A Renewed Approach to the Connection Between Economic Complexity and Environmental Degradation Considering the Energy Innovation Process in the Five Major European Economies

Abstract

This study examines the potential for long-term sustainable development in highly developed and sophisticated economies by analyzing the N-shaped environmental Kuznets curve in the relationship between the economic complexity index and the ecological footprint of the five largest European economies from 1994 to 2020. It also assesses the joint impact of R&D expenditure, renewable energy consumption, environmental taxation, and the moderating effect of the R&D effort-economic complexity index linkage on the economic complexity-ecological footprint nexus. Long-run estimation was conducted using Driscoll-Kraay standard error estimators. Subsequently, the panel-corrected standard errors robustness test was performed. The results confirm the existence of the N-shaped EKC in the economic complexity-ecological footprint relationship and reveal a significant inverse relationship between R&D expenditure, renewable energy consumption, and environmental taxation in their effect on the ecological footprint. The R&D expenditure dampening effect is also validated, indicating pathways toward long-term sustainability. Therefore, when considered together, enhancing renewable energy consumption, environmental taxes, and R&D expenditure can alleviate long-term ecological depletion in the highly developed and complex five largest European economies analyzed. These findings have essential economic and environmental policy implications.

1. Introduction

Since the late 19th century, and particularly from the 1950s onwards, the world has experienced significant economic, demographic, and urban growth, driven by technological advances and improved public health and living conditions [1]. However, such progress and heightened well-being in many modern societies have come at a considerable cost. The extensive and often uncontrolled exploitation of energy and natural resources, alongside the generation of pollution and waste, has severely compromised the integrity of the Earth’s ecosystems and biodiversity [2]. Human economic activities have contributed to global environmental degradation and rapid climate change [3,4], inflicting significant material and human damage and disproportionately affecting the most vulnerable populations [5].

Today, most societies acknowledge that economic and social progress is only possible if the planet’s survival is ensured. Therefore, sustainable development is a collective responsibility, particularly for the most advanced nations. Due to their significant political and economic influence, these nations’ development trajectories, actions, and environmental policies significantly impact and shape the capacity of other nations to pursue growth paths that align with nature [6]. Thus, this research examines the potential for sustainable development in Europe’s five large advanced economies, considering economic complexity, technological efforts, the promotion of energy transition through renewable resources, and environmental taxation.

Recent years have seen growing recognition of the ecological footprint (EFP) as a valuable metric for analyzing the environmental implications of economic development. Numerous studies have adopted the EFP to facilitate a more complete understanding of ecological degradation [7,8,9,10,11,12]. The EFP was initially introduced by [13] and subsequently elaborated upon by [14]. This metric quantifies the requisite natural assets and resources, in terms of biocapacity areas, to fulfill human consumption and activities, encompassing waste and emission absorption [15,16]. This metric has certain limitations, the most important of which is its focus on the pressure humans place on the environment without simultaneously considering the availability of biocapacity, that is, nature’s capacity to absorb these pressures at any given moment [17]. It remains highly relevant, however, in the analysis of the environmental constraints facing economic development. Furthermore, by representing a stock variable, it fails to reflect the impact of significant technological advances [18]. Nevertheless, it is crucial to recognize that the ongoing biocapacity deficit is related to biosphere regeneration, being a critical limiting factor of the human economy [19].

The environmental Kuznets curve (EKC) [20] has been widely utilized in studies examining the environmental impact of economic activities [21,22]. Although the EKC has usually had as many detractors as supporters and has been historically recognized and critically evaluated [23], it remains an undeniably valuable methodological tool in the economic development literature for exploring and assessing the medium- and long-term dynamics of human-induced environmental impacts [24,25,26]. The traditional environmental EKC model posits a quadratic relationship between economic growth and environmental degradation, characterized by an inverted U-shaped parabola [20]. Under this EKC hypothesis, as economies develop and per capita income rises, production methods, consumer demand, and societal priorities alter the productive structure, efficiency levels, and environmental pressures. Initially, this process increases environmental degradation, which, however, declines as economic progress continues, decreasing in line with elevated levels of income per capita and development [27].

The relationship between economic growth and its environmental ramifications is well documented in the literature [28,29,30]. It is worth noting that while economic complexity (ECI) may not directly correspond to economic growth, it serves as a more accurate indicator of a nation’s development and prosperity. ECI accounts for the intricate nature of a country’s productive structure and knowledge capabilities [31,32], encapsulating the advanced skills and knowledge that facilitate the production of diverse, high-value-added goods. Therefore, ECI provides a more precise portrayal of the productive structure in highly developed economies, wherein factors such as human capital, regulatory frameworks, and intangible assets play substantial roles in long-term growth and competitiveness [33].

Several studies have recently examined the linkages between ECI and EFP, with most following the logic of the EKC hypothesis [34,35,36,37,38]. These studies compellingly argue against the assumption that increased ECI correlates with a higher demand for resources and energy to produce new goods and services, which can initially exacerbate pressure on environmental quality. Adopting more environmentally efficient technologies can replace outdated practices as knowledge expands and is effectively integrated. This transition and the transformation and modernization of the productive structure may alleviate environmental pressures. Consequently, high levels of ECI can enable the pursuit of sustainable development pathways [39,40].

Drawing on the premises above and inspired by recent research conducted by [38,41,42,43,44,45], this study endeavors to explore the nexus between ECI and EFP across the five largest European economies (EU-5): Germany, France, Spain, Italy, and the United Kingdom. Hence, it seeks to validate the presence of an N-EKC relation between ECI and EFP, taking into consideration the role of research and development effort (R&D), renewable energy consumption (RNW), and environmental taxation (ETX). Moreover, the work aims to investigate the potential moderating impact of technological innovation on the environment when combined with economic complexity (ECI×R&D) in the abovementioned countries from 1994 to 2020.

Two primary considerations underlie our selection of the economic panel (EU-5), namely, Germany, France, the UK, Italy, and Spain. Firstly, Europe’s unwavering commitment to achieving a climate-neutral economy in the medium term, guided by the European Green Deal [46], makes it an optimal region for assessing the viability of economic and climate policies and instruments in advancing sustainable development pathways. Secondly, these five economies collectively host the highest population and command significant economic influence within Europe. Moreover, they boast high levels of economic development at both the European and global scale, underscoring their ECI. Notably, while these economies are home to around 4% of the world’s population, they are responsible for close to a fifth of the global EFP, with an environmental impact nearing 90% on a European scale, exerting substantial environmental pressure within their respective borders and across the broader European landscape [47,48,49,50]. (See Figure 1).

This study investigates the potential existence of a non-linear relationship, specifically the presence of an N-shaped EKC (N-EKC) when ECI and EFP are linked in the EU-5. In so doing, this research contributes to the needs of policymakers and society at large and seeks to elucidate the rationale behind medium- and long-term sustainable development paths for the entire EU-5 as principal representatives of the European climate issue. In this regard, we posit that the failure to validate the N-EKC between ECI and EFP in the presence of other key climate-related variables, such as technological innovation expenditure (R&D), renewable clean energy consumption, and environmental taxation (ETX), will corroborate, within the outlined constraints, the contribution of ECI and these three primary environmental instruments to mitigating pressure on climate, ecosystems, and biodiversity. Furthermore, this will support the notion that economic development at highly advanced stages, or in other words, with increased ECI, helps establish growth trajectories consistent with the goals of urgently needed climate neutrality and sustainability in the short and medium term.

This research uses the economic complexity index (ECI) to measure economic development in modeling the EKC. This choice is predicated on the ECI’s indicating the level of sophistication of an economy’s productive structure, reflecting the advanced skills and knowledge attained, which, in turn, enable the production of diverse, high-value-added goods. Consequently, the ECI provides a more accurate depiction of the productive structure of highly advanced economies, where factors such as human capital, regulations, and intangible assets significantly influence long-term growth and competitiveness [33]. For instance, in 2020, Germany was a global leader in economic complexity, ranking fourth, while the UK secured the eighth position. Additionally, France, Italy, and Spain were among the top 32 of 133 economies worldwide [50]. (See Figure 1).

This study uses ECI as a gauge of a nation’s developmental progress alongside other significant variables affecting the environment. These variables align with the primary strategic priorities of European climate policy in the EU-5. Within this framework, the study represents decarbonization, energy transition, and the commitment to technological innovation and regulation goals through the variables of RNW, R&D, and ETX, which are crucial in guiding and equitably distributing the costs and benefits of climate endeavors. Using the ECI is key given its association with a modern, dynamic, and technologically advanced productive structure that facilitates the integration and expansion of new processes, innovations, and improved efficiency. Moreover, these economies typically demonstrate openness to adopting measures and regulations for environmental sustainability [52,53,54].

