Can China's trade and investment facilitate renewable energy consumption in Africa?

Abstract

Although several studies have investigated the impact of international trade and foreign direct investment (FDI) on energy consumption in Africa, extensive analyses from a renewable energy perspective are scarce. Moreover, the influence of Chinese FDI and trade on renewable energy consumption is quite inexistent. To fill the gap in the energy economics literature, this study empirically investigates the impact of China's trade and FDI on renewable energy consumption in 40 African countries over the period 2003 to 2016. The panel corrected standard errors estimation method is adopted. Additionally, panel causality tests are used to show the long-run relationship among the variables. The results revealed that both imports and outward direct investment from China had been positively associated with renewable energy consumption. However, exports to China were negatively related to renewable energy consumption. The results further reveal that economic growth and financial market developments are found to be ideal for promoting renewable energy consumption. The panel causality results show that there are significant bidirectional causalities between renewable energy consumption and economic growth. The results suggest that trade and investment policies should be integrated with renewable energy policy in order to achieve sustainable development. Besides, it is also important to adopt sound macroeconomic policies which promote economic growth and financial market development.

Keywords

Renewable energy consumption trade foreign direct investment China Africa

Introduction

De-carbonization of energy sources through promotion of renewable energy consumption is critically important to meet environmentally sustainable development goals (SDGs). Due to globally increased concerns over the issues of climate change, environmental pollutions and global warming, reliance on renewable energy consumption has become an alternative solution.¹ According to the Energy Information Administration projection, renewable energy is expected to be the fastest growing energy source in the coming years due to global environmental challenges and technological advancements. It is estimated that global renewable energy consumption will be expected to account for about 27% of global energy consumption in 2050. It will be more than double between 2020 and 2050, and renewable energy consumption nearly equals liquid fuel consumption by 2050.² However, despite huge potential, the state of renewable energy consumption and transition in Africa is still low and insignificant (see Figure 1). Lack of adequate access to modern energy sources is one of the main obstacles to poverty reduction and socio-economic development. The International Energy Agency report¹ shows that despite some progress, about 600 million people in Africa still have no access to electricity, and close to 780 million people use traditional energy sources for cooking (Figure 2).

Figure 1.

The proportion of people access to clean fuels and technologies for cooking in Africa.

Figure 2.

Share of primary energy consumption by source, world.

In order to achieve the transition to renewable energy, it is important to transfer technologies to developing countries. In this regard, international trade and foreign direct investment (FDI) have become one major source, facilitating the transfer of renewable energy technology to host countries. Several existing studies have indicated that FDI and international trade effectively promote renewable energy developments.^3–5 For example, Doytch and Narayan⁶ showed that FDI induces the consumption of renewable energy consumption and discourages the consumption of non-renewable energy sources. In the same fashion, Ji and Zhang⁵ indicated that financial market development and FDI positively and significantly contribute to renewable energy growth in China. Murshed⁴ also documented that international trade promotes the consumption of renewable energy sources in South Asian economies (Figure 1).

Although several studies in Africa have related international openness through trade and FDI to energy consumption from the general context, extensive analyses which examine the dynamics from renewable energy transitions perspective are still scarce.⁷^,⁸ With this, the impacts of international trade and FDI on renewable and non-renewable energy consumption need to be separated. Hence, in the light of achieving the SDGs of United Nation and Agenda 2063 of African union, this study tried to bridge this gap by empirically investigating the effects of Sino-African trade and investment relations on renewable energy consumption in Africa. The study attempts to address the following questions: Does Sino-African trade and investment relations contribute to renewable energy consumption in Africa? What are the main driving factors for renewable energy transition in Africa? How do economic growth and other macroeconomic factors affect renewable energy consumption on the continent? What are the casual associations between renewable energy consumption and its determinant factors? (Figure 3).

Figure 3.

The trend of China's FDI and trade in Africa.

The stock of Chinese investment in Africa has significantly increased in the last few years, rising from just $490 million in 2003 to $46 billion in 2018, making China the largest developing country investor in Africa. Similarly, China has become the largest trading partner of African countries recently, with total bilateral trade volume reaching about $185 billion in 2018 (from barely less than $2 billion in 1992). China surpassed the US in 2009 to become the largest trading partner of African countries.⁹ In 2014, China became the single largest source of Africa's imports. Trade and investment cooperation has expanded and been particularly stimulated after the establishment of the Forum on China-Africa Cooperation (FOCAC), which was first held in Beijing in 2000.¹⁰ Such strong economic relations have impacts and implications in terms of renewable energy consumption inducement on the continent. As shown above, studies have indicated that international trade and FDI impact energy consumption and host economies’ environmental conditions. Transition to renewable energy is important to comply with binding obligations for reducing CO₂ emissions and achieving sustainable development. In order to move to carbon efficient economies, the critical assessment of renewable energy consumption and its determinant factors are relevant. Hence, it is important for deeper understanding of the determinants of renewable energy consumption in developing countries like Africa. To our knowledge, empirical studies have yet well investigated the link between FDI, trade, economic growth and renewable energy transition in the case of African economies, particularly from Sino-African economic relation perspectives.

This paper contributes to the existing literature in a couple of ways. First, to our knowledge, this study is the first which comprehensively investigates the effects of Sino-African economic relations on renewable energy transition in Africa. The development of renewable energy sources is important to meet the continent's environmental and energy security. Both the SDG of United Nation and Agenda 2063 of African Union puts renewable energy development as main pillar areas for achieving resilient economies. In this case, the study sheds light on the determinants of renewable energy transition in Africa using a large sample of countries. Second, methodologically, the study employed appropriate analytical techniques of panel-corrected standard errors (PCSE) estimation approach with a sound testing framework. This is critical to resolve the cross-sectional dependency of the short panel data nature of the study. The PCSE estimation technique is particularly appropriate for panel data with a short time span and robust to handle cross-sectional dependency, group-wise heteroskedasticity and autocorrelation problems.¹¹ Finally, regression and co-integration results among the panel variables do not show the direction of casual relationship between those variables. To examine the direction of causality, the study adopted the recently developed panel granger causality test approach by Dumitrescu and Hurlin (DH).¹² This is critical to show the direction and feedback relationship between the main variables.

Literature review

FDI can affect renewable energy through direct knowledge transfer.⁶ According to the authors, energy-saving technologies have been found to be more leading by FDI inflows. Based on the underlying argument, the composition, scale and technique effect has been found to explain the relationship between FDI and renewable energy consumption. Moreover, energy efficiency can be improved by foreign companies as they have power to bring new technology through promoting structural change. According to Doytch and Narayan,⁶ FDI can contribute to renewable energy development through promoting marketing know-how. Further, the authors argued that FDI can be pointed out as major factor that can enhance energy efficiency. In the same vein, FDI has been documented to improve environmental quality through promoting renewable energy development.¹³ Furthermore, the rise of CO₂ emissions can be mitigated in the presence of FDI which has the power to positively promote renewable energy development. Similarly Zhang et al.¹⁴ argued that the interaction between FDI and renewable energy may lower CO₂ emissions thereby improving health quality.

