Retirement and Socioeconomic Characteristics of Aged Veterans: Differences by Education and Race/Ethnicity

Introduction

Veterans constitute sizable shares of the Social Security beneficiary population and the aged population as a whole. More than 9.2 million veterans received Social Security benefits in 2016, accounting for 18 percent of all adult beneficiaries (Social Security Administration 2017). Despite widespread concern among policymakers and the public about the economic well-being of aged veterans, empirical analysis of their socioeconomic outcomes remains limited. Moreover, prior studies of the socioeconomic characteristics of aged veterans have focused on comparisons between veterans and nonveterans (for example, Tamborini, Purcell, and Olsen 2016). Such studies, although useful, may overlook substantial within-group heterogeneity among veterans. In this study, we attempt to fill some of this gap by investigating trends in the retirement and socioeconomic characteristics of aged veterans. We examine differences both within subgroups of veterans and between veterans and nonveterans.

The well-being of veterans may differ along a number of lines, with education and race/ethnicity being two prominent dimensions. Studies of financial and social outcomes of the general population show sharp differentials across educational-attainment groups (Hout 2012; Tamborini, Kim, and Sakamoto 2015; Kim, Tamborini, and Sakamoto 2015). With this pattern in mind, one might expect similarly substantial differences among aged veterans by educational attainment. Race and ethnicity are also recognized as important indicators of access to economic and social resources (Hirsch and Winters 2014; Western and Pettit 2005). Accordingly, the socioeconomic conditions of aged veterans are likely to vary by race/ethnicity. Additionally, because military service can provide individuals from disadvantaged backgrounds a bridge to opportunity by improving their human capital (Kleykamp 2013; Sampson and Laub 1996; Teachman 2004; Teachman and Tedrow 2007), one might expect to find differences in outcomes between veterans and nonveterans with similar educational and racial/ethnic backgrounds.

This article assesses the extent of educational and racial/ethnic differences both among aged veterans and between aged veterans and nonveterans over the last two decades. We define “aged” as 55 or older, and examine three subgroups (55–61, 62–69, and 70 or older) to enable preretirement- and retirement-age comparisons. We focus on a range of socioeconomic indicators that include labor force participation and earnings, Social Security and other income sources, and poverty. The analysis is based on national data collected in 1995 and 2015 by the Census Bureau in its Current Population Survey Annual Social and Economic Supplement (CPS/ASEC). That source provides two large samples of aged male veterans 20 years apart. Consistent with past research, we examine only men, given the relatively small sample of aged female veterans in the CPS/ASEC to date.

The results help identify heterogeneity in the experiences of aged male veterans and between veterans and nonveterans. The analysis also enables a comparison of two snapshots of the aged veteran population spanning 20 years, which may be useful for designing policies and programs that address the changing needs of Americans with military service histories (Congressional Budget Office 2014; Government Accountability Office 2012). It is important to understand how the sociodemographics of aged veterans who enlisted during the all-volunteer force (AVF) period may differ from those of their predecessors, who served during periods of conscription. Social Security policy changes also could have important effects on aged veterans.

Background

We begin with a brief review of some of the major mechanisms that shape veterans' later-life circumstances. Conceptually, a useful starting point is to place veterans within a life-course context. A life-course perspective views a person's socioeconomic status in later life as the accumulation of earlier-life experiences within particular structural, institutional, and sociocultural contexts (Couch and others 2013; Elder 1998). This frame helps tie veterans' circumstances in later life to the cumulative effect of a series of prior experiences, including events and conditions before and during military service as well as in civilian life (London and Wilmoth 2016; Wilmoth and London 2013).

A life-course approach also emphasizes individuals' placement in cohorts with particular sociohistorical contexts (Tamborini, Couch, and Reznik 2015; Elder 1998). For veterans, one's birth cohort is critically important because it is associated with a specific military service period. A substantial line of research finds that the period in which one served in the military (such as the World War II, Korean War, Vietnam War, and Gulf War eras) is associated with a range of outcomes because it underlies divergent variables such as whether service was primarily conscripted or voluntary,¹ the extent and nature of combat exposure, the duration of service, and age at entry (Teachman and Tedrow 2007; Teachman 2007b; Wilmoth and London 2013). Different military service periods also reflect different recruitment methods, selection processes, and employers' attitudes toward veterans (MacLean and Kleykamp 2016). These factors, in turn, are likely to influence changes in veterans' socioeconomic characteristics and outcomes. By stratifying our study samples by age in 1995 and 2015, we are able to compare the experiences of veterans across distinct military service periods including the World War II era, the Korean War era, the Vietnam War era, and the AVF era.

