CMStatistics 2015: Start Registration
View Submission - CMStatistics
B1324
Title: Fast and robust bootstrap of robust estimators in SUR models Authors:  Kris Peremans - University of Leuven (Belgium) [presenting]
Stefan Van Aelst - University of Leuven (Belgium)
Abstract: Robust estimators of the seemingly unrelated regression model are considered. First, S-estimators are studied which can attain a high breakdown value, but their normal efficiency can be quite low. Therefore, MM-estimators are introduced to obtain estimators that have both a high breakdown value and a high normal efficiency. Furthermore, the problem of statistical inference is studied. Classical inference relies on assumptions which are hard to verify. Moreover, classical inference may be non-robust. Therefore, a fast and robust bootstrap procedure is developed. Bias corrected and accelerated confidence intervals of the estimated parameters are constructed and their performance is analyzed in simulation studies. The robust estimators and the fast and robust bootstrap procedure are illustrated on some examples.