Table 5.

Outputs of cointegration test results.

Panel A: Pedroni residual-based test for cointegration
Dependent variable	Panel ADF-Statistic	Panel PP-Statistic	Group ADF-Statistic	Group PP-Statistic
$REP$	-2.634∗∗∗ (0.004)	-7.014∗∗∗ (0.000)	-2.115∗∗ (0.017)	-2.844∗∗ (0.002)
Panel B: Westerlund Bootstrapped error correction based cointegration
Dependent variable	$G_{\tau }$	$G_{\alpha }$	$P_{\tau }$	$P_{\alpha }$
$REP$	-2.636∗∗ (0.022)	-7.360 (0.564)	-9.091 ∗∗∗ (0.000)	-14.899 ∗∗∗ (0.000)

∗∗∗ and ∗∗ imply significance at 1% and 5% levels. ADF is Augmented Dickey–Fuller. PP is Phillips-Perron. Gt and Gα are groups mean statistics while Pt and Pα are panel statistics Schwarz information criteria have been used to select the lag length. The probability values are presented in the parenthesis. For the Pedroni test, the bandwidth determination is premised on the Newey-West automatic and Bartlett kernel For the Westerlund [76] test, the number of bootstraps to generate the bootstrapped p values, which are robust is set to 100.