Navigating the paths to sustainable environments via energy security,renewable energy and economic complexity: Evidence from array of pollution metrics

Abstract

Empirical narratives pertaining to the implications of energy security (ENS) on environmental quality (EQ), particularly in the context of Emerging Seven (E7) countries, are currently understudied. Likewise, the influence of renewable energy (RWN), economic complexity (ECI) and environmental control technology (EVT) on EQ is still unsettled. Thus, based on arrays of pollution metrics (ecological footprint (ECF), carbon emission damage (CO₂D) and consumption-based carbon emissions (CCO₂) and annual panel series spanning 1990–2018, this study provides fresh insights into the implications of the listed variables on EQs in E7 countries. Accordingly, the following empirical narratives were established: (i) The effects of the enlisted panel series on EQ are sensitive to the peculiarities of each pollution metric. (ii) Estimates of the pool mean group autoregressive distributed lag (PMG-ARDL) model and those of cross-sectional augmented autoregressive distributed lag (CS-ARDL) underscore the important roles of ENS and RWN on EQ, particularly on ECF; and ECI on CO₂D. While EVT provides insignificant carbon emission reducing effects across all variants of EQ indicators, population and economic growth are all time-significant pollution enablers. (iii) The outcomes of the panel quantile regression demonstrate that ENS enhances EQ only at the 25th and 50th quantiles and accelerates pollution at the 75th and 95th quantiles. RWN enhances EQ across all quantiles while ECI deaccelerates ECF and CO₂D and accelerates CCO₂ across all quantiles. Furthermore, EVT accelerates ECF and CO₂D while it deaccelerates CCO₂ across all quantiles. Meanwhile, the unpleasant effects of growth and population are reemphasized. Policy strategies to ensure a sustainable environment in E7 have been highlighted.

Keywords

Environmental quality energy security renewable energy economic complexity environmental technology E7 countries

Introduction

Environmental decay and ecological deficit are among the toughest challenges that countries all over the globe are constantly grappling with. It is indeed very worrisome and an issue of critical concern how the global environmental ecosystem is being compromised uncontrollably chiefly on the instance of human activities. On account of several prior studies,^1–7 the quest for economic advancements and overall development seems to have blindfolded humankind, leading to compromised ecosystems both in developed and developing countries. To respond to the mounting challenges of ecological imbalances, global economies have entered into several international conventions/treaties that were conscientiously constituted and crafted to ensure comprehensive abatement of continuous environmental decay. Such treaties, including the Kyoto protocol in 1997, the Paris Agreement in 2015 and the Intergovernmental Panel on Climate Change (IPCC) in 2018 were all crafted to ensure a sustainable environment for the benefit of all humankind. In addition, a historic conference of Paris (COP21) provides critical objectives, including keeping global temperature within 2°C above the preindustrial level. The agreement also craves a further decline in global temperature to at least 1.5°C.⁸^,⁹ Likewise, the commitment to sustainable ecosystems and healthier environments occupies the 13th spot in the United Nations’ sustainable development goals (SDGs).

Unarguably, the Emerging seven (E7) countries consisting of China, India, Brazil, Turkey, Russian federation, Mexico and Indonesia are not just among the world's fastest growing economies but are prominently among the world's top carbon emitters. Specifically, the E7 countries are among the fastest growing economies in terms of population and economic growth¹⁰ and, at the same time, are among the world's most compromised ecosystems.¹¹ Likewise, based on the report of Earth System Science Data (ESSD [2020]), the E7 countries jointly contributed approximately 45% to global greenhouse emissions. It is also established that mainly due to high energy consumption occasioned by the quest for development, the E7 countries continue grappling with monumental environmental decay. Likewise, the E7 countries are ranked among the world top 10 on the basis of total ecological footprint (ECF) burden. Accordingly, China, India, Russian federations, Brazil, Indonesia and Mexico are ranked 1st, 3rd, 4th, 6th, 7th and 8th, respectively. Meanwhile, Turkey is ranked a distance of 14th position. Likewise, all the E7 countries except Brazil and the Russian Federation are biocapacity debtors. That is, a precarious situation whereby a country's ECF exceeds its biocapacity. Accordingly, China, India, Turkey, Mexico and Indonesia are indebted biocapacities of 311%, 171%, 156%, 106% and 38%, respectively. Brazil and Russian federation enjoy biocapacity reserves of 233% and 26%, respectively¹. In terms of energy security (ENS), there is an impressive rise in most E7 countries. For instance, access to electricity in India increased from 43% to almost 85%, while in Indonesia, it increased to 62% in 1990² (see Figure 1 for more insights). Obviously, for countries with strong population growth, such improvements in the share of the population with access to electricity are impressive owing to the potential environmental enhancing effects of ENS. Therefore, it is critical to understand the implications of ENS and other enlisted series on environment qualities (EQs) in E7 countries and accordingly provide a workable policy framework to reposition the E7 countries on the path of environmental sustainability. Additionally, the urgent need for the continuous reappraisal of available policy options in these countries to ensure long-lasting sustainable ecosystems and overall carbon neutrality is critical.

Figure 1.

Access to electricity.

Irrespective of the daunting environmental pollution experiences of the E7 countries, it is surprising that empirical enquiries pertaining to the constituent factors of environmental decay in these emerging countries are largely understudied. Furthermore, empirical expositions of the critical factors that could salvage the compromised ecosystem within the E7 countries are also not readily available. More challenging is the unconvincing positions and ambiguities in related prior empirical studies, which, to a very large extent, has made the issue bordering on the determinants of environmental decay an open debate. Accordingly, Akadiri et al.,¹² Li et al.¹³ and Mujtaba et al.¹⁴ reveal that renewable energy (RWN), economic growth and economic complexities (ECIs) contribute to growing ecological deficits in China, 147 countries and the Middle East and North Africa. Similar narratives were exposed in Awosusi et al.¹⁵ and Hamid et al.,¹⁶ suggesting that affluence, non-RWN and technological innovativeness are fundamental pollution enablers in Brazil, Russian Federation, India, China and South Africa (BRICS) countries. However, the reports of Akalin et al.¹⁷ and Chien et al.¹⁸ exonerate RWN from the negative implications of fossil fuel and economic growth on EQ in the newly industrialized nations and ASEAN countries, respectively. To further deepen the contradictions, Ali et al.¹⁹ rectify that economic growth does not contribute significantly to environmental decay in Malaysia at all times. Zeraibi et al.²⁰ suggest that higher renewable electricity and technological innovations are enhanced, while increasing economic growth and financial development mitigate EQ in ASEAN-5 nations. A similar assertion was extended by Zhao et al.,²¹ suggesting that energy poverty eradication (ENS) and technological advancement encouraged green growth in China.

Another critical incentive for the current evaluation is the near absence of prior studies that integrated and considered the implications of metrics such as ENS, RWNECI and environmental control technology (EVT) on EQ, particularly in E7 countries. On this premise, the study aims at unravelling the potential implications of the listed metrics as well as economic growth (affluence) and population on environmental health in these countries. To further provide clearer perspectives, we outlined the following research question: (i) Can the E7 countries achieve a clean environment by ensuring higher ENS? (ii) To what extent can the transition to RWN consumption and commitment to ECI ensure sustainable ecosystems in these economies? (iii) What is awaiting the E7 countries ecosystem when its EVT progresses? and (iv) Do affluence and population explosion constitute significant threats to EQ in E7? Overall, we conjectured that improvements in ENS, transition to RWN consumption, ECI and improved EVT will invoke a healthier environment while affluence and a growing population constitute a drag on EQ in these countries.

Notably, the modest contribution of this study to the literature related to EQ determinants stems from its ability to enlist such critical metrics listed above within the Stochastic Impacts by Regression on Population, Affluence and Technology (STIRPAT) framework of E7 countries. It is pertinent to highlight that prior studies did not trade this path. Meanwhile, the dearth of studies linking ENS to climate change is further highlighted in Jessel et al.,²² suggesting the urgent need for more empirical enquiries in that direction. To further extend the trajectory of knowledge, this study, unlike existing investigations, also engaged arrays of pollution indicators such as the ECF, carbon emission damage (CO₂D) and consumption-based carbon emissions (CCO₂). In addition, unlike prior studies, the current evaluation relied on the barrage of robust panel estimators, including the pool mean group autoregressive distributed lag (PMG-ARDL), the dynamic cross-sectional augmented autoregressive distributed lag (CS-ARDL) and the panel quantile regression (P-QREG) techniques. Expectedly, these holistic approaches lacking in related prior studies were consciously undertaken not only to extend the frontiers of knowledge but also to provide more formidable carbon neutrality pathways for the E7 countries. Consequently, it is expected that the inferences from this commitment will avail the E7 countries of a cutting-edge template that ensures total compliance with all treaties entered by E7 countries. Likewise, policies that will ensure that these countries remain within the track of SDGs 7 and 13 will emerge. Expectedly, beyond the E7 countries, other developing and emerging economies may draw inspiration from the inferences of this study to devise environmental sustainability strategies.

