中国制造业上市公司财务欺诈识别分析
本站原创 论文中文摘要：国内外研究财务风险和财务预警白勺模型很多,而研究财务欺诈白勺相对较少。其中一个原因是财务风险容易从财务指标上看出来,而财务欺诈相对来说要隐蔽很多,欺诈白勺手段也是五花八门,层出不穷。不同行业不同性质白勺企业手段各有各白勺招。本文选择了对国民经济影响最为重要白勺制造业来分析和识别财务。以中国上市公司每年公布白勺年报为分析依据,先通过分析手段和动因,提取和计算财务指标,作为模型白勺输入变量。运用基于连续性白勺主成分回归模型和基于离散性白勺决策树分类模型,计算其对样本白勺识别效率。通过分析结果,来证明模型对制造行业白勺上市公司财务欺诈是否具有识别作用
Abstract（英文摘要）：www.328tibEt.cn What’s the finance fraud? The finance fraud of listed company defines the responsibility persons make illusive accounting information in order to escape the tax, own the unlawful interest and other private profit.The accounting fraud of listed company is always troubling and important thing for location government, investors and audit departments of finance. If we can look for a good method to solve it, that will give large effect to the securities business of China, and bring advantage to all investors and government. It also advances the credit of securities business.The detection of fraudulent financial statements, along with the qualification of financial statements, he recently been in the limelight in China because of the increase in the number of companies listed on the Shanghai Stock Exchange and Shenzhen Stock Exchange(and raising capital through public offerings) and the attempts to reduce taxation on profits. In China, the public has been consistent in its demand for fraudulent financial statements and qualified opinions as warning signs of business failure. There is an increasing demand for greater transparency, consistency and more information to be incorporated within financial statements.This standard respects the auditor’s consideration of the risk that fraud and error may exist, and clarifies the arguments on the inherent limitations of an auditor’s ability to detect error and fraud, particularly management fraud. Moreover, it emphasizes the distinction between management and employee fraud and elaborates on the discussion concerning fraudulent financial reporting.Detecting management fraud is a difficult task when using normal audit procedures. First, there is a shortage of knowledge concerning the characteristics of management fraud. Secondly, given its infrequency, most auditors lack the experience necessary to detect it. Finally, managers deliberately try to deceive auditors. For such managers, who comprehend the limitations of any audit, standard auditing procedures may prove insufficient. These limitations suggest that there is a need for additional analytical procedures for the effective detection of management fraud. It has also been noted that the increased emphasis on system assesent is at odds with the profession’s position regarding fraud detection, since most material frauds originate at the top levels of the organization, where controls and systems are least prevalent and effective.Our sample contained data from all manufacturing firms (no financial companies were included). Auditors checked all the firms in the sample. We choose 89 fraud companies of these firms from 2003 to 2005, there was published indication or proof of involvement in issuing finance fraud by securities supervision bureau or other finance and financial audit departments. The classification of a financial statement as false was based on the following parameters: inclusion in the auditors’report of serious doubts as to the accuracy of the accounts, observations by the tax authorities regarding serious taxation intransigencies which significantly altered the company’s annual balance sheet and income statement, the application of China legislation regarding negative net worth, the inclusion of the company in the Shanghai and Shenzhen Stock Exchange categories of under observation and“negotiation suspended”for reasons associated with the falsification of the company’s financial data and, the existence of court proceedings pending with respect to serious taxation contrentions.Logistic regression model has become the standard method of analysis in the academic and practical fields of finance fraud measurement . Considering the characteristics of high correlation and high dimension of the credit data of listed companies in China , this paper presents a new approach combining PCA with logistic analysis to empirically predict the Chinese corporate fraud. This should show that the new approach is more stable and reliable than the simple logistic approach.These ratios are: Logarithm of total assets , working capital, the ratio of property plant & equipment (net fixed assets) to total assets , sales to total assets, current assets/current liabilities , net income/fixed assets , cash/total assets , quick assets/current liabilities , earnings before interest and taxes and long term debt/total assets and so on. In total, we compiled 15 financial ratios. In an attempt to reduce dimensionality, we ran t-test to test whether the differences between the two classes were significant for each variable. If the difference was not significant difference, the variable was considered non-informative. We depict the means, standard deviations, Sig-values for each variable.A Decision Tree (DT) is a tree structure, where each node represents a test on an attribute and each branch represents an outcome of the test. In this way, the tree attempts to divide observations into mutually exclusive subgroups. The goodness of a split is based on the selection of the attribute that best separates the sample. The sample is successively divided into subsets, until either no further splitting can produce statistically significant differences or the subgroups are too all to undergo similar meaningful division.Auditing practices nowadays he to cope with an increasing number of management fraud cases. Data Mining techniques, which claim they he advanced classification and prediction capabilities, could facilitate auditors in accomplishing the task of management fraud detection. The aim of this study has been to investigate the usefulness and compare the performance of three Data Mining techniques in detecting fraudulent financial statements by using published financial data. The methods employed were Decision Trees, PCA Regression. In terms of performance, the Decision Trees model achieved the best performance managing to correctly classify the validation sample in a 10-fold cross validation procedure. The accuracy rates of the PCA Regression model and the Decision Tree model were 74.3% and 82.1%, respectively. The Type I error rate was lower for Type 2 models.
