We describe an approximation method, based on binomial distribution adjustments, for constructing independent loss distributions.
KaneAffiliated Professor, EDHEC Business School
This method can handle both homogeneous and heterogeneous loss portfolios. We find that this simple algorithm provides an excellent fit to the exact distribution for a broad range of correlations and portfolio credit quality. For typical correlations, and homogeneous loss portfolios, the percentage error in the spread is typically 0.04% and usually less. It is at least 30 times faster than a full homogeneous loss recursion, and significantly faster for inhomogeneous loss portfolios.
|Research Cluster :||Finance|