Could Globalisation and Renewable Energy Contribute to a Decarbonised Economy in the European Union?

Abstract

This study investigates the impact of globalisation, renewable energy consumption, and economic growth on CO₂ emissions in 26 European Union (EU) countries, in the period 1990–2020. Second-generation panel unit root tests are applied, the Westerlund cointegration test is used, and a panel of fully modified least squares (FMOLS) and dynamic ordinary least squares (DOLS) techniques are employed to estimate the long-term relationship between variables. The causality relationship among the considered variables is identified using the heterogeneous Dumitrescu–Hurlin causality test. It was found that globalisation and renewable energy consumption contributed to the carbon emissions’ mitigation, while economic growth induced their increase. The results are robust when control variables (i.e., financial development, foreign direct investment, and urbanisation) are added to the model. Foreign direct investment and urbanisation are contributors to carbon emissions’ increase, whereas financial development induces their decrease. The effect of the considered variables on carbon emissions is differentiated by economic development and level of institutional quality. Unidirectional causality relationships were identified from globalisation to carbon emissions and from carbon emissions to foreign direct investment and bidirectional relationships were found between economic growth, renewable energy consumption, financial development, and carbon emissions. The policy implications of the results are also discussed.

1. Introduction

The process of globalisation links countries and nations economically, financially and politically, impacting economies, political systems, environment, culture and prosperity around the world. Despite the recent opinions of experts in relation to the phenomenon of “de-globalisation”, [1,2,3]), empirical research provides evidence in support of a sound foundation of globalisation ([4], pp. 13–24), and also asserts that the “world economy will need more globalisation” [5]. De-globalisation, or a “new world order”, is a newly introduced topic in political discourse and also a concern of scholars. The world is experiencing this shift due to the pandemic, Russia’s war on Ukraine, disruptions in global supply chains, China–US trade tensions, higher global risks, decoupling economies, and the rise of the Indian population. All these appear to be threats to globalisation, and may also seriously alter the geopolitical landscape. Some voices are arguing that the COVID-19 pandemic has slowed globalisation, based on the diminished values of the KOF globalisation index in 2020. This decrease is registered mainly in high-income countries, while in middle- and low-income countries, the value of the KOF index has remained unchanged. European countries remain highly globalised due to their free trade agreements and strong political efforts dedicated to economic integration [6].

In all countries, environmental pollution is caused by factors such as economic growth, globalisation, industrialisation, investment, and urbanisation. All these factors mean an increased demand for energy, which causes more air pollution; these are mainly carbon emissions, which represented around 79.04% of greenhouse gas emissions in the European Union (EU) in 2020 (computation based on World Development Indicators [7,8]). A possible solution for alleviating carbon emissions is to diversify the structure of the energy consumption mix by extending the share of renewable and other forms of non-polluting energy. A higher share of renewable energy in the consumption mix would lead to a decarbonised economy. The European Union has assumed an ambitious target for 2050, that of climate neutrality (no net emissions of greenhouse gas); also, it also aims to raise the share of renewable energy to 45% by 2030 [9]. Member States are engaged in reforms to implement climate policies and the green transition towards the goals of the European Green Deal [10], the long-term EU’s strategy for a sustainable future.

In this context, studies on the influence of globalisation and renewable energy on the environment could provide pertinent suggestions for designing EU energy policies.

In spite of a relatively rich body of literature on the impact of globalisation and renewable energy consumption on carbon emissions, only few studies have assessed the impact of these variables on carbon emissions, along with financial development, urbanisation, and foreign direct investment in European Union countries. As examples of studies, we can nominate the following. Addai et al. [11] analysed the link between decarbonisation technology, economic globalisation, economic growth, and energy use in the case of the German economy; Vatamanu and Zugravu [12] explored the impact of financial development, institutional quality, and renewable energy use on carbon emissions in the EU countries; Ali et al. [13] investigated the link between the carbon dioxide intensity of GDP and environmental degradation in Southern European countries; Horobeț et al. [14] evaluated the financial development–environmental quality nexus in the case of the European Union economy; Horobeț et al. [15] linked inward foreign direct investment to carbon emissions in 24 European Union countries; Destek [16] studied how globalisation and the environment are correlated in Central and Eastern European (CEE) countries; Rahman et al. [17] revealed the link between financial development, globalisation, and environmental degradation in CEE countries; Ayeche et al. [18] highlighted the relationship between economic growth, trade openness, financial development, and carbon emissions for 40 European countries; Al-Mulali et al. [19] estimated the influence of renewable electricity production, economic growth, financial development, trade openness, and urbanisation on carbon emissions in 23 selected EU countries; and finally, Sadorski [20] explored the nexus of financial development–energy consumption in CEE frontier economies. None of these studies address a whole set of proposed economic variables and their links with carbon emissions.

The present research aims to investigate the nexus between globalisation, renewable energy, and carbon emissions at the EU level, alongside other pollution determinants (financial development, foreign direct investment, and urbanisation), taking into consideration that European countries are highly globalised; these countries are some of the highest producers of renewable energy in the world, and aim to be climate neutral in 2050.

The contributions of our study to existing knowledge consist of the following: (1) employing heterogeneous panel estimation techniques that allow for cross-section dependences to model the impact of globalisation and renewable energy on carbon emissions in EU countries; (2) giving an overall image of the impact of globalisation and renewable energy use in the EU, and adding a fresh overview of the limited existing evidence on EU territories; (3) highlighting the role of the level of economic development and institutional quality in mitigating carbon emissions; (4) providing useful findings for researchers, academics, governments and policymakers, and also, support for environmental and energy policy designs at the EU level (i.e., implementation of the European Green Deal).

