Disability and Labor Market Earnings

Abstract

Over one million Americans aged 15 years and older are deaf or hard of hearing. These individuals may face barriers to and within the labor market, leading to lower employment rates and reduced earnings compared with their counterparts without a hearing disability. Our study contributes to the sparse literature on the relationship between hearing disability and labor market outcomes by examining “hearing earnings gaps,” namely, earnings gaps between individuals who are deaf or hard of hearing and their counterparts without a hearing disability. Using a sample of 25- to 40-year-old full-time year-round workers from the 2011 American Community Survey, we estimate separate earnings equations by hearing ability and gender using generalized estimating equations. For both men and women, Blinder–Oaxaca decompositions indicate that roughly 40% of the overall hearing earnings gap is attributable to differences in educational attainment, potential experience, race/ethnicity, and marital status. The remaining 60% may reflect differences in communication skills and other unobservable characteristics, occupational segregation, labor market discrimination, and stigma.

Keywords

accommodations ADA economics/social security employment labor policy

According to the U.S. Census Bureau, 1.1 million Americans aged 15 years and older were deaf or unable to hear a normal conversation in 2010 (Brault, 2012). Deaf adults are more likely than individuals without a disability to be unemployed or underemployed (Punch, Hyde, & Creed, 2004). For example, in 2010, among individuals aged 21 to 64 years in the United States, the employment rate for those reporting severe hearing difficulty was 41.4% compared with 79.1% for those without a disability (Brault, 2012). Among employed individuals in this age group, median monthly earnings for those with severe hearing difficulty were US$1,943, compared with US$2,724 for those without a disability (Brault, 2012). While the Census Bureau uses a self-reported measure of hearing disability, the Medical Expenditure Panel Survey (MEPS) identifies medical conditions such as hearing loss using the International Classification of Diseases code, a diagnostic tool developed by the World Health Organization (WHO; 2016). Despite differences in disability measures, statistics from these two sources are similar. In a sample of 18- to 64-year-old workers from the 2006 and 2008 MEPS, mean annual earnings were US$23,481 and US$31,272, respectively, among those with and without hearing loss (Jung & Bhattacharyya, 2012).

Although individuals with severe hearing disability have lower employment rates and lower median earnings than those without a disability, they enjoy higher employment rates but lower median earnings than individuals with severe difficulty seeing or those using a wheelchair. For example, in 2010, employment rates were 34.0% and 18.3% among individuals with severe difficulty seeing and those using a wheelchair, respectively, compared with 41.4% among those with severe difficulty hearing. In contrast, median monthly earnings were US$2,564 and US$2,404, respectively, among those with severe difficulty seeing and those using a wheelchair, compared with US$1,943 among those with severe difficulty hearing (Brault, 2012). Disparities in employment rates and earnings across disability categories suggest that labor market barriers facing individuals who are deaf or hard of hearing may differ from those facing individuals with other disabilities, highlighting the importance of separate analyses by disability category. In this study, we focus on labor market barriers facing those who are deaf or hard of hearing and the impact of these barriers on annual earnings.

Several factors may contribute to disparities in employment and earnings by hearing ability. Individuals who are deaf or hard of hearing face barriers to and within the labor market as a result of reduced educational attainment (Hogan, O’Loughlin, Davis, & Kendig, 2009; Woodcock & Pole, 2008). For example, based on data from the 2011 American Community Survey (ACS), 16.0% of working-aged individuals who were deaf or had serious difficulty hearing had completed a bachelor’s degree or more, compared with 31.2% of working-aged individuals without a disability (Erickson, Lee, & von Schrader, 2012). Other labor market barriers facing individuals who are deaf or hard of hearing include challenges communicating in the workplace (Winn, 2006), occupational segregation (Punch et al., 2004; Woodcock & Pole, 2008), stigma (Winn, 2006), and labor market discrimination (Bowe, McMahon, Chang, & Louvi, 2005; Woodcock & Pole, 2008). Like educational attainment, workplace communication skills are a form of human capital that may influence productivity and, in turn, earnings. While occupational segregation may partially reflect differences in preferences between individuals who are deaf or hard of hearing and their counterparts without a hearing disability, it may also reflect barriers to employment in certain occupations. Finally, stigma and discrimination may limit labor market opportunities for deaf and hard-of-hearing individuals. Some organizations and employers may not be willing to provide accommodations such as interpreters (Bowe et al., 2005; Punch et al., 2004; Woodcock & Pole, 2008). Unlike building a ramp to accommodate employees in wheelchairs—a one-time fixed cost—providing interpreters involves ongoing costs. Given the high cost of ongoing interpreting services, some employers might offer deaf employees less expensive but also less effective accommodations, reducing deaf employees’ productivity and, in turn, their earnings. In addition, misperceptions of individuals with disabilities in general and of deaf and hard-of-hearing individuals in particular (Winn, 2006) may influence hiring and promotion practices. Finally, in light of concerns about possible discrimination, deaf and hard-of-hearing employees may be reluctant to negotiate starting salaries and raises or to move from one organization to another given the difficulty finding jobs where they feel secure and where they receive accommodations (Prilleltensky & Gonick, 1996).