The rest of the study is organized as follows: Review of related literature section reveals the related literature while Methodology section presents the methodology. The penultimate section provides data analysis and discussions while the Conclusion and policy implications section unveils the concluding remarks and policy options.

Review of related literature

Theoretical literature

Environmental economists have proposed some theories pertaining to the dynamic interactions of economic metrics and EQ. Among such theories is the environmental Kuznets curve (EKC) hypothesis.²³ In a nut shell, the EKC theory predicts a U-shaped or an inverse interaction between economic growth and EQ. Technically, the theory suggests that at the initial stage of economic growth, environmental pollution increases to a certain threshold where continuous growth ensures a cleaner environment by reducing pollution.²⁴ Additionally, some researchers²⁵ connected the metrics of technology, innovations, RWN, ECI and ecological balance to the popular STIRPAT framework. Overall, the findings of these related studies are inconclusive. Moreover, there is a dearth of empirical narratives pertaining to the interactions of ENS and the other enlisted metrics with environmental performance in the context of E7 countries. For clarity, we have outlined several recent empirical investigations that considered the relevant factors influencing EQ.

Empirical literature survey

The empirical survey is presented in such a way that it clearly showcases the contradictory submissions of prior studies, which obviously demands further clarification. For instance, Bilgili et al.²⁶ considered the influence of ENS on CO₂D in select Asian economies between 1997 and 2017. Based on the outcomes of the panel quantile regression, they report a significant inverse interaction between pollution and access to electricity. Unarguably, their study did not consider the E7 countries and other pollution indicators. The study also failed to include metrics such as ECI and EVT within the model. Likewise, Li, et al.¹³ and Zhao et al.,²¹ on the basis of a panel of 30 Chinese provinces, document that energy poverty elimination (increasing ENS) and technological innovations contribute significantly to green growth. However, this study not only failed to consider E7 countries but also ignored other variants of pollution indicators and relied on a less robust econometrics algorithm.

For studies in the context of E7 countries, Tong et al.¹⁰ document that energy consumption is the cause of environmental decay in E7 countries, while ul Husnain et al.¹¹ buttressed further that while non-RWN mitigates, RWN enhances EQ in E7 countries. More conflicting narratives in the context of E7 countries emerged from the submissions of Gyamfi et al.²⁷ and Gyamfi et al.²⁸ While the former emphasizes the long-term deleterious effects of renewable and non-RWN on EQ the latter documents an appealing influence of RWN investment on EQ. Additionally, Huang et al.²⁹ report that information and communication technology (ICT) and ECI escalate pollution while RWN reduces pollution in E7 countries. Obviously, these few studies that considered the E7 countries not only failed to include both ENS, ECI and EVT in their analysis, but they also compounded the subsisting contradictions. Moreover, they also considered only a variant of the pollution indicator, therefore compelling the need for further evaluations.

Several other studies have, albeit divergent inferences, considered other determinants of EQs in many other economies. Accordingly, we X-ray such related prior and recent studies. For instance, Usman et al.³⁰ relied on the outcomes of the CS-ARDL model and inferred that electricity consumption spurs ecological deficits in advanced countries. Likewise, Usman and Hammar³¹ relied on the outcomes of AMG and CCEMG panel estimators to emphasize that improvement in technology mitigate EQ in Asian-Pacific Economic Corporation (APEC) countries. Conversely, Zeraibi et al.²⁰ document that renewable electricity and technological innovations enhance EQ in ASEAN-5. This opinion is further enthroned by the submissions of Abbasi et al.,³² Razzaq et al.³³ and Yang et al.³⁴ extolling the appealing impacts of technology on WQ in Pakistan and BICS economies, respectively. Likewise, Nguyen et al.³⁵ and Xin et al.³⁶ rectified that improvements in ECT and innovation expenditure improve EQ in the G-20 economies and US, respectively. However, Awosusi et al.¹⁵ and Umar et al.³⁷ emphasized that innovations spur CO₂ emissions in BRICS countries and China, respectively. More contradictory inferences emerged from the submissions of Ahmad et al.³⁸ and Ding et al.,³⁹ emphasizing the environmental deleterious effects of technological innovativeness and RWN on the environments in twenty-four OECD and G7 countries, respectively. Additionally, Amin et al.⁴⁰ report that for Asian nations, modernization accelerates ecological deficits.

Furthermore, prior studies still provide an open debate on the RWN–EQ nexus; meanwhile, as previously highlighted, the E7 countries were underrepresented. Accordingly, Chien et al.¹⁸ adopted the QARDL model and report that RWN and eco-innovations deaccelerate emissions in the United States. Corroborative inferences emerged from Chien et al.,⁴¹ Dingru et al.,⁴² Gormus and Aydin⁴³ and Hsu et al.⁴⁴ in the context of Asian, BRICS, top 10 innovative economies and China, respectively. Likewise, Chang et al.,⁴⁵ Dogan and Shah⁴⁶ and Zhao et al.²¹ document that RWN and environmental-related technology deescalate environmental decay in Gulf Cooporation Council (GCC) countries and the top five polluted countries, respectively. Furthermore, Atsu et al.,⁴⁷ Mujtaba et al.⁴⁸ and Wang and Dong⁴⁹ identified that RWN improves EQ in South Africa, OECD countries and sub-Saharan African countries, respectively. Iqbal et al.⁵⁰ further comment that RWN and ECT are desirable factors for clean environments in the context of OECD economies. Conversely, Chen and Lee,⁵¹ however, rectify that RWN usage and technological advancement have no significant CO₂ reducing effects globally but for high-income countries. Likewise, Destek et al.⁵² reported that biomass (RWN) increases the ECF but reduces CO₂ emissions in the top five biomass nations. Furthermore, while Murshed et al.⁵³ remark that RWN consumption degrades EQ in G7 countries, Wang et al.⁷ admit that RWN utilization provides a varying reducing influence on the ECF across different economic settings. Likewise, Emre et al.⁵⁴ extended the debate and considered RWN utilization and ECI as desirable EQ enhancing factors. Likewise, He et al.⁵⁵ employed the CS-ARDL for panel datasets spanning 1990–2018 for the top 10 energy transition economies. They rectified that RWN adaptation as well as ECIs deaccelerate carbon emissions.

Undoubtedly, irrespective of the bourgeoning literature pertaining to EQ determinants, the influence of ENS on EQ, particularly in the E7 countries context, is conspicuously lacking in prior studies. Meanwhile, the few existing ones failed to consider other pollution metrics, and they relied largely on conditional mean estimators that are not robust to non-linearities. Thus, the present study bridges the conspicuous lacuna in the literature and therefore makes a significant contribution to knowledge.

Methodology

Data description

For empirical analysis, this study relied on annualized panel datasets that span 1990–2018 cutting across E7 countries. Consequently, a macro panel framework with a larger time dimension (T) and a smaller cross-section (N) dimension are envisaged. In this instance, all relevant prerequisite steps peculiar to such an arrangement (T > N) were adopted in this study for robust inferences. Meanwhile, it is imperative to highlight that the scope of the study is selected due to data availability constraints. Table 1 provides more tables about the relevant panel series.

Table 1.

Data description and sources.

Series	Notation	Unit of measurement	Source
Ecological footprint	ECF	Total footprint per capita	GFN
Carbon emission damage	CO₂D	Adjusted savings: carbon dioxide damage – current US$	WDI
Consumption-based emission	CCO₂	KgCO₂ per GDP	GCA
Energy security	ENS	Access to electricity – % of population	WDI
Renewable energy consumption	RWN	% total final energy consumption	WDI
Economic complexity	ECI	Country complexity index	ECA
Environmental control technology	EVT	Patent on environmental control technology	OECD
Economic growth	AFL	GDP per capita – current US$	WDI
Population	POP	Population, total	WDI

Note: GFN, WDI, GCA, ECD and OECD depict Global Footprint Network, World Development Indicators, Global Carbon Atlas, Economic Complexity Atlas and Organization for Economic Co-operation and Development, respectively.

Model specification

This study considers a STIRPAT framework for E7 countries’ environmental performance when embellished with ENS, RWN, ECI and ECT. Accordingly, relying on insights from prior related studies,²⁶ the overall objective is to empirically ascertain the implications of the enlisted series on environmental performance in the context of E7 countries. On this premise, the typical IPAT framework is projected:

I_{i t} = α x P_{i t}^{a} x A_{i t}^{b} x T_{i t}^{c} x ε_{i t}

(1)

where I, P, A and T denote EQ, population, affluence and technological advancement, respectively, while е underscores the stochastic term.

