| Since the Construction Bank of China lent the first loan of residential mortgage in 1992, it has become the major financing way in which people realize the dream of having a house and has accepted by the masses of residents gradually. By the end of 2002, bank's loan balance of residential mortgage already exceeded 820 billion Yuan in China. And it still increases fast at the speed of 40 - 50% per year. Meanwhile, default risk rate of residential mortgage begins to soar constantly too. Average default risk rate of residential mortgage was about 1%, and potential risk has already appeared gradually (Liu Ping, 2002). Now, financial circles hi China have generally realized the necessity and urgency of strengthening residential mortgage default risk management, and academia in China has also paid attention to it. However, systemic theoretical researches and empirical researches are still blank in this field. Because of above-mentioned backgrounds, the dissertation regards the determinants of residential mortgage default risk as research object, and attempts to identify the major determinants, and carries on risk clustering. This research will help banks build theoretic and technological foundation to reduce default risk at the beginning of residential mortgages lending.In developed countries, financial institutions have already had a history of more than one hundred years to lend residential mortgages, and this business is very mature. Correspondingly, scholars of these countries have carried on a lot of theoretic and empirical researches to probe into residential mortgage default risk, and achievements in those researches are very abundant. Those achievements offer some theoretical reference for choosing variables and quantization of this research. Based on literature research, the author had interviewed with 15 residential mortgage managers or loan officers of 6 commercial banks. Finally, the author selected 13 sub-branches of one state-owned bank in Hangzhou city as sampling bank whose residential mortgage dada was relatively intact. 5855 normal loan samples and 4865 default loan samples were collected hi this bank. After getting rid of variables missed or filled out by mistake, 1025 effective normal loan samples and 836 effective default loan samples were obtained, so the total capacity of the samples was 1861. SPSS 10.0 software is used to do Factor Analysis, Discriminant Analysis, Cluster Analysis and Logistic model by using those investigated data. The main conclusions and innovations of the dissertation are as follows:(1) To design an empirical analysis framework by using Factor analysis,Discriminant analysis, Cluster analysis and Logistic model. Literatures indicate that Logit Model, Logistic Model, Probit Model, MLR, Cluster Model, Option Model, Proportion Risk Model, and Discriminant Analysis are the main effective methods and models when researching the determinants of residential mortgage default risk. Based on Literature researches, the dissertation designs an empirical analysis framework including several methods and models we mentioned before. The summaries of these models, methods, and the consequent empirical analysis framework provide a methodology foundation for researching residential mortgage default risk in our country.(2) To put forward a four-dimension framework for variable choosing. In literatures reading, the author found in their empirical analysis residential mortgage default risk that foreign scholars usually chose their research variables from two or three in the following four dimensions: loan characteristics dimension, borrower characteristics dimension, property characteristics dimension and regional characteristics dimension. Though some specific documents have mentioned to choose variables from the four dimensions, those literatures had not pointed out clearly the variable choice framework. Based on summarizing foreign scholars' experience of selecting variables in researching residential mortgage default risk, the dissertation proposes de...
|
PDF Full Text Request