R&D and ETX measures significantly influence the future of productive structure and ecological impact. We consider R&D to be a crucial factor in shaping environmental trajectories in the economies analyzed. New technologies enhance the efficiency and competitiveness of the productive system, thereby reducing pressure on the environment and resources, leading to environmental improvements [55,56]. Additionally, “green” innovations are more ecologically and energetically efficient, directly helping improve ecosystem health [57,58].

ETX is widely recognized as an effective mechanism for formulating sustainable strategies. Its primary objective is to internalize the external costs associated with pollution and waste, enhancing environmental integrity and overall well-being. Furthermore, it encourages the adoption of less polluting and more efficient technologies, while steering investments toward sectors that prioritize environmentally friendly production [59]. Consequently, investments in green technologies and energy-efficient sectors are increasingly attractive to prospective investors, primarily due to the potential for reduced tax liabilities and the availability of tax incentives and credits [60,61,62,63]. The current research incorporates the ETX variable as a proxy for environmental tax pressure, serving as a strategic tool to mitigate ecological and climate-related challenges.

This research also considers the consumption of renewable energy (RNW). Renewable energies and technological and ecological efficiency innovations are crucial in reducing carbon emissions and transitioning to sustainable energy, being essential for achieving sustainability goals and combating global warming. Additionally, given their disruptive nature, they drive technological developments and create job opportunities, making them strategic tools in the fight against climate change [64,65,66].

This study aims to investigate the moderating effect of technology innovation efforts (R&D) on the relationship between the ECI and EFP in the context of the the EU-5. Additionally, we consider the influences of R&D, ETX, and RNW on this nexus.

This work makes a significant contribution to the literature across four key areas. First, we intend to conduct the first empirical analysis utilizing the N-EKC hypothesis. This analysis will illuminate the relationship between ECI and EFP in five major European economies (EU-5). Second, we seek to provide insights into the relevance and applicability of the EKC as a methodological framework for evaluating the environmental impact of economic development, as reflected by ECI. Third, we aim to assess whether expenditures on R&D can effectively moderate the environmental pressures associated with economic complexity in the EU-5. Finally, we will employ robust estimators using the Driscoll-Kraay standard errors technique [67] to estimate the long-term relationships between the variables of interest. We seek to find conclusions pertinent to climate policy within the framework of the EU-5, anticipating these findings will serve as a foundation for future research initiatives beyond the region analyzed.

The following sections of the article are organized as follows: The literature review covers recent relevant studies directly related to the variables and hypotheses examined in this research. Next, we provide detailed descriptions of the data, modeling, and econometric techniques used to achieve the results, which are critically assessed in the subsequent section. Finally, the study concludes by summarizing the key findings, contributions, and policy implications.

2. Literature Review

The use of ECI as a proxy to measure the developmental status of productive structure has received considerable attention in contemporary research. Ref. [68] examined the influence of economic complexity on the EFP in the United States, establishing the validity of an inverted U-shaped relationship consistent with the EKC hypothesis. Ref. [53] investigated the nexus between ECI and EFP in a cohort of emerging economies, identifying a direct correlation between ECI growth and environmental degradation. Ref. [9] also explored the link between ECI and ecological deterioration in Japan. Through Quantile-ARDL modeling, they discerned a bidirectional and asymmetric relationship between both variables in the short and long term. Ref. [34] probed the ECI-EFP relationship in China, finding no support for the EKC hypothesis, positing that economic complexity exacerbates environmental damage in the long term. However, ref. [27] found evidence supporting the inverted U-shaped EKC, emphasizing that ECI can mitigate environmental degradation in G7 countries when accounting for other factors, such as clean and renewable energy innovation and democratic quality. Ref. [69] examined the ECI-EFP relationship in the G-20 economies. They endorsed the inverted U-shaped relationship, reporting that economic complexity alleviates medium- and long-term ecological damage in the 20 largest world economies. Ref. [70] studied the impacts of shocks in ECI, FDI, environmental technology, and renewable energy on carbon emission in the leading clean energy investment countries, finding that ECI and FDI alleviate ecological pressure. Ref. [38] examined the impact of ECI on EFP in G7 countries, finding empirical evidence of an inverted-U relationship between the two variables in G-7 countries. They also reported that human development, high innovation processes, and renewable energy consumption reduce the EFP, showing how the moderating effect of the relationship between human development and innovation processes on economic complexity reduces the global impact on EFP.

Recent research underscores the heterogeneous impact of ECI on the environment across countries and regions, depending on the variables considered and the theoretical and empirical specifications used. Therefore, further research into this relationship, as undertaken in this study, is imperative.

Numerous recent empirical studies have examined the environmental impact of economic activity, employing various EKC models and hypotheses [32,71,72,73,74]. The EKC model posits a quadratic relationship between economic development and its environmental impact, taking the shape of an inverted U-shaped parabola [20]. The classic EKC hypothesis holds that, as economies progress, changes in production methods, demand for goods, and societal priorities lead to alterations in the productive structure, efficiency levels, and environmental pressure. This results in an initial increase in environmental deterioration, which then decreases as economic progress continues, and ultimately falls in line with high levels of development [27,75,76].

Furthermore, studies have explored alternative hypotheses and incorporated several variables considering different shapes (U, N, M, etc.) and the N-EKC. They suggest that, at high levels of economic progress, frictions may arise due to technological obsolescence or regulatory deficiencies, putting stress on the environment [77]. Ref. [78] investigate the impact of financial development, non-renewable energy, renewable energy, and trade openness on EFP from 1990 to 2020 under the new hypothetical imitations of the ECI-induced EKC framework in BRICS-T economies. The authors find support for the existence of an inverted-U EKC relation.

Ref. [79] examined the EKC hypothesis across 74 countries, investigating the relationship between CO₂ emissions and GDP per capita, with findings supporting the N-shaped (N-EKC) relationship in most analyzed economies. Similarly, ref. [80] studied the link between emissions and GDP per capita in seven emerging economies, revealing support for the traditional inverted-U quadratic form rather than the N-shaped EKC. Ref. [81] delved into the validity of the N-EKC in the sub-Saharan region, factoring in GDP per capita, renewable and non-renewable energies, and their combined impact on environmental quality, evidencing this relationship in both the short- and long-term. Ref. [77] inspected the feasibility of nuclear energy mitigating environmental deterioration using a panel of major nuclear powers, finding support for the N-EKC relationship. These varied outcomes across different studies suggest that the N-EKC is a credible model with empirical backing, which, however, warrants further investigation. Ref. [82] explored the link between natural resources, economic complexity, and sustainable development in European Union (EU) member states using the N-EKC framework, finding evidence of the presence of the N-EKC relationship in the countries analyzed. The present study seeks to address this knowledge gap by examining the potential existence of the N-EKC in the EU-5.

The intersection of technological and energy innovation with renewable energy is widely acknowledged to have an environmental impact in conjunction with economic advancement. This consensus is drawn from a comprehensive review of relevant literature. For instance, ref. [83] leveraged FM-DOLS estimators to validate the positive influence of renewable energies on environmental amelioration in Indonesia. Likewise, ref. [84] revealed a positive correlation between reduced EFP and intensified use of renewable energies in Turkey. Ref. [85] employed sophisticated co-integration techniques for panel data, confirming the efficacy of technological innovations in mitigating ecological deterioration.

Meanwhile, ref. [86] employed a stochastic model (STIRPAT model) and standard econometric techniques to evaluate the relationship between renewable energy, technological innovation, and other variables in the Asia Pacific Economic Cooperation (APEC) countries. Their findings revealed contrasting long-term outcomes, with renewable energy showcasing positive environmental impacts, while technological innovation generated no similar effects. The present study aims to incorporate the concomitant influence of technological innovation and renewable energy consumption in investigating the nexus between environmental conditions and EFP under the N-EKC hypothesis.

