Selection, Structural Transformation, and the Cost Disease of Services

Labor productivity growth in the service sector may be mismeasured if workers with heterogeneous skills self-select into sectors. I document with US data that workers reallocated from manufacturing earn more than incumbent workers in professional services but less than incumbent workers in education, health, and public services. A generalized quantitative Roy model predicts a selection effect on labor productivity growth in professional services that is 10 percentage points higher than what a conventional selection model predicts. Overall, the selection effect contributes little to the cost disease of services, which contrasts sharply with the hypothesis in the literature. (JEL E24, J24, J31, J44, J45, L60, L80)