The paper is structured as follows: Section 2 discusses the current relevant literature; Section 3 describes our data and methodology; Section 4 presents the main findings; Section 5 provides a discussion of results; and Section 6 includes conclusions and policy implications.

2. Literature Review

This section discusses recent studies regarding the impact of globalisation, renewable energy, and economic growth on pollution, as well as other factors such as financial development, foreign direct investment, and urbanisation.

2.1. Globalisation and Pollution

An impressive amount of the recent literature is devoted to exploring the link between globalisation and environment. Studies regarding this link present varied results, with a part of them revealing a positive association, and another, a negative one.

Shabaz et al. [21] found that globalisation contributed to a decrease in CO₂ emissions in China within the period of 1970 to 2012. Similar results are reported by Patel and Mehta [22] for India and by Islam et al. [23] for Bangladesh. Taking a panel data approach, Lv and Xu [24] revealed that increases in globalisation reduced CO₂ emissions in 15 emerging countries within the period of 1970 to 2012; You and Lv [25] revealed similar results using a panel of 83 countries within the period of 1985 to 2013. Zaidi et al. [26] concluded that globalisation significantly reduced carbon emissions in APEC countries based on data from 1990 to 2016. Globalisation had a reducing effect on carbon emissions in Central and Eastern European countries within the period of 1980 to 2016, according to the results of Rahman et al. [17]; a similar effect was found in MENA countries from 1970 to 2015 [27], in 18 Latin American and Caribbean economies, using data from the period of 1990–2014 [28]. Yang et al. [29] revealed that globalisation reduced carbon emissions in a global sample of 97 countries, during 1990–2016. Aladejare [30] reported that globalisation reduced environmental degradation in the five richest African economies, from 1990 to 2019. Ansari et al. [31] conclude that globalisation had a reducing effect on carbon emissions in the case of ten carbon emitters of developing countries within the period 1980–2018. This is in line with the results of Rahman and Alam [32] for Asian countries. Using a global panel of 73 developing countries, Jahanger et al. [33] found that globalisation has an adverse impact on carbon emissions.

The level of economic development is a differentiating factor in analyses of the impact of globalisation on the environment. Globalisation plays a significant role in lower-income countries by mitigating carbon emissions, while it increases carbon emissions in high-, upper-, and middle-income countries [34]. According to the findings of Leal et al. [35], globalisation caused a 0.2% and 0.52% increase in environmental degradation in developing economies, and reduction of 0.88% and 0.85% in developed ones.

Another group of studies report a positive correlation between globalisation and carbon emissions. For example, an enhancing effect of globalisation on emissions has been found in Japan [36], India [37], China [38,39,40], Argentina [41], and Australia [42]. Globalisation has increased carbon emissions in 25 developed countries, according to the findings of Shahbaz et al. [43], and Huo et al. (2022) [44]. Economic globalisation has increased carbon emissions in the long term in OECD countries [45], in NAFTA countries [46], and in South-Asian economies, [47,48,49]. Similar findings were reported for the South Asian Association for Regional Cooperation (SAARC) countries within the period 1990–2018 by Azam et al. (2022) [50], as well as in G20 countries in the period from 2005 to 2018, by Tian and Li (2022) [51].

2.2. Renewable Energy Consumption and Carbon Emissions

Renewable energy use negatively affects CO₂ emissions [32,52,53,54,55,56]). Alola and Joshua [34] revealed that renewable energy use improves environmental quality by reducing carbon emissions only in the short run in high-, low- and upper-middle-income countries. Renewable energy consumption reduced carbon emissions in 18 Latin American and Caribbean economies [28], in Argentina [41], in Brazil [57], and also in the Nordic countries [58]. Ansari et al. [31] found that renewable energy has a negative and significant impact on carbon emissions in developing countries, in line with the results reported by Kwakwa [59] for 32 African economies, by Cao et al. [60] for 37 OECD countries, and Tian and Li [51] for G20 countries. Additionally, Sheraz et al. [61] found that renewable energy reduced carbon emissions in 64 Belt and Road Initiative countries; Amin and Song (2023) [62] reported similar findings for South Asia, as did Sun et al. [63] for MENA countries. In a study on 130 countries from 1992 to 2019, Li et al. [64] concluded that there is a negative relationship between renewable energy use and carbon emissions; when renewable energy consumption increases, the negative effect on carbon emissions becomes more significant. Furthermore, the effect is stronger in poor countries than in rich countries.

2.3. Economic Growth and Carbon Emissions

Economic growth is positively associated with an increase in carbon emissions [11,13,32,42,44,55,56,57,63,65,66,67,68,69,70,71]. GDP per capita and urbanisation growth determined environmental degradation in Asian Association for Regional Cooperation (SAARC) countries [50], and also in Bangladesh [23]. Economic growth caused an increase in carbon emissions in APEC economies over the period 2000–2019 [72]. Contrary to these results, recent studies have revealed that economic development is negatively linked with carbon emissions globally, in Europe, Africa, and Asia [73], and also in Singapore [74]. Developed and developing countries are following different paths in terms of economic development and environmental quality. Economic growth and financial development alleviate carbon emissions in high-income economies, but induce the opposite effect in middle- and low-income economies [75].

2.4. Other Variables with an Impact on Carbon Dioxide Emissions

2.4.1. Financial Development

The relevant literature regarding the nexus between financial development and carbon emissions provides mixed results. Some studies report a positive correlation, and others a negative one; some reveal dual effects of financial development on carbon dioxide emissions.