Numerous previous studies have investigated how openness to international markets through trade and FDI influences renewable energy consumption. Some of these studies have particularly focus on the analysis of a single country (for instance^5,15,16), while other investigate groups of countries.^4,17–20 Moreover, the findings regarding the relationship between FDI and renewable energy consumption are mixed and inclusive.⁶ For example, using panel data from 74 countries, and Narayan⁶ showed that FDI reduces the consumption of non-renewable energy consumption and induces consumption of renewable energy sources. The study argued that FDI inflows enhance technological innovation, which is imperative in triggering renewable energy transitions in host countries. Paramati et al.¹⁹ asserted that economic growth, FDI and stock market developments have significant positive impacts on clean energy consumption in emerging economies. Sadorsky¹⁷ explores the relationship between renewable energy consumption, CO₂ emissions and oil prices in the group of seven (G7) countries. The author shows that an increase in economic growth and environmental pollution through CO₂ emissions are the major drivers for demand for renewable energy consumption in those economies. Similarly, Sadorsky²¹ indicated that economic growth has positively and significantly affected renewable energy consumption in emerging economies.

A recent study by Murshed¹⁸ indicated that trade relations among the South Asian economies boost consumption of renewable energy, while FDI inflows are found to reduce the overall use of renewable energy. Using data from the period 1985–2017, Yilanci et al.³ examined the impact of trade openness and FDI on renewable energy consumption for BRICS (Brazil, Russia, India, China and South Africa). The results showed that trade openness has positively influenced renewable energy consumption in South Africa and negatively influenced it in Russia, China and South Africa. Moreover, FDI has a positive impact on clean energy use in Russia, while its effect is insignificant in China and South Africa. Ergun et al.⁸ in the case of 21 sample sub-Saharan African countries found that FDI has positively influenced renewable energy consumption.

Besides, Ji and Zhang⁵ highlighted that financial market development and FDI have positively and significantly contributed to renewable energy growth in China. The justification is that financial market development assists in raising capital for clean energy projects and allows investors to obtain higher risk-adjusted returns. Similarly, Eren et al.¹⁵ estimate that economic growth and financial market development have positively influenced renewable energy consumption in India. A study by Murshed⁴ indicates that trade openness promotes the consumption of renewable energy sources and energy efficiency in South Asian economies. Kim and Park²² provide additional evidence that financial market growth influenced the adoption of renewable energy technology in China.

In contrast, there are studies which argue that FDI and international trade have an insignificant or a negative impact on renewable energy consumption. For instance, a study by Lee,²³ for 19 of the group of 20 (G20) countries during the period 1971–2009, shows that FDI inflows have no significant effect on renewable energy consumption in those economies. Similarly, Kiliçarslan²⁴ mentioned that FDI has a negative effect on renewable energy consumption for the economies of BRICS and Turkey. Shahbaz et al.²⁵ also reported that FDI has an insignificant impact on renewable energy consumption. While prior research has examined the impact of global FDI and trade openness renewable energy consumption and found an inconclusive result, it is important to look at a specific FDI and trade. Recently, Chinese has strongly engaged in Africa by investing and doing business. As such it has affected African economy.¹³ Thus, showing the importance of Chinese investment and trade in promoting energy renewable in Africa need to be researched in order to provide a better policy for the governments in the continent. Hence, this study opted for Chinese FDI and trade to reflect on China's strong bilateral business relationship with the African countries.

Research methodology

Data sources

The study employed data mainly from China's Ministry of Commerce (MOFCOM) and World Bank's World Development Indicators (WDI) database. It utilized panel data for 40 sample African countries over the period from 2003 to 2016. Countries are selected based on the availability of all the data required for the analysis, and sub-regional representativeness is considered to incorporate diverse characteristics. It is important to note that most of data which concerned Chinese FDI and Chinese trade with Africa has been started to be collected in 2003. Moreover, our data has been ended in 2016 due to renewable energy consumption data.

Data on China's FDI and bilateral trade were collected from MOFCOM. The FDI data shows actual outward FDIs by Chinese firms and published in the China Commerce Yearbook by MOFCOM. The MOFCOM data capture the value of realized outward FDI by Chinese firms by country of destination². Admittedly, these data have their own limitations in describing China's outward investment in Africa, as there are various unreported investments by Chinese firms in many African countries¹⁰ (Table 1).

Table 1.

Sources and definitions of the main variables.

Variables	Definition	Data source
lnREC	Total renewable energy consumption as % of total primary energy consumption	WDI
lnCE	% of population access to clean fuels and technologies for cooking	WDI
lnY	Real GDP per capita growth rate	WDI
lnFDI	The stock of China's FDI in African countries	MOFCOM
lnEX	The total value of Africa's exports to China	MOFCOM
lnIM	The total value of Africa's imports from China	MOFCOM
LnFM	Domestic credit provided by financial sector as percentage of GDP	WDI
LnURB	Total urban population as a percentage of total population	WDI
lnPR	Real crude oil price ($brent/barrel)	WDI

Empirical approach

Model specification

Two widely used indicators of renewable energy transition are employed in this study. These are the percentage of renewable energy consumption out of total primary energy consumption and the percentage of population access to clean energy and technologies. Consistent with the broader literature^3,6,19,26 empirical models for determinants of renewable energy consumption for both models are formulated in log form as follows:

l n R E C_{i t} = α_{0} + β_{1} l n I M_{i t} + β_{2} l n E X_{i t} + β_{3} l n F D I_{i t} + β_{4} l n Y_{i t} + β_{5} l n U R B_{i t} + β_{6} l n P R_{i t} + β_{8} l n F M_{i t} + ε_{i t}

(1)

l n C E_{i t} = α_{0} + β_{1} l n I M_{i t} + β_{2} l n E X_{i t} + β_{3} l n F D I_{i t} + β_{4} l n Y_{i t} + β_{5} l n U R B_{i t} + β_{6} l n P R_{i t} + β_{8} l n F M_{i t} + ε_{i t}

(2)

Where i and t shows the individual cross-sectional unit (African countries in this case) and the time period, two indicators of renewable energy transition are used in the study as shown in model (1) and model (2). The dependent variables REC indicate the natural logarithm of shares of renewable energy in total final primary energy consumption for respective African countries, lnCE denotes the natural logarithm of the percentage of the population's access to clean fuels and technologies for cooking. The use of this variable as indicator of renewable energy transition is derived from the SDG of United Nations. Goal seven of the SDG (SDG 7) particularly focuses on the importance of promotion and adoption of clean energy technologies across developing countries. The same indicator variables are used by recent studies.^18,26,27

As far as the main explanatory variables are concerned, lnLM and lnEX stands for the natural logarithm of the total values of imports from China and exports to China respectively, lnFDI shows the natural logarithm of the stock of Chinese FDI in respective African countries and lnY denotes the natural logarithm of gross domestic product (GDP) per capita growth rate. Additionally, lnURB, lnPR and lnFM, respectively indicates the natural logarithm of urbanization level, natural logarithm of the real crude oil prices and the natural logarithm of domestic financial market development in respective African countries; while εit is the error term. The parameters α₀ and β_i (i = 1, 2, 3, 4, 5, 6, 7) indicate the intercept and the estimated elasticities.

The linear log transformation of the model is important to consider the smoothness of the data as well as to provide an elasticity effect of the determinant variables. This model assumed that economic structures, external and demographic factors determine renewable energy transitions in an economy.

