The Role of Green Innovation, Renewable Energy, and Institutional Quality in Promoting Green Growth: Evidence from African Countries

Abstract

Green growth exhibits an immense potential to transform economies and safeguard the planet as it creates a symbiotic relationship between economic progress and environmental protection. This study examines the impact of green innovation, renewable energy consumption, and institutional quality on green growth in African countries, controlling for GDP per capita, trade openness, foreign direct investment (FDI), population, and natural resource rent. The short- and long-run relationships are investigated using pooled mean group (PMG), mean group (MG), and dynamic fixed effects (DFE) models with panel data for 49 African countries from 2000 to 2021. The findings reveal that green innovation, renewable energy consumption, institutional quality, GDP per capita, trade openness, and population growth have positive long-run effects on green growth. In contrast, FDI and natural resource depletion have adverse effects. In the short run, only institutional quality and GDP per capita positively affect green growth, while natural resource rent has a negative impact. Considering these findings, this study recommends that policymakers in Africa promote green innovation and adopt energy-efficient technologies, increase the use of renewable energy resources, and improve institutional quality to achieve green growth.

1. Introduction

Following the dawn of industrialization in the mid-18th century, the global economy experienced remarkable growth and development [1]. However, there is growing concern about the negative environmental impacts of this growth. Rapid growth requires high resource consumption that causes severe ecological damage, including pollution and climate change [2]. In addition, ecological damage impedes the growth of the global economy, putting sustainable development at the top of the international agenda [3].

Various tools and programs have been introduced over the past few decades to reconcile high economic growth and environmental protection. In 2015, the United Nations (UN) launched the 2030 Agenda for Sustainable Development and introduced 17 Sustainable Development Goals (SDGs) to end poverty in all forms and protect the planet. In the same year, 196 countries signed the Paris Agreement, a legally binding international treaty on climate change whose main objective is to achieve zero emissions by 2050 and thereby limit the global average temperature to 2 °C above preindustrial levels [4].

However, the progress toward SDGs and the terms of the Paris Agreement is slow and uneven. Many countries still face severe poverty, income inequality, and environmental degradation. The main challenges preventing progress toward sustainable development are political hesitation, path dependency, and lack of cooperation among nations [5].

Green growth emerges as a viable approach to achieve the coincident objectives of fostering economic growth and development while preserving the services of natural resources and the environment [6]. Green growth strategies involve integrating environmental concerns into economic decision-making processes and restructuring industries to prioritize the environment. This includes investing in clean technologies, renewable energy, and resource efficiency. While innovation in green technologies is expected to drive efficiency gains, adopting cleaner technologies and renewable energy aims to mitigate environmental damage [7,8].

On the back of this background, this study aims to identify the determinants of green growth. We hypothesize that green innovation, renewable energy consumption, and institutional quality are essential factors that promote green growth. Green innovation involves developing products, services, and processes that improve energy efficiency and use natural resources efficiently without harming the environment. Hence, while low greenhouse gas emitting machinery, equipment, and processes help mitigate environmental damage, industries that adopt these innovations are transformed into sustainable growth paths [9,10].

Augmenting renewable energy sources is also crucial in promoting green growth. Investing in renewable energy sources such as solar, wind, and hydropower can help reduce dependence on fossil fuels while satisfying electricity demand and increasing energy security for countries worldwide [11]. Additionally, it aligns with the principles of green innovation, as it promotes the efficient utilization of natural resources and promotes environmentally friendly technologies.

Institutional factors, such as political stability, the rule of law and democracy, and bureaucratic quality, lead to economic growth and improve environmental quality [12,13,14]. On the other hand, poor institutional quality can hinder entrepreneurship and innovation, leading to the persistence of low-income areas [15] and low environmental quality. Moreover, institutional quality also affects the environmental quality of surrounding nations through spatial diffusion channels [16]. Thus, institutions may be pivotal in shaping societal rules, norms, and structures as crucial facilitators of green growth.

Accordingly, this study investigates the effects of green innovation, renewable energy consumption, and institutional quality on green growth in selected African countries. We focus on African economies for several reasons. First, Africa is currently the second fastest-growing continent in the world. Between 2000 and 2021, its average annual growth rate was 4.1% (compared to 3.06% of the world average growth rate) (see Figure A1), and it is expected to increase to nearly 5–6% in 2030 [17]. However, this rapid growth has come at a cost, leading to increased energy consumption, pollution, and environmental degradation [18]. This, coupled with the ongoing struggles with poverty, reveals the importance of understanding green growth in Africa.

Second, while contributing a relatively low share of global carbon emissions, estimated at 0.7–1.3 billion metric tons annually between 2000 and 2021 (see Figure A2), Africa bears a disproportionately heavy burden from the consequences of climate change. This disparity becomes evident when compared to major emitters like Asia Pacific (7.7–17.7 billion), North America (6.6–5.7 billion), and Europe (4.8–3.8 billion). This unequal distribution of emissions transforms into tangible hardships for Africa, with approximately 250 million people already experiencing water stress and an estimated 80% of African countries projected to lack sustainable water management by 2030 [19]. This situation underscores the urgent need for approaches that concurrently address environmental challenges and support sustainable economic growth.

Third, Africa’s commitment to achieving the SDGs by 2030 and the objectives of the Paris Agreement by 2050 requires a shift to sustainable resource management that prioritizes conservation, biodiversity protection, and renewable energy consumption [20]. Therefore, identifying the determinants of green growth and making policy recommendations in this direction will promote sustainable development on the continent and help African policymakers move forward toward their commitment to the SDGs and the objectives of the Paris Agreement.

While there is extensive literature on sustainability and green growth for developing and advanced economies, the literature on African countries is emerging. The closest study to ours is [21], which studies the impact of technological innovation, renewable energy consumption, and institutional quality on CO₂ emissions in 25 African countries. Our study strays from the typical carbon emissions approach in the environmental literature and highlights the importance of economic development, social well-being, the efficient use of natural resources, and ecological sustainability through green growth. To this end, we generate the green growth variable for 49 African countries using the framework proposed by [22].

The results indicate that institutional quality contributes to green growth in the short and long run. We reason that this is because solid institutions provide a stable and predictable environment and foster investment in sustainable technologies and practices. Strong institutions also enforce compliance with environmental regulations and sustainability programs. We also show that green innovation and renewable energy consumption have a negative and insignificant effect on green growth in the short term. In contrast, their long-term effect is positive and statistically significant. The short-term negative impact could be due to the entailed upfront investment and infrastructure demands. Yet, investment in green innovation and renewable technology production should be reinforced with a long-term perspective.

The control variables that we utilize also exhibit interesting results. Trade openness has a negative and insignificant effect on green growth in the short run but a positive and significant impact in the long run. This might be because of the initial specialization of highly polluting industries in African countries due to less stringent environmental measures. As exporters are exposed to stricter environmental standards imposed by major importers over time, they adopt cleaner production methods. The effect of FDI on green growth in the short run is negative and insignificant, while the negative impact is significant in the long run. This might again be due to the loose environmental regulations in African countries, which allow highly polluting foreign investments. Natural resource rent adversely impacts green growth and is statistically significant throughout the short and long run. The effect of GDP per capita on green growth is positive and statistically significant in the short and the long run. Finally, population growth exhibits a negative and statistically insignificant effect on green growth in the short run but a positive and statistically significant impact in the long run.

