Study of the Environmental Kuznets Curve in the EU27 Countries Taking into Account Socio-Economic Factors and GHG and PM Emissions

Abstract

The study of the relationship between human economic activity and the state of the environment in recent decades is reflected, among others, in the study of the environmental Kuznets curve (EKC) hypothesis. Numerous attempts have been made to examine the existence of the EKC by correlating various measures of environmental devastation with GDP per capita indicators. In this study, the aim and research gap were to compare and confirm/exclude the obtained results with the studies previously conducted by the authors, which referred to the analysis of the EKC in relation to general GHG emissions. In this analysis, GHG emissions per capita were assumed. In recent years, in the EU countries, more and more attention has been paid to research on the EKC focused on GHG emissions, and a certain research gap has been noticed in the direction of EKC analyses for PM_2.5 and PM₁₀. In the context of PM emissions, the very negative impact on human health should be emphasized. The authors decided to analyze the EKC hypothesis based on the current available data also in relation to PM emissions. In this study, a group of socio-economic variables in the form of GDP, gross electricity production, passenger cars, and population were used in relation to GHG, PM_2.5, and PM₁₀ emissions in the EU-27 countries. The study used multiple regression analysis to test the direction of the relationship between pollutant emissions and GDP. In the case of Germany, a negative correlation was obtained for GHG, PM_2.5, PM₁₀ emissions, and GDP. The EKC approach can be successfully used both in system awareness (qualitative) and quantitative studies to prepare scenarios of changes in greenhouse gas and PM emissions and to create strategic planning, manage resources, promote innovation, and in climate policy.

1. Introduction

Since the late 20th century, when researchers Grossman and Krueger [1] recommended using an approximation of Environmental Kuznets Curves (EKC) in the environmental and economic benefit-loss analysis to link economic growth and CO₂ emissions, this approach has been used in a very diverse way. Kuznett’s curve models in environmental economics theory were originally addressed to depict income variability, economic growth, and environmental pollution rates. Classically, the upward (industrialization) period increases pollution levels to excess when environmental consciousness emerges under certain economic conditions [2,3,4]. Initially, the EKC approach was built on several influencing factors or mechanisms. In these cases, the EKC was considered to materialize in the context of market liberalization, which promotes economic development. It should be considered that countries develop their economies in specific directions depending on resources, geographical dislocation advantages, and technological progress, including new and greener production methods and productivity efficiency [5,6].

The use of the EKC approach is ambiguous, although more support generally prevails. This is evidenced by an analysis of environmental and econometric indicators in groups of countries in the European region, such as a complex analysis in 28 European countries [7], five European Union countries [8], or at country level—France [9] and Finland [10]. The use of the EKC is much more intensive in various aspects in China [11,12,13], Pakistan [14], and the US [15]. Several other studies, such as in Australia [16] and Italy [17], have obtained results where mathematically precise correlations between econometric parameters and environmental indicators were impossible to obtain.

2. Literature Review

The spatial and temporal scale of research directions, fields, and geographical distribution in the use of the EKC is extremely broad: this is due to resource problems, variability of pollutant emissions and concentrations, waste, food safety, sustainable development, environmental management, and environmental policy, among others.

Several studies have identified that the EKC relationship should be adjusted according to the dataset and regional specificity. Moreover, about eight forms of association between economic and environmental indicators has been recognized, such as direct and inverted N-shaped, direct and inverted U-shaped, negative and positive monotonic, and the option of no relationship [18,19]. A more detailed understanding of the inverse U-shaped curves of the EKC relates to the analysis of the phases of economic development. The most common breakdown is in three phases [19,20]: (a) an agrarian economy characterized by low production and low income, sufficient environmental resources, and low waste [21], but where a lack of understanding of long-term environmental protection poses risks of environmental degradation; (b) rapid growth in production and average income, rustic rapid depletion of resources, and rapid accumulation of waste; and (3) high-level services and production, and high income.

Several studies have analyzed factors that could affect the shape of the inverse U-curve, for example as early as 1993. In that year, it was estimated that planned policies and monopolies, subsidies, and an abnormally low tax on natural resources impact the volume of GHG emissions, as in such cases, resources are not spared. In such cases, the slope of the U-shape curve changes [22]. Furthermore, several studies have highlighted that an effective environmental management system is essential; for example, a high risk of corruption has a significant impact on the time of reaching the turning point [23,24].

