Are Natural Resources Harmful to the Ecology? Fresh Insights from Middle East and North African Resource-Abundant Countries

Abstract

The Middle East and North African (MENA) region is among the regions most impacted by global warming and climate change. At the same time, the region accounts for 58% of global oil reserves and 43% of global natural gas reserves. It is, therefore, important to assess the role of natural resource abundance in the environmental degradation faced by MENA resource-abundant countries. This study contributes to this research area by exploring the short- and long-term repercussions of natural resources on the ecological footprint (EFP) of eight resource-rich MENA countries between 2000 and 2021. The research performs both aggregate and disaggregate assessments by considering the total resource rents, as well as specific rents of oil, natural gas, and minerals. The pooled mean group estimator indicates that a rise of 1% in total natural resources induces an increase of 0.053% in the EFP, implying that natural resources are harmful to the environment. The disaggregate analysis shows that oil rents have the most adverse environmental effects in the long run, followed by natural gas. Finally, mineral rents are determined to be neutral vis-à-vis the environment. In light of these findings, policy recommendations for reducing the adverse environmental impacts of natural resources are suggested.

1. Introduction

Unprecedented research and public attention have recently focused on the potential consequences of environmental deterioration. The excessive use of natural capital and the rise in greenhouse gas (GHG) emissions are among the most contributing factors to global warming, ocean acidification, desertification, and biodiversity loss [1,2,3,4]. Given the current ecological situation and the increasing significance of natural resources in shaping environmental quality, natural resource management has garnered considerable attention from scholars. Indeed, natural resources play a pivotal role in the economic growth and development of numerous countries. However, while natural resource exploitation can boost economic growth and improve well-being, it can also result in the degradation of environmental conditions. Discussions surrounding the linkage between economic growth and environmental deterioration, i.e., the Environmental Kuznets Curve (EKC) hypothesis, continue to be an ongoing focus in the global discourse on sustainable development. Therefore, many scholars highlighted that resource-based development strategies had reached their maximum capacity, given their detrimental repercussions on the ecology. Furthermore, the use of natural resources has the potential to cause significant harm to the environment. This includes the release of harmful pollutants into the air, the contamination of water and soil, the destruction of landscapes, and an adverse effect on ecosystems and biodiversity. These detrimental effects are a direct result of the extensive energy needed for resource extraction and processing. Another strand in the existing literature has instead confirmed that resource-rich countries can generate increased revenues through the exploitation of natural resources, which may speed up the energy transition process and improve the environment [5,6].

The previous discussion suggests there is an ongoing debate regarding the environmental repercussions of natural resources, with no definitive conclusions reached. This study investigates the impacts of natural resources on environmental degradation, as measured by the ecological footprint (EFP), in Middle East and North African (MENA) countries. By doing so, the study contributes to the current literature in different ways. First, despite the increasing number of studies on the subject, little is known about the environmental implications of natural resources in MENA countries. The selection of MENA countries is based on a number of considerations. On the one hand, MENA countries are well-endowed in natural resources, including oil and gas. According to ref. [7], the MENA region accounts for about 58% of global oil reserves and 43% of global natural gas reserves. In addition, statistics from ref. [8] suggest that, in 2021, total natural resource rents represented 61% of the gross domestic product (GDP) in Libya, 43.4% in Iraq, 27.2% in Qatar, and 25.5% in Saudi Arabia. This situation made the MENA region a significant global player in the provision of natural resources, which has the potential to impact international economic prosperity [9]. However, MENA resource-abundant countries have become increasingly reliant on natural resources and sensitive to global energy price fluctuations. Indeed, natural resources are extremely important in most MENA resource-rich countries, as their economies rely heavily on natural resource revenues. On the other hand, the MENA region is among the most polluted regions worldwide and is highly impacted by climate change [10]. Ref. [11] stated that several cities in the MENA region exhibit very high levels of air pollution, placing them among the most polluted worldwide. At the same time, MENA countries have higher rates of marine plastic pollution per capita. Considering simultaneously the natural resource abundance in MENA countries and the deterioration of ecological indicators, it is interesting to examine empirically whether natural resource exploitation has contributed to the current environmental degradation. Figure 1 depicts the EFP per capita in selected resource-abundant MENA countries and the world. As shown, the EFP in most MENA countries is higher than the world average (2.58 gha per person). Specifically, Qatar and the United Arab Emirates exhibited the highest EFP in 2022, with 13.13 and 8.71 gha per person, respectively.

The present study provides a second contribution by undertaking comprehensive examinations of both the total and specific environmental consequences of natural resources. Indeed, although the aggregate analysis provides a better understanding of the impacts of natural resources on ecology, it falls short in identifying the specific resource types that have the most significant implications on environmental conditions. This is important for MENA countries, which are endowed with various natural resources. Therefore, evaluating the ecological repercussions of each natural resource (oil, natural gas, and minerals) may provide more insightful findings, leading to the formulation of specific policy recommendations.

The rest of this paper is organized as follows. Section 2 synthesizes previous studies and the main gaps in the literature, while Section 3 describes the econometric approach and data. Section 4 discusses the empirical findings, and, finally, the concluding remarks and policy implications are provided in Section 5.

