn, 2002). The labor force participation rate of older women rose as well, following a remarkable increase in labor force participation among younger women over many decades.

A constellation of forces, some just now gaining momentum, has raised labor force participation at older ages at just the time it is needed. Age-related health declines and the reluctance of employers to hire and retain older workers present challenges, but the outlook for future gains in labor force participation at older ages is promising. The labor market is accommodating older workers to some degree, and older men and women are themselves adapting on a number of fronts, which could substantially lessen the economic impact of population aging.

The paper begins by documenting the striking shift in the population age distribution well under way, the slowdown in labor force growth, and the corresponding rise in the economic dependency ratio, which can be viewed as a measure of society’s consumption needs relative to its productive capacity. We document the historic turnaround in labor force participation and show how future increases in participation could significantly dampen the rise in the economic dependency ratio. In the second section, we turn to the most important factors behind the increase in labor force participation realized to date: the shift in the skill composition of the workforce, and technological change. In the third section, we argue that forces such as changes in the structure of employer-provided pensions and Social Security are likely to propel future increases in labor force participation at older ages. The fourth section illustrates the diversity of adaptations already at play in the labor market as older men and women seek to extend their working lives. Finally, we discuss the relatively less dramatic population aging in the United States compared to other high-income countries and how the United States is better poised than many countries to attain further gains in labor force participation at older ages.

Population Aging and the Workforce

The age distribution of the U.S. workforce has shifted dramatically in the last two decades, a shift that will continue in the next 20 years. Figure 1 presents snapshots of 1990, 2010, and 2030 based on actual and projected data from the Bureau of Labor Statistics. The height of each bar indicates the total population in a given age group, while the gray and black segments show how the age group is divided among workers (any labor force participation) and nonworkers (no labor force participation). The figure highlights three defining features of population and workforce change during this period.

Figure 1. U.S. Population by Age and Labor Force Status in 1990, 2010, and 2030.

Figure 1

Open in a new tab

Source: Authors’ tabulations of Bureau of Labor Statistics figures for civilian noninstitutional population reported in Toossi (2006) and at 〈http://www.bls.gov/emp/emplab1.htm〉.

The first defining feature is a striking shift in the shape of the population age distribution. In 1990, the distribution had a pronounced triangular shape; but by 2030, the distribution will become visibly rectangular. The primary force behind this structural change is fertility behavior, beginning with the birth of the postwar baby boom cohort between 1946 and 1964. The triangular shape in 1990 arose from the bulge created by the baby boom cohort’s presence in the younger age groups. The baby boom bulge is still evident in 2010, when the baby boomers will occupy the middle age groups, but is less evident in 2030 as the baby boomers reach the oldest age groups. The disappearing bulge reflects a force known as “population momentum.” In the absence of population momentum, the age distribution in 2030 would be the mirror image of its shape in 1990—triangular, rather than rectangular, with the tip of the triangle now on the right. Figure 2 illustrates the momentum effect by contrasting total births and birth rates over the twentieth century. Even though birth rates fell dramatically in the 1960s and have remained flat ever since, the annual number of births has risen steadily since the mid-1970s as the large baby boom cohort began having children, creating an “echo baby boom” (born 1980–1999). More recently, the year 2007 saw the highest number of births in U.S. history, surpassing even the peak of the postwar baby boom.

Figure 2. Live Births and Fertility Rates, 1930–2007.

Figure 2

Open in a new tab

Source: Centers for Disease Control and Prevention (CDC)/National Center for Health Statistics (NCHS), National Vital Statistics System.

Large younger birth cohorts will help slow population aging. The population grows older both as the baby boom bulge advances through the age structure and as life expectancy continues to rise. During the twentieth century, life expectancy in developed countries rose linearly, by 0.2 years per calendar year among those 50 and older, and shows no signs of decelerating (Christensen, Doblhammer, Rau, and Vaupel, 2009). Although fertility and mortality are the primary forces changing the age distribution, immigration will also play a role. Because immigration inflows mostly affect the bottom half of the age distribution, they accelerate the movement toward a rectangular shape, helping to slow population aging.

The second defining feature implied by the patterns in Figure 1 is a sharp slowdown in labor force growth. The labor force is projected to have grown by 24.4 percent between 1990 and 2010; the effect is visually apparent in Figure 1 as an expansion of the area occupied by the gray bars between the top and middle panels. However, labor force growth between 2010 and 2030 is projected to be just 10.5 percent—less than half the growth expected between 1990 and 2010. Labor force growth between 1990 and 2010 is in line with expected population growth of 26.0 percent during the same period, but labor force growth between 2010 and 2020 will be well below population growth of 17.5 percent during that period. This relative slowdown in labor force growth compared to population growth will lead to a corresponding slowdown in growth of GDP per capita absent an offsetting influence such as an increase in labor productivity, labor force participation, or immigration.

Another way of seeing this pattern is to consider the trend in the “economic dependency ratio,” here defined as the ratio of nonworkers 16 and older to workers 16 and older, multiplied times 100. The solid line in Figure 3 shows the economic dependency ratio implied by the Bureau of Labor Statistics’s estimates of labor force growth between 1990 and 2030. The economic dependency ratio will rise modestly from 50 adult nonworkers per 100 workers in 1990 to 52 nonworkers per 100 workers in 2010, then shoot up to 62 nonworkers per 100 workers by 2030 as the baby boomers retire from the labor force.

Figure 3. Economic Dependency Ratio under Two Future Labor Force Participation (LFP) Scenarios.

Figure 3

Open in a new tab

Source: Authors’ tabulations of Bureau of Labor Statistics (BLS) data reported in Toossi (2002, 2006) and at 〈http://www.bls.gov/emp/emplab1.htm〉.

Notes: The economic dependency ratio is the ratio of nonworkers 16+ to workers 16+ multiplied by 100. BLS labor force projections assume flattening growth in labor force participation (LFP) rates in most age groups after 2010 for men and 2020 for women. The “No flattening” scenario assumes labor force participation rates in all age categories would continue to rise at the same growth rate as between 2000 and 2010.

