Green Energy Finance and Agricultural Performance in MENA Region: Structural Pathways Toward Sustainability

Abstract

This study investigates the macroeconomic, institutional, and energy-related determinants of agricultural value added in Middle East and North Africa (MENA) countries over the period 2000–2023, with particular emphasis on whether international clean energy finance operates as a conditionally effective driver depending on energy endowments. Using a panel fixed-effects framework with Driscoll–Kraay standard errors to address cross-sectional dependence, heteroskedasticity, and serial correlation, the analysis incorporates an interaction term between clean energy finance and an oil-exporting dummy to capture structural heterogeneity. Robustness is ensured through Panel-Corrected Standard Errors (PCSEs), Granger causality tests, and System GMM estimation. The findings reveal that GDP per capita and clean energy finance are positively and significantly associated with agricultural value added, while trade openness negatively affects the sector. Importantly, the interaction results indicate strong asymmetry: the positive contribution of clean energy finance is concentrated in non-oil economies but becomes weak or insignificant in oil-exporting countries, consistent with diminishing marginal returns in energy-abundant contexts. Inflation captures nominal price effects, while short-run dynamics suggest the presence of adjustment costs. Overall, the study highlights that clean energy finance acts as a structurally conditional mechanism, offering nuanced and policy-relevant insights for sustainable agricultural transformation in MENA economies.

1. Introduction

Agriculture remains a strategic pillar of economic stability and food security in Middle East and North Africa (MENA) countries, despite its declining share in aggregate output. The sector remains a key source of employment, rural income, and food security, particularly in economies characterized by demographic pressures and uneven development patterns. However, agricultural systems in the region are increasingly exposed to structural challenges, including water scarcity, land degradation, and climate variability, which constrain productivity and threaten long-term sustainability.

These challenges are particularly acute in arid and semi-arid environments, where agricultural production is highly dependent on efficient resource management and technological adaptation. In this context, understanding the macroeconomic and structural drivers of agricultural performance has become essential for designing policies that enhance resilience and sustainability. Beyond traditional inputs such as land and labor, recent research emphasizes the growing importance of infrastructure, institutional quality, and access to modern technologies in shaping agricultural outcomes.

At the same time, the global shift toward sustainability has intensified interest in the role of green finance as a catalyst for structural transformation. International financial flows directed toward clean energy projects have emerged as a potential mechanism to support agricultural development. By improving access to reliable and affordable energy, especially in rural areas, clean energy investments can facilitate irrigation, storage, and processing activities that are critical for productivity gains.

Despite this growing relevance, empirical evidence on the relationship between clean energy finance and agricultural performance remains limited, particularly at the macroeconomic level and within the MENA region. Much of the existing literature relies on micro-level analyses or single-country case studies, which, while informative, do not capture broader structural dynamics or cross-country variation (Mei et al., 2022) [1]. Moreover, many empirical studies implicitly assume homogeneous effects across countries, thereby overlooking the extent to which structural differences condition the effectiveness of energy-related investments.

This limitation is especially important in the MENA context, where economies differ markedly in terms of energy endowments, institutional capacity, and economic structure. In particular, the distinction between oil-exporting and non-oil economies introduces a fundamental asymmetry that may shape how clean energy finance influences agricultural outcomes. Ignoring this heterogeneity risks masking important conditional relationships and may lead to incomplete or misleading policy conclusions.

Despite this growing body of research, existing studies remain limited in three critical respects. First, most empirical analyses assume homogeneous effects of energy-related investments, implicitly treating clean energy finance as a uniformly effective policy tool across countries. Second, the macroeconomic transmission mechanisms linking clean energy finance to agricultural performance remain insufficiently theorized, particularly in resource-constrained environments. Third, empirical evidence for the MENA region is scarce and fragmented, with limited attention to structural differences between oil-exporting and non-oil economies. These limitations suggest that the effectiveness of clean energy finance may be conditional rather than universal, depending on underlying economic structures and energy constraints.

This study addresses these gaps by examining the determinants of agricultural value added across MENA countries over the period 2000–2023 within a panel data framework. The empirical strategy is designed to capture both cross-sectional and temporal variation while addressing key econometric challenges, including cross-sectional dependence, heteroskedasticity, and serial correlation. Rather than aiming to establish strict causal relationships, the analysis focuses on identifying robust structural associations and short-run adjustment dynamics, providing a realistic and policy-relevant interpretation of the results.

The main contribution of this paper lies in demonstrating that the impact of clean energy finance on agricultural performance is not uniform but structurally conditional, depending critically on underlying energy endowments and economic characteristics. Rather than merely examining the relationship between clean energy finance and agriculture, the study specifically demonstrates that the effectiveness of clean energy finance differs substantially between oil-exporting and non-oil MENA economies. In particular, the findings reveal that positive effects are concentrated in non-oil economies, where energy constraints are binding, while no comparable gains are observed in oil-exporting countries. The interaction-based evidence further confirms that this heterogeneity is statistically significant, reflecting a diminishing marginal returns mechanism in energy-abundant contexts. Accordingly, the study advances the literature by highlighting that clean energy finance should not be viewed as a universally effective policy instrument, but rather as a context-dependent mechanism whose effectiveness varies according to structural economic conditions. This highlights the importance of explicitly accounting for structural asymmetry when evaluating the effectiveness of energy-transition policies.

By integrating macroeconomic determinants, energy-related financial flows, and interaction-based heterogeneity within a unified empirical framework, this study advances the literature beyond conventional homogeneous specifications. It provides a more nuanced and theoretically grounded understanding of agricultural transformation in resource-constrained environments. From a policy perspective, the results underscore the need for differentiated and context-specific strategies, emphasizing that the effectiveness of clean energy finance depends on country-specific structural conditions rather than representing a universally applicable policy instrument across MENA economies.

The remainder of this paper is structured as follows. Section 2 reviews the relevant literature on the determinants of agricultural value added. Section 3 describes the data and outlines the econometric methodology. Section 4 presents empirical results, including baseline estimations and robust checks. Section 5 discusses the findings considering the underlying economic mechanisms and existing literature. Finally, Section 6 concludes with policy implications, study limitations, and directions for future research.

2. Literature Review

2.1. Determinants of Agricultural Value Added

The determinants of agricultural value added have been widely analyzed within development and environmental economics, yet existing studies increasingly emphasize that traditional factor-based explanations are insufficient to capture productivity dynamics in resource-constrained regions such as MENA. Traditional approaches emphasize the role of factor endowments—such as land, labor, and capitals—as the primary drivers of agricultural output. However, more recent studies highlight the increasing importance of productivity-enhancing factors, including infrastructure development, technological adoption, and institutional quality, in shaping agricultural outcomes.

Economic development, commonly proxied by GDP per capita, is widely recognized as a key determinant of agricultural productivity and structural transformation dynamics Higher income levels enable greater investment in infrastructure, improve access to financial resources, and facilitate the adoption of modern technologies, thereby enhancing productivity (Fan, Gulati, & Thorat, 2008 [2]; Abid, 2025 [3]). At the same time, the relationship between economic growth and agriculture is not unidirectional. Structural transformation processes may reallocate resources away from agriculture toward more productive sectors, potentially reducing its relative contribution to the economy. However, in many developing regions, productivity gains may offset these reallocation effects, leading to more efficient agricultural systems.

Inflation represents another macroeconomic factor with theoretically ambiguous effects on agricultural performance. On one hand, increases in agricultural prices may raise nominal value added through price transmission mechanisms. On the other hand, inflation can introduce uncertainty, increase input costs, and distort investment decisions, thereby negatively affecting agricultural production (Swinnen, 2010 [4]; Anríquez & Daidone, 2010 [5]). This dual nature makes it essential to distinguish between nominal and real effects when interpreting empirical results.

Trade openness is also a critical determinant with context-dependent effects shaped by domestic competitiveness and structural constraints. While trade liberalization can improve efficiency, expand market access, and promote specialization, it may also expose domestic producers to international competition. In economies with structural disadvantages—such as low productivity or limited technological capacity—this exposure may lead to a decline in domestic agricultural output (Anderson, 2009 [6]). Consequently, the impact of trade on agriculture is highly context-dependent and shaped by domestic policy frameworks and competitiveness levels.

2.2. Clean Energy, Finance, and Agriculture

The relationship between energy, water, and agriculture has been widely explored within the energy–water–food nexus framework. This literature emphasizes the critical role of energy access in supporting irrigation systems, water management, and agricultural processing, particularly in resource-constrained environments (Jalilov et al., 2018 [7]; ElZein et al., 2022 [8]). Reliable energy supply is essential for improving agricultural productivity and ensuring the efficient use of scarce resources.

Renewable energy technologies, particularly decentralized solar systems, have been identified as effective solutions for addressing energy constraints in rural areas. By reducing dependence on fossil fuels and improving energy reliability, these technologies can enhance agricultural productivity and resilience (Burney et al., 2010 [9]). Their relevance is especially pronounced in regions where conventional energy infrastructure is limited or unreliable.

