ECONOMICS BULLETIN, vol.41, pp.1813-1825, 2021 (ESCI)
The purpose of this study is to propose a new residual-based unit root test and then apply it to examine the stationarity
of gross domestic product (GDP) for EU membership countries. For this purpose, the CIPS test proposed by Pesaran
(2007) has been extended to a structure that takes into account the knowledge of the non-normally distributed
residuals. For this, the residual augmented least squares (RALS) estimators proposed by Im and Schmidt (2008) were
included in the CIPS test. The second and third moments of the error terms are added to the cross-sectionally
augmented ADF (CADF) regression that constitutes the CIPS test process. When calibrated under the behavior of the
residues non-normally distributed residuals during the data generation process, it is seen that the panel unit root test
specific to the series in which the residuals are not normally distributed has higher power and more appropriate size
than CIPS test. According to the results of empirical analysis, it was concluded that the CIPS test was stationary only
at the 10% level, while according to the RALS-CIPS test it was concluded that it was stationary at the 1% significance
level. It can be interpreted that the RALS-CIPS test is stronger because it used additional information consisting of
residual moments. The test offers a simple way to have good size and power properties for non-normal errors.