Decarbonization Pathways in Selected MENA Countries: Panel Evidence on Transport Services, Renewable Energy, and the EKC Hypothesis

Abstract

This study investigates the relationship between economic growth and environmental performance in selected Middle East and North Africa (MENA) countries through the lens of the Environmental Kuznets Curve (EKC) hypothesis. Due to data availability constraints, our sample includes Algeria, Egypt, Lebanon, Mauritius, Morocco, and Oman, covering the period 1990–2022. Using annual panel data, we apply panel cointegration techniques alongside Fully Modified Ordinary Least Squares (FMOLS) and Dynamic Ordinary Least Squares (DOLS) estimators, complemented by Granger causality tests, to examine the interaction among GDP per capita, renewable energy consumption, and transport service exports in determining CO₂ emissions per unit of GDP. The empirical findings provide only partial support for the EKC: while the DOLS results confirm an inverted U-shaped income–emissions relationship, the FMOLS estimations contradict it, suggesting a more complex and nonlinear pattern. Beyond testing the EKC, this study contributes two novel dimensions to the literature. First, it shows that renewable energy exerts a statistically significant negative effect on carbon intensity in the long run, despite weak short-run causality, highlighting the delayed but durable environmental benefits of clean energy adoption. Second, it introduces transport service exports as a proxy for structural economic transformation, capturing the role of trade-driven diversification in reducing emissions. By embedding renewable energy deployment and service-based trade dynamics into the EKC framework, the study advances a more policy-relevant and region-specific understanding of the growth–environment nexus in the selected MENA economies. The results underscore the importance of scaling renewable energy, promoting low-carbon service sectors, and aligning trade and environmental policies to ensure that economic growth supports long-term climate objectives.

1. Introduction

The Middle East and North Africa (MENA) region is experiencing significant economic transformation along with intensifying environmental challenges. As regional economies strive to reconcile growth objectives with the imperatives of energy transition and climate change mitigation, monitoring key indicators becomes essential for assessing progress and guiding policy. Metrics such as carbon emissions, gross domestic product (GDP), renewable energy penetration, and the share of transport services in commercial service exports offer valuable insights into economic structure, resource utilization, and development priorities. These indicators not only reflect national trajectories and endowments but also reveal persistent structural barriers, including regulatory inefficiencies and institutional fragmentation, that impede effective policy implementation. While some MENA countries have taken notable steps toward expanding renewable energy capacity and diversifying their economies, the region remains heavily reliant on fossil fuel exports and is disproportionately vulnerable to the impacts of climate change, including water scarcity, extreme heat, and socioeconomic instability [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35].

Carbon dioxide emissions in MENA have risen substantially over the past decades, largely due to fossil fuel dependency, population growth, and urbanization. According to the International Energy Agency [16,17,18,28], MENA accounts for nearly 7% of global CO₂ emissions, despite representing a smaller share of the global population. Countries such as Saudi Arabia, the United Arab Emirates, and Iran have some of the highest per capita emissions in the world. While several nations have announced climate action plans, such as Nationally Determined Contributions (NDCs), progress remains uneven. A recent regional review by Al-Zayer and Al-Barjas [2] highlights that many MENA countries lack robust emissions monitoring and enforcement mechanisms, which hampers long-term mitigation efforts [26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48].

Gross domestic product (GDP) in the MENA region reflects the dependence of several economies on hydrocarbon exports. Oil-rich states, particularly those in the Gulf Cooperation Council (GCC), have experienced sustained growth but remain highly vulnerable to fluctuations in global energy markets [36,37]. In response, many are pursuing ambitious diversification strategies, including Saudi Arabia’s Vision 2030, the UAE’s Energy Strategy 2050, and Egypt’s Sustainable Development Strategy: Vision 2030 [38,39,40]. At the same time, non-oil-exporting countries face structural challenges such as water scarcity, youth unemployment, and fiscal constraints, which reduce their resilience to external shocks and limit their policy space for green transition [37,38,39,40,41,42].

Renewable energy has emerged as a key policy focus in the MENA region, driven by the dual goals of reducing carbon emissions and conserving fossil fuel resources for export. Solar and wind projects are growing in scale, particularly in Morocco, Egypt, Jordan, and the UAE. Morocco, for instance, aims to generate over 50% of its electricity from renewables by 2030 and has already commissioned large-scale solar plants such as Noor Ouarzazate. However, a review by Razi and Inser [49] notes that institutional inertia, weak grid integration, and inconsistent regulatory frameworks continue to constrain the full deployment of clean energy technologies across the region [19,20,21,22,23,24,27,28,29,30,31,32,38,39,40,50].

Transport services, measured as a percentage of commercial service exports, serve as a proxy for assessing the structural composition of the service economy. This indicator reflects the extent to which a country’s commercial service exports are reliant on transport-related activities, thereby illustrating the role of the transport sector within the broader services trade portfolio. In the context of the Middle East and North Africa (MENA) region, this variable provides valuable insight into the degree of economic diversification, particularly in economies traditionally dependent on resource-based exports. A higher or growing share of transport services may suggest progress toward more balanced and service-oriented economic structures, while a lower share could indicate continued concentration in a narrow range of export activities. More specifically, countries like Egypt (through the Suez Canal), the UAE (via Dubai’s aviation and logistics hubs), and Morocco (with its growing shipping sector) are increasingly leveraging their geographic advantages to expand their service economies. Transport and logistics not only contribute to GDP growth but also play a strategic role in employment and international connectivity. Yet, this sector also poses environmental challenges, including emissions from maritime and air traffic, as discussed by Omar et al. [9] in a comparative analysis of decarbonization pathways for transport in MENA countries [1,2,3,4,7,8,9,10,11,12,14,17,18,19,20,22,23,24,25,26,27,28,29,30,31,32,33,34,39,40,41,42,50].

Collectively, these indicators—carbon emissions, GDP, renewables, and transport services—offer a comprehensive framework for evaluating sustainability and economic performance in the MENA region. They highlight both shared regional trends and country-specific trajectories, emphasizing the need for tailored policy responses that align with national capabilities, resource constraints, and long-term development goals.

Within this framework, the present work makes an effort to identify the interlinkages among key macroeconomic and structural variables—namely, GDP per capita, trade openness (exports), renewable energy share, and CO₂ emissions per unit of GDP—with the overarching aim of assessing their combined and individual impacts on environmental degradation. By employing a panel data econometric approach supported by visual analytics, this study contributes to a deeper understanding of how economic growth and structural transformation influence environmental performance across selected MENA countries over time.

Due to data availability constraints, our sample includes Algeria, Egypt, Lebanon, Mauritius, Morocco, and Oman. While geographically diverse, these countries share several structural characteristics prevalent across the MENA region—such as resource dependency, vulnerability to climate change, and ongoing transitions in energy, agriculture, and environmental policy—rendering them broadly representative of key MENA dynamics [41,42]. The inclusion of Mauritius, though not geographically part of MENA, is justified by its participation in Afro-Arab and Arab League cooperation frameworks, as well as similar socioeconomic challenges, particularly in relation to sustainable development, food security, and energy transitions [51,52]. These countries thus provide a suitable lens for understanding broader trends in the region, especially where data limitations restrict full regional representation.

The selection of Algeria, Egypt, Morocco, Lebanon, Oman, and Mauritius as research subjects provides a representative and balanced framework for analyzing key environmental and economic dynamics across the MENA region. Algeria, Egypt, and Morocco typify the North African subregion, while Lebanon and Oman represent the Eastern Mediterranean and the Gulf, respectively. Although Mauritius is geographically outside the core MENA area, its inclusion is justified by its membership in the Arab League and its economic participation in Afro-Arab cooperation frameworks, as well as its exposure to similar environmental and policy challenges [51,52]. This selection encompasses both oil-exporting (Algeria, Oman) and oil-importing (Lebanon, Morocco) countries, providing insights into the varied economic structures and energy dependencies within the region [53]. Despite differing levels of development and state intervention, these countries share critical challenges typical of the MENA region: widespread water scarcity, food insecurity, high youth unemployment, import dependency, and increasing vulnerability to climate change [54,55]. Moreover, all are active participants in regional initiatives such as the Arab League, the Union for the Mediterranean, and collaborative platforms under the FAO and ESCWA, reinforcing their relevance in comparative policy studies and regional sustainability dialogues [51,56].

The novelty of this study lies in its integrated and multi-dimensional analytical framework, which combines visual trend diagnostics with formal panel econometric techniques to test the Environmental Kuznets Curve (EKC) hypothesis. Specifically, the research is guided by three hypotheses:

H1: 

Economic growth in selected MENA countries follows the EKC pattern, where rising GDP per capita initially increases carbon intensity; however, beyond a threshold income level, further growth reduces emissions.

H2: 

Renewable energy consumption contributes to long-run reductions in carbon intensity, even if its short-run effects are weak or insignificant.

H3: 

Transport service exports, as a proxy for structural economic transformation, reduce carbon intensity by shifting growth toward low-carbon service activities.

By incorporating renewable energy consumption and transport service exports into the EKC framework, the analysis captures critical but often underexplored dimensions of energy transition and trade-related diversification. This dual-method approach—merging graphical exploration with robust econometric modeling—moves beyond mere validation of existing theories by offering a refined understanding of the nonlinear dynamics that underpin the relationship between economic development and environmental performance across heterogeneous economies.

Unlike many earlier studies that either adopt single-country designs or treat economic and environmental indicators in isolation, this research employs a regionally comparative and temporally dynamic perspective. Using panel cointegration techniques, it evaluates long-run equilibrium relationships between income, emissions, and structural variables in selected MENA countries—Algeria, Egypt, Lebanon, Mauritius, Morocco, and Oman. These countries represent diverse energy profiles, trade exposure, and development pathways, making them an ideal sample for assessing regional dynamics. Renewable energy consumption is employed as a proxy for each country’s environmental sustainability trajectory, while GDP per capita (PPP, constant 2017 international dollars) serves as an indicator of economic development. In parallel, transport service exports are used as a measure of service-sector dynamism and economic diversification—factors increasingly central to low-carbon transition strategies.

The objective of the study is therefore twofold: first, to empirically test the EKC hypothesis in the context of selected MENA economies (H1), and second, to evaluate whether renewable energy (H2) and service-sector trade (H3) act as moderating forces that accelerate the transition from the upward to the downward slope of the curve. The insights generated aim to inform the design of effective, evidence-based policy interventions for climate-resilient, low-carbon, and inclusive development in the region.

