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
. 2017 Jul 8;74(7):1200–1210. doi: 10.1093/geronb/gbx096

Trajectories of Work Disability and Economic Insecurity Approaching Retirement

Kim M Shuey 1,✉, Andrea E Willson 1
Editor: Philippa Clarke
PMCID: PMC6748769  PMID: 28977512

Abstract

Objectives

In this article, we examine the connection between trajectories of work disability and economic precarity in late midlife. We conceptualize work disability as a possible mechanism linking early and later life economic disadvantage.

Methods

We model trajectories of work disability characterized by timing and stability for a cohort of Baby Boomers (22–32 in 1981) using 32 years of data from the Panel Study of Income Dynamics and latent class analysis. Measures of childhood disadvantage are included as predictors of work disability trajectories, which are subsequently included in logistic regression models predicting four economic outcomes (poverty, asset poverty, home ownership, and pension ownership) at ages 54–64.

Results

Childhood disadvantage selected individuals into five distinct classes of work disability that differed in timing and stability. All of the disability trajectories were associated with an increased risk of economic insecurity in late midlife compared to the never work disabled.

Discussion

This study contributes to the aging literature through its incorporation of the early life origins of pathways of disability and their links to economic outcomes approaching retirement. Findings suggest work disability is anchored in early life disadvantage and is associated with economic insecurity later in life.

Keywords: Disability trajectories, Economic inequality, Early-life disadvantage, Health disparities

Despite the protections offered by civil rights and employment legislation and the efforts of disability activists over the past few decades, research has demonstrated that persons living with a disability often face labor market disadvantages including lower rates of labor force participation, higher rates of unemployment and involuntary part-time work, greater job insecurity, and lower levels of pay and benefits (Baldwin & Schumacher, 2002; Schur, Kruse, Blasi, & Blanck, 2009). Depending on the timing of disability onset, barriers to educational attainment, discrimination in hiring, a lack of workplace accommodation, and discriminatory attitudes of employers present obstacles to career advancement and job retention (Schur, 2003; Schur et al., 2009). Labor market disadvantages have contributed to a poverty rate two times higher for working-age people with disabilities than among those without (Erickson, Lee, & von Schrader, 2015).

One complicating factor in disability research is that disability encompasses a heterogenous group of people who experience various types of physical and mental health conditions, including chronic disease, functional limitations, and intellectual disabilities, at various points in the life course (Burkhauser, Daly, Houtenville, & Nargis, 2002; Verbrugge & Jette, 1994). A social model of disability highlights that impairments become disabling within the context of social roles and institutions, such as in a labor market lacking adequate workplace supports, and these processes vary over the life course and according to social structural location (Nagi, 1991; Priestley, 2005; Verbrugge & Jette, 1994). This model of disability points to the socioeconomic consequences, such as material disadvantage and poverty, resulting from the limits placed by the social environment on an individual’s ability to engage in life activities, such as work (Kelley-Moore, 2010). This conceptual orientation can be contrasted to a medical model of disability, which has dominated research and tends to focus on specific impairments or conditions and their progression, as well as on the process of successful aging (Brandt & Pope, 1997; Oliver, 1996). In addition, although scholars agree that disability is a process that unfolds over time, which for some begins in early life, most studies examine the onset and progression of disability in late life rather than as a part of a longer-term process of social stratification and inequality. The focus on the prevalence of disability or disability trajectories and functional impairments in late life results in a “siloing of age-specific disability research,” in part because there are few longitudinal, nationally representative data sources that capture early and midlife (Verbrugge, 2016:1141).

Also relevant to understanding the consequences of disability are theories on the accumulative nature of disadvantage across the life course (Dannefer, 2003; Ferraro & Shippee, 2009; O’Rand, 1996), which suggest that disability during the working years is both an outcome of earlier socioeconomic disadvantage and a source of future disadvantage that continues into old age. Labor market participation shapes wealth accumulation, home ownership, and access to employer pensions, Social Security, and other benefits, and late middle age represents peak earning years for many workers which sets the stage for either economic security or precarity that carries forward into later life. Although the labor market inequalities experienced by persons with disabilities remain a major source of concern among social policy makers, work scholars, and disability activists, currently there is little research that examines the long-term pathways of work disability across individual lives, their roots in early life disadvantage, and their relationship with economic well-being as individuals approach later life. To address these gaps, in this analysis we use data from the Panel Study of Income Dynamics (PSID) and latent class analysis to construct trajectories of work disability that differ in timing and pathway, examine their early life anchors, as well as their relationship with multiple indicators of economic precariousness for those approaching traditional retirement age.

Trajectories of Disability

Within the social sciences, disability is conceptualized as a dynamic life course process that unfolds over time and is often measured as a long-term trajectory (Verbrugge & Jette, 1994). This conceptualization has gained prominence in recent decades partly as a response to the challenges faced by early attempts to understand health inequality using static point-in-time estimates or short-term longitudinal data capturing only a small portion of the life course. To this end, studies of disability have addressed a variety of questions ranging from cohort differences in trajectories of disability and their associated disease profiles to disability progression (Clarke & Latham, 2014; Nusselder, Looman, & Mackenbach, 2006; Pais, 2014; Taylor & Lynch, 2004; Taylor & Lynch, 2011; Wolf, Freedman, Ondrich, Seplaki, & Spillman, 2015). Most examine the experiences of older adults using data such as the Health and Retirement Study or regional studies of aging and measures of functional limitations and instrumental activities of daily living. Some examples include studies using growth curve models to examine trajectories of functional limitations (Liang et al., 2008; Warner & Brown, 2011) as well as disability onset and progression and potential mediating mechanisms, such as education and financial capital (Taylor, 2010; Taylor, 2011). Other studies use similar methodologies to group respondents into disability trajectories based on indicators such as activities of daily living and chronic conditions (Deeg, 2005; Liang et al., 2009; Lynch, 2015; Nusselder et al., 2006; Verbrugge, Latham, & Clarke, 2017; Wolf et al., 2015). Methodological variations include survival analysis of disability onset and microsimulations of active life expectancy in midlife and beyond (Laditka & Laditka, 2015; Latham, 2012), latent class analyses of cohort variations in the long-term experiences of disablement (Taylor & Lynch, 2011) and race differences in trajectories of work-limiting health impairments (Pais, 2014).

