The effect of black population on the electoral performance of the democratic party : evidence from Alabama and Mississippi

Abstract

This thesis aims to estimate the effect of the percentage of black population in a county on the Democratic Party’s performance in presidential elections in Alabama and Mississippi. The argument highlights the relevance of race when determining the political preference of southern voters in contrast to other explanatory variables, and also assesses its possible endogeneity within a model that ascertains the primary factors that drive the Democrat vote. The identification strategy consists of an IV model under exact identification. The instrument of choice is the percentage of a county’s area whose soil belongs to the suborder Uderts, whose validity is sustained using historical and empirical evidence. The main takeaway is that the explanatory power of black population is 10.1% higher when instrumented through Uderts soils compared to an OLS estimation, where an increase of 1 p.p. in the black population of a county results in a rise of 0.958 p.p. in the votes the Democrats receive in presidential elections. This result is robust to variations in the measures of the dependent and independent variables, the inclusion of controls, and a series of robustness checks.