Dynamic drivers of carbon emission reduction in Lithuania under the European Green Deal: a wavelet analysis of synergies in waste management, energy efficiency, R&D, and international legal compliance Authors: Jingchi Wen, Babatunde Sunday Eweade, Huseyin Ozdeser, Berna Uzun Published: 2026 ABSTRACT Reducing carbon emissions remains a major challenge for Lithuania under the European Green Deal, particularly due to the dominance of energy-related emissions and evolving economic dynamics. This study investigates the key drivers of CO₂ emissions, focusing on energy intensity, research and development (R&D), waste management, international legal compliance, economic growth, and trade openness over the period 1996Q1–2024Q4. To capture nonlinear, time-varying, and frequency-dependent relationships, the study employs advanced wavelet-based techniques, including Wavelet Power Spectrum, Wavelet Coherence, and Partial Wavelet Coherence. The findings reveal that energy intensity is the most dominant and persistent driver of emissions, exhibiting strong long-run effects, while economic growth also contributes positively, confirming the presence of scale effects. In contrast, R&D and institutional quality show conditional and time-dependent emission-reducing impacts, whereas waste management plays a relatively weak and indirect role. Based on these results, the study recommends prioritizing energy efficiency improvements, strengthening green innovation policies, enhancing institutional effectiveness, and accelerating the transition toward renewable energy systems to achieve sustainable emission reductions. Introduction Climate change mitigation has become one of the most pressing global challenges of the 21st century, with the reduction of carbon dioxide (CO₂) emissions recognized as a central objective for achieving environmental sustainability and long-term economic resilience [1–3]. Rising greenhouse gas emissions, largely driven by fossil fuel consumption and industrial activity, continue to accelerate global warming and intensify climate-related risks. In response, governments and international organizations have introduced ambitious frameworks to decarbonize economies and transition toward sustainable energy systems [4]. Among these, the European Union (EU) has taken a leading role through the European Green Deal, which establishes a legally binding commitment to achieve climate neutrality by 20501. This policy framework requires coordinated emission-reduction strategies across key sectors, including energy, industry, transportation, and waste management. Lithuania represents a particularly relevant case within this transition. As a small, open, post-transition economy, it has made significant progress in reducing greenhouse gas emissions, achieving a decline of over 60% since 1990. Despite this progress, emissions remain uneven across sectors, with transport and energy continuing to account for a substantial share2. Moreover, Lithuania’s transition is shaped by evolving institutional frameworks, energy restructuring, and integration into EU markets, making its emission dynamics more complex than those of more mature economies. This context makes Lithuania an important yet underexplored case for understanding how structural, economic, and policy-related factors jointly influence carbon emissions in smaller EU member states3. A substantial body of literature has examined the determinants of carbon emissions, highlighting the roles of energy consumption, economic growth, technological innovation, and globalization. Energy intensity has consistently been identified as a key driver, as higher energy use per unit of output leads to increased emissions [5–7]. Economic growth has also been widely linked to environmental degradation, particularly in early development stages, as described by the Environmental Kuznets Curve (EKC) hypothesis [8]. At the same time, research and development (R&D) promotes cleaner technologies and energy efficiency, although its effects are often nonlinear and dependent on institutional and economic conditions [9, 10]. Additional factors such as waste management, institutional quality, and trade openness further shape environmental outcomes through regulatory, structural, and technological channels [11, 12]. Recent empirical studies have emphasized that carbon emission dynamics are inherently complex and shaped by multidimensional interactions among economic, technological, and institutional factors. For example [13], demonstrated that the drivers of CO₂ emissions vary significantly across countries due to differences in energy structure and development stages. Despite these advances, several important limitations remain. First, most existing studies focus on large economies or cross-country panels, thereby overlooking the specific dynamics of smaller post-transition EU economies such as Lithuania. Second, prior research often examines emission drivers in isolation, limiting the ability to capture their combined and interactive effects. Third, conventional linear econometric approaches dominate the literature, making it difficult to account for nonlinear, time-varying, and frequency-dependent relationships. As a result, existing studies provide only a partial understanding of how emission drivers evolve over time, particularly under policy frameworks such as the European Green Deal. Against this backdrop, this study makes several important contributions to the literature. First, it provides a novel country-specific analysis of Lithuania, addressing a clear gap in the literature where smaller post-transition EU economies remain underexplored. Second, the study develops an integrated analytical framework that simultaneously examines multiple key drivers of CO₂ emissions—energy intensity, research and development, waste management, international legal compliance, economic growth, and trade openness—thereby offering a more comprehensive understanding of emission dynamics. Third, it introduces a methodological innovation by applying advanced wavelet-based techniques, including Wavelet Power Spectrum, Wavelet Coherence, and Partial Wavelet Coherence, to capture time-varying, nonlinear, and frequency-dependent relationships [14, 15]. Importantly, the significance of this study extends beyond the Lithuanian case. By uncovering how emission drivers interact over time within a post-transition economy, the findings provide insights that are highly relevant to other EU countries and emerging economies undergoing similar structural transformations. This enhances the study’s appeal to a broader academic and policy-oriented readership, particularly those interested in climate policy, sustainable development, and energy transition. In line with these objectives, this study seeks to answer the following research questions: (i)How do energy intensity, research and development, waste management, international legal compliance, economic growth, and trade openness influence CO₂ emissions in Lithuania?