There are fewer examples of longitudinal studies that examine the relationship between disability trajectories and subsequent outcomes. To this end, research using the PSID found significant declines in earnings and hours worked following disability onset that persisted a decade later (Mok, Meyer, Charles, & Achen, 2008) and that people aging with disabilities (experiencing a work-limiting condition in four waves) are disadvantaged in terms of employment status, hours worked, and household income (Clarke & Latham, 2014). Missing from this small body of work are two conceptually and methodologically important issues. First, the timing of life course transitions is important because it affects both their meaning and their potential consequences, as the same event can affect individuals differently depending on when it occurs (Elder, Johnson, & Crosnoe, 2003; George, 1993). Timing reflects the social context, comprised of age-graded social roles and their intersection with disabling social institutions and culture, which individuals with disabilities are required to navigate (Putnam, 2002; Shuey, Willson, & Bouchard, 2016). Disability onset occurs earlier in life for some groups than for others as part of a long-term pathway that reflects the accumulation of disadvantage. Within sociology, this accumulative process is elaborated by theories of Cumulative Inequality (CIT) and cumulative advantage/disadvantage (CAD), which provide key frameworks for understanding socioeconomic status and mobility across the life course at the individual level and growing inequality with age within cohorts at the population level (Dannefer, 2003; Ferraro & Shippee, 2009). CIT/CAD describe an accumulative process whereby early-life advantage provides opportunities to access additional resources as well as to avoid many sources of adversity, while early disadvantage generates exposure to risk and further losses (O’Rand, 2006). For example, the life course of individuals with early onset disability is influenced by the treatment of disability within educational institutions, which impacts attainment, labor market entry, and future wages. The systematic barriers experienced as a result of disability onset earlier in life are potentially different than for individuals encountering disability toward the end of their work lives. In recognition of these differences, aging scholars sometimes draw a distinction between the experience of aging with disability and aging into disability, although this dichotomy has been criticized as limited in its ability to capture timing-related differences in individual experience within institutional contexts such as the labor market (Verbrugge & Yang, 2002; Verbrugge et al., 2017).

Second, endogenous selection and the relationship between disability and other systems of social stratification that begin early in life are important aspects of the accumulative relationship between work disability and later life outcomes (for discussion, see Pais, 2014). Early-life disadvantage disproportionately places individuals on health trajectories that affect labor force participation and the accumulation of resources, which in turn affects future health in a reciprocal manner across the life course (Haas, 2006; Haas, Glymour, & Berkman, 2011; Warren, 2009). Work disability serves as both an outcome and a mechanism of inequality that is embedded in broader systems of stratification whereby childhood socioeconomic disadvantage may initiate a pathway of lower educational attainment, employment in jobs that negatively impact health and provide lower wages and disadvantageous living conditions, and result in the earlier onset of work-limiting health conditions. People are selected into work disability as a result of structures of inequality associated with social location, and work disability is in turn related to the stratification of later life outcomes. An exclusive focus on disability as an end of life outcome results in processual truncation that separates disability from the larger systems of social inequality of which it is both a cause and effect.

To gain an understanding of disability and its relationship with inequality across the life course into old age requires that we begin our inquiry with the early-life socioeconomic anchors of disability and the extent to which pathways of work-limiting disability have long-term implications for economic well-being in later life. We conceptualize work disability as a mechanism of stratification which operates within the labor market to connect early life disadvantage with later-life outcomes through a “chain of risk” whereby early exposures to disadvantage increase the likelihood of further exposures that have negative consequences for economic security in late midlife and beyond (e.g., Haas et al., 2011; Kuh & Ben-Shlomo, 2004; O’Rand & Hamil-Luker, 2005). In this study, we do not attempt to simultaneously model the relationship between disability and socioeconomic conditions across the life course, or other co-occurring adult processes, such as family formation and dissolution. Rather, we examine the connection between work disability as an outcome of selection processes that begin in early life and its relationship with economic precarity in late midlife. In doing so, we extend the literature in three ways. We first determine whether clear pathways of work disability (characterized by timing and stability or change in work disability status from early adulthood to late midlife) are discernable across a 32-year period beginning in young adulthood (age 22–32) to late midlife (54–64). Second, we model the process of endogenous selection that begins with childhood disadvantage and influences an individual’s pathway of work disability. Third, we examine the relationship between work disability trajectories and four key economic indicators measured in late midlife, as individuals approach traditional retirement age (54–64). The four commonly used economic indicators include poverty, asset poverty (a family’s lack of a safety net against the short-term loss of income, indicating a vulnerable economic position [Haveman & Wolff, 2004; Oliver & Shapiro, 1997]), pension ownership (occupational pensions are an indicator of stable employment in a job with benefits and an important source of income security in old age [O’Rand, 2006]), and home ownership (the single largest asset of most Americans which can be liquidated for income or other needs and which represents a “fundamental dividing line in the well-being of older adults” [Fisher, Johnson, Marchand, Smeeding, & Torrey, 2007:S126; Robert & House, 1996]). All four measures of economic insecurity in midlife indicate a risk of inadequate accumulation of resources to meet economic needs in later life.

Data and Methods

Data

The PSID began as an annual survey in 1968 with a representative sample of approximately 4,800 American households and information on almost 18,000 individuals residing in those households (PSID, 2016). The PSID became biennial in 1997 and the latest available survey wave at the time of this analysis was 2013. Its design maintains representativeness of the nonimmigrant U.S. population, as children and adults who split-off from original PSID households are followed in their new household (Fitzgerald, 2011). This study examines patterns of work disability from 1981 (the first year that work disability questions were consistently included for women) to 2013 for Baby Boom respondents aged 22–32 in 1981 (N = 2,063). We then further limit the sample to respondents who participated in at least three survey waves from 1981 to 2013 and have a valid observation on the dependent variables, observed in 2013 when respondents were 54–64 (N = 1,738).

