Investor Reactions to Generative AI Usage in MD&A Disclosures

Abstract

This study examines the impact of generative artificial intelligence (GenAI) usage on investor reactions by focussing on the use of GenAI to create or modify ‘Management’s Discussion and Analysis of Financial Condition and Results of Operations’ (MD&A) sections of 10-K filings. GenAI applications such as ChatGPT have quickly become integral to various fields, including accounting. Analyzing 6,977 firm-year observations from filing years 2021 to 2023, we find that using GenAI for MD&A texts is associated with adverse short-term investor reactions measured by signed and absolute cumulative abnormal returns (CAR). Using causal mediation analysis, we show that GenAI enhances financial disclosures’ readability, leading to greater CAR. Tone, in contrast, remains unaffected, suggesting firms prioritize simplifying over tone manipulation. However, the mediation effect is outweighed by the negative direct effect, leading to a significantly negative impact of GenAI usage in narrative disclosures on investor reactions. These insights have implications for policymakers and practitioners. Regulatory frameworks guiding GenAI use in financial reporting could mitigate unintended effects, while preparers should carefully weigh efficiency gains against potential risks to investor trust. Our study provides first evidence on capital market implications of GenAI usage in financial disclosures, contributing to research on technology-driven shifts in financial reporting.

Keywords: Generative AI, Investor Reactions, MD&A, Financial Reporting

JEL Classification: C45, G14, G15, G41, M41