摘要
鉴于现阶段我国上市公司的信用数据具有高维性和高相关性的特点,已有企业财务危机预警研究多采用能够有效降维和消除共线性的主成分logistic模型。然而,这种模型定式在提取主成分时没有考虑解释变量与被解释变量之间的相关性,可能导致与企业财务状况关系密切的解释变量信息的丢失,从而削弱模型的预测能力。考虑到这一缺陷,本文在分析中首次引入偏最小二乘方法,并对我国沪深两市上市公司的经营失败进行了实证研究,结果表明偏最小二乘logistic模型不仅具有较优的拟合度,而且具有较高的企业经营失败预测能力。Bootstrap检验显示模型还具有较强的稳健性,预警效果更为可靠。
Considering the characteristics of high correlation and high dimension of the credit data of listed companies in China,previous literatures often utilize PCA logistic model to realize variables independency and decreasing dimension.However,this approach only takes into account the information from independent variables when extracting PCA.This paper,for the first time,draws PLS method into the traditional logistic model to empirically predict the Chinese corporate distress.Our findings show that the PLS logistic model has better fitness,high distress prediction ability,and bootstrap test further shows the new approach has greater robustness.
出处
《管理工程学报》
CSSCI
北大核心
2010年第4期100-103,共4页
Journal of Industrial Engineering and Engineering Management
基金
教育部人文社会科学重点研究基地重大项目(08JJD63008)
国家自然科学基金资助项目(70802032)
中国博士后科学基金(20090460681)