Researchers have identified a range of individual and institutional factors that may affect veterans' socioeconomic circumstances in later life. Work history influences veterans' well-being in several ways, including its association with lifetime earnings, pension savings, and Social Security benefit levels. The labor market consequences of military service have been featured prominently in the literature (Angrist and Krueger 1994; Kleykamp 2013; Sampson and Laub 1996; Teachman and Tedrow 2004). Evidence suggests that veterans of World War II and the Korean War earned more in postservice employment than nonveterans (MacLean and Elder 2007), although their higher earnings may have been driven in large part by the positive health and education characteristics that enhanced their likelihood of selection into service (Angrist and Krueger 1994; Teachman and Tedrow 2004). Other research shows that the postservice earnings of Vietnam War veterans were lower than those of nonveterans, especially for white veterans (Angrist 1990; Angrist, Chen, and Song 2011). The labor market outcomes of AVF-era veterans appear mixed (Kleykamp 2013; Tamborini, Purcell and Olsen 2016; Teachman and Tedrow 2007).

Another factor of aged veterans' well-being is their health. Substantial research demonstrates that veterans are at greater risk of poor health than are nonveterans (Bedard and Deschênes 2006; Black and others 2004; Heflin, Wilmoth, and London 2012; Teachman 2011; Wilmoth, London, and Heflin 2015), especially among those who experienced combat (Dobkin and Shabani 2009; MacLean and Elder 2007). Wilmoth, London, and Parker (2010) found greater health declines after retirement among veterans than among nonveterans. Wilmoth and London (2011) found higher risks of impairment and substance abuse among some cohorts of veterans (for example, Vietnam-era servicemembers) than for nonveterans. In another study, active-duty members of the AVF self-reported worse overall health than reserve-duty servicemembers and nonveterans who passed the military's physical entrance exam (Teachman 2011). Work-limiting health conditions among male veterans nearing retirement age have become more prevalent in recent years, as the AVF servicemembers have aged (Tamborini, Purcell, and Olsen 2016).

Aged veterans' well-being is also linked to the U.S. retirement system and a variety of veterans benefit programs (Street and Hoffman 2013; Tamborini, Purcell, and Olsen 2016; Wilmoth and London 2011). A key source of retirement income for veterans and nonveterans alike is Social Security (Olsen and O'Leary 2011). In the 2015 CPS/ASEC, veterans and their families composed 31 percent of the adult Social Security beneficiary population. Social Security coverage was first extended to active-duty military servicemembers in 1957. Those who served during the period 1957–2001 receive special credits (up to $1,200 a year) that augment the earnings used to calculate their Social Security benefits. Veterans whose active-duty service concluded on or before December 31, 1956 may receive special monthly service credits (Social Security Administration 2018). In addition, veterans may be eligible for Social Security Disability Insurance benefits if they are unable to work because of a medical condition that is expected to last at least 1 year or result in death.

Aged veterans also may receive a variety of military service-based benefits (Congressional Budget Office 2014; Niebuhr and others 2011). A key resource for career veterans is dual entitlement to civilian and military pensions (Street and Hoffman 2013). The Department of Defense (DOD) provides retirement and health insurance benefits to military veterans who retired from active duty or served in the Armed Forces Reserves or National Guard for a specified time. In general, individuals with 20 years of military service are eligible for retirement-plan coverage, depending on the timing of their military entrance, their base pay at retirement, their years of service, and other factors.² DOD also provides disability pensions to servicemembers who have a disability that meets a severity threshold (called a 30 percent rating) and who have been determined to be unfit for continued service.

In addition, the Department of Veterans Affairs (VA) provides a range of resources for persons who served on active duty but were not full military career veterans, including tuition and other education benefits, life insurance, and home loans. The agency's pension program provides means-tested cash benefits for low-income veterans aged 65 or older³ and its veterans' compensation program provides service-related disability benefits.⁴ In addition, many veterans are eligible for medical services provided at more than 150 VA Medical Centers and approximately 1,400 community-based outpatient clinics across the United States.

Finally, the importance of selection processes into the military bears noting when evaluating the well-being of older veterans. The characteristics of individuals with military experience are not distributed as randomly as are those of the general population. Individuals must pass rigorous physical and medical examinations as well as the Armed Services Vocational Aptitude Battery to enter the military, establishing selective health and human capital characteristics for this group. During voluntary-enlistment service periods, selection also involves an individual's judgment regarding the costs and benefits of joining the military. The differences in tendencies between individuals who enlist and those who do not may drive differences in later-life circumstances between veterans and nonveterans. Moreover, changes in selection processes across service periods are also likely to influence changes in veterans' socioeconomic outcomes over time.

