Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2024 Jun 5;79(8):gbae104. doi: 10.1093/geronb/gbae104

Bridge Employment or Encore Career? Examining Predictors That Distinguish Later-Life Career Transitions

Yun taek Oh 1,✉
Editor: Markus Schafer2
PMCID: PMC11226993  PMID: 38837332

Abstract

Objectives

Bridge employment and encore careers are 2 prevalent retirement pathways that have different goals and outcomes. Yet, “changing jobs in later life” is the shared prequel that blurs the distinction between them in empirical studies. This study proposes a set of criteria—voluntariness of career transition and the duration of work in the posttransition job—to distinguish various retirement pathways and investigates the predictors that distinguish the workers’ choice of these pathways.

Methods

I conducted multinomial logistic regression to examine the predictors that distinguish between bridge employment, encore career, and direct workforce exit using the longitudinal sample of respondents with full-time career jobs in the Health and Retirement Study 1992–2020 (HRS, N = 2,038). To examine the predictors that distinguish between bridge employment and encore careers, I conducted logistic regression on the subsample of respondents who chose either bridge employment or encore careers (n = 927).

Results

The results show that the accumulated human capital from career jobs, physical and mental health conditions before leaving career jobs, and self-identified retirement status when transitioning to new jobs distinguish the workers’ choices of taking on different retirement pathways.

Discussion

Maintaining the labor force participation of older workers is an important human resource agenda for policymakers. This study suggests that increasing the number of quality jobs for older workers would promote bridge employment and encore careers by raising the benefits of making career transitions as well as improving older workers’ health.

Keywords: Retirement, Retirement pathway, Retirement transition, Workforce aging

The increased longevity and prolonged work lives contributed to the diversification of how to exit the workforce. The traditional “lockstep” retirement—exiting the workforce straight from career jobs—is outdated and unattainable for most older workers (Kojola & Moen, 2016). Instead, these workers choose various retirement pathways to fill the transition between ending lifetime career job and complete workforce exit (Cahill et al., 2006, 2012). Starting some form of bridge employment or beginning a new “encore” career is increasingly observed in the employment patterns of contemporary cohorts of older workers (Quinn, 2010).

Bridge employment is defined as participation in some type of paid work after ending a career job and before complete workforce exit. Researchers report multiple forms of bridge employment such as switching to different occupations, working for different employers, and reducing work hours (Bishop, 2022; Cahill et al., 2012). Previous studies have documented the reasons and motivations of older workers undertaking bridge employment: deriving more life satisfaction from working, having control over their lives, engaging in challenging new tasks, reducing physical job demands, and increasing their retirement-related financial resources (Ekerdt, 2010; Gustman & Steinmeier, 1991; Kim & Feldman, 2000; Pengcharoen & Shultz, 2010).

Compared to bridge employment, an encore career is an emerging retirement pathway that refers to the activities of pursuing a new career in later life, largely motivated by social purposes: “finding meaning in the workplace” and “improving one’s communities” (Freedman, 2006, p. 44). Boveda and Metz (2016) propose that encore careers can be distinguished from bridge employment because the implication of beginning a job is to continue labor force participation in a new career where bridge employment is to “phase out” of the workforce (p. 157). These studies suggest that although the motivation to pursue an encore career might be similar to those of bridge employment, the end goal is different: remaining in the workforce versus phasing out.

Cahill et al. (2018) ask whether the empirical studies in the bridge employment literature actually investigate the “transition to workforce exit” role of bridge employment; instead, the bridge job activities in the data might be “another series of job changes” (p. 330). Among these series of job changes, some of them may be to start new career jobs; as a “new phase of work” described in Freedman (2006, p. 44), starting a new career also requires “job changes” as bridge employment does. It means that “changing jobs in later life” is not a distinct prequel of bridge employment but a common starting point in pursuing both retirement pathways. This is problematic for the researchers because available quantitative data may not provide sufficient information, such as work attitudes and/or labor market contexts, to distinguish between these pathways.

In this study, I propose two criteria—the voluntariness of career transition (job changes) and the duration of work in the posttransition job—to distinguish between various later-life retirement pathways, including direct workforce exit, involuntary career transitions, bridge employment, and encore careers. Then, I examine the predictors that distinguish between these pathways, based on the theories from multiple disciplines.

This study provides three contributions to the retirement literature. First, this is one of the first studies suggesting how to distinguish between bridge employment and encore careers using criteria that are testable using quantitative data and largely available in longitudinal data. Most of the previous studies investigated encore careers as a subjectively defined pathway that is determined using qualitative approaches. In addition, as Cahill et al. (2018) address, some studies failed to distinguish bridge employment from other retirement pathways that have different motivations and outcomes. This study opens the possibilities of making clear distinctions between the two pathways that share the same prequel and conducting quantitative research on encore careers by imposing quantitatively testable conditions. Second, this is one of the first studies investigating the role of self-identified (self-reported) retirement status on the choice of retirement pathways of older workers. Although the importance of this status in retirement research was first addressed several decades ago by Murray (1979), its influence on later-life outcomes is not well-documented. Instead, previous studies used this status as a key element in directly classifying one’s “partial retirement” status. Unlike these studies, I use this status as a predictor of a certain pathway, “encore careers” to be specific, based on retirement identity theory, suggested in Szinovacz and DeViney (1999). Third, this study provides a deeper understanding of sociodemographic and economic mechanisms of choosing different retirement pathways in later work lives based on the theories from multiple disciplines.

Theoretical Framework

Categorizing Retirement Pathways

Feldman (1994) defines bridge employment as a work activity to bridge between career jobs and complete workforce exit, which “usually lasts for a shorter duration.” Feldman also described the 10-year duration of work in the same job as a “long-term career job” that is used by labor economists (e.g., Lawson, 1991). These definitions suggest that the duration of work in the new job may indicate the purpose of career transition: bridging between career jobs and complete workforce exit or pursuing a new career. In addition, Freedman (2006) suggests “10- or 15-year career moving” that “might not be as long as midlife work” as a duration of encore careers (p. 46). Furthermore, the use of a “10-year duration of work in the same job” to identify workers’ careers is not a new concept in the retirement literature. For instance, the full-time career (FTC) job is defined as fulfilling “working full-time (1,600+ annual work hours) and 10+ years of tenure” in the same job (Cahill et al., 2006, 2018; Quinn, 1999). Thus, the use of “10-year duration of work in the posttransition job (new job after leaving career job)” to distinguish between bridge employment and encore careers reflects the goals of these pathways and simplifies the distinctions of these pathways by retrospectively tracking back from the time of respondents’ workforce exit (or another job change) to the time of their leaving from original career jobs.

