Identification of the strength of human capital externalities at the aggregate level is still not fully understood. The existing method may yield positive or negative externalities even if wages reflect marginal social products. We propose an approach that yields positive average human capital externalities if and only if the marginal social product of workers with aboveaverage human capital exceeds their wage. As an application, we estimate the strength of average-schooling externalities in US cities between 1970 and 1990.