When expanded, studies often include economic parameters such as Gross National Product (GNP) and the relationship to specific environmental pollution indicators such as changes in SO₂ concentration and dynamical changes in particulate matter emissions. They also include attempts to understand and model future climate change scenarios to achieve climate neutrality, limit global warming, or even reach pre-industrialization levels. In such cases, the EKC models include a more comprehensive range of environmental indicators such as industrial development, waste management, land use and change, and land management practices. Theoretical studies of any kind need to be validated. In the case of EKC models, square and cubic econometric regression equations are used, the choice of which is determined by the data sets used and the purpose of the study. In most cases, square regression equations are used for singular environmental pollutants, while cubic correlation testing is used for multiparameter cases [25].

The most frequent findings and relationships resulting from the analysis of scientific publications include the following: there is a strong relationship between economic and environmental indicators, most often in N-shaped form or inverted U-shaped form; GNP can be successfully used as a parameter for economic growth; and the EKC approach can be successfully used both in system awareness studies (qualitative) and quantitatively to prepare greenhouse gases (GHG) change scenarios and to undertake strategical planning [25].

Unlike other environmental polluting activities, atmospheric pollution is characterized by spatial distribution, transboundary transmission and impacts. In the case of particulate matter (PM₁₀, PM_2.5), not only are primary emissions associated with economic activity, but local secondary aerosol formation due to specific microclimatic conditions and other indirect sources of pollution, such as sources of emissions of volatile organic compounds, are often essential. As identified in several studies, such circumstances frequently make it impossible to see a specific relationship between particulate matter emissions and economic indicators [11]. Several studies have identified that PM_2.5 concentrations cannot be explained solely by population changes [26], changes in production processes, energy load, or variability in transport unit flows. Some authors suggest adding secondary industry criteria, including urbanization rates, to these patterns of variance relationships. One of the most sophisticated methodological tools for Kuznet includes regional particulate matter concentration changes, specific gross domestic product income levels, production structure indicators, and energy consumption indicators [11].

Several studies in the Nordic countries (Sweden [27], Norway [28], and Finland [10]), characterized by progressive environmental policies and support for the EKC approach, have shown that sustainable development can be planned by reducing emissions and promoting renewable energy resources. A positive example is the Swedish environmental policy, introducing a so-called carbon tax on fossil fuels. When analyzing the results obtained, it was possible to confirm that the “critical point” of the EKC curve was reached more quickly, even in the case of relatively low income levels. On the other hand, the environmental policy introduced in Japan (stringent energy efficiency norms [29]), based on technological innovation and energy efficiency and the development and implementation of state-of-the-art technologies to reduce both energy consumption and environmental pollution, has proven its effectiveness. These activities resulted in lower dependency on environmentally polluting energy sources while ensuring high economic development. The experience of Germany [30], where the transition to a circular economy is slower, shows that various initiatives (investments in renewable energy infrastructure, energy efficiency) significantly reduce the use of fossil fuels. As a result, progress towards the “critical point” is faster. The policy pursued in Germany is characterized by gradual growth and the achievement of an environmental quality balance.

Less positive examples are known about the desire to develop too rapidly and without thought, ignoring environmental requirements. In some studies, in China, where the focus is on the fastest possible economic growth and although ecological requirements have been introduced (investments in the use of renewable energy sources, increased use of solar energy), environmental quality indicators have been reduced, contributing to environmental degradation [31]. Data analysis showed that reaching the “critical point” occurred over an extended period. The US also shows skepticism, only supporting partially implementing the EKC approach [32]. Attitudes towards environmental demands and the introduction of a higher environmental consciousness were clearly demonstrated when the US withdrew from the Paris Agreement. Interestingly, 107 days after the withdrawal, the US rejoined the Paris Agreement. However, such inconsistencies weaken international cooperation on climate change mitigation. Such inconsistencies in implementing environmental policy are also observed in India, where industrialization and urbanization are very rapid, and the demand for energy, mainly provided using fossil fuels (coal), is increasing accordingly, leading to rapid environmental degradation and rising pollution levels. While India has seen several renewable energy policy initiatives (e.g., solar panel farms), the non-systematic and slow implementation of environmental policy leads to stagnation, and the achievement of the “critical point” is extremely slow [33].

