企业破产预测实证模型评述  

The Enterprise Bankruptcy Forecast Empirical Model

作　　者：王利娜[1] 

机构地区：[1]中国社会科学院经济研究所,北京100083

出　　处：《河北经贸大学学报》2012年第3期51-54,共4页Journal of Hebei University of Economics and Business

摘　　要：随着计量方法的改进,对企业破产的实证预测,重新成为理论界和实务界关注的重点。40年来代表性的破产预测实证模型包括专家评分系统、分离模型、多元分离模型、Logit模型、生存分析模型、神经网络模型等。对于每种模型的假设前提、所选择的解释变量、预测能力以及可能的局限性进行评述,可以得出结论:多元分离模型虽然传统,但实用性最强;分时风险模型因为改善了分离模型的静态性,预测准确率也更高,是目前应用最广泛的模型。With the improvement of the measurement method,the empirical prediction of the enterprise bankruptcy again becomes the critical concern of theory and practice.The representative bankruptcy predicted empirical model of 40 years includes experts marking system,separation model,multiple separation model,Logit model,survival analysis model,the neural network model,etc.By conducting recitation for each model hypothesis,the choice of explanatory variables,and the limitation of the prediction ability,we can draw the conclusion that the multiple separation model is traditional,but its practical applicability is the strongest.Time-share risk model can improve the quiescent condition of the separation model,its predict accuracy is higher too,and is now one of the most widely applied model.

关 键 词：破产预测 多元分离模型 分时风险模型 

分 类 号：F275.2[经济管理—企业管理]