Miguel de Carvalho (Pontificia Universidad Católica de Chile)

In this talk I will discuss a few ‘subtleties’ on statistical modeling of risk, with a special emphasis on methods for statistics of multivariate extremes. As- sessment of the risk of such events involves the estimation of small probabilities and hence entails extrapolation into the tails of multivariate distributions, often beyond any existing data. The mathematical basis for such extrapolation is the statistics of extremes, which is an active domain of current research. I will start by discussing basic concepts and common fallacies on statistical modeling of risk. The main goal of the talk is on discussing methods based on families of spectral measures of multivariate extreme value distributions, which provide a natural way to model nonstationary dependence structures.

Department of Mathematics

Pontifical Catholic University of Chile (PUC-Chile)

Av. Vicuña Mackenna 4860, Macul,

Santiago – Chile

(+56 2) 2354 5779

Center for Mathematical Modeling (CMM)

Faculty of Physical and Mathematical Sciences (FCFM)

Universidad de Chile

Beauchef 851, Edificio Norte, Piso 7,

Santiago – Chile