In the above equations, if β₃ > 0, outward FDI inflows from China induce or increase the consumption of renewable energy sources in respective African countries, otherwise β₃ < 0. Bilateral trade relations are positively related to renewable energy consumption if β₁ and β₂ > 0 otherwise β₁ & β₂ < 0. Similarly, economic growth or per capita income growth has a positive impact on renewable energy consumption (which might be expected due to a rise in public awareness about environmental quality), when β₄ > 0 otherwise β₄ < 0. The same explanation is true for the remaining variables.

To account for a non-linearity relationship between renewable energy consumption and determinant factors, the following specifications are used which incorporate the square of GDP per capita growth rate in addition to the above variables:

l n R E C_{i t} = α_{0} + β_{1} l n I M_{i t} + β_{2} l n E X_{i t} + β_{3} l n F D I_{i t} + β_{4} l n Y_{i t} + β_{5} l n F M_{i t} + β_{6} l n U R B_{i t} + β_{7} l n P R_{i t} + β_{8} l n Y_{i, t}^{2} + ε_{i t}

(3)

l n C E_{i t} = α_{0} + β_{1} l n I M_{i t} + β_{2} l n E X_{i t} + β_{3} l n F D I_{i t} + β_{4} l n Y_{i t} + β_{5} l n F M_{i t} + β_{6} l n U R B_{i t} + β_{7} l n P R_{i t} + β_{8} l n Y_{i, t}^{2} + ε_{i t}

(4)

Where Y² is the squared term of the real per capita GDP growth rate of sample African countries. It is expected that β₃ < 0; β₄ > 0 if U-shaped relationship exists between economic growth and renewable energy consumption in Africa, otherwise inverted-U shaped relationship is likely expected, that is, β₃>0; β₄ < 0. This sort of specification is important to consider the non-linear relationship between renewable energy consumption and its determinant factors.

In addition to FDI, imports and exports; economic growth, urbanization, domestic financial development and energy price coefficients are expected to influence renewable energy consumption in sample African countries on both models.

Estimation strategy

Panel data usually suffer cross-sectional dependence (CD), group-wise heteroskedasticity and autocorrelation issues. Given the relatively larger sample size of the data unit (N = 40) with shorter time dimensions (14 years, i.e. from 2003 to 2016), first, there is a need to examine the existence of CD among panel variables. To conduct this, Pesaran²⁸ cross-sectional dependency tests were employed and the results are reported in Table 2. According to the test results shown in Table 2, there is evidence of CD in the panel variables. In order to address the CD, the present study adopted the panel corrected standard errors (PCSE) estimation technique, which is widely considered as the most appropriate technique in such circumstances.^11,29,30

Table 2.

Unit root test results for panel variables.

Variable	lnREC	lnCE	lnIM	lnEX	lnY	lnFDI	lnURB	lnPR	lnFM
At levels
LLC	−3.668***	−10.754***	−10.914***	−6.862***	−6.458***	−13.055***	5.176	−7.019***	−12.699***
Breitung	2.8011	−0.887	7.088	2.919	10.120	7.905	11.738	−1.919**	2.764
IPS	2.6237	−3.99***	−6.580	−4.558***	0.135	−4.529***	2.201	−5.486***	−0.971
Fisher-ADF	2.655	13.97***	−9.162***	−6.457***	−1.925**	−9.512***	−6.837***	−6.000***	−1.056
First difference
LLC	−9.240***	−6.99***	−9.590***	−13.511***	−8.671***	−26.125***	−10.86***	−8.624***	−25.60***
Breitung	−10.48***	13.60***	−7.622***	−6.845***	−6.853***	−4.300***	18.92***	−13.52***	−9.291***
IPS	−10.001***	−8.50***	−9.220***	−11.485***	−6.828***	−11.518***	10.71***	−9.09***	−9.342***
Fisher-ADF	−16.409***	40.02***	−13.937***	−20.689***	−9.946***	−22.383***	−25.37***	−12.69***	−15.39***

Note: LLC and IPS indicate Levin-Lin-Chu and Im-Pesaran-Skin tests.

*^, **

and *** represents 10%, 5% and 1% significance level, respectively.

PCSE estimation technique is particularly suited for the nature of panel data which have CD among variables.¹¹ Moreover, it is most appropriate when the panel dataset has larger cross-sectional units (N) than time period dimensions (T). In the present study, cross-sectional units (40 countries) were greater than the time period (14 years). Hence, the PCSE model was adopted in this study to produce efficient results. Recently, this estimation approach has been widely used by various scholars in different areas.^26,27

The PCSE estimation approach basically employs a sandwich estimator to include CD in the process of computing standard errors.³¹ In other words, PCSE estimates are calculated for linear cross-sectional models where the parameters are estimated by either OLS or Prais–Winsten regression. In the process of calculating the standard errors and the variance-covariance estimates under PCSE, disturbances are assumed to be heteroskedastic or that each panel unit has its own variance. It also assumed that each pair of panels had its own covariance (contemporaneously correlated across panels) and also had auto-correlated of order one.¹¹

The Dumitrescu–Hurlin panel causality test

To examine the direction of causality among the panel variables, the study adopted the panel granger causality test approach by DH.¹² The DH test approach was designed to test the possible causality in heterogeneous panel data models. According to DH¹² and Aydin,³² the DH test has the following advantages compared to other conventional causality test approaches. First, the DH test method can account for cross-sectional dependency and slope heterogeneity of panel variables. Second, the test method is flexible in nature, which can be applied in heterogeneous panels and also in a situation where the time dimension is higher than or less than the cross-sectional unit. This is a major limitation of the existing conventional causality estimation techniques. Finally, the DH test technique can effectively be applied to unbalanced panel data.

Given these facts, it is important to discover the directions of causality among the panel variables, once long-run relations prevail. The DH causality test is represented in a linear form as follows:

Y_{i, t} = β_{i} + \sum_{k = 1}^{q} δ_{i}^{(k)} Y_{i, t - k} + \sum_{k = 1}^{q} β_{i}^{(k)} X_{i . t - k} + ε_{i, t}

(5)

Where Y and X are the two panel variables that follow the stationarity property for a given cross-sectional unit at particular time period.

δ_{i}^{(k)}

is lag parameters, k = lag length,

β_{i}^{(k)}

indicates the slope parameters.

The DH¹² causality approach relaxes the strong assumption of Granger³³ which states slope homogeneity across the cross-sectional units. Instead, the DH method estimates a z-bar statistic by allowing the slope coefficients to vary across the cross-sectional units.³⁴

This mean value of Wald test statistic is predicted under the null hypothesis of non-causality between a pair of stationary panel variables, compared to an alternative hypothesis of the existence of a causal relationship between those variables in at least one of the cross-sectional units. The average Wald statistic can be computed as follows:

W_{N, T}^{H N C} = \frac{1}{N} \sum_{i = 1}^{N} W_{i, T}

(6)

Where

W_{N, T}^{H N C}

shows the mean value of the individual Wald statistics and

W_{i, T}

indicates the individual Wald statistics of cross-sectional unit.

Given that the individual residuals are independently distributed across all the cross-sectional units and the corresponding covariance equal to zero, DH¹² showed that the average Wald statistic converges to the following values under larger cross-sectional dimensions:

Z_{N, T}^{H N C} = \sqrt{\frac{N}{2 K}} (W_{N, T}^{H N C} - K)_{T, N \to \infty}^{d}, N (0.1)

(7)

Where

Z_{N, T}^{H N C}

= the Z-bar statistic and K refers to the optimal lag length. To compute the standardized z-bar statistic

({\bar{Z}}_{N, T}^{H N C}

)

{\bar{Z}}_{N, T}^{H N C} = \frac{\sqrt{N}}{\sqrt{v a r ({\overset{⌣}{W}}_{i, T})}} [W_{N, T}^{H N C} - E {\overset{⌣}{W}}_{i, T}]

(8)

The corresponding statistical significance of the z-bar statistic under this method employs a causality test running from the independent to the dependent variable.

