Asymmetric effects of foreign direct investment and trade openness on economic growth in Somalia: Evidence from a non-linear ARDL approach

Figures & data

Figure 1. Trade openness and GDP growth rate of Somalia from 1990 to 2020. Sources: UNTCAD (2020) and SESRIC (2020).

Table 1. Descriptive information and correlation matrix.

Panel A: Summary of descriptive statistics
	lnGDP	lnK	lnLF	lnTO	lnFDI
Mean	3.067	2.399	0.274	1.843	0.294
Median	3.069	2.397	0.273	1.895	0.106
Maximum	3.206	2.519	0.460	2.125	1.471
Minimum	2.937	2.287	0.130	1.070	−0.732
Standard deviation	0.085	0.073	0.110	0.264	0.842
Skewness	−0.002	0.125	0.138	−1.428	0.116
Kurtosis	1.858	1.722	1.690	4.534	1.315
Jarque-Bera	1.684	2.190	2.315	13.568	3.737
Probability	0.431	0.335	0.314	0.001	0.154
Observations	31	31	31	31	31
Panel B: Correlation matrix
lnGDP	1
lnK	0.906	1
lnLF	0.851	0.574	1
lnTO	0.413	0.032	0.797	1
lnFDI	0.885	0.661	0.930	0.701	1

Table 2. Unit root tests.

ADF Test
	Level		At first difference
Variables	Constant	Constant with trend	Constant	Constant with trend
lnGDP	−0.0013	−18.589***	−4.469***	−5.243***
lnK	1.571	−1.961	−4.887***	−3.914**
lnLF	−0.876	−8.563***	−4.288***	−3.012
lnTO	−2.180	−1.520	−4.947***	−4.124**
lnFDI	−0.139	−3.056	−6.608***	−6.442***
PP Test
	Level		At first difference
Variables	Constant	Constant with trend	Constant	Constant with trend
lnGDP	−0.507	−4.390***	−4.610***	−5.385***
lnK	−1.593	−6.110***	−4.976***	−6.043***
lnLF	1.655	−5.358***	−2.277	−1.173
lnTO	−4.313***	−0.479	−4.936***	−6.018***
lnFDI	−0.067	−3.059	−6.589***	−6.423***

Note: ** and *** symbolize significance at 5% and 1%, respectively.

Table 3. Nonlinearity of BDS test.

	lnGDP		lnGFCF		lnLF		lnTO		lnFDI
Dimension	BDS Stat	z-Stat	BDS Stat	z-Stat	BDS Stat	z-Stat	BDS Stat	z-Stat	BDS Stat	z-Stat
2	−0.013	−0.868	−0.003	−0.273	0.197	9.111	0.127	5.672	0.019	0.846
3	0.07	2.363	−0.003	−0.137	0.342	9.369	0.202	5.458	0.053	1.42
4	0.151	3.344	−0.005	−0.159	0.445	9.654	0.236	5.201	0.076	1.643
5	0.227	3.795	0.052	1.126	0.515	10.132	0.243	4.945	0.085	1.712
6	0.295	3.999	0.106	1.798	0.561	10.766	0.228	4.649	0.075	1.503

Table 4. F-bounds test.

Model specification	Lower bound	Upper bound	F-statistics	Conclusion
Linear	2.62	3.9	3.303	Inconclusive
Non-linear	2.04	3.24	7.306	Cointegration

Note: the critical values, at the 5% significance level, are based on Narayan (2005).

Table 5. Long-run coefficient estimates.

Variable	Coefficient
lnK	1.309***
	(47.289)
lnLF	−1.168**
	(-3.008)
lnFDI^+	0.103***
	(3.713)
lnFDI^—	0.172***
	(2.726)
lnTO^+	0.084
	(1.072)
lnTO^—	−0.380*
	(-1.828)

Note: *, **, *** signifies significance at 10%, 5% and 1%, respectively. Numbers in parenthesis indicate the t-statistic.

Table 6. Short-run dynamic effect and error correction model.

Variable	Coefficient
ΔlnGDPt-1	−0.015
	(-0.170)
ΔlnLF	2.713*
	(2.098)
ΔlnLFt-1	−2.291
	(-1.663)
ΔlnK	0.631***
	(9.500)
ΔlnFDI^+	0.036**
	(2.656)
ΔlnFDI^+t-1	0.006
	(0.676)
ΔlnFDI^—	−0.009
	(-0.292)
ΔlnFDI^–t-1	0.074*
	(1.911)
ΔlnTO^+	−0.082**
	(-2.780)
ΔlnTO^–	0.078
	(0.872)
ECT(-1)	−0.905***
	(-3.330)
R^2	0.929

Note: *, **, *** present significance at 10%, 5% and 1%, respectively. Numbers in parenthesis signify t-statistic.

Figure 2. Model stability CUSUM test and CUSUM square test.

Table 7. Diagnostic tests.

Type	Statistic
Serial Correlation – LM test	1.384 [0.320]
Heteroscedasticity – BPG test	1.424 [0.329]
Normality – JB test	0.106 [0.947]
Ramsey RESET test	0.938 [0.379]
WLR – FDI	6.110 [0.0294]
WSR – FDI	1.214 [0.3349]
WLR – TO	17.491 [0.0013]
WSR – TO	2.588 [0.1260]

Note: values in the [.] are the p-values. W_LR and W_SR are the long-run and short-run asymmetries of FDI and TO, respectively.

Figure 3. Dynamic multiplier effect graphs.

Table 8. Johansen cointegration test.

No. of CE(s)	Eigenvalue	Statistic	Critical Value	Prob.
Trace test
None *	0.8263	117.3888	69.8189	0.0000
At most 1 *	0.7371	68.3832	47.8561	0.0002
At most 2 *	0.4527	30.9748	29.7971	0.0364
At most 3	0.3312	14.0999	15.4947	0.0802
At most 4	0.0964	2.8379	3.8415	0.0921
Max-Eigen
None *	0.8263	49.0056	33.8769	0.0004
At most 1 *	0.7371	37.4084	27.5843	0.0020
At most 2	0.4527	16.8749	21.1316	0.1780
At most 3	0.3312	11.2620	14.2646	0.1416
At most 4	0.0964	2.8379	3.8415	0.0921

Note: * denotes rejection of the null hypothesis of no cointegration at the 5% significance level.

Table 9. VECM causality tests.

	Short-run causality					Long-run causality
	ΔlnGDP	ΔlnLF	ΔlnK	ΔlnFDI	ΔlnTO	ECTt-1
ΔlnGDP		3.455*	0.568*	0.009	−0.085*	0.286
ΔlnLF	0.014		−0.018	−0.0004	0.006	−0.010
ΔlnK	−1.166**	6.297***		0.003	−0.120**	0.075
ΔlnFDI	−14.068***	−89.830***	11.990***		−1.265***	12.014***
ΔlnTO	0.153	−10.688	−0.085	0.045		−0.235

Note: *, **, *** signify significance at 10%, 5%, and 1%, respectively.

Narayan, P. K. (2005). The saving and investment nexus for China: Evidence from cointegration tests. Applied Economics, 37(17), 1979–1990. https://doi.org/10.1080/00036840500278103

 Web of Science ®Google Scholar

Data availability

The datasets used and/or analyzed during the current study are available from the following links: https://data.worldbank.org/indicator/SL.TLF.TOTL.IN?locations=SO; https://www.sesric.org/query.php; and https://unctadstat.unctad.org/wds/TableViewer/dimView.aspx.