The EKC mathematical models consist of several, mainly regression equations. The most common square mathematical model describes the inverse U-shape relationship in which environmental degradation (such as pollution levels or environmental impacts) depends on income (GDP per capita) over a given period and coefficients that characterize the effects of regressors by the impact ratio between economic growth and environmental pollution. Typically, these models also consider the error component [34]. However, a logarithmic model may also be used if the relationship between variables is logarithmic. Extended regression equations typically include a more comprehensive range of parameters such as technological innovations (an indicator of the intensity, development, and contribution of innovation), energy consumption (share of renewable energy), economic activity (share of different sectors in GDP), and institutional quality (environmental policy, environmental legislation). Sometimes, so-called panel data models are used to compare different countries with each other. These models then include parameters that characterize national developments as compared to global trends [35].

The potential for future use of the EKC is broad and prospective, as such analysis offers a structured framework for research and analysis of the balance of environmental and economic development based on which sustainable policy development can be built. This approach of the EKC treaty is expected to be very practical in the future for climate policy planning, innovation promotion, and resource management. For example, the EC approach can be applied to (1) countries to set up specific support programs to stimulate clean technologies in the early stages of growth, making it possible to reach the “critical point” of the EKC faster; (2) countries could identify stages of development for when to launch more robust emission control measures and move to renewable energy in order to achieve carbon neutrality by 2050; (3) this approach can help to understand when to introduce new technologies that reduce pollution, identify financial resource investments in renewable energy, energy efficiency, and circular economy technologies to help achieve the “critical point” more quickly; (4) to facilitate the transition to a circular economy based on resource efficiency, minimization and recycling of waste; in this case, pollution is mitigated at an early stage of development by promoting the use of renewable or recycled resources rather than the investment of new resources; (5) as an instrument for urban development, health policy planning, and pollution reduction in agglomerations; and (6) the implementation of ambitious international environmental policy and climate agreement initiatives at the national level; for example, higher income countries could set higher environmental targets. Such adaptive policies could be more effective in mitigating global climate change.

Significant challenges in using the EKC in atmospheric pollution, particularly particulate matter pollution, relate to data variability and stochastic dispersion of pollution data. Identifying and linking atmospheric levels of aerosol pollution with economic growth is only sometimes possible, as even countries with very high levels of income pollution can remain at very high levels due to cross-border and long-range pollution transport and market mobility (economic mobility). For example, factories and production can be moved to less environmentally friendly countries. It has been repeatedly found that there is a risk that the EKC theory may prove ineffective in the case of aerosol pollution and that any future policy and stricter hypothetical environmental instruments should be rethought very carefully.

Despite the many positive aspects and widespread application of the ECK, this method also has several weak points that should be assessed when applied: (1) measurement data could be nonhomogeneous, and inequality problems within different data sets and regions could arise; (2) nonrobustness is essential because, in some cases, unique situations could occur when the EKC approach cannot explain relationships; (3) external factors, such as globalization or international policy, could unexpectedly influence the situation; (4) irreversible environmental damage that cannot be recovered even if the environmental consciousness is high and the critical point has been reached; (5) a practical framework for environmental regulation and governance often needs to catch up to economic growth and environmental consciousness, hampering the achievement of the critical point; and (6) temporal growth and structural complexity: complex interactions between inequality, economic growth, and environmental degradation can complicate the legacies of EKC models.

Recently, Mohammed et al., 2024 [18] published a comprehensive study on the diversity of EKC use. A survey of econometric literature on EKC implementation shows that: (a) wide spatial research areas were covered, and analysis was also carried out within different economic alliances (e.g., ASEAN, BRICS, EU, OECD, PIIGS); (b) temporal analysis covering data from the late 1970s until the present; (c) there is a strong dominance of CO₂ pollution variability analysis; (d) there is a wide diversity of input data, e.g., GDP, trade, human capital index, natural resources rents, economic complexity index, urbanization process GDP, Brexit and crisis episodes, governmental expenditures, globalization and institutional quality, population aging, and income inequality; and (e) consequently, extensive outputs were obtained, e.g., showing different EKC output shapes and validating previous studies connected to the EKC relationship between renewable energy, GDP, and CO₂.

