授信業務是銀行獲利的主要來源之一,因此授信業務對於金融體系而言益形重要。然而授信業務有其信用風險,一旦授信戶逾期違約,將對銀行本身甚至金融市場產生莫大的衝擊,因此有效的風險控管係一重要的議題,本研究利用1982年至2009年期間台灣地區之財務危機公司作為樣本資料,檢視影響公司經營危機的變數資訊。本研究分成兩個模式來分析:模式一,採原始資料模式;模式二,採變動模式(變動即指前一年的財務比率減前二年的財務比率)。為避免自變數之間有共線性的問題存在,因此,本研究利用逐步迴歸的方式篩選出具有區別力的變數,並利用篩選出最具區別力的變數,以羅吉斯迴歸分別建立財務比率單一指標預警模型與包含財務比率、銀行借款變數及非財務比率的多指標預警模型,並比較此二類模型之區別力。研究結果發現,多指標財務危機預警模型之整體區別正確率高於單一指標預警模型,且實證顯示模式一的正確分類率,並不能從變動模式中獲取額外資訊。
Being a main source of profit for banks, the lending business is vital to the financial system. However, the lending business carries credit risk. When debtors default or violate the terms of their loans it can have a great impact on individual banks in particular and the financial markets in general. It is for this reason that effective risk management is such an important issue. This study uses companies in Taiwan who experienced financial distress from 1982-2009 as samples to investigate variables in companies’ crisis management. This study refers to the companies as sample that come through financial crisis during 1982-2009 in Taiwan in order to investigate variables of crisis management. The study analyzes two type of models, one as original , another as variational (The financial ratio of variation subtract previous two years from preceding year.) To avoid the Collinearity problem of indepent variable, the study sifts through the material to gather variables with higher discriminatory power, using logistic regression to establish a prediction model based on financial ratios alone, as well as a prediction model based on multiple indicators, including financial ratios, bank loan variables and non-financial ratios. Then the study compares the discriminatory power of the two models. The study finds the prediction model based on multiple financial distress indicators has a higher overall accuracy rate than the prediction model based on a single indicator.