Employment Discrimination, Segregation, and Health

Abstract

The author examines available evidence on the effects of exposure to joblessness on emotional well-being according to race and sex. The impact of racism on general health outcomes also is considered, particularly racism in the specific form of wage discrimination.

Perceptions of racism and measured exposures to racism may be distinct triggers for adverse health outcomes. Whether the effects of racism are best evaluated on the basis of self-classification or social classification of racial identity is unclear.

Some research sorts between the effects of race and socioeconomic status on health. The development of a new longitudinal database will facilitate more accurate identification of connections between racism and negative health effects.

BEING UNEMPLOYED OR OUT of the labor force has extensive destructive effects on psychological well-being. Social psychologists use expectancy theory as the basis for their view that individuals who possess a more internal locus of control—who view themselves as more in control of their own destiny—lead healthier and more successful lives than those with a more external locus of control. An internal locus of control is associated with stronger motivation; experimental studies have demonstrated that people with a more internal locus of control display higher levels of concentration, better access to working memory, more effort, and greater persistence than their peers with a more external locus of control.1 Exposure to unemployment and longer durations of unemployment tend to increase the likelihood that an individual will have a more external locus of control.

In the initial phase of an empirical research project, Arthur Goldsmith, Jonathan Veum, and I used the 1979 routine of the National Longitudinal Survey of Youth (NLSY), which included a locus of control score for each respondent, in an attempt to assess the psychological effects of joblessness.2 Scores were based on respondents’ answers on an abbreviated form of Rotter’s instrument3 for measuring expectancy. During that year the sample of respondents was relatively young, with a mean age of 20 years, and the total number of respondents at the time of the first interviews in 1979 was 12 686. We found that, even in this young sample, there was mild evidence of a negative effect of adverse work histories on degree of internalization, and the effect was stronger for the young women in the sample than for the male respondents. There were no differences in effects associated with race.

Subsequently, we used the 1980 NLSY routine to examine the effects of work histories on self-esteem, measured in this instance via the 10-item Rosenberg instrument.3 Self-esteem measures seek to capture people’s assessment of their personal worth or value, and higher levels of self-esteem appear to have a moderate association with psychological health.4 The Rosenberg instrument also can be decomposed into subsections that capture anxiety (apprehensiveness), alienation (depersonalization or loss of identity), and depression (hopelessness–helplessness). In the case of the latter subsection, there is a correlation between responses to questions that fall strongly into the depression category and clinical diagnoses of depression.

In 1980, with the mean age of our sample 1 year older, we found that self-esteem declined with both spells of unemployment and absence from the labor force. In fact, we found that the psychological effects of the 2 conditions—both forms of joblessness—were largely indistinguishable in our research. Again, there were no racial differences in these effects. In the case of this young sample, having experienced a spell of unemployment or absence from the labor force during 1979–1980 was associated systematically with elevated levels of alienation, anxiety, and depression.

In instances in which joblessness occurred earlier (i.e., 1978–1979), there was no evidence of an adverse effect. If individuals experienced joblessness in both periods, however, the impact on alienation, anxiety, and depression was even greater than the effect of being unemployed during the most recent period. It should be noted that women generally were more depressed and anxious than men and that Blacks displayed mildly greater anxiety than Whites; however, the direction of the effects of joblessness on mental health was the same regardless of sex or race.

In a further study, we used the 1992 routine of the NLSY to examine the effects of receiving an efficiency wage (a wage greater than the mean for individuals with similar productivity-linked characteristics) on motivation.1 We considered effects of joblessness on another locus of control (or motivation) measure, the Pearlin Mastery Scale5, which was administered to respondents in the 1992 NLSY. By this time the mean age of the respondents was 30–31 years, and they had a more extended work history. While we found that receipt of an efficiency wage was associated with higher scores on the Pearlin Mastery Scale, Blacks had the lowest wages in the sample and were least likely to receive an efficiency wage.

