Abstract

This paper examines the application of LDA topic modelling to risk disclosures in FTSE350 firms’ annual reports. We show that LDA implementation choices significantly impact topic representations and subsequent inferences. Using a corpus of FTSE350 annual reports, we show that preprocessing decisions, multiword expressions, and labelling strategies materially affect topic interpretability and granularity. Our analysis reveals that while risk reporting addresses key business risks at an aggregate level, the degree of firm-specific commentary is sensitive to topic granularity. Hierarchical linear modelling suggests that 27% of topic variation is within firms for broad topics, increasing to 75% for granular topics. We leverage GPT to enhance topic labelling, showcasing the potential of LLMs in financial text analysis. These findings challenge the claims of LDA objectivity and highlight the importance of domain expertise. We propose a practical checklist for LDA implementation in accounting and finance research emphasizing transparency and robustness checks.

Professor Steven Young

Is a Professor of Accounting, Deputy Dean, and Associate Dean for Resources (Finance) at Lancaster University Management School. His research interests cover topics at the interface between accounting and corporate finance, including executive remuneration, narrative reporting and natural language processing applications, and financial reporting quality. Professor Young’s research has featured in leading academic journals including The Accounting Review, Review of Accounting Studies, and Accounting, Organizations and Society. He is an Associate Editor for Journal of Business Finance and Accounting, an editorial board member of several journals including Accounting, Organizations and Society and Accounting and Business Research, and Chair of the European Accounting Association Publications Committee.