Asymmetric Effects of Digital Trade on Environmental Sustainability: Evidence from GCC Economies

Abstract

Rapid digital transformation is reshaping global trade and raising important questions about its environmental impact, particularly in energy-intensive GCC economies. Despite growing interest, existing evidence remains inconclusive and often overlooks potential nonlinear effects. This study explores how digital trade influences environmental sustainability in Gulf Cooperation Council (GCC) countries over the period 2010–2024. Using a balanced panel dataset for the six economies, the analysis applies a fixed-effects approach with Driscoll–Kraay standard errors to account for cross-sectional dependence and other econometric concerns. To better capture the complexity of the relationship, the study also adopts an asymmetric framework that distinguishes between positive and negative changes in digital trade. The findings show that digital trade does not have a significant effect in the linear model. However, once asymmetry is considered, a clearer pattern emerges. Increases in digital trade are associated with lower CO₂ emissions, while decreases tend to raise emissions. Energy consumption remains the primary driver of emissions, while technological readiness helps reduce environmental pressure. Urbanization and political stability, on the other hand, are linked to higher emissions, reflecting ongoing structural challenges in the region. Overall, the results highlight the importance of sustaining digital trade growth and strengthening technological capabilities to support environmental sustainability in GCC economies.

1. Introduction

Climate change remains one of the most pressing challenges of the twenty-first century. The continuous rise in greenhouse gas emissions, particularly carbon dioxide (CO₂), has raised serious concerns about environmental degradation, global warming, and long-term sustainability. As economies continue to grow and global trade becomes more interconnected, policymakers face the difficult task of balancing economic development with environmental protection. In this context, increasing attention has been given to the role of digital transformation and technological innovation in supporting sustainable development and reducing environmental pressure [1].

Over the past two decades, digital trade has expanded rapidly and become a key feature of the global economy. It includes a wide range of activities enabled by digital technologies, such as e-commerce, digital services, online marketplaces, and cross-border data flows. These developments have reshaped traditional trade by improving access to international markets, lowering transaction costs, and increasing the efficiency of production and distribution. As a result, digital trade has emerged as an important driver of economic growth and global integration. Recent global evidence highlights the scale of digital transformation, with global e-commerce sales reaching approximately USD 27 trillion in 2022 [2].

At the same time, digital trade has important implications for environmental sustainability. On the one hand, digital technologies can help reduce emissions by improving efficiency, optimizing supply chains, and encouraging cleaner production processes [3,4,5]. On the other hand, the expansion of digital infrastructure, such as e-commerce and logistics, may raise energy demand and electricity consumption, potentially leading to higher emissions [6,7]. This mixed impact has led to an ongoing debate in the literature about whether digital trade ultimately benefits or harms the environment, with recent studies providing evidence of both emission-reducing and emission-increasing effects depending on the economic and technological context [6,8].

A growing number of studies have explored the relationship between digitalization and environmental outcomes. Some argue that digital trade contributes to emission reductions through innovation and better resource allocation, while others contend that it may initially increase emissions due to higher energy demand associated with digital infrastructure [2,9]. Recent empirical evidence highlights that these effects are often heterogeneous and may vary across regions and development levels [6,10,11]. Moreover, emerging studies suggest that digital trade may influence environmental outcomes indirectly through technological innovation, energy efficiency, and trade-related regulatory frameworks [12,13].

In addition to economic conditions, institutional conditions also play an important role in shaping environmental outcomes. The design and implementation of environmental policies, as well as the ability to support investment in sustainable technologies, depend on the quality of governance and political stability. Countries with stronger institutions are generally better positioned to promote long-term environmental sustainability.

These issues are particularly relevant for GCC economies, including Saudi Arabia, which are undergoing rapid digital transformation while facing increasing environmental challenges. As part of its Vision 2030 development strategy, Saudi Arabia has invested heavily in digital infrastructure, technological innovation, and economic diversification [14]. However, the region remains highly dependent on energy-intensive activities, making environmental sustainability a key concern. This situation makes the GCC an important context for examining whether digital trade can contribute to improving environmental performance.

Given this background, the present study examines the relationship between digital trade and carbon emissions while accounting for key economic and institutional factors. Specifically, it considers economic growth, energy consumption, urbanization, and political stability to provide a more comprehensive understanding of the drivers of environmental performance. Accordingly, the study addresses the following question: Does digital trade improve environmental quality, and does its impact differ between positive and negative changes in digital trade?

Despite the growing body of research, several gaps remain. First, most existing studies focus on developed economies or large emerging markets, while GCC countries have received relatively limited attention. Second, previous research often examines the direct link between digitalization and emissions without fully considering institutional factors such as political stability [1]. Third, many studies rely on linear approaches, which may overlook the possibility that positive and negative changes in digital trade have different environmental effects, despite recent evidence suggesting that the relationship between digital trade and environmental outcomes may be nonlinear and asymmetric [11].

This study addresses these gaps in three ways. First, it provides new evidence from GCC economies, a region that remains underexplored in the digital sustainability literature. Second, it adopts an asymmetric approach that allows positive and negative changes in digital trade to have different impacts on CO₂ emissions. Third, it brings together economic and institutional factors within a single empirical framework, offering a more complete understanding of the drivers of environmental performance. The remainder of this paper is organized as follows: Section 2 reviews the relevant literature. Section 3 outlines the data and empirical methodology. Section 4 presents and discusses the empirical results. Finally, Section 5 concludes with policy implications and study limitations.