It should be noted that some of the studies, e.g., in Malaysia (1970–2016), did not recognize the conventional EKC pattern between GDP, natural resources, and CO₂. In another study where income inequality and CO₂ analysis was performed [36], substantial conclusions were reached about the double threshold effect on economic growth and carbon per capita emission.

The configuration of dependent and independent variables and the Kuznets curve research methodology make each approach and studies original and innovative, because it usually concerns different variables and each time brings new observations and information complementing knowledge in this area. It should be emphasized that the studies need to be updated as the data are subject to change, which directly affects the results of the EKC analysis. In the previous publication by the authors [37], the explained variable was the total GHG emission in selected European countries. In this study, the aim and research gap were to compare and confirm/exclude the obtained results with studies previously conducted by adopting GHG emissions per capita. The survey has been extended to all EU-27 countries. In recent years, in the EU countries, research on the EKC has been increasingly focused on GHG emissions, and a certain research gap is observed in the direction of EKC analyses for PM_2.5 and PM₁₀. In the context of PM emissions, it should be emphasized that there is a very negative impact on human health. The authors decided to analyze the EKC hypothesis based on the current available data also in terms of PM emissions. In the context of both GHG and PM emissions, any complementary information is a valuable source of knowledge.

Despite prevailing support for the suitability of EKC theory, scientific publications focus on two different groups of countries: specific countries with a high level of development (e.g., USA, Germany, France) and countries with a moderate level of development over a certain period (e.g., India, Eastern European countries). Systematic empirical surveys, acquiring up-to-date data, and adding datasets are necessary to validate the results obtained, which is the aim of this study. The strategic role of such a verification process is also confirmed by the statements made in the publication [38] on validating the EKC. As one of the most successful options, an extended dataset is proposed to implement cross-validation data that would confirm the relevance of the EKC hypothesis. An essential aspect that should be considered is the use of the EKC methodology in different geographical regions. The 2023 analytical review publication [3] on the application of the EKC method concluded that publications produced by scientists from China, Turkey, Pakistan, Malaysia, and Saudi Arabia, which have, in fact, also determined the rapid development of this idea, are highly dominant. Relatively fewer publications concern the European region, with the highest number for Italy (14), the UK (15), and Spain (12), 5.5 times less overall than for China. This disproportionality points to an acute need to align research sectors and trends, mainly because of the globalization of interconnected economic and environmental processes.

The research hypothesis of this study is the assumption that among the analyzed EU-27 countries there are countries for which the hypothesis of the existence of the EKC is achieved. The analysis is intended to indicate whether countries increase or decrease emissions as GDP increases.

3. Materials and Methods

3.1. Data Source and Methodological Framework

The paper investigated the intertemporal link in the energy–environment–income relationship. Energy is one of the important variables. Including fuels and/or energy in EKC studies helps decision makers understand the factors that may affect energy consumption and/or CO₂ and PM emissions in the long term.

The explanatory variables include CO₂, PM_2.5, and PM₁₀. PM poses a serious threat to human health and in many European countries its permissible standards are constantly exceeded. CO₂, as a greenhouse gas, especially in the EU-27, is subject to various forms of monitoring and assessment of emission levels and, in accordance with the regulations in force in the EU, constant efforts are being made to reduce its emissions. GDP per capita is a variable that is obligatorily examined when undertaking Kuznets curve studies. In most European countries, energy production and emissions from transport are associated with the emission of both PM and CO₂. Therefore, it was decided to choose EP and passenger cars as explanatory variables (they constitute the largest source of emissions and the largest group of all vehicle categories). The size of the population in individual countries is a complementary variable, often considered in research in another form, e.g., human capital index or population aging [18].

The goal of this paper was to test the hypothesis of the environmental Kuznets curve (EKC) in EU-27 countries. The research sample consisted of Member States of the European community EU27. Data from the international databases of public statistics OECD [39] and Eurostat [40] in the years 2000–2022 were used. Due to the availability of data in the Eurostat database, the analysis taking into account CO₂ emissions was carried out for the period 2000–2022, and for PM_2.5 and PM₁₀ in the years 2000–2021.The multivariate framework included: the group of economic and social variables constituting explanatory (independent) variables consisted of GDP data, electricity production (EP), passenger cars (cars) and the size of the population in individual countries (pop) and the explained (dependent) variables were GHG emission per capita and national total emission for the entire territory PM_2.5 and PM₁₀. The analyses were carried out separately for individual countries of the European Community, justifying this step with separate, differing economic and environmental policies pursued in these countries.