2. Literature Review

2.1. The Previous Literature

The environmental consequences of natural resources have recently attracted the attention of scholars [12,13,14,15,16,17]. The existing studies have employed various methodologies and econometric techniques. Theoretically speaking, natural resource abundance may positively or negatively affect the environment. On the one hand, the majority of studies have demonstrated the adverse implications of natural resources on the environment [18,19,20]. Indeed, exploring and extracting natural resources requires many activities that put pressure on the environment and affect the ecology. Instead, another strand in the literature suggests that natural resources may benefit the environment. Ref. [21] argued that natural resources deter fossil fuels and promote the transition towards low-polluting energy sources. Therefore, natural resource abundance can deteriorate or enhance environmental conditions [22].

Empirical studies on the environmental implications of natural resources have reached mixed findings. The results remain inconclusive as there is no agreement on the nature of these impacts. For some authors, the effect is negative; for others, natural resources positively affect environmental quality. For example, ref. [18] indicated that natural resources increased CO₂ emissions and environmental deterioration in 16 European countries. The study explores the long-term causal linkages between renewable and nonrenewable energy use, natural resources, and income. The PMG-ARDL model indicates a positive correlation between resource rents and long-term emissions. This confirms that the use of resources harms ecological sustainability. Additionally, ref. [23] analyzed the effects of natural resources on EFP in Pakistan from 1970 to 2014. The investigation showed that resource abundance raised the EFP and deteriorated environmental conditions. Ref. [24] confirmed these results using data from 1971 to 2017 in Pakistan to check the implications of natural resources on environmental quality. Three environmental indicators are used: CO₂ emissions, carbon footprint, and EFP. The dynamic ARDL technique suggested that resource rents increased CO₂ emissions, carbon footprint, and EFP in the long run. More specifically, a 1% rise in natural resource rents leads to a long-term increase of 0.206% in CO₂ emissions, a 0.788% increase in the EFP, and a 0.021% increase in the carbon footprint. Ref. [25] also investigated the implications of natural resource depletion on CO₂ emissions and the EFP in Pakistan using the dynamic ARDL technique. In contrast to the findings of ref. [24], the results demonstrate that the utilization of natural resources is associated with a reduction in emissions and an improvement in environmental quality in the long run. Furthermore, ref. [26] explored the repercussions of several factors, including natural resources, on the environmental deterioration in China, as measured by EFP. The outcomes indicate that resource rents have positive implications on the EFP only in the long run. Ref. [27] also examined the contribution of natural resources to environmental degradation in China using the Quantile ARDL developed by ref. [28]. The findings confirm a positive connection between natural resources and emissions, particularly in higher quantiles of the distribution. Finally, ref. [29] examined the contribution of resource rents to CO₂ emissions in a sample of SSA nations from 1994 to 2020 based on the GMM estimator. The authors concluded that resource rents raise CO₂ levels and lead to environmental damage. These findings have been previously reached by ref. [30], who concluded that natural resource abundance increases the EFP and deteriorates the environmental quality in SSA countries.

Another viewpoint in the literature has demonstrated that natural resources can enhance environmental conditions. Ref. [21] conducted research to determine the repercussions of resources on CO₂ emissions in a selection of European nations. The analysis revealed a negative association, suggesting that countries endowed with extensive natural resources may witness a decline in CO₂ emissions. In a similar vein, ref. [31] explored the implications of natural resource rents on the EFP in BRICS nations from 1992 to 2016. The results show that resource rent diminishes the EFP and improves environmental quality. The case of BRICS countries has also been analyzed by ref. [32] between 1995 and 2018. Total resource rents are found to reduce CO₂ emissions, whereas forest and oil rents raise emissions. Ref. [33] concentrated on the repercussions of resource rents on the EFP in the US between 1970 and 2015. The outcomes suggested that natural resources, among others, reduce the EFP. The US has also been investigated by ref. [12], who analyzed the implications of natural resources on CO₂ emissions and the EFP from 1971 to 2016. Using different econometric techniques, the authors confirmed the findings of ref. [33], as natural resources reduce both emissions and the EFP. Recently, ref. [34] investigated the contribution of natural resources to CO₂ emissions in the US using the dynamic ARDL model. The outcomes are different from previous studies and suggest a positive linkage between natural resources and emissions. Ref. [5] checked the effects of fiscal decentralization and natural resource rents on CO₂ emissions in a group of Organisation for OECD nations. The researchers concluded that the presence of resource rents and fiscal decentralization leads to the long-term improvement in environmental conditions through the reduction in CO₂ emissions. Recently, ref. [35] explored the implications of resource rents on the EFP pressure index in ASEAN-5 over the period of 1990 to 2018. Resource rents increase the environmental quality and decrease the EFP for all conditional quantiles. These outcomes align with those of ref. [36], who concluded that natural resources increase CO₂ emissions in ASEAN countries. Finally, ref. [16] analyzed the impacts of natural resources on the load capacity factor (LCF) in 17 Asian-Pacific Economic Cooperation nations between 1990 and 2019. The findings show that natural resources increase the LCF and improve environmental quality in the long term.