Result and discussion

Cross-sectional dependency test

CD may prevail in panel data due to unobserved common shocks or trends that turn out to be components of the error terms. When CD existed in panel data and failure to account for this dependence in empirical estimation may result in inconsistent standard errors of the estimated parameters. Pesaran²⁸ cross-sectional dependency tests were performed to assess CD within the panel variables. The Pesaran²⁸ CD test can be represented as follows:

C D = \sqrt{\frac{1}{N (N - 1)}} \sum_{I = 1}^{N - 1} \sum_{j = i + 1}^{N} (T {\overset{⌣}{ρ}}_{i, j}^{2} - 1)

(9)

When the cross-sectional unit (N) is larger than the time dimension of the panel data (T), Pesaran²⁸ has suggested to use the following test statistic:

C D = \sqrt{\frac{2 T}{N (N - 1)}} (\sum_{I = 1}^{N - 1} \sum_{j = i + 1}^{N} ρ ⌢_{i, j}^{2})

(10)

The term

ρ ⌢_{i, j}

indicates the pairwise correlations in residuals or errors. Accordingly, the null hypothesis and the alternative hypothesis used to test the cross-sectional dependency of the panel variables are stated as follows:

Null hypothesis – no cross-sectional dependency – H_O: cov (ɛit ɛ_ij) = 0

Alternative hypothesis – there is cross-sectional dependency – H₁: cov (ɛit ɛ_ij) ≠ 0

Finally, p-values are computed to make a decision regarding the null hypothesis. If the calculated probability values are smaller than the significance values, the null hypothesis is rejected. Otherwise, if the calculated probability values are higher than the significance values, the null hypothesis cannot be rejected.

The test results in Table 3 clearly show that the null hypothesis of cross-sectional independence was rejected for all the panel variables. It reveals that all the panel variables exhibit CD. Hence, the empirical estimation approach must include appropriate tools which are robust and avoid size distortion bias arising from CD.

Table 3.

Cross-sectional dependency test results.

Variables	Test statistic	p-value
lnREC	41.617***	0.000
lnCE	58.303***	0.000
lnY	74.401***	0.000
lnFDI	96.192***	0.000
lnEX	100.464***	0.000
lnIM	67.616***	0.000
lnURB	77.026***	0.000
lnFM	35.729***	0.000
lnPR	111.714^***	0.000

Note: In CD test, the null hypothesis of cross-section independence, CD ∼ N (0, 1).

***Significance at the 1% level of significance.

Unit root and co-integration tests

After the CD test, the next step is the unit root and co-integration tests to prove the stationary and long-run relationships of the variables, respectively. Unit root test is performed to check the stationarity of the panel variables. There are various methods of testing panel unit root or stationarity of the panel variables. This includes the Levin-Lin-Chun (LLC) test recommended by Levin et al.,³⁵ the Im-Pesaran-Shin (IPS) test method proposed by Im et al.,³⁶ the Augmented Dickey-Fuller (ADF) unit root test approach proposed by Maddala and Wu,³⁷ the Breitung test by Breitung and Das,³⁸ the PP test recommended by Choi³⁹ and the one proposed by Hadri.⁴⁰

To overcome the limitations of one test method over the other and incorporate a diverse test approach, the study used the following four types of unit root test tools; the LLC test, the Breitung test, the IPS test and the Fisher-ADF test. The unit root test results for all the panel variables are reported in Table 2. According to the estimated test results in Table 4, some of the panel variables are non-stationary at levels (i.e. we cannot reject the null hypothesis, which states that the panel variables have a unit root). However, the test results for the first difference of all the panel variables indicate that they are stationary, or can be rejected as the null hypothesis at the 1% level. Hence, the first differences of the panel variables are stationary or have no unit root.

Table 4.

Panel co-integration tests for model (1).

	Statistic	p-value
Augmented Dickey-Fuller test	−3.3536^***	0.0004

Note: The null hypothesis is no cointegration among panel variables.

***

represents a 1% significance level.

After the stationarity test, a panel cointegration test is performed to show whether the panel variables are co-integrated or whether there is a long run equilibrium relationship among the variables. To perform this test, the Kao⁴¹ residual co-integration test was employed and the estimated results are reported in Table 4. The test results for ADF show that the panel variables are co-integrated and hence there are long-run relationships between the variables during the study period. Similar co-integration tests were conducted for model (2) and reported in Appendix (Table A3 in the appendix part).

Determinants of renewable energy consumption

The estimation results from PCSE in the case of model 1 and model 3 are shown in Table 5. In this case, the dependent variable is the ratio of renewable energy consumption to total final primary energy consumption. The estimated coefficients for per capita GDP and square of per capita GDP growth rates are negative and positive, respectively. These results indicate that economic growth initially enhances the consumption of non-renewable energy sources and discourages renewable energy utilization in sample African countries. However, at higher levels of growth and development, economies can promote the consumption of renewable energy sources. It suggests that there is U-shaped non-linear association between renewable energy consumption and economic growth. This is consistent with Doytch and Narayan⁶ which reported that economic growth initially promotes mainly non-renewable energy consumption.

Table 5.

The PCSE estimation results for determinants of renewable energy consumption in Africa.

Variables	Full sample				Sub-Saharan African countries’ sample
	(1)		(2)		(1)		(2)
	Coefficients	SE	Coefficients	SE	Coefficients	SE	Coefficients	SE
lnY	−0.7808***	0.010	−0.780***	0 .0103	−0.3522***	0.0085	−0.059***	0.0062
lnFDI	0.0247***	0.0061	0.025***	0.0062	0.0149***	0.0051	0.0172***	0.005
lnY²			0.115***	0.018			0.530***	0.0919
lnEX	−0.1675***	0 .0219	−0.1671***	0 .022	−0.0439***	0.0114	−0.058***	0.0125
lnIM	0.0852***	0.0122	0.0849***	0.0121	0.0344***	0.004	0.0363***	0.0049
lnURB	−0.0798***	0.0155	−0.0796***	0.0157	−0.1201***	0.0139	−0.1102***	0.0132
lnFM	0.1146***	0.0188	0.007	0.005	0.288***	0.0467	0.297***	0.0461
lnPR	0.0637**	0.0273	0.0628**	0.0275	−0.006	0.0112	−0.0055	0.0108
Constant	10.25***	0.2327	10.226	0.228	5.671	0.2937	2.630	0.4618
N	640		640		544		544
Number of countries	40		40		34		34
R ²	0.4436		0.4439		0.6457		0.6575
Wald chi2 p -value	19923.33(0.0000)		20958.37(0.0000)		10505.56(0.0000)		33494.55(0.0000)

Note: The dependent variable is the log of share of renewable energy consumption in total primary energy consumption, standard errors are panel corrected standard errors, and *^, ** and *** are respectively significant levels at 10%, 5%, and 1%.

