Human capital accumulation is usually held to generate positive externalities. However, the empirical evidence on human capital externalities has been scarce and inconclusive. The missing evidence appears to be caused by misleading comparisons between private and social rates of return, by incorrect interpretations of regression coefficients, or by an inappropriate level of aggregation. Our own estimates point to relatively robust empirical evidence for a human capital externality that is about twice as large as a benchmark estimate, which is derived from a standard parameterization of a simple growth model. We ponder a possible causal link from human capital to social capital in the context of alternative views of long run development and argue that the grand transition view (Paldam 2002) is compatible with the existence of a large human capital externality.