CMStatistics 2021: Start Registration
View Submission - CMStatistics
Title: Extreme expectile estimation from short-tailed data Authors:  Boutheina NEMOUCHI - ENSAI (école nationale de la statistique et de l\'analyse de l\'information, France) (France)
Abdelaati Daouia - Toulouse School of Economics (France)
Gilles Stupfler - ENSAI - CREST (France)
Boutheina Nemouchi - ENSAI (France) [presenting]
Abstract: Expectile estimation has been recently investigated extensively from the perspective of extreme value theory. Attention has been, however, restricted to heavy-tailed distributions. We focus on the less-discussed problem of estimating extreme expectiles from short-tailed data. We present extrapolated expectile estimators based on extreme quantiles and develop their asymptotic theory. Then, we provide Monte Carlo evidence of their superiority relative to competing estimators extrapolating intermediate asymmetric least squares expectiles. Finally, we discuss concrete examples of application.