Financial development is seen as a determinant of carbon emissions [65,66,76]. Financial development reduced CO₂ emissions in MENA countries during the period of 1970 to 2015 [27]. Kirikkaleli and Adebayo [55] found that global financial development has a significant positive effect on environmental quality through decreasing the carbon emissions. Guo and Hu [77] reported similar results in the case of the Chinese economy. Sheraz et al. [69] report that financial development decreased carbon emissions in G20 countries from 1986 to 2018. Financial development decreased CO₂ emissions in the long and short term in the Nordic countries (1980–2020), according to Wu et al. [58]. Hung [78] found that financial development is an important factor of environmental degradation, and a decrease in CO₂ emissions could predict negative financial development in Vietnam. Ahmad et al. [79] concluded that human capital can reduce the effect financial development has in increasing carbon emissions. Khan and Ozturk [80] provided evidence in support of the pollution-inhibiting role of financial development within a sample consisting of 88 developing countries during 2000–2014.

A significant positive effect of financial development on carbon dioxide emissions was identified in Nigeria [67], in G8 and D8 countries [81], in the South Asian Association for Regional Cooperation (SAARC) countries [50], in 64 Belt and Road Initiative (BRI) countries [61], and in EU countries [14]. Acheampong et al. [82] proved that financial development reduced carbon emissions within the period of 1980–2015 in developed countries, and had the opposite effect on developing countries. Similarly, Jiang and Ma [83] found that in developing economies, financial development has a positive influence on carbon dioxide emissions, and no obvious effect in developed economies.

Liu et al. [84] showed that financial development had dual effects on carbon dioxide emissions in BRI countries within the period of 1997–2019: a restraining effect and a rebounding effect. The restraining effect decreases with time, leading to a blocking point. Financial development led to decrease carbon emissions in eighteen APEC countries from 2000 to 2019, according to the results of Hasni et al. [72]. The same result was reported by Patel and Mehta [22] in the case of the Indian economy.

Contrary to these findings, Rahman et al. [17] found no significant link between financial development and carbon emissions for CEE countries over the period of 1980–2016, similar to the results of Ayeche et al. [18] for 40 European countries in the period 1985 to 2014.

2.4.2. Urbanisation

Urbanisation was revealed to be a stimulating factor of carbon emission in Argentina [51] and Bangladesh [23], in SAARC countries [50], in the MENA region [63], and in Singapore [74]. Urbanisation was found to have increased CO₂ emissions in the long and short run in South Asia [62]. Differing from these conclusions, Aladejare [30] reported a positive impact of urbanisation on environmental quality in the five richest African countries, as did Lv and Xu [85] for middle-income countries, and Kwakwa et al. [86] for Ghana.

2.4.3. Foreign Direct Investment

Foreign direct investment (FDI) caused more pollution in MENA countries from 1970 to 2015 [27], whereas other studies revealed that foreign direct investment has a negative effect on carbon dioxide emissions [12,87,88]. Similarly, Horobeț et al. [15] concluded that inward FDI positively impacted reductions in carbon dioxide emissions in 24 EU countries, as did Jahanger et al. [33] for 73 developing economies. Abdul-Mumuni et al. [89] revealed an asymmetrical link between FDI and carbon emissions in the long run: a positive shock of FDI will lead to an increase in carbon emissions, while a negative shock will induce a decrease. FDI had a positive impact on carbon emissions in Ghana in 1971–2018 [86], and in Italy for the period 1971–2019 [71]. The study developed by Wang et al. [90] for 67 countries shows that FDI has a positive impact on carbon dioxide emissions for countries with a GDP per capita lower than 541.87 USD, and a negative impact when GDP per capita exceeds this level.

We can observe from the above discussion that there is a scarcity of studies on the European area, and also that European countries have been less investigated regarding the nexus between globalisation, renewable energy, and carbon dioxide emissions, in the presence of other pollution stimulating factors such as financial development, urbanisation, and foreign direct investment. The present study intends to cover this gap by building on the literature focused on globalisation and the environment in the European Union, given the need for effective energy policies, to overcome climate change and environmental degradation issues, that are conducive to achieving the ambitious goal of a decarbonised economy by 2050.

3. Data and Methodology

3.1. Data

The data used in this study cover the period 1990 to 2020 for 26 European Union countries (Austria, Belgium, Bulgaria, Croatia, Czech Republic, Cyprus, Denmark, Estonia, Germany, Greece, Hungary, Finland, France, Ireland, Italy, Latvia, Lithuania, Luxembourg, Netherlands, Poland, Portugal, Romania, Spain, Slovakia, Slovenia, and Sweden).

The study was motivated by our existing relevant theoretical foundation and empirical analyses, the selected variables being consistent with them. Table 1 displays the study’s variables, measurement, and their source.

The dependent variable is carbon dioxide emissions per capita (in metric tons), as a proxy for air pollution, meaning emissions stemming from fuels (fossil, liquid and gas), consistent with Soaib et al. [81], Hafeez et al. [91], Gyamfi et al. [92], Muhammad and Khan [93], Muhammad et al. [94], Sun et al. [95], Zhuo and Qamruzzaman [96], and Jiang et al. [97].

The independent variables are globalisation, gross domestic product per capita and renewable energy consumption as a variable of interest; financial development, foreign direct investment, and urbanisation are the control variables.

The KOF index of globalisation, as a measure of globalisation, was introduced by Dreher (2003) at the KOF Swiss Economic Institute; it has since been updated in 2008, 2018, and 2019 [98,99,100,101]. The study uses the overall index of globalisation (including all dimensions: economic, social, and political) [6]. This variable was employed in previous studies focusing on the link between globalisation and pollution [23,43,49,50,60,93].

Gross domestic product (GDP) per capita is the proxy for economic growth in our study. GDP per capita based on purchasing power parity (PPP) is the economic output (in international dollars, using purchasing power parity rates) divided among the population, and it reflects the state of development of a given country. The nature of its relationship with carbon emissions (negative or positive) depends on the share of energy from renewable sources in the energy consumption mix required for economic development.