(ii)Which of these factors act as the most dominant drivers of emissions, and how do their effects vary over short-, medium-, and long-run periods? (i)How do energy intensity, research and development, waste management, international legal compliance, economic growth, and trade openness influence CO₂ emissions in Lithuania? How do energy intensity, research and development, waste management, international legal compliance, economic growth, and trade openness influence CO₂ emissions in Lithuania? (ii)Which of these factors act as the most dominant drivers of emissions, and how do their effects vary over short-, medium-, and long-run periods? Which of these factors act as the most dominant drivers of emissions, and how do their effects vary over short-, medium-, and long-run periods? The remainder of the paper is structured as follows. The next section reviews the relevant literature and theoretical framework, followed by data and methodology. The empirical results are then presented and discussed, and the final section concludes with policy implications and limitations. Theoretical framework and hypotheses development This study is anchored in an integrated theoretical framework that explains how economic, technological, institutional, and structural factors influence carbon dioxide (CO₂) emissions, which is treated as the dependent variable. The analysis considers energy intensity (ENI), research and development (RD), waste management (WAM), international legal compliance (ILC), economic growth (GDP), and trade openness (TRA) as key determinants of environmental outcomes. The framework draws on several complementary theories, including the Environmental Kuznets Curve (EKC), Endogenous Growth Theory, Ecological Modernization Theory, Institutional Theory, and the Trade–Environment Nexus. The relationship between economic growth and environmental degradation is primarily explained by the Environmental Kuznets Curve (EKC) hypothesis, which posits that environmental degradation initially increases with economic growth due to scale effects, industrial expansion, and rising energy demand, before eventually declining at higher income levels as a result of technological advancement and structural transformation [8]. In the case of Lithuania, a post-transition economy, this decoupling process remains incomplete, suggesting that economic growth continues to exert upward pressure on emissions. Accordingly, economic expansion is expected to increase CO₂ emissions. Energy intensity represents a core structural driver of emissions, as it reflects the amount of energy required to produce a unit of economic output. The energy–environment nexus suggests that higher energy intensity leads to increased reliance on fossil fuels, thereby elevating CO₂ emissions [16, 17]. Conversely, improvements in energy efficiency reduce energy consumption and contribute to emission mitigation. However, in economies where fossil fuels still dominate the energy mix, the positive relationship between energy intensity and emissions is likely to persist [20]. The role of research and development is grounded in Endogenous Growth Theory, which emphasizes innovation as a key driver of technological progress and long-term economic performance [21]. Investment in R&D facilitates the development of energy-efficient technologies, cleaner production processes, and environmentally friendly innovations, thereby reducing carbon emissions [22]. Nevertheless, the environmental benefits of R&D may not be immediate, as they often depend on the level of technological diffusion, institutional support, and economic conditions. Despite these limitations, the long-run effect of R&D is expected to contribute to emission reduction. Waste management is explained through Pollution Control Theory and Circular Economy Theory, which emphasize reducing waste generation, increasing recycling, and minimizing landfill dependency to limit environmental degradation [23, 23]. In this study, waste management is proxied by hazardous waste per capita, where higher values reflect inefficient waste practices and greater environmental pressure. Poor waste management contributes to increased emissions through methane release and inefficient disposal systems, suggesting a positive relationship between WAM and CO₂ emissions. Institutional quality, represented by international legal compliance, is rooted in Institutional Theory, which highlights the importance of governance, regulatory effectiveness, and rule of law in shaping environmental outcomes. Strong institutions facilitate the enforcement of environmental regulations, enhance compliance with international climate agreements, and promote the adoption of sustainable technologies [24–26]. Therefore, improved institutional quality is expected to reduce CO₂ emissions by strengthening environmental governance and policy effectiveness. Finally, the effect of trade openness is explained by the Trade–Environment Nexus, which operates through scale, composition, and technique effects. Trade expansion can increase emissions through higher production and energy consumption (scale effect), alter the industrial structure toward either cleaner or more polluting sectors (composition