Table 2. Nonlinear ARDL estimation...

Table 2.

Nonlinear ARDL estimation results for asymmetric effect of CEPU on trade from China

Variable name	Nigeria	S. Africa	Kenya	Ghana	Angola	Tanzania
Panel A – estimated long-run coefficients
$ρ$	−0.32***	−0.21***	−0.55***	−0.15***	−0.74***	−0.38***
$β^{+}$ (CEPU)	0.44	−0.18	0.01	0.01	−0.56	0.38***
$β^{-}$ (CEPU)	−0.74*	0.15	0.47***	0.55	−0.78***	0.35*
Exchange rate	3.90**	3.08***	1.57***	0.31	−0.52***	1.85
Gross capital formation	6.52***	1.95	−0.82	1.24*	1.17***	−1.08
Labour force participation	101.53***	17.53	21.38***	−37.27*	6.14	14.49*
Real GDP	−6.89	−12.76*	0.17	−3.77	−7.19***	−7.21
Panel B – estimated short-run coefficients
$\sum_{j = 0}^{q} π_{j}^{+}$ (CEPU)	0.00	0.20**	0.20**	0.00	−2.68***	0.16**
$\sum_{j = 0}^{q} π_{j}^{-}$ (CEPU)	0.00	0.00	0.15*	0.00	0.40**	0.00
$\sum_{j = 0}^{q} δ$ (exchange rate)	−4.77**	−0.16*	1.53***	−0.85***	−0.40**	−1.93***
$\sum_{j = 0}^{q} τ$ (gross capital formation)	1.97***	1.54***	0.93***	0.00	−1.73***	−0.97***
$\sum_{j = 0}^{q} ψ$ (labour force participation)	110.17***	−7.60**	−34.84**	−94.88**	−50.46**	0.00
$\sum_{j = 0}^{q} ω$ (real GDP)	−3.78**	0.00	0.00	−5.38***	0.76**	0.00
Panel C – symmetry tests
$H_{0} : β^{+} = β^{-}$	8.01***	2.08	39.98***	1.23	12.52***	0.03
$H_{0} : \sum_{j = 0}^{q} π_{j}^{+} = \sum_{j = 0}^{q} π_{j}^{-}$	0.00	4.74**	0.19	0.00	26.89***	4.25**
Panel D – diagnostic tests
F_PSS	5.60***	3.80***	8.17***	2.67**	10.54***	7.39***
t_BDM	−4.60***	−3.43**	−6.83***	−3.00*	−8.35***	−3.58**
BG test (NR²) (serial correlation)	10.28***	0.10	7.13**	0.38	2.40	0.07
BGP test (heteroscedasticity)	36.32***	26.74***	13.43	9.74	34.50*	10.39
${\bar{R}}^{2}$	0.60	0.44	0.77	0.54	0.73	0.75

Variable name	Nigeria	S. Africa	Kenya	Ghana	Angola	Tanzania
Panel A – estimated long-run coefficients
$ρ$	−0.32***	−0.21***	−0.55***	−0.15***	−0.74***	−0.38***
$β^{+}$ (CEPU)	0.44	−0.18	0.01	0.01	−0.56	0.38***
$β^{-}$ (CEPU)	−0.74*	0.15	0.47***	0.55	−0.78***	0.35*
Exchange rate	3.90**	3.08***	1.57***	0.31	−0.52***	1.85
Gross capital formation	6.52***	1.95	−0.82	1.24*	1.17***	−1.08
Labour force participation	101.53***	17.53	21.38***	−37.27*	6.14	14.49*
Real GDP	−6.89	−12.76*	0.17	−3.77	−7.19***	−7.21
Panel B – estimated short-run coefficients
$\sum_{j = 0}^{q} π_{j}^{+}$ (CEPU)	0.00	0.20**	0.20**	0.00	−2.68***	0.16**
$\sum_{j = 0}^{q} π_{j}^{-}$ (CEPU)	0.00	0.00	0.15*	0.00	0.40**	0.00
$\sum_{j = 0}^{q} δ$ (exchange rate)	−4.77**	−0.16*	1.53***	−0.85***	−0.40**	−1.93***
$\sum_{j = 0}^{q} τ$ (gross capital formation)	1.97***	1.54***	0.93***	0.00	−1.73***	−0.97***
$\sum_{j = 0}^{q} ψ$ (labour force participation)	110.17***	−7.60**	−34.84**	−94.88**	−50.46**	0.00
$\sum_{j = 0}^{q} ω$ (real GDP)	−3.78**	0.00	0.00	−5.38***	0.76**	0.00
Panel C – symmetry tests
$H_{0} : β^{+} = β^{-}$	8.01***	2.08	39.98***	1.23	12.52***	0.03
$H_{0} : \sum_{j = 0}^{q} π_{j}^{+} = \sum_{j = 0}^{q} π_{j}^{-}$	0.00	4.74**	0.19	0.00	26.89***	4.25**
Panel D – diagnostic tests
F_PSS	5.60***	3.80***	8.17***	2.67**	10.54***	7.39***
t_BDM	−4.60***	−3.43**	−6.83***	−3.00*	−8.35***	−3.58**
BG test (NR²) (serial correlation)	10.28***	0.10	7.13**	0.38	2.40	0.07
BGP test (heteroscedasticity)	36.32***	26.74***	13.43	9.74	34.50*	10.39
${\bar{R}}^{2}$	0.60	0.44	0.77	0.54	0.73	0.75

Note(s):

*, **and *** signify significance at the 10, 5 and 1% level, respectively. The symmetry tests results are derived from standard Wald tests. The BGP test stands for the Breusch–Pagan–Godfrey heteroscedasticity test, whereas the BG test stands for Breusch–Godfrey serial correlation test. F_PSS stands for the test statistic derived from the bounds testing approach established by Pesaran et al. (2001), while t_BDM denotes the test statistic that follows the cointegration test developed by Banerjee et al. (1998). The critical values associated with the t_BDM statistic are as follows: −2.91 at the 10 level, −3.22 at the 5 level and −3.82 at the 1% level

Source(s): Authors’ own work

[ViewLarge]

Sharing Unavailable