These mechanisms operate through indirect and multi-dimensional transmission channels linking energy access to irrigation efficiency, input utilization, and post-harvest processing, thereby influencing agricultural productivity at the macro level.

More recently, the focus has shifted toward the role of financial mechanisms in enabling the adoption of clean energy solutions. International financial flows directed toward renewable energy projects can alleviate investment constraints, support infrastructure development, and facilitate long-term sustainability transitions. However, empirical evidence on the macroeconomic impact of such financial flows remains limited, particularly in developing regions.

The relationship between clean energy finance and agricultural performance operates primarily through indirect infrastructure and energy-access channels rather than through direct agricultural subsidies. Importantly, the clean energy finance variable used in this study captures broad international financial flows directed toward renewable-energy development and clean energy projects, rather than financing specifically allocated to the agricultural sector. These investments can nevertheless influence agricultural productivity indirectly by improving energy availability, supporting irrigation systems, reducing energy costs, and enhancing storage and processing capacities, particularly in rural and resource-constrained areas. Through these mechanisms, clean energy finance contributes to a more reliable and sustainable energy environment that facilitates agricultural production and post-harvest activities. This transmission mechanism is especially relevant in non-oil MENA economies, where energy constraints represent a significant barrier to agricultural efficiency and productivity growth.

Recent studies provide growing empirical evidence on the macroeconomic role of green finance. For example, Wang and Xu (2025) [10] show that green finance significantly improves energy efficiency through technological upgrading and structural transformation, highlighting its role as a catalyst for productivity-enhancing investments. Similarly, CİTİL (2024) [11] demonstrates that green finance constitutes a key prerequisite for green growth in macroeconomic settings, reinforcing its function as a structural driver rather than a sector-specific input. In addition, Lee and Hussain (2022) [12] emphasize the critical link between energy consumption and carbon-neutral development, suggesting that sustainability-oriented financial flows operate through energy-related transmission channels. These contributions collectively support the view that clean energy finance influences economic performance indirectly by alleviating structural constraints and enabling long-term sustainability transitions.

Recent finance literature further reinforces the macroeconomic importance of green finance in facilitating sustainability-oriented structural transformation. For example, Alharbi, Al Mamun, Boubaker, and Rizvi (2023) [13] provide international evidence that green finance significantly supports renewable-energy development by improving financing conditions for clean energy infrastructure and accelerating energy-transition mechanisms. Similarly, Al Mamun, Boubaker, and Nguyen (2022) [14] show that green finance contributes to decarbonization and sustainability-oriented economic restructuring through investments that enhance environmental efficiency and long-run sustainable development. These studies provide additional theoretical support for the argument that international clean energy finance may indirectly improve agricultural performance through enhanced energy access, irrigation capacity, storage infrastructure, and processing efficiency, particularly in resource-constrained economies.

In addition, existing studies often overlook the role of structural heterogeneity in shaping the effectiveness of clean energy finance. Differences in energy endowments, institutional capacity, and economic structure may lead to heterogeneous outcomes across countries. In the MENA region, the contrast between oil-exporting and non-oil economies represents a critical dimension that has not been sufficiently addressed in empirical literature.

Despite the growing body of literature on the energy–water–food nexus and renewable energy adoption, existing studies remain limited in two important respects. First, most empirical analyses are conducted at the micro or country-specific level, which restricts their ability to capture broader macroeconomic dynamics and cross-country heterogeneity. Second, most studies implicitly assume homogeneous effects of energy-related investments, thereby overlooking the role of structural differences in shaping outcomes. In the context of MENA economies, where energy endowments vary significantly between oil-exporting and non-oil countries, this limitation is particularly critical. By explicitly incorporating structural heterogeneity within a macro-panel framework, this study extends the existing literature and provides new evidence on the conditional nature of clean energy finance in agricultural transformation.

This study extends the literature by explicitly incorporating structural heterogeneity within a macro-panel framework, an aspect largely overlooked in existing empirical research.

2.3. Hypothesis Development

Building on the theoretical and empirical insights discussed above, this study develops a set of testable hypotheses linking macroeconomic, structural, and energy-related factors to agricultural value added in the MENA region. The hypotheses are grounded in both structural transformation theory and the energy–water–food nexus framework.

These hypotheses are empirically tested using a panel data framework that accounts for cross-sectional dependence, unobserved heterogeneity, and dynamic adjustments.

Economic development is expected to enhance agricultural performance through capital deepening, infrastructure expansion, and technological adoption. Higher income levels facilitate access to credit, improve input quality, and enable investments in productivity-enhancing technologies. Previous empirical studies have documented that economic development contributes positively to agricultural productivity through modernization and infrastructure improvement (Fan et al., 2008 [2]; Gollin et al., 2014 [15]).

H1. 

Economic development (GDP per capita) has a positive effect on agricultural value added.

The effect of inflation on agriculture is theoretically ambiguous. While higher output prices may increase nominal agricultural value added through price transmission mechanisms, inflation can also generate uncertainty, raise input costs, and distort investment decisions. Empirical evidence remains mixed. For example, Swinnen (2010) [4] and Anríquez and Daidone (2010) [5] show that inflationary pressures may simultaneously increase nominal agricultural revenues while reducing real sectoral efficiency through higher production costs and market uncertainty.

H2. 

Inflation has an ambiguous effect on agricultural value added.

Trade openness may influence agriculture through two opposing channels. On one hand, it can improve efficiency and market access; on the other, it may expose domestic producers to international competition, particularly in structurally weak agricultural systems. Several empirical studies report that trade liberalization may negatively affect domestic agricultural sectors in developing economies characterized by weak competitiveness and import dependence (Anderson, 2009 [6]; Fuglie & Wang, 2013 [16]).

H3. 

Trade openness is expected to exert a negative effect on agricultural value added in structurally constrained economies, reflecting competitiveness limitations and exposure to external pressures.

Clean energy finance is expected to influence agriculture indirectly through the energy–water–food nexus. By improving access to reliable energy, particularly in rural areas, it appears to support irrigation, storage, and processing activities, thereby enhancing productivity. Recent empirical research highlights that renewable-energy investments and green finance mechanisms contribute positively to agricultural sustainability and productivity through improved energy access and infrastructure development (Burney et al., 2010 [9]; Wang & Xu, 2025 [10]; Lee & Hussain, 2022 [12]).

H4. 

Clean energy finance is positively associated with agricultural value added.

The effectiveness of clean energy finance is likely to depend on underlying energy constraints. In energy-scarce (non-oil) economies, clean energy investments can alleviate binding constraints, while in energy-abundant (oil-exporting) economies, the marginal impact is expected to be limited. This argument is consistent with the structural heterogeneity literature, which emphasizes that the effectiveness of sustainability-oriented investments depends on country-specific economic structures, institutional conditions, and resource endowments (Jalilov et al., 2018 [7]; ElZein et al., 2024 [8]).

H5. 

The positive effect of clean energy finance on agricultural value added is stronger in non-oil economies than in oil-exporting economies.

2.4. Contributions of the Study

This study contributes to literature in several important ways. First, it provides new macro-level evidence on the determinants of agricultural value added in MENA countries, a region that remains underexplored in empirical research despite its structural vulnerability to environmental and resource constraints. Second, the study advances the literature by explicitly incorporating structural heterogeneity, demonstrating that the impact of clean energy finance is not homogeneous but depends critically on underlying energy endowments.

Third, unlike much of the existing literature—which relies on micro-level analyses or assumes uniform effects, this paper adopts a macro-panel framework that captures both cross-country variation and temporal dynamics. For example, several previous studies on the energy–agriculture nexus primarily focus on household-level adoption, farm-level efficiency, or single-country case studies (Burney et al., 2010 [9]; Jalilov et al., 2018 [7]; Mei et al., 2022 [1]). In contrast, cross-country macro-panel evidence for the MENA region remains limited. This allows for a more comprehensive understanding of how economic, institutional, and energy-related factors jointly shape agricultural outcomes.

Most importantly, the study identifies a structural asymmetry between oil-exporting and non-oil economies, showing that clean energy finance enhances agricultural performance only where energy scarcity constitutes a binding constraint. This finding challenges the implicit assumption of uniform policy effectiveness and highlights the importance of context-specific strategies in energy-transition and agricultural policies.

Overall, the study advances literature by shifting the focus from uniform policy effects to conditional and context-dependent relationships, thereby offering a more robust framework for understanding agricultural transformation in resource-constrained and structurally heterogeneous economies.

3. Data

The dataset consists of a panel of Middle East and North Africa (MENA) countries observed over the period 2000–2023, constructed to examine the determinants of agricultural value added in the region. The sample includes both oil-exporting and non-oil economies, enabling a structured analysis of cross-country heterogeneity in energy endowments and economic structure. All continuous variables are transformed into natural logarithms to reduce heteroskedasticity and allow for elasticity interpretation, except institutional quality. First differences are used in short-run specifications (PCSE) to capture dynamic adjustments.

Table 1. Data description.