Tempering this drag on labor force growth is the third defining feature: a notable rise in labor force participation at older ages. Throughout most of the twentieth century, the labor force participation rate of men, and particularly older men, declined (Moen, 1994; Costa, 1998). Figure 4 shows historical and projected labor force participation rates by age group and sex from 1950–2030. For men ages 55–64, the rate of decline slowed in the 1980s and then flattened. For men 65 and older, the rate of decline also slowed in the 1980s and then, remarkably, the trend reversed in the 1990s. The Bureau of Labor Statistics assumes a flattening of the trend by 2010 for men ages 65–74 and by 2020 for men 75 and older.

Figure 4. Labor Force Participation Rates by Sex and Age, 1950–2000, and Projected 2010–2030.

Figure 4

Open in a new tab

Source: Bureau of Labor Statistics (BLS) data reported in Toossi (2002, 2006) and at 〈http://www.bls.gov/emp/emplab1.htm〉. Dashed lines denote BLS projections.

Just as remarkable is the rise in labor force participation among women. Between 1950 and 2000, the labor force participation rate of women in the prime age groups (25–34, 34–44, and 45–54) approximately doubled from 35 to 70 percent, as shown in the bottom panel of Figure 4. Labor force participation among women 55–64 doubled as well, rising from roughly 25 to 50 percent. Labor force participation of women 65 and older remained low during the period, but began rising between 1990 and 2000, and is projected to rise further through 2020 and then flatten.

Population predictions for 2030 are reasonably firm; after all, most people who will be alive then have already been born. But estimates of future labor force participation in Figure 4 are less certain and are made by advancing the population through the age structure and assuming past trends in participation will level out. If the Bureau of Labor Statistics assumed labor force participation rates would continue to rise at the same rate after 2020 as before—and we believe there is good reason to think they might continue to rise—our calculations show that the economic dependency ratio in 2030 would be significantly lower, rising slightly to 53 instead of 62 nonworkers per 100 workers. The dotted line in Figure 3 illustrates this effect. In short, the economic impact of population aging on our future standard of living depends a great deal on the evolution of labor force participation among older workers.

The next section discusses explanations for the turnaround in the labor force participation rates of older workers, while the following section discusses how these factors and others will shape future labor force participation rates of older workers.

Explaining the Trend Reversal

The cause of the reversal in labor force participation rates for older men has been the subject of intense interest. Changes in the level of labor force participation over time can result from changes in either entry or exit behavior. For example, an increase in labor force entry at younger ages could lead to a higher fraction of the population employed at older ages, as could a decrease in the rate of labor force exit at older ages. Recent evidence favors the latter explanation. Labor force participation rates for younger men have actually fallen over time (as shown in Figure 4), and retirement rates are indeed lower among younger cohorts (for example, Hurd and Rohwedder, 2008). In contrast to men, the rise in female labor force participation at older ages is a consequence of both increased labor force entry at younger ages and decreased retirement rates at older ages.

Why have men’s retirement rates declined? The principal supply-side explanation for the trend reversal centers on the change in the skill composition of the workforce. According to Goldin and Katz (2007), years of completed schooling rose on average by two years for every 25 successive birth-year cohorts born between 1876 and 1950 (and subsequently rose by only 0.68 years for the 25 cohorts born between 1950 and 1975). As a result, completed schooling (by age 30) among those in the 65–74 age group will rise from 10.4 years in 1990 to 13.3 years in 2030. Educated people work more at least in part because they are paid more, have more fulfilling jobs, and face fewer physical demands. Blau and Goodstein (forthcoming) illustrate how, absent this compositional shift to a more educated workforce, the labor force participation rate of older men (ages 55–69) might have continued to decline; in fact, they argue, changes in the educational composition of the work-force can account for the entire trend reversal.

A second explanation for the trend reversal centers on the historic rise of the dual-earner family. Indeed, the percentage of married couples ages 51–56 with both spouses working rose from 53 percent in 1992 to 60 percent in 2004.1 Schirle (2008) attributes 25 percent of the rise in labor force participation among older men to the rising labor force participation of their wives. A likely mechanism is complementarity in the leisure time of spouses, one reason for the empirical tendency of couples to retire around the same time (for example, Gustman and Steinmeier, 2000; Maestas, 2001; Coile, 2004). The flip side of leisure complementarity is that retirement leisure is valued less if one’s spouse is still working. Schirle’s result, which is corroborated by Blau and Goodstein, suggests that the education effect may, at least in part, operate through the spousal labor supply channel; after all, higher-educated men tend to marry higher-educated women (Mare, 1991).

Perhaps surprisingly, reforms to the Social Security program have emerged as secondary forces. Three reforms in particular altered the return to work at older ages; a gradual increase in the age at which full benefits are payable (the full retirement age), from age 65 to 67 for people born after 1937; elimination of a penalty associated with claiming benefits after the full retirement age; and repeal of an implicit tax on earnings after the full retirement age (the so-called earnings test). Gustman and Steinmeier (2009) attribute only one-sixth of the increase in labor force participation between 1998 and 2004 among men ages 65–67 to the Social Security reforms, and Blau and Goodstein (forthcoming) reach a similar conclusion. In fact, Social Security policy changes may have even reduced labor force participation for some individuals, as they appear to have resulted in a small increase in both applications and enrollment in the Social Security Disability Insurance program (Duggan, Singleton, and Song, 2005; Li and Maestas, 2008). Overall, these policy changes do not appear to have been powerful enough by themselves to have reversed the long-term downward trend in labor force participation, in part because some of the changes were not effective until 2000, well after the leveling of the downward trend in labor force participation among older men began.2

Finally, mortality decline is a supply-side explanation for the trend reversal that has not received much empirical attention. Life expectancy in developed countries has increased steadily over the past century and a half, but only since the late 1970s have the gains been fueled by increases in life expectancy at older ages. This change was both unprecedented and unexpected (Christensen, Doblhammer, Rau, and Vaupel, 2009). Furthermore, as we live longer, we also live better (Cutler, 2001). Most evidence points to deferred onset of disability despite an increase in chronic conditions, with the latter in part due to earlier diagnosis (Christensen, Doblhammer, Rau, and Vaupel, 2009). As a result, most retirees in their 60s and 70s are now physically able to work (Weir, 2007). Still, health does decline on average with age, bringing challenges unique to later-life careers.

