Workplace Automation and Employment Outcomes: an Empirical Exploration

Abstract

I utilize longitudinal individual-level information extracted from the Panel Study of Income Dynamics (PSID) to explore how automation risk at the occupation level impacts subsequent income and employment, motivated by the contemporary surge in automation anxiety as well as by recent literature that documents the damaging aggregate employment trends attributable to the adoption of machine labor. I follow a task-based approach that explains the mechanism of automation and characterizes automation risk as a disproportionally high concentration of routine labor for a given occupation. I find counterintuitive evidence suggesting that automation risk is positively associated with income, and that a modestly elevated risk of involuntary job displacement appears to be the only meaningful negative impact of automation risk. Even after involuntarily displaced, individuals previously employed in highly automatable professions do not face negative income or employment shocks that are significantly different from those felt by workers displaced from other careers. I attempt to reconcile existing labor theory with the lack of major negative automation impacts estimated in this paper by citing the preference by employers to lay off workers rather than decrease wages, the skill substitution and complementation effects of automation, as well as the transferability of routine work experience due to the slow pace of automation. Finally, I find evidence of a reversal of the income effect of routine labor from positive in the 1980s and 1990s to a weakly negative association today, prompting additional research into the changing nature of automation.