Date of Award

1-1-2014

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

College/School/Department

Department of Economics

Content Description

1 online resource (xiv, 146 pages) : illustrations (some color)

Dissertation/Thesis Chair

Kajal Lahiri

Committee Members

Huaming Peng, Zhongwen Liang

Keywords

initial values, local power, panel data, unit root tests, Panel analysis, Econometrics, Weibull distribution, Gaussian distribution

Subject Categories

Economics

Abstract

This dissertation investigates and compares the performance of the panel unit root tests in terms of local power. The tests considered include the Gaussian power envelope, the pooled t-tests based on the Dickey-Fuller (DF) least squares demeaning and on the initial value demeaning, the IPS test proposed in Im, Pesaran, and Shin (2003) , and the Fisher-P type tests. Cases where the panel data have no fixed effects and where the panel data contain fixed effects are studied. The model allows for heterogeneity not only in the auto-regressive coefficients and the error terms but also in the fixed effects. The study is conducted firstly assuming zero initial values. Then the impact of non-zero, random initial values on the local power is investigated. The results show that the assumption concerning the initial values is extremely important and the impact of the non-zero random initial values on the local power of the panel unit root tests is very different when individual fixed effects are present in the model. This is demonstrated both in asymptotic theory and in finite-sample simulations.

Included in

Economics Commons

Share

COinS