We investigate the prevalence and determinants of ethnic favoritism, i.e., preferential public policies targeted at the political leader's ethnic group. We are the first to study ethnic favoritism in a global sample and to use a broad measure – nighttime light intensity -- that allows capturing the distributive effects of a wide range of policies. We construct two panel datasets with several thousand ethnographic regions from around 140 multi-ethnic countries and annual observations from 1992 to 2013. We find robust evidence for ethnic favoritism: ethnographic regions enjoy 7%-10% more intense nighttime light and 2%-3% higher GDP when being the current political leader’s ethnic homeland. We further document that ethnic favoritism is a global phenomenon prevalent both within and outside of Africa; that economic development and better political institutions have at best weak effects on ethnic favoritism; that ethnic favoritism is partly motivated by electoral concerns and extends to linguistically close groups; and that ethnic favoritism does not contribute to sustainable development.