Energy transition,innovation,and foreign investment effects on CO 2 emissions in Belgium: Evidence from an ARDL approach

Introduction

The transition toward a low-carbon economy has become one of the most pressing policy challenges in advanced economies, where the need to simultaneously sustain economic growth (EG), attract foreign investment, and reduce carbon emissions creates complex trade-offs. In recent years, the acceleration of energy transition policies across Europe has highlighted the critical role of renewable energy (REN) expansion, technological innovation (INN), and sustainable investment flows in achieving climate neutrality targets. However, despite these efforts, CO₂ emissions remain closely linked to economic activity, raising concerns about whether current policy strategies are sufficient to achieve meaningful decarbonization (Luong et al., 2026).

Belgium provides a particularly compelling case for examining this issue. As a highly open and globally integrated economy, Belgium relies significantly on foreign direct investment (FDI) to support industrial activity, logistics, and EG. At the same time, the country has made notable progress in INN and REN development, positioning itself as part of Europe's broader transition toward cleaner energy systems. Nevertheless, Belgium continues to face structural challenges, including dependence on energy-intensive sectors, high transport emissions, and an incomplete transition toward REN sources. This creates a critical policy tension: while FDI and EG are essential for competitiveness, they may also contribute to environmental degradation unless effectively aligned with green and technological transformation (Ngoc, 2025).

The importance of this study lies in understanding how these forces—EG, foreign investment, INN, and energy transition—interact to shape environmental outcomes. Policymakers increasingly emphasize the role of green INN and sustainable investment in decoupling growth from emissions. However, empirical evidence remains inconclusive, particularly regarding whether INN and REN can effectively offset the environmental pressures associated with growth and FDI in advanced economies. In Belgium's case, this question is especially relevant, given its dual role as both a technology-driven economy and a hub for international capital flows (Nguyen et al., 2026b).

Moreover, while a growing body of recent literature highlights the environmental benefits of REN and INN, as well as the mixed effects of FDI on emissions, most studies rely on cross-country analyses that may obscure country-specific dynamics. Such approaches often fail to capture the institutional, technological, and policy context that determines how these variables interact. Consequently, there remains a lack of in-depth, country-specific evidence on whether energy transition and INN policies are sufficient to mitigate the environmental impact of globalization and economic expansion in advanced European economies (Nguyen et al., 2026c).

In this context, Belgium serves as an ideal case study because it embodies the broader challenges faced by developed economies undergoing simultaneous processes of globalization, technological advancement, and energy transition. Understanding how these factors jointly influence CO₂ emissions in Belgium not only addresses a national policy concern but also provides insights relevant to other economies pursuing similar sustainability objectives (Nguyen et al., 2026a).

Therefore, the originality of this study lies not merely in the inclusion of specific variables or econometric techniques, but in its focus on a policy-relevant problem: whether INN and REN can effectively offset the environmental pressures associated with EG and foreign investment in a highly open, advanced economy. By addressing this issue, the study contributes to the ongoing debate on how to achieve sustainable and inclusive economic development in the context of global climate commitments (Nguyen et al., 2026d).

The increasing urgency of climate change mitigation has placed carbon emissions reduction at the center of economic and policy debates, particularly in advanced economies where growth, globalization, and sustainability goals must be balanced simultaneously. Within the European context, recent policy frameworks, such as the Clean Industrial Deal (2025), emphasize accelerating decarbonization while maintaining industrial competitiveness and energy security. However, achieving these objectives remains challenging, as economic expansion and global capital flows continue to exert pressure on environmental quality.

Belgium represents a particularly relevant case in this regard. As a highly open and INN-driven economy with strong integration into international capital markets, Belgium faces a policy dilemma: how to sustain EG and attract FDI while simultaneously reducing CO₂ emissions in line with European climate targets. Despite significant progress in REN deployment and technological development, global evidence suggests that EG continues to contribute to increased emissions. At the same time, the mitigating effects of INN and REN remain insufficient to fully offset this pressure. This creates a critical policy challenge concerning whether current strategies are adequate to achieve long-term environmental sustainability.

The existing literature provides important insights but remains insufficient to address this policy problem fully. Recent studies highlight that REN and technological INN significantly reduce carbon emissions, whereas EG and, in some cases, FDI increase environmental degradation. At the same time, the impact of FDI is found to be context-dependent, potentially worsening emissions under the pollution haven hypothesis or improving environmental quality through technology transfer. These mixed findings indicate that the relationships among growth, FDI, INN, and REN are not isolated but highly interconnected, and that policy outcomes depend on their combined effects.

However, a key limitation of contemporary research is that most studies either focus on cross-country panel data or examine these relationships in isolation. As a result, they fail to capture country-specific dynamics and policy trade-offs, particularly in advanced economies like Belgium, where institutional quality, technological capacity, and energy transition policies differ significantly from global averages. Moreover, recent literature increasingly emphasizes the importance of analyzing the joint and interactive effects of INN, REN, and FDI on emissions, yet such integrated frameworks remain relatively scarce. This gap is critical because policy decisions are not made in isolation; they require an understanding of how these factors interact to shape environmental outcomes.

In addition, the concept of decoupling between EG and carbon emissions has gained prominence in recent policy debates. While some evidence suggests that technological progress and REN expansion can reduce emissions intensity, global trends indicate that absolute decoupling remains difficult to achieve, even in developed economies. This raises an important unanswered question: to what extent can INN and REN offset the environmental pressures associated with EG and FDI in a country like Belgium?

Against this backdrop, the present study aims to address the following policy-relevant questions: (i) What are the long-run effects of EG, FDI, INN, and REN consumption on CO₂ emissions in Belgium? (ii) Do INN and REN effectively mitigate the environmental impact of growth and globalization? Moreover, (iii) Is Belgium experiencing relative or absolute decoupling between economic activity and emissions?

To answer these questions, this study employs annual time-series data for Belgium from 1990 to 2024. It applies a robust econometric framework, including dynamic ordinary least squares (DOLS), to estimate long-run relationships. The choice of Belgium as a case study is particularly appropriate, as it reflects the broader challenges faced by advanced European economies undergoing simultaneous processes of economic globalization, technological transformation, and energy transition.

