中国上市公司财务困境预测的实证研究 ――以制造业为例.pdfVIP

摘 要 近年来，我国不少上市公司由于经营管理不善频频陷入财务困境，给投资者 和债权人带来了重大损失，也给证券市场的健康发展带来了不良影响，因此对财 务困境进行预测研究显得非常必要。本文以中国制造业上市公司为研究对象，对 财务困境公司和非财务困境公司按照 1：1 的比例进行配对，共选取了 218 家公 司作为的研究样本。首先比较检验了财务困境公司与正常公司在财务困境发生之 前两者在财务指标上的差异，从中选出具有显著性差异的指标，然后进行多重共 线型检验，去掉了存在高度相关的指标，得到了 18 个财务指标变量，作为预测 两类公司的基本变量。在此基础上，本文用单变量分析法、多元线型判别分析法 和多元逻辑回归分析法，三种不同方法构建相应的模型，来预测上市公司在未来 发生财务困境的可能性大小。研究表明，多元逻辑判别分析法的预测效果最好。 另外，本文还特地从公司治理的角度研究两类公司的差异，选取了反映股权结构 的变量结合财务指标变量共同构建预测模型，使模型在中长期预测准确性有了一 定的提高。 关键词：财务困境预测 多元线型判别分析 多元逻辑回归 上市公司 I Abstract In recent years, many listed companies run into financial distress frequently in China, coursing great damages to investors and creditors, and dampening security markets development. So, it is necessary to do some predictive studies on financial distress. This paper takes Chinas manufacturing industry as sample. We match each ST company with one financially healthy company of equal size, and select 218 companies in total. First, we use t-test to compare financial indexes between healthy companies and distress companies before they go distress, and find some significant variables. In order to avoid the multicollinearity, we omit variables with large correlation coefficient, and 18 variables remain, which are basic variables to predict. After these, Single Variable Discriminant (SVD), Multivariable Discriminant Analysis (MDA), and Multivariable Logistic Analysis (MLA) are used to predict the probability of financial distress of listed companies in this paper. Research shows that the MLA’s correct percent of prediction is the highest among three methods given above. Finally, we study the differences between two type