Cross country variations in the impact of schooling on growth: some simple tests on obvious modifications *

Many existing cross -country analyses of the impact of schooling on growth find low economic returns to schooling, with the coefficient estimates being sensitive to modelling approach and sample selection. This paper argues that coefficient instability is to be expected when regression samples include countries that vary greatly in their ability to use schooling productively. Data on corruption, the black market premium on foreign exchange and the extent of the brain drain for developing countries are used as indicators of an economy’s productive use of schooling. These data are used to classify countries into sub-samples on the basis of those more or less likely to incorporate schooling into productive human capital. The results indicate that in sub-samples of countries with lower black market premia, corruption and ‘brain drain’ the impact of schooling on economic growth is substantially higher.