B1493
Title: Threshold selection in univariate extreme value analysis
Authors: Laura Fee Schneider - University of Goettingen (Germany) [presenting]
Andrea Krajina - Rabobank (Netherlands)
Tatyana Krivobokova - Georg-August-Universitaet Goettingen (Germany)
Abstract: Threshold selection plays a key role for various aspects of statistical inference of rare events. Most classical approaches tackling this problem for heavy-tailed distributions crucially depend on tuning parameters or critical values to be chosen by the practitioner. To simplify the use of automated data-driven threshold selection methods we introduce two new procedures not requiring the choice of any parameters. The first method measures the deviation of the log-spacings from the exponential distribution and works well for estimating high quantiles. The second approach estimates the asymptotic MSE of the Hill estimator unbiasedly if $\rho=-1$, and we illustrate that the approach still performs well for other values of the second order parameter.
