Projections of Disability in the Department of Defense Workforce Through 2031: Estimating Future Assistive Technology Requirements for Department of Defense Civilian Employees and Service Members

Issue

The U.S. Department of Defense (DoD) requires both current and projected estimates of the size of its workforce population with disabilities. These estimates support the strategic goals outlined in the 2018 National Defense Strategy to cultivate “a motivated, diverse, and highly skilled civilian workforce” (U.S. Department of Defense, 2018, p. 8), as well as a range of executive orders directing DoD to hire employees with disabilities and provide them with reasonable accommodations. Furthermore, estimates by type of disability are necessary to determine the assistive technology (AT) required and its anticipated costs through 2031. Overall federal employee disability projections are also derived to provide information pertinent to the Rehabilitation Act of 1973 and multiple executive orders, including Executive Order 14035. Inasmuch as AT may aid in the recovery and retention of injured service members or may aid in the postservice employment of service members medically separating, we also seek to estimate potential demand for AT from these groups. As such, this study develops projections of the DoD civilian employee population, and of injured and wounded service members, with specific disabilities already categorized by DoD's centralized AT procurer (hearing, vision, cognitive, and dexterity disabilities), as well as the potential anticipated requests for AT by these populations and their costs, between 2021 and 2031.

Approach

We take two general approaches to estimate the flow of potential requests: where possible, we use new diagnoses or new disability ratings by disability category among service members as direct estimates of potential requests. For certain service member disability categories, as well as for all DoD civilian workers, new diagnoses or ratings were not available to us, and hence we estimated the population, or “stock,” of all individuals in these categories. Although it is conceivable that, in any given year, this entire population would make AT requests, the academic literature on accommodation requests indicates that only a subset of individuals with particular disabilities requests accommodations in any particular year; as such, we convert our estimated stock of all individuals in these disability categories into estimated flows of potential requests by applying rates of accommodation requests drawn from this literature. These estimated flows still represent the potential requests, but under assumptions we consider more realistic than assuming all individuals annually make new AT requests.

To derive projections of DoD civilians with disabilities from 2021 to 2031, this study uses a combination of two large, long-running nationally representative survey sources (the American Community Survey and the National Health Interview Survey), U.S. Census Bureau population projection data, and administrative data from the Defense Manpower Data Center. We first derive baseline (2021) disability prevalence by age, sex, and race/ethnicity subgroups. Next we derive two projections of the overall federal and DoD civilian workforces through 2031 based on U.S. Bureau of Labor Statistics (BLS) workforce projections and observed trends in the DoD civilian workforce. Baseline demographic composition of each workforce is adjusted throughout the period to reflect anticipated population composition changes in age and race/ethnicity through 2031. Finally, disability prevalence rates by age, sex, and race/ethnicity are applied to the projections to derive estimates of the number of federal and DoD civilian workers with specific disabilities through 2031. We present two scenarios to provide a range of estimates: one based on the BLS's anticipated changes (reduction) in the size of the overall federal workforce, and one based specifically on observed (status quo increasing) trends in the size of the DoD civilian workforce.

Anticipated costs for AT are then derived by first estimating the number of workers who will request accommodations among all workers with each type of disability. Administrative data from the Computer/Electronic Accommodations Program (CAP) are then used to estimate the average number and the average cost of accommodations for a given civilian worker with a given type of disability. Annual and total 2022–2031 cost estimates of AT are presented for both the shrinking workforce and status quo (increasing) workforce scenarios.

CAP procures technologies for two categories of service members. The first is those with health conditions interfering with their deployability and thus resulting in their referral to the Disability Evaluation System (DES) process. We note that entering the DES process begins with a referral to a Medical Evaluation Board; CAP refers to this status as wounded service member-referred to a Medical Evaluation Board (WSM-MEB). CAP also procures technologies for service members with health conditions who have not been referred to the DES, which it refers to as wounded service member-continuing on active duty (WSM-COAD).