Imports from China have been found to be ideal for promoting renewable energy consumption in Africa. According to the estimates, imports from China have significantly contributed to boosting renewable energy utilization in sample African countries. The coefficients are statistically significant across all the specifications and sample groups. For instance, the corresponding elasticity estimate for the full sample group indicates that a 1% increase in imports from China will approximately account for about 0.0852% rise in renewable energy consumption, given other factors constant. The positive relationship between trade and renewable energy consumption is consistent with the finding of Murshed¹⁸ and Murshed⁴ who reported that improvement in trade intensities among Asian economies as a promoting factor for renewable energy transition.

The estimate of the elasticity, however, reveals a negative association between renewable energy consumption and exports to China. The results show that exportation of goods and services from Africa mainly utilized non-renewable energy sources during the study period. The estimated elasticity reveals that other things remain constant; a 1% increase in exports to China will approximately account for about 0.0439% reduction in renewable energy consumption in sub-Saharan African countries. The coefficient is statistically significant at the 1% level.

It is also evident from the table that exogenous energy price shocks are expected to induce the consumption of renewable energy. The estimated coefficients are positive and statistically significant for the full sample group. The elasticity estimates for the full sample group show that a 1% increase in energy prices will account for about 0.0637% rise in renewable energy consumption, taking factors constant. The finding shows that crude oil and renewable energy sources are considered substitutes. Similar findings are reported by Sadorsky,¹⁷ who indicated that higher oil prices are associated with a transition to renewable energy consumption in developed countries (G-7 countries). Omri and Nguyen²⁰ also showed the same results for a group of both developed and developing countries. The effect of energy prices on renewable energy consumption is negative and insignificant in the sub-Saharan region.

Additionally, FDI inflows from China were found to be stimulants of renewable energy consumption in Africa. The results indicate that FDI induces a technological spillover to the host countries, which in turn promotes the development of renewable energy projects.⁴² The coefficients are statistically significant across all the specifications and the sample groups. Other things remain constant: a 1% increase in the stock of China's FDI in Africa will approximately result in about 0.0247% and 0.0149% rise in renewable energy consumption for the full and sub-Saharan sample groups, respectively. The results precisely postulate that international investment flows from China can contribute to renewable energy development on the continent. This finding is consistent with previous studies in many countries (for instance, ^6,16,19,42). Doytch and Narayan⁶ found that higher FDI inflows increase industrial level of renewable energy utilization and reduce non-renewable energy consumption. Similarly, Fan and Hao¹⁶ reported that FDI inflows have a long-run significant impact on renewable energy consumption in China. However, Murshed¹⁸ found the opposite results, which reveal that FDI inflows are reducing the overall use of renewable energy in South Asian countries. Lee²³ also reports that FDI inflows have no statistically significant effect on clean energy consumption in the group of 20 (G20) countries.

Finally, according to the regression results, domestic financial market development was found to be statistically significant in explaining the variation in renewable energy consumption in Africa. The result implies that as the financial market develops, there is a tendency to shift from non-renewable to renewable energy consumption. In contrast, urbanization has been found to negatively affect renewable energy consumption. This somewhat might be interpreted as an indication of lower and polluted urban growth in Africa, which basically uses non-renewable energy sources.

The PCSE regression results based on model 2 and model 4 are shown in Table 6. In this case, the dependent variable is the percentage of the population with access to clean fuels and technology for cooking.

Table 6.

The PCSE estimation results in the case of access to clean energy and technology.

Variables	Full sample group				Sub-Saharan African sample group
	(1)		(2)		(1)		(2)
	Coefficients	SE	Coefficients	SE	Coefficients	SE	Coefficients	SE
lnY	1.2939***	0.0186	1.294***	0.0185	1.240***	0.0128	1.184***	0.0169
lnY²			0.3584***	0.0343			0.4762***	0.0280
lnFDI	0.0064	0.0097	0.0060	0.0097	0.0263***	0.0096	0.0052	0.0091
lnEX	−0.0464***	0.0098	−0.046***	0.0098	−0.0394***	0.0088	−0.0157	0.0101
lnIM	0.0821***	0.0200	0.0832	0.0197	0.0226	0.0155	0.0069	0.0084
lnURB	0.3588***	0.0336	0.0089	0.0099	−0.4650***	0.0553	−0.0416**	0.0210
lnFM	0.0707***	0.0205	0.0715***	0.0208	0.0042	0.0139	0.070***	0.0174
lnPR	−0.0565**	0.0284	−0.0592**	0.0286	0.0773***	0.0255	0.0739***	0.0263
Constant	−9.209***	0.1926	−9.246***	0.190	−6.988	0.1885	−7.6178***	0.1980
N	560		560		490		490
Number of countries	40		40		35		35
R ²	0.7175		0.7178		0.7199		0.7056
Wald chi2 p-value	61902.88 (0.0000)		63246.52 (0.0000)		48392.48 (0.0000)		255279.23 (0.0000)

Note: The dependent variable is the log of share of the percentage of population access to clean fuels and technology for cooking; standard errors are panel corrected standard errors, and *^, ** and *** are respectively significant levels at 10%, 5% and 1%.

One interesting piece of evidence from Table 6 is the effect of economic growth on access to clean energy and technologies. In this case, the coefficients for both GDP growth rates and squared GDP growth rates are positive. These results indicate that there is an inverted U-shaped relationship between economic growth and access to clean fuels and technology in sample African countries. The corresponding elasticity estimate for the full sample shows that a 1% increase in the economic growth or real GDP per capita growth rate will approximately account for a rise in access to clean energy and technology by 1.2939%, taking other factors remain constant. This finding implies that economic growth can enhance the purchasing power of households, which in turn enables them to afford relatively better, cleaner cooking fuels and technologies.

It is also apparent from the estimation results that FDI inflows from China have been positively associated with access to clean energy and technology in Africa. The coefficient was statistically significant for the sub-Saharan African countries sample group. A possible explanation behind the positive association between FDI and access to clean technology could be due to the fact that the advanced investment from China can possibly be contributed for production and importation of relatively cleaner energy sources, ultimately enhancing the overall consumption of clean energy. This result echoed the findings of Murshed et al.²⁶ in South Asian economies, which reveal the positive impacts of FDI on access to clean energy and technology.

Similarly, imports from China also stimulated access to cleaner energy and technology across sample African countries. The estimated coefficient for full sample indicates that, keeping other factors remain constant, a 1% increase in imports from China will approximately result in increase of clean fuel and technology accessibility by 0.0821% in Africa. The coefficient is statistically significant at the 1%level. However, though positive, the estimated coefficients are statistically not significant for the sub-Saharan African countries sample group. In contrast, exports to China have been negatively associated with access to clean energy and technologies in Africa.

These two segments of empirical results have critically important implications for agrarian economies of African countries where households are predominantly utilized traditional energy sources (such as biomass, unclean solid cooking fuels, charcoals, etc) which contributed for environmental degradation. Hence, expansion of trade and investment relations with China could be an ideal source of technological capacity which will facilitate the production and distribution of clean energy facilities in African countries.

More importantly, financial market development is significantly contributed for the promotion of clean energy accessibility in sample African countries. According to statistical estimates, an improvement of domestic financial market is found to be statistically significant in explaining the variation in access to clean energy and technology. Keeping other factors remain constant, a 1% increase in domestic financial market development will contribute to about 0.0715 and 0.070% increase in access to clean fuel and technology for the full sample and sub-Saharan African countries sample groups, respectively. This empirical finding falls in similar lines with Ji and Zhang⁵ and Eren et al.¹⁵ Ji and Zhang⁵ indicated that financial market development significantly contributes to renewable energy growth in China. Using several econometric techniques, Eren et al.¹⁵ also found that the growth of the financial market has a positive impact on clean energy consumption in India.