Renewable energy consumption is expressed as the share of energy from renewable sources (solar, wind, hydropower, geothermal, bioenergy and marine energy) among the total final energy consumption. We expect a negative relationship between carbon dioxide emissions and renewable energy consumption (i.e., renewable energy use induces a reducing effect of environmental degradation). This metric for renewable energy consumption was also used by: Rahman and Alam [32]; Wang et al. [45]; Sadiq et al. [49], Kirikkaleli and Adebayo [55]; Kwakwa [59], Amin and Song [63], and Li et al. [64].

Financial development is a multidimensional process influencing all of society, including evolving financial institutions and markets. Our study uses the financial development indicator developed by International Monetary Fund (IMF) [102,103]. The IMF index is defined by a combination of the characteristics of financial institutions (access and efficiency) and financial markets (size and liquidity) [102]. Several studies analysing the impact of financial development on environmental sustainability work with this index [12,14,67,72,79,82,104].

As a variable impacting pollution in the EU countries, foreign direct investment is measured by the net inflow of foreign investment as percentage of GDP. This measure is present in the empirical analyses of factors influencing carbon emissions [15,71,89,105,106].

Urbanisation, as a potential factor influencing CO₂ emissions’ increase, is measured by the percentage of the total population made up by urban populations. This indicator has been used in related studies on the link between globalisation and carbon dioxide emissions, such as those by Islam et al. [23], Aladejare [30], Murshed et al. [41], Azam et al. [50], Kwakwa et al. [86] Amin and Song [62], and Sun et al. [63].

Table 1. Variables: symbol, measurement, and source.

Symbol	Variable	Measurement	Source
Dependent Variable
CO2	Carbon dioxide emissions	Metric tons per capita	World Bank [107]
Variables of interest
GDPpc	Gross Domestic Pro- duct per capita	Gross Domestic Product per capita on Purchasing Power parity (PPP) constant international 2017 international USD)	World Bank [108]
KOF	Globalisation	Overall Globalisation Index Overall	KOF Swiss Economic Institute [6]
REC	Renewable Energy Consumption	Share of renewable energy in the total final consumption (%)	World Bank [109]
Control variables
FinDev	Financial Development	Financial Development Index: an aggregate of Financial Institutions Index (Banking sector) and Financial Markets Index (market capitalisation)	International Monetary Fund (IMF) [103]
FDI	Foreign Direct Investment	Foreign Direct Investment Inflows, as % of GDP	World Bank [110]
URB	Urbanisation	Share of urban population in the total population (%)	World Bank [111]

Table 1. Variables: symbol, measurement, and source.

Symbol	Variable	Measurement	Source
Dependent Variable
CO2	Carbon dioxide emissions	Metric tons per capita	World Bank [107]
Variables of interest
GDPpc	Gross Domestic Pro- duct per capita	Gross Domestic Product per capita on Purchasing Power parity (PPP) constant international 2017 international USD)	World Bank [108]
KOF	Globalisation	Overall Globalisation Index Overall	KOF Swiss Economic Institute [6]
REC	Renewable Energy Consumption	Share of renewable energy in the total final consumption (%)	World Bank [109]
Control variables
FinDev	Financial Development	Financial Development Index: an aggregate of Financial Institutions Index (Banking sector) and Financial Markets Index (market capitalisation)	International Monetary Fund (IMF) [103]
FDI	Foreign Direct Investment	Foreign Direct Investment Inflows, as % of GDP	World Bank [110]
URB	Urbanisation	Share of urban population in the total population (%)	World Bank [111]

Table 2. Descriptive statistics of variables.

	lnCO2	lnFinDev	lnFDI	lnGDPpc	lnKOF	lnRENE	lnURB
Mean	1.984917	−0.804342	0.781727	10.34840	4.334920	2.362099	4.239015
Median	1.989818	−0.692153	0.879344	10.40894	4.380615	2.450573	4.230368
Maximum	3.413184	−0.104631	4.313947	11.70063	4.509831	3.968025	4.585386
Minimum	1.074004	−4.806411	−10.61131	9.169518	3.721731	−1.094043	3.869429
Std. Dev.	0.418699	0.539448	1.338030	0.504522	0.148314	0.943458	0.168917
Observations	806	806	806	806	806	806	806

Source: authors’ computation based on EViews 12.0 software.

provides a summary of the descriptive statistics of the variables used. All variables are in ln. Over the period (1990–2020), we note that overall standard deviations are generally low. The highest value is recorded by lnGDP per capita (10.34840), while the minimum value is registered for lnFDI (−10.61131).

3.2. Methodology and Econometric Strategy

3.2.1. Methodology

Following previous research on the impact of globalisation and renewable energy on carbon dioxide emissions [63,112], the mathematical representation of our model is as follows:

$${ln{CO}_2}_{i,t}=\alpha +{\beta }_1·{lnGDP}_{pci,t}+{\beta }_2·{lnKOF}_{i,t}+{\beta }_3·{lnREC}_{i,t}+{\beta }_4·{CV}_{i,t}+{\epsilon }_{i,t}$$

(1)

where i denotes the country and t the time, respectively; CO₂ represents the carbon emissions per capita, GDPpc is the gross domestic product per capita, KOF denotes the KOF globalisation index (overall), and REC stands for the renewable energy consumption (as a % of the total final consumption). CV expresses a set of control variables: financial development (FinDev), foreign direct investment (FDI) and urbanisation (URB). ${\beta }_1$, ${\beta }_2$, ${\beta }_3$, ${\beta }_4$ are coefficients to be estimated, and ${\epsilon }_{i,t}$ is the stochastic error term.