Another factor is the voluntariness of career transitions. This is also important in distinguishing retirement pathways because the purpose, career job circumstances, and outcomes of involuntary transitions are different from those of voluntary transitions; these differences may introduce further heterogeneity in the estimations and interpretations of outcomes (Van Solinge & Henkens, 2007). For instance, Dingemans and Henkens (2014) show that although involuntary career transitions might partly compensate aspects of job-related and life satisfaction lost due to the involuntary leave from career jobs, voluntary career transitions, intentionally initiated by workers, contribute to increased life satisfaction. This result suggests that involuntary and voluntary career transitions have different social and psychological consequences and therefore must be treated as a separate retirement pathway (Hershey & Henkens, 2014).

Therefore, in this study, I use two criteria—“duration of work in the posttransition jobs” and “voluntariness of career transition”—to distinguish the retirement pathways in later life into four groups. First, direct workforce exit is defined as exiting the workforce directly from career jobs without making any career transition. Second, involuntary career transition is defined as any career transition that is done in an involuntary manner regardless of the duration of work after leaving career jobs. Third, bridge employment is defined as a voluntary career transition in which the duration of work in the posttransition job is less than 10 years. Last, an encore career is defined as a voluntary career transition in which this duration is 10 years or longer. For the 10-year duration of the posttransition job, I do not impose the “full-time work schedule” condition to identify one’s retirement pathway as encore careers because previous studies on bridge employment and encore careers (e.g., Feldman, 1994; Freedman, 2006; Lawson, 1991) did not necessarily impose a “full-time work schedule” condition to identify one’s “new” career. Furthermore, Bank et al. (2011) show that a flexible work schedule is an important aspect of pursuing encore careers, suggesting that imposing a “full-time work schedule” condition may not be relevant in identifying one’s “new career” in later life. Figure 1 illustrates these career transitions in detail.

Figure 1.

Alt Text: The retirement pathways are categorized into four groups: “direct workforce exit,” “involuntary career transition,” “bridge employment,” and “encore career.” These categorizations are done based on (1) whether to leave the workforce or to start new jobs, (2) leaving the career jobs voluntarily or not, and (3) whether to maintain the new jobs for 10+ years or not.

Open in a new tab

Typology of later-life career transition.

Trade-off Between Human Capital and Socioeconomic/Psychological Benefits

When making career transitions, one or both of the following are lost: occupation-specific and firm-specific human capital. Occupation-specific human capital is accumulated through occupational training and working in career occupations for a considerable duration, and it plays a significant role in higher performance, labor income, and employee benefits (Gibbons & Waldman, 2004). Similarly, firm-specific human capital is accumulated through firm-specific training and working in career jobs for a considerable duration and plays the same role as occupation-specific human capital (Loewenstein & Spletzer, 1999). These losses lead to the reduction in utility from working. Yet, in return, workers gain social and psychological benefits, such as life satisfaction and sense of control (Kim & Feldman, 2000; Pengcharoen & Shultz, 2010), leading to the increase in utility of working.

In aggregation, voluntary career transition involves a trade-off between the loss of accumulated human capital from career jobs and the gain of social and psychological benefits from working in new jobs. There are two implications for this trade-off. First, it suggests that workers change their jobs only if the utility gain from the benefits of working in new jobs is greater than the utility loss from the loss of human capital. Second, it implies that workers who accumulate a higher level of human capital will experience greater losses when leaving career jobs, and therefore these workers are less likely to make career transitions.

How do we measure human capital? One way is to use labor income and employee benefits; these are not only forms of financial capital but also representations of the level of human capital. The hedonic model of compensation suggests that workers with more human capital, associated with higher productivity, receive both higher wages and more employee benefits—such as employer-provided health insurance, job security, and flexible retirement plans—because employers are willing to pay more to attract and retain these workers to their businesses (Eriksson & Kristensen, 2014; Lazear & Shaw, 2007; Oh & Kleiner, 2024). In other words, higher wages and more employee benefits represent higher human capital.

Thus, the cost of leaving career jobs is higher for workers with higher wages and more employee benefits and therefore these workers are less likely to make career transitions (Robinson, 2018; Shaw, 1987). Therefore, I hypothesize that the workers with higher labor income and employee benefits are less likely to take on bridge employment or encore careers (Hypothesis 1).

Health as a Determinant of Later-Life Retirement Pathways

The role of physical and mental health on later-life work outcomes has been studied in multiple disciplines. Economists explain the effect of health on older workers’ retirement decisions through mathematical modeling by treating health as a stock that depreciates with age but can be restored by health stock investment (Grossman, 1972; Oh, 2023; Wolfe, 1985). The majority of economic studies reported that good health increases the likelihood of working longer and making later-life career transitions. For instance, Bound et al. (1999) show that both poor health at the time of the survey and the declining trend in health are strongly correlated with earlier workforce exit. Studies on later-life career transitions also show that poor health is associated with a lower likelihood of making career transitions (e.g., Cahill et al., 2006, 2018).

Although psychologists and sociologists provided robust empirical evidence on how good health increases the likelihood of working longer and making later-life career transitions (e.g., Damman et al., 2011; van Solinge & Henkens, 2005), Beehr (2014) points out that “none of the major theoretical approaches to retirement (such as continuity theory and role theory) focuses on health” (p. 1095). One pioneering theory that explains the role of health in retirement transitions is the resource-based dynamic perspective; this theory explains that older adults’ available resources before making career transitions, such as retirement savings and good health conditions, have direct effects on choosing career transitions instead of direct workforce exit by allowing the “ease of adjustment” (more apt to adapt to the new jobs) during the career transition process (Wang et al., 2011).