In the case of both Blacks and non-Hispanic Whites, exposure to bouts of unemployment as well as longer durations of unemployment lowered motivation, producing what can be referred to as a “scarred” worker effect.3,6 The experience of unemployment undermines the worker’s will to perform, resulting in the individual being less productive and less employable thereafter. However, there was no systematic effect of exposure to or duration of unemployment on the motivation levels of Hispanic respondents.

It is important to note that we found evidence, among Black respondents, of differences in the effect on motivation of one’s own exposure to unemployment and the presence of higher unemployment in one’s local labor market. The former had a negative effect on effort, while the latter was associated with higher scores on the Pearlin Mastery Scale. Thus, a higher local unemployment rate actually spurred individual Blacks to be more motivated, but being subjected to unemployment oneself had a destructive effect on effort levels. Local unemployment rate did not have a statistically significant effect on the measured effort of Whites and Hispanics.

Again, we found no racial differences in the effects of unemployment bouts or unemployment duration on motivation. There was modest evidence of higher levels of motivation among Black NLSY respondents than among White or Hispanic respondents in 1992. What was vastly different, however, was the extent of exposure to unemployment among Blacks. It is established that, historically and nationally, the Black unemployment rate is at least 2 times as high as the White rate, and the differential is wider still for younger workers.

In the case of the 1992 NLSY sample, Black respondents lived in areas where, on average, the local unemployment rate was 3 times as high as the White rate and almost twice as high as the Hispanic rate. Of course, as noted earlier, this generally would have prompted greater effort on the part of Blacks. But Blacks averaged more episodes of unemployment than Hispanics or Whites. Blacks also experienced the highest mean duration of unemployment: 32 weeks, 11 weeks longer than the mean duration for Hispanics and 14 weeks longer than the mean duration for Whites. Thus, while the adverse effects of exposure to and duration of unemployment were similar across the groups, incidence rates varied strongly. In addition, Blacks were likely to incur more psychological damage from joblessness, because they are subjected to joblessness more often.

It has also been shown that prolonged unemployment is associated with emotional damage that can overlap and intersect with harm to one’s own and others’ physical well-being:

Cross-national data indicate that periods of high unemployment are associated with increased rates of suicide and increased spouse abuse. There also is evidence suggesting that crime rates rise with the incidence of unemployment. A decrease in self-respect [and an increase in] sleeplessness, depression, and . . . mortality, not just via suicide, but due to what [Amartya] Sen terms “clinically identifiable illnesses” all accompany unemployment.7^(p494),8

Moreover, emotional decline among adults can inflict adverse effects on their children. Ahluwalia et al. found, in a sample of US welfare recipients, that mothers with depressive symptoms displayed patterns of impaired parenting.9 Their children demonstrated not only problem behaviors in school but lower school achievement.

The “last in, first out” phenomenon associated with discriminatory employment practices appears to be a central factor dictating greater exposure to joblessness among Blacks. In a study involving detailed information about the characteristics of workers unemployed as a result of mass layoffs and plant closings in North Carolina, Field and Winfrey found that Black workers were laid off in numbers disproportionate to their presence in the labor force.10 After controlling for differences in individual characteristics with respect to skill as well as industry and occupational differences, they still found strong evidence that employment practices differed according to race.

LABOR MARKET DISCRIMINATION AND HEALTH

A recent development in the literature on labor market discrimination is the claim that the relevant factors influencing wage differentials between Blacks and Whites are premarket or extramarket factors. Several researchers investigating the NLSY sample have claimed that evidence of in-market discrimination against Black workers is all but eliminated once cognitive skill differences are taken into account. To control for such differences, these researchers used Armed Forces Qualifications Test (AFQT) scores recorded in the 1980 routine of the NLSY. When they included these scores in a garden-variety wage equation, they found that very little of the wage gap between racial/ethnic groups remains unexplained by factors other than the productivity-linked characteristics of workers.

