LAUSR

The predominant approach for studying macro-financial linkages is employing standard econometric techniques to link bank-specific risk parameters to macroeconomic and financial indicators. A single model is estimated using historical data and is subsequently used to forecast their evolution under various scenarios for the explanatory variables. However, the implicit assumption of invariable relationship structure across different economic conditions might not necessarily hold therefore introducing important model risk. Focusing on Greece, Italy, Ireland, Portugal, and Spain, this study examines the dynamic nature of the relationship between credit risk and macroeconomic and market-based indicators under stress. The empirical results indicate that the relationship structure is changing during crises and a significant number of models is affected by this change. Model combination techniques are utilized to mitigate the impact of this structural change. The results show that model combination outperforms individual models under changing economic conditions.