Across the disciplines, empirical studies consistently showed that workers with poor physical and mental health are more likely to exit the workforce earlier and less likely to take on career transitions. Thus, I hypothesize that workers with good health are more likely to take on bridge employment or encore careers than exit the workforce directly (Hypothesis 2). In particular, good health is associated with the longer work duration; therefore, I hypothesize that workers with good health are more likely to take on encore careers than bridge employment (Hypothesis 3).

Self-Identified Retirement Status as Identity in Career Transition

Previous studies investigating the predictors of retirement transitions often miss the role of self-identified retirement status (named “retirement identity”). Retirement identity helps in understanding older worker’s perception of retirement status that shapes the work and nonwork roles after transition, work attachment, job satisfaction, and level of self-esteem (Ekerdt & DeViney, 1990; Kim & Moen, 2002), which consequently influence the duration of work in the posttransition jobs. In other words, “not retired” as an identity when transitioning to new jobs may indicate the continuation of work roles and strong attachment to new jobs while “partly or completely retired” as identities may indicate the opposite (Szinovacz & DeViney, 1999), implying that retirement identity may influence the duration of work in the posttransition jobs. Thus, I hypothesize that the workers who identify themselves as “not retired” when making career transitions are more likely to choose encore careers than bridge employment (Hypothesis 4).

Method

Data

I draw on a sample of respondents from the Health and Retirement Study (HRS, 2023) using the RAND-HRS Longitudinal File 2020 (V1) from 1992 to 2020. The HRS is a biennial longitudinal survey that provides abundant information on the nationally representative sample of Americans aged 50 and older (Sonnega et al., 2014). For this study, I use the sample of respondents born between 1931 and 1953, grouped into three cohorts—the initial HRS participants recruited in 1992 (1992 cohort, born between 1931 and 1941), the so-called War Babies (1998 cohort, born between 1942 and 1947), and the Early Baby Boomers (2004 cohort, born between 1948 and 1953)—to control for the potential birth-cohort difference in the choice of retirement pathways. In the estimation, for each respondent, I select the wave that is one wave before making career transitions to capture the pretransition socioeconomic and health conditions such as weekly earnings and employer-provided insurance from career jobs, physical and health conditions before making career transitions, etc.

Because this study aims to examine the transitions from career jobs, it is important to sample the respondents who remain in their career jobs when entering the survey. To do so, I follow the definition of FTC jobs provided in the previous literature (Cahill et al., 2006, 2018; Quinn, 1999): 10+ years of tenure and 1,600+ annual work hours when they enter the survey. Accordingly, I excluded the respondents who do not maintain their FTC status when entering the survey (N = 1,224). I also excluded the respondents who returned to the workforce after leaving their career jobs for 2+ years, often called “unretirement” (N = 404), because they did not transition from “career jobs” but from “outside the workforce,” and therefore its processes and outcomes are possibly different from other retirement pathways (Cahill et al., 2011; Gonzales et al., 2017).

Operationalization of Various Retirement Pathways

Leaving career jobs is defined as moving from one’s FTC job to another jobs or workforce exit (N = 2,142). Career transition, a subset of leaving career jobs, is defined as moving from one’s FTC job to another job, operationalized as leaving one’s FTC employer to another employers or self-employed (or vice versa if the FTC job is self-employed). Other types of job changes, such as switching to different occupations and reducing work hours, are also considered as career transitions only if they involve leaving FTC employers. I impose this restriction because the HRS does not provide the reasons for other job changes but the reason why a respondent left the previous employer.

The voluntariness of career transition is one of the key elements that distinguish involuntary career transitions from bridge employment and encore careers. Any career transitions due to involuntary leaves (business closed, laid off, or poor health or disabled) and family reasons (family care, respondents or family moved, or handed over responsibilities to other family members) are defined as involuntary career transitions (N = 104). The sample of respondents who made involuntary career transition is excluded from the final sample because involuntary career transition is defined as a distinct retirement pathway from bridge employment and encore careers.

Among the workers who voluntarily took on career transitions, I measure the duration of work in the posttransition jobs to distinguish the choice of their retirement pathways. To examine this duration, I use three criteria: change in occupations, employers, and workforce exit within or after 10 years since the latest career transition. The description in Figure 1 provides detailed information about each criterion. Given these three criteria, 843 respondents are defined as taking on bridge employment. The respondents whose voluntary career transitions do not belong to any of the three criteria are considered to take on encore careers (n = 129). Because the HRS data are available until 2020, making career transitions after 2012 systematically does not allow workers to fulfill the “10+ years of working in the posttransition job” condition. Thus, the respondents who made career transitions after 2012 are excluded from the sample.

After excluding ineligible respondents—those who remain in the workforce without career transition until 2020, those who made career transitions after 2012, and those who take on involuntary career transitions—the sample includes 2,038 respondents: 1,082 men and 956 women. Table 1 provides the descriptive statistics of the study variables.

Table 1.

Characteristics of HRS Participants With Three Retirement Pathways

Variable Direct workforce exit Bridge employment Encore career
Mean % SD Mean % SD Mean % SD
Weekly earnings(t − 1) (US$) 737 1,013 528 1,380 685 1,019
Household wealth(t − 1) (US$) 117,529 260,476 86,684 276,469 111,985 482,988
Physical and mental health(t − 1)
 # of health conditions 1.22 1.11 1.05 1.05 0.81 0.88
 # of functional limitations 2.21 2.68 1.97 2.63 1.59 2.68
 CES-D score 1.23 1.64 1.23 1.65 0.81 1.07
Age of work adjustment(t) 61.47 4.31 60.31 5.06 59.78 4.83
Employer-provided insurance(t − 1) 77.02 42.09 58.01 49.38 52.71 50.12
Retirement identity(t) (1 = retired) N/A N/A 40.22 49.07 25.00 43.48
Receiving retirement benefits(t) 49.72 50.02 43.06 49.55 34.88 47.85
Receiving Medicare(t) 21.67 41.22 22.06 41.49 14.73 35.58
Gender (1 = women) 47.84 49.98 46.50 49.91 41.86 49.53
Birth cohort
 Initial HRS 66.89 47.08 64.41 47.91 62.02 48.72
 War Babies 17.26 37.81 18.74 39.05 16.28 37.06
 Early Baby Boomers 15.85 36.54 16.84 37.45 21.71 41.38
Non-White or Hispanic 25.52 43.62 24.44 43.00 23.26 42.41
Education
 Less than high school 21.58 41.15 15.90 36.59 13.18 33.96
 High school graduate/GED 48.22 49.99 50.30 50.03 38.76 48.91
 Some college and above 30.21 45.94 33.81 47.33 48.06 50.16
Spouse presence and work status(t − 1)
 Spouse present and working 42.87 49.51 47.21 49.95 55.81 49.85
 Spouse present and not working 29.83 45.77 22.78 41.96 19.38 39.68
 Spouse not present 27.30 44.57 30.01 45.86 24.81 43.36
Self-employed(t − 1) 6.66 24.95 13.17 33.83 13.95 34.79
Observations 1,066 843 129
Open in a new tab