2. Literature Review

2.1. Digital Trade and Carbon Emissions

Academic interest in the relationship between digital trade and environmental sustainability has grown significantly in recent years. As digital technologies expand and international trade increasingly relies on e-commerce platforms, digital services, and online systems, researchers have begun to examine whether digital trade helps improve environmental quality or contributes to environmental pressure. Existing literature suggests that digital trade can influence carbon emissions through several channels, including improved supply chain management, technological innovation, and more efficient production and transportation systems [15,16,17].

A large number of empirical studies point to the environmental benefits of digital trade. For example, Li et al. [18], using data from 46 countries, find that digital trade reduces carbon emissions by improving resource allocation and increasing production efficiency. Digital platforms can lower transaction costs and enhance the performance of global supply chains, which in turn supports more efficient energy use. Similarly, Wang et al. [19] and Hu et al. [20] argue that digital trade and technological innovation contribute to emissions reductions by encouraging the adoption of cleaner technologies and more energy-efficient production processes.

Cross-country evidence further supports these findings. Ma et al. [21] show that the digital economy reduces trade-adjusted carbon emissions by improving technological efficiency. In a similar vein, Ke [22] finds that digitalization, when combined with financial development, helps reduce carbon emissions in developing countries. Other studies also suggest that digital trade improves productivity and economic efficiency, which can lower carbon intensity and enhance environmental performance [23].

However, the existing evidence is not entirely consistent. While many studies highlight the environmental benefits of digital trade, others point to a potential increase in energy demand associated with digital infrastructure, particularly in economies that rely heavily on fossil fuels. This suggests that the environmental impact of digital trade is context-dependent and may vary across countries depending on their level of development, energy structure, and technological capacity. Overall, while there is a broad consensus that digital trade can improve environmental performance through efficiency gains, the evidence remains mixed, likely due to differences in countries’ energy structures, levels of digital maturity, and methodological approaches across studies.

2.2. Digital Trade, Innovation, and Environmental Sustainability

The relationship between digital trade and environmental sustainability is closely linked to technological advancement and strength of institutional frameworks. A growing body of literature suggests that digital trade supports more sustainable production by encouraging green innovation and improving financial development. For example, Fu et al. [24] show that digital trade contributes to both financial development and ecological innovation, which are key to enhancing environmental sustainability. In a similar context, Han et al. [25] discover that digital trade promotes green innovation in the BRICS economies, leading to noticeable improvements in environmental quality.

At the same time, institutional frameworks and trade agreements play an important role in shaping how digital trade affects the environment. Li et al. [11] show that digital trade provisions in regional trade agreements can help reduce CO₂ emissions by improving regulatory transparency and encouraging cooperation across countries. Similarly, Liu and Gao [12] demonstrate that digital trade regulations support cleaner production methods and reduce export-related carbon emissions. In addition, Qiu and Wan [26] highlight that green innovation strengthens the positive impact of digital trade on environmental sustainability. Supporting this, Ma et al. [27] demonstrate that digital services trade can improve carbon productivity through technological progress and structural changes.

Beyond innovation and institutions, digital technologies also contribute to environmental sustainability by improving logistics and supply chain efficiency. Ji et al. [28] find that digital trade reduces regional carbon emissions by enhancing logistics performance and resource allocation. Similarly, Zhu et al. [8] report that digital trade supports more efficient production systems and sustainable consumption patterns, leading to lower emissions. Wen and Zhu [29] further show that digital trade promotes low-carbon development in the transportation sector by optimizing logistics and reducing fuel consumption. Evidence from the European Union also confirms that digital trade lowers the environmental costs associated with traditional trade processes [30]. In summary, although the literature largely agrees on the positive role of digital trade in fostering innovation and sustainability, variations in institutional quality and regulatory frameworks may explain the differing magnitude of its environmental impact across countries.

2.3. Nonlinear Effects and Environmental Risks of Digital Trade

While a growing body of research highlights the environmental benefits of digital trade, several studies suggest that this relationship is not always straightforward. In particular, digital trade may have a nonlinear or conditional effect on carbon emissions. For example, Kwilinski [31] argues that the early stages of digital expansion may increase energy consumption due to rising demand for digital infrastructure and economic activity. However, as digital systems mature and become more efficient, their environmental impact may improve, leading to a reduction in emissions over time.

Similar patterns are reported in other empirical studies. Chen and Jiang [32] find that the effect of digital service trade on carbon emissions depends on a country’s economic structure and level of technological advancement. Likewise, Zhou and Guo [33] demonstrate that the environmental impact of digital trade varies across cities and regions depending on their degree of digital development. Supporting this view, Zhang [34] identifies a threshold effect, suggesting that digital transformation begins to reduce emissions only after a certain level of development is reached.

The literature also highlights significant differences across countries. Li et al. [18] show that although digital trade generally contributes to lower carbon emissions, its impact varies depending on each country’s stage of development. In addition, İmamoğlu [35] points out that the expansion of digital technologies may initially increase electricity consumption due to the growing number of data centers and the rapid development of digital infrastructure.

