Automation and the Phillips curve
Could the surge in automation explain the muted sensitivity of inflation to unemployment observed in advanced economies until the COVID-19 pandemic? This is the question Basso and Rachedi attempt to answer in this paper. Their empirical approach consists of building a panel of non-tradable goods inflation, wage inflation, unemployment rate, and robot adoption at the U.S. metropolitan area (MSA) level. To measure automation, they combine the robot installation for each industry at the U.S. national level with the employment share of each industry at the MSA level. Their panel includes data across 384 MSAs at an annual frequency from 2008 and 2018. Their theoretical approach consists of extending a standard New Keynesian model with two key features: search frictions in the labor market and the possibility of robot adoption. The nominal interest rate is set by a Taylor rule that responds to changes in inflation and the unemployment gap. They find:
- First, automation shrinks workers’ bargaining power, dampening the sensitivity of wages to unemployment. Second, automation reduces the labor share, decoupling output changes from unemployment variation. Both channels flatten the price Phillips curve.
- An increase in robot adoption by one standard deviation reduces the sensitivity of price inflation and wage inflation to unemployment by 17% and 9%, respectively.
- When boosting automation is costly, the threat of robot adoption is no longer effective in curtailing workers’ bargaining power amidst large expansionary shocks, leading to a steeper Phillips curve.
Robot Adoption and Inflation Dynamics
Authors: Henrique S. Basso, Omar Rachedi
From: Banco de España, ESADE Business School