effect), and facilitate the transfer of cleaner technologies (technique effect). In many developing and transition economies, the scale effect tends to dominate, resulting in increased emissions [44, 54]. As such, trade openness is expected to exert an upward influence on CO₂ emissions. Based on the above theoretical arguments, the study proposes the following hypotheses. The fig. 1 shows the conceptual framework of the study under investigation. H1 Economic growth (GDP) has a positive effect on CO₂ emissions. H2 Energy intensity (ENI) has a positive effect on CO₂ emissions. H3 Research and development (RD) reduce CO₂ emissions. H4 Waste management (WAM) has a positive effect on CO₂ emissions. H5 International legal compliance (ILC) reduces CO₂ emissions. H6 Trade openness (TRA) has a positive effect on CO₂ emissions. Fig. 1Conceptual framework of the study Conceptual framework of the study Literature review [20] examined the role of supply chain digitization in shaping technological progress bias and its implications for energy intensity in China’s industrial sectors. Using a supply-side economics framework and a three-factor nested CES production function over the 2008–2021 period, they found that digitization induced capital-biased technological progress, reducing the relative share of energy inputs. The study further showed that this process lowered energy intensity, with varying effects across sectors depending on competition levels and supply chain concentration [21]. examined the effects of renewable energy and tourism demand on SDGs in the seven most visited countries from 2003 to 2019 using IV-GMM, CS-ARDL, AMG, CCEMG, and Panel Quantile Regression. They found that renewable energy positively affected SDGs, while tourism demand was insignificant in linear models. However, quantile results showed significant relationships for both variables across different levels. [22] examined the impact of economic, social, governance factors and climate policy uncertainty on CO₂ emissions globally from 2001 to 2020 using ARDL models. They found that economic factors increased emissions, while social factors helped reduce them, and climate policy uncertainty had no significant effect. The results highlighted the need to align economic growth with social development to achieve environmental sustainability [23]. investigated how economic growth, FDI, institutional quality, and democracy influenced the low-carbon energy transition in the USA and China using a quantile-on-quantile regression approach. They found that the effects of economic growth and institutional quality varied across consumption levels and between countries, reflecting differences in governance structures. The study emphasized the need for targeted policy reforms to support sustainable energy transitions in line with SDGs. [33] examined the effects of energy transition and energy-related R&D investments on CO₂ emissions in Nordic countries using a wavelet local multiple correlation model with monthly data from 2000 to 2021. They found that energy transition and R&D investments had mixed effects on emissions across frequencies, while energy efficiency R&D generally reduced emissions in some cases. The study highlighted that different types of R&D investments play varying roles, suggesting the need for targeted decarbonization policies [34]. investigated the causal relationships between renewable energy consumption, economic growth, and CO₂ emissions in G7 countries from 1970 to 2021 using the Bootstrap Rolling Window Granger Causality approach. They found no consistent causal relationship across the full sample due to parameter instability, although significant linkages appeared in certain sub-periods. The study suggested that these variations reflect structural economic changes and help explain inconsistent findings in the literature. [35] examined the effects of green bonds, clean energy production, oil prices, and geopolitical risk on sectoral decarbonization from January 2019 to July 2024 using the KRLS approach. The findings showed that green bonds and nuclear energy did not significantly support decarbonization, while solar, wind, and hydro energy contributed positively in specific sectors. Additionally, oil prices and geopolitical risk enhanced decarbonization outcomes, with results varying across sectors and demonstrating strong predictive accuracy [36]. systematically evaluated China’s industrial decarbonization policy framework using bibliometric analysis and the Policy Modeling Consistency (PMC) Index on 58 national policy documents. The results showed high consistency in carbon emission abatement (9.07), energy efficiency (8.14), and technological innovation (8.12), but weaker performance in socio-economic risk mitigation (6.97) and the circular economy (6.77). The study highlighted policy misalignment across pathways, particularly the limited integration of just transition and circularity, and recommended strengthening circular economy governance and social support mechanisms to improve overall policy coherence. [37] analysed the determinants of consumption-based CO₂ emissions in Asian countries from 1991 to 2020, focusing on financial development, green technology innovation, urbanization, and trade openness. Using the CS-ARDL model to addressed cross-sectional dependence and heterogeneity, the study found that financial development increases emissions, while green technology significantly reduces them. It also revealed that green innovation helps offset the environmental impact of financial growth, whereas urbanization and trade openness have limited effects. Research gap Despite extensive research on the determinants of carbon emissions, several gaps remain. First, existing studies largely focus on large economies or panel analyses, with limited attention to small post-transition countries like Lithuania, where structural and institutional dynamics differ. Second, prior research often examines emission drivers in isolation, neglecting a comprehensive framework that jointly considers energy intensity, R&D, waste management, institutional quality, economic