Robots and Industrial Labor: Evidence from Japan

39 Pages Posted: 23 Sep 2020

See all articles by Robert Dekle

Robert Dekle

University of Southern California - Department of Economics

Date Written: August 9, 2020


This paper provides one of the first studies of the impact of the introduction of robots to equilibrium labor demand in an entire economy. We use long-term (1979-2012) industry-level panel data from Japan, a country that is over 10 times as intensive in the use of robots as in the U.S.

Our model, which is derived from Acemoglu and Restrepo (2017), shows that there are three effects on labor demand in the aggregate economy from the introduction of industrial robots. The first is the negative displacement effect of robots taking over the tasks of humans. The second is the positive industry productivity effect of robots lowering costs in a particular industry. This lowering of costs draws demand to the industry and expands output and employment in that industry. The third is the positive general equilibrium effect. Robots raise productivity and expands demand in all industries, thereby increasing the product and labor demand for the industry introducing the robots.

In our estimation, we find that the displacement effect is insignificant, the productivity effect is sometimes positively significant, and the macroeconomic general equilibrium effect is always highly positively significant. Overall, the introduction of robots has been beneficial to the demand for labor in Japan.

Our results are robust to various controls and instrumental variables.

Keywords: Robots and Labor, Wages, Japan

JEL Classification: E1, Fo

Suggested Citation

Dekle, Robert, Robots and Industrial Labor: Evidence from Japan (August 9, 2020). Available at SSRN: or

Robert Dekle (Contact Author)

University of Southern California - Department of Economics ( email )

3620 South Vermont Ave. Kaprielian (KAP) Hall, 300
Los Angeles, CA 90089
United States
213-740-8335 (Phone)

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
PlumX Metrics