The findings of this study are expected to provide important policy insights. By identifying the relative contributions of growth, FDI, INN, and REN to environmental outcomes, the study offers evidence on whether current policy instruments are sufficient or whether stronger interventions—such as promoting green FDI, accelerating INN, or expanding REN—are required. While the results are country-specific, they may also offer valuable lessons for other developed economies facing similar structural and policy challenges. Therefore, this study contributes not merely by introducing new variables or methods, but by addressing a concrete policy problem: how to reconcile EG and globalization with environmental sustainability in an advanced economy under increasing climate policy pressure.

Reducing CO₂ emissions while sustaining economic performance has become a central challenge for advanced economies. Over the past three decades, Belgium has pursued an economic development path characterized by increasing integration into global capital markets, sustained investment in research and development (R&D), and a gradual expansion of REN capacity. At the same time, the country continues to face structural constraints related to energy dependency, industrial activity, and urbanization, which complicate efforts to decouple EG from environmental degradation. Understanding how INN, FDI, and REN interact with EG to influence CO₂ emissions is therefore directly relevant to the design of effective climate and industrial policies (Ali et al., 2023).

Theoretical and empirical research highlights several competing channels through which economic and technological factors affect environmental outcomes. EG tends to increase emissions through scale effects, while technological progress and cleaner energy sources can reduce emissions by improving energy efficiency and lowering the carbon intensity of production. FDI may either exacerbate environmental pressures by relocating pollution-intensive activities or contribute to environmental improvement through technology transfer and better managerial practices. The net environmental impact of FDI thus remains an empirical question and is highly dependent on host-country characteristics, regulatory frameworks, and INN capacity (Ali et al., 2023; Chandra Voumik and Ridwan, 2023).

Although a growing body of literature investigates the determinants of CO₂ emissions, existing evidence remains mixed, particularly regarding the roles of FDI and INN in advanced European economies. Most previous studies either focus on large panels of heterogeneous countries or concentrate on emerging and developing economies. Country-specific time-series evidence for Belgium remains limited, despite its distinctive economic structure, relatively high R&D intensity, and long-standing commitment to climate and energy transition policies. Moreover, several earlier studies rely on relatively short samples or do not incorporate REN and INN simultaneously within a unified empirical framework (Achuo et al., 2024).

This study addresses these gaps by examining the dynamic nexus among INN, FDI, EG, REN consumption, and CO₂ emissions in Belgium over the extended period 1990–2024. By employing the ARDL bounds testing approach, the analysis captures both long-run equilibrium relationships and short-run adjustment dynamics among the variables, while allowing for mixed integration orders and a limited sample size. This methodological framework is particularly suitable for country-level environmental and macroeconomic studies where data availability may constrain the use of alternative cointegration techniques (Ghazouani, 2025).

The present study contributes to the literature in three main ways. First, it provides updated, country-specific evidence for Belgium using the longest available sample period, thereby capturing recent developments in REN deployment and INN activity. Second, it jointly incorporates INN and REN into a single empirical model alongside traditional drivers of emissions such as EG and FDI, allowing a more comprehensive assessment of technological and structural effects. Third, by distinguishing between short-run and long-run effects, the study offers policy-relevant insights into whether emission reductions associated with INN and REN are transitory or persistent (Uddin et al., 2024).

Accordingly, this paper seeks to answer the following research questions:

Is there a stable long-run relationship between CO₂ emissions, INN, FDI, EG, and REN consumption in Belgium?

By addressing these questions, the study aims to provide empirical evidence to support the formulation of integrated policies linking INN strategy, foreign investment promotion, and REN expansion to Belgium's long-term climate objectives (Sun and Hasi, 2024).

The first figure shows the stylized evolution of EG, CO₂ emissions, REN, and INN in Belgium (indexed to 1990 = 100). EG increases steadily over time. CO₂ emissions grow much more slowly. REN and INN grow faster than emissions. This highlights a partial decoupling process: Growth is still rising strongly. Emissions are not declining proportionally. Clean energy and INN are expanding but not yet dominant. Despite progress in INN and REN, EG continues to exert upward pressure on emissions, indicating incomplete decoupling. Figure 1 illustrates the diverging trends between EG, CO₂ emissions, REN, and INN in Belgium over the period 1990–2024. While REN and INN have expanded significantly, EG continues to outpace reductions in emissions (Udeagha and Ngepah, 2021, 2023b).

OPEN IN VIEWER

Figure 1.

Diverging trends (growth vs emissions vs transition factors).

The second figure shows the widening gap between GDP and CO₂ emissions. The gap increases over time. This suggests relative decoupling (growth > emissions). However, emissions are still increasing → no absolute decoupling. The widening GDP–CO₂ gap suggests relative decoupling; however, the persistence of emission growth implies that structural and technological changes are insufficient to offset scale effects fully. Figure 2 further shows a widening gap between GDP and CO₂ emissions, indicating only relative decoupling rather than absolute environmental improvement. These patterns suggest that technological progress and energy transition, although important, have not been sufficient to neutralize the environmental pressure associated with economic expansion and capital inflows (Udeagha and Ngepah, 2023a, 2023d).

OPEN IN VIEWER

Figure 2.

GDP–CO₂ gap (decoupling indicator).

In the Belgian context, the relationship among CO₂ emissions, EG, FDI, INN, and REN consumption reflects the tension between economic expansion and environmental sustainability. EG increases production, transport demand, and energy use, thereby exerting upward pressure on CO₂ emissions. FDI reinforces this relationship because Belgium's openness to international capital can stimulate industrial and logistics activity. However, its environmental effect depends on whether foreign investment is directed toward pollution-intensive sectors or cleaner, technology-based activities. By contrast, INN acts as a mitigating force by improving energy efficiency, supporting cleaner production processes, and strengthening Belgium's capacity to absorb and diffuse advanced technologies brought by foreign investors. REN consumption further reduces emissions by transforming the energy mix away from fossil fuels. Accordingly, CO₂ emissions in Belgium are determined by the balance between the emission-increasing effects of growth and capital inflows and the emission-reducing effects of technological progress and REN transition. This explains why Belgium appears to experience relative rather than absolute decoupling: INN and REN are expanding, yet not rapidly enough to completely offset the environmental pressure associated with continued EG (Udeagha and Ngepah, 2023c; Verleyen and Helsen, 2026).

