Estimating Skilled Labor Efficiency Using Trade and Industry Data∗

An important task in the growth and development literature has been to characterize the nature of cross-country labor efficiency differences. In this paper, I develop a new method to estimate the relative efficiency of skilled and unskilled labor across countries using disaggregated industry and trade data. The method assigns a country a high efficiency of skilled labor if it has low relative unit production costs in skill-intensive industries, conditional on factor prices. For estimation, I use international trade data. I first document that the share of exports in skill-intensive industries rises sharply with income levels. Interpreted through the lens of a gravity model, this pattern suggests that rich countries have low relative unit production costs in skill-intensive industries, and low effective costs of skilled labor services. I show that for standard trade elasticities, these low effective costs of skilled labor services cannot be rationalized by differences in skilled wage premia, which leads me to infer that skilled labor is relatively efficient in rich countries. Integrating these findings into a development accounting exercise in manufacturing, I find that accounting for skill-specific efficiency differences reduces the size of uniform labor efficiency differences: the difference in skill-neutral TFP differences between rich and poor countries in manufacturing falls from a factor of 4.3 to a factor of 2.6. ∗This paper was previously circulated under the name ”Human Capital and Development Accounting Revisited”. I am grateful to Timo Boppart, Axel Gottfries, John Hassler, Karl Harmenberg, Per Krusell, Erik Öberg, and Torsten Persson for extensive comments on the project. Moreover, I am grateful for helpful discussions with Ingvild Almås, Konrad Burchardi, Francesco Caselli, Saman Darougheh, Niels-Jakob Harbo Hansen, Chad Jones, Loukas Karabarbounis, David Lagakos, Bo Malmberg, Kurt Mitman, Jon de Quidt, Jósef Sigurdsson, Lena Sommestad, David Strömberg, and David Weil. †Postdoctoral Fellow. SIEPR, Stanford University. Email: [email protected]. 1