Point-in-Time PD Term Structure Models with Loan Credit Quality as a Component

Most point-in-time PD term structure models used in industry for stress testing and IFRS9 expected loss estimation apply only to macroeconomic scenarios. Loan level credit quality is not a factor in these models. In practice, credit profile at assessment time plays an important role in the performance of the loan during its lifetime. A forward-looking point-in-time PD term structure model with loan credit quality as a component is widely expected. In this paper, we propose a forward-looking point-in-time PD term structure model based on forward survival probability, extending the model proposed in [8] by including a loan specific credit quality score as a component. The model can be derived under the Merton model framework. Under this model, the forward survival probability for a forward term is driven by a loan credit quality score in addition to macroeconomic factors. Empirical results show, the inclusion of the loan specific credit score can significantly improve the performance of the model. The proposed approaches provide a tool for modeling point-in-time PD term structure in cases where loan credit profile is essential. The model can be implemented easily by using, for example, the SAS procedure PROC NLMIXED.