Overall, these findings suggest that the relationship between digital trade and carbon emissions is not only nonlinear but also potentially asymmetric. Increases in digital trade may enhance efficiency and reduce emissions, while declines may disrupt production processes and increase environmental pressure. Such differences cannot be fully captured using a standard linear framework, highlighting the importance of adopting an empirical framework that allows for asymmetric effects. Taken together, the literature suggests a general agreement on the presence of nonlinear and conditional effects, while discrepancies in findings can be attributed to differences in development stages, threshold levels, and model specifications used in empirical analyses.

2.4. Digital Trade and Political Stability

Political stability plays an important role in shaping how effectively digital trade can support environmental sustainability. In stable governance, policymakers are better able to enforce environmental regulations and implement long-term strategies for renewable energy and green innovation. Stability also encourages private investment in cleaner technologies and sustainable practices. In addition, governments in such settings are more capable of developing and maintaining the digital infrastructure required for low-carbon economic activities.

A growing body of research highlights the importance of governance quality and institutional stability in determining the environmental impact of digital transformation. For example, Ahmad and Satrovic [36] find that countries with stronger institutional quality and political stability tend to experience lower carbon emissions, as effective governance improves policy enforcement. Similarly, Yang et al. [37] emphasize that strong legal systems and consistent governance frameworks are essential for achieving carbon neutrality and long-term sustainability goals.

These findings suggest that the environmental benefits of digital trade are closely linked to institutional conditions, which supports the inclusion of political stability as a key control variable in the empirical analysis. Overall, there is strong consensus that political stability enhances the environmental benefits of digital trade, although differences in governance effectiveness and policy implementation capacity may account for variations in observed outcomes across countries.

2.5. Digital Trade and Sustainable Development in Emerging and GCC Economies

Recent studies have increasingly examined the role of digital trade in promoting sustainable development in developing and emerging economies. Pariyar et al. [38] show that digital trade improves environmental quality by supporting technology adoption and enhancing production efficiency. Similarly, Djermoun et al. [39] find that digital trade contributes to both environmental performance and sustainable economic growth in emerging economies. In a related context, Liu and Wang [40] demonstrate that trade in digital services enhances carbon total factor productivity, suggesting that economies can improve efficiency while reducing emissions. Evidence from regional economic groups, such as RCEP countries, also confirms that digital trade supports environmental sustainability through technological progress and trade efficiency [41].

The importance of digitalization is particularly evident in GCC economies. Asmyatullin and Glavina [42] show that digitalization plays a key role in improving energy efficiency and supporting green economic transformation in the region. Similarly, Touati and Ben-Salha [43] find that digitalization, alongside industrialization and financial development, has a significant impact on long-term environmental sustainability in GCC economies. However, despite these contributions, existing studies on GCC countries remain limited and largely rely on linear approaches, leaving the asymmetric effects of digital trade on environmental outcomes insufficiently explored, which represents an important gap that this study aims to address.

Overall, the literature suggests that digital trade can influence environmental outcomes through multiple interconnected channels, including efficiency gains, technological innovation, and institutional quality. In this study, these mechanisms are reflected in the selection of key variables, where energy consumption captures the scale and efficiency effects, technological readiness (FTRI) represents innovation capacity, and political stability reflects the institutional environment.

It is important to note that the proxy used for digital trade, namely digitally deliverable services, primarily captures the intangible dimension of digitalization and may not fully reflect the environmental impact of digitalized goods trade and e-commerce logistics. While this measure is widely adopted due to data availability and cross-country comparability, it may underestimate certain emission channels related to physical trade activities.

To provide a clearer analytical foundation, this study draws on the environmental economics literature that links economic activity, technological change, and institutional factors to environmental outcomes. In this context, digital trade can affect carbon emissions through several interconnected channels. From an efficiency perspective, digital trade can reduce transaction costs, improve supply chain management, and enhance resource allocation, which may help lower energy use and emissions. From a technological perspective, digitalization supports innovation and encourages the adoption of cleaner and more energy-efficient production processes. In addition, from a structural and institutional perspective, the environmental impact of digital trade depends on factors such as economic structure, energy dependence, and governance quality, which influence how digitalization translates into environmental performance. Together, this framework provides the basis for examining the asymmetric effects of digital trade on carbon emissions.

3. Data and Methodology

3.1. Data

This study uses a balanced panel dataset covering the six GCC countries: Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and the United Arab Emirates, for the period 2010–2023. The data were obtained from the World Bank’s World Development Indicators (WDI) and the United Nations Conference on Trade and Development (UNCTAD). Both sources are widely recognized in empirical research and provide consistent and comparable indicators across countries, which enhances the reliability of the analysis.

The study period was selected based on the availability of complete and consistent data for all variables across the six GCC countries. This makes it possible to construct a balanced panel dataset and avoids potential biases that may arise from missing observations or the need to interpolate data. In addition, the use of annual data follows common practice in cross-country environmental and macroeconomic research, where annual observations are typically used to examine long-run patterns and dynamic relationships among variables.

Table 1. Variables definition and data sources.

