公司營運良窳攸關銀行授信決策，乃至投資人之投資意願，然而因為資訊的不對稱性，使得外部人員對於公司實際營運概況無法全盤了解，為協助外部人員熟悉公司營運、獲利與投資決策之優缺點，亟待建立一具預測與鑑別能力之財務預警模型。過去許多關於公司財務危機預警之研究，所考慮的投入變數僅限於一般的財務變數或非財務變數，而本研究則探討在傳統的公司財務危機預警系統中，利用羅吉斯模型(Logit)及倒傳遞類神經網路模型(BPN)，加入風險值(VaR)這個新的變數，以建立新的財務危機預警模型，期望能提高公司之預警能力，研究期間為2001年1月1日至2005年12月31日，共5年資料，參酌股市觀測站與台灣經濟新報資料，選取35家具財務危機公司，並比對35家產業、規模相仿之正常公司。研究結果顯示，加入風險值變數，確實可以提高危機預警的能力；而在與加入非財務變數中的會計師簽證作比較時，利用風險值更可以提早提供公司財務危機之預測。 A company's operational performance usually affects the loan decision-making of it's funding bank and the anticipation of investors. Because the information's asymmetry causes the exterior persons unable to survey the actual situation of a business,therefore,for assisting them to get familiar with the companies'management,and the profit of the investment, a precautionary early finance-warning model with the ability to forecast must to be established. Among those past studies on companyies' financial distress prediction models,the input parameters considered are only limited to general financial or non-financial factors. This study establishes a new early warning system by utilizing a new variable,Value at Risk (VaR), into the traditional Logit and Back-Propagation Network (BPN) models. The study obtained 35 distressed firms and 35 regular firms in the same industry and the time span is from 2001 to 2005. The result shows that the prediction ability is indeed improved by using the VaR as well as financial variables in company's financial distress prediction models. Furthermore,the application of VaR offers earlier financial crisis prediction while compared to the use of the opinions of certified public accountants.