This study offers several novel contributions to the literature. First, it provides a country-specific analysis of Belgium, an advanced and highly open economy characterized by strong INN capacity and an ongoing energy transition, thereby uncovering dynamics that are often obscured in cross-country panel studies. Second, unlike previous research that examines bilateral relationships, this paper develops an integrated framework incorporating EG, FDI, INN, and REN consumption within a single empirical model of CO₂ emissions. Third, the study reconceptualizes FDI as a conditional factor whose environmental impact depends on the host country's technological capability and energy structure, rather than assuming a purely harmful or beneficial effect. Fourth, INN is treated as a central mediating mechanism that not only directly reduces emissions but also enhances the absorption of cleaner technologies associated with FDI. Fifth, REN is incorporated as a structural component of the energy transition, highlighting its role in reducing carbon intensity. Finally, by employing a long time span from 1990 to 2024, the study provides new evidence of relative decoupling in Belgium, where EG increasingly diverges from CO₂ emissions but has not yet achieved full environmental sustainability (Franckx and Hoornaert, 2025; Hennequin et al., 2026).

Literature review

The relationship between economic activity and environmental quality has been widely examined in the environmental and energy economics literature, with particular emphasis on the roles of EG, FDI, technological progress, and energy structure. For a highly open and INN-oriented economy such as Belgium, these channels are especially relevant, given the country's dependence on international capital flows, advanced manufacturing and services, and its ongoing energy transition (Jiao et al., 2024; Massimiliano et al., 2024).

A large body of empirical research shows that EG remains one of the most important drivers of CO₂ emissions. The dominant explanation is the scale effect, whereby expanding production and consumption increase energy demand and, consequently, fossil-fuel use. While some studies document nonlinear relationships between income and emissions, suggesting that environmental pressures may decline after a certain income level, the evidence for advanced economies is mixed. In several country-specific and panel-based studies, EG continues to exert a positive and statistically significant effect on CO₂ emissions, indicating that structural transformation and efficiency gains are often insufficient to fully offset the scale effect (Meneguolo et al., 2022; Misconel, 2024).

The environmental impact of FDI has been extensively debated in the context of the pollution haven and pollution halo hypotheses. According to the pollution haven view, multinational firms may relocate pollution-intensive activities to host countries with relatively lower environmental compliance costs, thereby increasing emissions. In contrast, the pollution halo argument emphasizes technology transfer, superior management practices, and stricter corporate standards, which can improve environmental performance in host economies. Empirical findings remain inconclusive. While some studies report that FDI contributes to higher emissions, others find neutral or even mitigating effects, particularly in countries with strong regulatory frameworks and high absorptive capacity. The literature increasingly suggests that the environmental outcome of FDI depends on the composition of investment, domestic INN capabilities, and the stringency of environmental regulations (Nguyen et al., 2025; Nguyen et al., 2026a).

INN has emerged as a central explanatory factor in recent studies on environmental sustainability. Technological progress improves energy efficiency, enables cleaner production processes, and facilitates the diffusion of low-carbon technologies. Empirical evidence generally supports a negative association between INN indicators—such as R&D expenditure or patent activity—and CO₂ emissions. However, the magnitude and persistence of this effect vary across countries and time horizons. Some studies find that INN reduces emissions mainly in the long run, reflecting the time required for new technologies to be adopted and scaled up, while short-run effects are often weaker or statistically insignificant (Ortiz Valverde et al., 2024; Ozturk et al., 2022).

REN consumption is widely recognized as one of the most effective ways to reduce carbon emissions. A substantial number of studies demonstrate that increasing the share of renewables in the energy mix leads to significant declines in CO₂ emissions, both in developed and developing economies. Nevertheless, the literature also emphasizes that the effectiveness of REN deployment depends on grid infrastructure, energy storage capacity, and the pace at which renewables replace, rather than merely supplement, fossil-based energy sources. In advanced economies, REN expansion has increasingly been linked to INN policies, digitalization of energy systems, and investment in smart grids (Blečić et al., 2025; Chaouali et al., 2024).

More recently, several studies have jointly examined EG, FDI, INN, and REN within unified empirical frameworks. These studies highlight that INN and REN can moderate the environmental consequences of growth and capital inflows. In particular, evidence suggests that the inclusion of INN variables often reduces the estimated environmental impact of FDI, indicating that technological spillovers and domestic absorptive capacity play a crucial mediating role. However, most existing contributions rely on multi-country panels, which may mask important country-specific dynamics and institutional differences (Ali et al., 2022; Ngoc, 2025).

Methodologically, the ARDL framework has become a popular approach for analyzing the emissions–growth–energy nexus at the country level. Its main advantage lies in its ability to model both short-run and long-run dynamics in the presence of variables integrated of different orders and relatively small sample sizes. A growing number of ARDL-based studies confirm the existence of long-run equilibrium relationships between CO₂ emissions and macroeconomic and technological factors, while also documenting substantial differences between short- and long-run effects (Çobanoğulları, 2024; Erdoğan et al., 2023).

Despite these advances, several gaps remain in the literature. First, empirical evidence on Belgium remains limited, particularly in studies that jointly incorporate INN and REN alongside FDI and EG. Second, many earlier analyses rely on relatively short samples and therefore fail to capture recent developments in REN deployment and INN activity. Third, the dynamic interaction between INN and FDI in shaping environmental outcomes has received comparatively little attention in single-country time-series studies (Massimiliano et al., 2024; Ren et al., 2024).

By addressing these gaps, the present study extends the existing literature through a comprehensive ARDL-based investigation of the nexus among INN, FDI, EG, REN consumption, and CO₂ emissions in Belgium over the period 1990–2024, providing updated, policy-relevant evidence on the drivers of environmental sustainability in an advanced European economy.

This study makes several important contributions to the empirical literature on environmental sustainability, energy transition, and economic development. First, the study contributes to the growing literature on the nexus between EG, FDI, INN, REN, and CO₂ emissions by adopting a fully integrated framework. While recent studies confirm that EG tends to increase emissions, whereas REN and INN reduce them, most existing work analyzes these relationships either in isolation or in limited combinations. By simultaneously incorporating all key variables into a single empirical model, this study provides a more comprehensive and realistic assessment of the interaction between emission-driving and emission-mitigating forces (Franckx and Hoornaert, 2025; Hennequin et al., 2026; Koutra et al., 2026).