Despite evidence of employment and earnings gaps between individuals who are deaf or hard of hearing and those who are not, only a few empirical studies examine the effect of hearing disability on employment after controlling for other relevant characteristics such as educational attainment. Thus, the literature provides scant evidence on differences in employment across individuals with different hearing abilities but otherwise similar characteristics. Using data from the 1978 Survey of Disability and Work and the 1979 Health Interview Survey, Stern (1989) measured the effects of disability on labor force participation. His models include a broad range of disabilities including deafness and hearing disability. Using data from the 2000–2001 Canadian Community Health Survey (CCHS), Woodcock and Pole (2008) estimated logistic models of three dimensions of labor force status, namely, any employment over the past 12 months, underemployment, and employment in a professional or managerial occupation. In addition to a dummy variable indicating deafness or hearing loss, all three of their models control for age and sex, while the third also controls for educational attainment. Using 2006 and 2008 MEPS data, Jung and Bhattacharyya (2012) estimated a logistic model of employment to examine the relationship between hearing loss and employment after controlling for demographic characteristics and comorbidity.

While evidence on employment gaps by hearing ability is sparse, to our knowledge, only one econometric analysis provides evidence on the relationship between hearing ability and earnings, after controlling for other relevant characteristics. With their subsample of employed individuals, Jung and Bhattacharyya (2012) estimated the earnings gap between those with and without hearing loss after controlling for demographic characteristics and comorbidity. They reported that adults with hearing loss earn significantly less than those without hearing loss.

Our study contributes to this literature by providing further evidence on the earnings gap between individuals who are deaf or hard of hearing and those without a hearing disability—what we call the hearing earnings gap. In particular, our research addresses the following pair of related questions: What proportion of the hearing earnings gap can be explained by differences in educational attainment and other observable characteristics? What proportion of the gap might be attributable to less visible barriers such as challenges communicating in the workplace, occupational segregation, stigma, and labor market discrimination? The answers to these questions have important policy implications. For example, if differences in educational attainment explain a substantial portion of the gap, policies that enhance educational opportunities for deaf and hard-of-hearing individuals may be an effective way to reduce the hearing earnings gap. On the contrary, the greater the portion of the gap that is not explained by educational attainment and other observable characteristics, the greater the evidence in favor of other explanations for the hearing earnings gap and, in turn, the greater the need for policies that facilitate communication in the workplace between deaf and hard-of-hearing individuals and their colleagues, customers, and clients without a hearing disability as well as policies that reduce occupational segregation, stigma, and labor market discrimination.

To address these questions, we estimate earnings equations by hearing ability and gender. Using our parameter estimates, we decompose the overall male and female hearing earnings gaps into explained and unexplained components using the Blinder–Oaxaca method (Blinder, 1973; Oaxaca, 1973). In addition, we present estimated earnings for a set of prototypical individuals of the same gender and race/ethnicity but different hearing abilities. Finally, we offer insights on public policies designed to reduce labor market barriers for individuals who are deaf and hard of hearing.

Method

Data

We use data from the 2011 ACS to examine earnings gaps between individuals who are deaf or hard of hearing and those who are not. The 2011 ACS is a nationally representative sample of 2,128,104 households with a response rate of 97.6%. The household survey includes questions pertaining to each household member’s demographic characteristics, labor market activity, and health status. Our analysis relies on data collected on individuals living in households but not on those living in group quarters. In light of increased rates of hearing loss in middle age (Holt, Hotto, & Cole, 1994; Stevens et al., 2013) and the possibility that those who are already established in their careers may experience smaller earnings penalties as their hearing deteriorates than do those with congenital hearing loss or with hearing loss that occurs prior to adulthood, we restrict the sample to household members aged 25 to 40 years. The ACS provides a dichotomous measure of hearing disability, namely, whether or not the individual is deaf or has serious difficulty hearing. Given the limited number of individuals in the deaf and hard-of-hearing subsample who identify their race and ethnicity as other than White, Black, Asian, or Hispanic, we restrict our sample to these four racial and ethnic groups. In addition, given our goal of examining earnings differences by hearing ability and the reporting of earnings in the ACS on an annual basis (wages, salary, commissions, bonuses, tips, and self-employment income during the past 12 months), we restrict our sample to full-time year-round (FTYR) workers. We define FTYR workers as individuals who report positive earnings over the past year, who worked at least 40 of the past 52 weeks (including paid vacation, paid sick leave, and military service), and who worked at least 35 hr per week in a usual work week over this period. Our sample consists of 1,443 deaf or hard-of-hearing men, 172,048 men without a hearing disability, 728 deaf or hard-of-hearing women, and 134,877 women without a hearing disability.

Table 1 displays descriptive statistics by gender and hearing ability for our selected sample. For both men and women, mean earnings are lower among respondents who are deaf or hard of hearing than among those who are not. These differences are significant both statistically (p < .001) and economically: While men who are deaf or hard of hearing earn 16.4% less on average than men without a hearing disability (US$45,244 vs. US$54,137), the hearing earnings gap is 14.2% (US$37,181 vs. US$43,318) among women.

Table 1.

Descriptive Statistics by Gender and Hearing Status.

