The effects of energy transition and technological innovation on environmental quality: a CS-ARDL analysis of G7 economies

Recent years have seen profound changes, including global market integration and rapid industrial development, while climate change and environmental degradation increasingly challenge policymakers and researchers. Overexploitation of resources and rising pollution underscore the limitations of the current unsustainable development model. In this context, the energy transition (ET) has gained attention as a strategy to reduce pollution and improve resource management (Wang, Ouyang, & Wang, 2023a; Gielen et al., 2019).

Carbon neutrality, a central goal of sustainable development, requires both technological innovations in energy production and transformative changes in consumption. Achieving this target entails managing energy demand efficiently while minimizing emissions, with research highlighting technological, technical, and socio-economic dimensions of the transition (Wang, Sami, Khan, Alamri, & Zaidan, 2023b). Achieving climate neutrality by 2050 demands ambitious energy transitions across EU Member States, though substantial investments may raise energy prices and exacerbate energy poverty (Śmiech, Karpinska, & Bouzarovski, 2025).

While energy consumption supports economic growth and living standards, it also drives environmental degradation. Expanding renewables mitigates emissions, reduces fossil fuel dependence, and improves competitiveness, with sources such as wind and solar often more cost-effective than fossil fuels (Wei, Jiandong, & Saleem, 2023; Ram et al., 2018). TI supports socio-economic progress and environmental sustainability, particularly when private-sector investment targets eco-friendly technologies (Ullah, Ozturk, Majeed, & Ahmad, 2021; Khan & Majeed, 2019). It enables smart energy networks, integrates renewable sources, enhances energy efficiency, reduces emissions, and fosters economic resilience through job creation, while environmental monitoring supports regulatory compliance (Majeed, Xie, Gao, Du, & Muniba, 2025; Khajeh Naeeni, 2023).

Environmental technological innovation (TI), reinforced by strong institutions, is crucial for mitigating CO2 emissions and promoting greener production systems (Khan, Weili, & Khan, 2022; Zhang, Ozturk, & Ullah, 2022; Obobisa et al., 2022). In Organisation for Economic Co-operation and Development (OECD) economies, such innovation enhances environmental sustainability, though its effects on CO₂ emissions are heterogeneous, while circular economy practices further reduce non-renewable resource use and foster socio-economic-environmental synergies (Rosa, Sassanelli, Urbinati, Chiaroni, & Terzi, 2020).

Building on this context, the present study examines the combined impact of ET and TI on environmental quality in G7 countries from 1999 to 2023 using the cross-sectional autoregressive distributed lag (CS-ARDL) model. Results indicate that TI significantly reduces CO₂ emissions, though its effect is smaller than that of ET. The study contributes by analyzing the joint impact of energy and technological transitions, applying a CS-ARDL framework that accounts for cross-country heterogeneity and dependencies, and providing policy-relevant insights on investment in renewable energy, energy efficiency, and green technologies.

The remainder of the article is structured as follows: Section 2 reviews the literature and develops research hypotheses; Section 3 outlines the methodology; Section 4 presents results and discussion; Section 5 offers robustness checks; and Section 6 concludes with policy implications and directions for future research.

ET and TI are key drivers of environmental quality. While renewable energy adoption reduces fossil fuel dependence, TI improves energy efficiency and mitigates emissions. These strategies are interdependent, as the success of the ET relies on the deployment of innovative technologies. Their environmental impact, however, depends on implementation speed, investment levels, and institutional context.

Energy consumption, which rises with economic development, is a central concern in climate policy, as emphasized by the Paris Agreement (2015) and COP-26. Renewable energy and green innovations are crucial for reducing CO₂ emissions, enhancing sustainable development, and improving the energy mix, supply security, and economic growth (Vural, 2020; Saleem, Khan, & Mahdavian, 2022; Shahbaz, Loganathan, Zeshan, & Zaman, 2015). Empirical evidence is mixed: some studies report negligible effects (Alola, Adebayo, & Onifade, 2022), while others find significant emission reductions (Saleem, Khan, & Shabbir, 2020; Chien et al., 2021; Adebayo & Kirikkaleli, 2021; Soylu, Adebayo, & Kirikkaleli, 2021).

Environmental modeling and case studies highlight both benefits and limitations of renewable energy adoption. Gençer, Torkamani, Miller, Wu, and O’Sullivan (2020) link renewables to lower CO₂ emissions, Cardoso and González (2019) show that insufficient household energy efficiency increases environmental costs in Argentina, Kokkinos, Karayannis, and Moustakas (2020) emphasize urban energy availability for sustainable development, and Poruschi and Ambrey (2019) report potential negative effects of widespread solar adoption in Australia. In emerging economies, renewable energy and green technologies reduce fossil fuel dependence, enhance energy security, and promote sustainable development (Yang, Luo, Liu, Hua, & Liu, 2024; Zhang & Kong, 2022; Majeed & Luni, 2019; Dong, Sun, Jiang, & Zeng, 2018).