The negative coefficients of urbanization for sub-Saharan region probably might be due to relatively lower capacity of renewable energy infrastructure in those economies, predominant utilization traditional energy sources and poor access to basic services.

Robustness check

In order to check the robustness of the above results, additional estimation has been conducted using the fully modified ordinary least square (FMOLS) estimation method. The results for both models are reported in Table 7. The estimated coefficients for the main variables are consistent with the PCSE estimation results for both models, though slight differences are observed regarding the significance level of some variables.

Table 7.

The FMOLS estimation results for both models.

Dependent variable is lnREC					Dependent variable is lnCE
Variables	Full sample		Sub-Saharan sample		Full sample		Sub-Saharan sample
Variables	(1)	(2)	(1)	(2)	(1)	(2)	(1)	(2)
lnY	0.325(0.212)	−0.010(0.497)	0.408**(0.201)	0.089(0.088)	0.325(0.211)	0.091(0.081)	0.408**(0.201)	0.056(0.075)
lnY²		0.081***(0.019)		0.056(0.075)		0.375*(0.215)		0.426**(0.203)
lnFDI	0.631***(0.226)	0.406**(0.183)	0.545**(0.218)	0.589***(0.218)	0.631***(0.225)	0.668***(0.225)	0.146(0.112)	0.145(0.112)
lnEX	−0.112(0.213)	−0.113(0.095)	−0.447**(0.201)	−0.475**(0 .200)	−0.112(0.213)	−0.127(0.212)	−0.447**(0.201)	−0.474**(0.200)
lnIM	0.065(0.098)	0.076(0.077)	0.075(0.089)	0.426**(0.203)	0.065(0.098)	0.074(0.098)	0.545**(0.218)	0.589***(0.218)
lnURB	2.228***(0.564)	0.283(0.617)	2.218***(0.517)	2.21***(0.514)	2.22***(0.564)	2.23***(0.563)	2.228***(0.517)	2.21***(0.514)
lnFM	0.057(0.119)	0.076(0.077)	0.145(0.112)	0.145(0.112)	0.057(0.119)	0.061(0.119)	0.075(0.089)	0.089(0.088)
lnPR	0.154(0.634)	0.127(0.166)	0.446(0.606)	0.429(0.603)	0.154(0.634)	0.150(0.633)	0.446 (0.606)	0.429(0.603)
Constant	−7.989***(3.42)	−4.743*(2.69)	−4.33(3.24)	−4.209(3.24)	−7.98**(3.423)	−8.25**(3.43)	−4.33(3.24)	−4.21(3.25)
N	559	559	489	489	559	559	489	489

Note: Values in parenthesis are standard errors and *^, ** and *** are respectively significant levels at 10%, 5% and 1%

The FMOLS estimates confirm the above results that FDI inflows from China have positively influenced the renewable energy transition in Africa. The coefficients are statistically significant for both indicators of renewable energy transitions. It is also evident from the results that imports from China have been positively associated with access to clean energy and technologies. In contrast, similar to PCSE estimates, exports to China have been negatively related to renewable energy consumption in sample African countries during the study period. Thus, in line with the negative sign of the coefficient, it implies that exports to China can be largely utilizing overall non-renewable energy consumption in sample African economies.

Generally, it is evident from the above results that although the regression results from FMOLS estimates are mostly matching in terms of the predicted signs of coefficients with PCSE, the predictor power of some variables is statistically insignificant under FMOLS. This might be due to the fact that FMOLS estimates are appropriate for panels with large time dimensions, which is not the case in the present study. However, the PCSE estimate provides results which are appropriate for short panel data, account for cross-sectional dependency, autocorrelations and are also less sensitive to outliers. Hence, such differences in estimated results justify the choice of the PCSE estimation technique, which simultaneously accounts for several econometric issues. Moreover, a sensitivity analysis has been performed by using GMM estimation technique to account for endogeneity. For brevity, these results are not reported but are available on request.

Panel causality estimation results

Although the regression results predict the average effects of the explanatory variables on the dependent variables, it inherently fails to show the reverse possibility of average effects of the dependent variable on the independent variables. Therefore, it is important to perform the causality analysis to indicate the potential reverse causations between the variables. The panel causality tests were conducted using the newly developed test approach by DH¹² and the results are reported in Table 8 in the context of both models.

Table 8.

Dumitrescu–Hurlin causality test results.

Null hypothesis	W-Stat.	Z-bar Stat.	p-value	Conclusion
For lnREC model
lnlnY does not granger cause lnREC	3.9499	13.1925***	0.0000	LnY ↔ lnREC
lnREC does not granger cause lnY	2.2610	5.6395***	0.0000	LnY ↔ lnREC
lnIM does not granger cause lnREC	5.1009	18.3398***	0.0000	LnLM → lnREC
lnREC does not granger cause lnIM	0.8249	−0.7829	0.4337	LnLM → lnREC
lnEX does not granger cause lnREC	3.1902	9.7948***	0.0000	LnEX ↔ lnREC
lnREC does not granger cause lnEX	1.7572	1.7708*	0.0766	LnEX ↔ lnREC
lnFDI does not granger cause lnREC	3.6956	12.0551***	0.0000	LnFDI ↔ lnREC
lnREC does not granger cause lnFDI	2.8635	8.3336***	0.0000	LnFDI ↔ lnREC
lnURB does not granger cause lnREC	3.3216	10.3826***	0.0000	LnURB ↔ lnREC
lnREC does not granger cause lnURB	5.9178	21.9929***	0.0000	LnURB ↔ lnREC
lnFM does not granger cause lnREC	2.5289	6.8374***	0.0000	LnFM ↔ lnREC
lnREC does not granger cause lnFM	1.8154	1.9559*	0.0505	LnFM ↔ lnREC
lnPR does not granger cause lnREC	2.1159	4.9904***	0.0000	LnPR → lnREC
lnREC does not granger cause lnPR	0.9359	−0.2867	0.7743	LnPR → lnREC
For lnCE model
lnlnY does not granger cause lnCE	2.1231	5.0224***	0.0000	LnY ↔ lnCE
lnCE does not granger cause lnlnY	3.6883	12.0225***	0.0000	LnY ↔ lnCE
lnIM does not granger cause lnCE	1.6830	3.0543***	0.0023	LnIM ↔ lnCE
lnCE does not granger cause lnIM	1.9549	2.0590**	0.0395	LnIM ↔ lnCE
lnEX does not granger cause lnCE	1.2574	1.1513	0.2496
lnCE does not granger cause lnEX	2.6391	7.3302***	0.0000	LnCE → lnEX
lnFDI does not granger cause lnCE	1.8261	3.6942***	0.0002	LnFDI ↔ lnCE
lnCE does not granger cause lnFDI	9.5219	38.1110***	0.0000	LnFDI ↔ lnCE
lnURB does not granger cause lnCE	5.9012	21.9188***	0.0000	LnURB ↔ lnCE
lnCE does not granger cause lnURB	12.4845	51.3602***	0.0000	LnURB ↔ lnCE
lnFM does not granger cause lnCE	1.2162	0.9669	0.3336
lnCE does not granger cause lnFM	3.1424	9.5809***	0.0000	LnCE → lnFM
lnPR does not granger cause lnCE	1.1676	0.7495	0.4535
lnCE does not granger cause lnPR	0.3659	−2.8359***	0.0046	LnCE → lnPR

Note: → shows unidirectional causality, ↔ indicates bidirectional causality, *^, ** and *** are respectively significant levels at 10%, 5% and 1%.

