Artificial Intelligence, Robots and Unemployment: Evidence from OECD Countries

Abstract

Investigating the relation between artificial intelligence, robots and unemployment on a panel of 33 OECD countries covering the 2005—2017 period, we find that a 10% increase in the stock of industrial robots is associated with a 0.42 point increase in the unemployment rate. For artificial intelligence (AI), we use patents as a proxy of AI-related technological capabilities and find a positive correlation with the aggregated unemployment rate, albeit statistically weaker than the one found for robots. We then run the regressions on unemployment rates differentiated by education and age, and observe highly heterogeneous effects between groups. For example, the effect of robots is 2.5 times greater for 25-34 year-olds below upper secondary education levels than for the 55-64 year-olds with a tertiary degree. Lastly, the effect of robots is strongest on the unemployment rate of people with a medium level of education, providing some evidence that robots could contribute to the polarization of the labor market. A similar effect is found with AI, but the results are less robust than for robots.