Although the PSID sample has declined over time, multiple studies have monitored the effects of attrition and analyses comparing the PSID with a nationally representative repeated cross-sectional survey show that the weighted sample maintains its representativeness on many characteristics (Fitzgerald, 2011). From our investigations of the potential effect of attrition in this analysis, we conclude that our findings are likely conservative estimates of the association of disability pathways and economic security (see Supplementary Material Section 1.1 for details).

Measures

Economic insecurity

Four dependent variables, poverty status, asset poverty, pension ownership, and home ownership are measured in 2013. The official U.S. poverty measure, one of the most closely scrutinized and widely cited indicators used by policymakers and researchers, compares the annual income of a family to a minimum annual income standard (Meyer & Sullivan, 2012). Respondents were considered to be living in poverty if the family’s total household income fell below 125% of the official U.S. Census needs standard (threshold values based on family size, the number of persons under age 18, and age of the householder). We use 125% of the U.S. poverty threshold because the PSID consistently finds higher reported incomes than the Census Bureau (Bane & Ellwood, 1983; Wagmiller, Lennon, Kuang, Alberti, & Aber, 2006). The concept asset poverty compliments poverty indicators by assessing a family’s asset holdings as a safety net to meet basic needs when income is not available for a period of time (Oliver & Shapiro, 1997). Asset poverty measures whether households have accrued enough asset value to allow them to remain above the poverty line for 3 months without a stream of income (Haveman & Wolff, 2004; Rank, 2009). Using the PSID-constructed financial wealth variable, which excludes housing equity, we coded asset poverty = 1 if financial wealth is less than the household’s annual Census needs standard divided by four. Pension ownership = 1 if respondents reported a pension through their current or previous job. Homeownership = 1 if respondents owned or were buying their home. Further detail on measurement of these variables is provided in the Supplementary Material Section 1.2.

Work disability

Respondents were coded “1” if they answered yes to the question, “Do you have any physical or nervous condition that limits the type of work or the amount of work you can do?” Potential work-limiting impairments include chronic conditions, short-term physical injuries, long-term disabilities, mental health issues, and any other health-related issue that prevents someone from working to the extent he or she would have without the condition. Patterns of responses over 12 waves between 1981 and 2013 were evaluated using latent class analysis, described in detail below.

Early life conditions

Information about childhood circumstances was gathered through retrospective questions asked of respondents in 1985. Childhood poverty = 1 if the respondent answered “poor” in response to the question “were your parents poor when you were growing up, pretty well off, or what?” Childhood poor health = 1 if the respondent reported that his/her health before age 16 was poor. Parents’ education (using the parent with the highest level of education) compared less than a high school degree, a high school degree but less than a college degree, and don’t know (in order to retain respondents who were unable to answer the question) to respondents whose parent(s) had a college degree or post-graduate degree (reference category).

Covariates

Respondents’ education was measured similar to parents’ education (although a “don’t know” category was not necessary). Race/ethnicity compares White (=1) and minority respondents, which consist primarily of African American as well as a small number of Hispanic and Asian sample members. Sex (male = 1), age in 2013 (continuous 54–64 in 2013), and marital status in 2013 (married = 1) were also included as covariates in analyses of economic outcomes. To control for the potential confounding effects of economic status in early adulthood, discussed below, we include a measure of poverty in 1981 consistent with the coding of the 2013 measure.

Analytic Strategy

In the first stage of the analysis, we model trajectories of work disability using indicators of disability measured from 1981 to 2013 and repeated-measures latent class analysis (RMLCA). RMLCA is a finite mixture model that identifies subgroups of individuals who are homogenous in their pattern of experience on an outcome as they age (Collins & Lanza, 2010; Lanza, Dziak, Huang, Xu, & Collins, 2015). RMLCA detects associations among variables due to an unmeasured, latent source of variation and sorts individuals into mutually exclusive and exhaustive latent classes for which the posterior probability is the highest based on patterns of observed responses (Collins & Lanza, 2010; McCutcheon, 1987; Nagin, 2005). Because RMLCA does not require the specification of a functional form for individual change over time, it allows for the identification of unique patterns of work disability. An important step in RMLCA is the inclusion of covariates into models to identify characteristics that predict membership in the various latent classes. Our conceptual model, which suggests that early-life environment affects economic outcomes in later life in part through pathways of work disability, motivates the inclusion of childhood circumstances (poverty, health, parents’ education), as well as controls for respondents’ education, sex and race/ethnicity. In the second stage of analysis, we include the classes of work disability estimated with these covariates as independent variables in logistic regression models predicting the four economic outcomes (see Supplementary Material Section 1.3 for methodological details). To control for the possibility that the association between work disability pathway and economic insecurity in 2013 is due to a correlation between work disability and economic security earlier in the life course, we include poverty measured in 1981, which coincides with the first observation of work disability. With the inclusion of poverty in 1981, the variance in the 2013 economic insecurity measures that is predicted by the work disability latent classes is less likely driven by earlier economic insecurity (Selig & Little, 2012).

Results

Identifying Work Disability Trajectories

Our first goal is to identify trajectories of work disability and specify the number of latent classes. Model selection is theoretically motivated and based on a combination of fit statistics, parsimony and the interpretability of sequentially estimated models (Collins & Lanza, 2010). The life course perspective and previous literature (e.g., Pais, 2014) framed our interpretation of post hoc tests of model fit, from which we determined that the model with five latent classes provided the best fit to the data (see Supplementary Material Section 1.4 for methodological details).

Item-response probabilities show the association between the indicators of work disability at each time point and the latent classes of work disability, and for each class, they indicate the probability of work disability in each year. See Figure 1 for the item-response probabilities for the measurement of five classes of work disability and estimates of the proportion of the sample in each latent class (weighted). A large proportion of the sample (67.5%) had a very low probability of experiencing a work disability from young adulthood to late middle age (labeled “Never Work Disabled”). Another 11.8% of the sample had a low probability of experiencing a work disability until their 50s and 60s, (the “Late Onset” class). In contrast, respondents in the “Early Onset” latent class (4%) had a high likelihood of experiencing a work disability in their 20s that quickly increased and was maintained at a high level for the duration of the period. Respondents in the “Midlife Onset” class (8.3%) saw a steep increase in the probability of work disability in their 40s that continued to climb to high levels by the end of the study period. Finally, respondents in the “Transient Work Disability” class (8.4%) experienced episodic health conditions that limited their ability to work. We plotted a random subset of cases from this class and their trajectories of work disability evidence frequent movement in and out of work disability over the analysis period (not shown). This movement results in an average probability of work disability that is consistently elevated in the middle years of the observation period. These findings are similar to the health impairment trajectories identified by Pais (2014) using the NLSY79.