Variable	Men		Women
Variable	Deaf or hard of hearing	Not deaf or hard of hearing	Deaf or hard of hearing	Not deaf or hard of hearing
Earnings in US$, M (SD)	45,244^a*** (34,237)	54,137 (48,475)	37,181^a*** (32,236)	43,318 (33,593)
Educational attainment (%)
High school degree	31.1^b***	24.4	20.6^b***	16.3
Some college	28.5	21.5	25.5	20.6
Associate’s degree	10.8	8.7	11.4	11.1
Bachelor’s degree	13.4	24.4	21.8	29.7
Master’s degree	4.2	8.2	10.2	13.5
Professional degree	0.8	2.4	1.8	2.8
Doctoral degree	0.9	1.5	1.0	1.5
Race/ethnicity (%)
White	77.0^b***	69.1	74.0^b***	67.0
Black	5.1	7.4	10.4	11.9
Asian	2.4	6.2	3.0	6.4
Hispanic	15.6	17.4	12.5	14.6
Age in years, M (SD)	33.6^a*** (4.6)	32.9 (4.6)	33.4^a*** (4.7)	32.6 (4.6)
Married (%)	59.1	60.3	47.8^c**	53.3
Sample size	1,443	172,048	728	134,877

Note. Hypothesis tests compare deaf or hard of hearing men (women) with men (women) without a hearing disability.

The t test for difference between two means. ^bχ² test of independence. ^cTest for difference between two proportions.

p < .05. **p < .01. ***p < .001.

Among both men and women, chi-square tests indicate that educational attainment is not independent of hearing ability (p < .001). Consistent with evidence reported in the literature (e.g., Hogan et al., 2009; Woodcock & Pole, 2008), men and women who are deaf or hard of hearing display lower overall levels of educational attainment than their counterparts without a hearing disability.

Among both men and women, chi-square tests also indicate that race/ethnicity is not independent of hearing ability (p < .001). Consistent with evidence that hearing disability is more prevalent among Whites than other racial and ethnic groups (Holt et al., 1994; Lin et al., 2012), the deaf and hard-of-hearing subsamples are disproportionately White.

The mean age is roughly 33 years, ranging from 32.6 to 33.6 years, in all four subsamples. Among men and women, individuals who are deaf or hard of hearing are less likely to be married, but the difference is statistically significant only in the case of women.

Econometric Model

Following the convention in labor economics (e.g., Choi, Joesch, & Lundberg, 2008; Hanushek & Woessmann, 2008), the dependent variable in our model is the natural logarithm of annual earnings. This functional form, known as a log-level model, corrects for skewness in the earnings distribution. A log-level model is also appropriate in situations where percentage differences in the dependent variable (e.g., earnings) across groups (e.g., high school graduates vs. individuals without a high school degree) are more relevant than unit (e.g., dollar) differences.

Our model includes two measures of human capital, namely, educational attainment and potential experience. In particular, our models distinguish among eight levels of educational attainment: less than a high school degree (the reference category), a high school degree or its equivalent, some college, an associate’s degree, a bachelor’s degree, a master’s degree, a professional degree, or a doctoral degree. In the absence of data on actual labor market experience, our models include age as a measure of potential labor market experience. Although potential experience may differ from actual experience, actual experience is endogenous if those with high earnings potential gain more experience. Age enters the model in a quadratic form; specifically, the model includes age in years and age squared divided by 100. Given evidence of differences in labor market outcomes by race and ethnicity (e.g., Caiazza, Shaw, & Werschkul, 2004), our models distinguish among non-Hispanic White (the reference category), non-Hispanic Black, non-Hispanic Asian, and Hispanic individuals. Finally, given the possibility that marriage influences productivity, that employers discriminate on the basis of marital status, or that marital status signals unobservable characteristics valued by employers (Chiodo & Owyang, 2002, 2003), our models control for marital status.

Thus, for each of our four subsamples, we estimate a model of the form:

ln Earnings = α + X ’ β + Z ’ γ + ε,

where X represents a vector of dummy variables indicating educational attainment and Z represents a vector that consists of age, age squared divided by 100, dummy variables indicating race/ethnicity, and marital status.

Estimation Method

We estimate separate earnings equations for four subsamples of FTYR workers: men who are deaf or hard of hearing, men who are not deaf or hard of hearing, women who are deaf or hard of hearing, and women who are not deaf or hard of hearing. This approach allows for the possibility that earnings equations vary by hearing ability and gender. For example, individuals who are deaf or hard of hearing may enjoy smaller premia for higher education than their counterparts without a hearing disability.

Given concerns about self-selection into the labor market as well as employment barriers facing those who are deaf or hard of hearing, our initial model controlled for selection into FTYR employment using Heckman’s two-step procedure (Heckman, 1979). The first step of this procedure involves estimation of a selection equation—in our case, selection into FTYR employment. The inverse Mills ratio, a term obtained from the selection equation, enters the outcome equation—in our case, the log earnings equation—to control for selection into FTYR employment.

Our employment equation included the same measures of human capital and demographic characteristics as our log earnings equation. In addition, for identification purposes, our employment equation included complementary household income, defined as household income excluding the individual’s own earnings, a variable that may influence the individual’s labor force participation decision but presumably not his or her earnings (De Coulon, 2001).