Variable	Abbreviation	Definition	Source
The Dependent variable
	CO2	Carbon dioxide (CO2) emissions excluding LULUCF per capita (t CO2e/capita)	WB-WDI
The Independent variable
	DigTrade	International trade in digitally deliverable services (%_total_trade_in_services)	UNCTAD
The Control Variables
	GDP	GDP per capita (constant 2015 US$)	WB-WDI
	Euse	Energy use (kg of oil equivalent per capita)	WB-WDI
	FTRI	Frontier Technology Readiness Index	UNCTAD
	URB	Urban population (% of total population)	WB-WDI
	PS	Political Stability and Absence of Violence/Terrorism: Estimate	WB-ESG

presents the variables used in this study, including their definitions, measurement units, and corresponding data sources.

3.2. Methodology

To analyze the relationship between digital trade and carbon emissions in GCC economies, this study adopts a panel time-series econometric methodology. The methodology follows a progressive empirical strategy comprising preliminary diagnostics, unit root testing, and dynamic panel estimation.

First, descriptive statistics and pairwise correlation analysis are undertaken to understand the basic properties of the data and detect potential multi-collinearity among variables. Second, cross-sectional dependence tests are undertaken because macroeconomic panels, including countries, are often affected by shared shocks such as oil price variations, regional economic integration, or global technological advances. Ignoring such reliance may lead to biased and inconsistent estimators [44].

To discover cross-sectional dependence, numerous complementary tests are employed, including Pesaran’s CD test, the Breusch–Pagan LM test, the Friedman test, and Frees’ test. These tests offer a detailed examination of whether the error terms across countries are connected [45].

Following the detection of cross-sectional dependence, the stationarity features of the variables are investigated using second-generation panel unit root tests. Specifically, the Cross-Sectional Augmented Dickey–Fuller (CADF) and the Cross-Sectional Augmented IPS (CIPS) tests developed by Pesaran [46] are applied. These tests account for cross-sectional dependence by adding cross-sectional averages into the unit root regressions and are consequently more reliable than first-generation tests in panels where cross-country correlations exist.

3.2.1. Model Specification

To evaluate the relationship between digital trade and environmental quality in GCC economies, this study utilizes a panel data framework that accounts for country-specific variability. Panel data approaches are particularly beneficial when evaluating macroeconomic interactions between countries because they allow researchers to control for unobserved characteristics that may fluctuate across cross-sectional units but remain relatively stable across time [47,48]. The baseline empirical model is specified as follows:

$$ln{CO}_{2it}={\alpha }_i+{\beta }_1{lnDigTrade}_{it}+{{\beta }_2{lnGDP}_{it}+\beta }_3{lnEUse}_{it}+{\beta }_4{FTRI}_{it}{+\beta }_5{lnURB}_{it}+{\beta }_6{lnPS}_{it}+{\epsilon }_{it}$$

(1)

where $i$ denotes the country and t represents the time period. LnCO₂ represents carbon dioxide emission per capita and serves as the dependent variable. The key explanatory variable is digital trade (lnDigTrade), while additional control variables include economic growth (lnGDP), energy consumption (lnEUse), frontier technology readiness index (FTRI), urbanization (lnURB), and political stability (PS). The term (${\alpha }_i$) captures country-specific fixed effects, while (${\epsilon }_{it}$) represents the error term.

These control variables are commonly employed in the environmental economics literature to capture economic activity, energy consumption, technological progress, and institutional settings that may influence environmental outcomes [1,3].

3.2.2. Asymmetric Effect of Digital Trade

While many empirical studies assume that economic variables affect environmental outcomes in a symmetric way, more recent research suggests that positive and negative shocks may have different impacts [49]. In the case of digital trade, increases and decreases can influence carbon emissions through various channels, such as improvements in efficiency, the substitution of digital services, and changes in production and logistics processes.

To capture these differences, this study adopts an asymmetric approach based on partial sum decomposition. Specifically, digital trade is separated into positive and negative changes as follows:

$${DigTrade}_{it}\left(+\right)=\max \left({∆DigTrade}_{it},0\right)$$

$${DigTrade}_{it}\left(-\right)=\mathrm{m}\mathrm{i}\mathrm{n}({∆DigTrade}_{it},0)$$

The symmetric specification is therefore given as follows:

$$ln{CO}_{2it}={\alpha }_i+{\beta }_1{DigTrade}_{it}\left(+\right)+{\beta }_2{DigTrade}_{it}\left(-\right)+{{\beta }_3{lnGDP}_{it}+\beta }_4{lnEUse}_{it}+{\beta }_5{FTRI}_{it}{+\beta }_6{lnURB}_{it}+{\beta }_7{lnPS}_{it}+{\epsilon }_{it}$$

(2)

This approach enables us to test whether positive and negative changes in digital trade have asymmetric effects on carbon emissions. It is important to note that the implementation of the asymmetric model involves first differencing the digital trade variable, which results in a loss of one time observation per country. Consequently, the number of observations decreases from 90 in the linear model to 84 in the asymmetric specification.

3.2.3. Model Selection

To find the optimal panel estimator, both fixed-effects (FE) and random-effects (RE) models are calculated. The Hausman [50] specification test is then utilized to evaluate whether country-specific effects are linked with the explanatory variables. The test findings reject the null hypothesis of no connection, showing that the fixed-effects estimator offers more consistent estimates. Therefore, the fixed-effects model is used as the baseline definition.

