Specifying Racial Inequality in Employment Earnings:



Specifying Racial Inequality in Employment Earnings:

Has the Process Changed Since 1976?

Richard Hogan (hoganr@purdue.edu)

Tyrell Connor

Sarah A. Mustillo

Paper submitted to Social Problems

July 2009

Comments and suggestions by Carolyn C. Perrucci were particularly helpful.

Abstract

Data for household heads in the 1976, 1985, and 2005 panels of the Panel Study of Income Dynamics indicate that racial inequality in employment income has not diminished to any appreciable extent, but neither did it increase appreciably between 1976 and 1985 (as suggested by Cancio et al. 1996). We find little change over time in either our ability to explain income inequality or in the robust effect of race. Access to class or occupational privilege changed fairly dramatically over the three decades of this analysis, but the long term pattern is fairly stable and durable racial inequality, which we view as the product of class, race, and gender relations, producing and reproducing inequality (Hogan and Perrucci 1998; Tilly 1998) within an institutional context of declining real wages, unionization rates, and secure employment opportunities, particularly for black men (Perrucci and Perrucci 2009).

Specifying Racial Inequality in Employment Earnings:

Has the Process Changed Since 1976?

Status attainment scholars believed that racial inequality was essentially about the inheritance of status, which is why racial inequality in employment earnings could be specified by the inherited advantages of education and occupation, with a residual or unexplained degree of inequality, attributed to discrimination (Duncan 1969, p. 100; Sowell 1981, pp. 290-294). Duncan (1969) reported both intercept and slope differences in black and white status attainment models, indicating the double disadvantage of race: both inherited status and the inability to translate education into occupation and income. Then Wright and Perrone (1977) found that black male managers gained comparable earnings advantages from increased education. Stoltzenberg (1975a) found similar results within occupational and industrial segments. The problem, of course, was that most blacks lacked the education, the managerial/professional positions, and the access to employment opportunities in the core industrial sector. Thus Wilson (1978, p. 1) concluded that blacks no longer suffered "racial oppression" so much as "economic class subordination."

Since then, Tomaskovic-Devey et al. (2006) have documented the racial desegregation of occupations, 1966-1980, indicating that progress has been stalled since 1980. Cancio et al. (1996) argue, in a similar vein, that between 1975 and 1984 racial discrimination increased, due to declining federal support for racial justice. Both Tomaskovic-Devey et al. and Cancio et al. argue, against Wilson (1978) and others who would reduce to race to class (e.g., Stolzenberg 1975a; Wright and Perrone 1977), that racial discrimination increased as the threat of the Civil Rights Movement and the opportunity of divided elites (particularly economic versus political elites) subsided (see Tarrow 1998, p. 76, on political opportunities; see Tilly 1978, pp. 133-138, on opportunity/threat). Tomaskovic-Devey et al. (2006) acknowledge that there are, in addition, effects of the shift to a service economy, but the thrust of their conclusions is that progress toward racial equality has stalled for what are, essentially, political reasons.

We are more inclined to support the class theorists, while conceding that race cannot be reduced to class (Hogan 2005a). Still we are inclined to argue that better measures of class and more attention to how race and class are reproduced through gender relations will illuminate efforts to disaggregate the components of racial inequality. We propose to begin this task by considering three panels from the Panel Study of Income Dynamics (PSID). First, we will look at the Reagan years (as Cancio et al. [1996] have done) and consider the possibility that racial inequality in employment earnings (in these data) increased between 1976 and 1985. Then we will look at more recent data (2005) to consider the extent to which what Cancio et al. (1996) consider to be the effect of the Reagan years was really a secular trend in the postmodern capitalist era (Harvey 1990), in which the decline of manufacturing and of traditional New Deal union benefits (Hogan 2005b) has resulted in increasing racial inequality that cannot be specified by the effects of education, marital status, occupation, or class. Simply stated, the most important changes are changes in the occupational or class structure (Hauser et al. 1975).

In conclusion, we will claim that a better specification of class relations illuminates the nature of these changes. Specifically, as we document here, the concomitant growth of the black professionals, managers, and self-employed or supervisorial classes with the even more spectacular growth of the black non-union working class, between 1976 and 2005, explains the paradox of increasing employment opportunities for black men and increasing racial inequality in earning.

As a first step toward specifying racial inequality in employment earnings, we compare data for household heads in the PSID, using the 1976, 1985, and 2005 panels, to indicate, first, that racial inequality in employment income has not diminished to any appreciable extent, but neither had it increased appreciably between 1976 and 1985. In fact, the gap between black and white male household head earnings increased between 1976 and 2005, while the gap between white male and black female heads declined modestly, but in both cases the change was most dramatic between 1985 and 2005. Second, we find that even before Reagan took office, our best model of class, race, and gender inequality still does not specify (or explain away) significant racial inequality. Third, we find little change over time in either our ability to explain income inequality or in the robust effect of race. It is clear that access to class or occupational privilege changed fairly dramatically over the three decades of this analysis, but the earnings advantages have not always been comparable. Ultimately, then, we conclude that the civil rights and equal opportunity efforts of the Sixties did bear fruit in providing access to "middle class" employment opportunities, but the long term pattern is fairly stable and durable racial inequality, which we view as the product of race, class, and gender relations, producing and reproducing inequality (Hogan and Perrucci 1998; Tilly 1998) within an institutional context of declining real wages, unionization, and secure employment, particularly for black men (Perrucci and Perrucci 2009).