2. Literature Review

Developing countries in the Middle East and North Africa (MENA) region are facing increasing pressure to decarbonize their transport systems while sustaining economic growth. Rapid urbanization, outdated infrastructure, and growing vehicle ownership have led to rising greenhouse gas emissions, particularly from the transport sector—a sector that remains heavily fossil fuel-dependent. Unlike many developed economies that have prioritized transport electrification and progressed toward this, most MENA countries continue to face significant constraints due to limited renewable energy capacity, institutional inertia, and insufficient investment in sustainable mobility infrastructure. In this context, understanding the relationship between economic development and environmental degradation is critical. One widely referenced framework that provides such insight is the Environmental Kuznets Curve (EKC) hypothesis [1,2,3,4,5,6,7,8,9,10,32,40,57,58].

The EKC posits that environmental degradation initially worsens as income rises but begins to decline after surpassing a certain income threshold, producing an inverted-U-shaped curve [43,50,57,58,59,60,61,62]. In the early stages of economic development, growth is often accompanied by industrialization, higher energy use, and pollution. However, beyond a turning point, further income growth is assumed to enable technological improvements, better regulation, and increased public demand for environmental quality, thus reducing environmental harm. Following the initial wave of research on the Environmental Kuznets Curve (EKC), numerous subsequent studies—ranging from single-country analyses [15,16,17,18,19] to cross-country investigations [20,21,22,23,24,25,26]—have produced mixed empirical results using a variety of econometric techniques. While several studies lend strong support to the EKC hypothesis [15,17,18,19,21], others provide only partial evidence [19,25], and some find no empirical validation of the EKC relationship at all [16]. These divergent findings highlight the context-dependent nature of the EKC and suggest that its validity may vary significantly based on country-specific characteristics, methodological choices, and the selection of environmental and economic indicators [63,64,65,66,67,68,69,70,71,72].

Ang [66], in a study of France, used time series analysis and found empirical support for the EKC, observing that emissions increased at lower income levels but declined with continued economic growth, reflecting a shift toward cleaner technologies and structural transformation. Similarly, Esteve and Tamarit [67], employed nonlinear cointegration techniques on long-run Spanish data and confirmed the EKC by identifying income thresholds beyond which environmental quality improved. In contrast, Wang et al. [34,73] found a U-shaped relationship in China, indicating that economic growth may not necessarily lead to environmental improvement, particularly in contexts with high industrial intensity and weak environmental governance [7,8,9,10,11,12,58,59,60,61,62].

Developing countries have also been a focus of EKC research. Shahbaz et al. [68] found strong support for the EKC in Pakistan using cointegration and Granger causality techniques. Developing countries have increasingly become the focus of Environmental Kuznets Curve (EKC) research due to their growing environmental challenges alongside their economic development. Shahbaz et al. [68] provided robust empirical support for the EKC hypothesis in Pakistan by employing cointegration and Granger causality techniques, revealing a nonlinear relationship between economic growth and carbon emissions. Similarly, Ozturk and Acaravci [74] confirmed the validity of the EKC in Turkey, emphasizing the critical roles of energy consumption and trade openness as long-term determinants of CO₂ emissions. Focusing on India, Tiwari et al. extended the EKC framework by incorporating coal consumption, finding evidence of long-run cointegration and supporting the EKC hypothesis in the Indian context. However, empirical findings remain mixed, and the EKC pattern is not universally observable. Aydin et al. [75], in a panel study of European countries, found that the EKC was only supported in a subset of nations—such as Denmark and Italy—highlighting the heterogeneity of environmental outcomes even within geographically and economically integrated regions. These divergent results underscore the importance of country-specific factors, such as energy policy, institutional quality, and the structure of economic activities, which can significantly influence the shape and trajectory of the environmental impact–income relationship. Consequently, while the EKC provides a useful heuristic for exploring the interplay between development and environmental degradation, its applicability must be assessed cautiously and contextually [48,58,63,64,65,66,67].

Contradictory findings are also prevalent. Al-Mulali et al. [76], in a study on Vietnam, found a persistent positive correlation between income and emissions in both the short and long run, suggesting that growth alone does not guarantee environmental improvement. Babba employed the GMM methodology for higher-, middle- and lower-income groups, and found an N-shaped relationship, implying that emissions might increase again at very high income levels, complicating the simple EKC narrative [58,69,70].

Panel-based studies provide broader regional insights. Azmin et al. [77] confirmed the EKC in ASEAN countries, while Pao and Tsai [78] validated it in the BRIC nations. In Africa, Fu et al. [79] found partial EKC support for air and water pollutants, with results varying by environmental indicator.

The MENA region has gradually emerged as a subject of EKC testing, though studies remain relatively scarce and show mixed outcomes [1,2,3,4,5,6,7,8,9,10,11,39,40,50,58,62,72]. Al-Rawashdeh et al. [28] confirmed the EKC for certain pollutants in MENA countries, suggesting that income growth can lead to reduced pollution, but only under specific economic conditions. Onafowora and Owoye [80] also tested the EKC across eight countries, including some in the MENA region, and found only partial support, reinforcing the need to consider national-level differences in industrial structure, policy enforcement, and energy dependence

Jebli and Yussef [81] focused on Tunisia and validated the EKC by revealing long-run relationships between emissions, output, trade openness, and energy consumption. Their study underscores the critical role of complementary policy reforms—particularly in trade and energy—to achieve both growth and sustainability. Jebli et al. [82] found evidence from Central and South America showing that renewable energy, tourism, and FDI contribute to long-run emission reductions, while trade and economic growth tend to increase emissions, underscoring the importance of green tourism and renewable energy in climate policy design. While not MENA-based, Lau et al.’s [64] work on Malaysia is methodologically relevant, as it integrates FDI and trade into the EKC framework. Their findings suggest that globalization dynamics—such as foreign investment—can either mitigate or intensify emissions, depending on the nature of industrial transfer and regulatory standards, aligning closely with the Pollution Haven Hypothesis.

Further enriching the debate, Osabuohien et al. [83] examined African countries and introduced governance and institutional quality into the EKC framework. Their results confirmed that strong institutions and effective policy environments are essential to achieving the downward slope in the EKC. This insight is highly relevant to the MENA context, where institutional fragmentation and regulatory gaps often delay or dilute environmental progress.

Collectively, these studies highlight that while the EKC hypothesis may hold under specific conditions, its applicability is neither universal nor automatic. Notably, in the MENA region, the shape and validity of the EKC depend on a host of interacting variables—including trade integration, FDI inflows, energy mix, infrastructure quality, and institutional strength. Consequently, testing the EKC in this context requires a nuanced, multi-variable approach rather than relying solely on income and emissions data.

In line with this understanding, the present study seeks to assess whether the EKC holds for MENA countries by incorporating additional variables that reflect the broader economic-environmental nexus. In particular, we integrate the share of renewable energy as an indicator of a country’s “greening” capacity, and include a variable representing transport infrastructure and services, given the sector’s central role in emissions. Alongside GDP and the square of GDP (to capture nonlinearity), these variables allow for a more comprehensive examination of how economic structure, energy transition, and mobility systems influence environmental outcomes in the region. This extended approach offers a more robust and context-specific test of the EKC hypothesis in MENA economies, which are at a critical crossroads between fossil fuel dependency and the imperative for sustainable development [65].

3. Research Area

Environmental performance in the MENA region is increasingly shaped by the nexus of energy use, transport development, and economic growth, yet the pace and scope of progress remain uneven across countries. Many MENA nations face the dual challenge of decarbonizing their energy and transport sectors while sustaining economic development and addressing population growth and urbanization [62,83,84,85]. Renewable energy adoption, though expanding in certain areas, reflects stark regional disparities. Morocco, for example, has emerged as a leader in clean energy and low-carbon transport systems, with over 42% of its electricity now generated from renewable sources, primarily wind and solar [14,15,16,17,18]. This progress supports large-scale initiatives such as the electrification of railways and the introduction of electric buses in major cities like Casablanca and Rabat, which are backed by fiscal incentives and green infrastructure funding. Tangier’s port, a strategic node in North African logistics, has been modernized to include rail electrification, further reducing emissions linked to freight transport [19,20,21,22,23,24,25,26,27].

Egypt presents another case of ambitious electrification efforts, particularly in Cairo, where the metro system significantly mitigates road congestion and emissions. The government’s partnership with UAE and Chinese developers to localize EV production, alongside solar-powered infrastructure along green corridors, reflects an integrated approach to energy and transport reform. However, outside the capital, the transition remains slow due to infrastructure and financing limitations [10,24,86,87,88,89]. Oman, meanwhile, has prioritized green hydrogen within its Vision 2040 framework. Through massive investments in solar farms and hydrogen fueling infrastructure, the country aims to become a regional export hub for green fuels. While Oman’s overall transport emissions are lower than those of Egypt or Morocco, its per capita emissions remain high due to dispersed settlement patterns, low public transport usage, and vehicle dependence [87,88,89].

In contrast, Algeria and Lebanon illustrate the structural barriers many MENA countries face. In Algeria, institutional inertia, subsidy lock-ins, and underinvestment hinder both energy diversification and transport modernization. Despite recent policy shifts to integrate solar power and EV charging infrastructure, progress remains limited. Lebanon’s chronic fiscal instability, electricity shortages, and outdated vehicle fleets have stalled meaningful decarbonization, though donor-supported projects and solar microgrids are emerging in urban centers like Beirut and Tripoli. The lack of a stable regulatory framework continues to deter large-scale investment [27,89,90].

Mauritius, while not part of continental MENA, offers a notable counterpoint as a small-island developing state with advanced plans for clean transport and energy. Its integration of solar-powered EV fleets and maritime mobility systems, alongside dynamic electricity tariffs, shows how targeted policies can yield tangible environmental gains. The Metro Express electric rail project has already reduced urban transport emissions and serves as a model for sustainable transit in small economies [91].

These country cases highlight a broader reality: environmental performance in the MENA region is strongly mediated by national policy frameworks, energy strategies, and the degree of infrastructure modernization. Countries with coherent plans, investment in renewables, and alignment between economic and environmental policy—such as Morocco, Egypt, and Oman—are advancing more rapidly toward decarbonization. In contrast, states hampered by political instability, institutional fragmentation, or economic hardship face greater difficulties aligning environmental and development goals [84,85]. The role of transport infrastructure is central to this dynamic, as it not only contributes significantly to emissions but also offers a key area where renewable energy integration and emissions reduction strategies can converge. As such, any attempt to assess or improve environmental outcomes in MENA must consider the interconnected impacts of energy generation capacity, public transit electrification, regulatory strength, and economic context on both emissions trajectories and sustainable development prospects [91].

