거시변화 및 차주 특성을 반영한 주택담보대출 위험 예측모형

This paper makes a macro-micro combined model to forecast mortgage defect probability. The model consists of two-step procedure in which borrower’s characteristics and macroeconomic variables are included as independent variable. At the first step, we develope a binary choice model with time dummies, assuming that the defects probability is affected by both borrower’s characteristics and periodical variations. At the second step, we identify a time series model that forecasts the time dummy coefficients with macroeconomic independent variables. Since reflecting both macro and micro environmental information, the model has advantages in both sorting out a risk group at micro level and measuring a size of macroeconomic shock at aggregated level. The estimation results are statistically stable. It means that macroeconomic forecasting model, which is based on the laws of large number, seems to be showing mortgage borrower’s defect process on a limited way. Simulation study also shows that defect probability in a case of considering both macro level and micro level effects is approximately two times higher than in a case of considering only micro level effects. Therefore, both the authorities and individual companies have an important role to handle mortgage market risks.