Figure 1.

Figure 1.

Open in a new tab

Item-response probabilities for a five-class longitudinal latent class model of work disability trajectories, Panel Study of Income Dynamics, 1981–2013 (N = 2,063).

Early Life Anchors of Work Disability Trajectories

Table 1 presents weighted means and proportions by work disability class. There is little difference in mean age across the classes, and no significant differences in their racial makeup, which ranges from 79% (Early Onset) to 89% White (Transient). Those in the Midlife Onset class are predominantly women (75.9%), while the composition of the other classes is more gender balanced (48% male Late Onset and 44% male Early Onset and Never Work Disabled). The identification of a class with a high proportion of women with work disability onset in midlife is consistent with research demonstrating higher rates of disability and functional limitations among women in middle age (Gorman & Read, 2006; Verbrugge & Liu, 2014). About one-fifth of the Early Onset class reported they had poor health in childhood, significantly higher than all the other classes. Proportions of respondents reporting poor childhood health in the Late Onset and Never Work Disabled classes were not significantly different. All the classes except for the Transient class were more likely than the Never Work Disabled to report that their family was poor while they were growing up, with the highest proportion among the Early Onset class (almost 40%). Fewer than 40% of individuals in the Midlife and Early Onset classes compared to 72% of the Never Work Disabled had highly educated parents. Similarly, respondents in the Midlife and Early Life Onset classes were themselves the most educationally disadvantaged. Seventy-one percent of the Never Work Disabled class was married in 2013 compared to just over one-third of the Early Onset class and roughly half of respondents in the other disability classes.

Table 1.

Proportions and Means by Disability Trajectory, Weighted, Panel Study of Income Dynamics

Never work disabled Transient Late onset Midlife onset Early onset
Independent variables f
Age in 2013 59.1b,e 59.1b,e 60.1a,c 59.6 60.1a,c
Race/ethnicity
 White 86.3 88.5 81.0 85.1 79.4
 Non-White 13.7 11.5 19.0 14.9 20.6
Sex
 Male 44.0d 46.0d 48.2d 24.1a,b,c,e 43.8d
 Female 56.0 54.0 51.8 75.9 56.2
Childhood health
 Fair/poor 1.3c,d,e 11.4a,b,e 2.9c,d,e 12.4a,b,e 21.0a,b,c,d
Childhood poverty 17.6b,d,e 21.5b,d,e 29.4a 35.7a,c 39.0a,c
Parents’ education
 % 12 or more years 72.3d,e 73.7 65.7 62.5a 62.1a
Education
 % 12 or more years 91.6c,d,e 84.4a,d 90.0d,e 75.1a,b,c 76.9a,b
Marital status 2013 70.9b,c,d,e 54.1a,e 57.5a,e 48.1a 35.7a,b,c
Dependent variables g
Poverty 4.6b,c,d,e 11.7a,b 18.8a,c 17.8a 18.7a
Asset poverty 14.3b,c,d,e 22.2a,d,e 23.7a 40.7a,c 42.9a,c
Home ownership 83.3b,d,e 76.5b 65.6a,c 68.0a 69.2a
Pension 53.9b,c,d,e 40.1a 32.4a 29.6a 30.5a
Open in a new tab

Note: aSignificantly different from Never Work Disabled, p < .05. bSignificantly different from Late Onset, p < .05. cSignificantly different from Transient, p < .05. dSignificantly different from Midlife Onset, p < .05. eSignificantly different from Early Life Onset, p < .05. fN = 2,063. gN = 1,738.

In Table 2, odds ratios report individuals’ likelihood of experiencing a particular trajectory of work disability relative to the Never Work Disabled (reference category). Early-life disadvantage is significantly related to work disability in adulthood, as individuals who experienced poverty in childhood have almost twice the likelihood of membership in the Midlife (OR 1.856) and almost three times the likelihood of membership in the Early Onset latent class (OR 2.908) compared to the Never Work Disabled class. Poor childhood health also increased the likelihood of membership in the Transient (OR 4.789), Midlife (OR 6.364) and Early Onset (OR 15.284) latent classes compared to the Never Work Disabled class. Although parents’ education did not predict work disability trajectory, respondents’ low education is associated with both Early (OR 3.185) and Midlife Onset (OR 4.981). Overall, the findings demonstrate that early-life conditions select individuals into work disability trajectories, with early disadvantage disproportionately associated with early and midlife onset.

Table 2.

Predictors of Membership in Latent Classes of Work Disability Trajectories, Panel Study of Income Dynamics, 1981–2013

Transient (8.4%) Late onset (11.8%) Midlife onset (8.3%) Early onset (4.0%)
Intercept 0.032 0.188 0.134 0.005
Male 1.181 1.140 0.488** 1.183
White 1.201 0.697 0.813 1.157
Education
 <high school 1.634 1.310 4.981*** 3.185*
 High school  (vs. college degree) 0.979 1.276 1.533 1.130
Parents’ education
 <high school 2.356 1.475 0.739 3.710
 High school 2.652 1.177 0.782 3.556
 Don’t know  (vs. college degree) 1.232 0.598 0.660 4.111
Childhood poverty 1.711 1.560 1.856* 2.908*
Childhood poor health 4.789** 1.479 6.364*** 15.284***
Open in a new tab

Note: N = 2,063. Reference category = Never Work Disabled (67.5%).