For this study, the modified STIRPAT framework, where energy poverty, RWN consumption, eco-innovation and ECT were enlisted with affluence and population to rectify their specific influences on environmental performance in E7 countries, is presented. Meanwhile, three EQ indicators (ECF, CO₂D and CCO₂) were selected for comparative expositions and robust policy narratives.

E Q_{i t} = f (E N S_{i t}, R W N_{i t}, E C I_{i t}, E V T_{i t}, A F L_{i t}, P O P_{i t})

(2)

where EQ_it, f, ENS_it, RWN_it, ECI_it, EVT_it, AFL_it and POP_it represent EQ, functional notation, ENS, RWN consumption, ECT, affluence and population, respectively. Notably, several prior studies¹⁸ related to the determinants of environmental performance across several global economies considered most of these metrics significant predictors of EQ. Furthermore, to provide more empirical details, equation (2) is transformed into econometric specifications to represent equation (3).

\ln E Q_{i t} = ϕ_{0} + ϕ_{1} \ln E N S_{i t} + ϕ_{2} \ln R W N_{i t} + ϕ_{3} \ln E C I_{i t} + ϕ_{4} \ln E V T_{i t} + ϕ_{5} \ln A F L_{i t} + ϕ_{6} \ln P O P_{i t} + ε_{i t}

(3)

where the series that have been defined earlier retain their notations, while the subscripts i and t imply cross-sectional dimension and time dimension, b_0, b_{1 –} b₇ and

ε_{i t}

imply coefficients of the constant term, the explanatory variables and the stochastic term, respectively. It is imperative to highlight that all the panel series have been transformed into natural logarithmic values to enable interpretations on the basis of elasticities (see ^3,4,56).

Estimation techniques

Several enhanced panel estimators peculiar to the macro panel (T > N) specification have been adopted in this study given that the study aims to provide cutting-edge policy strategies that will ensure the attainment of SDG7 in E7 countries. In this study, the Pesaran⁵⁷ cross-sectional dependency (CD) procedure and the Pesaran⁵⁸ weak-CD techniques were selected for the CD test Furthermore, the study relied on the Hashem Pesaran and Yamagata⁵⁹ slope heterogeneity (SH) test process to consider the effects of transnational heterogeneity. As earlier identified, these steps became necessary to select the appropriate generation of panel stationarity test procedures that will avail robust estimates.²⁴ The formulae for the CD and SH tests are provided in eqautions (4) and (5a) and (5b), respectively.

C D = \sqrt{\frac{2}{B (B - B)} \sum_{n = 1}^{B - 1}} \sum_{m = n + 1}^{B} A_{n m} {B_{n}^{2}}_{m} \to B (0, 1)

(4)

Δ H T = (N) \frac{1}{2} (2 k) - \frac{1}{2} (\frac{1}{N} S - k)

(5a)

Δ H T = (N) \frac{1}{2} (\frac{2 k - (T - K - 1)}{T + 1}) \frac{1}{2} (\frac{1}{N} S - k)

(5b)

where

B_{n m}^{2}

is the residual-related statistics and A and B denote the time and cross-sectional dimensions, respectively. SH test,

Δ H T

and

Δ A H T

represent the delta tilde and the adjusted delta tilde, respectively.

First-generation panel unit root procedures assume cross-sectional independence of the panel series; as such, the procedures perform poorly amidst CD and heterogeneous slopes among the panel series. Therefore, this study adopted sets of second-generation panel unit root procedures, including the cross-sectional Im, Pesaran and Shin (CIPS)⁶⁰ and the Pesaran cross-section augmented Dickey–Fuller (PesCADF)⁵⁷ processes for its inferences. Accordingly, CIPS and CADF inspection structures are given in equations (6) and (7), respectively.

C I P S (N, T) = \frac{\sum_{i = 1}^{N} t_{i} (N, T)}{N}

(6)

Δ γ_{i t} = α_{i} + ρ_{i} γ_{i t - 1} + β_{i} γ_{i t - 1} + \sum_{j = 0}^{K} θ_{i j} Δ γ_{i t - 1} + \sum_{j = 0}^{K} \partial_{i j} γ_{i t - 1} + ε_{i t}

(7)

Additionally, to extend the trajectory of knowledge in this context, a third-generation panel unit root process⁶¹ that accounts for structural breaks among series was employed. Notably, macroeconomic variables are frequently susceptible to shocks that may arise from uncertainties and other economic cycles. Therefore, neglecting the effects of such outcomes on panel series may lead to inaccurate diagnosis that may result in inconsistent estimates and biased policy prescriptions. The structural breaks (third-generation) unit root test scheme is provided in equation (8).

θ = (\sum_{i = 1}^{N} θ_{t - 1}^{1} Q_{m}^{(λ)} θ_{i - 1})^{- 1} (\sum_{i = 1}^{N} θ_{t - 1}^{1} Q_{m}^{(λ)} θ_{i - 1}), m = (M 1, M 2)

(8)

where

Q_{m}^{(λ)}

represents the (TxT) ‘within’ transformation (annihilator) matrix of the time series of the panel

Υ_{i t}

.⁶¹

Furthermore, the study relied on two enhanced cointegration procedures, Pedroni⁶² and Westerlund,⁶³ to examine if the enlisted panel series coevolve in the long term. Accordingly, these cointegration techniques are Lagrange multiplier (LM) and bootstrap-based cointegration procedures, which also account for heterogeneous slope, time trend, heteroscedastic and individual-specific intercepts, as well as unknown breaks within the cointegrated series.⁶⁴ The functional notation of these error-correction base panel cointegration procedures is provided herewith in equation (9).

Δ χ γ_{i t} = θ_{j}^{i} d_{t} + θ_{i} (χ_{i t} - β_{i}^{1} γ_{i t_{1}} - 1) + \sum_{j = 1}^{k j} θ_{i j} Δ χ_{i t - j} + \sum_{j = q 1}^{k i} γ_{i j} Δ γ_{i t - j} + ε_{i t}

(9)

where $θ_{i}$ is the coefficient of the stochastic term that shows the speed at which the equilibrium is restored and $Δ$ is the difference operator that shows the changes in the panel series over the long run. Accordingly, this procedure avails four critical statistics, namely, the group (G_t, G_a) and panel (P_t, P_a) statistics. The null hypothesis suggests the absence of long-run stable coevolution among the enlisted series. Subsequently, the estimation schemes of the four (4) critical statistics are provided herewith in equations (10) to (13), respectively.

G_{t} = \frac{1}{N} \sum_{i = 1}^{N} \frac{ϖ_{i}}{S_{e} (ϖ_{i})}

(10)

G_{t} = \frac{1}{N} \sum_{i = 1}^{N} \frac{T ϖ_{i}}{1 - \sum_{j = 1}^{k} ϖ_{i j}}

(11)

P_{t} = \frac{ϖ}{s_{e} (ϖ)}

(12)

P_{a} = T_{ϖ}

(13)

In equations (10) and (11), $G_{t}$ and $G_{a}$ represent the statistics that show that there is cointegration in at least one group of the panel series. In equations (12) and (13), $P_{t}$ and $P_{a}$ are statistics that show the evidence of cointegration within the panel series. Instructively, the statistical significance of one or all the statistics provides enough evidence to reject the null of no long-run coevolution among the enlisted series.

To understand the implications of individual explanatory variables on each of the selected response variables (EQ indicators), the study carefully explored arrays of robust and enhanced panel estimators. Among such are the PMG-ARDL, the CS-ARDL and the P-QREG. In specific terms, the PMG-ARDL⁶⁵ and CS-ARDL⁶⁶ are enhanced panel estimators that are robust to heterogeneous effects and endogeneity problems in panel series.³⁰ The procedures are formidable even in the face of cross-sectional exigencies and provide empirical details for both short- and long-term effects.⁶⁷ Furthermore, the PMG-ARDL panel estimator performs efficiently even when the relevant panel series are fractionally integrated. However, the models are inefficient when the series are second-order unit root processes. In line with the aforementioned previous studies, we present the functional form of the PMG-ARDL procedure herewith in equation (14).

\ln Δ γ_{i t} = β_{0} + \ln β_{1} γ_{i t - 1} + \ln β_{2} χ_{i t - 1} + \sum_{i = 1}^{n} θ_{1} \ln Δ γ_{i t - 1} + \sum_{i = 1}^{n} θ_{2} \ln Δ χ_{i t - 1} + ε_{i t}

(14)

where

γ

_it is the dependent variable,

x_{i t} is the independent

variable,

Δ

is the difference operator, ln is the natural logarithm notation and ɛ_it is the stochastic term of the panel series.