In real research, the issue of assessing quantitative relationships between various aspects of phenomena is often taken up. The purpose of such analyses is usually the desire to get to know them better, confirm or refute the hypotheses formulated in the theory, the ability to predict the development of the studied processes or phenomena, or the use of knowledge of quantitative relationships for simulations. In this study, the multiple regression method was used. The general goal of multiple regression is to quantify the relationships between multiple independent (explanatory) variables and a dependent (criterion, explained) variable. With the regression equation at hand, it is easy to predict the processes of interest to us.

In order to examine the correlation of independent variables with the dependent variable, the Pearson linear correlation coefficient was used. The relationship of all economic and social variables with respect to the dependent variable was examined based on multiple regression analysis, which also allowed us to test the hypothesis of the EKC.

The modeling used multiple regression, which makes it possible to search and quantitatively describe complex relationships. The constructed multiple regression model allows for examining the influence of many independent variables (X₁. X₂. … X_n) on one dependent variable (Y). It is an extension of regression models based on the Pearson linear correlation coefficient. It assumes the existence of a linear relationship between the studied variables [41]. The multiple regression model takes the following form:

$$Y={\alpha }_0+{\alpha }_1{X}_1+{\alpha }_2{X}_{2 }+\dots +{\alpha }_n{X}_n+ϵ$$

(1)

where:

• Y—dependent variable;
• X₁. X₂. … X_n—independent variables;
• α₁. α₂. … α_n—parameters;
• ε—random component (the rest of the model).

The research was carried out using the Statistica v. 13.3 statistical package.

3.2. Data Description

Descriptive statistics of the variables for the analyzed countries are presented in Table S1 (Supplementary Materials). Descriptive statistics were performed for the arithmetic mean of the data in the years 2000–2022 and for PM_2.5 and PM₁₀ from the period 2000–2021.

The average GHG emissions per capita in the EU-27 in 2022 was 7.6 tons. The highest emissions were recorded in Luxembourg (14.5), Ireland (13.1), Czechia (11.4), Estonia (10.7), and Cyprus (10.2). The average of the overall national GHG emissions in the EU-27 totaled 129,054.29 tons. The highest overall annual emissions were recorded in Germany (777,380.09), Italy (419,467.07), France (409,732.55), Poland (383,434.04), and Spain (309,286.15) (Figure 1).

The average PM_2.5 emissions per capita in the EU-27 in 2021 totaled 0.0033 tons. The highest emissions occurred in the following countries: Latvia (0.0094), Poland (0.0079), Croatia (0.0067), Romania (0.0060), and Slovenia (0.0048). In terms of total PM_2.5 emissions in the EU-27, the average amount was 49,375.92 tons, with the highest emissions recorded in Poland (297,282) [42], France (189,218), Italy (149,106), Spain (135,005), and Romania (116,136) (Figure 2a). In regard to PM₁₀ emissions, the average emission per capita was 0.005 tons, with the highest emissions in Latvia 0.015, Croatia 0.012, Poland 0.010, Estonia 0.009, and Lithuania 0.008. The average total PM₁₀ emissions amounted to 74,804.5 tons, with the highest emissions in Poland 38,7843, France 270,346, Spain 215,303, Italy 199,661, and Germany 183,993 (Figure 2b).

The leading countries with the highest GDP per capita in 2022 were: Luxembourg, Ireland Denmark, Sweden, and the Netherlands (Figure 3). At the same time, it is noted that some of these countries are characterized by the highest GHG emissions per capita (Figure 1).

Some accuracy has been observed in Figure 4 and Figure 5. The gross electricity production per capita data in individual countries reflect a certain trend in relation to the data on the total number of passenger cars.

The population in individual countries is presented in Figure 6.

4. Results and Discussion

The significance of statistical variables was determined based on Pearson correlation coefficients. The variables statistically significant for p < 0.05 are indicated in bold font (Supplementary Materials, Table S2). The correlation in Table S2 was performed for data from 2021. Statistically insignificant variables were not considered in the next step of the research. For some countries, all the considered variables turned out to be statistically insignificant (Austria for GHG, Hungary for GHG, Lithuenia for GHG and PM_2.5, Poland, Portugal for GHG, Romania for PM_2.5 and PM₁₀).

