In addition, Social Security provides a higher income replacement rate for those in low-SES households (Kahn et al., 2017), reducing the economic impact of leaving the labor force. The health and work opportunity barriers to employment mean that working longer may not be a realistic solution to declines in economic status at older ages for this group (Munnell, Webb et al., 2018), suggesting that working longer is not simply a matter of choice for many Americans. In fact, those who would benefit the most economically from remaining in the labor force often face constraints to participation that reduce both their preference for working longer and the viability of this as an option for improving their economic circumstances (Solem et al., 2014; Szinovacz et al., 2014; Siegrist et al., 2007).

Systemic Disparities

Constraints on work at older ages are not distributed equally throughout the population but fall along existing lines defined by social and economic inequalities such as gender, race-ethnicity, immigration status, socioeconomic status, and geographic region. Health, financial security, employment, and retirement at older ages are shaped not only by current characteristics but also by the opportunity structures experienced throughout the life course (Moen et al., 2020, 2021; Fisher, Chaffee et al., 2016; Warner and Brown, 2011; Brown, 2009). Thus, the impacts of discrimination and structural inequality experienced in early adulthood and midlife continue to affect the work and retirement opportunities available to older adults. The effects of these inequities accumulate over the life course, determining the resources available for retiring or reducing employment in later life (Rothstein, 2017; Brown, 2016). Then, transitions between work and retirement reproduce and reinforce these social and economic inequalities at older ages. How these factors work together to shape work and nonwork in later life has implications for the well-being of older adults.

COMMITTEE FORMATION AND STATEMENT OF TASK

In 2019, the Arthur P. Sloan Foundation asked the National Academies of Sciences, Engineering, and Medicine (the National Academies) to produce a consensus study on the aging workforce and employment at older ages. The specific charge to the National Academies was stated as follows:

The National Academies appointed the Committee on Understanding the Aging Workforce and Employment at Older Ages to carry out this task. Ten prominent scholars representing a broad array of disciplines—including economics, psychology, organizational psychology, labor relations, sociology, and social work—were included on the committee. The committee met virtually, six times over a 10-month period, to produce this report.

SITUATING INDIVIDUAL-LEVEL CHARACTERISTICS WITHIN CONTEXTS

The statement of task calls for a focus on the ways in which individual-level characteristics are associated with work at older ages. However, it is not possible to understand the role of individual-level characteristics without considering their relationship to the contexts in which work occurs.

Some of these characteristics, such as income or wealth, might be measured at the individual level as an individual’s earnings or assets, but individuals are often situated within families in which the income and assets are a shared resource. Relatedly, married couples tend to make joint retirement decisions that lead both spouses to retire at close to the same time (Angrisani et al., 2017; Coile, 2015), which leads women, who are generally

younger than their spouses, to retire at younger ages (Maestas, 2018). These decisions may be driven by caregiving needs, which are more likely to be shouldered by women. In fact, women are more likely than men to reduce their employment to meet the informal caregiving needs of a spouse, parent, child, or other relative. The gender gap in earnings, which widens with age (Goldin et al., 2017), also means that women remain more dependent on their spouse’s earnings to maintain their standard of living. This greater dependence is why women are more likely to increase their labor market participation after the death of a spouse than are men (Fadlon and Nielsen, 2017).

Individual characteristics, such as age, race-ethnicity, and education, can shape opportunities for work as well as the experience of work within workplaces. Workplace policies and job characteristics often play a decisive role in decisions to retire. Preferences for certain job characteristics can change over the life course, and these preferences can affect retirement transitions. For example, older workers place a higher value on non-pecuniary job characteristics than do younger workers (Maestas et al., 2018). Characteristics such as moderate physical activity, sitting, team-based evaluation, schedule flexibility, and work autonomy are disproportionately valued by older workers (Maestas et al., 2018). Those who work longer in their career job or exit their career job for bridge employment tend to work fewer hours, have a flexible schedule, and receive lower hourly wages (Ameriks et al., 2018). Workers employed by employers who are more willing to accommodate a lighter workload for older employees are more likely to remain employed full-time and less likely to retire (Ameriks et al., 2018).

