企業信用模型建置與驗證—使用乏析應變數以塑化業及食品業為例

Abstract

台灣上市公司不預警地宣布重整，跳票、全額交割或下市，造成投資大眾的損失，因此，必須建立企業信用模型來偵測其經營狀況。本研究發現財務比率自企業危機前五年起逐漸惡化，表示財務比率在危機發生前有惡化現象，另外危機發生後幾年財務比率仍有影響，故本研究視企業危機為一逐年遞增或遞減的變數，使用模糊數轉化，加入危機發生前後的總體變數，並且結合統計多變量分析和資料探勘中的乏析理論建立模型，使用窮舉法找出解釋力最佳之企業信用模型，結果顯示，採用模糊數轉化之應變數相當顯著。

The listed companies in Taiwan suddenly announced restructuring, bankruptcy or out of stock, and their investors lost a lot. Therefore, we must set up the enterprise credit model to detect and examine their management states. We discover that the financial ratios decrease gradually since the past five years of enterprise`s crisis. Besides, financial ratios still diminish after the crisis take place. Therefore, this research regards enterprise`s crisis as one parameter, and we transform the parameter by fuzzy numbers. In addition, we use the macro economical parameters and combine multivariate analysis and fuzzy logic theory to find out a higher significant model. The result shows it is high significant to adopt the fuzzy number dependent variable.