【摘要】 为了克服直接使用财务指标建立Logit模型存在的估计和检验不可靠问题,在所有的财务比率指标基础上,通过删除完全共线、意义相同和缺失数据指标,建立海选指标集.利用t检验筛选指标,删除t检验不显著的指标,保留t检验显著的指标,构建财务危机预警指标体系.通过提取互不相关的Pls因子,估计Pls-logit模型对因子赋权.将Pls因子回归代入Pls-logit模型对指标赋权,建立基于Pls-logit模型的财务危机预警模型.以2010年381家制造业上市公司作为建模样本进行了实证分析,表明Pls-logit模型中31个系数在5%的显著性水平下显著,说明建立的财务预警指标体系能够显著解释财务危机.以2010年和2011年随机抽取的制造业公司为检验样本进行回溯检验,检验结果表明预警总准确率达到91.7%以上.2010年危机预警准确率为66.7%;2011年危机预警率为100%.Pls-logit模型的总体预警准确率明显优于财务指标logit模型.更多还原

【Abstract】 Based on the modeling sample of 381 manufacturing listed companies in 2010,collinear index,same meaning index and missing data index was removed off the financial ratio index set,the rest index were remained as the original financial index set.The t test was applied to index selection based on the original financial index set.The financial index of t test not significant was deleted.The rest set of financial index was refered to the prediction financial index system,which had the ability to identify the financial crisis.The Pls-logit model was estimated with the unrelated Pls factor which were extracted by the Pls method.The financial crisis prediction model was deduced by substituting the relation of Pls factor and financial index into Pls-logit model.Based on the random sample in 2010 and 2011,prediction accuracy rate of Pls-logit model was checked by the return test.The t test of 31 coefficient was significant under 5% level,which suggested that the prediction financial index system could significantly predict the financial crisis.On the contrary,it was unreliable of the coeffecient esimation and test for the financial indicators logit model due to the multi-collinearity.For the Pls-logit model,the total prediction accuracy was more than 91.7%;Crisis prediction accuracy was 66.7% in 2010;Crisis prediction accuracy was 100%in 2011.It was suggested by the comparative analysis that the prediciton accuracy of Pls-logit model was superior to the financial index logit model.更多还原