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

Economic Insecurity

Next, we examine whether the work disability trajectories identified above are associated with economic insecurity as individuals approach traditional retirement age. Descriptive results presented in the bottom panel of Table 1 show the economic advantages of respondents who never experienced a work disability. A significantly lower proportion of the Never Work Disabled class experience poverty or asset poverty compared to all the other classes, and a significantly higher proportion of the Never Work Disabled have a pension than all of the other classes. Similarly, the Never Work Disabled are significantly more likely to own a home than all the classes except the Transient class. Table 3 presents odds ratios from logistic regression models of the four measures of economic insecurity including basic demographic controls in Model 1 (age, sex, race/ethnicity) and the addition of education, marital status and poverty in 1981 in Model 2 as potential confounders of the relationship between work disability and economic outcomes. In analyses not shown here, we tested for significant differences in the effects of the work disability trajectories on each outcome and report the results below.

Table 3.

Logistic Regressions of the Odds of Home Ownership, Poverty, Asset Poverty and Pensions, Panel Study of Income Dynamics, 2013

Owns home Poverty Asset poverty Pension ownership
Variables Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Model 1 Model 2
Intercept 0.048* 0.122 6.505 0.149* 22.279* 1.113 19.465** 3.211*
Disability pathways
 Transient 0.849 0.926 2.574** 1.932** 1.785* 1.274 0.530** 0.668**
 Late onset 0.439*** 0.505*** 3.712*** 3.449*** 2.102*** 1.909*** 0.464*** 0.422***
 Midlife onset 0.392*** 0.850 4.764*** 3.910*** 3.893*** 2.662*** 0.317*** 0.363***
 Early onset (vs. Never) 0.397** 0.702 4.045*** 2.161* 3.551*** 2.043* 0.399** 0.541*
Age 1.065** 1.039 0.941* 0.531 0.940** 0.963 0.944*** 0.926***
Male (vs. female) 1.111 0.839 0.786 0.464 0.700** 0.849 1.171 1.043
White (vs. minority) 4.081*** 2.316*** 0.173*** 0.335*** 0.232*** 0.406*** 1.944*** 1.419**
Education
 <High school 0.404*** 4.260*** 3.408*** 0.283***
 High school (vs. college degree) 0.944 2.862** 1.982*** 0.600***
Married 2013 7.459*** 0.172*** 0.294*** 1.298**
Poverty 1981 0.750 1.986*** 2.045*** 0.590***
Open in a new tab

Note: N = 1,738; *p < .05; **p < .01; ***p < .001.

Home ownership

In Model 1, respondents in each of the work disability trajectories except for the Transient class are significantly less likely to own their home compared to the Never Work Disabled. From comparisons not shown, respondents in the Transient class are more likely to own their home than those in the Late, Midlife, and Early Onset classes. Model 2 suggests that the effect of work disability on home ownership operates through lower levels of education and marriage (earlier poverty exposure was not statistically significant), for all but those with late work disability onset. Consistent with research on home ownership, respondents who lack a high school diploma are significantly less likely than those with a postsecondary degree to own a home, while those who are married are almost eight times more likely than those who are not to own their home.

Poverty

All the work disability trajectories led to an increased risk of poverty compared to the Never Work Disabled class (Model 1). For example, respondents in the Midlife Onset class are almost 5 times as likely as the Never Work Disabled to live in poverty (OR 4.764) and those in the Early Onset class are 4 times as likely (OR 4.045). With the inclusion of controls in Model 2 the work disability coefficients are slightly reduced but maintain their statistical significance. Further analysis (not shown) indicated that the work disability trajectories are not significantly different from each other in their likelihood of increasing poverty, demonstrating that a health condition that disrupts employment even temporarily or later in life is detrimental for economic security approaching traditional retirement age.

Asset poverty

Experiencing a work disability at any point in adulthood increases the likelihood of asset poverty (Model 1). Asset poverty is twice as likely for respondents in the Late Onset class, over three times as likely for respondents in the Early Onset class, and almost four times more likely for those in the Midlife Onset class compared to the Never Work Disabled class (OR 2.102, 3.551, and 3.893 respectively). With the inclusion of education, marital status, and 1981 poverty in Model 2, the Transient class coefficient loses statistical significance and the odds ratios of the other work disability trajectories are slightly reduced but remain statistically significant. Respondents in both the Midlife and Early Onset classes have a significantly higher likelihood of asset poverty than the other classes but were not significantly different from one another (results not shown).

Pension ownership

Those who experience a work disability are significantly less likely to have an employer pension than the never work disabled as they approach retirement age (Model 1). This is not a function of differences in education, marital status or earlier poverty, as both the magnitude and statistical significance of the odds ratios are maintained in Model 2. The effects of the work disability trajectories on the likelihood of having a pension do not significantly differ from one another (results not shown).

Discussion

This study examined how trajectories of work disability from early adulthood to late midlife are associated with disadvantage in key economic outcomes as individuals approach traditional retirement age. Results showed four distinct trajectories of work disability that differ in their timing and stability, and compared to the never work disabled, all trajectories placed individuals at an increased risk of economic insecurity across the four measures. Consistent with theories of CIT/CAD and existing research demonstrating the childhood origins of life course health, early life circumstances selected individuals into disadvantaged work disability trajectories. Early life disadvantage increased the likelihood of experiencing early or midlife onset of persistent work disability. For example, those who first experienced work disability in early adulthood or midlife were more likely to have experienced childhood poverty, to have low educational attainment, and to have had poor childhood health.

Further, the findings demonstrate that work disability contributes to future inequality. Work disability, regardless of timing and duration, is associated with an increased risk of economic insecurity. Net of poverty status at the beginning of the observation period, all the work disability classes were more likely to experience poverty in late midlife and were less likely to have a pension. All except the transient class also were more likely to experience asset poverty. Findings regarding home ownership were less clear, although the lower likelihood of home ownership among the late onset class suggests that families may liquidate assets as a source of financial resources upon the onset of a work disability in late mid-life. The differential effect of the work disability trajectories on economic precarity was not as distinct as we expected based on theories of CIT/CAD. For example, individuals whose health limited their labor force attachment in their 50s and 60s (Late Onset) were as economically insecure as those experiencing the onset of a work-limiting impairment a decade earlier (Midlife Onset), across all of the economic outcomes. This suggests that employment disruption, even of a transient nature or occurring late in one’s work life, significantly diminishes the ability to accumulate resources, demonstrating the importance of full labor force participation throughout the adult life course.

