The Spillover Effects of Auditor Sanctions on Client Firms: Evidence From Deep Learning on Investors’ Online Voice in China

This study examines the spillover effects of auditor sanctions on client firms and validates the extent to which social media communications empower traditionally weak investors’ voice. In the context of China which offers unique data from regulated interactive platforms, the study examines the effects of auditor sanctions on investors’ online voice, employing the DID model. Results suggest that auditor sanctions increase the number of questions and number of words in the questions asked by investors about client firms on online interactive platforms. The deep learning Pre-trained BERTBASE model for sentiment analysis demonstrates that auditor sanctions also significantly increase the number of questions with negative sentiments. The results still hold after multiple robustness tests. Such adverse spillover effects of auditor sanctions are more pronounced among non-state-owned client firms and firms with higher shareholding concentration. This study contributes by enriching the application of deep learning models in text analysis in the social sciences, validating the importance that investors attach to auditor sanctions and thus the effectiveness of government regulation.