Indicator Name	Code	Description	Source
Agriculture, forestry, and fishing, value added (constant 2015 US$)	AGR	Net output of agriculture, forestry, and fishing sectors (adjusted for inflation).	World Bank (WDI)
International finance received for clean energy (All renewables) Millions of constant 2021 US$	LGF	International financial flows to developing countries for clean energy R&D and renewable energy production.	Organisation for Economic Co-operation and Development (OECD)- International Renewable Energy Agency (IRENA)
Arable land (% of land area)	LAN	Share of total land area used for temporary crops, meadows, gardens, or left fallow.	World Bank (WDI)
GDP per capita (constant 2015 US$)	GDP	Gross domestic product per person, inflation adjusted.	World Bank (WDI)
Trade (% of GDP)	TRD	Sum of exports and imports of goods/services as share of GDP.	World Bank (WDI)
Consumer price index (2010 = 100)	CPI	Measures the average change in prices paid by consumers (inflation index).	World Bank (WDI)
Institutional Quality (Composite Governance Indicator)	IQR	Average of six WGI components: political stability, voice/accountability, regulatory quality, effectiveness, rule of law, and corruption control.	World Bank (WGI)

Notes: Variable definitions and data sources are reported. AGR denotes agricultural value added; LGF represents international clean energy finance flows; GDP is GDP per capita; TRD denotes trade openness; CPI is the consumer price index; IQR represents institutional quality. All monetary variables are expressed in constant U.S. dollars.

provides a summary of the variables, including definitions and data sources.

The dependent variable, agriculture, forestry, and fishing value added (AGR), measured in constant 2015 U.S. dollars, reflects the real economic contribution of the primary sector to national output. This measure captures sectoral productivity and economic performance while controlling inflationary effects. It is particularly relevant in the MENA context, where agriculture remains a key source of employment and livelihood despite increasing environmental constraints.

To investigate the underlying drivers of agricultural performance, the dataset includes six explanatory variables capturing economic, structural, institutional, and energy-related dimensions. International financial flows for clean energy (LGF) represent external investments directed toward renewable energy development. Rather than acting as a direct input, LGF is conceptualized as a macro-structural variable influencing agriculture indirectly through energy availability, infrastructure development, and cost-reduction channels.

Arable land (LAN), expressed as a percentage of total land area, reflects the natural capacity for crop production. However, in the MENA region, where land and water constraints are binding, this variable primarily captures structural limitations rather than expansion potential.

GDP per capita (GDP), measured in constant 2015 U.S. dollars, serves as a proxy for economic development and structural transformation. It reflects the capacity of an economy to invest in infrastructure, technology, and agricultural modernization. Trade openness (TRD), defined as the ratio of exports and imports to GDP, captures the degree of integration into global markets. This variable reflects both access to inputs and exposure to external competition.

The consumer price index (CPI), indexed to 2010, is used to measure inflation. Its inclusion allows us to distinguish between nominal price effects and real changes in agricultural performance.

Finally, institutional quality (IQR) is introduced as a composite governance indicator, constructed as the average of six World Bank Worldwide Governance Indicators: political stability, voice and accountability, regulatory quality, government effectiveness, rule of law, and control of corruption. This variable captures the broader institutional environment affecting investment, policy implementation, and resource allocation in agriculture.

Overall, the dataset provides a comprehensive macro-level framework that integrates economic, institutional, and energy-related factors, allowing for a multidimensional analysis of agricultural performance across countries and over time. The panel structure—characterized by a relatively small cross-sectional dimension and a longer time span—motivates the use of estimators robust to cross-sectional dependence, heteroskedasticity, and serial correlation, ensuring reliable inference.

4. Methodology

4.1. Panel Data Regression Models: Fixed Effects and Random Effects

To investigate the determinants of sustainable agricultural value added in MENA countries, we employ panel data regression models, specifically Fixed Effects (FE) and Random Effects (RE). Panel techniques are particularly suitable in this context as they exploit both cross-sectional and temporal variation while controlling for unobserved heterogeneity across countries.

The Fixed Effects (FE) model is adopted as the baseline specification, as it controls for unobserved country-specific heterogeneity that may be correlated with the explanatory variables. This is especially relevant in the MENA region, where structural characteristics—such as institutional quality, geographic conditions, and resource endowments—jointly influence agricultural performance and its determinants. The FE model is specified as:

$$Y_{it}={\alpha }_i+\beta X_{it}+{\epsilon }_{it}$$

(1)

where $Y_{it}$ represents agricultural value added for country i at time t, ${\alpha }_i$ captures country-specific effects, $X_{it}$ denotes the vector of explanatory variables, and ${\epsilon }_{it}$ is the error term.

The Random Effects (RE) model is estimated as a benchmark specification, assuming that individual-specific effects are uncorrelated with the regressors:

$$Y_{it}=\alpha +\beta X_{it}+u_i+{\epsilon }_{it}$$

(2)

To determine the appropriate specification, the Hausman test (Hausman, 1978 [17]) is conducted. The results strongly favor the FE model, confirming that unobserved heterogeneity is correlated with the regressors and must be explicitly controlled for.

4.2. Robust Estimation: Driscoll–Kraay Standard Errors

Several diagnostic tests are performed to assess the validity of the panel estimates. The Wooldridge test indicates autocorrelation, the modified Wald test reveals heteroskedasticity, and Pesaran-type tests confirm cross-sectional dependence.

Given the presence of these econometric issues, conventional standard errors may lead to biased statistical inference. Therefore, the study employs Driscoll–Kraay standard errors, which provide consistent and robust inference in the presence of heteroskedasticity, serial correlation, and cross-sectional dependence. This estimator is particularly appropriate for macro-panel datasets with relatively large time dimensions.

Accordingly, the FE model with Driscoll–Kraay standard errors constitutes the primary empirical specification. This approach ensures that the estimated coefficients reflect robust structural associations rather than spurious relationships driven by cross-sectional dependence.

4.3. Heterogeneity Analysis

4.3.1. Oil vs. Non-Oil Economies

To account for structural heterogeneity within the MENA region, the analysis is extended by estimating the model separately for oil-exporting and non-oil economies. The classification divides countries into two groups based on their structural dependence on hydrocarbon production and energy-export revenues. The oil-exporting group includes Algeria, Iran, Iraq, Libya, and Yemen, while the non-oil group includes Djibouti, Egypt, Jordan, Lebanon, Morocco, the Syrian Arab Republic, and Tunisia. This classification is intended to capture differences in energy endowments and structural economic conditions that may condition the effectiveness of clean energy finance.

This subsample approach allows for a direct assessment of how differences in energy endowments condition the effectiveness of clean energy finance, without imposing restrictive parametric assumptions. It also enhances the interpretability of results by explicitly capturing structural asymmetries across country groups.

4.3.2. Interaction Model

To complement the subsample analysis and provide a more rigorous test of conditional effects, an interaction term between clean energy finance (LGF) and an oil-exporting country dummy is introduced. The model is specified as follows:

$$AG{R}_{it}={\beta }_0+{\beta }_{1}LG{F}_{it}+{\beta }_{2}OilDumm{y}_i+{\beta }_3(LG{F}_{it}\times OilDumm{y}_i)+X_{it}\gamma +{\mu }_i+{\epsilon }_{it}$$

(3)

where $AG{R}_{it}$ denotes agricultural value added, $LG{F}_{it}$ represents clean energy finance, $OilDumm{y}_i$ is a binary variable equal to one for oil-exporting countries, and $X_{it}$ is a vector of control variables. The term ${\mu }_i$ captures unobserved country-specific effects, while ${\epsilon }_{it}$ is the idiosyncratic error term.

The coefficient ${\beta }_3$ measures the differential impact of clean energy finance in oil-exporting economies relative to non-oil economies. A negative or statistically insignificant coefficient would indicate that the effectiveness of clean energy finance diminishes in energy-abundant contexts, consistent with a diminishing marginal returns mechanism.

Compared to the subsample approach, this specification provides a unified and statistically more efficient framework for testing structural heterogeneity within a single model.

The same FE–Driscoll–Kraay framework is applied to ensure comparability across subsamples.

4.4. Short-Run Dynamics: PCSE Estimation

To complement the baseline specification, short-run dynamics are examined using Panel-Corrected Standard Errors (PCSEs), following Beck and Katz (1995) [18].

PCSE estimation is particularly suitable in this context given the presence of contemporaneous correlation, heteroskedasticity, and relatively long-time dimensions (T > N). Unlike dynamic panel estimators, PCSE avoids instrument proliferation and bias when the research objective is to capture short-run associations rather than structural causality (Beck & Katz, 1995 [18]; Hoechle, 2007 [19]).

The model is estimated in first differences:

$$\Delta Y_{it}=\alpha +\beta \mathrm{D}·X_{it}+\gamma \mathrm{D}·Z_{it}+u_i+{\epsilon }_{it}$$

(4)

where D. denotes the first-difference operator.