However, we observed high correlations between the inverse Mills ratio and the regressors in our log earnings equations, particularly in the female subsamples, suggesting that characteristics influencing FTYR employment are highly correlated with those influencing earnings. As discussed in Puhani (2000), in situations characterized by a high correlation between the exogenous variables in the outcome and selection equations, subsample ordinary least squares (OLS) may be a more robust estimator than Heckman’s limited- or full-information maximum likelihood estimator (Heckman, 1979). Accordingly, our earnings equations do not control for selection into FTYR employment.

As discussed earlier, the ACS was administered to households, but the unit of observation in our model is an individual. Thus, our data are clustered at the level of the household. In cases where two or more members of a household are included in one of our subsamples, the unobservables influencing earnings may be correlated across members of the household. More generally, in the case of clustered data, coefficient estimates are inefficient and standard errors may be underestimated. However, the method of generalized estimating equations (GEEs) yields efficient estimates and empirical (robust) standard error estimates (Hanley, Negassa, Edwardes, & Forrester, 2003). Intuitively, “the GEE approach uses weighted combinations of observations to extract the appropriate amount of information from correlated data” (Hanley et al., 2003, p. 365). Accordingly, we use GEEs rather than OLS to estimate the parameters of our earnings equations.

Blinder–Oaxaca Decompositions

Following the approach developed by Blinder (1973) and Oaxaca (1973), we decompose the male and female hearing earnings gaps into explained and unexplained components. The explained component of the gap represents the portion of the gap that stems from differences in observable characteristics—in this case, differences in educational attainment, potential experience, race/ethnicity, and marital status—between the average individual with and without a hearing disability. The unexplained component represents the portion of the gap that stems from differences in intercepts and coefficients across earnings equations (Yun, 2008). For example, as a result of differences in workplace communication skills, occupational segregation, stigma, and labor market discrimination, individuals without a hearing disability may enjoy greater returns to a college education than their deaf and hard-of-hearing counterparts.

Adjusted Hearings Gaps

Following the Blinder–Oaxaca decompositions, we report and discuss predicted earnings for a set of prototypical individuals to provide intuition on adjusted hearing gaps, namely, differences in earnings between individuals with different hearing abilities but otherwise identical characteristics.

Results

Parameter Estimates

Table 2 presents estimates of our log earnings equations for each subsample. As expected, our estimates suggest that earnings are positively associated with educational attainment for men and women with and without a hearing disability. As illustrated in Figure 1, with one exception (men with a doctoral degree), the estimated earnings premium for each level of educational attainment relative to the lowest level—no high school degree—is higher among men without a hearing disability than among their deaf or hard-of-hearing counterparts. For example, among men who are not deaf or hard of hearing, those with a master’s degree earn on average about 142% more than those without a high school degree after controlling for age, race/ethnicity, and marital status; among men who are deaf or hard of hearing, the estimated premium for a master’s degree is approximately 103%. The difference between these two premia is statistically significant (p < .05). Similarly, the premium associated with a professional degree is statistically significantly (p < .05) greater among men without than among men with a hearing disability (227% vs. 134%). Among men, differences in earnings premia associated with other levels of educational attainment are not statistically significant at the 5% level. As shown in Figure 2, the estimated earnings premium for each level of educational attainment is higher among women who are not deaf or hard of hearing than among their deaf or hard-of-hearing counterparts, but none of the differences in estimated premia is statistically significant.

Table 2.

Estimated Log (Earnings) Equations by Gender and Hearing Status.

Parameter estimate	Men		Women
Parameter estimate	Deaf or hard of hearing	Not deaf or hard of hearing	Deaf or hard of hearing	Not deaf or hard of hearing
Intercept	10.39*** (0.80)	8.48*** (0.08)	8.84*** (1.13)	7.34*** (0.43)
Educational attainment
High school	0.14* (0.06)	0.23*** (0.01)	0.17 (0.10)	0.25*** (0.01)
Some college	0.27*** (0.06)	0.38*** (0.01)	0.23* (0.10)	0.40*** (0.01)
Associate’s	0.35*** (0.06)	0.44*** (0.01)	0.44*** (0.11)	0.54*** (0.01)
Bachelor’s	0.60*** (0.07)	0.73*** (0.01)	0.66*** (0.10)	0.78*** (0.02)
Master’s	0.71*** (0.09)	0.88*** (0.01)	0.82*** (0.10)	0.95*** (0.01)
Professional	0.85*** (0.16)	1.18*** (0.01)	1.33*** (0.26)	1.34*** (0.03)
Doctoral	1.00*** (0.16)	0.91*** (0.01)	1.08*** (0.19)	1.13*** (0.01)
Potential experience
Age (years)	−0.03 (0.05)	0.07*** (0.00)	0.05 (0.07)	0.13*** (0.02)
Age squared/100	0.08 (0.08)	−0.07***(0.01)	−0.05 (0.11)	−0.16*** (0.03)
Race/ethnicity
Black	−0.16* (0.08)	−0.19*** (0.01)	−0.05 (0.07)	−0.06*** (0.01)
Asian	−0.22* (0.09)	−0.02** (0.01)	0.04 (0.13)	0.10*** (0.01)
Hispanic	−0.12** (0.04)	−0.16*** (0.00)	−0.06 (0.06)	−0.04*** (0.01)
Married	0.14*** (0.03)	0.17*** (0.00)	0.10* (0.04)	0.07*** (0.02)
Sample size	1,443	172,048	728	134,877

Note. Estimates are obtained using generalized estimating equations. Robust standard errors are in parentheses. Bold font indicates that coefficients are statistically significantly different for men with and without hearing disability.

p < .05. **p < .01. ***p < .001.

