【作者】 吴娟；

【导师】 陆珩瑱；

【作者基本信息】 南京航空航天大学 ， 金融学， 2008， 硕士

【摘要】 近年来,我国住房抵押贷款市场得到了快速发展,提前还款风险也随之不断放大,而银行业对提前还款行为的忍痛放任或被动限制将会严重影响资金的使用效率和收益;同时,随着2005年住房抵押贷款支持证券(MBS)的发行,解决其合理定价问题已成为决定其发展前景的重要因素,而准确预测提前还款是合理定价的关键所在,可见,加强对住房抵押贷款提前还款风险的研究已成为当务之急。借鉴国内外的研究成果,本文在搜集、整理住房抵押贷款数据的基础上,引入定量分析法,通过因子分析、判别分析等多元统计方法寻找提前还款的主要影响变量,并据此建立模型来预测提前还款风险。通过研究发现:影响我国住房抵押贷款提前还款的主要因素涵盖贷款特征因素、经济环境特征因素、借款人特征因素和房屋特征因素,具体分为8大因子变量——利率因子、部分提前还款因子、贷款期限价值比因子、时间过程因子、借款人性别因子、借款人年龄因子、借款人财力因子和借款人单位类别因子;在构建预测提前还款模型的过程中发现,非线性Logistic模型比线性判别函数的总体预测准确率更高,但判别函数对提前还款这类贷款的判别正确率更高,可以说前者属于保守型预测模型,后者属于激进型预测模型。更多还原

【Abstract】 In recent years, the mortgage market in China has been witnessing a fast development, at the same time, prepayment risk is enlarged continuously. In this situation, the banking either endures the risk silently or takes restriction passively, which will decrease the efficiency and income of funds seriously. Simultaneously, with the issue of mortgage-backed security in 2005, the reasonable pricing of it has become a decisive factor in its development, moreover, the key step of pricing lies in estimating prepayment accurately. It is thus clear that, strengthening the study on the prepayment risk of mortgage has become an urgent matter of the moment. Based on the international and domestic research result, this paper uses the quantitative method to study on prepayment. According to factor analysis and discriminant analysis of multivariate statistics, the paper tries to find out the major influencing factors of mortgage’s prepayment of China, and then establish econometric model of prepayment to predict the prepayment risk. From the study, we discover that the main influencing factors of our mortgage’s prepayment include the characteristic factors of mortgage, the characteristic factors of economic environment, the characteristic factors of mortgage’s borrower and the characteristic factors of house, we can divide them into 8 factors, namely, the interest rate factor, the partial prepayment factor, the loan to value factor, the time process factor, the borrower’s gender factor, the borrower’s age factor, the borrower’s financial position factor and the category factor of borrower’s work; From the established prepayment models, we find out that nonlinear logistic function can predict more accurately than linear discriminant function as a whole, but discriminant function can predict more accurately on prepayment mortgage, therefore we can say that the former is a conservative model and the latter is a radical one.更多还原