The PCSE estimator is particularly suitable in this context as it accounts for contemporaneous correlation and panel-level heteroskedasticity, while avoiding the limitations of dynamic panel estimators in relatively small samples. This specification enables a clear distinction between short-run adjustment effects and medium-run structural relationships identified in level estimations.

4.5. Panel Causality: Dumitrescu and Hurlin (2012) [20]

To explore temporal relationships, the Dumitrescu and Hurlin (2012) [20] panel Granger causality test is employed.

It is important to emphasize that this test identifies predictive (Granger) causality rather than structural causality. Accordingly, the results are interpreted as indicating temporal precedence rather than definitive causal relationships.

4.6. Panel Cointegration Analysis

To assess long-run relationships, Kao (1999) [21], Pedroni (1999, 2004) [22,23], and Westerlund (2007) [24] cointegration tests are applied.

Given the presence of mixed integration orders and cross-sectional dependence, the interpretation of cointegration results is approached with caution. The evidence of cointegration is not fully consistent across tests; therefore, the analysis prioritizes structural associations in levels and short-run dynamics rather than relying exclusively on long-run equilibrium estimation.

4.7. Robustness Check: Dynamic Panel Estimation Using System GMM

To further assess robustness and address potential endogeneity, the System GMM estimator (Arellano and Bover, 1995 [25]; Blundell and Bond, 1998 [26]) is employed.

This approach accounts for potential reverse causality and dynamic persistence by using internal instruments derived from lagged variables. However, given the relatively small cross-sectional dimension of the dataset, particular care is taken to avoid instrument proliferation and ensure the validity of the estimates.

The model includes a lagged dependent variable and treats LGF, GDP, and TRD as potentially endogenous.

Importantly, System GMM is used as a complementary robustness check rather than the primary identification strategy. The main conclusions are therefore based on the FE–Driscoll–Kraay estimates, which are better aligned with the structure of the data and provide more reliable inference in this context.

5. Results

5.1. Data Analysis

5.1.1. Correlation and Multicollinearity Analysis

The correlation matrix and Variance Inflation Factor (VIF) results (

Table 2. Correlation and VIF Table.

Variable	AGR	LAN	LGF	GDP	TRD	IQR	CPI	VIF	1/VIF
AGR	1.0000							1.1500	0.8679
LAN	0.5297	1.0000						1.4600	0.6863
LGF	0.0274	0.1034	1.0000					1.3300	0.7512
GDP	0.0742	0.3066	−0.2137	1.0000				1.1600	0.8653
TRD	−0.9014	−0.4511	0.0376	−0.1288	1.0000			1.5300	0.6553
IQR	−0.1724	0.0425	0.1657	−0.0352	0.1182	1.0000		1.3200	0.7576
CPI	0.1321	0.0657	0.0103	0.3875	−0.1643	−0.1069	1.0000	1.3300	0.7512

Notes: Pairwise correlation coefficients are reported. VIF denotes the Variance Inflation Factor used to assess multicollinearity. Values below 10 indicate no serious multicollinearity concerns.

) provide preliminary insight into linear associations among variables.

Arable land (LAN) exhibits a moderate positive correlation with AGR, suggesting that countries with greater agricultural resource endowments tend to record higher levels of agricultural output. Trade openness (TRD) shows a strong negative correlation with AGR; however, this bivariate relationship primarily reflects cross-country structural differences rather than within-country dynamics over time.

The relatively weak pairwise correlations observed for LGF, GDP, IQR, and CPI indicate that simple bivariate relationships do not adequately capture the underlying multivariate interactions explored in the regression analysis.

The VIF values are all well below the conventional threshold of 10 with a mean 1.32, indicating no serious multicollinearity concerns. This confirms that the explanatory variables can be jointly included without compromising the stability of coefficient estimates.

5.1.2. Descriptive Statistics

The descriptive statistics reveal substantial cross-country variation, particularly in arable land and clean energy finance flows (

Table 3. Descriptive statistics.

Variable	Mean	Std. Dev.	Min	Max
AGR	22.4835	1.5983	17.0164	24.6207
LAN	1.5205	1.4432	−3.1433	3.1679
LGF	15.2694	3.1234	9.2103	21.4472
GDP	8.1940	0.4312	6.5685	9.1000
TRD	4.2862	0.4637	3.3964	5.8522
IQR	3.2845	0.8165	0.5501	4.1685
CPI	4.6641	0.6107	2.8991	7.4150

Notes: Summary statistics are based on the unbalanced panel dataset for MENA countries over the period 2000–2023. Variables are expressed in logarithmic form where applicable.

). This dispersion supports the use of panel estimators capable of controlling unobserved heterogeneity across countries.

Agricultural value-added shows reasonable variation across the panel, suggesting meaningful differences in agricultural productivity among the countries studied. Among the independent variables, arable land displays significant dispersion. This is expected in the MENA region, where some countries, such as Egypt and Morocco, have substantial agricultural areas, while others in the region have very limited arable land. The log transformation captures this disparity effectively, particularly given the skewed distribution of land endowments across MENA countries.

International financial flows for clean energy also vary widely across the region. This reflects differing national strategies and levels of engagement with international donors and investors in renewable energy. Although these flows are not directly aimed at agriculture, they may influence agricultural performance indirectly through improvements in infrastructure, energy access, and environmental sustainability. GDP per capita shows moderate variation, consistent with the diverse levels of economic development across MENA countries. The log transformation here allows us to observe proportional differences in income levels and assess how they relate to agricultural sector performance.

Trade openness, institutional quality, and inflation, each show enough variation to support robust empirical analysis. Trade openness is particularly relevant in the context of the MENA region’s integration into global markets, where economies with greater exposure to international trade may undergo structural shifts that reduce reliance on agriculture. Institutional quality reflects the governance environment, which can influence agricultural investment, land tenure security, and resource management. Lastly, inflation levels, though relatively stable in some countries, differ enough across the panel to allow exploration of how price stability, or the lack of it, affects agricultural outcomes.

Overall, the transformed dataset captures a rich set of dynamics, enabling an in-depth assessment of how economic, institutional, and environmental factors influence agricultural performance in the MENA region. The observed dispersion across countries also supports the later heterogeneity analysis between oil-exporting and non-oil economies.

5.1.3. Unit Root and Cross-Sectional Dependence

The stationarity results of the panel data series, assessed through unit root testing, reveal important insights about the nature of each variable and their suitability for econometric modelling (

Table 4. Fisher-type Unit Root Test Results.

Variable	Z-Stat	p-Value	Variable	Z-Stat	p-Value
AGR	0.5486	0.7084	dAGR	−7.6026	0.0000
LAN	−1.6393	0.0506	dLAN	−8.0605	0.0000
LGF	−4.5006	0.0000	dLGF	−10.936	0.0000
GDP	−2.2362	0.0127	dGDP	−4.1165	0.0000
TRD	−1.9191	0.0275	dTRD	−6.8866	0.0000
IQR	0.1995	0.5791	dIQR	−7.496	0.0000
CPI	2.8587	0.9979	dCPI	−2.4228	0.0000

Notes: Fisher-type panel unit-root tests are reported. Variables prefixed with “d” denote first differences. The null hypothesis is the presence of a unit root.

).

The stationarity results indicate a mixed order of integration. AGR, IQR, and CPI are non-stationary in levels but become stationary after first differencing, while LAN, LGF, GDP, and TRD are stationary in levels.

The use of differenced forms for inflation and institutional quality in selected specifications reflects both econometric and conceptual considerations. Unit-root tests indicate that CPI and IQR exhibit non-stationary behavior in levels, whereas their first differences are stationary. Accordingly, differencing these variables helps avoid spurious inference and ensures consistency with the short-run dynamic framework. From a conceptual perspective, changes in inflation and institutional quality are more relevant for capturing short-term adjustments in agricultural performance than their absolute levels. Therefore, the coefficients on dCPI and dIQR should be interpreted as reflecting the effects of short-run changes in inflationary conditions and institutional quality, whereas variables retained in levels primarily capture medium-run structural associations.

This mixed integration structure justifies the empirical strategy that distinguishes between level-based structural associations and short-run dynamics estimated in first differences.

Table 5. Pesaran’s Cross-sectional Dependence (CD) Test Results.

Variable	CD Statistic	p-Value
AGR	20.1910 ***	0.0000
LAN	1.4370	0.1510
LGF	2.0490 **	0.0410
GDP	10.3100 ***	0.0000
TRD	10.6320 ***	0.0000
IQR	7.5790 ***	0.0000
CPI	26.4580 ***	0.0000

Notes: Pesaran’s CD test examines cross-sectional dependence across panel units. *** and **, indicate statistical significance at the 1% and 5% levels, respectively.

presents the results of Pesaran’s Cross-sectional Dependence (CD) test, which examines whether residuals across panels are correlated, indicating the presence of cross-sectional dependence.

Pesaran’s CD test reveals strong cross-sectional dependence across most variables. This finding suggests the presence of common regional shocks—such as energy price fluctuations, trade linkages, and macroeconomic spillovers—affecting MENA economies simultaneously.