Equation (1) is estimated in ln in order to minimise variations in variables under consideration. The control variables will be successively added, one by one, thus generating additional models.

We propose to test the following hypotheses in our study: (H1) globalisation and renewable energy consumption have a negative impact on carbon emissions; (H2) economic growth has an the effect of increasing carbon emissions; (H3) the level of economic development can differentiate the impact of globalisation, renewable energy, and economic growth on carbon emissions; (H4) the effect of globalisation, renewable energy and economic growth on carbon emissions is differentiated by institutional quality.

3.2.2. Econometric Strategy

The methodology consists of the following steps: (1) a cross-sectional dependence test; (2) a stationarity check of the considered variables; (3) a cointegration test; (4) an estimation of the long-term coefficients of FMOLS and DOLS models; (5) a robustness check of the results; (6) a test of the causality between variables.

Cross-Sectional Dependence

Usually, cross-sectional dependence in the data series can be detected due to unobserved mutual factors, spillover effect, or common shock. If cross-sectional dependence is detected, traditional unit root tests may provide bias outcomes. For reliable results, we used a Breusch–Pagan LM test [113], a Pesaran Scaled LM test, a bias-corrected scaled LM test, and a Pesaran CD test [114]. The null hypothesis of cross-sectional dependence is accepted in the case that the p value is lower than 0.01. The statistic of Pesaran CD is given by the following equation:

$$D=\sqrt{\frac{2}{N(N-1)}}\ast \left(\sum _{i=1}^{N-1}\sum _{j=\mathrm{i}+1}^N{T}_{ij}{p}_{ij}\to N(0.1)\right)$$

(2)

where $p_{ij}$ stands for the correlation coefficients of residuals, N denotes the number of countries, and T the time.

Stationarity of Variables

Before applying the cointegration test and estimation techniques, we intend to identify the integrated properties of the variables under consideration. We propose to use second-generation panel unit root tests that are based on the heterogeneity assumption, thus avoiding the shortcoming of the cross-sectional dependence of first-generation unit root tests. We apply two types of second-generation panel unit root tests, namely those proposed by Pesaran [115]: the cross-sectional augmented Dickey–Fuller (PES-CADF) test and the cross-sectional Im, Pesaran and Shin (CIPS).

The CADF test consists of standard Dickey–Fuller (DF) regressions that are augmented with the cross-sectional average of lagged series at their levels, and the first difference series of the i-th cross-section in the panel are run as follows:

$$∆y_{it}={\alpha }_i+{\rho }_i{y}_{i,t-1}+{\delta }_i{y}_{t-1}+\sum _{j=0}^k{\delta }_{ij}∆y_{i,t-j}+\sum _{j=0}^k∆y_{i,t-j}+{\epsilon }_{it}$$

(3)

where $y_{t-1}=\frac{1}{N}\sum _{i=1}^N{y}_{i,t-1}$; $∆y_t=\sum _{i=1}^N{y}_{it}$; ${\alpha }_i$ is a constant; k specifies the lag; and $t_i$ (N,T) is the t-statistic of the estimated ${\rho }_i$.

CIPS is computed as the mean of the individual CADF statistics for individual cross-sections:

$$CIPS=\frac{1}{N}\sum _{i=1}^N{t}_i(N,T)$$

(4)

where $t_i$ (N,T) denotes the CADF statistics for the i-th cross section in the CADF regression.

For both tests, the null hypothesis of homogeneous unit root states that ”all sections in the panel are nonstationary”, while under the alternative hypothesis, ”at least, one individual section is stationary”. Both tests are largely used in the literature regarding the impact of various economic, financial, or energy variables such as globalisation, economic growth, financial development, renewable/non-renewable energy on carbon emissions and environmental quality [104,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135].

Cointegration Test

When the cross-sectional dependence is identified, the error correction-based panel cointegration test, as proposed by Westerlund [136], provides robust results [137]. The test is currently used in the current literature focusing on the influence of economic and energy variables on pollution [13,30,31,45,51,62,73,79,104,138].

Within this test, the null hypothesis of no cointegration is based on checking the presence of a unit root of residuals. The alternative hypothesis under this test states that some panels are cointegrated based on computing the variance–ratio (VR) statistic. The test is based on two assumptions regarding the presence of the cointegration of variables (1) in some of the panels, and (2) in all the panels. Based on the p-value of the VR statistic, a rejection or acceptance of the null hypothesis of no cointegration is carried out. If the p-value is under the chosen significance level, the null hypothesis is rejected in favour of the alternative, that at least some panels or all panels are cointegrated.

Panel Model Estimation

We estimated Equation (1) using the panel models developed by Pedroni [139,140], namely fully modified ordinary least square (FMOLS) and dynamic ordinary least square (DOLS).

Within the panel FMOLS model, the following regression is used:

$$y_{it}={\alpha }_{it}+{\delta }_{it}t+\beta x_{it}+{\mu }_{it}$$

(5)

$$x_{it}=x_{it-1}+e_i$$

(6)

where $y_{it}$ represents the dependent variable and $x_{it}$ the independent variable, ${\alpha }_{it}$ denotes the constant effects, and β stands for the long-term cointegration coefficient that will be estimated under the assumption of no panel cross-sectional dependence.

