Understanding Long-Run African Growth: Colonial Institutions or Colonial Education? Evidence from a New Data Set

Long-term growth in developing countries has been explained in four frameworks: ‘extractive colonial institutions’ (Acemoglu et al., 2001), ‘colonial legal origin’ (La Porta et al., 2004) ‘geography’ (Gallup et al., 1998) and ‘colonial human capital’ (Glaeser et al., 2004). In this paper we test the ‘colonial human capital’ explanation for sub-Saharan Africa, controlling for legal origins and geography. Utilizing freshly collected data on colonial-era population density and education, we find that in sub-Saharan Africa, high European population mortality did not lead to low European population densities, contra Acemoglu et al., (2001). Further, we find that instrumented human capital explains long-term growth better, and shows greater stability over time, than instrumented measures for extractive institutions. We therefore suggest that the impact of the disease environment on African long-term growth runs through a human capital channel rather than an extractive-institutions channel. The effect of education is robust to including variables capturing legal origin and geography, which have additional explanatory power. We also find some evidence that institutions are endogenous to education.