【作者】 程守红；

【导师】 左小德；

【作者基本信息】 暨南大学 ， 企业管理， 2004， 硕士

【摘要】 以是否被特别处理为财务危机标志，本文利用我国上市公司近几年的年报财务数据，选取了50家财务危机企业和50家健康企业作为开发样本。借助SPSS统计分析软件，首先研究了开发样本财务指标的数据特征和财务危机出现前3年内这两类企业28个财务指标的显著性差异，其后利用因子分析法筛选了8个指标作为建立模型的自变量，构建了多元线性判别模型和逻辑斯蒂回归模型，最后应用逻辑斯蒂回归模型对估计样本的40家企业进行了检验。 实证分析结果表明：其一，我国上市公司财务比率不服从正态分布，因而不适宜使用多元线性判别分析方法建立财务危机预测模型；其二，尽管在只有十年历史的我国资本市场中，会计数据质量不尽人意，但财务数据仍具有一定的信息含量，对企业利益相关者预测公司的财务危机有较大的帮助，只是财务数据的有效预警期为财务危机前两年；其三，对于同一信息集，逻辑斯蒂回归模型优于多元线性判别模型；其四，利用开发样本构建的逻辑斯蒂回归模型在财务危机发生前1年和前2年分别有95.87％和84％的回判准确率，而对于估计样本在财务危机发生前1年和前2年则分别有95％和82.5％的预测准确率。更多还原

【Abstract】 With the sign of whether the corporation is "special treated" or not, the thesis uses the financial datas of China’s listed companies in recent years and selects the initial samples which are composed of 100 enterprises with 50 firms in each of the two groups. In virtue of SPSS, firstly, the text discusses the data characterictics and the differences of predictor variables between the two groups for up to three years prior to the failure event. Then the study uses Factor Analysis to reduce datas, and establishes Linear Multiple Discrimant Analysis Model and Logistic Regression Model. Lastly, the research tests 40 firms of the holdout samples by Logistic Regression Model.The results are as the following: firstly, financial ratios aren’t identical to the normal distribution, so it’s not suitable to conduct the early-warning model by Linear Multiple Discrimant Analysis. Secondly, the accounting datas are not imprecise, but they show some information. Thirdly, for the same datas, the model by Logistic Regression is more precise in predicing financial distress. Lastly, the model by Logistic Regression can yield an overall correct classification accuracy of 95.87% one year prior to failure and of 84% two year prior to failure to the initial samples. The model is validated by using the holdout samples, the overall correct classification of the model for the first year to two year prior to distress is 95% and 82.5%.更多还原