近年來由於金融風暴與歐債危機等事件對全球的金融市場造成嚴重的衝擊，有鑑於此議題的重要性，國內外學者紛紛進行深入的探討與財務風險相關之課題，然而前述研究之焦點大多集中於數值型態之資訊，卻忽略了非數值型態資訊之重要性，即財務報表可讀性，而預測模型的主軸也是建構財務危機預警機制，研究指出百分之九十九的財務危機起因於經營績效不佳，所以本研究將發展一個企業經營績效預測模型，在企業經營績效的評估上有別於過去僅採用財務數值，本研究將採用平衡計分卡(balanced scorecards, BSC)為基礎，但此機制無法給予決策者一個明確之方向，為了可系統性的解決此缺失，將進一步採用資料包絡分析法(data envelopment analysis, DEA)對企業績效進行評估。此外，過去的預測模型大多以統計為基礎，但此模型的建構必須符合一些嚴格的統計假設，大大降低其實際應用之範疇，為了降低此問題，本研究將採用一種較為新穎的人工智慧技術，即支援向量機(support vector machine, SVM)，此模型不僅不須符合傳統的統計假設，對於離群值或極端值亦有較佳的容忍度，且具備相當優越的泛化能力，以促使預設之效率更加提升。本研究資料取自台灣經濟新報資料庫(Taiwan economic journal, TEJ)，研究對象以2016年台灣上櫃電子業為主。實證結果顯示同時考慮財務變數與財務報表可讀性對於經營績效具有重要性，而研究之成果可進一步提供給管理者或投資人對其企業有更深的了解。
Due to the recent financial turmoil and European debt crisis, evaluate the corporate’s financial risk turns out to be an essential and attractive research topic. In compared with well-examined research works on financial crisis prediction and credit risk prediction, the research project on corporate’s operating performance evaluation is quite rare. However, close to 99% of financial crisis cases are due to bad operating performance. Most previous researches utilized financial ratios (such as return on assets: ROA, return on equity, ROE) to determine the corporate’s operating status. Unfortunately, aforementioned financial ratios belong to one input and one output variable that can’t represent the corporate’s real economic substance. To overcome this obstacle, this study proposed a novel decision making mechanism which incorporated balanced scorecards (BSC) and data envelopment analysis (DEA) to determine the corporate’s operating performance. The analyzed result was then fed into support vector machine (SVM) to construct the model for corporate operating performance forecasting. The model, examined by real cases, was a promising alternative for corporate operating performance forecasting.
