Unveiling Themes in 10-K Disclosures: A New Topic Modeling Perspective

We investigate the topics in the Management’s Discussion and Analysis (MD&A) section of 10-K filings. Our approach extracts MD&A topics by clustering words around anchor words that broadly define potential themes. The resulting topics are intelligible, distinct and multi-faceted, shedding light on why classical topic models applied to 10-K filings might lack interpretability. We extract two loading series from the MD&As: topic prevalence and topic sentiment. We find that topic prevalence exhibits significant variation throughout the sample period, while sentiment displays marked heterogeneity across topics. Linking MD&A topics to stock returns, we document non-uniform market perceptions toward the topic sentiment.