The asymmetric effect of financial development on energy consumption in sub-Saharan Africa

Figures & data

Figure 1. Asymmetry hypothesis between financial development and energy consumption.

Table 1. Summary of variable descriptions, data sources, and expected signs

Variables	Symbol	Measure	A priori sign (±)	Data source
Energy consumption	EC	(Kg of oil equivalent) per capita	N/A	WDI & IEA database
Financial development	FD	A comprehensive financial development index proposed by Svirydzenka (2016)	±	IMF database
Economic growth	GDPK^2	GDP per capita (log) (Constant 2010 US$)	+	WDI database
Energy access	EA	TPES per capita (toe)	+	IEA
Poverty	POV	Poverty headcount ratio	-	Economic survey-WDI
Industrialization	IND	Manufacturing, value added (% of GDP)	±	WDI
Foreign Direct Investment	FDI	a share of GDP in country i at period t	-	FDI markets database
Inflation	INFL	The overall increase in prices/ increase in the cost of living in a country	+	WDI database

Source: Authors

Table 2. Summary of descriptive statistics

Variables	N	Sum	Mean	SD	CV	Min	Max	Kurtosis	Skewness
EC	567	3.759	6.629	5.839	0.88	1.009	31.291	8.882	2.533
FD	567	9.061	15.98	11.12	0.70	1.883	61.781	6.501	1.893
GDPK	567	3.908	6.893	1.117	0.16	4.631	9.288	2.134	0.269
EA	567	20.404	35.99	26.12	0.73	.01	99.435	2.851	0.782
POV	567	24.208	42.69	22.04	0.52	0.287	95.06	2.582	0.0899
IND	567	6.406	11.30	5.770	0.51	2.541	33.346	4.141	0.979
FDI	567	1.732	3.055	5.114	1.67	−8.703	50.00	29.52	4.242
INFL	567	44.181	77.92	1.023	0.013	−55.312	23,773.13	509.9	22.11

Variables are signified as EC—Energy consumption, FD -Financial development, GDPK—Growth domestic product per capita, EA—Energy Access, POV—Poverty IND—Industrialisation, FDI—Foreign Direct Investment, INFL—Inflation. Source: Source: Authors’ computations

Table 3. Results from the Pesaran (2004) cross-sectional Independence

	EC	FD	GDPK	EA	POV	IND	FDI	INFL
CD test value	0.291	−0.450	−5.560	0.149	−0.300	−0.446	−0.655	−0229
P value	0.771	0.652	0.575	0.881	0.765	0.656	0.513	0.819

Notes: Level of significance at the ***1%, **5%, and *10%. If P-Value is <0.05 it is significant. If it is >0.05 it is insignificant. Source: Authors’ computations

Table 4. Unit root tests

VAR	Level		1^st difference
	LLC (adj. t)	IPS(W-stat)	LLC (adj. t)	IPS (W-stat)
EC	0.547(0.292)	1.021(8.464)	−8.040***(0.000)	−8.057***(0.000)
FD	−3.177***(0.000)	−2.053**(0.020)	-	-
GDPK	−0.428(0.334)	0.864(0.806)	−5.088***(0.000)	−6.221***(0.000)
EA	−1.461**(0.072)	−3.374***(0.000)	-	-
POV	2.0393 (0.979)	−1.2710 (0.102)	−8.9567 ***(0.000)	−4.2865 ***(0.000)
IND	−1.349(0.087)	0.384(0.649)	−7.000***(0.000)	−8.262***(0.000)
FDI	−3.926(0.000)	−3.0194(0.000)	-	-
INFL	−5.9060(0.000)	−3.4366(0.000)	-	-

Notes: Level of significance at the ***1%, **5%, and *10%. If P-Value is <0.05 it is significant. If it is >0.05 it is insignificant. Source: Authors’ computations

Table 5. Pedroni cointegration test results

	Statistics	Probability	Weighted Statistics	Probability
Within-dimension
Panel v-statistic	1.855	0.032	−5.237	1.000
Panel p-statistic	0.898	0.815	−1.092	0.138
Panel PP-statistic	−4.032	0.000	−6.253	0.000
Panel ADF-statistic	−5.660	0.000	−6.489	0.000
Between-dimension
Group p-statistic	2.509	0.994
Group PP-statistic	−3.140	0.000
Panel ADF-statistic	−2.904	0.002

Notes: Level of significance at the ***1%, **5%, and *10%. If P-Value is <0.05 it is significant. If it is >0.05 it is insignificant. Between-dimension refers to group mean statistics that provide average results of an individual country, while panel statistics pool the statistics within-dimension. Source: Authors’ computations.