According to the results, there are statistically significant bidirectional causalities between renewable energy consumption and economic growth in both models. This finding suggests that not only does higher economic growth enhance the transition to renewable energy, but renewable energy promotion is also vital to achieve sustainable economic growth in Africa. A similar bidirectional causality was reported by Eren et al.¹⁵ in the case of India. Ergun et al.⁸ also found bidirectional causal relationship between economic growth and renewable energy consumption for sample of 21 African countries. Such feedback effect shows the significance of stable economic progress for renewable energy consumption. However, a study by Menyah and Wolde-Rufael⁴³ showed a unidirectional causality stemming from economic growth to renewable energy consumption for USA.

Moreover, a strong unidirectional causality from imports to renewable energy consumption was found. The result reveals that there is significant unidirectional causality stemming from imports to renewable energy consumption. The causality result supports the corresponding regression estimate. The finding agrees with Zeren and Akkuş³⁴ who indicate that there is unidirectional causality stemming from trade openness to renewable energy consumption in developing countries.

There is bidirectional causality between FDI inflows from China and renewable energy consumption in Africa in both models. This finding again illustrates that not only does the FDI inflows induce renewable energy consumption, but also that the development of renewable energy consumption also attracts investments in African countries.⁴⁴ The finding is consistent with Murshed¹⁸ who found a causality between FDI and renewable energy transitions in South Asian countries.

Consistent with the regression results, the causality estimates show there is unidirectional causality stemming from energy prices to renewable energy consumption. This finding again indicates that an increase in crude oil prices is expected to induce the consumption of renewable energy, as the two energy sources are substitutable. This is particularly true for oil importing African countries, where continuous fluctuations in oil prices affect macro economies and force countries to find alternative energy sources.

Finally, bidirectional casual association was found between urbanization and renewable energy consumption for both models. These causality results reveal that further development of urbanization in Africa has a feedback effect in terms of promoting renewable energy consumption.

Conclusion and policy implications

De-carbonization of energy sources through promotion of renewable energy consumption is critically vital to meet environmentally sustainable development goals. International trade and FDI inflows to developing countries are considered to be an important source of capital which will facilitate the transfer of knowledge and technologies related to renewable energy transitions. This study empirically investigates the effect of Sino-African economic relations through trade and investment on renewable energy consumption in Africa. Two indicators of renewable energy transition are used in this study. These are the percentage of renewable energy consumption out of total primary energy consumption and the percentage of the population access to clean energy fuels and technologies for cooking. PCSE estimation techniques are employed. Additionally, the study employed Dumitrescu–Hurlin (2012) panel causality analysis to show the long-run casual association between the variables.

The results revealed that FDI inflows from China have significantly contributed to renewable energy consumption in Africa. Imports from China have also been found to be ideal for promoting renewable energy consumption. The results confirm that international economic links with China in terms of imports and investment can have spillover effects in terms of stimulating renewable energy transition in Africa. In contrast, exports to China have been negatively associated with renewable energy consumption, indicating that exports are mainly utilizing non-renewable energy sources. According to the regression results, an improvement in the domestic financial market was found to be statistically significant in explaining the variation in access to clean energy technologies in African countries. Moreover, higher levels of per capita income or economic growth have been positively related to renewable energy consumption.

The panel causality test results show that there are significant bidirectional causalities between renewable energy consumption and GDP per capita growth rates in both models. There is also bidirectional causality between FDI inflows from China and renewable energy consumption. The finding illustrates that not only do the flow of FDI or economic growth induces renewable energy growth, but also the promotion of renewable energy consumption is critical to achieve sustainable economic growth, attract investment and also expand international trade. Moreover, the causality test results further indicate the importance of urbanization and financial market development in the long run.

Finally, given the above results and discussions, the following policy implications are drawn from the study. First, governments and policymakers in African countries need to further strengthen trade and investment relations with China, which will maximize renewable energy technology transfer on the continent. An incentive structures like auctions, feed-in tariffs should be adopted by African governments in order to increase Chinese FDI. Moreover, African governments should minimize political risk which seems to hamper the attractiveness of Chinese FDI. Trade and investment policies should be integrated with renewable energy policy in order to achieve sustainable development. As China is becoming the largest investors and trading partner in Africa, and China's investment and trade are expected to increase in the coming years, there is a need to adopt suitable policies and strategies which will maximize the technological spillover effects of economic cooperation in terms of enhancing renewable energy transition in the continent. Consistent with this, there is a need to maximize Belt and Road Initiative opportunities, which ruminate about the possibility of a green belt of connectivity through renewable energy installations through advanced technological grids. This is particularly important to meet both the SDGs (2030) of United Nations and agenda 2063 of African Union, which stresses renewable energy transition as core target.

Additionally, the development of domestic financial markets on the continent is important to promote renewable energy projects and technologies in Africa. Incentive measures must also be adopted to encourage the utilization of renewable energy sources in export sectors and governments need to make renewable energy technologies accessible in this sector.

Finally, sound macroeconomic policy, which promotes sustainable economic growth and per capita income, is important to expand renewable energy consumption in Africa.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Mesfin Welderufael Berhe

Toyo Amègnonna Marcel Dossou

Abel Dula Wedajo

Notes

Mesfin Welderufael Berhe holds PhD in Development Economics at Southwestern University of Finance and Economics. His research interest includes development economics, technological spillover effect of investment and trade in Africa, China-Africa economic relations, macro-economic policy analysis, energy and environmental economics and tourism economics. He has published various academic papers in top rated journals. He has previously worked at University of Gondar as Assistant professor of Economics.

Toyo Amègnonna Marcel Dossou is an independent researcher at the Public Economics Laboratory of the University of Abomey-Calavi in Benin. He obtained PhD in Regional Economics (2022) at Southwestern University of Finance and Economics, the People's Republic of China. He has published in renowned journals such as Tourism Economics, African Development Review, Current Issues in Tourism, Research in Globalization, Energy& Environment, Renewable energy, International Social Science Journal and Journal of Environmental Management. His research interests include environ- mental economics, Tourism Economics, Institutional Economics, Development Economics, and applied econometrics. He can be contacted at dossoumarcel863@yahoo.fr.

Junbing Huang is currently a professor at School of Economics, Southwestern University of Finance and Economics. Currently, his main research interests include digital economy, environmental policy, ecological economics, and energy economics. Dr. Huang published over 50 papers in peer review journals including: Renewable and Sustainable Reviews, Energy, Energy Economics, Energy Policy, etc.

Abel Dula Wedajo is an academic with a diverse background in business, economics, and technology. He holds a PhD from the Southwestern University of Finance and Economics in China. Abel's research interests include Innovation, Foreign Direct Investment, and Entrepreneurship. He has worked as an instructor in Ethiopian Ministry of Agriculture Colleges, an enterprise development officer in Wolkite Town Trade and Development Office, and a college dean in Business and Economics College at Wolkite University. He has also served as a Senior Product Engineer in Cloud Computing at Huawei Technologies Ethiopia. Currently, Abel is working as an Assistant Professor and Scholar at Wolkite University, while also teaching at Saxo Fintech Business School Sanya, China.

