Beyond the Lines: Uncovering Private Information from Fund Managers' Disclosures

Abstract

We identify private information from mutual fund managers' narratives in their letters to shareholders with the help of large language models (LLM). Informed funds identified by LLM deliver superior abnormal returns and are more likely to receive an upgrade in Morningstar ratings. When interpreting the textual features associated with private information in fund disclosure, we find that informed funds present more forward-looking discussions and issue more modest and cautious statements. In most cases, investors recognize such qualitative disclosure and reward funds that disclose private information with greater flows, but only if those investors actively acquire the disclosure. Furthermore, we show that managers' narrative disclosures shed particular light on their skill when funds' quantitative information is hard to evaluate, suggesting the importance of qualitative disclosures.

Keywords: JEL Classification: C45, G11, G14, G23 Machine Learning, Large Language Models, Mutual Fund Performance, Disclosure, Textual Analysis, Shareholder Letters, Fund Flows

JEL Classification: C45, G11, G14, G23