Investigating institutional quality and carbon mitigation drive in Sub-Saharan Africa: Are growth levels,energy use,population,and industrialization consequential factors?

Abstract

In view of the United Nations’ Sustainable Development Goals on clean and responsible energy consumption, climate change mitigation, and sustainable economic growth (UN-SDGs-7, 11–13), this study examines institutional quality (IQ)–carbon emissions nexus in the framework of the Environmental Kuznets Curve (EKC) hypothesis. Six dimensions of IQ from the World Governance Indicators (WGIs) were used while focusing on Sub-Saharan African (SSA) countries between 1996 and 2019. After controlling for growth, energy use, and industrialization levels, the empirical results validated the EKC hypothesis for the SSA as a unit increase in economic growth initially worsens the environment while further economic expansion eventually improves the environment. However, mixed results were obtained on the effects of IQ indicators. CO2 emissions are only substantially reduced by corruption control, regulatory quality, and the rule of law among other IQ measures. Furthermore, the causality analysis showed a unidirectional causality between growth and environmentally detrimental energy consumption levels coupled with a two-way emission-population growth causality flow as well as a two-way emissions—IQ causality channel. While economic growth, energy use, and industrialization levels undermine environmental sustainability in the SSA region via increased carbon emissions, the overall findings signal the moderating roles of IQ. Hence, the strengthening of institutions is recommended for environmental sustainability enhancement in the SSA region.

Keywords

Institutional quality CS-ARDL SSA countries EKC

Introduction

Sub-Saharan Africa (SSA) economies are constantly being faced with global warming issues and there are increasing challenges on how to mitigate climate-related menace in the region over the last couple of years. Hence, authorities often try to combat environmental degradation and a major aspect of this combat involves greenhouse gas (GHG) emissions reduction. Decisively reducing GHG emissions such as CO₂ emissions tends to be a hard task as policy makers are often concerned with how to ensure that economic growth is not jeopardized while mitigating climate-related menace. Institutional discussion often arises at this point since institutions are comprised of laws and regulations that can act as accelerators for long-term development in general. Although institutions have a role in addressing the emissions reduction–economic growth concerns,¹ their influence is, however, difficult to quantify and different institutional indicators can have different effects.^2,3

Institutional failure can harm the environment while on the contrary good institutional structure such as well-enforced environmental regulations can improve the quality of an environment as highlighted by Inrahim and Law and Lau et al.^4,5 Efficient institutions can properly manage industrial layouts in countries to support economic growth while minimizing environmental damage. As a result, institutional involvement can improve air quality through appropriate laws. While there are empirical findings highlighting bad environmental conditions to be the result of inadequate institutions such as Cole et al. and Welsch, ^6,7 other studies have also attributed proper environmental management to better institutions.^8,9 Invariably, the conclusions about the impact of institutional quality (IQ) measures on environmental deterioration tend to be based on the institutional and environmental condition assessment indicators.

Some studies have recently begun to draw the attention of society to the position of institutions as a possible deciding factor for environmental pollution levels unlike the direct focus on energy consumption levels sorely. Most of these studies focused on developed regions of the world, leaving the case of developing regions with little or no documentation, specifically the SSA region. In recent years, efforts have been made to investigate the African institutional experience. The existing literature mainly focused on a wide range of hypotheses for why Africa's institutional performance has lagged relative to its developing world counterparts. However, the function of institutions in carbon emission discussion is the focus of this present study. According to Ndulu et al.,¹⁰ a study carried out by the AERC revealed that institutions are given considerable attention in explaining the development of African economies and this is also corroborated by the study of Appiah et al..¹¹ In a societal context or a more formal perspective, North¹² described institutions as the rule of the game, the humanly created restrictions that shape human collaboration. Good institutions are crucial in creating an atmosphere for the juice of economic growth to filter down to SSA's poorest segments of the population.¹³ However, the condition is paradoxical in Africa as the continent suffers more from various consequences of environmental change than other continents considering the transition from agrarian to manufacturing economies among African countries leading to a rise in environmental degradation. IQ is a crucial factor in economic progress, and this applies in developed and less developed economies. Strong institutional frameworks fight corruption, assist in establishing the rule of law, reduce military participation in politics, and improve imprudence in public financing. On the other hand, dysfunctional institutions have a long-term impact on a country's economic development as noted by Sherani.¹⁴

In general, IQ is linked to tactics used by the county's institutions to establish cultural and legal structures that facilitate socioeconomic and financial activities and can influence the right intentions to control pollution. The levels of IQ can have a big effect on environmental indicators like CO₂ emission. Ineffective financing from inadequate institutions can be a curse on the public sector, among other things like corruption, shoddy bureaucratic procedures, and weak protocols.¹⁵ Researchers and scientists are now focusing on IQ in relation to the environment^16,17 and other studies have confirmed that an effective and impartial governmental organization can play important roles in fostering cooperation including those relating to environmental deals.^18,19 As a result, institutions are currently essential in solving environmental issues. Comparatively, if institutional flaws exist, businesses are prone to ignore necessary efforts for the control of pollutants like CO₂ emissions.

Studies on environmental economics ^20–22 are becoming more prevalent, and some of them have also concentrated on IQ and policies because of the crucial roles that these elements have in environmental attributes. Bekun et al.²³ concluded that IQ contributes to lowering the environmental cost of economic growth while examining the relationship between IQ and economic performance in a cross-country analysis employing alternative institutional indicators. Additionally, Huynh and Hoang²⁴ suggested that the existence of functional markets, social institutions, and government policies determine when and how environmental quality will improve. Similar to this, Zhang et al.²⁵ reiterate the significance of IQ and the rule of law by contending that if institutions have poor IQ, businesses are more likely to disregard environmental externalities and CO2 emission regulation. According to a recent study by Sadik-Zada and Loewenstein,²⁶ some elements, such political freedoms and civil liberties, that are frequently strongly influenced by IQ may have a detrimental impact on per-person carbon emissions.

Thus, this study intends to analyze the connection between IQ, CO₂ emission, economic growth, energy usage, industrialization, and population among SSA countries. The study further addresses whether IQ is vital to the environmental challenges in the SSA. To create effective policies that would support sustainable economic growth, particularly in SSA countries, policy makers must understand the effects of consumption patterns of renewable energy and IQ on CO₂ emissions. This study differs from previous research in that it extends the Environmental Kuznets Curve (EKC) framework by incorporating the combined effects of population expansion, industrialization, energy use, IQ, and economic growth. In most studies,^18,27,28 IQ is assessed using selected governance indicators, including political stability, government efficiency, regulation quality, rule of law, and corruption control. This research is distinct and contributes to the body of literature in a way that, it examines the function of IQ as a driver of carbon emissions using several modalities of IQ indicators using all six IQ indicators separately that were not covered in previous studies. We also present a variety of second-generation panel method econometric analyses, which are thought to be superior to first-generation methods in terms of providing reliable empirical coefficients and outcomes for future policy direction. These methods deliver a feasible solution in terms of accuracy, efficiency, and robust outputs for the main detected problems relating to cross-sectional dependence and heterogeneity.

Literature review

A limited but growing body of research has been done to examine the link between IQ and environmental performance. Literature has covered a variety of potential outcomes, such as political and economic ones, in this regard. Most of the earlier research has concentrated on the factors that contribute to the growth of air pollution emissions, the most important aspect of the environment. We have evaluated the literature on institutions in this area, both theoretical and empirical on CO₂ emissions.

Institutions and environment relationship

According to institutional theories, institutions play a significant role in the quality of the environment.¹⁸ After showing how institutions were receiving less support for engaging in constructive activity, North¹² concentrated on the importance of institutions for developing and underdeveloped nations. Democracy has a key role in ensuring higher environmental quality.²⁹ Kirkpatrick and Parker³⁰ noted the significance of institutions in addressing problems in developing and transitioning economies. Aron³¹ identified bad institutional performance to be the result of ineffective enforcement. Emission is connected to institutions through legal processes and potential economic contracts. To control emissions, Makdissi and Wodon³² emphasized the significance of environmental legislation. Effective political institutions are guided by good governance which is important for the state of the environment.³³ The key routes that appear to be distorting environment governance are bureaucratic ineptitude, poor institutional management, and financial mismanagement.^7,34 Controlling abatement costs is a crucial strategy for political institutions to influence environmental quality.³⁵ Although there is scanty data for emerging nations, the quality of institutions and their impact on emissions have also been explored. According to Torras and Boyce,³⁶ political systems and institutions in least-developed nations have a favorable impact on the quality of the air and water. Midlarsky³⁷ has also observed that democracy's influence on environmental quality is uncertain.

In a variety of developing nations, Gani³⁸ and Gani⁹ have also looked at how institutional indicators including political stability, government effectiveness, regulatory quality, rule of law, and corruption affect carbon emissions. According to the empirical findings, IQ affects the environment. Similar findings were previously reported.³⁶ After considering 61 countries, Akhbari and Nejati³⁹ discovered conflicting results on the impact of IQ indicators on corruption and concluded that both developed and developing nations had negligible effects from increased corruption control. IQ has been observed to be reducing carbon emissions.⁴⁰ There is disagreement over how political institutions affect carbon emissions, which is a significant limitation of earlier research.

