History of Subminimum Wage in the United States

FLSA Section 14(c) was introduced in the 1930s in an attempt to ensure people with disabilities were indeed paid based on their performance (Maurer, 2011; Whittaker, 2005). However, since its introduction, Section 14(c) has been rife with conflict. For example, in 1967, at the urging of Congress, the United States Department of Labor conducted an in-depth examination of subminimum wage practices. The report found that sheltered workshops themselves, rather than people with disabilities, were largely to blame for reduced productivity (Whittaker, 2005). Similarly, in 1977, the Department of Labor also testified in a Congressional hearing that subminimum wage was problematic because "'many of the managers of the sheltered workshops are not sufficiently knowledgeable in costing and pricing… essential elements of a successful workshop operation'" (Whittaker, 2005, p. 13).

A few years later (1980), in a House Subcommittee on Labor Standards hearing about Section 14(c), the Department of Labor also,

confirmed that its oversight of the [Section 14(c)] program had been inadequate… The wage rate determination process, it was argued, was inadequate… The Wage and Hour Division [of the Department of Labor] did not have an accurate barometer against which to measure the productivity of the worker [with disabilities]… Nor did employers of [people with disabilities] provide a totally clear picture of who they were and what their mission was. (Whittaker, 2005, p. 14)

Around the same time, a representative from the National Federation for the Blind (NFB), as well as others, including House Representatives themselves, critiqued sheltered workshops for portraying people with disabilities as both clients (i.e. rehabilitation service recipients) and employees. The NFB representative argued,

the sheltered institutions were basically industries that 'have covered their business activities with a veil of 'social services'' and labeled their workforce as 'clients' thereby allowing them to argue for the benefits of special workshops, yet they do 'very little rehabilitation and a whole lot of business and industrial activity.' (Whittaker, 2005, p. 18).

This practice is also problematic as these companies may benefit from Medicaid funding and other public subsidies, while also serving as competitive private sector corporations (Whittaker, 2005).

In 1981, the United States General Accounting Office issued a report on sheltered workshops which found less than one-twentieth of sheltered workshops were investigated annually. Yet of those investigated, 60% underpaid their workers even according to subminimum wage standards (Whittaker, 2005). In an attempt to remedy some of these problems, Section 14(c) of the FLSA was amended and restructured in 1986 so that wage floors (wage minimums) could be removed. This amendment aimed to improve the wages of people with disabilities, so they actually reflected their contributions. Unfortunately, the move away from a wage floor did not go as planned and did not result in people with disabilities being paid more; as a result, it failed to address what ultimately was a structural problem (Whittaker, 2005).

As there were still problems, in 1994, Congressional hearings declared productivity-based subminimum wage inequitable and unworkable; however, ultimately, no changes were made (Whittaker, 2005). Another attempt at reform occurred in 2001 when the General Accounting Office released a new study reporting the Department of Labor had "not effectively managed the special minimum wage program to ensure that 14(c) workers receive the correct wages" (Whittaker, 2005, p. 34); yet still, no changes were made.

Almost a decade later, another attempt was made to remedy subminimum wage problems, this time through legislation. The Fair Wages to Workers with Disabilities Act (H.R. 3086), was introduced to discontinue the issuing of special wage certificates, phase out existing special wage certificates, and repeal Section 14(c). The bill had bipartisan support with 82 cosponsors (65 Democrats, 17 Republicans) and was widely supported by major national disability groups that advocate for many different types of people with disabilities; however, the bill died in the House in 2011 (Civic Impulse, 2016; Madrigal, 2011).

Subminimum Wage Today

Section 14(c) is "rooted in the premise that, in order 'to prevent curtailment of opportunities for employment,' a wage rate 'lower than' the otherwise applicable FLSA minimum may be justified" (Whittaker, 2005, pp. 35-36). Yet, findings from numerous Department of Labor, and General Accounting Office reports, and Congressional hearings have revealed that not only do employers not adequately pay people under subminimum wage, subminimum wage is also discriminatory (Whittaker, 2005). In theory, subminimum wage is supposed to be based on workers' productivity; however, research has found that companies utilizing these certificates always ends up paying below the minimum wage, regardless of their employees' productivity (Maurer, 2011). In fact, there is little evidence of subminimum wage's effectiveness, other than to the bottom line of organizations who wish to profit off cheap labor. For example, Goodwill Industries, which has a total revenue of $5.59 billion, gets donations and government contracts that supplement their business model, while still paying people with disabilities subminimum wage (Dejean, 2017). Moreover, some companies pay people with disabilities a subminimum wage while doing the exact same job as nondisabled workers who are paid above minimum wage (Elk, 2013). This discrimination is often based on misconceptions about the abilities and contributions of people with disabilities (Maurer, 2011).

