Abstract
Hate crimes are a serious problem in the U.S., particularly since the COVID-19 pandemic. Policymakers disagree as to whether the problem should be tackled before or after the fact. Social scientists have trouble studying hate crimes themselves since hate crimes are underreported and political. This paper analyzes socioeconomic, political, and demographic determinants of hate crimes in the United States over the past thirty years based on individual hate crime data from the Federal Bureau of Investigation (FBI) from 1991 through 2021, which is aggregated by county and year to construct a 31-year county-level panel. The economic variables are unemployment rates, poverty rates, real income per capita, and a recession year dummy variable; the political variables are Republican-voting percentage, a Confederacy dummy, and a 9/11 and COVID-19 year dummy variable; and the demographic variables include Bachelor’s degree-holding percentage, percentage aged 15 to 24 years, male percentage of the population, and White percentage of the population. Pooled ordinary least squares (OLS) regressions, fixed effects regressions, and negative binomial and Poisson regression models are estimated. In the best-fit model, the time fixed effects model, Republican-voting percentage, Bachelor’s degree-holding percentage, and the Confederacy dummy were all significant at the 1% level, and coefficients were negative, positive, and negative, respectively. Negative binomial and Poisson models account for the overdispersion of hate crimes and find significance with poverty rates, real income per capita, White percentage, and Republican-voting percentage, while coefficients of the first two are negative and the latter two are positive.