The employment landscape for workers with disabilities is marked by significant challenges, including disparities in wages, health insurance coverage, and retirement benefits. These inequities contribute to broader social and economic disadvantages, making it crucial to explore mechanisms to enhance this demographic’s employment conditions. One such mechanism is union representation. Unions have long been associated with securing better wages, comprehensive health insurance, and robust retirement plans for their members. This paper delves into the union advantage for workers with disabilities, analyzing how union representation impacts these critical aspects of employment compared to their non-unionized counterparts and workers without disabilities.
This analysis seeks to introduce several new elements to the existing literature. First, it examines a sample that includes years before and since the beginning of the COVID-19 pandemic and includes adjustments for the pandemic era in regression models. Second, it analyzes both wages and access to benefits. Wages are not the only compensation that matters, nor are they the only compensation for which unions must fight. In addition to mean hourly wages, this paper also examines the relationship between union representation and the availability of employer-sponsored health and retirement benefits for workers with disabilities. Finally, recognizing that people with disabilities are not a monolith, this paper disaggregates by type of difficulty whenever possible.
This paper examines several questions about the relationship between unions and compensation for workers with disabilities, including:
The concept of disability can be understood through various models, each providing a different perspective on the nature, causes, and implications of disability. The three primary models are the social, medical, and universal models of disability.
The social model of disability views disability as a result of the interaction between individuals with impairments and societal barriers, rather than the impairment itself being the cause of disability. This model primarily focuses on the individual’s environment rather than the individual themself. Societal barriers may include physical barriers (e.g., inaccessible buildings), attitudinal barriers (e.g., prejudice and stereotypes), and systemic barriers (e.g., discriminatory policies and practices).
The medical model of disability views disability as a problem caused by disease, injury, or health conditions residing primarily in the individual that needs to be cured or managed. The dominant objective is to diagnose, treat, and rehabilitate the individual to reduce or eliminate their impairment. This pathology-oriented model often emphasizes the biological and physiological aspects of disability.
Finally, the universal model of disability, also known as the human rights model, integrates principles from the social and medical models while strongly emphasizing human rights and the dignity of individuals with disabilities. It conceptualizes disability as something that exists on a continuum and may be contingent on environmental factors. Disability is understood within the context of human rights, emphasizing equality, non-discrimination, and the inherent dignity of all individuals, with the understanding that nearly everyone will experience some form of disability at some point throughout their life. Within the universal model, disability may be permanent, temporary, or relapsing and remitting. This model advocates for the removal of societal barriers and the provision of support to ensure full and equal participation in all aspects of life. It focuses on the empowerment and self-determination of individuals with disabilities, ensuring they are fully integrated into society and can exercise agency in their lives.
There is also some debate over person-first versus identity-first language among those with disabilities. Person-first language emphasizes the individual before their disability (e.g., “person with a disability”), while identity-first language places the disability-related identity before the person (e.g., “disabled person”). While some prefer the former as a way to reduce stigma and emphasize shared humanity, it has also been interpreted as implying that there is something negative about having a disability and criticized by some disabled people as diminishing a core part of their identity. In recognition of the passionate arguments for each type of language, this paper uses both forms interchangeably.
The social and universal models of disability helped inform the development of legislation such as the Americans with Disabilities Act (ADA), which aimed to create more inclusive and accommodating systems of employment, infrastructure, and public services. However, while the ADA is an invaluable resource, it has proven insufficient for the task of creating economic justice for workers with disabilities. Employment levels for working-age people with disabilities remain far below those of their nondisabled peers, and workers with disabilities make just $0.77 per dollar earned by workers without disabilities. People with disabilities are also more likely to experience poverty than people without disabilities.
Unions offer a potential salve. First, previous research has found high union wage premiums for disabled workers. A study by Ameri et al. found a union wage premium of 29.8 percent for workers with disabilities, compared to a 23.9 percent union wage premium for workers without disabilities. They also find a small disability pay gap in union workplaces. Another study by Pettinicchio and Maroto found that the union effect on the weekly earnings of workers with disabilities was twice that for workers without disabilities. Disabled union members saw, on average, a 28 percent earnings premium compared to otherwise similarly situated nonunion workers with disabilities. The premium was closer to 35 percent for workers who experienced difficulty running errands or otherwise living independently. They further found that the earnings gap between union and nonunion workers with disabilities was greater than the unionized-nonunionized earnings gap among women, Black, or Hispanic workers.
To that end, unions have also been shown to reduce inequality and improve wages and benefit access across many other groups, including but not limited to younger workers, women, low-income workers, Black workers, and Hispanic workers. While the relationship has varied over time, union density among workers with disabilities is similar to that among workers without disabilities (Figure 1).
