From extended regular variation to regular variation with application in extreme value statis\-tics (Q1018349)

This paper offers a unified treatment of the asymptotic properties for most estimators of extreme value characteristics to be found in the literature. A relation between regular variation and extended regular variation of second order is established. The study is organized in five sections, the first two preparing the background necessary for the proof of the main theorem, in section three. The results from the fourth section allow to foresee useful ramifications of the main theorem, and in the last section an example of applications is presented. The maximum likelihood estimator is obtained as a shift-invariant estimator.