Table 6. Hausman test for long-run homogeneity

H0: PMG estimator is efficient and consistent, but MG is inefficient

Ha: PMG is inconsistent, but MG is consistent and efficient

Regression model without the decomposed variable of financial development: χ2(5) = 10.19 & Prob. > χ2 = 0.1779

Regression model including the decomposed variable of financial development: χ2(5) = 11.80 & Prob. > χ2 = 0.1602

Based on the χ2 test statistics for the two regression models, we cannot reject H0, hence the PMG is more consistent and efficient than the MG estimator.

Notes: Level of significance at the ***1%, **5%, and *10%. If P-Value is <0.05 it is significant. If it is >0.05 then it is insignificant. Source: Authors’ computations

Table 7. Linear effect of financial development on energy consumption in 21 SSA countries

Dependent Variable (EC)	PMG	Std. Error	t-stat	Prob
Part A: Long-run coefficients
FD	0.121***	0.024	5.02	0.000
GDPK	−0.176*	0.095	−1.84	0.065
EA	0.085***	0.008	10.05	0.000
POV	−0.005	0.003	−1.38	0.166
IND	−0.007	0.014	−0.49	0.622
FDI	−0.034***	0.012	−2.88	0.004
INFL	0.014***	0.003	4.07	0.000
Part B: Short-run coefficients
ECT	−0.115***	0.045	−2.54	0.011
FD	0.055	0.040	1.38	0.169
GDPK	0.396	0.244	1.62	0.104
EA	−0.103	0.069	−1.48	0.139
POV	0.111	0.068	1.64	0.102
IND	0.111	0.078	1.42	0.155
FDI	0.008	0.013	0.60	0.546
INFL	−0.003	0.005	−0.49	0.623
CONS.	0.662	0.339	1.95	0.051
No. of countries	21
No. of observations	567
Log-likelihood	240.806

***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively. Or *p-value<0.10; **p-value<0.05; ***p-value<0.01. Source: Authors’ computations.

Table 8. Non-linear effect on financial development & energy consumption in 21 SSA countries

Dependent variable (EC)	PMG	Std. Error	t-stat	Prob
Part A: Long-run coefficients
FD_POS	0.049***	0.015	3.05	0.002
FD_NEG	0.055***	0.014	3.95	0.000
GDPK	0.471***	0.121	3.90	0.000
EA	0.024***	0.004	6.33	0.000
POV	−0.032***	0.006	−5.69	0.000
IND	−0.147***	0.019	−7.53	0.000
FDI	−0.199***	0.023	−8.76	0.000
INFL	0.008***	0.002	3.87	0.000
Asymmetric hypothesis
–γ^−1 | β1 ^=—γ^+2 | β1	3.10 H0			0.0782*
Part B: Short-run coefficients
ECT	−0.109***	0.040	−2.73	0.006
FD_POS	0.071	0.055	1.29	0.197
FD_NEG	0.063	0.044	1.45	0.148
GDPK	0.292	0.194	1.51	0.132
EA	−0.104	0.072	−1.45	0.147
POV	0.076	0.075	1.02	0.309
IND	0.124	0.079	1.57	0.117
FDI	0.015	0.014	1.11	0.268
INFL	0.001	0.005	0.03	0.974
CONS.	1.061	0.569	1.87	0.062
No. of countries	21
No. of observations	567
Log-likelihood	255.522
Asymmetric hypothesis
–γ^−1 | β1 ^=—γ^+2 | β1	0.72 Ha			0.3975

***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively. Or *p-value<0.10; **p-value<0.05; ***p-value<0.01. Financial development = (FD_POS & FD_NEG). H₀ = Null hypothesis is rejected: confirms the presence of asymmetry, H_a = Null hypothesis cannot be rejected: No asymmetry confirmed. Source: Authors’ computations

Svirydzenka, K. (2016). Introducing a new broad-based index of financial development. IMF Working Papers, 16(5), 1. https://doi.org/10.5089/9781513583709.001

 Google Scholar

Pesaran, M. H. (2004). General diagnostic test for cross-section dependence in panels. Working Paper, University of Cambridge & University of Southern California (USC). https://doi.org/10.17863/CAM.5113

 Google Scholar