Given the mixed order of integration and cross-sectional dependence, the empirical strategy prioritizes robust inference over strict long-run equilibrium estimation. Consequently, the use of Driscoll–Kraay standard errors is required to obtain consistent and reliable inference.

5.2. Regression Analysis: Fixed and Random Effects Models

Table 6. Panel Regression Results.

Fixed-Effects (Within) Regression				Random-Effects GLS Regression
Variable	Coefficient	Std. Error	p-Value	Variable	Coefficient	Std. Error	p-Value
LAN	0.2230	0.1940	0.2530	LAN	0.5710 ***	0.1040	0.0000
LGF	0.0180 ***	0.0060	0.0040	LGF	0.0172 **	0.0079	0.0300
GDP	1.1600 ***	0.1360	0.0000	GDP	0.9140 ***	0.1660	0.0000
TRD	−0.3450 ***	0.0900	0.0000	TRD	−0.5500 ***	0.1110	0.0000
dIQR	0.0370	0.1200	0.7560	dIQR	0.0551	0.1550	0.7230
dCPI	0.7060 ***	0.1470	0.0000	dCPI	0.6960 ***	0.1890	0.0000
_cons	13.7540 ***	1.2490	0.0000	_cons	15.9680 ***	1.4650	0.0000
Model Summary		Value		Model Summary		Value
R-squared (within)		0.4101		R-squared (within)		0.3662
R-squared (between)		0.3997		R-squared (between)		0.5482
R-squared (overall)		0.2958		R-squared (overall)		0.4222
F-statistic		19.5800		Wald Chi-squared		142.1700
Prob > F		0.0000		Prob > Chi2		0.0000

Notes: Fixed-effects and random-effects estimations are reported. Country fixed effects are included in the fixed-effects specification. Number of countries = 12; Number of observations = 280. *** and ** indicate significance at the 1% and 5% levels, respectively.

reports the panel regression results estimated using both fixed effects and random effects models to examine the relationship between the dependent and independent variables across the panel dataset.

The comparison between FE and RE models reveals consistent patterns across specifications. GDP per capita (GDP) shows a strong and highly significant positive association with agricultural value added, indicating that higher income levels are linked to improved agricultural performance. Similarly, changes in inflation (dCPI) are positive and highly significant, suggesting that short-run increases in price levels are associated primarily with nominal valuation effects rather than real productivity improvements.

International clean energy finance (LGF) is statistically significant in both models. This suggests that external financial flows are associated with improved agricultural performance through indirect infrastructure and energy-related channels. Trade openness (TRD) remains negative and significant. This may reflect structural transformation effects, where increased integration into global markets reduces the relative importance of agriculture. Arable land (LAN) is not significant in the FE model but significant in RE. This divergence indicates that land operates primarily as a cross-country structural factor rather than a within-country dynamic driver. Short-run changes in institutional quality (dIQR) are not statistically significant in most specifications, suggesting that institutional effects are likely to operate gradually over longer horizons rather than through immediate adjustments.

5.3. Diagnostic Tests

Table 7. Hausman Test Results.

Test	Chi2 Statistic	p-Value
Hausman Test	76.2300 ***	0.0000

Notes: The Hausman test compares fixed-effects and random-effects estimators. A statistically significant statistic indicates that the fixed-effects model is preferred. *** p < 0.01.

presents the results of the Hausman specification test, which is used to determine the appropriate model between fixed effects and random effects. A significant test statistic indicates that the fixed effects model is preferred due to systematic differences in coefficients.

The Hausman test strongly rejects the RE specification (p < 0.01), confirming that fixed effects are the appropriate estimator. Accordingly, subsequent interpretation focuses on the fixed-effects specification and its robust variants.

The results for both the Wooldridge test for autocorrelation and the Modified Wald test for groupwise heteroskedasticity are presented in

Table 8. Diagnostic Tests Results.

Test	Statistic	p-Value
Wooldridge Test for Autocorrelation	23.6010 ***	0.0009
Modified Wald Test for Groupwise Heteroskedasticity	349.8400 ***	0.0000

Notes: The Wooldridge test examines first-order autocorrelation, while the Modified Wald test evaluates groupwise heteroskedasticity in the fixed-effects model. *** p < 0.01.

.

Wooldridge and Modified Wald tests confirm the presence of serial correlation and heteroskedasticity. Combined with cross-sectional dependence, this justifies using Driscoll–Kraay standard errors, which are robust to all three issues. For this reason, the fixed-effects model with Driscoll–Kraay standard errors, is treated as the main empirical specification.

5.4. Main Estimation: Driscoll–Kraay

The fixed-effects model is estimated using Driscoll–Kraay standard errors, which provide robust and reliable inference under these conditions (

Table 9. Fixed-effects with Driscoll–Kraay standard errors.

Variable	Coefficient	Standard Error	p-Value
LAN	0.2230	0.1519	0.1570
LGF	0.0178 **	0.0077	0.0300
GDP	1.1602 ***	0.1712	0.0000
TRD	−0.3450 **	0.1351	0.0180
dIQR	0.0372	0.0952	0.7000
dCPI	0.7059 ***	0.1210	0.0000
_cons	13.7544 ***	1.5511	0.0000
Model Summary		Value
Within R-squared		0.4101
F-statistic		17.1600
Prob > F		0.0000

Notes: Fixed-effects estimation with Driscoll–Kraay standard errors robust to heteroskedasticity, serial correlation, and cross-sectional dependence. Country fixed effects are included. Number of countries = 12; Number of observations = 280. *** and ** indicate significance at the 1% and 5% levels, respectively.

).

The Driscoll–Kraay results confirm the baseline findings. Clean energy finance (LGF) has a positive and statistically significant effect, supporting the hypothesis that energy-related investments indirectly enhance agricultural performance. GDP remains a strong and robust determinant, reinforcing the role of economic development in supporting agriculture. Trade openness (TRD) has a negative and significant effect. This result is consistent with the presence of structural reallocation mechanisms and external competition pressures. Inflation (dCPI) remains positive and highly significant. This result should be interpreted cautiously, as it likely reflects nominal valuation effects rather than real productivity improvements.

5.5. Heterogeneity Analysis

5.5.1. Oil vs. Non-Oil Economies

To account for structural heterogeneity within the MENA region, the baseline fixed-effects model with Driscoll–Kraay standard errors was re-estimated separately for oil-exporting and non-oil economies. The results reveal a clear asymmetry in the role of clean energy finance (

Table 10. Fixed-effects with Driscoll–Kraay standard errors: Oil vs. Non-Oil Economies.

	Oil Economies			Non-Oil Economies
Variable	Coefficient	Std. Error	p-Value	Coefficient	Std. Error	p-Value
LAN	0.0465	0.6320	0.9420	0.2885	0.2202	0.2040
LGF	−0.0169	0.0182	0.3650	0.0277 ***	0.0056	0.0000
GDP	0.8234 *	0.4328	0.0710	1.2716 ***	0.0825	0.0000
TRD	−0.3816	0.3730	0.3180	−0.4357 ***	0.1226	0.0020
dIQR	−0.0556	0.1266	0.6650	0.0471	0.2722	0.8640
dCPI	1.4411 ***	0.3111	0.0000	0.5062 ***	0.0611	0.0000
_cons	18.1511 ***	4.5897	0.0010	12.6709 ***	1.1402	0.0000
Model Summary
Statistic	Oil model			Non-oil model
Within R2	0.3130			0.6033
F-statistic	6.2500			48.6500
Prob > F	0.0007			0.0000

Notes: Fixed-effects estimations with Driscoll–Kraay standard errors are reported separately for oil-exporting and non-oil economies. Country fixed effects are included in all specifications. Number of countries 5 for Oil model with 117 observations and 7 for Non-oil model with 163 observations. *** and * indicate significance at the 1% and 10% levels, respectively.

).

The results reveal a clear asymmetry between oil and non-oil economies. In oil-exporting countries, clean energy finance has no statistically significant effect, indicating limited marginal impact in energy-abundant environments. In contrast, in non-oil economies, LGF is positive and highly significant, suggesting that energy constraints play a crucial role in determining the effectiveness of clean energy investments.

This asymmetry reflects a structural constraint mechanism, whereby the marginal productivity of clean energy investments depends on the extent to which energy availability constitutes a binding constraint. In non-oil economies, where energy access limits agricultural production, additional energy-related investment generates substantial productivity gains. In contrast, in oil-exporting economies, where energy supply is already abundant, the marginal contribution of such investments is significantly reduced, leading to diminishing returns.

5.5.2. Interaction Term Model

Table 11. Fixed-effects with Driscoll–Kraay standard errors: Interaction Model Results.