Notes: CES-D = Center for Epidemiological Studies—Depression scale; GED = General Educational Development; HRS = Health and Retirement Study; SD = standard deviation. Weekly earnings, employer-provided insurance, physical and mental health, household wealth measures, and self-employed status are obtained from one wave before the career transitions, and the rest of measures are obtained from the wave of making career transitions. The subscripts—(t − 1) and (t)—indicate when the measures are drawn; (t − 1) indicates the time before making career transitions, and (t) indicates the time when making career transitions. The covariates without these subscripts are the time-invariant measures.

Predictors and Covariates of Career Transitions

I use three sets of main variables in the estimation procedures. First, I use log-transformed weekly earnings and employer-provided insurance from career jobs (before making career transitions) to examine Hypothesis 1: evaluating how human capital is associated with later-life career transitions. These measures are obtained from one wave before making career transitions. Second, I use a set of physical and mental health before making career transitions to examine Hypotheses 2 and 3: evaluating how health is associated with later-life career transitions. For physical and mental health measures, I use the number of health conditions, number of functional limitations, and the HRS adaptation of the Center for Epidemiological Studies—Depression (CES-D) score (Sonnega et al., 2014; Steffick et al., 2000); higher numeric values for these covariates indicate poor health conditions. Last, I use the self-identified retirement status when making career transitions to examine Hypothesis 4: evaluating the role of retirement identity in career transition. I recode the trinomial retirement identity variable into a binomial indicator: “not retired” as a reference, and “partly or completely retired” as a comparison variable. This measure is obtained from the wave when making career transitions.

In addition, I use three sets of covariates: time-invariant measures, measures before making career transitions, and measures at the time of making career transitions. The time-invariant measures include study cohorts, gender, race and ethnicity, and education. The measures before making career transitions include presence and work status of respondent’s spouse before career transitions, log-transformed net nonhousing household wealth, self-employed status in career jobs, occupations of FTC jobs categorized into 17 groups, and Census region of residency to control for the variations between sociodemographic and economic groups (Bugliari et al., 2023). The measures at the time of making career transitions include age of leaving career jobs (work adjustment) and the receipt of retirement-related benefits (Social Security and pension) and Medicare. Table 2 provides more information about these predictors and covariates.

Table 2.

Operationalization of Study Variables

Study variable Range List of measures
Receiving retirement benefits Yes or no Receiving at least one of the following benefits: Social Security (amount of Social Security benefit > $0) and/or pension (amount of pension benefit > $0)
# of health conditions 0–8 Sum of (health condition = yes). The health condition measures include: high blood pressure, diabetes, lung, heart condition, stroke, emotional/psychiatric problem, and arthritis.
# of functional limitations 0–9 Sum of (functional limitation = yes). The functional limitation measures include: walking several blocks, sitting 2 hr, getting up from chair, climbing stairs, stooping, kneeling, or crouching, lifting or carrying weights over 10 pounds, picking up a dime from a table, reaching or extending arms above shoulder level, and pulling or pushing large objects.
CES-D score 0–8 Sum of (mental health = yes/no). The mental health measures include:
Yes: felt depressed, felt activities were efforts, was sleep restless, felt lonely, felt sad, and could not get going
No: was happy and enjoyed life
Open in a new tab

Note: CES-D = Center for Epidemiological Studies—Depression scale.

Analysis Strategy

The estimation strategy is twofold. First, I examine the predictors distinguishing the two voluntary career transitions—bridge employment and encore career—from the direct workforce exit using multinomial logistic regression. Second, I examine the predictors distinguishing between the two voluntary career transitions using logistic regression. The binary retirement identity indicator is used in the second estimation only to evaluate the role of retirement identity on the choice between the two voluntary career transitions. I use STATA/MP 18 for the entire process, including data cleaning and analyses.

Results

Comparison of Voluntary Career Transitions With Direct Workforce Exit

Table 3 shows the estimated coefficients, standard errors, relative risk ratios, and 95% confidence intervals of the relative risk ratio from multinomial logistic regression. Using the “direct workforce exit” group as a reference, the first five columns show the results using “bridge employment” as a comparison group, and the next five columns show the results using “encore career” as a comparison group.

Table 3.

