影響房屋貸款信用評等因素之個案研究

A Case Study on the Factors Influencing Credit Grading of the Housing Loan in Taiwan

何佳霖(Ho Chia-Lin)

指導教授 ： 劉欽泉

亞洲大學/管理學院/經營管理學系碩士班/碩士(2004年)

摘要

授信業務是金融機構收入的主要來源，掌握最正確的貸款人背景資訊，降低日後違約風險，是銀行重要的課題。因此本文旨在探討影響房屋貸款的信用評等因素，選擇地段、屋況、貸款額度佔放款值成數、所得、職業、每月平均收入與每月攤還本息倍數比、年齡、抵押品類型、屋齡、有無其他不動產等評等因素作為研究變數；以類神經網路分析為研究方法進行實證分析。實證結果發現:年齡愈大傾向低貸款等級，職業類別為醫師、律師、建築師、會計師、軍警公教主管級以上人員、大專院校副教授級以上人員等是高貸款等級，職業類別為其他類者是第3類貸款額度，抵押品為透天型有較高貸款額度，屋齡愈大有低貸款等級，地段愈差為較低貸款額度，已無其他不動產者傾向第4類(401萬~600(含)萬)貸款等級，尚有其他不動產者為最高貸款等級，貸款額度佔放款值成數7成(含)以下者傾向高貸款額度，年所得低為低貸款等級，借款人每月收入達攤還本息金額1.5倍(含)至2倍(不含)者為較低貸款額度，屋況愈佳有最高等級貸款額度。最後，將低、中、高三類貸款額度等級的敏感度相關係數取最大的前五名做排序，發現「貸款額度佔放款值成數」與「抵押品類型」在三類等級中都是在核貸時會先行考慮的因素。在最低與最高等級中，前五名的輸入變數完全相同，顯示在影響低額與高額的貸款額度中，同樣認為屋齡、屋況、年所得、抵押品類型、貸款額度佔放款值成數等是決定貸款額度高低的主要原因。

關鍵字

房屋貸款 ； 信用評等 ； 類神經網路

並列摘要

Loan is a main earnings of the financial institution. How to grasp most correct background information of creditor, and reduce the default risk in the future is the important subject of the banks. The main purpose of the thesis is to analyze the factors that influence credit grading of the housing loan. 10 grading factors have been selected to be the analysis variables: house location, house conditions, loan-to-value, creditor income, occupation, the reimbursement ratio of the creditor monthly income, age, collateral type, housing age, the other real estates. The Neural Network analysis has been applied as the research instrument. We find that lower grade of loan incline to older age; the occupation type 3(doctor, lawyer, architect, accountant, the commissioner of civil servant and above, associate professor and above) has the first grade of loan; collateral type is villa which has higher grade of loan; the region location is worse which has lower grade of loan; people Who doesn’t have the other real estate incline to fourth grade of loan(4 million 10 thousand dollars~ 6 million(include)dollars); loan-to-value which is less than seventy percent(include)inclines toward higher grade of loan; income is lower which has lower grade of loan; the reimbursement ratio of the creditor monthly income is 1.5(include)~2(not include)which has lower grade of loan; the better condition has higher grade of loan the financial institution will evaluate. Finally, we choose the top five coefficients of the lowest, middle and highest grade of loan of the sensitivity analysis. We find that loan-to-value and collateral type are considered in advance factors. In the lowest and highest grade, the top five input variables are all the same. It’s showing that housing age, house conditions, income, collateral type and loan-to-value are the main reasons for the financial institutions to determine the amount grade of loan.

並列關鍵字

housing loan ； credit grading ； neural networks

參考文獻

劉展宏、張金鶚(2001)，購屋貸款提前清償行為之研究。住宅學報，10(1)，29-49。

江百信、張金鶚(1995)，我國購屋貸款放款條件之研究。住宅學報，3，1-20。

李桐豪、呂美慧(2000)，金融機構房貸客戶授信評量模式分析-Logistic迴歸之應用。台灣金融財務季刊，1(1)，1-20。

Archer, W. R., David, C. L., & Gary, A. M. (1996). The effect of income and collateral constraints on residential mortgage terminations. Regional Science and Urban Economics , 26 , 235-261.

Deng, Y. H., John, M. Q., & Robert, V. O. (1995). Mortgage Default and Low Downpayment Loans：The Costs of Public Subsidy. NBER Working Paper,