The panel FMOLS estimator for the i-th cross-section is given by

$${\widehat{\beta }}_{FM}^{\ast }=n^{-1}{\displaystyle \sum _{i=1}^n{\widehat{\beta }}_{FM,i}^{\ast }}$$

(7)

The T-statistic for the panel cointegration coefficient is computed as follows:

$$t_{{\widehat{\beta }}_{FM}^{*}}^{*}=n^{-1}{\displaystyle \sum _{i=1}^n{t}_{{\widehat{\beta }}_{FM,i}^{*}}}$$

(8)

The DOLS estimator is the result of the following regression estimation:

$$y_{it}={\alpha }_i+{\beta }_i{x}_{it}+{\displaystyle \sum _{k=-K_i}^{K_i}{\gamma }_{it}}\Delta x_{it-K}+{\epsilon }_{it}$$

(9)

The above equation is estimated for each panel cross-section. Further, the cointegration coefficient of the overall panel is calculated as the average of the DOLS coefficients of each section.

The panel DOLS estimator is given by the following formula:

$${\widehat{\beta }}_D^{\ast }=n^{-1}{\displaystyle \sum _{i=1}^n{\widehat{\beta }}_{D,i}^{\ast }}$$

(10)

The t-statistic for the panel cointegration coefficient is computed as follows:

$$t_{{\widehat{\beta }}_D^{*}}^{*}=n^{-1}{\displaystyle \sum _{i=1}^n{t}_{{\widehat{\beta }}_{D,i}^{*}}}$$

(11)

The panel FMOLS and DOLS approach was also used by Sahoo and Sethi [141] for analysing the impact of renewable and non-renewable energy, globalisation, natural resources, and human capital on environmental quality in developing countries; this approach was also used by Kirikkaleli and Adebayo [55] for a global sample of countries. The FMOLS model was employed by Al-Mulali et al. [19] to explore the effect of economic growth, financial development, renewable energy, trade openness, and urbanisation on pollution in selected European countries, as well as by Wang et al. [45] to examine the linkage between CO₂ emissions, human development, financial development, and globalisation in OECD countries. This model was also used by Wang et al. [130] to analyse the impact of FDI on environmental sustainability in 67 countries.

Robustness Check of Results

The robustness check of our results will be performed in two parts: (1) adding the control variables to the variables of interest in Equation (1); and (2) dividing the panel of 26 EU countries into two subpanels according to two criteria (the level of economic development and the level of institutional quality).

The level of economic development creates the capacity of a country to address problems related to mitigating pollution, openness for trade and foreign investment, financial development, and diversifying the energy consumption mix. Based on this criterion, we divided the 26 EU countries in two subpanels, developed and developing countries, meaning a set of 15 Western developed countries (the old Member States of Austria, Belgium, Cyprus, Denmark, Germany, Finland, France, Greece, Italy, Ireland, Luxembourg, Netherlands, Spain, Portugal and Sweden) and 11 countries from Central and Eastern Europe (the Czech Republic, Poland, Slovakia, Slovenia, Estonia, Latvia, Lithuania, Hungary, Bulgaria, Romania and Croatia).

As institutions shape economic, energy, and environmental policies, it is crucial to incorporate institutional variables when analysing the nexus of CO₂ emissions, renewable energy, and economic growth to prevent variable omission bias. Institutional quality entails good governance, the rule of law, and the quality of bureaucracy or corruption. The role of institutional quality in mitigating pollution has been revealed in several papers as a moderating factor of the influence of various pollution determinants. As Kim et al. [142] highlighted, high institutional quality decreases energy use and carbon emissions. Ahmad et al. [79] concluded that institutional quality reduces the ecological impact of financial development in emerging countries. Simionescu et al. [143] found that institutional quality contributed to environmental quality and renewable energy consumption in the long term within the period of 2002–2008, by reducing the level of GHG emissions in CEE countries. Sheraz et al. [61] pointed out that institutional quality is related to carbon emissions, with a low level (bad governance, corruption, low quality of bureaucracy) being associated with environmental degradation. Corruption as a dimension of institutional quality was found to increase carbon emissions in Asian countries [32]. Institutional quality and governance positively influence renewable energy consumption [12]. Kwakwa [59] reports the same in the case of African countries based on data covering 2002–2021. Carbon emissions are substantially reduced by corruption control, regulatory quality, and the rule of law [144]. Institutional quality reduced CO₂ emissions in G-7 countries [145] and 45 sub-Saharan African countries [146]. Khan and Rana [147] revealed that better economic institutions helped in reducing pollution emissions in 41 Asian economies from 1996 to 2015, while institutional quality can moderate the negative impact of financial development on carbon emissions [148] in South Asian Economies. Islam et al. [149] found that institutional quality stimulates renewable energy consumption in Bangladesh. Jiang et al. [97] also revealed that improvements in institutional quality can curb carbon emissions using a panel of 57 Belt and Road (B&R) countries over the period of 1995 to 2018, in line with the results of Jahanger et al. [33] for 73 developing countries (from Asia, Africa and Latin America).

In order to differentiate the effect of globalisation and renewable energy consumption on carbon emissions in EU economies, based on the institutional quality of countries, we used indicators from the Worldwide Governance Indicators database [150]. It includes six components of governance: voice and accountability, political stability and lack of violence, rule of law, government effectiveness, regulatory quality, and control of corruption [151]. We computed the mean of these governance indicators for EU countries within the period of 1996–2020. We split the panel of EU 26 countries into two: 13 countries with a mean above 1 (meaning a high level of institutional quality) (Austria, Belgium, Cyprus, Germany, Denmark, Finland, France, Ireland, Luxembourg, Netherlands, Estonia, Portugal, and Sweden) and 13 countries with a mean under 1 (meaning a low level of institutional quality (Greece, Italy, Spain, Bulgaria, Croatia, Czech Republic, Latvia, Lithuania, Poland, Hungary, Romania, Slovakia and Slovenia).