growth, and trade openness. Third, most studies rely on conventional linear methods, which fail to capture the time-varying and nonlinear relationships among variables. These limits understanding of how these factors influence emissions across different time horizons. Finally, the application of advanced wavelet-based approaches remains limited in country-specific EU contexts, particularly in relation to the European Green Deal. This study addresses these gaps by integrating multiple emission drivers within a unified framework and employing wavelet-based techniques to capture dynamic and frequency-dependent relationships in Lithuania. Data and methods Data This study analyzes the impact of energy intensity (ENI), research and development (RD), waste management (WAM), international legal compliance (ILC), economic growth (GDP), and trade openness (TRA) on carbon emissions (CO₂) in Lithuania. The analysis uses quarterly data from 1996Q1 to 2024Q4, where CO₂ emissions (metric tons per capita) serve as the dependent variable, and ENI, RD, WAM, ICC, GDP, and TRA are the independent variables. Energy intensity (ENI) is measured as primary energy consumption per unit of GDP, while RD is proxied by R&D expenditure (% of GDP), and WAM by hazardous waste per capita (kg), all sourced from Our World in Data (2025). ICC is represented by the rule of law indicator, while GDP (constant 2015 US$) and TRA (% of GDP) are obtained from the World Bank (2025). Annual data were converted into quarterly frequency using linear interpolation to ensure suitability for wavelet analysis. Table 2 presents the data sources. Table 2Data sourcesVariablesCodesUnitsSourcesCO₂ emissionsCO₂Metric tons per capitaOWIDEnergy intensityENIPrimary energy consumption per unit of gross domestic product (GDP), in kilowatt-hours per dollarOWIDResearch & developmentRD% of GDPOWIDWaste managementWAMHazardous waste generated per capita (kg)OWIDInternational legal complianceILCRule of lawWBEconomic growthGDPGDP per capita (constant 2015 US$)WBTrade opennessTRATrade (% of GDP)WBWB: World Bank; OWID: Our World in Data Table 2Data sourcesVariablesCodesUnitsSourcesCO₂ emissionsCO₂Metric tons per capitaOWIDEnergy intensityENIPrimary energy consumption per unit of gross domestic product (GDP), in kilowatt-hours per dollarOWIDResearch & developmentRD% of GDPOWIDWaste managementWAMHazardous waste generated per capita (kg)OWIDInternational legal complianceILCRule of lawWBEconomic growthGDPGDP per capita (constant 2015 US$)WBTrade opennessTRATrade (% of GDP)WBWB: World Bank; OWID: Our World in Data Data sources WB: World Bank; OWID: Our World in Data Methods Wavelet signal processing, originally proposed [36], has been widely applied in economic and financial analysis to examine time-frequency relationships between variables. Unlike traditional techniques such as the Fourier transform, wavelet analysis effectively handles non-stationary time series and captures both temporal and frequency dynamics simultaneously [14, 15]. In this study, the Morlet wavelet is employed due to its suitability in balancing time and frequency localization. The mother wavelet function is defined as:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\psi\left(t\right)={\pi}^{-\frac{1}{4}}{e}^{i{\omega}_{0}t}{e}^{-\frac{1}{2}{t}^{2}}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\omega}_{0}$$\end{document} represents the central frequency, commonly set to 6 to ensure an optimal trade-off between time and frequency resolution [14, 37]. The continuous wavelet transform (CWT) is obtained by convolving the time series with scaled and translated versions of the mother wavelet:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{W}_{x}\left(a,b\right)={\int\:}_{-{\infty\:}}^{{\infty\:}}\:x\left(t\right)\frac{1}{\sqrt{\left|b\right|}}{\psi\:}^{\mathrm{*}}\left(\frac{t-a}{b}\right)dt$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:a$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:b$$\end{document} denote the location and scale parameters, respectively, and * indicates the complex conjugate. The Wavelet Power Spectrum (WPS) is computed as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\left|{W}_{x}(a,b)\right|}^{2}$$\end{document} and is used to analyze the distribution of variance across time and frequency, thereby identifying periods of high volatility and structural changes in the series [15]. The statistical significance of the wavelet power is evaluated against a red-noise background spectrum using Monte Carlo simulations. To investigate the co-movement between variables, the cross-wavelet transform is defined as:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{W}_{xy}\left(a,b\right)={W}_{x}\left(a,b\right){W}_{y}^{\mathrm{*}}\left(a,b\right)$$\end{document} which measures the local covariance between two time series [14]. However, since cross wavelet power does not account for normalization, wavelet coherence is employed to capture the localized correlation between series. The squared wavelet coherence is given by:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}_{xy}^{2}\left(a,b\right)=\frac{{\left|S\left({b}^{-1}{W}_{xy}\left(a,b\right)\right)\right|}^{2}}{S\left({b}^{-1}{\left|{W}_{x}\left(a,b\right)\right|}^{2}\right)S\left({b}^{-1}{\left|{W}_{y}\left(a,b\right)\right|}^{2}\right)}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:S$$\end{document} denotes a smoothing operator in both time and scale. The coherence values range from 0 to 1, with higher values indicating stronger co-movement [38]. To further isolate the direct relationship between variables, Partial Wavelet Coherence (PWC) is applied. This method removes the influence of a third variable and captures the net interaction between two series. Following Mihanović et al. [39], the squared PWC is defined as:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\left(R{P}_{y{x}_{1}\cdot\:{x}_{2}}\right)}^{2}=\frac{{\left|{R}_{y{x}_{1}}-{R}_{y{x}_{2}}{R}_{{x}_{2}{x}_{1}}\right|}^{2}}{\left(1-{R}_{y{x}_{2}}^{2}\right)\left(1-{R}_{{x}_{2}{x}_{1}}^{2}\right)}$$\end{document} which ranges between 0 and 