How ICT and Green Technologies Shape the Nexus Between Financial Development and Carbon Footprint: Evidence from an N-Shaped EKC

Abstract

This study explores the effect of financial development, economic growth, ICT, green technologies, and strict environmental policies on environmental sustainability in the states of the European Union from 1996 to 2022. It also evaluates the EKC hypothesis and examines how ICT and green technologies moderate the linkage between financial development and carbon footprint. The Westerlund-Durbin-Hausman cointegration test is used for the long-run relationship. The FMOLS and CUP-FMOLS estimators are used to estimate the long-run elasticity coefficients, providing reliable results. The results reveal an inverted N-shaped linkage between GDP and carbon footprint in EU states, validating the EKC hypothesis. Furthermore, financial development has been found to increase carbon footprints, whereas green technologies, ICT, and stringent environmental regulations have been shown to mitigate these effects. Additionally, the interaction effects of ICT and green technologies with financial development demonstrate a reduction in the carbon footprint. These findings indicate that the EU should integrate the moderating role of innovation into policies addressing the pollution caused by financial development to achieve net-zero emission goals.

1. Introduction

The expansion of various economic activities has resulted in greenhouse gas (GHG) emissions surpassing the environment’s carrying capacity. This global rise in GHG emissions is causing increasingly severe environmental damage. The accumulation of these gases is widely recognized as a major contributor to climate change, making environmental pollution and protection critical issues on the international political agenda. Policymakers strive to create effective strategies for a sustainable future, aiming to balance economic development with environmental preservation while addressing the adverse effects of climate change [1,2,3].

The fact that developed countries prioritize environmental concerns only at advanced stages of their development processes makes it difficult for developing countries to adopt environmentally friendly practices. While the “grow first, clean later” approach is based on the idea that economic growth can solve environmental problems, this view is supported by the arguments of the Environmental Kuznets Curve (inverted-U model). In this framework, while some countries have focused on sustainable development, others have preferred to prioritize economic growth while ignoring environmental policies [4]. Much research has examined the coupling between economic growth and environmental quality, often using Grossman and Krueger’s [5] Environmental Kuznets Curve (EKC) hypothesis. This hypothesis posits that environmental degradation intensifies as per capita income rises but eventually diminishes once a critical income level is reached, creating a characteristic inverted U-shaped curve [6,7,8]. This concept implies that environmental harm is more pronounced during the early phases of economic growth but tends to decline as economic development reaches a more advanced stage.

This phenomenon is shaped by three main mechanisms: composition, scale, and technical effects. First, the scale effect, observed in the initial stages of growth, suggests that environmental degradation increases as the economy grows because economic and technological structures have not changed sufficiently in this process. The composition effect emphasizes that environmental improvements are possible as the economic structure shifts from highly polluting to less polluting sectors. Finally, technical effects show the critical role of clean technological innovations that emerge after industrial development in improving environmental quality [9]. Additionally, the N-shaped EKC suggests that the inverted U-shaped EKC may not hold in the long run. Increased wealth can further trigger environmental degradation once income growth exceeds a certain level. This is associated with the dominance of scale effects over composition and technical impact, limited capacity to improve industry dispersion, or insufficient incentives for technological innovation [10].

Additionally, many researchers argue that financial development can promote economic growth and reduce environmental degradation [11]. Therefore, financial development is essential when discussing the relationship between economic growth and the environment. One of the primary reasons for this is that financial liberalization and development accelerate economic growth and affect environmental performance by attracting R&D investments and foreign direct investment. Moreover, financial development facilitates developing countries’ access to new technologies and supports clean and environmentally friendly production, improving the global environment and sustainable development. However, it is noted that financial development, although it boosts economic growth, can also trigger industrial pollution and environmental degradation [12]. In other words, the coupling between the environment and financial development is analyzed from two perspectives. Accordingly, financial development can improve the quality of the environment by realizing environmentally friendly projects through research and development activities and encouraging clean technology investments. However, financial development can also lead to increased environmental degradation by supporting credit facilities to purchase products such as mechanical machinery, electrical appliances, automobiles, and housing [13].

Modern economic growth theory states that the key factors of economic growth are technology and science. Technological progress is characterized by its essential role in overcoming growth constraints, reducing ecological pollution, and increasing the efficiency of other elements. In addition to addressing ecosystem concerns in growth theories, these advances offer practical solutions in areas such as green energy and ecological policies. They are seen as a key tool in combating environmental degradation. The coupling between green technologies and economic growth can be understood through endogenous growth theory, which suggests that innovation leads to increasing returns to scale, and ecological modernization theory, which emphasizes the role of environmentally focused innovations in promoting green growth. Thus, cooperation in technology development plays a crucial role in tackling regional pollution and global climate change. Innovations in green technology contribute to environmental protection by supporting the efficient use of natural resources and the emergence of energy-saving products [14,15,16]. Intensifying investments and innovations to realize green growth requires detailed research to understand the formation of the green economy and its effect on sustainable development. In this process, the business community is interested in economic benefits, authorities are setting environmental targets, and the public, representing social interests, plays an essential role. At the same time, innovation and green knowledge management processes play a critical role in achieving sustainable development goals. These processes enable the development of high-quality products to reduce environmental impacts through environmentally friendly technologies and innovations [17]. However, while providing environmental efficiency at the micro level, technological progress can increase energy consumption at the macro level and lead to environmental degradation through the rebound effect. However, when technological innovation supports economic growth and environmental protection, it contributes to sustainable development and creates a win-win situation [18].

Feng et al. [19] emphasize that sustainable development has long faced significant underfunding and that the digital age presents a critical opportunity to address environmental and ecological challenges through the rapid advancement of digital technology. In the digital age, the rapid development of information and communication technologies (ICT) has led to significant transformations in economic growth, environmental impacts, and the financial sector. Thanks to low transaction costs, ICT reduces costs by facilitating communication and coordination through online banking, e-commerce, and smartphone applications. It also makes the industrial sector more productive, enables more efficient allocation of financial assets, and increases financial inclusion. The global financial sector, the largest recipient of ICT since the 1990s, has developed based on digital information. This has deeply affected the coupling between economic growth and financial development [20,21]. Conversely, ICT has different effects on energy use. ICT has the potential to reduce carbon emissions through energy efficiency and smart applications. However, increased ICT use may lead to a rebound effect and negatively affect the initial carbon reductions. Moreover, different social and economic stages of countries cause variations in the impacts of ICT on environmental pollution [22].

Zhou et al. [23] argue that technological innovations can significantly increase eco-efficiency due to innovation compensation, adjustment cost, and energy rebound effects. However, inappropriate environmental regulations may reduce the marginal benefits of these innovations. He also states that when rebound effects occur, moderate environmental regulations effectively mitigate the negative impact of technology. Thus, the stringency of environmental policies is seen as a measure to combat environmental degradation. According to a standard view, countries can reduce their production costs and increase their export capacity by implementing lax environmental regulations, but this approach risks turning countries into pollution havens. In contrast, Porter and Van der Linde [24] argue that strict environmental standards can provide a dynamic competitive advantage through technological innovation. However, the high costs of strict environmental policies may make it difficult for industries to engage in environmentally friendly investments and limit environmental improvement efforts. To avoid these costs, industries in developed countries may move their dirty production to countries with more lax environmental regulations. The Race to the Bottom hypothesis suggests that developing nations lower environmental standards to attract foreign capital [25,26].

Based on the above discussion, this research examines how ICT and green technologies moderate the coupling between financial development and the carbon footprint in EU nations. The study also considers the connection between economic growth and carbon footprint from an EKC perspective and incorporates stringent environmental policies into the model. Aligned with the European Green Deal and the European Climate Act, the EU has set ambitious targets to cut greenhouse gas emissions by 55% by 2030 and achieve net zero emissions by 2050. To support these goals, the European Parliament approved a series of amendments to the EU Treaty on 22 November 2023, aiming to enhance the implementation of sustainable development objectives both within the EU and globally. These amendments include more rigorous commitments to combat climate change, protect biodiversity, eliminate discrimination, promote diversity, improve public health and education, ensure full employment, and accelerate social progress. Despite these efforts, significant gaps remain in addressing challenges such as sustainable food systems, climate change, and responsible production and consumption [27]. Additionally, the Global Footprint Network (GFN) highlights that some EU states face ecological resource deficits, and only 10% of materials used in Europe come from recycling [28,29]. These findings underscore that while the EU leads in sustainable development initiatives, greater efforts are required to meet its net-zero emission goals.

In this regard, the study contributes to the literature in the following ways. First, the study focuses on the relationship between economic growth (EG) and carbon footprint (CF). In empirical studies conducted in this context, some findings have confirmed the existence of a linear coupling between EG and environmental pollution [30,31,32,33], while some findings have confirmed an EKC relationship [6,7,8,32]. Conversely, some studies do not support the EKC hypothesis [10,33,34]. Recent studies suggest that the inverted U-shaped EKC may not remain constant in the long run. GDP growth over time may negatively affect environmental quality due to the underutilization of efficient technologies. Therefore, it is argued that EG and environmental degradation have an N-shaped EKC relationship [8,35]. Based on these findings, we test the hypothesis of an N-shaped EKC linkage between EG and CF in EU states. This will provide a new perspective on the literature by re-examining these ambiguous relationships.

Second, the study focuses on coupling financial development (FD) and CF. FD is seen as an essential component of EG. In addition, FD has a complex structure that has the potential to provide opportunities for sustainability while creating impacts that may harm the environment [36,37]. Previous studies generally ignore the EKC hypothesis in the relationship between FFD and the environment. Therefore, when examining the coupling between EG and the environment, FD stands out as a critical element, and considering these variables together offers more comprehensive policy implications for a sustainable environment. Third, the study focuses on the moderating role of ICT and green technologies in the connection between FD and CF. The financial system plays a crucial role in enabling green economic growth by allowing more efficient allocation of social resources and accelerating green technology innovations, but the financial sector relies heavily on the ICT sector for effective service delivery [38]. However, the debate on the impact of ICT on environmental quality continues. While there are arguments that ICT improves environmental quality [39,40], there is also empirical evidence that ICT leads to environmental degradation through the rebound effect [21,41]. In this context, revealing how the interaction of relevant variables impacts environmental quality will offer a new perspective on the innovation-finance-environment relationship. Thus, it will provide a holistic perspective on sustainable development. Finally, the study focuses on the effect of strict environmental policies on CF. Therefore, various strategies for reducing CF in EU states can be evaluated together.

The study consists of five sections. Section 1 provides the theoretical background and introduction. Section 2 contains the literature summary and the literature gap. Section 3 includes methodology, findings, and discussion. Section 4 includes the results of that analysis. Section 5 presents conclusions and policy recommendations.

2. Literature Review and Theoretical Underpinning

2.1. Economic Growth and Environmental Sustainability

The connection between economic growth (EG) and environmental quality is based on a reciprocal interaction. Economic activities, including consumption and production processes, are directly linked to the environment in which they occur. Therefore, as economic growth increases, so do its environmental impacts. The conflict between the cumulative nature of EG and the permanent effect on non-renewable resources results in an unbalanced relationship [4]. The continuous deterioration of the environment due to the overuse of natural resources has led to increased research on the linkage between environmental pollution and economic activity [42]. There are several studies in the literature on the impact of EG on environmental pollution using different measures of environmental sustainability, such as CO₂, ecological footprint (EF), and carbon footprint (CF) [32,43,44]. The effects of EG on pollution are attributed to various reasons. Firstly, using primitive technologies to boost production can lead to increased pollution. In addition, while pollution increases as the economic structure shifts from agriculture to industry, this increase tends to decrease as the service sector develops. Finally, developing and applying environmentally friendly technologies significantly reduces pollution [45,46]. On the contrary, theoretical and empirical research on this issue is based on the EKC framework. The EKC hypothesis assumes an inverted U-shaped connection between EG and environmental pollution [5,47]. Based on the EKC, pollution increases until EG reaches a certain point, but environmental quality improves once EG surpasses this threshold. This can be explained by the economy’s evolution from an industrial to a service-oriented structure, the increase in education, and the resulting increase in environmental awareness. In this context, the relationship between per capita income (economic development) and environmental degradation forms an inverted U-shaped curve [48]. However, the N-shaped EKC approach suggests that the inverted U-shaped EKC does not remain constant in the long run. Over time, the increase in GDP may begin to degrade the quality of the environment. According to the researchers, the N-shaped EKC occurs when the scale effect dominates the composition effect caused by the underutilization of efficient technologies [49].