This splitting is motivated by the fact that pollution, globalisation, economic growth, the energy mix (i.e., the share of energy that comes from renewable sources), financial development, and foreign direct investment are subjects of national policies, designed and implemented by institutions. Institutional quality is important for the adoption of renewable energy, the stimulation of green investment, and the effectiveness of environmental rules; it can stimulate or impede globalisation, financial development, and the level of urbanisation.

Panel Causality Test

In order to identify the direction of the causal relationship between the variables, the Dumitrescu and Hurlin [152] test will be used. This is appropriate for heterogeneous panel data and widely used in analyses of the impact of the economy on pollution [18,21,30,45,49,50,61,62,66,104,145,153].

Under the null hypothesis of no causality running from x to y, we have

$$\mathrm{H}0:{\beta }_i=0 \mathrm{for} \forall i=1,\dots ,n; {\beta }_i=({\beta }_i^{(1)},{\beta }_i^{(2)},\dots ,{\beta }_i^{(k)})$$

The alternative hypothesis assumes that there are $n_1$ < n individual processes with no causal relationship from x to y:

$$\mathrm{H}1: {\beta }_i=0 \mathrm{for} \forall i=1,\dots ,n_1$$

$${\beta }_i\ne 0\mathrm{for} \forall i=n_1,n_1+1,n_1+2,\dots ,n$$

where $0\le \frac{n_1}{n}$ < 1. When $n_1=n$, no causality is identified for any section in the panel.

When $n_1=0$, causality is identified for all sections in the panel. When $n_1>0$, the causality relationship is heterogeneous. The Dumitrescu–Hurlin test first computes the individual Wald statistics to identify the causality relationship in each section, and then computes the overall Wald statistic as their average:

$$W_{n,T}=\frac{1}{n}{\displaystyle \sum _{i=1}^n{W}_{i,T}}$$

(12)

The null hypothesis of non-causality states that each individual Wald statistic will converge to a Chi-squared distribution:

$$W_{i,T}\stackrel{d}{\to }{\chi }^2(K),\forall i=1,\dots ,n$$

(13)

where K = freedom degrees.

When T→ $\infty$, the individual Wald statistics are identically distributed, assuming that individual residuals are independently distributed across sections.

When $T<n$, the Z-statistic is computed as follows:

$$Z_{n,T}=\sqrt{\frac{n}{2K}}\left(W_{n,T}-k\right)\to n\left(0,1\right)$$

(14)

when the value of the Z-statistic is above the critical value of a given risk level, the null hypothesis of homogeneous non-causality is rejected.

4. Main Findings

4.1. Cross-Sectional Dependence Test

The results displayed in

Table 3. Results of the cross-sectional dependence test.

	lnCO2	lnGDPpc	lnKOF	lnREC	lnFinDev	lnFDI	lnURB
Breusch-Pagan LM	3417.781 *	8177.36 *	7100.18 *	5708.25 *	5573.02 *	1026.37 *	6614.57 *
Pesaran Scaled LM	121.308 *	305.64 *	265.74 *	211.14 *	205.84 *	27.51 *	246.69 *
Bias-Corrected Scaled LM	120.86 *	305.19 *	265.29 *	210.70 *	205.39 *	27.06 *	246.24 *
Pesaran CD	43.66 *	89.30 *	80.66 *	72.37 *	69.99 *	21.22 *	16.98 *

Note: * p < 0.01; Source: authors’ computation based on EViews 12.0 software.

suggest that the null hypothesis of cross-sectional independence is rejected for the 1% significance level for all the variables under consideration, indicating the presence of cross-sectional dependence among variables.

As a consequence, we used second-generation unit root tests for the levels and the first-order differences of variables, namely CIPS and CADF.

4.2. Stationarity Test

We notice from

Table 4. Stationarity test results.

Variable	PES-CADF Test		CIPS Test
	z (t-Bar)		CIPS Statistic
	Constant	Constant and Trend	Constant	Constant and Trend
lnCO2	−2.155 **	−2.097	−2.238 **	−2.409
ΔlnCO2	−3.657 *	−3.971 *	−4.711 *	−5.127 *
lnGDPpc	−2.142 **	−2.492	−2.031 *	−2.441 *
ΔlnGDPpc	−3.200 *	−3.220 *	−3.922 *	−3.916 *
lnKOF	−2.011	−2.186	−2.456	−2.774
ΔlnKOF	−3.395 *	−3.848 *	−4.996 *	−5.391 *
lnREC	−2.624 *	−2.741 *	−2.164 **	−2.431 **
ΔlnREC	−3.554 *	−3.833 *	−4.946 *	−5.242 *
lnFinDev	−2.966 *	−3.277 *	−2.971 *	−3.390 *
ΔlnFinDev	−4.584 *	−4.565 *	−5.402 *	−5.524 *
lnFDI	−3.285 *	−3.535 *	−4.735 *	−5.022 *
ΔlnFDI	−5.270 *	−5.304 *	−6.186 *	−6.402 *
lnURB	−1.536	−1.996	−0.865	−0.801
ΔlnURB	−2.158 **	−2.305 **	−2.028 *	−2.305 *

Note: * p < 0.01; ** p < 0.05; Source: authors’ computation based on Stata 15.1 software.

that lnREC, lnFinDev, lnFDI are stationary at their levels and also at their first difference, while lnCO₂, lnGDPpc, lnKOF and lnURB are stationary only their first difference for a minimum of 5% statistical significance. Thus, we can conclude that all variables are integrated at their first order (I (1)).

4.3. Cointegration Test

The results of the Westerlund cointegration test for the EU panel (

Table 5. Westerlund cointegration test.

