Description:
Using a unique panel survey of final-year undergraduates at six of the largest universities in Mozambique, we study the wage premium associated with completing an undergraduate degree. Conditional on a very rich set of controls, including pre-degree earnings, objective measures of ability, and academic performance, we find heterogeneity in 'sheepskin effects' across different kinds of firms. We propose a simple model of wage-setting in which productivity is only partially observable in some firms and fully observable in others. In this setting, education serves both to enhance productivity and as a productivity signal. Consistent with the theory, positions where productivity is likely to be less observable offer larger sheepskin effects.