Figure 1.

Estimated premia associated with a high school degree or beyond: Men.

Figure 2.

Estimated premia associated with a high school degree or beyond: Women.

In our subsample of men without a hearing disability and in both subsamples of women, predicted earnings increase at a decreasing rate within the age range of our sample, namely, 25 to 40 years; in our subsample of deaf and hard-of-hearing men, predicted earnings increase at an increasing rate within this range. The coefficients on age and its square are statistically significant in the large subsamples of men and women without a hearing disability.

In all four subsamples, married men and women earn statistically significantly more than their unmarried counterparts, a finding that is consistent with earlier findings in the case of men (Chiodo & Owyang, 2002) but inconsistent with earlier findings in the case of women (Chiodo & Owyang, 2003).

As expected, for both subsamples of men, racial and ethnic minorities earn statistically significantly less than Whites with the same educational attainment, age, and marital status. At nearly 20%, the estimated earnings gap between Asian and White men is particularly pronounced among deaf and hard-of-hearing men; this gap is statistically (p < .05) and economically significantly greater than the estimated gap (roughly 2%) between Asian and White men without a hearing disability. For both subsamples of women, our estimates suggest that Black and Hispanic women earn less than their White counterparts, whereas Asian women earn more; however, among women, earnings differences between racial and ethnic minorities and Whites are statistically significant only in the subsample of women without a hearing disability.

Blinder–Oaxaca Decompositions

Table 3 displays the steps in our Blinder–Oaxaca decompositions including the explained and unexplained portions of the male and female hearing earnings gaps. The first row of the table provides the arithmetic mean of log earnings for each of the four subsamples. These means are obtained by evaluating the estimated log earnings equations presented in Table 2 at the means of the independent variables for the subsample in question (see Table 1). For example, the arithmetic mean log earnings is 10.5047 for men who are deaf or hard of hearing and 10.6749 for men who are not. Exponentiating the arithmetic mean of log earnings provides an arguably more intuitive statistic, namely, the geometric mean earnings for the subsample in question. For example, as indicated in the second row of Table 3, the geometric mean earnings for our subsample of deaf and hard-of-hearing men is e^10.5047 or approximately US$36,487 compared with US$43,257 for men who are not deaf or hard of hearing. Thus, as shown in the third row of Table 3, the overall male hearing earnings gap is 15.7% when measured in terms of geometric means, similar in magnitude to the gap based on arithmetic means reported earlier (16.4%). In the case of women, the overall hearing earnings gap based on geometric mean earnings is 16.0% (US$30,506 vs. US$36,323).

Table 3.

Blinder–Oaxaca Decompositions.

Measure	Men		Women
Measure	Deaf or hard of hearing	Not deaf or hard of hearing	Deaf or hard of hearing	Not deaf or hard of hearing
Arithmetic mean log earnings: Own earnings equation	10.5047	10.6749	10.3257	10.5002
Geometric mean earnings: Own earnings equation (US$)	36,487	43,257	30,506	36,323
Total gap in geometric mean earnings	15.7%		16.0%
Arithmetic mean log earnings: Not deaf or hard of hearing earnings equation	10.6115	10.6749	10.4334	10.5002
Geometric mean earnings: Not deaf or hard of hearing earnings equation (US$)	40,597	43,257	33,975	36,323
Explained pap	6.1% gap in earnings≈39% of total gap		6.5% in earnings≈40% of total gap
Unexplained gap	9.6% gap in earnings≈61% of total gap		9.5% gap in earnings≈60% of total gap

Note. The arithmetic mean of log earnings is the predicted log earnings evaluated at the subsample means. The geometric mean earnings is equivalent to e^{(arithmetic mean log earnings)}. Gaps are computed relative to the subsample without a hearing disability.

Evaluating the estimated log earnings equation for men who are not deaf or hard of hearing at the means of our independent variables for the subsample of men who are deaf or hard of hearing yields the arithmetic mean log earnings (10.6115) for our subsample of deaf and hard-of-hearing men in the hypothetical case where deaf and hard-of-hearing men enjoyed the same earnings equations as men without a hearing disability. These arithmetic means are displayed in the fourth row of Table 3. Exponentiating the arithmetic mean log earnings again yield the comparable geometric means for each subsample. For example, as shown in the fifth row of Table 3, the geometric mean earnings of men who are deaf or hard of hearing would be US$40,597 in this hypothetical case. If men who are deaf or hard of hearing enjoyed the same earnings equation as men who are not, our Blinder–Oaxaca decompositions (Blinder, 1973; Oaxaca, 1973) thus indicate that predicted earnings for a deaf or hard-of-hearing man with average characteristics would be approximately 6.1% lower than predicted earnings of the average man without a hearing disability (US$40,597 vs. US$43,257). Likewise, if women who are deaf or hard of hearing enjoyed the same earnings equation as women without a hearing disability, the average deaf or hard-of-hearing woman would earn 6.5% less than the average woman without a hearing disability. Thus, our findings suggest that approximately 40% of the overall male and female hearing earnings gaps is explained by differences in educational attainment, potential experience, race/ethnicity, and marital status. The remaining 60% of these gaps reflects differences in intercepts and coefficients stemming from differences in unobservable characteristics such as communication skills as well as occupational segregation, labor market discrimination, and/or stigma.