In this framework, the link between EG and the environment has been examined in the previous literature, focusing on a linear relationship and the EKC hypothesis. Majeed and Mazhar [13], in their study on OECD, MENA, BRICS, G7, and B&R country groups, state that GDP increases EF. Abid et al. [30] examined the impact of GDP on the EF in low-income, lower-middle-income, upper-middle-income, and high-income country groups. Based on the findings, GDP increases the environmental quality by reducing the EF in all groups except the lower-middle-income country group. But, in lower-middle-income nations, GDP increases the EF. The authors emphasize that differences in regional and income levels cause variations in levels of environmental sensitivity. Shahbaz et al. [50] state that GDP increases the EF in the 10 countries with the highest EF and suggest that policymakers should direct financial resources towards renewable energy.

Among the studies examining the non-linear relationship within the framework of the EKC hypothesis, the studies conducted by Saboori and Sulaiman [51] in Malaysia and Fan and Zheng [52] in the Sichuan Province of China do not support the EKC. The authors emphasize that the EKC is invalid when there is insufficient per capita income. However, Shahbaz et al. [53] confirmed the validity of the EKC hypothesis for India and Javid and Sharif [54] for Pakistan. Similarly, Can and Gozgor [55] found an inverted U connection between GDP and CO₂ emissions in France. This study suggests that green energy sources should be promoted to support EG. Conversely, Numan et al. [56] analyzed the 2001–2020 data from high-income and lower-middle-income nations, excluding upper-middle-income countries. It is emphasized that the N-shaped EKC relationship depends on the countries’ income levels. Based on the study’s results, the authors suggest that policies should be tailored to each country’s income level and environmental degradation status, and additional measures should be implemented to increase the use of renewable energy. Similarly, Jahanger et al. [7] found that nations producing the most nuclear energy support the N-shaped EKC hypothesis. In line with this finding, the authors emphasized that the use of renewable energy and nuclear energy sources should be increased.

Luo and Sun [8] stated that energy efficiency reduces the EG in G20 countries and confirms the EKC hypothesis. The authors argue that increasing energy efficiency and providing support and incentives are necessary to reduce environmental pollution. Similarly, Saud et al. [35] confirmed the N-shaped EKC hypothesis in EU states from 1990 to 2019. The authors stated that EU states heavily depend on fossil fuel energy sources to increase their EG and try to achieve these goals through the manufacturing sector. The authors argue that improving energy efficiency and providing support and incentives are necessary to reduce pollution.

H1. 

There is an N-shaped relationship between economic growth and carbon footprint.

2.2. Financial Development and Environmental Sustainability

Financial development (FD) is seen as an integral part of EG. Securities and trading provide an effective tool to ensure that available funds are used more efficiently, allowing funds to be channeled to areas experiencing resource shortages. According to this approach, the financial sector bridges the deficit and surplus sectors of the economy. In this way, FD improves the mobilization, efficient use, and tracking of funds. FD contributes to the economy by enabling a more efficient allocation of resources, reducing transition costs, increasing the amount of credit available to households and businesses, and encouraging high-yield investments [38,57,58]. FD supports economic activity by facilitating individuals’ access to credit and reducing poverty by lowering costs. This process paves the way for developing new financial markets, institutions, and instruments. Kashyap [59] states that rapid changes are occurring in the new era of financial innovation, with many new actors from different sectors involved in the financial ecosystem. The author also emphasizes that significant opportunities for innovative solutions have emerged as traditional institutions, technology giants, and startups compete in global markets. As the FD increases, financial costs decrease. This situation can encourage businesses and individuals to invest more and purchase new equipment and machinery, thereby increasing energy demand. It may promote switching to expensive products that lead to high energy emissions and consumption. Moreover, the development of stock markets reduces firms’ financial costs, increases their liquidity, and enables them to increase their productivity. Financial development minimizes the problem of asymmetric information, thereby providing low-cost financing and expanding financing channels. Accordingly, production volume increases. However, this situation may lead to CO₂ and other harmful emissions being released into the environment.

In this context, Hafeez et al. [60] analyzed empirical findings and found that FD increased CF in BRI countries between 1990 and 2017. Similarly, Baloch et al. [61] found that FD increased EF in Belt and Road countries between 1990 and 2016. The authors attributed this to FD increasing individuals’ needs and the financial sector directing resources to firms. This process leads to the intensification of production activities, causing an increase in industrial waste and environmental degradation. The authors also argue that FD supports infrastructure projects by providing medium- and long-term development loans, which may require large-scale land, water, and air resources, thereby potentially increasing EF. Sharma et al. [62] found that FD increased EF and CF in eight developing countries in Southeast and South Asia during the period 1990–2015. They also found that the development of the financial sector plays an essential moderating role in the link between energy and environmental footprints. Shoaib et al. [63] found that the financial development of D8 countries from the developing country group and G8 countries from the developed country group increased CO₂ emissions. Shahbaz et al. [64] investigated the relationship between FD and carbon emissions for G7 countries from the developed country group. An M-shaped relationship between FD and carbon emissions was found in Canada, Japan, and the US; an inverted N-shaped relationship in France, Italy, and the UK; and a W-shaped relationship in Germany. Wen et al. [65] examined the relationship between the financial development and CO₂ emissions of G20 countries from developed and developing country groups. The authors found an inverted U-shaped relationship between financial development and CO₂ emissions for developing countries, and a U-shaped relationship for developed countries. They explained this situation by stating that the level of financial development increases carbon emissions until a certain point. Still, beyond this level, CO₂ emissions are suppressed due to the impact of technological development. The authors stated that differences in the level of financial development among heterogeneous country groups affected the results. Ashraf et al. [66] found an inverse U-shaped relationship between FD and EF in a global sample. The authors explain this relationship with the decreasing scale effect of FD and the increasing technological and compositional effects that transform the economic structure. They also argue that the successful practices of financial institutions in making environmentally friendly investment decisions should be transformed into a global conscious investment culture. Sun et al. [67] confirmed an inverse U-shaped relationship between FD and CF in South Asian countries between 2000 and 2018. The authors explained that the financial sectors in South Asian countries have not yet reached a sufficient level and that FD policies in these countries are not fully aligned with environmental sustainability goals. Uddin et al. [68] stated that the financial development level of 119 developed and developing countries increased EF. Saqib et al. [69] found that FD hindered green growth in the ten countries with the highest EF between 1990 and 2019. The authors note that the allocation of financial resources is generally seen as an element that encourages EG, but at the same time creates a significant environmental burden. Uzar and Eyuboglu [70] found that the increase in FD in Türkiye between 1990 and 2021 led to an increase in EF, while the decrease in FD tended to increase EF at a higher rate. The authors suggest that this implies, on one hand, that FD supports activities that increase EF. On the other hand, a decline in FD, interpreted as economic stagnation or financial instability, has more serious negative effects on environmental sustainability. Horobet et al. [71] examined the relationship between financial development and environmental degradation in 40 European countries. The study found that as countries’ level of financial development increased, so did environmental degradation.

In contrast, there is also a view that FD has a positive impact on the environment. Some researchers argue that FD may have positive effects by encouraging technological investments to reduce environmental pollution. This demonstrates that FD has a complex structure that can both harm the environment and offer opportunities for sustainability [36,37]. In contrast, Dogan et al. [72] found that FD reduced EF in MINT countries during the period 1971–2013. The authors explained that the financial systems of host countries can effectively direct financial resources to environmentally sensitive sectors. Batala et al. [73] found that FD improved environmental quality in BRI countries. Jahanger et al. [74] analyzed a sample of 73 developing countries over the period 1990–2016. They found that FD reduced EF in Asian countries but did not have a similar effect in African, Latin American, and Caribbean countries. From a different perspective, Omoke et al. [75] show that increases in FD (positive shocks) significantly reduced EF, thereby improving environmental sustainability in Nigeria from 1971 to 2014. Conversely, declines in FD (negative shocks) significantly increase EF, negatively affecting environmental sustainability.

Based on the above discussion, we formulate the following hypothesis for EU countries:

H2. 

Financial development increases the carbon footprint.

2.3. Green Technology and Environmental Sustainability

Technological advances and innovative approaches have great potential in solving the pressing problems of energy transition and ecological sustainability. Green technology (GT) theory argues that it should include pollution control, waste management, recycling, monitoring, and various evaluation techniques. In this context, GT innovations are critical for all countries. However, the adoption and diffusion of GT innovations vary significantly from country to country [76]. Bergougui [77] emphasizes that GT plays an essential and multifaceted role in efforts to achieve ecological sustainability. According to Bergougui [77], green technology helps to overcome growth limits, improves adaptation and mitigation strategies for environmental pollution, and increases the efficiency of various factors such as green energy, the service sector, and environmental policies. Saqib et al. [69] state that the impact of green innovations on climate targets remains controversial due to the rebound effect, a phenomenon widely discussed in the literature. They also note that although green innovations are key elements of green growth policies, their impact on environmental goals is still controversial. The rebound effect is the idea that environmental innovation generates economic gains by increasing resource efficiency, but these gains may lead to greater resource consumption. These arguments focus on the perception that savings from the efficient use of resources are seen as a price reduction, which triggers an increase in resource demand [78]. In this framework, Lin and Ma [79] found that GT innovations had heterogeneous effects in different types of cities in China from 2006 to 2007. They also found that GT innovations effectively reduced CO₂ emissions after 2010, but this effect was not evident in Chinese cities before 2010. In addition, the authors emphasized that the CO₂ emission reduction effect of the GT innovation becomes significant when a city’s human capital reaches a certain level. Sharif et al. [80] found that GT reduced environmental degradation in G7 nations from 1995 to 2019. In contrast, Chang et al. [81] found that GT had no significant effect on carbon emissions in China from 2003 to 2019, while green knowledge and process innovations reduced environmental degradation. Balsalobre-Lorente et al. [82] found that high-innovation processes reduced environmental degradation in G7 nations from 1991 to 2018. The authors relate this finding to the endogenous growth theory, stating that, according to the theory, investments in the R&D sector contribute to nations’ achievement of environmental and economic prosperity through locally realized technological innovations. They also state that technological innovation in the energy sector is critical in transitioning from dirty to clean energy sources. Nketiah et al. [83] found that GT reduced environmental degradation from 1996 to 2022. The authors stated that policymakers should encourage the diffusion and implementation of GT and support the development of strong environmental policies.

Moreover, some argue that FD and GT development are linked. The financial system, through its functions such as information gathering and analysis, risk management, and credit allocation, contributes to sustaining technological innovations and increasing their efficiency. Moreover, enabling finance to allocate social resources more efficiently and accelerate the country’s green technology innovations is critical in achieving green economic growth [84]. Bergougui [77] concluded that increased FD increases the EF in Algeria from 1990 to 2024. In contrast, higher levels of GT reduce the negative effects of FD on the EF. Based on the above discussion, we formulate the following hypotheses:

H3. 

Green technologies reduce carbon footprint.

H4. 

Green technologies have a moderating role in the relationship between financial development and environmental quality.