Appendix

Table A3.

Panel co-integration tests for model (2).

	Statistic	p-value
Augmented Dickey-Fuller test	2.6534***	0.0040

Note: the null hypothesis is no cointegration among panel variables.

***1% significance level.

References

Aguirre

Ibikunle

. Determinants of renewable energy growth: a global sample analysis. Energy Policy 2014; 69: 374–384.

Energy Information administration. International Energy Outlook 2021 with projection to 2050; U.S. Energy Information administration; https://www.eia.gov/outlooks/ieo/pdf/IEO2021_Narrative.pdf. 2021.

Yilanci

Ozgur

Gorus

. The asymmetric effects of foreign direct investment on clean energy consumption in BRICS countries: a recently introduced hidden cointegration test. J Cleaner Prod 2019; 237: 117786.

Murshed

. Does improvement in trade openness facilitate renewable energy transition? Evidence from selected South Asian economies. South Asia Econ J 2018; 19: 151–170.

Zhang

. How much does financial development contribute to renewable energy growth and upgrading of energy structure in China? Energy Policy 2019; 128: 114–124.

Doytch

Narayan

. Does FDI influence renewable energy consumption? An analysis of sectoral FDI impact on renewable and non-renewable industrial energy consumption. Energy Econ 2016; 54: 291–301.

Aïssa

MSB

Jebli

Youssef

. Output, renewable energy consumption and trade in Africa. Energy Policy 2014; 66: 11–18.

Ergun

Owusu

Rivas

. Determinants of renewable energy consumption in Africa. Environ Sci Pollut Res 2019; 26: 15390–15405.

Chen

Dollar

Tang

. Why is China investing in Africa? Evid Firm Level 2018; 32: 610–632.

10.

Brautigam

Diao

McMillan

, et al.

Chinese Investment in Africa: how much do we know?

PEDL Synthesis Series 2017; 2: 43.

11.

Bailey

Katz

. Implementing panel corrected standard errors in R: the PCSE package. J Stat Softw 2011; 42: 1–11.

12.

Dumitrescu

Hurlin

. Testing for Granger non-causality in heterogeneous panels. Econ Model 2012; 29: 1450–1460.

13.

Zakari

Oryani

Alvarado

, et al. Assessing the impact of green energy and finance on environmental performance in China and Japan. Econ Change Restruct 2023; 56: 1185–1199.

14.

Zhang

Zhao

Cai

, et al. Influence of renewable energy infrastructure, Chinese outward FDI, and technical efficiency on ecological sustainability in belt and road node economies. Renew Energy 2023; 205: 608–616.

15.

Eren

Taspinar

Gokmenoglu

. The impact of financial development and economic growth on renewable energy consumption: empirical analysis of India. Sci Total Environ 2019; 663: 189–197.

16.

Fan

Hao

. An empirical research on the relationship amongst renewable energy consumption, economic growth and foreign direct investment in China. Renew Energy 2020; 146: 598–609.

17.

Sadorsky

. Renewable energy consumption, CO2 emissions and oil prices in the G7 countries. Energy Econ 2009a; 31: 456–462.

18.

Murshed

. Can regional trade integration facilitate renewable energy transition to ensure energy sustainability in south Asia? Energy Reports 2021; 7: 808–821.

19.

Paramati

Ummalla

Apergis

. The effect of foreign direct investment and stock market growth on clean energy use across a panel of emerging market economies. Energy Econ 2016; 56: 29–41.

20.

Omri

Nguyen

. On the determinants of renewable energy consumption: international evidence. Energy 2014; 72: 554–560.

21.

Sadorsky

. Renewable energy consumption and income in emerging economies. Energy Policy 2009b; 37: 4021–4028.

22.

Kim

Park

. Financial development and deployment of renewable energy technologies. Energy Econ 2016; 59: 238–250.

23.

Lee

. The contribution of foreign direct investment to clean energy use, carbon emissions and economic growth. Energy Policy 2013; 55: 483–489.

24.

Kilicarslan

. The Relationship between Foreign Direct Investment and Renewable Energy Production: Evidence from Brazil. Int J Energy Econ Policy 2019; 9: 291–297.

25.

Shahbaz

Destek

Polemis

. Do foreign capital and financial development affect clean energy consumption and carbon emissions? Evidence from BRICS and next-11 countries. SPOUDAI-J Econ Business 2018; 68: 20–50.

26.

Murshed

Abbass

Rashid

. Modelling renewable energy adoption across south Asian economies: empirical evidence from Bangladesh, India, Pakistan and Sri Lanka. Int J Finance Econ 2021; 26: 5425–5450.

27.

Nguyen

Bui

, et al. Financial development and energy poverty: global evidence. Environ Sci Pollut Res 2021; 28: 35188–35225.

28.

Pesaran

. General diagnostic tests for cross-sectional dependence in panels. Empir Econ 2021; 60: 13–50.

29.

Marques

Fuinhas

. Is renewable energy effective in promoting growth? Energy Policy 2012a; 46: 434–442.

30.

Marques

Fuinhas

. Are public policies towards renewables successful? Evidence from European countries. Renew Energy 2012b; 44: 109–118.

31.

Beck

Katz

. What to do (and not to do) with time-series cross-section data. Am Political Sci Rev 1995; 89: 634–647.

32.

Aydin

. Renewable and non-renewable electricity consumption–economic growth nexus: evidence from OECD countries. Renew Energy 2019; 136: 599–606.

33.

Granger

CWJ

. Investigating causal relations by econometric models and cross-spectral methods. Econometrica 1969; 37: 424–438.

34.

Zeren

Akkuş

. The relationship between renewable energy consumption and trade openness: new evidence from emerging economies. Renew Energy 2020; 147: 322–329.

35.

Levin

Lin

Chu

CSJ

. Unit root tests in panel data: asymptotic and finite-sample properties. J Econom 2002; 108: 1–24.

36.

Pesaran

Shin

. Testing for unit roots in heterogeneous panels. J Econom 2003; 115: 53–74.

37.

Maddala

. A comparative study of unit root tests with panel data and a new simple test. Oxford Bull Econ Stat 1999; 61: 631–652.

38.

Breitung

Das

. Panel unit root tests under cross-sectional dependence. Statist Neerlandica 2005; 59: 414–433.

39.

Choi

. Unit root tests for panel data. J Int Money Finance 2001; 20: 249–272.

40.

Hadri

. Testing for stationarity in heterogeneous panel data. Econom J 2000; 3: 148–161.

41.

Kao

. Spurious regression and residual-based tests for cointegration in panel data. J Econom 1999; 90: 1–44.

42.

Doytch

. The impact of foreign direct investment on the ecological footprints of nations. Environ Sustain Indicat 2020; 8: 100085.

43.

Menyah

Wolde-Rufael

. CO2 Emissions, nuclear energy, renewable energy and economic growth in the US. Energy Policy 2010; 38: 2911–2915.

44.

Ekwueme

Zoaka

Alola

. Carbon emission effect of renewable energy utilization, fiscal development, and foreign direct investment in South Africa. Environ Sci Pollut Res 2021; 28: 41821–41833.