The efficacy of ETX in enhancing environmental conditions has not been fully determined, as evidenced by [87] comprehensive review of recent theoretical and empirical literature on the subject, with further research thus being warranted. For example, ref. [88] conducted a study harnessing data from 29 OECD economies. They observed that ETX policies exhibit can potentially mitigate ecological strain in the long term, albeit with less distinct effects in the short term. Similarly, ref. [52] utilized various estimators to show that ETX plays a role in diminishing ecological degradation in G-10 world economies. Furthermore, ref. [89] investigated the interplay between ETX, RNW, and EFP in European countries, with their findings emphatically supporting the implementation of environmental taxes in conjunction with renewable energy to enhance environmental quality. Ref. [60] investigated the causal relationships between environmental taxes and environment-related technological innovation, concluding that environmental taxes stimulate technological innovation in high- and middle-income nations.

The findings from previous studies suggest several noteworthy conclusions. Firstly, it is imperative to further scrutinize the validity and significance of the N-EKC hypothesis, specifically in exploring the ecological impact in highly progressive and developed economies, exemplified by the five largest European economies examined herein. Our study evaluates the N-EKC within the interface of ECI and EFP for the EU-5. Furthermore, technological innovation, the adoption of renewable energy, and environmental regulations consistently demonstrate a favorable influence on sustainable development, regardless of the empirical methodologies employed or the geographic areas analyzed. As such, this research assigns significant importance to these three variables, alongside ECI, in their collective impact on environmental degradation in the EU-5.

The gaps detected in our review of the literature lead us to propose the following hypotheses to be validated in the next phase of the study:

H₁: 

N-shaped EKC relationship between ECI and EFP in EU-5.

H₂: 

R&D decreases EFP in EU-5.

H₃: 

RNW decreases EFP in EU-5.

H₄: 

ETX decreases EFP in EU-5.

H₅: 

ECI×R&D decreases EFP in EU-5.

3. Materials and Methods

The objective of this research is to examine the environmental impact of ECI on EFP, with a focus on the N-EKC hypothesis. Additionally, this study incorporates the combined influences of R&D expenditure, RNW consumption, ETX, and the moderating effect of R&D expenditure on the EFP-ECI relationship. The analysis centers on the five largest European economies (EU-5) from 1994 to 2020.

Following the specifications established by [22,34,68,77,90], the foundational model that consolidates and organizes the relationships between the variables of interest is as follows:

$${\mathrm{E}\mathrm{F}\mathrm{P}}_{\mathrm{i}\mathrm{t}}=\mathrm{f}\left({\mathrm{E}\mathrm{C}\mathrm{I}}_{\mathrm{i}\mathrm{t}},{\mathrm{E}\mathrm{C}\mathrm{I}}_{\mathrm{i}\mathrm{t}}^2,{\mathrm{E}\mathrm{C}\mathrm{I}}_{\mathrm{i}\mathrm{t}}^3,{\mathrm{R}\&\mathrm{D}}_{\mathrm{i}\mathrm{t}-1},{{\mathrm{R}\mathrm{N}\mathrm{W}}_{\mathrm{i}\mathrm{t}},\mathrm{E}\mathrm{T}\mathrm{X}}_{\mathrm{i}\mathrm{t}}, {{\mathrm{E}\mathrm{C}\mathrm{I}}_{\mathrm{i}\mathrm{t}}\times \mathrm{R}\&\mathrm{D}}_{\mathrm{i}\mathrm{t}-1}\right)$$

(1)

The general specification presented in Equation (1) underlies the empirical model to be tested in this study. To mitigate potential issues of heteroscedasticity, the variables included in Equation (1) were recalculated using the natural logarithm, except the ECI variable, which is represented as an index number. Consequently, the empirical model to be econometrically estimated is articulated as follows:

$$\begin{gathered} \mathrm{l}\mathrm{n}{\mathrm{E}\mathrm{F}\mathrm{P}}_{\mathrm{i}\mathrm{t}}={\mathrm{\vartheta }}_1{\mathrm{E}\mathrm{C}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\vartheta }}_2{\mathrm{E}\mathrm{C}\mathrm{I}}_{\mathrm{i}\mathrm{t}}^2+{\mathrm{\vartheta }}_3{\mathrm{E}\mathrm{C}\mathrm{I}}_{\mathrm{i}\mathrm{t}}^3 \\ +{\mathrm{\vartheta }}_4\mathrm{l}\mathrm{n}{\mathrm{R}\&\mathrm{D}}_{\mathrm{i}\mathrm{t}-1}+{\mathrm{\vartheta }}_5\mathrm{l}\mathrm{n}{\mathrm{R}\mathrm{N}\mathrm{W}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\vartheta }}_6\mathrm{l}\mathrm{n}{\mathrm{E}\mathrm{T}\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\vartheta }}_7\mathrm{l}\mathrm{n}\left({{\mathrm{E}\mathrm{C}\mathrm{I}}_{\mathrm{i}\mathrm{t}}\times \mathrm{R}\&\mathrm{D}}_{\mathrm{i}\mathrm{t}-1}\right)+{\mathrm{\epsilon }}_{\mathrm{i}\mathrm{t}} \end{gathered}$$

(2)

The present analysis uses LnEFP to represent the EFP variable, which assesses the environmental degradation associated with the ECI. In the equation formulated for the model, ECI, ECI², and ECI³ aim to substantiate the presence of an N-shaped EKC-like relationship regarding EFP in the panel under analysis (see Figure 2).

Moreover, in Equation (2), LnR&D captures the efforts invested in technological innovation, quantified through R&D expenditure relative to GDP. LnRNW denotes energy consumption derived from clean and renewable sources, while LnETX indicates an environmental tax. The penultimate term presented on the right-hand side of the equation, Ln(ECI×R&D), illustrates the interaction between ECI and the commitment to technological innovation efforts. This term represents the moderating impact of greater ECI, together with R&D expenditure, on environmental outcomes. This term describes how the interplay between greater ECI and increased investment in R&D influences environmental outcomes. Specifically, the moderating effect of R&D expenditure would explain why increases in the level of ECI in the economies analyzed do not lead to greater pressure and environmental deterioration, thereby facilitating sustainable development.

Graphically represent this in the rightward shift of the N-EKC curve (see Figure 3).

In Equation (2), the parameters ${\mathrm{\vartheta }}_1$ to ${\mathrm{\vartheta }}_7$ reflect the long-term elasticities between the model’s dependent variable (LnEFP) and the various explanatory variables incorporated into the analysis. Furthermore, ε signifies the error term. Subscript i denotes the cross-section within the data panel on the economies analyzed. In contrast, subscript t indicates the time dimension, covering the period from 1994 to 2020. This specific time frame was selected for two main reasons: the availability of data on certain variables, and to enable the inclusion of the United Kingdom in the analysis, which, in turn allowed us to form the group of the five largest economies in Europe (EU-5). The UK formally exited the European Union (EU) in January 2019, with a de facto transition occurring between 2019 and 2020. Therefore, 2020 was the final year in which the UK operated under EU policies, particularly as regards community energy and environmental policies, which are relevant to this research.

Table 1. Data sources (period of research from 1994 to 2020).

Variables	Symbol	Unit of Measurement	Source [Period of Research: 1994–2020]
Ecological Footprint	EFP	Global hectares per person (gha)	GFN a
Economic Complexity Index	ECI	Index	The Atlas of Economic Complexity b
Environmental taxation	ETX	% Total tax revenue	OECD c
Research & Development expenditure	R&D	% Total GDP	OECD d
Renewable energy consumption	RNW	% Total final energy consumption	World Bank e

Note: own preparation, ^a [49], ^b [91], ^c [92], ^d [93], ^e [94].

delineates the variables used in the model, detailing their acronyms, units of measurement, and the sources of the data.

The panel selected for this research includes five major European economies: France, Germany, Italy, Spain, and the UK. The study covers the period from 1994 to 2020, encompassing 700 records. These nations rank among the most developed and advanced worldwide; notably, all the countries, except Spain, are members of the G7, seven of the world’s most powerful economic powers. In 2020, the final year examined in this study, these five economies collectively accounted for approximately 73% of Europe’s total population, nearly 80% of its Gross Domestic Product (GDP), and almost 20% of global GDP. Notably, while this group comprises only 4% of the world’s population, it is responsible for close to a fifth of the global EFP, with an environmental impact nearing 90% on a European scale. Germany is a leader in economic sophistication, ranking fourth worldwide in the 2020 ECI. Additionally, France, Italy, and the UK appear among the top twenty countries in terms of economic sophistication. In comparison, Spain is ranked 32nd of the 133 countries assessed. These insights highlight the significance of these economies in analyzing the relationship between the level of sophistication of their productive structures and their environmental impact, which is the focus of this research (see

Table 2. Indicators of The Five Largest European Economies (2020).

