The Empirical Study Of Individual Housing Loan Credit Risk

Personal housing loan is a high-quality asset. It is a effective method that adjust the credit quality of the bank. With the rapid development of real estate industry and the rapidly number of individual housing loan, the credit risk exposed gradually. The U.S. subprime mortgage crisis destroyed financial markets. If personal housing loan credit risk accumulated to a certain degree, the outbreak will trigger a domino type chain reaction and finally harm economic security. Analysis individual housing loan default influence factors and improve the credit risk management method is helpful for Banks to increase the safety of credit assets.This article makes empirical research on the individual housing loan credit risk according to the individual housing loan data of the commercial bank in Xi'an from 2005 to 2008 based on the summary of the individual housing loan credit risk research results and related theory which made by domestic and foreign scholars. Firstly, this article analyses the affecting factors of individual housing loan default risk and makes the variance analysis on the selected indexes. This article describes the distribution of the breach and each index through the statistical analysis of the sample data. Secondly, this article carries on the forecast of the individual housing loan credit risk with the application of the proximal support vector machine (PSVM) and proves the practicality and applicability in the prediction of individual housing loan credit risk by constructing a test matrix and attribute matrix and the model test. Finally, proximal support vector machine is evaluated and made model analysis and error analysis limitations, and the suggestions for improvement are put forward.This article draws the following conclusion through the analysis on large amounts of real and effective personal housing loan data:The accuracy rate which predicts whether a personal housing loan borrower breaches by proximal support vector machine is 85.6%, high accuracy and the first kind of the error rate is low. Proximal support vector machine is more rapid and accurate in dealing with a large sample problem than logistic model. Through the evaluation and test on model, this article suggests that proximal support vector machine can give forecast on individual housing loan credit risk.