Demand-side explanations for the trend reversal among men have received comparatively little study, mostly due to a scarcity of nationally representative employer data. One potentially important explanation is the nature of work itself. Technological advancement since 1960 has favored cognitive and analytic skills over manual and routine skills (Katz and Murphy, 1992; Autor, Levy, and Murnane, 2003), although the shift in job content appears to have favored women more than men (Black and Spitz-Oener, forthcoming). Nonetheless, as technological change has made jobs less physically demanding and the Americans with Disabilities Act has required employers to provide reasonable accommodations for disabilities, it is now more possible than ever to work in the presence of health limitations. Moreover, continued growth in service-related industries implies a growing availability of less physically demanding jobs, which are thus more suitable for older workers.

A second demand-side explanation is the shift in the structure of employer pensions from traditional defined benefit plans to defined contribution plans, such as 401(k) plans. Defined benefit pensions are usually based on age, final salary, and job tenure, so that once an employee reaches the plan’s normal retirement age (usually somewhere between 55 and 65), the employee is eligible for full benefits, which vary with job tenure and final salary. Defined benefit plans usually have early retirement ages (that may also depend on years of service) at which point reduced benefits become available. Benefits often increase a lot in present value in the years right before eligibility for early retirement benefits but less so afterward, in many cases becoming actuarially unfair after the normal retirement age (Gustman, Steinmeier, and Tabatai, forthcoming).3 This accrual pattern creates strong incentives for continued work before eligibility for early retirement, but weaker incentives and even disincentives for work beyond that point. On the other hand, defined contribution pension values are not directly tied to age and tenure, can increase at rates approximating actuarial fairness depending on market returns and availability of an employer match, and may be claimed penalty-free as early as age 55 if the individual leaves the employer or at 59½ if the employee continues to work.

But the transition in pension arrangements is not likely a primary explanation for the reversal in labor force participation trends for older men in the 1990s. Defined benefit plan coverage was never close to universal. Even at their height in the late 1980s, only 50 percent of the workforce was covered by such a plan, and this figure fell only modestly to about 45 percent by the end of the 1990s (Blau and Goodstein, forthcoming). Although many employers stopped offering defined benefit plans during this period, data from the Pension Benefit Guarantee Corporation indicate that plan closures were concentrated among small employers (Perun and Valenti, 2008). Furthermore, plan closures have affected younger workers more than older workers—a nontrivial share of defined benefit plans were not terminated altogether, just closed to new entrants. Even among terminated plans, about 95 percent have been terminations of healthy plans, in which the full value of accrued benefits is set aside to be administered by an insurer as an annuity contract or paid out in lump sum (Perun and Valenti, 2008).

Coincident with the shift in pension arrangements has been a steady reduction in employer-provided retiree health insurance benefits, both in terms of coverage and generosity (Kaiser Family Foundation and Hewitt Associates, 2006). Retiree health benefits are a form of subsidized health insurance for those who retire prior to Medicare eligibility, and unlike pensions, reductions in benefit generosity have affected current as well as future retirees.

Finally, the passage of the 1968 Age Discrimination in Employment Act (ADEA) imposed new constraints on employers in hiring, firing, laying-off, and compensating older workers. Although the law was passed well before the turnaround in male labor force participation, researchers have generally agreed that the federal law did not have much effect until much later—in 1978, an amendment extended the law’s protected age from 40–65 to 40–70; in 1979, enforcement became the responsibility of the Equal Employment Opportunity Commission; and in 1986, an amendment eliminated mandatory retirement (Neumark, 2001). While empirical evidence suggests the law may have, perversely, worsened employment outcomes for older workers (Lahey, 2008), the elimination of mandatory retirement could have indirectly affected labor force participation at older ages by changing social perceptions of the “normal” retirement age.

The United States is not unique in the changing labor force participation patterns of older men and women. Figure 5 shows that in OECD countries, average labor force participation among older women (50–64) has risen dramatically since the 1980s, and downward trends in labor force participation of men reversed between 1994 and 2004 (more specifically, downward trends reversed across two-thirds of OECD countries) (OECD, 2006). Despite tremendous gains in labor force participation among women in most OECD countries, including the United States, women continue to participate in the labor force at lower rates than men at all ages, they retire at younger ages than men, and conditional upon working, they work fewer hours. The similarity of labor force participation patterns in the United States and other OECD countries across time and space points to common forces, like the growth in educational attainment, technological change, and the labor force participation of married women, as the primary drivers of the reversal in male retirement behavior; and to reforms to Social Security, private pensions, and employment protection law, which have tended to be unique to the U.S. context, as secondary forces. Although many OECD countries have also sought to reform their social insurance systems, often by eliminating implicit subsidies for early retirement, most changes have been only recently implemented or will be implemented in the near future. Similarly, owing to a 2006 European Union directive, most European countries have only very recently passed age discrimination legislation, and surprisingly, these new laws do not prohibit mandatory retirement (OECD, 2006).

Figure 5. Trends in Labor Force Participation at Ages 55–64 in the United States and OECD Countries.

Figure 5

Open in a new tab

Source: OECD Employment and Labour Market Statistics, LFS by Sex and Age 〈http://titania.sourceoecd.org/vl=4696454/cl=14/nw=1/rpsv/~3962/v123n1/s8/p1〉.

Note: The OECD average participation rate is the population-weighted average of the participation rates in Canada, France, Germany, Italy, Japan, the Netherlands, Norway, Spain, and the United States.