Choice of Retirement Pathways (Three Pathways), Multinomial Logistic Regression

Variable Bridge employment Encore career
EST SE RRR 95% CI EST SE RRR 95% CI
Weekly earnings(t − 1) −0.276*** 0.019 0.758 0.729 0.789 −0.234*** 0.035 0.791 0.738 0.848
Employer-provided insurance(t − 1) −0.797*** 0.122 0.451 0.355 0.573 −1.182*** 0.220 0.307 0.199 0.472
Physical and mental health(t − 1)
 # of health conditions −0.130* 0.055 0.878 0.787 0.979 −0.262* 0.115 0.769 0.614 0.963
 # of functional limitations −0.016 0.030 0.984 0.927 1.044 −0.147* 0.066 0.863 0.758 0.983
 CES-D score −0.001 0.034 0.999 0.935 1.068 −0.165* 0.080 0.848 0.725 0.992
Household wealth(t − 1) −1.656** 0.532 0.191 0.067 0.542 −1.053 0.916 0.349 0.058 2.099
Received retirement benefits(t) −0.276* 0.129 0.759 0.589 0.978 −0.482 0.262 0.617 0.369 1.032
Received Medicare(t) 0.106 0.199 1.111 0.753 1.640 −0.675 0.469 0.509 0.203 1.276
Gender (1 = women) −0.170 0.123 0.844 0.663 1.074 −0.279 0.232 0.757 0.480 1.193
Cohort (base = HRS core)
 War Baby 0.080 0.206 1.083 0.723 1.622 −0.130 0.426 0.878 0.381 2.024
 Early Baby Boomer −0.153 0.311 0.858 0.466 1.580 −0.254 0.612 0.775 0.234 2.574
Race/ethnicity (1 = non-White/Hispanic) −0.064 0.130 0.938 0.726 1.211 0.052 0.248 1.054 0.649 1.712
Education (reference = less than high school)
 High school graduate/GED 0.608*** 0.156 1.836 1.353 2.491 0.450 0.326 1.568 0.829 2.969
 Some college or higher 0.642*** 0.195 1.899 1.297 2.782 0.972** 0.375 2.643 1.266 5.516
Spouse presence and work status(t − 1) (reference = present and working)
 Present and not working −0.275* 0.134 0.760 0.585 0.988 −0.479 0.264 0.620 0.369 1.040
 Not present 0.291 0.221 1.338 0.868 2.063 −0.148 0.469 0.863 0.344 2.163
Age of leaving career jobs(t) (reference = 51–61)
 62–66 −0.939*** 0.154 0.391 0.289 0.529 −0.661* 0.305 0.516 0.284 0.939
 67+ −0.647* 0.307 0.524 0.287 0.955 0.033 0.649 1.033 0.289 3.691
Self-employed(t − 1) (1 = yes) 0.043 0.191 1.044 0.718 1.519 0.010 0.321 1.010 0.538 1.895
Open in a new tab

Notes: CES-D = Center for Epidemiological Studies—Depression scale; GED = General Educational Development; HRS = Health and Retirement Study. The above table presents the results from multinomial logistic regression: the estimated coefficients (EST), standard error (SE), relative risk ratio (RRR), and the 95% confidence interval of the relative risk ratio (95% CI). Direct workforce exit is set as a reference-dependent variable; bridge employment and encore career decisions are used as multinomial-dependent variables. The estimation controlled for career occupations, year of leaving career jobs, and region of residency. Weekly earnings and household wealth are log-transformed. The subscripts—(t − 1) and (t)—indicate when the measures are drawn; (t − 1) indicates the time before making career transitions, and (t) indicates the time when making career transitions. The covariates without these subscripts are the time-invariant measures. N = 2,038, log pseudo-likelihood = −1,511.424, likelihood ratio χ2 = 559.19, and pseudo-R2 = 0.1561.

***p < .001. **p < .01. *p < .05.

In the columns for bridge employment, the estimations for the covariates weekly earning (β = −0.28), employer-provided insurance (β = −0.80), number of health conditions (β = −0.13), household wealth (β = −1.66), receiving retirement benefits (β = −0.28), high school graduate or General Educational Development (β = 0.61), some college or higher (β = 0.64), spouse present and not working (β = −0.28), and leaving career jobs at ages 62–66 and ages 67+ (β = −0.94 and −0.65, respectively) are statistically significant. These results suggest that workers are more likely to take on bridge employment than direct workforce exit if they received lower labor income and no employer-provided insurance from career jobs, had less number of health conditions, accumulated less amount of household wealth, did not receive any retirement-related benefits at the time of leaving career jobs, more educated, had a working spouse, and left career jobs at younger ages.

In the columns for encore careers, similarly, the estimations for the covariates weekly earning (β = −0.23), employer-provided insurance (β = −1.18), number of health conditions (β = −0.26), number of functional limitations (β = −0.15), and the CES-D score (β = −0.17), some college or higher education (β = −0.97), and leaving career jobs at ages 62–66 (β = −0.66) are statistically significant. These results suggest that workers are more likely to take on encore careers than direct workforce exit if they received lower labor income and no employer-provided insurance from career jobs, had good physical and mental health conditions, more educated, had working spouse, and left career jobs at younger ages.

These results are consistent with Hypotheses 1 and 2. As hypothesized, workers receiving higher labor income and employee benefits (employer-provided insurance) from their career jobs are less likely to take on bridge employment or encore careers but directly exit the workforce. In particular, the magnitudes of estimation for employer-provided insurance are particularly large for both comparison groups, suggesting that employee benefits are one of the key drivers that limit older workers from making career transitions.

The other estimated coefficients that are statistically significant are also consistent with the findings from previous literature. For instance, the negative and statistically significant estimations for household wealth and receipt of retirement benefits for bridge employment suggest that older workers with less retirement resources take on bridge employment to accumulate these resources during their extended work lives (Bishop, 2022; Cahill et al., 2012). Also, the negative and statistically significant estimations for the spouse’s presence and work status in both career transitions provide evidence of the couple’s joint retirement patterns: spouse’s work status influences the choice of retirement pathways (Pienta & Hayward, 2002; Yorgason et al., 2020). In particular, the magnitude of estimations for education level and age of leaving career jobs are particularly large, suggesting that higher education is one of the key drivers of making career transitions in later life and the timing of making career transitions is also important.

Predictors Distinguishing Between Bridge Employment and Encore Careers

Table 4 shows the estimated coefficients, standard errors, odds ratios, and 95% confidence intervals of the odds ratios from logistic regression using the subsample of workers who take on bridge employment or encore careers. The estimated coefficients for the number of functional limitations (β = −0.16) and the CES-D score (β = −0.18) are negative and statistically significant, suggesting that workers with good physical and mental health are more likely to pursue encore careers than bridge employment. This is consistent with the prediction in Hypothesis 3; the positive association between good health and longer duration of work suggests that the workers with good health are more likely to choose encore careers than bridge employment. Furthermore, the estimated coefficient for retirement identity (β = −0.71) is negative and statistically significant, suggesting that workers who identify themselves as “partly or completely retired” when making career transitions are less likely to choose encore careers, consistent with the prediction in Hypothesis 4.

Table 4.