The results of the multiple regression analysis are presented in Table S3 (Supplementary Materials). The regression analysis was conducted for data from the period 2000–2022 for CO₂ emissions, and for PM_2.5 and PM₁₀ from the period 2000–2021. The R² coefficient value refers to the model fitted to the data. For further verification, the Adjustice R² parameters and the Fisher test result (F), which is particularly important when comparing models, are used. The higher the F-test value, the better the model fit. The values of the F statistics and the corresponding level of test probability p indicate a significant linear relationship. The statistically significant variables are indicated in bold font. The coefficients of determination for the numerous countries are very high, e.g., for GHG Adj. R² ˃ 0.9 (Belgium, Denmark, France, Greece, Ireland, Italy, Luxembourg, Malta, Netherland) or high (Adj. R² ˃ 0.8 Cyprus, Czechia, Hungary, Latvia, Romania).

Due to the low adjustice R² coefficient indicating weak, very weak, or lack of linear correlation, the following countries were not considered in the further part of the study: Bulgaria for PM_2.5, Croatia for GHG and PM₁₀, Estonia for PM₁₀, Finland for GHG, Italy for PM₁₀, Slovenia for GHG, and Sweden for GHG (Table S3).

Due to the number of countries analyzed, an interpretation of the regression results and coefficients was made, for example, for Austria (

Table 1. Regression equations for GHG, PM_2.5, and PM₁₀ emissions.

	Country	Regression Equation
1	Austria	Y1 = 0.167X1 − 0.525X2 + 90759.59 Y3 = 0.264X1 − 0.621X2 + 108235.3
2	Belgium	Y1 = 0.230X1 + 1.099X2 − 2.295X4 + 80.652
3	Bulgaria	Y1 = 0.717X3
4	Croatia	Y1 = 0.432X1 + 4.038
5	Cyprus	Y1 = 1.328X2 − 2.179X4 + 39.433 Y2 = 0.245X3 − 1.085X4 + 7499.7 Y3 = 0.312X3 − 1.103X4 + 14049.9
6	Czechia	Y1 = 0.901X1 − 0.868X4 + 65.298 Y2 = −1.020X2 + 0.221X4 Y3 = −1.351X2 + 0.364X4
7	Denmark	Y1 = 0.207X3 + 62.360 Y2 = −2.978X2 Y3 = 0.677X1
8	Estonia	Y1 = 0.628X3
9	Finland	Y1 = 0.639X3 Y2 = −1.631X2 + 0.117X3 Y3 = −1.451X2 + 0.163X3
10	France	Y1 = 0.260X1 − 1.003X4 + 30.336 Y2 = −0.860X4 + 1963269 Y3 = −0.977X4 + 2403857
11	Germany	Y1 = −0.765X4 + 28.503 Y2 = −0.920X1 + 441758.8 * Y3 = −0.899X1 + 622720.9
12	Greece	Y1 = 0.276X1 − 0.567X2 + 0.203X3 + 0.294X4 − 24.012 Y2 = 0.380X1 − 0.703X2 Y3 = 0.497X1 − 0.677X2
13	Hungary	Y1 = 1.588X1 + 2.375X4 − 95.547 Y3 = −0.987 X4 + 622808.2
14	Ireland	Y1 = 0.242X1 + 0.924X2 + 0.238X3 − 2.274 Y2 = 2.077X2 − 3.060X4 + 68213.40 Y3 = 3.705X2 − 4.709X4 + 188393.8
15	Italy	Y1 = −0.584X2 + 0.375X3 − 0.292X4 + 26.560 Y2 = −0.680X2 + 0.535X3 Y3 = −0.625X2
16	Latvia	Y1 = −0.981X4 + 36.332 Y2 = 1.295X4 − 66007.1 Y3 = 0.928X1 + 1.626 − 77783.5
17	Lithuania	Y3 = 0.322X2
18	Luxembourg	Y1 = 0.494X3 Y2 = 0.987X4 Y3 = 0.818X4
19	Malta	Y1 = −2.452X2 + 0.258X3 − 1.296X4 + 51.834
20	Netherlands	Y1 = −1.16824 X4 + 69.550 Y2 = 0.263X1 − 2.048X2 + 0.134X3 + 0.701X4 Y3 = 0.253X1 − 1.855X2 + 0.562X4
21	Poland	**
22	Portugal	Y2 = −0.687X3 + 122834.1 Y3 = −0.670X3 + 195269.4
23	Romania	Y1 = 2.206X1 − 1.720X2 + 1.400X4 − 26.797
24	Slovakia	Y1 = 0.893X1 − 1.753X2 Y2 = 0.692X1 Y3 = 0.769X1
25	Slovenia	Y2 = 2.484X2 − 2.497X4 + 241994.9 Y3 = −1.949X4 + 261530.8
26	Spain	Y1 = −0.689X2 + 20.447 Y2 = −0.840X2 + 304514.1 Y3 = 0.671X3 − 1.438X4 + 922699.5
27	Sweden	**