Veterans' Later-Life Circumstances by Education and Race/Ethnicity

Although some socioeconomic characteristics may typify older veterans as a whole, there is likely to be substantial within-group variation. In this article, we focus on two key dimensions, education and race/ethnicity.

One would expect differences in well-being among older veterans by education because educational attainment is predictive of economic and social resources over the life course, notably including lifetime earnings (Tamborini, Kim, and Sakamoto 2015). Serving in the military may itself have varying consequences on educational attainment. On one hand, military service may disrupt educational careers, given that the common age of service entry among men is 18 to 24. On the other hand, military service may enhance educational attainment and skills (Angrist 1993; Bound and Turner 2002; Stanley 2003), particularly among disadvantaged enlistees (Teachman 2007a). For example, the federal government has historically provided financial assistance for veterans' education-related expenses under legislation such as the Servicemen's Readjustment Act of 1944 (the “GI Bill”) and the Post-9/11 Veterans Educational Assistance Act of 2008. Moreover, educational attainment is a potentially significant selection factor for military service because of higher enlistment rates among individuals with disadvantaged backgrounds (Teachman 2005, 2007a, 2007b).

Additionally, we expect differences in the socioeconomic outcomes experienced by aged veterans across racial and ethnic groups. Such differences may be driven, for example, by reduced employment and educational opportunities among minorities (Huffman and Cohen 2004) and differences in key indicators of those outcomes, such as earnings and wealth, by race and ethnicity (Killewald, Pfeffer, and Schachner 2017).

Differences also may emerge between veterans and nonveterans of similar characteristics. For instance, Angrist (1990) found that the civilian earnings of white Vietnam-era veterans after discharge were lower than those of white nonveterans in the same period. Greenberg and Rosenheck (2007) found that unemployment among white AVF veterans was higher than that of their nonveteran peers, but unemployment among black veterans was lower than that of their nonveteran peers.⁵ One reason why black (or Hispanic) veterans may exhibit improved economic outcomes relative to black (or Hispanic) nonveterans is that military service provides a bridge to those from disadvantaged backgrounds to build human capital and social networks over the life course (Kleykamp 2013; Sampson and Laub 1996; Teachman 2004; Teachman and Tedrow 2007).

In this analysis, we extend current research by drawing from nationally representative data to assess the degree to which the socioeconomic outcomes of older male veterans vary across educational and racial/ethnic groups in 1995 and 2015, and how conditions differ between veterans and nonveterans.

Data and Methods

We use data from the 1995 and 2015 CPS/ASEC. Administered by the Census Bureau, the CPS/ASEC (sometimes called the March Supplement) is a nationally representative survey of around 75,000 resident civilian noninstitutionalized households. The survey covers demographic and economic characteristics, including labor force status, income, poverty, and government program participation, among many other variables. CPS/ASEC respondents report demographic information as of the time of the survey interview and income and employment information for the preceding year.⁶ A key advantage of the CPS/ASEC is that it includes data for distinct and relatively large samples of male veterans collected at different points in time, permitting us to assess the well-being of aged veterans across education and race/ethnicity subgroups using surveys fielded two decades apart.

We restrict our analytic sample to male respondents aged 55 or older when they responded to the CPS/ASEC. We exclude women because of insufficient sample sizes among aged veterans.⁷ We split the analytic sample into veterans and nonveterans. Veteran status is indicated by self-reports of previous active-duty service in the armed forces. Respondents reporting otherwise are deemed to be nonveterans. To account for cohort and age effects, we stratify the sample by three age groups (55–61, 62–69, and 70 or older).

Measures

We categorize educational attainment by highest level completed: less than high school, high school graduate, some college, and college graduate. For race/ethnicity, we classify respondents into one of four subgroups: non-Hispanic white, non-Hispanic black, Hispanic (any race), and other. Non-Hispanic respondents self-identifying multiple race categories are included in the “other” category. Hereafter, the “non-Hispanic” qualifier is to be assumed when we refer to white and black individuals.