These decompositions provide insight on the possible extent of labor market barriers—challenges communicating in the workplace, occupational segregation, stigma, and discrimination—but they must be interpreted with caution. For example, to the extent that differences in occupational preferences between those who are deaf or hard of hearing and those who are not contribute to hearing earnings gaps, the unexplained component overstates the barriers facing deaf and hard-of-hearing workers. However, if individuals who are deaf or hard of hearing experience discrimination in educational opportunities, the explained component partially reflects differences in opportunities, or pre-market discrimination.

Adjusted Hearing Earnings Gaps

As indicated above, hearing earnings gaps for both men and women stem partially from differences in characteristics and partially from differences in earnings equations. A comparison of predicted earnings for a few prototypical individuals highlights the magnitudes of the adjusted hearing earnings gaps—the estimated earnings gaps after controlling for human capital and demographic characteristics. For example, consider 33-year-old, unmarried high school graduates. As indicated in Table 4, among White men, predicted earnings are US$32,369 for those without a hearing disability and US$29,696 for deaf and hard-of-hearing men, translating to an adjusted hearing earnings gap of about 8%. Among White women, comparable figures are US$25,126 and US$21,340, again corresponding to an adjusted gap of about 8%. Turning to 33-year-old unmarried college graduates, predicted earnings are about 11% lower for deaf or hard-of-hearing White men (women) than for White men (women) without a hearing disability. Deaf and hard-of-hearing Asian men experience large earnings gaps relative to Asian men without a hearing disability. Among unmarried 33-year-old high school graduates, predicted earnings for deaf and hard-of-hearing Asian men are approximately 25% lower than their counterparts without a hearing disability. Among unmarried 33-year-old college graduates, the gap is roughly 27%.

Table 4.

Predicted Earnings (US$) of Prototypical Individuals Aged 33 Years.

Educational Attainment & Race/Ethnicity	Unmarried men		Unmarried women		Married men		Married women
Educational Attainment & Race/Ethnicity	Deaf or hard of hearing	Not deaf or hard of hearing	Deaf or hard of hearing	Not deaf or hard of hearing	Deaf or hard of hearing	Not deaf or hard of hearing	Deaf or hard of hearing	Not deaf or hard of hearing
High school graduates
White	29,696	32,369	23,140	25,126	34,022	38,180	25,469	27,046
Black	25,272	26,900	21,933	23,559	28,954	31,728	24,140	25,359
Asian	23,820	31,751	23,969	27,805	27,290	37,450	26,382	29,929
Hispanic	26,341	27,600	21,760	24,139	30,178	32,554	23,950	25,982
College graduates
White	47,078	53,001	37,810	42,667	53,937	62,515	41,616	45,926
Black	40,066	44,045	35,837	40,006	45,902	51,952	39,444	43,061
Asian	37,762	51,988	39,165	47,215	43,264	61,320	43,107	50,822
Hispanic	41,759	45,192	35,555	40,990	47,843	53,304	39,133	44,120

Discussion

We contribute to the sparse literature on hearing earnings gaps. In particular, we estimate separate earnings equations by gender and hearing ability and decompose the observed gaps into the component that is explained by differences in educational attainment, potential experience, race/ethnicity, and marital status and the component that is not explained by these characteristics. Although most differences are not statistically significant, our parameter estimates suggest that individuals without a hearing disability enjoy greater returns to education than their deaf and hard-of-hearing counterparts, after controlling for potential experience, race/ethnicity, and marital status. For both men and women, differences in educational attainment, age, race/ethnicity, and marital status explain roughly 40% of the hearing earnings gap. Thus, evidence suggests that up to 60% of the hearing earnings gap may stem from less visible barriers facing deaf and hard-of-hearing individuals such as challenges communicating in the workplace, occupational segregation, stigma, and labor market discrimination. Although our findings do not indicate the relative importance of these less visible barriers, these four explanations are not completely distinct. For example, discrimination in hiring on the basis of hearing ability may contribute to occupational segregation.

Overall, our findings suggest a need for policies that are more effective at reducing disparities in educational attainment as well as those that reduce workplace communication challenges, occupational segregation, and labor market discrimination. Increasing accommodations and opportunities in high schools, colleges, and universities could increase overall educational attainment and workplace communication skills among deaf and hard-of-hearing individuals, in turn increasing overall earnings in this population. For example, increased access to high-quality educational interpreters and other educational resources as well as professional enrichment programs such as internships and after-school activities could enhance the returns to education for individuals who are deaf and hard of hearing. In addition, incentives for schools to offer programs that educate students with disabilities about their rights in the workplace and that enhance disabled students’ abilities to advocate for themselves (Punch et al., 2004) may reduce hearings earnings gaps as current students enter the labor market.