\sum_{i = 1}^{n} φ_{i} l n Δ γ_{i t - i}

signifies the short-run dynamics and

β_{1} γ_{i t - 1}

denotes the long-run interactions. Accordingly, the PMG-ARDL process for the current investigation is presented in equation (15).

Δ l n E Q_{i t} = β_{0} + β_{1} l n E Q_{i t - 1} + β_{2} l n E N S_{i t - 1} + β_{3} l n R W N_{i t - 1} + β_{4} l n E C I_{i t - 1} + β_{5} l n E V T_{i t - 1} + β_{6} l n A F L_{i t - 1} + β_{7} l n P O P_{i t - 1} + \sum_{i = 1}^{n 1} θ_{1} Δ l n E Q_{i t - i} + \sum_{i = 0}^{n} θ_{2} Δ l n E N S_{i t - i} + \sum_{i = 0}^{n} θ_{3} Δ l n R W N_{i t - i} + \sum_{i = 0}^{n} θ_{4} Δ l n E C I_{i t - i} + \sum_{i = 0}^{n} θ_{5} Δ l n E V T_{i t - i} + \sum_{i = 0}^{n} θ_{6} Δ l n A F L_{i t - i} + \sum_{i = 0}^{n} θ_{7} Δ l n P O P_{i t - i} + ε_{i t}

(15)

To provide a robustness check on the outcome of the traditional PMG-ARDL technique, the study also adopts the CS-ARDL technique. The CS-ARDL panel estimator proposed by Chudik and Pesaran⁶⁶ also accounts for heterogeneity in panel series, as well as long- and short-run dynamics interactions of relevant variables.^4,67 Other advantages of the CS-ARDL include its ability to produce efficient estimates among fractionally integrated panel series. The model also avails reliable estimates amidst cross-sectional dependences.^3,30 Likewise, the procedure is equally an efficient technique in the case of weak exogeneity that may appear due to the lagged response variable in the specified equation. Furthermore, the technique prevents the endogeneity problem by adding the lagged cross-sectional averages in the equation.³⁰ However, much like the PMG-ARDL, the CS-ARDL breaks down amidst the second-difference panel series, and the model may not adequately address the issues of non-linearities and structural breaks. The baseline functional notation of the CS-ARDL is thus presented in equation (16).

Δ E Q_{i t} = C_{i} + λ_{i} (E Q_{i t - 1} - β_{i}^{1} χ_{i t} - δ_{1 j} E Q_{i t - 1} - δ_{2 j} χ_{t - 1}) + \sum_{j = 1}^{p - 1} θ_{i j} Δ E Q_{i t j} + \sum_{i = 1}^{q - 1} θ_{i j} Δ χ_{i t j} + δ_{1 i} Δ E Q + δ_{2 i} Δ χ_{t} + ε_{i t}

(16)

where

E Q_{i t}

represents the various response variables (ECF, CO₂D and CCO₂), while

χ_{i t}

denotes the long-run coefficients of all the listed explanatory variables. The short-run dynamic coefficients of the explanatory and response variables are denoted by

θ

and

Φ

, respectively, while the long-term coefficients are denoted by

β

, while

ε_{i t}

represents the stochastic error term.

Following recent studies,²⁶ the panel quantile regression technique of Ponomareva⁶⁸ was also adopted in this study to adequately account for potential non-linearity and structural breaks peculiar to economic dynamics. Ideally, economic variables suffer from shocks that arise through economic cycles such as financial crises and the recent COVID-19 global pandemic. Unfortunately, traditional econometric algorithms are incapable of providing a consistent account of such non-linear dynamics among variables. The panel quantile regression model enjoys several advantages over the traditional linear models. Accordingly, unlike the conventional linear models that account for only the conditional mean of the response variable, the quantile regression process allows for the identification of other relevant information within the conditional distributions.⁶⁸ The model also accounts for the influence of any unobserved heterogeneity within the panel series and controls for the unobserved individual impasse.²⁹

Unlike traditional linear models, the quantile regression specification is based on different quantitative points. The characteristics of each point in the conditional distribution can be ascertained by providing a conditional estimate through the quantile regression procedure.²⁶ The quantile regression estimates the conditional median or other quantiles at different percentiles. Equation (17) describes the quantile regression with fixed effects that this study adopted following prior studies.²⁶ Ponomareva⁶⁸ further explained that the combination of quantile regression models with a fixed effects panel data model provides robust outcomes given that it allows for the study of the causal effect of covariates on the distribution of the explained variable (quantile regression) while controlling for some unobserved heterogeneity (fixed effects).

Q_{τ} (m_{i t} | n_{i t}, β_{i o (τ)}) = β_{i o (τ)} + n_{i t}^{k} γ_{o (τ)} \equiv N_{i t}^{k} θ_{i (τ),} t + 1 \dots, k, i = 1, \dots, n a n d τ ε (0, 1)

(17)

where

m_{i t}

denotes the predicted variable,

n_{i t}

signifies the k dimension vector of the independent variable,

γ_{0 (τ)}

is the slope coefficient and

β_{i 0} (τ)

is the individual unique fixed effect term.

In quantile estimations, the estimated coefficients are not assumed to be constants but are regarded as functions with a dependency on several quantiles.²⁶ The sequence of arriving at the estimated coefficients at a particular quantile value is similar to the sequence of arriving at the estimated coefficients of a conventional linear regression model. The main difference is that in quantile regression estimations, the median absolute deviation is minimized. Following the above arguments, the quantile regression framework of the interactions of the dependent variable (EQ) and the enlisted explanatory variables have been outlined in equation (18).

Q_{τ} (E Q_{i t}) = q (α_{i}) + δ_{1} (q) E N S,_{i t} + δ_{2} (q) R W N,_{i t} δ_{3} (q) E C I,_{i t} + δ_{4} (q) E V T,_{i t} + δ_{5} (q) A F L,_{i t} + δ_{6} (q) P O P,_{i t} + n (q,_{i t})

(18)

where Q signifies the conditional quantile level of the response variable and i and t denote cross-sectional and time dimensions, respectively. ɑ indicates the fixed effect of a pure location adjustment effect on the conditional quantiles of the dependent variable, q represents the quantile level of the conditional distribution and n denotes residuals of the estimated quantiles regression. In this study, the relevant parameters were estimated at various quantiles, including the 25th, 50th, 75th and 95th percentiles.

Data estimation and discussions

The outcomes of the summary statistics and correlation matrix have been projected on Table 2 for empirical overview.

Table 2.

Summary statistics and correlation matrix.

Series	lnECF	lnCO₂D	lnCCO₂	lnENS	lnRWN	lnECI	lnEVT	lnAFL	LnPOP
Panel A:
Mean	2.46	23.48	0.36	4.02	2.96	0.39	2.11	8.01	19.33
SD	1.51	1.24	0.52	1.41	0.89	0.43	0.42	1.08	1.06
Minimum	0	21.24	0.12	0	1.15	−0.35	0.84	5.71	17.80
Maximum	6.79	26.72	1.36	4.60	4.07	1.39	3.50	9.67	21.05
Kurtosis	2.83	2.91	7.89	7.16	2.29	2.55	3.86	2.15	1.85
Panel B:
	lnECF	lnCO₂D	lnCCO₂	lnENS	lnRWN	lnECI	lnAFL	lnPOP	lnEVT
lnECF	1.00
lnCO₂D	0.13*	1.00
lnCCO₂	0.11	0.52***	1.00
lnENS	0.07	0.43***	0.10	1.00
lnRWN	−0.77***	−0.33***	−0.32***	−0.08	1.00
lnECI	0.25***	0.34***	0.01	0.24***	−0.29***	1.00
lnAFL	0.68***	0.06	−0.31***	0.04	−0.54***	0.44**	1.00
lnPOP	−0.50***	0.67***	0.42***	0.33***	0.37***	0.03	−0.57***	1.00
lnEVT	0.25***	0.10	0.04	0.09	−0.27***	0.04	0.27***	−0.16**	1.00

Note: ***, ** and * depict statistically significant correlations at the 1%, 5% and 10% significance levels, respectively.

Source: Authors’ computations.

CCO₂: consumption-based carbon emissions; CO₂D: carbon emission damage; ECI: economic complexity; ECF: ecological footprint; ENS: energy security; EVT: environmental control technology; RWN: renewable energy.