2.4. ICT and Environmental Sustainability

Advances in digital technologies have significantly impacted business models, the economy, and living standards. Digitalization offers new revenue and value creation opportunities by transforming business processes. Digitalization contributes to energy savings through industrial transformations and increased energy efficiency. Information and communication technologies (ICTs), a form of digitalization, have made life easier for individuals and professionals while providing significant benefits for society and the economy. These technologies have reduced transaction and logistics costs, enabled businesses to reduce production costs thanks to the Internet, and enabled individuals to minimize risks by accessing information more easily. The use of ICTs can allow countries to improve energy efficiency. While the impact of digitalization on the economy’s orientation towards the service sector is not fully known, there is some evidence that it can reduce energy consumption. ICTs can potentially minimize transport costs, raise environmental awareness, and reduce environmental degradation through the use of green technologies. However, the environmental impacts of ICTs have recently started to be better understood. The increasing global demand for ICT products and services has increased energy demand and environmental costs. While the production and use of digital devices contain toxic substances such as chlorine, bromine, and lithium that can harm the environment, it also increases the use of valuable and limited resources such as copper, gold, and silver [85,86,87].

In this context, empirical evidence on the effects of digitalization on the environment varies. Zhang and Liu [39] found that ICT reduced environmental degradation in China from 2000 to 2010. Similarly, Batool et al. [40] found that ICT helped reduce environmental degradation in the South Korean economy in the medium and long term from 1973 to 2016. Differently, Khan et al. [88] found that ICT supports environmental sustainability in developed countries but has the opposite effect in developing countries across 91 countries from 1990 to 2017. The authors state that in developed countries, ICT reduces environmental pollution by increasing production efficiency and supports environmental sustainability. Moreover, it is noted that environmental degradation can be decoupled from economic growth thanks to the technological effects of ICT. However, it is emphasized that the potential of ICT to reduce environmental pollution in developing countries is limited; environmental pollution increases due to insufficient technological development and the scale effect of ICT. This situation reveals the negative impacts of ICT production, utilization, and waste management on the environment.

Huang et al. [21] found that information and communication technologies increased environmental degradation in E7 countries but decreased it in G7 countries from 1995 to 2018. Emphasizing that compensation and rebound effects are more dominant in developing countries, the authors state that since ICT infrastructure, such as mobile telephony, broadband, and internet, has increased significantly in E-7 countries, it is critical to determine how these technologies can be used most effectively to reduce environmentally damaging pollution. From a different perspective, Zhang et al. [41] found that universal connectivity services in China have increased environmental degradation. However, this negative impact can be offset by the presence of strict environmental policies. Shobande and Asongu [89] found that ICT has supported environmental sustainability in South Africa from 1980 to 2017, but its effects in Nigeria are more uncertain. Accordingly, the authors emphasize implementing emergency response programs targeting environmental infrastructure investments to reduce Nigeria’s climate change risk. Differently, Yadou et al. [90] found that ICT reduced the negative effect of remittances on environmental pollution from 2000 to 2021. Additionally, Le and Pham [91], in their study on a sample of 38 countries from 2006 to 2020, found that digitalization reduces environmental pollution. The researchers stated that by adopting policies supporting digital technologies, policymakers can promote financial inclusion, foster economic growth, and reduce environmental pollution. Adeshola et al. [92] state that in 23 European Union (EU) states between 2000 and 2017, the effect of ICT on greenhouse gas emissions is not positive and significant during periods of low environmental taxes. However, ICT has a reducing effect on greenhouse gas emissions when environmental taxes exceed the threshold value. They also state that ICT, renewable energy, and environmental taxes can be used effectively in the EU region to achieve long-run environmental sustainability. In this context, we formulate the following hypotheses based on the above discussions:

H5. 

ICT the carbon footprint.

H6. 

ICT moderates the connection between financial development and carbon footprint.

2.5. Environmental Policy Stringency and Environmental Sustainability

It is now generally accepted that environmental degradation creates negative externalities. Therefore, increasing the stringency of environmental policies (EPS) is crucial to ensure a clean and sustainable ecosystem. In this context, many countries have adopted environmental taxes on transportation and energy, as well as strategies to promote renewable energy consumption and production [26]. EPS can be defined as bearing the direct or indirect costs of environmentally harmful activities at the business or household level. Stringent environmental regulations can change producer and consumer behavior by increasing the cost of environmentally damaging activities. In this context, by setting higher taxes and lower emission targets for polluting industries, governments encourage economic actors to move away from fossil fuels and towards clean energy sources such as solar, wind, geothermal, and hydroelectricity. At the same time, policies that support the development of environmentally friendly technologies are an effective tool to reduce environmental degradation. Moreover, raising ecological standards for FDI allows for the prevention of environmentally damaging projects. Such regulations play a critical role in minimizing the negative impacts of socio-economic development on the environment [93,94]. In developing countries, however, environmental regulations are generally more flexible, leading to regional differences in the pricing and taxation of emissions. This flexibility in environmental policies, combined with trade liberalization, can facilitate the shift in pollution-intensive sectors to less regulated regions [95]. Kongbuamai et al. [96] found that stringent environmental policies reduced environmental degradation in BRICS nations from 1995 to 2016. The authors argue that stringency in environmental policy has been essential in improving environmental quality through strong legislation and effective governance, especially in China. Wen et al. [93] found that the probability of pollution-intensive FDI increased with lower levels of stringency in environmental regulations when these regulations were lax in BRICS countries between 2000 and 2020. The authors argue that strict environmental regulations may encourage pollution-intensive FDI, but this is not always the case at lower levels of regulation. The same polluting activity may be considered socially and economically inappropriate in developed regions, while it may be preferable in less developed areas. It is emphasized that these differences form the basis for forming pollution havens.

Li et al. [97] found that EPS reduced environmental degradation in BRICS nations from 1990 to 2019. Dai and Du [98] showed that EPS reduced environmental degradation in BIRCS-T countries from 1995 to 2021. The authors suggest that BRICS countries adopt strict environmental laws and regulations to minimize ecological degradation. They also state that environmental regulations should be strengthened, and legal measures should be taken to prevent companies from avoiding these regulations to produce products with higher carbon emissions. Similarly, Sohag et al. [99] found that EPS reduced the EF from 1990 to 2018. The authors argue that stronger environmental policies, supported by technological advances and clean energy initiatives, should be implemented to reduce the negative impact of human activities on natural resources. Luo et al. [100] also found that EPS reduced the EF from 1992 to 2020. The authors suggest that the combined impacts of macroprudential policies and EPS can be significant in ensuring a sustainable environment. In contrast, Dmytrenko et al. [101], in their study for the period 2000–2019 in Western Europe and Central Europe, found that the EPS has a significant impact only in Western Europe. The authors note that Western European countries are often taken as a reference point for Central European and Eastern European states. However, Central and Eastern European states have different resources and historical processes and face historical burdens, including the denial of environmental issues. Therefore, applying models that have been successful in Western Europe to these regions may have limited results.

Based on the above discussion, we formulate the following hypothesis:

H7. 

Strict environmental policies reduce carbon footprint.

2.6. Research Gap and Research Questions

Achieving sustainable development goals and reducing pollution are becoming increasingly important globally. With their strong environmental policies, EU states are leading the world in green transformation and environmental sustainability. This requires a different perspective on the interaction between environmental quality and financial development. However, no research exists on how ICT and GT shape the relationship between FD and CF in EU states. Moreover, previous studies generally focused on the ecological footprint and CO₂ when investigating the direct impact of FD on environmental pollution. This study is the second in the existing literature to examine the effects of FD on CF, and the first to be conducted in EU states. While providing comprehensive findings on how the moderating impact of ICT and GT shapes the relationship between FD and environmental pollution, this study offers a new perspective to the literature. It examines the contribution of GT to environmental sustainability in the context of FD and how ICT is transforming financial systems. In this context, this study is the first to analyze the indirect effects of ICT on the environment for EU states, link the CF reduction capacity of GT with the dynamics of FD, and assess the environmental impact of FD with a more holistic approach.

While the validity of the traditional U-shaped or inverted U-shaped EKC hypothesis has been tested in most studies in the literature, this study considers the N-shaped EKC hypothesis and finds that economic growth can increase CF after a particular stage. This finding makes an essential contribution to the analysis of the limits of sustainable economic growth. Moreover, this study empirically tests the N-shaped EKC for developed EU states by examining the effects on carbon emissions at different stages of economic development. In this respect, the results of this study fill an essential gap in the literature by helping policymakers better guide their sustainable growth strategies.

The focus of this study on EU states is based on four main factors. Firstly, the EU targets a 55% reduction in greenhouse gas emissions by 2030 and aims to reach net zero emissions by 2050. However, some EU states have an imbalance of ecological resources, with only 10% of their materials being recycled. This study is the first to examine the impact of financial development on the carbon footprint of the EU region. Despite being a leader in sustainable development goals, the EU has implemented the world’s most stringent environmental policies to achieve net zero emissions. Secondly, EU climate policies, high fossil fuel prices, and the need to increase renewable energy sources and reduce fossil fuel consumption have prioritized green innovation. In addition, EU states have increased investments in digitalization to ensure a better future for their economies [28,29,102,103,104]. The moderating effect of ICT and green technologies on environmental sustainability in the context of financial development has not been previously investigated in the literature. Therefore, the current study is the first in the literature to consider this. Thirdly, CF varies significantly across EU member states depending on their economic growth levels and industrial structures. Therefore, a region analysis fills the gap in the literature by contributing to a better understanding of FD’s environmental impact. Fourthly, EU states provide a suitable opportunity for the empirical analysis of EKC. The EU is one of the world’s most developed and organized regions, known for its high-income level, economic stability, and strengthening of environmental protection regulations. However, existing studies typically analyze outdated or limited data sets using traditional methods. This study uses econometric models to present new findings based on comprehensive, up-to-date data sets, thereby filling a gap in the literature. The study examines the subject in greater depth by employing advanced, robust econometric methods for heterogeneous EU countries with varying characteristics. In conclusion, this study analyzes the impact of different levels of GDP, FD, ICT, GT, and EPS on EU states’ CF for the first time. Moreover, it aims to address the unique context of heterogeneous EU states and fill the existing research gap by revealing how FD, moderated by ICT and GT, affects the FD-CF relationship using advanced econometric methods.

Based on these discussions, the following research questions are posed for EU states:

Q1:

Is there an N-shaped connection between economic growth and carbon footprint?

Q2:

Does financial development increase carbon footprint?

Q3:

Do green technologies reduce carbon footprint?

Q4:

Do green technologies have a moderating role in the connection between financial development and environmental quality?

Q5:

Does ICT reduce carbon footprint?

Q6:

Does ICT moderate the connection between financial development and carbon footprint?

Q7:

Do strict environmental policies reduce carbon footprint?

3. Data Discussion, Model Construction, and Estimation Strategy

3.1. Data Discussion

The current study focuses on the effects of GDP, FD, ICT, GT, and EPS on environmental pollution. CF is considered an indicator of environmental pollution. CF is a measure that represents the total amount of carbon dioxide emissions caused directly or indirectly by an activity or accumulated throughout the life cycle of a product [105]. Therefore, CF is considered an important indicator of environmental pollution. In this context, it makes sense to use carbon footprint data instead of CO₂ emissions data, because the carbon footprint not only considers emissions but also includes the capacity to sequester carbon through natural means. Thus, environmental impact can be assessed more comprehensively based on the amount of ecologically productive land required for both the production and absorption of CO₂ [67]. Factors such as financial development, green technology, ICT, and economic growth affect not only production processes but also consumption habits and trade volumes. Therefore, considering CF when focusing on the environmental impacts of these variables provides a holistic approach.

The existing literature demonstrates varying impacts of FD on environmental pollution [7,66,67]. Therefore, this study re-examines this ambiguous relationship regarding EU states. Additionally, it incorporates ICT and GT, which are considered opportunities for environmental sustainability. The importance of ICT was clearly emphasized in the 2018 report published by the Interagency Task Team on Science, Technology, and Innovation, in line with the Sustainable Development Goals (SDGs) [106]. ICT is defined as a general-purpose technology that enables technical services and new technological developments [107]. When examining the relationship between ICT and environmental quality in the current literature, some studies [108,109] argue that ICT increases environmental pollution due to higher energy consumption. In contrast, references [110,111,112] support the idea that ICT improves environmental quality by enabling efficient energy use, reducing asymmetric information problems related to environmental sustainability, and developing transportation systems. Empirical findings [80,81,82] generally indicate that GT enhances environmental quality. Moreover, ICT is linked to the development of the financial system [113]. Conversely, a well-developed financial system can support the advancement of GT. From these perspectives, the study examines the direct effects of ICT and GT on environmental pollution and their moderating roles in the relationship between FD and environmental pollution. The study evaluates the relationship with EKC and includes stringent environmental policies in the model to provide a more comprehensive perspective.