Figure 1
The protections enshrined into law by the ADA become hollow if not enforced, and reports suggest that they are regularly flouted. Unions act as a source of employer accountability and may provide an additional enforcement mechanism, ensuring that these hard-won legal protections become a reality. Specifically, unions can help enforce ADA provisions and ensure that accommodation needs that might otherwise go ignored are met in accordance with both the law and collective bargaining agreements (CBAs). Unions are required to duly represent members in disputes, and they provide a structure that allows employees to have issues addressed without personally incurring costly legal expenses. Unions can also assist with workers’ compensation claims, offer emotional support, uphold contact with workers who are away and rehabilitating, offer guidance to employers about the ADA, and take other actions to reduce employer animosity that often accompanies disruptions to the existing state of affairs.
Unions and CBAs give disabled workers access to important resources and opportunities. While there have been instances of accommodation needs conflicting with CBA-established hierarchies, centering the needs of disabled workers during both bargaining and enforcement can circumvent such issues. Unions can also take a universal design approach by reframing accommodations as universal benefits for all staff whenever possible. Flexible work schedules, for example, can benefit and raise productivity for workers with and without chronic illnesses. Remote work opportunities have proven popular among workers with and without disabilities during the COVID-19 pandemic. In fact, the pandemic-era increase in such opportunities revealed how much more accommodating employers could have been before the pandemic.
The number of disabled working-age people in the US has increased since the COVID-19 pandemic began. While the economy has made a strong recovery, the public health crisis—the pandemic part of the pandemic—remains ongoing. Given the risk of Long COVID, some health experts have expressed concern that the ongoing and now largely unmitigated spread of COVID-19 risks disabling a substantial portion of the workforce. The latter is especially salient for people with disabilities, many of whom are at increased risk of negative health complications should they be infected with SARS-CoV-2, the virus that causes COVID-19. This makes the potential role of unions in increasing access to health care and retirement coverage especially salient – the former to address COVID-related health issues and the latter to cushion the financial blow should the person need to stop working.
The analysis uses pooled public-use microdata from the Current Population Survey Outgoing Rotation Group (CPS ORG) and Annual Social and Economic Supplement (CPS ASEC) for 2016 through 2023. The wage advantage portion of the analysis utilizes merged CPS ORG data from the US Census with CPS ORG extracts from the Economic Policy Institute. The benefit coverage portion of the analysis uses harmonized CPS ORG and CPS ASEC extracts from IPUMS.
The Current Population Survey (CPS) determines disability status using a series of six yes-or-no questions from the Department of Health and Human Services (HHS) known as the Standard Short Set. The respondent is asked whether they experience “serious difficulty” (1) hearing, (2) seeing, (3) concentrating or making decisions, (4) walking or climbing stairs, (5) dressing or bathing, and (6) doing errands alone. Research suggests that large swathes of the disabled population are uncaptured by both the Standard Short Set and its closest competitor, the Washington Group Short Set (WG-SS). A comparison across survey sets suggests that the CPS in particular may underestimate disability prevalence, even compared to other national US surveys that use the same question set. However, while other publicly available US surveys contain information on disability status, the CPS is one of the few that also includes detailed information on employment, union representation, wage and salary income, and employer-sponsored health insurance and retirement plan coverage.
CPS ORG, also known as the Earner Study, utilizes a rotation pattern whereby households are included in the survey for four consecutive months, exit the survey for eight consecutive months, and then rotate back into the survey for another final four consecutive months. Workers in the survey are asked questions about their earnings and union representation when they rotate out in months four and eight.
CPS ASEC, also known as the March CPS, is an annual supplement to the CPS’ basic monthly labor force data. CPS ASEC includes information on work experience, income, noncash benefits, and poverty status, among other things. The reference period for questions regarding health insurance and retirement plan coverage in CPS ASEC is the previous calendar year.
All regressions include adjustments for age, age squared, education (less than high school, high school, some college, bachelor's degree, and advanced), gender (when men and women appear in the same regression), race/ethnicity (when those of different races/ ethnicities appear the same regression), year, state, COVID-19 pandemic status, US citizenship, type of disability (when those with different types of difficulty appear in the same regression), presence of multiple types of disability (when disabled people are included in a regression), part-time hours, public sector, and two-digit occupation and industry. Though related works on this subject, such Ameri et al. (2019) and Pettinicchio and Maroto (2020), use more detailed four-digit occupation codes, the use of such codes would not be feasible to assess union benefit premiums for disabled workers given sample size limitations.