However, this body of work neglected to control for psychological capital or workers’ motivation/effort levels. In 2 articles involving data from the 1992 and 1993 routines of the NLSY, Goldsmith, Veum, and I reestimated the garden-variety wage equation, including both AFQT scores and scores on the Pearlin Mastery Scale as right-hand-side (independent) variables.1,5 This was an especially powerful exercise, because Black workers’ effort scores would have been influenced by their previous work histories, which generally were less salutary than those of Whites or Hispanics in the NLSY sample. Nonetheless, once we controlled for motivation, there was a strong, negative race effect in terms of wages among Black respondents, despite the presence of AFQT scores in the regression analysis.

Indeed, full sample results from the NLSY indicate evidence of both sex discrimination against women and racial discrimination against Blacks in regard to wages. Moreover, among subsamples of the respondents, there was evidence of significant intrasex earnings discrimination against Black men as well as Black women, although such discrimination against Black men appeared to be greater than that against Black women.

Reinforcing the evidence of discriminatory processes in labor markets are the results we uncovered when we estimated a labor force participation equation using the 1993 NLSY.5 Again, we found strong negative race effects among both male and female Blacks after controlling for motivation and AFQT scores as well as other standard factors (education, work experience, age, marital status, presence of children, urban or rural residence). These findings were entirely consistent with the “last in” side of the employment ledger.

We reported in the same article5 that individuals with lower motivation levels, regardless of race, are less likely to be labor force participants, less likely to engage in on-the-job training programs, more likely to experience job separation, and more likely to be unsatisfied with their job if they have one. It has been shown that Blacks actually have higher levels of motivation than Whites with comparable work histories. One can only imagine the levels of motivation Black workers might possess if labor market discrimination somehow vanished and the mean profile of their work histories was similar to that of Whites.

Does exposure to labor market discrimination have adverse health effects on Blacks residing in the United States? No existing data set permits us to address this question directly. A study conducted among ethnic minority groups in England and Wales revealed that “a marked independent association existed between reported experience of racism [verbal abuse or physical attack] and perceptions of Britain as a ‘racist society’ ” and adverse health effects, including greater self-reported incidence rates of high blood pressure, higher risks of heart attack, more respiratory illnesses, higher estimated weekly prevalence rates of depression, and higher estimated annual prevalence rates of psychosis.11

Findings from studies conducted in the United States also suggest that not only does the perception of exposure to racism affect health, but also the way in which an individual anticipates responding to racism. It has been shown that Black Americans who describe themselves as inclined to lodge a complaint about or confront racism are likely to have lower blood pressure levels than Blacks who report that they would not report or challenge racism.12,13 However, no existing studies have directly connected employment-based acts of racism to the health status of Black Americans.

In this context, it also is essential to be alert to the important distinction between perceptions of discrimination and actual exposure to discrimination. One of the few data sets permitting explicit consideration of this distinction is the Multi-City Study of Urban Inequality (MCSUI), a survey in which data were gathered from 8000 households and 3000 employers in 4 major cities: Atlanta, Boston, Detroit, and Los Angeles. In a collaborative study, Major Coleman, Rhonda Sharpe, and I (unpublished data, 2002) examined responses to 2 of the questions included in this survey: (1) During the past year, were you discriminated against at your place of work because of your race/ethnicity? and (2) Have you ever felt in the past that others at your place of employment got promotions or pay raises faster than you did because of your race or ethnicity?

We found that 20% of Black male MCSUI respondents answered affirmatively in response to the first question, while 16% of Black female respondents answered affirmatively. In the case of the second question, 27% of Black male respondents answered affirmatively, while 20% of Black female respondents indicated that they had experienced pay or promotion discrimination. Among Whites, 5% of men and 7% of women reported that they had experienced discrimination at their place of work, while 8% of men and 7% of women indicated that they had experienced discrimination in the area of raises and promotions (M. Coleman, W. A. Darity Jr., and R. Sharpe, unpublished data, 2002).

