此一研究的重要為探討房屋貸款提前還款再貸款之變數。研究的變數包含1.性別、2.教育程度、3.婚姻狀況4.申貸時年齡、5.月收入、6.現職工作年資、7.職業、8.還款年限、9.房屋使用用途別、10.貸款金額、11.貸放成數、12.還款方式、13.有無保證人、14.有無政府優惠貸款、15.貸款月付金佔總收入比率、16.借保人有無關係、17.房屋類型等十七變項因素。 研究結果發現,對於房屋貸款提前還款再貸款戶中。性別、月收入、房屋使用用途別、貸款金額、還款方式、有無保證人、有無政府優惠貸款等七個變項因素無顯著性相關性。而在 「婚姻狀況」、「職業」、「貸款月付金佔總收入比率」等三項因素具邊際顯著相關性。教育程度、申貸時年齡、現職工作年資、還款年限、貸放成數、借保關係(無保人或有直屬親屬關係)及房屋類型等七項則具有顯著性相關。 在將上述所發現之七個顯著因子導入Logistic及倒傳遞類神經,其分析結果顯示,在Logistic模型之整體辨識正確率為79.82%,而在倒傳遞類神經模型之整體辨識正確率為96.9%。其二種分析結果以類神經網路整體辨識正確率較準確。 本研究結果顯示房屋提前還款戶之“再貸款”與“不再貸款”兩類型之房屋貸款戶中特徵玆列如下: 一.「再貸款」房屋貸款戶—年齡在30歲以上、貸放成數在80%以下及借保關係(無保人或有直屬親屬關係)等者。 二.「不再借款」房屋貸款戶—年齡在30歲以下、貸放成數在80%以上及借保關係(有保人且非直屬親屬關係)等者。
Abstract The main purpose of this research is to discuss the variables attributing to the refinance of the early- repayment mortgage loan. 17 variables are defined as: 1. Gender 2. Education level 3. Marital status 4. Age 5. Monthly income 6. Working years for current job 7. Occupation 8. The period of repayments 9. The use of the property 10. Loan amount 11. Loan percentage of appraisal 12. Methods of repayment 13. With or without guarantee 14. With or without government subsidy 15.The proportion of monthly income to total income 16. The relationship between borrower and guarantee 17.The type of the property. The result of the research identities that seven variables which includes gender, monthly income, the use of the properties, loan amount, methods of repayment, with or without guarantee and with or without government subsidy have no significant relationship with the refinance of the early- repayment mortgage loan. However, three variables of marital status, occupation and the proportion of repayment to total income have marginally significant impact on the behavior of refinance. The rest of the variables including education level of borrowers, age, working years for current job, the period of repayments, loan percentage of appraisal, the relationship between borrower and guarantee and the type of the property have significant relationship with the behavior of refinance. The outcome of the analysis reveals that the overall accuracy rates are 79.82% and 96.9% in Logistic model and Back Propagation Artificial Neural Network model respectively while putting the seven significant factors as mentioned above into two models. Obviously, the latter model has higher accuracy rate. The research outcome also indicates that the characteristics of two types of early repayment mortgage loan borrowers- the refinance and non-refinance are shown as following: 1. Refinancing borrowers- over 30 years old, loan percentage of appraisal is below 80% and there are relationship between borrowers and guarantees (no guarantee or direct relatives). 2. Non-refinancing borrowers- below 30 years old, loan percentage of appraisal is higher than 80% and there are no relationship between borrowers and guarantees (with guarantees or non direct relatives).