Predicting defaults in China's credit bond market
Key researchers
- Bei Cui
- Priscila Grecov
Project background and aim
For this paper, researchers at the Monash Centre for Financial Studies in Melbourne developed a new and innovative analytical model focused specifically on assessing default risk in the Chinese credit bond market. Extensive testing of the model reveals its dramatic outperformance of both the Chinese domestic ratings agencies and major international agencies in its ability to predict which Chinese credit bonds will default.
Key findings:
- The exclusive use of internal detailed data on Chinese bond-issuing companies, joined to machine learning techniques and models, leads to substantially more accurate predictions of bond defaults than existing methods.
- Our proposed method is substantially more reliable than the classical Z”-Score method, achieving an accuracy rate of more than 90% across a sample group of 5,533 bonds.
- Our method potentially fills a major knowledge gap in the Chinese bond market, which is currently beset by low coverage by international credit rating agencies.
- Our method can work as an early warning and default-risk flag tool for foreign investors considering adding Chinese domestic credit bonds to their portfolios.
- Investors can also run this modelling internally to perform periodic assessments of the issuer’s default risk based on timely financial data.