Future Trends in Labor Force Participation at Older Ages

The prospect for future increases in labor force participation at older ages is quite good. Although the growth in educational attainment that fueled the turnaround for older men slowed in the 1970s (Goldin and Katz, 2007), other forces are gaining momentum and could propel future increases.

Will Workers Want to Work Longer? The Supply of Older Workers

Changes to Social Security were not a major force behind the turnaround in participation by older men in the 1990s. One reason for this may be that the most important policy change, the scheduled increase in the full retirement age, has not yet been fully realized. The phased increase in the Social Security full retirement age will not be fully implemented until the 1960 birth cohort turns 62 in 2022, and given the system’s solvency issues, more reforms seem likely in the future. Early evidence on the policy’s effect thus far suggests it may be sizable (Mastrobuoni, 2009).

In addition, women increasingly face direct incentives for continued work at later ages because additional work may raise their Social Security benefits. A woman’s Social Security benefit is the higher amount of the benefit she is entitled to based on her own work history or 50 percent of the benefit her husband would receive based on his work history.4 As the labor force participation rate of younger women has risen over time, more and more women qualify for benefits on the basis of their own work history rather than their husband’s. In particular, for an older woman with early-career gaps in earnings due to childbearing, the marginal increase in earnings and pension wealth coming from another year of work could be larger than that of a same-aged man with a continuous earnings history. As noted earlier, when women work longer, so might their husbands.

In the future, workers may not only want to work longer, but they may need to work longer in order to support consumption over a longer lifespan. The linear increase in life expectancy over the past 150 years or so has not yet shown any deceleration of progress or imminent limit to human lifespan. Continued progress toward better health during longer life will depend in part on the adoption of healthy behaviors such as less smoking, more exercise, and better diets, but evidence is mounting that at least on some dimensions of health, age 60 is the new 50 (for example Martin, Schoeni, Freedman, and Andreski, 2007).

One potential constraint on labor supply in the future is that with increasing life expectancy, older Americans will feel the pull of family care-giving responsibilities from many directions—their elderly parents, their spouse, and even their grandchildren. In deciding whether to work, older workers face tradeoffs that interestingly mimic those often faced by younger married women; the opportunity cost of time is relatively high and technology may not provide a close substitute.

Will Employers Want Older Workers? The Demand for Older Workers

One of the most important forces gaining momentum is the fundamental restructuring of private pensions over the last two decades. The full effect of the movement toward defined contribution plans has yet to fully manifest—even among the oldest baby boomers, about 60 percent of their pension wealth resides in defined benefit pension plans (Gustman, Steinmeier, and Tabatatbai, this issue). Friedberg and Webb (2005) estimate that workers with defined contribution plans retire almost two years later than those with defined benefit plans, all else equal, and Hurd and Rohwedder (2008) show how the change in the structure of pensions accounts for a significant portion of changes in future anticipated employment at ages 62 and 65 among members of the baby boom cohort born between 1950 and 1953.

In addition, legal barriers to phased retirement have only recently been partially relaxed. The Pension Protection Act of 2006, while stopping short of explicit guidelines, effectively permits pension payments to employees while working beginning at age 62 even if they have not reached their plan’s normal retirement age. The extent to which employers and thus employees will choose to make use of this option and its effect on total hours remains to be seen.5 On one side, greater use of phased retirement may keep older workers at least partially in the workforce; on the other side, phased retirement may lead some workers who would have been full-time to choose partial retirement sooner (Gustman and Steinmeier, 2008). Nonetheless, employers have a new tool with which to manage the pace of transitions into retirement.

Over time, older workers are becoming closer skill substitutes for younger workers, which in turn should raise demand for older workers, particularly in jobs where the productivity return to experience is high. For example, data from Goldin and Katz (2007) show that in 1990 a retiring 65 year-old had on average 10.9 years of schooling, while an entering 25 year-old in that year had 13.5 years of schooling, a gap of 2.6 years. In 2010, a retiring 65 year-old will have on average 12.6 years of schooling, while an entering 25 year-old will have about 13.9 years, dramatically narrowing the education gap to just 1.3 years. By 2030, a retiring 65 year-old will have 13.5 years of schooling, while, if current trends continue, an entering 25 year-old might have around 14.0 years, further closing the education gap to just half a year. The slowdown in growth of educational attainment implies that for the first time in history, new labor force entrants are not substantially more educated than those retiring from the labor force.

Might employers believe that older workers are less productive than younger workers? Conventional wisdom suggests yes, but direct evidence on this point is scant. Research in cognitive neuroscience finds that many aspects of cognition, including processing speed, long-term memory, and reasoning abilities decline from about age 20, and by some estimates the decline accelerates after age 60, although considerable heterogeneity exists across people (Hedden and Gabrieli, 2004). However, other valuable skills such as emotional processing, semantic knowledge, autobiographical memory, and short-term memory remain relatively stable as one ages—and the extent to which individuals can use accumulated knowledge and experience to compensate for any limitations is not well understood. Productivity need not decline steeply with age if compensatory behavior is possible or if skills that do not decline with age are primarily used.

Evidence continues to mount that statistical discrimination, judgment based on a group’s characteristics rather than the individual’s, is an important factor in the hiring and retention of older workers (Johnson and Neumark, 1997; Neumark, 2001; Lahey, 2008). Based on an experimental labor market study, Lahey (2008) finds that a younger woman is 40 percent more likely to be offered an interview for an entry-level job than an older woman. Notwithstanding innovations that make it easier to test a worker’s productivity or ability to learn, cognitive declines with age at the population level make it likely that this form of discrimination will continue to present a barrier to work at older ages.