There are many heterogenous experiences of disability. This analysis focused on the stratification of outcomes associated with work disability pathways rather than on the disablement process itself. Although the measure of disability used in this analysis does not incorporate the specific health condition or type of disability experienced, the measure supports a social disability model by focusing on the intersection of impairment and social context, rather than on the impairment itself. Our measure captures the experience of a physical or mental health condition that is perceived by the respondent as significant enough to impair the ability to work in his or her given occupation. Health conditions of differing severity and type may prevent employment, as what is limiting in one occupation may not be in another, or in a workplace that provides adequate accommodations compared to one that does not. One avenue for future research involves an investigation of the types of health problems that lead to different work disability pathways and their intersection with occupational location, as well as further conceptualization and systematic examination of the socioeconomic consequences of disability across the entire life course from birth to death. In addition, the contribution of gender and marital status to the socioeconomic effects of work disability pathways requires greater exploration. Although our finding that marriage is protective and that women are overrepresented in the midlife onset pathway is consistent with existing research, research also suggests that there are gender and marital status differences in the likelihood of leaving the labor market following disability, suggesting that the socioeconomic consequences of work disability are more nuanced than described here (Flippen & Tienda, 2000; McDonough & Amick, 2001).

Additionally, variation in the experiences of work disability across birth cohorts is also an important consideration. Recent decades have seen advances in medicine and technology that have extended longevity for persons with early life impairments, as well as the enactment of legislation, most notably the Americans with Disabilities Act (1990) and the Family and Medical Leave Act (1993), aimed at increasing labor force participation and return to work following disability. Legislative requirements for workplace disability accommodations provided by employers address the lack of fit between individuals and their work environment and serve to protect workers from the potentially harmful effects of disability on employment outcomes. The Baby Boom cohorts in our analysis were among the first to experience a significant portion of their work lives under legislative protections and we would expect significant variation in the relationship between pathways of work disability and economic precarity among earlier and later cohorts.

In sum, work disability and its limits on labor force participation may be one important mechanism through which disadvantaged childhood circumstances are linked to later life economic insecurity. These life course pathways are complex and warrant further research, involving the relationship between early life disadvantage, disability onset and duration, educational attainment, marriage, and occupational location. For example, to the extent that marriage is protective against economic insecurity for those with a work disability, as we see in this analysis, lower rates of marriage for those with early and midlife onset contribute to the precariousness associated with disability. Overall the current study demonstrates the need to reconsider the “silos” created by a focus on age-specific measures and studies of disability. Nearly 20% of our sample experienced a pathway of persistent work limiting disability that began in midlife, which is further evidence for the importance of considering disability, as well as its socioeconomic origins and consequences, as a process that extends across the entire life course, rather than one relegated to study during two distinct life stages, early life and old age (also see Verbrugge et al., 2017).

Supplementary Material

Supplementary data is available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online.

Funding

This work was supported by an operating grant from the Canadian Institutes of Health Research (120311). The collection of data used in this study was partly supported by the National Institutes of Health (R01 HD069609) and the National Science Foundation (1157698).

Conflict of Interest

None reported.

Supplementary Material

gbx096_suppl_Supplemental_material
Click here for additional data file. (22.3KB, docx)

