企業營運狀況會定期反應在財務報表上,但卻需等到財務揭露之後,投資人才能預警公司是否出現狀況,但是若企業刻意粉飾財報資訊則無法直接由財務資訊得知企業營運情形,而許多非財務資訊經先前學者證實可幫助預測企業預測財務危機的發生,但股價資訊或信用評等狀況等指標因素卻鮮少學者深入研究。 本研究結合企業財務比率指標和非財務資訊指標建構出企業財務危機預警模式。研究樣本取36家財務危機公司,以一比一配對方式選取36家正常公司,共72家公司樣本,資料蒐集法令規章和前人學著經驗法則使用30項財務比率和7項非財務指標作為輸入變數。 本研究使用資料探勘的決策樹方法和類神經網路分析企業財務危機預警模式,發現財務危機發生前一年的預測準確率,以類神經網路分析法有較佳的預測能力,前二年和前三年則是決策樹分析法較佳;本研究預警模式證實準確率優於先前學者所使用的非財務指標預警模式,且更適用於台灣企業。研究發現額外加入股價資訊指標和信用評等指標的預警模型確實能增加財務危機預警的準確率,而且有效降低型一錯誤率發生。
Enterprise operating status will be disclosed periodically on financial statement and investors can get fully information once the formal financial statement is disclosed and published. If exectives of firms intentionally dress financial statements up , investors can not get real enterprise operating status from it. However, non-financial information was proved to predict financial distress by former researchers. But few studies exploit stock and credit ranking information to construct financial crisis prediction model. The study uses financial and non-financial information to predict corporate financial distress. We get 36 financial distress and 36 normal firms for sampling data. Regarding to our data gathering and former researchers’ experiences, we exploit 30 financial indicators and 7 non-financial indicators for our input data. Decioion tree and neural network analysis were used by the study to construct financial crisis prediction model and found neural network analysis obtains better prediction accuracy in the year before financial crisis.Howere,decision tree obtains better prediction accuracy in the former two and three year of financial crisis. The study was approved to have better prediction accuracy than former researchers and is much more suitable for Taiwan firms’ financial crisis prediction.The result found stock and credit ranking information of non-financial indicators can improve the accuracy of financial crisis prediction model and effectively lower Type 1 error.