Abstract
This paper aims to estimate joint default probabilities under the structural-form model with a random environment; namely stochastic correlation. By means of a singular perturbation method, we obtain an asymptotic expansion of a two-name joint default probability under a fast mean-reverting stochastic correlation model. The leading order term in the expansion is a joint default probability with an effective constant correlation. Then we incorporate an efficient importance sampling method used to solve a first passage time problem. This procedure constitutes a homogenized importance sampling to solve the full problem of estimating the joint default probability with stochastic correlation models. © Springer-Verlag Berlin Heidelberg 2012.