We decided that, in the case of both Black and White respondents, it would be useful to compare these patterns of responses against direct statistical measures of wage discrimination. Using a variant of the Blinder–Oaxaca–Blau–Duncan decomposition procedure, we were able to construct individual estimates of wage discrimination for each respondent in the MCSUI sample (M. Coleman, W. A. Darity Jr., and R. Sharpe, unpublished data, 2002).

We found that, among more than 97% of Black male and female MCSUI respondents who reported experiencing discrimination at their place of work, there was statistical evidence of wage discrimination, while the opposite was the case among Whites who reported workplace discrimination (97% of White women and 100% of White men displayed no evidence of wage discrimination). The results with respect to the second question were similar. In the case of 96% of Black men and 98% of Black women who reported discrimination in regard to promotions and raises, there was statistical evidence of wage discrimination, while about 15% of White male and female respondents who included such reports exhibited evidence of wage discrimination (M. Coleman, W. A. Darity Jr., and R. Sharpe, unpublished data, 2002).

More interesting still, Coleman, Sharpe, and I found that more than 90% of the Black men and women who did not report being subjected to discrimination in their responses to either question also exhibited statistical evidence of wage discrimination, while the rate was below 3% for White male and female interviewees with similar responses. If the statistical evidence is taken seriously, it would suggest that the vast majority of Blacks either do not recognize the discrimination to which they are being subjected or are engaged in some form of cognitive dissonance or denial (note the evidence described earlier suggesting that the adverse health effects of exposure to labor market discrimination are more pronounced among Blacks who do not report their experiences).12,13

Nevertheless, we cannot determine whether Blacks who perceive or acknowledge that they have experienced employment-based discrimination are healthier psychologically or healthier in other ways than Blacks who report that they have not been subjected to discrimination. Is it the perceptual response that is linked to health outcomes, or is it the discriminatory practices themselves? We know that discriminatory practices leading to joblessness result in psychological harm, but previous research has linked such harm to the exposure to joblessness itself rather than the associated racist processes. We need further investigations and new data sets to examine whether either the perception or the practice of discrimination leads to physiological damage.

SKIN SHADE, JOB DISCRIMINATION, AND HEALTH

Phenotypical variations within racial/ethnic groups also are associated with different social and economic outcomes. The alleged “one-drop rule” for creating racial boundaries in the US context (metaphorically, “one drop of Black blood makes one Black”) does not operate in the unequivocal pattern asserted by social scientists who seek to construct a contrast with Latin American norms of race relations.

Seltzer and Smith, using data from the 1982 General Social Survey sample of 510 Blacks distinguished as dark, medium, or light skinned, found that dark-skinned Blacks fared worse on all social and economic dimensions.14 Similarly, Keith and Herring, using data from the 1979–1980 National Survey of Black Americans (NSBA), developed by James Jackson at the University of Michigan, found that Blacks with darker complexions had less education, less prestigious occupations, lower personal incomes, and lower family incomes than Blacks with lighter complexions.15 Telles and Murguia found, in their study involving data from the 1979 National Chicano Survey, that Chicanos with lighter skin tones and more European features had higher earnings, were of higher socioeconomic status (SES), and faced less discrimination.16

It is not clear, however, that lighter-skinned Blacks consistently face less employment discrimination. In preliminary research that Darrick Hamilton and I have conducted with the MCSUI data set, results suggest that lighter-skinned blacks, while earning more than their darker-skinned brethren, actually face higher discriminatory deficits in terms of wages. The updated NSBA, the year 2000 National Survey of American Life, also will enable researchers to further explore the social and economic consequences of skin shade variations.

However, these data sets are not rich in information about health outcomes, although the National Survey of American Life data set will contain all of the Pearlin Mastery Scale questions (the NSBA included an abbreviated Rotter instrument). Without a still newer data set, it will not be feasible to explore the relationships between race/ethnicity, phenotype, labor market outcomes, and health outcomes.