The choice of an appropriate estimator is especially crucial in panel datasets with a small cross-sectional dimension and a moderate temporal period, as in our work. The panel comprises six GCC countries studied from 2010 to 2024, yielding a small-N, moderate-T structure. In such cases, conventional estimators may provide biased results if the error components exhibit cross-sectional dependency, heteroscedasticity, or serial correlation. The diagnostic tests used in the study demonstrate the presence of cross-sectional dependence, groupwise heteroscedasticity, and first-order serial correlation. Ignoring these econometric concerns could result in inconsistent standard errors and inaccurate statistical significance. To address these concerns, the study adopts Driscoll–Kraay standard errors, which allow robust inference even when spatial and temporal dependence exists between panel units.

The Driscoll–Kraay estimator is particularly useful for macro-panel datasets since it corrects the covariance matrix for heteroscedasticity, autocorrelation, and cross-sectional dependency all at once. Unlike typical clustered standard errors, this approach is dependable when cross-sectional units are interdependent, which is prevalent in economically associated regions like the Gulf Cooperation Council (GCC). Furthermore, earlier research has shown that Driscoll–Kraay standard errors work well in panels with few cross-sectional units and moderate time dimensions [51,52]. Therefore, using the fixed-effects model with Driscoll–Kraay standard errors ensures more reliable statistical inference and supports the empirical findings provided in this work.

3.2.4. Robustness Tests

To confirm the veracity of the empirical findings, many robustness checks are done. First, the baseline model is re-estimated in the Driscoll–Kraay estimator with different lag structures to test the sensitivity to serial correlation assumptions. Second, temporal fixed effects are used to account for common shocks that hit all countries concurrently. Finally, a leave-one-country-out technique is used to ensure that the outcomes are not influenced by a single GCC country.

The following section reports the empirical findings of the study. The analysis begins with descriptive statistics and correlation results to provide an initial overview of the dataset. Next, several diagnostic tests are performed to assess cross-sectional dependence and the stationarity properties of the variables. The baseline fixed-effects model is then estimated, followed by an asymmetric specification that captures the potentially different environmental impacts of positive and negative digital trade shocks. To further ensure the reliability of the results, several robustness checks are conducted, including alternative Driscoll–Kraay lag specifications, the inclusion of time fixed effects, and a leave-one-country-out analysis.

4. Results and Discussion

This section presents and discusses the empirical findings of the study. The analysis begins with descriptive statistics to provide an overview of the distribution and variability of the variables across GCC countries. The correlation matrix is then examined to identify preliminary relationships between the key variables and assess potential multicollinearity concerns. Next, a series of diagnostic tests are conducted, including cross-sectional dependence and panel unit root tests, to ensure the statistical validity of the panel dataset. Following these preliminary analyses, the baseline panel regression results are reported based on the fixed effects estimator. Building on the baseline specification, an asymmetric model is estimated to investigate whether positive and negative shocks in digital trade exert different effects on environmental outcomes. Finally, several robustness checks are performed using alternative Driscoll–Kraay lag structures, time fixed effects, and a leave-one-country-out approach to verify the stability and reliability of the results.

Descriptive statistics for the variables used in the analysis are presented in

Table 2. Descriptive Statistics.

Variable	Obs.	Mean	Std. Dev.	Min	Max
lnCO2	90	3.2082	0.3177	2.8865	3.9815
lnDigTrade	90	7.9749	1.4324	5.1475	10.8061
lnGDP	90	25.7518	1.0093	24.0303	27.5046
lnEUse	90	9.1397	0.3257	8.5378	9.9738
FTRI	90	0.5867	0.0767	0.4	0.8
lnURB	90	4.5040	0.1020	4.3196	4.6052
PS	90	0.2009	0.6526	−1.3354	1.2236

. The results show moderate variation across the panel, reflecting differences in economic activity, energy use, and institutional conditions among GCC countries. In particular, the relatively higher dispersion in digital trade and political stability suggests heterogeneous development patterns across the region, which justifies the use of a panel estimation framework that accounts for country-specific effects.

Table 3. Correlation Matrix.

Variable	lnCO2	lnDigTrade	lnGDP	lnEUse	FTRI	lnURB	PS
lnCO2	1.0000
lnDigTrade	−0.0012	1.0000
lnGDP	−0.1342	0.2134	1.0000
lnEUse	0.9387	0.2050	−0.1168	1.0000
FTRI	−0.4078	0.5794	0.4530	−0.2482	1.0000
lnURB	0.6343	0.3171	−0.4316	0.7503	−0.1688	1.0000
PS	0.4102	−0.0304	0.1700	0.2196	−0.1718	−0.2328	1.0000

presents the correlation matrix among the study variables. The results indicate that most variables exhibit moderate correlations, suggesting no serious multicollinearity concerns. CO₂ emissions show a strong positive correlation with energy use and urbanization, highlighting the central role of energy consumption and urban development in shaping environmental outcomes in GCC economies.

Table 4. Cross-Section Dependence Tests: Variable Level.

Variable	CD Statistics	p Value
lnCO2	5.90	0.000
lnDigTrade	11.48	0.000
lnGDP	13.91	0.000
lnEUse	3.37	0.001
FTRI	8.24	0.000
lnURB	-	-
PS	1.88	0.060

reports the results of the cross-sectional dependence tests. The Pesaran CD statistics indicate significant cross-sectional dependence for most variables, suggesting that shocks affecting one GCC country may spill over to others. The CD statistics for urbanization (lnURB) could not be computed due to limited cross-sectional variation in the variable across the GCC countries. Overall, these findings support the use of second-generation panel techniques that account for cross-sectional dependence in the empirical analysis.

