Machines Against Workers? - The Heterogeneous Impact of Robots on Union Strength - May 25, 2026

Sep 4, 2026

Does workplace automation weaken labor unions and thereby reduce workers’ voice in democratic politics? Dominant research claims a uniformly negative effect, yet standard economic automation theory omits unions and political science often targets average estimates. This paper argues automation’s relationship with organized labor is heterogeneous. It integrates task-based automation into strategic worker mobilization and collective bargaining, implying unions can be disempowered or empowered. Empirically, robotization’s effect varies across US regional labor markets, largely due to unobserved characteristics, emphasizing strategic mobilization.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages49

Reading Time2 h 3 min

AccessFree to download

Introduction

Automation and union strength

Robots, AI, and political economy debates

Task-based theory and bargaining

Related empirical and theoretical work

Machines Against Workers? The Heterogeneous Impact of Robots on Union Strength∗ Michael Becher† Daniel Stegmueller‡ This version: May 25, 2026 Comments welcome Abstract Does the increasing automation of the workplace weaken labor unions? The dominant hypothesis in current political economy research is that automation weakens unions and, as a result, undermines worker voice in democratic politics. However, the standard economic theory of automation does not consider unions, and recent political science studies focus on estimating average effects. We argue that the relationship between automation and organized labor is heterogeneous. We integrate task-based automation into a model of strategic worker mobilization and collective bargaining. Our new model suggests that automation can disempower but also empower unions. Empirically, we use quantile instrumental variable regression and find that the impact of robotization on unionization significantly varies across regional labor markets in the US. Consistent with the theoretical perspective, this heterogeneity is mostly due to unobserved characteristics. Our results suggest that research on the political economy impacts of automation should pay more attention to strategic worker mobilization. ∗We gratefully acknowledge extremely helpful feedback on previous versions of this paper from John Ahlquist, Carles Boix, Timm Betz, Michael Donnelly, Aina Gallego, Matthias Haslberger, Giovanna Invernizzi, Nikitas Konstantinidis, Daniel Kselman, Eddy Malesky, Erica Owen, Piero Stanig, participants at the 2nd IE-Rochester Political Economy Workshop, APSA 2024, EPSA 2024, University of Barcelona Workshop on AI & Politics, the Munk School at the University of Toronto, the Duke Political Economy seminar, the Comparative Politics Workshop at Princeton, and the PRG seminar at IE University. †IE University, School of Politics, Economics & Global Affairs, Madrid, Spain, michael. becher@ie. edu ‡Duke University, [daniel. stegmueller@duke. edu](daniel. stegmueller@duke. edu) 1 Introduction Does the increasing automation of the workplace undermine the capacity of workers to organize in labor unions? The answer to this question is relevant for scholars of the workplace and scholars of democratic politics. In the workplace, unions shape wages, working conditions, and wage inequality (Doellgast 2022; Farber et al. 2021; Pontusson, Rueda, and Way 2002) . In democratic politics, unions have been linked to preference formation, elections, policymaking, and political equality (Ahlquist 2017; Becher and Stegmueller 2021; Flavin 2018; Hartney and Kogan 2025; Hertel-Fernandez 2025; Kim and Margalit 2017; Leighley and Nagler 2007; Schlozman 2015; but see Rosenfeld 2014; Yan 2025) . Crucially, the strength of unions is not set in stone. It has varied widely over time and across space, within and across countries (Naidu 2022; Wallerstein and Western 2000) . Technological change may contribute to union decline, foster a revival, or not matter. The adoption of industrial robots is a leading example of automation technology that substitutes tasks performed by labor with tasks performed by machines. Since the 1990s, the number of robots at work has increased at least four-fold. A booming literature in the social sciences is dedicated to untangling the economic (Acemoglu and Restrepo 2020; Graetz and Michaels 2018) and political consequences of robotization (Anelli, Colantone, and Stanig 2021; Chaudoin and Mangini 2026; Frey, Berger, and Chen 2018; Schll and Kurer 2024) . More recently, the adoption of artificial intelligence (AI) in the workplace and the rise of industrial AI has spurred further debate about the functioning of democratic capitalism (Boix 2019; Gallego and Kurer 2022; Magistro et al. 2026; McElheran, Brynjolfsson, and Kroff 2025; Tan and Thelen 2026) . While scholars have paid less attention to how automation affects labor unions, the dominant view in the nascent literature is tha

FAQs

What central question does the paper address about automation and unions?

Whether increasing workplace automation undermines workers’ capacity to organize in labor unions, affecting worker voice in democratic politics.

How does the paper’s theory differ from the standard economic view of automation?

It incorporates unions by linking automation exposure to strategic worker mobilization and collective bargaining, predicting heterogeneous effects rather than a single average direction.

What does the empirical evidence show about robotization’s impact on unionization?

The impact of robotization on unionization significantly varies across regional labor markets in the United States.

Why does the paper argue this heterogeneity occurs?

It suggests the variation is mostly due to unobserved characteristics, consistent with the theoretical perspective on strategic mobilization.

Sep 4, 2026

Uploaded by Aria Callaghan