The rest of the paper is structured as follows: Section 2 presents the literature review. Section 3 addresses the methods and data. Section 4 provides the empirical results and discussion. Section 5 concludes the results with some policy options.

2. Literature Review

2.1. Green Innovation and Green Growth

Ref. [23] challenged the conventional view that there is a tradeoff between environmental protection and economic competitiveness. The authors argue that this approach assumes a static world where environmental regulations would increase firms’ costs that already established their cost-minimizing production plans. However, according to the authors, the new paradigm of international competitiveness is based on the firm’s ability to improve and innovate continually. Hence, they suggest that properly drafted and more stringent environmental regulations can stimulate innovation and enhance firms’ competitiveness. Similarly, ref. [24] argues that ecological innovation can serve sustainability by reconciling economic growth and environmental preservation at the firm level. While securing efficient use of natural resources and decreasing environmental damage, ecological innovation reinforces economic growth by increasing firm competitiveness.

In this vein of research, ref. [25] developed an endogenous growth model where clean and dirty technologies compete in production and innovation. The authors show that while research may be directed to either type of technology, the more advanced the dirty technology, the more difficult it is to move toward clean technologies. Accordingly, the authors suggest that research subsidies toward clean technologies would be optimal for welfare improvements. The results of [25] also imply the importance of institutions in green growth. As they would not allow the implementation of dirty technologies, stronger institutions would contribute to green growth in the short and the long run.

On the other hand, ref. [26] notes that energy efficiency improvement may save less energy than expected due to rebound or backfire effects of energy use. The rebound effect refers to less than one-for-one energy correspondence between energy efficiency gains and reduced energy use. In contrast, backfire relates to cases where energy efficiency gains increase energy use. The authors develop a dynamic general equilibrium macroeconomic model to identify the macroeconomic effects of environmental rebound and backfire effects. They find that in terms of pollution, the macroeconomic rebound effect for the economy may be substantial in several situations, such as when the elasticity between clean and dirty energy consumption is lower or the environmental efficiency of clean energy is lower.

Considering these competing theoretical predictions, several researchers have empirically analyzed the relationship between green innovation and green growth. For instance, ref. [7] assessed the effect of green development on economic expansion using data from 32 OECD countries from 1990 to 2013. The authors find that sustainable technology transfer and innovation positively contribute to green growth and economic development. Specifically, the authors suggest that innovations and technology contribute to green growth by supporting environmentally friendly transportation, promoting clean consumption, and optimizing corporate supply chains.

Ref. [27] investigated the impact of innovation on economic development in developing countries from 1990 to 2018. The authors find that increased technological innovation leads to short-term and long-term economic growth, and innovation is crucial in attaining sustainable development in developing countries. Ref. [28] analyzed the effect of green innovation on the green economy in the BRICS (Brazil, Russia, India, China, and South Africa) economies from 1992 to 2016. Their findings demonstrate that innovation in environmental technologies is crucial in promoting green growth. Ref. [29] investigated the effects of green innovation and financial globalization on green growth in the BRICS economies. Their findings also support that green innovations are instrumental in attaining green growth.

While green innovation is usually found to contribute to economic development, the results may vary with the type of innovation (or the rebound and backfire effects of innovations as noted above) and the economic structure of the countries examined. For instance, ref. [30] shows that technology patent applications for renewable energy positively affected emissions in emerging Asian economies from 1990 to 2019. The authors suggest that the positive relation might be because the innovation of new technologies is concerned with increasing energy production rather than making a greener environment. Ref. [31] finds that while technological innovation reduces carbon emissions in the US, the effect is not statistically significant in China. Given such asymmetries, we take a holistic approach in our analysis and study the positive and negative aspects of green innovation on green growth in Africa.

2.2. Renewable Energy and Green Growth

Transitioning to renewable energy sources is critical in mitigating climate change and fostering sustainable development. Unlike fossil fuels, which contribute significantly to greenhouse gas emissions, renewable sources such as solar, wind, and geothermal power offer clean and sustainable energy alternatives. Beyond their environmental benefits, renewable energy sources also hold the potential to stimulate economic growth as they are cost-effective, provide energy security, and help alleviate poverty by extending electricity to remote areas due to their decentralized nature [32]. In this regard, studies have extensively explored the relationship between renewable energy and sustainable development using various theoretical and empirical frameworks.

Theoretically, two prominent frameworks have been employed to analyze the nexus between renewable energy and economic growth: the environmental Kuznets curve (EKC) hypothesis and endogenous growth theory. The EKC hypothesis postulates an inverted U-shaped relationship between economic growth and ecological degradation [33]. This implies that environmental degradation initially increases with economic growth, as economies prioritize rapid development and rely heavily on traditional energy sources with significant ecological consequences. However, as economies mature and reach a certain level of development, environmental awareness and technological advancements shift towards more sustainable practices, including adopting renewable energy sources. This transition contributes to a decline in environmental degradation, resulting in the descending part of the EKC curve.

In endogenous growth theory, technological development occurs either as a byproduct of the production process or as a consequence of intentional research and development activities [34]. This theory implies that renewable energy can promote economic growth by enhancing the economy’s productivity and efficiency and creating new markets and opportunities for innovation.

Regarding the effects of renewable energy production on economic growth, the International Energy Agency (IEA) found that renewable energy could create 28 million new jobs worldwide by 2050 [35]. Ref. [36] surveyed the studies on renewable energy and economic growth, revealing a positive and significant relationship between the two variables in developing countries. Ref. [37] identified a positive correlation between renewable energy consumption and resource efficiency across 123 industrialized and developing nations. Ref. [8] showed that higher levels of renewable energy consumption were associated with increased economic growth rates in South Asian countries from 2000 to 2018.

While most studies support the positive contribution of renewable energy to green growth, a few studies have presented contrasting findings. For instance, ref. [38] noted no causal relationship between renewable energy consumption and economic growth in 27 European countries between 1997 and 2007. Ref. [39] suggested that renewable energy may impede economic growth due to high capital investment and limited technological advancements. However, the author argued that this negative impact will diminish as the economy advances. Ref. [40] found that increased use of combustible renewables led to higher carbon emissions in five North African nations between 1980 and 2011, potentially hindering green growth. Ref. [41] cautioned about the potential environmental consequences of large-scale renewable energy adoption, such as land use changes, resource extraction impacts, and manufacturing waste. Given the inconclusive empirical results, we aim to identify the relationship between renewable energy and green growth in the African context.

2.3. Institutional Quality and Green Growth

Strong institutions enable the establishment of comprehensive environmental regulations and catalyze the widespread adoption of cleaner technologies, reducing pollution emissions and safeguarding natural resources [42,43]. They promote sustainable resource management by delineating property rights and instituting governance frameworks that discourage overexploitation [44,45]. Moreover, institutions initiate innovation in green technologies by providing financial support, formulating supportive policies, and fostering collaboration between research institutions and businesses [46].

A growing body of empirical research has delved into the influence of institutional quality on green growth progress. For instance, ref. [37] argued that institutional dimensions such as government effectiveness and the rule of law could be crucial in promoting environmental quality and green growth in developing countries. Ref. [47] found that institutional quality and human capital were instrumental in reducing the ecological footprint in emerging countries between 1984 and 2017. Similarly, ref. [43] revealed that the quality of institutions plays a vital role in enhancing environmental quality through the effective implementation of economic and environmental policies, including the provision of supportive financial mechanisms. Ref. [46] investigated the role of environmental regulation on green total factor productivity (GTFP) in China’s manufacturing industry from 2003 to 2016 and found that environmental regulation positively affects GTFP.