Variable	Coefficient	Std. Error	p-Value
LAN	0.2695 *	0.1563	0.0990
LGF	0.0301 ***	0.0074	0.0010
LGF × Oil	−0.0443 *	0.0247	0.0870
GDP	1.0835 ***	0.1380	0.0000
TRD	−0.3055 **	0.1340	0.0330
dIQR	0.0289	0.0954	0.7650
dCPI	0.6675 ***	0.1215	0.0000
_cons	14.1492 ***	1.1802	0.0000
Model Summary
Statistic	Interaction model
Within R2	0.4447
F-statistic	30.37
Prob > F	0.0000

Notes: Fixed-effects estimation with Driscoll–Kraay standard errors. The interaction term captures the differential effect of clean energy finance in oil-exporting economies relative to non-oil economies. Country fixed effects are included. Number of countries = 12; Number of observations = 280. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

presents the results of the fixed-effects model estimated with Driscoll–Kraay standard errors, incorporating the interaction term between clean energy finance (LGF) and the oil-exporting country dummy to formally assess structural heterogeneity across MENA economies.

The interaction model provides strong evidence of structurally differentiated effects of clean energy finance across MENA economies. The coefficient on clean energy finance (LGF) is positive and highly significant (β = 0.030, p < 0.01), indicating that, in non-oil economies, clean energy finance exerts a robust positive effect on agricultural value added.

However, the interaction term between LGF and the oil-exporting dummy is negative and marginally significant (β = −0.044, p < 0.10), suggesting that this positive effect is significantly attenuated in oil-exporting economies. This implies that the marginal impact of clean energy finance is conditional on underlying energy endowments and is substantially weaker in energy-abundant contexts.

To quantify this difference, the total effect of clean energy finance in oil-exporting economies can be computed as the sum of the baseline and interaction coefficients (0.030 − 0.044 ≈ −0.014), indicating that the net effect becomes statistically negligible or even slightly negative. This finding provides direct empirical support for a diminishing marginal returns mechanism; whereby additional energy-related investment yields limited productivity gains in economies where energy constraints are not binding.

These results reinforce the presence of structural asymmetry and confirm that the effectiveness of clean energy finance is not uniform but depends critically on country-specific structural conditions. Compared to subsample estimations, the interaction model offers a more rigorous and statistically efficient validation of this heterogeneity within a unified empirical framework.

The statistical significance of the interaction term confirms that the difference in coefficients between oil and non-oil economies is not merely descriptive but statistically meaningful.

In terms of economic significance, the estimated coefficients indicate that the magnitude of the effects is economically meaningful, particularly in non-oil economies. Since the variables are expressed in logarithmic form, the coefficients can be interpreted as elasticities. The coefficient of clean energy finance (LGF) in the non-oil economies model (0.0277) implies that a 1% increase in international clean energy finance is associated with approximately a 0.028% increase in agricultural value added. Although the elasticity may appear modest, its cumulative impact becomes substantial over time, especially in resource-constrained economies where agricultural productivity is highly sensitive to energy availability and infrastructure quality. For example, a 10% increase in clean energy finance would be associated with nearly a 0.28% increase in agricultural value added in non-oil MENA economies. In contrast, the effect becomes statistically negligible in oil-exporting countries, reinforcing the argument that the effectiveness of clean energy finance depends on underlying structural energy constraints rather than representing a universally effective policy instrument. These findings therefore suggest that clean energy finance can generate economically relevant productivity gains in energy-scarce agricultural systems through improved irrigation, storage, processing, and rural energy access.

5.6. Cointegration Tests

The results of the cointegration tests, Kao, Pedroni, and Westerlund, are summarized in

Table 12. Cointegration Tests Results.

Test	Statistic	Value	p-Value
Kao Test	Modified Dickey–Fuller t	1.9936 **	0.0231
	Dickey–Fuller t	2.0832 **	0.0186
	Augmented Dickey–Fuller t	1.4615 *	0.0719
	Unadjusted Modified Dickey–Fuller t	−1.5070 *	0.0659
	Unadjusted Dickey–Fuller t	−1.2550	0.1047
Pedroni Test	Modified Phillips-Perron t	3.0312 ***	0.0012
	Phillips-Perron t	−0.1830	0.4276
	Augmented Dickey–Fuller t	0.1893	0.4249
Westerlund Test	Variance ratio	1.4238 *	0.0773

Notes: Kao, Pedroni, and Westerlund panel cointegration tests are reported to examine long-run equilibrium relationships among variables. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

, providing evidence on the existence of long-run relationships among the panel variables.

The Kao, Pedroni, and Westerlund tests provide mixed and weak evidence of cointegration. Given cross-sectional dependence and mixed integration orders, the overall conclusion is that evidence of a stable long-run equilibrium relationship remains weak and inconclusive. Given mixed integration orders and weak, inconsistent cointegration evidence, the study refrains from estimating long-run cointegration coefficients and instead focuses on structural associations in levels and short-run dynamics in first differences.

5.7. Short-Run Dynamics (PCSE)

Table 13. Panel-Corrected Standard Errors (PCSEs).

Variable	Coefficient	Std. Error	p-Value
LAN D1.	0.1677 **	0.0750	0.0250
LGF D1.	−0.0049 **	0.0020	0.0150
GDP D1.	0.4673 ***	0.1560	0.0030
TRD D1.	−0.0920 *	0.0534	0.0850
IQR D1.	0.0810 ***	0.0256	0.0020
CPI D1.	0.1707	0.1167	0.1440
_cons	0.0276 ***	0.0056	0.0000
Model Summary		Value
R-squared		0.0745
Wald Chi2		41.80
Prob > Chi2		0.0000

Notes: Panel-Corrected Standard Error (PCSE) estimation is reported to capture short-run dynamics under contemporaneous correlation and heteroskedasticity. Number of countries = 12; Number of observations = 268. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

presents the results of the linear regression using Panel-Corrected Standard Errors (PCSEs). This model accounts for contemporaneous correlation across panels and panel-level heteroskedasticity. The coefficients represent short-run dynamics of the differenced variables, providing insights into the immediate effects of the independent variables on the dependent variable.

The PCSE results highlight short-run adjustment effects. In the short run, clean energy finance (D.LGF) exhibits a negative and significant effect, suggesting transitional adjustment costs or reallocation effects.

GDP growth remains positive and significant in the baseline specification; however, heterogeneous short-run effects across subsamples suggest instability in dynamic adjustments. Institutional improvements (D.IQR) become significant in the short run, indicating that governance changes may have more immediate effects compared to structural variables.

The explanatory power is relatively low, which is expected in first-difference models. These models capture short-term fluctuations rather than long-run structural relationships.

5.8. Short-Run Heterogeneity

The subgroup PCSE estimates further reinforce the heterogeneity identified in the baseline model (

Table 14. Panel-Corrected Standard Errors (PCSEs) (Oil vs. Non-Oil Countries).

	Oil Economies			Non-Oil Economies
Variable	Coefficient	Std. Error	p-Value	Coefficient	Std. Error	p-Value
LAN D1.	0.0434	0.0669	0.5170	0.2442 ***	0.0119	0.0000
LGF D1.	−0.0311 *	0.0168	0.0630	0.0601 ***	0.0082	0.0000
GDP D1.	0.8782 **	0.4161	0.0350	−0.7487 ***	0.0424	0.0000
TRD D1.	−1.2050 ***	0.2352	0.0000	−2.7421 ***	0.0765	0.0000
IQR D1.	−0.1265	0.2284	0.5800	1.4955 ***	0.3934	0.0000
CPI D1.	2.8284 ***	0.4379	0.0000	0.2425 *	0.1254	0.0530
_cons	21.1028 ***	4.1692	0.0000	38.9172 ***	0.4583	0.0000
Model Summary
Statistic	Oil model			Non-oil model
R-squared	0.8128			0.8867
Wald Chi2	203.11			3052.47
Prob > Chi2	0.0000			0.0000

Notes: Panel-Corrected Standard Error (PCSE) estimation is reported to capture short-run dynamics under contemporaneous correlation and heteroskedasticity. Number of countries 5 for Oil model with 111 observations and 7 for Non-oil model with 157 observations; ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

).

The subgroup analysis reinforces earlier findings. In non-oil economies, clean energy finance has a strong and positive short-run effect, while in oil economies the effect is negative or insignificant. This confirms that both the direction and magnitude of the relationship depend on structural conditions and time horizons.

5.9. Causality Analysis

Table 15. Dumitrescu & Hurlin (2012) [20] Granger non-causality Test Results.

Direction	Z-bar	p-Value
AGRICULTURE → LAN	−0.6440	0.5196
LAN → AGRICULTURE	2.5716 ***	0.0100
AGRICULTURE → LGF	0.6294	0.5291
LGF → AGRICULTURE	2.6787 ***	0.0074
AGRICULTURE → GDP	3.7246 ***	0.0002
GDP → AGRICULTURE	−0.5740	0.5658
AGRICULTURE → TRD	1.6985 *	0.0894
TRD → AGRICULTURE	4.0241 ***	0.0001
AGRICULTURE → IQR	−0.9200	0.3578
IQR → AGRICULTURE	0.2774	0.7815
AGRICULTURE → CPI	1.0968	0.2727
CPI → AGRICULTURE	0.9537	0.3402

Notes: Dumitrescu–Hurlin panel Granger causality tests are reported. The results indicate predictive temporal precedence rather than structural causality. *** and * indicate significance at the 1% and 10% levels, respectively.

presents the results of the Dumitrescu and Hurlin (2012) [20] panel Granger non-causality test. This test examines whether lagged values of one variable can predict current values of another variable across panel units, while allowing for heterogeneity in causal relationships across cross-sections. The null hypothesis assumes no Granger causality for all panel members.