Perhaps the biggest constraint on employer demand for older workers in the future will be health care costs. As the workforce ages, its average health declines and health care costs rise. The rising prevalence of disease, chronic conditions, and obesity among people aged 65–69 suggests that an aging workforce will surely increase the cost of providing health care to older workers in the near future (Christensen, Doblhammer, Rau, and Vaupel, 2009; Lakdawalla, Goldman, and Shang, 2005). Thus far, medical innovation has been a double-edged sword, both lengthening life and increasing costs. The extent to which expected healthcare costs depress employer demand for older workers is unknown but may be substantial. Age discrimination laws prevent employers from varying health insurance coverage by employee age. Even when the employee is Medicare-eligible, the employer is the primary payer and Medicare is the secondary payer, paying only what the employer plan does not cover (up to coverage limits). An alternative would be to make Medicare the primary payer under all circumstances, decreasing the cost of employing an older worker relative to a younger worker but increasing Medicare’s expenditures.

A Diversity of Adaptations at Work

As labor force participation at older ages has risen, the ways in which men and women work at older ages has been changing, too. The traditional view of retirement as a permanent and complete exit from the labor force and from a full-time job held since middle age does not describe most workers today.

For many workers, an exit from the labor force after age 50 is temporary. Between the ages of 51 and 61, only 40 percent of those who stop working remain out of the labor force; 40 percent return to work; and 20 percent pass through a period of partial retirement (part-time work following retirement) during the next six years or more (Maestas, forthcoming). Sixty-one percent of re-entering workers change (three-digit) occupations, with a notable shift from managerial and professional specialty occupations to sales, administrative support, and services positions. A parallel shift in industry is evident, as individuals move from the manufacturing sector to the services sector (Maestas, forthcoming).

Retired workers from earlier birth cohorts were less likely to return to work. Only 25 percent of retirees from the 1905 to 1911 birth cohort re-entered the labor force (Ruhm, 1990). But the fraction of older male workers reporting partial retirement is only slightly higher among recent retirees than among earlier birth cohorts (Gustman and Steinmeier, 2000; Maestas, 2007). In fact, hours worked per week for workers age 50 and over have changed very little over time. Data from the Current Population Survey reveal only a small increase in part-time work (1 to 34 hours per week) from 24 percent in 1969 to 27 percent in 2001 (Karoly and Zissimopoulos, 2004) followed by a small decline over the last decade (authors’ calculations from Purcell, 2008). Nonetheless, older workers apparently want part-time work. Zissimopoulos and Karoly (2007a) find that 58 percent of older workers looking for a job state they are looking for part-time work and that approximately half of older workers state a desire to reduce hours as they age.

So why don’t more older workers work part-time? Several potential explanations have been offered for the stickiness of work hours. As mentioned earlier, until recent legislative changes, employees were restricted by law from collecting a pension while working if they had not yet reached the plan’s normal retirement age. Thus, an employee could not offset a reduction in work hours with pension income. Moreover, leaving one’s main, full-time job for part-time work more often than not results in a lower hourly wage and loss of health insurance and other benefits. Even though part-time workers need not be paid expensive health and pension benefits, employer demand for part-time older workers may be low in part because high turnover rates among workers nearing retirement make it difficult to spread fixed costs associated with hiring and training across a long horizon. While shorter payback periods are certainly an issue for employers, the concern applies to both full-time and part-time workers and increasingly applies to younger workers as well as older workers, as the gap in expected tenure between older and younger workers appears to be falling (Friedberg and Owyang, 2004).

For workers desiring a reduction in work hours, self-employment may be an alternative to labor force withdrawal and offer an opportunity to reduce hours. In fact, self-employment increases rapidly with age—nearly one in four 65-year-old workers is self-employed—and part-time work is common. About 41 percent of the self-employed over age 50 worked part-time in 2001, and this is up from 33 percent in 1969 (Karoly and Zissimopoulos, 2004). By our calculations, older employees who become self-employed reduce hours over a ten-year period from an average of 41 weekly hours to 34 weekly hours.

Is self-employment for older workers a form of retirement leisure? Its diversity suggests otherwise. Just over one-third of the self-employed are women and 44 percent have employees. Similar to wage and salaried employees, their occupations are varied. According to data on workers over age 50 from the 2001 March Current Population Survey (Karoly and Zissimopoulos, 2004), just over 35 percent of both self-employed workers and wage and salaried employees report their occupation as executive, managerial, or professional. Another 23 percent of self-employed workers are in sales, while administrative support is the next most common occupation for wage and salaried employees (16 percent). Self-employed workers, like wage and salaried employees, work in all industries (with the exception of public administration) and increasingly, more in the service industry. The most common industry for both classes of work (self-employed, wage and salary) is professional services followed by trade for self-employed older workers and manufacturing for wage and salaried workers.

As shown in Figure 6, self-employment rates start off very low among the youngest workers, increase steadily through age 64, and then begin to rise more sharply at ages 65 and above. The rise in self-employment rates at the end of the labor market career is driven both by lower rates of retirement from self-employment than from wage and salary work as well as new movements into self-employment; nearly one third of self-employed older workers became so at or after age 50 (Zissimopoulos and Karoly, 2007a). A greater number of older workers may be self-employed in the future because the baby boom cohort will reach retirement with considerably more wealth and education than prior birth cohorts.

Figure 6. Percent of Workers Who Are Self-Employed by Age Group.

Figure 6

Open in a new tab

Source: Tabulations of Current Population Survey 2002 presented in Zissimopoulos and Karoly (2007b).

The dynamics of labor force behavior at older ages are illustrated in Table 1. These data show the percent of men and women ages 51 to 67 in the Health and Retirement Study (HRS) who transition between self-employment, wage and salary work, and nonemployment during a two-year period. About one-quarter of men and 20 percent of women change labor force status over the course of two years.6 Many of these changes in status are between wage and salary work and self-employment. About 6 percent of self-employed men and women are working as employees two years later, and 2 percent of male employees and 1 percent of female employees become self-employed. Employees ages 51–55 are approximately twice as likely as those ages 62–67 to become self-employed. Empirical studies reveal both push and pull factors at work in movements between these classes of work. Wage and salary workers with work-limiting health conditions are pushed into self-employment, possibly as a way to accommodate their condition, while wealthy wage and salary workers are pulled into it (Zissimopoulos and Karoly, 2007b). Low-wealth, self-employed workers are more likely than those with high wealth to transition to wage and salary work.