According to Jung,⁴¹ an increase in IQ in developing economies causes pervasive technology upgrading effects in advanced economies, which results in increases in overall productivity. In addition, institutional is described as the rules that govern interactions in social, political, and economic contexts that reflect human behavior.⁴² The legal and political systems, the rule of law, conventions, values, and traditions are only a few examples of the formal and informal rules that make up society. There is conflicting evidence in the literature that relates IQ to the environment. While some studies have discovered that IQ has a favorable effect on the environment, others have found the exact reverse. Additionally, the IQ index differs across studies. In 66 developing nations, Azam et al.⁴³ found that IQ increases energy consumption and environmental degradation. They discovered that, in the current period of globalization, IQ has no bearing on reducing pollution. Additionally, the aforementioned study utilized the conventional indicator of environmental degradation (i.e. CO₂ emissions). On the other hand, in the study of Ali et al.⁴⁴ on 47 developing countries, it has been found that better institutions can help to improve the quality of the environment. The present study discovered that, in contrast to the first study cited in the aforementioned correlation, higher IQ can help reduce CO₂ emissions and enhance environmental quality. Additionally, Awais et al.⁴⁵ examined how governance affected the reduction of CO₂ emissions from 1996 to 2017. According to their findings, improved governance can help the BRICS economies reduce their CO₂ emissions, which will lessen environmental devastation there. The relationship between IQ and CO₂ emissions has also been explained by subsequent studies although this was not considered in the earlier investigations.^16,46 We, therefore, aim to contribute to the SSA literature and close this research gap by examining the relationship between IQ and CO₂ emission specifically in the SSA.

Data, models, and methodology

Data

The data used comprises annual data on institutional indices, economic growth, and environmental measures. The IQ measures include all the variables stipulated by the World Governance Indicators (WGIs) and other controlling variables such as growth, industrialization, energy consumption, and populace growth, with CO₂ emissions as the dependent variable from the British Petroleum (BP) database. From 30 countries in the SSA¹, the analysis and data cover the period 1996–2019, as environmental and institutional matters are crucial in this setting. There is an appropriate cross-state range of variations in institutional and ecological variables. The table below gives an extensive discussion of the features of the variables involved.

The descriptive statistics of the individual variables are presented in Table 1, and it shows that emissions of CO₂ have a mean of.8906 mt with a SD of 1.7672. For the study period, the average economic growth of SSA is 1.4713%, with an SD of 4.7927, while the average energy intake is 9.3680 kt with an SD of 6.9400. With an SD of.9678, the average population size is 2.5699, while industrialization has a modest growth rate of 26.1928% with an SD of 12.9052. The full meaning of IQ indices is provided in the list of nomenclature in Table A1 in the Appendix. The variables of the institutions, including COC, GE, PS, RQ, ROL, and VA, have a mean (and a SD) of −.6159 (.62584), −.6578 (.8706), −.6578 (.8706), −.5529 (.8386), −.6067 (.6329), and −.5792 (.7299), respectively. Also, there are associational coefficients between the primary variables employed in the analysis. Based on the World Bank specifications of the WGIs, the six IQ indicators may have overlapping impacts. Table 1 shows that, except for the control variables that are negatively linked in the model, the dependent variable CO₂ is positively but less correlated to the six governance indicators. Since the six variables of governance correlate in the model, the empirical model is estimated by regressing the six governance metrics and control variables separately from the others to prevent conflicting effects.

Table 1.

Variable features.

Variables	CO₂	IND	GR	ENE	POP	COC	GE	PS	RQ	ROL	VA
Mean	0.8906	26.1928	1.4713	9.3680	2.5699	−0.615	−0.6578	−0.552	−0.564	−0.606	−0.579
SD	1.7672	12.9052	4.7927	6.9400	0.9678	0.62584	0.8706	0.8386	0.6162	0.6329	0.7299
Min	−8.5464	6.0940	−47.503	1.6527	−6.766	−2.681	−17.221	−2.713	−2.375	−2.575	−2.880
Max	9.9795	77.4137	37.5355	50.1347	8.1179	1.2167	1.1097	1.2192	1.1273	1.0771	3.2092
	Correlation analysis
CO₂	1
IND	0.2590	1
GR	−0.0352	−0.076	1
ENE	−0.1448	−0.0890	−0.0046	1
POP	−0.3097	0.0146	−0.0108	0.1187	1
COC	0.2384	−0.0540	0.1363	−0.2375	−0.265	1
GE	0.2419	0.0221	0.0996	−0.1875	−0.198	0.6411	1
PS	0.1851	0.1142	0.0443	−0.1998	−0.141	0.7178	0.5491	1
RQ	0.3058	−0.0224	0.1343	−0.3269	−0.182	0.8355	0.6254	0.6755	1
ROL	0.2333	−0.0350	0.1133	−0.3249	−0.174	0.8628	0.6010	0.7672	0.8780	1
VA	0.3047	−0.0071	0.1022	−0.2367	−0.219	0.7375	0.5802	0.6654	0.8130	0.8018	1

ENE, energy usage; IND, industrialization; POP, population growth.

Model

To structure the framework of the long-term association and causality between CO₂ emissions and institutional efficiency, the study proposed the following structure and the EKC model, which is intended to achieve the objectives. Pollution levels such as CO₂ emissions, IQ indicators, EKC variables (Growth and Growth Squared) with industrialization (IND), energy usage (ENE), and population growth (POP) as control variables (Z) are the significant variables involved in the model construction. The following econometric model based on the EKC theory is proposed based on the above assumptions:

C O_{2 i t} = β_{0} + β_{1} I Q_{i t} + β_{2} G R_{i t} + β_{3} G R_{i t}^{2} + β_{4} Z_{i t} + μ_{i t}

(1)

β₀ implies an unnoticed individual effect and β₁ captures the effect of IQ indicators on CO2 with some control variables in the presence of EKC.

μ

_it represents the stochastic term. The stochastic term is distributed with the assumption of zero mean and constant variance, and superscript t is the time range. The above model is specified in the EKC theory and is proposed as

C O_{2 i t} = β_{0} + β_{1} I Q_{i t} + β_{2} G R_{i t} + β_{3} G R_{i t}^{2} + β_{4} I N D_{i t} + β_{5} E N E_{i t} + β_{6} P O P_{i t} + μ_{i t}

(2)

The IQ indicators include voice and transparency, political solidity, corruption prevention, quality of regulation, the rule of law, and government success.

Methods

To achieve the desired results, our study utilized different econometric procedures and the subsequent techniques: cross-sectional dependence to determine sectional properties, and slope homogeneity checking to analyze the existence of heterogeneity along with the order. Furthermore, to discuss the long-term relationship and causality, Westerlund and Edgerton⁴⁷ LM approach is used for cointegration analysis among variables under review, while Dumitrescu and Hurlin⁴⁸ for causality analysis.

Test of CSD

There is a greater probability that data obtained for the understudy nations display cross-sectional dependencies based on their trade connection and other economic activities. Therefore, consideration was given to the bias-corrected LM test of^49–52 where their equations can be shown as:

y_{i t} = α_{i} + β_{1}^{'} X_{i t} + e_{i t}

(3)

X_it, as well as β, are vector regressors and coefficients to be predicted while α_i, describes individual constant factors. Cross-sectional units are considered independently distributed (i.i.d.) under the null principle eit may be correlated by cross sections as an alternative; however, the fact that there is no serial association remains unchanged. Therefore, H_o: _ij =

ρ

_ji = cor (μ_i, μ_jt) = 0 where

i \neq j

, and on the other hand, H_o: _ij =

ρ

_ji

\neq

0. Also, ρij is the noise estimate of the product-moment link and is provided in equation (4).

ρ_{i j} = ρ_{j i} = \frac{\sum_{i = 1}^{T} μ it, μ jt}{(\sum_{i = 1}^{T} {μ it}^{2})^{1 / 2} (\sum_{i = 1}^{T} {μ jt}^{2})^{1 / 2}}

( 4)

where the LM = T

\sum_{i = 1}^{N - 1} \sum_{j = i + 1}^{N} {\hat{ρ}}^{2} ij

and the

\hat{ρ}

ij is a sample estimate of the correlational residual as sets (5).

{\hat{ρ}}_{i j} = {\hat{ρ}}_{i j} = \frac{\sum_{i = 1}^{T} \hat{u} i t, \hat{u} j t}{(\sum_{i = 1}^{T} \hat{u} i t^{2})^{1 / 2} (\sum_{i = 1}^{T} \hat{u} j t^{2})^{1 / 2}}

(5)

And

\hat{u}

it is the measurement of uit in (4). On the assumption of null interest, LM is asymptotically scattered as chi-square distribution with degrees of freedom whereas if N is high with a finite T, this exam is expected to experience significant size biases- a situation commonly experienced in analytical applications, primarily since the LM statistics are not appropriately focused to finite T. With N high, the bias is possible and would worsen. An alternative to that was proposed,

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

(6)

Moreover, it showed that under the null principle CD →d N (0, 1). If data has nonstationary models and homogeneous/heterogeneous dynamic techniques, the CD statistics have an actual mean, with T and N having specific values of zero, unlike the LM statistics. The standard FE/RE techniques are biased toward homogeneous and heterogeneous dynamic models.^53,54 Moreover, the CD technique is still helpful since, given the limited model bias of the framework of the factors, the FE/RE residuals will have precisely the mean of zero if the instabilities are distributed symmetrically.⁵⁰ suggests, for unbalanced panels, a slightly altered form of equation (3) given by

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

(7)

where Tij = # (Ti ∩ Tj) (i.e. number of similar observations among units i and j in a time series),

{\hat{ρ}}_{i j} = {\hat{ρ}}_{i j} = \frac{\sum t ϵ T i \cap T j (\hat{u} i t - \hat{u} i) (\hat{u} j j - \hat{u} j)}{{\sum_{t} ϵ T i \cap {T j (\hat{u} i t - \hat{u} i)}^{2}}^{1 / 2} {\sum_{t} ϵ T j \cap T j {(\hat{u} i t {\hat{u}}_{j j})}^{2}} s s^{1 / 2}}

