Consistent Assumptions for Modeling Credit Loss Correlations

We consider a single period portfolio of n dependent credit risks that are subject to default during the period. We show that using stochastic loss given default random variables in conjunction with default correlations can give rise to an inconsistent set of assumptions for estimating the variance of the portfolio loss. Two sets of consistent assumptions are provided, which it turns out, also provide bounds on the variance of the portfolio's loss. An example of an inconsistent set of assumptions is given