Research has found many people who at one time were paid subminimum wage demonstrate their ability to work successfully in the community on normal wages once their job is appropriately matched with their skills (Butterworth, Hall, Hoff, & Migliore, 2007). Moreover, supported employees in integrated employment actually generate fewer costs for organizations than those in sheltered workshops (Cimera, 2011). They also do so while making more money and costing businesses less (Cimera, 2011).

In an individualized neoliberal society, such as the United States, subminimum wage is also a barrier to true community participation and citizenship. While people's value should not be tied to their ability to work and it is important to note the role economic structures play in disability oppression (Barnes, 1997; Charlton, 1998; Duggan, 2003; Taylor, 2004), people have the right to work if they want to. People should also have the right to be paid fairly for that work. Yet, in addition to being socially devalued, people on subminimum wages often get stuck in a cycle of poverty. People with disabilities making subminimum wage also lack purchasing power, and as a result, have no effective way to get out of this system and into the integrated work market.

In addition, it is important to recognize that disability and ableism do not exist in a vacuum – not only do people with disabilities have intersecting identities, intersecting systems of oppression also operate simultaneously and often intertwine, not only within the employment arena, but also in the United States at large (Annamma, Connor, & Ferri, 2013; Bell, 2006; Darity & Mason, 1998; Erevelles, 2011; Erevelles & Minear, 2010; Garland-Thomson, 2005; Kumari-Campbell, 2009; National Women's Law Center, 2017; O'Leary, 2017; O'Toole, 2013; Pager & Shepherd, 2008; Quillian, Pager, Hexel, & Midtbøen, 2017; Yearby, n.d.). For example, slavery not only played a key role in shaping and growing America's economy prior to reconstruction, but it also has a lasting legacy on the United States economy today (Beckert & Rockman, 2016). Moreover, people of color, especially people of color with disabilities, continue to face widespread and systemic racism (Annamma et al., 2013; Bell, 2006; Erevelles & Minear, 2010; O'Toole, 2013). In the United States employment discrimination is tied to racism and sexism, in addition to ableism (Bureau of Labor Statistics, 2018a, 2018b; National Women's Law Center, 2017). For example, different people are also funneled into different kinds of work (Reskin, 1988). These different 'types' of work, such as gendered work, "reinforce already existing inequalities in the labour market" (Hughes, McKie, Hopkins, & Watson, 2005, p. 261). Moreover, in addition to a larger systemic pay gap between men and women in general, pay gaps intensify for White people with disabilities, and women of color with and without disabilities (National Women's Law Center, 2017). Ableism, racism, and sexism impact not only who is hired, but in what positions, how much people are paid, how they are treated at their job, and their advancement opportunities.

As subminimum wage is a prominent and problematic issue affecting the lives of many people with disabilities, and as it is linked to discrimination, the aim of this study was to explore how prejudice impacts the use of special wage certificates and subminimum wage across the United States. Since ableism, racism, and sexism intersect, including regarding employment discrimination, we also were interested in exploring how those factors may impact the use of special wage certificates. We had the following research questions:

1. How do ableism, racism, sexism, population size, wealth, and political orientation impact if special wage certificates are used in a region?
2. How do ableism, racism, sexism, population size, wealth, and political orientation impact how many subminimum wage certificates are used in a region?
3. How do ableism, racism, sexism, population size, wealth, and political orientation impact how many people are paid subminimum wage in a region?

To explore these questions, we analyzed data about the use of special wage certificates around the United States, and prejudice data from 4.70 million (M) people.