2.2. Empirical Studies on MENA Countries

There has been an increasing amount of research dedicated to understanding the consequences of resources on the environment in the MENA region. However, most of the focus is on the Gulf Cooperation Council (GCC). For example, ref. [22] explored the consequences of natural resources on CO₂ emissions in GCC nations from 1990 to 2018 and confirmed that resource rents lower emissions. In addition, ref. [37] checked the implications of total resource rents on the EFP in GCC nations from 1995 to 2017. The analysis shows that natural resources reduce the EFP and promote environmental sustainability. More specifically, a 1% increase in natural resource rents induces an EFP fall by 0.076%. The case of GCC countries has also been analyzed by ref. [38], who focused on the effects of resource rents on CO₂ emissions using the MG, AMG, and CCEMG estimators. Unlike ref. [37], both estimators suggest that natural resource rents increase CO₂ emissions and deteriorate the environment in GCC countries. Finally, ref. [20] focused on the repercussions of resource rents on the EFP in Saudi Arabia. The authors revealed that total resource rents increase the EFP in the long run. Upon disaggregating natural resources, the authors concluded that environmental deterioration is only caused by oil and gas rents. The case of MENA nations has also been investigated by ref. [39], who assessed the spatial effects of oil and natural gas rents on CO₂ emissions in 17 MENA countries. The authors confirmed the presence of spatial effects, as a 1% increase in oil rents reduces CO₂ emissions by 0.212% in local economies while it increases emissions by 0.533% in neighboring economies. Furthermore, ref. [40] analyzed the implications of oil, mineral, and forest rents on CO₂ emissions in the MENA region. The study suggests that oil and mineral rents deteriorate the environment while forest rents improve it. Finally, ref. [41] assessed the impacts of natural resources on the EFP in selected MENA countries for the period of 1994–2018. The outcomes confirm that natural resource rent has a positive correlation with the EFP.

2.3. Research Gaps

The preceding debate suggests that the empirical body of work on the ecological consequences of natural resources in resource-rich MENA nations has yielded diverse results. In addition, most studies focusing on the environmental impacts of resources in the MENA region have mainly concentrated on GCC countries and ignored some resource-abundant North African countries, such as Algeria and Libya. In addition, most studies conducted an aggregate analysis considering total natural resource rents. Exceptions are ref. [20,39,40], which considered different types of natural resources. Finally, studies on the MENA region used CO₂ emissions as an environmental indicator [39,40]. However, CO₂ emissions are a metric of air quality and do not reflect overall environmental conditions because they do not consider other environmental factors such as land-use transformation, soil degradation, and water contamination.

3. Data and Methodology

3.1. Data

This study analyzes the impacts of aggregate and disaggregated natural resources on the EFP of eight resource-abundant MENA countries (Algeria, Iraq, Kuwait, Libya, Morocco, Qatar, Saudi Arabia, and the United Arab Emirates) from 2000 to 2021. The equation to be estimated is derived from a modified version of the Stochastic Impacts by Regression on Population, Affluence, and Technology (STIRPAT) model developed by ref. [42], which can be formulated as follows:

$$LNEF{P}_{i,t}=\alpha +{\beta }_{1}LNNR{R}_{i,t}+{\beta }_{2}LNGD{P}_{i,t}+{\beta }_{3}LNGD{P^2}_{i,t}+{\beta }_{4}LNTE{C}_{i,t}+{\beta }_{5}LNUR{B}_{i,t}+{\epsilon }_{i,t}$$

(1)

where $LNEFP$ denotes the natural logarithm of the ecological footprint, and $LNNRR$ represents the natural logarithm of resource rents. $LNGDP$, ${LNGDP}^2,LNTEC,$ and $LNURB$ denote the natural logarithm of GDP per capita, GDP per capita squared, total energy use, and urbanization rate, respectively. ${\beta }_1$, ${\beta }_2$, ${\beta }_3$, ${\beta }_4$, and ${\beta }_5$ are coefficients to be estimated, while $i$ and $t$ represent the countries and time, respectively. Finally, ${\epsilon }_{i,t}$ is the error term. The environmental implications of total natural resources, oil, natural gas, and mineral rents are separately estimated in the study. Therefore, four different models are estimated:

Model 1

$$\mathit{L}\mathit{N}\mathit{E}\mathit{F}{\mathit{P}}_{\mathit{i},\mathit{t}}={\mathit{\alpha }}_1+{\mathit{\beta }}_{11}\mathit{L}\mathit{N}\mathit{T}{\mathit{R}\mathit{E}\mathit{N}\mathit{T}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{12}\mathit{L}\mathit{N}\mathit{G}\mathit{D}{\mathit{P}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{13}\mathit{L}\mathit{N}\mathit{G}\mathit{D}{{\mathit{P}}^2}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{14}\mathit{L}\mathit{N}\mathit{T}\mathit{E}{\mathit{C}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{15}\mathit{L}\mathit{N}\mathit{U}\mathit{R}{\mathit{B}}_{\mathit{i},\mathit{t}}+{\mathit{\vartheta }}_{\mathit{i},\mathit{t}}$$

(2)

Model 2

$$\mathit{L}\mathit{N}\mathit{E}\mathit{F}{\mathit{P}}_{\mathit{i},\mathit{t}}={\mathit{\alpha }}_2+{\mathit{\beta }}_{21}\mathit{L}\mathit{N}\mathit{O}{\mathit{R}\mathit{E}\mathit{N}\mathit{T}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{22}\mathit{L}\mathit{N}\mathit{G}\mathit{D}{\mathit{P}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{23}\mathit{L}\mathit{N}\mathit{G}\mathit{D}{{\mathit{P}}^2}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{24}\mathit{L}\mathit{N}\mathit{T}\mathit{E}{\mathit{C}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{25}\mathit{L}\mathit{N}\mathit{U}\mathit{R}{\mathit{B}}_{\mathit{i},\mathit{t}}+{\mathit{\theta }}_{\mathit{i},\mathit{t}}$$

(3)