Some studies, however, report conflicting results, showing that renewable energy may increase emissions (Nguyen & Kakinaka, 2019; Apergis, Payne, Menyah, & Wolde-Rufael, 2010; Bulut, 2017) or have negligible impact (Charfeddine & Kahia, 2019). Despite these inconsistencies, substantial evidence supports the emission-reducing role of renewable energy, as seen in G7 countries (Zafar, Zaidi, Sinha, Gedikli, & Hou, 2019) and across five South Asian economies, motivating the following hypothesis:

Protecting environmental quality has become a global priority, with TI recognized as a key tool to address environmental challenges (Ullah et al., 2021; Zhang, Peng, Ma, & Shen, 2017). Environmental sustainability is central to economic, social, and political agendas, prompting firms and governments to adopt proactive strategies supported by technology, structural reforms, and regulatory frameworks (Fraj, Matute, & Melero, 2015; Abdallh & Abugamos, 2017). Empirical evidence from G7 and OECD economies shows that eco-innovation, renewable energy, and environmental policies significantly reduce CO₂ emissions and improve environmental quality (Borgi, Alessa, Hamza, & Albitar, 2024; Wang et al., 2023a, b).

Environmental proactivity drives firms to continuously adapt products, processes, and technologies. Product innovation introduces new or enhanced goods and services, while process innovation implements novel production methods (Schumpeter, 1983), safeguarded by patents and trademarks (Sandner & Block, 2011). TI promotes the green economy by lowering environmental impact, enhancing competitiveness, and reinforcing regional green systems (Shabir, Hussain, Işık, Razzaq, & Mehroush, 2023; Xu, Wu, & Zhang, 2020; Ben Amara & Chen, 2020).

However, the impact of TI on environmental quality is mixed. Some studies find significant emission reductions in BRICS economies (Brasil, Russia, India, China and South Africa) and other countries (Adebayo et al., 2023; Ahmad, Youjin, Žiković, & Belyaeva, 2023), while others show limited effects in OECD economies (Cheng, Ren, Wang, & Yan, 2019). Renewable energy generally mitigates emissions more effectively, though economic growth and fossil fuel use often exacerbate environmental degradation (Raihan & Tuspekova, 2022; Chen & Lei, 2018). In some cases, innovation can initially worsen environmental pressures, as observed in African and Asian-Pacific Economic cooperation (APEC) countries, before delivering benefits (Usman & Hammar, 2021; Acemoglu, Aghion, Bursztyn, & Hemous, 2012; Dauda et al., 2021).

The Environmental Kuznets Curve (EKC) describes an inverted U-shaped relationship between economic growth and environmental degradation, where emissions decline beyond a certain income threshold due to technological progress and environmental regulations. This study highlights TI as a key mechanism for surpassing this threshold and enhancing environmental quality in G7 economies, motivating the formulation of the following hypothesis.

This study investigates the impact of ET and TI on environmental quality in G7 countries (Canada, France, Italy, Germany, Japan, the United Kingdom, and the United States) from 1999 to 2023, a period encompassing key phases of technological progress, ET, and major economic events such as the 2008 financial crisis and the Paris Agreement. This timeframe provides sufficient depth for the CS-ARDL framework to robustly estimate short- and long-term dynamics across heterogeneous units, while G7 economies offer a critical case study due to their historical emissions, technological leadership, and strong institutional frameworks.

To assess empirically the impact of ET and TI on environmental quality, we adopt the CS-ARDL (Cross-Sectionally Augmented Autoregressive Distributed Lag) approach as our baseline estimation strategy. The baseline specification is formulated as follows:

Where, CO_{2i, t} denotes the dependent variable (CO₂ emissions), X_{i, t} represents the explanatory variables, including (ET, technological innovations, information and communication technology (ICT), and gross domestic product (GDP), p_u and p_w refer to the respective numbers of lags, and e_{i, t} is the error term.

To account for cross-sectional dependence and to avoid inconsistent or misleading parameter estimates, the model is extended by including the cross-sectional averages of each regressor, as proposed by Chudik and Pesaran (2015).

V_{i, t−I} represents the cross-sectional averages of the dependent and independent variables (V_{i, t−I}= CO_{2i, t-I}, X_{i, t-I}) capturing the spillover effects across units. pv denotes the number of lags for these averages. This extension ensures that cross-sectional dependence is explicitly accounted for, which is not addressed in the basic specification.