References

  1. Baldwin M. L., & Schumacher E. J (2002). A note on job mobility among workers with disabilities. Industrial Relations, 41, 430–441. doi:10.1111/1468-232X.00255 [Google Scholar]
  2. Bane M., & Ellwood D (1983). The dynamics of dependence: The routes to self-sufficiency. Cambridge, MA: Urban Systems Research and Development. [Google Scholar]
  3. Brandt E. N. Jr & Pope A. M (1997). Enabling America: Assessing the Role of Rehabilitation Science and Engineering. Committee on Assessing Rehabilitation Science and Engineering, Division of Health Sciences Policy, Institute of Medicine.Washington D.C: National Academy Press. doi:10.17226/5799 [PubMed] [Google Scholar]
  4. Burkhauser R. V. Daly M. C. Houtenville A. J. & Nargis N (2002). Self-reported work-limitation data: What they can and cannot tell us. Demography, 39, 541–555. doi:10.1353/dem.2002.0025 [DOI] [PubMed] [Google Scholar]
  5. Clarke P. & Latham K (2014). Life course health and socioeconomic profiles of Americans aging with disability. Disability and Health Journal, 7, S15–S23. doi:10.1016/j.dhjo.2013.08.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Collins L. M., & Lanza S. T (2010). Latent class and latent transition analysis. Hoboken, NJ: Wiley. doi:10.1002/9780470567333 [Google Scholar]
  7. Dannefer D. (2003). Cumulative advantage/disadvantage and the life course: Cross-fertilizing age and social science theory. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 58, S327–S337. doi:10.1093/geronb/58.6.S327 [DOI] [PubMed] [Google Scholar]
  8. Deeg D. J. H. (2005). Longitudinal characterization of course types of functional limitations. Disability and Rehabilitation, 27, 253–261. doi:10.1080/09638280400006507 [DOI] [PubMed] [Google Scholar]
  9. Erickson W., Lee C., & von Schrader S (2015). Disability statistics from the 2013 American Community Survey (ACS). Ithaca, NY: Cornell University Yang Tan Institute (YTI). [Google Scholar]
  10. Elder G. H. Jr, Johnson M. K. & Crosnoe R (2003). The emergence and development of life course theory. In Mortimer J. & Shanahan M. (Eds.), Handbook of the life course (pp. 3–19). New York: Kluwer Academic/Plenum Publishers. doi:10.1007/978-0-306-48247-2_1 [Google Scholar]
  11. Ferraro K. F. & Shippee T. P (2009). Aging and cumulative inequality: How does inequality get under the skin?The Gerontologist, 49, 333–343. doi:10.1093/geront/gnp034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fisher J. D., Johnson D. S., Marchand J. T., Smeeding T. M., & Torrey B. B (2007). No place like home: Older adults and their housing. Journals of Gerontology: Social Sciences, 62B, S120–S128. doi:10.1093/geronb/62.2.S120 [DOI] [PubMed] [Google Scholar]
  13. Fitzgerald J. (2011). Attrition in models of intergenerational links in health and economic status in the PSID. The B.E. Journal of Economic Analysis & Policy, 11, 1–61. doi:10.2202/1935-1682.2868 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Flippen C. & Tienda M (2000). Pathways to retirement: Patterns of labor force participation and labor market exit among the pre-retirement population by race, Hispanic origin, and sex. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 55, S14–S27. doi:10.1093/geronb/55.1.S14 [DOI] [PubMed] [Google Scholar]
  15. George L. K. (1993). Sociological perspective on life transitions. Annual Review of Sociology, 19, 353–373. doi:10.1146/annurev.so.19.080193.002033 [Google Scholar]
  16. Gorman B. K. & Read J. G (2006). Gender disparities in adult health: An examination of three measures of morbidity. Journal of Health and Social Behavior, 47, 95–110. doi:10.1177/002214650604700201 [DOI] [PubMed] [Google Scholar]
  17. Haas S. A. (2006). Health selection and the process of social stratification: The effect of childhood health on socioeconomic attainment. Journal of Health and Social Behavior, 47, 339–354. doi:10.1177/002214650604700403 [DOI] [PubMed] [Google Scholar]
  18. Haas S. A. Glymour M. M. & Berkman L. F (2011). Childhood health and labor market inequality over the life course. Journal of Health and Social Behavior, 52, 298–313. doi:10.1177/0022146511410431 [DOI] [PubMed] [Google Scholar]
  19. Haveman R. & Wolff E. N (2004). The concept and measurement of asset poverty: Levels, trends and composition for the U.S., 1983–2001. Journal of Economic Inequality, 2, 145–169. doi:10.1007/s10888-004-4387-3 [Google Scholar]
  20. Kelley-Moore J. (2010). Disability and ageing: The social construction of causality. In Dannefer D. & Phillipson C. (Eds.), The Sage handbook of social gerontology (pp. 96–110), Thousand Oaks, CA: Sage. doi:10.4135/9781446200933.n7 [Google Scholar]
  21. Kuh D., & Ben-Shlomo B (2004). A life course approach to chronic disease epidemiology (2nd edn). New York: Oxford University Press. doi:10.1093/acprof:oso/9780198578154.001.0001 [Google Scholar]
  22. Laditka J. N. & Laditka S (2015). Associations of educational attainment with disability and life expectancy by race and gender in the United States: A longitudinal analysis of the Panel Study of Income Dynamics. Journal of Aging and Health, 28, 1403–1425. doi:10.1177/0898264315620590 [DOI] [PubMed] [Google Scholar]
  23. Latham K. (2012). Progressive and accelerated disability onset by race/ethnicity and education among late midlife and older adults. Journal of Aging and Health, 24, 1320–1345. doi:10.1177/0898264312459345 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lanza S. T., Dziak J. J., Huang L., Xu S., & Collins L. M (2015). Proc LCA and Proc LTA Users’ Guide (Version 1.3.2). University Park: The Methodology Center, Penn State; http://methodology.psu.edu. [Google Scholar]
  25. Liang J. Bennett J. M. Shaw B. A. Quiñones A. R. Ye W. Xu X. & Ofstedal M. B (2008). Gender differences in functional status in middle and older age: Are there any age variations?The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 63, S282–S292. doi:10.1093/geronb/63.5.S282 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Liang J. Xu X. Bennett J. M. Ye W. & Quiñones A. R (2009). Ethnicity and changing functional health in middle and late life: A person-centered approach. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 65, 470–481. doi:10.1093/geronb/gbp114 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lynch S. M. (2015). Commentary on “Disability Trajectories at the End of Life: A Countdown Model” the problems with time-to-death as a predictor of disability. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 70, 753–756. doi:10.1093/geronb/gbv001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. McCutcheon A. L. (1987). Latent class analysis. Quantitative applications in the social sciences, Series No. 64.Thousand Oaks, CA: Sage. [Google Scholar]
  29. McDonough P. & Amick B. C.. 3rd (2001). The social context of health selection: A longitudinal study of health and employment. Social Science & Medicine (1982), 53, 135–145. doi:10.1016/S0277-9536(00)00318-X [DOI] [PubMed] [Google Scholar]
  30. Meyer B. D., & Sullivan J. X (2012). Identifying the disadvantaged: Official poverty, consumption poverty, and the new supplemental poverty measure. Journal of Economic Perspectives, 26, 3. doi:10.1257/jep.26.3.111 [Google Scholar]