The comparative analysis reveals a significant disparity in progress toward transport decarbonization across the MENA region. Countries such as Morocco and Mauritius have made notable advances in integrating renewable energy into their transport systems, benefiting from cohesive policy frameworks, targeted investments, and international partnerships. In contrast, Algeria and Lebanon continue to face substantial obstacles due to structural weaknesses, political instability, and limited fiscal capacity, resulting in slow or stalled progress. These divergences underscore the critical role of governance, institutional strength, and long-term planning in enabling energy transitions [86,87,88,89,90,91,92,93].

Data from the International Energy Agency (IEA) and the Organisation for Economic Co-operation and Development (OECD) reinforce the importance of aligning transport electrification with grid decarbonization, particularly in MENA economies where renewable electricity generation remains low. Electrifying transport without a clean energy backbone risks shifting emissions from tailpipes to power plants, undermining overall environmental performance. To address this, policy coherence, regulatory incentives, and regional cooperation are essential [16,17,18,19,20,21,22,23,24,25,26,27,28,29].

Morocco’s issuance of green bonds and Egypt’s use of public–private partnerships (PPPs) for infrastructure development offer replicable financing mechanisms that could be adapted by other countries in the region, including Oman and Algeria. These instruments not only attract private capital but also align financial flows with climate objectives. By embedding sustainability criteria in national investment strategies, governments can accelerate the transition to cleaner transport systems [19,20,21,22,23,24,25,26,27,28,29,78,79,80,81,82,83,84,85,86,87].

Table 1. Comparative emissions and renewable generation metrics (2023).

Country	Transport CO2 Emissions (% of total)	Per Capita Transport CO2 (tons)	Renewable Electricity Share (%)	EV Penetration Rate (%)
Morocco	23%	0.9	42%	2.1
Egypt	15.8%	1.4	22%	0.5
Oman	19%	2.7	11%	1.8
Algeria	21%	1.5	<5%	0.3
Lebanon	25%	1.3	<3%	<0.1
Mauritius	14.9%	1.1	28%	3.2

Source: data derived from [94].

provides a comparative overview of key indicators—including transport’s contribution to national CO₂ emissions (%), per capita transport-related carbon output (tons), the share of renewable electricity (%), and EV penetration rates (%)—across six selected MENA countries. This data supports and contextualizes the patterns discussed in the literature, highlighting both progress and persistent gaps in the region’s energy and transport transition.

To accelerate the transition, governments should (1) adopt national transport electrification plans linked to renewable energy targets; (2) phase out fossil fuel subsidies gradually; (3) invest in EV charging infrastructure; and (4) create environments enabling public–private partnerships. Each country’s context will determine their specific roadmap, but the overarching objective remains the same: reduce CO₂ emissions from transport while expanding renewable energy generation capacity [94].

4. Data Methodology

4.1. Data Description

To achieve the objectives of this study, we relied on data primarily derived from the World Bank, covering the period 1990–2019. A key limitation of the analysis was data availability, which restricted the sample to six countries: Algeria, Egypt, Lebanon, Mauritius, Morocco, and Oman. Within this framework, the study examines the interplay between economic growth (proxied by GDP per capita), the greening of economies (captured through the share of renewable energy consumption), and the contribution of transport services within the broader services trade portfolio, assessing their combined impact on environmental degradation as measured by carbon emissions relative to GDP.

In this study, carbon emissions are expressed relative to GDP, i.e., as carbon intensity, rather than in per capita terms. While per capita emissions are widely used to assess the equity and distributional aspects of environmental pressure, they do not directly capture the efficiency of economic production. Since the Environmental Kuznets Curve (EKC) hypothesis is concerned with how economic growth influences environmental performance, the appropriate measure is emissions per unit of output, which reflects whether growth is accompanied by rising or declining environmental costs. Carbon intensity therefore provides a more consistent indicator for EKC testing, as it allows the analysis to evaluate the extent of decoupling between GDP growth and emissions, in line with recent studies emphasizing the role of efficiency gains in the growth–environment nexus [59,63,82].

As a robust consideration, per capita emissions may still be employed in complementary analyses to highlight distributional dimensions and cross-country disparities, although they are not the central focus of the EKC framework adopted here. To provide an overview of the dataset and establish empirical context, descriptive statistics were calculated and are presented in

Table 2. Descriptive statistics of the data.

	GDPCON1	ECO2_GDP	Renewables	TR_EXPO
Mean	9.397386	0.295680	8.362720	22.51979
Median	9.309852	0.284998	5.910000	19.80842
Maximum	10.60192	0.522801	47.06783	100.0000
Minimum	8.234913	0.138225	0.000000	0.000000
Std. Dev.	0.625313	0.081234	9.438161	18.64803
Skewness	0.348634	0.742166	1.803104	1.578419
Kurtosis	2.422393	3.314176	6.795193	7.527432
Jarque-Bera	6.148604	17.26460	205.5617	228.4745
Probability	0.046222	0.000178	0.000000	0.000000
Sum	1691.530	53.22235	1505.290	4053.563
Sum Sq. Dev.	69.99194	1.181223	15,945.12	62,247.06
Observations	180	180	180	180

.

The correlation matrix provided in

Table 3. Correlation matrix of the model variables.

Correlation	GDPCON1	GDPCON2	Renew	TR_EXPO	ECO2_GDP
GDPCON1	1.000000
GDPCON2	0.999301	1.000000
RENEW	−0.457914	−0.457442	1.000000
TR_EXPO	0.214932	0.228864	−0.180676	1.000000
ECO2_GDP	−0.031434	−0.024414	−0.321657	−0.112513	1.000000

reveals several important patterns. GDP per capita (GDPCON1) and its squared term (GDPCON2) are almost perfectly correlated, as expected, confirming that they capture the same underlying income dynamics in the EKC framework. A strong negative correlation exists between income and renewable energy consumption (around −0.46), suggesting that higher-income MENA economies tend to rely less on renewables, likely due to entrenched fossil-fuel systems, subsidies, or infrastructural lock-ins. Renewable energy, however, shows a moderate negative correlation with carbon emissions per unit of GDP (ECO2_GDP at −0.32), supporting the view that an increase in the deployment of renewables improves environmental efficiency. Trade exposure (TR_EXPO) is only weakly correlated with both income (positive) and emissions intensity (negative), indicating that trade openness alone does not strongly drive environmental performance in the region. Finally, the weak negative correlation between income and emissions intensity (−0.03) provides only limited descriptive support for the Environmental Kuznets Curve, highlighting the need for econometric testing to capture the hypothesized nonlinear relationship. Together, these correlations point to complex and sometimes counterintuitive linkages between growth, structural change, and environmental outcomes in the selected MENA countries.

The next step in the methodological sequence is to complement these descriptive insights with graphical illustrations of the temporal dynamics of the dataset. Using scatter plots, trend lines, and stacked area graphs, the analysis visually explores the evolution of economic activity, renewable energy use, trade exposure, and emissions performance across the sample countries from 1990 to 2019. Such visualizations provide an intuitive means of identifying nonlinear relationships, outliers, and structural breaks that may not be immediately evident from the correlation matrix alone, while also contextualizing country-specific trajectories within regional patterns.

The charts in Figure 1 provide a visual analysis of economic and environmental trends for a group of countries between 1990 and 2010. The GDP at constant prices (GDPCON1) chart shows a general upward trajectory across most countries, indicating sustained real economic growth. Some countries, like Oman and Morocco, exhibit sharp increases in output toward the end of the period, suggesting either successful economic reforms or benefits from resource-driven development. In contrast, countries such as Mauritius show flatter growth, pointing to slower structural progress.

The visual analysis of the data highlights heterogeneous trajectories of economic growth, carbon intensity, renewable energy adoption, and trade across the selected MENA countries from 1990 to 2019. GDP at constant prices (GDPCON1) shows an upward trajectory across the panel, with sharp increases in Oman and Morocco suggesting resource-driven expansion and reforms, while Mauritius displays flatter growth. CO₂ emissions per unit of GDP (ECO₂_GDP) generally declined over time, indicating improvements in carbon efficiency, though temporary spikes in some countries reflect industrial surges or policy lapses. Renewable energy shares remain consistently low, with Mauritius showing temporary peaks, underscoring the limited and uneven regional transition to clean energy. Trade exports (TR_EXPO) appear highly volatile, especially in resource-dependent economies like Oman, reflecting external shocks and commodity reliance, while other countries show modest increases in export capacity.

Before conducting the econometric analysis, we employed scatter plots to visually explore potential relationships between each explanatory variable and environmental performance across the sample of MENA countries. This graphical approach served as a preliminary diagnostic tool, helping to identify the direction, form, and strength of associations—such as linearity or potential nonlinearities—between economic, energy, and transport variables and CO₂ emissions intensity. These visual patterns provided initial insights and informed the specification of the econometric model by highlighting plausible interactions and guiding expectations regarding the underlying relationships.

The scatter plots in Figure 2a–c illustrate the relationships between CO₂ emissions intensity (Eco2_GDP) and three factors for the selected MENA countries. More specifically, the scatter plots for MENA countries indicate that GDP per capita (PPP, constant 2017 international dollar) does not display a strong or consistent relationship with CO₂ emissions intensity (Eco2_GDP). This suggests that higher income levels in the region do not automatically translate into lower carbon intensity, pointing to the absence of a clear decoupling effect between growth and emissions. In contrast, renewable energy shares exhibit a strong negative association with emissions intensity, showing that greater reliance on renewables is linked to reductions in CO₂ relative to GDP, with Morocco standing out as a regional leader in renewable adoption. Transport services, measured as a percentage of commercial service exports, show a weaker but still negative relationship with emissions intensity. This implies that countries where transport services contribute more significantly to exports may experience modest improvements in emissions efficiency, likely reflecting a gradual shift toward service-oriented economic activity as opposed to heavy reliance on energy-intensive sectors.

Given the vague results provided by the above plot, we also provide a graph that presents the interlinkages in terms of individual countries in Figure 3.

