Can social diversity explain the differences in poverty? Empirical evidence for US States 2000-2010

Abstract

Little research has been done on issues of underdevelopment in the developed countries. The onset of the financial crisis in 2008, and the subsequent macroeconomic instability in parts of Europe have sparked debate about income inequality and minimum wage. To date the existence of poverty alongside a well-structured welfare state system remains an open question. Literature provides evidence that income inequality and stagnant growth in median wages in USA are the main reasons for limited positive impacts of rising per capita income over time. We argue that the existing literature understates the role of differences in social diversity in explaining differential economic outcomes across the geographic regions in the US. In this paper, we use two measures of social diversity: linguistic diversity and racial diversity, and investigate if these diversities can be identified as deterministic factors for the differences in poverty rates across the US states. The empirical estimation of the ordered logit model indicates that, ceteris paribus, an increase in racial diversity increases the log-odds of being in a higher poverty-stricken group, while an increase in linguistic diversity decreases the log-odds of being in a higher poverty-stricken group. For the two groups with the lowest poverty rate, linguistic diversity exhibits a positive marginal effect on the probability of belonging to these groups while racial diversity shows a negative marginal effect. This suggests that states with higher linguistic diversity and/or lower racial diversity are more likely to have a lower poverty rate than otherwise. Considering the other control variables, our results suggest that states with relatively higher income inequality, lower shares of Hispanic and undereducated people, and whose gross state product per capita is high, tend to have lower poverty rates. The marginal effects of the statistically significant variables have the opposite signs for the higher poverty quintiles. A possible explanation for this contrasting impact of diversity measures can be traced back to the institutionalization of social segregation under slavery. We argue that boundaries across racial groups are likely to have an enduring effect on optimal provision of public goods, and thus are likely to have a negative impact on growth-creating economic opportunities. As the boundaries based on differences in the mother tongue only are less pronounced, the negative impact of linguistic barriers to communication is outweighed by the positive impact of complementarities of knowledge sets and skills of a diverse population.