Privacy-Preserving Information Sharing within an Audit Firm

This paper explores the possibility of sharing firm-level information within an audit firm in a privacy-preserving manner and demonstrates the benefits of doing this, under the assumption that the same audit firm serves multiple clients competing in the same industry. We observe significant improvements from sharing contemporaneous information from peer companies both in estimation accuracy and error detection performance. Additionally, we introduce an empirical approach for utilizing contemporaneous accounting information from peer companies without violating clients' confidentiality. We find that auditors can achieve comparable level of benefits from only sharing self-generated estimation residuals (errors) with that from sharing prediction or actual accounting numbers both in estimation accuracy and in error detection performance. To satisfy stricter privacy concerns, we propose a scheme based on sharing categorical information derived from prediction errors