Discussion Papers, Interdisciplinary Research Project 373: Quantification and Simulation of Economic Processes 2002,30
We propose a model of correlated multi-firm default with incomplete information. While public bond investors observe issuers' assets and defaults, we suppose that they are not informed about the threshold asset level at which a firm is liquidated. Bond investors form instead a prior on these thresholds. Stochastic dependence between default events is induced through correlated asset values and correlated default thresholds. The former results from dependence of firms on common macroeconomic factors, while the latter corresponds to direct inter-firm linkages. Having addressed this issuer interdependence, the predictions of our model are consistent with empirically well documented facts, in particular the clustering of defaults. We characterize joint conditional default probabilities as assessed by the imperfectly informed secondary market. The representation of dependence via (conditional) copulas is emphasized. We propose the default time copula as a consistent default correlation measure, which overcomes the limitations of existing covariance based measures. A case study is examined, where issuers' assets follow geometric Brownian motions and bond investors' threshold prior is uniform.
incomplete information correlated defaults default clustering joint default distribution copulas