Choice of Retirement Pathway—Bridge Employment or Encore Careers

Variable EST SE OR 95% CI
Retirement identity(t) (1 = retired) −0.711* 0.298 0.491 0.274 0.882
Weekly earnings(t − 1) 0.028 0.039 1.028 0.952 1.111
Employer-provided insurance(t − 1) −0.472* 0.223 0.624 0.402 0.966
Physical and mental health(t − 1)
 # of health conditions −0.063 0.118 0.939 0.746 1.183
 # of functional limitations −0.155* 0.074 0.857 0.741 0.991
 CES-D score −0.176* 0.083 0.839 0.713 0.987
Household wealth(t − 1) 1.039 0.937 2.827 0.451 17.72
Received retirement benefits(t) 0.044 0.323 1.045 0.555 1.968
Received Medicare(t) −1.052* 0.531 0.349 0.123 0.988
Gender (1 = women) −0.232 0.247 0.793 0.488 1.288
Cohort (base = HRS core)
 War Baby −0.217 0.454 0.805 0.331 1.960
 Early Baby Boomer −0.444 0.664 0.642 0.175 2.356
Race/ethnicity (1 = non-White/Hispanic) 0.207 0.256 1.229 0.744 2.031
Education (reference = less than high school)
 High school graduate/GED −0.074 0.343 0.929 0.474 1.819
 Some college or higher 0.329 0.395 1.389 0.641 3.011
Spouse presence and work status(t − 1) (reference = present and working)
 Present and not working −0.468 0.295 0.626 0.351 1.117
 Not present −0.451 0.474 0.637 0.252 1.613
Age of leaving career jobs(t) (reference = 51–61)
 62–66 0.602 0.345 1.827 0.930 3.589
 67+ 1.003 0.698 2.727 0.695 1.70
Self-employed(t − 1) (1 = yes) −0.141 0.314 0.868 0.469 1.605
Open in a new tab

Notes: CES-D = Center for Epidemiological Studies—Depression scale; GED = General Educational Development; HRS = Health and Retirement Study. The above table presents the results from logistic regression using bridge employment as a reference-dependent variable: the estimated coefficient (EST), standard error (SE), odds ratio (OR), and the 95% confidence interval of the odds ratio (95% CI). Forty-five respondents dropped from the estimation due to the missing retirement identities. The estimation controlled for career occupations, year of leaving career jobs, and region of residency. Weekly earnings and household wealth are log-transformed. The subscripts—(t − 1) and (t)—indicate where the measures are drawn from; (t − 1) indicates the time before making career transitions, and (t) indicates the time when making career transitions. The covariates without these subscripts are the time-invariant measures. n = 927, log pseudo-likelihood = −331.904, likelihood ratio χ2 = 65.71, and pseudo-R2 = 0.090.

***p < .001. **p < .01. *p < .05.

Note that the estimated coefficient for employer-provided insurance (β = −0.47) is negative and statistically significant, suggesting that workers who received employee benefits from their career jobs are less likely to take on encore careers. Considering the difference in the magnitude of the estimated coefficients for this variable in Table 3, the receipt of employer-provided insurance from career jobs may indicate the higher level of accumulated human capital from one’s career job that discourages pursuing a new career. However, further investigation is needed regarding the association between employee benefits and the choice of bridge employment and encore careers.

Discussion

In this study, I suggest two criteria to distinguish the retirement pathways based on how “later-life career” is defined in the previous retirement literature. Then, based on the theories from multiple disciplines, I hypothesize and examine a set of variables to predict the decisions of choosing these retirement pathways. The results in the previous section are consistent with the theoretical predictions. First, the results in Table 3 show that the higher cost of voluntary career transition due to the loss of human capital, represented by higher labor income and more employee benefits, discourages workers from making later-life career transitions. Second, the results in Tables 3 and 4 show that poor physical and mental health is negatively associated with making later-life career transitions and the duration of work in the posttransition jobs, which workers with poor health are more likely to choose bridge employment than encore careers. Last, the results in Table 4 show that how workers self-identify their retirement status when making career transitions predicts the duration of work in the posttransition jobs that allow distinguishing between bridge employment and encore careers.

Maintaining the labor force participation of older workers is an important agenda for policymakers. As previous studies suggest, promoting bridge employment and encore careers helps maintaining older workers’ labor force participation (e.g., Cahill et al., 2012; Freedman, 2006). This study suggests that increasing the number of quality jobs for older workers will contribute to maintain the level of older workers’ labor force participation by promoting bridge employment and encore careers.

One of the implications of this study is that workers with more human capital are less likely to make career transitions because the cost (human capital) of making career transitions is greater than the benefits (social and psychological benefits). Increasing the benefit of making career transitions reduces the gap between cost and benefit, which can be done by increasing the number of quality jobs available for older workers. The improvement in job quality for older workers would involve higher wages, more employee benefits, better work conditions, and more social and psychological benefits such as a sense of belonging, contribution to communities, etc. (Freedman, 2006; Kim & Feldman, 2000; Oh, 2023; Pengcharoen & Shultz, 2010). Improvement in job quality does not only increase the listed benefits but also contributes to improving older workers’ health (Berkman & Truesdale, 2023). Because the findings of this study also suggest that good physical and mental health is associated with making career transitions and a longer duration of work in the posttransition jobs, increasing the number of quality jobs also contributes to improving older workers’ health and therefore further prolongs the duration of their labor force participation.

Although this study thoroughly examined the influences of various sociodemographic and economic indicators with strong theoretical foundations, there are several limitations. First, although the generally agreed definition of bridge employment includes various types of job changes, this study only includes the changes in employers because the indicators to identify the voluntariness of career transition are available only for those who left their career employers. Second, although one of the goals of this study is to provide objective guidelines in distinguishing various retirement pathways, the use of a 10-year threshold and voluntary career transitions as distinction criteria of retirement pathways limit the subjective identification of each pathway. For instance, working 10+ years in the posttransition jobs may be challenging for those making career transitions after age 60; rather, workers aged 60 and older may identify their career transitions as encore careers even if they plan to work less than 10 years. Similarly, those who changed their jobs after leaving career jobs involuntarily may consider their new job as an encore career. As Freedman (2006) suggests, the purpose of making career transitions, which is rather subjective, is an important part of the extended work lives after the transitions.

