影響台灣房貸業務逾期繳款之授信因子研究

A Research for the Factors of Nonperforming Mortgages in Taiwan

周俊民(Chun-Mine Chou)

指導教授 ： 池祥麟

國立臺北大學/商學院/國際財務金融碩士在職專班/碩士(2010年)

摘要

房屋貸款長久以來一直是台灣金融機構獲利最穩定，授信風險最低的業務之一。因此已成為台灣各金融機構消費者貸款業務中所占比重最高的項目，故房貸之逾放比重大小，影響各金融機構之收益甚巨；且近期社會輿論不斷反映房價不斷上漲，已背離市場價值，為防止房屋市場泡沫化，中央銀行亦厲行各項政策指導各銀行，限縮對豪宅產品及房屋投資客之貸放條件，如：提高貸放利率，降低貸房成數，並一律取消寬限期，如此措施之下對未來金融機構之房貸業務影響極為深遠。
本次研究之目的主要為運用羅吉斯迴歸模型研判影響台灣各不同區域房屋貸款違約風險的各類授信因子，其中包括央行嚴格要求緊縮貸放條件的對象：投資客及豪宅(高總價房屋)。同時檢測央行之因應策略，如：提高貸放利率，降低貸款成數及取消寬限期等是否為影響房屋貸款違約風險的重要授信因子。
本研究拮取國內某家金融機構自民國80年成立，迄98年12月31日止所貸放之房屋貸款案件尚有現欠之95,442筆為研究資料。經捨棄借款人重複及資料不全之案件後最後以46,236筆借款案件為本次研究之對象。
經以本研究模型評估上開研究資料，所得到之實證結果為：第一、影響整體台灣房貸業務是否違約之最顯著授信變數為：利率、初貸成數、性別、是否有寬限期、是否有保證人。第二、影響台北市房貸業務是否違約之最顯著授信變數為：初貸成數、是否有寬限期。第三、影響台灣北部地區(不含台北市)房貸業務是否違約之最顯著授信變數為：利率、初貸成數、初貸金額、是否有寬限期。第四、影響台灣中部地區房貸業務是否違約之最顯著授信變數為：利率、性別、是否有寬限期。第五、影響台灣南部地區房貸業務是否違約之最顯著授信變數為：利率、初貸成數、職位、是否有寬限期、是否有保證人。第六、央行嚴控房貸之政策中：是否有寬限期是影響所有房貸是否違約之最顯著授信變數；且經實證借款人是否為投資客對房貸業務是否違約無存在顯著的相關性。
因此建議金融機構需建立涵蓋過去已清償結案及被駁回的所有案件的房貸授信資料檔案系統，以備後續查詢及研究；同時亦需建立定期篩選顯著影響放款成敗的授信變數之制度，做為調整授信審核架構之依據。

關鍵字

房貸 ； 違約率 ； 羅吉斯迴歸模型 ； 寬限期 ； 投資客

並列摘要

For a long time, the mortgage is one of the stable profit factors and lower credit risks for the financial institutions in Taiwan. As the mortgage occupies the highest proportion of various consumptive loan services for Taiwan financial institutions, the revenues of the Taiwan financial institutions is significantly related to the proportion of overdue mortgages. Moreover, recent public opinions continue to reflect rising house prices and the departure between house prices and the market value. To prevent the housing market bubble, the Central Bank strictly enforces the policy guidance to the banks and restricts the loan terms on luxury house products and housing investors, such as increasing loan rates, reducing the credit lines of mortgages, cancelling the grace periods, and it may have a great effect for mortgage business of Taiwan financial institutions in the future.

This research studies the various mortgaged credit factors affecting mortgaged default risks for different areas in Taiwan by logistic regression model. The credit factors include the targets, housing investors and the mansion (high total price house), which Central Bank strictly requests the restriction of the loan terms. Simultaneously, this research examines and tests the strategies such as increasing loan rates, reducing the credit lines of mortgages, cancelling the grace periods, adopted by Central Bank.

Based on the overdue mortgaged data files during 1991 to 2009 December 31 from one Taiwan financial institution established in 1991, 95,442 cases were initially identified the study samples. After eliminated the duplicate persons and insufficient information, the 46,236 cases were selected as samples in this research.

The results yielded by the analysis are as follows:
1. The major significant credited variables of mortgaged business in Taiwan overall areas are the interest rates, initial credit lines, the sex, whether to have the grace periods and whether to have the guarantors.
2. The major significant credited variables of mortgaged business in Taipei city are initial credit lines and whether to have the grace periods.
3. The major significant credited variables of mortgaged business in Taiwan north areas excluding Taipei city are the interest rates, initial credit lines, initial credit amounts and whether to have the grace periods.
4. The major significant credited variables of mortgaged business in Taiwan central areas are the interest rates, the sex and whether to have the grace periods.
5. The major significant credited variables of mortgaged business in Taiwan south areas are the interest rates, initial credit lines, the position, whether to have the grace periods and whether to have the guarantors.
6. Under strict mortgaged policies regulated by Central Bank, whether to have the grace periods is the most significant credited variables for Taiwan mortgaged business. In addition, after the real diagnosis, there is no significantly correlation between borrowers as the housing investors and mortgaged business.

As a result, it is suggested that the financial institutions have to establish
the mortgaged file system covering all of the settled and rejected cases in the past for the further inquiries and researches. At the same time, the financial institutions have to establish the credited system regularly screening the credited variable significantly affecting the results of success or failure mortgaged cases for the basis of the adjusting credited policies.

並列關鍵字

mortgage ； default rate ； logistic regression ； the grace periods ； investors

參考文獻

1.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,政治大學金融研究所碩士論文。

3.郭雲啟 (2006)，房屋貸款提前還款再貸款之研究-以 S 銀行為例，中原大學企業管理研究所碩士論文。

1.Gardner, M. J. and D. L. Mills (1989), “Evaluating the Likelihood of Default on

2.Steenackers, A. and M. J. Goovaets (1989), “A Credit Scoring Model for Personal

Loans,” Insurance Mathematics Economics, pp.31-34.