In this context, the study analyzes data from 20 EU states (Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, the Netherlands, Poland, Portugal, Slovakia, Slovenia, Spain, and Sweden) for the period 1996–2022. The study period was defined as the range of dates from the first to the last date for which data on the indicators used to construct the ICT Index were available. Information on data sources, measures, and variable descriptions for the selected variables is reported in Table 1. As an indicator of ICT, the study uses a composite index constructed using the principal component analysis technique, incorporating representatives of mobile cellular subscriptions, fixed telephone subscriptions, and internet users. Such indicators can be seen as a measure of a country’s digital infrastructure and technological sophistication. Indicators such as mobile tele-fund usage and internet access reflect the level of adoption of information and communication technologies [114].

Table 1. Data Description.

Variable	Symbol	Variable Definitions	Source
Carbon footprint	CF: log of CF	Carbon footprint (gha per person)	GFN
Information and communication technologies	ICT	The ICT index is constructed by principal component analysis based on three components: Mobile cellular subscriptions Fixed telephone subscriptions (per 100 people) Individuals using the Internet	WB
Environmental Policy Stringency Index	EPS: log of EPS	The index measures the stringency of 13 environmental policy tools, primarily related to climate change and air pollution	OECD
Financial development	FD: log of FD	Financial development index (Broad-based index of financial depth, access, and efficiency)	IMF
Green Technology	GT: log of GT	Patents in environment-related technologies	OECD
Gross domestic product	GDP: log of GDP	Per capita, constant 2015 US$	WB
Gross domestic product	GDP2: log of GDP2	Square per capita, constant 2015 US$	WB
Gross domestic product	GDP3: log of GDP3	Cube per capita, constant 2015 US$	WB

3.2. Model Construction

Based on Grossman and Krueger’s [5] EKC theory, the study explains how EG, FD, technology, and EPS affect environmental degradation in EU states. According to this theory, the start of the economic growth process leads to increased production activities, which cause environmental degradation, accompanied by increased income. However, they argue that environmental degradation will not increase as income levels rise due to the effects of technological developments. On the contrary, it will decrease [5]. This situation is considered U-shaped. Previous studies [6,7,8,32] have primarily focused on the U-shaped EKC. This implies that environmental pollution increases until EG reaches a specific limit. However, once EG reaches a certain level, environmental quality improves. Nevertheless, such linkages can be N-shaped, and environmental quality may deteriorate again once a particular point of EG is exceeded. This may be because scale effects outweigh compositional and technical effects in the later stages of EG [115]. In this respect, the study focuses on the validity of the N-shaped EKC in EU states. Furthermore, FD is essential when discussing the connection between EG and the environment. Financial liberalization and development accelerate economic growth and affect environmental performance by attracting foreign and R&D investments. However, FD can also trigger industrial pollution and environmental degradation [12]. FD promotes credit formation, investment, and economic growth, which increases energy consumption and reduces environmental quality [116,117]. Farooq et al. [118] examined the connection between carbon emissions and financial development within the scope of EKC theory. Therefore, we include FD in our model. Conversely, modern economic growth theory states that technology is crucial to EG. Technological advances play a critical role in overcoming growth limits, reducing ecological pollution, and increasing the efficiency of other elements. However, while technological progress can improve environmental efficiency at the macro and micro levels, it can increase energy consumption and lead to environmental degradation [14,82]. Yao et al. [119], who investigated the relationship between digitalization and zero carbon within the framework of EKC theory, emphasize that digitalization increases energy efficiency and reduces environmental impact. Again, within the scope of EKC theory, Ullah et al. [120] examined the relationship between digitalization and carbon emissions in their study. In addition, Li et al. [121] examined digitalization and environmental degradation within the framework of ecological modernization theory, which deals with environmental problems and technological development. Therefore, we consider the development of green technology and digitization for broader implications in the context of the EKC. Environmental degradation is now recognized to create negative externalities, so solving these problems should not be left to market forces. The effectiveness of environmental policies to maintain a clean ecosystem has become even more crucial. In this regard, policy instruments such as the imposition of environmental taxes and strategies to promote the use of renewable energy are being adopted by many countries [26]. When income and environmental degradation increase significantly, more stringent environmental regulations are imposed on the economy. This can lead to a shift in domestic production of polluting products to other countries with generally lower incomes and more lax environmental laws [122]. Nabi et al. [123] investigated the relationship between EPS and environmental degradation under the EKC theory. In addition, Sohag et al. [99] conducted a study on environmental policies within the framework of EKC theory. We also include EPS in our model within the scope of these arguments. Ultimately, our goal is to offer fresh insights into the interplay between innovation, finance, and the environment by examining how ICT and green technologies influence the link between FD and CF.

The environmental quality functions are tested under the model structure applied in this study; the CF measures environmental quality. At the same time, ICT, FD, EPS, GDP, and GT, which are hypothesized to affect CF, are used as explanatory variables. Additionally, the effects of the interactions of financial development with ICT (FD × ICT) and green technologies (FD × GT) on CF are also examined. Equations (1)–(3) show the functions tested for the effects of the explanatory variables on CF.

$${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}=f({{{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^2,{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^3,\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}},\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{E}\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}})$$

(1)

$${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}=f({{{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^2,{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^3,\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}},\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{E}\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}})$$

(2)

$${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}=f({{{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^2,{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^3,\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}},\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{E}\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}})$$

(3)

Logarithmic transformation calculates parameter values in the tested functions for the CF, FD, GDP, EPS, and GT series. In contrast, the ICT index series is analyzed using its natural structure. Equations (4)–(6) estimate CF functions using econometric models.

$${\mathrm{L}\mathrm{n}(\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}})={\mathsf{\alpha }}_0+{{{\mathsf{\alpha }}_1\mathrm{L}\mathrm{n}\left(\mathrm{G}\mathrm{D}\mathrm{P}\right)}_{\mathrm{i}\mathrm{t}}+\mathsf{\alpha }}_2{{\mathrm{L}\mathrm{n}\left(\mathrm{G}\mathrm{D}\mathrm{P}\right)}_{\mathrm{i}\mathrm{t}}^2+{{\mathsf{\alpha }}_3\mathrm{L}\mathrm{n}{\left(\mathrm{G}\mathrm{D}\mathrm{P}\right)}_{\mathrm{i}\mathrm{t}}^3+\mathsf{\alpha }}_4\mathrm{L}\mathrm{n}\left(\mathrm{F}\mathrm{D}\right)}_{\mathrm{i}\mathrm{t}}{+\mathsf{\alpha }}_5{\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_6{\mathrm{L}\mathrm{n}(\mathrm{E}\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}})+{{\mathsf{\alpha }}_7{\mathrm{L}\mathrm{n}(\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}})+\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}$$

(4)

$${\mathrm{L}\mathrm{n}(\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}})={\mathsf{\alpha }}_0+{{{\mathsf{\alpha }}_1\mathrm{L}\mathrm{n}\left(\mathrm{G}\mathrm{D}\mathrm{P}\right)}_{\mathrm{i}\mathrm{t}}+\mathsf{\alpha }}_2{{\mathrm{L}\mathrm{n}\left(\mathrm{G}\mathrm{D}\mathrm{P}\right)}_{\mathrm{i}\mathrm{t}}^2+{{\mathsf{\alpha }}_3\mathrm{L}\mathrm{n}{\left(\mathrm{G}\mathrm{D}\mathrm{P}\right)}_{\mathrm{i}\mathrm{t}}^3+\mathsf{\alpha }}_4\mathrm{L}\mathrm{n}\left(\mathrm{F}\mathrm{D}\right)}_{\mathrm{i}\mathrm{t}}{+\mathsf{\alpha }}_5{\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_6{\mathrm{L}\mathrm{n}(\mathrm{E}\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}})+{{\mathsf{\alpha }}_7{\mathrm{L}\mathrm{n}(\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}})+{\mathsf{\alpha }}_8\mathrm{L}\mathrm{n}\left({\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}}\right)\times {\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}$$

(5)

$${\mathrm{L}\mathrm{n}(\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}})={\mathsf{\alpha }}_0+{{{\mathsf{\alpha }}_1\mathrm{L}\mathrm{n}\left(\mathrm{G}\mathrm{D}\mathrm{P}\right)}_{\mathrm{i}\mathrm{t}}+\mathsf{\alpha }}_2{{\mathrm{L}\mathrm{n}\left(\mathrm{G}\mathrm{D}\mathrm{P}\right)}_{\mathrm{i}\mathrm{t}}^2+{{\mathsf{\alpha }}_3\mathrm{L}\mathrm{n}{\left(\mathrm{G}\mathrm{D}\mathrm{P}\right)}_{\mathrm{i}\mathrm{t}}^3+\mathsf{\alpha }}_4\mathrm{L}\mathrm{n}\left(\mathrm{F}\mathrm{D}\right)}_{\mathrm{i}\mathrm{t}}{+\mathsf{\alpha }}_5{\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_6{\mathrm{L}\mathrm{n}(\mathrm{E}\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}})+{{\mathsf{\alpha }}_7{\mathrm{L}\mathrm{n}(\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}})+{\mathsf{\alpha }}_8\mathrm{L}\mathrm{n}\left({\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}}\right)\times \mathrm{L}\mathrm{n}\left({\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}}\right)+\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}$$

(6)

In the equations, ${\mathsf{\alpha }}_0$ is the constant term, ${\mathsf{\alpha }}_1-{\mathsf{\alpha }}_8$ are the estimation parameters,${\mathsf{\mu }}_{\mathrm{i}\mathrm{t}}$ is the error term, and i and t are the nations and time in the panel data.

3.3. Estimation Strategy

This study applies panel data analysis, which allows the simultaneous analysis of time and cross-sectional data for EU states to examine the factors affecting environmental pollution. This method investigates the econometric functions created based on the study’s theoretical framework. First, various preliminary tests should be used to obtain consistent and reliable results. Correlation, endogeneity, Variance Inflation Factor (VIF), cross-section dependence (CSD), slope heterogeneity, and stationarity analyses were utilized in this context. Correlation analysis is used to calculate the possible levels of relationship between series. The method that provides precise results based on the findings of the normal distribution is preferred. The Spearman correlation test, selected in the absence of the normality assumption, calculates the correlation coefficients between the series. The calculation of correlation coefficients is shown in Equation (7).

$$r_s=1-{\displaystyle \frac{6{\sum }_{i=1}^n{d}_i^2}{n(n^2-1)}}$$

(7)

In the given equation, $X$ and $Y$ represent any two variables, while $i = 1, 2, \dots , n,$ ($x_i, y_i$) represent the $i\mathrm{t}\mathrm{h}$ measurement of these variables. Explanatory variables with similar characteristics should not be included in the models. In addition, interdependence between the explanatory variable and other variables can lead to endogeneity. It is crucial to check for multicollinearity and endogeneity in the model before estimation. The endogeneity and multicollinearity problems among the estimated models indicate inconsistent model estimation results. The study uses the panel two-stage least squares model, block exogenous Wald test, and Sargan-Hansen test to detect the endogeneity problem. VIF analysis is used to detect the multicollinearity problem. VIF determines how far the parameter estimates and variances deviate from their valid values due to the multicollinearity problem. The calculation of VIF and the tolerance coefficient is shown in Equation (8).

$$VIF={\displaystyle \frac{1}{1-R^2}}={\displaystyle \frac{1}{Tolerance}}$$

(8)

