Unpacking social exposure

Authors: Laurence van Lent, Grigory Vilkov, Zacharias Sautner, Mingyang Liu, Ruishen Zhang
Cited by: 0

Abstract

We develop a method to measure firms' exposure to social issues from earnings-call transcripts. The method applies large language models and decomposes social exposure into five categories: Employee Welfare, DEI, Stakeholder Outreach, Ethical Commitments, and Crisis Response. The measures cover over 15,000 firms across 93 countries from 2003 to 2024. These categories exhibit distinct, sometimes opposing, associations with labor outcomes, productivity, and ESG ratings. Aggregating them attenuates estimated effects, explaining weak findings in prior work. Exploiting the Dobbs decision as a shock to labor supply, we find that firms with higher pre-Dobbs DEI exposure have lower worker outflow rates in states enacting abortion bans.

Fields of Study & Subjects

Corporate Social Responsibility Reporting Computational and Text Analysis Methods Political Influence and Corporate Strategies

Publication details

Journal / source
OSF Preprints (OSF Preprints)
Publication date
28 January 2026
Document type
Preprint