Listed Companies Financial Reporting Fraud Detection Research

The dissertation focuses on the financial reports fraud and its detection. The main steps include the signification and motivation, fraud research retrospection /histories, fraud characteristics applying to the fraud measures/expression as well as the detection of fraud from the qualitative and quantitative analysis. The main contribution locates in: the basic steps and practical operation methods for qualitative analysis are determined and a four-standard examination for quatitative detection with a higher identification rate.In the int(?)oduction, the signification and motivation are expressed broadly. The motivation is directly from the "Enron" fraudulence case occurred in 2002 shocking the world cap(?)tal market. It, as an American outstanding company, deceived the commonalty investors, which concerned funds broke the historical record in US. The scandal attracted much of attention and urging the new research on the financial fraud. On the other side, China has the similar circumstances that frauds and developments are stepping forwards. From the world viewpoint, as a result, it is fairly important in investors' protection specially.The following chapters are the main contents. Chapter 2 provides the historical research retrospection and evaluation. I divide it into three parts-relevant concepts, theory headwaters and detection/identification. In relevant concepts, fraud and its relevant earnings manipulation, accounting information distortion, disrules, illusive comments, etc are explained respectively. For the theory headwaters, some different oppions to the positive and objective factors are concluded. In fraud detection and identification, I, according to the historical research time, divide it into "red flag" research and detection method research which are expanded by state. The red flag research inclines to qualitative analysis for the fraud and treats the abnormities as the research originals The detection method research based on the positivism focuses on improvements on the detection models and statistical differences analysis. Those give a good reference for following work. In chapter 3, a concept framework about financial reports fraud is determined at the base of research retrospection i.e. start from the motivation and procedure, the relevant earnings management, misstatement, accounting information distortion, etc are all directed into fraud of financial reports. Afterwards, the details of fraud histories are expanded according to the onshore and offshore. In foreign countries, financial fraud occurred early in 18.C embodying several apparent steps. Fraud fastigium accomplishes with economic booms. However, in China, fraud has no apparent watersheds due to the 10+ year's limited developments. It seems a admixture fusing every fraud measure and methods. After compare, no matter foreign or domestic, their fraud motivation and goals, deep influence and enormous losses are the same, without dying down under the government gradually increasement in supervision. The differences are only to Chinese un-mature capital market resulting in more direct frauds scarce of complex emulation methods.Chapter 4 tells the means and measures of financial fraud. It is the embodiment of Chinese financial fraud history after new understanding of the fraud. Also, it contains the most of fraud cases approved or unapproved. Several parts concerned are:Make full use of accounting standards leaks, such as relative transactions in related party, debt restructuring, non-monetary transactions, leasing, consolidated statements, accounting policy and estimation changes, etc.;Impose the capital M&A, including acquisition, merger, and subsidiary company set-up;Utilize the third power containing the government subsidy, tax allowance and even finance entrusted; andWords and language fraud including events disguise, avoidance of important and dwell on the trivial and information ahead or behind of schedule.Combined to fraud measures, chapter 5 states the fraud expression in financial reports. It starts out from the information carrier analyzing the possible itemsinfluenced. Relative items in balance sheet, income statement, cash flow statement and other appendix are explained adequately. This chapter 5 can be seen as the result and purpose of chapter 4, while chapter 4 can also be seen as the causes of chapter 5.The following chapters are about fraud qualitative/quantitative detection since above contents is prepared. Chapter 6 is a research by qualitative method for fraud detection. I provide several steps after the above red flags analysis to develop the qualitative method:Words and moods are important attention;Meanings from the words should be gradual reasoning; andConclude the doubtful points and contradictions.Doubtful points and contradictions are the real red flags which imply the possible fraud if we can connect the joints effectively. A listed company named "Chongqing Shiye (code: 000736)" is a good example which is illuminated according to the annual reports of 2003. Many of doubtful points flood into the reports and finally be discovered. Some important apocalypses are followed.Qualitative analysis method has a few restrictions although it is useful and practical. For example, it needs analysts' broad knowledge, deep analysis ability and penetrating observation. Time cost is