The figure illustrates the relationship between CO₂ emissions intensity (Eco2_GDP) and three explanatory factors: GDP per capita, PPP (constant 2017 international dollar) in its logarithmic form (GDP1), the squared term of GDP per capita (GDP2), and transport services as a share of commercial service exports (TR_EXPO), across six MENA countries (Algeria, Egypt, Lebanon, Mauritius, Morocco, and Oman).

The plots reveal heterogeneous dynamics. For most countries (Egypt, Mauritius, Morocco, Oman), GDP1 shows a downward trend with respect to emissions intensity, while GDP2 remains relatively flat, indicating weak evidence of a nonlinear Environmental Kuznets Curve (EKC) effect. Lebanon appears to deviate, showing a less systematic pattern, which may be explained by political and economic instability influencing both income and environmental performance. Transport services (TR_EXPO) show more variability across the sample: in some cases, such as Egypt and Oman, TR_EXPO tends to increase as emissions intensity falls, pointing to a modest shift toward services contributing to reduced carbon intensity. In other cases, such as Algeria and Morocco, the association is weaker, reflecting the dominance of resource-based activities.

The selection of Algeria, Egypt, Lebanon, Mauritius, Morocco, and Oman reflects the structural diversity within the MENA region, which makes them particularly suitable for panel investigation. Algeria and Oman represent resource-dependent economies with high reliance on fossil fuels, high emissions intensity, and weak renewable penetration. Egypt and Morocco illustrate middle-income economies undergoing gradual diversification, with Morocco in particular emerging as a regional leader in renewable energy deployment. Mauritius offers the perspective of a small island and service-oriented economy, where both renewable energy and transport services play a more visible role in shaping emissions outcomes. Lebanon, meanwhile, highlights the case of a service-dominated economy where political and structural constraints create less systematic environmental patterns. Together, these countries capture a broad spectrum of energy and economic profiles within the region, thereby providing a balanced panel that justifies the use of cointegration methods to uncover common long-run relationships while accounting for heterogeneity.

Overall, the figure underscores the necessity of panel data cointegration analysis. Country-specific plots alone cannot fully capture the long-run equilibrium relationships, as the individual time series are short (≈30 years) and affected by structural shocks. Panel cointegration techniques allow for pooling across countries, thereby improving statistical power, while also accounting for heterogeneity. When combined with FMOLS and DOLS estimators, the analysis provides robust long-run coefficients, confirming the limited role of income growth alone, the importance of renewable energy adoption, and the supplementary role of transport services in reducing CO₂ intensity in the MENA region.

Unlike conventional studies that focus narrowly on single-country cases or isolate variables without accounting for regional dynamics, this study adopts a regionally comparative and temporally expansive approach. It provides insight into how structural and policy-driven differences influence environmental performance across diverse economic contexts within selected MENA countries [59,94,95,96,97]. The core objective is to generate robust empirical evidence on whether economic development trajectories in the region can align with environmental sustainability targets, particularly through increased reliance on renewable energy and deeper integration into global trade.

The analysis seeks to determine whether higher income levels are associated with reduced emissions intensity, whether renewable energy expansion yields measurable environmental benefits, and how trade dynamics influence carbon efficiency. It also explores how these relationships vary across income tiers and development stages. In doing so, the study aims to clarify whether MENA economies can decouple economic growth from environmental harm, and under what specific policy and structural conditions such a transition becomes feasible [94,95,96].

The econometric methodology further strengthens this investigation by combining complementary techniques that capture both long-run equilibrium dynamics and short-run causal linkages. We began by testing for cross-sectional dependence (CSD) using Pesaran’s [98] diagnostic, since economic, trade, and energy shocks are likely to spill over across MENA economies. Accounting for CSD is crucial, as failure to do so can bias unit root and cointegration results in panel settings.

To assess the time-series properties of the data, both first- and second-generation panel unit root tests were applied [99]. While first-generation tests (e.g., Levin–Lin–Chu, Im–Pesaran–Shin) assume cross-sectional independence, second-generation methods relax this restriction [98,100,101,102,103]. Specifically, the CIPS test [104] and the PANIC method [105] were employed. These approaches allow for heterogeneous dynamics and cross-sectional dependence, producing more robust evidence of whether the variables are stationary in levels or differences. This ensures that subsequent estimations are not undermined by spurious results due to unmodeled cross-country correlations.

Once the integration order was established, we tested for cointegration using both the Pedroni [106,107,108,109] and Kao [110,111] residual-based approaches. These tests confirmed whether GDP, carbon intensity, renewable energy, and trade variables share a long-run equilibrium relationship across the panel.

Given the evidence of cointegration, long-run coefficients were estimated using Fully Modified Ordinary Least Squares [104] and Dynamic Ordinary Least Squares [110,111]. FMOLS corrects for endogeneity and serial correlation through non-parametric adjustments, while DOLS incorporates leads and lags in the regressors to absorb short-run dynamics. To complement these long-run relationships, panel Granger causality tests [112] were conducted, capturing short-run feedback loops and directional linkages among variables [58,78].

This layered approach—combining CSD diagnostics, second-generation unit root tests (CIPS and PANIC), panel cointegration, FMOLS/DOLS estimation, and causality testing—ensures robustness. It provides a comprehensive framework for analyzing the income–environment nexus while addressing structural heterogeneity, cross-country spillovers, and potential endogeneity [105,106,107,108,109,110,111].

This analytical approach provides a crucial extension to the visual assessments presented earlier through scatter plots and time-series area graphs. While the graphical analysis offered preliminary insights into the possible direction and strength of variable relationships—such as the inverse relationship between GDP and emissions intensity or the negative correlation between renewables and CO₂ emissions—the econometric analysis validates these patterns under stricter statistical conditions. It quantifies the magnitude and significance of these relationships, handles unobserved heterogeneity, and identifies structural consistencies or divergences across the sample. Together, the visual and econometric analyses complement each other, creating a more complete picture of the environmental–economic dynamics at play in the MENA region and informing more targeted, evidence-based policy interventions.

4.2. Methodology

4.2.1. Panel Data—Unit Root Tests

This empirical analysis, as mentioned above, was based on annual data for the period 1990–2019, yielding 30 observations per country and a total of 180 observations across six MENA economies (Algeria, Egypt, Lebanon, Mauritius, Morocco, and Oman). The relatively small cross-sectional dimension reflects strict data availability constraints, particularly for renewable energy and transport service exports. To ensure consistency, only countries with uninterrupted data series for the full set of variables were included. Although the panel is small-N, large-T, the 30-year time span supports the use of panel cointegration estimators such as FMOLS and DOLS, which are designed to handle heterogeneous panels with a moderate cross-sectional size but longer time series [106,107,108,109,110].

In panel data econometrics, testing for unit roots is an essential first step in determining the time-series properties of the variables and identifying their order of integration before proceeding to cointegration analysis. This study applies both first-generation and second-generation panel unit root tests to ensure the robustness of the findings, especially in light of the strong likelihood of cross-sectional dependence among MENA countries due to shared economic, geographic, and policy factors. More specifically, the cross-sectional dependence test was employed and validated this dependence. In addition to the implementation of the second-generation unit root tests that account for cross-sectional dependence, we also employed first-generation tests.

For the first-generation tests, the Levin, Lin, and Chu (LLC) and the Im, Pesaran, and Shin (IPS) tests were utilized [98,99,100,101,102]. The LLC test assumes a common autoregressive parameter across all panel units and tests the null hypothesis of a unit root against the alternative: that all series are stationary with a shared dynamic structure. While its pooled framework enhances statistical power, its assumption of homogeneity can be overly restrictive for structurally diverse economies. In contrast, the IPS test allows for heterogeneity in the autoregressive coefficients, testing whether at least some of the panel units are stationary. This makes it more flexible and appropriate for datasets where countries differ in structural and policy contexts. However, both LLC and IPS assume cross-sectional independence—an assumption that is often violated in macroeconomic panels, especially in regions like MENA, where global shocks and regional spillovers are common.

To address this issue, the study also employs second-generation unit root tests that account for cross-sectional dependence. The PANIC method by Bai and Ng [105] decomposes each series into common and idiosyncratic components, enabling a nuanced understanding of whether nonstationarity arises from shared global trends or country-specific dynamics. This distinction is particularly relevant in MENA, where both global energy prices and local governance influence macroeconomic indicators. PANIC offers greater diagnostic precision and accommodates a wide range of dependence structures.

In addition, Pesaran’s Cross-sectionally Augmented Dickey–Fuller (CADF) test is applied. This test directly incorporates cross-sectional averages of the variables and their lagged differences into the unit root regression, effectively capturing unobserved common factors such as synchronized policy shifts or external shocks. The CADF test is well-suited to macroeconomic panels with significant cross-country interlinkages and provides more robust inference when the panel includes a large number of countries [104,105].

By combining these first- and second-generation unit root tests, the analysis establishes a rigorous basis for evaluating the integration properties of the variables. The dual approach allows for both traditional benchmarking and correction for potential cross-sectional dependence, increasing confidence in the identification of long-run relationships. This methodological rigor ensures that subsequent cointegration analysis and long-run estimations are built upon solid ground, thereby enhancing the credibility and policy relevance of the empirical results.

4.2.2. Panel Data—Cointegration Tests

To determine whether a long-run equilibrium relationship exists among the non-stationary variables, this study employs panel cointegration tests. Such tests are critical in identifying stable, long-term associations among variables that are individually integrated of order one, I(1). Following confirmation of unit roots, two widely used residual-based panel cointegration tests were applied: the Kao test and the Pedroni tests. Both methods build upon the Engle–Granger two-step framework but differ in their assumptions about homogeneity and cross-sectional dynamics [107,108,109,110,111].

The Kao test assumes a homogeneous cointegrating vector and identical autoregressive dynamics across all cross-sectional units. It first estimates the long-run relationship using pooled ordinary least squares (OLS) and then assesses the stationarity of the residuals to determine cointegration [110,111]. If the residuals are stationary, this suggests that the variables share a common stochastic trend and thus move together over time. Despite its simplicity and interpretability, the Kao test imposes restrictive assumptions—namely, cross-sectional independence and homogeneity—which may not hold in structurally diverse macroeconomic panels like those of MENA countries.

To address these limitations, the Pedroni tests introduce greater flexibility by allowing for heterogeneity in both the cointegrating relationships and short-run dynamics across countries. Pedroni developed seven test statistics: four within-dimension (panel) statistics that pool autoregressive coefficients across units, and three between-dimension (group-mean) statistics that allow for coefficients to vary. This framework accommodates country-specific intercepts, slopes, and dynamics, making it particularly well-suited for analyzing panels with varied institutional structures, policy environments, and development trajectories. In both test categories, the null hypothesis is the absence of cointegration, while the alternative permits cointegration in at least some countries [107,108].