(8)

and

\hat{u} i = \frac{\sum_{t} ϵ T_{i} \cap T_{j}, \hat{u} i t}{# (T_{i} \cap T_{j})}

(9)

The revised equation reflects the fact that the t residuals do not amount to zero. Therefore, Breusch and Pagan⁴⁹ proposes the CD approach of the LM to process CSD as provided in equation (10).

y_{i j} t = α_{0 i} + α_{1 i} X_{i t} + μ_{i t}

(10)

where subscripts i range from 1 to N given time range t from 1 to T, the CS and time frame; X_it signifies the explained variables vector, and α_0i and α_1i denote individual intercepts and variables of slope throughout nations. Therefore, the Breusch–Pagan LM techniques indicators can be determined as

LM = \sum_{i = 1}^{N - 1} \sum_{j = i + 1}^{N} T_{i j} {\hat{ρ}}_{i j}^{2} \to X^{2} \frac{N (N - 1)}{2}

(11)

where

\hat{ρ}

ij represented the coefficient of pair correlation based on equation (10) by applying the ordinary least squares (OLS) techniques. Thereafter, the null assumption of no CD was also checked and compared with the alternative CD principle. Pesaran et al.⁵⁵ suggest that the LM technique is not adequate if the N is high since the mean populace link is near zero. To fix this matter, Pesaran and Yamagata⁵⁶ recommend utilizing actual mean and variance figures from the LM technique as a bias-adjusted version of the technique. The figures for the bias-accustomed LM (i.e. Pesaran LM) will therefore be considered as

L M = \sqrt{(\frac{2}{N (N - 1)}}) \sum_{i = 1}^{N - 1} \sum_{j = i + 1}^{N} {\hat{ρ}}_{i j} \frac{(T - K) {\hat{ρ}}_{i j}^{2} - μ}{\sqrt{ν^{2} T_{i j}}} \to N (0, 1)

(12)

where k states the number of regressors and

μ

Tij and

ν^{2}

Tij 2 denotes the exact mean and variance of (T-k)

{\hat{ρ}}^{2}

ij.

Test of SH

The complexities of the economic procedure in every one of the nations may be alike due to the existence of substantial CD,⁵⁷ thus recommending a rationale supposing homoscedasticity in panels. In contrast, N occurs comparatively to T to test the homogeneity of the slope. Therefore in large panels, N and T are,⁵⁷ test extensions, and are modeled as follows, by Pesaran and Yamagata⁵⁶

\tilde{S} = \sum_{i = 1}^{N} (\hat{β} i - {\hat{β}}_{W F E})^{'} \frac{X_{i}^{'} M τ X_{i}}{σ_{i}^{2}} (\hat{β} i - {\hat{β}}_{W F E})

(13)

where S˜ stands for accustomed Swamy test (i.e. Pesaran and Yamagata SH test) statistic, β^i stipulates the coefficient of joint OLS, ) indicates the weighted fixed effect joint estimator,

\tilde{σ}

² = ((yi -Xi

\hat{β}

i) ^′ M

τ

(yi -Xi

\hat{β}

i))/(T -k-1) embodies the evaluation of σ²i, xi denotes the matrix covering the explained factors in mean nonconformities, and the uniqueness matrix is signified by M

τ

. Thereafter, the null for slope homogeneity is (H0:

β

i =

β

j for all i). The Pesaran and Yamagata SH test is compared to the alternative assertion supporting heterogeneity (H1:

β

i ≠

β

j for all i). The standard dispersal statistics is given as

\tilde{Δ} = \sqrt{N (\frac{N^{- 1} \tilde{S} - k}{\sqrt{2 k}}})

(14)

The bias-adjusted form of the Δ˜ exam can be structured as surveys:

{\tilde{Δ}}_{adj} = \sqrt{N} [\frac{N^{- 1} \tilde{S} - E (\tilde{Z_{i}} T)}{\sqrt{Var} (\tilde{Z_{i}} T)}

(15)

where Var (

\tilde{z_{i}} t

) = (2k (T -k-1))∕(T + 1) and E (

\tilde{z_{i}} t

) = k stipulates the variance and the mean, compatibly.

Unit root test

When the order is CD and heterogeneous, conventional panel stationarity techniques like the Augmented Dickey Fuller (ADF) are ineffective. Pesaran⁵⁸ overcomes this problem by combining the classic ADF technique with CD using lagged phases and ordering one of the distinct structures, resulting in a new panel unit root test accommodating CD and heterogeneity. The CADF regression is depicted as follows:

Δ y_{it} = α_{i} + b_{i} {\bar{y}}_{i, t - 1} + c_{i} y_{t - 1} + d_{i} Δ {\bar{y}}_{t} + e_{it}

(16)

While subscriptions i = 1,2, … N and t = 1,2, … T signify the CS measurement and time, individually; y_t₋₁ = N⁻¹

\sum_{i = 1}^{N} y i

T-1 states the CS mean of y_i_,t−1; as well as Δyt = N-1 ∑N i = 1 Δy_it indicates the first order of Δy_it. The null principle of stationarity is (H0

β

i = 0 for all i), and the CADF technique evaluates the alternative statement of no stationarity (H1:

β

i < 0 for some i). Pesaran⁵⁸ developed a cross-sectional augmented CIPS stationarity technique established on the average of separate CADF analyses for the entire group. The following are considerations for the CIPS test statistics:

C I P S = N^{- 1} \sum_{i = 1}^{N} C A D F i

(17)

where CADFi specifies a CADF technique measurement for equation (16) cross-sectional unit ith. The cruciality of the unit root test has resonated in empirical studies and the application of this approach in particular has been broadly welcomed in many other relevant empirical studies.^59–62

Test of cointegration

The Westerlund and Edgerton⁴⁷ panel long-term analysis, noted for vigorous handling of CS, is utilized to analyze the long-term links in an econometric scheme between the variables in question. The CD is accounted for by the Westerlund and Edgerton⁴⁷ long-term approach by assessing the likelihood figures of the technique value using the bootstrapping method. Based on the null principle of no long-term link, for a minimum of one CS unit or long-term link surrounded by the entire panel, against the alternative principle of cointegration, two group mean approaches, and a two-panel approach are examined. The four experiments carried out under the cointegration panel style for the approach are defined in terms of an ECM, which can be written as

Δ Y_{i, t} = δ d_{t} + α_{i} (y_{i, t - 1} - β_{i}^{'} X_{1, t - 1}) + \sum_{j = 1}^{p i} α_{i j} Δ y_{i},_{t - j} + \sum_{- q i}^{p i} γ_{i j} Δ X_{i, t - j} + e_{i}, t

(18)

The deterministic components are dt, and the lag lengths and lead orders that vary across individual CD are pi and qi. Within this framework, the two group -mean calculated t stats G_t and G_a, and two-panel calculated t stats P_t and P_a, can be defined by Westerlund and Edgerton⁴⁷ long period analysis as

G_{t} = \frac{1}{N} \sum_{i = 1}^{N} \frac{\hat{a}}{S E (\hat{a_{i}})}

(19)

G_{t} = \frac{1}{N} \sum_{i = 1}^{N} \frac{T \hat{a}}{\hat{a_{i}} (1)}

(20)

P_{t} = \frac{\hat{a_{i}}}{S E (\hat{a_{i}})}

(21)

P_{a} = T \hat{α}

(22)

The significance of these test values terminates the null presumption of suggesting a long-run connection between the model's factors. Cointegrating activities are a requirement for long-run estimates to be tested utilizing proper techniques. The Westerlund approach has been utilized in many other relevant empirical studies.^63–65

Test of cross-sectionally augmented autoregressive distributed lag (CS-ARDL)

To analyze and present the long-run technique centered on the MG method, the research uses the CS-ARDL methodology^66,67 since it is the most accurate and efficient optimal in terms of sample accuracy and effectiveness. The CS-ARDL technique removes the excesses of evaluating the integration order ahead of time. The validity of this method is dependent on some conditions: the variables of interest should have a long-run relationship and the regressors should be weakly exogenous and serially uncorrelated to the residual. The CS-ARDL method effectively controls for cross-sectional dependencies when describing heterogeneity. Additionally, the CS-ARDL holds the following merits, namely (i) it makes available the best possible option in terms of accuracy, efficiency, and robust outcomes in panel data analysis, (ii) it eliminates the need to pretest the integration order, deals effectively with CS-ARDL issues and describes heterogeneous time series, (iii) it addresses the problem of slope homogeneity issues and feedback effects between the indicators, and (iv) it gives long and short-term effects.^54,66–71 The MG method, which analysis distinct equations for each nation and identifies the average of assessed factors across nations, is at one extreme. When the time series measurement of the data is necessarily high, the MG method provides consistent evaluations of the average of the parameters.⁵⁴ In one part, the MG approach calculates various models for each state and analyses the average of approximate estimates for the samples. When the time series measurement of the data is satisfactorily large, the MG system produces reliable evaluations of the average of the parameters.⁵⁴ The CS-ARDL approach proposed by Chudik et al.⁶⁸ is expressed in equation (23).

y_{i t} = C_{y i} * + \sum_{l = 1}^{p y} φ_{i j} y_{i t - 1} + \sum_{l = 0}^{p x} β_{i l}^{'} x_{i, t - 1} + \sum_{l = 0}^{p \bar{z}} Ψ_{i j} {\bar{z}}_{t - 1} + e_{i t} *

(23)

where

\bar{Z}

_t = (

\bar{y}

\bar{x}

ʹ_t),p

\bar{z}

= T^1/3 and two choices for the lingering lag orders are well-thought-out: ARDL(2,1) requirement, p_y = 2 and p_x = 1, and ARDL(1,0) requirement, p_y = 1 and p_x = 0. The CS-ARDL evaluations of the separate mean equal coefficient are then assumed by

