A hybrid model to estimate corporate default probabilities in China based on zero-price probability model and long short-term memory

ABSTRACT

This article proposes a hybrid model by combining zero-price probability model with long short-term memory (ZPP-LSTM) to estimate corporate default probabilities. The ZPP-LSTM model enhances the time-series data forecast by introducing LSTM in ZPP model, which can better estimate the corporate default probabilities in the industry sensitive to an uncertain environment. The full samples of Chinese listed companies in construction and real estate industries are selected to evaluate the performance of ZPP-LSTM model. The results show that our proposed model outperforms other benchmark models in terms of the default probability estimation.

KEYWORDS:

• Default probabilities
• zero-price probability model
• long short-term memory
• listed companies

JEL CODES:

• G30
• G33

Acknowledgments

We would like to thank an anonymous referee for providing many constructive comments and suggestions that helped us significantly improve the article. We are also grateful for the comments from Dr. Timothy Wang.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China (NSFC) [71872094, 71473145].