These cointegration tests are applied after confirming that the variables are non-stationary but integrated of the same order. Additionally, to account for potential cross-sectional dependence in the data, the analysis incorporates Pesaran’s Cross-sectionally Augmented Dickey–Fuller (CADF) test as part of the unit root testing framework. Pesaran’s approach enhances the reliability of subsequent cointegration testing by controlling unobserved common factors that might influence all panel members simultaneously, such as regional shocks or global economic cycles [104,105].

Together, the Kao and Pedroni tests offer complementary strengths: the former provides a straightforward benchmark, while the latter enables more realistic modeling of panel heterogeneity and dependence. The confirmation of cointegration using both methods lays a robust empirical foundation for the use of long-run estimators such as Fully Modified OLS (FMOLS) and Dynamic OLS (DOLS), which are only valid when a long-run equilibrium relationship is present. This methodological rigor enhances the credibility of the study’s findings on the environmental and economic dynamics within the MENA region [110,111,112,113,114,115,116,117,118].

4.2.3. FMOLS and DOLS Estimators for Long-Run Relationships

The current study employs three advanced cointegration estimation techniques to assess the long-run relationship among the study variables: Dynamic Ordinary Least Squares (DOLS), Fully Modified Ordinary Least Squares (FMOLS), and Canonical Cointegration Regression (CCR). These methodologies are selected for their capacity to provide consistent, unbiased, and asymptotically efficient estimates in the presence of non-stationary time series and potential endogeneity among regressors.

The DOLS technique, introduced by Stock and Watson [100] and extended to panel contexts by Kao and Chiang [110], addresses endogeneity and serial correlation by incorporating both leads and lags of the first-differenced independent variables. This parametric approach augments the standard cointegration regression to eliminate bias caused by feedback effects from the dependent variable to the regressors. The DOLS specification can be expressed as follows [110]:

$${\mathrm{y}}_{\mathrm{i}}\mathrm{t}={\mathsf{\alpha }}_{\mathrm{i}}+{\mathsf{\beta }}_{\mathrm{i}}{\mathrm{x}}_{\mathrm{i}}\mathrm{t}+{\mathsf{\Sigma }}_{\mathrm{j}=-{\mathrm{p}}_{\mathrm{i}}}^{{\mathrm{p}}_{\mathrm{i}}}{\mathsf{\gamma }}_{\mathrm{i}}\mathrm{j}\mathsf{\Delta }{\mathrm{x}}_{\mathrm{i},\mathrm{t}-\mathrm{j}}+{\mathsf{\epsilon }}_{\mathrm{i}}\mathrm{t}$$

(1)

where y_t is the dependent variable, x_t the non-stationary regressor, Δx_t the first differences in x_t leads and lags, and ε_t is the error term. The inclusion of both future and past changes in the explanatory variables helps mitigate simultaneity and serial correlation [105,106,107,108,109,110,111].

The FMOLS method, developed by Phillips [113], is a semi-parametric estimator that modifies the OLS estimator to correct for both endogeneity and serial correlation in the cointegrating regression. FMOLS achieves this by adjusting the error term using non-parametric estimates of the long-run covariance matrix. The FMOLS estimator is based on the following form:

$${\mathrm{y}}_{\mathrm{i}}\mathrm{t}={\mathsf{\alpha }}_{\mathrm{i}}+{\mathsf{\beta }}_{\mathrm{i}}{\mathrm{x}}_{\mathrm{i}}\mathrm{t}+{\mathrm{u}}_{\mathrm{i}}\mathrm{t}$$

(2)

with corrections applied to u_t via the long-run covariance matrix Ω (eq2), which is decomposed into

$$\Omega ={\Omega }_0+{\Sigma }_{\mathrm{j}=1}^{\mathrm{\infty }}{\Gamma }_{\mathrm{j}}$$

(3)

$${\mathrm{y}}_{\mathrm{i}}\mathrm{t}={\mathsf{\alpha }}_{\mathrm{i}}+{\mathsf{\beta }}_{\mathrm{i}}{\mathrm{x}}_{\mathrm{i}}\mathrm{t}+{\mathrm{u}}_{\mathrm{i}}\mathrm{t}\ast$$

(4)

Here, Ω₀ is the contemporaneous covariance and Γ_j are autocovariances at lag j. The FMOLS estimator uses kernel-based techniques to estimate these quantities and obtain corrected residuals, thereby yielding robust long-run coefficients [114].

All estimators—FMOLS (non-parametric) and DOLS (parametric)–address the key econometric challenges of cointegrated systems: endogeneity and serial correlation. While FMOLS and DOLS adopt different strategies to deal with these issues, both have been shown to outperform traditional OLS in cointegrated panels, especially under small-sample conditions. The DOLS estimator is particularly effective in time series forecasting due to its incorporation of deterministic trends and its ability to model dynamic adjustments through lag and lead terms. Meanwhile, FMOLS is robust to structural differences and cross-sectional dependence when extended to panel frameworks [105,106,107,108,109,113,114].

The combined application of FMOLS and DOLS estimators ensures a triangulated and consistent assessment of the long-run parameters. Utilizing both approaches enhances the robustness and credibility of empirical results by allowing cross-validation of outcomes under distinct estimation assumptions. The analysis, conducted in accordance with the specified methodology, yields results that are presented and critically discussed in Section 5.

5. Results

5.1. Panel Unit Root and Cointegration Test Results

The first step in our empirical analysis involves determining the order of integration for all variables included in the model through the application of panel unit root tests. To ensure the appropriateness and reliability of the chosen unit root tests, we initially conducted a cross-sectional dependence (CD) test. This preliminary step is essential, as the presence of cross-sectional dependence can significantly affect the performance and validity of traditional panel unit root tests [99,100,101,102,103]. By accounting for cross-sectional dependencies, the selected unit root tests provide more robust and accurate assessments of stationarity within the panel data framework.

The results of the cross-sectional dependence tests in

Table 4. Cross-section dependence test.

	ECO2_GDP	GDPCON1	Renewables	TR_EXPO
Breusch-Pagan LM	215.95 *** (0.000)	251.26 *** (0.0000)	NA	81.54 *** (0.000)
Pesaran scaled LM	36.689 *** (0.000)	43.136 *** (0.0000)	NA	12.15 *** (0.000)
Bias-corrected scaled LM	36.586 *** (0.000)	43.03 *** (0.000)	NA	12.04 *** (0.0000)
Pesaran CD	4.506 *** (0.0000)	14.69 *** (0.0000)	NA	0.839225 (0.4013)

*** rejection of null hypothesis of no cross-section dependence.

indicate a significant degree of interdependence among the variables across the MENA countries in the panel. For CO₂ emissions per unit of GDP (ECO₂_GDP), all applied tests—the Breusch–Pagan LM, Pesaran scaled LM, bias-corrected scaled LM, and Pesaran CD—yield highly significant results with p-values below 0.01. This strong evidence of cross-sectional dependence suggests that environmental performance is not determined in isolation within each country. Rather, it is likely influenced by regional spillovers, shared economic conditions, environmental policy diffusion, or exposure to common shocks such as oil market fluctuations or climate-related events.

GDP per capita at constant prices (GDPcon1) similarly shows significant cross-sectional dependence across all testing methods. The magnitude of the Pesaran CD statistic, in particular, indicates a high level of synchronization in economic activity among MENA economies. This may be attributed to their structural dependence on energy markets, regional trade agreements, or synchronized investment cycles. In practical terms, this means that economic developments in one MENA country are likely to be correlated with outcomes in others, reinforcing the need to account for such linkages in econometric modeling.

For the trade services variable (TRexpo), the results are somewhat mixed. While the LM-based tests point to significant cross-sectional dependence, the Pesaran CD [104] test yields an insignificant result, suggesting the presence of only moderate or variable cross-sectional effects. This discrepancy could be explained by heterogeneity in trade structures across countries in the region. Some economies may be more integrated into global trade networks, while others are more domestically focused or reliant on bilateral trade flows. The inconsistency also implies that while common trade dynamics exist to some extent, they are not as pronounced or uniform, as in the case of economic output or emissions efficiency.

No test results were reported for the renewable energy share variable, which may be due to data limitations such as unbalanced coverage, insufficient variation, or missing observations in the panel. This warrants further investigation and possibly data refinement, as the renewable energy component is critical to understanding the region’s environmental trajectory.

Overall, the results provide strong evidence that cross-sectional dependence is a prominent feature of the dataset, particularly concerning emissions and economic indicators. This finding underscores the necessity of employing second-generation methods that explicitly accommodate cross-sectional dependence, although first-generation unit root tests are also employed and provided for comparison purposes [99,100,101,102,103]. It is also worth noting that ignoring such interdependencies can lead to biased estimators and unreliable inference, especially in studies involving integrated global or regional economies. The observed cross-sectional dependence reflects the deeply interconnected nature of environmental and economic processes among the countries comprising the research area, likely driven by shared economic structures, energy–trade relationships, and regional environmental spillovers. These dynamics highlight the importance of coordinated regional strategies and policy frameworks, as unilateral national approaches may be less effective in addressing transboundary environmental challenges. Promoting regional cooperation could thus enhance both environmental sustainability and long-term economic development across the MENA region.

The results of the panel unit root tests as provided in

Table 5. First- and second-generation unit root tests results.

