Abstract:
As the new Basel Capital Accord (Basel II) was introduced since 2006, banks and other financial institutions are encouraged to calculate their own capital requirements for credit risk under the advanced rating based (A-IRB) approach. In order to be compliant with this approach, financial institutions need to estimate the Loss Given Default (LGD), the credit loss incurred as a fraction of the exposure if the borrower defaults. This paper studies LGD using a large set of historical facility-level default and recovery data of residential mortgages from a major bank in New Zealand. We find that the LGD is highly sensitive to the loan-to-value ratio (LVR), however, the empirical LGDs of each LVR band are significantly lower than the prescribed ones set by the Reserve Bank of New Zealand (RBNZ). The main gap is the regulator's model did not take into account of the fact that a portion of defaulted accounts would go back in order, while the empirical model has incorporated this fact. We also quantify the potential impact of mortgage loss severity in distressed housing markets, comparing to normal housing market conditions. We also find that the level of the change of the property value significantly affect LGD. These findings have important policy implications for several key issues in Basel II implementation.