Employers can signal the value they place on retaining older workers through the policies and practices they enact within the workplace. For example, workers who are provided with a health- or disability-related accommodation by their employer are less likely to file for disability and more likely to remain employed than workers who do not receive accommodation (Maestas et al., 2019; Hill et al., 2016). In contrast, employers may weight employee evaluations toward characteristics associated with younger workers, such as speed and innovation, while discounting those that are more commonly associated with older workers, such as knowledge, expertise, and loyalty (Van Dalen et al., 2010). Employers may also view older employees as less technologically proficient or less adaptable to new environments or practices and may invest less in training them. These forms of workplace discrimination affect workers’ decisions to remain in their positions (Angrisani et al., 2016).

Individual characteristics also shape the opportunities for changing jobs or returning to work. Historically, employment opportunities have narrowed at older ages as the physical ability to perform work tasks deteriorates (Rutledge et al., 2017). However, technological change and economic shifts to jobs that are less dependent on manual and routine skills and favor those with cognitive and analytic skills (Autor et al., 2003) have allowed more older adults to remain in the labor force despite health limitations at older ages. The shift away from physically demanding occupations creates working environments in which health limitations are more easily accommodated by employers (Maestas and Zissimopoulos, 2010), but the higher-skill demands of these jobs prevent less educated workers from benefiting from these changes. This means that employment opportunities at older ages have increased for women—who are less likely to be employed in physically demanding occupations than men—and more educated workers but have remained narrow for less educated men (Rutledge et al., 2017), restricting the latter’s ability to delay retirement.

Macroeconomic and local labor market conditions also play an important role in shaping opportunities for employment at older ages. Older workers who are near retirement age are more responsive to changes in the labor market, both positive and negative, than younger workers (Gorodnichenko et al., 2013; Maestas, Mullen, and Powell, 2013; Coile and Levine, 2011). Factors such as globalization and automation can change the demand for workers in specific industries and occupations (Lee and Angrisani, 2020; Maestas, 2010), and these changes can have much stronger effects on older workers. Moreover, older adults are less likely to relocate to areas with better employment prospects (U.S. Census Bureau, 2019), which leaves them more susceptible to changes in local economic conditions.

Negative labor market shocks, such as the onset of a recession, the ongoing COVID-19 pandemic, or the shuttering of a large employer, are more likely to affect the long-term employment behavior of older workers. As unemployment increases, full-time workers become more likely to transition into partial or full retirement, while those in partial retirement become more likely to enter full retirement (Papadopoulos et al., 2020; Gorodnichenko et al., 2013). Once unemployed, older workers face longer periods of unemployment than younger unemployed workers, are more likely to file unemployment claims than middle-aged workers (Neumark, 2018), and suffer

larger losses in earnings than younger unemployed workers (Davis and von Wachter, 2011). Unemployment-induced earlier retirement is also associated with lower income in retirement and earlier mortality (Coile et al., 2014; Coile and Levine, 2011).

Public policies can play an important role in individual retirement decisions by setting the context for retirement and work decisions. The important roles of Social Security and Medicare in providing individual financial support and health insurance coverage are well established (for example, see Bee and Mitchell, 2017). But a host of other policies, addressing retirement savings, paid and unpaid leave to address caregiving responsibilities, subsidies to provide formal care, and retraining may also improve financial resources or opportunities to continue working at older ages. Beyond these individual effects, public policies can affect the incentives for organizations to implement workplace changes that enable or inhibit work at older ages. These policy environments are not set in stone, but are ever in flux.

Thus, to fully understand how individual characteristics affect the experience of work, their effects must be considered within the broader contexts through which they influence individual preferences, expectations, and constraints for work and retirement. This will provide a deeper understanding of the relationship between context and individual outcomes, providing a guide for future policymaking.

BOUNDARIES ON THE SCOPE OF THIS REPORT

The scope of the statement of task assigned to the committee was broad, which meant that producing a concise and useful summary of the research literature required the committee to identify clear boundaries to its review. Based on its statement of task and discussions with the study sponsor, the committee understood that its review should be centered on the behavioral and social factors associated with continuing work at older ages, with an eye toward the development of a clear research agenda identifying key topics that should be targeted in future research.

There were two topics that were relevant to the committee’s task but so broad in scope that a thorough review of their content would not be feasible within the timeframe of the study. Moreover, because the committee’s primary task was to identify and highlight new research questions and areas that have not received adequate attention within the field to date, the extensive extant literature on these two topics made them a less promising source for identifying new research questions. The two topics are the relationship between health and work and the role of public policy in shaping retirement decisions. Both of these factors play such important roles in transitions between work and retirement at older ages that they could not be ignored in their entirety. For this reason, the committee set boundaries regarding the coverage of these topics within this report.