\hat{θ} C S - A R D L,_{I} = \frac{\sum_{l = 0}^{p x} \hat{β_{i l}}}{1 - \sum_{l = 1}^{p y} \hat{φ_{i l}}}

(24)

where the guesstimates of the short-run (

\hat{φ}

_il,

\hat{β}

) are estimated following equation (24). The mean long-run impacts are estimated as

N^{- 1} \sum_{i = 1}^{N} \hat{θ} C S - A R D L, i

, and the implication are built on the traditional nonparametric estimator of the asymptotic variance of the MG estimator. Also, i = 1, 2… N, t = 1, 2… T, and x_it is k _x × 1 vector relating to CS unit i at time t; c*_yi and c

Ψ

i are individual FE for unit i, git is k _x × 1 vector of covariates specific to unit i (not detected in the panel data model)

Causality test

The Dumitrescu and Hurlin⁴⁸ technique is an improved variant of the causation technique for heterogeneous tables. The method calculates the average of separate Wald analyses for cross-sectional units using the Granger causation test In the panel causation test by Dumitrescu and Hurlin,⁴⁸ three different statistical values are calculated as follows:

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

(25)

where

W_{N, T}^{Hnc}

is the mean value of the individual Wald figures. As T and N approach infinity, the mean statistics meet successively with equation (26), supposing that the separate residuals are originally diffused around the CS with the covariances summing up to zero, according to Dumitrescu and Hurlin⁴⁸:

Z_{N, T}^{H n c} = \sqrt{\frac{N}{2 K}} (W_{N, T}^{H n c} - K) ⟶_{N, T \to \infty}^{d} N (0, 1)

(26)

where

Z_{N, T}^{Hnc}

is the z-values, N is the sum of CS, and K is the lag's optimal length. Even, conferring to the DH causality method, if T slopes to infinity, the separate Wald data are dispersed initially equal, with the mean discrete Wald figures equivalent to K and the variance similar to 2 K. After that, a roughly standardized Z-values for the mean Wald figures of the HNC null statement is determined, described as:

Z_{N}^{H n c} = \frac{\sqrt{N} [W_{N, T}^{H n c} - N^{- 1} \sum_{i = 1}^{N} E (W_{i, T})]}{\sqrt{N^{- 1} \sum_{i = 1}^{N} V a r (W_{i, T})}} ⟶_{N, T \to \infty}^{d} N (0, 1)

(27)

The null statement and alternative argument for the calculated panel statistics are as follows:

H_{0} : β_{i} = 0 \forall i = 1, 2, \dots, N

H_{1} : β_{i} = 0 \forall i = 1, 2, \dots, N 1

β_{i} \neq 0 \forall i = N_{1} + 1, N_{1} + 2, \dots, N

normal panel data scheme.

Results and discussions

CSD, SH, SC, FM, and MC tests

The achieved results in Table 2 show the presence of CSD and SH as well as the elimination of the test hypothesis of Friedman in all variables using the^55,56,72 test for poor CSD, SH, and Friedman test. This is shown by the fact that all the probability values of CD statistics, LM statistics ( $\tilde{Δ}$ and $\tilde{Δ}$ Adj), in all the variables involved, and Friedman assessments are noteworthy at 1% and 5%. This means that in each of the states chosen in SSA, a shock appears to be conveyed to other states within the panel. The results continue to prove the absence of multicollinearity as well as serial autocorrelation among the datasets.

CIPS and CADF stationarity tests

Based on the outcomes, the stationarity of the data was established by employing the (CIPS and CADF) test that took CSD into account. The test outcomes are summarized in Table 3. According to the unit root test results, all variables are stationary for intercept process patterns at the level and first difference; thus, variables are stationary at both orders I (0) and I (1) as shown in Table 3. However, regression at their rate would also be spurious, which means that they must pass the cointegration test not to lose any long-term data. Therefore, the second-generation panel long haul test developed by Westerlund and Edgerton⁴⁷ was adopted to test the long-run connexion between variables in a heterogeneous panel of SSA countries.

Table 2.

SH, FM, CSD, SC, and MC findings.

Test	T-stats	Prob.
CSD test	5.866	0.0000^z
SH test ( $\tilde{Δ}$ test)	5.628	0.0000^z
SH test ( $\tilde{Δ}$ adj test)	7.477	0.0000^z
Serial autocorrelation	0.454	0.5062
Friedman	79.688	0.0000^z
Multicollinearity	-	2.94

Note: z, x, and y signifies p < 0.01; p < 0.05; p < 0.1.

Table 3.

CIPS and CADF units root results.

	CIPS		CADF
Variables	I(0)	I(1)	I(0)	I(1)
CO₂	−2.284	−5.025^z	0.610	−8.413^z
IND	−2.575	−5.386^z	0.378	−10.193^z
GR	−4.833	−6.342 ^z	−7.869	−19.722^z
GR²	−4.665	−6.201^z	−6.066	−17.199^z
ENE	−1.893	−5.192^z	2.717	−7.509^z
POP	−2.006	−1.621^z	−14.114	−8.014
COC	−3.439^z	−5.758^z	−1.442	−16.607^z
GE	−2.749	−5.392	1.337	−11.230^y
PS	−3.124	−6.075^z	1.006	−13.002
RQ	−3.245	−5.921^z	−1.735	−17.239^z
ROL	−2.747	−6.037^z	1.123	−15.961^z
VA	−2.408	−5.384^z	2.835	−14.523^z

Note: z, x, and y signifies p < 0.01; p < 0.05; p < 0.1.

ENE, energy usage; IND, industrialization; POP, population growth.

Cointegration test

However, the outcomes of the Westerlund and Edgerton⁴⁷ ECM cointegration tests are described in Table 4. The analysis revealed cointegration among the equations’ series; therefore, it is important to estimate the long-term relations between variables. Furthermore, we also provided the cointegration results of each country of the SSA, and the results are reported in Table 7. In the realization of the individual country cointegration results, the ARDL bounds test was applied to test the cointegration outcomes. The bounds test is favored over other tests because it applies to variables of varying order of integration.^73–75 The F-statistic is used such that if it is greater than the I(1) bound indicates that cointegration is possible in the interaction, whereas if the F-statistic is less than the I(0) bounds indicates that cointegration does not exist The results indicate that 12 SSA countries exhibit a long-run relationship with 12 countries also revealing no long-term link while 6 countries indicate an inconclusive pattern meaning the F-stats fall between the I(0) and I(1) bound. In a nutshell, there is greater evidence to support cointegration on an individual level than rejecting the presence of a cointegration relationship among variables. The study then proceeds to assess the long- and short-term estimates using the CS-ARDL system.

Table 4.

Westerlund long-run results.

Coint. test	IND	GR	ENE	POP	COC	GE	PS	RQ	ROL	VA
Gt	1.624	1.962	1.432	0.187	−1.828^y	1.216	1.301	0.507	0.758	1.342
Ga	−2.136^y	−1.288^x	−1.817^y	−3.105^z	−2.298^y	−2.343^z	−1.987^y	−1.987^x	−2.008^y	−1.907^y
Pt	−2.088^y	−2.489^z	−1.336^x	−3.289^z	−2.058^y	−3.289^z	−3.289^z	−2.404^z	−2.413^z	−1.262
Pa	−18.858^z	−19.996^z	−20.957^z	−26.285^z	−2.687^z	−21.316^z	−21.316^z	−18.523^z	−20.643^z	−17.961^z

Note: z, x, and y signifies p < 0.01; p < 0.05; p < 0.1.

ENE, energy usage; IND, industrialization; POP, population growth.

CS-ARDL long- and short-run effects

Potential CSD and heterogeneity are significant concerns that prior studies have disregarded when evaluating the carbon emission model.¹¹ Due to the model's dynamic nature, CSD and heterogeneity may appear. Additionally, these result from shocks and volatility that nations encounter as a result of the transmission of similar economic activities as well as their geographical position. The model cannot be estimated using the OLS approach due to the aforementioned econometric problems since OLS findings are inconsistent and biased in the presence of CSD and SH. Thus, to get unbiased and consistent results, the CS-ARDL estimation technique will be applied. CS-ARDL was utilized to extract reliable facts and overcome empirical problems. The method delivers a feasible solution in terms of accuracy, efficiency, and robust outputs for the main detected problems relating to cross-sectional dependence and heterogeneity while producing both long and short-run impacts.

Table 5 presents the estimated outcomes. The EKC theory is tested first in this study. According to the calculated coefficient, a 1% rise in economic growth will result in a 0.0063 increase in carbon emissions. This outcome is at odds with the decoupling between economic expansion and carbon emissions. It suggests that environmental pollution is being sacrificed for economic growth in SSA. A squared term of growth is incorporated into the model to examine if the relationship between economic growth and carbon emissions is curvilinear. This squared term's negative, statistically significant coefficient (−0.0408). It suggests that after a certain level of growth is reached, carbon emissions start to decrease. These findings support the EKC theory in SSA, which states that the link between wealth and carbon emissions in SSA is inverted U-shaped. These findings confirm finding of Grossman and Krueger finding that growth and pollution have an inverted U-shaped connection. This result of lower emissions at higher levels of growth has also been observed in previous research such as; Nasir et al.⁷⁶ for ASEAN states and Bekun et al.²³ for E7 countries among others. Similarly, Erdoğan et al.⁷⁷ investigated the case of Turkey and the Caspian countries and found that CO₂ emissions rise at the opening stages of growth but begin to decline as further economic development is achieved among these countries. Zoundi⁷⁸ attempted to quantify the nonlinear impacts of growth on CO₂ discharges in designated African states. The research employed a system of panel long-run approach and tested the robustness of the findings through different advanced methods of panel assessment. Over the stated era, the EKC proposition was found to be true in these designated African states.