The regressions use mutually exclusive categories for racial and ethnic groups. White includes those who identified exclusively as non-Hispanic white. Black encompasses non-Hispanics who identified as Black, including those who also identified as other races. Hispanic includes all respondents who indicated they were Hispanic, including those who identified as other races. Asian/ Pacific Islander refers to non-Hispanic Asians, Hawaiians, and Pacific Islanders who did not also identify as Black. Other comprises those who identified as American Indians or Alaskan natives and did not identify as Black, Hispanic, Asian, Hawaiian, or Pacific Islander, as well as those who identified exclusively as non-White, non-Black, and non-Asian races.
Types of difficulty are not mutually exclusive and each category may include those who also experience other types of difficulty. These categories of difficulty correspond to the six questions mentioned in the previous section. They include hearing difficulty, vision difficulty, cognitive difficulty, ambulatory difficulty, personal care difficulty, and independent living difficulty. Those categorized as having multiple types of disability must report at least two different types of difficulty, but may report more than two.
Regression-adjusted union wage premiums are calculated using ordinary least squares. Union wage premiums in percent are converted from log points by taking the antilog of regression coefficients and subtracting one. Union wage premiums in dollars are calculated relative to mean non-union wages for a given group. Dollars for all years are adjusted for inflation using the CPI-U-RS. Imputed wages are excluded from the analysis to avoid the match bias noted by Bollinger and Hirsch (2006). Coverage regression results for subgroups are reported if they are significant at the one percent level with a sufficient sample size to perform the analysis.
In 2023, the CPS introduced new wage rounding rules as a privacy protection measure. When these rounding rules were retroactively applied to other years in the sample, the difference in results was negligible; for the sake of consistency, the results discussed in this report reflect the application of such rules. Alongside the new rounding rules, the CPS also introduced new changes to its top-coding. This analysis adapts a set of adjustments developed by John Schmitt to create a consistent hourly wage series so that the adjustments reflect both the new top-coding rules and the new rounding rules. All wage data in this analysis reflect the application of said adjustments.
Employer-sponsored benefit coverage regressions are probits. Union health insurance and retirement coverage increases in percent terms are relative to the current coverage rates for non-union workers. Coverage regression results for subgroups are reported if they are significant at the five percent level with a sufficient sample size to perform the analysis.
Replication files and detailed appendix tables are available on GitHub.
Between 2016 and 2023, workers with disabilities enjoyed significant union premiums regarding mean hourly wages, employer-sponsored health insurance coverage, and employer-sponsored retirement plan coverage (Figure 2).
Figure 2
<
The Wage Advantage for Union Workers with Disabilities
Table 1 shows union density and the regression-adjusted percent and dollar mean hourly earnings premiums for workers with and without disabilities. The table includes this information for all employed adults and for various subgroups, including by gender, race/ethnicity, type of difficulty (for those with disabilities), and for adults of prime working age (ages 25–54). Results are only shown for subgroups with results significant at the one percent level and for which the sample size was sufficient to perform the analysis. Detailed results can be found in Appendix Tables 1 through 5.
Table 1
As Figure 3 shows, the regression-adjusted mean hourly wage for a disabled worker represented by a union was $31.99. This was substantially higher than the wage for a disabled worker not represented by a union ($27.18), but lower than the wage for a non-union worker without a disability ($34.78).
Figure 3
Union wage premiums persisted across types of difficulty among disabled workers (Figure 4). Those with personal care difficulties experienced the largest adjusted percent boost, at 23 percent, followed by those with hearing difficulties at 19.1 percent. This corresponded with a dollar increase of $6.51 for workers with personal care difficulties and $6.05 for workers with hearing difficulties. Those who experienced ambulatory difficulty had the smallest percent boost, at 15.4 percent, corresponding to a $4.02 increase in the mean hourly wage.
Figure 4
Among those of prime working age (ages 25 to 54), the regression-adjusted mean hourly wage was nearly 18 percent higher for a disabled worker with union representation than for a disabled worker without union representation. But as Figure 5 shows, the union wage bump was not sufficient to eliminate the disability pay gap for this age group.
Figure 5
Unions boosted wages for both men and women, though the premium was larger for men. Unions also increased wages for workers with disabilities across racial and ethnic groups. Compared to their non-union counterparts, Hispanic workers with disabilities saw a larger percent increase than any other racial or ethnic group. The effect was such that the regression-adjusted mean hourly wage for union Hispanic workers with disabilities was higher than the mean hourly wage for nonunion Hispanic workers without disabilities. (Figure 6)
Figure 6
The Benefits Advantage for Union Workers with Disabilities
As Figure 7 shows, even after adjusting for other characteristics, disabled union workers were much more likely than disabled nonunion workers to have employer-sponsored health insurance (69.7 percent vs. 49.3 percent) and retirement benefits (46.5 percent vs. 30.1 percent).
