Modelling default risk including idiosyncratic default risk in returns on corporate bond portfolios

Student thesis: Master's ThesisMaster of Philosophy


Returns on corporate bond portfolios are subject to both systematic and idiosyncratic default risks. Koivu and Pennanen (2010) derived two-factor return models for corporate bonds, using one of the risk factors to approximate e ects of systematic default risk on returns. Koivu and Pennanen (2014) similarly showed that modelling returns of index-linked bonds can be reduced to statistically modelling of two risk factors, one of which is an underlying index and could be used to model default losses. However it wasn't discussed further how this can be done. Our goal is to model stochastically default losses on returns of corporate bonds using the underlying index in Koivu and Pennanen (2014), whilst considering e ects of both systematic and idiosyncratic default risks. We rst give a more precise economic meaning to the underlying index for corporate bonds. We then express returns of corporate bonds as a function of default losses, which in turn are a function of the underlying index. Instead of approximating default losses using historical yield spreads like Koivu and Pennanen (2010), we model them over time as a compound (inhomogeneous) Poisson gamma process. Parameters of our default losses model are easily estimated using the Maximum likelihood method. Default losses simulated using our model are reasonably close to historical default losses of the Bank of America Merrill Lynch US High Yield index. Most importantly, using our proposed default losses model, our two-factor return model for corporate bonds is suitable for both well-diversi ed and non well-diversi ed bond portfolios.
Date of Award1 Apr 2020
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorTeemu Pennanen (Supervisor) & Cristin Buescu (Supervisor)

Cite this