Lucas, Douglas J., "Default Correlation and Credit Analysis", Journal of Fixed Income, Vol. 4, No. 4, (March 1995), pp. 76-87.
Abstract: The paper addresses the credit analysis of debt that has two separate sources of payment, such as an underlying obligor and a guarantor. In addressing this specific problem, the paper supplies the first theoretical and empirical quantification of default correlation. It develops an equation relating the probability that two credits will jointly default to default correlation and the probability that each individual credit will default. The paper also develops a methodology to calculate historic default correlations and presents historical default correlations using Moody's default data. But the paper points out the inability of pairwise default correlation or variance/co-variance matrices to describe default probabilities in a portfolio comprised of three or more credits. And it also points out the inability to distinguish between default correlation of changing default probability.
Download paper (722K PDF) 12 pages - Posted with the author's consent.