Various factors, such as country-specific latent indicators, standard shocks, externalities, geographic effects, and economic and financial assimilation, which may occur in the cross-sections comprising the panel data, can create dependencies between cross-sections. CSD can be defined as the interdependence of cross-sectional units or the effect of a shock on different units within one of the cross-sectional units [124]. In other words, CSD indicates the correlation between the error terms calculated for each cross-sectional unit, such as each country, enterprise, or region in the panel data model. Estimating without CSD can lead to inconsistent and erroneous parameters. The criterion for selecting the test to be used in the studies is the time and cross-sectional values of the available data set. The calculation of the Bias-corrected LM test, as proposed by [125] and used when both the size and time value of the data set are significant, is shown in Equation (9).

$${LM}_{BC}=\sqrt{{\displaystyle \frac{1}{n\left(n-1\right)}}}{\sum }_{i=1}^{n-1}{\sum }_{j=i+1}^n\left(T{\widehat{P}}_{ij}^2-1\right)-{\displaystyle \frac{n}{2(T-1)}}$$

(9)

The CDS test determines which first- or second-generation unit root tests will determine the series’ stationarity. The different characteristics of the EU states, even though they are in the same geographical region, indicate a heterogeneous structure. Slope heterogeneity is another test that decides which unit root test and cointegration method to use in the stationarity test. The Delta test developed by Swamy [126] is calculated as stated in Equations (10) and (11) [127].

$$\tilde{\Delta }=\sqrt{N}\left({\displaystyle \frac{N^{-1}\mathrm{Š}-k}{\sqrt{2k}}}\right)$$

(10)

Assuming the null hypothesis of slope parameters homogeneity, the error terms are normally distributed, while the Delta test statistic is also normally distributed. A corrected delta test statistic is developed for small samples.

$${\widetilde{\Delta }}_{adj}=\sqrt{N}\left({\displaystyle \frac{N^{-1}Š-E\left({\tilde{Z}}_{jt}\right)}{\sqrt{Var\left({\tilde{Z}}_{jt}\right)}}}\right)$$

(11)

$E\left({\tilde{Z}}_{jt}\right)$ and $Var\left({\tilde{Z}}_{jt}\right)$ denote $k$ and $2k(T-k-1)/T+1$ respectively.

Due to economic, political, and social factors, as well as globalization, the economies of countries have become more integrated. Given the existence of CSD and heterogeneity, it is essential to test the stationarity of the series. It is proposed that the integration sequence of panel data be analyzed using non-parametric and parametric methods. To check the stationarity of the series based on the evidence of CSD, the PANIC test developed by Bai and Ng [128] and the ADF-based second-generation CADF tests developed by Pesaran [129] are applied in this research. PANIC allows for the separate testing of common factors and error terms [130]. The deterministic component $D_{it}=\sum _{j=0}^p{\delta }_i{t}^j$ is the individual-specific fixed effect $D_{it}={\delta }_i$ when $p=0$ and the individual-specific time trend when $p=1$. When there is no deterministic term, $p=-1$. Therefore, the data construction process can be expressed as follows [130]:

$$X_{it}=D_{it}+{\lambda }_i^{\prime }{F}_t+e_{it}$$

(12)

$$\left(1-L\right)F_t=C\left(L\right){\mathrm{Ƞ}}_t$$

(13)

$$e_{it}={\rho }_{it}{e}_{it-1}+{\epsilon }_{it}$$

(14)

In Equations (12)–(14), $F_t$, is the vector of common factors in dimension $r\times 1$ that trigger correlation between cross-sectional units. ${\lambda }_i$ is the vector of factor loadings in $r\times 1$ dimension. $e_{it}$ is the error term and $C\left(L\right)$ is the $r\times r$ matrix containing the polynomials of the lag operator, which can be expressed as $C\left(L\right)= \sum _{j=0}^{\infty }{C}_j{L}^j .$ In obtaining Panic tests, the first differences of Equations (12)–(14) are taken and expressed as follows:

$$∆X_{it}={\lambda }_i^{\prime }∆F_t+∆e_{it}$$

(15)

Another second-generation unit root test is the ADF-based CADF unit root test developed by Pesaran [129], which includes the lagged values of the unit mean and their first differences in the ADF model of each unit. Moreover, the purpose of this regression is to eliminate correlation across units. The CADF test assumes that each nation has time effects to a different degree and takes autocorrelation into account. This test is used in the N > T and T > N cases. The calculation of the CADF is shown in Equation (16) [129].

$${∆y}_{it}=a_i+b_i{y}_{i,t-1}+c_i{\overline{y}}_{i,t-1}+d_i{∆\overline{y}}_{i,t}+e_{it}$$

(16)

In Equation (16), ${\overline{y}}_{i,t-1}={\displaystyle \frac{1}{N}}\sum _{i=1}^N{y}_{i,t-1}$ and ${∆\overline{y}}_{i,t}= {\displaystyle \frac{1}{N}} ∆y_{i,t}$. The average of the CADF statistics of all units in the panel gives the CIPS statistic, which tests the overall stationarity of the panel. The calculation of the CIPS statistics is shown in Equation (17).

$$CIPS={\displaystyle \frac{1}{N}}{\sum }_{i=1}^N{CADF}_i$$

(17)

After reviewing the trend components and the series’ integration level, panel cointegration investigates the long-run linkages between non-stationary series. The second-generation panel cointegration tests consider CSD and heterogeneity. The current study employs the Westerlund-Durbin-Hausman test to examine the cointegration connection. Westerlund [131] introduces two approaches for testing a cointegration connection under these assumptions. The first is the Durbin-H group test, which permits the autoregressive parameter to vary across cross-sections, and the second is the Durbin-H panel test, which assumes a uniform autoregressive parameter across all cross-sections [131]. The panel data model utilized for this test is expressed as follows:

$$y_{it}={\alpha }_i+{\beta }_i{x}_{it}+z_{it}$$

(18)

$$x_{it}={\delta }_i{X}_{it-1}+w_{it}$$

(19)

The disturbance term $z_{it}$ is modeled to follow a series of equations incorporating common factors, thereby accounting for CSD.

$$z_{it}={\lambda }_i^{\prime }{F}_t+e_{it}$$

(20)

$$F_{jt}=p_j{F}_{jt-1}+u_{jt}$$

(21)

$$e_{it}={\varphi }_i{e}_{it-1}+v_{it}$$

(22)

In Equations (20)–(22), $F_t$; is the k-dimensional common factor vector $(j=1,\dots ,k)$; ${\lambda }_i^{\prime }$ is the vector consistent with the factor loadings. Also, $p_j<1$ for each $j$. When the first differences in Equation (20) are taken,

$$∆z_{it}={\lambda }_i^{\prime }∆F_t+∆e_{it}$$

(23)

In this equation, $∆z_{it}$ is unknown. Therefore, Principal Component analysis is used instead of OLS estimation. Accordingly, the resulting equation is as follows:

$$∆{\widehat{z}}_{it}=∆y_{it}-{\widehat{\beta }}_i∆x_{it}$$

(24)

${\widehat{\beta }}_i$ in Equation (24) is calculated by regressing $∆y_{it}$ on $∆x_{it}$. For $∆F_t$, the principal component estimator $∆{\widehat{F}}_t$ is calculated as the Eigen vector $\sqrt{T-1}$ times the largest value of the $\left(T-1\right)\times \left(T-1\right)$ dimensional matrix $∆{\widehat{z}}_{it} ∆{\widehat{z}}_{it}^{\prime }$. Meanwhile, the matrix of estimated factor loadings is given by ${\lambda }_i={\displaystyle \frac{∆{\widehat{F}}^{\prime }∆\widehat{z}}{(T-1)}}$. ${\lambda }_i$ and $∆{\widehat{F}}^{\prime }$ are decomposed and the first difference in their residuals is calculated.

$$∆{\widehat{e}}_{it}=∆{\widehat{z}}_{it}-{\lambda }_i^{\prime }∆{\widehat{F}}_t$$

(25)

The ${\widehat{e}}_{it}$ in Equation (25) can be expressed as ${\widehat{e}}_{it}=\sum _{j=2}^t∆{\widehat{e}}_{ij}$. The kernel estimator required for the Westerlund Durbin-Hausman Test is shown in Equation (26).

$${\widehat{w}}_i^2={\displaystyle \frac{1}{t-1}}{\sum }_{j=M_i}^{M_i}\left(1-{\displaystyle \frac{j}{M_i+1}}\right){\sum }_{t=j+1}^T{\widehat{u}}_{it}{\widehat{u}}_{it-j}$$

(26)

In Equation (26), ${\widehat{u}}_{it}$ denotes OLS residuals and $M_i$ denotes bandwidth. The value of ${\widehat{w}}_i^2$ is consistent with the estimate of the long-run variance $w_i^2$ of ${\widehat{u}}_{it}$. Accordingly, the calculation of Durbin H group and Durbin H panel $({DH}_G, {DH}_P)$ statistics is as follows:

$${DH}_g={\sum }_{i=1}^n{\widehat{S}}_i{({\tilde{\varphi }}_i-{\widehat{\varphi }}_i)}^2{\sum }_{t=2}^T{\widehat{e}}_{it-1}^2$$

(27)

$${DH}_p={\sum }_{i=1}^n{\widehat{S}}_n{(\tilde{\varphi }-\widehat{\varphi })}^2{\sum }_{i=1}^n{\sum }_{t=2}^T{\widehat{e}}_{it-1}^2$$

(28)

Following the cointegration findings, the CUP-FMOLS estimator, which can produce consistent estimation results under CSD and heterogeneity, was used to examine the long-term impact of the regressors affecting the carbon footprint. The findings were also robustified by applying the FMOLS. The FMOLS developed by Pedroni [132], the FMOLS estimation of the population parameter $\beta$ for country $i$, is mathematically represented as follows:

$${\widehat{\beta }}_i^{\mathrm{*}}={\left(X_i^{\prime }{X}_i\right)}^{-1}(X_i^{\prime }{y}_i^{\mathrm{*}}-T\delta )$$

(29)

In Equation (29), $y_i^{*}$ is the transformed variable, $T$ is time, and $\delta$ is the correction parameter for serial correlation. Like the FMOLS estimator, it is a consistent and efficient estimator against endogeneity, autocorrelation, and variance problems. The most important difference in the CUP-FMOLS estimator is that it considers CSD and heterogeneity. The CUP-FMOLS estimator was introduced to the literature by Bai et al. [133]. The CUP-FMOLS model, which assumes that the series are stationary of the same degree and have common factors in a multifactor model, is as follows:

$$E_{jt}=\gamma Y_{jt}^2+\beta Y_{jt}+{\epsilon }_{jt}\to x_{jt}^{\prime }+{\epsilon }_{jt}$$

(30)

The variables used are calculated as follows:

$${\epsilon }_{jt}={\lambda }_{jt}^{\prime }{F}_t+u_{jt}$$

(31)

$$x_{jt}^{\prime }=x_{jt-1}^{\prime }+{\epsilon }_{jt}$$

(32)

$$F_t=F_{t-1}+{\mathrm{Ƞ}}_{jt}$$

(33)

In addition, the Jarque–Bera test is used to test the assumption of normality; Ramsey’s Reset test is used to determine whether there is an identification error in the models; LM test is used for the autocorrelation problem, which is defined as the presence of a significant relationship between successive dependence or successive error unit values; and finally, BPG test is used for the results of changing variance, which refers to the variance of the error term being different (Figure 1). All of these are used as diagnostic tests.

Figure 1. Flowchart of the research method.

