The Remainder Effect: How Automation Raises Returns to Detailed Skills
Boston Univ. School of Law Research Paper No. 22-3
62 Pages Posted: 25 Feb 2022 Last revised: 18 Aug 2022
Date Written: February 24, 2022
Abstract
Using help-wanted ad data, this paper argues that automation increases demand for detailed skills that are typically unobserved, but which are major determinants of pay. Following automation events, we find that employers request more detailed skills and they substantially increase pay offers (8.7%). Importantly, these increases are not limited to select occupational groups—they apply to both routine and non-routine jobs, to jobs requiring college and those that do not. To explain this phenomenon, we extend the Acemoglu-Restrepo task-based model of automation to consider labor quality, which depends on workers having task-specific skills. We obtain a Remainder Effect: when automation displaces labor on some tasks, it raises the returns to specific skills on the remaining tasks performed by diverse occupational groups. Because not all firms automate, this effect can raise income dispersion within occupations and between firms, including the sorting of skilled workers to high-paying firms. In contrast, labor displacement alone tends to increase between-occupation pay differences.
Keywords: automation, income inequality, skills, information technology, software
JEL Classification: J31, O33, J23
Suggested Citation: Suggested Citation