While there is increasing evidence linking institutional quality and green growth, some studies have found that strong institutions may not always result in positive outcomes. For instance, ref. [21] examined the impact of institutional quality on carbon emissions in 25 African countries and found that institutional quality can positively impact carbon emissions. Nevertheless, the authors note that the lack of stringent environmental regulations or the failure to enforce existing regulations may increase emissions. Ref. [48] find mixed impact of institutional quality on green economic growth in West African Economic and Monetary Union countries. Specifically, the study finds a positive effect of institutional quality on green economic growth in Côte d’Ivoire, Mali, Niger, Senegal, and Togo. However, the effect is negative in Benin and Burkina Faso.

In summary, while there is growing evidence of the link between institutions and green growth, more studies are needed to understand the specific mechanisms through which institutions influence green growth, the effects of different types of institutions and institutional reforms, and the factors that contribute to the variation in the impact of institutions across countries and regions. Additionally, there is a need for focused research on the empirical relationship between institutional quality and green growth in African countries.

3. Data and Modeling

3.1. Data

This paper studies the determinants of green growth in 49 African countries (see Appendix B for the list of countries) from 2000 to 2021. The countries and periods are chosen based on data availability. The dependent variable, green growth, takes into consideration developmental, social, and environmental aspects of growth and, as such, decouples economic growth from resource consumption and harmful environmental impacts [6,7]. Given the limited availability of data for Africa, we develop the green growth data employing the framework proposed by [6,22] that take into account the monetary value of gross domestic product (GDP), education expenditure, fossil fuel consumption, forest depletion, and carbon damage, which is represented in the following equation:

$${GG}_t=GD{P}_t+{EDU}_t-NR{D}_t-NF{D}_t-{CO}_{2\mathrm{t}}$$

(1)

where ${GG}_t$ represents green growth, $GD{P}_t$ indicates gross domestic product, ${EDU}_t$ indicates education expenses, $NR{D}_t$ represents natural resource depletion for the depletion of minerals such as coal, crude oil, and natural gas, $NF{D}_t$ denotes forest depletion, and ${CO}_{2\mathrm{t}}$ is carbon damage. The calculation of the monetary value of the carbon damage variable relies on the World Bank convention that employs a fixed unit damage cost of USD 40 per ton of carbon emitted in 2020, measured in 2017 USD (https://databank.worldbank.org/metadataglossary/world-development-indicators/series/NY.ADJ.DCO2.GN.ZS; accessed on 3 May 2023. This unit cost is a standardized metric for the economic damage associated with each ton of carbon.

Various studies have indicated that technological innovation, the adoption of renewable energy sources, and institutional quality, either together or separately, play a crucial role in reducing carbon emissions and fostering economic development [8,21,31,47]. This study investigates how these variables synergistically contribute to promoting a green economy in African countries. Following the convention in the literature, we use the share of renewable energy in total final energy consumption for the renewable energy variable. Several studies have utilized environmental-related technologies as a measure of green innovation. However, these data are available only for a few African countries. Therefore, we follow [21] and use total patents as a proxy for green technological innovation.

Regarding institutional factors, we utilize the Worldwide Governance Indicators (WGI) developed by [49], which encompass the following six dimensions of governance measures: voice and accountability, political stability and absence of violence/terrorism, government effectiveness, regulatory quality, rule of law, and control of corruption. However, the high correlation among these indicators introduces multicollinearity concerns when included as separate variables in a model. We adopt the methodology outlined in [37] to address this concern. We employed principal component analysis (PCA) to construct a composite index. The composite index accurately portrays institutional quality by transforming correlated variables into uncorrelated components and effectively mitigates issues related to multicollinearity. The results of the PCA are presented in

Table 1. PCA regression for the institutional quality variable.

Component	Eigenvalue	Difference	Proportion	Cumulative	Variables
1	4.79405	4.32816	0.7990	0.7990	Regulatory Quality
2	0.465888	0.105333	0.0776	0.8767	Government Effectiveness
3	0.360555	0.160399	0.0601	0.9367	Control of Corruption
4	0.200157	0.100561	0.0334	0.9701	Rule of law
5	0.0995957	0.0198444	0.0166	0.9867	Voice and Accountability
6	0.0797513	.	0.0133	1.0000	Political Stability
Kaiser–Meyer–Olkin measure of sampling adequacy					0.8927

Source: Authors’ compilation from WGI.

. The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy stands at 89%, surpassing the recommended 50% threshold, indicating the adequacy of our sample for PCA. Furthermore, Component 1 is selected as the principal based on its highest eigenvalue, suggesting that it explains the most variance in the data. As a result, various indicators of institutional quality are amalgamated into a singular indicator.

We control several variables that are known to impact green growth. Economic activity, a significant driver of environmental quality [21], is controlled using GDP per capita. This metric is crucial for sustainable economic development, as it is often achieved through increased productivity and infrastructure investments. However, heightened economic activity typically correlates with greater energy consumption, which can have adverse environmental effects [50]. Additionally, we incorporate FDI and trade to control globalization factors. These variables can facilitate technology transfer and sustainable practices promoting investments in renewable energy and eco-friendly infrastructure [51,52]. However, multinational companies may relocate pollution-intensive operations to developing countries, potentially hindering local environmental progress. We also account for population size and natural resource rents following insights from [37]. These factors help us mitigate the impacts of growing human activities and resource extraction on green growth. Therefore, our comprehensive approach ensures the robustness and reliability of our findings by rigorously controlling these variables.

Table 2. Study variables: code, measurement, and sources.

Variables	Code	Measurement	Source
Green growth	GG	It is measured by subtracting the monetary value of minerals, deforestation, and carbon dioxide emissions from the sum of GDP and education expenditure.	[22]
GDP per capita	GDP	Constant 2015 USD	WDI
Trade	TOP	Imports + Export (% of GDP)	WDI
Foreign direct investment	FDI	Net inflows (% of GDP)	WDI
Institutional quality	INST	Index	WGI
Green innovation	INV	Total patent	OECD
Renewable energy consumption	REE	% of total	WDI
Population	POP	Millions	WDI
Natural resources rents	NRS	% of GDP	WDI

presents the variables used in the study, along with their respective codes, measurements, and sources.

We draw independent variables from various sources. Green innovation is proxied by the total patent data from the OECD database. Data for GDP per capita (constant 2015 USD), trade openness (reflecting the total share of exports and imports in GDP), foreign direct investment (expressed as net inflows as a percentage of GDP), renewable energy consumption as a percentage of total energy, population in millions, and natural resource rents as a percentage of GDP, are extracted from the World Development Indicators (WDI) database.

3.2. Econometric Models

This paper examines the role of green innovation, renewable energy, and institutional quality on green growth, considering both short-term and long-term effects. The proposed green growth function, which represents the relationship between the dependent and independent variables, is as follows:

$${GG}_{it}=f\left({GDP}_{it},{TOP}_{it},{FDI}_{it},{INST}_{it},{INV}_{it}{,REE}_{it},{POP}_{it},{NRS}_{it}\right)$$

(2)

where $GG$ stands for green growth, $GDP$ per capita measures economic growth, $TOP$ measures trade openness, $FDI$ captures foreign direct investment, $INST$ assesses institutional quality, $INV$ focuses on green innovation, $REE$ is renewable energy, $POP$ denotes population, and $NRS$ stands for natural resource rent. To address concerns related to heteroscedasticity, a natural logarithm transformation is applied to all parameters except for $INST$ and $FDI$, which have small values. This transformation mitigates heteroscedasticity problems and allows the construction of an augmented multivariate function within the panel framework.