Table 1.

Employment Transitions for Men and Women Ages 51–67

Status at time t Status at time t + 2 (percent distribution)
Wage and salary Self-employed Not working
Men
 Wage and salary 72.8% 1.9% 25.3%
 Self-employed 6.5% 64.8% 28.7%
 Not working 5.0% 2.1% 92.9%
Women
 Wage and salary 71.8% 1.1% 27.1%
 Self-employed 6.1% 56.5% 37.4%
 Not working 4.4% 1.6% 94.0%
Open in a new tab

Source: Zissimopoulos and Karoly (2007a) using data from the Health and Retirement Study (HRS) 1992–2000.

Notes: Table 1 shows the percent of men and women ages 51 to 67 in the Health and Retirement Study (HRS) who transition between self-employment, wage and salary work, and nonemployment during a two-year period. Percents have been calculated using HRS sampling weights.

While exits from the labor force are common at these ages, men and women also enter the labor market. In a two-year period, 7 percent of men who were not working enter the labor force and among them, 30 percent become self-employed. Accumulation of capital, in the form of prior experience in self-employment and assets are by far the most important factors in the choice to return to work in self-employment (Zissimopoulos and Karoly, 2008).

While job changes at older ages, whether voluntary or involuntary, are associated with a decline in wages and loss of pension and health benefits, most are also associated with reduced work stress, physical effort, and managerial responsibilities (Johnson and Kawachi, 2007). Overall, older workers appear to enjoy work with 89 percent of older workers over age 50 reporting so (Zissimopoulos and Karoly, 2007a).

In fact, many older men and women would apparently like to work but cannot find jobs. Maestas and Li (2006) find that only about one-half of older job-searchers attain jobs within two years. Similarly, Chan and Stevens (2001) report that an older male worker who is laid off has a 61 percent chance of returning to work two years later and an older female worker has a 55 percent chance, suggesting older workers face many hurdles in the job search. Understanding why individuals are involuntarily out of the labor force, who chooses self-employment, and the consequences of both for retirement income and wealth take on added importance in a policy environment keen on encouraging work at older ages.

In short, older workers, like younger workers, move in and out of the labor force, change jobs, occupations, and industries, and move between wage and salary work and self-employment. A dynamic, experienced, highly educated, and older workforce presents opportunities for employers and older workers alike.

The Big Picture

The economic challenges posed by an aging society are significant, but not insurmountable. In fact, population aging will be less severe in the United States than in many other OECD countries for the simple reason that fertility rates in the United States never fell quite as far as they did in other OECD countries following the postwar baby boom. Indeed, the decline in fertility experienced in the United States was only half the average decline in OECD countries (OECD, 2009).

Figure 7 shows how the United States compares to other OECD countries on two critical dimensions: population aging and capacity for labor force adjustment. In terms of population aging, the United States is in the middle, on course to experience a relatively moderate change in the percent of its population over 65 between 2005 and 2030. The U.S. economy is also well positioned for further adjustment in labor force participation, having already attained labor force participation rates among those over 65 that are well above most OECD countries. In contrast, many European countries will face moderate to high population aging, yet virtually no one over age 65 works. While this certainly leaves ample room for adjustment in Europe, doing so may be particularly difficult given mutually reinforcing social norms and social insurance systems that facilitate early retirement through a variety of channels. Perhaps the main area in which the United States faces greater challenges than other countries is its health care system, with its high costs and its mixture of private-sector insurance with Medicare.

Figure 7. Population Aging and Labor Force Participation in OECD Countries.

Figure 7

Open in a new tab

Source: Source OECD Employment and Labour Market Statistics, LFS by Sex and Age (〈http://titania.sourceoecd.org/vl=4696454/cl=14/nw=1/rpsv/~3962/v123n1/s8/p1〉); OECD Factbook 2008, Population 65 and Over (〈http://oberon.sourceoecd.org/vl=1758417/cl=11/nw=1/rpsv/factbook2009/01/02/01/index.htm〉).

Note: Labor force participation rates are for men and women combined.

Demographics is destiny in the sense that the population age distribution is set decades before its effects occur. But the economic response is typically more adaptive than demographic determinism would suggest. Indeed, in the United States, a variety of economic forces—including changes in pensions, longer life expectancy, delay in disability onset, and rising labor force attachment among women—are likely to push labor force participation at older ages higher even if there are no further changes in policy.

As markets continue to adjust to the aging of the American workplace, more employers will no doubt find it advantageous to hire older workers and to devise new mechanisms for retaining older workers. Making work attractive may require firms to compete on the nonpecuniary dimensions that are valued by older workers, such as flexible hours and reduced responsibilities. The dynamic nature of labor force participation at older ages suggests a workforce that is capable of change in response to economic incentives and evolving personal preferences.

Public policy could support these forces and encourage still greater labor force participation by older Americans. For example, policymakers could undo the remaining work disincentives that are built into private and public pensions or work to resolve information-related labor market imperfections; a growing body of evidence suggests that people do not fully understand the rules governing their pension and Social Security benefits. Encouraging work at older ages serves a variety of social goals, including counteracting the slowdown in labor force growth and shoring up the finances of Social Security and Medicare. As men and women extend their working lives, they enhance their own retirement income security and may ease the strain of an aging population on economic growth.

Acknowledgments

We are grateful to David Autor, James Hines, Bruce Sacerdote, Timothy Taylor, Alan Gustman, and Joanna Lahey for many helpful comments and suggestions. We thank Larry Katz for generously providing his point estimates of years of schooling by birth cohort, Qian Gu for research assistance regarding declines in weekly hours of self-employed older workers over time, and Aviva Litovitz for assistance with OECD data. This publication was made possible by grant numbers 1 R01 AG025552-01 and 1 P01 AG022481-01 from the National Institute on Aging. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institute on Aging.

