Cross-national data show no associationbetween increases in human capital attributable to therising educational attainment of the labor force and therate of growth of output per worker. This implies that theassociation of educational capital growth with conventionalmeasures of total factor production is large, stronglystatistically significant, and negative. These are 'onaverage' results, derived from imposing a constantcoefficient. However, the development impact of educationvaried widely across countries and has fallen short ofexpectations for three possible reasons. First, theinstitutional/governance environment could have beensufficiently perverse that the accumulation of educationalcapital lowered economic growth. Second, marginal returns toeducation could have fallen rapidly as the supply ofeducated labor expanded while demand remained stagnant.Third, educational quality could have been so low that yearsof schooling created no human capital. The extent and mix ofthese three phenomena vary from country to country inexplaining the actual economic impact of education, or thelack thereof.