The functional form of the transformed model is as follows:

$$\begin{array}{c}{lnGG}_{it}={\alpha }_0+{\alpha }_1{lnGDP}_{it}+{\alpha }_2{lnTOP}_{it}+{\alpha }_3{FDI}_{it}+{\alpha }_4{INST}_{it}+{\theta }_5{lnINV}_{it}+{\alpha }_6{lnREE}_{it} \\ +{\alpha }_7{lnPOP}_{it}+{\alpha }_8{lnNRS}_{it}+{\epsilon }_{it}\end{array}$$

(3)

where ${\alpha }_0$, …., ${\alpha }_8$ signify the green growth parameter regarding $GDP$, $TOP$, $FDI$, $INST$, $INV$, $REE$, $POP,$ and $NRS$, respectively. The subscript $i$ corresponds to individual cross-section countries, while the subscript $t$ indicates periods ranging from 2000 to 2021.

3.3. Modeling Approach

Figure 1 illustrates the schematic representation of the econometric approach employed in this study. Initially, a cross-sectional dependency (CD) test is executed to gauge variable dependencies. Subsequently, the [53] test is used to discern whether the model demonstrates heterogeneity or homogeneity. A unit root test is executed to investigate the stationarity of panel data. Moving forward, a panel co-integration test is implemented to examine the potential existence of long-term relationships among the variables. The econometric approach further employs the pool mean group (PMG), mean group (MG), and dynamic fixed effects (DFE) tests to gauge both short- and long-term impacts of the variables.

To fortify the estimation’s robustness, additional scrutiny is applied through the generalized method of moments (GMM), fixed effects (FE), and augmented mean group (AMG) tests. Finally, a long-run relationship test is conducted using the Dumitrescu–Hurlin panel causality test.

3.4. Cross-Sectional Dependency

The CD is the correlation between the error terms of different cross-sectional units in a panel dataset [54]. CD is a common problem in panel data analysis. This is particularly true for African countries due to their similar socioeconomic and geographic characteristics. If CD is present, it can bias the results of panel data estimators. Therefore, to address the potential problems associated with CD, the first step in our analysis is to test the null hypothesis of no CD against the existence of CD. The present study adopts the CD test introduced by [55].

$$CD=\sqrt{{\displaystyle \frac{2T}{N(N-1)}}\left(\sum _{i=1}^{N-1} \sum _{j=i+1}^N {\widehat{\rho }}_{ij}^2\right)}\to {N\left(\mathrm{0,1}\right)}_{i,j}$$

(4)

$$\mathrm{C}\mathrm{D}=\mathrm{1,2},3\dots ..52..\mathrm{N}$$

(5)

$$CD=\sqrt{{\displaystyle \frac{2T}{N(N-1)}} \left(\sum _{i=1}^{N-1} \sum _{j=i+1}^N {\widehat{\rho }}_{ij}\right)}{\displaystyle \frac{\left( T-K\right){\widehat{\rho }}_{ij}^2-E(T-K){\widehat{\rho }}_{ij}^2}{Var(T-K){\widehat{\rho }}_{ij}^2}}$$

(6)

where ${\widehat{\rho }}_{ij}^2$ is the correlation coefficient between residuals.

3.5. Slope Homogeneity

Following the CD test, the study scrutinizes the homogeneity of slope coefficients using the [53] slope heterogeneity test. This test is favored over the seemingly unrelated regression equation (SURE) method because it eliminates the possibility of cross-sectional dependency [56]. The homogeneity model for this test is formulated as follows:

$${∆}_{hp}={\left(\mathrm{N}\right)}^{1/2}{\left(2\mathrm{K}\right)}^{-1/2}\left({\displaystyle \frac{1}{\mathrm{N}}}\mathrm{S}-\mathrm{K}\right)$$

(7)

$${∆}_{hp/I}={\left(N\right)}^{1/2}{\left({\displaystyle \frac{2K(T-K-1}{t+1}}\right)}^{-1/2}\left({\displaystyle \frac{1}{N}}S-K\right)$$

(8)

${∆}_{hp}$ indicates the tilde delta, while ${∆}_{hp/I}$ signifies the adjusted tilde delta.

3.6. Unit Root Test

After testing CD and slope heterogeneity, the next stage is to test the cointegration order of the relevant variables. In the presence of CD, first-generation unit root tests may incorrectly reject the null hypothesis and assume independence. This can lead to misleading inferences, as the tests may inaccurately indicate stationarity while ignoring the effects of CD [57]. To overcome these limitations, the present study adopted a second-generation unit root test of “Cross-Sectionally Augmented Dicky-Fuller (CADF) and the Cross-Sectionally Augmented IPS (CIPS) tests” by [54]. CADF and CIPS tests are designed to address CD and data stationarity. Unlike their first-generation counterparts, these tests account for the data’s possible interdependencies or cross-sectional dependence through parametric and nonparametric approaches [58]. The CADF and CIPS tests operate under the null hypothesis of a homogeneous unit root for all panel members. Nonetheless, the alternative hypothesis posits the existence of at least one stationary panel member. The CIPS equation is as follows:

$$\mathsf{\Delta }{Y}_{i,t}={\delta }_i+{\delta }_i{x}_{i,t-1}+{\delta }_i{\overline{x}}_{t-1}+\sum _{l=0} {\delta }_{il}\mathsf{\Delta }{\overline{Y}}_{t-1}+\sum _{l=0} {\delta }_{il}\mathsf{\Delta }{Y}_{i,t-1}+{\mu }_{it}$$

(9)

The subsequent CIPS statistics can be calculated:

$$\mathrm{C}\mathrm{I}\mathrm{P}\mathrm{S}=N^{-1}\sum _{i=1}^{N-1} t_i\left(N,T\right)$$

(10)

3.7. Cointegration Test

Cointegration analysis has gained increasing attention in recent years, particularly in panel data models. While it is unnecessary to test for the presence or absence of cointegration in estimating a panel ARDL model, the presence of cointegration can improve the accuracy and reliability of the model’s results [57]. We employ [59] cointegration test, designed to be robust against panel heterogeneity and CD. These tests examine cointegration without typical factor constraints and verify the presence of error correction for integrated series at order $0$ or $1$ by using two-panel means ($p_t$ and $p_a$) and two-group means ($G_t$ and $G_a$) [58]. The panel means test indicates whether the entire panel is cointegrated, while the group means test determines if at least one element exhibits cointegration. Acceptance of the alternative hypothesis indicates evidence of long-term cointegration among the variables. The following equation represents [59] test specifications:

$$∆{GG}_{it}={\alpha }_{0i}+\sum _{i=l}^p{\alpha }_i∆{GG}_{I t-i}+\sum _{i=l}^r{\phi }_i∆Y_{I t-i}+{\lambda }_i{ECM}_{I t-i}+{\mu }_{i,1}$$

(11)

where $Y$ represents the vector of independent variables and ${\lambda }_{i }$ denotes the error term adjustment speed. If ${\lambda }_i=0$, the variables are not cointegrated, whereas if ${\lambda }_i<0$, the variables are cointegrated.