Disentangling Disadvantage

"Among blacks in the twentieth century, the initial success of the nonviolent civil rights movement slowed perceptibly as more militant, direct action stiffened the resistance of the larger society ... The point here is not to definitively solve the question as to how much of intergroup differences ... have been due to the behavior and attitudes of the particular ethnic groups [versus] the behavior and attitudes of the larger society. The point is that this is a complex question, not a simple axiom." (Sowell 1981, p. 294).

One of the challenges of disentangling disadvantage is to avoid partitioning blame between the victims of discrimination and the discriminators. What Duncan (1969) calls the "vicious cycle of poverty" is frequently used by left leaning social critics (e.g., Michael Harrington 1963, quoted in Duncan 1969, p. 85) to blame the system or by conservatives to blame the victim, or their parents (Gallaway 1966, cited in Duncan 1969, p. 87). Although the liberal proponents of equal opportunity and affirmative action were relying on the best available social science data it was not clear to Duncan that they understood the difference between intercept and slope differences, or, in lay terms, between the initial disadvantage of being black (inherited status of parents) and the barriers to achievement that are associated with being black (blocked or unavailable mobility paths). Duncan (1969) found that black men fail to capitalize on their achievements and on the achievements of their parents. Not only do they start out at a disadvantage, but the disadvantage is compounded by the inability to translate father's occupation into occupational mobility (or even simple inheritance) or to translate occupation and education into income.

Featherman and Hauser (1978) later reproduced the racial difference in inheritance of occupation, both in the original Blau and Duncan (1967) data from 1962 and in their 1973 replication. In 1962, only 13.3% of black sons of upper non-manual fathers achieved upper non-manual status, but this increased to 43.9% in 1973. Among white men, however, 57% (in 1962) and 59% (in 1973) inherited upper non-manual occupational status (Kerbo 2009, pp. 391, 400). Thus we might argue that there has been progress but there continues to be racial inequality, both in the inheritance of status and in the status attainment process. That much of the story seems noncontroversial. Obviously, Civil Rights and Great Society legislation had a tremendous effect on poverty and on racial inequality. It seems equally clear that there has been little change in patterns of racial inequality since 1973, particularly in the earnings gap between white and black males. Although the gender gap has declined markedly since 1970 it appears that the racial gap has remained relatively stable, as has the pattern of unemployment. Black males earn about 60% of what white males earn, and they suffer unemployment rates of double the white figure (Hogan and Perrucci 2007; Kerbo 2009, p. 349).

Sociologists who focus on the segmentation of labor and industrial markets, most notably Stolzenberg (1975a), challenge the idea that blacks earn less as a function of their education and occupation, insisting that status attainment has failed to appreciate the extent to which the returns on education and experience vary across occupation and industry (Stolzenberg 1975a; Stolzenberg 1975b). Wright and Perrone (1977) similarly argue that class relations specify the previously observed differences in return on education. Black male managers, although they earn less than their white counterparts, receive comparable return on education. Wright (1997, pp. 67-68) explains that a large component of racial inequality can be specified by class. As he explains, "[J]ust under 30% of white men occupy privileged class locations, compared to 12.5% of white women, 8.4% of black men and 3% of black women." (Wright 1997, p. 69). At the opposite pole, "87% of black women, 77% of black men and 67% of white women are in the extended working class, compared to only about 51% of white men." (Wright 1997, p. 69).

Wright (1997) intends to subsume all forms of inequality under the rubric of "exploitation," but Tilly (1998) and Hogan (2001) argue that organizations (including unions, firms, and even extended families) create categorical inequality through either "exploitation" or "opportunity hoarding." Specifically, they effect inequality by including subordinates (or instituting subordination) to facilitate super-ordinate pursuit of profit (exploitation) or by excluding others from the pursuit of profitable enterprise (opportunity hoarding). As Hogan (2001) explains, racial inequality is rooted in endogamy rules that exclude blacks from the opportunities to capitalize on the resources that whites inherit or marry. This exclusion (opportunity hoarding) from the family is combined with exploitation at work and reproduced or institutionalized in various forms of patronage networks that exclude blacks from employment or business opportunities. In addition, the exclusion of blacks from the white family is reproduced in racially segregated housing, which becomes institutionalized into real estate and lending practices that further reduce the ability of blacks to accumulate wealth in the form of housing equity (Oliver and Shapiro 1997, pp. 19-23, 86).

Ultimately, we seem to have specified distributional components of racial inequality that we might associate with "opportunity hoarding" in racial and ethnic patronage, denying blacks access to managerial and professional positions in the core industrial sector, or to union jobs in manufacturing and construction industries. At the same time, the exploitation of labor, through employment and the lack of capital required for self-employment, and the opportunities for mobility from employee to supervisor (if not employer) are also critical components of the enduring racial inequality in employment earnings. In the analysis that follows we will include measures of occupation, industry, and class, along with the standard (often considered human capital) measures of education, hours/weeks worked and years of work experience, along with age, marital status, and region (since wage rates and employment opportunities vary across regions).

