论文部分内容阅读
我国上市公司一般具有较大的资产规模与较强的盈利能力,发生财务困境的概率较低,数据呈非平衡性。采用传统Logistic回归会受到因变量分布不平衡的影响。本文将西方学者在医学现象研究中普遍使用的稀有事件Logistic回归引入我国上市公司财务困境预测的研究中,根据财务困境发生实际概率确定样本观察单位的权重,构建Relogit回归模型。研究结果表明,Relogit模型预测效果优于传统Logistic模型。
Generally speaking, the listed companies in our country have larger assets scale and stronger profitability, the probability of financial distress is lower, and the data are unbalanced. The use of traditional Logistic regression will be affected by the unbalanced distribution of dependent variables. This paper introduces the Logistic regression of rare events commonly used by western scholars in the study of medical phenomena into the study of financial distress prediction of listed companies in China. The weight of sample observation units is determined according to the actual probability of financial distress and the Relogit regression model is constructed. The results show that the Relogit model is better than the traditional Logistic model.