QuarcHome
Model catalogueBack to the model index

Li’s Gaussian Copula and Nth-to-Default Swaps

A Gaussian copula joins individual default-time distributions through a dependence structure. In a one-factor version, defaults are conditionally independent once the common factor is fixed. This simplifies calculation of the chance that a basket reaches its nth default. Marginal default curves alone do not determine that chance or the contract’s sensitivities.

How it works

A Gaussian copula combines marginal default-time distributions using correlated latent variables. In a one-factor representation, conditioning on the common factor makes defaults independent. Computing the conditional distribution of the number of defaults across time then provides the building blocks for an nth-to-default swap.

Worked example

Take two borrowers with 2% default probability by a fixed date. Under independence, the probability of at least one default is 1 − 0.98² = 3.96%, and the probability of both defaulting is 0.04%. Under perfectly shared defaults these probabilities are both 2%. Marginal probabilities are identical in both cases, while the two notes react in opposite directions.

Validation and limitations

Recover individual survival curves from instruments with consistent settlement terms. Recheck integration accuracy in high correlation and stress the dependence structure as well as the assumed recovery. The value of the contract also depends on when the nth default occurs, not just its probability by a single date.

Further reading