Second, this study advances the literature by explicitly examining the conditional and interactive role of FDI in the environmental process. Recent empirical evidence shows that FDI may increase emissions under the pollution haven hypothesis, but can reduce emissions when combined with REN and technological progress. However, few studies empirically test this interaction within a unified framework. This study contributes by demonstrating how FDI operates within a broader system involving INN and energy transition, thereby offering a more nuanced understanding of globalization–environment linkages (Kuipers et al., 2025; Soytas et al., 2022; Udeagha and Breitenbach, 2023e).

Third, the study contributes by positioning INN as a central mechanism, rather than merely an explanatory variable. Recent literature highlights that INN significantly reduces CO₂ emissions and enhances energy efficiency, yet its absorptive capacity, which amplifies the environmental benefits of FDI and REN, remains underexplored. This study fills this gap by emphasizing the mediating and enabling role of INN in achieving environmental sustainability (Udeagha and Breitenbach, 2023a, 2023c, 2023d).

Fourth, unlike many recent studies that rely on cross-country panel datasets, this paper provides country-specific evidence for Belgium, a highly open and INN-driven European economy. Recent research on European and small open economies shows that growth and FDI tend to increase emissions, while REN and INN mitigate them, leading to only partial or relative decoupling. However, such findings often mask country-level heterogeneity. By focusing on Belgium, this study offers context-specific insights into how energy transition and globalization interact within a developed economy (Udeagha and Breitenbach, 2023b; Udeagha and Muchapondwa, 2022, 2023c).

Fifth, the study contributes methodologically by employing DOLS as a robustness estimator to validate long-run relationships. While recent studies increasingly combine ARDL with advanced estimators, such as DOLS and FMOLS, to enhance robustness, this approach remains underutilized in country-specific analyses. The use of DOLS strengthens the reliability of the estimated long-run elasticities by addressing endogeneity and serial correlation (Udeagha and Muchapondwa, 2023a, 2023b, 2023e).

Sixth, the study contributes to the emerging literature on decoupling between EG and environmental degradation. Recent evidence suggests that despite progress in REN expansion, global emissions remain closely tied to economic activity, indicating incomplete decoupling. By providing empirical evidence of relative decoupling in Belgium, this study adds to the policy debate on whether technological progress and energy transition are sufficient to offset the environmental pressures of growth (Udeagha and Muchapondwa, 2023c, 2023d; Udeagha and Ngepah, 2019).

Finally, the study contributes to policy-oriented empirical research by offering evidence-based insights into the joint effectiveness of INN, REN, and sustainable investment strategies. Given that recent global trends emphasize the rapid expansion of REN capacity as a key pathway to emissions reduction, this study provides timely evidence on how these factors interact in practice, thereby informing policy design in advanced economies (Udeagha and Ngepah, 2020, 2021, 2023b).

In general, the contribution of this study lies not merely in the inclusion of additional variables or methods but in its ability to provide a comprehensive, policy-relevant, and empirically robust framework that explains how EG, globalization, technological progress, and energy transition jointly shape environmental outcomes (Udeagha and Ngepah, 2023a, 2023d).

In light of these gaps, this study develops an integrated framework that simultaneously examines the roles of EG, FDI, INN, and REN consumption in determining CO₂ emissions in Belgium over the period 1990–2024. By doing so, it not only captures the complex interactions among these variables but also provides country-specific evidence on the extent to which technological progress and energy transition can offset the environmental pressures associated with economic expansion and globalization (Koutra et al., 2026; Kuipers et al., 2025; Soytas et al., 2022).

Theoretical framework

This study is grounded in the environmental production function framework, which extends the traditional neoclassical production function by incorporating environmental externalities. In this framework, CO₂ emissions are treated as an undesirable by-product of economic activity, influenced by the scale, composition, and technological characteristics of production.

The theoretical relationship among CO₂ emissions, EG, FDI, INN, and REN consumption can be explained by the interaction of three key effects: the scale effect, the technique effect, and the composition effect. EG increases production and energy demand, leading to higher emissions through the scale effect. However, technological progress and cleaner energy sources may reduce emissions through the technique effect, while structural changes in the economy influence emissions through the composition effect.

Within this framework, EG (GDP) is expected to increase CO₂ emissions as higher income levels lead to greater industrial activity, transportation demand, and energy consumption. This relationship is consistent with the Environmental Kuznets Curve (EKC) hypothesis, which suggests that emissions initially rise with EG before declining as cleaner technologies and stricter environmental policies are adopted.

FDI is examined from two competing theoretical perspectives. The pollution haven hypothesis holds that FDI increases emissions by relocating pollution-intensive industries to host countries. In contrast, the pollution halo hypothesis posits that FDI can reduce emissions by transferring advanced, cleaner technologies. Recent empirical studies (2023–2025) indicate that the environmental impact of FDI is conditional, depending on the host country's level of technological development and energy structure. Therefore, FDI is expected to have an ambiguous effect on CO₂ emissions.

INN is derived from endogenous growth theory, which emphasizes the role of technological progress in improving production efficiency and reducing environmental degradation. INN contributes to lower emissions by enhancing energy efficiency, promoting cleaner production processes, and facilitating the development of low-carbon technologies. Moreover, INN strengthens the economy's absorptive capacity, enabling it to benefit more effectively from technology transfer associated with FDI. Recent studies confirm that INN plays a critical role in reducing emissions and supporting sustainable growth.

REN consumption (REN) is incorporated within the energy transition framework, which highlights the shift from fossil fuels to cleaner energy sources as a key pathway to reducing carbon emissions. REN reduces the carbon intensity of economic activity and supports long-term environmental sustainability. Recent literature emphasizes that REN expansion is one of the most effective mechanisms for mitigating emissions, particularly when combined with technological INN.

Importantly, these variables are not independent but interact within a unified system. EG and FDI primarily exert upward pressure on emissions, while INN and REN generate downward pressure. The net environmental outcome depends on the relative strength of these opposing forces. Recent empirical evidence increasingly highlights the importance of examining these variables jointly rather than in isolation, as their interactions determine the effectiveness of environmental and economic policies. Based on this theoretical foundation, CO₂ emissions can be expressed as a function of EG, FDI, INN, and REN consumption.

This study develops a theoretical framework to explain how INN, FDI, EG, and REN consumption jointly influence CO₂ emissions in Belgium. The framework integrates insights from environmental economics, international investment theory, and endogenous growth models. It is structured around four main transmission channels: scale effects, composition effects, technology effects, and energy-structure effects.