The Granger Causality results reveal several important directional relationships between agriculture and other macroeconomic variables. There is evidence that land, green finance, and trade openness Granger-cause agricultural value added, suggesting that these variables help predict agricultural performance over time. Conversely, agriculture appears to be Granger-cause GDP per capita, implying that agricultural performance may contribute to broader economic activity.

Meanwhile, there is weak or no evidence of bidirectional Granger causality involving institutional quality and inflation. Overall, the results point to a predominantly unidirectional pattern in which agriculture is influenced by several structural variables. However, these findings should be interpreted as predictive rather than structural causality and therefore do not eliminate endogeneity concerns.

These findings suggest that agricultural performance is influenced by structural factors while also contributing to broader economic activity. However, these results reflect predictive relationships rather than structural causality and should be interpreted accordingly.

Overall, the findings indicate that the relationship between clean energy finance and agricultural performance is conditional, time-dependent, and structurally heterogeneous.

At the aggregate level, international clean energy finance is positively associated with added agricultural value; however, this relationship is neither uniform nor linear. The positive effect is primarily driven by non-oil economies, where structural reliance on agriculture and greater sensitivity to external financing enhance its impact. In contrast, the effect appears weak or statistically insignificant in oil-based economies, likely due to structural differences and lower dependence on the agricultural sector. Moreover, LGF exhibits a negative short-run effect, reflecting transitional adjustment costs, investment lags, and resource reallocation dynamics. Overall, these findings support a context-dependent and non-linear interpretation, indicating that the effectiveness of clean energy finance in promoting agricultural performance varies significantly across economic structures rather than operating as a one-size-fits-all policy instrument.

5.10. Robustness Check: Dynamic Panel Estimation Using System GMM

As an additional robustness check, a dynamic panel model is estimated using the System GMM estimator to address potential endogeneity and persistence in agricultural value added (

Table 16. System GMM Estimation Results.

Variables	Coefficient	Std. Error	p-Value
L.AGR	0.7385 ***	0.1416	0.001
LAN	0.0367	0.0514	0.493
LGF	0.0258	0.0145	0.108
GDP	−0.0721	0.1587	0.660
TRD	−0.7777 *	0.3971	0.082
IQR	0.1502	0.1339	0.291
CPI	0.2312	0.3786	0.556
_cons	9.3801	5.6845	0.133
Diagnostic tests	Value
Number of instruments	12
AR(1) p-value	0.0900
AR(2) p-value	0.7950
Hansen test (p-value)	0.9890
Sargan test (p-value)	0.9360

Notes: System GMM estimation is reported as a robustness check to address potential endogeneity and dynamic persistence. Robust standard errors are reported. AR(2) and Hansen tests are used to assess model validity. Number of countries = 12; Number of observations = 256. *** and * indicate significance at the 1% and 10% levels, respectively.

). This approach is particularly appropriate in macro-panel settings where reverse causality and dynamic adjustment processes may bias standard estimators. The results confirm a strong degree of persistence, as evidenced by the positive and highly significant coefficient on the lagged dependent variable.

The main findings remain broadly consistent with the baseline estimates. In particular, clean energy finance retains a positive coefficient, while trade openness continues to exhibit a negative effect, although both are only weakly significant in the dynamic specification. This pattern suggests that the effects of clean energy finance are gradual and may not fully materialize in short-run dynamic settings, while trade-related structural pressures remain persistent.

Diagnostic tests indicate no evidence of second-order serial correlation, while first-order autocorrelation is weakly significant, which is expected in first-differenced equations and does not invalidate the model. The Hansen test yields a high p-value, indicating that the instruments are valid; however, this result should be interpreted cautiously given the limited power of overidentification tests in panels with a relatively small number of cross-sectional units.

Overall, the System GMM results reinforce the direction and magnitude of the baseline relationships, supporting the robustness of the empirical findings. However, given the relatively small cross-sectional dimension of the dataset, System GMM is treated as a complementary robustness check rather than the primary identification strategy.

Accordingly, the primary conclusions of the study rely on the fixed-effects estimates with Driscoll–Kraay standard errors, which are more appropriate for the structure of the data and provide more reliable inference in the presence of cross-sectional dependence.

6. Discussion

This study investigates the macroeconomic, institutional, environmental, and energy-related determinants of agricultural value added (AGR) across MENA countries using panel econometric techniques. The analysis relies primarily on a fixed-effects specification with Driscoll–Kraay standard errors, complemented by PCSE estimates to capture short-run dynamics, thereby ensuring robust inference in the presence of cross-sectional dependence and distinguishing clearly between structural associations and short-run adjustments. Accordingly, the findings are interpreted as structural relationships rather than causal effects. Rather than reiterating empirical results, this section interprets the findings through underlying economic mechanisms and situates them within the existing literature, with a focus on context-specific dynamics in MENA economies.

6.1. Economic Development and Agriculture

The consistently strong and statistically significant positive relationship between GDP per capita and agricultural value added can be explained through capital deepening and structural transformation mechanisms. Economic development expands public and private investment in infrastructures such as irrigation systems, transport networks, and storage facilities—which directly enhance agricultural productivity (Fan, Gulati, & Thorat, 2008 [2]). In addition, higher income levels improve access to credit, modern input, and technology adoption, reinforcing productivity gains.

In the MENA context, these mechanisms are particularly relevant due to binding environmental constraints, especially water scarcity. Economic growth enables the diffusion of water-saving technologies (e.g., drip irrigation) and climate-resilient crops, which reduce vulnerability to environmental shocks (Hejazi et al., 2023 [27]). Thus, GDP per capita operates as a structural enabling factor rather than a purely demand-driven variable.

However, this relationship must be interpreted within the broader framework of structural change. While economic growth may shift labor and capital away from agriculture, the results suggest that productivity gains dominate reallocation effects in MENA. This indicates a transition toward efficiency-driven agricultural development rather than simple sectoral decline.

6.2. Inflation and Agricultural Output

The positive association between inflation and agricultural value added is best understood through price transmission mechanisms rather than productivity effects. In agricultural markets, especially in partially regulated or less integrated economies, increases in output prices may temporarily raise measured value added if producer prices adjust faster than input costs (Swinnen, 2010 [4]).

However, the absence of significance in short-run specifications highlights the instability and non-linear nature of inflation effects. Inflation volatility may distort production decisions, increase uncertainty, and raise the cost of key inputs such as fertilizers and energy (Anríquez & Daidone, 2010 [5]). This suggests that inflation reflects short-term valuation effects rather than a stable structural determinant.

Accordingly, the results reinforce the need to distinguish between nominal valuation effects and real productivity changes, which is often overlooked in macro-level analyses.

6.3. Clean Energy Finance and Agricultural Transformation

The positive association between international clean energy finance (LGF) and agricultural value added can be interpreted through a structural constraint relaxation mechanism. Specifically, in energy-constrained environments, agricultural production is limited by insufficient access to reliable and affordable energy for irrigation, storage, and processing. Clean energy finance alleviates these constraints by enabling infrastructure investment, thereby increasing the marginal productivity of complementary inputs such as land and labor (Jalilov et al., 2018 [7]; ElZein et al., 2022 [8]; Alharbi et al., 2023 [13]). However, this mechanism operates conditionally. In energy-abundant (oil-exporting) economies, where energy constraints are not binding, additional financial inflows toward clean energy generate limited incremental gains, resulting in lower marginal returns. This explains the observed heterogeneity and highlights the importance of context-dependent policy effectiveness (Al Mamun et al., 2022 [14]).

Beyond statistical significance, the magnitude of the estimated coefficients also indicates economically meaningful implications for agricultural performance. The positive coefficient of clean energy finance suggests that increases in renewable-energy-related financial flows are associated with measurable improvements in agricultural value added, particularly in non-oil economies where energy shortages constitute a major production constraint. Although the estimated elasticity remains moderate, the cumulative impact becomes economically relevant over time through improvements in irrigation efficiency, rural electrification, storage capacity, and agricultural processing infrastructure (Burney et al., 2010 [9]; Wang & Xu, 2025 [10]).

The interaction model provides formal econometric support for this mechanism. The estimated interaction term between clean energy finance and the oil-exporting country dummy is negative and statistically significant, confirming that the marginal effect of clean energy finance is substantially weaker in energy-abundant economies. This finding supports the presence of a diminishing marginal returns mechanism; whereby additional energy-related investments generate lower productivity gains when basic energy constraints are already alleviated. This interpretation is consistent with structural heterogeneity literature emphasizing that the effectiveness of sustainability-oriented investments depends critically on country-specific economic structures, institutional quality, and resource endowments (Malah Kuete & Asongu, 2022 [28]; Jalilov et al., 2018 [6]; Laderach et al., 2022 [29]).

