The Allocation of Talent, Economic Development and Skill Premia

Three central observations motivate this paper. First, there is little variation in the fraction of unskilled workers between rich and middle income countries. This occurs despite well-known, large differences in output per worker. Among a set of rich countries, unskilled workers 25 and older, defined as those with at most high school education, averaged 82 percent. Meanwhile, for a set of middle income countries, the average fraction is about 87 percent. Second, skill premia varies much more across the same set of countries. The skill premium in the poorer countries is about 62.5 percent higher than that in the rich group. Finally, data from the Programme for International Student Assessment (PISA) indicates substantial variation in the measured talent of young individuals across countries. We develop a parsimonious framework that connects these observations. We model the division of the labor force between unskilled and skilled workers, and map this division into observables such as output and skill premia, as well as unobservables such as `quality' of workers across countries. We discipline this framework with a host of aggregate and cross sectional observations, and then use it to investigate the extent to which the first two observations mentioned above can be accounted for by exogenous differences across countries. In our model, the differences in TFP and observed PISA scores imply that the fraction unskilled in poor countries would be 2.7 percent more than that in rich countries and the skill premium would be 10 percent more. The corresponding figures in the data are 9.3 percent and 64 percent. If we add country-specific impediments to the conversion of unskilled labor into skilled labor, and calibrate them to match the observed fraction of unskilled workers in each country, then our model generates 53 percent more skill premium in poor countries relative to rich.