Footnotes

1

Authors calculations using the Health and Retirement Study.

2

The gradual increase in the full retirement age from 65 to 67 was announced in 1983, but the first cohort to which it applied (born 1938) did not reach age 62 until 2000. The increase in the credit for delaying claiming past the full retirement age was announced in 1983 and phased in between 1987 and 2005. The earnings test was first reduced in 1990 (announced in 1983) and then partially eliminated in 2000 (announced in 2000).

3

The actuarial penalty is much smaller now than it was decades ago when firms could stop crediting benefits for work after age 65.

4

The same rule applies to men, but in practice very few men receive spousal benefits. A divorced individual may also claim a spousal benefit if the marriage lasted at least 10 years.

5

Without further changes to plan rules, many defined benefit plans would impose a penalty on phased retirement since earnings in the final years of employment, which determine benefits, would be reduced.

6

Percent of men and women changing labor force status and class of work represents a weighted average of the percent of wage and salaried workers, self-employed, and nonworkers changing labor force status.

Contributor Information

Nicole Maestas, Email: Nicole_Maestas@rand.org.

Julie Zissimopoulos, Email: Julie_Zissimopoulos@rand.org.

References

  1. Autor David H, Levy Frank, Murnane Richard J. The Skill Content of Recent Technological Change: An Empirical Exploration. Quarterly Journal of Economics. 2003;118(4):1279–1333. [Google Scholar]
  2. Black Sandra, Spitz-Oener Alexandra. Technological Change and the Skill Content of Women’s Work. Review of Economics and Statistics. Forthcoming. (NBER Working Paper 13116.) [Google Scholar]
  3. Blau David, Goodstein Ryan. Can Social Security Explain Trends in Labor Force Participation of Older Men in the United States? Journal of Human Resources Forthcoming. [Google Scholar]
  4. Chan Sewin, Stevens Ann H. Job Loss and Employment Patterns of Older Workers. Journal of Labor Economics. 2001;9(2):186–205. [Google Scholar]
  5. Christensen Kaare, Doblhammer Gabriele, Rau Roland, Vaupel James W. Ageing Populations: The Challenges Ahead. Lancet. 2009;374(9696):1196–208. doi: 10.1016/S0140-6736(09)61460-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Coile Courtney. Retirement Incentives and Couples’ Retirement Decisions. Topics in Economic Analysis and Policy. 2004;4(1):article 17. [Google Scholar]
  7. Costa Dora L. The Evolution of Retirement: Summary of a Research Project. American Economic Review. 1998;88(2):232–36. [Google Scholar]
  8. Cutler David. Declining Disability among the Elderly. Health Affairs. 2001;20(6):11–27. doi: 10.1377/hlthaff.20.6.11. [DOI] [PubMed] [Google Scholar]
  9. Duggan Mark, Singleton Perry, Son Jae. Aching to Retire? The Rise in the Full Retirement Age and Its Impact on the Disability Rolls. NBER Working Paper 11811 2005 [Google Scholar]
  10. Friedberg Leora, Owyang Michael. Explaining the Evolution of Pension Structure and Job Tenure. NBER Working Paper W10714 2004 [Google Scholar]
  11. Friedberg Leora, Webb Anthony. Retirement and the Evolution of Pension Structure. Journal of Human Resources. 2005;40(2):281–308. [Google Scholar]
  12. Goldin Claudia, Katz Lawrence F. Long-Run Changes in the Wage Structure: Narrowing, Widening, Polarizing. Brookings Papers on Economic Activity. 2007;(2):135–67. [Google Scholar]
  13. Gustman Alan L, Steinmeier Thomas L. Retirement Outcomes in the Health and Retirement Study. Social Security Bulletin. 2000;63(4):57–71. doi: 10.3386/w7588. [DOI] [PubMed] [Google Scholar]
  14. Gustman Alan L, Steinmeier Thomas L. Projecting Behavioral Responses to the Next Generation of Retirement Policies. In: Polachek Solomon W, Tatsiramos Konstantinos., editors. Research in Labor Economics, Vol. 28: Work Earnings, and Other Aspects of the Employment Relation. JAI Press; 2008. pp. 141–96. [Google Scholar]
  15. Gustman Alan L, Steinmeier Thomas L. How Changes in Social Security Affect Recent Retirement Trends. Research on Aging. 2009;31(2):261. [Google Scholar]
  16. Gustman Alan L, Steinmeir Thomas L, Tabatabai Nahid. Pensions in Health and Retirement Study. Harvard University Press; Forthcoming. [Google Scholar]
  17. Hamilton Brady, Martin Joyce A, Ventura Stephanie J. Births: Preliminary Data for 2007. National Vital Statistics Reports. 2009;57(12) [PubMed] [Google Scholar]
  18. Hedden Tray, Gabrieli John. Insights into the Ageing Mind: A View from Cognitive Neuroscience. Nature Reviews Neuroscience. 2004;5(2):87–96. doi: 10.1038/nrn1323. [DOI] [PubMed] [Google Scholar]
  19. Hurd Michael, Rohwedder Susann. Trends in Labor Force Participation: How Much Is Due to Changes in Pensions? 2008. Unpublished paper. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Johnson Richard W, Kawachi Janette. Job Changes at Older Ages: Effects on Wages, Benefits and Other Job Attributes. 2007 Available at SSRN: http://ssrn.com/abstract=1299189.
  21. Johnson Richard W, Neumark David. Age Discrimination, Job Separations, and Employment Status of Older Workers: Evidence from Self-Reports. The Journal of Human Resources. 1997;32(4):779–811. [Google Scholar]
