貝式定理在企業舞弊預警模型上之應用

企業舞弊在現今社會中層出不窮，為高知識份子的犯罪型態，身居要職之人士極力規避查緝，鋌而走險只為獲取高額的不法獲利，損害投資人利益也重創整體經濟市場，因此能確切偵測出異常訊號之模型已被各界所重視。本研究提出的財務報導舞弊之預警模型，有別於過往文獻中的羅吉斯迴歸，利用貝氏推論的概念進行預警模型的建構。本研究蒐集327家企業之資料建構貝式定理之預警模型，透過馬可夫鏈蒙地卡羅(MCMC)進行參數估計。而本研究也使用同一份資料進行羅吉斯迴歸分析，研究發現，貝式定理能有效避免羅吉斯迴歸普遍高估預測的情況，顯示出透過MCMC修正估計參數之偏誤後，能獲得比羅吉斯迴歸更為精確之預測結果。

關鍵字

財務報表舞弊；貝式定理；羅吉斯迴歸；馬可夫鏈蒙地卡羅

並列摘要

Corporate fraud, which takes many forms in contemporary society, is a crime mostly committed by highly educated people in positions of power. They attempt to evade prosecution and take huge risks for staggering large and illicit profits. The act of corporate fraud not only curtails the rights and privileges of the investors but also upsets the overall market economy. For this reason, the formulation of a model that could help detect any unusual market fluctuations would be essential for investors. Thus, in this paper, we propose a model that would act as an early warning system by predicting fraud associated with financial statements. Unlike the standard logistic model that has played a significant role in previous studies, we will utilize the Bayesian Probit model to create a visualized method for helping decision making. For this study, the data are from 327 businesses to establish and utilize a parametric estimation based on the Markov Chain Monte Carlo (MCMC). The comparison of both models indicated that the projection tends to be overvalued when adopting the standard logistic regression model, yet the Bayesian probit model can help reduce the overall level. This showed that after the bias of the parametric estimation is rectified; a more accurate forecast can be obtained from standard logistic regression.