We explore three domains of aged veterans' socioeconomic circumstances. The first domain is demographic, including period of most recent active-duty military service,⁸ marital status, and living arrangements. The second domain involves labor market outcomes: employment, median individual earnings, and work-limiting disability, defined as a self-reported health problem or disability that prevents work or limits the kind or amount of work.⁹

The third domain is economic resources, which we explore in more detail than the first two domains. We investigate the prevalence among veterans of income from certain sources, namely military pensions, veterans' benefits, and employer-provided pensions. We also consider the prevalence and median amount of Social Security income. Further, we calculate household reliance on Social Security by summing the annual amount of Social Security income for all members of the family and dividing that amount by the sum of annual family income from all sources. Finally, we look at veterans' income security by comparing total family income with the federal official poverty threshold.

Our empirical approach is descriptive, given that our goal is to document the extent to which circumstances differ among aged veterans by education and race/ethnicity, as well as between veterans and nonveterans of the same education and racial/ethnic groups. We do not attempt to examine the net effects of education and race/ethnicity, nor do we attempt to elicit the causal role of military service. All estimates are weighted to allow for generalization of the U.S. adult civilian noninstitutionalized population. Dollar amounts are adjusted to 2014 levels using the Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W).

We note that CPS/ASEC estimates contain both sampling and nonsampling error.¹⁰ Although the CPS/ASEC provides fairly large sample sizes, researchers should nonetheless be cautious in drawing inferences about small differences between subgroups or over time. All comparisons we make below between 1995 and 2015 and across population subgroups within the same year are statistically different at the 90 percent level.¹¹

Furthermore, recent research has found that income from employment-related pensions and retirement accounts tends to be underreported in the CPS/ASEC (Anguelov, Iams, and Purcell 2012; Czajka and Denmead 2012; Gustman, Steinmeier, and Tabatabai 2014; Iams and Purcell 2013; Munnell and Chen 2014). Consequently, estimates of reliance on Social Security among retirees based on CPS/ASEC data may be overstated. Bee and Mitchell (2017) used matched survey and administrative data to estimate the extent to which retirement income is underreported in the CPS/ASEC.¹²

Results

Table 1 compares the sociodemographic characteristics of aged male veterans and nonveterans in three age groups for 1995 and 2015. In 1995, the majority of aged veterans had served in the Korean War or earlier (including World War II). By 2015, veterans aged 55–69 had served mainly during the Vietnam War and the AVF periods. This aging of the AVF-era veterans has important consequences for the changing socioeconomic circumstances and demographic composition of aged veterans, as we discuss below.

The proportions of aged veterans who did not have a high school diploma were relatively low in the years observed. This is due, in large part, to the selective requirements of military service. The proportions of aged veterans and nonveterans who were college graduates increased over the study period, but the rate of growth was slower for veterans. As a result, by 2015, the proportions of college graduates among nonveterans aged 55–61 (32 percent) and 62–69 (38 percent) exceeded those of veterans (24 percent and 26 percent, respectively).

Table 1 also shows an increase in racial and ethnic diversity among aged veterans over time. In 1995, about 90 percent of aged veterans were white, whereas by 2015, the proportion had declined to 72 percent of veterans aged 55–61 and 83 percent of those aged 62–69. Black veterans accounted for most of the corresponding trend; by 2015, black men's share of veterans aged 55–61 (18 percent) exceeded their share of the overall population and of nonveterans in that age group (9 percent). By contrast, they had accounted for only 6 percent of veterans and for 11 percent of nonveterans aged 55–61 in 1995. Likewise, the percentage of aged veterans who were Hispanic roughly doubled between 1995 and 2015, reflecting the increase in the Hispanic share of the general population.

In terms of marital status, older veterans and nonveterans had somewhat similar odds of being married in 2015. Notably, the share of married veterans aged 55–61 declined sharply over the 20-year span, from 81 percent in 1995 to 65 percent in 2015, reflecting downward trends in the shares of married persons in the population at large (Iams and Tamborini 2012). Aged veterans were also more likely to live alone in 2015 than in 1995.

Table 2 presents results for the labor market domain. The estimates suggest notable differences in employment outcomes by education and race/ethnicity. Of veterans aged 55–61, college graduates reported the highest prior-year employment rate in 2015 (82 percent, compared with 63 percent for high school graduates). Of the race/ethnicity groups, black veterans reported the lowest employment rate in 2015, especially in the 55–61 age group (59 percent, compared with 71 percent to 75 percent for the other groups). Furthermore, median earnings among employed veterans aged 55–61 was lower among minorities ($40,000 for black and Hispanic veterans and $45,000 for other races) than for white veterans ($51,000). The table also reveals temporal changes in these patterns. One salient change was the decline over time in prior-year employment for all veterans aged 55–61, from 80 percent in 1995 to 69 percent in 2015.