Model 3

$$\mathit{L}\mathit{N}\mathit{E}\mathit{F}{\mathit{P}}_{\mathit{i},\mathit{t}}={\mathit{\alpha }}_3+{\mathit{\beta }}_{31}\mathit{L}\mathit{N}\mathit{G}{\mathit{R}\mathit{E}\mathit{N}\mathit{T}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{32}\mathit{L}\mathit{N}\mathit{G}\mathit{D}{\mathit{P}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{33}\mathit{L}\mathit{N}\mathit{G}\mathit{D}{{\mathit{P}}^2}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{34}\mathit{L}\mathit{N}\mathit{T}\mathit{E}{\mathit{C}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{35}\mathit{L}\mathit{N}\mathit{U}\mathit{R}{\mathit{B}}_{\mathit{i},\mathit{t}}+{\mathit{\delta }}_{\mathit{i},\mathit{t}}$$

(4)

Model 4

$$\mathit{L}\mathit{N}\mathit{E}\mathit{F}{\mathit{P}}_{\mathit{i},\mathit{t}}={\mathit{\alpha }}_4+{\mathit{\beta }}_{41}\mathit{L}\mathit{N}\mathit{M}{\mathit{R}\mathit{E}\mathit{N}\mathit{T}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{42}\mathit{L}\mathit{N}\mathit{G}\mathit{D}{\mathit{P}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{43}\mathit{L}\mathit{N}\mathit{G}\mathit{D}{{\mathit{P}}^2}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{44}\mathit{L}\mathit{N}\mathit{T}\mathit{E}{\mathit{C}}_{\mathit{i},\mathit{t}}+{\mathit{\beta }}_{45}\mathit{L}\mathit{N}\mathit{U}\mathit{R}{\mathit{B}}_{\mathit{i},\mathit{t}}+{\mathit{\phi }}_{\mathit{i},\mathit{t}}$$

(5)

As previously mentioned, the dependent variable in models 1–4 is represented by the ecological footprint. Indeed, recent studies have confirmed the superiority of the EFP compared to conventional environmental measures, including CO₂ emissions refs. [43,44]. The variable of interest, namely, natural resources, is quantified in this study using both aggregate and disaggregate measures. As an aggregate measure, we used total natural resource rents as a share of GDP. Then, we employed oil, natural gas, and mineral resource rents as a share of GDP. Finally, the control variables derived from the STIRPAT model are GDP per capita, urbanization, and total energy consumption.

Table 1. Data description.

Variable	Acronym	Definition	Source
Ecological footprint	EFP	Ecological footprint (gha)	GFN
Total resource rent	TRENT	Total natural resources rents (% of GDP)	WDI
Oil rents	ORENT	Oil rents (% of GDP)	WDI
Natural gas rents	NRENT	Natural gas rents (% of GDP)	WDI
Mineral rents	MRENT	Mineral rents (% of GDP)	WDI
Income	GDP	Real GDP per capita (constant 2015 US$)	WDI
Urbanization	URB	Urban population	WDI
Energy consumption	TEC	Total energy consumption (MTOE)	EIA

EIA: U.S. Energy Information Administration; GFN: Global Footprint Network; WDI: World Development Indicators.

provides a presentation of the variables used in the empirical investigation.

Table 2. Descriptive statistics.

Variable	Mean	Median	Maximum	Minimum
EFP	6.03 × 107	4.82 × 107	2.24 × 108	6.36 × 106
TEC	63.170	38	309	10
GDP	2.26 × 104	1.46 × 104	7.35 × 104	1.95 × 103
URB	1.34 × 107	1.17 × 107	3.28 × 107	6.22 × 105
TRENT	31.481	31.838	66.059	0.194
ORENT	29.165	28.211	65.157	0.0008
GRENT	1.966	1.095	12.011	0.002
MRENT	0.300	0.000	7.204	0.000

reports some descriptive statistics of the different variables. The findings indicate that the average natural resource rents account for 31.481% of GDP. At the same time, Libya experienced the highest total resource rents of 66.059% in 2008. The table also suggests the dominance of oil as the primary natural resource that provides rent to MENA countries. Oil rents represent about 29.165% of GDP in the selected countries.

On the other hand, average natural gas rents are about 2% of GDP, with a maximum of 12% in Qatar in 2021. Finally, mineral rents are relatively weak, representing only 0.3% of GDP. The maximum mineral rents were recorded in Morocco in 2011. It is worth noting that Morocco ranked 11th and 17th in cobalt and silver production, respectively [45]. Thus, it can be argued that MENA countries exhibit a certain degree of diversity in the composition of natural resources, with a significant predominance of oil as the primary resource.

3.2. Empirical Methodology

The empirical study consists of many steps. First, a preliminary analysis is conducted to assess the existence of slope homogeneity (SH) and cross-section dependence (CSD) between countries. Second, we test the stationarity based on the cross-sectionally augmented IPS (CIPS) unit root test. The use of the CIPS panel unit root tests is mandatory in the presence of CSD. Then, we check on long-run relationships between the different variables using the demeaned Pedroni cointegration tests proposed by ref. [46,47] and the ECM-based panel cointegration test developed by ref. [48]. Finally, we employ three distinct methodologies to estimate the short-run and long-term impacts of resource rents on EFP: dynamic fixed effects (DFE), pooled mean group (PMG), and mean group (MG) estimators. The Hausman test is employed to identify the appropriate technique. Figure 2 summarizes the flowchart of the empirical investigation.

4. Empirical Results

4.1. Slope Homogeneity Test Results

We employ the SH test for the panel data of ref. [49] to examine the homogeneous behavior between countries and the selection of appropriate estimation procedures.

