基于多元判别分析和神经网络技术的公司财务困境预警  

Financial Distress Early Warning Based on MDA and ANN Technique

作　　者：张玲[1] 陈收[1] 张昕[1] 

机构地区：[1]湖南大学工商管理学院,湖南长沙410082

出　　处：《系统工程》2005年第11期49-56,共8页Systems Engineering

基　　金：国家自然科学基金资助项目(70172018)

摘　　要：应用我国上市公司的财务报表数据,采用多元判别分析(M DA)技术和神经网络(ANN)技术对我国上市公司进行财务困境预警研究。实证结果显示这两种方法都具有较好的预测效果,其中ANN模型的短期预测效果优于M DA模型的预测效果,但无明显的优势。研究同时证明在现有会计制度和会计准则下,财务报表能提供预测财务困境的大量有用信息。财务危机有踪可寻。This paper makes use of the financial statements of listed companies in China Stock Exchange and build a successful model for corporate failure diagnosis by MDA and ANN technique. The results show that both models can predict corporate financial distress very well, ANN＇s prediction ability is superior than MDA for short term up to two years, but no significant improvement. The research also proves that under current accounting system and rules in China, financial statements are able to provide sufficient information for the prediction of financial distressed corporations.

关 键 词：财务困境 多元判别分析技术 人工神经网络 

分 类 号：F275[经济管理—企业管理]