Therefore, future research should extend these findings to evaluate multiple criteria, including subjective identification, for distinguishing various retirement pathways. Both qualitative and quantitative research can help establish more concrete and quantitatively testable definition of encore careers so that the differences in their characteristics from bridge employment can be used as determinants distinguishing these two retirement pathways. Future research should also include various types of work adjustment, including switching occupations and reducing work hours, as career transitions and investigate the similarities and differences in the sociodemographic and economic characteristics among those who take on encore careers via different work adjustments. Last, fruitful would be the investigation of the short- and long-term outcomes of involuntary career transitions and unretirement in relation to financial, psychosocial, and health gains and losses using the HRS or other longitudinal data sets to gain a deeper understanding on the predictors of their heterogeneous outcomes.

Acknowledgments

The Health and Retirement Study (HRS) is supported by the National Institute on Aging (NIA U01AG009740). I really appreciate Jacqui Smith, Marina Larkina, and the members of the HRS Psychosocial Aging Work Group for constructive feedback. I also appreciate Phyllis Moen for mentorship and inspiration in starting and completing this research.

Contributor Information

Yun taek Oh, Institute for Social Research, University of Michigan, Ann Arbor, Michigan, USA.

Markus Schafer, (Social Sciences Section).

Funding

Y. Oh is supported by the National Institute on Aging (R01 AG051142: PI: J. Smith).

Conflict of Interest

None.