Table 2 also shows that there are considerable labor market differences between veterans and nonveterans, especially among the two younger age groups, many or some of whom have not yet retired. Overall, aged veterans reported lower prior-year employment rates than nonveterans in 2015 (for example, 78 percent of nonveterans aged 55–61 were employed, compared with 69 percent of veterans). These between-group differences were larger at lower education levels. Most notably, among men aged 55–61 without a high school diploma, 62 percent of nonveterans were employed in 2014, compared with 48 percent of veterans. Among individuals who worked, prior-year median earnings of college-graduate and white veterans aged 55–61 in 2015 ($70,000 and $51,000, respectively) were slightly lower than those of their nonveteran peers ($77,000 and $55,000, respectively). Interestingly, the gaps in median earnings by education (college graduate versus high school graduate) and race/ethnicity (white versus black or Hispanic) were smaller for veterans than for nonveterans in the preretirement (55–61) age group in 2015.

Another indicator of the work domain is the prevalence of a work-limiting disability. As shown in Table 2, the prevalence of work-limiting disabilities among aged male veterans varies sharply by education and race/ethnicity. Among veterans aged 55–61 in 2015, a work-limiting condition was reported by 43 percent of those without a high school diploma, 28 percent of high school graduates, and 8 percent of college graduates; and by 29 percent of black veterans, versus 22 percent of white veterans.

In addition, we observe an increasing prevalence of work-limiting disability among all male veterans aged 55–61 between 1995 (15 percent) and 2015 (23 percent). Notably, this increase was concentrated among veterans without a college degree. For example, among veterans aged 55–61 who did not finish high school, the work-limiting disability rate rose from 29 percent in 1995 to 43 percent in 2015. For those with some college, it increased from 15 percent to 26 percent. The growth in the prevalence of work-limiting disability between 1995 and 2015 was greater for veterans than for nonveterans of the same age. For instance, in 1995, the work-limiting disability rate among men aged 55–61 was 4 percentage points higher for nonveterans than for veterans, but in 2015, the prevalence was 8 percentage points higher for veterans. This pattern is consistent with administrative data showing that the number of veterans with a service-connected disability rose from 2.2 million in 1986 to 3.7 million in 2013, even as the total veteran population decreased (VA 2014b).

Table 3 examines prior-year income sources for men aged 55 or older, focusing on military pensions, veterans' benefits, and employer-provided pensions. Overall, a relatively small share of aged veterans reported income from a DOD military pension (in 2015, around 4 to 5 percent). This is because most individuals with military experience are not career servicemembers and thus do not serve long enough (typically, 20 years) to qualify for a DOD pension. A higher proportion of veterans reported receipt of VA-administered veterans' benefits, which are available to noncareer veterans. For example, almost 18 percent of veterans aged 62–69 in 2015 reported income from veterans' benefits, more than four times the proportion that reported military pension income.

Income-source patterns vary by education and race/ethnicity among veterans. In 2015, veterans aged 55–61 with some college or a college degree were more likely to report prior-year income from a military pension than less educated veterans. Veterans with lower levels of education were also less likely to report income from an employer-provided pension in 2015. This pattern is consistent with the notion that pensions are more likely to be offered at jobs requiring higher education (Tamborini, Purcell, and Iams 2013). Among racial/ethnic groups, black and Hispanic veterans were more likely to report income from military pensions than were white veterans in most age groups in both 1995 and 2015.

Table 4 reports Social Security income characteristics of aged veterans. In 2015, prior-year Social Security income was more prevalent among less educated veterans aged 55–69, and black veterans aged 55–61 were more likely to receive Social Security income than similarly aged white veterans. In the 55–61 age range, Disability Insurance benefits—paid to people who cannot work because they have a medical condition that is expected to last at least a year—likely are the predominant type of Social Security income. Among men of typical retirement-benefit claiming age (62–69), both the prevalence of Social Security beneficiary status and the median family-level Social Security income amounts were modestly higher among white veterans.

With respect to differences between veterans and nonveterans, the estimates show that a higher percentage of veterans lived in families receiving Social Security benefits than did nonveterans. For example, in 2015, almost 74 percent of veterans aged 62–69 reported Social Security income, compared with 56 percent of nonveterans. As expected, the percentage of individuals in families with Social Security income increased with age for both veterans and nonveterans.

