Date Added: May 2011
The authors consider a structural model for the estimation of credit risk based on Merton's original model. By using Random Matrix Theory they demonstrate analytically that the presence of correlations severely limits the effect of diversification in a credit portfolio if the correlations are not identically zero. The existence of correlations alters the tails of the loss distribution tremendously, even if their average is zero. Under the assumption of randomly fluctuating correlations, a lower bound for the estimation of the loss distribution is provided.