Subminimum Wage

Data about Section 14(c) special wage certificates were obtained from Department of Labor (2018a, 2018b, 2018c, 2018d). The Department of Labor details every employer in the nation that utilizes special wage certificates, including how many workers are paid subminimum wage under those certificates. Employers' locations were utilized to aggregate the special wage certificates by region – metropolitan statistical area (MSA). The United States Census Bureau (2016) defines an MSA as

a geographic entity delineated by the Office of Management and Budget for use by federal statistical agencies. Metropolitan statistical areas consist of the county or counties (or equivalent entities) associated with at least one urbanized area of at least 50,000 population, plus adjacent counties having a high degree of social and economic integration with the core as measured through commuting ties. (n.p.)

There were 366 MSAs and 48 nonmetropolitan areas (one nonmetropolitan for each state, excluding New Jersey and Rhode Island, which did not have nonmetropolitan areas) in the dataset. The Department of Labor data were then used to determine if each region (MSAs and nonmetropolitan areas) utilized special wage certificates (yes (1); no (0)), how many special wage certificates they used, and how many people they paid subminimum wage.

Ableism, Racism, and Sexism

There are two levels of attitudes: explicit (conscious) attitudes and implicit (unconscious) attitudes (Amodio & Mendoza, 2011; Antonak & Livneh, 2000). As people may feel pressured to conceal their biases, or may be unaware they hold biased attitudes, there are concerns that explicit measures do not capture all attitudes (Amodio & Mendoza, 2011; Antonak & Livneh, 2000). This may be especially true for topics where it is socially taboo to divulge having negative attitudes, such as against people with disabilities. For this reason, much attitude research has shifted towards examining implicit attitudes. Implicit attitudes can relate to automatic processes triggered by external cues and reflect associations between attitudes and concepts; "'implicit' refers to [lack of] awareness of how a bias influences a response, rather than to the experience of bias or to the response itself" (Amodio & Mendoza, 2011, p. 359).

One of the most prevalent methods to measure implicit attitudes is the implicit association test (IAT). The IAT examines people's associations and attitudes by measuring reaction time when items are sorted in stereotype congruent and incongruent ways; the quicker the reaction time, the stronger the association between groups and traits (Karpinski & Hilton, 2001). Scores of 0.15 to 0.34 reveal a slight preference for the social majority group, 0.35 to 0.64 a moderate preference, and 0.65 and greater a strong preference (Aaberg, 2012; Greenwald, Nosek, & Banaji, 2003). Negative values of the same values above reveal preferences for the social minority group, and scores from -0.14 to 0.14 reveal no prejudice (Aaberg, 2012; Greenwald et al., 2003).

Data about implicit ableism, racism, and sexism were obtained from Project Implicit (Xu, Nosek, & Greenwald, 2014), a database where people can test their implicit attitudes. A total of 7.60 M people from all 50 states and the District of Columbia participated in the racism IAT (Black-White IAT), 2.00 M in the sexism IAT (gender-career IAT), and 0.70 M in the ableism IAT (disability attitudes IAT) between 2002 and 2017. Of those 10.28 M total participants, half of those participants (45.9%) also completed demographic information about their residency (region). Those without this information or who lived in United States territories were removed. This resulted in a total n of 4.7 M people (ableism = 0.3 M; racism = 3.5 M; sexism = 0.9 M). There was an average of 3,798 participants per MSA (ableism M = 789; racism M = 8,461; sexism M = 2,143). Participants' residency was then used to aggregate IAT scores by region (MSA), with the MSA's mean score serving as the MSA's ableism, racism, and sexism scores.

Political Orientation

We included political orientation as a variable because of the link between conservatism, business interests, and the status quo in the United States. Research has found, not only do conservatives tend to be more favorable towards corporations than liberals, they are also more likely to favor maintaining the status quo and resist attempts to change it by rationalizing and accepting inequality (Graham, Haidt, & Nosek, 2009; Jost, Nosek, & Gosling, 2008).

Data about participants' political orientation was also obtained from the Project Implicit dataset. Participants were asked to select their political orientation on a seven-point Likert scale from strongly conservative (1) to strongly liberal (7). We followed the same procedure as with the ableism, racism, and sexism scores to create an average political orientation of each MSA using this participant data. The average political orientation per region was 4.23 (SD = 0.38) (neutral), ranging from 3.20 (slightly conservative) (Provo-Orem, UT) to 5.43 (slightly liberal) (Santa Cruz-Watsonville, CA).

