Loss Given Default modelling for corporate bonds

Details

Supervisors

Hainaut, Donatien

Faculty

Faculté des sciences

Degree label

Master [120] en sciences actuarielles, à finalité spécialisée

Abstract

The estimation of Loss Given Default (LGD) is a critical aspect of credit risk assessment, influencing loan pricing, capital allocation, and risk management strategies in financial institutions. This master's thesis aims to develop a rigorous and regulation-compliant methodology when modelling LGD or the Recovery Rate. The research is guided by five key objectives. Firstly, the study embarks on an exploration of preprocessing techniques tailored specifically to LGD datasets. The goal is to develop a standardised and effective methodology that addresses the characteristics and challenges of LGD data. Secondly, the thesis investigates various modelling approaches for LGD estimation, leveraging advanced techniques such as glassbox models. By capturing the technicalities of credit risk dynamics, these models aim to enhance predictive accuracy and reliability. In addition, the research expands the existing LGD literature by exploring alternative loss functions beyond the traditional Mean Squared Error (MSE). These alternative functions are evaluated for their efficacy in capturing the underlying risk factors associated with LGD. Furthermore, a comprehensive economic comparison of different models and loss functions is conducted to assess their performance and suitability in real-world financial applications. This analysis provides valuable insights into the economic implications of LGD modelling decisions. Finally, state-of-the-art interpretability methods are employed to elucidate the results obtained from the LGD models. By enhancing transparency and regulatory compliance, these methods contribute to a deeper understanding of the underlying risk factors driving LGD estimates. The goal of this research is to build a robust and regulation-compliant framework for LGD modelling, offering valuable insights for financial institutions and regulators alike.