An important socioeconomic indicator is reliance on Social Security income. Table 4 shows that Social Security benefits constitute a substantial share of the family income of the typical veteran beneficiary, and the share increases with age. Among black men aged 62–69 in 2015, veterans had a median family-income reliance on Social Security of 34 percent, compared with 56 percent among nonveterans; at ages 70 or older, those figures were 56 percent for veterans and 77 percent for nonveterans. The pattern was similar in 1995.

Table 5 presents median prior-year family income and poverty (and near-poverty) rates of men aged 55 or older. Veterans' economic well-being is strongly related to their education level. Aged veterans with at least some college fared much better than those with less education. The poverty rate among all veterans aged 55–61 rose from 7 percent in 1995 to 11 percent in 2015. In addition, the gap in family income by education had widened among veterans, mirroring increasing income differentials by education in the population at large (Crystal, Shea, and Reyes 2017; Hout 2012).

The income and poverty characteristics of aged veterans also vary by race/ethnicity. In 2015, black veterans of all three age groups exhibited the lowest median family income of the racial/ethnic groups. The percentages of veterans in poverty and near poverty were generally higher for minority veterans than for white veterans. Between 1995 and 2015, poverty and near-poverty rates rose among both black and white veterans aged 55–61.

There are also interesting patterns between veterans and nonveterans. Among veterans aged 55–61 in 2015, those with a high school diploma had lower family income and higher poverty and near-poverty rates than similarly educated nonveterans. Yet across racial/ethnic groups, black and Hispanic veterans aged 55–61 in 2015 had slightly higher family income than their nonveteran peers. Further, black and Hispanic veterans had comparatively lower poverty rates than their nonveteran peers in all age groups. For example, 21 percent of black nonveterans aged 62–69 in 2015 were poor, compared with 13 percent of black veterans. By contrast, the poverty rate for white veterans aged 55–61 in 2015 was 10 percent, compared with 9 percent among white nonveterans of the same age.

Conclusions

Policymakers have long been concerned about the well-being of veterans. In this article, we highlight within-group differences in selected indicators of well-being among aged veterans, most notably across educational and racial/ethnic subgroups. Taken together, the results offer a more mixed and nuanced picture of aged veterans than can be provided by analyzing aged veterans as a whole. With increasing female and minority enlistment in recent decades, the veteran population will become increasingly diverse as the AVF-era cohorts age. Thus, it becomes more important to account for the heterogeneity of the veteran population when addressing concerns about their economic well-being.

Our results reinforce previous findings that prior education is strongly associated with well-being in later life for veterans as well as for the population overall. It may not be surprising that veterans' employment, earnings, and work-limiting disability rates differ substantially across education levels. However, this pattern is increasingly relevant given the declining prevalence of postsecondary education among veterans in recent cohorts of near-retirees and the increasing importance of a college degree for positive lifetime outcomes. The results also shed light on important differences across racial/ethnic groups. On average, aged black and Hispanic veterans exhibit lower employment, higher work-limiting disability, and greater poverty rates than white veterans of the same age.

Our results also show notable differences between aged veterans and nonveterans. For example, aged black and Hispanic veterans appear to fare somewhat better than their nonveteran counterparts in terms of income and poverty. In comparison, aged white veterans tend not to fare quite as well as their nonveteran peers, particularly in the 2015 CPS/ASEC. This finding is consistent with the argument that military service may provide a bridge for individuals from disadvantaged communities into higher-paying jobs after they complete their military service. Nonetheless, one noteworthy outcome is an estimated rise in the poverty rate among all veterans aged 55–61, from 6 percent in 1995 to 11 percent in 2015. This pattern might be related to the military transition to an AVF in 1973 and the long-term sociodemographic changes that resulted.

In closing, we note some limitations of our study. Our descriptive analysis shows relevant socioeconomic differentials among aged veterans, and between veterans and nonveterans, and by education and race/ethnicity; but it does not allow causal conclusions about the effect of military service across these different groups. Further, we have not tested hypotheses of why these differences emerge, nor have we elicited the net effects of the variables under study. The results presented herein may serve as a baseline for future work that accounts for complex interactions between variables. Additionally, the socioeconomic indicators we examined were not exhaustive. For example, we did not explore wealth or total pension resources. Moreover, using panel data would allow researchers to trace the sequence of lifetime events and outcomes for veterans as they transition into later life. Future research also would benefit from addressing why economic insecurity among veterans nearing retirement seems to have increased in recent years, particularly for those from the AVF service era. Including women in the sample of veterans would also be useful, and should become more practical as more recent cohorts with substantial female enlistment reach older ages.