	Population			GDP			Economic Complexity Index b		Ecological Footprint a
Country	(million)	% Europe	% World	(billion $US)	% Europe	% World	(ECI)	Ranking position	(gha pc)	(million gha)	% Europe	% World
France	67.10	15.00%	0.86%	2.78	16.50%	3.30%	1.24	15	4.00	268.40	16.50%	3.30%
Germany	83.17	18.60%	1.07%	4.08	24.20%	4.80%	1.82	4	5.30	440.80	27.00%	5.70%
Italy	60.25	13.50%	0.78%	2.05	12.20%	2.40%	1.45	12	4.70	283.20	17.00%	3.70%
Spain	47.33	10.60%	0.61%	1.40	8.30%	1.60%	1.23	32	3.80	179.90	11.00%	2.30%
UK	67.03	15.00%	0.86%	3.09	18.30%	3.70%	1.78	13	4.60	308.30	18.30%	4.00%
Top-5	324.87	72.70%	4.18%	13.40	79.50%	15.80%				1480.60	89.80%	19.00%
Europe	447.71	100.00%	5.75%	168.50	100.00%	19.80%			4.50	2020.60	100.00%	25.00%
World	7794.80		100.00%	852.40		100.00%			2.80	21,825.40		100.00%

Note: own preparation, data obtained from ^a [49], ^b [50].

).

Econometric Techniques

This research utilizes econometric modeling that incorporates panel data. However, undetected cross-sectional dependence and other panel data issues may result in unreliable and biased estimates. To address these concerns, it is crucial to conduct a series of tests before undertaking econometric estimations based on this type of data. This proactive approach enhances the reliability and robustness of the findings (Figure 4) illustrates the methodological framework implemented in our econometric analysis.

Calculating the Variance Inflation Factors (VIF) before estimating econometric models is critical to identify and quantify increases in parameter variance stemming from multicollinearity among predictor variables [95]. Specifically, VIF measures how much the variance of an estimate for a regressor, denoted by ${\mathrm{\vartheta }}_{\mathrm{j}}$, is “inflated” due to the interaction between variable ${\mathrm{k}}_{\mathrm{j}}$ and the other predictors ${\mathrm{k}}_{\mathrm{i}\ne \mathrm{j}}$ in the model. Additionally, tolerance is defined as the proportion of variance in a predictor that is unexplained by the other predictors in the model.

$$\mathrm{V}\mathrm{I}\mathrm{F}={\displaystyle \frac{1}{1-{\mathrm{R}}_{\mathrm{k}}^2}}$$

(3)

$$\mathrm{T}\mathrm{o}\mathrm{l}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}={\displaystyle \frac{1}{\mathrm{V}\mathrm{I}\mathrm{F}}}=1-{\mathrm{R}}_{\mathrm{k}}^2$$

(4)

Running a unit root test is crucial for accurate and reliable estimations with panel data. This step is key because estimating non-stationary variables may generate inefficiencies unless co-integration exists within the panel data. This study uses the second-generation Dickey-Fuller augmented cross-section test (CADF) developed by [96]. This method effectively accounts for autoregressive structures and heterogeneity within the dataset. Our null hypothesis assumes the presence of non-stationarity in the data.

Figure 4. Flowchart of the econometric modeling approach. Own preparation, refs. [96,97].

Figure 4. Flowchart of the econometric modeling approach. Own preparation, refs. [96,97].

$${∆\mathrm{C}\mathrm{A}}_{\mathrm{i}\mathrm{t}}={\phi }_{\mathrm{i}\mathrm{t}}+{\phi }_{\mathrm{i}\mathrm{t}}{\mathrm{Z}}_{\mathrm{i}\mathrm{t}-1}+{\phi }_{\mathrm{i}\mathrm{t}}{\overline{\mathrm{C}\mathrm{A}}}_{\mathrm{i}\mathrm{t}-1}+\sum _{\mathrm{l}=0}^{\mathrm{p}}{\phi }_{\mathrm{i}\mathrm{l}}∆{\overline{\mathrm{C}\mathrm{A}}}_{\mathrm{t}-\mathrm{l}}+\sum _{\mathrm{l}=1}^{\mathrm{p}}{\phi }_{\mathrm{i}\mathrm{l}}∆{\mathrm{C}\mathrm{A}}_{\mathrm{i}\mathrm{t}-\mathrm{l}}+{\mathrm{\mu }}_{\mathrm{i}\mathrm{t}}$$

(5)

where ${\overline{\mathrm{C}\mathrm{A}}}_{\mathrm{i}\mathrm{t}-1}$ and $∆{\overline{\mathrm{C}\mathrm{A}}}_{\mathrm{t}-\mathrm{l}}$ are the cross-section averages.

This study implements the [97] co-integration test to investigate the long-term co-integration relationships among the variables under examination, which include ECI, ECI², ECI³, LnR&D, LnRNW, LnETX, Ln(ECI×R&D), and LnEFP. The general specification of this test is outlined as follows:

$${\mathrm{G}}_{\mathrm{t}}={\displaystyle \frac{1}{\mathrm{N}}}\sum _{\mathrm{i}-1}^{\mathrm{N}}{\displaystyle \frac{{\mathrm{\alpha }}_{\mathrm{i}}}{\mathrm{S}\mathrm{E}\left({\mathrm{\alpha }}_{\mathrm{i}}\right)}}$$

(6)

$${\mathrm{G}}_{\mathrm{a}}={\displaystyle \frac{1}{\mathrm{N}}}\sum _{\mathrm{i}-1}^{\mathrm{N}}{\displaystyle \frac{{\mathrm{T}\mathrm{\alpha }}_{\mathrm{i}}}{{\mathrm{\alpha }}_{\mathrm{i}}\left(1\right)}}$$

(7)

$${\mathrm{P}}_{\mathrm{t}}={\displaystyle \frac{\mathrm{\alpha }}{\mathrm{S}\mathrm{E}\left(\mathrm{\alpha }\right)}}$$

(8)

$${\mathrm{P}}_{\mathrm{\alpha }}=\mathrm{T}\mathrm{\alpha }$$

(9)

The group mean statistics are calculated in Equations (6) and (7), and the panel statistics are calculated in Equations (8) and (9).

The Driscoll-Kraay standard errors estimation (DKSE) [67] is an econometric technique specifically designed to address autocorrelation, heteroscedasticity, and cross-sectional dependence in panel data [98,99,100]. It is particularly appropriate for panel data sets whose cross-sectional dimension is larger than the temporal dimension.

Following [101], to estimate the standard error matrix using the Driscoll-Kraay technique, a linear model is needed:

$${\mathrm{y}}_{\mathrm{i}\mathrm{t}}={\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\prime }\mathrm{\theta }+{\mathrm{\epsilon }}_{\mathrm{i}\mathrm{t}},\hspace{1em}\hspace{1em}\hspace{1em}\mathrm{i}=1,\dots ,\mathrm{N},\hspace{1em}\hspace{1em}\hspace{1em}\mathrm{t}=1,\dots ,\mathrm{T}$$

(10)

where y_it is a scalar type, x_it is a (K + 1) × 1 vector of independent variables, with the first element being 1, and θ being a (K + 1) × 1 vector of unknown coefficients. Subscript i refers to the units in the cross-section, while t indicates the time. The set of observations is grouped as follows:

$${y=\left[y_{{1t}_{11}}\dots y_{{1T}_1}{y}_{{2t}_{21}}\dots y_{{NT}_N}\right]}^{\prime }{ and X=\left[x_{{1t}_{11}}\dots x_{{1T}_1}{x}_{{2t}_{21}}\dots x_{{NT}_N}\right]}^{\prime }$$

(11)

where unbalanced panel data are allowed, since, for individual i, there is only one subset t_i1, …, T_i, with 1 ≤ t_i1 ≤ T_i ≤ T among all possible T observations. Furthermore, regressors x_it are assumed to be uncorrelated as regards the scalar disturbance ε_is for any s, t (implying substantial homogeneity). However, autocorrelation, heteroscedasticity, and cross-sectional dependence in disturbances ε_is are allowed. Under these assumptions, θ can be consistently estimated by ordinary least squares (OLS), resulting in:

$$\widehat{\theta }={\left(X^{\prime }X\right)}^{-1}{X}^{\prime }y$$

(12)

The standard errors of the coefficients, estimated using the Driscoll-Kraay method, are calculated as the square root of the diagonal elements from the asymptotic and robust covariance matrix.