The outcomes of the two pre-estimation diagnostic tests (summary statistics and correlation matrix) (Table 2) highlight some important characteristics of the enclosed panel series. Accordingly, the expected values of all the series fall between their minimum and maximum values. This denotes a possible convergence of the enlisted panel series. Furthermore, lnECF is the most dispersed, followed by lnCO₂D while lnEVT is the least dispersed series. The kurtosis statistic, which indicates whether a series is normally distributed or otherwise unveils that among the panel series, lnCO2 and lnAFL deviate from a normal distribution. In this case, a conventional linear econometric technique may not produce consistent estimates when enlisted series are not normally distributed, therefore justifying the application of non-linear techniques that can adequately capture the effects of such abnormal distributions. In addition, the correlation matrix (Panel B) provides preliminary insights about the strength of correlation among the enlisted series. The matrix suggests that lnENS, lnECI, lnAFL, lnEVT and lnPOP are potential positive predictors while lnRWN is a negative predictor of all the EQ indicators. Meanwhile, the problem of multicollinearity is ruled out given that all the correlation coefficients are below 50%.⁵⁶

To provide deeper insights into the characteristics of the enlisted variable and possible interdependences of the cross-sections, the study proceeded with CD and SH tests. Thus, the outcomes of these tests are illustrated in Table 3.

Table 3.

Estimates of cross-sectional dependency (CD) and slope heterogeneity (SH) tests.

Variable	CD	P-value	W-CD	P-value
lnECF	2.07**	0.038	24.01***	0.000
lnCO₂D	24.16***	0.000	24.67***	0.000
lnCCO₂	3.04***	0.002	24.07***	0.000
lnENS	9.85***	0.000	21.60***	0.000
lnRWN	17.09***	0.000	24.62***	0.000
lnECI	7.04***	0.000	11.95***	0.000
lnAFL	21.83***	0.000	24.63***	0.000
lnPOP	11.49***	0.000	24.67***	0.000
LnEVT	2.12**	0.034	23.90***	0.000
Panel B: SH
Model 1:	Delta	p value
	7.084***	0.000
Adj.	8.324***	0.000
Model 2:	Delta	p value
	12.085***	0.000
Adj.	14.202***	0.000
Model 3:	Delta	p value
	8.639***	0.000
Adj.	10.153***	0.000

Note: *** and ** imply significant relationships at the 1% and 5% levels of significance.

Source: Authors’ computations.

Estimates of CD and SH tests (Table 3) rectified that the panel series are cross-sectional dependent. This implies the rejection of the null hypothesis of cross-sectional independent panel series. The SH test (Panel B) also provides overwhelming evidence of transnational heterogeneity of the relevant panel series. This also suggests the rejection of the null hypothesis of a homogeneous slope among the cross-sections. Given the outcomes of the CD and SH tests, the study adopts two second-generation unit root techniques and a third-generation panel unit root procedure. Expectedly, these selected panel techniques are formidable amidst cross-sectional exigencies and structural breaks.

The second-generation panel stationarity test estimates (CIPS and CADF – Table 4)⁶¹ provide some interesting outcomes. Accordingly, the panel datasets are mutually cointegrated between order-zero and order-one. This outcome highlights the suitability of the adopted empirical estimation models (PMG-ARDL) and (CS-ARDL). Specifically, all the EQ variables (lnECF, lnCO₂D and lnCCO₂) as well as lnRWN, lnECI and lnPOP are all first differenced stationary series, while lnENS, lnAFL and lnEVT are level stationary series. Additionally, the outcome of the third-generation panel unit root procedure provides evidence that some of the panel series are stationary processes, while some are unit root processes. The outcome also suggests evidence of structural breaks in the panel series. Specifically, structural breaks affected the series lnECF and lnEVT in 1991, lnCCO₂ in 1996, lnAFL in 2005 and lnENS in 2008. Consequently, the application of the enlisted panel estimators, including the panel quantile regression, has been conscientiously selected to adequately account for the influence of such a phenomenon, thereby leading to robust policy prescriptions. Before the application of the selected panel empirical estimators, efforts were made to confirm the potential long-run coevolution of the relevant panel series. The outcomes of the panel cointegration test based on the Pedroni and Westerlund procedures are illustrated in Table 5.

Table 4.

Panel unit root test results.

Series	CIPS		CADF
Series	CIPS (level)	$Δ$ IPS	CADF (level)	$Δ$ CADF	MinZ stat	P-value	Break date
lnECF	−1.932	−5.197***	−1.633	−3.033***	−15.71***	0.000	1991
lnCO₂D	−1.883	−4.375***	−1.871	−2.389**	1.167	1.000	Na
lnCCO₂	−2.072	−4.859***	−2.014	−2.525**	−3.420***	0.000	1996
lnENS	−4.553***	-	−2.763***	-	−10.92***	0.000	2008
lnRWN	−1.790	−4.311***	−2.126	−2.661***	−0.617	1.000	Na
lnECI	−2.071	−5.439***	−1.739	−2.635***	−0.51	1.000	Na
lnAFL	−2.318*	-	−2.092	−3.263***	−1.357***	0.000	2005
lnPOP	−1.378	−2.312*	−0.274	−3.515***	2.148	1.000	Na
lnEVT	−4.275***	-	−3.163***	-	−18.24***	0.000	1991

Note: ***, ** and * imply rejection of the null hypothesis at the 1%, 5% and 10% levels of significance, respectively. while Na implies an unknown structural break date.

Source: Authors’ computations.

CIPS: cross-sectional Im, Pesaran and Shin; CCO₂: consumption-based carbon emissions; CO₂D: carbon emission damage; ECI: economic complexity; ECF: ecological footprint; ENS: energy security; EVT: environmental control technology; RWN: renewable energy.

Table 5.

Panel cointegration tests.

Westerlund			Pedroni
Parameter	Value	P-value	Parameter	Statistic	P-value
Model 1: lnecf = (lnens lnrwn lneci lnevt lnafl lnpop)
Gt	−7.014***	0.000	Modified Phillips–Perron t	1.89***	0.028
Ga	−4.44	0.999	Phillips–Perron t	−3.52***	0.000
Pt	−75.23***	0.000	Augmented Dickey–Fuller t	−6.51***	0.000
Pa	−27.65	0.000
Model 2: lnco₂d = (lnens lnrwn lneci lnevt lnafl lnpop)
Gt	−2.06	0.828	Modified Phillips–Perron t	2.68***	0.003
Ga	−0.90	1.00	Phillips–Perron t	0.37	0.353
Pt	−25.63*	0.096	Augmented Dickey–Fuller t	0.33	0.361
Pa	−12.72*	0.086
Model 3: lncco₂ = (lnens lnrwn lneci lnevt lnafl lnpop)
Gt	−5.75*	0.060	Modified Phillips–Perron t	1.82**	0.034
Ga	−0.79	1.000	Phillips–Perron t	−1.83**	0.032
Pt	−0.73	1.000	Augmented Dickey–Fuller t	−2.32**	0.010
Pa	−0.37	0.998

Note: ***, ** and * imply the rejection of the null hypothesis at the 1%, 5% and 10% significance levels, respectively.

Source: Authors’ computations.

Table 5 illustrates the estimates of the two cointegration procedures adopted in this study for all the models. Consequently, the results underscore that the enlisted panel series in all the models representing EQ have a long-run coevolution. Therefore, this indicates the rejection of the null hypothesis of no cointegration. However, it is imperative to highlight that except for model 1 (lnecf = (lnens lnrwn lneci lnevt lnafl lnpop), there is evidence of weak cointegration among the panel series judging from the Westerlund cointegration procedure. This outcome further highlights the need for the application of the non-linear (PQREG) model, which will provide robust outcomes with far-reaching implications for environmental performance in E7 countries. Furthermore, in Table 6, we provide the results of the PMG-ARDL and CS-ARDL procedures that account for both the long- and short-term impacts of individual explanatory variables on the response variable.

Table 6.

PMG-ARDL and CS-ARDL estimates.

Variables	PMG-ARDL			CS-ARDL
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Long-run effects
	Coefficient	Coefficient	Coefficient	Coefficient	Coefficient	Coefficient
Lnens	−0.091**	−0.044	−0.160	−0.104	1.118	−0.871
Lnrwn	−0.734***	−0.085***	−0.143	−2.490	0.653	−0.076**
Lneci	0.377	−0.244**	−0.078	0.208	0.369	0.039
Lnevt	−0.115	−0.039	0.010	0.692	−0.014	0.023
Lnafl	0.475***	0.092*	0.075	−0.003	0.136	0.090**
Lnpop	−0.622	3.845***	0.114	16.560	3.627**	2.039**
Short-run effects
$Δ$ lnens	−34.66	13.108	−0.672	−0.180	−34.099	−0.086
$Δ$ lnrwn	0.255	−0.726***	−0.147***	−1.110*	−0.454**	−0.147***
$Δ$ lneci	0.048	0.011	0.009	0.282	0.024	−0.010
$Δ$ lnevt	0.043	−0.013***	0.005	0.916	−0.034*	0.040
$Δ$ lnafl	0.394***	0.039	−0.010	0.158	−0.006	0.118**
$Δ$ lnpop	−7.022	3.900	−1.628	3.127	4.348	−0.983
Ect	−0.348***	0.114***	−0.126**	0.263***	−0.067***	−0.422**

Note: ***, ** and * depict rejection of the null hypothesis of no significant relationship at the 1%, 5% and 10% significance levels, respectively.