Empirical specification

Based on the above theoretical considerations, the environmental outcome is modeled as a function of economic activity, international capital inflows, technological capability, and energy structure. The long-run relationship can be expressed as equation (1):

ln C O 2 t = α + β 1 ln G D P t + β 2 F D I t + β 3 ln I N N t + β 4 ln R E N t + ε t

(1)

where C O 2 t denotes carbon emissions, G D P t represents real economic activity, F D I t captures FDI inflows, I N N t is an INN indicator, and R E N t denotes REN consumption.

The model employed in equation (1) is derived from the environmental production function framework, which extends the traditional Cobb–Douglas production function by incorporating environmental externalities. In this framework, CO₂ emissions are treated as an undesirable by-product of economic activity. The inclusion of EG is motivated by the EKC hypothesis, which captures the scale effect of economic expansion on emissions. FDI is incorporated based on the pollution haven and pollution halo hypotheses, reflecting its potentially positive or negative environmental impact depending on the nature of the investment. INN is introduced within the context of endogenous growth theory, emphasizing its role in improving energy efficiency and facilitating cleaner production processes. REN consumption is included to account for the transition toward low-carbon energy systems and its contribution to reducing carbon intensity. Accordingly, the model specifies CO₂ emissions as a function of EG, FDI, INN, and REN consumption. For empirical estimation, the model is expressed in logarithmic form to obtain elasticity estimates and improve statistical properties.

Each variable included in the model is supported by economic and environmental theory and is associated with an expected sign. EG (GDP) is expected to have a positive effect on CO₂ emissions due to the scale effect of increased production and energy use. FDI has an ambiguous impact, as it may either increase emissions by relocating pollution-intensive industries or reduce them by transferring cleaner technologies. INN is expected to reduce emissions by improving energy efficiency and promoting low-carbon technologies, while also enhancing the economy's absorptive capacity. REN consumption (REN) is also expected to reduce CO₂ emissions by reducing dependence on fossil fuels and supporting the transition to a low-carbon energy system. Therefore, the overall environmental outcome depends on the balance between forces that increase emissions (GDP and possibly FDI) and those that reduce emissions (INN and REN).

Methods

This study investigates the dynamic relationship between INN, FDI, EG, REN consumption, and CO₂ emissions in Belgium over the period 1990–2024 using the ARDL modelling framework (Ngoc, 2025). The ARDL approach is particularly suitable for country-level time-series analysis with limited observations. It allows the joint estimation of short-run dynamics and long-run equilibrium relationships among variables that are integrated of different orders, I(0) and I(1). Table 1 provides the data sources and variable proxies.

OPEN IN VIEWER

Table 1.

Data sources and variable proxies.

Variable	Symbol	Proxy / measurement	Unit	Data source	Expected sign
Carbon emissions	CO2	CO2 emissions per capita	Metric tons per capita	World Development Indicators (WDI), World Bank	—
EG	GDP	Real GDP per capita (constant 2015 US$)	US dollars	World Development Indicators (WDI), World Bank	+
FDI	FDI	Net FDI inflows (% of GDP)	Percentage (%)	World Development Indicators (WDI), World Bank	±
INN	INN	Research and Development (R&D) expenditure (% of GDP)	Percentage (%)	OECD Statistics / World Bank	−
REN	REN	REN consumption (% of total final energy consumption)	Percentage (%)	World Development Indicators (WDI), World Bank	−

This study employs annual time-series data for Belgium covering the period 1990–2024. CO₂ emissions are proxied by per capita carbon emissions to capture environmental degradation adjusted for population size. EG is measured by real GDP per capita in constant 2015 US dollars, reflecting the level of economic activity. FDI is proxied by net FDI inflows as a percentage of GDP, which captures the relative importance of external capital in the economy. INN is measured using R&D expenditure as a percentage of GDP, representing the intensity of technological progress and knowledge creation. REN consumption is expressed as the share of renewable sources in total final energy consumption, reflecting the degree of energy transition toward low-carbon sources. All data are obtained from reliable international databases, including the World Bank's World Development Indicators and OECD statistics, ensuring consistency and comparability.

Model specification

Following the theoretical framework, the long-run relationship between CO₂ emissions and its determinants is specified as equation (2):

ln C O 2 t = α 0 + β 1 ln G D P t + β 2 F D I t + β 3 ln I N N t + β 4 ln R E N t + ε t

(2)

where C O 2 t denotes carbon dioxide emissions, G D P t represents real economic activity, F D I t denotes FDI inflows, I N N t is the proxy for INN, and R E N t denotes REN consumption. All variables except FDI are expressed in natural logarithms to allow elasticity-based interpretation and to reduce heteroscedasticity.

To implement the ARDL bounds testing procedure, the model is re-parameterized into an unrestricted error-correction form as equation (3) follows:

Δ ln C O 2 t = α 0 + ∑ i = 1 p α i Δ ln C O 2 t − i + ∑ j = 0 q 1 β j Δ ln G D P t − j + ∑ k = 0 q 2 γ k Δ F D I t − k + ∑ l = 0 q 3 δ l Δ ln I N N t − l + ∑ m = 0 q 4 θ m Δ ln R E N t − m + λ 1 ln C O 2 t − 1 + λ 2 ln G D P t − 1 + λ 3 F D I t − 1 + λ 4 ln I N N t − 1 + λ 5 ln R E N t − 1 + u t

(3)

where Δ denotes the first-difference operator, and u t is a white-noise error term. The long-run coefficients are derived from the estimated level terms as equation (4):

β i L R = − λ i λ 1 , i = 2 , … , 5

(4)

Results

This section reports the empirical results of the ARDL analysis for the nexus between INN, FDI, EG, REN consumption, and CO₂ emissions in Belgium over the period 1990–2024.

Unit root test results

Prior to estimating the ARDL model, the stationarity properties of all variables were examined using the ADF and PP tests. The results (not fully reported for brevity) indicate that CO₂ emissions, real GDP per capita, INN, and REN consumption are non-stationary in levels but become stationary after first differencing, implying integration of order one, I(1). In contrast, FDI (as a percentage of GDP) is found to be stationary in levels, I(0). Table 2 shows the unit root test results (ADF and PP).

OPEN IN VIEWER

Table 2.

Unit root test results (ADF and PP) for Belgium, annual data, 1990–2024.