For WSM-MEBs, we rely on DES administrative records and publicly available reports to measure the annual numbers of these service members who are potentially eligible to receive CAP AT by disability category given a disability rating for specific health conditions. We use these data sources to construct a measure of each CAP disability category under study, and then use historical case statistics to provide an estimated range for service members undergoing disability evaluation over the coming decade. Our primary scenario is an annual average across the available data from 2011 to 2018, and we provide two alternative annual estimates for each condition: the pre-2016 peak of cases during relatively high-conflict service experiences, and the post-2016 trough of cases during relatively less conflict-intensive periods. We assume that each newly disability rated WSM-MEB may potentially seek out AT, and we apply the CAP-provided average number and cost of technologies per disabling health condition, as in the civilian analysis.

Measuring the prevalence of accommodation-sensitive health conditions among WSM-COADs requires stronger assumptions, given that, unlike those separating via the DES, there are no disability rating system records. We thus focus on diagnoses among active service members of specific conditions that represent the most frequent conditions for which WSM-COADs receive AT from CAP. Using existing publicly accessible documents, including military medical surveillance publications, RAND Corporation reports, and the military medical research literature, we estimate the average, peak, and trough prevalence of these diagnoses from fiscal year (FY) 2016 to FY 2020. As in the civilian analysis, we estimate the number of new requests for accommodations from each medical diagnosis group among WSM-COADs. We then draw on CAP administrative records to apply an assumption of average number of accommodations per request, and the cost per accommodation unit.

Key Findings

Overall, we find that projected annual costs for estimated potential requests are between $14.5 and $18.1 million in 2021 dollars. This range is between three and four times as large as recent annual costs for each of the populations—DoD civilian workers, WSM-COADs, and WSM-MEBs—that CAP procures AT for, but we note again that, by construction, these estimates are larger than requests that any given DoD agency would receive, since they attempt to estimate these direct fulfilled requests, fulfilled requests to any potential accommodation provider, and unfulfilled requests. That is, given historical data, we estimate many more individuals who are accommodation sensitive—those with health conditions that interfere with their work and for whom accommodations would lessen this interference—than CAP alone has procured technology for. Given that we focus on four specific categories of disability that are just a subset of all disabilities, the implication is that the accommodation-sensitive DoD civilian workers and service members are substantially greater in number than indicated by historical CAP procurement.

Our DoD civilian projections include two scenarios to provide a range of estimates: first, on the low end, that there will be a 2.4-percent decrease in the overall size of the workforce under the shrinking workforce scenario (in pace with the overall federal workforce projected by the BLS); or, second (based on observed trends since 2007 in the size of the DoD civilian workforce specifically), that there will be an increase of 12.4 percent in the growing workforce scenario. In both scenarios, the compositional change in the workforce with specific disabilities is the same between 2021 and 2031: we project that the proportion of the DoD civilian workforce with hearing disabilities will decline by 2.4 percent, the proportion with vision disabilities will decline by 0.7 percent, the proportion with cognitive disabilities will increase by 1.2 percent, and the proportion with dexterity disabilities will increase by 1.3 percent. Note that DoD civilian workers with dexterity disabilities are already the largest group of the four and represent more than half of all recent requests for AT by DoD civilian workers from CAP. These trends are generally consistent with the trends projected for the federal workforce, with the exception that the proportion of the federal workforce with vision disabilities is projected to remain constant over time. We use the two workforce size projections to derive a low-cost scenario and a high-cost scenario for projected disabled DoD civilian workers.

In our projections for WSM-COAD and WSM-MEB costs, we also construct a range of scenarios, with a low-cost scenario based on lower prevalence after 2015, a high-cost scenario based on higher prevalence before 2015, and an average of the two to provide a middle-cost scenario.

We provide estimated annual costs in 2021 dollars, and then provide ten-year costs based on Congressional Budget Office (CBO) inflation projections through 2031. Our results are summarized in Table 1, alongside CAP's own average annual costs from 2016 to 2019, inflated to 2021 dollars with the Consumer Price Index for All Urban Consumers.

We note again that our estimates are of potential requests, although they are based on empirical evidence of rates of actual requests in other populations. CAP itself is not the sole provider of accommodations to DoD civilian workers and service members with these disabilities. If the rate at which a given DoD civilian worker requests an accommodation differs from our assumptions (based on estimates from other civilian populations of any requests by workers with a given disability, as well as the assumption that the frequency of these requests are once every two years for hardware or every year for software and services), our estimates would be directly and substantially affected.