Table 5.

CS-ARDL long- and short-run results.

Variables	Short-run Est.		Long-run Est.
Variables	Coef.	z-Stats	Coef.	z-Stats
L.CO₂	−0.8523	−7.63^z	−1.8522	−16.58^z
IND	−0.0010	−0.23	0.0113	2.15^x
GR	0.4218	1.25	0.0063	2.02^x
GR²	0.0001	0.05	−0.0408	−1.87^y
ENE	−0.0010	−0.06	0.0038	0.05^y
POP	−0.1002	−0.74	0.0715	0.99^y
COC	−0.0798	−0.49	−0.0128	−0.23^y
GE	0.1619	0.49	0.0193	0.17^y
PS	0.4321	1.28	0.2123	1.29^x
RQ	−0.0044	−0.03	−0.2863	−1.23^x
ROL	−0.6257	−1.30	−0.00002	−0.00^x
VA	0.2835	0.97	0.1074	0.76
R² (MG)	0.91		Prob > F	0.00

Note: z, x, and y signifies p < 0.01; p < 0.05; p < 0.1.

ENE, energy usage; IND, industrialization; POP, population growth.

The nations of SSA import a substantial amount of energy. According to the estimated value of the energy consumption coefficient, SSA's per capita carbon emissions will rise by 0.0038 for every 1% increase in energy consumption. This result backs up the claim made by Appiah et al.⁷⁹ that the majority of the carbon emissions may be explained by including an energy variable in the model. The rise in energy usage is related to greater levels of income, leading to a higher general level of consumption that has been triggering energy demand and this opinion has been reinforced in the empirical literature in middle-income states, sub-Saharan areas, and Europe.^17,80–83 Wu et al.⁸⁴ have reacted in a mixed dimension to the energy-emission tie, noting that increasing energy use contributes a vital role in inducing CO₂ emissions, however, as environmental regulatory levels grow, the rising impact decreased. Hence, they further asserted that CO₂ emissions increases are restricted by effective environmental regulation.

According to estimates, industrialization and carbon emissions are directly related. Industrialization's coefficient is statistically significant and positive. According to the calculated value of the coefficient, carbon emissions rise by 0.0113 when industrialization rises by 1%. It shows that the environment in SSA nations has gotten worse as a result of industrialization. This conclusion holds up under many equation parameters. It confirms that industrialization damages the environment in poor countries, according to certain earlier studies. The report from⁸⁵ confirms that industrial development promotes CO₂ emission, but this motivating impact decreases following environmental regulations. By (i) contributing to growth in the overall growth, (ii) having a large-scale impact on energy use, and (iii) using carbon-intensive manufacturing processes, industrialization has a positive influence on CO₂ emissions. Growth is the primary origin of sharp increases in CO₂ emission; growth generally has an increasing impact on CO₂ emission.^86,87 When capital accumulates, more possessions are diverted from agriculture to manufacturing toward increasing the production of domestic manufactured goods. The transition from agronomy, extraction, and light production to resource-related heavy production happens early in the industrialization process, with changes primarily in the scope and mix of production instead of the pace of technology advancement.⁸⁸

POP, a population indicator, has a long-term impact on carbon emissions. POP has been used by Bahizire et al.²⁷ as a measure of population increase. The statistically significant positive coefficient for POP indicates that an increase in POP does not affect CO2 emissions. From this, it can be said that population growth is considered one of the vital factors responsible for SSA CO₂ emissions. This correlation is consistent with several other studies.^89–92 Population expansion harms the environment via the economy's scale, structure, and composition of output and usage (increasing CO₂ emissions) according to Zhu and Peng.⁹⁰ Also, Shi⁸⁹ stated that demographic change in emissions could also be influenced by consumption trends correlated with the population's age composition. To capture this process, the labor-age population ratio was applied to the model. Countries with a higher percentage of the working-age population have been proposed to consume more energy and resources, thus producing more emissions. This coefficient verifies a positive relationship between the proportion of the working-age population and emissions. However, the effect of population change on emissions remains substantially high as the population increases.

According to the stated results in Table 4, the outcomes show mixed results, with CO2 emissions decreasing by −0.0128, −0.2863, and −0.00002, respectively, for every 1% rise in corruption control, regulatory quality, and the rule of law. Theoretically, this is the outcome that the IQ should have produced. Numerous methods may be used to explain how institutions help environmental quality. Because corruption is reduced, regulations are of high quality, and the rule of law encourages understanding of environmental issues among citizens of countries and environmental interest groups, the quality of institutions is anticipated to help reduce pollution. As a result, stricter environmental laws are encouraged. Environmental pollution is decreased as a result of increased public awareness and effective application of environmental regulation legislation. Better IQ fosters economic freedom and market economies, which in turn foster environmental quality. It also displays respect for the rule of law and human life. Strong and effective institutions promote the adoption of renewable energy technology and aid in the proper implementation of energy regulations. These findings validate the empirical results of Gani and Abid.^38,93 They suggested that in SSA, IQ measures such as corruption prevention, quality of regulation, and the rule of law substantially reduce CO₂ discharges. The findings have shown that these variables have a minimal positive influence on the efficiency of the atmosphere and cause suboptimal government policy effects. The proof that regulatory quality significantly limits the growth of CO₂ discharges affirms the notional claim that regulatory quality increases the strictness of environmental rules and thereby reduces pollution.

Unsimilar to this, political stability, government efficacy, and voice and transparency all have a positive and significant impact on CO2 emissions, indicating that when these factors rise, emissions also rise. One explanation for the IQ variable's beneficial influence on CO2 emissions may be the greater levels of political stability and government effectiveness in these nations, which give the populace more civil rights and political freedom.⁵ The findings may also show that these countries’ political environments are less concerned with concerns affecting the common public, such as pollution from CO2 emissions. Although the political and administrative structures in these nations may have taken steps to prioritize environmental quality, it may still take some time for the bureaucracy's thinking on environmental issues to change.

Causal nexus assessment

The research uses Dumitrescu and Hurlin⁴⁸ to analyze the proposition of Granger non-causality which is given in Table 6 to test for the track of causation in the heterogeneous panel results. Evidence from the analysis shows that in almost all pairings, the null proposition of Granger non-causality is dismissed at a 1% significance level, except for the causality of carbon dioxide emissions from economic development. The revelation from the test of causality proves that a one-route causal pathway from industrialization to CO₂ emission is located on the causal route. An analysis by Wang et al.⁹⁴ found that industrialization and CO₂ emissions are unidirectionally related. They explained that R&D recommendations were made for basic research in the manufacturing and energy industries.

Table 6.

DH causality results.

Causal test	W-bar stats	Z-bar tilde stats	Causality outcome
CO₂ → IND IND → CO₂	1.1523 2.3514	0.2233 4.1087^z	One-way
CO₂ → GR GR →CO₂	1.6196 1.4791	1.7375^x 1.2821	One-way
CO₂ → ENE ENE →CO₂	3.3615 3.2539	7.3811^z 7.0327	One-way
CO₂ → POP POP →CO₂	2.0402 14.2295	3.1003^z 42.5932^z	Two-way
CO₂ → COC COC →CO₂	2.9078 1.9376	5.9114^z 2.7679^z	Two-way
CO₂ → GE GE →CO₂	2.4041 1.8841	4.2794^z 2.5944^z	Two-way
CO₂ → PS PS →CO₂	2.8667 1.9389	5.7779^z 2.7722^z	Two-way
CO₂ → RQ RQ →CO₂	2.4633 2.8934	4.4711^z 5.8644^z	Two-way
CO₂ → ROL ROL →CO₂	2.0871 1.9691	3.2521^z 2.8697^z	Two-way
CO₂ → VA VA →CO₂	2.0387 2.2628	3.0952 ^z 2.8697^z	Two-way

Note: z, x, and y signifies p < 0.01; p < 0.05; p < 0.1.

ENE, energy usage; IND, industrialization; POP, population growth.

The outcome also presented a unidirectional causality regarding income and emission. This relationship aligns with the empirical evidence found in South Africa and Malaysia.^95,96 The conclusion to be drawn here is that any measures to reduce the emissions of pollutants would affect growth. However, this is not to say that SSA countries must compromise their environmental targets for growth. Therefore, the option is to increase the level of cleaner production and consumption in the SSA countries, not only in the energy sector but also in other sectors such as industry, transport, and services, which are closely linked to environmental degradation.

The most interesting finding, however, is possible that there is a sign of a single-route causality in SSA that runs from emissions to energy use. Intuitively, the opposite may appear to be predicted, as energy use is the utmost substantial basis of carbon dioxide emissions. This may be because 30% of SSA CO₂ emissions are accounted for by the energy production and mining industries, with only a marginally higher portion of the manufacturing sector accounting for it.⁹⁷ Emissions of carbon dioxide tend to precede energy consumption, given that the bulk of the emissions results from the production of electricity. Also, SSA is a net importer of petroleum and natural gas, accounting for a large portion of electricity production. Therefore, raising predictions of energy usage is not counterintuitive to past emissions. In their report, Soytas and Sari⁹⁸ confirm and record these findings’ implications by suggesting a single-path causality flowing from CO2 emissions to energy use in the long term and due to the overuse of fossil fuel energy.