After estimating the elasticity coefficients of the explanatory variables on CF, the unidirectional or bidirectional causal relationships between the variables were examined using the method of Emirmahmutoglu and Kose [134] (E-K). The estimation findings were detailed to determine the relationship between the explanatory variables and CF. This test evaluates CSD and heterogeneity by calculating the Fisher test statistics using the LA-VAR method. Equation (34) estimates the null hypothesis H0: A_1,2,ij = 0 with (ki + dmax), while Equation (35) computes the Fisher test statistics.

$$z_{i,t}={\mu }_i+A_{i1}{z}_{i,t-1}+...+A_{{ik}_i}{z}_{i,t-k_{i^{\prime }}}+{\sum }_{1=k_i+1}^{k_i+d {max}_i}{A}_{i1}{z}_{i,t-1}+u_{i,t}$$

(34)

$$\mathsf{ƛ}=-2{\sum }_{i=1}^{N}ln(p_i) i=\mathrm{1,2},\dots ,N$$

(35)

Equations (36) and (37) determine the Fisher test statistics using the bootstrap method.

$$x_{i,t}={\mu }_i^x+{\sum }_{j=1}^{k_i+d {max}_i}{A}_{11,ij}{x}_{i,t-j}+{\sum }_{j=1}^{k_i+d {max}_i}{A}_{12,ij}{y}_{i,t-j}+u_{i,t}^x$$

(36)

$$y_{i,t}={\mu }_i^y+{\sum }_{j=1}^{k_i+d {max}_i}{A}_{21,ij}{x}_{i,t-j}+{\sum }_{j=1}^{k_i+d {max}_i}{A}_{22,ij}{y}_{i,t-j}+u_{i,t}^y$$

(37)

In the VAR model with k_i + dmax lags, dmax_i indicates the highest integration degree across cross-sections.

4. Results

Table 2 exhibits the summary statistics of the series. The mean of CF is 1.20, EPS is 2.50, FD is 0.59, ICT is 0.21, GT is 10.93, and GDP is 10.24. All the series are left-skewed except for CF and GT, while CF and GT are right-skewed. Based on the findings of the Jarque–Bera, the series is not normally distributed. Box plots and scatter plots have been presented in Table 2 to show the asymmetric movement and random distribution of the series. The distribution in the scatter plot indicates no multicollinearity.

Table 2. Descriptive statistics.

Box Plots
	CF	EPS	FD	ICT	GT	GDP
Mean	1.205	2.509	0.598	0.212	10.931	10.247
Med.	1.167	2.639	0.650	0.349	10.736	10.394
Max.	2.518	4.889	0.901	2.140	26.616	11.630
Min.	0.496	0.056	0.192	−2.117	0.901	8.694
Std. Dev.	0.359	0.916	0.177	0.907	4.102	0.616
Skew.	1.402	−0.376	−0.554	−0.120	0.755	−0.133
Kurt.	6.151	2.743	2.181	2.451	4.252	2.596
J-B	400.162 ***	14.214 ***	42.740 ***	8.088 ***	86.545 ***	5.266 *
Obs.	540	540	540	540	540	540
Scatters

* and *** indicate significance at 10% and 1%.

The heat plot, VIF, and tolerance values showing the correlation relationship between the series in the model are shown in Table 3. The heat-plot correlation table strongly links the CF and the explanatory variables. Tolerance and VIF tests were applied to detect the multicollinearity problem in this context. According to the results of this test, a VIF value less than 5 indicates no multicollinearity [135]. In this study, the average VIF value was found to be 1.868. This shows that there is no multicollinearity. Financial development, economic growth, ICT, and green technology are macroeconomic variables that are theoretically strongly interacting and determined simultaneously. The primary purpose of the endogeneity tests performed in the study is, however, to confirm that there is no statistical or econometric inconsistency or systematic deviation between the dependent variable and the explanatory variables in the panel data analysis process. To this end, the present study employed the Panel Two-Stage Least Squares (2SLS) method to assess the endogeneity issue comprehensively and utilized lagged variables. In this approach, under the assumption that explanatory variables may be endogenous, appropriate instrumental variables were defined, and the model was estimated in two stages. In addition, the Block Exogeneity Wald test was employed to assess the model’s exogeneity assumption. The results of this test indicated the absence of a systematic, simultaneous relationship between the dependent variable and the explanatory variables.

Table 3. Correlation matrix.

1/VIF	VIF	CF	1.000	−0.072	0.133	0.322	−0.122	0.463
0.625	1.601	EPS	−0.072	1.000	0.277	0.137	0.524	0.377
0.430	2.328	FD	0.133	0.277	1.000	0.576	0.056	0.649
0.587	1.703	ICT	0.322	0.137	0.576	1.000	−0.082	0.541
0.717	1.395	GT	−0.122	0.524	0.056	−0.082	1.000	0.103
0.432	2.315	GDP	0.463	0.377	0.649	0.541	0.103	1.000
Mean VIF	1.868		CF	EPS	FD	ICT	GT	GDP

Furthermore, the Sargan–Hansen over-specification test was conducted to evaluate the validity of the instruments utilized. The statistical outcomes substantiated the validity of the instruments and the model’s adequate specification. In view of these findings, the objective of the endogeneity tests is to ascertain the absence of statistical or econometric inconsistency between the dependent variable and the explanatory variables in the panel data analysis process. It does not purport to predict a causal direction. According to the results of the endogeneity test in Table 4, there is no endogeneity.

Table 4. Endogeneity test results.

Panel Two-Stage Least Squares
Variable	Coefficient	t-Statistic	Prob.
EPSRESID	−0.023	−0.234	0.814
FDRESID	0.006	0.086	0.930
ICTRESID	−0.015	−0.858	0.391
GTRESID	−0.002	−0.609	0.542
GDPRESID	0.010	0.774	0.438
Block exogenous Wald
Hypothesis—H0: Exogenous		X2(1)	Prob.
EPS	FD	1.340	0.247
	ICT	0.536	0.464
	GT	0.488	0.484
	GDP	0.031	0.859
FD	EPS	1.702	0.192
	ICT	0.153	0.695
	GT	0.757	0.384
	GDP	0.562	0.453
ICT	EPS	0.544	0.460
	FD	0.033	0.855
	GT	0.023	0.878
	GDP	0.658	0.417
GT	EPS	0.183	0.668
	FD	0.004	0.948
	ICT	1.201	0.273
	GDP	0.017	0.896
GDP	EPS	0.640	0.423
	FD	0.379	0.537
	ICT	0.007	0.930
	GT	0.054	0.815
Sargan-Hansen Test for Exogeneity of Instruments
Instrument specification:	Instrument validity	Sargan–Hansen J statistic	Prob(J-statistic)
@DYN(CF, -2) EPS(-1) FD(-1) ICT(-1) GT(-1) GDP(-1)	Model A	18.411	0.188
@DYN(CF, -2) EPS(-1) FD(-1) ICT(-1) GT(-1) GDP(-1) FD × ICT(-1)	Model B	16.576	0.219
@DYN(CF, -2) EPS(-1) FD(-1) ICT(-1) GT(-1) GDP(-1) FD × GT(-1)	Model C	18.054	0.659
H0: The instruments used in this model are valid

Table 5 shows the findings of the CSD and slope heterogeneity. These findings indicate that the variables have a heterogeneous structure except for EPS and ICT, and there is a CSD between cross-sections for all series and models. This can be interpreted as a shock between cross-sections affecting others.

Table 5. Results of CSD and slope heterogeneity tests.

Variable	Pesaran Bias-Corrected LM		Delta Tests
	Stat.	Prob.	$\tilde{\mathsf{\Delta }}$	Prob.	${\tilde{\mathsf{\Delta }}}_{\mathit{a}\mathit{d}\mathit{j}}$	Prob.
CF	136.414	0.000	2.575	0.005	2.731	0.003
EPS	206.467	0.000	−1.465	0.929	−1.554	0.940
FD	78.713	0.000	3.971	0.000	4.212	0.000
ICT	81.795	0.000	−1.530	0.937	−1.623	0.948
GT	71.592	0.000	2.606	0.005	2.764	0.003
GDP	115.260	0.000	4.879	0.000	5.175	0.000
Model A	9.353	0.000	10.903	0.000	13.235	0.000
Model B	10.948	0.000	10.735	0.000	13.389	0.000
Model C	13.608	0.000	9.551	0.000	11.913	0.000

The CIPS result is presented in Table 6. These results show that all the series have a unit root. Taking the first differences in the series, it is found that they are stationary at I(1). Moreover, when the results of the PANIC test are analyzed, similar results are obtained with the CIPS statistics. It is understood that the series contains a unit root at the level, but when the first difference is taken, the series is stationary at I(1).

Table 6. Unit root test results.

	Intercept
Variables	CIPS		PANIC
	CIPS t-Stat.	Prob.	PCe Choi	Prob.	PCe MW	Prob.
CF	−1.994	≥0.10	−0.413	0.660	36.299	0.637
ΔCF	−2.747	<0.01	10.151	0.000	130.794	0.000
EPS	−1.756	≥0.10	0.005	0.497	40.049	0.468
ΔEPS	−2.591	<0.01	11.167	0.000	139.885	0.000
FD	−1.836	≥0.10	−0.146	0.558	38.690	0.529
ΔFD	−3.079	<0.01	10.345	0.000	132.530	0.000
ICT	−1.928	≥0.10	1.108	0.133	47.660	0.135
ΔICT	−2.501	<0.01	13.103	0.000	157.204	0.000
GT	−1.933	≥0.10	0.770	0.220	46.889	0.210
ΔGT	−3.607	<0.01	12.799	0.000	154.480	0.000
GDP	−1.132	≥0.10	−2.188	0.985	20.429	0.995
ΔGDP	−3.574	<0.01	3.431	0.000	70.693	0.002
	Intercept and trend
Variables	CIPS		PANIC
	CIPS t-stat.	Prob.	PCe Choi	Prob.	PCe MW	Prob.
CF	−2.181	≥0.10	0.557	0.288	44.984	0.271
ΔCF	−3.963	<0.01	9.950	0.000	128.999	0.000
EPS	−1.748	≥0.10	−1.350	0.911	27.924	0.925
ΔEPS	−4.565	<0.01	10.159	0.000	130.872	0.000
FD	−1.992	≥0.10	−1.307	0.904	28.303	0.917
ΔFD	−4.496	<0.01	8.105	0.000	112.499	0.000
ICT	−1.960	≥0.10	0.916	0.179	45.988	0.175
ΔICT	−3.803	<0.01	13.416	0.000	160.001	0.000
GT	−2.354	≥0.10	−0.381	0.648	36.588	0.624
ΔGT	−4.969	<0.01	10.520	0.000	134.097	0.000
GDP	−1.527	≥0.10	−1.988	0.976	22.212	0.989
ΔGDP	−4.148	<0.01	2.694	0.003	64.099	0.009

Maximum lags 2. Critical values-intercept 1% (−2.39) 5% (−2.20) 10% (2.11), intercept and trend 1% (−2.90) 5% (−2.72) 10% (−2.63).

Table 7 shows the findings of the Durbin-H group statistics, which express the heterogeneity of the slope coefficients in the cointegration equation. Analyzing these results, it is possible to conclude that the cointegration link exists in Models A, B, and C.

Table 7. WDH cointegration results.

Model A		Value	Prob.	Model B		Value	Prob.	Model C		Value	Prob.
	DHg	3.181	0.001		DHg	2.662	0.004		DHg	3.107	0.001
	DHp	2.264	0.012		DHp	2.614	0.004		DHp	4.347	0.001

Table 8 shows the long-term coupling between the variables estimated by the FMOLS and CUP-FMOLS approaches. Diagnostic tests for autocorrelation, normal distribution, and heteroskedasticity prove that the estimation results are reliable, robust, and consistent. Moreover, the Ramsey Reset test shows that the explanatory variables do not have non-linear or higher-order effects on the CF. GDP, square, and cube are positively, negatively, and positively related to CF in all models. In this context, Hypothesis 1 is accepted. This finding reveals that GDP directly impacts CF and increases environmental degradation in the initial development stage. In the second stage, GDP growth decreases environmental degradation due to compositional and technical factors. However, in the third stage, environmental degradation increases again due to the technical obsolescence effect, but this increase is not as significant as in the first stage. In other words, the N-shaped EKC approach shows that the inverted U-shaped EKC does not remain constant in the EU states in the long run. Over time, the increase in GDP may start to deteriorate environmental quality. The N-shaped EKC occurs when the scale effect dominates the composition effect due to the underutilization of efficient technologies [49]. Therefore, it can be stated that efficient technologies are not used effectively in EU states. This finding is consistent with the findings of Numan et al. [56] for high-income countries and Saud et al. [35] for EU states. Saud et al. [35] note that EU states heavily depend on fossil fuel energy to achieve their economic growth targets, primarily through the manufacturing sector. Our findings support this view. Therefore, it is essential to encourage innovation and R&D investment in these countries to reduce the impact of technical obsolescence.