Source: Authors’ computations.

CS-ARDL: cross-sectional augmented autoregressive distributed lag; PMG-ARDL: pool mean group autoregressive distributed lag.

The analysed results emanating from the PMG-ARDL and CS-ARDL highlight some interesting features that are critical for devising robust environmental sustainability pathways in E7 countries. Accordingly, each of the three selected EQ indicators (ECF, CO₂D and CCO₂) are represented as separate models (1–3) in that other. On account of the PMG-ARDL, it is established that ENS provides significant long-term impediments to carbon emissions in the E7 countries, mostly on ECF. Specifically, a one-per cent improvement in ENS results in an approximately 9% significant decline in the ECF. However, when the other two EQ indicators (CO₂D and CCO₂) are considered, ENS provides insignificant declining effects on them. This finding buttresses that its effects are sensitive to variants of EQ indicators, which also marks the significant contribution of this study. This outcome aligns with the reports of Bilgili et al.²⁶ and Zhao et al.,²¹ which buttress that ENS enhances EQ in selected Asian countries and China, respectively. These two studies failed to consider other variants of pollution indicators. RWN consumption enhances EQ in E7 countries irrespective of the variant of the EQ indicator considered. However, it is pertinent to highlight that RWN consumption failed to produce significant reducing effects on CCO₂. Accordingly, this outcome invalidates the report of Akadiri et al.;¹² however, it corroborates the reports of several other studies.^17,18

Furthermore, ECI and EVT also provide varying effects when different pollution indicators are considered. For instance, ECI provides a significant carbon neutrality pathway when CO₂D is considered, while it provides an insignificant reducing effect on CCO₂. However, the empirical analysis indicates that ECI impedes ecological balance when the ECF is considered. EVT also provides insignificant carbon neutrality pathways in E7 countries when the ECF and CO₂D are considered, while it produces insignificant but long-term positive environmental pollution effects. This suggests the inadequacy and lack of penetration of EVT in mitigating environmental pollution. This poor outcome of EVT highlighted above validates some prior studies (Awosusi et al.¹⁵); however, it is at variant with the submissions of Nguyen et al.³⁵ and Xin et al.³⁶ in the context of G-20 and the US economies, respectively. The outcome also highlights the appealing attributes of considering different variants of EQ indicators that promise streamlined policy strategies leading to comprehensive carbon neutrality pathways in E7 countries. Likewise, affluence and population constitute significant EQ degrading effects in E7 countries with varying effects across the variants of pollution indicators. However, on average, their EQ degrading effects outweigh their insignificant appealing effects. Furthermore, the short-term effects are largely inconsistent and vary considerably across variants of pollution indicators. However, it is pertinent to highlight the consistent EQ enhancing attributes of RWN that cut across all the selected pollution indicators, except the ECF. Overall, there is evidence of long-run steady-state convergences to equilibria given that all the error-correction terms are negative and significant.

The outcomes of the CS-ARDL for each model, which serves as a robustness check on the PMG-ARDL estimates, are largely consistent with the estimates of the PMG-ARDL, but there are minor dissimilarities that are not strong enough to dispute the outcomes of the PMG-ARDL. The observed consistencies in the estimates of the two models buttress the robustness of the study and consequential robustness of policy prescriptions thereof. Regardless of the outcomes of these estimates, the study elected to probe further principally to ensure that the influence of structural breaks is adequately captured. In this end, the PQREG was adopted accordingly given its appealing attributes, as highlighted earlier.

The estimates of the non-linear model (PQREG) that accounts for the effects of structural breaks and non-linear interactions among the explanatory and response variables are shown in Table 7. The estimate is based on the effects of the explanatory variables of each EQ indicator over the 25th, 50th, 75th and 95th quantiles. The PQREG analysis depicts several interesting outcomes highlighting its efficiency in stimulating robust empirical imperatives relevant for cutting-edge policy prescriptions. Accordingly, ENS possesses environmental enhancing attributes across the quantiles except at the 75th and 95th quantiles, where it escalates ecological deficits significantly. In essence, caution must be applied not to allow the ENS nosedive to thresholds where it cannot guarantee a cleaner environment. Therefore, policymakers must ensure that ENS is sustained or surpasses the 50th quantile, where it provides optimal environmental enhancing effects.

Table 7.

Panel quantile regression estimates.

Variable/Quan.	Model 1		Model 2		Model 3
	Coefficient	P-value	Coefficient	P-value	Coefficient	P-value
25th quantile
Lnens	−0.002	0.562	−0.032	0.436	−0.016***	0.000
Lnrwn	−0.151***	0.000	−0.795***	0.000	−0.128***	0.000
Lneci	−0.316***	0.000	−0.220***	0.000	0.019***	0.000
Lnevt	0.199***	0.000	0.014	0.325	−0.002***	0.000
Lnafl	0.299***	0.000	0.529***	0.000	−0.048***	0.000
Lnpop	0.078***	0.000	0.584***	0.000	0.087***	0.000
50th quantile
Lnens	0.015***	0.000	−0.020	0.279	−0.014***	0.000
Lnrwn	−0.176***	0.000	−0.758***	0.000	−0.134***	0.000
Lneci	−0.238***	0.000	−0.207***	0.000	0.013**	0.034
Lnevt	0.251***	0.000	0.021***	0.001	−0.002**	0.043
Lnafl	0.304***	0.000	0.541***	0.000	0.048***	0.000
Lnpop	0.047	0.104	0.386***	0.000	0.074***	0.000
75th quantile
Lnens	0.014***	0.000	−0.032***	0.000	−0.014***	0.000
Lnrwn	−0.219***	0.000	−0.764***	0.000	−0.139***	0.000
Lneci	−0.365***	0.000	−0.175***	0.000	0.017***	0.000
Lnevt	0.242***	0.000	0.021***	0.000	−0.003***	0.000
Lnafl	0.338***	0.000	0.537***	0.000	0.047***	0.000
Lnpop	0.059***	0.001	0.378***	0.000	0.069***	0.000
95th quantile
Lnens	0.016***	0.000	−0.029***	0.000	−0.013***	0.000
Lnrwn	−0.158***	0.000	−0.709***	0.000	−0.144***	0.000
Lneci	−0.396***	0.000	−0.181***	0.000	0.018***	0.000
Lnevt	0.258***	0.000	0.036***	0.000	−0.003***	0.000
Lnafl	0.383***	0.000	0.548***	0.000	−0.047***	0.000
Lnpop	0.089***	0.000	0.371***	0.000	0.069***	0.000

Note: ***, ** and * depict rejection of the null hypothesis of no significant relationship at the 1%, 5% and 10% significance levels, respectively.

Source: Authors’ computations.

Similar to the outcomes of the PMG-ARDL and CS-ARDL, the PQREG estimates reinforce the environmental enhancing attributes of RWN across all the quantiles considered. Thus, policies that could ensure the continuous development, deployment and consumption of RWN in these countries are critical for sustainable environments. Furthermore, ECI also possesses notable significant EQ enhancing attributes at all quantiles of ECF and CO₂D; however, it contributes positively and significantly to CCO₂. EVT provides varying positive and negative effects on EQ across the quantiles. Specifically, it deescalates CCO₂ significantly at the 25th, 50th and 95th quantiles and CO₂D at the 75th quantile. However, it significantly impedes EQ across the quantiles, and all pollution indicators are listed herein. This largely underscores the divergent submissions of prior studies in terms of the varying effects of EVT on EQ. This further reinforces the need to consider different pollution indicators for streamlined policy strategies. Thus, varied policy options are essential if E7 countries must reap the full potential of EVT at all times. Moreover, if the variable must deliver its full potential, it is pertinent to remark that more investments are needed to develop and deploy more ECTs in E7 countries. Economic growth and population maintained their negative effects on the environment across all quantiles except for a significant reducing influence on CCO₂ at the 25th and 95th quantiles. This demonstrates that E7 countries have continuously traded sustainable environments for rising affluence and population. Thus, to attain the relevant SDGs 7, 11, 12 and 13 and in compliance with the relevant treaties, E7 countries must vigorously pursue and outlined policies that ensure a balance between affluence, population and sustainable environment.