Region Size

In order to examine relationships with, as well as control for, region size, each region's population was utilized as a variable. Population data was obtained from the United States Census Bureau (2018) and Kaiser Family Foundation (2017) (nonmetropolitan areas). We utilized the most recent data available (2016). The average region had a population of 755,112 people (SD = 1.52 M), ranging from 54,742 (Carson City, NV) to 20.15 M (New York-Northern New Jersey-Long Island, NY-NJ-PA).

Region Wealth

A common argument in support of subminimum wage is that companies paying subminimum wage cannot afford to pay people minimum wage or above (Whittaker, 2005). Thus, regional wealth was used as a variable. Regional total personal income was used to represent regional wealth. Total personal income is:

…the income received by, or on behalf of, all persons from all sources: from participation as laborers in production, from owning a home or business, from the ownership of financial assets, and from government and business in the form of transfers. It includes income from domestic sources as well as the rest of the world. It does not include realized or unrealized capital gains or losses. (Bureau of Economic Analysis, 2016, n.p.)

Personal income for each of the regions was obtained from the United States Bureau of Economic Analysis (2018). Data from 2016 was utilized as it was the latest available. The average personal income per MSA was $75.14 M (SD = 159.04), ranging from $2.23 M (Carson City, NV) to $1,193.45 M (TX nonmetropolitan area).

Analysis

This study had three research questions. The first question was: How do ableism, racism, sexism, population size, wealth, and political orientation impact if special wage certificates are used in a region? To explore this research question, a binary logistic regression was utilized with the dependent variable (DV) use of special wage certificates (yes (1); no (0)). The independent variables (IVs) included the following: ableism; racism; sexism; political orientation; region wealth (personal income in M); and, population (in 100,000).

The second research question was: How do ableism, racism, sexism, population size, wealth, and political orientation impact how many subminimum wage certificates are used in a region? We ran a multiple linear regression to determine correlates of the number of special wage certificates used in a region per capita (per 100,000) (DV). The IVs included: ableism; racism; sexism; political orientation; and, region wealth (personal income in M). We also controlled for region size by embedding the population into the DV – per capita calculation – to minimize collinearity.

The third research question was: How do ableism, racism, sexism, population size, wealth, and political orientation impact how many people are paid subminimum wage in a region? To explore this research question, a multiple linear regression was utilized. The DV was the number of people paid subminimum wage per 100,000 people in the region; per capita calculations (people paid subminimum wage divided by region population in 100,000) were used to control for region size and minimize collinearity. The IVs were: ableism; racism; sexism; political orientation; and, wealth (personal income in M).

Utilization of Special Wage Certificates

We ran a logistic regression to determine what factors (IVs: ableism, racism, sexism, political orientation, personal income (M), population (100,000)) impacted if a region utilized special wage certificates (DV: use of special wage certificates). The model was significant, χ²(6) = 53.59, -2LL = 366.66, p < .001. The model correctly categorized 77.2% of cases and predicted 19.2% of variance (Nagelkerke R²).

Table 1
Binary Logistic Regression: Odds of Utilizing Special Wage Certificates

Variable	B	S.E.	Wald	df	Odds ratio	95% confidence interval
Ableism	2.94	2.61	1.27	1.00	18.93	0.11	3,141.49
Racism	5.33	3.20	2.78	1.00	207.05	0.39	109,859.04
Sexism	-1.52	4.68	0.10	1.00	0.22	0.00002	2136.58
Political orientation	-0.25	0.35	0.54	1.00	0.78	0.39	1.53
Population (in 100,000)***	0.26	0.07	14.32	1.00	1.30	1.13	1.48
Personal Income (M)	0.00	0.00	0.00	1.00	1.00	0.99	1.01
Constant	-1.34	3.38	0.16	1.00	0.26

According to univariate statistics, population was significant (p < 0.001) (Table 1). Controlling for all other variables, the larger the region, the more likely they were to utilize special wage certificates (figure 1). For example, a region with 100,000 residents has a 25.4% probability of utilizing special wage certificates. In comparison, a region with 1 M residents has a 77.9% probability of utilizing special wage certificates.

