Tasks and Heterogeneous Human Capital

This paper proposes a new approach to modeling heterogeneous human capital using task data from the Dictionary of Occupational Titles. The key feature of the model is that it departs from the Roy model, which treats occupations as distinct categories, and conceives of occupations as bundles of tasks. The advantages of this approach are that it can accommodate many occupations without computational burden and provide a clear interpretation as to how and why skills are differently rewarded across occupations. The model is structurally estimated by the Kalman filter using the NLSY79.