A Study of Bank’s Credit-Granting Model to Evaluate Small-scale Consumer Loan

• Main Authors: Lung-Hsien Huang, 黃隆憲
• Other Authors: Ying-Shing Lin
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/68727764854143928196

碩士 === 國立高雄第一科技大學 === 財務管理所 === 94 === Abstract In the past five years,average Taiwan gross national income are less than NT$15,000; however, individual consuming debts are more than NT$40,000.People getting into debts are more than people earning money. Particularly, the credit debts and small-scale consumer loan have risen sharply among these liabilities (debts). The increase of consuming debts accelerates overdue cases of bank loans.This study collects small-scale consumer loan cases in a branch of a southern Taiwan bank during 2003 and 2004 as samples. This research selected 500 cases in total; 400 cases of sample are regular ones that repay on time and 100 cases are overdue cases. Based on the results of this study, the researcher concludes that consumers’ gender, number of children, educational background, work seniority, annual income, occupation, and amount of loan seven significant variables influencing small-scale consumer loan. The most important variables are gender, annual income, work seniority, educational background, the amount of loan, the number of children, and the occupation is less important. The results indicate that the prediction accuracy of overdue cases is 67%, regular cases 95%, and the correctness in total 89.4% by using Logistic Regression. Based on the results, the Goodness is significant at of Fit χ2=251.825（p=0.000<0.05）. And the Hosmer-Lemeshow is not significant at 4.698（p=0.789>0.05）, which indicates that the assumption is correct through using Bank’s Credit-Granting Model.This model is practically useful in predicting the probability of the cases to be regular one or overdue one according to application information in a short period. This study employs the model as follows, Z =－8.106＋1.995 X1－0.429 X3＋0.887 X4＋0.902 X6＋0.938 X7－0.287 X8＋0.606 X10 X1 means gender; X3 means the number of children; X4 means educational background; X6 means work seniority; X7 means annual income; X8 means occupation; X10 means the iii amount of loan. In conclusion, bank’s profit rely more on should control the risk management than granting credit. If banks are able to sort out the high-risk cases through using Logistic Regression before granting consumer loans, they will be more profitable.If bank maintain high credit-granting quality, they can also avoid social and economic problems are well.