One such boundary concerned the relationship between work and health. Health is one of the most important predictors of labor force participation at older ages, both because of the direct causal relationship between work and health and because of its relationship to other characteristics that shape opportunities for work, such as educational attainment (Jason et al., 2017; Zajacova et al., 2014; Cahill et al., 2006). The causal relationship between work and health can run in both directions. Health status can affect work and retirement decisions; poor health can lead an individual to discontinue working and leave the labor force (Zajacova et al., 2014; Warner and Brown, 2011), while improvements in health can enable individuals to continue working longer (Coile et al., 2017). At the same time, work can also affect health, both positively, by promoting physical, cognitive, and social engagement (Carr et al., 2021; Fitzpatrick and Moore, 2018; Berkman et al., 2014), and negatively, through physical demands and injuries and mental stress (Fisher, Chaffee et al., 2016). This complex relationship makes identifying the effect of health on work inherently difficult and has spawned an extensive literature that seeks to establish its causal nature.

The committee viewed health as one of the individual-level factors that predicts labor force participation at older ages, and therefore provides a brief review of the research establishing a causal effect of health on work. Beyond this, the committee considered the effects of health only insofar as they were related to other factors under consideration, such as caregiving or workplace accommodations. In particular, a review of the effect of work on health was considered outside the scope of the committee’s statement of task.

The second boundary the committee established in its review focused on the relationship between public policy and work at older ages. As noted above, public policies play an important role in setting the contexts in which individuals’ preferences, expectations, and constraints are defined. For example, the age-based eligibility

requirements of older-age programs, such as Social Security and Medicare, can have strong effects on retirement decisions (Fadlon and Deshpande, 2020; Coile et al., 2019; Burtless and Moffitt, 1986). A wide variety of programs have been introduced to encourage saving behaviors at younger ages in order to financially prepare individuals and families for retirement (e.g., 401(k) plans and IRAs). In addition, many other age-neutral programs and policies, such as family leave and job retraining programs, can help older workers remain in the labor force. The wide-ranging variety and scope of such programs made a thorough review of the literature evaluating the effects of these programs infeasible. Such a review would constitute its own report. The committee compromised by including a chapter that provides a broad overview of both age-targeted and age-neutral public programs that may affect the work behaviors of older workers.

Although the statement of task explicitly charges the committee with a focus on what is known about the United States, population aging is occurring within many countries, and this has led to the creation of a rich and informative international literature on working longer that can help inform our understanding of working longer in the U.S. context. For this reason, the committee draws on this literature in places where similar research is lacking in the United States, when we wish to indicate that findings from the United States have shown broad applicability across countries, or when a comparative focus can reveal information about the specific U.S. context. However, the committee regarded research addressing how the U.S. context differs from that of other countries and how these differences explain international differences in work patterns at older ages as falling outside the scope of the report, and was mindful of these limits when discussing research conducted outside the United States.

UNDERSTANDING WORK AT OLDER AGES

The remainder of this introductory chapter briefly discusses what defines an older worker and introduces the theoretical concepts of expectations, constraints, and preferences for work and retirement, which the committee relies on as a frame for coverage of issues of an aging workforce.

Normative Expectations about Work at Older Ages

At what age do workers become “older workers?” There is no single, clear answer to this question to which the committee can point, because the relevant age range often depends on the topic of study. Federal laws targeting age discrimination in employment identify workers ages 40 and over as potential victims of age discrimination, while federal retirement policies are usually focused on adults in their 60s. The Health and Retirement Survey (HRS), a longitudinal panel study conducted biennially since 1992 by the University of Michigan, which is one of the most commonly used data sources for studying both retirement and preretirement behaviors of older adults in the United States, initially identified eligible households as those with at least one member who was between the ages of 51 and 61, in order to capture work behaviors prior to the initiation of retirement (Sonnega et al., 2014). As such, research on age discrimination in the workplace may consider “older workers” to include a younger age range than research focused on transitions to retirement or on re-entries into the labor force from retirement. Even when considering transitions to retirement, the employment experiences of and opportunities available to workers approaching retirement may differ substantially from those who are well past conventional retirement ages.

