為因應詭譎多變的經濟環境，避免公司因產生逾期放款（Non-performing Loans）而發生呆帳損失，故影響房屋貸款之信用風險因子一直是各家金融業建構信用評分之依據。本研究主要以人（貸款客戶條件及信用狀況）與物（擔保品座落及價格）為評估重點，以國內某金融業之房屋貸款為主要研究基礎，探討造成客戶發生信用風險之主要原因。研究資料以1994年至2019年間申貸房屋貸款之客戶為主軸，於資料去識別化後，先運用Logit模型、Probit模型為主軸並搭配LPM模型，再運用逐步迴歸模式研究各個變數影響信用風險之顯著程度，進而得出影響房屋貸款之顯著信用風險因子。本研究結果發現，主要的二種解釋模型（Logit及Probit）顯示出影響房屋貸款較為顯著之17個信用風險因子，分別為貸款金額、性別、婚姻狀況、公司排名、教育程度（國中、高中職及專科）、年收入、有信用卡循環、承作區域、其他借款（短期借款、車貸及學貸）、有延遲繳款及次數、服務單位、申貸年限等；此外，藉由時間因子的加入，實證分析結果顯示過去與現在之違約率確實在不同時間有相當差異。

In response to the treacherous and capricious economic environments and in order to avoid bad debt losses due to non-performing loans, the credit risk factors that affect housing loans have always been the basis for the construction of credit scores in various financial institutions. In this study, we discuss this issue by two perspectives: 1. People, i.e., the conditions of loan customers and credit status of the borrowers, and 2. Objects, i.e., the location and market price of the collateral, and explore the major reason of customer credit risks on the basis of mortgage in a domestic financial institution. The collected data was based on the housing loans applicants over the last 25 years from 1994 to 2019. After de-identifying the data, we use the Logit model and the Probit model as the main tools along with the LPM model as the supplementary tool. We then use the stepwise regression model to study the significance of each variable affecting credit risks. Through this process, the significant credit risk factors that influence housing loans were obtained. Our results from the Logit and Probit models show that there are 17 significant factors that affect housing loans, which include the loan size, gender, marital status, company ranking, education level, bank grading, annual income, the number of credit card circulation, other loans (short-term loans, car loans and student loans), the number of delayed payments, service units, loan application period. In addition, when time factor is added, it is observed that the default rate indeed varies across different time periods.

參考文獻
一、 中文文獻
1. 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」
2. 陳逸文 (2000)，台灣金融財務季刊銀行授信之風險管理
3. 林建州 (2001），「銀行個人消費信用貸款授信風險評估模式之研究」，國立中山大學財務管理研究所碩士論文
4. 劉長寬 (2003)，「應用Logit模型於消費者擔保貸款違約行為之實證研究」，陽科技大學財務金融系碩士論文
5. 王愛儉 (2005)，中國社區銀行發展模式研， 中國金融出版社, 2005.12
6. 王思評 (2006），「房屋抵押貸款授信風險評估研究-以Ｘ銀行為例」，大同大學事業經營研究所碩士論文
7. 葉彩蓮、翁家君 (2011），「投資客與風險變數對於房屋貸款信用之影響」，東吳經濟商學學報第七十三期（2011.06）:1-28
8. 鄭宗志 (2011)，應用經濟學快報，2011,18,1437-1442台灣表演藝術出席終結者：常態概率分析。國立政治大學統計系
9. 蘇佳翎 (2015），「台灣上市櫃電子製造業 財務危機預警模型之實證研究Logistic模型KMV模型之比較」，台灣科技大學研究所碩士論文
10. 邱政憲 (2017），「影響壽險業房貸違約因素之研究」，國立政治大學地政學系碩士在職專班論文
11. 程濤 (2003)，財務預警模型綜述.山西財經大學學報，2003,25（5）
12. 東吳經濟商學學報第四十八期（2005.03）:103-126 黃嘉興、謝永明、劉宗哲「房屋抵押貸款客戶違約預測模型之比較研究」
13. 汪瓊，第五講,第四節.教育技術研究方法-課程講義 北京大學教育學院
14. 金融監督管理委員會網站資訊109.03.12 逾期放款辦法
15. 商管科技季刊第十卷第三期（2009）Commerce & Management Quarterly, Vol. 10, No.3, pp. 465- 489 (2009)
二、 英文文獻
1. Updegrave (1987), “How Lender Size You Up” , Money
2. Crook, Edelman, Thomas (2007), “Recent developments in consumer credit risk assessment”, European Journal of Operational Research, vol. 183, pp. 1447-1465
3. Azam, Rehan Muhammad, Danish Syed Akbar, Suleman (2012),“The significance of socioeconomic factors on personal loan decision a study of consumer banking local private banks　in Pakistan”
4. Cuiqing Jiang, Zhao Wang, Ruiya Wang & Yong Ding (2017), “Loan default prediction by combining soft information extracted from descriptive text in online peer-to-peer lending”
5. Bennouna and Tkiouat (2019), “Scoring in microfinance: credit risk management tool –Case of Morocco-“, Procedia Computer Science Volume 148，2019，pp. 522-531
6. Brooks（3ed）Chap 12
7. Wooldridge, 6/e, Chap 17