A further cautionary note is in order: phenotypical information is vital, especially in examining labor market and health linkages among Latinos, but such information is not available in the NLSY. The reason is the wide gulf between racial social classifications and self-classifications among Latinos. Many Mexicans, Puerto Ricans, and Cubans share an aversion to self-identification as Black to such an extent that members of each group with dark complexions frequently will self-report their race as White, bypassing “intermediate” categories (e.g., “brown,” “mulatto”).17 Thus, it is important that interviewers also provide their own coding of an individual’s race and phenotypical attributes to establish how he or she is most likely to be perceived in the general society.

RACE, CLASS, SEGREGATION, AND HEALTH

Sorting between the effects of race and SES on health outcomes is a tricky business as well. An excellent recent study conducted by Darrick Hamilton (unpublished data, 2000) sought to explain racial differences in infant mortality rates. The Black rate persistently is more than double the White rate. Hamilton found that socioeconomic attributes and demographic and geographical differences did not provide an explanation of the higher Black infant mortality rate. Indeed, more than 4 times the disparity in rates remained unexplained by this set of factors, leading Hamilton to conclude that there are racial differences in access to prenatal care that must reside at the heart of the infant mortality gap.

Similarly, David Williams points out that although “lowincome whites have rates of heart disease that are about twice as high as those of highincome blacks . . . racial differences in heart disease persist at every level of SES.”18 According to Williams, this is suggestive of a discriminatory environment that operates to create a Black health disadvantage in 2 ways.18

First, current economic circumstances of Black men and women do not provide a gauge of their exposure to economic disadvantage over the course of their lifetime. This effect, associated with life cycle conditions, would be reinforced by intergenerational transmission of resources and prospects for well-being. Using the Integrated Public Use Micro-Sample, which includes all available censuses from 1850 to the present, Jason Dietrich, David Guilkey, and I have shown that powerful group-based legacy effects are present in American labor markets.19 We found that the human capital characteristics and degree of labor market discrimination faced by groups’ ethnic ancestors during the period 1880 to 1910 had powerful effects on the occupational outcomes of their descendants a full century later. One might expect such a pattern to translate into intergroup differences in health outcomes as well.

Second, Williams observed that, as a result of the processes of discrimination, “blacks are less able than whites to translate economic status into desirable resources in society.” He continued: “In 1990, for example, among persons working full time in the labor force, white males who graduated from high school had a median annual income that was $6000 more than black males and $5000 more than Hispanic males. . . . Research also reveals that blacks and Hispanics are more likely to live in neighborhoods that are lower in SES and desirable neighborhood amenities.”18^(p4)

Typically, however, SES measures involve controls for personal or family income but not the measure of class difference that is most pronounced for Blacks and Whites: wealth. Wealth is a stock concept rather than the flow concept associated with income, and it includes property holdings (e.g., real estate) as well as financial assets (e.g., stocks and bonds). The highest reported estimate of the mean ratio of Black wealth to White wealth is 0.25. Some studies place the ratio of median Black to White wealth at zero.20^{(pp179–180)} Williams made a similar point:

[A]lthough there are large racial differences in income, focusing on income differences grossly understates the true racial differences in economic circumstances. . . . [R]acial differences in wealth are much larger than racial differences in income and these disparities persist at all levels of income. . . . [A]mong persons who fall into the lowest quintile of income, the median net worth of whites is $10,000 compared to one dollar for African Americans.18^(p4)

Thus, an important but unanswered set of questions includes the following: How can we improve the caliber of our attempts to control for SES in research on racial disparities by incorporating information about wealth? How long (for how many generations) would racial differences in health outcomes such as infant mortality or heart disease persist if the wealth gap were closed by raising the Black wealth profile? Is wealth redistribution according to race a vital strategy for improving health outcomes among Blacks? To what extent would discriminatory processes continue to inhibit improvement in Black health outcomes if the wealth gap were closed?