Table 8. Results of estimation.

	Robustness			Main Estimation			Post Estimation
	FMOLS			Cup-FMOLS
Model A Regressor	Coeff.	t-stat.	Prob.	Coeff.	t-stat.	Prob.	Test	F/t-stat.	Prob.
GDP	0.011	8.584	0.000	0.017	4.607	0.000	Jarque–Bera	0.746	0.688
GDP2	−0.021	−3.797	0.000	−0.046	−8.053	0.000	Ramsey’s Reset	0.130	0.896
GDP3	0.009	9.479	0.000	0.012	5.226	0.000	LM	1.520	0.157
FD	0.406	12.985	0.000	0.219	2.150	0.032	BPG	0.688	0.682
ICT	−0.125	−22.558	0.000	−0.115	−8.053	0.000	${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}=f({{{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^2,{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^3,\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}},\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{E}\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}})$
EPS	−0.135	−18.113	0.000	−0.116	−3.391	0.001
GT	−0.043	−3.048	0.002	−0.036	−7.166	0.000
Adj. R2	0.853 ***
Model B Regressor	Coeff.	t-stat.	Prob.	Coeff.	t-stat.	Prob.	Test	F/t-stat.	Prob.
GDP	0.008	8.010	0.000	0.016	3.236	0.002	Jarque–Bera	0.928	0.628
GDP2	−0.013	−6.761	0.000	−0.051	−8.210	0.000	Ramsey’s Reset	0.118	0.905
GDP3	0.006	14.417	0.000	0.013	5.792	0.000	LM	1.550	0.147
FD	0.431	17.158	0.000	0.228	2.057	0.040	BPG	0.672	0.716
ICT	−0.246	−9.727	0.000	−0.216	−6.626	0.000	${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}=f({{{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^2,{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^3,\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}},\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{E}\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}})$
EPS	−0.133	−18.221	0.000	−0.076	−10.598	0.000
GT	−0.036	−5.767	0.000	−0.027	−4.499	0.000
FD × ICT	−0.208	−5.609	0.000	−0.104	−6.325	0.000
Adj. R2	0.855 ***
Model C Regressor	Coeff.	t-stat.	Prob.	Coeff.	t-stat.	Prob.	Test	F/t-stat.	Prob.
GDP	0.009	6.069	0.000	0.014	3.063	0.004	Jarque–Bera	0.942	0.624
GDP2	−0.011	−5.355	0.000	−0.047	−8.557	0.000	Ramsey’s Reset	0.156	0.875
GDP3	0.007	11.264	0.000	0.011	4.894	0.000	LM	1.512	0.160
FD	0.619	20.742	0.000	0.310	2.407	0.026	BPG	0.602	0.776
ICT	−0.100	−21.235	0.000	−0.113	−5.745	0.000	${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}=f({{{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^2,{\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i}\mathrm{t}}^3,\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}},\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{E}\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}},{\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}},{\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{G}\mathrm{T}}_{\mathrm{i}\mathrm{t}})$
EPS	−0.085	−11.293	0.000	−0.057	−13.341	0.000
GT	−0.205	−13.089	0.000	−0.103	−3.418	0.001
FD × GT	−0.033	−16.332	0.000	−0.021	−5.781	0.000
Adj. R2	0.864 ***

*** indicates significance at 1%.

FD increases CF in all models. In this context, Hypothesis 2 is accepted. This finding is consistent with the findings of Sharma et al. [62] for Asian countries and Saqib et al. [69] for countries with the highest EF. This confirms the view of Baloch et al. [61] that development increases human needs, and the financial sector’s intensification of production activities by transferring resources to firms leads to increased industrial waste and environmental degradation. Wang et al. [136] discovered that financial development is associated with an increase in environmental degradation in OECD countries. The study posits that this phenomenon can be attributed to the robust financial sectors in these regions, which have been instrumental in providing substantial financial support through various channels, including business establishment and expansion, infrastructure investments, and low-interest credit opportunities. These processes have the potential to increase natural resource consumption and contribute to environmental pollution. Moreover, domestic and foreign investment activities encouraged by financial development are also among the key determinants of environmental degradation. While financial development has been shown to support research and investment in clean technology and environmentally friendly projects, the study results indicate that its adverse effects on environmental quality exceed its potential benefits in OECD countries. Wen et al. [65] examined the relationship between financial development and CO₂ emissions; they determined that this relationship follows an inverted U shape in developing countries and a U shape in developed countries. The authors of the study posited that disparities in financial development levels among heterogeneous groups significantly impact the observed outcomes. Horobet et al. [71] found that financial development in EU countries increases environmental degradation. This effect occurs in the long term through energy consumption and direct foreign investment inflows. It also appears in the short term through FDI and trade openness channels. The deleterious impact on the environment has been attributed to several factors, including increased energy consumption, a surge in pollution and waste production, and resulting environmental degradation. Moreover, it has been asserted that investors in select developing European countries have exploited the more lenient environmental regulations in these regions to operate, thereby engendering adverse environmental impacts. Consequently, it has been emphasized that policymakers should take this situation into account when developing environmental regulations. The present findings lend support to these views. Therefore, it can be posited that the correlation between financial development and environmental quality in EU countries depends on the countries’ environmental policies, potentially yielding divergent outcomes within heterogeneous groups. In this context, it can be concluded that FD policies in EU countries are not compatible with sustainable environmental goals. In conclusion, it is imperative to raise awareness among financial institutions regarding environmentally friendly investment decisions. Regulations can be implemented for investment projects to support environmental sustainability in the financial sector in EU countries. The financing of sustainable development in the renewable energy sector can be facilitated through the issuance of green bonds. Furthermore, policymakers should consider offering more favorable financing conditions to support green investments.

The impact of ICT and GT on CF is negative in all models. In this context, Hypothesis 3 and Hypothesis 5 are accepted. These findings are consistent with those of Khan et al. [88] and Huang et al. [21] for the sample of developed countries, those of Shobande and Asongu [89] for South Africa, and those of Adeshola et al. [92] for the EU states. On the contrary, it contradicts the findings of Khan et al. [88] and Huang et al. [21] for developing countries. Our findings suggest that in EU states, ICT reduces environmental pollution and promotes environmental sustainability by increasing production efficiency. In contrast, Adeshola et al. [92] argue that ICT reduces environmental pollution if strict environmental policies are implemented in EU states. They also suggest that ICT and environmental taxes can be used together to achieve environmental sustainability in the long term for the EU region. Our findings support this view. Moreover, the findings underline that GT significantly impacts the sustainable environment. Therefore, GT is essential in developing environmental pollution adaptation and mitigation strategies in EU states. GT can support sustainability in multifaceted areas such as energy efficiency, renewable energy development, and transportation. Therefore, supporting innovation development in EU states can be an effective tool for reducing environmental degradation. Similarly, EPS reduces CF in all models. In this context, Hypothesis 7 is accepted. This suggests that EPS minimizes environmental degradation by promoting renewable energy and GT. This finding is consistent with the findings of Li et al. [97] and Dai and Du [98] for BRICS nations and Dmytrenko et al. [101] for Western European countries. Sohag et al. [99] emphasize that to reduce environmental damage, it is necessary to implement more effective environmental policies supported by technological innovation and clean energy projects. Our findings support this view and show that technological innovation and the implementation of EPS are essential for a sustainable environment in EU states. Dmytrenko et al. [101] state that Western European countries often serve as a model for Central and Eastern European states. However, Central and Eastern European states have different resources and historical backgrounds and are influenced by historical burdens, including the neglect of environmental issues. Therefore, it is essential to implement policies in EU states, considering the differences in practice between countries, to achieve sustainable development goals.

In addition to the direct effect of the relevant variables, the study also focuses on the impact of the FD × ICT interaction and the FD × GT interactions on CF. Accordingly, in contrast to the direct effect of FD increasing CF, the interaction of FD and ICT and the interaction of FD and GT decrease CF (Model B and Model C). In this context, Hypothesis 4 and Hypothesis 6 are accepted. In this group of countries, innovation developments can mediate the negative effects of FD on environmental degradation. The interaction between financial development and ICT has the potential to transform the financial sector in terms of innovation and efficiency, thereby reducing environmental pollution by improving resource allocation and energy efficiency. Furthermore, enhancing financial inclusion through digitalization has the potential to stimulate investment in environmentally sustainable technologies and synchronize economic growth with environmental sustainability. The digitalization process has the potential to accelerate the development of green financial technologies in the financial sector, thereby facilitating the flow of capital to sustainable projects. Moreover, the enhanced transparency and accessibility to information engendered by digitalization can enable more effective evaluation of environmental risks and the allocation of investments toward environmentally friendly projects. Furthermore, financial development has the potential to enhance innovation’s role in environmental sustainability by facilitating the financing of research and development (R&D) activities and clean technologies. In this process, improving resource allocation efficiency and leveraging technologies that augment energy efficiency can play a pivotal role in curbing environmental degradation. Consequently, the findings of this study demonstrate that the integration of technological innovations within the EU financial sector has the potential to foster environmental sustainability. This assertion is supported by several positive effects, including the reduction of CO₂ emissions, the facilitation of green project financing, and the protection of environmental standards. Conversely, as posited by Feng et al. [19], the progression of digitalization has the potential to foster the development of clean energy sources by means of augmenting household credit size and income levels. Consequently, financial system fund transfers can be allocated toward clean technology and energy. Additionally, FD has the capacity to provide financial support for the development of GT. The adverse consequences of FD can be mitigated by transitioning investments from polluting projects to supporting GT (see Figure 2).

Figure 2. Graphical summary of results. Note: “+” indicates a positive relationship. “−“ indicates a negative relationship. "---" indicates an N-shaped relationship.

Empirical support is provided for the key findings on the N-shaped EKC relationship reported in Table 8; income turning points were calculated using a third-degree regression cubic polynomial applied to the annual GDP growth rates for each country in the sample. The location of the countries in the final upward-sloping portion of the N-shaped EKC is presented in Table 9. Among the twenty EU countries, Austria, Belgium, the Czech Republic, France, Greece, Hungary, Italy, Luxembourg, Slovenia, Spain, and Sweden are located in the final upward-sloping part of the N-shaped form in the EKC hypothesis. This phenomenon suggests that environmental enhancement in these nations has not yet achieved sustainability over an extended timeframe, indicating a reemergence of scale effects. Moreover, the failure of these nations to adequately reduce their reliance on fossil fuels in their energy transition processes indicates a persistent high energy intensity in their industrial and chemical sectors. Consequently, investments in energy transition and clean production have not yet reached a sufficient level. Conversely, Estonia, Finland, the Netherlands, Portugal, and Slovakia exhibit a U-shaped trend. These countries are European nations that have demonstrated notable advancements in clean production and energy efficiency. The completion of structural transformation in these countries and the high share of renewable energy may be contributing factors. In contrast, Denmark, Germany, Ireland, and Poland exhibit additional cubic forms. The underlying factors contributing to this phenomenon may include unstable environmental-economic relations, economic and structural fluctuations, energy structure, and industrial dependency factors in these countries.

Table 9. N-shaped EKC results for EU countries.