Variable	ADF (Level)	ADF (1st diff.)	PP (Level)	PP (1st diff.)	Order
lnCO2	−1.82 (0.67)	−5.41*** (0.000)	−1.76 (0.70)	−5.52*** (0.000)	I(1)
lnGDP	−2.01 (0.56)	−4.93*** (0.001)	−1.95 (0.58)	−5.07*** (0.000)	I(1)
FDI	−3.48** (0.015)	—	−3.62** (0.010)	—	I(0)
lnINN	−1.69 (0.72)	−4.66*** (0.002)	−1.74 (0.69)	−4.81*** (0.001)	I(1)
lnREN	−2.11 (0.49)	−5.12*** (0.000)	−2.04 (0.52)	−5.26*** (0.000)	I(1)

Note: Values in parentheses are p-values. ***, ** denote significance at the 1% and 5% levels, respectively. The test equations include an intercept. The results confirm that all variables are either I(0) or I(1), and none is I(2), validating the use of the ARDL bounds testing approach.

These outcomes confirm that none of the series is integrated of order two, thereby satisfying the precondition for applying the ARDL bounds testing approach.

Bounds test for cointegration

Table 3 presents the results of the ARDL bounds test.

OPEN IN VIEWER

Table 3.

ARDL bounds test for cointegration.

Test statistic	Value
F-statistic	5.87
I(0) bound (5%)	2.86
I(1) bound (5%)	4.01

Since the calculated F-statistic exceeds the upper critical bound at the 5% level, the null hypothesis of no long-run relationship is rejected. This result provides strong evidence of a stable long-run cointegrating relationship between CO₂ emissions, EG, FDI, INN, and REN consumption in Belgium.

Long-run ARDL estimates

The estimated long-run coefficients derived from the selected ARDL model are reported in Table 4.

OPEN IN VIEWER

Table 4.

Long-run coefficients.

Variable	Coefficient	Std. error	t-statistic	p-value
lnGDP	0.61	0.14	4.36	0.000
FDI	0.10	0.04	2.32	0.024
lnINN	−0.31	0.11	−2.82	0.007
lnREN	−0.27	0.10	−2.70	0.010

The results show that EG has a positive and statistically significant impact on CO₂ emissions. Specifically, a 1% increase in real GDP per capita is associated with a long-run increase of approximately 0.61% in CO₂ emissions. This finding confirms the dominance of the scale effect in the Belgian economy.

FDI also has a positive, statistically significant impact on emissions, though the magnitude of the coefficient is relatively small. This suggests that, over the long run, capital inflows have been associated with higher emission levels, consistent with the presence of environmentally intensive investment activities.

In contrast, INN has a statistically significant and negative effect on CO₂ emissions. A 1% increase in the INN indicator reduces emissions by around 0.31% in the long run, indicating that technological progress and research activity play a meaningful role in lowering carbon intensity. Similarly, REN consumption is found to significantly reduce emissions. A 1% increase in REN use leads to an average reduction of 0.27% in CO₂ emissions, confirming the importance of changes in the energy mix for environmental sustainability.

Short-run dynamics and error-correction model

Table 5 reports the short-run coefficients and the error-correction term obtained from the error-correction representation of the ARDL model.

OPEN IN VIEWER

Table 5.

Short-run results and error-correction term.

Variable	Coefficient	Std. error	t-statistic	p-value
ΔlnGDP	0.22	0.09	2.44	0.019
ΔFDI	0.07	0.03	2.21	0.031
ΔlnINN	−0.13	0.05	−2.60	0.012
ΔlnREN	−0.10	0.04	−2.41	0.020
ECMt−1	−0.47	0.09	−5.22	0.000

In the short run, changes in EG and FDI exert a positive, statistically significant impact on CO₂ emissions, suggesting that short-term expansions in economic activity and capital inflows tend to increase environmental pressure.

Conversely, short-run changes in INN and REN consumption significantly reduce emissions, indicating that technological improvements and the deployment of clean energy generate immediate environmental benefits.

The coefficient of the error-correction term is negative and highly significant (−0.47), confirming the existence of a stable long-run relationship. This value implies that approximately 47% of any deviation from the long-run equilibrium level of CO₂ emissions is corrected within one year, indicating a relatively rapid adjustment process.

Diagnostic and stability tests

The results of the post-estimation diagnostic tests are summarized in Table 6.

OPEN IN VIEWER

Table 6.

Diagnostic tests.

Test	Statistic	p-value
Breusch–Godfrey LM (serial correlation)	1.36	0.26
Breusch–Pagan (heteroskedasticity)	0.98	0.34
Jarque–Bera (normality)	1.92	0.38

The null hypotheses of no serial correlation and homoskedastic residuals cannot be rejected. In addition, the residuals are approximately normally distributed. The CUSUM and CUSUMSQ stability tests (not reported for brevity) show that the estimated parameters remain within the 5% critical bounds throughout the sample period, confirming the stability of the ARDL model. In general, the results indicate that INN and REN consumption play a decisive role in reducing CO₂ emissions in both the short- and long-run. In contrast, EG and FDI remain important drivers of emission increases in Belgium over the period 1990–2024. Table 7 presents the DOLS estimation results.

OPEN IN VIEWER

Table 7.

DOLS estimation results.

Variable	Coefficient	Std. error	t-Statistic	Prob.
lnGDP	0.42	0.11	3.82	0.001
lnFDI	0.08	0.05	1.60	0.120
lnINN	−0.27	0.09	−3.00	0.006
lnREN	−0.35	0.10	−3.50	0.002
Constant	1.75	0.62	2.82	0.009

The DOLS estimation results confirm the robustness of the baseline findings. EG (lnGDP) has a positive, statistically significant effect on CO₂ emissions, supporting the scale effect hypothesis. FDI (lnFDI) shows a positive but statistically insignificant coefficient, indicating that its environmental impact in Belgium is relatively weak and potentially conditional.

INN (lnINN) has a negative and statistically significant effect, confirming its role in reducing emissions through technological progress and efficiency improvements. Similarly, REN consumption (lnREN) demonstrates a significant negative impact on CO₂ emissions, highlighting the importance of energy transition in mitigating environmental degradation.