Finally, the long-run coefficients are derived from the short-run estimates. The mean estimator for the long-run relationship is calculated as follows:

Carbon dioxide (CO₂) emissions, typically measured in metric tons per capita, primarily arise from fossil fuel combustion and industrial activities and serve as a widely used proxy for environmental quality (Chang, Liu, Luo, & Xing, 2023; Kafeel et al., 2024). ET is captured by the ratio of renewable energy to final energy consumption, reflecting the replacement of fossil fuels with renewable sources such as solar and wind, and serving as a reliable indicator of structural changes in energy systems (He et al., 2023; Lee & Wang, 2024).

TI is measured by the share of patents held by residents and non-residents in innovative technologies, reflecting sustainable growth potential and contributing to emission reductions through improved efficiency in fossil fuel use (Li, Li, & Wang, 2022; Bai et al., 2020). Control variables include ICT adoption, which influences CO₂ emissions via enhanced energy efficiency and smart energy management, and economic growth, measured by annual GDP growth, capturing overall economic activity.

The selected variables are grounded in the EKC framework, which posits an inverted-U relationship between economic growth (GDP) and environmental degradation. ET and TI are incorporated as key drivers facilitating emission reductions, while ICT enhances sustainable practices. Collectively, these variables provide a robust basis for analyzing the determinants of environmental quality.

Table 1 presents a comprehensive description of the variables employed in the empirical analysis, including their definitions, measurement approaches, proxy specifications, and corresponding data sources (see Table 1).

To analyze the short- and long-term effects of ET and TI on environmental quality, accounting for individual heterogeneity and cross-sectional dependence, we employ the CS-ARDL methodology (Chudik & Pesaran, 2015; Chudik, Mohaddes, Pesaran, & Raissi, 2017). This approach captures temporal dynamics and controls for unobserved common shocks via cross-sectional averages, providing robust estimates. The procedure consists of five steps:

In recent years, increasing economic, social, and cultural interdependence has exposed countries to common shocks that may affect estimation results. To detect cross-sectional dependence, we employ Pesaran's (2004, 2015) CD test and Juodis and Reese's (2022) CD_W test. This step is crucial for identifying correlations arising from strong economic linkages and for selecting appropriate estimators to ensure robust and reliable results. The CD test statistics are reported below.

Where:

Where $ρˆij$ represent the pair-wise residual correlation

Pesaran, Ullah, & Yamagata (2008) developed an endogeneity test to account for biases and fixed dimensions related to both sample and time. To overcome this type of problem, we employ the Pesaran et al. (2008) method. The homogeneity test statistics are presented as follows:

The ΔSH _adj statistic corresponds to the adjusted version of delta (Δ).

The classic Im-Pesaran-Shin (IPS) unit root estimation techniques are effective in the absence of cross-sectoral dependency issues. Thus, the alternative second-generation methods of the cross-sectionally augmented Im and cross-sectionally augmented Dickey-Fuler (CIPS and CADF) unit root tests proposed by Pesaran (2007) are the appropriate method for addressing this type of problem. The test equation is as follows:

The average cross-sectional values correspond respectively to $y̅t−1$ and $yt−1̅.$ The statistics associated with the CIPS test are then expressed as follows:

In order to overcome the weakness of the classical cointegration method, Westerlund and Edgerton (2008) proposed a more flexible and robust method for the treatment of transverse dependence. Thus, this method generates four distinct test statistics (Gt, Ga, Pt and Pa). The Gt and pt statistics are determined through the standard deviation in $αˆi$ a standard way. Ga and Pa are calculated based on the covariance variance estimator of Newey-West (1994). These statistics are presented as follows:

Under the null hypothesis, we conclude that the variables are not cointegrated and that, consequently, the data generation process is not an error correction model.

The first step is to test for interindividual dependence. As shown in Table 2, both Pesaran's (2004, 2015) and Juodis and Reese (2022) tests reject the null hypothesis of independence, with statistics significant at the 1% and 5% levels, confirming the presence of cross-sectional dependence.

The results in Table 3 indicate that the Pesaran et al. (2008) tests reveal significant heterogeneity of slopes (Δ = 9.518; Δ adj = 10.917, p < 0.01), confirming that the impact of ET and TI on environmental quality varies from one unit to another.

The stationarity of the variables was assessed using the CADF and CIPS tests. Results reported in Table 4 indicate that all variables contain a unit root at levels but become stationary after first differencing, confirming they are integrated of order one (I (1)).

Table 5 shows that the Gt, Ga, Pt, and Pa statistics are significant at the 1% level, confirming the existence of a cointegration relationship between all variables. Therefore, from a methodological point of view, it is necessary to apply robust estimators to cross-dependence and heterogeneity, such as CS-ARDL methods (Chudik & Pesaran, 2015).

