Concentration Test: Johansen's Fisher Panel Cointegration Test

According to the cointegration analysis established in the mid-80s, the non-stationarity of the underlying time series does not always imply the non-stationarity of the linear combinations of these series. Hence, it is a prerequisite to ensure that all variables are integrated of order one in levels before employing the panel cointegration tests.
Table 3 presents the results performed by Im, Pesaran and Shin unit root test of all variables for the four panels of economies tested in both levels and in first difference. Panel A, B, C and D represents all income economies, high-income economies, upper-middle-income economies and lower-middle-income economies respectively.It can be inferred from Table 3 that for Panel A and B, real gross domestic
The Johansen procedure detect the number of cointegrating variable by two likelihood tests namely the trace test and maximum eigenvalue test. This paper adopted the Johansen’s Fisher panel cointegration test because it is developed from Johansen’s time-series cointegration test which allows the use of a mixture of I(1) and I(0) variables in the test (Johansen,1995). Thus, this could imply that when conducting the panel cointegration test, using a set of panel data variables which consists of different orders of integration, would not produce biased results.Table 4 reports both the trace test and maximum eigenvalue test statistics of the Johansen’s Fisher panel cointegration analysis for the four panels of different economies. Except for the Fisher trace test and maximum eigenvalue test statistics of Panel D which does not reject the null hypothesis of no cointegration at the 1% significance level, other statistics in Panel A, B and C significantly confirm there is one variable cointegrating with GDP in the long
Yet, real gross fixed capital formation and labor force participation rate have relatively larger effect on the countries’ economic growth instead of nuclear energy consumption.Table 6 reports the result of Granger causality test for each panel data set. From Panel A, which includes all countries that utilized nuclear energy, real GDP have a positive and significant effect to real GFCF in short-run; long-run unidirectional relationship from nuclear energy consumption to real GDP and real GFCF. For Panel B, which is high-income economy, it appears that there is a unidirectional causality from nuclear energy consumption to real GDP in short-run and a long-run unidirectional causality between nuclear energy consumption and real GDP, real GFCF and labor force participation rate. Besides, from Panel C, which is the upper-middle countries, shows that real GDP, real GFCF and labor force participation rate have unidirectional causality effect to nuclear energy consumption in short-run. For Panel D, which is the lower-middle income countries, Granger causality test shows that there is a bi-directional causality between real GDP and real GFCF in short-run; while labor force participation rate have a long-run unidirectional causality effect to real GDP and real GFCF. However, nuclear energy consumption does not have significant effect to GDP in the