Assessing CDOs under alternative copulas

Synthetic collateralized debt obligations are popular vehicles for trading portfolios of credit risks. We present a copula based Monte Carlo simulation procedure for pricing them. Using the Gaussian copula of joint default times, we assess the risks of CDOs and their sensitivity to model parameters. Joint defaults are rare; many studies suggest Gaussian copula has limited ability to capture extreme events. We use the copula to assess the risks of misspecifying tail dependence. The choice of copula is shown to significantly affect tranche prices.