With regard to policies outside of the formal education arena, current programs focus primarily on helping individuals with severe disabilities secure and maintain employment (Ross, 2012). For example, the Division of Vocational Rehabilitation in the Washington State Department of Social and Health Services provides assistance with resumes, interview practice, and job skills (Washington State Department of Social and Health Services, n.d.). While efforts to increase employment are appropriate given the relatively high unemployment rates in the deaf population (Punch et al., 2004), our findings highlight the need for policies that address employment as well earnings disparities. Hearing earnings gaps among those with higher levels of educational attainment also highlight the need for policies that foster professional development and salary negotiation skills among the highest functioning members of the deaf and hard-of-hearing community.

Although the Americans With Disabilities Act (ADA) requires employers to provide accommodations for disabled employees, where an accommodation is defined as “a reasonable adjustment to a job or work environment that makes it possible for an individual with a disability to perform job duties” (U.S. Department of Labor, n.d.), the ADA does not specifically require employers to provide interpreters for deaf employees. With accommodations that are less expensive but also less effective than interpreters, employers may fulfill their legal obligations, but deaf employees may not reach their full potential. Given the high costs of interpreters and other accommodations, the Corporation for National and Community Service, a United States federal agency, offers its partner agencies a variety of disability inclusion resources including online courses, advising, and technical assistance (Corporation for National and Community Service, n.d.). More resources and incentives for organizations and businesses to provide expensive or ongoing accommodations for employees with disabilities might alleviate hearing earnings gaps. For example, grants for small businesses to provide accommodations or to offer disability sensitivity training may enhance the performance, job satisfaction, and ultimately the earnings of deaf and hard-of-hearing employees.

The ADA and the Rehabilitation Act forbid discrimination in “any aspect of employment including hiring, firing, pay, job assignments, promotions, layoffs, training, fringe benefits, and any other condition of employment” (U.S. Equal Employment Opportunity Commission [EEOC], n.d.). Although the EEOC, the body charged with enforcement of these anti-discrimination laws, investigates claims relating to disability discrimination, evidence suggests that many deaf and hard-of-hearing individuals are unaware of the process of filing a discrimination claim with the EEOC or of the resources available to them (Bowe et al., 2005). Moreover, employers prevail in most employment discrimination allegations filed by deaf and hard-of-hearing individuals and investigated by the EEOC (Bowe et al., 2005). To the extent that labor market discrimination contributes to the hearing earnings gap, increasing deaf and hard-of-hearing employees’ understanding of their rights, the process of filing a discrimination claim with the EEOC, and strategies for filing a successful petition with the EEOC may reduce earnings disparities.

Although this study provides evidence on the magnitudes of hearing earnings gaps, several limitations are worth acknowledging. First, our measure of hearing disability is self-reported and therefore somewhat subjective. Second, we do not observe grades or severity of hearing loss and thus rely on a dichotomous measure of hearing ability, namely, whether or not an individual has “serious difficulty hearing.” By combining all individuals with a hearing disability into a single category, our estimates probably understate the earnings gap between deaf individuals and those without a hearing disability, while overstating the earnings gap between hard-of-hearing individuals and those without a hearing disability. Similarly, the ACS data do not distinguish between unilateral and bilateral hearing loss, although those with bilateral hearing loss may experience greater earnings penalties (Jung & Bhattacharyya, 2012). Third, we do not observe which deaf or hard-of-hearing individuals use hearing aids or have cochlear implants; thus, we cannot examine the effect of these treatments on earnings (Jung & Bhattacharyya, 2012). Fourth, although the ACS questionnaire inquires whether each individual “speaks a language other than English at home,” the data set does not include a code for American Sign Language (ASL). However, earnings gaps may depend on the extent to which workers with hearing disabilities rely on ASL versus spoken English.

Other potential limitations concern the measurement of earnings. Given our focus on FTYR workers and the relatively low employment rates among deaf and hard-of-hearing individuals, our results probably understate differences in labor market outcomes by hearing ability. Moreover, to protect the privacy of respondents, the Census Bureau top codes earnings. In 2011, the earnings thresholds for top coding ranged from US$141,000 in Puerto Rico to US$607,000 in Washington, D.C., with a mean of US$331,077 across the 50 states, the District of Columbia, and Puerto Rico (U.S. Census Bureau, 2015). Because deaf and hard-of-hearing individuals are less likely than those without a hearing disability to have earnings above these thresholds, our estimates of the hearing earnings gaps may be biased downward. However, the effect is probably minimal given the small proportion of young adults, regardless of hearing ability, with earnings above these levels.

Finally, given high correlations between the exogenous variables in our selection and outcome equations and the resulting concerns about robustness of Heckman’s limited information maximum likelihood (LIML) and full information maximum likelihood (FIML) estimators, our earnings equations do not control for selection into FTYR employment (Puhani, 2000).

Footnotes

Acknowledgements

We thank Vaishali Agrawal, Nicholas Woo, and Annie Zhang for their valuable research assistance; participants at the 2014 Intersectionality Research, Policy and Practice Conference in Vancouver and at the 2015 annual meeting of the European Society for Population Economics in Izmir for their comments and insights; the Albers School of Business and Economics for support in the form of work study and Albers Fellowship grants; and Seattle University for support in the form of a summer faculty fellowship.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Blinder

A. S.

(1973). Wage discrimination: Reduced form and structural estimates. The Journal of Human Resources, 8, 436–455.

