The information value of news in predicting defaults of corporate bond issuers

Speaker：Bu Hui ,Associate Professor/Department Chair of Finance, School of Economics and Management, Beihang University

Host：Zeng Yan , Professor, Lingnan College

Time and Date：14:30, Dec. 4, 2023

Venue：Lingnan Hall (214) 

Language:  Chinese

Abstract：

This study investigates the information value and predictive power of news headlines for predicting the first-time defaults of corporate bond issuers in the Chinese market. By integrating domain knowledge-derived topics with cause-effect logic into a newly proposed credit risk dictionary, we formulate firm-level news-related variables extracted from news headlines. This study explores whether news offers incremental information value compared to financial ratios and economic variables, examining the specific information covered, identifying the media categories providing this incremental information, and evaluating the predictive power and early warning capabilities of news-related variables. The discussions are made by adopting the static logistic regression model, an interpretable AI model, based on all default events since 2014 to 2022 in China. The findings reveal that news headlines encapsulate firm-specific incremental information beyond conventional financial ratios and economic variables, showcasing predictive power. The integration of cause-effect domain knowledge-derived topics yields more informative insights than individual words in text analysis. News can unveil information related to business operations, company management, and associated risks, aspects challenging to measure otherwise. Our results exhibit robustness. Furthermore, news variables exhibit early warning capabilities with forecasting horizons of 3 months or longer, offering valuable insights for credit risk management practices. These discoveries deepen our comprehension of news-driven credit risk assessment, bearing implications for financial practitioners and policymakers.

Research Area:

Financial Engineering, Financial Market Microstructure, Empirical Asset Pricing, Forecasting and Risk Management, etc.