* A negative correlation between the GDP and pollutant emissions, ** Statistically insignificant variables.

). The significance and interpretation of the impact of the analyzed variables on pollutant emissions, except for GDP, was not analyzed, as the aim of the study was to attempt to confirm or reject the existence of the EKC hypothesis.

Analyses were conducted separately for individual countries of the European Community, justifying this step with separate, different economic and environmental policies implemented in these countries.

The obtained modeling results were interpreted as follows using the first example from Table 1 for Austria:

-

when GDP increases/decreases by 1 euro per capita, PM_2.5 emissions will increase/decrease by 0.167 tons, assuming that other variables remain unchanged;

-

when the number of passenger cars decreases/increases by 1 thousand then PM_2.5 emissions will increase/decrease by 0.525 tons, assuming other variables remain unchanged.

In some cases, the probability statistics in the obtained regression equation for all independent variables are p > 0.05, and are therefore not included. Such examples can be observed for Austria, Portugal, and Slovenia in relation to GHG emissions; for Belgium, Bulgaria, Croatia, Estonia, Malta, and Romania for PM_2.5 and PM₁₀ emissions; for Hungary for PM_2.5 emissions; for Lithuania for GHG and PM_2.5; and for Poland and Sweden for all studied emissions (Table S3). It should be assumed that other, perhaps non-linear methods would give more satisfactory results.

It is possible that some other explanatory variables, which were not included in this study, may have a significant impact on the analyzed GHG, PM_2.5, and PM₁₀ emissions. It should also be assumed that there may be a relationship between the studied variables and emissions, but it is not a linear relationship. The use of multiple regression analysis in this case did not bring satisfactory results.

The results indicating a negative correlation between the GDP variable and pollutant emissions are marked in bold (Table 1).

In a few EU-27 countries, statistical significance was achieved for all or most of the independent variables considered in the study: Belgium, Greece, Ireland, Italy, Malta, and Romania for GHG, and the Netherlands for PM_2.5. In terms of the dependence of explained and explanatory variables on parameters other than GDP, they are presented differently in the case of different countries. The regression equations show that in some countries, as the electricity production variable increases, the emissions of the dependent variables will increase accordingly: in Denmark, Estonia, Finland, Greece, Ireland, Luxembourg, Malta, and Spain GHG, in Cyprus PM_2.5 and PM₁₀, in the Netherlands PM_2.5, and in Italy GHG and PM_2.5, assuming, of course, that the other variables remain unchanged. In the vast majority of cases, this dependence concerns GHG emissions. In the case of Portugal, for example, the X₃ variable negatively correlates with PM emissions and explains only approximately 54% of the variability of the dependent variables (Table S3). In other cases, e.g., Austria, Czechia, Finland, Denmark, Greece, and Italy, the model indicates a negative relationship between the variable passenger cars and the explained variables, which may be conditioned by the bringing to the market of increasingly modern, lower-emission, and more ecological modes of transport. In the analyzed countries, in which the dependent variable in the form of population demonstrated statistical significance, in most cases it was a negative correlation and related to GHG emissions for Belgium, Czechia, the Netherlands, Italy, Latvia, Romania, and Malta, emissions of all variables in Cyprus and France, PM emissions for Hungary, Spain for PM₁₀, and Ireland for PM_2.5 and PM₁₀. In other cases, e.g., Poland or Sweden, none of the analyzed variables demonstrated statistical significance. This results from the specificity of the economic structure and energy system of a given country. In other cases, e.g., Poland or Sweden, none of the analyzed variables showed statistical significance. This results from the specificity of the economic structure and energy system of a given country. For example, in Poland, the municipal and residential sector is responsible for the main PM emissions, including mainly households—PM_2.5 (85%) and PM₁₀ (67%) [43]. In Sweden, households are responsible for about 60% of Sweden’s consumption-based GHG emissions, with emissions from the public sector and investments making up the remaining 40% [44].

