Modelling default contagion using multivariate phase-type distributions (Q539143)

In this paper the author does two things. On one hand he proposes a contagion model to model the distribution of default times of an arbitrary collection of obligors. On the other he calibrates the model for a given collection of ten obligors. The contagion model is in the class of multivariate phase-type, in which the individual default intensities are coupled by supposing that the default intensity of any given individual may increase (or decrease) each time other individual defaults. Using this model, a Markov chain is constructed such that default times correspond to hitting times of a special class of sets. With this joint and conditional distributions can be computed in terms of the transition matrix of the chain. Once that is achieved, the author examines the calibration issue in a specific situation, two portfolios with ten obligors in two economic sectors.