基于截尾分位数回归模型的上市公司财务困境影响因素研究

Analysis of Factors Affecting Financial Distress of Listed Companies in China Based on Censored Quantile Regression Model

用生存分析方法研究财务困境,通常用Cox比例风险模型来建模,但Cox比例风险模型的应用需要满足两个假定,现实中难以同时满足。为解决Cox比例风险模型的不足,将截尾分位数回归模型应用到生存分析中。该模型对生存时间直接进行建模,增加了模型的可解释性,为分析生存数据提供了一种有效的方法,并可借此对上市公司发生财务困境的影响因素做出判断。实证结果表明,在上市公司财务困境研究中,将截尾分位数回归模型和Cox比例风险模型的实证结果相比较,发现截尾分位数回归模型的结果更为准确。

The Cox proportional hazards model for survival analysis is often applied to studying the problem of financial distress. However, this model needs two difficult assumptions on the behavior of conditional survival functions. So the censored quantile regression model is introduced to survival analysis. The modeling on survival times is made directly with interpretable regression coefficients, which improves the interpretability of the model. It not only provides an effective way for survival a- nalysis, but also is able to identify the key factors affecting the financial distress of listed companies correctly. The results of empirical analysis on listed companies in China show that the censored quan- tile regression model performs better than the Cox proportional hazards model in the study of financial distress.