A Framework for Stress Testing Bank's Credit Risk

This paper develops a framework for stress testing the credit exposures of Hong Kong's retail banks to macroeconomic shocks. It involves the construction of macroeconomic credit risk models, each consisting of a multiple regression model explaining the default rate of banks, and a set of autoregressive models explaining the macroeconomic environment estimated by the method of seemingly unrelated regression. Specifically, two macroeconomic credit risk models are built. One model is specified for the overall loan portfolios of banks and, to illustrate how the same framework can be applied for stress testing loans to different economic sectors, the other model is specified for the banks' mortgage exposures only. The empirical results suggest a significant relationship between the default rates of bank loans and key macroeconomic factors including Hong Kong¡¦s real GDP, real interest rates, real property prices and Mainland China's real GDP. Macro stress testing is then performed to assess the vulnerability and risk exposures of banks' overall loan portfolios and mortgage exposures. By using the framework, a Monte Carlo method is applied to estimate the distribution of possible credit losses conditional on an artificially introduced shock. Different shocks are individually introduced into the framework for the stress tests. The magnitudes of the shocks are specified according to those occurred during the Asian financial crisis. The result shows that even for the Value-at-Risk (VaR) at the confidence level of 90%, banks would continue to make a profit in most stressed scenarios, suggesting that the current credit risk of the banking sector is moderate. However, under the extreme case for the VaR at the confidence level of 99%, banks' credit loss would range from a maximum of 3.22% to a maximum of 5.56% of the portfolios, and if a confidence level of 99.9% is taken, it could range from a maximum of 6.08% to a maximum of 8.95%. These estimated maximum losses are very similar to what the market experienced one year after the Asian financial crisis shock. However, the probability of such losses and beyond is very low.