基于Logistic回归的上市公司亏损逆转识别模型研究

Research on Recognition Model of Loss Reversal of Listed Company Based on Logistic Regression

为了建立合理有效的上市公司亏损逆转识别模型,本文分别从亏损历史状态、公司财务状况、股利支付情况及投资者预期因素四个层面选取18个指标作为备选变量,并运用Logistic过程中逐步回归的方法从中选择了7个对因变量影响显著的自变量,然后以2005年度发生亏损的我国127家上市公司为样本,分别对亏损逆转样本组(88家)和非逆转样本组(39家)进行Logistic回归检验,并建立了最终的基于Lo-gistic回归的上市公司亏损逆转识别模型,最后通过检验样本组回代判定,发现该模型对识别亏损公司在亏损后第二年度是否会发生逆转情形具有较高的预测准确性。

In order to establish a rational and effective loss reversed recognition model of listed company,this paper chooses 18 selected independent variables from the four factors including the loss of state history,the company＇s financial situation,dividends paid and investors expected,and uses the method of Gradual return in Logistic process to choose the seven independent variables affecting significantly the dependent variable. Then it makes the 127 loss listed companies in 2005 as samples,and tests the loss reversal sample group（88） and the non-reversal samples（39）group in Logistic regression,and establishes the ultimate recognition model of loss reversal based on Logistic regression for the listed company. Through the final judgment,it finds that this model to identify loss reversal in the second year after the loss have a high forecast accuracy.