论文关键词： 财务欺诈；决策树；主成分分析；逻辑回归；
Abstract（英文摘要）：www.328tibEt.cn What’s the finance fraud? The finance fraud of listed company defines the responsibility persons make illusive accounting information in order to escape the tax, own the unlawful interest and other private profit.The accounting fraud of listed company is always troubling and important thing for location government, investors and audit departments of finance. If we can look for a good method to solve it, that will give large effect to the securities business of China, and bring advantage to all investors and government. It also advances the credit of securities business.The detection of fraudulent financial statements, along with the qualification of financial statements, he recently been in the limelight in China because of the increase in the number of companies listed on the Shanghai Stock Exchange and Shenzhen Stock Exchange(and raising capital through public offerings) and the attempts to reduce taxation on profits. In China, the public has been consistent in its demand for fraudulent financial statements and qualified opinions as warning signs of business failure. There is an increasing demand for greater transparency, consistency and more information to be incorporated within financial statements.This standard respects the auditor’s consideration of the risk that fraud and error may exist, and clarifies the arguments on the inherent limitations of an auditor’s ability to detect error and fraud, particularly management fraud. Moreover, it emphasizes the distinction between management and employee fraud and elaborates on the discussion concerning fraudulent financial reporting.Detecting management fraud is a difficult task when using normal audit procedures. First, there is a shortage of knowledge concerning the characteristics of management fraud. Secondly, given its infrequency, most auditors lack the experience necessary to detect it. Finally, managers deliberately try to deceive auditors. For such managers, who comprehend the limitations of any audit, standard auditing procedures may prove insufficient. These limitations suggest that there is a need for additional analytical procedures for the effective detection of management fraud. It has also been noted that the increased emphasis on system assesent is at odds with the profession’s position regarding fraud detection, since most material frauds originate at the top levels of the organization, where controls and systems are least prevalent and effective.Our sample contained data from all manufacturing firms (no financial companies were included). Auditors checked all the firms in the sample. We choose 89 fraud companies of these firms from 2003 to 2005, there was published indication or proof of involvement in issuing finance fraud by securities supervision bureau or other finance and financial audit departments. The classification of a financial statement as false was based on the following parameters: inclusion in the auditors’report of serious doubts as to the accuracy of the accounts, observations by the tax authorities regarding serious taxation intransigencies which significantly altered the company’s annual balance sheet and income statement, the application of China legislation regarding negative net worth, the inclusion of the company in the Shanghai and Shenzhen Stock Exchange categories of under observation and“negotiation suspended”for reasons associated with the falsification of the company’s financial data and, the existence of court proceedings pending with respect to serious taxation contrentions.Logistic regression model has become the standard method of analysis in the academic and practical fields of finance fraud measurement . Considering the characteristics of high correlation and high dimension of the credit data of listed companies in China , this paper presents a new approach combining PCA with logistic analysis to empirically predict the Chinese corporate fraud. This should show that the new approach is more stable and reliable than the simple logistic approach.These ratios are: Logarithm of total assets , working capital, the ratio of property plant & equipment (net fixed assets) to total assets , sales to total assets, current assets/current liabilities , net income/fixed assets , cash/total assets , quick assets/current liabilities , earnings before interest and taxes and long term debt/total assets and so on. In total, we compiled 15 financial ratios. In an attempt to reduce dimensionality, we ran t-test to test whether the differences between the two classes were significant for each variable. If the difference was not significant difference, the variable was considered non-informative. We depict the means, standard deviations, Sig-values for each variable.A Decision Tree (DT) is a tree structure, where each node represents a test on an attribute and each branch represents an outcome of the test. In this way, the tree attempts to divide observations into mutually exclusive subgroups. The goodness of a split is based on the selection of the attribute that best separates the sample. The sample is successively divided into subsets, until either no further splitting can produce statistically significant differences or the subgroups are too all to undergo similar meaningful division.Auditing practices nowadays he to cope with an increasing number of management fraud cases. Data Mining techniques, which claim they he advanced classification and prediction capabilities, could facilitate auditors in accomplishing the task of management fraud detection. The aim of this study has been to investigate the usefulness and compare the performance of three Data Mining techniques in detecting fraudulent financial statements by using published financial data. The methods employed were Decision Trees, PCA Regression. In terms of performance, the Decision Trees model achieved the best performance managing to correctly classify the validation sample in a 10-fold cross validation procedure. The accuracy rates of the PCA Regression model and the Decision Tree model were 74.3% and 82.1%, respectively. The Type I error rate was lower for Type 2 models.
论文关键词： 财务欺诈；决策树；主成分分析；逻辑回归；