New block bootstrap methods: Sufficient and/or ordered

Beyaztas B. H., Firuzan E., Beyaztas U.

COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION, vol.46, no.5, pp.3942-3951, 2017 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 46 Issue: 5
  • Publication Date: 2017
  • Doi Number: 10.1080/03610918.2015.1066808
  • Journal Name: COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.3942-3951
  • Keywords: Block bootstrap, Sufficient bootstrap, Time series, 37M10, 60G25, 62F40, REGRESSION-MODELS, JACKKNIFE
  • Marmara University Affiliated: No

Abstract

In this study, we propose sufficient time series bootstrap methods that achieve better results than conventional non-overlapping block bootstrap, but with less computing time and lower standard errors of estimation. Also, we propose using a new technique using ordered bootstrapped blocks, to better preserve the dependency structure of the original data. The performance of the proposed methods are compared in a simulation study for MA(2) and AR(2) processes and in an example. The results show that our methods are good competitors that often exhibit improved performance over the conventional block methods.