PANIC in the Presence of Uncertainty about the Deterministic Trend*
Author | Joakim Westerlund,Johan Blomquist |
Date | 01 February 2013 |
Published date | 01 February 2013 |
DOI | http://doi.org/10.1111/obes.12008 |
123
©Blackwell Publishing Ltd and the Department of Economics, University of Oxford 2012. Published by Blackwell Publishing Ltd,
9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 75, 1 (2013) 0305-9049
doi: 10.1111/obes.12008
PANIC in the Presence of Uncertainty about the
Deterministic TrendÅ
Joakim Westerlund† and Johan Blomquist‡
†Financial Econometrics Group, School of Accounting, Economics, and Finance, 70 Elgar Road,
Burwood Highway, VIC 3125, Deakin University, Melbourne, Australia
(e-mail: j.westerlund@deakin.edu.au)
‡AgriFood Economics Centre, Box 730, 220 07 Lund, Sweden (e-mail: johan.blomquist@slu.se)
Abstract
Most macroeconomic and financial panel variables are trending. However, because of the
well-known power problem in the presence of incidental trends, many researchers gamble
that their unit root test regressions can be ran without such trends, thereby running the risk
of obtaining spurious results. This article takes one of the most general and popular panel
unit root tests, known as PANIC, and shows how it can be modified to account for the
uncertainty regarding the deterministic trend.
I. Introduction
In this article, we consider the problem of testing for a unit root in the panel data variable
Xi,t, which is observable for t=1, ...,Ttime series and i=1, ...,Ncross-section units.
While reasonably straightforward in its absence, the presence of cross-section dependence
has been shown to add significantly to the complexity of the testing problem. As a response
to this, several panel unit root tests that are robust to such dependence have recently been
proposed (see Breitung and Pesaran, 2008). The starting point of this paper is the panel
analysis of non-stationary idiosyncratic and common components (PANIC) method of Bai
and Ng (2004), in which Xi,tis assumed to be composed of two components, one that is
common and one that is idiosyncratic. The basic idea is to infer the order of integration of
Xi,tby testing for unit roots in each component separately.
The reason for focusing on PANICis that it has a number of distinct features that makes
it very attractive, especially from an applied point of view. The main advantage is that in
contrast to most, if not all, other tests around, in PANIC the common and idiosyncratic
components are not restricted to have the same order of integration. Another advantage
ÅPrevious versions of this article were presented at the 5th Nordic Econometric Meeting in Lund and at a seminar
at Lund University. The authors would like to thank seminar participants and in particularAnindya Banerjee, David
Edgerton, Randi Hjalmarsson, Matthew Lindquist, Peter Lindstr¨om, Hashem Pesaran and two anonymous referees
for may valuable comments and suggestions. The authors gratefully acknowledge financial support from the Jan
Wallander and Tom Hedelius Foundation, research grant numbers W2006–0068:1 and P2009–0189:1.
JEL Classification numbers: C32, C33.
To continue reading
Request your trial