But a general finding from our estimates are that we find that costs for civilian workers and service members are higher than historical CAP procurements (in particular, among WSM-MEBs), indicating that the prevalence of disabling conditions in these populations is much higher than rates of current CAP requests.

Key Assumptions

The estimates and projections in this study are necessarily based on a large number of simplifying assumptions based on data limitations and a ten-year projection horizon. Estimates are likely sensitive to all the assumptions. Here we highlight key assumptions for which our projections are particularly sensitive.

For the civilian DoD and federal worker projections, the most fundamental assumptions relate to the projected workforce size trends over time. For the federal workforce, we adopt the BLS projection of a decline in total federal workforce size through 2031. For the civilian DoD workforce size, we assumed two scenarios that reflect hypothetical lower and upper estimates to provide a range; the lower estimate applies the BLS decrease assumption, and the upper estimate is based on the observed status quo trend in DoD civilian workforce size between 2007 and 2021. These two scenarios provide a range of total DoD civilian workforce sizes of between 747,000 and 860,000. To the extent that actual DoD workforce size diverges from either scenario, our estimates will similarly over- or underestimate the number of disabled workers; however, although the assumption of the overall growth of the DoD civilian workforce affects our estimates, the relative turnover in this workforce results in a narrow range for our final cost estimates, as shown in the first row of Table 1.

Another consequential assumption is that the demographic composition of federal and DoD civilian workforces will shift at pace with the demographic shifts projected by the Census Bureau, with decreasing proportions of non-Hispanic white employees and a general shift toward older ages. Differences in disability prevalence by age and race/ethnicity will affect disability prevalence in the workforce as these demographics shift over time. If workforce demographic composition shifts diverge from population level shifts, our estimates will over- or underestimate the number of disabled workers depending on the demographic characteristic and the type of disability.

A third important limitation for our projections of DoD civilians with disabilities is that we have assumed that the prevalence of disabilities in the DoD civilian workforce is the same as in the overall federal workforce. If disability rates differ between DoD civilians and federal workers overall, our estimates will not reflect this. We note that our analysis is limited to full-time civilian DoD workers, which does not include the roughly 4 percent who are part-time, seasonal, or participating in internship programs. Furthermore, if federal or DoD employment policy substantially changes with regard to the hiring or continued employment of workers with disabilities, our projections may diverge from actual growth in employment of these workers.

A pivotal assumption for both our DoD civilian estimates and for the WSM-COAD estimates is how we estimate the fraction of accommodation-sensitive individuals that will request an accommodation each year, and how often. We drew on research based on rates of accommodation in the broader workforce for our disability categories, which itself provides limited information on how often an individual with a given disability seeks an accommodation. To arrive at the frequency of requests, we worked in consultation with CAP to arrive at our assumption of every other year for hardware and every year for software and services; to the extent these rates systematically differ for the populations under study here or the rates of accommodation differ systematically from our estimation approach, our estimates will be biased accordingly.

For our WSM-COAD estimation, we relied on diagnoses to indicate the presence of an accommodation-sensitive health condition. If conditions are systematically underdiagnosed, or if specific diagnoses are indicative of conditions that are less likely to interfere with work, our findings on prevalence would be biased in one direction or the other. To the extent that diagnoses and rates of requests for accommodations differ substantially (e.g., if individuals seek out CAP services instead of medical treatment), then our estimates may be biased accordingly. These biases would particularly affect our dexterity and cognitive disability categories, since these categories represent the majority of costs.

Finally, we note one continuing source of uncertainty: the long-term impacts of COVID-19—namely, postacute sequelae of the virus (so-called long COVID) on rates of disability in the federal and DoD civilian workforces. Applying recent published estimates based on U.S. Department of Veteran Affairs patients to our populations, we find that the overall increase in cost would be approximately 3 percent in each of our scenarios, but we note that the evidence base on the consequences of long COVID for disability rates is rapidly evolving.