Before proceeding with the estimation, the stationarity properties of the variables are examined to ensure the validity of the empirical approach. The findings of the CIPS and CADF panel unit root tests in

Table 5. Unit Root Tests.

	CIPS		CADF
Variable	Level	First Dif.	Level	First Dif.
lnCO2	−2.023	−3.026 ***	−2.261	−3.416 ***
lnDigTrade	−3.090 ***	−4.284 ***	−3.859 ***	−3.759 ***
lnGDP	−2.779 **	−3.654 ***	−2.018	−2.817 ***
lnEUse	−2.565 **	−4.107 ***	−2.456 **	−2.477 **
FTRI	−3.094 ***	−4.114 ***	−2.583 **	−3.011 ***
lnURB	−1.938	0.777	−0.222	0.817
PS	−1.454	−2.568 **	−1.701	−1.980

Notes: *** p < 0.01, ** p < 0.05.

indicate that the variables exhibit mixed orders of integration, with most variables being stationary at levels or after first differencing. In contrast, lnURB appears to be non-stationary, likely reflecting its gradual and persistent evolution over time. Such behavior is not uncommon for structural variables in macro-panel datasets.

Given that the primary objective of this study is to examine asymmetric short-run responses, the empirical analysis employs a fixed-effects model with Driscoll–Kraay standard errors. This approach provides robust inference in the presence of cross-sectional dependence, heteroskedasticity, and serial correlation. In addition, the asymmetric specification, based on decomposing digital trade into positive and negative changes, is designed to capture short-run dynamics rather than long-run equilibrium relationships.

Table 6. Baseline Panel Estimation and Model Selection.

Variables	FE		RE
	Coefficient	SE	Coefficient		SE
lnDigTrade	0.0122	0.0207	−0.0419 ***		0.0073
lnGDP	−0.2377 **	0.0884	0.0080		0.0096
lnEUse	0.5393 ***	0.0886	0.8234 ***		0.0420
FTRI	−0.2701 **	0.1232	−0.1242		0.1415
lnURB	1.7581 **	0.7860	0.3288 **		0.1562
PS	0.0693 **	0.0310	0.1100 ***		0.0135
C	−3.4721	3.7665	−5.6196 ***		0.5922
Observations	90		90
Countries	6		6
R2	0.53		0.46
Test		Statistic		p value
F Test for Individual Effects		8.2		(0.000)
Hausman Test		47.80		(0.000)

Notes: *** p < 0.01, ** p < 0.05.

reports the baseline panel estimation results using fixed-effects (FE) and random-effects (RE) models. The f test for individual effects is statistically significant, indicating the presence of country-specific effects in the panel. Furthermore, the Hausman test strongly rejects the random-effects specification, suggesting that the fixed-effects model provides more consistent estimates. Based on these results, the fixed-effects model is selected as the preferred baseline specification for the subsequent analysis.

Table 7. Diagnostic Tests.

Variable	CD Statistics	p Value
Pesaran CD test	2.042 **	0.041
Breusch–Pagan LM test	40.089 ***	0.000
Modified Wald test for heteroscedasticity	89.97 ***	0.000
Wooldridge test for autocorrelation	6.63 **	0.050
Mean VIF	2.96	-

Notes: *** p < 0.01, ** p < 0.05.

presents the results of several diagnostic tests for the panel model. The Pesaran’s CD and Breusch–Pagan LM tests indicate the presence of cross-sectional dependence among the panel units. In addition, the Modified Wald test confirms heteroscedasticity, while the Wooldridge test suggests the presence of serial correlation in the error terms. The mean VIF value remains well below the commonly accepted threshold, indicating that multicollinearity is not a major concern. Overall, these findings justify the use of robust estimation techniques that account for cross-sectional dependence and serial correlation.

Table 8. Main Estimation Results: Linear VS Asymmetric Model.

Variables	Linear Model		Asymmetric Model
	Coefficient	SE	Coefficient	SE
lnDigTrade	0.0122	0.0156	-	-
d.DigTrade (+)	-	-	−0.0343 **	0.0140
d.DigTrade (−)	-	-	0.0883 **	0.0378
lnGDP	−0.2377 **	0.1001	−0.4472 ***	0.0717
lnEUse	0.5393 ***	0.0704	0.4924 ***	0.0994
FTRI	−0.2701 **	0.0827	−0.2163 **	0.0983
lnURB	1.7581 ***	0.4649	2.7451 ***	0.3328
PS	0.0693 **	0.0251	0.1094 ***	0.0226
C	−3.4721	3.3679	−2.0203	2.4577
Observations	90		84
Countries	6		6
R2	0.53		0.62

Notes: *** p < 0.01, ** p < 0.05.

presents the main estimation results comparing the linear and asymmetric specifications. In the linear model, digital trade does not show a statistically significant relationship with CO₂ emissions. This finding reflects the complex and potentially offsetting effects of digital trade in the GCC context. On the one hand, digital trade may enhance efficiency and reduce emissions through technological improvements. On the other hand, these gains may be offset by increased energy consumption associated with digital infrastructure, such as data centers and digital services, particularly in economies characterized by energy-intensive production structures and a high reliance on fossil fuels. This ambiguity suggests that the overall effect of digital trade may be masked when positive and negative changes are aggregated within a linear framework.

