我国上市公司财务危机预警实证研究——基于主成分分析模型、线性判别模型和逻辑回归模型的比较分析

Empirical Research on Forewarning of Financial Crisis of Chinese Listed Companies: Comparison among Principle Component Analysis Model,Linear Discriminating Model,and Logical Regression Model

以我国上市公司为研究对象，根据行业分类和总资产规模（相差在5％以内）选取了2001～2002年被ST的公司和正常公司各28家作为训练样本，2004年被趼的公司和正常公司各28家作为检验样本，运用3种独立的建模方法，分别建立了我国上市公司的主成分分析预警模型、线性判别预警模型和逻辑回归预警模型。通过模型比较发现，逻辑回归模型的效果要优于另外两种模型。具有较高的判别准确率。实证研究表明，这种预警模型适用于各个行业，对中国企业的财务危机的预测与防范能起到一定的作用。

Based on industry categories and general asset scale(within 5% margin),this paper regards Chinese listed companies as the research objectives.Twenty-eight companies which have been specially treated and the same number of normal companies from 2001 to 2002 have been selected as the training samples.By using three kinds of independent methods of setting up models,the principle component analysis model,the linear discriminating model,and the logical regression model have been established.It is then found out that the logical regression model is better than the other two models with higher degree of accuracy.According to the empirical research,such model can be applied to different industries and can be used to predict financial crisis of Chinese enterprises.