Analyst Reports and Corporate Financial Distress Prediction

Security analysts publish research reports on publicly listed companies periodically, providing crucial information for capital market participants. By extracting quantitative variables of stock recommendations and earnings forecasts and qualitative variables of readability and negative tone from analyst reports, this study examines the impact of analyst report information on corporate financial distress prediction (FDP). Using the asymmetric bagging method for handling imbalanced data and a light gradient boosting machine ensemble classifier, we construct several FDP models by stepwise incorporating the quantitative and qualitative variables from analyst reports, alongside benchmark variables such as financial ratios, corporate governance and analyst report attention, to investigate the incremental effect of analyst report information on corporate FDP. Empirical results based on a sample of Chinese listed firms from 2013 to 2021 show that the inclusion of analyst report variables significantly improves the performance of corporate FDP models. We further find that these variables exhibit heterogeneous abilities to identify the different types of financial distress arising from distinct reasons. Feature importance results indicate that qualitative attributes of analyst reports, particularly negative tone and readability, receive higher relative importance than traditional quantitative signals. Further interaction effect analysis indicates that high‐quality analyst reports exhibit relatively stronger predictive power for financial distress. Following the convergence of International Financial Reporting Standards, the predictive power of quantitative information in analyst reports weakened, while qualitative content demonstrated a stronger ability to predict financial distress.