A dual causality move between population and emissions of CO₂ exists. Therefore, population growth is seen as one of Africa's focal drivers of CO2 discharges.⁹⁹ A high birth rate leads to a high level of energy consumption that causes environmental deterioration. A study by Li et al.¹⁹ found that the rate of energy consumption has a positive bearing on CO₂ emissions as the population grows. They suggested in their study that while the rate of CO₂ emissions is growing due to population growth, policy makers should develop energy conservation policies and strategies for reducing carbon to help protect the environment from emissions. The current result also confirms the study of Asumadu-Sarkodie and Owusu .⁹⁹

Regarding the causal ties and the relationship between IQ indicators and emissions, the findings specify a significant dual causal correlation. The levels of CO₂ discharges may result in institutional reformations and restructuring to address climate change by governments and other stakeholders. For example, the health and well-being of people in China have been adversely affected by high CO₂ emissions in addition to its rising economic costs for all stakeholders being the most notable emitter of CO₂ globally. Hence, there are institutional-backed reformations and restructuring in China toward addressing its energy and industrial policies to cope with high CO₂ emissions as noted by Zheng et al.¹⁰⁰ These include engaging in international initiatives on climate change governance, investing in the renewable energy industry, constructing energy-efficient buildings, designing renewable energy-driven transport systems, and switching industries from hydrocarbon-based to clean energy.¹⁰¹ These structural changes led to the plateauing of CO₂ emissions from 2012 onwards.¹⁰² The overall causality nexus is presented in Figure 1.

Figure 1.

Causal ways, ↔, → signifies a bidirectional and unidirectional causal way, respectively.

Conclusion

Through the deployment of advanced panel data series spanning over 23 years (1996–2019), this study examined the connections between IQ measures and CO2 emissions among the SSA countries in an active EKC environment. The CS-ARDL technique was deployed to analyze the long-term connection between IQ measures and CO2 emissions after the verification of cointegration. To evaluate the long- and short-run effects, the analysis used the CS-ARDL test since the variables are combined at a mixed stage of integration. The Westerlund cointegration test demonstrates the cointegration between the dependent and independent variables. Also, in the long run, the DH causality is used as a robustness test for the CS-ARDL. Findings from the CS-ARDL first reveal the existence of the EKC. Additionally, the effects of IQ variables evidenced mixed results with the empirical findings, observing that CO2 emissions are substantially reduced by corruption control, regulatory quality, and the rule of law. Further explanations specify that instead of exhibiting CO2 reduction functions, an index surge in the level of political stability, the voice of accountability of the populace, and government effectiveness propels emissions levels. The DH causality test revealed a single-way causal lane connecting CO2 emissions to industrialization, economic development, and energy consumption. There are also two-lane causal flows from CO2 emissions to the environment and all institutional efficiency indices. Thus, according to the findings, SSA countries’ industrial policies are critical in addressing environmental issues caused by CO2 emissions. Also, findings show that the abysmal performance of IQ indices in SSA countries would contribute proportionately to increased CO2 emissions.

Policy recommendations

Following the findings, some distinctive policy implications and recommendations emerge from the study. First, a healthy governance framework is fundamental for environmental protection. The state of institutions determines the implementation and effects of government policies that represent the ability to manage environmental problems. For environmental sustainability, well-defined institutional structures are important because they act as an interceding force to produce a win-win situation. Thus, policy makers need to concentrate on different facets of governance and their relationships to efficiently regulate emissions and decide how their operational consistency can be enhanced. Furthermore, it is recommended that authorities and stakeholders should take steps to safeguard and strengthen environmental protection. This includes enacting tighter environmental regulations on crucial sectors like industry and transport sectors. However, the authorities would need to also ensure that there is adequate investment in cleaner transport modes like green urban mass transport systems. Furthermore, the study advises that government interventions should be directed toward the opening of new villages and cities in the African subregions to achieve an equitably spatial population distribution. This would address the level of the dense population associated with major African cities thereby, reducing the undue demand on existing resources. Generally, the practice of geopolitical favoritism in citing economic resource centers in urban areas should be abandoned in favor of ensuring even population distribution and reducing heavy population clusters and massive rural–urban migration in search of economic opportunities and industrial benefits. For the objective of fully embracing the international birth control movement, Africa should look inward and examine itself. African subregions should impose stricter limits on the operations of foreign investors engaged in polluting businesses in their territories. To ensure that polluting industries bear the full cost of their negative externalities and are accountable to their environment, it is necessary to build a strong and responsive institutional structure. This has been shown to be a significant hindrance to the achievement of a better atmosphere in African cities throughout time. Multinational corporations feel that they can always control current policies by bribing, rather than investing considerably in environmental remediation. So, to keep pollution havens under control in the subregions, stricter environmental rules are needed, as well as the creation of sustainable institutional systems. In the same way, organized, sustained, and strategic investment in sustainable energy industry is required in the SSA economies. In addition, to promote environmental consciousness in African cities, it is necessary to consciously instill environmental concerns into the basic learning curriculum. This will educate the public on the importance of basic environmental stewardship, resulting in a reduction in the degree of misuse and pollution. Lastly, following the validity of the EKC, we recommend a more equitable distribution of income to facilitate the EKC advantage in the SSA region.¹⁰³

Table 7.

Individual countries’ cointegrations analysis results.

Country	F-stats	10% I(0) I(1)	5% I(0) I(1)	1% I(0) I(1)	Decision
Botswana	3.335	2.03 3.13^y	2.32 3.50	2.96 4.26	Cointegration
Gabon	0.691	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
Mauritius	1.800	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
Namibia	6.021	2.03 3.13	2.32 3.50	2.96 4.26^z	Cointegration
South Africa	2.127	2.03 3.13	2.32 3.50	2.96 4.26	İnconclusive<
Zambia	6.832	2.03 3.13	2.32 3.50	2.96 4.26^z	Cointegration
Sudan	3.169	2.03 3.13^y	2.32 3.50	2.96 4.26	Cointegration
Naija	1.687	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
Ghana	3.431	2.03 3.13	2.32 3.50	2.96 4.26^z	Cointegration
Cote d’voire	2.280	2.03 3.13	2.32 3.50	2.96 4.26	İnconclusive
Angola	8.687	2.03 3.13	2.32 3.50	2.96 4.26	Cointegration
Cameroon	1.995	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
Kenya	1.799	2.03 3.13	2.32 3.50	2.96 4.26	No ointegration
Congo DR	7.273	2.03 3.13	2.32 3.50	2.96 4.26^z	Cointegration
Togo	19.943	2.03 3.13	2.32 3.50	2.96 4.26	İnconclusive
Uganda	2.351	2.03 3.13	2.32 3.50	2.96 4.26	İnconclusive
Zimbabwe	5.323	2.03 3.13	2.32 3.50	2.96 4.26^z	Cointegration
Madagascar	5.250	2.03 3.13	2.32 3.50	2.96 4.26^z	Cointegration
B. Faso	1.980	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
Malawi	1.423	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
Mozambique	2.108	2.03 3.13	2.32 3.50	2.96 4.26	İnconclusive
Ethiopia	2.718	2.03 3.13	2.32 3.50	2.96 4.26	İnconclusive
Niger	12.822	2.03 3.13	2.32 3.50	2.96 4.26^z	Cointegration
Guinea	1.466	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
Rwanda	1.373	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
The Gambia	4.575	2.03 3.13	2.32 3.50	2.96 4.26^z	Cointegration
Senegal	1.973	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
Tanzania	1.705	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
Sierra Leone	1.912	2.03 3.13	2.32 3.50	2.96 4.26	No cointegration
Mauritania	21.185	2.03 3.13	2.32 3.50	2.96 4.26^z	Cointegration

Footnotes

Author’s contribution

The first author (Michael Appiah) alongside the second author (Mingxing Li) were responsible for the conceptual construction of the study's idea. The third author (Stephen Taiwo Onifade) alongside the fourth author (Bright Akwasi Gyamfi) handled the introduction and literature sections. The data gathering, preliminary analysis, simulation, and interpretation of the simulated results were carried out by the first and second authors while proofreading and general manuscript editing were joint efforts of all of the authors.

Availability of data and materials

The data for this present study are sourced from the database of the World Governance Indicators (WGIs) (http://info.worldbank.org/governance/wgi/) and British Petroleum (BP) database ().

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 author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Michael Appiah

Stephen Taiwo Onifade

Notes

Appendix

Table A1.

List of nomenclature.

Abbreviations	Meaning
EKC	Environmental Kuznets Curve
AERC	African Economic Research Consortium
SSA	Sub-Saharan Africa
DH	Dumitrescu–Hurlin
CS-ARDL	Cross-Sectionally Augmented Auto Regression Distributed lag
WGIs	World Governance Indicators
IQ	Institutional Quality
LM	Lagrange Multiplier
CH₄	Methane
CS	Cross Section
CSD	Cross Section Dependence
RE	Random Effects
FE	Fixed Effects
OLS	Ordinary Least Squares
SH	Slope homogeneity
CADF	Cross-Sectional Augmented Dickey–Fuller
CIPS	Cross-Sectionally Im, Pesaran and Shin
ECM	Error Correction Model
MG	Mean Group
ARDL	Auto Regression Distributed lag
FMOLS	Fully Modified Ordinary Least Squares
ASEAN	Association of Southeast Asian Nations
AMG	Augmented Mean Group
PVAR	Panel Vector Auto Regression
MENA	Middle East and North Africa
AIC	Akaike Information Criterion
BIC	Bayesian Information Criterion
HQIC	Hannan Quinn Information Criterion
COC	Control of Corruption
GE	Government Effectiveness
PS	Political Stability
RQ	Regulatory Quality
ROL	Rule of law
VA	Voice of Accountability

Michael Appiah, is currently a postdoc researcher at the School of Management, Department of Finance & Economics at Jiangsu University, China where he focuses his research on Sustainable Development & Assessment; with a broad interest in Financial, Developmental, Environmental, and Energy Economics including Econometric Modeling. He is passionate about developmental issues in Africa; especially among sub-Saharan African countries and has a substantial number of papers in top journals.