Thus, in many cases, the relevant definition of “older worker” is dependent on what is meant by “retirement age.” Unlike some countries, the United States does not have an officially recognized mandatory retirement age; however, age-based eligibility requirements for older-age programs such as Social Security and Medicare have often served as benchmarks for establishing conventional retirement age (Coile et al., 2019; Gruber and Wise, 1999). This is in part because the benefits offered by these programs enable retirement (Song and Manchester, 2007), but also because normative expectations about what constitutes the appropriate age to retire underlie the establishment of their eligibility ages.

As a growing proportion of adults continues to work beyond conventional retirement ages, normative expectations about what constitutes retirement age could change. Such a change may be further incentivized by the need to bolster the solvency of social programs directed at older adults (U.S. Social Security Administration, 2020c), because policy changes such as raising the age at which enrollees receive partial or full benefits can reduce the

overall costs of these programs (Olsen, 2012). In fact, fears that the dramatic increases in life expectancy that occurred in the mid-twentieth century (coupled with the effects of high inflation and unemployment rates in the late-1970s and early-1980s) would lead to the Social Security program’s insolvency motivated the 1982 passage of legislation that gradually raised the age at which individuals became eligible for its full benefits, from age 65–67 (McSteen, 1985). However, due to concerns that not all workers benefited equally from these gains in longevity and that many adults faced difficulty remaining in the labor force at older ages, this legislation left the original age at which individuals became eligible for partial benefits at age 62. This latter decision did protect workers who were less able to continue working at older ages. Yet it also potentially widened economic inequality among older adults by constraining the qualification for full benefits for the most vulnerable older workers, that is, those with limited employment opportunities or with health conditions that restrict their activity. Some of these workers may retain eligibility under the Social Security Disability Insurance provisions, subject to establishing the presence of a qualifying disability.

Though many policies have aimed to improve the financial stability of older adults through the introduction of targeted retirement savings vehicles, research has consistently shown that working longer has a greater impact on financial outcomes than changing savings behaviors (Bronshtein et al., 2019). In addition, much of the research on work at older ages is focused on factors that enable these adults to remain in the labor force longer or on barriers that prevent them from doing so (e.g., Fast et al., 2020; Keating et al., 2019; Stoilko and Strough, 2019; Kalleberg, 2018; Gustafson, 2017; Fisher, Chaffee et al., 2016; Feldman and Beehr, 2011; Moen et al., 2006; Raymo and Sweeney, 2006; Dentiger and Clarkberg, 2002). One might plausibly conclude from this formulation of the research that there is an underlying normative expectation that these adults should be working longer. But this need not be the case; drawing this conclusion largely depends on how the research question is framed and presented. For example, research on factors that enable or prevent continued work at older ages could, where possible, more clearly identify and target older adults who express an interest in continuing to work.

Older adults who continue to work beyond traditional retirement ages are disproportionately those who are in good health, have more education, have better employment opportunities, and prefer to remain in the labor force (Moen, 2016b; Cahill et al., 2006)—precisely those who are most likely to experience work as a positive contribution to their well-being. This means that often research on older workers, particularly on those that remain in the labor force beyond traditional retirement ages, is based on population data, such as the HRS, in which the experiences of these well-off adults are well represented but, because less affluent adults are less likely to continue working, those of less affluent workers at these ages are not. This can obscure the work outcomes of less affluent older workers those of whose experiences in the labor force may be substantively different, creating a more positive representation of the effects of work on quality of life that could shift normative expectations about work and retirement at older ages. Thus, in presenting research on the older workforce, it is crucial to be clear about the characteristics of the study population to whom findings apply. Moreover, research that focuses on explicating the heterogeneity of work experiences at older ages would provide a more nuanced understanding of the role of work and reduce the likelihood of the ecological fallacy—in which the experiences of more economically advantaged adults will be seen as representative of the experiences of the older worker population as a whole—being applied to this population.

A Framework for Understanding Work at Older Ages

At the outset of our task, the committee members developed a framework for understanding the experience of work and transitions between work and nonwork at older ages. This framework relied on conceptual elements that were common across their respective research disciplines and provided a common language for presenting the research in this report. It is discussed in greater detail in Chapter 3, but at its core it conceptualizes work at older ages as being shaped by an individual’s preferences for work and expectations about the future, as well as constraints on his or her ability to realize preferred work relationships. These three dimensions—preferences, expectations, and constraints—each shape the others and are mutually reinforcing.

