Default probabilities in credit risk management: estimation, model calibration and backtesting

This PhD thesis is devoted to the estimation and examination of default probabilities (PDs) within credit risk management. Assigning an appropriate PD is a widely employed strategy by many financial institutions as well as the supervisory authorities, and providing accurate estimates of PDs is considered as one of the key challenges in credit risk management. False estimation of these probabilities leads to, among other things, unreasonable ratings and incorrect pricing of financial instruments. As a matter of fact, these issues were among the key reasons for the global financial crisis (GFC) as undervaluation of the default risks in the mortgage market almost caused the collapse of the financial system. The first research paper, titled Structural Credit Risk Models with Subordinated Processes, analyses structural models based on the Merton framework. First, we observe that the classical distributional assumption of the Merton model is rejected. Second, we implement a structural credit risk model based on stable non-Gaussian processes as a representative of subordinated models in order to overcome some drawbacks of the original Merton approach. Finally, following the Moody’s KMV estimation methodology, we propose an empirical comparison between the results obtained from the classical Merton model and the stable Paretian one. In particular, we suggest alternative parameter estimation techniques for subordinated processes, and we optimise the performance for the stable Paretian model. Our results indicate that PDs are generally underestimated by the Merton model and that the stable Lévy model is substantially more sensitive to the period of the financial crises. The second study, Prediction of U.S. Commercial Bank Failures via Scoring Models: The FFIEC Database Case, is devoted to examining the performance of static and multi-period credit-scoring models for determining PDs of financial institutions. We use an extensive database for the U.S. provided by the Federal Financial Institutions Examination Council (FFIEC). Our sample contains more than 7,000 U.S. commercial banks with 405 default events. Our analysis also focuses on evaluating the performance of the considered scoring techniques. We apply a substantial number of model evaluation techniques, including methods that have not yet been applied in the literature on credit scoring. We also provide an overall ranking of the models according to the different evaluation criteria and find that the considered scoring models provide a high predictive accuracy in distinguishing between defaulting and non-defaulting financial institutions. Despite the difficulty of predicting defaults in the financial sector, as has been mentioned in the literature, the proposed models also perform very well in comparison to results on scoring techniques for the corporate sector. Finally, the last research paper, titled Distress Risk and Stock Returns of U.S. Renewable Energy Companies, investigates the question of whether distressed risk is priced in the renewable energy sector. Using the Expected Default Frequency (EDF) measure obtained from Moody’s KMV, we demonstrate that there is a positive cross-sectional relationship between distress risk and returns of equally-weighted (EW) portfolios, providing evidence for a distress risk premium in the U.S. renewable energy sector. The positively priced distress premium is also confirmed by investigating returns corrected for common Fama and French, and Carhart risk factors. We further show that raw and risk-adjusted returns of EW portfolio that takes a long position in the 10% most distressed stocks, and a short position in the 10% safest stocks, generally outperform the S&P 500 index throughout our sample period (2002–2013).