Copula Based Synthetic CDO Pricing

This thesis focuses on the pricing of synthetic CDO’s using different copulas. For each copula, we compare the estimated tranche spreads for different homogeneous asset correlations, how the tranche spreads change when a heterogeneous asset correlation is imposed and finally what the implied correlation curve looks like. The Variance Gamma leads to the highest tranche spreads in most cases while the spreads under the Gaussian, Student’s t and Clayton’s are quite similar. We see furthermore that the introduction of heterogeneous asset correlations has different effects on the tranche spreads compared to the case with a homogeneous asset correlation. This difference in tranche spreads is significantly larger when the asset correlation matrix is wider. Finally, the Variance Gamma and Clayton’s copula fit the market tranche prices much better. The Student’s t copula has the worst fit.