Publication Type

Journal Article

Version

acceptedVersion

Publication Date

1-2010

Abstract

This article shows that when applied to nonstationary time series, the conventional Regression Error Specification Test (RESET) leads to severe size distortion and its asymptotic distribution involves a mixture of noncentral chi(2) distributions. Nonstationarity introduces bias terms in the limit distribution, and appropriate corrections for the bias are presented leading to a modified RESET test that has a central chi(2) limit distribution. In simulations, this modified test is shown to have power not only against nonlinear cointegration but also against the absence of cointegration. In an empirical illustration, the linear purchasing power parity (PPP) specification is tested using five Organization for Economic Cooperation and Development (OECD) countries.

Keywords

Noncentral chi(2) distribution, Nonlinear cointegration, RESET test, Specification test

Discipline

Econometrics

Research Areas

Econometrics

Publication

Journal of Business and Economic Statistics

Volume

28

Issue

1

First Page

96

Last Page

114

ISSN

0735-0015

Identifier

10.1198/jbes.2009.07182

Publisher

Taylor & Francis: SSH Journals

Copyright Owner and License

Authors

Additional URL

https://doi.org/10.1198/jbes.2009.07182

Included in

Econometrics Commons

Share

COinS