We are not particularly interested in contrasting "human capital" with class analysis or in arguing about how to measure class (Grusky and Sørensen 1998; Hogan 2005a; Robinson and Kelley 1979; Wright 1997). Similarly, we are inclined to include whatever information is available about unions and firm size (Kalleberg and Van Buren 1996), as well as the correlated measure of industrial core and periphery (Beck et al. 1978). Our intention is to specify the most efficient model for explaining income inequality by race, class and gender in 1976, 1985, and 2005 panels of the PSID, so that we can consider the extent to which the specification has changed or anything else has changed across three cohorts of household heads. Although we shall offer separate specification for black and white males and females we are particularly interested here in racial differences, which is why we are limiting our attention to heads of households (not including wives). Although this limits our ability to examine gender inequality (since the vast majority of household heads, following the 1968 convention for the first wave of the PSID, are married men) it gives us a better estimate of changes in racial inequality across cohorts, since households tend to reproduce race and class through gender relations, most notably marriage and child rearing. Here we shall ignore the details of this reproduction and look only at the outcomes, which reflect unequal life chances and business opportunities.

Data and Methods

The PSID offers forty years of fairly detailed information about employment, assets, and family finances. Here we are limiting our attention to three waves of data. The data from 1976 and 1985 were used by Cancio et al. (1996) to support their claim that racial discrimination (which had declined between 1962 and 1972: Featherman and Hauser 1978; Hout 1984) actually increased between 1976 and 1985 due to "the government's retreat from anti-discrimination initiatives in the 1980s" (Cancio et al. 1996, p. 541). We add data from the 2005 panel and estimate Ordinary Least Squares (OLS) Regression Models predicting head of household logged annual earnings from employment (including self-employment: called "head's labor income" in PSID). We do not include wives' earnings, capital gain/losses, government or private pensions or entitlements and we limit our sample to black and white heads of households who reported that they were currently working at the time of the interview.

In order to compare the results across waves we use weights (provided by PSID) to approximate population parameters in comparing means and in OLS models for all household heads. This also corrects for the over-sampling of black households. In the OLS models we use a series of dummy variables to distinguish white, male, married, core (industrial sector), manager, professional (occupational titles), supervisor (with authority for pay and promotions), proprietor (self-employed with less than five employees), employer (five or more employees), union (member) and region (North East [NE], Mid West [MW], West and South [excluded/reference category in OLS models]). We also use continuous measures of age (in years) and education (scaled 1-9 by PSID in 1976 and 1985; scaled 0-6, following Wright and Perrone 1977, in 2005), experience (years worked fulltime since age 18), hours/week and weeks/year worked during previous year.

The class and occupational measures (following Hogan 2005a) exclude managers, professionals, and the self-employed from "union workers" and excludes the self-employed from managerial but not professional classifications. In other words, respondents might be self-employed professionals but not self-employed managers. The same is true for supervisors. There are no self-employed supervisors, only employers and proprietors. Thus we might consider employer, proprietor, supervisor, and worker to be the mutually exclusive and exhaustive class categories, with worker as the reference (excluded) category. Similarly, we might consider professional, manager, union worker, and non-union worker as mutually exclusive and exhaustive occupational categories, with non-union worker as the reference (excluded) category.

Supervisor could not be coded for 2005, nor could experience. Similarly, firm size (more than fifty workers) was not available for the earlier panels. Within these limits we estimated the same models for all three years—identical models for 1976 and 1985, for all black and white, male and female heads of households, combined and analyzed separately. We also estimated the mean income (weighted) for all and separately for white and black men and women and used these to calculate racial and gender earnings gaps (percent of average earnings for black men and for white and black women/average white male earnings). These means (and gaps) provide the most direct evidence of increasing or declining racial and gender inequality, while the OLS models provide evidence of changes in the specification of race, class, and gender inequality.

Results

Table 1 reports weighted mean (and standard deviation) annual earnings for white and black male and female household heads in the PSID in 1976, 1985, and 2005. For each year, the percent of white male income is presented for white females and for black males and females. Also reported ("Total") are means (and standard deviations) for all heads of households (along with N for total sample and subsamples).

(table 1 about here)

Clearly white men claimed the highest income (labor income from head) and black women claimed the lowest. Also, the income (in nonstandard or "real" dollars) rises dramatically over time. Average total income more than doubles between 1975 and 1984, and it more than doubles again between 1984 and 2004. As seen in the "% of white male" figures, white female household heads claim a relatively stable 54-57% of white male income, rising slightly to indicate a modest decline in the gender gap during the Carter-Reagan years and a return to the baseline in 2004. Black males also experience a modest increase in relative income in 1984 but a much more substantial decline by 2004, averaging around 68% but exhibiting more variation over time than we see for white females. In fact, for black men, the 1975-2004 decrease in relative earnings, from 68% to 63% of white male earnings, is significant (z=2.16, p ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download