联系我们

论文查询结果	返回搜索

论文编号：	1878
作者编号：	041847
上传时间：	2010/5/18 10:34:28
中文题目：	企业财务预警模型的比较研究--基于我国制造业上市公司数据的实证分析
英文题目：	Financial Early Warning Model Fisher Dicriminant Model
指导老师：	黄菊珊
中文关键字：	财务预警模型 Fisher二类判别模型 Logistic回归模型
英文关键字：	Financial Early Warning Model Fisher Dicriminant Model
中文摘要：	随着我国加入WTO，企业经营环境日益复杂多变，经营风险越来越大，若经营不善，就可能陷入财务危机。就上市公司而言，证券监管部门的管理制度连同1986年颁布的《破产法》构成了上市公司财务预警模型研究的制度背景。同时企业陷入财务危机通常是个逐渐的过程，大多数的企业从正常到逐步恶化，都是有先兆的，而且可以预警。因此，对于投资者、监管部门、企业管理者和债权人而言，如果能够借助一定的方法，建立一套科学的财务预警模型，无疑就可以把握住证券市场的命脉。研究财务预警模型的方法越来越多，但哪种方法建立的模型更有效？本文就是通过对财务预警模型的概述，选择了Fisher二类判别模型、Logistic回归模型和BP网络模型，首先对三种模型进行理论分析，随后建立和检验了三种模型，最后对三种模型进行了比较研究。 本文采用了实证和规范相结合的研究方法，以我国制造业A股上市公司因“财务状况异常”而被特别处理的公司作为研究对象，选择2003-2005年65家财务危机公司，同时采用配对的方法逐年选择65家财务健康公司；初步选定53个变量指标并通过柯尔莫哥洛夫-米诺夫检验、曼-惠特尼-威尔科克森检验、逐步判别分析进行筛选，建立和检验了Fisher二类判别模型、Logistic回归模型和BP网络模型，并对其进行了比较研究。 本文的研究结论如下：（1）财务类指标在我国企业财务预警模型中具有显著地影响，而现金流量类指标和市场收益率指标在我国并不适用。（2）ST前1年，BP网络模型具有最强的预测能力，准确率达到95%。ST前2年，三种模型的预测准确率分别为87.5%、94.6%、85%，说明Logistic具有最强的预测能力。ST前3年、ST前4年，Fisher二类判别模型比Logistic回归模型具有更好的预测能力。BP网络模型的有效性还需进一步的实证检验。
英文摘要：	With China’s accession to WTO, business environment is more and more complex, and business risk is increasing. If businessing badly, the enterprise may fall into the financial crisis. Speaking of public company, the strict management system of Securities Supervisory department and “Bankruptcy Law” promulgated in 1986 constitute the system background of corporate financial early warning model research. Meanwhile, it’s a gradual process for enterprise to fall into financial crisis. There are omen for enterprise becoming worsen and enterprise may be forewarned. Therefore, if we could establish a set of scientific financial early warning models by certain method, it’s no doubt that related persons such as investors, supervisory department, management and creditor may grasp the securities market’s life. There are more and more financial early warning models, but which model is most effective? First, the paper outlined the financial early warning models, and selected Fisher Discriminant Model, Logistic Regression Model and BP Network Model. Then, the paper researched the three models theoretically. Finally, the paper established and tested three models and compared them. The paper employed empirical and normative research methods. First, the manufacturing sector A-share listed companies which were specially treated because of the “exceptional financial situation” in our country were recognized as a research. Then, I chose 65 financial crisis companies and 65 financial health companies each year in 2003-2005 by matching method, and 53 preliminary variable indicators which were screened by K-S test, MWW test and Progressive Discriminant Analysis. Finally, I established and tested Fisher Discriminant Model, Logistic Regression Model and BP Network Model, and compared them. The paper has made the conlusions as follows: (1) financial indicators have evident effect in our country’s financial early warning models, but indicators of cash flow and market yield are not suitable in our country. (2) 1 year before ST, BP Network Model had the strongest projected capacity of 95% accuracy rate. 2 years before ST, forecast accuracy rate of three models were 87.5%, 94.6%, 85%, respectively. This indicated that Logistic Regression Model had the strongest projected capacity. 3 years and 4 years before ST, Fisher Discriminant Model was better than Logistic Regression Model. The effectiveness of BP Network Model needs further examination.
查看全文：	预览  下载(下载需要进行登录)