The lower tail dependence copula for bi-extremal copula

Sala 6.4.30, FCUL, Lisboa

Por Marco Aurélio Sanfins (DE-IME, Universidade Federal Fluminense).

Research on extreme events are mainly devoted to investigate the behavior of the higher order statistic. In this work we advocate that the second highest statistic also contain substantial information, when combined with the previous one. We merge these two statistics through a new copula model, denoted here as Bi-Extremal Copula, or BIX. Through algebraic development, it is possible to show that the Lower Tail dependence copula LTDC of BIX copula reduces to the Copula product under some mild conditions. Under this assumption, we propose its use as an Independence test of random variable. At the end, it is provided theoretical and simulated results which profs its benefits for i.i.d. continuous random variables.

CEAUL - Centro de Estatística e Aplicações da Universidade de Lisboa