Mingxing Li, is an Associate Professor at the School of Management, Jiangsu University, China. He lectures at the School of Management at Jiangsu University, China, and focuses his research on enterprises' intellectual property management and strategy. He instructs the graduate students and does some other research.

Stephen Taiwo Onifade (PhD) is a researcher and lecturer at KTO Karatay University, Konya-Turkey, where he focuses on research in Energy and Environmental Economics, Development Studies, with further interest in Econometric Modeling. He is passionate about environmental and developmental issues in Africa; especially among sub-Saharan African countries and some of his research works have been directed towards addressing related challenges in these nations. His research works have appeared in various reputable journals including, Sustainable Development, Journal of cleaner production, the International Journal of Sustainable Development & World Ecology, Resources Policy, and Environmental Science & Pollution Research among others.

Bright Akwasi Gyamfi (PhD) is a researcher in the Faculty of Economics and Administrative sciences in Cyprus International University. His research interest is in the area of Tourism economics, Energy economics, Sustainability and Agriculture economics. His work has appeared in journal such as Environmental Science and Pollution, Heliyon and more recently, Journal of cleaner production, Resources Policy and Energy and Environment.

References

IPCC. AR5 Climate Change 2014: Impacts, Adaptation, and Vulnerability. Available at: https://www.ipcc.ch/report/ar5/wg2/ (accessed October 2022). 2014.

Tateishi

Bragagnolo

de Faria

. Economic and environmental efficiencies of greenhouse gases’ emissions under institutional influence. Technol Forecast Soc Change 2020; 161: 1–2.

Zheng

Abbasi

Salem

, et al. How technological innovation and institutional quality affect sectoral energy consumption in Pakistan? Fresh policy insights from novel econometric approach. Technol Forecast Soc Change 2022; 183: 121900.

Ibrahim

Law

. Social capital and CO2 emission—output relations: a panel analysis. Renewable Sustainable Energy Rev 2014; 29: 528–534.

Lau

L-S

Choong

C-K

Eng

Y-K

. Carbon dioxide emission, institutional quality, and economic growth: empirical evidence in Malaysia. Renewable Energy 2014; 68: 276–281.

Cole

Elliott

Fredriksson

. Endogenous pollution havens: does FDI influence environmental regulations? Scand J Econ 2006; 108: 157–178.

Welsch

. Corruption, growth, and the environment: a cross-country analysis. Environ Dev Econ 2004; 9: 663–693.

Fredriksson

Svensson

. Political instability, corruption and policy formation: the case of environmental policy. J Public Econ 2003; 87: 1383–1405.

Gani

. The effect of trade and institutions on pollution in the arab countries. J Int Trade Law Policy 2013; 12: 154–168.

10.

Ndulu

Soludo

Collier

, et al. The political economy of economic growth in Africa, 1960–2000. In: Frontmatter. Vol. 1. Cambridge: Cambridge University Press, 2007, pp. i–iv.

11.

Appiah

Onifade

Gyamfi

. Building critical infrastructures: evaluating the roles of governance and institutions in infrastructural developments in sub-Saharan African countries. Eval Rev 2022; 0: 1–25.

12.

North

. Institutions. J Econ Perspect 1990; 5: 97–112.

13.

Thorbecke

. Institutions for Inclusive Growth and Development in Sub-Saharan Africa. A paper presented at the JICA Conference, SOAS, University of London. 2013.

14.

Sherani

. In: Aneel

Haroon

Niazi

(eds) Institutional reforms in Pakistan: missing piece of the development puzzle. 70. Lahore: Sang-e-Meel Publisher, 2019, pp. 39–61.

15.

Dada

Ajide

Sharimakin

. Shadow economy, institutions and environmental pollution: insights from Africa. World J Sci Technol Sustainable Dev 2021.

16.

Hassan

Khan

SU-D

Xia

, et al. Role of institutions in correcting environmental pollution: an empirical investigation. Sustainable Cities Soc 2020; 53: 101901.

17.

Khan

Weili

Khan

. Recent advances in energy usage and environmental degradation: does quality institutions matter? A worldwide evidence. Energy Rep 2021; 7: 1091–1103.

18.

Appiah

Frowne

. Financial development, institutional quality and economic growth: evidence from ECOWAS countries. Organ Markets Emerging Econ 2020; 11: 6–17.

19.

Appiah

Dodoo

RNA

. A co-integration approach to the study of economic growth and environmental impact in Ghana. NMIMS Manege Rev 2020; 38: 31–47.

20.

Stavropoulos

Wall

. Environmental regulations and industrial competitiveness: evidence from China. Appl Econ 2018; 50: 1378–1394.

21.

Zhu

, et al. Energy and environmental efficiency measurement of China’s industrial sectors: a DEA model with non-homogeneous inputs and outputs. Energy Econ 2019; 78: 468–480.

22.

Abbasi

Shahbaz

Zhang

, et al. Analyze the environmental sustainability factors of China: the role of fossil fuel energy and renewable energy. Renewable Energy 2022; 187: 390–402.

23.

Bekun

Gyamfi

Onifade

, et al. Beyond the environmental kuznets curve in E7 economies: accounting for the combined impacts of institutional quality and renewables. J Cleaner Prod 2021; 314: 127924.

24.

Huynh

Hoang

. Foreign direct investment and air pollution in Asian countries: does institutional quality matter? Appl Econ Lett 2019; 26: 1388–1392.

25.

Zhang

Ozturk

Ullah

. Institutional factors-environmental quality nexus in BRICS: a strategic pillar of governmental performance. Econ Res-Ekonomska Istraživanja 2022: 1–13.

26.

Sadik-Zada

Loewenstein

. Drivers of CO2-emissions in fossil fuel abundant settings:(pooled) mean group and nonparametric panel analyses. Energies 2020; 13: 3956.

27.

Bahizire

Fanglin

Appiah

, et al. Research on the advance role of institutional quality on achieving environmental efficiency in sub-Saharan Africa. Front Environ Sci 2022; 10.

28.

Udemba

. Mitigating environmental degradation with institutional quality and foreign direct investment (FDI): new evidence from asymmetric approach. Environ Sci Pollut Res 2021; 28: 43669–43683.

29.

Neumayer

. Do democracies exhibit stronger international environmental commitment? A cross-country analysis. J Peace Res 2002; 39: 139–164.

30.

Kirkpatrick

Parker

. Regulatory impact assessment and regulatory governance in developing countries. Public Adm Dev: Int J Manage Res Pract 2004; 24: 333–344.

31.

Aron

. Growth and institutions: a review of the evidence. World Bank Res Obs 2000; 15: 99–135.

32.

Makdissi

Wodon

. Environmental regulation and economic growth under education externalities. J Econ Dev 2006; 31: 45–51.

33.

Barrett

Graddy

. Freedom, growth, and the environment. Environ Dev Econ 2000; 5: 433–456.

34.

Ward

. Liberal democracy and sustainability. Env Polit 2008; 17: 386–409.

35.

D-J

Zha

D-L

, et al. A nonparametric distance function approach with endogenous direction for estimating marginal abatement costs of CO2 emissions. J Manage Sci Eng 2021.

36.

Torras

Boyce

. Income, inequality, and pollution: a reassessment of the environmental Kuznets curve. Ecol Econ 1998; 25: 147–160.

37.

Midlarsky

. Democracy and the environment: an empirical assessment. J Peace Res 1998; 35: 341–361.

38.

Gani

. The relationship between good governance and carbon dioxide emissions: evidence from developing economies. J Econ Dev 2012; 37: 77–93.

39.

Akhbari

Nejati

. The effect of corruption on carbon emissions in developed and developing countries: empirical investigation of a claim. Heliyon 2019; 5: 1–9.

40.

Osabuohien

Efobi

Gitau

. Beyond the environmental kuznets curve in Africa: evidence from panel cointegration. J Environ Policy Plann 2014; 14: 517–538.

41.

Jung

. Institutional quality, FDI, and productivity: a theoretical analysis. Sustainability 2020; 12: 7057.

42.

Hussain

Usman

, et al. Re-investigation of the resource curse hypothesis: the role of political institutions and energy prices in BRIC countries. Resour Policy 2020; 69: 101833.

43.

Azam

Liu

Ahmad

. Impact of institutional quality on environment and energy consumption: evidence from developing world. Environ Dev Sustainability 2021; 23: 1646–1667.

44.

Ali

Zeqiraj

Lin

, et al.

Does quality institutions promote environmental quality?

Environ Sci Pollution Res 2019; 26: 10446–10456.

45.

Awais

Mahmood

. Science of the total environment effect of natural resources, renewable energy and economic development on CO2 emissions in BRICS countries. Sci Total Environ 2019; 678: 632–638.

46.

Asif

Majid

. Institutional quality, natural resources and FDI: empirical evidence from Pakistan. Eurasian Bus Rev 2018; 8: 391–407.

47.

Westerlund

Edgerton

. A panel bootstrap cointegration test Econ Lett 2007; 97: 185–190.

48.

Dumitrescu

E-I

Hurlin

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

49.

Breusch

Pagan

. The Lagrange multiplier test and its applications to model specification in econometrics. Rev Econ Stud 1980; 47: 239–253.

50.

Pesaran

. General diagnostic tests for cross-sectional dependence in panels. University of Cambridge, Cambridge Working Papers in Economics, 435. 2004.

51.

Baltagi

Feng

Kao

. A Lagrange multiplier test for cross-sectional dependence in a fixed effects panel data model. J Econom 2012; 170: 164–177.

