Estimates of the e¤ect of education on GDP (the social return to education)have been hard to reconcile with micro evidence on the private return. We present a simple explanation that combines two ideas: imperfect substitution between worker types and endogenous skill biased technological...