LLM-based Methodology for Measuring MD&A Disclosure Quality and Its Application to Credit Rating Prediction

Key Points

• Quantifying MD&A quality reveals significant positive relationships with credit ratings, impacting financial assessments.
• The regression analysis indicates that the 'financial condition' and 'disclosure volume' are key predictors of credit ratings.
• Using machine learning, the XGBoost model with disclosure quality variables shows improved performance metrics like F1 score and recall.
• These findings support the use of LLM for enhancing the efficiency and accuracy of credit risk predictions based on text data.