The test statistics are:

$$G_t={\displaystyle \frac{1}{N}}\sum _{i=l}^{\mathrm{N}}{\displaystyle \frac{{\alpha }_{\mathrm{i}}}{SE\left({\alpha }_{\mathrm{i}}\right)}}$$

(12)

$$G_a={\displaystyle \frac{1}{N}}\sum _{i=l}^{\mathrm{N}}{\displaystyle \frac{T\alpha i}{\alpha i\left(1\right)}}$$

(13)

$$P_t={\displaystyle \frac{\alpha }{SE\left(\alpha \right)}}$$

(14)

$$P_a=T\alpha$$

(15)

3.8. Estimation Technique

Once it is established that all variables are integrated, the subsequent step involves computing the long-run and short-run coefficients using an appropriate method. In this study, we employ the autoregressive distributed lag (ARDL) approach applied within the framework of “maximum likelihood estimation (MLE)” as developed by [60]. The panel ARDL estimator offers three different methods for estimation: the mean group (MG) estimator, the pooled mean group (PMG) estimator, and the dynamic fixed effect (DFE) estimator. These estimators address the issue of heterogeneity bias in dynamic panels by allowing for differences in parameter estimates across countries. Notably, these estimators provide accurate forecasts for variables integrated at order 0 or 1 and deal well with CD and slope heterogeneity issues. While the panel ARDL addresses endogeneity bias and serial correlation by incorporating sufficient lags of independent and dependent variables, prior research [61,62] have demonstrated that these estimators maintain validity, efficiency, and consistency even in the presence of nonstationary and endogeneity concerns, outperforming other models.

The MG estimator is designed to account for heterogeneity across countries in both the short-run and long-run parameters [63]. It is less restrictive, and it computes individual regressions for each country in the panel. It derives the long-run parameters from the ARDL model and allows for slope and intercept variation across countries. This makes the MG estimator suitable when significant heterogeneity exists among the subject countries, allowing for more accurate conclusions about individual country dynamics. The PMG estimator imposes homogeneity on the long-run parameters while allowing for heterogeneity in the short-run [57]. It works by combining pooling and averaging coefficients, suitable under the assumption that similar long-run impacts occur across all countries within the panel [64]. The DFE estimator is an upgraded version of the PMG estimator that assumes uniformity in both the short run and long run across all countries in the panel [57]. The choice of estimator depends on the results of the Hausman test.

The panel ARDL estimation equation is specified as follows:

$$\begin{array}{cc}\hfill ∆{LNGG}_{it}={\alpha }_i+& {\alpha }_{1i}{LNGG}_{it-1}+ {\alpha }_{2i}{LNGDP}_{I,-1}+{\alpha }_{3i}{TOP}_{I,-1}+{\alpha }_{4i}{FDIgdp}_{I,-1}\hfill \\ & +{\alpha }_{5i}{INST}_{I,-1}+{\alpha }_{6i}{LNINV}_{I,-1}+{\alpha }_{7i}{REE}_{I,-1}+{\alpha }_{7i}{LNPOP}_{I,-1}\hfill \\ & +{\alpha }_{8i}{NRS}_{I,-1}+{\displaystyle \sum _{j=1}^{P_i}} {\beta }_{1j}\Delta {LNGGDP}_{it-1}+{\displaystyle \sum _{j=1}^{P_2}} {\beta }_{2j}\Delta {LNGDP}_{I,-1}\hfill \\ & +{\displaystyle \sum _{j=1}^{P_3}} {\beta }_{3j}\Delta {TOP}_{I,-1}+ {\displaystyle \sum _{j=1}^{P_4}} {\beta }_{4j}\Delta {FDI}_{I,-1} \hfill \\ & + {\displaystyle \sum _{j=1}^{P_5}} {\beta }_{5j}\Delta {INST}_{I,-1} + {\displaystyle \sum _{j=1}^{P_6}} {\beta }_{6j}\Delta {LNINV}_{I,-1}\hfill \\ & + {\displaystyle \sum _{j=1}^{P_7}} {\beta }_{7j}\Delta {REE}_{I,-1} + {\displaystyle \sum _{j=1}^{P_8}} {\beta }_{8j}\Delta {LNPOP}_{I,-1}\hfill \\ & + {\displaystyle \sum _{j=1}^{P_9}} {\beta }_{9j}\Delta {NRS}_{I,-1}+e_{it} \hfill \end{array}$$

(16)

where $∆$ represents the first differences, ${\alpha }_i$ is a constant, ${\beta }_{nij}$ (n = 1, …, 9) represents short-run coefficients, ${{\alpha }_{\mathrm{m}}}_i$ $(m=1,\dots ,9)$ shows long-run coefficients, and $e_{it}$ denotes the error term. The optimal lag orders for the first differencing are determined using the Schwarz and Akaike information criterion. By including error correction terms, the equation reads:

$$\begin{array}{c}\Delta {LNGG}_{it}\hfill \\ = {\lambda }_i+{\displaystyle \sum _{j=1}^{P_i}} {\beta }_{1j}\Delta {LNGG}_{it-1}+{\displaystyle \sum _{j=1}^{P_2}} {\beta }_{2j}\Delta {LNGDP}_{I,-1}+ {\displaystyle \sum _{j=1}^{P_3}} {\beta }_{3j}\Delta {TOP}_{I,-1} \hfill \\ + {\displaystyle \sum _{j=1}^{P_4}} {\beta }_{4j}\Delta {FDI}_{I,-1} + {\displaystyle \sum _{j=1}^{P_5}} {\beta }_{5j}\Delta {INST}_{I,-1} + {\displaystyle \sum _{j=1}^{P_6}} {\beta }_{6j}\Delta {LNINV}_{I,-1} \hfill \\ + {\displaystyle \sum _{j=1}^{P_7}} {\beta }_{7j}\Delta {REE}_{I,-1} + {\displaystyle \sum _{j=1}^{P_8}} {\beta }_{8j}\Delta {LNPOP}_{I,-1} + {\displaystyle \sum _{j=1}^{P_9}} {\beta }_{9j}\Delta {NRS}_{I,-1}\hfill \\ +{\mathsf{\mu }}_i{ECT}_{it,-1}\hfill \\ +e_{it}\hfill \end{array}$$

(17)

where ${\mathrm{E}\mathrm{C}\mathrm{T}}_{it,-1}$ is the error correction term.

3.9. Robustness Check Test

To check the robustness of the models employed in this study, three methodological approaches are applied: the generalized method of moments (GMM), fixed effects (FE), and the augmented mean group (AMG) method. The selection of these methods is motivated by their versatility in addressing various challenges inherent in empirical analyses. Primarily, these methodologies prove helpful in scenarios marked by heterogeneity, no stationarity, endogeneity, and cross-section dependence. Their robustness allows for effectively modeling diverse individual characteristics and mitigating potential endogeneity challenges. Moreover, these methods are well suited to address correlations, especially among cross-sections. By incorporating GMM, FE, and AMG, the study aims to enhance the reliability of its analysis by systematically tackling issues related to heterogeneity and correlation among variables.

3.10. Causality Test

This study employs the panel Granger causality test introduced by [65] to investigate the cause-and-effect dynamics among variables and their respective directions. This test proves especially useful when error terms exhibit cross-sectional dependence. The DH Granger causality test assumes a null hypothesis stating the absence of a causal relationship between variables. In contrast, the alternative hypothesis suggests the presence of a causal link among the variables.

