A growing body of literature tests the effects of different tax structures on long-run economic growth. We argue that these tests do not properly account for endogeneity between supposedly independent variables. We run several cross-country ordinary least squares tests with special attention to human capital, and show how education choice behaviors are affected by different tax mixes. The results obtained by microeconomic theory are validated, and they imply that accumulation rates of human capital cannot be deemed independent from savings taxation. Our results also show that more progressive labor taxation does not appear to be correlated with lower investments in education, contrary to what one would expect from microeconomic theory. We discuss possible implications, and suggest that a likely explanation lies in the outcome of redistribution policies reducing credit constraints of poorer households, thus allowing them easier access to education and, consequently, higher aggregate human capital accumulation.