longer as well. Compared to it, quantitative analysis applying the statistical models is advanced at objectivity without subjective differences or/and influences. And conclusion errors are limited which is more practical for middle-small sized investors.Chapter 7 is the beginning of detection models as research designs. The whole consideration and samples selected are determined firstly. I prefer the miscarriage of justice for non-fraud companies to the misjudging fraud companies. In samples, all evidenced fraud companies are supposed fraud while all un-evidenced companies, even if those are fraud detected in future, are still treated as un-fraud. Meanwhile, the fraud companies are reference of fraud occurring at some period otherwise un-fraud. Samples are contributed from the 500 un-fraud companies and dozens of frauds covering the period between 2000 and 2004. CCER database is available due to the mass data information. Secondly, relative variables are set-up under theanalysis of marco- and micro-factors. Thirdly, for variables, some selection principles are determined which contains public information channel, individual analysis, from appearance to connotation, concise and simple and measurable. These principles determined are for the possibility and feasibility of quantitative analysis. Accordingly, some important variables disobeying the above principles are erased form the models pitifully, such as management moral levels. Finally, variables are expressed in four fields: financial index, corporate governance, financial risk & pressure and assistant related transactions. A series of index are enriched according to the research achievements before and needs. For the financial index, I decide it in "time point", "industry belonged", "index in income statement", "index in balance sheet", "index in cash flow statement" and "index from statements cross". In corporate governance, it is decided in "time point", "governance conferences", "shares contribution", "management structure" and "management promotion". In financial risk and pressure, it contains "bargaining state", "auditing opinion", "market evaluation" and "inner risk and pressure" from last year. Related transactions are separated as "bargaining state", "controlling relationship in related parties", "related party character", "related party ownership percentage", "related party transaction amounts", "related transactions character", "whether occupy funds of listed companies", "occupying funds scale". In the last part, I expressed that the statistical research method including "classify" and "logistic regression" under software SPSS10.0.The above preparation is followed by the chapter 8, detection models built up, which are expanded by financial index, corporate governance, financial risk& pressure, and related transactions. The first step is to select the variables after classifying analysis according to the significance level. The second step is to build up the models form Standardization, Un-standardization, Bayes and Logistic functions respectively.After classifying the financial indexes, some significant indexes occurs, such as "other incomes/ profits before tax", "time of point", "other operating cash/cash from the sales", "current liabilities/inventories", "inventories turnover", and "taxes andcharges/main operating net revenue". For the corporate governance indexes, "time point", "first-shareholder holding percentage", and "supervisor amounts holding shares" are significant. For the financial risk& pressure, significant indexes are "time point", "auditing opinion", "P/E ratio" and "current liabilities/inventories". Related transactions, as references, have significant indexes shown as "bargaining state", "time point", "related party equity occupying percentage", "related party relationship character" and "weather occupy funds of listed companies". For the above significant variables, Standardization, Un-standardization, Bayes and Logistic functions build up different forecasting models respectively.Chapter 9 provides the main conclusions and compare of detection results. In general speaking, for the fraud, Logistic function has the higher reliability than Bayes' arriving at 100% level although its identification rate is much lower. For the non-fraud, Bayes' function exceeds the Logistic in reliability although its identification rate is not high as Logistic. At the same time, among the three parts of indexes, financial risk& pressure hits the peak of identification rate, corporate governance leading the second and financial indexes occupying the last. The detection models make the red flags brief only depending on the external annual reports and receive the satisfied identification rate 70% or so. After the practice of same listed company, the conclusion is the same as one from the qualitative analysis.The last one, as the ending words, some important restrictions and shortcomings are mentioned before the future research prospects.The contributions from this dissertation are located at:New explanation for the reports fraud covering all measurements and styles which not limited in statements or appearance truthfulness;A deep qualitative analysis method is created in accordance with the new fraud concept and every measurements/expressions;A quantitative analysis system is developed with the better identification rate expressed as four parts and their indexes intercrossing.