However, once the asymmetric specification is introduced, a clearer pattern emerges. The results indicate that the positive changes in digital trade are associated with a reduction in CO₂ emissions, whereas negative changes tend to increase emissions, and both effects are statistically significant. This finding suggests that the environmental impact of digital trade is not uniform. Expansions in digital trade will likely enhance efficiency, encourage digital service substitution, and support technological improvements, all of which contribute to lower emissions. In contrast, contractions in digital trade may disrupt these efficiency gains and increase reliance on more carbon-intensive production processes. These findings are consistent with studies emphasizing the role of digitalization in improving environmental efficiency Li et al. [18] and Wang et al. [19], while also aligning with the nonlinear adjustment framework highlighted by Shin et al. [49], which emphasizes that positive and negative shocks can affect economic and environmental outcomes differently.

With regard to the control variables, energy consumption remains the strongest driver of emissions, showing a consistently positive and significant effect across both models. This result reflects the continued reliance of GCC economies on energy-intensive activities and is consistent with earlier findings in the environmental economics literature [1,3]. Economic growth exhibits a negative coefficient, suggesting that improvements in economic performance may be associated with better environmental efficiency, possibly due to technological progress or structural transformation. In addition, frontier technology readiness (FTRI) has a negative and significant effect, indicating that technological advancement contributes to reducing emissions. By contrast, urbanization and political stability show positive effects on emissions, which may reflect the environmental pressures associated with rapid urban development and expanding economic activity in the region. In the GCC context, where urbanization is closely tied to construction, transportation, and industrial activity, these effects may dominate potential efficiency gains in the short run.

Overall, the asymmetric model provides a better fit than the linear specification, highlighting the importance of considering nonlinear dynamics when evaluating the environmental implications of digital trade.

Table 9. Wald Test for Digital Trade Asymmetry.

Hypothesis	Statistics	p Value
d. DigTrade (+) = d.DigTrade (−)	7.57 **	0.017

Notes: ** p < 0.05.

reports the Wald test results for the equality of positive and negative digital trade shocks. The null hypothesis of symmetry is rejected at the 5% significance level, indicating that the effects of positive and negative changes in digital trade are statistically different. This finding provides formal evidence of an asymmetric relationship, supporting the use of a nonlinear specification and confirming that digital trade influences environmental outcomes in a non-uniform manner.

Table 10. Robustness Checks Using Alternative Driscoll–Kraay Specifications.

Variable	Baseline FE–DK	DK lag(1)	DK lag(3)	DK + Time FE
d.DigTrade (+)	−0.0343 **	−0.0343 **	−0.0343 **	−0.0322 *
	(0.0140)	(0.0155)	(0.0133)	(0.0181)
d.DigTrade (−)	0.0883 **	0.0883 **	0.0883 **	0.0967 **
	(0.0378)	(0.0401)	(0.0333)	(0.0344)
lnGDP	−0.4472 ***	−0.4472 ***	−0.4472 ***	−0.1638
	(0.0717)	(0.0771)	(0.0625)	(0.2486)
lnEUse	0.4924 ***	0.4924 ***	0.4924 ***	0.3482 ***
	(0.0994)	(0.0951)	(0.0962)	(0.1437)
FTRI	−0.2163 **	−0.2163 **	−0.2163 **	−0.1092
	(0.0983)	(0.0990)	(0.0881)	(0.1459)
lnURB	2.7451 ***	2.7451 ***	2.7451 ***	3.0238 ***
	(0.3328)	(0.5246)	(0.2747)	(0.2858)
PS	0.1094 ***	0.1094 ***	0.1094 ***	0.1012 ***
	(0.0226)	(0.0234)	(0.0218)	(0.0308)
Observations	84	84	84	84
Countries	6	6	6	6
R2	0.62	0.62	0.62	0.71

Notes: Driscoll–Kraay standard errors are reported in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.10.

presents the robustness checks based on alternative Driscoll–Kraay specifications. The results remain largely consistent across different lag structures and after controlling for time effects. In particular, the symmetric effects of digital trade persist, with positive shocks reducing emissions and negative shocks increasing emissions, confirming the stability of the main findings. The control variables also retain their expected signs and significance, especially energy use and urbanization, although some coefficients slightly weaken with the inclusion of time fixed effects, the overall conclusions remain unchanged. These findings suggest that the results are robust and not sensitive to model specification or estimation choices.

Table 11. Robustness Checks Using Leave-One-Country-Out Analysis.

