点击次数：

论文类型：会议论文

发表时间：2009-01-01

收录刊物：CPCI-S、CPCI-SSH

页面范围：148-153

关键字：mixed logit; factor analysis; credit risk; prediction

摘要：This paper used mixed logit model to predict credit risk of listed companies in China. In order to reduce the difficulty in dealing with the facts of correlation and multi-dimension of the financial indexes and meanwhile to ensure that the information is not lost, we introduced factor analysis to the mixed logit equation and thereby constructed a Factor Analysis Mixed Logit Model. Fifteen factors were extracted from original financial indexes, and four main factors which affect dramatically were selected to substitute the original financial indexes as explanatory variables. The empirical results showed that though prediction accuracy of the new method is a little lower than mixed logit model's, it can reach more than 80%. The new approach presented in this paper is more reliable because it eliminates correlation of explanatory variables and reduces information loss effectively.