	Carbon Emissions		GDPcon1		GDPcon2		Renewables		TRexpo
	Non-cross-sectionally dependent unit root tests
	Level	First Diff.	Level	First Diff.	Level	First Diff.	Level	First Diff.	Level	First Diff.
Levin, Lin and Chu t *	−1.079	−4.455 *** (0.000)	1.861 *** (0.969)	−0.801 (0.212)	1.72217 (0.96)	−0.932 (0.176)	0.501 (0.692)	−4.558 *** (0.000)	-0.470 (0.319)	−4.375 *** (0.00)
Im, Pesaran and Shin W-stat	1.34052 0.12126	−5.72 *** (0.00)	−0.68 (0.2484)	−4.07 *** (0.000)	0.31953 0.6253	−4.012 *** (0.00)	−0.09757 (0.46)	−4.53 *** (0.000)	−1.03096 (0.15)	−6.60 *** (0.000)
ADF-Fisher Chi-square	1.84363 (0.97)	53.1 *** (0.000)	4.21580 (0.98)	42.32 *** (0.000)	13.8661 0.31	41.8 *** (0.000)	8.643 (0.567)	37.7 *** (0.000)	19.2545 * (0.0826)	61.64 *** (0.000)
PP-Fisher Chi-square	−8.355 −0.09	126.1 *** (0.000)	17.1788 (0.1430)	105.6 *** (0.000)	21.5334 ** (0.043)	102.18 *** (0.00)	16.9821 * (0.075)	111.7 *** (0.000)	23.45 * (0.024)	321.24 *** (0.00)
Cross-sectionally dependent unit root tests
Bai and NG-Panic	−1.691 (0.09)	1.737 *** (0.082)	−1.039 (0.29)	Inf *** (0.000)	1.008 (0.313)	2.89 *** (0.004)	6.3544 (0.999)	Inf *** (0.00)	−1.75 (0.08)	9.812 *** (0.000)
Pesaran -CIPS	−1.848 ≥ 0.10	−3.72 *** < (0.01)	−1.077 ≥ 0.10	−2.259* < 0.10	0.5016 ≥ 0.10	−2.164 * < 0.10	−1.69 ≥ 0.10	−2.738 ***< 0.01	−1.418 *< 0.10	−2059 *** < 0.01

***, **, and * indicate rejection of the null hypothesis (unit root process) at the 1%, 5%, and 10% levels of significance, respectively.

indicate that the majority of variables employed in the EKC model—including carbon emissions, GDP per capita (GDPcon1), squared GDP per capita (GDPcon2), renewable energy consumption, and transport services exports (TRexpo)—are non-stationary when level but become stationary after first differencing, implying that they are integrated in order one, I(1). This finding is consistent across a range of first-generation tests, such as Levin, Lin and Chu (LLC), Im, Pesaran and Shin (IPS), ADF-Fisher, and PP-Fisher (individual root), all of which produce statistically significant results at the first-difference level for most variables (e.g., IPS W-stat for GDPcon² = −4.071, p < 0.01). However, the presence of cross-sectional dependence among the panel units—particularly in emissions and economic indicators—necessitates the application of second-generation unit root tests. The Bai and NG PANIC and CIPS tests, which account for cross-sectional correlation, offer more nuanced outcomes. While some variables, notably GDPcon², show strong evidence of stationarity in first differences (CIPS = −2.259, p < 0.10), others, such as TRexpo and renewables, exhibit weaker or borderline significance levels, suggesting potential heterogeneity in integration orders or structural breaks. While first-generation unit root and cointegration tests are reported for comparison with earlier studies, all substantive interpretations and policy conclusions in this paper rely on the more appropriate second-generation tests, which explicitly account for cross-sectional dependence [99,100,101,102,103]. Consequently, the application of panel cointegration techniques and estimators such as Fully Modified OLS (FMOLS) and Dynamic OLS (DOLS) is justified, as these approaches provide robust inference in the presence of cross-sectional dependence and heterogeneous dynamics [98,99,100,101,102,103,104,105]. The confirmation that the variables are I(1) or I(0) supports the validity of proceeding with cointegration analysis to explore long-run relationships within the Environmental Kuznets Curve framework.

The results from the Pedroni Residual Cointegration Test and the Johansen Fisher Panel Cointegration Test in

Table 6. Pedroni residual cointegration test.

Alternative Hypothesis: Common AR Coefs. (Within-Dimension)
				Weighted
		Statistic	Prob.	Statistic	Prob.
Panel v-Statistic		−1.431950	0.9239	−0.800103	0.7882
Panel rho-Statistic		1.723443	0.9576	1.101340	0.8646
Panel PP-Statistic		1.197190	0.8844	−0.300100	0.3821
Panel ADF-Statistic		−0.164569	0.4346	−1.897748	0.0289
Alternative Hypothesis: Individual AR Coefs. (Between-Dimension)
		Statistic	Prob.
Group rho-Statistic		1.284895	0.9006
Group PP-Statistic		−0.770325	0.2206
Group ADF-Statistic		−2.092631	0.0182

and

Table 7. Johansen Fisher panel cointegration test.

Hypothesized	Fisher Stat. *		Fisher Stat. *
No. of CE(s)	(from Trace Test)	Prob.	(from Max-Eigen Test)	Prob.
None	126.7	0.0000	91.53	0.0000
At most 1	50.35	0.0000	26.23	0.0034
At most 2	29.86	0.0009	16.86	0.0775
At most 3	19.69	0.0324	12.41	0.2584
At most 4	15.28	0.1221	15.28	0.1221

* Probabilities are computed using asymptotic Chi-square distribution.

respectively offer complementary insights into the long-run relationships among the variables under study: CO₂ emissions per unit of GDP (ECO₂_GDP), GDP per capita levels (GDPCON1 and GDPCON2), renewable energy share, and transport services export share (TR_EXPO) across the MENA countries.

The Pedroni test, which evaluates cointegration under both within-dimension and between-dimension frameworks, yields mixed results. Most statistics (Panel v, rho, and PP) have high p-values, failing to reject the null hypothesis of no cointegration. However, one notable exception is the Panel ADF-Statistic (weighted) and the Group ADF-Statistic, both of which are statistically significant at the 5% level (p = 0.0289 and p = 0.0182), indicating cointegration under specific assumptions. These isolated rejections suggest weak evidence of cointegration, possibly due to heterogeneity across countries or insufficient power in the residual-based framework given the small panel size (five cross-sections).

By contrast, the Johansen Fisher panel cointegration test provides strong and consistent evidence of cointegration. Both the trace and maximum eigenvalue tests reject the null hypothesis of no cointegration at the 1% significance level for up to three cointegrating vectors. The Fisher tests are validated as statistically significant confirming that a stable long-run equilibrium relationship exists among the variables. These results are more robust and reliable in this context because the Johansen approach allows for multiple cointegrating relationships and explicitly models the joint dynamics of the entire system. The contrast between the Pedroni and Johansen test outcomes reflects methodological differences. While the Pedroni test may suffer from limited power in panels with few countries and possible cross-sectional dependence, the Johansen method aggregates individual countries’ test results, enhancing its overall power and robustness [105,106,107,108,109,110,111,112,113,114].

The confirmation of cointegration, especially through the Johansen test, lends robust empirical support to the validity of the Environmental Kuznets Curve (EKC) hypothesis in the sample countries of the MENA region. It indicates that CO₂ emissions intensity, income levels (GDPCON1 and GDPCON2), the share of renewable energy, and the structure of the service economy as proxied by transport service exports (TR_EXPO) are not independent or randomly correlated but instead move together over the long term. This implies that environmental performance in these countries is shaped by structural economic and energy dynamics rather than short-term variations, reinforcing the view that growth, energy transition, and trade composition are deeply interconnected.

The cointegration findings provide a sound justification for the application of long-run estimators such as Fully Modified Ordinary Least Squares (FMOLS) and Dynamic Ordinary Least Squares (DOLS), which are only valid when a stable long-term relationship among variables exists. These methods allow for a more accurate estimation of the elasticities and directions of influence between economic activity, renewable energy penetration, trade orientation, and environmental degradation. Importantly, the presence of cointegration confirms that the observed nonlinear associations and causality results are not the result of temporary shocks or data artifacts but rather reflect meaningful long-run dynamics consistent with EKC theory [73,74,75,76,77,78,79,80,81,82,83,84,85,86,87].

This outcome suggests that economic growth, coupled with appropriate structural shifts—such as greater reliance on renewable energy and the transformation of services through cleaner transport systems, can reduce emissions intensity and improve environmental outcomes over time. The policy implication is clear: the transition to sustainability in the specific sample of MENA countries requires not just growth, but deliberate efforts to green that growth through institutional, technological, and sectoral reforms. Without these targeted measures, the environmental benefits associated with income gains may remain elusive. Therefore, the long run cointegration between these variables highlights the importance of integrating climate objectives into economic development strategies, ensuring that the pathways to growth are aligned with long-term environmental goals.

5.2. FMOLS and DOLS Estimation Results

The results of the panel Dynamic Ordinary Least Squares (DOLS) regression offer strong empirical support for the Environmental Kuznets Curve (EKC) hypothesis within the MENA countries studied. The negative and statistically significant coefficient of GDP per capita (GDPCON1) indicates that at lower levels of income, economic growth leads to reductions in CO₂ emissions per unit of GDP, reflecting early-stage environmental improvements due to factors such as improved efficiency, modernization, and initial environmental reforms. In contrast, the positive and significant coefficient on the squared term of GDP per capita (GDPCON2) confirms the presence of an inverted U-shaped relationship, where emissions intensity begins to rise again as income surpasses a certain threshold [103,104,105]. This result substantiates the EKC framework, implying that environmental degradation initially intensifies with economic growth, then declines, but may increase again without targeted interventions.

The results of the panel Dynamic Ordinary Least Squares (DOLS) regression in

Table 8. The DOLS estimation results in the long-run relationship between carbon intensity and its determinants.

Variable	Coefficient	Std. Error	t-Statistic	Prob.
GDPCON1	−3.785578 ***	1.095444	−3.455748	0.0009
GDPCON2	0.192848 ***	0.057937	3.328563	0.0014
RENEWABLES	−0.006349 ***	0.001456	−4.359853	0.0000
TR_EXPO	−0.001188 **	0.000533	−2.227317	0.0291
R-squared	0.979744	Mean dependent var		0.287121
Adjusted R-squared	0.961201	S.D. dependent var		0.078510
S.E. of regression	0.015465	Sum squared resid		0.016980
Long-run variance	0.000178

Notes: ** and *** represent rejection of null hypothesis for 5 and 1% levels of significance.

offer strong empirical support for the Environmental Kuznets Curve (EKC) hypothesis within the selected MENA countries but with the reverse sign. The negative and statistically significant coefficient of GDP per capita (GDPCON1) indicates that, at lower levels of income, economic growth leads to reductions in CO₂ emissions per unit of GDP, reflecting early-stage environmental improvements due to factors such as improved efficiency, modernization, and initial environmental reforms. In contrast, the positive and significant coefficient on the squared term of GDP per capita (GDPCON2) confirms the presence of an inverted U-shaped relationship, where emissions intensity begins to rise again as income surpasses a certain threshold. This result substantiates the EKC framework, implying that environmental degradation initially intensifies with economic growth, then declines, but may increase again without targeted interventions [35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50].

