Modelling long run PD for SME portfolio

Through-the-Cycle (TTC) PD is the long-run probability of default through a full credit cycle. It is a more suitable risk factor for the capital purpose than Point-in-time (PIT) PD as it is less sensitive to the fluctuations of the economic environment. Accurate forecasting of the TTC PD is an industry challenge for Australian banks due to the lack of data in sufficient long period of time. Furthermore, low default rates are often observed in the book of small and medium enterprises (SME), which imposes extra difficulty for the TTC PD estimation. To meet these challenges, this research proposes a TTC PD model constructed in a way similar to the Cox Proportional Hazard modeling framework by incorporating the long-term survivors. The variations of the hazard are explained by four components including the non-linear life-cycle effect, credit quality of the portfolio, macro-economic environment and seasonality effect. In particular, the origination dates (vintages) are used as a proxy of the changes of credit policy. Back-casting with historical economic data and simulation techniques are used to provide an estimate of TTC PD after the model is built.