$$V\left(\widehat{\theta }\right)={\left(X^{\prime }X\right)}^{-1}{\widehat{S}}_T{\left(X^{\prime }X\right)}^{-1}$$

(13)

where ${\widehat{S}}_T$ is defined, as in [102]:

$${\widehat{\mathrm{S}}}_{\mathrm{T}}={\widehat{\mathrm{\Omega }}}_0+\sum _{\mathrm{j}=1}^{\mathrm{m}\left(\mathrm{T}\right)}\mathrm{w}\left(\mathrm{j},\mathrm{m}\right)\left[{\widehat{\mathrm{\Omega }}}_{\mathrm{j}}+{\widehat{\mathrm{\Omega }}}_{\mathrm{j}}^{\prime }\right]$$

(14)

where m(T) refers to the interval for which the residuals may be auto-correlated, and the modified Barlett weights are specified as follows:

$$\mathrm{w}\left(\mathrm{j},\mathrm{m}\right)=1-{\displaystyle \frac{\mathrm{j}}{\left\{\mathrm{m}\left(\mathrm{T}\right)+1\right\}}}$$

(15)

The method ensures that matrix $\widehat{\mathrm{S}}$ remains positive semi-definite while refining the sample autocovariance function. This reduces the emphasis on higher-order lags. Hence, matrix ${\widehat{\mathrm{\Omega }}}_{\mathrm{j}}$, with dimensions of (K + 1) × (K + 1), is defined as follows:

$${\widehat{\mathrm{\Omega }}}_{\mathrm{j}}=\sum _{\mathrm{t}=\mathrm{j}+1}^{\mathrm{T}}{\mathrm{h}}_{\mathrm{t}}\left(\widehat{\mathrm{\theta }}\right){\mathrm{h}}_{\mathrm{t}-\mathrm{j}}{\left(\widehat{\mathrm{\theta }}\right)}^{\prime }\mathrm{w}\mathrm{i}\mathrm{t}\mathrm{h} {\mathrm{h}}_{\mathrm{t}}\left(\widehat{\mathrm{\theta }}\right)=\sum _{\mathrm{i}=1}^{\mathrm{N}\left(\mathrm{t}\right)}{\mathrm{h}}_{\mathrm{i}\mathrm{t}}\left(\widehat{\mathrm{\theta }}\right)$$

(16)

In Equation (16), the summation of the individual ${\mathrm{h}}_{\mathrm{i}\mathrm{t}}\left(\widehat{\mathrm{\theta }}\right)$ conditions at time t is defined to range from 1 to 1 a N(t), where N is allowed to change according to t. This small adjustment to the original Driscoll-Kraay estimator modifies it for effective use with unbalanced panels. For pooled OLS estimation, the orthogonality conditions associated with each ${\mathrm{h}}_{\mathrm{i}\mathrm{t}}\left(\widehat{\mathrm{\theta }}\right)$ in Equation (16) are detailed as follows:

$${\mathrm{h}}_{\mathrm{i}\mathrm{t}}\left(\widehat{\mathrm{\theta }}\right)={\mathrm{x}}_{\mathrm{i}\mathrm{t}}{\widehat{\mathrm{\epsilon }}}_{\mathrm{i}\mathrm{t}}={\mathrm{x}}_{\mathrm{i}\mathrm{t}}\left({\mathrm{y}}_{\mathrm{i}\mathrm{t}}-{\mathrm{x}}_{\mathrm{i}\mathrm{t}}^{\prime }\widehat{\mathrm{\theta }}\right)$$

(17)

The Driscoll-Kraay covariance matrix estimator is effectively analogous to the heteroscedasticity- and autocorrelation-consistent covariance matrix estimator proposed by [102], especially when applied to time series data using cross-sectional averages, as defined in Equation (17).

This estimator, based on cross-sectional averages, provides consistent standard errors, regardless of the panel’s cross-sectional dimension. Using this method, the consistency has been shown to hold even as N → ∞ approaches infinity [67]. Moreover, using this method for the covariance matrix estimation yields standard errors that are robust to various forms of cross-sectional and temporal dependence.

To validate and enhance the precision of the DKSE estimates for the parameters analyzed in this study, a robustness test is conducted using the Panel-Corrected Standard Errors (PCSE) technique, proposed by [103]. This non-parametric approach yields standard error estimators that are weighted according to the spatial and temporal dimensions of the data panel. Consequently, this methodology provides more accurate estimates [104]. It effectively mitigates biases associated with issues such as autocorrelation, cross-sectional dependence, and heteroscedasticity, which can lead to inefficient and biased estimates [67].

The PCSE method begins by transforming the data to eliminate serial correlation, and subsequently applies Ordinary Least Squares (OLS) to this transformed dataset. Finally, to improve the efficiency of the parameter estimates, the standard errors are exhaustively examined and corrected in the case of autocorrelation, cross-sectional dependence, or heteroscedasticity [105]. This analysis commences with the following linear relationship:

$${\mathrm{y}}_{\mathrm{i}}={\mathrm{i}}_{\mathrm{N}\mathrm{T}}{\mathrm{\beta }}_0+{\mathrm{X}}_{\mathrm{i}}{\mathrm{\beta }}_{\mathrm{x}}+{\mathrm{\epsilon }}_{\mathrm{i}}$$

(18)

where N and T represent, respectively, the cross-section and the period series, ${\mathrm{y}}_{\mathrm{i}}$ is a T × 1 vector of observations of the dependent variable for the i-unit of cross-section; ${\mathrm{i}}_{\mathrm{N}\mathrm{T}}$ is a T × 1 vector of ones; ${\mathrm{X}}_{\mathrm{i}}$ is a T × 1 vector of observations of the explanatory variable; ${\mathrm{\beta }}_0$ and ${\mathrm{\beta }}_{\mathrm{x}}$ are scalars, and ${\mathrm{\epsilon }}_{\mathrm{i}}$ is a T × 1 vector of error terms, where ε~N(0, Ω_NT). The error structure Ω_NT is based on the Parks model [106] and assumes (i) heteroscedasticity, (ii) first-order serial correlation, and (iii) time-invariant cross-section correlation.

Consequently, the PCSE robustness test validates the results obtained in the DKSE test, improves the reliability of the study’s main findings, and bolsters conclusions of the research.

4. Results

In our empirical analysis (see Equation (2)), with the exception of the ECI, data for all the variables were logarithmically transformed.

Table 3. Descriptive statistics.

	LnEFP	ECI	LnR&D	LnRNW	LnETX
Mean	1.622	1.533	0.487	1.994	1.854
Median	1.665	1.501	0.480	2.224	1.855
Maximum	1.812	2.335	1.153	2.913	2.230
Minimum	1.242	0.700	−0.260	−0.163	1.459
Std. Dev.	0.129	0.401	0.365	0.819	0.184
Asymmetry	−0.782	0.073	−0.147	−1.270	0.027
Kurtosis	−0.259	−0.618	−0.932	0.779	−1.089
Prob.	0.022 *	0.068 **	0.062 **	0.139	0.031 *

Note: * & ** represent 5% and 10% level of significance respectively.

summarizes the key descriptive statistics derived from this data set.

The descriptive statistics analysis reveals that the average logarithmic value of the EFP (LnEFP) is 1.622, with a maximum value of 1.812 and a minimum of 1.242, resulting in a standard deviation of 0.129. These metrics suggest that the environmental degradation in the countries analyzed followed a similar trend over time. This observation is further corroborated by the average (1.622) proximity to distribution of the median (1.665) of this variable. Moreover, the distribution of LnEFP shows a slight negative skewness, indicating that many of these economies made progress in reducing their environmental impact over time. This improvement may be attributed to the ecological commitments and national and European environmental policies implemented.

The statistics of the variable measuring the level of sophistication of the productive structure (ECI) reveal significant variation across the economies. This is reflected in the wide range between the minimum value (0.7) and the maximum value (2.335), as well as the relatively high standard deviation (0.401). While all these economies are classified as highly sophisticated compared to global standards, Germany consistently maintained its position as one of the world’s leading economies across the period under review. It remained considerably ahead of the UK, France, and Italy and is particularly distanced from Spain, which is positioned significantly below the intermediate group of economies. Moreover, the positive value of the asymmetry statistic (0.073) indicates that most of the economies studied slightly enhanced their level of sophistication over time, likely attributable to ongoing investments and the incorporation of new technologies into their productive frameworks.