Bowe

F. G.

McMahon

B. T.

Chang

Louvi

(2005). Workplace discrimination, deafness, and hearing impairment: The national EEOC ADA research project. Work: A Journal of Prevention, Assessment & Rehabilitation, 25, 19–25.

Brault

M. W.

(2012). Americans with disabilities: 2010 (Current Population Reports, P70–131). Washington, DC: U.S. Census Bureau.

Caiazza

Shaw

Werschkul

(2004). Women’s economic status in the states: Wide disparities by race, ethnicity, and region. Washington, DC: Institute for Women’s Policy Research.

Chiodo

A. J.

Owyang

M. T.

(2002, April). For love or money: Why married men make more. The Regional Economist, pp. 10–11.

Chiodo

A. J.

Owyang

M. T.

(2003, April). Marriage, motherhood, and money: How do women’s life decisions influence their wages? The Regional Economist, pp. 4–9.

Choi

H. J.

Joesch

J. M.

Lundberg

(2008). Sons, daughters, wives, and the labour market outcomes of West German men. Labour Economics, 15, 795–811.

Corporation for National and Community Service. (n.d.). Disability inclusion resources. Retrieved from http://www.nationalservice.gov/resources/disability-inclusion

De Coulon

(2001). Wage differentials between ethnic groups in Switzerland. Labour, 15, 111–132.

10.

Erickson

Lee

von Schrader

(2012). 2011 disability status report: United States. Ithaca, NY: Cornell University Employment and Disability Institute.

11.

Hanley

J. A.

Negassa

Edwardes

M. D. D. B.

Forrester

J. E.

(2003). Statistical analysis of correlated data using generalized estimating equations: An orientation. American Journal of Epidemiology, 157, 364–375.

12.

Hanushek

E. A.

Woessmann

(2008). The role of cognitive skills in economic development. Journal of Economic Literature, 46 607–668.

13.

Heckman

J. J.

(1979). Sample selection bias as a specification error. Econometrica, 47, 153–161.

14.

Hogan

O’Loughlin

Davis

Kendig

(2009). Hearing loss and paid employment: Australian population survey findings. International Journal of Audiology, 48, 117–122.

15.

Holt

Hotto

Cole

(1994). Demographic aspects of hearing impairment: Questions and answers. Washington, DC: Gallaudet Research Institute.

16.

Jung

Bhattacharyya

(2012). Association of hearing loss with decreased employment and income among adults in the United States. Annals of Otology, Rhinology & Laryngology, 121, 771–775.

17.

Lin

F. R.

Maas

Chien

Carey

J. P.

Ferrucci

Thorpe

(2012). Association of skin color, race/ethnicity, and hearing loss among adults in the USA. Journal of the Association for Research in Otolaryngology, 13, 109–117.

18.

Oaxaca

(1973). Male-female wage differentials in urban labor markets. International Economic Review, 14, 693–709.

19.

Prilleltensky

Gonick

(1996). Polities change, oppression remains: On the psychology and politics of oppression. Political Psychology, 17, 127–148.

20.

Puhani

P. A.

(2000). The Heckman correction for sample selection and its critique. Journal of Economic Surveys, 14, 53–68.

21.

Punch

Hyde

Creed

P. A.

(2004). Issues in the school-to-work transition of hard of hearing adolescents. American Annals of the Deaf, 149, 28–38.

22.

Ross

(2012). The current state of vocational rehabilitation services. In Vickerstaff

Phillipson

Wilkie

(Eds.), Work, health and well being: The challenges of managing health at work (pp. 95–118). Bristol, UK: Policy Press.

23.

Stern

(1989). Measuring the effect of disability on labor force participation. The Journal of Human Resources, 24, 361–395.

24.

Stevens

Flaxman

Brunskill

Mascarenhas

Mathers

C. D.

Finucane

(2013). Global and regional hearing impairment prevalence: An analysis of 42 studies in 29 countries. European Journal of Public Health, 23, 146–152.

25.

U.S. Census Bureau. (2015). American Community Survey: PUMS technical documentation. Retrieved from https://www.census.gov/programs-surveys/acs/technical-documentation/pums/documentation.2011.html

26.

U.S. Department of Labor. (n.d.). Disability resources. Retrieved from http://www.dol.gov/general/topic/disability/jobaccommodations

27.

U.S. Equal Employment Opportunity Commission. (n.d.). Disability discrimination. Retrieved from https://www.eeoc.gov/laws/types/disability.cfm

28.

Washington State Department of Social and Health Services. (n.d.). Rehabilitation administration. Retrieved from https://www.dshs.wa.gov/ra/division-vocational-rehabilitation/assistive-technology-services

29.

Winn

(2006). Is there a link between hearing aid use, employment, and income? American Annals of the Deaf, 151, 434–440.

30.

Woodcock

Pole

J. D.

(2008). Educational attainment, labour force status and injury: A comparison of Canadians with and without deafness and hearing loss. International Journal of Rehabilitation Research, 31, 297–304.

31.

World Health Organization. (2016). International Classification of Diseases (ICD). Retrieved from http://www.who.int/classifications/icd/en

32.

Yun

(2008). Blinder-Oaxaca decomposition technique. In International encyclopedia of the social sciences. Retrieved from http://www.encyclopedia.com/doc/1G2-3045300211.html