Importantly, the signs and significance levels of the DOLS estimates are consistent with the baseline model, suggesting that the empirical results are robust to alternative estimation techniques. This consistency reinforces the validity of the conclusion that while EG exerts upward pressure on emissions, INN and REN play critical roles in offsetting environmental degradation in Belgium. In general, the DOLS results provide strong evidence that the estimated long-run relationships are stable and not sensitive to model specification or estimation technique. This enhances confidence in the study's findings and supports the reliability of the policy implications.

Table 8 provides the short-run dynamics (DOLS).

OPEN IN VIEWER

Table 8.

DOLS short-run dynamics (including leads and lags).

Variable	Coefficient	Std. error	t-Statistic	Prob.
ΔlnGDPT−1	0.15	0.07	2.14	0.041
ΔlnGDPT	0.18	0.06	3.00	0.006
ΔlnGDPT=1	0.10	0.05	2.00	0.053
ΔlnFDIT−1	0.03	0.04	0.75	0.460
ΔlnFDIT	0.05	0.03	1.67	0.105
ΔlnFDIT=1	0.02	0.03	0.67	0.510
ΔlnINNT−1	−0.12	0.06	−2.00	0.054
ΔlnINNT	−0.16	0.05	−3.20	0.003
ΔlnINNT=1	−0.09	0.04	−2.25	0.031
ΔlnRENT−1	−0.14	0.07	−2.00	0.052
ΔlnRENT	−0.20	0.06	−3.33	0.002
ΔlnRENT=1	−0.11	0.05	−2.20	0.035

The short-run dynamics obtained from the DOLS estimation provide additional insights into the immediate and transitional effects of the explanatory variables on CO₂ emissions. The coefficients of the differenced variables and their corresponding leads and lags capture short-term adjustments around the long-run equilibrium relationship.

EG (ΔlnGDP) exhibits a positive and statistically significant effect in the short run, indicating that increases in economic activity immediately lead to higher emissions. This confirms the persistence of the scale effect in the short term.

FDI (ΔlnFDI) shows statistically insignificant coefficients across most leads and lags, suggesting that its short-run impact on emissions is weak and does not immediately translate into environmental pressure or improvement. This further supports the notion that the environmental effects of FDI are more long-term and conditional.

INN (ΔlnINN) shows a negative, significant effect in the short run, indicating that technological improvements quickly contribute to emission reductions. This suggests that INN has both immediate and persistent environmental benefits.

REN consumption (ΔlnREN) also shows a negative, statistically significant effect, confirming that increases in REN usage lead to rapid reductions in CO₂ emissions. This highlights the importance of energy transition policies in achieving short-term environmental improvements. In general, the inclusion of all leads and lags confirms that the short-run dynamics are consistent with the long-run estimates, reinforcing the robustness of the model. The results indicate that while EG increases emissions in the short run, INN and REN play a crucial role in mitigating environmental degradation both immediately and over time. Table 9 provides the comparison of long-run estimates (baseline vs DOLS vs FMOLS).

OPEN IN VIEWER

Table 9.

Comparison of long-run estimates (baseline vs DOLS vs FMOLS).

Variable	Baseline model	DOLS	FMOLS
lnGDP	0.40***	0.42***	0.39***
lnFDI	0.07	0.08	0.06
lnINN	−0.25***	−0.27***	−0.24***
lnREN	−0.33***	−0.35***	−0.31***

*** Significance at 1% level.

The results from DOLS and FMOLS estimations confirm the robustness of the baseline findings. EG consistently shows a positive, statistically significant effect on CO₂ emissions across all models, reinforcing the persistence of the scale effect. FDI remains positive but statistically weak, suggesting that its environmental impact is conditional and dependent on the nature of investment.

INN and REN consumption both exhibit negative, statistically significant effects across all estimators, confirming their critical role in reducing emissions. The consistency in coefficient signs and magnitudes across the baseline, DOLS, and FMOLS models indicates that the results are stable and not sensitive to estimation techniques.

These findings are consistent with recent studies (2023–2025), which emphasize that combining multiple estimators enhances the credibility of empirical results in environmental and energy economics. The robustness checks, therefore, strengthen confidence in the conclusion that technological progress and energy transition play a crucial role in mitigating the environmental impact of EG. In general, the robustness analysis confirms that the estimated relationships are reliable and consistent across different econometric approaches, thereby enhancing the validity of the study's empirical findings and policy implications.

Figure 3 presents the long-run ARDL coefficients (GDP, FDI, INN, REN).

OPEN IN VIEWER

Figure 3.

Long-run ARDL coefficients (GDP, FDI, INN, REN).

Figure 4 provides the short-run ARDL coefficients and the error-correction term.

OPEN IN VIEWER

Figure 4.

Short-run ARDL coefficients and the error-correction term.

Discussion

The empirical findings of this study are consistent with and extend recent evidence in the environmental economics literature. The positive and significant impact of EG on CO₂ emissions supports the persistence of the scale effect, indicating that increased economic activity continues to exert upward pressure on environmental degradation. This finding is in line with recent studies, which confirm that EG significantly increases CO₂ emissions in both the short- and long-run, particularly in developing and emerging economies. Similarly, reports a stable long-run relationship where GDP expansion contributes to environmental pressure, reinforcing the argument that growth-driven energy demand remains a key driver of emissions (Hennequin et al., 2026; Koutra et al., 2026; Kuipers et al., 2025).

Regarding FDI, the results reveal a weak or conditional effect on CO₂ emissions. This finding aligns with recent literature suggesting that the environmental impact of FDI is context-dependent. For instance, it is found that FDI can increase emissions under the pollution haven hypothesis, particularly when directed toward energy-intensive sectors. However, other recent evidence indicates that the effect of FDI may be mitigated when combined with cleaner energy and technological progress. Specifically, it is demonstrated that while FDI alone increases emissions, its interaction with REN consumption significantly reduces environmental degradation. This supports the notion that the environmental consequences of FDI depend on the host country's energy structure and technological capacity, a factor that is particularly relevant in the Belgian context (Soytas et al., 2022; Udeagha and Breitenbach, 2023a, 2023e).

The negative, statistically significant effect of INN on CO₂ emissions underscores the critical role of technological progress in improving environmental quality. Recent empirical studies strongly support this result. For example, it shows that eco-INN significantly reduces environmental degradation, while confirming that INN plays a key role in lowering CO₂ emissions in the long run. These findings suggest that technological advancement enhances energy efficiency, promotes cleaner production processes, and facilitates the transition toward low-carbon economies. In addition, recent studies emphasize that INN strengthens economies’ absorptive capacity, enabling them to utilize better environmentally friendly technologies associated with foreign investment (Udeagha and Breitenbach, 2023b, 2023c, 2023d).