  22. Kaiser Family Foundation and Hewitt Associates. Retiree Health Benefits Examined: Findings of the Kaiser/Hewitt 2006 Survey on Retiree Health Benefits. 2006 http://www.kff.org/medicare/7587.cfm#at/
  23. Karoly Lynn A, Zissimopoulos Julie. Self-Employment among Older U.S. Workers. Monthly Labor Review. 2004;27(7):24–47. [Google Scholar]
  24. Katz Lawrence F, Murphy Kevin M. Changes in Relative Wages, 1963–1987: Supply and Demand Factors. The Quarterly Journal of Economics. 1992;107(1):35–78. [Google Scholar]
  25. Lahey Joanna. Age, Women, and Hiring: An Experimental Study. Journal of Human Resources. 2008;43(1):30–56. [Google Scholar]
  26. Lakdawalla Darius N, Goldman Dana P, Shang Baoping. Health Affairs. 2005. The Health and Cost Consequences of Obesity among the Future Elderly; pp. W5-R30–W5-R41. Web Exclusive, September 26, 2005. [DOI] [PubMed] [Google Scholar]
  27. Li Xiaoyan, Maestas Nicole. Michigan Retirement Research Center Working Paper 98. 2008. Does the Rise in the Full Retirement Age Encourage Disability Benefits Applications? Evidence from the Health and Retirement Study. [Google Scholar]
  28. Maestas Nicole. Labor, Love, and Leisure: Complementarity and Timing of Retirement by Working Couples. 2001 http://client.norc.org/jole/SOLEweb/maestas.pdf.
  29. Maestas Nicole. Cohort Differences in Retirement Expectations and Realizations. In: Madrian Brigitte, Mitchell Olivia, Soldo Beth., editors. Redefining Retirement: How Will Boomers Fare? Vol. 2. Oxford University Press; 2007. [Google Scholar]
  30. Maestas Nicole. Back to Work: Expectations and Realizations of Work after Retirement. Journal of Human Resources. doi: 10.1353/jhr.2010.0011. Forthcoming. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Maestas Nicole, Li Xiaoyan. Michigan Retirement Research Center Working Paper 133. 2006. Discouraged Workers? Job Search Outcomes of Older Workers. [Google Scholar]
  32. Mare Robert. Five Decades of Educational Assortative Mating. American Sociological Review. 1991;56(1):15–32. [Google Scholar]
  33. Martin Linda, Schoeni Robert, Freed-man Vicki, Andreski Patricia. Feeling Better? Trends in General Health Status. Journal of Gerontology. 2007;62B(1):S11–S21. doi: 10.1093/geronb/62.1.s11. [DOI] [PubMed] [Google Scholar]
  34. Mastrobuoni Giovanni. Labor Supply Effects of the Recent Social Security Benefit Cuts: Empirical Estimates Using Cohort Discontinuities. Journal of Public Economics. 2009;93(11–12):1224–33. [Google Scholar]
  35. Moen Jon R. Rural Nonfarm Households: Leaving the Farm and the Retirement of Older Men, 1860–1980. Social Science History. 1994;18(1):55–75. [Google Scholar]
  36. Munnell Alicia, Sass Steven. Working Longer. The Solution to the Retirement Income Challenge. Washington, DC: Brookings Institution Press; 2008. [Google Scholar]
  37. Munnell Alicia, Sass Steven, Soto Mauricio. Work Opportunities for Older Americans Series 3. Center for Retirement Research; 2006. Employer Attitudes toward Older Workers: Survey Results. [Google Scholar]
  38. Neumark David. NBER Working Paper 8512. 2001. Age Discrimination Legislation in the United States. [Google Scholar]
  39. OECD. Live Longer, Work Longer. Paris: OECD; 2006. [Google Scholar]
  40. OECD. [accessed October 21, 2009];Total Fertility Rates. 2009 http://www.oecd.org/dataoecd/44/34/37962718.pdf.
  41. Perun Pamela, Valenti Joseph J. Defined Benefit Plans: Going, Going, Gone?. Paper presented at the Thirtieth Annual APPAM Research Conference; November 7, 2008; Los Angeles, CA. 2008. [Google Scholar]
  42. Purcell Patrick. Older Workers: Employment and Retirement Trends. Congressional Research Service Report for Congress.2008. [Google Scholar]
  43. Quinn Joseph. Changing Retirement Trends and Their Impact on Elderly Entitlement Programs. In: Altman Stuart H, Shactman David I., editors. Chapter 13 of Policies for an Aging Society. Baltimore: Johns Hopkins University Press; 2002. [Google Scholar]
  44. Ruhm Christopher J. Bridge Jobs and Partial Retirement. Journal of Labor Economics. 1990;8(4):482–501. [Google Scholar]
  45. Schirle Tammy. Why Have the Labor Force Participation Rates of Older Men Increased since the Mid-1990s? Journal of Labor Economics. 2008;26(4):549–94. [Google Scholar]
  46. Toossi Mitra. A Century of Change: The U.S. Labor Force, 1950–2050. Monthly Labor Review. 2002 May;:15–28. [Google Scholar]
  47. Toossi Mitra. A New Look at Long-term Labor Force Projections to 2050. Monthly Labor Review. 2006 Nov;:19–39. [Google Scholar]
  48. Weir David. Are Baby Boomers Living Well Longer? In: Madrian Brigette, Mitchell Olivia S, Soldo Beth J., editors. Redefining Retirement: How Will Boomers Fare. New York: Oxford University Press; 2007. pp. 95–111. [Google Scholar]
  49. Zissimopoulos Julie, Karoly Lynn A. AARP Public Policy Institute Publication #2007-04. 2007a. Work and Well-Being among the Self-Employed at Older Ages. [Google Scholar]
  50. Zissimopoulos Julie, Karoly Lynn A. Transitions to Self-Employment at Older Ages: The Role of Wealth, Health, Health Insurance and Other Factors. Labour Economics. 2007b;14(2):269–95. [Google Scholar]
  51. Zissimopoulos Julie, Karoly Lynn A. Labor Force Dynamics at Older Ages: Movements in Self-Employment for Workers and Nonworkers. Research on Aging. 2008;31(1):89–111. doi: 10.1177/0164027508324642. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES