臺灣博碩士論文加值系統

財務報表舞弊每年皆造成龐大損失，若能利用公開資訊偵測財報舞弊將對公司利害關係人有莫大幫助。本研究以資料探勘方法羅吉斯迴歸、類神經網路以及決策樹三者並搭配變數組合－全部、篩選、財務、公司治理、z-score共五種，分別建立財報舞弊預測模式。
舞弊公司樣本為1998年至2010年間受金融監督管理委員會證期局起訴案件與證券及期貨投資者保護中心公布的團體訴訟及仲裁案件之公司，以1:4的方式搭配正常公司做為研究樣本。考量財務變數乃基本評斷公司財務狀況好壞的指標，但公司治理為能即時反應經營狀況的先行指標，故選用財務以及公司治理兩者做為研究變數。
透過總正確率、型一誤差、型二誤差以及錯誤歸類成本衡量模型績效，並利用十折法確立模型穩健程度，尋求一個最適合偵測財報舞弊的模型。
研究結果羅吉斯迴歸於舞弊發生前一年利用篩選變數建構的模型最能偵測出潛在舞弊公司，其次為利用全部變數之類神經網路，各年度全部模型相互比較下，使用舞弊前一年篩選變數的羅吉斯迴歸還是最適合用來偵測財務報表舞弊。

Financial statement fraud is a serious threat to company stakeholders and often results in huge losses every year, using public information to detect financial statement fraud will be a huge benefit to interested parties. This study uses 3 data mining techniques to develop fraud detection models, techniques includes logistic regression, neural network and decision tree, and with 5 variable combination - all, filtered, financial, corporate governance, z-score.
Fraud samples are those between 1998's to 2010 and is prosecuted by Financial Supervisory Commission of Securities and Futures Bureau and prosecuted cases filed in the Securities and Futures Investors Protection Center, study samples are fraud companies in a 1:4 mix of normal companies and further separated into training data and testing data. Financial variables could basically judge whether the target company's financial situation is good or bad, but to have a immediate response of business conditions, corporate governance variables should also be considered, thus, research variable take both financial and corporate governance aspects into account.
To measure model performance, we use total correct rate, type I error, type II error and misclassification cost, and 10-fold cross-validation approach is established to prove model robustness, by comparing different prediction models, this study find the most suitable model for detecting financial statement fraud.
As for the outcome, logistic regression with filtered variables results to be most suitable to detect financial statement fraud 1 year prior the outburst of fraud event, neural network with all variables results to be most suitable to detect financial statement fraud 2 year prior the outburst of fraud event, in overall, logistic regression with filtered variable still to be the most suitable model to detect financial statement fraud 1 year prior the outburst of fraud event.

壹、緒論 1
一、 研究背景與動機 1
二、 研究目的 3
三、 研究貢獻 4
四、論文架構 5
貳、文獻探討 6
一、舞弊定義 6
二、 公司治理 8
三、公司治理與舞弊 10
四、舞弊偵測相關研究 12
五、研究流程 19
叁、研究方法 20
一、樣本蒐集與研究期間 20
二、變數選定 20
三、指標前置整理 23
四、效益評估 24
五、交叉驗證 26
肆、實驗結果與分析 28
一、資料前置處理 28
二、建立分類模型 33
三、模型分類能力之比較與評估 38
伍、結論與建議 45
一、研究結論 45
二、研究貢獻與建議 46
參考文獻 48
附表 51

上市上櫃公司治理實務守則，民國99年11月10日修正，證券暨期貨法令。
林豐澤，「演化式計算上篇：演化式演算法的三種理論模式」，智慧科技與應用統計學報 3卷1期，民國94年。
陳惠玲，「財務舞弊專題」，TEJ台灣經濟新報，民國94年。
許伯彥，「財務報表舞弊風險評量模式研究」，國立台灣大學會計學研究所未出版碩士論文，台灣台北市，民國92年。
葉怡成，「應用類神經網路」，儒林圖書公司，民國96年。
蘇木春、張孝德，「機械學習：類神經網路、模糊系統以及基因演算法則，」全華科技圖書股份有限公司，民國98年。
AICPA. 2002, Considerations of Fraud in a Financial Statement Audit. Statement of Auditing Standards No. 99. American Institute of Certified Public Accountants: New York, NY.
Albrecht, W. S., C. C. Albrecht, and C.O. Albrecht, and M.F. Zimbelman. 2008, Fraud Examination . South－Western College Pub; 3 edition
Arnold, V., and S. G. Sutton. 2002, Researching Accounting as an Information Systems Discipline, Published by the American Accounting Association Information Systems, p273－283.
Association of Certified Fraud Examiners （ACFE）. 2008, 2008 Report to the Nation on Occupational Fraud and Abuse. Austin, TX: ACFE.
Beasley, M.S., J.V. Carcello, and D.R. Hermanson. 1999, Fraudulent financial reporting: 1987–1997. An analysis of public companies. New York, NY: COSO;
Beasley, M.S. 1996, An Empirical Investigation of the Relation between Board of Director Composition and Financial Statement Fraud. Accounting Review, 71（4）: 443–460.
Cotton D.L. 2002, Fixing CPA ethics can be an inside job. Available at http://www.washingtonpost.com/ac2/wpdyn/A50649－2002Oct19?Language=pringter
D’Agostino, D., and O. Williams. 2002, Financial Statement Restatements: Trends, Market Impacts, Regulatory Responses, and Remaining Challenges. United States General Accounting Office （GAO）: Washington, DC.
David Olson, Yong Shi. 2006.,Desicion Tree, Introduction to Business Data Mining p161－180 Irwin:McGraw－Hill.
Dechow, P. M., R. G. Sloan, and A. P. Sweeney. 1996, Causes and Consequences of Earnings Manipulation: An Analysis of Firms Subject to Enforcement Actions by the SEC, Contemporary Accounting Research, 13（Spring）: 1–36.
Eccles, R. G., and J. P. Philip. 1992, Creating a Comprehensive System to Measure Performance, Management Accounting, 74（4）: 41－44
Fanning K, and K. Cogger. 1998, Neural network detection of management fraud using published financial data. International Journal of Intelligent Systems in Accounting, Finance and Management 7: 21–41.
Fanning K, K. Cogger, and R. Srivastave. 1995, Detection of Management Fraud: A Neural Network Approach. International Journal of Intelligent Systems in Accounting, Finance and Management 4: 26–113
Fidelis, M.V., H.S, Lopes, and A. Freitas, 2000, Discovering Comprehensible Classification Rules with a Genetic Algorithm, in Proceedings of the IEEE Congress on Evolutionary Computation, 1: 805－810.
Freeman, J.A., and D. Skapura. 1992. Neural Networks: Algorithms, Applications, and Programming Techniques, Addison－Wesley Publishing Company, Reading, Massachusetts, pp. 213 － 262.
Hoogs B., T. Kiehl, and C. LaComb and D. Senturk. .2007, A Genetic Algorithm Approach to Detecting Temporal Patterns Indicative of Financial Statement Fraud. International Journal of Intelligent Systems in Accounting, Finance and Management 15, 41–56
Karpoff J., and J. Lott. 1993, The Reputational Penalty Firms Bear from Committing Criminal Fraud. Journal of Law and Economics 36（2）:757–802.
Kim, J.B. Yi, C.H. 2006, Ownership Structure, Business Group Affiliation, Listing Status,
and Earnings Management: Evidence from Korea Contemporary Accounting
Research 23（2）: 427－464
Kirkos, S., Spathis, C., & Manolopoulos, Y. 2007, Data mining techniques for the detection of fraudulent financial statements. Expert Systems with Application, 32(4), 995-1003
Kwon T., and E. Feroz. 1996, A Multilayered Perceptron Approach to Prediction of the SEC’s Investigation Targets. IEEE Transactions on Neural Networks 7（5）: 1286–1290.
Lee T., R. Ingram, and T. Howard. 1999, The Difference between Earnings and Operating Cash Flow as an Indicator of Financial Reporting Fraud. Contemporary Accounting Research 16（4）: 749–786.
Man, K.F., K.S. Tang, and S. Kwong. 1999, Genetic algorithms: Concepts and Designs, Springer－Verlag, London
Mitchell, M. 1996. An Introduction to Genetic Algorithms. Massachusetts Institute of Technology: Cambridge, MA.
OECD Principles of Corporate Governance，2004
www.oecd.org/dataoecd/32/18/31557724.pdf
Persons, O. 1995, Using Financial Statement Data to Identify Factors Associated with Fraudulent Financial Reporting. Journal of Applied Business Research 11（3）: 38–46.
Prodromidis, A. L. and Stolfo, S., Agent-Based Distributed Learning Applied to Fraud Detection,”Technical Report CUCS-014-19, Department of Computer Science, Columbia University, 1998.
Sharma, V. D. 2004, Board of Director Characteristics, Institutional Ownership, and Fraud:
Evidence from Australia. Auditing: A Journal of Practice & Theory , Vol23, NO.2,
105－117
Wells, J. 2004a, New Approaches to Fraud Deterrence. Journal of Accountancy Online Issues, found at http://www.aicpa.org/PUBS/jofa/feb2004/wells.htm.
Wells, J. 2004b, Corporate Fraud Handbook: Prevention and Detection. John Wiley & Sons, Inc. Hoboken, New Jersey.
Worldwide: Looters have a foothold. 2007, Journal of Accountancy., 32－38.