本研究係以國內某家金融機構,2002年1月到2005年12月底貸放之房屋貸款案件中,抽取正常戶200件、違約戶200件,共400件作為樣本,採用Logistic 迴歸分析,希望能找出影響房屋貸款逾期的主要因素。並且對於在2006年底開始實施新版「巴賽爾資本協定」以前,採用內部評等法方式進行資本列提的金融機構,能參考本研究所建立的違約機率量化流程,著手建置本身的信用風險模型。 經Logistic迴歸分析後在所投入13種自變數中,採正向逐步選擇法得出八項與假設相符分別為:年齡、婚姻、貸款用途、職業、職位、區位、借款金額、貸款成數。另有有五項與假設不符分別為,性別、年資、型態、估價值、保證人。研究發現,現許多以往研究列為顯著之變數在本研究有不顯著的情形,是因為受到景氣好壞以及政府法令變遷的影響。最後、本研究建議審查人員必須透過即時的統計資訊,適時修正審查的標準。
This study is based on a financial institution’s housing 400 loan clients, covering the period from January 2002 to December 2005. Among the samples selected are 200 normal clients and 200 delinquent clients. This study applies Logistic Regression analysis to find out the determinants for delinquent cases. This study also provides a reference to financial institutions that are currently using internal evaluation model or capital rating model to develop their respective credit risk evaluation model before the implementation of International Convergence of Capital Measurement and Capital Standards by the year end of 2006. The result from Logistic Regression model applying forward stepwise method on the 13 independent variables shows that 8 items are in accord with the hypotheses. The 8 items are: Age, Marriage, Purpose of Loan, Occupation, Position, Location, Loan Amount and Percentage of Loan Approval. The 5 items that are not in accord with the hypotheses are Sex, Seniority, Type of Property, Revaluation Value and Guarantor. This study shows that some variables that were tested to be significant in previous study has demonstrated insignificant character in this study. This may be due to the economic cycle where the market is experiencing as well as the changes in government policies. This study recommends that credit evaluation officers to be updated with the latest and current statistical information to assess their standards of the evaluation mechanism from time to time.