Portfolio credit risk models are grasping the important independent variables such as default rate by estimating the values of long-term mean based on actual occurrence, but these values of long-term mean may seriously distort the short-term forecast which is dependent on the economic conditions at a point of specific time.
Using CreditRisk+ model, this paper examines if there is a kind of meaning among the trends of Credit VaR values of three portfolios which are composed of the bankrupted companies at different period for five years before default and if the model has the power of forecasting the actual amount of risk.
The results of analysis are as follows. There is no consistency among the trends of Credit VaR values of three portfolios under confidence level of 99% which is used generally. However, there is consistency among them under confidence level of 99.9%. And it appears that the movement of Credit VaR value does not precede the actual amount of risk such as the increment of default rate. On the contrary, it appears that the change of default rate precedes the movement of Credit VaR value.
