Makroobezřetnostní zátěžové testování solventnosti bank: modely a metodiky pro kreditní riziko

This thesis deals with models and methodologies for macroprudential stress testing of bank solvency, focusing on the area of credit risk. From a theoretical standpoint, the thesis presents a comprehensive framework for macroprudential stress testing, from stochastic scenario generation using structural Bayesian VAR models to advanced techniques for modelling key credit risk parameters. Specifically, a Bayesian model averaging (BMA) methodology is introduced for the robust estimation of probability of default (PD), a structural model for loss given default (LGD), and a transition probability matrix model for estimating expected credit losses under the IFRS 9 accounting standard. Furthermore, the thesis introduces a granular approach to modelling risk-weighted assets (RWA) based on the regulatory ASRF model. These methodologies are subsequently used not only to quantify impacts on capital adequacy but also to analyze broader topics in supervision and financial stability, such as the disclosure of stress test results, the procyclicality of the IFRS 9 standard, the calibration of the Countercyclical Capital Buffer (CCyB), and an innovative approach to measuring the macroprudential policy stance.