Finally, Thomas LaViest outlined the connection between residential segregation and adverse health outcomes among Black Americans:

A study of the effect of racial residential segregation on race differences in infant mortality in U.S. cities found that the Black infant mortality rate was higher and the White rate lower in highly segregated cities. . . . Although the empirical link between segregation and mortality is relatively straightforward, the specific mechanisms that connect these variables are less direct. Previous studies have demonstrated a greater prevalence of negative social conditions in many highly segregated African American urban communities. These studies have shown that such communities are highly toxic environments . . . that are not well served by basic city services . . . and that lack adequate medical services. . . . A recent study also demonstrates an excessive availability of alcohol in greatly segregated cities. . . . Thus, racial segregation is an indication of the magnitude of the differences in material conditions in which African Americans and Whites live in the same city. A wide racial divide in exposure to these conditions measured by racial segregation is manifested in racial differences in health outcomes.21^(p151)

Such segregation is maintained through discrimination in housing and real estate markets, documented most comprehensively in audit studies conducted by the US Department of Housing and Urban Development.22

Part of the impairment in terms of Black American health resulting from segregation can be attributed to the lower income levels and greater incidence of concentrated poverty in many predominantly Black communities. However, there is a salient dimension linked to race and lack of political power. The absence of accessible, high-quality hospital or health care services is a function of metropolitan political decisions regarding allocation of resources. And race plays a role in determinations of which communities are given priority, with Black communities frequently on the short end of the stick. That stick becomes shorter still as the growing diversity of American society expands the range of groups with legitimate claims for redress of their grievances.21

In concluding, a companion issue that needs to be addressed is that of advocacy. In the case of monitoring of Black health affairs, there is no federal watchdog unit that possesses credibility and influence. Clovis Semmes recently advocated restoration of a version of Booker T. Washington’s National Negro Health Week movement and the renewed publication of a modern equivalent of the National Negro Health News, first published at Tuskegee Institute in the early part of the 20th century and then taken over by the US Public Health Service until its demise.23

The eventual demise of National Negro Health News was associated with the elimination of the Office of Negro Health Work, an event being examined by psychologist Shawn Thompson, a postdoctoral fellow at the University of North Carolina at Chapel Hill. According to Thompson, the Office of Negro Health Work maintained a spotlight not only on health conditions among Blacks but on strategies for improving Black health from its inception through its subsumption by the US Public Health Service during the interval between 1932 and 1950. In 1950 it was closed, ostensibly because, in the gravitation toward a color-blind society, there was no need to maintain a federal unit focused on the health needs of a single group. The disparities between Blacks and the larger society had not been eliminated, however, and the social conditions facing Blacks remained distinctive. In a color-fair society, the Office of Negro Health Work would have been maintained and its activities expanded.

Thus, one part of a new research agenda must involve a focus on the issue of how to design and establish an effective federal advocacy unit for Black health that will incorporate a broad view of the causes and consequences of racial disparities in health. There is also a need to develop a new longitudinal data set that encompasses from the outset individuals in a wider age range than, for example, the NLSY (in its first round of interviews, the NLSY targeted youths, therefore interviewing individuals aged 14 to 22 years). Such a data set would provide researchers with detailed information about respondents’ economic characteristics (work histories, income and wealth status), demographic characteristics (age, sex, neighborhood, race, ancestry), self-reported physical health, scores on standardized tests of achievement, and mental well-being.

Careful attention should lead to the design of a survey that provides information about individuals’ perceptions of their exposure to discrimination in a variety of arenas—especially the workplace and the housing market—as well as their phenotypical attributes, particularly their skin shade. An ideal survey design would also enable researchers to assess respondents’ educational backgrounds as well as obtain information regarding their parents’ and grandparents’ occupations, educational status, and income levels. In addition, such a survey would provide data on individual’s (and their parents’) economic characteristics.

Thus, there is a need for a single data set that would enable us to trace the connections between health outcomes and discrimination, both perceptually and in terms of statistical measures of discrimination, while controlling for a wide range of factors that we now consider pertinent. My recommendations of future directions involve developing a new institution whose focus is on Black health conditions and developing a new database that will facilitate a comprehensive investigation of the links between race, ethnicity, labor market outcomes, and mental and physical well-being.