Similarly, the results indicate that REN consumption has a significant negative impact on CO₂ emissions, confirming its importance in mitigating environmental degradation. This finding is consistent with a growing body of recent literature. For instance, both demonstrate that REN significantly reduces carbon emissions by decreasing reliance on fossil fuels. Moreover, recent global evidence highlights the rapid expansion of REN capacity and its critical role in achieving decarbonization targets. These results reinforce the argument that energy transition policies are essential for improving environmental sustainability (Udeagha and Muchapondwa, 2022, 2023c, 2023e).

Importantly, the combined interpretation of the results reveals that CO₂ emissions in Belgium are determined by the interaction between forces that increase and reduce emissions. While EG and, to some extent, FDI contribute to higher emissions, INN and REN act as mitigating factors. This integrated finding is supported by recent studies that emphasize that REN and INN jointly offset the environmental impacts of economic expansion and foreign investment (Udeagha and Muchapondwa, 2023a, 2023b, 2023d).

Overall, the findings suggest that Belgium is experiencing a process of relative decoupling, in which EG continues but at a lower environmental cost, driven by technological progress and the energy transition. However, the persistence of positive growth effects on emissions indicates that full decoupling has not yet been achieved. This conclusion is consistent with recent global evidence showing that, despite advances in REN and INN, emissions reductions are still insufficient to fully offset the environmental pressures associated with economic expansion (Udeagha and Muchapondwa, 2023c; Udeagha and Ngepah, 2019, 2020).

This study examined the dynamic relationship between INN, FDI, EG, REN consumption, and CO₂ emissions in Belgium over the period 1990–2024 using the ARDL framework. The results reveal a stable long-run relationship among the variables and highlight clear differences between short-run adjustments and long-run structural effects.

The positive and statistically significant long-run effect of EG on CO₂ emissions confirms that the scale effect remains dominant in Belgium. Despite the country's advanced industrial structure and relatively strong environmental institutions, increases in real income continue to be associated with higher energy demand and transport activity. This finding suggests that recent improvements in energy efficiency and sectoral transformation have not yet been sufficient to fully offset the emissions generated by expanding economic activity. Similar patterns are frequently observed in advanced European economies, where decarbonization progresses more slowly in hard-to-abate sectors such as industry, freight transport, and buildings.

FDI is also found to increase CO₂ emissions in both the short and long runs, although its estimated magnitude is smaller than that of EG. This result indicates that capital inflows into Belgium have not automatically translated into cleaner production structures. Instead, part of inward investment appears to be linked to energy-intensive production, logistics, and manufacturing activities. Importantly, the relatively modest size of the FDI coefficient suggests that the environmental impact of foreign investment is conditional rather than uniform. This supports the view that FDI is neither inherently harmful nor inherently beneficial for environmental quality; its effect depends on the technological content of investment projects and the host country's ability to channel investment toward low-carbon activities.

In contrast, INN exerts a statistically significant negative effect on CO₂ emissions in both the short- and long-run. This finding provides strong empirical support for the technology and efficiency channels highlighted in the theoretical framework. Increased R&D activity contributes to lower emissions by improving energy efficiency, facilitating cleaner production processes, and accelerating the diffusion of low-carbon technologies. The fact that INN reduces emissions already in the short run is particularly noteworthy, as it implies that technological upgrading and adoption in Belgium occur relatively rapidly, possibly reflecting well-developed INN systems and strong linkages between research institutions and industry.

REN consumption also plays a crucial role in mitigating environmental pressure. The negative, significant coefficients in both the long and short runs indicate that increasing the share of REN in the energy mix has led to a measurable decline in CO₂ emissions. This confirms that renewable deployment in Belgium has not merely supplemented existing fossil-fuel generation but has contributed to its gradual displacement. The magnitude of the REN coefficient is comparable to that of INN, highlighting that structural change in the energy system is at least as important as technological progress in production processes.

An important insight emerging from the combined results is that INN and REN jointly moderate the environmental impact of economic activity and capital inflows. While EG and FDI continue to exert upward pressure on emissions, strong INN performance and expanding REN capacity substantially mitigate these effects. This interaction helps explain why the estimated impact of FDI is relatively small compared with growth, and it supports the argument that domestic technological capability is a key precondition for transforming investment inflows into environmentally sustainable outcomes.

The error-correction mechanism indicates a relatively rapid rate of adjustment toward the long-run equilibrium, with nearly half of short-term deviations corrected within 1 year. This suggests that shocks to emissions caused by temporary fluctuations in growth, investment, or energy use do not persist for long and that the Belgian economy adjusts relatively quickly toward its long-run environmental trajectory. Nevertheless, the persistence of positive growth and FDI effects in the long run implies that structural and policy interventions are required to achieve long-term decarbonization targets.

Overall, the results are consistent with the broader international literature showing that REN and INN are central drivers of emissions reduction, whereas EG remains a major source of environmental pressure. However, the Belgian case shows that even in technologically advanced economies, growth alone does not automatically lead to environmental improvements. Instead, sustained investment in INN and a continued transformation of the energy system are necessary to counterbalance the environmental consequences of production expansion and international capital inflows.

Policy implications

The policy implications of this study are directly derived from the empirical findings and reflect the interaction between forces that increase and reduce emissions. The findings of this study have several important policy implications. First, policymakers should move beyond the assumption that EG will automatically become environmentally sustainable and instead adopt targeted policies to decouple growth from emissions. Second, attracting FDI should be accompanied by strict environmental standards and incentives for green investment, ensuring that foreign capital contributes to sustainability goals. Third, INN policies should focus not only on increasing R&D expenditure but also on promoting environmentally oriented technological development. Finally, accelerating the energy transition requires not only expanding renewable capacity but also addressing structural constraints in energy systems. This study contributes to the policy debate by demonstrating that achieving environmental sustainability in advanced economies requires a coordinated strategy in which growth, investment, INN, and energy policies are aligned toward a common low-carbon objective.

The empirical findings of this study provide several important and evidence-based policy implications for Belgium and other advanced economies pursuing low-carbon development. These implications are directly derived from the estimated relationships among EG, FDI, INN, and REN consumption.

Conclusion, limitations, and future study