A0255
Title: Optimal choice of bootstrap block length for periodically correlated time series
Authors: Anna Dudek - AGH University of Krakow (Poland) [presenting]
Abstract: The focus is on the problem of choosing the optimal block length for block bootstrap methods designed for periodically correlated processes, such as the Generalized Seasonal Block Bootstrap, the Extension of Moving Block Bootstrap, and the Generalized Seasonal Tapered Block Bootstrap. We consider two estimation problems: the overall mean and the seasonal means. The obtained optimal block lengths are of the same order as for the corresponding approaches in the stationary case. They are obtained directly by minimizing the mean squared error of the corresponding bootstrap variance estimator or by exploiting the relationship between bootstrap and jackknife variance estimators. Finally, we present the results of the performed simulation study.
