A Geometric Interpretation of Default Correlation in CDO Tranches

We develop a geometric framework for understanding default correlation in collateralized debt obligations (CDOs). Under the standard one-factor Gaussian copula model, the latent Gaussian dependence structure of N obligors has ellipsoidal covariance contours defined by the correlation matrix. As the pairwise correlation rho increases from 0 toward 1, the ellipsoid deforms from a sphere to a line segment along the diagonal of R^N. We partition R^N into 2^N default zones and illustrate how probability mass is redistributed toward extreme zones (all-default and no-default) as correlation increases. This provides geometric intuition for why, in standard homogeneous or endpoint-comparison settings, equity tranches exhibit long-correlation behavior while senior tranches exhibit short-correlation behavior, whereas total expected loss is invariant. We extend the visualization to three dimensions and derive a polynomial endpoint-balance condition characterizing when the expected loss of an elementary one-default-thick mezzanine tranche coincides at the two extreme correlation regimes. The geometric approach complements algebraic treatments and is particularly relevant for small-N portfolios where the zone structure remains computationally tractable.