References

  1. Bank, D., Emerman, J., & Halvorsen, C. (2011). Encore career choices: Purpose, passion, and a paycheck in a tough economy [full report]. https://encore.org/wp-content/uploads/files/EncoreCareerChoices.pdf [Google Scholar]
  2. Beehr, T. A. (2014). To retire or not to retire: That is not the question. Journal of Organizational Behavior, 35(8), 1093–1108. 10.1002/job.1965 [DOI] [Google Scholar]
  3. Berkman, L. F., & Truesdale, B. C. (2023). Working longer and population aging in the U.S.: Why delayed retirement isn’t a practical solution for many. The Journal of the Economics of Ageing, 24, 100438. 10.1016/j.jeoa.2022.100438 [DOI] [Google Scholar]
  4. Bishop, C. E. (2022). Economics of aging: New insights. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 77(4), 735–738. 10.1093/geronb/gbac047 [DOI] [PubMed] [Google Scholar]
  5. Bound, J., Schoenbaum, M., Stinebrickner, T. R., & Waidmann, T. (1999). The dynamic effects of health on the labor force transitions of older workers. Labour Economics, 6(2), 179–202. 10.1016/s0927-5371(99)00015-9 [DOI] [Google Scholar]
  6. Boveda, I., & Metz, A. J. (2016). Predicting end-of-career transitions for Baby Boomers nearing retirement age. The Career Development Quarterly, 64(2), 153–168. 10.1002/cdq.12048 [DOI] [Google Scholar]
  7. Bugliari, D., Carroll, J., Hayden, O., Hayes, J., Hurd, M., Lee, S., Main, R., McCullough, C., Meijer, E., Pantoja, P., & Rohwedder, S. (2023). RAND HRS longitudinal file 2020 (V1) documentation. RAND Center for the Study of Aging. [Google Scholar]
  8. Cahill, K. E., Giandrea, M. D., & Quinn, J. F. (2006). Retirement patterns from career employment. Gerontologist, 46(4), 514–523. 10.1093/geront/46.4.514 [DOI] [PubMed] [Google Scholar]
  9. Cahill, K. E., Giandrea, M. D., & Quinn, J. F. (2011). Reentering the labor force after retirement. Monthly Labor Review, 134, 34. https://www.bls.gov/opub/mlr/2011/article/reentering-the-labor-force-after-retirement.htm [Google Scholar]
  10. Cahill, K. E., Giandrea, M. D., & Quinn, J. F. (2012). Bridge employment. In Wang M. (Ed.), The Oxford handbook of retirement. Oxford University Press. 10.1093/oxfordhb/9780199746521.013.0111 [DOI] [Google Scholar]
  11. Cahill, K. E., Giandrea, M. D., & Quinn, J. F. (2018). Is bridge job activity overstated? Work, Aging and Retirement, 4(4), 330–351. 10.1093/workar/way006 [DOI] [Google Scholar]
  12. Damman, M., Henkens, K., & Kalmijn, M. (2011). The impact of midlife educational, work, health, and family experiences on men’s early retirement. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 66(5), 617–627. 10.1093/geronb/gbr092 [DOI] [PubMed] [Google Scholar]
  13. Dingemans, E., & Henkens, K. (2014). Involuntary retirement, bridge employment, and satisfaction with life: A longitudinal investigation. Journal of Organizational Behavior, 35(4), 575–591. 10.1002/job.1914 [DOI] [Google Scholar]
  14. Ekerdt, D. J. (2010). Frontiers of research on work and retirement. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 65(1), 69–80. 10.1093/geronb/gbp109 [DOI] [PubMed] [Google Scholar]
  15. Ekerdt, D. J., & DeViney, S. (1990). On defining persons as retired. Journal of Aging Studies, 4(3), 211–229. 10.1016/0890-4065(90)90023-2 [DOI] [Google Scholar]
  16. Eriksson, T., & Kristensen, N. (2014). Wages or fringes? Some evidence on trade-offs and sorting. Journal of Labor Economics, 32(4), 899–928. 10.1086/676662 [DOI] [Google Scholar]
  17. Feldman, D. C. (1994). The decision to retire early: A review and conceptualization. The Academy of Management Review, 19(2), 285–311. 10.5465/amr.1994.9410210751 [DOI] [Google Scholar]
  18. Freedman, M. (2006). The social-purpose encore career: Baby boomers, civic engagement, and the next stage of work. Generations, 30(4), 43–46. https://www.ingentaconnect.com/content/asag/gen/2006/00000030/00000004/art00008 [Google Scholar]
  19. Gibbons, R., & Waldman, M. (2004). Task-specific human capital. American Economic Review, 94(2), 203–207. 10.1257/0002828041301579 [DOI] [Google Scholar]
  20. Gonzales, E., Lee, Y., & Brown, C. (2017). Back to work? Not everyone. Examining the longitudinal relationships between informal caregiving and paid work after formal retirement. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 72(3), 532–539. 10.1093/geronb/gbv095 [DOI] [PubMed] [Google Scholar]
  21. Grossman, M. (1972). On the concept of health capital and the demand for health. Journal of Political Economy, 80(2), 223–255. 10.1086/259880 [DOI] [Google Scholar]
  22. Gustman, A. L., & Steinmeier, T. L. (1991). The effects of pensions and retirement policies on retirement in higher education. The American Economic Review, 81(2), 111–115. https://www.jstor.org/stable/2006836 [Google Scholar]
  23. Health and Retirement Study. (2023). RAND HRS longitudinal file 2020 (V1) public use dataset. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740). [Google Scholar]
  24. Hershey, D. A., & Henkens, K. (2014). Impact of different types of retirement transitions on perceived satisfaction with life. Gerontologist, 54(2), 232–244. 10.1093/geront/gnt006 [DOI] [PubMed] [Google Scholar]
  25. Kim, J. E., & Moen, P. (2002). Retirement transitions, gender, and psychological well-being: A life-course, ecological model. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 57(3), 212–222. 10.1093/geronb/57.3.p212 [DOI] [PubMed] [Google Scholar]
  26. Kim, S., & Feldman, D. C. (2000). Working in retirement: The antecedents of bridge employment and its consequences for quality of life in retirement. The Academy of Management Journal, 43(6), 1195–1210. https://www.jstor.org/stable/1556345 [Google Scholar]
  27. Kojola, E., & Moen, P. (2016). No more lock-step retirement: Boomers’ shifting meanings of work and retirement. Journal of Aging Studies, 36, 59–70. 10.1016/j.jaging.2015.12.003 [DOI] [PubMed] [Google Scholar]
  28. Lawson, K. (1991). What happened to normal retirement age? Employee Benefit Plan Review, 46, 12–15. [Google Scholar]
  29. Lazear, E., & Shaw, K. (2007). Personnel economics: The economist’s view of human resources. Journal of Economic Perspectives, 21(4), 91–114. 10.1257/jep.21.4.91 [DOI] [Google Scholar]
  30. Loewenstein, M. A., & Spletzer, J. R. (1999). General and specific training: Evidence and implications. The Journal of Human Resources, 34(4), 710–733. 10.2307/146414 [DOI] [Google Scholar]
  31. Murray, J. (1979). Subjective retirement. Social Security Bulletin, 42(11), 20–24. https://pubmed.ncbi.nlm.nih.gov/531727/ [PubMed] [Google Scholar]
  32. Oh, Y. (2023). Three essays on bridge jobs of American midlife workers [doctoral dissertation]. University of Minnesota. [Google Scholar]
  33. Oh, Y., & Kleiner, M. M. (2024). The influence of occupational licensing on workforce transitions to retirement. (NBER Working Paper No. 32292). National Bureau of Economic Research. 10.3386/w32292 [DOI] [Google Scholar]
  34. Pengcharoen, C., & Shultz, K. S. (2010). The influences on bridge employment decisions. International Journal of Manpower, 31(3), 322–336. 10.1108/01437721011050602 [DOI] [Google Scholar]
  35. Pienta, A. M., & Hayward, M. D. (2002). Who expects to continue working after age 62? The retirement plans of couples. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 57(4), S199–S208. 10.1093/geronb/57.4.s199 [DOI] [PubMed] [Google Scholar]
  36. Quinn, J. F. (1999). Retirement patterns and bridge jobs in the 1990s. EBRI Issue Brief, 206, 1–22. [PubMed] [Google Scholar]
  37. Quinn, J. F. (2010). Work, retirement, and the encore career: Elders and the future of the American workforce. Generations, 34(3), 45–55. https://www.ingentaconnect.com/content/asag/gen/2010/00000034/00000003/art00011 [Google Scholar]
  38. Robinson, C. J. (2018). Occupational mobility, occupation distance, and specific human capital. Journal of Human Resources, 53(2), 513–551. 10.3368/jhr.53.2.0814.6556r2 [DOI] [Google Scholar]
  39. Shaw, K. (1987). Occupational change, employer change, and the transferability of skills. Southern Economic Journal, 53(3), 702–719. 10.2307/1058765 [DOI] [Google Scholar]
  40. Sonnega, A., Faul, J. D., Ofstedal, M. B., Langa, K. M., Phillips, J. W., & Weir, D. R. (2014). Cohort profile: The Health and Retirement Study (HRS). International Journal of Epidemiology, 43(2), 576–585. 10.1093/ije/dyu067 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Steffick, D. E., Wallace, R. B., Herzog, A. R., Ofstedal, M. B., Steffick, D., Fonda, S., & Langa, K. M. (2000). Documentation of affective functioning measures in the Health and Retirement Study. Survey Research Center, Institute for Social Research, University of Michigan. [Google Scholar]
  42. Szinovacz, M. E., & DeViney, S. (1999). The retiree identity: Gender and race differences. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 54(4), S207–S218. 10.1093/geronb/54b.4.s207 [DOI] [PubMed] [Google Scholar]
  43. Van Solinge, H., & Henkens, K. (2005). Couples’ adjustment to retirement: A multi-actor panel study. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 60(1), S11–S20. 10.1093/geronb/60.1.s11 [DOI] [PubMed] [Google Scholar]
  44. Van Solinge, H., & Henkens, K. (2007). Involuntary retirement: The role of restrictive circumstances, timing, and social embeddedness. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 62(5), S295–S303. 10.1093/geronb/62.5.s295 [DOI] [PubMed] [Google Scholar]
  45. Wang, M., Henkens, K., & van Solinge, H. (2011). Retirement adjustment: A review of theoretical and empirical advancements. American Psychologist, 66(3), 204–213. 10.1037/a0022414 [DOI] [PubMed] [Google Scholar]
  46. Wolfe, J. R. (1985). A model of declining health and retirement. Journal of Political Economy, 93(6), 1258–1267. 10.1086/261359 [DOI] [Google Scholar]
  47. Yorgason, J. B., Segel-Karpas, D., Wheeler, B. E., Malig Canlas, J., & Smith, J. (2020). Change in retirement plans among midlife couples during an economic recession. The Journals of gerontology, Series B: Psychological Sciences and Social Sciences, 75(4), 827–836. 10.1093/geronb/gbz076 [DOI] [PubMed] [Google Scholar]

Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press

RESOURCES