The central theme of the thesis is the construction of multivariate probability distributions in large dimensions. Special focus is put on the study of multivariate exponential laws. In particular, those exponential families with conditionally independent and identically distributed components are characterized. Subsequently, hierarchical dependence structures are constructed from one-factor building blocks. Besides a generic recipe for such structures, specific examples that are discussed comprise hierarchical Archimedean copulas, multivariate exponential distributions, as well as combinations thereof. Finally, some closely related topics and an application to portfolio credit risk modeling are sketched.
«
The central theme of the thesis is the construction of multivariate probability distributions in large dimensions. Special focus is put on the study of multivariate exponential laws. In particular, those exponential families with conditionally independent and identically distributed components are characterized. Subsequently, hierarchical dependence structures are constructed from one-factor building blocks. Besides a generic recipe for such structures, specific examples that are discussed compr...
»