The negative and highly significant coefficient on the renewable energy variable confirms that increasing the share of renewables in the energy mix is associated with substantial long-run reductions in carbon intensity. This result emphasizes the importance of expanding clean energy sources as a primary strategy for mitigating emissions in the region. Furthermore, the negative and statistically significant coefficient for TR_EXPO, which serves as a proxy for the structural composition of the service economy by measuring transport services as a percentage of commercial service exports, is a finding that should not be interpreted as direct evidence of low-carbon diversification, since aviation and shipping are themselves carbon-intensive; rather, the result may reflect broader structural and efficiency effects associated with service-sector expansion [87,88,89,90,91,92,93,94,95,96].

The exceptionally high R-squared value of approximately 0.98 confirms that the model explains nearly all of the variation in emissions intensity, enhancing confidence in the reliability of these results. Collectively, the findings confirm the validity of the EKC in the MENA context, with clear implications for policy design. First, MENA countries should not assume that economic growth alone will yield sustained environmental improvements; instead, complementary policies targeting energy transition and service sector modernization are essential. Second, renewable energy development must be prioritized not just as a climate strategy, but as an economic policy lever for long-term sustainability. Finally, the role of structural transformation in the service economy—particularly through the greening of transport services, should be integrated into broader climate and trade policy agendas. These results underscore the need for coordinated and forward-looking governance that aligns development with environmental protection, ensuring that long-run growth paths in the region are both economically robust and ecologically responsible.

The threshold for this estimated model is provided by the following formula [57]:

$$\mathrm{Turning} \mathrm{point} \mathrm{in} \mathrm{G}\mathrm{D}\mathrm{P}\mathrm{C}\mathrm{O}\mathrm{N}1=-{\displaystyle \frac{{\beta }_{\mathrm{GDPCON}1}}{2{\beta }_{\mathrm{GDPCON}2}}}$$

By substituting the values estimated, the result was

$$\mathrm{X}=-{\displaystyle \frac{-3.785578}{22\times 0.19284}}=9.82$$

Since GDPCON1 is negative and GDPCON2 is positive and significant, this confirms a U-shaped relationship, not the inverted-U of the standard EKC. In this case, at low income levels, higher GDP per capita reduces carbon intensity. After the threshold (~9.82), further income growth increases emissions intensity.

The results of the Fully Modified Ordinary Least Squares (FMOLS) as shown in

Table 9. FMOLS estimation results for the long-run relationship between carbon intensity and its determinants.

Variable	Coefficient	Std. Error	t-Statistic	Prob.
GDPCON2	−0.219509 ***	0.059668	−3.678818	0.0004
RENEWABLES	−0.008405 ***	0.001034	−8.131598	0.0000
GDPCON1	3.993197 ***	1.139407	3.504628	0.0007
TR_EXPO	−0.000998 ***	0.000222	−4.490533	0.0000
R-squared	0.935897		Mean dependent var	0.287817
Adjusted R-squared	0.921577		S.D. dependent var	0.086497
S.E. of regression	0.024223		Sum squared resid	0.055154
Long-run variance	0.000180

*** indicate rejection of the null hypothesis at the 1%, level of significance, respectively.

reverses the signs: it reports a positive coefficient for GDPCON1 and a negative coefficient for GDPCON2, implying a U-shaped relationship. This result contradicts the conventional EKC narrative and may indicate either model sensitivity or underlying nonlinear dynamics that differ across the MENA countries studied. In the FMOLS regression, the positive and significant coefficient of GDP per capita, coupled with a negative and significant coefficient on its squared term, indicates an inverted U-shaped relationship between income and CO₂ emissions per unit of GDP. This supports the idea that emissions initially rise with income growth but begin to decline after a certain income level is surpassed, in line with the EKC theory. Importantly, the coefficient on renewable energy is negative and highly significant, implying that a higher share of renewables in the energy mix is associated with long-term reductions in emissions intensity. This effect appears even stronger in the FMOLS model than in the DOLS estimates, suggesting that after correcting for potential endogeneity and serial correlation, renewable energy’s contribution to environmental efficiency becomes more pronounced. The negative and statistically significant coefficient for transport services as a share of commercial service exports (TR_EXPO) is a finding that can be interpreted wrongly. More specifically, we note that transport modes such as shipping and aviation are themselves carbon-intensive. For example, Gray et al. [29] document how decarbonizing these sectors requires major fuel and technology shifts due to their high-energy-density needs and operational constraints. Thus, the negative coefficient may more plausibly reflect the structural and efficiency gains associated with broader service-sector expansion rather than a pure low-carbon transition in transport exports.

A significant step in our analysis is the estimation of the EKC turning point. To obtain the EKC turning point (income threshold), use the quadratic rule:

$$\mathrm{Turning} \mathrm{point} \mathrm{in} \mathrm{G}\mathrm{D}\mathrm{P}\mathrm{C}\mathrm{O}\mathrm{N}1=-{\displaystyle \frac{{\beta }_{\mathrm{GDPCON}1}}{2{\beta }_{\mathrm{GDPCON}2}}}$$

Plugging in your FMOLS estimates

(GDPCON1 = 3.993197, GDPCON2 = −0.219509):

$$x^{*}=-{\displaystyle \frac{3.993197}{2(-0.219509)}}\approx 9.096$$

The interpretation is as follows (pick the one that matches how GDPCON1 is defined in your data):

Given that GDPCON1 is the natural log of GDP per capita, the income threshold is $\mathrm{e}\mathrm{x}\mathrm{p}\left(9.096\right)\approx 8900$ (≈$8.9 k per capita, in constant 2017 PPP universal dollars).

Since both ${\beta }_{\mathrm{GDPCON}1}$ and ${\beta }_{\mathrm{GDPCON}2}<0$ (and highly significant ***), the inverted-U EKC shape is satisfied, and the economy transitions to decline in carbon intensity once GDP per capita crosses this threshold.

The empirical examination of the Environmental Kuznets Curve (EKC) hypothesis with the two different methodologies yields mixed results across estimation techniques. The DOLS model supports the EKC, as evidenced by the negative coefficient on GDP per capita (GDPCON1) and positive coefficient on its squared term (GDPCON2), consistent with an inverted U-shaped relationship between income and environmental degradation. This suggests that at early stages of economic development, CO₂ emissions rise with income growth but eventually decline after surpassing a certain income threshold.

Conversely, the FMOLS estimation reverses the signs: it reports a positive coefficient for GDPCON1 and a negative coefficient for GDPCON2, implying a U-shaped relationship. This result contradicts the conventional EKC narrative and may indicate either model sensitivity or underlying nonlinear dynamics that differ across the MENA countries studied.

The two estimators reveal contrasting income–emissions profiles. The FMOLS model supports an inverted-U shaped EKC, with a turning point at approximately 9.10 (equivalent to about USD 8900 if GDPCON1 is log GDP per capita), suggesting that beyond this income level, further growth is associated with declining emissions intensity. By contrast, the second specification (with β₁ < 0 and β₂ > 0) implies a U-shaped trajectory, with a turning point closer to 9.82 (about USD 18,400), where emissions initially fall with income but begin rising again after the threshold. This divergence is not unusual and reflects the estimator design, specification sensitivity, and heterogeneity of the sample. FMOLS emphasizes long-run equilibrium by correcting for endogeneity and serial correlation through a non-parametric covariance structure, while DOLS incorporates leads and lags in regressors to capture short-run dynamics, potentially shifting its curvature depending on deterministic assumptions, lag length, bandwidth selection, or omitted structural variables such as institutions or energy prices. Moreover, cross-country heterogeneity in the MENA region means that national turning points may vary substantially, and pooled models risk averaging across distinct developmental and environmental regimes.

This divergence highlights the importance of methodological robustness and appropriate model specification when testing the EKC hypothesis. It may also reflect structural heterogeneity in how economic growth translates into environmental pressure across MENA countries, given their diverse energy systems, levels of regulatory enforcement, and industrial compositions. Applying heterogeneous panel models, threshold regressions, or quantile-based approaches could better capture these nuances and provide a more comprehensive understanding of the income–emissions relationship, thereby helping to overcome limitations of this type.

From a policy perspective, the results stress that economic growth alone cannot guarantee improved environmental outcomes. To avoid locking into fossil fuel-intensive development paths, MENA countries must integrate growth strategies with investments in renewable energy deployment, efficient transport infrastructure, and diversification into low-carbon services. The evidence that both models, despite their differences, point to nonlinear income–emissions dynamics highlights the potential for region-wide strategies that link structural transformation with climate objectives. By embedding clean energy, regulatory quality, and sectoral modernization into development agendas, MENA economies can achieve more sustainable growth trajectories while aligning with global decarbonization goals.

5.3. Granger Causality Test Results

To further investigate the dynamic interactions among the core variables, we employ pairwise Granger causality tests using annual panel data for the period 1990–2019 with one lag specification. This approach allows us to assess whether the past values of one variable contain predictive information about the current values of another, thereby shedding light on the direction of influence within the growth–energy–emissions nexus.

Table 10. Pairwise Granger causality test results (sample: 1990–2019; lags: 1).

Null Hypothesis:	Obs	F-Statistic	Prob.
GDPCON1 does not Granger Cause ECO2_GDP	174	5.38553	0.0215
ECO2_GDP does not Granger Cause GDPCON1		0.22641	0.6348
GDPCON2 does not Granger Cause ECO2_GDP	174	5.65462	0.0185
ECO2_GDP does not Granger Cause GDPCON2		0.41733	0.5191
RENEWABLES does not Granger Cause ECO2_GDP	174	2.59004	0.1094
ECO2_GDP does not Granger Cause RENEWABLES		0.37279	0.5423
TR_EXPO does not Granger Cause ECO2_GDP	174	1.63224	0.2031
ECO2_GDP does not Granger Cause TR_EXPO		0.52930	0.4679
GDPCON2 does not Granger Cause GDPCON1	174	1.51288	0.2204
GDPCON1 does not Granger Cause GDPCON2		1.76762	0.1854
RENEWABLES does not Granger Cause GDPCON1	174	5.77376	0.0173
GDPCON1 does not Granger Cause RENEWABLES		3.59725	0.0596
TR_EXPO does not Granger Cause GDPCON1	174	2.70462	0.1019
GDPCON1 does not Granger Cause TR_EXPO		3.08963	0.0806
RENEWABLES does not Granger Cause GDPCON2	174	5.69126	0.0181
GDPCON2 does not Granger Cause RENEWABLES		3.38093	0.0677
TR_EXPO does not Granger Cause GDPCON2	174	2.77569	0.0975
GDPCON2 does not Granger Cause TR_EXPO		3.29735	0.0711
TR_EXPO does not Granger Cause RENEWABLES	174	0.05041	0.8226
RENEWABLES does not Granger Cause TR_EXPO		0.67076	0.4139

Notes: A probability value (p-value) below 0.05 indicates rejection of the null hypothesis at the 5% significance level, implying evidence of predictive causality in the direction specified.

reports the results, including the number of observations, F-statistics, and probability values, where a p-value below 0.05 indicates rejection of the null hypothesis of no Granger causality [103,112].

