Stein's method and zero bias transformation for CDO tranche pricing (Q964669)

The authors propose an efficient approximation method, by adopting the framework of the Stein method and the zero bias transformation, to evaluate the credit portfolio product, which is named Collateralized Debt Obligation (CDO). The main term to compute has the structure of a call function for the cumulative loss on a portfolio of financial assets susceptible to default risk. The defaults are supposed to be conditionally independent given a common market factor. In this case the conditional loss can be written as a sum of independent random variables and the classical approximation problem arises for the expectation of functions of such sum variables. The authors propose to combine Gaussian and Poisson approximations, since the Stein method can be applied in both the cases. Also, a first-order corrector is proposed in each case. The two approximations are applied to CDO tranche pricing, and an empirical threshold is proposed between the two approximations. Some explicit estimations are presented.