Table 6 presents the lag-order selection criteria. The choice of the maximum number of lags is based on the selection criteria of the baseline model. Across multiple criteria Akaike information criterion (AIC), Hannan-Quinn Information Criterion (HQIC), Schwarz Bayesian Information Criterion (SBIC) and Final Prediction Error (FPE), the results indicate that the optimal lag length is three periods.

Table 7 shows that the ET significantly reduces CO₂ emissions at the 1% level in both the short and long term (β = −0.3790; −1.379, p < 0.05), with a one-unit improvement associated with a meaningful decline in emissions, consistent with prior studies (Ullah, Adebayo, Irfan, & Abbas, 2023; Majeed & Luni, 2019; Dong et al., 2018).

Economically, this underscores the high returns of investing in renewable energy and efficiency technologies, while politically it highlights the need for policies that accelerate the shift to low-carbon systems. TI also reduces emissions (β = −0.008; −0.007, p < 0.05), supporting evidence that advances in technology enhance energy efficiency and cleaner production (Li et al., 2022; Bai et al., 2020; Raihan & Tuspekova, 2022; Mughal et al., 2022; Hasan & Du, 2023). Promoting R&D, green patents, and technology diffusion can therefore yield tangible environmental benefits.

The relatively modest impact of TI may stem from using total patents rather than green-focused patents, diluting its effect on CO₂ reduction. In contrast, the ET delivers more direct and sustained benefits by replacing fossil fuels, whereas the effects of TI are indirect and may be partially offset by rebound effects.

For information and communication technologies, no short-term effect is detected, while a significant long-term reduction in emissions is observed (β = −0.707, p < 0.05).

This finding is consistent with earlier studies (Zhang et al., 2017), suggesting that the environmental benefits of ICT materialize gradually as initial investment costs are absorbed and efficiency gains emerge. In contrast, GDP has a positive and significant impact on CO₂ emissions in both the short and long run (β = 0.099; 0.070, p < 0.05), indicating that economic growth remains carbon-intensive and highlighting the need for ET policies that promote renewable energy adoption and sustainable growth.

To reinforce the validity and robustness results obtained using the CS-ARDL model, we employed two complementary methods: Augmented Mean Group (AMG) and Common Correlated Effects Mean Group (CCEMG). These two techniques enable us to examine the stability of long-term relationships in a context of cross-sectional dependence and heterogeneity between countries.

Table 8 indicates that both the AMG and CCEMG estimators confirm a significant negative effect of the ET on CO₂ emissions (β = −0.2181, p < 0.05; β = −0.5149, p < 0.01), consistent with prior studies (Ullah et al., 2023; Majeed & Luni, 2019; Dong et al., 2018), suggesting that expanding renewable energy adoption can effectively mitigate climate change. TI also significantly reduces emissions (β = −0.0002, p < 0.05; β = −0.0007, p < 0.01), supporting evidence that innovation enhances energy efficiency and cleaner production (Li et al., 2022; Bai et al., 2020; Raihan & Tuspekova, 2022; Mughal et al., 2022; Hasan & Du, 2023). ICT adoption exerts a marginal emission-reducing effect (β = −0.114, p < 0.10), whereas GDP growth significantly increases CO₂ emissions (β = 0.0623, p < 0.01; β = 0.0464, p < 0.01), highlighting the environmental trade-offs of economic expansion.

This study examines the effects of ET and TI on environmental quality, proxied by per capita CO₂ emissions, in G7 countries from 1999 to 2023. Employing CS-ARDL, AMG, and CCEMG estimators to account for cross-sectional dependence and slope heterogeneity, results indicate that ET and TI significantly reduce emissions, while ICT adoption has a marginal long-term effect. The CS-ARDL approach mitigates potential endogeneity from reverse causality through cross-sectional averages, and AMG and CCEMG provide robustness, enhancing result credibility.

The study is academically distinctive for applying a dynamic CS-ARDL framework that captures short- and long-term effects and for focusing on G7 economies, which are globally influential yet underexplored. Policy implications include sustained investment in renewable energy and green innovation, targeted subsidies, tax incentives for green R&D, stricter environmental standards, and capacity-building programs to accelerate low-carbon technology adoption.

CO₂ emissions remain a major driver of environmental degradation, underscoring the need for renewable energy infrastructure and strategies promoting green technological development. Limitations include using CO₂ emissions as the sole environmental proxy, total patent counts as a broad measure of innovation, and the omission of external shocks such as COVID-19 or geopolitical tensions. Future research could employ broader environmental indicators, more targeted innovation metrics (e.g. green patents or environmental R&D), and extend the analysis to emerging economies to better assess the role of innovation in sustainability.