Conclusion and policy implications

A clean environment remains a top priority of all modern economies and therefore cannot be overemphasized. Meanwhile, the E7 countries must not be left behind. Several critical factors have been considered determinants of EQ in several economies, but the issue is still unresolved. Furthermore, very few efforts have been extended to unravelling the relevant enablers of EQ in the context of E7 countries. Furthermore, variables such as ENS, RWN and EVT are largely understudied, mostly as they affect EQ in E7 countries. Against this backdrop, this study took several holistic steps to provide fresh insights in this direction. First, the study enlisted the above explanatory variables as well as economic growth and population to provide a modified STIRPAT framework for E7 countries on the instances of three EQ indicators that were not considered in prior studies. Second, the study provides robust empirical narratives by analysing annual panel series spanning 1990–2018 with three enhanced panel estimators, including the PMG-ARDL, CS-ARDL and PQREG.

In the above premise, several empirical discoveries emerged. First, the estimates from all the models rectified that ENS possesses varying environmental enhancing attributes peculiar to different EQ indicators. However, its most significant soothing effects occurred within the 50th quantile based on the outcome of the PQREG. Therefore, it is instructive to ensure more commitment to ensuring ENS in E7 countries. Second, RWN consumption provides a robust decarbonization pathway for E7 countries, thereby necessitating the compelling need for more investments in that direction. Third, ECI possesses some varying environmental enhancing attributes peculiar to the EQ indicator considered, while it remains inimical in some other indicators. Therefore, pleasant environmental outcomes await the ecosystem in E7 if more efforts are channeled to ensuring a more transformed economic outlook through the adoption of environmentally friendly production techniques. Fourth, the E7 countries have yet to reap the full benefits of EVT given its unimpressive effects on EQ. It is therefore imperative to ensure that the depth of EVT in E7 countries improves such that its full benefits are appropriated. Last, economic growth and population density maintained all-time negative influences on EQ in E7 countries.

Given the above discoveries and peculiarities of these emerging countries, policymakers are therefore encouraged to provide workable strategies that will ensure that the increasing affluence and teeming population of E7 countries respond to the environmental sustainability drive at all times. Therefore, investing a greater proportion of such income in green technologies and building human capital could salvage the ugly situation. In summary, the outcomes of this empirical study remain robust within the context of the E7 countries and the period investigated. However, more robust estimates could emerge when other economies are considered or when the data scope improves. Therefore, future research in this direction is encouraged.⁶⁹

Footnotes

Author’s contribution

All the authors contributed significantly in varying proportions in the various sections of the study. Likewise, all the authors read and approved the final draft of the manuscript.

Declaration of conflicting interests

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

Funding

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

Data availability statement

The datasets generated during and/or analysed during the current study are available from the corresponding author upon reasonable request.

ORCID iD

Emmanuel Uche

Notes

Emmanuel Uche (The Corresponding Author) is presently a Lecturer II and Researcher with the Abia State University, Uturu, Abia State, Nigeria. Dr. Uche has special interests in Development Economics, Environmental Economics, Econometrics Modelling and Macroeconomics. He has authored and co-authored several research articles in high-impact factor journals indexed in both Scopus and Web of Science.

Narasingha Das is Research Associate with the Economists for Peace and Security, Australia Chapter. Dr Das has special interests in Energy Economics, Econometrics Modelling and Environmental Economics. Das has also authored and co-authored several research articles in high-impact factor journals indexed in both Scopus and Web of Science.

Nnamdi Chinwendu Nwaeze is presently a Senior Lecturer in the Department of Economics Uturu, Abia State, Nigeria. Dr. Nwaeze has special interests on Development Economics and Macroeconomics. He has authored and co-authored several research articles in high-impact factor journals indexed in both Scopus and Web of Science.

Pinki Bera is currently with a faulty in the Department of Economics, Vidyasagar University, India. She has special interest in Energy Economics, Development Economics and Macroeconomics. She has featured in peer reviewed research articles published in Scopus and Web of Science indexed journals.

References

Jena

Mujtaba

Joshi

DPP

, et al. Exploring the nature of EKC hypothesis in Asia’s top emitters: Role of human capital, renewable and non-renewable energy consumption. Environ Sci Pollut Res 2022: 0123456789. https://doi.org/10.1007/s11356-022-21551-w

Huo

Hameed

Sharif

, et al. Recent scenario and nexus of globalization to CO2 emissions: Evidence from wavelet and quantile on quantile regression approach. Environ Res 2022; 212: 113067.

Opuala

Omoke

Uche

. Sustainable environment in West Africa: The roles of financial development, energy consumption, trade openness, urbanization and natural resource depletion. Int J Environ Sci Technol 2022: 0123456789 . https://doi.org/10.1007/s13762-022-04019-9

Obiakor

Uche

Das

. Technology in society is structural innovativeness a panacea for healthier environmentsEvidence from developing countries. Technol Soc 2022; 70: 102033.

Uche

Effiom

. Financial development and environmental sustainability in Nigeria: Fresh insights from multiple threshold nonlinear ARDL model. Environ Sci Pollut Res 2021; 28: 39524–39539.

Wang

. The asymmetric impact of renewable and non-renewable energy on total factor carbon productivity in 114 countries: Do urbanization and income inequality matter? Energy Strategy Rev 2022a; 44: 100942.

Wang

Zhang

, et al. The impact of renewable energy on decoupling economic growth from ecological footprint – an empirical analysis of 166 countries. J Cleaner Prod 2022b; 354: 131706.

United Nations Framework Convention on Climate Change (UNFCCC). Conference of the parties twenty-first session: Adoption of the Paris agreement. 2015. https://unfccc.int/resource/docs/2015/cop21/eng/l09r01.pdf (retrieved 19 September 2021)

Zwickl-Bernhard

Huppmann

Golab

, et al. Disclosing the heat density of district heating in Austria in 2050 under the remaining European CO2 budget of the 1.5 °C climate target. Sustainable Energy, Grids Networks 2022; 31: 100775.

10.

Tong

Ortiz

, et al.

Economic growth, energy consumption, and carbon dioxide emissions in the E7 countries: A bootstrap ARDL bound test

Energy Sustainability Soc 2020; 10: 1–17.

11.

ul Husnain

Syed

Bashir

, et al. Do geopolitical risk and energy consumption contribute to environmental degradation? Evidence from E7 countries. Environ Sci Pollut Res 2022; 29: 41640–41652.

12.

Akadiri

Adebayo

Asuzu

, et al. Testing the role of economic complexity on the ecological footprint in China: A nonparametric causality-in-quantiles approach. Energy Environ 2022: 0958305X2210945. https://doi.org/10.1177/0958305×221094573

13.

Wang

. The impact of energy efficiency on carbon emissions: Evidence from the transportation sector in Chinese 30 provinces. Sustainable Cities Soc 2022; 82: 103880.

14.

Mujtaba

Jena

Mishra

, et al. Do economic growth, energy consumption and population damage the environmental quality? Evidence from five regions using the nonlinear ARDL approach. Environ Challenges 2022; 8: 100554.

15.

Awosusi

Adebayo

Kirikkaleli

, et al. Role of technological innovation and globalization in BRICS economies: Policy towards environmental sustainability. Int J Sustainable Dev World Ecol 2022; 00: 1–18.

16.

Hamid

Alam

Kanwal

, et al. Decarbonization pathways: The roles of foreign direct investments, governance, democracy, economic growth, and renewable energy transition. Environ Sci Pollut Res 2022; 29: 49816–49831.

17.

Akalin

Erdogan

Asumadu

. Do dependence on fossil fuels and corruption spur ecological footprint? Environ Impact Assess Rev 2021; 90: 106641.

18.

Chien

Hsu

Sibghatullah

, et al. The role of technological innovation and cleaner energy towards the environment in ASEAN countries: Proposing a policy for sustainable development goals. Econ Res 2021a; 0: 1–16.

19.

Ali

Rahman

Zahid

, et al. Do technology and structural changes favour environment in Malaysia: An ARDL-based evidence for environmental kuznets curve. Environ, Dev Sustainability 2020; 22: 7927–7950.

20.

Zeraibi

Balsalobre-Lorente

Murshed

. The influences of renewable electricity generation, technological innovation, financial development, and economic growth on ecological footprints in ASEAN-5 countries. Environ Sci Pollut Res 2021; 28: 51003–51021.

21.

Zhao

Sinha

Inuwa

, et al. Does structural transformation in economy impact inequality in renewable energy productivity? Implications for sustainable development. Renewable Energy 2022; 189: 853–864.

22.

Jessel

Sawyer

Hernández

. Energy, poverty, and health in climate change: A comprehensive review of an emerging literature. Front Public Health 2019; 7. https://doi.org/10.3389/fpubh.2019.00357

23.

Grossman

Krueger

. Economic growth and the environment author (s): Gene M. Grossman and Alan B. Krueger reviewed work (s): published by: Oxford university press. Q J Econ 1995; 110: 353–377.

