A0639
Title: Extreme value copulas based on Freund's multivariate lifetime model
Authors: Sandor Guzmics - University of Vienna (Austria) [presenting]
Abstract: The exponential distribution and its multivariate generalizations are widely used in lifetime modeling. There are numerous models that explicitly incorporate a dependence structure among the components, Freund's bivariate distribution is such one. Its copula has been presented previously. We also provided previously the corresponding bivariate extreme value copula and discussed a natural multivariate generalization of the model. The basic idea is that the remaining lifetime of any entity in a multivariate system is shortened when one of the other entities defaults. We investigate the dependence structure in Freund's multivariate lifetime model, assuming a symmetric parameter setting, i.e., when the initial lifetime intensities, as well as the shock parameters, are all the same. We present some remarkable properties of this multivariate, parametric copula family, and examine the corresponding extreme value copulas.
