Bootstrap-Assisted Unit Root Testing with Piecewise Locally Stationary Errors

• Activity Number: 242
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Yeonwoo Rho* and Xiaofeng Shao
• Companies: Michigan Technological University and University of Illinois at Urbana-Champaign
• Keywords: Heteroscedasticity ; Locally stationary ; Unit root testing ; Wild bootstrap

Abstract

To deal with nonstationary errors in unit root testing, we adopt the piecewise locally stationary process, which allows for both smooth and abrupt changes in second or higher order properties. Under this framework, the limiting null distributions of the conventional unit root test statistics contain a number of unknown parameters that are difficult to estimate. We propose to use the dependent wild bootstrap to approximate the non-pivotal limiting null distributions, and the bootstrap consistency is rigorously justied. Through nite sample simulations, we demonstrate the size accuracy of our procedure as compared to the block bootstrap-based counterpart.