  31. Mok W. K., Meyer B. D., Charles K. K., & Achen A. C (2008). A note on “The longitudinal structure of earnings losses among work-limited disabled workers.” The Journal of Human Resources, 43, 3, 721–728. doi:10.1353/jhr.2008.0012 [Google Scholar]
  32. Nagi S. (1991). Disability concepts revisited: Implications to prevention. In Pope A. M. & Tarlove A. R. (Eds.) Disability in America: Toward a National Agenda for Prevention (pp. 307–327). Washington, DC: National Academy Press. [Google Scholar]
  33. Nagin D. S. (2005). Group-based modeling of development.Cambridge, MA: Harvard University Press. [Google Scholar]
  34. Nusselder W. J. Looman C. W. & Mackenbach J. P (2006). The level and time course of disability: Trajectories of disability in adults and young elderly. Disability and Rehabilitation, 28, 1015–1026. doi:10.1080/09638280500493803 [DOI] [PubMed] [Google Scholar]
  35. Oliver M. (1996). Understanding disability: From theory to practice. Basingstoke: Macmillan. [Google Scholar]
  36. Oliver M. L. & Shapiro T. M (1997). Black wealth/white wealth. New York: Routledge Press. [Google Scholar]
  37. O’Rand A. M. (1996). The precious and the precocious: Understanding cumulative disadvantage and cumulative advantage over the life course. The Gerontologist, 36, 230–238. doi:10.1093/geront/36.2.230 [DOI] [PubMed] [Google Scholar]
  38. O’Rand A. M. (2006). Stratification and the life course: Life course capital, life course risks, and social inequality. In Binstock R. H. & George L. K. (Eds.), Handbook of aging and the social sciences (6th edn, pp. 145–162). Burlington, MA: Academic Press. doi:10.1016/B978-012088388-2/50012-2 [Google Scholar]
  39. O’Rand A. M. & Hamil-Luker J (2005). Processes of cumulative adversity: Childhood disadvantage and increased risk of heart attack across the life course. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 60 Spec No 2, 117–124. doi:10.1093/geronb/60.Special_Issue_2.S117 [DOI] [PubMed] [Google Scholar]
  40. Pais J. (2014). Cumulative structural disadvantage and racial health disparities: The pathways of childhood socioeconomic influence. Demography, 51, 1729–1753. doi:10.1007/s13524-014-0330-9 [DOI] [PubMed] [Google Scholar]
  41. Panel Study of Income Dynamics. (2016). Produced and distributed by the Survey Research Center, Institute for Social Research, University of Michigan, Ann Arbor, MI. [Google Scholar]
  42. Priestley M. (2005). Disability and social inequalities. In Romero M. & Margolis E. (Eds.), The Blackwell companion to social inequalities (pp. 372–395). Oxford: Blackwell. doi:10.1002/9780470996973.ch17 [Google Scholar]
  43. Putnam M. (2002). Linking aging theory and disability models: Increasing the potential to explore aging with physical impairment. The Gerontologist, 42, 799–806. doi:10.1093/geront/42.6.799 [DOI] [PubMed] [Google Scholar]
  44. Robert S. & House J. S (1996). SES differentials in health by age and alternative indicators of SES. Journal of Aging and Health, 8, 359–388. doi:10.1177/089826439600800304 [DOI] [PubMed] [Google Scholar]
  45. Rank M. R. (2009). Measuring the economic racial divide across the course of American lives. Race and Social Problems, 1, 57–66. doi:10.1007/s12552-009-9009-z [Google Scholar]
  46. Schur L. (2003). Barriers or opportunities? The causes of contingent and part-time work among people with disabilities. Industrial Relations, 42, 589–622. doi:10.1111/1468-232X.00308 [Google Scholar]
  47. Schur L., Kruse D., Blasi J. & Blanck P (2009). Is disability disabling in all workplaces? Workplace disparities and corporate culture. Industrial Relations, 48, 381–410. doi:10.1111/j.1468-232X.2009.00565.x [Google Scholar]
  48. Selig J. P., & Little T. D (2012). Autoregressive and cross-lagged panel analysis for longitudinal data. In Laursen B., Little T. D., & Card N. A. (Eds.), Handbook of developmental research methods (pp. 265–278). New York: Guilford Press. [Google Scholar]
  49. Shuey K. M., Willson A. E., & Bouchard K (2016). Disability and social inequality in Canada.” In Grabb E. & Guppy N. (Eds.), Social inequality in Canada: Dimensions of disadvantage (6th edn, pp. 258–273), Oxford University Press. [Google Scholar]
  50. Taylor M. G. (2010). Capturing transitions and trajectories: The role of socioeconomic status in later life disability. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 65, 733–743. doi:10.1093/geronb/gbq018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Taylor M. (2011). The causal pathway from socioeconomic status to disability trajectories in later life: The importance of mediating mechanisms for onset and accumulation. Research on Aging, 33, 84–108. doi:10.1177/0164027510385011 [Google Scholar]
  52. Taylor M. G. & Lynch S. M (2004). Trajectories of impairment, social support, and depressive symptoms in later life. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 59, S238–S246. doi:10.1093/geronb/59.4.S238 [DOI] [PubMed] [Google Scholar]
  53. Taylor M. G. & Lynch S. M (2011). Cohort differences and chronic disease profiles of differential disability trajectories. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 66, 729–738. doi:10.1093/geronb/gbr104 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Verbrugge L. M. (2016). Disability experience and measurement. Journal of Aging and Health, 28, 1124–1158. doi:10.1177/0898264316656519 [DOI] [PubMed] [Google Scholar]
  55. Verbrugge L. M. & Jette A. M (1994). The disablement process. Social Science & Medicine (1982), 38, 1–14. doi:10.1016/0277-9536(94)90294-1 [DOI] [PubMed] [Google Scholar]
  56. Verbrugge L. M. & Liu X (2014). Midlife trends in activities and disability. Journal of Aging and Health, 26, 178–206. doi:10.1177/0898264313508189 [DOI] [PubMed] [Google Scholar]
  57. Verbrugge L. M., & Yang L (2002). Aging with disability and disability with aging. Journal of Disability Policy Studies, 12, 253–267. doi:10.1177/104420730201200405 [Google Scholar]
  58. Verbrugge L. M., Latham K., & Clarke P. J (2017). Aging with disability for midlife and older adults. Research on Aging, 39, 741–777. doi:10.1177/0164027516681051 [DOI] [PubMed] [Google Scholar]
  59. Wagmiller R. L., Lennon M. C., Kuang L., Alberti P. M., & Aber J. L (2006). The dynamics of economic disadvantage and children’s life chances. American Sociological Review, 71, 847–866. doi:10.1177/000312240607100507 [Google Scholar]
  60. Warner D. F. & Brown T. H (2011). Understanding how race/ethnicity and gender define age-trajectories of disability: An intersectionality approach. Social Science & Medicine (1982), 72, 1236–1248. doi:10.1016/j.socscimed.2011.02.034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Warren J. R. (2009). Socioeconomic status and health across the life course: A test of the social causation and health selection hypotheses. Social Forces; A Scientific Medium of Social Study and Interpretation, 87, 2125–2153. doi:10.1353/sof.0.0219 [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Wolf D. A. Freedman V. A. Ondrich J. I. Seplaki C. L. & Spillman B. C (2015). Disability trajectories at the end of life: A “Countdown” Model. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 70, 745–752. doi:10.1093/geronb/gbu182 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

gbx096_suppl_Supplemental_material
Click here for additional data file. (22.3KB, docx)

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

RESOURCES