1 and provides a clearer interpretation of direct relationships after controlling for confounding effects. Similar to wavelet coherence, its statistical significance is assessed using Monte Carlo simulations [40]. Figure 2 shows the workflow of the study. Fig. 2Workflow of the study Workflow of the study Empirical findings and discussion Statistical summary of variables Figure 3 reports the descriptive statistics of the variables. The mean values indicate that economic growth (GDP) (9.339) is the highest, followed by waste management (WAM) (4.126) and CO₂ emissions (CO₂) (3.858), while research and development (RD) (0.845) and international legal compliance (ILC) (0.877) are the lowest. Standard deviations show that WAM (0.79) and ENI (0.741) are the most volatile, whereas TRA (0.043) and RD (0.213) are the most stable. Most variables are positively skewed, except CO₂ (-1.076), GDP (-0.396), and TRA (-0.445), while WAM and RD are nearly symmetric. Kurtosis suggests mostly platykurtic distributions, except ILC (9.452), which is highly leptokurtic. The Jarque–Bera results confirm that normality is rejected for almost all variables, as their p-values are near zero. Only RD (p = 0.116) appears normally distributed, while WAM (0.014) and TRA (0.012) also reject normality at the 5% level. Fig. 3Descriptive statistics Descriptive statistics Figure 4 presents the scatter plot matrix alongside the correlation coefficients of the variables. The results reveal strong positive relationships among most variables, particularly between waste management (WAM) and lnWAM (0.981), research and development (RD) and lnGDP (0.948), and lnWAM and lnGDP (0.933), indicating a high degree of comovement. Similarly, RD is strongly correlated with WAM (0.896) and lnTRA (0.877), while lnGDP and lnTRA (0.914) also exhibit a strong positive association. These relationships are clearly reflected in the upward-sloping scatter plots. In contrast, energy intensity (ENI) shows strong negative correlations with several variables, including RD (–0.901), lnWAM (–0.901), and lnGDP (–0.959), consistent with the downward patterns observed in the scatter plots. CO₂ emissions (CO₂) display moderate positive correlations with most variables, such as lnGDP (0.609), lnTRA (0.615), and lnWAM (0.587), but are negatively related to ENI (–0.477). Meanwhile, international legal compliance (ILC) shows moderate positive correlations with other variables, including WAM (0.694), lnWAM (0.654), and lnGDP (0.612), while its relationship with ENI is negative (–0.527). Overall, the results indicate strong interdependence among the variables, with particularly high correlations suggesting potential multicollinearity. Fig. 4Scatter plot matrix of the variables. ***p < 0.01 **p < 0.05 *p < 0.10 Scatter plot matrix of the variables. ***p < 0.01 **p < 0.05 *p < 0.10 Since this study employs wavelet-based methods, the stationarity of the variables is examined on the decomposed series rather than the raw data. Specifically, the original series are decomposed into multiple frequency components using the Maximal Overlap Discrete Wavelet Transform (MODWT), after which the Augmented Dickey–Fuller (ADF) unit root test is applied to each scale. The results of the wavelet-based ADF unit root test are presented in Fig. 4. As shown, for all variables—including CO₂ emissions (CO₂), research and development (RD), energy intensity (ENI), international legal compliance (ILC), waste management (lnWAM), economic growth (lnGDP), and trade openness (lnTRA)—the estimated ADF statistics across all scales (D1–D5) lie below the 5% critical value, indicated by the dotted line. This indicates a consistent rejection of the null hypothesis of a unit root at each decomposition level. These findings confirm that all variables are stationary across different frequency bands, capturing short-, medium-, and long-run dynamics. Overall, the wavelet ADF results validate the suitability of the series for subsequent wavelet-based time–frequency analysis by ensuring stationarity across all time horizon (see Fig. 5) for the wavelet ADF unit root test. Fig. 5Wavelet ADF unit root test Wavelet ADF unit root test To further assess the dynamic structure of the series, the nonlinear characteristics of the variables are evaluated using the wavelet BDS test, following the approach of [41]. Figure 6 illustrates that the computed BDS Z-statistics for all variables—CO₂ emissions (CO₂), research and development (RD), energy intensity (ENI), international legal compliance (ILC), waste management (lnWAM), economic growth (lnGDP), and trade openness (lnTRA)—consistently lie above the dotted black line representing the 5% critical value across all wavelet scales (D1–D5) and embedding dimensions. As a result, the null hypothesis of independent and identically distributed (i.i.d.) behavior is rejected for all variables. This finding confirms the presence of nonlinear dependence in the wavelet-decomposed components of CO₂ emissions, R&D, energy intensity, legal compliance, waste management, economic growth, and trade openness, indicating that their dynamics are complex and cannot be adequately captured by purely linear models. Fig. 6Wavelet BDS nonlinearity test Wavelet BDS nonlinearity test Wavelet power spectrum We begin the empirical analysis with the wavelet power spectrum (WPS), which provides a detailed view of the time-varying volatility structures embedded in the selected variables for Lithuania. By decomposing the variance of each series across different time–frequency bands, the WPS captures how shocks evolve over short-, medium-, and long-term horizons, revealing both temporary fluctuations and persistent cycles. The results, presented in Fig. 7, reveal notable heterogeneity in volatility patterns across variables. CO₂ emissions in Lithuania exhibit strong power concentrations, particularly at medium- to long-term frequencies, especially during the mid-2000s and post-2010 periods. This suggests persistent environmental pressures linked to structural economic changes and energy use transitions in the country. Similarly, energy