The Relationship between Technical Progress and Employment

We show that Autor and Salomons’ (2017, 2018) analysis of the impact of technical progress on employment growth is problematic. When they use labor productivity growth as a proxy for technical progress, their regressions are quasi-accounting identities that omit one variable of the identity. Consequently, the coefficient of labor productivity growth suffers from omitted-variable bias, where the omitted variable is known. The use of total factor productivity (TFP) growth as a proxy for technical progress does not solve the problem. Contrary to what the profession has argued for decades, we show that this variable is not a measure of technical progress. This is because TFP growth derived residually from a production function, together with the conditions for producer equilibrium, can also be derived from an accounting identity without any assumption. We interpret TFP growth as a measure of distributional changes. This identity also indicates that Autor and Salomons’ estimates of TFP growth’s impact on employment growth are biased due to the omission of the other variables in the identity. Overall, we conclude that their work does not shed light on the question they address.