The classical multivariate Pareto model, which was referred to by Arnold [Arnold, B.C., 1983. Pareto Distributions. International Co-operative Publishing House], and is used to fit heavy tailed random variables, has serious disadvantages. First, each of its marginals has the same distribution up to location and scale parameters. Secondly, this model has a rigid dependence structure. Furthermore, the independent Pareto marginals do not belong to this model. In this paper, we introduce two multivariate models, whose marginals have different shape parameters and a more flexible dependence structure. Moreover, the independent Pareto marginals model is a special case of one of the suggested models. We also discuss regression and a measure of dependence for these models, along with some relevant inferences. The paper concludes with a numerical study.
Bibliographical noteFunding Information:
The authors wish to thank the Israel Zimmerman Foundation for financial support.
ASJC Scopus subject areas
- Statistics and Probability
- Statistics, Probability and Uncertainty