4. Results and Discussion

4.1. Descriptive Statistics

Table 3. Descriptive statistics.

Variables	LNGG	LNGDP	LNTOP	FDI	INST	LNINV	LNREE	LNPOP	LNNRS
Obs	1078	1078	1078	1078	1078	1078	1078	1078	1078
Mean	2.251	7.217	3.866	4.262	−1.1	3.461	3.717	1.595	1.87
SD	1.64	1.004	0.485	7.819	2.19	0.486	1.243	1.595	1.542
Min	−2.526	5.542	2.055	−26.64	−1.8	1.609	−2.813	11.29	6.749
Max	6.256	9.740	5.234	86.99	1.9	4.357	4.588	19.17	4.484
Skewness	0.037	0.539	0.044	5.244	0.447	−1.012	−2.426	−0.707	−2.409
Kurtosis	2.791	2.411	2.891	44.541	2.725	4.269	9.378	3.302	11.766
Jarque–Bera	2.40	63.48	0.80	879.83	31.71	128.05	420.59	63.46	444.10
Probability	0.3010	0.0000	0.6689	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000

presents a comprehensive summary of descriptive statistics for the variables employed in this study. The mean logarithm of green growth (LNGG) is 2.251, with a corresponding standard deviation of 1.64. FDI displays the highest standard deviation, while LNTOP registers the lowest value.

The criteria for a series conforming to a normal distribution involve a skewness value of 0 and a kurtosis value of 3 [66]. Upon analyzing specific variables, LNGG, LNGDP, LNTOP, FDI, and INST exhibit positive skewness, indicating a right-leaning tendency compared to a normal distribution. Conversely, LNINV, LNREE, LNPOP, and LNNRS show negative skewness, suggesting a left-leaning distribution. Furthermore, the kurtosis values for LNGG, LNGDP, LNTOP, and INST are below 3, indicating a platykurtic distribution characterized by a flatter peak. In contrast, FDI, LNINV, LNREE, LNPOP, and LNNRS display kurtosis values exceeding the standard threshold, signifying a leptokurtic distribution with heavier tails. Considering both kurtosis and skewness values, none of the variables meet the conditions for a normal distribution. Additionally, evidence from the Jarque–Bera test supports the conclusion that, except for LNGG and LNTOP, none of the series is normally distributed, as the null hypothesis of normality is rejected for the remaining variables.

Table 4. Cross-sectional dependence and slope homogeneity test.

Variables	CD-Test	Correlation	Slope Homogeneity Test
			[53]
LNGG	115.021 ***	0.72	Test Statistics
LNGDP	113.515 ***	0.71	Delta	9.042 ***
LNTOP	19.667 ***	0.12	Adj. Delta	13.479 ***
FDI	10.704 ***	0.07
INST	0.332	0.00
LNINV	47.264 ***	0.29
LNREE	64.434 ***	0.40
LNPOP	157.882 ***	0.98
LNNRS	40.052 ***	0.25

Note: *** denotes statistical significance at 1%.

displays the findings of the CD test. The result shows that the null hypothesis of no CD is rejected for all variables, indicating that the variables in the panel dataset are interdependent. This implies that an economic shock occurring in one country can spill over to the economies of other countries within the region. The slope heterogeneity test outcomes also reveal statistically significant values for delta and adjusted delta. These results indicate the presence of a slope heterogeneity issue. Consequently, CD and slope heterogeneity highlight the need for second-generation unit root tests to assess variable stationarity in panel data.

Table 5. Stationarity: CADF and CIPS test results.

Variables	CADF		CIPS		Integration Order
	Level	First Difference	Level	First Difference
LNGG	−0.725	−2.462 ***	−0.983	−3.176 ***	$l\left(1\right)$
LNGDP	−0.845	−2.378 ***	−1.185 **	−3.372 ***	$l\left(0\right)$
LNTOP	−1.420	−2.535 ***	−1.3 **	−4.443 ***	$l\left(1\right)$
FDI	−1.839 **	−2.500 ***	−2.478 ***	−5.118 ***	$l\left(0\right)$
INST	−1.829 **	−2.090 ***	−0.900	−3.883 ***	$l\left(0\right)$
LNINV	−2.125 ***	−2.828 ***	−1.942 ***	−4.494 ***	$l\left(0\right)$
LNREE	−1.962 ***	−-2.188 ***	−1.462 *	−3.853 ***	$l\left(0\right)$
LNPOP	−2.253 ***	−2.523 ***	−0.519	−2.230 ***	$l\left(0\right)$
LNNRS	−2.227 ***	−3.190 ***	−2.258 ***	−4.021 ***	$l\left(0\right)$

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

presents the results of the CADF and CIPS tests, wherein lagged variables are incorporated to address potential serial correlations. The findings indicate that all observed variables exhibit stationarity at the mixed level. Specifically, while some variables are stationary in their original form, all variables demonstrate stationarity when differenced once. Consequently, the null hypothesis of nonstationary series is rejected, indicating that these series possess an integration order of either 0 or 1. Therefore, using the ARDL method for estimating the underlying model does not result in misleading regression outcomes, as none of them are integrated of order 2.

The Westerlund cointegration test results in

Table 6. Cointegration test results.

	T Statics	p-Value	Decision
Gt	−1.6928 **	0.0453	Cointegration
Ga	−2.3388 ***	0.0097	Cointegration
Pt	−1.9805 **	0.0238	Cointegration
Pa	−1.8169 **	0.0346	Cointegration

Note: *** and ** denote statistical significance at the 1% and 5% levels, respectively.

show that the study variables are cointegrated, indicating a long-run relationship. Therefore, it is essential to assess the short- and long-run impacts of the independent variables on green growth.

4.2. Findings and Discussion

The long-run and short-run relations among variables are estimated through subsequent cointegration analysis.

Table 7. PMG, MG, and DFE results (1,1,1,1,1,1,1,1,1).

VARIABLES	PMG	MG	DFE
Short-Run Impact
$\mathsf{\Delta }$LNGDP	0.198 ***	0.0261	0.00958 ***
	(0.0417)	(0.0755)	(0.000471)
$\mathsf{\Delta }$LNTOP	−0.000662	0.000457	−0.00151 *
	(0.000697)	(0.00129)	(0.000914)
$\mathsf{\Delta }$FDI	0.000393	0.00574	7.08 × 10−05
	(0.00178)	(0.00498)	(0.00140)
$\mathsf{\Delta }$INST	0.0390 **	0.00635	−0.00492
	(0.0183)	(0.0278)	(0.0301)
$\mathsf{\Delta }$LNINV	−0.00690	0.0691	−0.00501
	(0.0416)	(0.121)	(0.0452)
$\mathsf{\Delta }$LNREE	−0.0501	−0.0431	−0.150
	(0.158)	(0.276)	(0.0964)
$\mathsf{\Delta }$LNPOP	1.331	24.04 **	−1.128
	(3.236)	(9.666)	(1.086)
$\mathsf{\Delta }$LNNRS	−0.0900 ***	−0.0725 *	−0.0736 **
	(0.0228)	(0.0389)	(0.0294)
Short run adjustment	−0.0642 ***	−0.620 ***	−0.131 ***
	(0.0166)	(0.139)	(0.0320)
Long-Run Impact
LNGDP	0.00515 ***	0.192 ***	0.00588 ***
	(0.000699)	(0.0491)	(0.00217)
LNTOP	0.00643 ***	0.00296	0.000109
	(0.00211)	(0.00510)	(0.00613)
FDI	−0.0135 ***	−0.0335	0.00383
	(0.00405)	(0.0458)	(0.0108)
INST	0.122 ***	0.0298	0.177
	(0.0295)	(0.0818)	(0.109)
LNINV	0.234 ***	−0.489	0.289
	(0.0656)	(0.475)	(0.281)
LNREE	1.095 ***	1.802 *	0.556
	(0.228)	(1.049)	(0.392)
LNPOP	1.244 ***	1.767 *	0.514
	(0.171)	(0.947)	(0.707)
LNNRS	−0.169 ***	−0.0102	−0.186
	(0.0513)	(0.180)	(0.161)
Constant	−0.437 ***	−1.822	0.281
	(0.162)	(2.438)	(0.287)
No. Observations	1028	1028	1028
No. Countries	49	49	49
Housman test 1 (PMG or MG) 0.460 (prob> chi2 = 0.99)
Housman test 2 (PMG or DFE) 0.26 (Prob > chi2 = 0.753)

