Does Learning to Add up Add up? The Returns to Schooling in Aggregate Data

The theoretical, conceptual, and practical difficulties with the use of cross-national data on schooling are so severe using aggregate data for any purpose for which individual level data would do should be avoided. There are, however, three questions for which the use of cross-national data on schooling could potentially help answer interesting questions for which individual data is insufficient. First, do differences in the evolution and dynamics of schooling help explain the big facts about the evolution and dynamics of output growth? Largely, no. Second, the existence and magnitude of output externalities to schooling is an important question with possible normative policy implications, and evidence for externalities requires at least some level of spatial aggregation. Does the cross-national data provide support for output externalities? Largely, no. Third, cross-national (or more broadly spatially aggregated) data allows the exploration of the impact on returns to schooling (or in the gap between private and social returns) of differences in economic environments. This last question seems a promising line for future research.