Table 3. SH test findings.

	Statistic	p-Value
Model 1. LNTRENT	3.130 ***	0.000
Model 2. LNORENT	2.400 **	0.010
Model 3. LNGRENT	2.980 ***	0.000
Model 4. LNMRENT	2.600 ***	0.000

*** and ** denote the rejection of the null hypothesis at the 1% and 5% level, respectively.

reports the homogeneity test results for the four models previously discussed. As shown, all test statistics are statistically significant, at least at 5%. Consequently, it can be inferred that there is heterogeneity among the countries being examined.

4.2. CSD Analysis

Next, we need to check the CSD for all variables. Indeed, ignoring CSD could result in the implementation of inadequate methodologies and inaccurate statistical findings. Indeed, the presence of CSD renders the standard techniques, such as ordinary least squares (OLS), inaccurate.

The interdependence between cross-sections may arise due to various factors, such as geographical connections and linkages in the social and economic spheres. This research implements the CD test of ref. [50] and the bias-corrected scaled LM test of ref. [51]. The outcomes are reported in

Table 4. CD test results.

	CD Test		Bias-Corrected Scaled LM Test
	Statistic	p-Value	Statistic	p-Value
LNEFP	21.575 ***	0.000	471.432 ***	0.000
LNTEC	15.12 ***	0.000	419.974 ***	0.000
LNGDP	0.952	0.340	214.005 ***	0.000
LNURB	23.713 ***	0.000	563.244 ***	0.000
LNTRENT	15.236 ***	0.000	294.787 ***	0.000
LNORENT	19.807 ***	0.000	401.624 ***	0.000
LNGRENT	13.569 ***	0.000	215.844 ***	0.000
LNMRENT	NA	NA	NA	NA

*** denotes the rejection of the null hypothesis at the 1% level.

. The CD test suggests rejecting the null hypothesis of no CSD for all series except GDP per capita. Therefore, one could confirm the presence of mutual linkages between the MENA countries under study and the necessity of using techniques that account for CSD in what follows.

4.3. Stationarity Analysis

When data exhibiting CSD are present, second-generation unit root tests should be implemented. The CIPS unit root test is performed for the series at the level and first difference. The findings are reported in

Table 5. Unit root test results.

Variables	Level		1st Difference
	C	C + T	C	C + T
LNEFP	−2.18	−2.54	−4.16 ***	−4.28 ***
LNTEC	−2.30 *	−2.25	−4.77 ***	−5.23 ***
LNGDP	−1.96	−2.12	−4.10 ***	−4.56 ***
LNGDP2	−1.97	−2.14	−4.12 ***	−4.56 ***
LNURB	−3.08 ***	−2.38	−4.12 ***	−4.61 ***
LNTRENT	−2.07	−2.10	−3.21 ***	−3.66 ***
LNORENT	−1.93	−2.11	−4.96 ***	−4.95 ***
LNGRENT	−1.73	−2.14	−3.30 ***	−4.10 ***
LNMRENT	0.59	−0.38	−4.48 ***	−3.96 ***

C and T denote the constant and time trend, respectively. The critical values are −2.21 (10%), −2.33 (5%), and −2.57 (10%) for constant specification, and −2.73 (10%), −2.86 (5%), and −3.10 (1%) for constant and trend specification. *** and * denote the statistical significance at 1% and 10%, respectively.

.

The findings suggest that there are no I(0) variables, except for urbanization and total energy consumption, which demonstrate weak stationarity, particularly for the specification with a constant. Nevertheless, all variables exhibit stationarity when differenced, regardless of whether the model includes a constant or both a constant and a trend. Therefore, one could argue that all series are I(1). Based on the unit root test findings, it is important to assess the cointegration between the variables and estimate the ecological effects of resource rents in the short and long term.

4.4. Cointegration Analysis

Given the unit root test results, a cointegration test can be employed to identify whether there are long-run linkages between the variables. In this research, two panel cointegration techniques are employed. First, we implement the panel cointegration test suggested by ref. [46,47]. Although this test is a first-generation cointegration test and is mainly used for cross-sectionally independent panels, it allows accounting for CSD by demeaning the data before implementing it. Furthermore, we employ the panel cointegration of ref. [48], which is an ECM-based panel data cointegration test. This test has the advantage of considering both heterogeneity and CSD. The outcomes of the different models are shown in

Table 6. Cointegration test results.

	Model 1 Total Resource Rents		Model 2 Oil Rents		Model 3 Natural Gas Rents		Model 4 Mineral Rents
	Statistic	p-Value	Statistic	p-Value	Statistic	p-Value	Statistic	p-Value
Demeaned Pedroni panel cointegration test
Modified Phillips–Perron t	1.63 **	0.050	1.74 **	0.040	1.68 **	0.040	1.89 **	0.020
Phillips–Perron t	−2.65 ***	0.000	−2.25 **	0.010	−2.73 ***	0.000	−2.22 **	0.010
Augmented Dickey–Fuller t	−2.17 **	0.010	−1.72 **	0.010	−1.94 **	0.020	−1.56 *	0.060
Westerlund ECM-based panel cointegration test
Variance ratio	−3.19 ***	0.000	−4.13 ***	0.000	−4.26 ***	0.000	−4.25 ***	0.000

Notes: ***, **, and * denote the rejection of the null hypothesis at 1%, 5%, and 10% level, respectively.