intensity (ENI) and economic growth (lnGDP) display pronounced power at lower frequencies, indicating long-run cyclical behavior. This reflects Lithuania’s gradual economic transformation and energy restructuring, where shocks tend to have prolonged effects rather than short-lived impacts. International legal compliance (ILC) shows strong and sustained power across medium- and long-term scales, highlighting the lasting influence of institutional and regulatory frameworks, particularly in the context of Lithuania’s integration into the European Union and adherence to international environmental standards. In contrast, trade openness (lnTRA) and waste management (lnWAM) exhibit relatively weaker and more dispersed power, with limited concentrations mainly at long-term frequencies, suggesting smoother adjustments and fewer abrupt fluctuations over time. Meanwhile, research and development (RD) demonstrates intermittent and scattered power, especially at short- to medium-term frequencies, indicating that innovation activities in Lithuania are more responsive to episodic policy changes, funding cycles, and external economic conditions. Inclusively, the WPS results suggest that CO₂ emissions, ENI, ILC, and lnGDP in Lithuania are driven by persistent long-run dynamics, whereas RD, lnTRA, and lnWAM exhibit more transient and less stable fluctuations, underscoring the diverse temporal behavior of the variables in the Lithuanian context. Fig. 7Wavelet power spectrum Wavelet power spectrum Wavelet coherence Figure 8 presents the wavelet coherence results between CO₂ emissions and selected variables (ENI, lnGDP, lnTRA, lnWAM, and RD), illustrating their co-movement across time and frequency domains. The results show that CO₂ and energy intensity (ENI) exhibit strong coherence, particularly in the medium- to long-run periods (8–32 scales), indicating a persistent relationship over time. Similarly, CO₂ and economic growth (InGDP) display significant coherence, especially in the long-run horizon, suggesting that economic growth is closely associated with carbon emissions over extended periods. These findings are consistent with studies such as [42, 43] which documented strong long-run linkages between economic growth, energy use, and environmental degradation. For CO₂ and trade openness (InTRA), the coherence is moderate but becomes stronger in the medium-run frequencies, reflecting intermittent but meaningful interactions. This aligns with the findings of [44, 45], who argued that trade openness influences emissions through scale, technique, and composition effects over time. In contrast, CO₂ and waste management (InWAM) show relatively weaker and less consistent coherence, although some significant regions appear in the short- to medium-run periods, supporting findings by [46, 47] that environmental management practices tend to have more localized and short-term impacts. Additionally, CO₂ and research and development (RD) demonstrate notable coherence, particularly in the short- to medium-run periods (4–12 scales), indicating that innovation activities intermittently influence CO₂ emissions. This resonates with studies such as [8], which highlight the role of technological innovation and R&D in improving environmental quality, albeit with time-varying effects. Overall, the results indicate that the relationship between CO₂ emissions and the explanatory variables is time- and frequency-dependent, with stronger and more stable linkages observed in the medium- and long-run horizons, consistent with existing empirical and theoretical literature. Fig. 8Wavelet coherence estimates Wavelet coherence estimates Figure 9 illustrates the partial wavelet coherence (PWC) between CO₂ emissions and energy intensity (ENI), controlling for RD, InWAM, InTRA, InGDP, and ILC, thereby providing a refined understanding of the energy–emissions nexus in Lithuania after filtering out confounding effects. In the first panel (CO₂ ~ ENI | RD), strong and persistent coherence is observed from the mid-2000s to late-2010s at medium- to long-term scales [8–32]. The arrows indicate predominantly in-phase movements, suggesting a positive relationship where higher energy intensity leads to increased CO₂ emissions, even after accounting for innovation effects. This finding is supported by recent evidence showing that higher energy intensity contributes significantly to increased CO₂ emissions, particularly when technological progress is insuffic ient to offset energy use [48]. In the second panel (CO₂ ~ ENI | InWAM), strong coherence persists across medium-run frequencies, indicating that waste management does not substantially alter the direct energy–emissions relationship. This aligns with recent studies emphasizing that core energy structure remains a dominant determinant of emissions compared to environmental management practices. The third panel (CO₂ ~ ENI | InTRA) shows sustained but slightly fragmented coherence across time–frequency domains, suggesting that trade openness partially mediates but does not eliminate the energy–emissions link. This is consistent with recent findings that trade influences emissions indirectly through structural and energy-related channels rather than acting as a primary driver [49]. In the fourth panel (CO₂ ~ ENI | InGDP), coherence remains strong at lower frequencies, confirming that the relationship between energy intensity and CO₂ emissions is robust beyond economic growth effects. Recent empirical studies similarly show that energy intensity remains a key factor influencing both emissions and growth dynamics, highlighting its structural importance in environmental outcomes [50]. Finally, in the fifth panel (CO₂ ~ ENI | ILC), significant coherence persists, particularly during the mid-sample period, indicating that institutional and regulatory factors do not fully offset the impact of energy intensity on emissions. This supports broader empirical evidence that policy and institutional improvements alone are insufficient without structural reductions in energy intensity [51]. Lastly, the PWC results confirm that the CO₂–ENI relationship