Abstract

This study examines how occupational AI exposure affects employment at the intensive margin, i.e., workday length. Using individual-level time diary data from 2004–2023, we find that greater AI exposure, whether from the ChatGPT shock or broader AI developments, is associated with longer work hours and less leisure, particularly non-screen-based leisure. Guided by a principal-agent framework, we find evidence for three channels: higher marginal productivity from AI-human complementarity, improved contracting efficiency from AI-enabled monitoring, and lower worker reservation utility, reflected in declining job satisfaction unrelated to salaries or employment risk. The workday extension is most pronounced in competitive labor and product markets, where weak worker bargaining power shifts productivity gains to firms or consumers. These findings challenge the view that technology reduces labor burdens, showing that it can erode work-life balance.

...read more at papers.ssrn.com