The results of the pairwise Granger causality tests offer valuable insights into the dynamic interactions among economic activity, renewable energy consumption, trade openness, and environmental performance across the selected MENA countries. These findings bear directly on the evaluation of the Environmental Kuznets Curve (EKC) hypothesis, which suggests a nonlinear relationship between economic growth and environmental degradation. Notably, the absence of a statistically significant short-run causal relationship from renewable energy consumption to CO₂ emissions indicates that the environmental benefits of renewables may be cumulative and primarily manifest over the long term. This supports the interpretation that renewable energy’s impact on emissions reduction is gradual, reinforcing the long-run findings from cointegration and FMOLS/DOLS estimations [100,101,102,103,104,105,106,107,108,109,110,111,112,113].

The most robust finding is that GDP per capita (both GDPCON1 and GDPCON2) Granger-causes CO₂ emissions per unit of GDP (ECO2_GDP), with statistically significant F-statistics (p-values of 0.0215 and 0.0185, respectively). However, the reverse causality—from ECO2_GDP to GDP per capita—is not supported, suggesting that economic development influences environmental performance, but not vice versa, at least in the short run. This one-way causality is consistent with the Environmental Kuznets Curve (EKC) hypothesis, which argues that emissions initially rise with income but decline after a threshold. This result highlights income as a driver of emissions intensity, confirming that growth precedes environmental improvements rather than the other way around. In contrast, renewable energy is not Granger-cause ECO2_GDP, nor does ECO2_GDP Granger-cause renewables, as both directions are statistically insignificant. This finding reflects the relatively limited role of renewables in the MENA energy mix, which remains heavily dependent on fossil fuels [22,23,24,25,26,27,28,29]. Structural delays, investment bottlenecks, and the mismatch between renewable capacity expansion and its integration into emission-intensive sectors may also weaken the short-run link. By comparison, studies in other emerging regions, such as Asia and Latin America [82,93], often identify stronger renewable–emissions linkages, suggesting that selected MENA countries lag behind in leveraging renewables for decarbonization. The inclusion of transport service exports adds a structural dimension, recognizing that environmental performance depends not only on income and energy composition but also on the nature of economic activities.

Although causality tests did not reveal significant short-run effects, transport services remain strategically important in MENA’s economies, where logistics and trade corridors underpin competitiveness [8,91]. Existing evidence shows that service-oriented economies generally achieve lower emissions intensity compared to manufacturing-led growth [85], yet the impact in the sample MENA countries depends on the carbon intensity of transport modes, as aviation and trucking may offset gains from rail or maritime logistics. Disaggregating these categories in future research would refine the interpretation of their environmental role. Overall, the results validate key elements of the EKC hypothesis while underscoring its limitations when applied in isolation, aligning with broader critiques that emphasize the conditional role of structural and institutional factors [86,87].

For the MENA region studied, this implies that environmental improvement cannot be left to income dynamics alone but requires deliberate policy interventions. Economic growth can be leveraged as a pathway to sustainability only if resources are directed to clean technologies and infrastructure rather than fossil fuel lock-ins. The absence of a strong renewable–emissions link highlights the urgency of scaling up renewable deployment, ensuring grid integration and aligning targets with decarbonization in energy-intensive industries. Similarly, transport infrastructure investments should prioritize low-carbon modes such as rail and maritime corridors, the electrification of freight and passenger systems, and intermodal logistics. Trade and industrial policies need to incorporate environmental standards to prevent competitiveness from reinforcing emissions, while institutional reforms are necessary to strengthen enforcement and policy coherence [78,79,80,81,82,83,84,85,86,87].

Taking together, the evidence suggests that the MENA countries studied will not achieve environmental improvements solely by reaching higher income levels, but by coupling growth with proactive energy and trade transitions. To our knowledge, this is the first study to integrate transport service exports into the EKC framework for MENA, offering new insights into how structural shifts in the service sector interact with environmental efficiency. Moreover, by combining GDP dynamics with renewable energy penetration and trade composition, the study advances the EKC debate beyond income–emissions relationships, providing a more comprehensive lens for analyzing sustainability transitions. This contribution is particularly relevant for policy design, as it identifies diversification into knowledge-intensive, low-carbon services, alongside renewable deployment, as dual engines for decarbonization in emerging economies.

Together, the findings imply that the MENA countries selected in the present work cannot rely solely on economic growth to solve environmental challenges. Instead, they must integrate environmental policies into development planning from the outset, focusing on renewable energy promotion, regulatory reforms, green trade facilitation, and investments in environmental data systems. Such measures are essential to ensure that development paths are both economically productive and ecologically sustainable, aligned with the objectives of climate resilience and long-term emissions reductions.

6. Conclusions-Policy Implications

This study examined the long-run relationship between economic growth and environmental performance in six selected MENA countries—Algeria, Egypt, Lebanon, Mauritius, Morocco, and Oman—over the period 1990–2019, testing the Environmental Kuznets Curve (EKC) hypothesis. Using advanced panel econometric techniques, including cointegration analysis, Fully Modified Ordinary Least Squares (FMOLS), Dynamic Ordinary Least Squares (DOLS), and Granger causality tests, the findings provide robust evidence of nonlinear income–emissions dynamics. Specifically, FMOLS suggests an inverted-U shaped EKC, with a turning point of around USD 8900 per capita (within the observed income range of the sample), indicating that middle-income economies may be approaching the point where growth begins to reduce emissions intensity. In contrast, the DOLS model produces a U-shaped profile, with a turning point at approximately USD 18,400 per capita, which lies outside the income levels of most sample countries, implying that growth at the current stages could still exacerbate emissions.

Based on these empirical findings, the policy implications are multifaceted and critical for the MENA region’s sustainable development agenda. The EKC hypothesis receives partial support: while one model indicates that growth can eventually deliver environmental improvements beyond a certain threshold, the other suggests that emissions may rebound at higher income levels. This contradictory evidence highlights that economic growth alone does not guarantee environmental sustainability, and that the growth–emissions nexus depends heavily on structural and institutional conditions. It also underscores the need for robust model specification and careful interpretation when using EKC evidence to guide policy.

Beyond income dynamics, the analysis reveals additional structural drivers. Renewable energy consumption has a significant long-run negative effect on emissions intensity, although no short-run causality was found, confirming that its environmental benefits are delayed and cumulative [115,116]. Policymakers must therefore adopt a long-term vision when investing in renewables, recognizing that immediate returns may be limited, but sustained deployment is crucial for structural decarbonization. Similarly, transport service exports—a proxy for structural transformation—consistently reduce emissions intensity, suggesting that their integration into global markets and the expansion of low-carbon, service-oriented activities can help mitigate emissions. However, these gains depend on decarbonizing transport systems and improving trade logistics to avoid offsetting effects [115,116,117,118,119].

From a policy perspective, the results make clear that MENA countries cannot rely solely on economic growth to address environmental challenges. Instead, environmental sustainability must be integrated into development planning from the outset. Priority actions include scaling up renewable energy deployment, strengthening regulatory frameworks, promoting service-sector diversification, and investing in low-carbon transport and trade infrastructure. Equally important is the development of environmental data and monitoring systems, which can guide evidence-based policy and track progress toward climate goals. These measures will ensure that development trajectories are not only economically productive but also environmentally sustainable, aligned with global decarbonization pathways and the region’s long-term climate resilience.

At the same time, the current study is not without limitations. The relatively small sample, in terms of both countries and years, restricts generalizability to the broader MENA region. Data constraints prevented the inclusion of institutional quality, energy prices, and clean technology investments, which could refine the analysis of structural drivers. The use of transport service exports as a proxy for structural transformation, while innovative, does not fully capture the complexity of the service sector’s environmental footprint. Moreover, the unbalanced nature of the panel introduces potential estimation bias, although the use of FMOLS and DOLS mitigates this concern.

A further limitation concerns the use of transport service exports as a proxy for structural transformation. While this variable provides a consistent measure across the selected MENA economies, it does not exclusively capture low-carbon or knowledge-intensive activities, since aviation and shipping remain highly carbon-intensive [29]. The negative and significant coefficient observed in our models may therefore reflect broader efficiency gains and structural shifts rather than direct evidence of decarbonization.

Future research should build on this study in several directions. First, the model could be re-estimated using broader service export categories, or by disaggregating between transport, ICT, financial, and professional services, to more accurately assess the environmental implications of service-sector diversification. Second, expanding country coverage and disaggregating emissions by sector (particularly transport and industry) would provide a clearer picture of the sources of carbon intensity. Incorporating institutional and governance indicators [116,117] could also shed light on the policy mechanisms linking growth, structural transformation, and environmental outcomes. Finally, the use of methodological extensions such as dynamic panel GMM to address endogeneity, nonlinear approaches like NARDL to capture asymmetries, and sector-specific renewable energy indices (e.g., solar, wind) would enhance robustness. Together, these refinements would not only advance EKC research but also increase its policy relevance for guiding sustainable, low-carbon transitions in MENA and beyond.

In sum, the findings show that while the EKC hypothesis receives partial support, its realization in MENA countries is contingent on structural reforms and proactive policies. Economic growth can contribute to decarbonization only if it is accompanied by investments in renewable energy, green transport, and institutional strengthening. Without such measures, the region risks locking into a fossil fuel-intensive trajectory, delaying or even reversing the EKC turning point. Coordinated regional strategies that integrate growth with climate objectives are therefore essential to ensure sustainable, inclusive, and resilient development pathways [119,120,121,122]. Empirical contributions in recent years provide corroborating evidence: for instance, the study by Al-Ayouty [4] affirms that renewable energy significantly reduces CO₂ emissions in MENA while validating the EKC in a broader panel context, and the work by Kong et al. [6] finds that government effectiveness and renewable energy accelerate progress toward carbon neutrality in the MENA region under heterogeneous panel techniques.