Much of the research about work at older ages focuses on the role of constraints or barriers to labor force participation at older ages (Fisher, Chaffee et al., 2016). As noted, this focus can be accompanied by an

often-unacknowledged assumption: that older adults should be working if they are able to do so. This, however, is a subjective value judgment rather than an empirically established fact. Incorporating the roles that preferences and expectations also play in shaping work outcomes acknowledges that work decisions are not based on work capacity or opportunities alone (James et al., 2016; Moen, 2016a). Individuals also weigh the relative value they place on engaging in other activities, such as leisure, travel, or volunteer work, when considering whether to continue working at older ages. These preferences shape and are shaped by expectations about how they will affect outcomes, such as financial stability, health, and quality of life (Boehm et al., 2014; Cahill et al., 2013; Wang and Shultz, 2010). Of course, constraints also play a significant role in whether individuals are able to fully realize their preferences for work at older ages, and these constraints also shape preferences for work and expectations for the future (Carr et al., 2016). Individuals face barriers to employment that can take many forms, such as health limitations on work capacity, caregiving demands, or lack of employment opportunities (Fast et al., 2020; Keating et al., 2019; Stoilko and Strough, 2019; Kalleberg, 2018; Gustafson, 2017; Feldman and Beehr, 2011; Moen et al., 2006; Raymo and Sweeney, 2006; Dentiger and Clarkberg, 2002).

Though the transition from work to retirement is often thought of as a single transition into an all-absorbing state, the reality is more complex and dynamic. Full retirement between ages 62 and 65 remains the most common pathway out of the labor force (Maestas, 2010), but a growing proportion of older Americans remain fully or partially employed in “bridge” employment (Ruhm, 1990) beyond these traditional retirement ages or return to work after a brief period out of the labor force (Fry, 2019; Quinn et al., 2019; Ameriks et al., 2018; Cahill et al., 2018; Maestas, 2010). Each of the three dimensions of preferences, expectations, and constraints can change over time, leading to multiple transitions between work and nonwork activities at older ages. Understanding how these forces affect the experience of work, as well as transitions into and out of retirement, can provide insight into the ways in which work contributes to well-being at older ages.

ORGANIZATION OF THE REPORT

The remainder of the report is organized into three parts. Part I provides background describing the aging workforce in the United States and further explicates the committee’s conceptual framework for understanding pathways between work and retirement. It begins with a description of the aging workforce that includes the demographic and health characteristics of those who remain in the labor force at older ages and the types of work they perform (Chapter 2). It then discusses the potential pathways between work and nonwork and a number of commonly used conceptual models for understanding transitions within this framework (Chapter 3).

Part II of the report examines what is known about the experience of work at older ages, focusing on factors that affect older workers’ preferences and their expectations for work and retirement, as well as the constraints that shape opportunities to realize these preferences and expectations. Chapters 4 through 8 are organized by the level of analysis at which the factors under consideration affect work at older ages. Chapter 4 begins with the most proximal factors, examining the individual and family-level characteristics associated with transitions between work and nonwork. Chapters 5 through 8 take a broader view, examining the ways in which context shapes and constrains the experience of work and pathways between work and nonwork at older ages. Chapter 5 focuses on the workplace and examines how workplace policies and practices can serve to enable or restrict the ability of older workers to remain in the labor force. Chapter 6 examines the role of age bias and discrimination in work-related experiences. Chapter 7 examines the role of labor markets in creating demand and opportunities for older workers to remain in the labor force. Chapter 8 focuses on the role of public policy in encouraging or enabling or, conversely, discouraging or operating as a barrier to labor force participation at older ages.

Part III of the report is a summary of the conclusions drawn from the preceding chapters and identifies key themes that cut across these chapters in order to outline a future research agenda on work at older ages. The final chapter (Chapter 9) draws from the findings of the earlier chapters. It identifies key conclusions that can be drawn from the extant research and proposes a detailed research agenda that, if enacted, would provide a comprehensive understanding of the social and economic role of work at older ages and the barriers older adults face in realizing their preferred work and retirement relationships.

This page intentionally left blank.