24.

Jin

Razzaq

Saleem

, et al. Asymmetric effects of eco-innovation and human capital development in realizing environmental sustainability in China: Evidence from quantile ARDL framework. Econ Res 2021; 0: 1–24.

25.

Shobande

Ogbeifun

. The criticality of financial development and energy consumption for environmental sustainability in OECD countries: Evidence from dynamic panel analysis. Int J Sustainable Dev World Ecol 2021; 00: 1–11.

26.

Bilgili

Ozturk

Kocak

, et al. The nexus between access to electricity and CO 2 damage in Asian countries: The evidence from quantile regression models. Energy Build 2021; 256: 11761.

27.

Gyamfi

Adedoyin

Bein

, et al. Environmental implications of N-shaped environmental kuznets curve for E7 countries. Environ Sci Pollut Res 2021; 28: 33072–33082.

28.

Gyamfi

Bein

Udemba

, et al. Renewable energy, economic globalization and foreign direct investment linkage for sustainable development in the E7 economies: Revisiting the pollution haven hypothesis. Int Soc Sci J 2022; 72: 91–110.

29.

Huang

Haseeb

Usman

, et al.

Dynamic association between ICT, renewable energy, economic complexity and ecological footprint: Is there any difference between E-7 (developing) and G-7 (developed) countries?

Technol Soc 2022; 68: 101853.

30.

Usman

Ozturk

Naqvi

SMMA

, et al. Revealing the nexus between nuclear energy and ecological footprint in STIRPAT model of advanced economies: Fresh evidence from novel CS-ARDL model. Prog Nucl Energy 2022; 148: 104220.

31.

Usman

Hammar

. Dynamic relationship between technological innovations, financial development, renewable energy, and ecological footprint: Fresh insights based on the STIRPAT model for Asia pacific economic cooperation countries. Environ Sci Pollut Res 2021; 28: 15519–15536.

32.

Abbasi

Hussain

Haddad

, et al. The role of financial development and technological innovation towards sustainable development in Pakistan: Fresh insights from consumption and territory-based emissions. Technol Forecast Soc Change 2022; 176: 121444.

33.

Razzaq

Sharif

Ahmad

, et al. Asymmetric role of tourism development and technology innovation on carbon dioxide emission reduction in the Chinese economy: Fresh insights from QARDL approach. Sustainable Development 2021; 29: 176–193.

34.

Yang

and Jahanger

. Remittance inflows affect the ecological footprint in BICS countries: do technological innovation and financial development matter?. Environ Sci Pollut Res 2021; 28: 23482–23500.

35.

Nguyen

Pham

TAT

Tram

HTX

. Role of information and communication technologies and innovation in driving carbon emissions and economic growth in selected G-20 countries⋆. J Environ Manag 2020; 261: 110162.

36.

Xin

Ahmad

Lei

, et al.

Do innovation in environmental-related technologies asymmetrically affect carbon dioxide emissions in the United States?

Technol Soc 2021; 67: 101761.

37.

Umar

Kirikkaleli

, et al.

COP21 Roadmap: Do innovation, financial development, and transportation infrastructure matter for environmental sustainability in China?

J Environ Manag 2020; 271: 111026.

38.

Ahmad

Khattak

Khan

, et al. Innovation, foreign direct investment (FDI), and the energy–pollution–growth nexus in OECD region: A simultaneous equation modelling approach. Environ Ecol Stat 2020; 27: 203–232.

39.

Ding

Khattak

Ahmad

. Towards sustainable production and consumption: Assessing the impact of energy productivity and eco-innovation on consumption-based carbon dioxide emissions (CCO2) in G-7 nations. Sustainable Prod Consumption 2021; 27: 254–268.

40.

Amin

Song

Khan

. Dynamic linkages of financial inclusion, modernization, and environmental sustainability in south Asia: A panel data analysis. Environ Sci Pollut Res 2022; 29: 16588–16596.

41.

Chien

Sadiq

Nawaz

, et al. A step toward reducing air pollution in top Asian economies: The role of green energy, eco- innovation, and environmental taxes. J Environ Manag 2021b; 297: 113420.

42.

Dingru

Ramzan

Irfan

, et al.

The role of renewable energy consumption towards carbon neutrality in BRICS nations: Does globalization matter?

Front Environ Sci 2021; 9: 1–12.

43.

Gormus

Aydin

. Revisiting the environmental Kuznets curve hypothesis using innovation: New evidence from the top 10 innovative economies. Environ Sci Pollut Res 2020; 27: 27904–27913.

44.

Hsu

Zhang

, et al. A step towards sustainable environment in China: The role of eco-innovation renewable energy and environmental taxes. J Environ Manag 2021; 299: 113609.

45.

Chang

Saydaliev

Meo

, et al. How renewable energy matter for environmental sustainability: Evidence from top-10 wind energy consumer countries of European Union. Sustainable Energy, Grids Networks 2022; 31: 100716.

46.

Dogan

Shah

. Analyzing the role of renewable energy and energy intensity in the ecological footprint of the United Arab Emirates. Sustainability 2022; 14: 1–12.

47.

Atsu

Adams

Adjei

. ICT, energy consumption, financial development, and environmental degradation in South Africa. Heliyon 2021; 7: e07328.

48.

Mujtaba

Jena

Bekun

, et al. Symmetric and asymmetric impact of economic growth, capital formation, renewable and non-renewable energy consumption on environment in OECD countries. Renewable Sustainable Energy Rev 2022a; 160: 112300.

49.

Wang

Dong

. What drives environmental degradation? Evidence from 14 Sub-Saharan African countries. Sci Total Environ 2019; 656: 165–173.

50.

Iqbal

Abbasi

Shinwari

, et al.

Does exports diversification and environmental innovation achieve carbon neutrality target of OECD economies?

J Environ Manag 2021; 291: 112648.

51.

Chen

Lee

. Does technological innovation reduce CO2 emissions? Cross- country evidence. J Cleaner Prod 2020; 263: 121550.

52.

Destek

Sarkodie

Asamoah

. Does biomass energy drive environmental sustainability? An SDG perspective for top five biomass consuming countries. Biomass Bioenergy 2021; 149: 106076.

53.

Murshed

Saboori

Madaleno

, et al. Exploring the nexuses between nuclear energy, renewable energy, and carbon dioxide emissions: The role of economic complexity in the G7 countries. Renewable Energy 2022; 190: 664–674.

54.

Emre

Wasif

Victor

, et al. Determinants of CO2 emissions in the BRICS economies: The role of partnerships investment in energy and economic complexity. Sustainable Energy Technol Assess 2022; 51: 101907.

55.

Ramzan

Awosusi

, et al. Does globalization moderate the effect of economic complexity on CO2 emissions? Evidence from the top 10 energy transition economies. Front Environ Sci 2021; 9: 1–11.

56.

Uche

Das

Bera

. Re-examining the environmental Kuznets curve (EKC) for India via the multiple threshold NARDL procedure. Environ Sci Pollut Res 2022: 0123456789. https://doi.org/10.1007/s11356-022-22912-1

57.

Pesaran

. General diagnostic tests for cross section dependence in panels. 2004, CESifo Working Paper Series No. 1229.

58.

Pesaran

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

59.

Hashem Pesaran

Yamagata

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

60.

Pesaran

. A simple panel unit root test in the presence of cross-section dependence. J Appl Econom 2007; 22(2): 265–312.

61.

Karavias

Tzavalis

. Local power of fixed-T panel unit root tests with serially correlated errors and incidental trends. J Time Ser Anal 2016; 37: 222–239.

62.

Pedroni

. Critical values for cointegration tests in heterogeneous panels with multiple of economics and statistics. Oxf Bull Econ Stat 1999; 61: 653–670.

63.

Westerlund

. Testing for error correction in panel data. Oxf Bull Econ Stat 2007; 69: 709–748.

64.

Erdoğan

Yıldırım

DÇ

, et al. The effects of innovation on sectoral carbon emissions: Evidence from G20 countries. J Environ Manag 2020; 267: 1–10.

65.

Pesaran

Shin

, et al. Pooled mean group estimation of dynamic heterogeneous panels. J Am Stat Assoc 1999; 94: 621–634.

66.

Chudik

Pesaran

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

67.

Effiom

Uche

. The political economy of industrialization in Sub-Saharan Africa: Evaluating the role of institutions. Inst Econ 2022; 14: 77–108.

68.

Ponomareva

. Identification in Quantile Regression Panel Data Models with Fixed Effects and Small T. 1–28. 2011.

69.

Wang

Liu

, et al. Per-capita carbon emissions in 147 countries: The effect of economic, energy, social, and trade structural changes. Sustainable Prod Consumption 2021; 27: 1149–1164.