The literature on schooling externalities in U.S. cities and states is rather mixed: positive external effects of average education levels are hardly found while positive externalities from the share of college graduates are more often identified. We propose a simple model to reconcile this mixed evidence. Our model predicts positive externalities from increased college education and negligible external effects from high school education. Using compulsory attendance/child labor laws, push-driven immigration of highly educated workers, and the location of land-grant colleges as instruments for schooling attainments, we test and confirm the model predictions with data on U.S. states for the period 1960–2000.