Does the GARCH Structural Credit Risk Model Make a Difference?

Abstract: In this study, we empirically investigate and evaluate various approaches to structurally assess credit risk using a panel of European banking groups. We consider not only the standard approaches in the literature, but also include models that allow the asset volatility to be stochastic and models that allow for short- and long-term components of default risk. Models are evaluated by comparing their ability to correctly and timely identify changes in risk indicators. Surprisingly, we find that the GARCH structural credit risk model, despite its more sophisticated modeling approach, typically underperforms more basic models. Importantly for macro-prudential policy, the combined Merton/GARCH-MIDAS model performs best and reflects important market events earlier than the other approaches.