. The different statistics of the demeaned Pedroni cointegration test suggest rejecting the null hypothesis for all models at different statistical levels. This finding is supported by the Westerlund cointegration test, which confirms the presence of long-run linkages for all models. Therefore, one could confirm a long-run cointegrating linkage between the different variables. This conclusion is valid for all considered models.

4.5. Short- and Long-Run Effects

4.5.1. Total Resource Rents and EFP

Table 7. Total resource rents and EFP.

Variables	PMG		MG		DFE
	Coeff.	p-Value	Coeff.	p-Value	Coeff.	p-Value
Long-run estimation
LNGDP	6.851 ***	(1.425)	−2.154	(19.86)	−3.765	(3.899)
LNGDP2	−0.129 ***	(0.028)	0.056	(0.391)	0.075	(0.078)
LNURB	0.269 ***	(0.052)	0.570 **	(0.268)	0.409 *	(0.219)
LNTRENT	0.053 ***	(0.019)	0.0008	(0.063)	0.019	(0.051)
LNTEC	0.414 ***	(0.057)	0.462 ***	(0.141)	0.405	(0.310)
Constant	−40.10 ***	(8.386)	−72.160	(158.0)	13.210	(11.22)
Short-run estimation
ECT	−0.506 ***	(0.105)	−0.953 ***	(0.097)	−0.233 ***	(0.060)
D.LNGDP	−5.279	(21.03)	−5.412	(21.56)	1.586	(3.038)
D.LNGDP2	0.092	(0.409)	0.094	(0.419)	−0.027	(0.061)
D.LNURB	−0.098	(1.557)	−0.931	(3.988)	0.761 ***	(0.192)
D.LNTRENT	0.001	(0.022)	−0.016	(0.030)	0.034 *	(0.017)
D.LNTEC	0.069	(0.089)	−0.052	(0.079)	0.129	(0.082)
Hausman test
Ho: PMG is more efficient than DFE/H1: DFE is more efficient than PMG					1.60 (0.750)
Ho: PMG is more efficient than MG/H1: MG is more efficient than PMG					1.65 (0.770)

The symbols ***, **, and * indicate the statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors are reported between parentheses. D is the first difference operator. ECT is the error correction term. For the Hausman test, p-values are reported between parentheses.

reports the estimation outcomes of the impact of total resource rents on the EFP using the PMG, MG, and DFE. Following much previous research, including ref. [52,53], we apply the Hausman test to ascertain whether a difference exists among the estimators. The PMG estimator is, therefore, chosen if the null is not rejected since it is efficient. Based on the findings at the bottom of Table 7, we cannot reject the null hypothesis that the PMG estimator is the more efficient one. Hence, the Hausman test results indicate that the PMG estimator is preferred over the DFE and MG estimators. Consequently, our analysis of the effects of resources on the EFP is based on the PMG estimator findings. To begin with, the ECT is negative and statistically significant when using the three estimators. This finding confirms those in Table 6 and suggests the existence of a long-run equilibrium association between the different variables with an adjustment speed of −0.502. In other words, about 50% of the disequilibria from the long-run connection are corrected during the current year. In terms of long-term consequences, the PMG estimator reveals a positive correlation between the GDP per capita and EFP, but the effect of the squared GDP per capita is negative. Consequently, this result validates the EKC and confirms that resource-rich MENA nations exhibit an inverse U-shaped relationship between the GDP per capita and environmental deterioration. In addition, urbanization and total energy consumption have positive long-run coefficients, which implies that they increase the EFP and deteriorate environmental quality.

A positive and statistically significant coefficient is also generated by the PMG estimator for total resource rents. Indeed, the EFP increases by 0.053% for every 1% increase in total natural resource revenues in the long run. These outcomes corroborate ref. [38] for GCC countries and ref. [20] for Saudi Arabia. Unlike the long-run effects, Table 7 suggests that all short-run coefficients, including the control variable and total resource rents, are not statistically significant. Consequently, the adverse effects of income, urbanization, energy use, and resource rents are only detected in the long term.

4.5.2. Disaggregate Resource Rents and EFP

We now assess the environmental impacts of oil, natural gas, and mineral rents, using the PMG, MG, and DFE estimators. The findings associated with the three models are summarized in

Table 8. Disaggregate resource rents and EFP.

