Occupational Race Segregation, Globalization, and White Advantage: White-Black Earnings Inequality in U.S. Metropolitan Areas

ABSTRACT

Addressing the need to systematically assess the materialist foundations of color-blind racism, we use insights from critical race theory to investigate the metropolitan-level racial inequality at the turn of the century. Namely, we examine the association between occupational race segregation and white advantage (i.e., white-black earnings inequality) for men and women in 202 U.S. metropolitan statistical areas in the year 2000. We find that occupational race segregation exacerbates white advantage for both male and female workers, supporting the tenets of the materialist conception of color-blind racism. We also consider how processes of globalization and labor market transformation impact white advantage. Our findings indicate that global capital increases white advantage for males, whereas foreign direct investment and casualization serve to decrease it. They also indicate that exports decrease white advantage for females, whereas percent foreign born increases it.

Notes

¹Differences between men’s and women’s incomes have been shrinking (Bernhardt, Morris, and Handcock 1995), but the disparities are still substantial. Because gender has such a pervasive effect in the labor market, we run separate analyses for men and women (Grant and Parcel 1990).

²Tomakovic-Devey et al. (2006) noted that while gender segregation in the workplace continued to decline into the 1990s and beyond, race segregation has not declined since about 1980.

³Details of the data set are available from the authors.

⁴MSAs are defined as metropolitan areas with at least one central urban area of 50,000 or more people. MSAs may comprised a single county or multiple counties. Multiple-county MSAs are formed when adjacent counties have a high degree of economic and social integration with one or more of the central urban areas and may span two or more states.

⁵For instance, Grant and Parcel (1990) included the largest 100 MSAs for 1980, and Jaret et al. (2003) included a random sample of 112 MSAs for 1990.

⁶Due to the relatively few degrees of freedom in our models, we checked their robustness by estimating reduced models that dropped nonsignificant variables from the final model. The significance of the remaining variables was not altered by this procedure.

⁷As a reviewer pointed out, estimates of median earnings for black males in the United States after 1990 are adversely influenced by high rates of mass incarceration, which reduce the supply of black workers and reduce their earnings power when they return to the community (Western 2006).

⁸Following previous research on black–white earnings inequality in urban areas (Grant and Parcel 1990; Hill 1974; McCall 2001b; Reid et al. 2007) and because of long-standing, deep gender differences in occupational structures and labor market outcomes, we conduct separate analyses of earnings ratios for men and women.

⁹Substitution of isolation and exposure, alternative measures of residential segregation identified by Massey and Denton (1988), yielded few substantive differences to those we report using the index of dissimilarity.

¹⁰The occupational segregation measures were calculated from occupational data in the 2000 U.S. Census. These measures were limited to full-time workers because this is the most stable portion of the labor market and controls for extraneous factors associated with part-time employment (Smith 1991; Volscho and Fullerton 2005).

¹¹Two MSAs were omitted from the sample due to inadequate occupational data: Pittsburgh, PA, and Fort Smith, AR-OK.

¹²While Jaret et al. (2003) use difference measures, we use ratio measures so that the independent variables are consistent with how we measure the dependent variables. This method has several advantages over alternative methods for including race- or gender-specific measures. By using ratio measures, we save degrees of freedom, avoid multicollinearity, and create a more straightforward interpretation of the results.

¹³In supplementary analyses, we determined that the inclusion of union membership mostly accounts for the loss of significance for the South dummy in both the male and female models.

Additional information

Notes on contributors

Travis Scott Lowe

Travis Scott Lowe is an Assistant Professor of Sociology at the University of Tulsa. His primary research areas are urban sociology, sociology of work and occupations, and social stratification. His current research focuses on changes in perceived job and labor market insecurity from the 1970s to the present.

Michael Wallace

Michael Wallace is Professor of Sociology at University of Connecticut. His primary areas of research interest are the sociology of work and organizations, social stratification and inequality, and the political economy of U.S. capitalism. His past research has centered on workplace issues such as earnings inequality, deskilling, deindustrialization, job security, and labor markets. Other work has focused on historical and contemporary studies of the U.S. labor movement, including strike activity, unionization, and union organizational effort. Recently, he has rekindled an earlier interest in the causes and consequences of military spending in the United States. Also, he has begun a new line of research in urban sociology that looks at metropolitan areas as arenas for the exploration of inequality in the U.S. political economy.