Country	Shape	Coeff.GDP	Coeff.GDP2	Coeff.GDP3
Austria	N-shaped (final upward)	0.001	−0.036	0.186
Belgium	N-shaped (final upward)	0.001	−0.036	0.222
Czechia	N-shaped (final upward)	0.001	−0.053	0.624
Denmark	Not N-shaped	0.000	0.012	−0.370
Estonia	Not N-shaped	0.001	−0.025	−0.188
Finland	Not N-shaped	0.001	−0.009	−0.329
France	N-shaped (final upward)	0.001	−0.027	0.091
Germany	Not N-shaped	0.000	0.011	−0.131
Greece	N-shaped (final upward)	0.005	−0.162	0.925
Hungary	N-shaped (final upward)	0.002	−0.065	0.460
Ireland	Not N-shaped	−0.001	0.083	−1.618
Italy	N-shaped (final upward)	0.002	−0.053	0.229
Luxembourg	N-shaped (final upward)	0.001	−0.039	0.140
The Netherlands	Not N-shaped	0.001	−0.004	−0.323
Poland	Not N-shaped	0.000	0.010	−0.279
Portugal	Not N-shaped	0.001	−0.004	−0.453
Slovakia	Not N-shaped	0.001	−0.031	0.302
Slovenia	N-shaped (final upward)	0.002	−0.050	0.199
Spain	N-shaped (final upward)	0.002	−0.070	0.342
Sweden	N-shaped (final upward)	0.001	−0.028	0.150

In the present study, Emirmahmutoglu and Kose’s [134] method (E-K) was employed to support and elaborate the long-term coefficient estimation results between explanatory variables and CF, and to capture country-specific heterogeneity (see Table 10). When all EU countries were evaluated collectively, it was determined that all explanatory variables were causes of CF. This finding serves to substantiate the notion that the carbon footprint is subject to the influence of macroeconomic, financial, and environmental variables. The robust causality observed across the panel suggests a collaborative influence of economic growth, financial development, ICT, environmental policy, and green technology on environmental indicators. However, at the country level, while relationships weaken in developed EU countries like Denmark and Germany, they strengthen in emerging EU countries such as Portugal and Slovakia. This phenomenon reveals a heterogeneity of the EKC hypothesis, with different patterns observed across EU countries.

Table 10. Causality results for the panel and country.

Causality	GDP ⟶ CF		FD ⟶ CF		ICT ⟶ CF		EPS ⟶ CF		GT ⟶ CF
	Fisher/Wald	Prob.	Fisher/Wald	Prob.	Fisher/Wald	Prob.	Fisher/Wald	Prob.	Fisher/Wald	Prob.
Panel	140.987	0.020	133.240	0.000	114.502	0.000	108.193	0.000	114.323	0.000
Austria	14.768	0.002	0.885	0.347	0.012	0.912	4.103	0.129	2.793	0.095
Belgium	7.089	0.029	1.767	0.184	17.265	0.016	4.588	0.101	18.844	0.000
Czechia	4.707	0.030	7.188	0.066	15.642	0.029	3.313	0.346	0.359	0.549
Denmark	0.898	0.343	0.001	0.973	6.923	0.328	9.620	0.022	2.926	0.087
Estonia	6.085	0.014	6.639	0.036	7.730	0.357	0.877	0.831	0.447	0.504
Finland	3.233	0.072	11.247	0.010	3.429	0.843	9.507	0.009	10.177	0.001
France	13.311	0.001	1.542	0.214	0.637	0.425	6.474	0.039	2.760	0.097
Germany	9.414	0.002	14.802	0.001	7.381	0.194	3.683	0.298	1.716	0.424
Greece	4.916	0.086	3.213	0.073	8.014	0.331	11.250	0.004	0.929	0.335
Hungary	0.66	0.417	6.275	0.043	2.500	0.287	0.200	0.654	0.403	0.526
Ireland	4.06	0.131	9.915	0.007	0.153	0.695	2.717	0.099	7.565	0.006
Italy	11.638	0.003	7.521	0.057	1.434	0.231	14.323	0.001	7.405	0.007
Luxembourg	2.733	0.098	0.262	0.609	59.451	0.000	5.493	0.019	0.253	0.969
The Netherlands	7.931	0.019	36.977	0.000	12.527	0.085	1.059	0.304	13.442	0.001
Poland	1.151	0.765	0.023	0.880	9.915	0.042	0.024	0.876	1.188	0.276
Portugal	13.307	0.004	0.275	0.600	18.091	0.012	6.089	0.014	1.232	0.267
Slovakia	0.511	0.475	0.257	0.612	18.673	0.009	0.119	0.730	0.098	0.755
Slovenia	2.676	0.102	2.188	0.139	8.816	0.266	2.160	0.340	4.256	0.235
Spain	12.321	0.000	12.296	0.002	3.672	0.817	13.110	0.000	11.113	0.004
Sweden	2.959	0.085	0.209	0.648	10.391	0.167	2.803	0.423	6.728	0.035

Specifically, the results show that Austria, Belgium, Czechia, Estonia, Finland, France, Germany, Greece, Italy, Luxembourg, the Netherlands, Portugal, Spain, and Sweden confirm the negative impact of economic growth on the carbon footprint, consistent with the first rising phase of the EKC. Across the EU, economic growth has led to increased environmental pressures in southern countries like Portugal, Spain, and Italy, as well as in central European countries such as Austria, the Czech Republic, and Germany. However, in northern and Baltic European countries such as Denmark, Ireland, and Sweden, this relationship has been eliminated due to the implementation of robust environmental management strategies. The analysis indicates that financial development is the primary driver of the carbon footprint in the following countries: the Czech Republic, Estonia, Finland, Germany, Greece, Hungary, Italy, Ireland, the Netherlands, and Spain. This phenomenon may indicate a nexus between financial systems and environmental degradation, characterized by the escalating demand for energy due to increased production and consumption. In contrast, a causal relationship between financial development and carbon footprint is not evident in Austria, Belgium, Denmark, France, Luxembourg, Poland, Portugal, Slovakia, Slovenia, and Sweden. The allocation of financial resources to environmentally friendly investments, the existence of strong regulations, and the prevalence of green finance instruments are factors that can explain this phenomenon. A statistically significant impact of ICT on the carbon footprint has been observed in Belgium, the Czech Republic, Luxembourg, the Netherlands, Poland, Portugal, and Slovakia. The findings indicate that the impact of ICT on the carbon footprint exhibits variation across nations. While the expansion of ICT in Central European countries has been shown to increase carbon intensity, this effect is counterbalanced by environmentally conscious policies in Western and Northern European countries with advanced digital infrastructure. The environmental policies of these nations have exerted a substantial influence on the carbon footprint in Belgium, Estonia, Finland, France, Greece, Italy, Ireland, Luxembourg, Portugal, and Spain. It is plausible to posit that implementing more stringent environmental policies in these countries can reduce carbon intensity by modifying energy structures, production technologies, and investment decisions. Conversely, in Austria, the Czech Republic, Denmark, Finland, Germany, Hungary, Poland, Slovakia, Slovenia, and Sweden, the EPS-CF relationship is not statistically significant, suggesting that shifts in environmental policy do not engender novel disparities in carbon performance. A statistical analysis reveals that green technology is a substantial factor in the carbon footprint of Austria, Belgium, Denmark, France, Germany, Ireland, Italy, the Netherlands, Spain, and Sweden. It can be posited that investments in green technology in these countries result in a reduction of carbon emissions. Conversely, no significant causality was identified in the Czech Republic, Estonia, Finland, Greece, Hungary, Luxembourg, Poland, Portugal, Slovakia, and Slovenia. This phenomenon may be attributed to the limited adoption of green technologies or the incomplete integration of environmental policies into the production system.

5. Conclusions

This research examines the effects of financial development, economic growth, ICT, green technology, and stringent environmental policies on environmental sustainability in EU states between 1996 and 2022. It also assesses the validity of the EKC hypothesis and investigates the moderating roles of ICT and green technology in the connection between financial development and carbon footprint. The results of the WDH cointegration test reveal a robust cointegration relationship between variables in all models. The CUP-FMOLS method is used to estimate the long-run coefficients of the variables reliably. The findings show an inverted N-shaped link between GDP and carbon footprint in EU states, thus confirming the EKC hypothesis. It has also been shown that financial development increases carbon footprint, while green technology, ICT, and stringent environmental policies contribute to its reduction. Finally, interactions between ICT and financial development, as well as green technology and financial development, are found to reduce the carbon footprint.

According to the results of the E-K [134] panel causality analysis, all explanatory variables were identified as causes of the carbon footprint across the EU. However, at the country level, a heterogeneous structure is observed. An analysis of the factors contributing to carbon footprints in select European countries reveals that economic growth is a primary contributing factor in Austria, Belgium, the Czech Republic, Estonia, Finland, France, Germany, Greece, Italy, Luxembourg, the Netherlands, Portugal, Spain, and Sweden. The financial development of the countries above has been shown to significantly impact the carbon footprint in the following nations: the Czech Republic, Estonia, Finland, Germany, Greece, Hungary, Italy, Ireland, the Netherlands, and Spain. ICT has a significant impact in Belgium, the Czech Republic, Luxembourg, the Netherlands, Poland, Portugal, and Slovakia. The carbon footprint in Belgium, Estonia, Finland, France, Greece, Italy, Ireland, Luxembourg, Portugal, and Spain is influenced by environmental policies. A statistical analysis reveals that green technology significantly contributes to the carbon footprint in Austria, Belgium, Denmark, France, Germany, Ireland, Italy, the Netherlands, Spain, and Sweden. In general, while the relationship is weakening in developed EU countries, stronger causal relationships have been identified in emerging economies.

5.1. Policy Implications

The paper draws some policy conclusions based on the findings. (1) The N-shaped EKC hypothesis has been confirmed across the EU, showing that economic growth cannot sustain environmental improvement beyond a certain point. Consequently, it is imperative to promote innovation, increase R&D investments, and strengthen energy transition policies to mitigate the negative effects of technical obsolescence in these countries. (2) Financial development has been found to increase the carbon footprint. The widespread use of green bonds and sustainable financing instruments in these countries has the potential to transform the financial sector and promote environmental responsibility. Policy instruments formulated under the EU Sustainable Finance Action Plan should be implemented more effectively in these countries. (3) The implementation of green technology has been shown to reduce the carbon footprint within the EU. Therefore, funds in these countries should be effectively directed towards green R&D, energy transition, and industrial modernization projects. Furthermore, ICT reduces the carbon footprint at the group level. It is important to leverage digital infrastructure to strategically increase energy efficiency in these countries. (4) Findings show that the interaction between green technology and ICT reduces the carbon footprint and thus mitigates the negative environmental impacts of financial development. Consequently, innovation development should be encouraged by focusing on digital transformation and the financial sector to promote green practices in these countries. It is recommended that funds be directed towards green technology development. Additionally, to ensure the sustainability of these applications, it is important to include renewable energy consumption in technology development to prevent rebound effects. (5) The implementation of strict environmental policies within the EU has contributed to reducing the carbon footprint of EU countries. However, differences in implementation between countries persist. Eliminating these differences and strengthening innovation to reduce pollution caused by financial development are necessary for achieving the net-zero target.

5.2. Limitations and Directions for Future Research

This study focuses on specific factors affecting the carbon footprint in EU countries. However, the study also has some limitations. First, the study covers the period between 1996 and 2022. In the future, updating the data or extending the analysis period could increase the timeliness of the findings. Second, the model includes a limited number of explanatory variables. Including additional variables such as the share of renewable energy consumption, industrial value added, trade openness, and human capital could facilitate more comprehensive results for EU countries. Third, the study focuses on carbon footprint as an indicator of environmental sustainability; different environmental indicators could also be considered in the future. Fourth, although the study comprehensively examines the relationship between financial development and environmental sustainability, data access constraints should be considered. Specifically, the Digital Economy and Society Index (DESI) data, which represents digitalization, only covers the period 2014–2022. This 8-year period is insufficient for long-term effects, preventing the inclusion of the long-term effects of digitalization in the model. Similarly, green finance indicators were not considered as they are only available for limited time periods for many countries. Due to the inability to use a comprehensive digitalization measure in assessing the environmental impacts of financial development, the moderating role of ICT has been examined. Furthermore, the potential effects of green finance markets on reducing the carbon footprint are indirectly reflected in the current results. Future research could include DESI or similar digitalization indices and green finance data in the study to draw broader findings and policy implications. Finally, the variables used in the study (financial development, economic growth, ICT, and technological innovation) may be affected by potential endogeneity issues. In future research, the use of methods such as generalized moments could contribute to addressing these issues and strengthening evidence of causal relationships.