This paper points out that modeling automation as factor-augmenting technological change has several unappealing implications. Instead, modeling it as the process of machines replacing tasks previously performed by labor is both descriptively realistic and leads to distinct and empirically plausible predictions. In contrast to factor-augmenting technological change, the substitution of machines for labor in additional tasks always reduces the labor share in national income and can reduce the equilibrium wage (for realistic parameter values). This approach to automation also enables a discussion of several new forces at work, including the introduction of new tasks, changes in the comparative advantage of labor relative to capital, the deepening of automation (whereby machines become more productive in tasks that are already automated), and the role of the elasticity of substitution and capital accumulation in the long-run adjustment of the economy.
We gratefully acknowledge financial support from Google, Microsoft, the Sloan Foundation and the Toulouse Network on Information Technology. The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.
Daron Acemoglu & Pascual Restrepo, 2018. "Modeling Automation," AEA Papers and Proceedings, vol 108, pages 48-53. citation courtesy of