Abstract:
We study a 2018 reform in South Korea that reduced tax credits for automation investments. This reform increased the tax cost of investing in robots and thus resembles a robot tax. Exploiting this natural experiment with industry-level data on robot installations and firm-level data from Orbis, we document a sharp decline in automation investments after the reform in industries with a large share of affected firms. At the firm level, we find that affected firms increased employment, consistent with the notion that robots replaced workers. The effects are heterogeneous: financially constrained firms cut investment overall, while unconstrained firms substituted away from robots, hired more workers, and reallocated resources toward more productive uses. For the latter group, we find improvements in various measures of investment quality, suggesting that the tax credit induced inefficient overinvestment in automation. Our evidence informs ongoing debates on robot taxation and the efficiency of tax incentives.