Number of Special Wage Certificates Utilized

We ran a multiple linear regression to determine correlates of the number of special wage certificates used in a region per capita (per 100,000) (DV). The IVs included ableism, racism, sexism, political orientation, and state wealth. We also controlled for region size by embedding the population into the DV (per capita calculation). The model was significant, F(5, 380) = 3.79, p = 0.002, R² = 0.047. Racism was significant at t = 2.53, p = 0.012 (Table 2). Wealth (personal income; M) was also significant at t = 2.08, p = 0.038. The more racist a region, the more special wage certificates they utilized (figure 2). For example, two regions of the same average size (755,111), political orientation (4.23), wealth (75.14 M), ableism (0.50), and sexism (0.39) but with different levels of racism (0.35 versus 0.45) are expected to utilize 0.74 and 1.00 subminimum certificates per 100,000 people in the region respectively.

Table 2
Multiple Linear Regression: Certificates Utilized

Variable	B	S.E.	t	95% confidence interval
(Constant)	-0.38	1.10	-0.34	-2.55	1.79
Ableism	0.69	0.88	0.78	-1.04	2.42
Racism*	2.61	1.03	2.53	0.58	4.64
Sexism	1.67	1.57	1.06	-1.42	4.77
Political orientation	-0.20	0.11	-1.80	-0.41	0.02
Personal income (M)*	0.0005	0.0003	2.08	0.00003	0.001

Moreover, the more the region's wealth, the more special wage certificates they utilized (figure 3). For example, two regions of the same average size (755,111), political orientation (4.23), ableism (0.50), racism (0.33), and sexism (0.39) but with different levels of wealth (100 M versus 500 M) are expected to utilize 0.70 and 0.92 subminimum certificates per 100,000 people in the region respectively.

Number of People Paid Subminimum Wage

To examine which factors impact the number of people paid subminimum wage in a region, we also ran a linear regression model between the number of people paid subminimum wage in the region per capita (per 100,000 people) (DV) and ableism, racism, sexism, political orientation, and wealth (IVs). The model was significant, F(5, 380) = 4.26, p = < 0.001, R² = 0.053. The ableism score was significant at t = 2.56, p = 0.011 (Table 3). The racism score was also significant at t = 3.79, p <; 0.001.

Table 3
Multiple Linear Regression: Workers Paid Subminimum Wage

Variable	B	S.E.	t	95% confidence interval
(Constant)	-189.88	93.17	-2.04	-373.08	-6.68
Ableism*	190.22	74.25	2.56	44.23	336.22
Racism***	330.51	87.13	3.79	159.19	501.82
Sexism	132.41	132.86	1.00	-128.82	393.64
Political orientation	-2.16	9.17	-0.24	-20.18	15.87
Personal Income (M)	0.01	0.02	0.24	-0.04	0.05

According to the model, the more ableist and racist a region, the more people that will be paid with subminimum wage (figure 4). For example, two regions of the same average size (755,111), political orientation (4.23), wealth (75.14 M), racism (0.33), and sexism (0.39) but with different levels of ableism (0.35 versus 0.45) are expected to pay 26.66 and 45.68 people subminimum wage per 100,000 people in the region respectively. Moreover, two regions of the same average size (755,111), political orientation (4.23), personal income (75.14 M), ableism (0.50) and sexism (0.39) but with different levels of racism (0.35 versus 0.45) are expected to pay 63.43 and 96.48 people subminimum wage per 100,000 people in the region respectively.

Limitations

When interpreting these findings, it should be noted that people volunteered to participate in the prejudice IATs and, therefore, there is a chance of selection bias. It should also be noted that this was a secondary data analysis and we did not have the ability to ask the participants additional questions. Moreover, we did not explore interactions. It should also be noted our aim was not to examine all potential factors related to subminimum wage.

Conclusion

Subminimum wage hinges on stereotypes about ability and productivity. This study was one of the first of its kind to examine the relationship between subminimum wage and prejudice, revealing a quantitative link between the use of subminimum wage practices, and racism and ableism. This study adds to decades of research, including from the Department of Labor and other government agencies (Whittaker, 2005), as well as calls from advocates with disabilities, highlighting that subminimum wage practices are discriminatory. Since focus on disability is often about how it affects potential workers, normalizing disability oppression helps maintain existing structures of work (Abberley, 1987; Barnes, 1997). These practices must be dismantled in order to end the unjust treatment of people with disabilities.