in Lithuania is strong, positive, and persistent across time and frequency domains, even after controlling for other variables. These findings are well supported by recent literature, which consistently identifies energy intensity as a fundamental and dominant driver of carbon emissions, with effects that remain robust across different model specifications and contexts. Fig. 9Partial wavelet coherence of CO₂ emissions with energy intensity controlling for RD, WAM, TRA, GDP, and ILC Partial wavelet coherence of CO₂ emissions with energy intensity controlling for RD, WAM, TRA, GDP, and ILC Figure 10 presents the partial wavelet coherence (PWC) between CO₂ emissions and waste management (InWAM), controlling for RD, InTRA, InGDP, ILC, and ENI, thereby isolating the direct relationship across time and frequency domains in Lithuania. In the first panel (CO₂ ~ InWAM | RD), the coherence appears weak and scattered, with only limited significant regions mainly in the short- to medium-run frequencies, suggesting that waste management has a limited direct impact on CO₂ emissions once innovation effects are accounted for. This finding is consistent with recent studies [52], which show that technological innovation tends to play a more dominant role than waste-related policies in mitigating emissions. In the second panel (CO₂ ~ InWAM | InTRA), coherence becomes slightly more pronounced in certain periods, particularly in the medium-run scales, indicating that trade openness may interact with waste management in shaping emissions, although the relationship remains inconsistent and time-varying. This aligns with [44, 53, 54], who find that trade affects environmental quality indirectly through structural and scale effects rather than through direct waste management channels. The third panel (CO₂ ~ InWAM | InGDP) shows moderate coherence in the medium- to long-run frequencies, especially around the mid-sample period, suggesting that waste management influences emissions more when conditioned on economic growth. This supports findings by [55], which emphasize that environmental improvements through waste and sustainability policies are often linked with economic development and structural transformation. In the fourth panel (CO₂ ~ InWAM | ILC), the coherence is more visible and persistent, particularly at lower frequencies, implying that institutional and regulatory frameworks enhance the effectiveness of waste management in influencing environmental outcomes. This resonates with [56], who highlight the importance of governance quality and environmental regulations in improving the effectiveness of waste management policies. Finally, in the fifth panel (CO₂ ~ InWAM | ENI), coherence is relatively weak and fragmented, indicating that once energy intensity is controlled for, the direct influence of waste management on CO₂ emissions becomes minimal. This supports evidence from [57], showing that energy intensity and energy structure remain the primary drivers of emissions, overshadowing the role of waste management alone. Generally, the PWC results suggest that the CO₂–waste management relationship in Lithuania is weak, indirect, and highly time-dependent, with stronger effects emerging only when combined with economic growth and institutional quality, consistent with recent empirical literature emphasizing the conditional effectiveness of environmental policies within broader energy and structural contexts. Fig. 10Partial wavelet coherence of CO₂ emissions with waste management controlling for RD, TRA, GDP, ILC, and ENI Partial wavelet coherence of CO₂ emissions with waste management controlling for RD, TRA, GDP, ILC, and ENI Figure 11 presents the partial wavelet coherence (PWC) between CO₂ emissions and research and development (RD), controlling for InWAM, InTRA, InGDP, ILC, and ENI, thereby isolating the direct relationship across time–frequency domains in Lithuania. In the first panel (CO₂ ~ RD | InWAM), the coherence appears relatively weak and dispersed, with only a few significant regions concentrated in the medium-run frequencies. This suggests that once waste management effects are controlled for, the direct influence of R&D on CO₂ emissions remains limited and unstable over time. The scattered nature of the coherence implies that innovation alone does not consistently translate into environmental improvements when waste management dynamics are accounted for. This finding aligns with recent evidence indicating that the environmental benefits of innovation are often indirect and contingent on complementary policies [52]. In the second panel (CO₂ ~ RD | InTRA), coherence becomes slightly more pronounced, particularly within medium- to long-run scales during specific periods. This indicates that trade openness may interact with R&D in shaping environmental outcomes, although the relationship remains time-varying and not uniformly strong. The results suggest that the effectiveness of innovation in reducing emissions may depend partly on trade-related structural changes and technology diffusion channels, consistent with prior studies highlighting the indirect role of trade in environmental quality [25, 54]. The third panel (CO₂ ~ RD | InGDP) reveals more consistent and stronger coherence, especially in medium- to long-run frequencies around the central part of the sample. This implies that R&D exerts a more substantial influence on CO₂ emissions when conditioned on economic growth. The finding suggests that innovation-driven environmental improvements are more effective within a growing economy, where resources and structural transformation support cleaner technologies. This supports earlier findings that economic growth enhances the effectiveness of environmental innovation [58, 59]. In the fourth panel (CO₂ ~ RD | ILC), coherence becomes more visible and persistent, particularly at lower frequencies, indicating a stronger long-run relationship. This suggests that institutional and regulatory quality enhances the effectiveness of R&D in influencing emissions. Strong governance frameworks likely facilitate the adoption and