中文摘要：

样本选取2011年和2012年我国沪深A股制造业中因财务困境陷入ST的公司和按照1：2比例配比的正常公司作为研究对象,并选取反映企业盈利能力、股东获利能力、现金流量能力、营运能力、发展能力、偿债能力的30个财务指标以及股权结构、管理结构、公司所在地的8个定性指标,以2011年的样本作为训练集,2012年样本作为测试集,在主成分分析的基础上构建以Logit模型为基础的传统预警模型和引入TOPSIS法后的二重分类模型。结果表明,引入TOPSIS法后构建的Logit模型能显著提高模型的预警准确度：对ST公司的预警准确度能提高18.5%,对样本总体的预警准确度能提高11.1%,这说明二重分类法可以构建有效的风险预警模型。

英文摘要：

Taking the indebted ST listed companies and twice as many the non-ST of manufacturing industry in Shanghai and Shenzhen A-share from 2011 to 2012 as sample,30 financial indicators that can reflect the business profitability,profitability of shareholders,cash flow capacity,operational capacity,development capacity and debt solvency and other 8 qualitative indicators including ownership structure,management structure and location of the companies are selected.Considered the sample of 2011 as a training set while the 2012 as testing set,a traditional warning model and double classification model in which TOPSIS method are introduced on the basis of the principal component analysis are built.The results show that the introduction of TOPSIS in risk early-warning model can increase the accuracy of early-warning significantly.It can improve the accuracy of early warning system 18.5%for ST companies and 11.1%for samples in general.This suggests that the method of double classification can be applied to build an effective risk early-warning model.