On robust tail index estimation under random censorship
Author(s) -
A. Sayah,
Djabrane Yahia,
Brahim Brahimi
Publication year - 2014
Publication title -
afrika statistika
Language(s) - French
Resource type - Journals
ISSN - 2316-090X
DOI - 10.16929/as/2014.671.61
Subject(s) - censorship , index (typography) , estimation , statistics , mathematics , econometrics , computer science , economics , political science , law , world wide web , management
In this paper, we propose a new robust tail index estimation procedure for Pareto-type distributions in the framework of randomly censored samples, based on the ideas of Kaplan-Meier integration using the huberized M-estimator of the tail index. We derive their asymptotic results. We illustrate the performance and the robustness of this estimator for small and large sample size in a simulation study. Dans cet article, nous proposons une nouvelle procedure de l'estimation robuste de l'indice de la queue pour les distributions de type Pareto dans le cas d'echantillons censures, sur la base des idees de l'intgrale de Kaplan-Meier en utilisant le huberized M-estimateur de l'indice de la queue. Nous derivons leurs resultats asymptotiques. Nous illustrons dans l'etude de la simulation la performance et la robustesse de cet estimateur pour un echantillon de petite et grande taille. Key words : Heavy-tailed distributions; Hill estimator; Random censorship; Regular variation; Robust estimation; Tail index.
Accelerating Research
Robert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom
Address
John Eccles HouseRobert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom