Energy Consumption, Economic Growth, and Climate Change Dynamics in Southeast European Countries

Abstract

This paper explores the evolving relationship between energy consumption, economic activity, and carbon emissions in Southeast European countries over the period 2000–2024. Using a dynamic panel framework, the analysis focuses primarily on short-run interactions, given the lack of evidence for a stable long-term equilibrium. The findings reveal that changes in energy consumption remain the main driver of fluctuations in CO₂ emissions, confirming the carbon-intensive nature of energy use across the region. In contrast, economic growth and industrial production do not show a statistically significant direct effect on emissions in the short term. Renewable energy plays a mitigating role, although its impact becomes more visible only when cross-country interdependencies are taken into account. This suggests that regional factors, such as shared policies and energy market shocks, shape environmental outcomes. Overall, the results indicate that emissions are influenced more by immediate changes in energy use than by persistent dynamics, highlighting the ongoing challenge of reducing environmental pressure without fundamentally transforming the energy structure.

1. Introduction

The interaction between economic growth, energy consumption, and environmental sustainability has become one of the defining challenges of contemporary economic policy. Despite decades of progress in environmental regulation and technological innovation, economic expansion remains closely tied to rising energy demand and carbon emissions, particularly in regions where production structures are still energy-intensive and heavily reliant on fossil fuels [1,2,3,4]. This persistent linkage raises a fundamental question: to what extent can economies sustain growth without exacerbating environmental degradation? This question is especially relevant in Southeast Europe, a region undergoing structural transformation while simultaneously facing mounting environmental and energy-related pressures. For the purposes of this study, Southeast Europe refers to Albania, Bosnia and Herzegovina, Croatia, Greece, Italy, Malta, Montenegro, North Macedonia, Portugal, Serbia, Slovenia, and Spain. These countries represent a particularly relevant case study due to their ongoing integration into the European energy and climate framework and their shared exposure to Mediterranean and Balkan energy market dynamics. The region is characterized by relatively high energy intensity, continued dependence on fossil fuel-based energy systems, and a slower and uneven energy transition process. The sample includes both EU member states and Western Balkan economies in order to capture broader regional energy and climate trends. However, important differences remain between these groups, particularly in terms of institutional capacity, renewable energy development, and alignment with EU climate and energy policies [5,6,7]. At the same time, the region is increasingly exposed to common external shocks, including energy price volatility [8], European climate policy adjustments [9], and, more recently, geopolitical disruptions that reshape both energy markets and environmental outcomes [10]. In particular, the energy crisis triggered by the Russia–Ukraine war has revealed structural vulnerabilities in energy supply systems and has reinforced the urgency of diversification and transition strategies [11,12]. A large body of empirical literature has explored the energy growth environment nexus, often focusing on long-run relationships and equilibrium dynamics. While many studies confirm a positive association between economic activity, energy consumption, and CO₂ emissions, the evidence remains far from uniform [13,14]. Increasingly, recent contributions suggest that aggregate measures of energy use are insufficient to explain environmental outcomes [15,16,17]. The composition of the energy mix plays a critical role in shaping emission trajectories. Whether renewable energy substitutes for or merely complements fossil fuels is especially consequential in this regard [18,19,20,21]. Moreover, emerging research highlights that these relationships are inherently dynamic and influenced by cross-country interdependencies, especially in regions exposed to common shocks and shared policy frameworks [22,23]. Despite a growing body of literature, three critical gaps remain. First, empirical evidence for Southeast European economies is still limited and fragmented. Second, existing studies predominantly focus on long-run equilibrium relationships, overlooking short-run dynamics that are particularly relevant in periods of economic and energy instability. Third, insufficient attention has been paid to cross-sectional dependence, despite the region’s exposure to common shocks and shared policy frameworks. This limitation has become increasingly relevant in the aftermath of recent energy market disruptions, which have altered the context in which energy consumption, economic activity, and emissions interact. These developments have further reinforced the importance of energy diversification, renewable energy investment, and regional policy coordination in Southeast Europe, where economies remain highly exposed to external energy shocks. Against this backdrop, this paper investigates the dynamic relationships between energy consumption, economic growth, renewable energy, and CO₂ emissions in Southeast European countries over the period 2000–2024. Unlike much of the existing literature, the analysis explicitly focuses on short-run dynamics, given the absence of a stable long-run equilibrium relationship. The empirical strategy employs a fixed-effects model with Driscoll–Kraay standard errors to account for cross-sectional dependence, complemented by a System GMM estimator to address potential endogeneity and dynamic panel bias. The results provide clear evidence that short-run fluctuations in energy consumption are the primary driver of CO₂ emissions across the region, confirming the carbon-intensive nature of existing energy systems. In contrast, economic growth and industrial activity do not exert a statistically significant direct effect on emissions, while the mitigating role of renewable energy becomes statistically relevant only when cross-sectional interdependencies are properly accounted for. These findings suggest that environmental outcomes in Southeast Europe are shaped less by persistent adjustment mechanisms and more by immediate changes in energy use and common regional shocks.

This paper contributes to the literature in three main ways. First, it provides novel empirical evidence for Southeast European countries, a region that remains underexplored in the energy growth environment nexus. Second, it shifts the analytical focus from long-run equilibrium relationships to short-run dynamics, capturing the effects of recent energy market disruptions. Third, it explicitly accounts for cross-sectional dependence and dynamic endogeneity using Driscoll–Kraay standard errors and System GMM, offering more robust and policy-relevant estimates.

The remainder of the paper is structured as follows. Section 2 reviews the relevant literature. Section 3 presents the methodology. Section 4 describes the data and variables. Section 5 discusses the empirical results, and the Section 6 concludes with policy implications.

2. Literature Review

The relationship between energy consumption, economic growth, and environmental sustainability has become a central research theme in energy economics and environmental policy, particularly in the context of climate change mitigation, energy security, technological transformation, and sustainable development transitions [1,2,3,24]. The theoretical foundation of this debate is commonly anchored in the Environmental Kuznets Curve (EKC) hypothesis, originally articulated by [25], which posits a non-linear relationship between economic growth and environmental degradation. According to this framework, environmental pressures tend to intensify during the early stages of industrialization due to fossil fuel dependence and energy-intensive production structures but may decline as economies adopt cleaner technologies, stricter environmental regulations, and more efficient production systems [1,26]. Recent evidence suggests that decarbonization outcomes increasingly depend on technological innovation, energy security, and coordinated climate policies [24]. However, empirical validation of the EKC remains inconclusive and highly context-dependent. Recent evidence continues to show that the growth emission relationship remains highly sensitive to the energy mix and the pace of structural change, even across European economies pursuing decarbonization strategies [19]. A growing body of evidence suggests that economic growth alone does not guarantee environmental improvement, particularly in emerging and transition economies where institutional constraints, fossil fuel dependence, and rapid urbanization may delay or even offset potential environmental gains [2,3,7]. In the case of European transition economies, empirical findings consistently indicate that increases in GDP and energy consumption are associated with higher carbon emissions, while renewable energy expansion plays a mitigating role. These results challenge the automatic convergence implied by the EKC and point instead to the importance of structural transformation in the energy sector as a prerequisite for sustainable growth [6,18]. Recent empirical studies also suggest that the success of decarbonization strategies depends on the ability of economies to combine renewable-energy expansion with improvements in energy efficiency and institutional capacity [27,28].

More recent European evidence also suggests that decarbonization depends not only on income growth, but on the composition of electricity generation and the extent to which renewable energy replaces fossil-based supply rather than simply complements it [19,20,21]. Beyond the EKC framework, alternative theoretical approaches provide a more comprehensive explanation of the energy-economy-environment nexus. The STIRPAT model (Stochastic Impacts by Regression on Population, Affluence, and Technology) [29,30] extends the IPAT identity by allowing empirical estimation of the joint effects of population, affluence, and technology on environmental impact, with energy intensity often serving as a proxy for technological efficiency. Relatedly, the concept of decoupling distinguishes between relative and absolute reductions in environmental pressure, emphasizing that sustained emissions reductions require not only economic growth but also structural transformation of energy systems and technological upgrading. In this context, energy transition theory highlights the long-term shift from fossil fuel-based systems toward renewable and low-carbon energy sources as a fundamental mechanism for achieving sustainability [31,32]. Complementing these perspectives, green growth and ecological economics offer contrasting views on whether economic expansion and environmental sustainability are compatible, highlighting the tension between technological optimism and biophysical constraints. Within this broader framework, the nexus between energy consumption and economic growth has been extensively examined using causality-based approaches (energy-growth nexus hypothesis) that distinguish between growth, conservation, feedback, and neutrality hypotheses [33,34,35]. Empirical findings, however, remain far from uniform. Studies focusing on OECD countries frequently identify bidirectional causality between energy use and GDP, suggesting a mutually reinforcing relationship between economic activity and energy demand [36,37]. In contrast, evidence from developing and transition economies tends to reveal more asymmetric and context-specific relationships, reflecting differences in economic structure, technological progress, and policy environments [37]. Research on European transition economies confirms the existence of a bidirectional relationship between GDP growth and energy consumption, although excessive energy use may negatively affect economic competitiveness by increasing production costs [38]. Nevertheless, empirical evidence does not always indicate that short-run changes in output translate directly into higher emissions once energy use is controlled, suggesting that the composition and intensity of energy consumption may matter more than aggregate income dynamics alone [2,18]. These findings reinforce the growing consensus that energy efficiency policies are not only environmentally beneficial but also economically strategic. More recent contributions have extended the energy–growth–environmental nexus by incorporating sectoral dimensions, highlighting how specific economic activities can amplify energy demand and environmental pressures. Tourism and industry have been increasingly identified as both drivers of economic growth and potential sources of increased energy consumption and emissions [39]. The expansion of tourism-related activities such as transportation, accommodation, and infrastructure development tends to intensify energy demand, especially in economies that remain dependent on fossil fuels [40,41]. Several empirical studies based on ARDL and panel econometric approaches report a positive association between tourism development, energy consumption, and CO₂ emissions [42,43,44] although some studies suggest that its environmental impact can be mitigated through renewable energy adoption and sustainable infrastructure investment [3,45]. Nevertheless, the effectiveness of such strategies remains uneven and highly dependent on institutional quality, regulatory enforcement, and the structure of the energy system [41]. Recent literature increasingly identifies digitalization and artificial intelligence (AI) as emerging drivers of electricity demand. According to the International Energy Agency [46], global data center electricity consumption is projected to rise from around 415 TWh in 2024 to approximately 945 TWh by 2030, with AI identified as a key driver of this increase. This rapid growth reflects the expanding computational requirements of AI models and digital infrastructure. Complementary evidence suggests that the increasing energy intensity of large-scale AI systems may significantly amplify electricity consumption and associated emissions [47]. These developments indicate that future energy demand may grow substantially even in economies pursuing efficiency improvements, thereby complicating decarbonization efforts unless accompanied by substantial expansion in renewable energy, grid infrastructure, and low-carbon power generation. Recent scholarship has also underlined the growing role of geopolitical shocks in shaping energy demand, supply security, and decarbonization pathways. The wars in Ukraine and the Middle East, in particular, exposed Europe’s dependence on imported fossil fuels and intensified volatility across gas and electricity markets. At the same time, emerging evidence suggests that geopolitical risk can accelerate renewable energy transition under certain conditions, especially when countries respond by reducing exposure to fossil-fuel dependence and restructuring energy trade networks. More broadly, renewed instability in the Middle East has reinforced concerns over supply-chain vulnerability and external energy dependence, highlighting that the energy–growth–environment nexus is increasingly shaped by geopolitical as well as economic factors [12]. Within this broader context, Southeast European economies represent a particularly relevant case due to their ongoing structural transformation and evolving environmental governance frameworks [6]. Recent European evidence further suggests that decarbonization dynamics are increasingly shaped by the interaction between climate regulation, technological innovation, energy prices, and energy security concerns. The European Green Deal, carbon-pricing mechanisms, and innovation-oriented energy policies are becoming central drivers of energy-transition strategies across Europe [24,48]. At the same time, geopolitical instability and persistent fossil-fuel dependence continue to complicate decarbonization pathways, particularly in regions exposed to external energy shocks and volatile energy markets [49]. Emerging evidence also indicates that technological innovation, digital infrastructure expansion, and AI-related electricity demand may substantially reshape future energy consumption patterns and environmental outcomes [47,50]. These countries are typically characterized by relatively high energy intensity and continued reliance on fossil fuels, which complicates efforts to decouple economic growth from environmental degradation [2,7]. At the same time, differences in institutional capacity, renewable-energy deployment, and integration into European climate frameworks contribute to substantial heterogeneity in regional energy-transition pathways. Empirical evidence indicates that both economic growth and energy consumption contribute positively to CO₂ emissions in the region, while renewable energy development plays a mitigating role [38].

Recent empirical literature on the energy–growth–environment nexus has increasingly emphasized the use of advanced econometric approaches to better capture complex and dynamic relationships among key variables. Studies employing ARDL models, panel cointegration techniques, nonlinear specifications, and VAR-based frameworks highlight the importance of distinguishing between short-run adjustments and long-run equilibrium dynamics [22,51]. This body of research underscores that the relationship between economic activity, energy consumption, and environmental outcomes is inherently dynamic and context-dependent, particularly in emerging and transition economies. At the same time, a growing strand of literature points to the importance of accounting for cross-sectional dependence in panel analyses, especially in regional settings where countries are jointly exposed to common shocks, such as energy price volatility, EU climate policies, technological diffusion, and geopolitical disruptions linked to war-related energy insecurity and changes in international energy trade patterns [23,52]. Ignoring these interdependencies may lead to biased estimates and to an underestimation of the environmental role of renewable energy. Recent studies also show that failing to account for nonlinear adjustment processes and regional spillovers may significantly distort estimates of the energy–growth–environment nexus in transition economies [15,21]. Despite these advances, several important research gaps remain. Southeast Europe and the Western Balkans continue to be underrepresented in empirical literature compared to Western European, Asian, and OECD economies [2,6]. Moreover, existing studies often rely on relatively short time spans and therefore fail to fully capture recent structural transformations related to the energy transition, post-crisis adjustments in energy markets, and emerging electricity demand pressures associated with digitalization and AI-related infrastructure. This limitation has become more important after the Russia–Ukraine war, which altered energy security priorities, accelerated renewable energy debates, and changed the strategic context in which European economies pursue decarbonization [11]. In addition, relatively few contributions jointly examine short-run dynamics and long-run relationships within a unified empirical framework, limiting the ability to fully understand adjustment processes in the energy–environment nexus. More broadly, the literature highlights that this nexus is shaped by complex interactions among economic structure, technological progress, institutional quality, and policy interventions. In transition economies, while economic growth can generate significant welfare gains, it may also intensify energy demand and environmental pressures in the absence of effective energy transition policies [45].

3. Methodology

This study investigates the dynamic interrelationships between energy consumption, economic growth, and climate-related environmental indicators in Southeast European economies. In the absence of panel cointegration, the analysis focuses on short-run dynamics rather than long-run equilibrium relationships. This implies that the empirical specification does not impose a long-run equilibrium structure but instead captures contemporaneous and dynamic adjustments in emissions. Given the relatively small cross-sectional dimension (N = 12) and moderate time dimension (T = 25), the fixed-effects estimator is adopted as the baseline approach, as it provides consistent estimates in panels of this structure. In particular, fixed effects allow for controlling unobserved country-specific heterogeneity, such as differences in energy mix, institutional quality, and structural characteristics of the economy. To address potential persistence and endogeneity concerns arising from the inclusion of a lagged dependent variable, the System GMM estimator is employed as a robustness check. This approach helps mitigate dynamic panel bias and potential simultaneity between emissions and explanatory variables. All variables are transformed into first differences to ensure stationarity and avoid spurious regression issues. This transformation is consistent with the integration properties of the data, as unit root tests indicate that the variables are non-stationary in levels but stationary in first differences. Standard errors are computed using the Driscoll–Kraay correction to account for heteroskedasticity, serial correlation, and cross-sectional dependence. Before model estimation, cross-sectional dependence was assessed using the Pesaran CD test, which did not indicate strong cross-sectional correlation across countries. However, the Modified Wald and Wooldridge tests revealed the presence of heteroskedasticity and first-order serial correlation. Therefore, Driscoll–Kraay standard errors were employed to ensure robust inference while also accounting for potential weak forms of residual cross-sectional dependence associated with common regional shocks and interconnected energy markets. This is particularly important in a regional context such as Southeast Europe, where countries are jointly exposed to common shocks, including energy price fluctuations, EU climate policies, and regional economic developments. System GMM results are reported solely for robustness purposes, confirming that the baseline findings are not driven by dynamic panel bias.

ΔlnCO_it = α_i + β₁ΔlnCO_i,t−1 + β₂ΔlnEN_it + β₃ΔlnREN_it + β₄ΔlnGDP_it + β₅ΔlnIPI_it + ε_it

where

• $\Delta \mathrm{l}\mathrm{n} C{O}_{it}$: change (first difference in logs) of CO₂ emissions for country $\mathrm{i}$ at time $\mathrm{t}$;
• $\Delta \mathrm{l}\mathrm{n} C{O}_{i,t-1}$: lagged dependent variable (persistence in emission growth);
• $\Delta \mathrm{l}\mathrm{n} E{N}_{it}$: change in energy consumption;
• $\Delta \mathrm{l}\mathrm{n} RE{N}_{it}$: change in renewable energy use;
• $\Delta \mathrm{l}\mathrm{n} GD{P}_{it}$: change in income/output;
• $\Delta \mathrm{l}\mathrm{n} IP{I}_{it}$: change in Industrial Production;
• ${\alpha }_i$: country-specific fixed effects (time-invariant heterogeneity);
• ${\epsilon }_{it}$: idiosyncratic error term.

ΔlnCO_it = β₁ΔlnCO_i,t−1 + β₂X_it + η_i + ε_it

where

• $X_{it}=(\Delta \ln E{N}_{it},\Delta \ln RE{N}_{it},\Delta \ln GD{P}_{it},\Delta \mathrm{l}\mathrm{n}{IPI}_{it})$;
• ${\eta }_i$ is eliminated by first-differencing;
• $\Delta \mathrm{l}\mathrm{n} C{O}_{i,t-1}$ is instrumented using internal GMM instruments (lags 2–3).

Diagnostic tests support the validity of the System GMM specification. The Arellano-Bond test indicates no evidence of second-order serial correlation in the differenced residuals, thereby validating the moment conditions. The Hansen test does not reject the null hypothesis of instrument validity, while the difference-in-Hansen test further supports the exogeneity of the instrument subsets. However, the Hansen test excluding the level instruments yields a borderline result (p = 0.051), suggesting weak evidence of potential concerns regarding this subset. Overall, the diagnostic tests indicate that the instrument set is broadly valid. The empirical strategy proceeds in several steps. First, panel unit root tests, including Levin-Lin-Chu [53] and Im, Pesara & Shin [54], are conducted to assess the time-series properties of the variables. The results indicate that the series are non-stationary in levels but become stationary after first differencing, implying integration of order one, I(1). Panel cointegration tests, based on Pedroni [54,55,56] and Westerlund [57], are then applied to the level variables, and the findings do not support the existence of a long-run equilibrium relationship. Considering these results, the empirical model is specified in first differences, following the standard approach to avoid spurious regression [57]. A dynamic specification is adopted by including the lagged dependent variable, allowing for short-run persistence in emission dynamics. This specification is consistent with the objective of capturing short-term adjustments rather than long-run convergence. Given the panel structure and the likelihood of cross-sectional dependence and serial correlation, inference is based on Driscoll–Kraay standard errors. As an additional robustness check, the model is re-estimated using fixed effects with standard errors clustered at the country level. Finally, for descriptive purposes, the model is also estimated in levels using fixed effects to illustrate cross-country heterogeneity. However, these results are interpreted cautiously and are not used for inference, due to the non-stationarity of the variables and the absence of cointegration. Given the presence of complex feedback mechanisms across economic and environmental systems, the empirical analysis employs advanced panel econometric techniques capable of capturing both long-run equilibrium relationships and short-run dynamics [1,2,50,58]. Such an approach is particularly relevant for Southeast Europe, where structural transformation, evolving environmental governance, and energy transition processes occur simultaneously [3,6].

4. Data and Variables

The empirical analysis is based on a balanced panel dataset of 12 Southeast European countries, comprising both European Union member states and Western Balkan economies, with a total of 240 observations and an average time dimension of approximately 20 years per country. For the purposes of this study, Southeast Europe refers to Albania, Bosnia and Herzegovina, Croatia, Greece, Italy, Malta, Montenegro, North Macedonia, Portugal, Serbia, Slovenia, and Spain. The selection of these countries reflects their shared exposure to Mediterranean and Balkan energy-market dynamics, relatively high energy intensity, continued dependence on fossil fuels, and ongoing efforts to align with the European climate and energy framework. Although the sample combines both EU member states and Western Balkan economies, these countries face similar challenges related to energy security, decarbonization, and renewable-energy transition. Southern European countries such as Italy, Spain, and Portugal are included because of their strong economic and energy interconnections with the broader Mediterranean region and their active role in regional climate and energy-policy coordination [24]. This regional focus allows for capturing heterogeneity in economic structures, energy systems, and stages of development, which is particularly relevant in the context of ongoing structural transformation and environmental policy convergence associated with European integration [6].

The dataset includes key macroeconomic and energy-related variables commonly used in the energy–growth–environment literature. Specifically, the analysis focuses on CO₂ emissions, energy consumption, renewable energy use, real economic activity (proxied by GDP), and industrial production. These variables are widely recognized as central to understanding the interaction between economic development, energy demand, and environmental pressure [1,18,34]. CO₂ emissions are used as the primary indicator of environmental degradation, given their dominant role in measuring anthropogenic climate impact in empirical macroeconomic studies [2,7]. The data used in this study were collected from internationally recognized databases, mainly the World Bank’s World Development Indicators, the International Energy Agency. CO₂ emissions are measured in metric tons per capita, while energy consumption refers to total primary energy use. Renewable energy is defined as the share of renewable energy in total final energy consumption. GDP is expressed in constant prices to improve comparability across countries and over time, and industrial production is included as an indicator of industrial activity. To ensure consistency and avoid spurious regression problems, all variables were transformed into natural logarithms and first differences before estimation. To ensure comparability and consistency, all variables are transformed into natural logarithms and subsequently expressed in first differences. This approach is consistent with standard practice in panel data analysis, particularly when variables exhibit non-stationarity, as it helps avoid spurious regression results and allows for elasticity-based interpretation of coefficients [57,58,59].

Panel unit root tests, including Im, Pesaran & Shin [54], indicate that the variables are integrated of order one, I(1). However, panel cointegration tests based on Pedroni [55,56] and Westerlund [57] do not provide evidence of a long-run equilibrium relationship among the variables. Considering these results, the empirical model is specified in first differences, following the standard approach to avoid spurious regression [58]. A dynamic specification is adopted by including the lagged dependent variable, allowing for short-run persistence in emission dynamics. This specification is consistent with the objective of capturing short-term adjustments rather than long-run convergence. Given the panel structure and the likelihood of cross-sectional dependence and serial correlation, inference is based on Driscoll–Kraay standard errors. As an additional robustness check, the model is re-estimated using fixed effects with standard errors clustered at the country level. Finally, for descriptive purposes, the model is also estimated in levels using fixed effects to illustrate cross-country heterogeneity. However, these results are interpreted cautiously and are not used for inference, due to the non-stationarity of the variables and the absence of cointegration. Given the presence of complex feedback mechanisms across economic and environmental systems, the empirical analysis employs advanced panel econometric techniques capable of capturing both long-run equilibrium relationships and short-run dynamics [2,51,59]. Such an approach is particularly relevant for Southeast Europe, where structural transformation, evolving environmental governance, and energy transition processes occur simultaneously [3,6].

Consequently, the empirical analysis focuses on short-run dynamics rather than long-run relationships.

The baseline estimation is conducted using the fixed-effects estimator, which effectively controls unobserved country-specific heterogeneity [60]. Given the inclusion of a lagged dependent variable and the potential for endogeneity, the System GMM estimator is employed as a robustness check, following the standard approach in dynamic panel data models [61,62]. In addition, Driscoll–Kraay standard errors are used to ensure robust inference in the presence of heteroskedasticity, serial correlation, and cross-sectional dependence (see Appendix A) [63].

Overall, this empirical framework is consistent with established practices in the energy economics literature and provides a reliable basis for examining the short-run relationship between energy consumption, renewable energy, economic activity, and environmental pressure in Southeast European countries.

5. Results

The results consistently show that CO₂ emissions are primarily driven by short-run fluctuations in energy consumption, confirming the carbon-intensive nature of energy use across countries. The absence of strong persistence in the lagged dependent variable suggests that short-run changes in emissions are not heavily path-dependent.

The coefficient of energy consumption exceeds unity across all specifications, suggesting that a 1% increase in energy consumption is associated with an increase in CO₂ emissions of more than 1%. This elasticity indicates that the energy systems of Southeast European economies remain heavily dependent on carbon-intensive sources, including coal, oil, and natural gas. The magnitude of the coefficient further suggests limited improvements in energy efficiency and incomplete decoupling between economic activity and environmental degradation. In this context, increases in energy demand continue to translate almost proportionally and in some cases more than proportionally into higher emissions, reflecting the structural rigidity of regional energy systems and the relatively slow pace of the energy transition. The findings are consistent with the broader literature on emerging and transition economies, where fossil-fuel dependence and industrial energy intensity remain significant determinants of environmental degradation.

The renewable-energy coefficient becomes statistically significant only when heteroskedasticity, serial correlation, and potential cross-sectional dependence are explicitly addressed through Driscoll–Kraay standard errors. This finding suggests that the environmental effectiveness of renewable energy in Southeast Europe cannot be interpreted solely at the national level, as it is increasingly influenced by common regional and global dynamics. These include EU climate policies, regional electricity market integration, technology diffusion, and synchronized energy price shocks. The result implies that renewable energy transitions in Southeast Europe exhibit spillover effects across countries, where policy developments and technological adoption in one economy may indirectly affect neighbouring countries through integrated energy systems and regulatory convergence. In this sense, the result supports the view that renewable energy development in Southeast Europe should be understood as part of a regionally integrated transition process, rather than a set of isolated national policy outcomes.

In contrast, income and industrial dynamics do not exert a statistically significant influence on emissions. The absence of a statistically significant GDP effect may indicate that short-run fluctuations in economic activity are not the primary drivers of emission dynamics in the region. Instead, emissions appear to depend more directly on the composition and intensity of energy consumption than on aggregate income growth itself. This finding may reflect structural transformation processes already underway in some Southeast European economies, where output growth is becoming gradually less emission intensive. Similarly, the lack of significance for industrial production suggests that industrial fluctuations alone are insufficient to explain short-run changes in emissions once aggregate energy consumption is controlled for. This may imply that the environmental impact of industrial activity operates primarily through the energy channel rather than through production dynamics per se.

The weak persistence in emission dynamics provides limited support for strong path dependence and only indirect evidence consistent with convergence in CO₂ emissions. Differences across estimators suggest that adjustment processes may be heterogeneous across countries. Dynamic panel presentet in

Table 1. Results.

Variables	FE (Cluster)	p Value	Driscoll–Kraay	p Value	System GMM	p Value
LdlnCO2	−0.037	0.657	−0.022	0.729	0.1126	0.820
Dlnen	1.177 ***	0.000	1.182 ***	0.000	1.250 **	0.022
Dlnren	−0.06	0.305	−0.064 **	0.029	−0.104	0.144
Dlngdp	−0.075	0.468	−0.048	0.748	−0.116	0.623
Dlnipi	−0.009	0.944	0.005	0.927	−0.097	0.542
Constant	−0.004	0.134	−0.005 **	0.019	−0.000	0.970

Standard significance levels are denoted as follows: *** p < 0.01, ** p < 0.05.

results indicate relatively low, though not negligible, inertia in environmental degradation, implying that emissions respond with partial but not complete persistence over time. Instead, changes in emissions are largely associated with variations in energy consumption, while renewable energy becomes statistically significant once Driscoll–Kraay standard errors are applied, suggesting that its estimated effect is sensitive to the treatment of heteroskedasticity, serial correlation, and potential cross-sectional dependence in the error structure.

Country Effect

Following DK estimation, country-specific fixed effects are retrieved using the predict, u option in Stata v.18, which provides estimates of the unobserved time-invariant heterogeneity for each country. These estimated effects capture unobserved time-invariant heterogeneity across countries after controlling for observable determinants of emissions. To facilitate interpretation, countries are grouped into four qualitative clusters according to the relative magnitude and sign of the estimated fixed effects. The cluster classification is descriptive rather than statistical and is intended to illustrate broad patterns of heterogeneity across Southeast European economies. The cluster categorization relies on an interpretative grouping based on the relative distribution of the estimated country fixed effects. Specifically, countries are ordered according to their estimated fixed effects and then partitioned into four groups based on relative thresholds, distinguishing between countries with systematically higher-than-average and lower-than-average unobserved emission propensities. This cluster classification is therefore intended as an interpretative device to summarize persistent cross-country heterogeneity in a transparent manner, rather than as an inference-driven segmentation technique. It allows for the identification of broad regional patterns across Southeast European economies, highlighting whether countries tend to share comparable emission profiles or whether there is substantial dispersion in unobserved country-specific effects even after controlling for observable determinants.

The estimated fixed effects reveal substantial heterogeneity across countries in their baseline emission levels. Countries with positive fixed effects exhibit structurally higher emissions, likely reflecting carbon-intensive energy systems and industrial composition, whereas negative effects indicate relatively cleaner structural characteristics. This heterogeneity suggests that, beyond observable factors such as energy consumption and renewable energy, unobserved country-specific characteristics play a non-negligible role in shaping emission dynamics.

As can be seen in

Table 2. Clusters.

Cluster 1—High persistent emitters	Cluster 2—Moderate emitters
Greece, Bosnia	Malta
	North Macedonia
	Italy
Cluster 3—Neutral emitters	Cluster 4—Stable low emitters
Serbia	Albania
Portugal	Montenegro
	Slovenia
	Croatia
	Spain

presented above, Greece and Bosnia and Herzegovina exhibit positive and relatively large effects, indicating a persistent carbon-intensive structure, whereas Slovenia and Montenegro display strongly negative values, reflecting more favorable structural conditions. Countries such as Spain, Croatia, and Albania also show negative effects, suggesting comparatively cleaner emission profiles. In contrast, Italy, Malta, and North Macedonia appear close to neutrality, indicating that their emission levels, like Serbia and Portugal, are largely explained by observable factors. Overall, these findings highlight that, beyond energy consumption, renewable energy, and income dynamics, unobserved country-specific characteristics play a key role in shaping emission patterns.

6. Conclusions

This paper provides new evidence on the short-run dynamics of the energy–economy–environment nexus in Southeast European countries, emphasizing the importance of moving beyond long-run equilibrium frameworks in contexts characterized by structural transformation and regional interdependencies. By combining fixed-effects estimation with Driscoll–Kraay corrections and System GMM robustness checks, the analysis offers a consistent and methodologically robust assessment of emission dynamics in the region.

The results reveal a clear and robust pattern: short-run changes in energy consumption are the primary driver of CO₂ emissions, with elasticity estimates consistently exceeding unity. This finding highlights the strongly carbon-intensive nature of energy use in Southeast Europe and points to the absence of effective decoupling between energy demand and environmental pressure. In contrast, economic growth and industrial production do not exert a statistically significant effect on emissions, suggesting that environmental outcomes are shaped less by the scale of economic activity and more by the structure and intensity of energy use.

The role of renewable energy emerges as conditional but non-negligible. Its mitigating effect becomes statistically significant only when cross-sectional dependence is explicitly accounted for, indicating that its impact is mediated by regional spillovers, common policy frameworks, and shared exposure to external shocks. This underscores the importance of incorporating cross-country interdependencies in empirical analyses of energy transitions, particularly in integrated regional settings.

Moreover, the absence of persistence in the lagged dependent variable suggests that emission dynamics are largely contemporaneous, with limited evidence of path dependence. This may indicate that environmental outcomes respond more directly to short-run changes in energy consumption rather than evolving through gradual adjustment processes. At the same time, the presence of substantial country-specific heterogeneity indicates that structural factors, such as energy mix composition and institutional characteristics, remain central in explaining cross-country differences in emission levels.

From a policy perspective, these findings suggest that meaningful decarbonization in Southeast Europe requires structural transformation of the energy system rather than reliance on incremental or growth-driven adjustments alone. Policy priorities should distinguish between short-term energy-security concerns and longer-term decarbonization objectives. In the short term, particularly in fossil-fuel-dependent Western Balkan economies, policy efforts should focus on improving energy efficiency, reducing vulnerability to external energy-price shocks, and strengthening the resilience of electricity markets. Over the longer term, accelerating renewable-energy investment, expanding regional electricity interconnections, and improving alignment with the EU Green Deal will be essential for supporting a more coordinated and sustainable energy transition across the region.

Future research could extend this framework by incorporating emerging drivers such as digitalization, artificial intelligence, and geopolitical risk, which are likely to play an increasingly important role in shaping energy demand and environmental outcomes.

Some limitations should nevertheless be acknowledged. First, the analysis focuses mainly on short-run dynamics, as the results do not support a stable long-run relationship among the variables. Second, structural differences across Southeast European economies may limit the broader applicability of the findings. Finally, the study does not explicitly account for factors such as institutional quality, technological innovation, or geopolitical risk, which may also influence energy-transition dynamics and environmental outcomes.