An analysis of the logarithmic statistics related to technological innovation efforts (LnR&D) and the consumption of energy from renewable sources (LnRNW) provides valuable insights. The variable LnRNW serves as an indicator of progress in energy transition processes. The results indicate considerable heterogeneity among the economies under study, particularly concerning RNW. This suggests that some economies significantly rely on non-renewable energy sources while others made substantial advances in transitioning from fossil fuels to clean energy alternatives.

Regarding R&D expenditure, the maximum logarithmic value recorded is 1.153, with a minimum of −0.260. The accompanying asymmetry statistic is −0.147, and the kurtosis is −0.932. These figures imply that while certain countries increased their investments in innovation, others experienced a decline over time. If this trend is substantiated, it has significant implications, as adopting new technologies typically enhances energy efficiency and environmental sustainability once integrated into productive practices.

Moreover, the analysis of environmental taxation (LnETX) shows that the economies collectively embraced fiscal instruments to mitigate the negative externalities associated with pollution. The ETX policies aim to boost productive and technological investments by offering incentives for environmentally sustainable practices, such as tax breaks for businesses pursuing green technologies. Meanwhile, they impose tax penalties on industries that significantly pollute, creating a framework that rewards eco-friendly behavior while discouraging harmful activities. These instruments seek to promote both economic development and environmental quality. The maximum value for LnETX is 2.230, with a minimum of 1.459—both positive indicators. The statistics reflect a nearly equal average and median, a low standard deviation of 0.184, an asymmetry approaching 0, and a significant negative kurtosis of −1.089. These findings suggest a trend toward harmonizing and intensifying ETX policies among the economies analyzed.

Furthermore,

Table 4. Correlation matrix.

	LnEFP	ECI	LnR&D	LnRNW	LnETX
LnEFP	1
ECI	0.505 **	1
LnR&D	0.087	0.632 **	1
LnRNW	−0.623 **	−0.517 **	0.066	1
LnETX	0.146	0.195 **	−0.423 **	−0.549 **	1

Note: significance levels, ** p < 0.01.

shows the correlation matrix derived from the data set used in our empirical analysis. The analysis reveals a significant relationship between the ECI and the logarithm of renewable energy consumption (LnRNW), with the logarithm of ecological footprint (LnEFP) as the dependent variable. A stronger ECI correlates with a larger EFP, suggesting complex interactions between economic development and environmental outcomes. In contrast, most of the other explanatory variables show weak correlations with LnEFP, except for a notable negative correlation between LnEFP and clean energy consumption (LnRNW). This suggests that increased renewable energy consumption is linked to a reduced EFP, highlighting the environmental benefits of cleaner energy sources. Furthermore, only the stronger correlations—specifically between ECI and LnEFP and between LnRNW and LnEFP—are statistically significant, suggesting these relationships deserve further investigation to understand their impacts on sustainability efforts.

The variance inflation factor (VIF) is calculated to assess the potential presence of multicollinearity among the explanatory variables in the empirical modeling.

Table 5. VIF analysis.

Variable	VIF	1/VIF
ECI	4.01	0.249492
LnR&D	3.76	0.265710
LnETX	2.37	0.421888
LnRNW	2.03	0.492608
Mean VIF	3.04

summarizes the findings. The results show that the VIF for all the variables is below 5, suggesting there are no multicollinearity in the modeling process.

Table 6. Ref. [96] CADF unit root test.

Variable	At I(0)		At I(1)
	t-Bar	p-Value	t-Bar	p-Value
LnEFP	−2.194	0.159	−3.475 *	0.000
ECI	−2.475 ***	0.050	−3.004 *	0.002
LnR&D	−1.674	0.579	−3.025 *	0.002
LnRNW	−0.947	0.970	−3.594 *	0.000
LnETX	−1.639	0.610	−2.489 **	0.047

Note: significance levels, * p < 0.01. ** p < 0.05. *** p < 0.1.

shows the results of the unit root test conducted. Specifically, the second generation CADF test developed by [96] was used to analyze each variable. The results indicate that, at level (I(0)), only the ECI variable is stationary. In contrast, all the other variables assessed were stationary at the first difference (I(1)). The results of this analysis clearly evidence that no statistical issues compromise the reliability and accuracy of the estimates obtained from the empirical models. This highlights the robustness of the findings, indicating that the methodologies employed are sound and produced reliable estimates.

Finally, the [97] co-integration test was conducted to assess the presence of co-integration among the variables in the empirical model. The null hypothesis asserts that no long-term relationship exists between the dependent and explanatory variables. The findings summarized in

Table 7. Ref. [97] test for co-integration.

Stats	Value	Z Value	p Value	Robust p Value
Gt	−1.787	1.541	0.938	0.000 *
Ga	−3.631	2.721	0.997	0.000 *
Pt	−3.166	1.512	0.935	0.000 *
Pa	−2.131	2.116	0.983	0.000 *

Note: significance level, * p < 0.01.

indicate the presence of co-integration among the analyzed variables. Due to the substantial significance of the Gt and Pt statistics, we can confidently reject the null hypothesis. This outcome supports the alternative hypothesis, confirming a long-term relationship between the variables in the model.

This study uses the Driscoll-Kraay Standard Errors (DKSE) approximation [67] to estimate the long-run coefficients of the empirical model under analysis. Specifically, it examines whether, during the period under study, the impact on the environment—measured by ecological footprint (LnEFP)—attributable to the level of complexity and sophistication of the productive structures of the five principal European economies (EU-5) can be elucidated using the N-EKC. Furthermore, this analysis considers the collective effects of several critical factors, including the commitment to technological innovation relative to GDP (LnR&D), advances in the energy transition process as indicated by the consumption of renewable energy (LnRNW), environmental taxation (LnETX), and the moderating effect of investments in R&D in conjunction with economic complexity (LnECI×R&D) on the EFP. The study examines whether ECI significantly influences the environment (LnEFP) using the N-shaped EKC hypothesis, while accounting for these supplementary variables. To ensure the robustness of the estimators derived from the Driscoll-Kraay Standard Errors regression, an additional test using Panel-Corrected Standard Errors (PCSE), developed by [103], was conducted. The findings of the DKSE and PCSE methodologies are detailed in

Table 8. Long-run estimations with DKSE.

Variable	Driscoll−Kraay Standard Errors (DKSE) Estimation
	Coef.	Std. Err.	t	P > |t|	[95% Conf. Interval]
ECI	0.0965003 **	0.0385351	2.50	0.020	0.0165834	0.1764172
ECI2	−0.0515496 **	0.020257	−2.54	0.018	−0.09356	−0.0095391
ECI3	0.0135448 **	0.0050371	2.69	0.013	0.0030985	0.0239912
LnR&D	−0.3906719 *	0.1185483	−3.30	0.003	−0.636526	−0.1448178
LnRNW	−0.0935813 *	0.0253206	−3.70	0.001	−0.1460931	−0.0410695
LnETX	−0.1450468	0.0903697	−1.61	0.123	−0.3324622	0.0423686
Ln(ECI×R&D)	−0.0230644 **	0.0095957	−2.40	0.025	−0.0429647	−0.003164
Cons,	−0.0572402 ***	0.0303445	−1.89	0.073	−0.1201708	0.0056904

Note: significance levels, * p < 0.01. ** p < 0.05. *** p < 0.1.

and

Table 9. Robustness check with PCSE.

Variable	Panel Corrected Standard Errors (PCSE) Estimation
	Coef.	Std. Err.	z	P > |z|	[95% Conf. Interval]
ECI	0.0938349 **	0.047174	1.99	0.047	0.0013755	0.1862943
ECI2	−0.0463693 **	0.0232777	−1.99	0.046	−0.0919927	−0.000746
ECI3	0.0119816 **	0.0057471	2.08	0.037	0.0007175	0.0232458
LnR&D	−0.4027006 *	0.1323677	−3.04	0.002	−0.6621366	−0.1432646
LnRNW	−0.0945518 **	0.0419288	−2.26	0.024	−0.1767307	−0.012373
LnETX	−0.1411995 ***	0.0813229	−1.74	0.083	−0.3005895	0.0181906
Ln(ECI×R&D)	−0.0241805 ***	0.0142114	−1.70	0.089	−0.0520342	0.0036733
Cons.	−0.0584177	0.0381156	−1.53	0.125	−0.133123	0.0162876

Note: significance levels, * p < 0.01. ** p < 0.05. *** p < 0.1.

respectively.

The regression analysis utilizing the DKSE approximation indicates that all estimated coefficients for the variables in the model, except the logarithm of environmental taxation (LnETX), are statistically significant. Notably, the coefficients associated with the three levels of economic complexity—ECI, ECI², and ECI³—achieved significance at the 0.05% level. Furthermore, the coefficient representing the moderating effect, denoted by Ln(ECI×R&D), demonstrates the same significance level. Additionally, the coefficients for both LnR&D and LnETX are significant at the 0.01% level, although the coefficient for LnETX alone is not statistically significant. In the subsequent PCSE robustness test, the estimation of each coefficient for all the model variables, including LnETX, was found to be substantial, with LnETX achieving a significance level of 0.1%. The coefficients for the three levels of ECI were significant at the 0.05% level, while the coefficient for the logarithm of investment in R&D (LnR&D) was significant at the 0.01% level. On the other hand, the comprehensive results of the DKSE and PCSE tests show that all the coefficients estimated for the variables in this analysis are statistically significant.

The magnitudes of the coefficients estimated are notably similar across both regression analyses. In the estimation using the DKSE approach, the coefficients corresponding to the three levels of the ECI variable (ECI, ECI², and ECI³) show slightly higher values. In contrast, the PCSC approach estimation reveals that the magnitude of the coefficients for LnR&D, LnRNW, and the moderating effect (Ln(ECI×R&D)) exceed those for the DKSE approach. Furthermore, the coefficient for LnETX is greater than those of the DKSE results, although it attains statistical significance solely under the PCSE approach. Additionally, the regression model employing the DKSE technique and the robustness test using PCSE demonstrate consistent signs for all the coefficients estimated. All the findings align with our proposed empirical hypothesis (Figure 5) summarizes the key findings of this research.

5. Discussion

The results of estimating the econometric model (see Table 8 and Table 9) show that the ECI notably impacts the logarithm of EFP (LnEFP). Based on the estimation results using the DKSE methodology, a 1% increase in ECI is associated with a corresponding 0.0965003% rise in LnEFP. That is, when the ECI for the economies under study rises by one unit, the associated environmental degradation, as quantified by the carbon footprint, increases by 0.0965003%, assuming all other factors remain constant. These findings suggest that during the early stages of sophistication in these European countries’ productive structures, advances in modernization and the complexity of the knowledge-based economic framework may adversely affect the environment, increasing carbon emissions. This finding aligns with those of previous research, including the studies by [22] which focused on the EKC nexus for 48 highly developed economies, and by [107] regarding China. It is further corroborated in the works by [38,108], analyzing ECI and EKC in G7 countries. It contradicts, however, the conclusions of other studies, such as that by [109], which suggests that economic complexity can alleviate environmental deterioration in OECD countries, and the study by [27], which finds that ECI helps reduce the carbon footprint in G7 countries.

The second finding of this research highlights that ECI² plays a significant role in alleviating environmental pressure in the economies analyzed. According to model estimates derived from the DKSE framework, a 1% increase in ECI² corresponds to a 0.0515496% reduction in LnEFP. This indicates, having successfully navigated the initial stages of industrialization and transitioned from outdated techniques to more advanced processes, these economies achieve a status that enables the adoption of energy-efficient and less polluting technologies and innovations. While productive diversification and sophistication continue to progress, the overall growth in economic performance and well-being occurs at a slower pace than the decrease in environmental pressure, which gradually lessens over time, in line with the findings of [38,68,110]. The results align with studies suggesting that, from a social and institutional perspective, individual and societal preferences in more advanced and economically developed economies notably shift towards a demand for superior goods. This shift includes a greater emphasis on peace, security, and improved quality of life and environmental conditions [27,111]. As these preferences evolve, they encourage the creation of policies and systems aimed at environmental regulation, control, and protection [112]. Thus, as economic complexity increases, the depletion of environmental resources is mitigated.

Thirdly, all factors being equal, a 1% increase in ECI³ results in a 0.0135448% increase in LnEFP. This finding suggests that various tensions and pressures may emerge as production systems in the countries under study continue to develop, modernize and reach greater complexity. In some cases, this might hinder or even reverse the positive environmental effects observed in earlier stages. This confirms our primary hypothesis of the analysis (H₁): the existence of an N-EKC among the advanced economies examined in exploring the relationship between ECI and environmental pressure measured by EFP. This finding carries significant implications, as it identifies tensions and limitations on sustainable development in the studied economies.

This finding runs counter to those of [80], who studied the relationship between emissions and GDP per capita in seven emerging economies, finding support for the traditional inverted U-shaped quadratic form rather than the N-EKC. Our findings align with those of [113], who confirmed the presence of the N-EKC in India. Similarly, ref. [114] explored this relationship among OECD countries, while [79] examined the EKC hypothesis across 74 countries, finding evidence supporting the N-EKC relationship in most economies analyzed. Ref. [77] also conducted a study on a panel of major nuclear powers, validating the N-EKC relationship.

Furthermore, ref. [82] examined the connection between natural resources, ECI, and sustainable development in European Union (EU) member states, using the N-EKC framework. They found support for the N-EKC relationship in all the countries examined. There is a need for more conclusive research on this phenomenon [77]. The current literature presents various explanations, a particularly compelling one being that productive systems of all types and complexities ultimately reach a saturation point at which the mechanisms that promote growth and sustainable development weaken [115]. Rapid technological obsolescence may further exacerbate this phenomenon [112]. In highly advanced economies, innovations are often introduced at an unprecedented rate, leading to swiftly replaced cutting-edge technologies. Consequently, this rapid turnover means existing technologies may not reach their full potential. Scholars have proposed that the rise in economic and social inequality—stemming from rapid technological advancements and the uneven distribution of their costs and benefits—plays a significant role in the emergence of environmental stresses in highly developed economies [116].

In summary, the present study explores the complex N-shaped relationships described by empirical evidence of the EKC in terms of the ECI and LnEFP in the EU-5 economies. The findings strongly corroborate the presence of an N-EKC, showing that when ECI increases beyond a certain point, environmental performance is negatively impacted. This pattern suggests a transition from scale to technique and composite effects. However, without appropriate improvements, such transition could lead to long-term obsolescence. Distinct phases are observed in this relationship. These results highlight the importance of considering additional influencing factors while examining how increased ECI may affect these countries’ environmental conditions. Therefore, this research assesses the environmental implications of ECI, taking into consideration the effects of R&D investment relative to GDP (LnR&D), renewable energy consumption (LnRNW), environmental taxation (LnETX), and the moderating role of R&D expenditure on the ECI and EFP nexus (Ln(ECI×R&D)).

The second hypothesis initially established (H₂) was also confirmed. The findings of the empirical model reveal a significant inverse relationship between LnR&D (the natural logarithm of R&D expenditure) and LnEFP (the natural logarithm of EFP). Specifically, the DKSE technique analysis reveals that a 1% increase in R&D expenditure relative to GDP (LnR&D) corresponds to a 0.3906719% reduction in EFP (LnEFP) across the panel of the economies analyzed. Furthermore, this positive environmental effect rises to 0.4027006% when considering the coefficient estimated for LnR&D in the PCSE test comparison. In conclusion, the proposed model substantiates the hypothesis that investments in technological innovation (LnR&D) yield positive environmental outcomes, assuming other factors remain constant. This finding aligns with results from several studies, including those by [11,117,118,119,120,121]. These works highlight that new technologies are more efficient and environmentally sustainable, helping alleviate the environmental pressures associated with economic development.

Our study examines the impact of renewable energy consumption (LnRNW) on the EFP (LnEFP), highlighting the importance of renewable energy and the energy transition process in promoting environmental sustainability. Focusing on the five largest European economies, the findings indicate that a shift toward cleaner energy reduces EFP and supports long-term ecological health, making renewable energy essential for achieving sustainable development and mitigating the effects of traditional energy practices. In this research, the coefficient estimated by the DKSE for the relationship between LnRNW and LnEFP is 0.0935813. This finding indicates that, in the economies analyzed, a 1% increase in renewable energy consumption leads to a 0.0935813% reduction in EFP, assuming all other factors remain constant. These results align with a significant body of research highlighting the commitment to renewable energy as one of the most effective and efficient strategies for reducing negative environmental impacts and addressing climate change. This is supported by the studies conducted by [11,118,119,122,123,124,125]. Supporting the energy transition by increasing the use of renewable energy alleviates identified environmental pressures [126]. Ultimately, hypothesis H₃ is confirmed based on the results obtained.