Clean energy investments—particularly in decentralized solar systems—reduce energy constraints in rural areas, enabling irrigation, storage, and processing activities that are critical for agricultural productivity (Alaofè et al., 2016 [30]; Burney, Naylor, & Postel, 2013 [31]). Moreover, renewable-energy infrastructure can improve resilience against climate-related disruptions while reducing dependence on conventional energy sources, thereby contributing to long-term agricultural sustainability (ElZein et al., 2024 [7]; Mrabet, 2023 [32]).

The negative short-run effect observed in PCSE estimates can be explained by adjustment and investment reallocation mechanisms. Large-scale energy projects are capital-intensive and may initially divert financial resources away from immediate agricultural inputs, leading to temporary output reductions (Prasad et al., 2025 [33]). This reflects a temporal mismatch between investment costs and delayed productivity gains.

Overall, both the subsample and interaction-model results consistently demonstrate that the effectiveness of clean energy finance is structurally conditional rather than universally applicable across MENA economies. The positive contribution of clean energy finance is concentrated primarily in non-oil economies, where energy scarcity remains a binding constraint on agricultural productivity.

6.4. Trade Openness and Sectoral Vulnerability

The negative relationship between trade openness and agricultural value added can be interpreted through competitive pressure and structural vulnerability mechanisms. In many MENA countries, domestic agriculture faces structural disadvantages, including low productivity, fragmented landholdings, and limited technological adoption. Trade liberalization exposes these sectors to international competition, often reducing domestic output (Anderson, 2009 [6]).

Additionally, trade openness may induce resource reallocation toward more competitive sectors, accelerating structural transformation away from agriculture. This finding is consistent with structural transformation theories emphasizing that developing economies frequently experience declining agricultural shares as labor and capital move toward industry and services during economic modernization (Diao et al., 2010 [34]; Timmer, 2009 [35]; McCullough, 2024 [36]). Thus, the observed relationship reflects structural adjustment processes rather than an inherently adverse role of trade.

Accordingly, the findings suggest that the impact of trade is conditional on domestic competitiveness and policy support, rather than uniformly negative.

6.5. Arable Land and Institutional Quality

The weak and inconsistent effects of arable land (LAN) suggest that agricultural performance in MENA is increasingly driven by productivity rather than factor expansion. Given binding land and water constraints, expansion-based growth strategies are limited, reinforcing the importance of technological intensification (Abou Zaki et al., 2022 [37]; Mrabet, 2023 [33]).

Similarly, the limited short-run effect of institutional quality (IQR) reflects delayed transmission mechanisms. Institutional improvements affect agriculture indirectly and over longer time horizons (Robinson & Acemoglu, 2012 [38]). Recent evidence further suggests that institutional quality influences agricultural value added primarily through governance efficiency, policy implementation, and investment climate improvements rather than immediate short-run adjustments (Gelgo et al., 2023 [39]; Barszczewski, 2024 [40]). Short-term variations may therefore fail to capture their cumulative impact.

6.6. Synthesis and Contribution

Taken together, the findings indicate that agricultural performance in MENA is primarily shaped by structural and macroeconomic conditions rather than traditional factor endowments. Economic development acts as a key enabling mechanism, while trade openness introduces adjustment pressures that may weaken domestic agriculture in the absence of supportive policies. Clean energy finance contributes positively, but its effects are conditional, heterogeneous, and time dependent.

This study contributes to the literature by providing macro-level empirical evidence that the effectiveness of clean energy finance depends critically on country-specific structural conditions. Unlike much of the existing literature—largely based on micro-level or single-country analyses, this study demonstrates that in MENA economies, the interaction between energy constraints, financial flows, and agricultural systems produces differentiated outcomes.

These findings are consistent with recent studies emphasizing the heterogeneous nature of energy-transition dynamics and sustainability outcomes across resource-abundant and resource-constrained economies (Abid, 2026 [41]).

The identification of a structural asymmetry between oil and non-oil economies represents a central contribution, showing that energy-transition finance enhances agricultural performance only where energy scarcity constitutes a binding constraint. This provides a more context-sensitive and policy-relevant framework for understanding agricultural transformation in resource-constrained environments.

6.7. Methodological Limitations

Despite the robustness of the empirical framework, several limitations should be acknowledged. First, the Kao, Pedroni, and Westerlund cointegration tests provide mixed and weak evidence of long-run equilibrium relationships. Therefore, the study focuses primarily on structural associations and medium-run dynamics rather than definitive long-run equilibrium effects.

Second, the Dumitrescu–Hurlin panel Granger causality test identifies predictive temporal relationships rather than strict structural causality. Accordingly, the findings should be interpreted as evidence of temporal linkages rather than conclusive causal mechanisms.

Third, the short-run PCSE estimations show that clean energy finance exerts a negative and significant short-run effect, likely reflecting transitional adjustment costs, resource reallocation, and implementation frictions associated with clean energy investments. This suggests that the benefits of clean energy finance may emerge gradually rather than instantaneously.

Overall, the findings support a nuanced interpretation in which the effectiveness of clean energy finance depends on structural conditions and the time horizon considered.

7. Conclusions

This study provides macro-level evidence that the relationship between clean energy finance and agricultural performance in MENA is conditional, heterogeneous, and time dependent. Rather than functioning as a direct sectoral input, clean energy finance operates as a context-dependent and structurally mediated mechanism, within the energy–water–food nexus, with effects that depend on underlying energy constraints and economic structures. By distinguishing between short-run adjustments and medium-run structural associations, the analysis explains why energy-transition finance may not generate immediate agricultural gains but can produce gradual improvements over time.

The empirical results highlight economic development, clean energy finance, and trade openness as key drivers of agricultural value added. GDP per capita emerges as a central enabling factor, supporting productivity through infrastructure development and technological adoption. Clean energy finance shows a positive association with agricultural performance; however, both subsample and interaction-based evidence indicate that this effect is primarily driven by non-oil economies, where energy constraints are binding. In contrast, the weaker or insignificant effects observed in oil-exporting countries point to diminishing marginal returns in energy-abundant contexts.

The interaction model provides formal confirmation of this result. Specifically, the negative and statistically significant interaction term indicates that the marginal effect of clean energy finance is substantially attenuated in oil-exporting economies. This finding reinforces the presence of a structural constraint mechanism and confirms that the effectiveness of clean energy finance depends critically on underlying energy conditions, rather than operating as a universally effective policy instrument. This interpretation is consistent with recent literature emphasizing asymmetric energy-transition dynamics and heterogeneous sustainability effects across oil-dependent and non-oil economies (Chaabouni & Abid, 2025 [42]).

Trade openness exhibits a negative relationship, pointing to structural vulnerabilities in domestic agriculture when exposed to international competition.

Inflation displays a positive association in level specifications, which should be interpreted as a nominal price effect rather than a real productivity gain. Arable land and institutional quality do not show robust short-run effects, suggesting that agricultural performance in MENA is increasingly shaped by structural and macroeconomic conditions rather than traditional factor endowments. Moreover, weak cointegration evidence indicates that short- to medium-run dynamics dominate long-run equilibrium relationships, while Granger causality results confirm that agricultural outcomes are primarily driven by macroeconomic and structural factors.

From a policy perspective, the findings emphasize the importance of targeted and context-specific strategies. Investments in decentralized clean energy infrastructures such as solar irrigation and cold storage can enhance agricultural productivity, particularly in energy-constrained economies. However, policies must be differentiated across countries, as the effectiveness of clean energy finance varies with energy endowments. Trade liberalization should be complemented by measures that strengthen agricultural competitiveness, including access to finance, rural infrastructure, and technological upgrading. More broadly, sustained and inclusive economic growth remains a fundamental condition for agricultural modernization, while improvements in governance and institutional frameworks are essential for long-term sustainability. Inflation should not be considered a policy tool, and macroeconomic stability remains a priority.

Despite the use of robust econometric techniques, several limitations should be acknowledged. Data constraints limit the inclusion of climate-related variables, and potential endogeneity—particularly involving GDP and energy finance—cannot be fully eliminated, although partially addressed through robustness checks including System GMM. In addition, structural heterogeneity across countries may extend beyond the oil vs. non-oil classification, and recent global shocks may introduce short-term distortions. Future research should incorporate climate indicators, explore nonlinear and interaction effects, and extend dynamic modeling frameworks to better capture the complexity of energy–agriculture linkages.

This study ultimately demonstrates that clean energy finance should be understood as a structurally conditional mechanism rather than a universally effective policy tool, with its impact fundamentally shaped by energy constraints, economic structure, and adjustment dynamics across countries. These findings further support recent evidence showing that renewable-energy transitions and sustainability-oriented investments generate differentiated outcomes depending on countries’ structural characteristics, energy dependence, and macroeconomic conditions (Hechmi & Abid, 2026 [43]).