Sensitivity analysis of CDO pricing models

Abstract

We compare some popular CDO pricing models and their sensitivity to default probabilities and correlation parameters. Here, the dependence between default times is modeled through Gaussian, stochastic correlation, Student t, Double t, and Clayton copulas. First, we compare these models’ sensitivity with respect to the credit spreads of obligors in the underlying portfolio. This sensitivity analysis has an important implication on hedging. We will investigate whether the models that fit market quotes well agree on the sensitivity to default probability. Moreover, we investigate the sensitivity of tranche premiums with respect to correlation parameters. More specifically, we investigate whether the models that fit market well share a correlation sensitivity pattern. We then conclude about the characteristics of models that are able to fit market quotes.