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

displays the ARDL model results with error corrections (p, q) using PMG, MG, and DFE. The findings suggest that the PMG estimator is the most appropriate model for estimating long- and short-term relationships among the variables of interest. The Hausman-1 and Hausman-2 test results support this. The error correction term (ECT) coefficient indicates the system’s tendency to approach its equilibrium level in the long run. Specifically, an ECT coefficient falling between 0 and −1 signifies convergence to the equilibrium level [57]. The ECT coefficient of −0.06 in the PMG results suggests that the system converges to its long-run equilibrium, implying that any deviations from equilibrium will be corrected over time.

The empirical results indicate that the coefficients of GDP per capita, institutional quality, and population are positive and statistically significant in the short and long run. However, the coefficient of natural resource rents is negative and statistically significant. Additionally, while the coefficients associated with green technological innovation, renewable energy consumption, and trade openness are positive in the long run, their short-run counterparts are negative. Conversely, the FDI coefficient is harmful in the long run, while its short-run counterpart is positive. Notably, all variables have a statistically significant impact on green growth in the long run. In contrast, only economic growth and institutional quality have a statistically significant effect on green growth in the short run.

Institutional quality has a positive and significant effect on green growth in the short and long term. This is likely because robust institutions provide a stable and predictable business environment and foster investment in sustainable technologies and practices. As a result, this enhances efficiency, reduces waste, and ultimately promotes higher levels of green growth. Moreover, strong institutions can enforce and ensure compliance with environmental regulations, thereby mitigating negative externalities associated with economic activities. High-quality institutions are also crucial in adopting and monitoring policies that reduce carbon emissions and encourage climate-friendly investments. By doing so, they contribute to cultivating a climate change adaptation culture and fostering sustainable development. This result aligns with the recent findings that institutional quality plays a significant role in determining green growth [13,47].

While green innovation has a negative and statistically insignificant short-run effect on green growth, its long-term impact is positive and statistically significant. This suggests that green innovation may not yield immediate benefits, but it is crucial in promoting sustainable development and achieving long-term environmental goals. The observed negative and insignificant impact of green innovation can likely be attributed to the time needed for new technologies and practices to be adopted and integrated into the economy. Additionally, green technologies often demand substantial upfront investments, and costs associated with implementing these innovations may discourage some companies from investing in them.

In the case of Africa, technological innovation gaps and lagging behind the technological frontiers hinder the ability of African firms to make innovations in green growth [21]. Nevertheless, as the benefits of green innovation become more apparent in time, significant improvements in environmental sustainability and economic growth will be realized. By promoting the development of zero-emission technologies through green patents and technological innovation, African countries can transform their industrial and manufacturing structures into more sustainable and environmentally friendly ones. This assertion is supported by the aforementioned empirical evidence [8,27] and the theory of environmental modernization, as proposed by [23].

The PMG result further reveals that long-term renewable energy is strongly associated with green growth, while the short-term impact is insignificant. The coefficient findings show that a 1% rise in renewable energy corresponds to a significant 1.16% annual growth in the green sector within the examined economy. This indicates a strong positive relationship between adopting renewable energy sources and promoting green growth. This compelling finding underscores a robust positive relationship between transitioning to renewable energy sources and promoting sustainable economic growth initiatives. Initially, implementing renewable energy may not immediately yield considerable green growth outcomes due to the entailed upfront investments, infrastructure developments, and the time required for materializing efficiency gains. However, in the long term, the benefits of renewable energy would unfold as production costs are reduced and environmental benefits are realized. In this regard, the findings of [67] show that certain African countries have recently embraced renewable energy as a viable alternative to conventional fossil fuels, contributing to favorable outcomes regarding green growth.

The control variables also exhibit significant effects on green growth. Trade openness has a negative and insignificant influence on green growth in the short run but a positive and significant impact in the long run. This finding contradicts the findings of [37], who found a negative and significant impact of trade on green growth in the short run for a sample of developed and developing countries. The negative short-run effect of trade openness on green growth may be due to the initial specialization of highly polluting export industries in African countries due to less stringent environmental regulations. This can increase pollution and environmental degradation, threatening ecosystems and green growth.

However, trade openness can positively affect green growth in the long run. African exporters are exposed to stricter environmental standards imposed by major importers over time, incentivizing them to adopt cleaner production methods. Trade also facilitates the exchange of knowledge and technology, introducing more efficient and environmentally friendly practices. These findings are consistent with those of [67], who conclude that trade openness positively and significantly impacts long-term growth in Africa.

While the effects of FDI on green growth in the short run are negative and insignificant, the negative effect is significant in the long run. This may be associated with the pollution haven hypothesis, which suggests that many foreign companies operating in African nations are highly polluting due to relaxed environmental regulations. Additionally, FDI in Africa often focuses on resource extraction and export-oriented industries, potentially causing ecological degradation and the depletion of natural resources. Additionally, multinational corporations may prioritize profits over sustainable practices, leading to further environmental harm. Our findings are aligned with those of [68].

Natural resource rent exerts an adverse impact on green growth, remaining statistically significant throughout the short and long run. This effect might be due to the limited economic diversification of African economies. Directing resources toward environmentally damaging extraction activities impedes the progress of sustainable and environmentally friendly sectors in the region. This study’s findings are aligned with the work of [69].

Our findings reveal a positive and statistically significant relationship between GDP per capita and green growth in the short and the long term. The positive impact of GDP per capita on green growth is linked to increased government spending on environmental protection, higher consumer demand for sustainable products, and improved environmental awareness in the short term. In the long term, economic growth drives technological innovation, economic transformation, and institutional development, all contributing to a cleaner and more sustainable economic model. It can also enable countries to implement policies and initiatives that promote sustainable development and address environmental challenges.

In contrast, population growth exhibits a negative and statistically insignificant effect on green growth in the short run but a positive and statistically significant impact in the long run. One possible explanation for this finding is that population growth in the short run may lead to increased resource consumption and environmental degradation, which can offset any potential benefits for green growth. However, population growth may stimulate innovation and technological advancements in the long run, leading to more sustainable and environmentally friendly practices, ultimately promoting green growth. The empirical evidence presented by [37] also supports the positive relationship between economic growth and green growth performance.

