No. 229: When Cheap Talk Has Bite

We ask when voluntary disclosure about hard-to-verify risks predicts later adverse outcomes. We study this in human-rights due-diligence (HRDD) discussion on UK earnings calls, where direct verification is weak and delayed labor-rights enforcement outcomes provide noisy independent checks on earlier talk. Across UK earnings-call transcripts from 2007–2024, within-firm increases in HRDD discussion predict next-quarter labor-rights controversies: a one-standard-deviation increase raises next-quarter incident probability by 1.33 pp and expected incident counts by 7.46%. The same discussion also predicts Employment Tribunal cases two years ahead, while workplace-safety prosecutions serve as a placebo with no predictive content. The predictive content is strongest in executive HRDD intensity and in concrete sub-topics (supply chain, remediation); generic stakeholder language and analyst HRDD intensity carry none. Predictive content strengthens after COVID-19 and after the 2020 Boohoo scandal for treated fashion firms, and repeated supply-chain language loses content as it becomes routine. These findings identify two conditions under which voluntary disclosure about hard-to-verify risks carries information: external scrutiny that makes misrepresentation costly, and language concrete enough to be verifiable. The findings imply that mandating more disclosure without strengthening the institutions that verify it may produce more talk without more information.