Variables	Model 2. Oil Rents			Model 3. Natural Gas Rents			Model 4. Mineral Rents
	PMG	MG	DFE	PMG	MG	DFE	PMG	MG	DFE
Long-run estimation
LNGDP	7.121 ***	3.659	−3.520	6.406 ***	−22.47	−5.470	7.000 ***	−7.712	−3.655
	(1.437)	(17.840)	(3.671)	(1.281)	(29.03)	(3.836)	(1.439)	(19.810)	(3.847)
LNGDP2	−0.136 ***	−0.055	0.070	−0.122 ***	0.462	0.109	−0.132 ***	0.168	0.0726
	(0.028)	(0.353)	(0.073)	(0.0254)	(0.577)	(0.077)	(0.028)	(0.389)	(0.077)
LNURB	0.305 ***	0.405 *	0.437 **	0.339 ***	0.301	0.380 *	0.270 ***	0.396	0.419 *
	(0.050)	(0.237)	(0.207)	(0.044)	(0.441)	(0.208)	(0.052)	(0.381)	(0.215)
LNTEC	0.466 ***	0.588 ***	0.371	0.419 ***	0.536 **	0.355	0.428 ***	0.446 ***	0.423
	(0.053)	(0.188)	(0.300)	(0.042)	(0.238)	(0.297)	(0.051)	(0.146)	(0.307)
LNRENT	0.117 ***	−0.352	0.059	0.040***	0.172 *	0.280 **	−0.004	0.005	0.042
	(0.042)	(0.566)	(0.093)	(0.013)	(0.0969)	(0.114)	(0.026)	(0.061)	(0.072)
Constant	−43.630 ***	−112.400	12.990	−38.520***	−16.28	17.840 *	−42.27 ***	7.788	12.710
	(9.119)	(178.200)	(11.240)	(9.079)	(187.100)	(10.660)	(9.031)	(154.100)	(11.070)
Short-run estimation
ECT	−0.532 ***	−0.972 ***	−0.244 ***	−0.523***	−0.960 ***	−0.226 ***	−0.524 ***	−0.914 ***	−0.230 ***
	(0.111)	(0.112)	(0.059)	(0.123)	(0.147)	(0.056)	(0.112)	(0.082)	(0.059)
D.LNGDP	1.485	−6.783	1.666	−2.731	−13.42	−0.469	−12.80	0.057	1.724
	(21.940)	(24.020)	(3.062)	(17.010)	(25.25)	(2.925)	(19.390)	(22.420)	(3.050)
D.LNGDP2	−0.041	0.113	−0.028	0.042	0.246	0.014	0.241	−0.012	−0.0305
	(0.429)	(0.470)	(0.061)	(0.328)	(0.491)	(0.058)	(0.377)	(0.435)	(0.061)
D.LNURB	0.339	1.425	0.747 ***	3.205	−2.406	0.878 ***	1.636 **	1.467	0.748 ***
	(2.081)	(4.450)	(0.194)	(2.301)	(1.869)	(0.185)	(0.814)	(2.232)	(0.195)
D.LNTEC	0.051	−0.100	0.142 *	−0.006	−0.171 *	0.147 *	0.0572	−0.034	0.121
	(0.092)	(0.077)	(0.083)	(0.099)	(0.104)	(0.077)	(0.094)	(0.077)	(0.082)
D.LNRENT	−0.180	−0.052	0.024	−0.009	0.0673	0.038	0.008	−0.014	0.0322
	(0.201)	(0.365)	(0.041)	(0.052)	(0.0438)	(0.029)	(0.021)	(0.030)	(0.020)
Hausman test
Ho: PMG is more efficient than DFE/H1: DFE is more efficient than PMG			1.44 (0.830)			0.88 (0.980)			3.29 (0.650)
Ho: PMG is more efficient than MG/H1: MG is more efficient than PMG			1.51 (0.790)			0.95 (0.910)			3.30 (0.660)

The symbols ***, **, and * indicate the statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors are between parentheses. D is the first difference operator. ECT is the error correction term. For the Hausman test, p-values are reported between parentheses. LNRENT refers to natural resource rents (oil, natural gas, or minerals).

. First, the Hausman test presented at the bottom of the table reveals that the PMG estimator is better than the DFE and MG estimators in both models. Consequently, our analysis will focus on the PMG estimator results. In the long run, GDP per capita has a positive implication on the EFP for the three models. In addition, the coefficient of the squared GDP per capita is negative and significant at 1%. Consequently, the EKC hypothesis is validated in resource-abundant MENA countries. This conclusion has been previously reached when conducting the aggregate analysis reported in Table 7. This could confirm the inverted U-shaped relationship between income and environmental deterioration and validate the EKC hypothesis in resource-abundant MENA countries. Urbanization also has a positive environmental repercussion in the long run, suggesting that the rapid urbanization observed in MENA countries in recent decades has been associated with increased GHG emissions, land-use changes, soil degradation, water contamination, and other related ecological factors. Total energy use also has detrimental repercussions on the EFP in the long run, regardless of the model considered. These findings are expected as MENA countries rely heavily on fossil-fuel energy sources, particularly oil and natural gas. These energy sources are acknowledged as the principal causes of GHG emissions and environmental deterioration.

Data from ref. [54] indicates that, in resource-rich MENA countries, the proportion of the final energy consumption attributable to renewable sources is exceedingly low. In 2020, it was 0.06% in Saudi Arabia and Qatar, 0.1% in Kuwait and Algeria, 1% in Iraq and the United Arab Emirates, and 3.1% in Libya. Table 8 provides information on the effects of oil, gas, and mineral rents on the EFP. The PMG estimator demonstrates that there are long-term positive and statistically significant coefficients for oil and natural gas rents. On the contrary, mineral rents have no coefficient. More specifically, the negative consequences of oil rents outweigh those of natural gas. Indeed, a 1% increase in oil rents causes the EFP to increase by 0.117%, whereas the same increase in natural gas rents raises the EFP by only 0.040% in the long run. These outcomes are expected given that resource-abundant MENA countries are predominantly endowed with oil. According to ref. [7], the OPEC Member MENA nations have 840 billion barrels of proven crude oil reserves, which represent 58% of total global oil reserves.

At the same time, the natural gas reserve is about 80 trillion cubic meters and represents 43% of the global reserve. Furthermore, the descriptive analysis provided in Table 2 indicates that oil rents are of greater importance than other natural resources in MENA nations. Indeed, the average oil rents account for 29.165% of GDP, while the average natural gas rents represent about 2% of GDP. These disparities in oil and gas endowments and exploitation in MENA nations may explain why oil rents have been found to have more detrimental effects on the EFP than natural gas. ECTs are negative and significant at 1% in all models. Additionally, the adjustment speed is moderate, as it ranges between 52% and 53% for the different models. These findings align with Table 6 and suggest the existence of long-run linkages between the EFP and resource rents in the different models. On the contrary, no substantial effects have been observed in the short term. These outcomes corroborate the aggregate analysis and indicate that resource rents only have adverse long-term environmental effects in resource-abundant MENA countries.

