Research on the Measurement of Subject Credit Risk of Chinese Port Enterprises by Constrained Logistic Regression.

This paper studies the subject credit risk of Chinese port enterprises. Since the impact of cash flow ability on credit risk measurement will be increased under extreme case, ordinary logistic regression methods may lack explanatory power for port enterprise default under extreme cases. Considering the characteristics of cash flow in port industry, we introduce the constrained logistic regression method to establish a default probability model which can describe the credit risk level of the industry with higher accuracy in the extreme case where an enterprise’s quick ratio is lower than a cutoff point, For empirical study, we leverage the data of more than 900 companies in port and transportation industry in 2016–2018. The constrained logistic regression splits the data into two subspaces based on quick ratios with the cutoff of 1.8. Then logistic regression is built on the two subspaces, respectively. The recall ratios show that the constrained logistic regression method performs better than the ordinary logistic regression on the study of corporate default probability in port and transportation industry.