Race, inequality, and social capital in the U.S. counties

ABSTRACT

This study examines how the interplay between racial diversity and economic inequality affects variations of social capital in the U.S. counties. In general, racial and economic heterogeneity is assumed to provide a negative environment for the growth of social capital. Building on this, we argue the effect of economic inequality is weaker than that of racial diversity because increased economic heterogeneity is felt less visibly and acutely than racial heterogeneity. Moreover, economic inequality can positively condition the adverse impact of racial diversity on social capital when the two interact. Based on the crosscutting cleavages theory, income inequality in a racially fragmented community works as an additional cleavage that crosscuts the different racial groups, mitigating the negative impact of racial diversity on social capital. The data analysis of 3,140 U.S. counties in 2009–2014 provides strong evidence for our arguments. Our findings offer important implications in understanding inequality, race and American democracy.

KEYWORDS:

• Social capital
• Racial diversity
• Income inequality
• U.S. counties
• American democracy

Notes

¹ Other county-level variables used in our analysis are provided by the American Community Survey, which is available only since 2009. See for a source of county-level social capital data utilized in this study, https://aese.psu.edu/nercrd/community/social-capital-resources. Rupasingha, A., Goetz, S. J., and Freshwater, D. (2006, with updates), “The Production of Social Capital in U.S. Counties,” Journal of Socio-Economics, 35, 83–101.

² Edgefield County in South Carolina scores 17.44 in 2009 and 21.8 in 2014 for social capital index.

³ The seven categories include non-Hispanic White, Hispanic, Black, American Indians, Asian, Pacific Islander, and others.

⁴ These are counties that either have urban population more than 20,000 or are located adjacent to a metro area.

⁵ As all variables are measured at the county level, one might see that a regression analysis with state fixed effects is a more straightforward option. Theoretically, this method is identical to the MLM with a random intercept model adopted in this paper. However, we believe the MLM with a random intercept is a more rigorous empirical technique that controls for the variations of the dependent variable across the states. As we can see in Table A2 in the appendix, the regression coefficients are slightly different from Table 1. However, it does not change the conclusions we are making here.