Race Inequities in Women’s Retirement: An Evaluation of the Intervening Effects of Health, Wealth, and Work and Family Histories

Tyson Brown, University of Florida

The literature has shown that there are race differences in retirement behavior. However, most of the research on race and retirement has overlooked women, with only few exceptions (e.g. Belgrave). Belgrave’s (1988) finding that African American women have more continuous patterns of work throughout the life course than white women, a finding that is opposite that found among men, suggests that the race-retirement behavior relationship may vary by gender, and underscores the need for further research on race differences in women’s retirement behavior. Furthermore, the limited number of studies that have explored race differences in women’s retirement behavior have employed different measurement strategies and shown mixed results. For example, labor force participation rates have been used to study retirement behavior and have masked important race differences that are revealed by further classifying individuals who are out of the labor force into disabled and nondisabled groups (Hayward, Friedman, and Chen, 1996). The purpose of the present study is to investigate race differences in women’s retirement behavior, using a wider range of retirement outcomes including retired and disabled. Furthermore, we explore whether life course factors such as sociodemographic characteristics, health, wealth, and work history and family history intervene in the race-retirement behavior relationship. A life course framework is used to investigate racial differences in sociodemographic characteristics, work and family circumstances, health, wealth, and subsequent retirement behavior. Specifically, we examine racial differences in these life course variables, and look at how important this is for retirement. In particular, we are interested in whether health concerns that are proximate to the retirement years are a consequence of greater lifelong socioeconomic disadvantaged positions of African American women. African American women have poorer health than white women. Specifically, African American women have a higher prevalence of diabetes, hypertension, heart disease, strokes, and functional loss. Having a chronic disease and/or functional limitations increases an individual’s odds of being work disabled. Therefore, we speculate that excess disability rates among African American women may be due to greater morbidity and functional loss. Likewise, race differences in wealth, and work and family circumstances are expected to play a role in race differences in retirement behavior. For example, African American women are less likely than White women to be married. African American women are especially less likely to have a retired spouse. Thus, given that unmarried women are less likely to be retired, African American women may work longer than White women. Panel data are employed from the Health and Retirement Study (HRS), 1992-2000. The HRS is a nationally representative sample of Americans between the ages 51-61 in 1992, with oversamples of African Americans, Latinos and Floridians. Data were collected every two years via face-to-face (baseline) and telephone interviews (1994, 1996, 1998, 2000). The HRS is an ideal source for investigating retirement behavior because it has extensive measures of known predictors of retirement behavior such as sociodemographic factors, work and family histories, and measures of health and wealth. Analyses are restricted to women between the ages of 51-61 in 1992 because this is the only age group in the sample that is nationally representative of both married and unmarried women. Because of this study’s focus on retirement transitions, analyses are further restricted to African American (n=446) and White (n=1821) women who are in the labor force (working ³ 1 per week) and who do not self identify as retired in 1992. Measurement of the Dependent Variable (Retirement Behavior). Women’s retirement behavior is self-reported at each wave and distinguishes respondents as being in one of the following three following mutually exclusive categories: (1) in the labor force (working ³ 1 hour per week), (2) not working (i.e. 0 hours per week) as a result of a disability, and (3) not working and not disabled (the majority of this group self-identifies as retired). This measure in constructed for each wave of data. Life Course Measures. Early life sociodemographic factors include parent’s education, respondent’s education, and race. Age (at baseline) is also included as a control variable. Family history variables include single parenting experiences (non-marital birth and post divorce spells of single parenting), marital status (married or partnered w/spouse in labor force; married or partnered w/spouse not in labor force; divorced or separated; widowed; never married), and present dependents in the household (1=yes; 0=no). Work history characteristics include work tenure, number of jobs, current occupation (white collar; blue collar; service), self-employment (1=currently self-employed; 0=otherwise), firm size, union/non-union status, pension eligibility status (currently receiving or eligible; covered; no pension), self-reported pension wealth, and health insurance coverage (employer-provided; non-employer-provided; uninsured). In the event that a respondent has more than one job, these work characteristics refer to their primary job. Measures of economic well being include household net worth and household income. Measures of physical health include hypertension, stroke, heart disease, diabetes, and cancer. These are dummy variables that are coded according to how the respondent answers the question in 1992 “Has a doctor ever told you that you have (had a) [condition].” In subsequent years, onset of new chronic conditions is measured. Hence health status is measured both at baseline (a fixed covariate) and with a time varying covariate. Functional health is measured by the presence of an ADL (1=difficulty in getting up from the bed, bathing, eating, or dressing; 0=otherwise). Analytic strategy. Descriptive statistics are presented for white and African American women. Racial differences in descriptive characteristics are calculated using t-test (continuous variables) and chi-square (categorical variables) statistics. Second, retirement rates of women are computed for (1) the full sample, and (2) by race. Third, discrete time hazard models of worker’s risks of becoming disabled or retired are estimated. A series of nested models are estimated in order to evaluate the direct and indirect effects of race and life course variables. An initial set of models includes race and other sociodemographic variables. Family, work, economic, and physical and functional health measures are added sequentially allowing comparisons of major effects across the models in order to establish direct and indirect effects of life course variables on retirement behavior. Analyses are weighted with respondent-level weights provided by the HRS staff. Results. Preliminary results from labor force transitions between waves 1 and 2 indicate that: (1) African American women are much more likely than White women to exit the labor force as a result of a disability; and work characteristics operate as suppressors of race differences in retirement behavior, with African American women being less likely to retire, (2) the race gap in work disability is due to compositional differences across the African American and White samples (i.e. African American women’s excess morbidity and functional loss), and (3) family and economic factors are important predictors of retirement behavior. References Belgrave, Linda L., 1988. “The Effects of Race Differences in Work History, Work Attitudes, Economic Resources, and Health on Women’s Retirement.” Research on Aging, 10 (3), 383-398. Hayward, Mark D., Samantha Friedman, and Hsinmu Chen. 1996. "Race Inequities in Men's Retirement." Journal of Gerontology: Social Sciences, 51B:S1﷓S10.