4.3. Robustness Check with Alternative Models

This study also considers two-step GMM, fixed effects, and AMG estimation to check the signs of variables and the robustness of the PMG approach. While the GMM method confirms the variables’ signs in the short run, the fixed effect and AMG methods ensure the coefficients’ signs in the long run.

Table 8. GMM, FE, and AMG results.

VARIABLES	GMM	FE	AMG
L.LNGGDP	0.567 ***
	(0.0418)
LNGDP	0.00309 ***	0.00496 ***	0.196 ***
	(0.000434)	(0.000339)	(0.0425)
LNTOP	0.00235 ***	−0.00157 *	−0.00163
	(0.000681)	(0.000940)	(0.00123)
FDI	−0.00402 ***	0.000362	0.00153
	(0.00142)	(0.00168)	(0.00310)
INST	0.101 ***	0.110 ***	0.0506 *
	(0.0242)	(0.0185)	(0.0282)
LNINV	−0.108	−0.177 ***	−0.0127
	(0.0700)	(0.0408)	(0.0402)
LNREE	−0.105 ***	−0.336 ***	0.315
	(0.0196)	(0.0659)	(0.488)
LNPOP	0.302 ***	2.411 ***	−0.198
	(0.0474)	(0.0764)	(0.157)
LNNRS	0.0252	0.0668 ***	−0.0889 ***
	(0.0205)	(0.0247)	(0.0242)
Constant	0.867 ***	−1.282 ***	0.555
	(0.275)	(0.363)	(2.290)
Number of ID	49	49	49
No. Instruments	31
AR (1)	0.000
AR (2)	0.653
Hansen Test	0.599
Observations	1029	1078	1078
R−squared		0.683

Note: *** and * denote statistical significance at 1%, and 10%, respectively.

presents the results of GMM, FE, and AMG estimations. The findings for most variables closely align with the PMG estimates, irrespective of specific coefficient values.

We also utilized Granger causality tests to investigate the long-term relationships between explanatory variables and African green growth. The results in

Table 9. Results of the Dumitrescu–Hurlin panel causality test.

Null Hypothesis:	W-Stat.	Zbar-Stat	Result	Conclusion
LNGG ≠ GDP	2.5371 ***	7.6084 ***	YES	LNGG causes LNGDP
GDP ≠ LNGG	2.616 ***	7.9986 ***	YES	LNGDP causes LNGG
LNGG ≠ LNTOP	2.7224 ***	8.5253 ***	YES	LNGG causes LNTOP
LNTOP ≠ LNGG	2.9931 ***	9.8654 ***	YES	LNTOP causes LNGG
LNGG ≠ FDI	1.79 ***	3.9103 ***	YES	LNGG causes FDI
FDI ≠ LNGG	2.2451 ***	6.1631 ***	YES	FDI causes LNGG
LNGGDP ≠ INST	1.762 ***	3.7715 ***	YES	LNGG causes INST
INST ≠ LNGG	4.104 ***	15.3641 ***	YES	INST causes LNGG
LNGG ≠ LNINV	2.5008 ***	7.4284 ***	YES	LNGG causes LNINV
LNINV ≠ LNGG	2.1931 ***	5.9053 ***	YES	LNINV causes LNGG
LNGGDP ≠ LNREE	1.4715 ***	2.3337 ***	YES	LNGG causes LNREE
LNREE ≠ LNGG	3.1527 ***	10.6554 ***	YES	LNREE causes LNGG
LNGG ≠ LNPOP	4.8883 ***	19.2459 ***	YES	LNGG causes LNPOP
LNPOP ≠ LNGG	15.9711 ***	74.1031 ***	YES	LNPOP causes LNGG
LNGG ≠ LNNRS	2.1386 ***	5.6359 ***	YES	LNGG causes LNNRS
LNNRS ≠ LNGG	2.3399 ***	6.6321 ***	YES	LNNRS causes LNGG

Note: *** denotes statistical significance at 1%.

indicate bidirectional causal relationships between the explanatory variables examined in the study and green growth. Therefore, it can be inferred that green innovation, institutional quality, renewable energy consumption, GDP per capita, trade openness, FDI, population, and natural resource rents affect green growth and vice versa. These findings are consistent with previous studies [8,70].

5. Conclusions and Policy Implications

By fostering economic prosperity while safeguarding the environment, green growth generates a holistic strategy for economic development. As such, green growth has developed as a critical paradigm for sustainable development. Therefore, understanding the factors that drive green growth is crucial. This study investigates the short- and long-term factors influencing green growth in Africa, focusing on institutional quality, green innovation, and renewable energy consumption. Additionally, this research considers the influence of GDP per capita, trade liberalization, FDI, population size, and natural resource rents. The robustness and reliability of the findings are ensured by addressing cross-dependence, slope heterogeneity, and long-run relationships between variables.

The finding shows that solid institutions are pivotal for fostering short- and long-term green growth. While green innovation exhibits a negative and statistically insignificant short-term effect, its long-term impact is positive and statistically significant, highlighting the need for long-term investment strategies. Similarly, long-term cost reductions that drive the use of renewable energy are positively related to green growth. However, its short-term impact could be adverse due to upfront investment and infrastructure demands. The impact of trade presents a complex picture. Short-term trade liberalization might contribute to environmental degradation, but long-term benefits emerge due to stricter environmental standards and knowledge transfer facilitated by trade. FDI exhibits a negative and significant long-term impact, aligning with the “pollution haven hypothesis”. Meanwhile, natural resource rents consistently hinder green growth across both timeframes. GDP per capita displays a positive and significant relationship with green growth in the short and long term. Despite exhibiting a negative and insignificant effect in the short term, population growth surprisingly shows a positive and significant impact in the long run, suggesting the potential for population pressure to stimulate innovation and sustainable practices, ultimately promoting green growth.

However, these general findings may have country-specific aspects. The variable “green growth” includes GDP, education spending, fossil fuel use, deforestation, and carbon impact, encompassing the most important aspects of green growth. While policymakers in lower-income countries like Ethiopia and Nigeria may prioritize increasing GDP, policymakers in higher-income countries such as Seychelles and Mauritius may be equally concerned about green growth’s social and environmental aspects. Therefore, future research could investigate non-linear models in GDP per capita and conduct cross-country comparative analysis to provide a more comprehensive understanding of green growth.

Based on our findings, the policy recommendations are summarized as follows. First, investment in robust and transparent institutions is crucial to creating a stable and predictable environment for long-term green growth strategies. This could involve strengthening regulatory frameworks, promoting good governance, and fostering public–private partnerships for sustainable development. Second, despite initial challenges, long-term investment in green innovation is crucial to bridging the innovation gap and promoting sustainable practices. Nevertheless, especially policymakers in lower-income countries should not rely on firms to invest in green innovation; instead, appropriate incentives such as tax breaks, grants, and public–private partnerships focused on green technologies should be fostered. Third, while acknowledging potential short-term challenges associated with upfront costs and infrastructure needs, prioritizing long-term investments in renewable energy infrastructure is essential. Targeted policies addressing cost concerns, such as subsidies or feed-in tariffs, can accelerate the transition. Fourth, promoting trade agreements that prioritize environmental sustainability is crucial to mitigate potential short-term environmental degradation from trade liberalization. These agreements should encourage knowledge transfer and stricter environmental standards among trading partners. Finally, a cautious approach to FDI is recommended. Governments should meticulously assess the environmental impact of potential FDI projects and ensure alignment with national green growth strategies to avoid attracting polluting industries.