Title: Bootstrapping the log-periodogram estimator of the long-memory parameter: is it worth weighting?

Authors: Saeed M. Heravi; Kerry D. Patterson

Addresses: Cardiff Business School, University of Cardiff, Cardiff CF10 3EU, UK; Department of Economics, University of Reading, Reading RG6 6AA, UK ' Cardiff Business School, University of Cardiff, Cardiff CF10 3EU, UK; Department of Economics, University of Reading, Reading RG6 6AA, UK

Abstract: Estimation of the long-memory parameter from the log-periodogram (LP) regression, due to Geweke and Porter-Hudak (GPH), is a simple and frequently used method of semi-parametric estimation. However, the simple LP estimator suffers from a finite sample bias that increases with the dependency in the short-run component of the spectral density. More recently, Arteche and Orbe (2009a, 2009b), in the context of the GPH estimator, suggested a promising bootstrap method, based on the frequency domain, to obtain the RMSE value of the frequency bandwidth, m, that avoids estimating the unknown parameter. We extend this bootstrap method to the AG and WLP estimators and to consideration of bootstrapping in the frequency domain (FD) and the time domain (TD) and, in each case, to 'blind' and 'local' versions. We undertake a comparative simulation analysis of these methods for relative performance on the dimensions of bias, RMSE, confidence interval width and fidelity.

Keywords: long memory; bootstrap; log-periodogram regression; variance inflation; weighted log-periodogram regression; time domain; frequency domain.

DOI: 10.1504/IJCEE.2021.116384

International Journal of Computational Economics and Econometrics, 2021 Vol.11 No.3, pp.201 - 221

Received: 05 Nov 2018
Accepted: 16 Nov 2019
Published online: 22 Jul 2021 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article