Figure 7
Table 2 and Table 3 show union density, the regression-adjusted percentage point union premium, and the percent increase in employer-sponsored health insurance and retirement coverage, respectively, for workers with and without disabilities. The table includes this information for all employed adults and for various subgroups, including by gender, race/ethnicity, type of difficulty (for those with disabilities), and for adults of prime working age (ages 25–54). Results are only shown for subgroups with results significant at the five percent level and for which the sample size was sufficient to perform the analysis. Detailed results are available in Appendix Tables 6 through 10 for health insurance coverage and Appendix Tables 11 through 15 for retirement coverage.
Table 2
Table 3
Union representation was associated with regression-adjusted coverage increases of 40.4 percent for employer-sponsored health insurance and 80.7 percent for employer-sponsored retirement plans among disabled workers of prime working age, those aged 25 to 54. The union boost nearly eliminated the health insurance coverage gap (Figure 8) and reversed the retirement coverage gap (Figure 9) between workers with and without disabilities in this age group.
Figure 8
Figure 9
As with wages, unions boosted benefits coverage across gender lines, but the effect was larger for men. There was not a statistically significant relationship between union representation and benefit coverage for Black workers. For Hispanic workers, however, there were substantial and significant regression-adjusted increases in both employer-sponsored health insurance and retirement plan coverage (Figure 10). No significant disability benefits gap existed among union or nonunion workers in this subgroup, but the union benefit advantage differential meant that unionized disabled Hispanic workers were more likely to have benefits coverage than their non-disabled counterparts.
Figure 10
These results demonstrate clear union wage and benefit premiums for workers with disabilities. However, previous work by Ameri et al. (2019) and Pettinicchio and Maroto (2020) found higher wage premiums using earlier sample years. Pettinicchio and Maroto (2020) also found a larger differential between the union wage premium for workers with and without disabilities, with the former exceeding the latter to such an extent that it significantly mitigated the gap in pay between the two groups. The change in the relative union premiums for these two groups is especially relevant to unions’ ability to mitigate economic inequality with respect to disability. Further research should explore why the disabled union wage premium has both declined overall and declined relative to the union premium for workers without disabilities. In particular, this research should make use of additional sample years to better interrogate the potential impact of the COVID-19 pandemic, both as an ongoing public health event and as a period of economic crisis and recovery.
While these results show a clear association between union representation and increased wages and benefit access, qualitative research involving specific contract language could shed additional light on the mechanisms underlying these premiums. In particular, the extent and effect of universal design practices in contributing to these premiums and in fostering more inclusive workplaces should be explored.
This analysis of the relationship between union representation and compensation for workers with disabilities between 2016 and 2023 reveals clear advantages regarding wages, health insurance coverage, and retirement benefits. The regression-adjusted mean hourly wage for unionized disabled workers was significantly higher at $31.99 compared to $27.18 for their non-union counterparts. This wage advantage was consistent across various types of disabilities, with those experiencing personal care and hearing difficulties receiving the most substantial wage premiums of 23.0 percent and 19.1 percent, respectively. Among prime working-age individuals (ages 25 to 54), unionized disabled workers enjoyed an 18 percent higher mean hourly wage than their non-union peers, though this was not enough to eliminate the disability pay gap.
Union representation also substantially increased the likelihood of disabled workers having employer-sponsored health insurance and retirement benefits. Even after adjusting for other characteristics, disabled union workers were more likely to have access to health insurance (69.7 percent vs. 49.3 percent) and retirement benefits (46.5 percent vs. 30.1 percent) than non-union disabled workers. Additionally, union representation was linked to significant increases in employer-sponsored health insurance (40.4 percent) and retirement plan coverage (80.7 percent) among disabled workers of prime working age.
The benefits of union representation extended across gender and racial/ethnic lines, although the magnitude of these benefits varied. Disabled men experienced larger wage and benefit premiums than disabled women. Though unions corresponded with an increase in regression-adjusted wages for all racial and ethnic groups, the impact on benefit coverage was less uniform. Hispanic workers with disabilities saw the most pronounced gains in both wages and benefit coverage, mitigating or reversing disability-based gaps among this group. However, while unions boosted wages among Black workers with disabilities, there was no statistically significant relationship between union representation and benefit coverage for disabled workers in this racial group.
Union representation provides a substantial economic advantage for workers with disabilities, enhancing wages and access to essential benefits such as health insurance and retirement plans. These findings underscore unions' critical role in promoting workplace equity and economic security for disabled workers. Future policies and advocacy efforts should consider strengthening union support and representation as a strategy to address the persistent disparities faced by disabled workers in the labor market.
The author would like to thank John Schmitt and Sylvia Allegretto for their immensely helpful feedback and guidance.