4.6. Discussion of Results

The objective of this paper is to explore the effects of aggregate and disaggregated natural resource rents on the ecological footprint in selected MENA countries. The empirical investigation yields several findings. First, the analysis shows that natural resource rents have detrimental effects on the EFP in the long term. The findings are consistent with those of numerous preceding studies, including ref. [55] Saudi Arabia, ref. [38] for GCC countries, ref. [34] for the United States, ref. [36] for a sample of ASEAN countries, ref. [29,30] for SSA countries, and refs. [41] for MENA countries. Indeed, resource-abundant countries are still dependent on natural resource rents as the main source of revenues for the government budget [56]. The importance of natural resources, particularly oil and natural gas, for all economic sectors and the weak energy transition process in most countries have resulted in an increased demand for fossil-fuel energy sources. This could represent an easy way for MENA resource-abundant countries to finance public investment, increase per capita income, and improve well-being. However, this could lead to a rise in GHG emissions and the degradation of environmental indicators. Second, the empirical investigation suggests that various types of natural resources affect the environment differently. Indeed, oil rents have the most detrimental effects in the long run, followed by natural gas. On the contrary, mineral rents have no significant effects on environmental quality. These findings corroborate those of ref. [20] for Saudi Arabia. The authors confirmed that environmental quality is only worsened by oil and natural gas rents. Ref. [57] also concluded that mineral resource rents have no significant effects on CO2 emissions in Saudi Arabia. However, ref. [40] confirmed that oil and mineral rents deteriorate the environment in MENA countries. Oil and natural gas extraction activities are known to be harmful to the environment. These resources are the main emitters of methane gas. According to ref. [58], about 40% of methane emissions are sourced from oil extraction, while the remaining 60% comes from leaks along the natural gas value chain. Additionally, methane accounts for about 70–90% of natural gas [59]. Furthermore, ref. [60] highlighted that methane is approximately ten times more potent than CO₂ as a GHG in terms of its capacity to warm the planet. In such a situation, the promotion of energy transition and the use of sustainable and clean energy sources may be helpful in mitigating the long-run adverse effects of the natural resource abundance in MENA countries.

5. Conclusions and Policy Recommendations

A discussion is taking place over the implications of natural resources on the environment in resource-abundant countries. MENA countries have an abundance of natural resources; yet, they also experience rapid environmental degradation. This study examines the contribution of total and disaggregated natural resources, including oil, gas, and minerals, on the EFP of eight resource-abundant MENA countries from 2000 to 2021. To do so, the empirical investigation estimates four models by accounting for the CSD and slope heterogeneity using a battery of second-generation panel techniques. The implications of natural resource rents on environmental degradation, both in the short term and long term, are subsequently estimated using the PMG, MG, and DFE estimators.

The analysis suggests the existence of cointegration for all considered models; i.e., there are long-run linkages between EFP and resource rents. In addition, the PMG estimator has been proven to be more efficient than the MG and DFE estimators in assessing the short- and long-run environmental repercussions of natural resources. The outcomes may be summarized as follows. First, the PMG estimator results show an inverted U-shaped association between income and EFP, which supports the EKC hypothesis in resource-rich MENA nations. The analysis also reveals that environmental quality is worsened by energy use and urbanization only in the long run. Total natural resource rents are found to have a negative long-term impact on the EFP, with a 1% increase in rents resulting in a 0.053% increase in the EFP. The investigation also reveals that natural resource rents and the other explanatory variables have no significant effects in the short run. The disaggregate analysis suggests that environmental quality reacts differently to the considered natural resource. Indeed, oil and gas pose a long-term threat to the environment, while the extraction of minerals does not have any impact. Moreover, oil rents have a more substantial adverse long-term impact than natural gas rents, as a 1% increase in oil rents leads to a rise in the EFP by 0.117%.

The results of this study can be valuable in formulating policy recommendations to conserve the environment in MENA countries. First, resource-rich MENA countries are urged to diversify their economies and lessen their reliance on fossil-fuel energy, notably oil and natural gas, as their primary source of government revenue. In addition to the economic advantages of diversification, this will enable resource-abundant MENA countries to mitigate the detrimental effects of oil and natural gas extraction on environmental conditions. Second, policymakers in the resource-abundant MENA countries may exhibit interest in expediting the energy transition by utilizing natural resource revenues. This is important since renewable and nuclear energy projects often require substantial investment for their installation. By doing so, MENA countries may make substantial progress in the energy transition based on their resource endowments. Finally, policymakers in MENA countries may levy a specific tax on companies engaged in the extraction and exploration of natural resources. The revenue generated from this tax could be allocated to a specified fund, which could be utilized to finance environmental protection initiatives and rehabilitation efforts.

Although the present study provides some fresh evidence on the environmental implications of natural resources, it could be improved in many ways. First, the empirical investigation focuses on a sample of MENA countries. It would be of interest to consider expanding the geographic scope to include, for example, oil-exporting OPEC countries. Furthermore, the analysis may be improved by analyzing the asymmetric implications of natural resources on environmental degradation, which may be useful for the design of policy implications. This could be achieved by estimating a nonlinear ARDL model.