52.

Pesaran

. Testing weak cross-sectional dependence in large panels. Econom Rev 2015; 34: 1089–1117.

53.

Nickell

. Biases in dynamic models with fixed effects. Econom: Journal of the Econom Society 1981; 49: 1417–1426.

54.

Pesaran

Smith

. Estimating long-run relationships from dynamic heterogeneous panels. J Econom 1995; 68: 79–113.

55.

Pesaran

Ullah

Yamagata

. A bias-adjusted LM test of error cross-section independence. Econom J 2008; 11: 105–127.

56.

Pesaran

Yamagata

. Testing slope homogeneity in large panels. J Econom 2008; 142: 50–93.

57.

Swamy

. Efficient inference in a random coefficient regression model. Econom: J Econom Soc 1970; 38: 311–323.

58.

Pesaran

. A simple panel unit root test in the presence of cross-section dependence. J Appl Econom 2007; 22: 265–312.

59.

Abbasi

Adedoyin

Radulescu

, et al. The role of forest and agriculture towards environmental fortification: designing a sustainable policy framework for top forested countries. Environ Dev Sustainability 2022; 24: 8639–8666.

60.

Bozkaya

Onifade

Duran

, et al.

Does environmentally friendly energy consumption spur economic progress: empirical evidence from the nordic countries?

Environ Sci Pollution Res 2022; 29: 1–11.

61.

Gyamfi

Onifade

Ofori

. Synthesizing the impacts of information and communication technology advancement and educational developments on environmental sustainability: a comparative analyses of three economic blocs—BRICS, MINT, and G7 economies. Sustainable Dev 2022; 30: 1–15.

62.

Khatir

Onifade

, et al. Reviewing trade openness, domestic investment, and economic growth nexus: contemporary policy implications for the MENA region. Revista Finanzas y Política Econ 2022; 14(2).

63.

Onifade

Retrospecting on resource abundance in leading oil-producing African countries: how valid is the environmental Kuznets curve (EKC) hypothesis in a sectoral composition framework?

Environ Sci Pollution Res 2022; 29(1).

64.

Khan

Zakari

Zhang

, et al. A study of trilemma energy balance, clean energy transitions, and economic expansion in the midst of environmental sustainability: new insights from three trilemma leadership. Energy 2022; 248: 123619.

65.

Onifade

Gyamfi

Ilham

, et al.

Re-examining the roles of economic globalization on environmental degradation in the E7 economies: are human capital, urbanization, and total natural resources essential components?

Resour Policy 2021; 74(102435).

66.

Chudik

Mohaddes

Pesaran

, et al. Long-run effects in large heterogeneous panel data models with cross-sectionally correlated errors. Emerald Group Publishing Limited, 2016.

67.

Chudik

Pesaran

. Common correlated effects estimation of heterogeneous dynamic panel data models with weakly exogenous regressors. J Econom 2015; 188: 393–420.

68.

Chudik

Mohaddes

Pesaran

, et al.

Is there a debt-threshold effect on output growth?

Rev Econ Stat 2017; 99: 135–150.

69.

Onifade

Bekun

Agboola

, et al.

How do technological innovation and renewables shape environmental quality advancement in emerging economies: an exploration of the E7 bloc?

Sustainable Dev 2022; 30: 1–13.

70.

Gyamfi

Bekun

Onifade

, et al. Significance of air transport to tourism-induced growth hypothesis in E7 economies: exploring the implications for environmental quality. Tourism: Int Interdiscip J 2022; 70: 339–353. https://hrcak.srce.hr/279096

71.

Onifade

Alola

. Energy transition and environmental quality prospects in leading emerging economies: the role of environmental-related technological innovation. Sustainable Dev 2022; 30: 1766–1778.

72.

Friedman

. The use of ranks to avoid the assumption of normality implicit in the analysis of variance. J Am Stat Assoc 1937; 32: 675–701.

73.

Çevik

Onifade

Erdoğan

, et al. An empirical retrospect of the impacts of government expenditures on economic growth: new evidence from the Nigerian economy. J Econ Struct 2020; 9: 6.

74.

Asongu

Onifade

, et al. Revisiting the trade and unemployment nexus: empirical evidence from the Nigerian economy. J Public Affairs 2020; 20: e2053.

75.

Hakan

Onifade

Savaş

. Modeling the impacts of MSMEs’ contributions to GDP and their constraints on unemployment: the case of African’s most populous country. Stud Bus Econ 2022; 17: 154–170.

76.

Nasir

Huynh

TLD

Tram

HTX

. Role of financial development, economic growth & foreign direct investment in driving climate change: a case of emerging ASEAN. J Environ Manag 2019; 242: 131–141.

77.

Erdoğan

Alagöz

Onifade

, et al. Renewables as a pathway to environmental sustainability targets in the era of trade liberalization: empirical evidence from Turkey and the Caspian countries. Environ Sci Pollut Res 2021; 28: 1–12.

78.

Zoundi

. CO2 Emissions, renewable energy and the environmental Kuznets curve, a panel cointegration approach. Renewable Sustainable Energy Rev 2017; 72: 1067–1075.

79.

Appiah

Korankye

. Modeling the linkages among CO2 emission, energy consumption, and industrialization in sub-Saharan African (SSA) countries. Environ Sci Pollut Res 2021; 28: 38506–38521.

80.

Alola

Acet

Onifade

, et al. Environmental aspect of energy transition and urbanization in the OPEC member states. Environ Sci Pollut Res 2021; 28: 17158–17169.

81.

Erdoğan

Onifade

Altuntaş

, et al.

Synthesizing urbanization and carbon emissions in Africa: how viable is environmental sustainability amid the quest for economic growth in a globalized world?

Environ Sci Pollut Res 2022; 29: 1–14.

82.

Gyamfi

Bekun

Balsalobre-Lorente

, et al.

Beyond the environmental kuznets curve: do combined impacts of air transport and rail transport matter for environmental sustainability amidst energy use in E7 economies?

Environ Dev Sustainability 2022; 24: 11852–11870.

83.

Alola

Onifade

Energy innovations and pathway to carbon neutrality in Finland. Sustainable Energy Technol Assess 2022; 52: 102272.

84.

Ren

, et al. How do energy consumption and environmental regulation affect carbon emissions in China? New evidence from a dynamic threshold panel model. Resour Policy 2020; 67: 101678.

85.

Cheng

. Urbanization, technology innovation and economic growth. Stat Res 2009; 26: 40–46.

86.

Abbasi

Hussain

Redulescu

, et al. Does natural resources depletion and economic growth achieve the carbon neutrality target of the UK? A way forward towards sustainable development. Resour Policy 2021; 74: 102341.

87.

Zhu

Liu

Tian

, et al. CO2 Emissions from the industrialization and urbanization processes in the manufacturing center Tianjin in China. J Cleaner Prod 2017; 168: 867–875.

88.

Wang

Zhang

. Does increasing investment in research and development promote economic growth decoupling from carbon emission growth? An empirical analysis of BRICS countries. J Cleaner Prod 2020; 252: 1–17.

89.

Shi

. The impact of population pressure on global carbon dioxide emissions, 1975–1996: evidence from pooled cross-country data. Ecol Econ 2003; 44: 29–42.

90.

Zhu

Peng

. The impacts of population change on carbon emissions in China during 1978–2008. Environ Impact Assess Rev 2012; 36: 1–8.

91.

Zheng

Walsh

. Economic growth, urbanization and energy consumption—a provincial level analysis of China. Energy Econ 2019; 80: 153–162.

92.

Awan

Abbasi

Rej

, et al. The impact of renewable energy, internet use and foreign direct investment on carbon dioxide emissions: a method of moments quantile analysis. Renewable Energy 2022; 189: 454–466.

93.

Abid

. Impact of economic, financial, and institutional factors on CO2 emissions: evidence from sub-Saharan Africa economies. Util Policy 2016; 41: 85–94.

94.

Wang

Rasool

Zhang

, et al. Dynamic linkage among industrialisation, urbanisation, and CO₂ emissions in APEC realms: Evidence based on DSUR estimation. Structural Change and Economic Dynamics 2020; 52: 382–389.

95.

Ang

. Economic development, pollutant emissions and energy consumption in Malaysia. J Policy Model 2008; 30: 271–278.

96.

Menyah

Wolde-Rufael

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

97.

Tunc

Türüt-Aşık

Akbostancı

. CO2 Emissions vs. CO2 responsibility: an input–output approach for the Turkish economy. Energy Policy 2007; 35: 855–868.

98.

Soytas

Sari

. Energy consumption, economic growth, and carbon emissions: challenges faced by an EU candidate member. Ecol Econ 2009; 68: 1667–1675.

99.

Asumadu-Sarkodie

Owusu

. Energy use, carbon dioxide emissions, GDP, industrialization, financial development, and population, a causal nexus in Sri Lanka: with a subsequent prediction of energy use using neural network. Energy Sources, Part B: Econ Plann Policy 2016; 11: 889–899.

100.

Zheng

Yonglong

Jingjing

, et al. Drivers of change in China’s energy-related CO2 emissions. Proceedings of the National Academy of Sciences 2020; 117(1): 29–36.

101.

Zheng

Coffman

, et al. Regional development and carbon emission in China. Energy Econ 2019; 81.

102.

Guan

Meng

Reiner

, et al. Structural decline in China’s CO 2 emissions through transitions in industry and energy systems. Nat Geosci 2018; 11: 551–555.

103.

Zhu

, et al. Energy and environmental efficiency of China’s transportation sectors under the constraints of energy consumption and environmental pollutions. Energy Econ 2020; 89: 104817.