The hierarchy of Pareto models, described in a paper written by Arnold in 2015, admit representations in terms of independent gamma variables. Utilizing general bivariate beta(2) models introduced in a paper written by Arnold and Ng in 2011, which also involve independent gamma components, it is possible to identify flexible bivariate generalized Pareto models which can exhibit a wide range of correlation values. Multivariate extensions of the models are described. Simplified submodels of these are amenable to standard inference strategies.
keywords
Gamma Components, Beta(2) Distribution, Feller-Pareto Distribution, Gini Index
Author biography
Department of Statistics - University of California - RIVERSIDE, USA (e-mail: barry.arnold@ucr.edu).