Furthermore, hypothesis 4 (H₄) is validated. The analysis of environmental taxation (LnETX) relative to the ecological footprint (LnEFP) reveals an inverse relationship between these two variables. However, this relationship is not fully evidenced by the DKSE estimations, where the estimated coefficient (−0.1450468) for the EU-5 does not achieve statistical significance. Notably, however, a statistically significant result emerges in the model estimated using the PCSE technique. A 1% increase in environmental taxation (LnETX) corresponds to a 0.1411995% reduction in the ecological footprint (LnEFP). These findings indicate that ETX may be a viable strategy to address environmental degradation. This conclusion is grounded in the inherently distorting effects of ETX, specifically designed to target and discourage activities and consumption patterns detrimental to the environment. In so doing, these taxes facilitate the transition towards more sustainable practices and help reduce our EFP.

Additionally, ETX generates supplementary resources for the public sector, which can be devoted to financing environmental initiatives and investing in green technologies and renewable energy sources. Recognizing that any tax represents a cost that may be partially transferred to prices in production and demand chains is crucial. This added cost might counteract the positive impacts of these fiscal measures, generating inflationary pressures, decreased productive activity, and diminished private investment [127]. This dual nature of ETX policy may help clarify the significant issues identified in the results of both the DKSE model estimations and the PCSE model. Nevertheless, our findings are consistent with recent literature supporting the beneficial effects of ETX on environmental outcomes [125,128,129].

This study evaluates the moderating effect of technological expenditure (denoted by Ln(ECI×R&D)) on the relationship between ECI and ecological footprint (LnEFP). The empirical modeling results reveal significant implications for environmental preservation, mainly based on the findings of the DKSE. Notably, a 1% increase in the technological moderating effect (Ln(ECI×R&D)) is associated with a 0.0230644% reduction in environmental deterioration (LnEFP). The validation of hypothesis 5 (H₅) is extremely significant and holds important implications for the overall analysis and its core objectives. This outcome is significant as it promotes ECI while preserving environmental quality. Graphically, the negative moderating effect indicates a rightward shift of the N-shaped curve that represents the relationship between ECI and EFP (see Figure 3). Consequently, increasing expenditure on research and new technologies is a critical factor in pursuing sustainable development pathways in highly advanced economies, such as those examined in this study. These findings align with the results presented by [69,130,131,132].

6. Conclusions

Recent research has examined the environmental challenges faced by European economies, which feature diverse productive structures with knowledge and innovation as key development drivers, indicating high ECI.

Despite the substantial resources and efforts of governments, institutions, companies, and individuals in the quest for sustainable development, the degradation of natural environments and the prevalence of extreme weather events continue to pose serious challenges to health, the economy, and overall well-being. Consequently, it is imperative to better understand the factors underlying the mechanisms and phenomena that disrupt the delicate balance between economic development, well-being, and environmental sustainability across many economies, including the most developed ones.

This research examines the environmental implications of ECI within the five largest European economies known for their advanced and sophisticated productive structures. The findings suggest these implications may contribute to long-term environmental degradation. To facilitate this analysis, we apply the EKC hypothesis, which allows us to propose a novel N-shaped pattern as an empirical framework for identifying the symptoms and underlying causes of this issue.

The study introduces a model to evaluate whether a Kuznets-like N-shaped pattern exists in the relationship between ECI and environmental impact, observed explicitly through EFP. Furthermore, the model considers additional interacting variables that may affect the cubic relationship between ECI and environmental impact. These variables include innovation expenditure as a percentage of GDP (R&D); renewable energy consumption, serving as a proxy for the energy transition (RNW); environmental taxation, which reflects the discretionary and overarching role of environmental tax policy (ETX); and the moderating effect of technological innovation (ECI×R&D) on the relationship between ECI and EFP. The aim of this comprehensive approach was to provide a more nuanced understanding of the dynamics at play in this critical research area. The model was estimated using Driscoll-Kraay Standard Errors (DKSE) and Panel Corrected Standard Errors (PCSE) techniques for a panel dataset that includes the five major European economies: France, Germany, Italy, Spain, and the UK. This analysis covers the period from 1994 to 2020. The main findings and policy recommendations arising from this study are summarized below.

All the hypotheses developed during the model construction phase were fully validated through model analysis. As a result, this study confirms the presence of an N-shaped EKC in the relationship between ECI and EFP, while controlling for other variables. An inverse relationship was also found between R&D, RNW, ETX, and EFP, while controlling for other variables. Finally, it is confirmed that the moderated and adverse effect is an inverse relationship between the moderating effect of R&D (ECI×R&D) and EFP, while controlling for other variables. These findings show that environmental tensions emerge at high levels of economic complexity. The underlying factors that might contribute to this phenomenon are the accelerated technical exhaustion and obsolescence in high technification and the advanced economic structure and increased socioeconomic inequalities in highly developed economies. This disparity is evidenced in widening gaps in wealth, access to technology, and overall well-being in the population.

Technological innovation, fueled by R&D investment as a proportion of GDP, is essential in tackling critical challenges. It markedly enhances the efficiency of production systems in terms of energy consumption and environmental impact. Additionally, it promotes the development of new energy sources and accelerates the transition toward sustainable energy solutions. Moreover, it plays a vital role in maintaining the technological replacement rate, particularly the concerns raised by the N-EKC, which emphasizes the rapid depletion and obsolescence of existing technologies. This study empirically supports a significant inverse relationship between the moderating effect (ECI×R&D) and the EFB, which enabling long-term sustainable development.

Improving and refining the energy transition process is imperative to promote long-term sustainable development pathways in advanced economies. This involves a robust focus on gradually adopting renewable energy sources. This perspective is supported by research illustrating a significant inverse relationship between RNW and EFP. The transition to renewable energy must be carefully aligned with an economy’s specific economic, social, and cultural contexts. Given that development of renewable energy technologies is ongoing and may entail higher relative costs, their implementation must avoid disrupting traditional sectors and exacerbating existing economic and social inequalities. These inequalities often arise from the complexities inherent in specialized economies.

Consequently, as set out in the European Green Deal [46], launched in 2019, the energy transition should be undertaken with a sense of selectivity and social responsibility, adhering to the principle of no one being left behind. ETX in the economies examined was validated as an effective mechanism for promoting ecological sustainability. However, akin to the policies governing energy transition, such taxation must be painstakingly designed and implemented to avoid creating regressive effects on households and hindering economic activity. The emphasis should pivot away from solely generating revenue, and focus on precisely defining the mechanisms of fiscal incentives and disincentives. Neglecting this may lead to increased costs, which could be transferred to consumers or result in business relocation. Ultimately, this could contribute to inflation and exacerbate disparities in household wealth without achieving a progressive approach to taxation. Contemporary empirical research findings on issues such as energy, fiscal matters, and technology should be used to shape environmental policies. The effectiveness of these policies is contingent upon a full understanding of the challenges we face and their underlying causes.

This study offers several important recommendations for policymakers. Advanced economies are increasingly susceptible to complex threats from their reliance on technology and knowledge. Therefore, early identification of these challenges is critical to mitigating the effects of tensions and rigidities that may adversely affect the environment. It is vital to recognize that promoting innovation and investing in technology are essential for economic and social advancement. In technologically advanced economies, a persistent demand for innovative solutions necessitates a swift pace of technological development to avoid bottlenecks and potential stagnation. To cultivate a strong culture of innovation, enhancing the quality of the educational system and promoting the generation, retention, and attraction of human and intangible capital (knowledge) is imperative.

Additionally, aligning energy and ecological transition processes with the existing economic, technological, social, and cultural contexts is crucial. A failure to achieve this alignment could lead to the marginalization and swift decline of specific sectors, thereby intensifying inequalities. Furthermore, environmental policies must be consistent with social realities, ensure equitable distribution of wealth, and strengthen the adaptive capacities of households and businesses. The principle of inclusivity, that is, no one being left behind, should guide how these policies are formulated and implemented. The increasing inequalities within advanced economies represent significant long-term risks to environmental sustainability.

Research development limitations have been identified. The first involves the short analysis period, affected by Brexit on one side and the complete paralysis of European and global economies due to the COVID-19 pandemic on the other. Additionally, although the five analyzed economies form a cohesive group of highly sophisticated Western economies, various economic, technological, and cultural factors set them apart. Recognizing these differences within the EU-5 group and lengthening the analysis period are potential areas for future exploration.