Variable	Exclude BHR	Exclude KWA	Exclude OMN	Exclude QTR	Exclude KSA	Exclude UAE
d.DigTrade (+)	−0.0758 ***	−0.0217	−0.0339 *	−0.0098	−0.0387 **	−0.0282 *
	(0.0240)	(0.0140)	(0.0180)	(0.0187)	(0.0148)	(0.0152)
d.DigTrade (−)	0.0788 *	0.0869 **	0.1077 **	0.1372 ***	0.1046	0.0440
	(0.0391)	(0.0346)	(0.0408)	(0.0200)	(0.0928)	(0.0301)
lnGDP	−0.4847 ***	−0.4384 ***	−0.2781 ***	−0.5890 ***	−0.4771 ***	−0.2912 ***
	(0.1008)	(0.0650)	(0.0601)	(0.0913)	(0.0866)	(0.0831)
lnEUse	0.5000 ***	0.3742 **	0.5368 ***	0.7440 ***	0.4787 ***	0.5142 ***
	(0.0878)	(0.1416)	(0.0882)	(0.0734)	(0.1065)	(0.0814)
FTRI	−0.2155 *	−0.1940 *	−0.0599	−0.3209 ***	−0.2201 *	−0.1188
	(0.1118)	(0.0993)	(0.1542)	(0.0669)	(0.1066)	(0.1352)
lnURB	3.0236 ***	2.2543 ***	−3.1127 **	4.5296 ***	2.8238 ***	2.7711 ***
	(0.5149)	(0.3287)	(1.3063)	(0.9222)	(0.3708)	(0.3368)
PS	0.1004 ***	0.1251 ***	0.0732 ***	0.1661 ***	0.1111 ***	0.0763 ***
	(0.0252)	(0.0253)	(0.0176)	(0.0430)	(0.0212)	(0.0220)
Observations	70	70	70	70	70	70
Countries	5	5	5	5	5	5

Notes: Driscoll–Kraay standard errors are reported in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.10.

presents the results of the leave-one-country out robustness analysis. The results show some variation in statistical significance, which is expected given the small sample size and heterogeneity across GCC economies. However, the overall pattern of the asymmetric effects remains broadly consistent, indicating that the main findings are not driven by any single country.

Taken together, the empirical results provide consistent evidence that the relationship between digital trade and environmental outcomes in GCC economies is inherently asymmetric. While the linear model fails to capture a significant effect, the asymmetric specification reveals that expansions in digital trade contribute to reducing emissions, whereas contractions tend to increase environmental pressure. This emphasizes the necessity of considering nonlinear dynamics when evaluating the environmental implications of digitalization. Across all model specifications and robustness checks, energy consumption remains the dominant driver of emissions, while technological readiness plays a mitigating role. The stability of these findings across alternative estimations reinforces their reliability and suggests that the results are not sensitive to model specification or sample composition. Overall, the evidence points to the importance of digital transformation as a potential pathway toward improving environmental performance in the GCC region, provided that supportive policies and sustainable practices are effectively implemented.

5. Conclusions, Policy Perspectives and Limitations

This study offers insight into how digital trade affects environmental outcomes in GCC nations, with a special emphasis on nonlinear impacts. The results indicate that the linkage is not straightforward. Growth in digital trade reduces CO₂ emissions, but drops increase environmental pressure. This disparity emphasizes the significance of continuing to advance digital trade to preserve its environmental benefits. The findings also confirm that energy consumption remains the main driver of emissions in the region, reflecting the ongoing reliance on energy-intensive activities. At the same time, improvements in technological readiness play an important role in reducing emissions, suggesting that innovation and digital infrastructure can support cleaner growth. Urbanization and political stability, although essential for development, appear to place additional pressure on the environment, indicating that economic expansion in the GCC still comes with environmental costs.

From a policy perspective, the asymmetric nature of the results implies that policies should not treat digital trade as a uniform driver of environmental outcomes. Instead, differentiated strategies are required. First, since increases in digital trade are associated with reductions in CO₂ emissions, policymakers should prioritize policies that sustain and accelerate digital trade expansion, including investments in digital infrastructure, cross-border e-commerce facilitation, and regulatory frameworks that support digital integration. Second, the finding that declines in digital trade increase environmental pressure highlights the importance of ensuring the resilience and continuity of digital systems. This requires strengthening digital infrastructure reliability, reducing disruptions in digital services, and enhancing cybersecurity and institutional support to prevent setbacks that may reverse environmental gains.

Third, the strong role of energy consumption suggests that the environmental benefits of digital trade are conditional on the energy structure. Therefore, GCC countries should align digital expansion with energy transition strategies by increasing the share of renewable energy used to power digital infrastructure, particularly data centers and logistics systems. Finally, the results indicate that policy responses should account for country-specific conditions within the GCC. Differences in technological readiness, energy dependence, and institutional quality suggest that a uniform policy approach may not be effective. Countries with higher digital maturity may focus on innovation-driven efficiency gains, while others may prioritize foundational digital infrastructure and regulatory reforms to unlock the environmental benefits of digital trade.

Despite these contributions, the study has some limitations. First, the analysis is based on a relatively small sample of GCC countries, which may limit the generalizability of the findings. Second, the measurement of digital trade relies on available aggregate indicators, which may not fully capture the complexity of digital economic activities, particularly the environmental impact of digitalized goods trade and e-commerce logistics. Third, some econometric challenges arise from the data. In particular, the urbanization variable (lnURB) shows non-stationarity and very limited variation across countries, which affects the performance of unit root and cross-sectional dependence tests. For this reason, the results related to urbanization should be interpreted with caution. Finally, although robust estimation techniques are applied, the focus on short-run dynamics and the absence of a cointegration framework may limit the ability to capture long-run relationships, and potential endogeneity and omitted variable bias cannot be completely ruled out. Future research could address these issues by using alternative measures of urbanization, expanding the sample, employing broader indicators of digital trade, and applying more advanced empirical approaches, such as cointegration-based models. In addition, future research may explore hybrid approaches that combine econometric and machine learning techniques to better capture complex nonlinear dynamics [53,54].