As a result of the negative correlation of the dependent variables with the GDP variable, the hypothesis of the environmental Kuznets curve was confirmed in the case of Germany for GHG PM_2.5 and PM₁₀ emissions (Table 1). In other studies carried out by the authors for a different set of variables, the existence of the EKC hypothesis for CO₂ emissions was also confirmed in the case of Germany [45]. Analysis based on statistical data from the Eurostat database [40] allowed for the justification of the obtained research results. Due to the considered variable of passenger cars, the final energy consumption in road transport was analyzed by fuel type. In this category in the electricity group, Germany had the highest value in the EU-27 of 211.084 thousand tons of oil equivalent (toe), as well as in the category other liquid biofuels, as one of two European countries, Germany, has the highest consumption of 1.283 toe in 2022. In the other category—blended biogasoline, Germany was ranked second after France with the value of 748.027 toe. In relation to the next analyzed variable, gross electricity production, in the group of gross available energy by product—renewables and biofuels, Germany has the highest value in the EU-27 at the level of 49,207.958 toe. In the category final energy consumption by product—renewables and biofuels, Germany ranks first with 19,111.317 toe.

5. Conclusions

The study attempted to take into account multicollinearity between the analyzed variables: GDP, gross electricity production, passenger cars and population in relation to GHG, and PM_2.5 and PM₁₀ emissions. The annual data from EU-27 countries for the period 2000–2022 were used to evaluate the EKC hypothesis.

At the beginning of economic development, environmental quality will deteriorate with economic growth. However, after reaching a peak point (turning point), environmental quality may improve with subsequent economic growth [46]. Our research shows the negative correlation of GDP with CO₂, PM_2.5, and PM₁₀ emissions in the case of Germany. This could suggest that it has probably reached the turning point, after which pollutant emissions began to decline despite a constant increase in GDP.

Based on the obtained results, it was concluded that the constructed model makes it possible to explain approximately 78% of the variability of the modeled dependent variable for GHG, 87% for PM_2.5, and 80% for PM₁₀. The values of the F statistic and the corresponding probability statistics p confirm a significant statistical linear relationship (Table S3). For the analyzed GHG, PM_2.5, and PM₁₀ emissions, the results of the multiple regression indicate a negative correlation. Based on the above research results, it can be concluded that the existence of the hypothesis about the environmental Kuznets curve is confirmed. This means that environmental degradation increased with the increase in per capita income in the early stages of economic growth, and then declined with per capita income after arriving at a threshold.

The availability of data in a broader dimension could certainly provide a more plausible description for testing the EKC hypothesis. However, the limited dataset may still provide preliminary inferences about the validity of the EKC hypothesis.

Current levels of pollutant emissions resulting from the use of fossil fuels are unsustainable and pose a threat to the environment. The adopted energy policies in the analyzed EU-27 countries should aim to reduce the intensity of atmospheric pollutant emissions per unit of energy used. However, as long as the main energy mix remains focused on fossil fuels, these goals will be difficult to achieve. The analysis conducted for the EU-27 countries can be summed up with the conclusion that currently economic growth does not guarantee a solution to environmental problems in most European countries. The EKC hypothesis is also criticized because it is believed to be merely a statistical result and not a common path of development [37].

The importance of updating research in this area is emphasized, as the variable parameters are constantly changing.

The results of this study open up new possibilities for decision makers in the field of designing comprehensive financial, economic, and energy policies to minimize the harmful impact of environmental pollution. When determining the allocation of quotas, e.g., GHG emissions for individual EU countries, it is necessary to consider, among others, population dynamics, GDP, transport development, and the structure and intensity of energy production and consumption, because the results of the study indicate that these variables significantly but in different ways affect the size of CO₂ and PM pollutant emissions in the countries studied.