implementation of green technologies, thereby strengthening the environmental benefits of innovation. This observation is consistent with studies emphasizing the role of institutional quality in environmental sustainability [60]. Finally, in the fifth panel (CO₂ ~ RD | ENI), coherence appears weaker and more fragmented across both time and frequency domains. This indicates that once energy intensity is controlled for, the direct impact of R&D on CO₂ emissions diminishes significantly. The result highlights that energy consumption patterns and efficiency remain dominant drivers of emissions, potentially overshadowing the independent contribution of innovation. This finding is in line with empirical evidence suggesting that energy intensity is a primary determinant of emissions dynamics [9, 10, 61]. Fig. 11Partial wavelet coherence of CO₂ emissions with research and development (RD) controlling for WAM, TRA, GDP, ILC, and ENI Partial wavelet coherence of CO₂ emissions with research and development (RD) controlling for WAM, TRA, GDP, ILC, and ENI Conclusion of the study This study examined the dynamic relationships between CO₂ emissions and key mitigation drivers—energy intensity (ENI), research and development (RD), waste management (WAM), economic growth (GDP), trade openness (TRA), and institutional quality (ILC) in Lithuania using advanced wavelet-based techniques, including Wavelet Coherence (WC), Partial Wavelet Coherence (PWC), and Multiple Wavelet Coherence (MWC). Unlike conventional static models, this approach captures time-varying and frequency-dependent interactions, providing a more comprehensive understanding of the evolving nature of emission drivers. The findings reveal that energy intensity remains the most dominant and persistent driver of CO₂ emissions, exhibiting strong and stable coherence across medium- and long-run frequencies. Economic growth also shows a significant positive relationship with emissions, confirming the presence of scale effects, particularly in the long run. In contrast, renewable energy and energy efficiency contribute to emission reductions, although their effects vary across time horizons. The results further indicate that R&D plays a conditional and time-dependent role, with stronger emission-reducing effects observed when combined with favorable economic growth and institutional quality. Similarly, waste management exhibits a relatively weak and indirect influence on emissions, becoming more effective only under supportive structural and regulatory conditions. Trade openness demonstrates moderate and intermittent effects, primarily influencing emissions through indirect channels such as technology transfer and structural transformation. Policy recommendations Based on the empirical findings, several important policy implications emerge: First, reducing energy intensity should remain a top policy priority, as it is the most consistent driver of emissions. Policymakers should accelerate investments in energy-efficient technologies, building renovation programs, and industrial efficiency improvements to achieve sustained emission reductions. Second, strengthening innovation systems is essential, but R&D policies should be strategically aligned with environmental goals. Public and private R&D investments should prioritize green technologies, renewable energy innovation, and energy-efficient production processes to maximize their environmental impact. Third, economic growth strategies must be aligned with sustainability objectives. Lithuania should promote green growth by encouraging low-carbon industries, sustainable infrastructure, and environmentally friendly production systems to decouple economic expansion from emissions. Fourth, institutional quality and regulatory frameworks should be enhanced, as strong governance significantly improves the effectiveness of environmental policies. Transparent regulations, enforcement mechanisms, and alignment with EU climate policies can facilitate the successful implementation of green initiatives. Fifth, while waste management plays a supportive role, its effectiveness can be strengthened through integration with broader environmental and circular economy strategies. Policies should focus on recycling, waste-to-energy technologies, and reducing landfill dependency. Finally, energy transition policies should be accelerated, particularly through increased adoption of renewable energy sources and reduced reliance on fossil fuels. Since energy structure strongly influences emissions, transitioning to cleaner energy systems is critical for achieving long-term climate goals. Limitations of the study Despite its contributions, this study has several limitations that should be acknowledged. First, the analysis focuses solely on Lithuania, which may limit the generalizability of the findings to other countries with different economic structures and institutional settings. Future studies could extend the analysis to a panel of countries or conduct comparative studies across EU economies. Second, although wavelet-based methods provide rich time–frequency insights, they do not fully establish causal relationships. While lead–lag patterns offer useful indications, more robust causal inference techniques could complement the analysis. Third, the study relies on proxy variables such as R&D expenditure and energy intensity, which may not fully capture the complexity of innovation quality or energy system transformation. Future research could incorporate more detailed indicators, such as green patents or sector-specific energy data. Fourth, data limitations required the conversion of annual data into quarterly frequency using interpolation, which may introduce minor distortions in the analysis. Access to higher-frequency data would improve the robustness of future studies. Finally, the study does not explicitly account for external shocks such as financial crises, geopolitical events, or COVID-19, which may influence emission dynamics. Future research could integrate such structural breaks to provide deeper insights into policy effectiveness under uncertainty.