University of Twente Student Theses


Performance of corporate bond indices: an approach inspired by Merton’s model

Breukelen, N. van (2013) Performance of corporate bond indices: an approach inspired by Merton’s model.

[img] PDF
Abstract:This thesis explores the application of using option pricing methodology in a firm value model in the evaluation of several different corporate bond types. More specifically, the aim of this thesis is to evaluate whether the application of market implied volatilities in a generalization of the Merton model results in a useful risk management tool for financial institutions. The corporate bonds returns have been obtained by modeling the corporate bonds as a nominal risk free corporate bond and equity derivatives. In order to capture the characteristics of the different types of corporate bonds put and call options have been used. The three types of bonds which have been modeled are investment grade, collateralized and high yield corporate bonds. Currently used models require a lot of factors which are not directly observable in the financial markets and therefore need to be estimated. The proposed model requires none of these factors and still tries to capture the main characteristics of the risk distribution of corporate bonds by extending the Merton model. This extension involves a risk free component, a nominal bond, and a combination of equity derivatives for modeling the three different types of corporate bonds. As these three types of corporate bonds have different characteristics it is required to model each corporate bond with different equity derivatives. We have tested the proposed model to three corporate bond benchmarks for portfolios of institutional investors. The three IBoxx benchmarks are a good representation of the complete spectrum of available bonds. The three types range from risky high yield bonds, investment grade bonds to low risk bonds protected by high quality collateral. An In-Sample test has been performed over the period 2006-2010 in order to evaluate the proposed model used for capturing the risk distribution of the three types of corporate bonds. To test whether the model is not just fitted on the in sample period, also an Out-of-Sample test is performed over the year 2011. The outcome of the In- and Out-Of-Sample test shows that, although the results vary over the three types of corporate bonds, there is evidence that modeling corporate bonds with the proposed model adequately captures the behavior of the investment grade bond index and the high yield bond index, even in the volatile time frame of the last couple of years. The contribution of the model is however foremost the replacement of unobserved inputs with market factors, thereby reducing model risk. The performance of the model differs between the corporate bond types. As we do not use specific bond factors, but try to capture the risk in the bonds using equity price movements, the model might not be able to capture specific risks of the bonds. An explanation for the differences in performance therefore might be found in the difference in riskiness of the three bond types. The model performs best with the high yield corporate bonds, which might be explained by the fact that these bonds have a high sensitivity to the equity price movements. The outcome of the proposed model for investment grade corporate bonds shows that in comparison to the high yield corporate bond model the behavior of this type of corporate bond is more difficult to capture. The behavior of collateralized corporate bonds is not adequately captured by the proposed model. This outcome shows that the proposed model needs an additional factor for capturing the specific risks of low risk bonds. The performance of the proposed model has also been compared against one of the most common ways to model corporate bonds in an ALM context. The proposed model has the benefit in comparison to the current used methods that it needs no market factors to be estimated while it still has a strong intuitive appeal. A regression analysis showed that both the proposed model and the returns of the Euro Stoxx 50 index and risk free interest rates showed that they explain the same level of variance and the standard errors are also similar. However, the proposed model has the big advantage of a better description of the left tail of the distribution, important for risk measures. This shows that the proposed model could be a good alternative for analyzing portfolios of corporate bonds in an ALM or other risk management setting.
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:85 business administration, organizational science
Programme:Business Administration MSc (60644)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page