A0862
Title: Bayesian analysis of peaks over threshold
Authors: Stefano Rizzelli - University of Padova (Italy) [presenting]
Simone Padoan - Bocconi University (Italy)
Clement Dombry - Universite de Franche Comte (France)
Abstract: The peaks over threshold (POT) method is one of the most popular approaches to statistical analysis of univariate extremes. Even if Bayesian methods in this context have found increasing use in applications, no asymptotic guarantee has been established. Contraction rates and asymptotic normality of posterior distributions for Bayesian inference are established using POTs, providing contraction rates for predictive distributions and also allowing (random) covariates to be included in the analysis. Simulations show that the properties established in the large sample theoretical study can be observed in practice for finite samples of moderate size.