Variables	Assumptions:
	Some Panels Are Cointegrated		All Panels Are Cointegrated
	Statistic	p-Value	Statistic	p-Value
LnCO2, LnGDPpc, lnKOF, lnREC	−2.751	0.0030	−1.804	0.0329
LnCO2, LnGDPpc, lnKOF, lnREC, lnFinDev,	−2.298	0.0108	−1.466	0.0500
LnCO2, LnGDPpc, lnKOF, lnREC, lnFDI	−1.8131	0.0349	−1.346	0.0500
LnCO2, LnGDPpc, lnKOF, lnREC, lnURB	−3.394	0.000	−1.910	0.0281

Source: authors’ computation based on Stata 15 software.

) reveal the existence of a long-term relationship between the considered variables, at 5% significance. The cointegration relationship is maintained when the control variables are added (rows 2–4 from Table 5).

Given the identified cointegration relationship between the considered variables, we further proceeded to estimate the long-run coefficients of Equation (1).

4.4. Long—Run Coefficients Estimation

We ran the FMOLS and DOLS estimations for four models by successively adding, one by one, the control variables (lnFinDEv, lnFDI, lnURB) to our interest variables (lnGPDpc, lnKOF, lnREC). The results are displayed in

Table 6. Regression estimation using the EU panel (26 countries). Dependent variable: lnCO₂.

	Model 1		Model 2		Model 3		Model 4
	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS
Variables	-Coefficients-
lnGDPpc	0.5310 *	0.5317 *	0.5596 *	0.5558 *	0.5244 *	0.5279 *	0.5065 *	0.4987 *
lnKOF	−0.7078 *	−0.7096 *	−0.7821 *	−0.7735 *	−0.7023 *	−0.7075 *	−0.9162 *	−0.9184 *
LnREC	−0.1883 *	−0.1870 *	−0.1892 *	−0.1879 *	−0.1808 *	−0.1823 *	−0.1776 *	−0.1753 *
lnFinDev			−0.0377 *	−0.0378 *
lnFDI					0.0323 *	0.0244 *
lnURB							0.2674 *	0.2878 *
R-squared	0.4904	0.4848	0.4926	0.4870	0.4984	0.4908	0.5017	0.4975

Source: authors’ computation based on EViews 12.0 software. Notes: * p < 0.01; Model 1 includes the variables of interest; In Models 2, 3, and 4, respectively, control variables are added, one by one.

.

Economic growth induced an increase in carbon emissions, the coefficient of lnGDPpc being statistically significant at a 1% significance threshold in all four models. Globalisation has a negative significant impact on carbon emissions alongside renewable energy use (hence the value of Prob. for all coefficients is less than 0.01). We can thus confirm the first and the second hypotheses of our study. Moreover, financial development caused a reduction in carbon dioxide emissions, while foreign direct investment and urbanisation led to an increase in carbon dioxide emissions (for a threshold statistical significance of 1%) (Table 6).

4.5. Robustness Analysis

We conducted an additional analysis in order to check the robustness of our findings. This analysis has two components. The first one refers to adding three control variables: the financial development index (FinDev), foreign direct investment (FDI), and urbanisation (URB). The cointegration relationship is also validated when the control variables are added, as shown in

Table 7. Regression estimation using a panel of Western and developed EU countries (Austria, Belgium, Cyprus, Denmark, Germany, Finland, France, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, and Sweden). Dependent variable: lnCO₂.

	Model 1		Model 2		Model 3		Model 4
	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS
Variables	-Coefficients-
lnGDPpc	0.7386 *	0.7389 *	0.7505 *	0.7433 *	0.7146 *	0.7193 *	0.7209 *	0.7162 *
lnKOF	−1.2446 *	−1.2460 *	−1.2854 *	−1.2642 *	−1.1915 *	−1.203 *	−1.3995 *	−1.3703 *
LnREC	−0.1353 *	−0.1337 *	−0.1334 *	−0.1324 *	−0.1309 *	−0.1299 *	−0.1300 *	−0.1289 *
lnFinDev			−0.1086 *	−0.0653 **
lnFDI					0.0173 *	0.0148 *
lnURB							0.1992 *	0.1801 *
R-squared	0.5634	0.5564	0.5677	0.7433	0.5653	0.5589	0.5699	0.5612

Note: * p < 0.01; ** p < 0.05. Source: authors’ computation based on EViews 12.0 software.

(rows 3–4). In Table 7,

Table 8. Regression estimation panel of Central and Eastern (developing) EU countries (Bulgaria, Estonia, Czech Republic, Croatia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia). Dependent variable: lnCO₂.

	Model 1		Model 2		Model 3		Model 4
	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS
Variables	-Coefficients-
lnGDPpc	0.6074 *	0.5846 *	0.6154 *	0.5937 *	0.6365 *	0.6163 *	0.5838 *	0.5408 *
lnKOF	−0.7845 *	−0.7356 *	−0.7860 *	−0.7419 *	−0.8644 *	−0.8135 *	−0.9502 *	−0.8708 *
LnREC	−0.3391 *	−0.3314 *	−0.3342 *	−0.3308 *	−0.3395 *	−0.3396 *	−0.3219 *	−0.3101 *
lnFinDev			0.0682 *	0.0518 *
lnFDI					0.0567 *	0.0443 *
lnURB							0.2169 *	0.2310 *
R-squared	0.3793	0.3837	0.3846	0.3879	0.4007	0.4018	0.3884	0.3935

Note: * p < 0.01. Source: authors’ computation based on EViews 12.0 software.

,

Table 9. Regression estimation using a panel of high-institutional-quality EU countries (Austria, Belgium, Cyprus, Denmark, Germany, Finland, France, Ireland, Luxembourg, Netherlands, Portugal, Sweden, and Estonia). Dependent variable lnCO₂.

	Model 1		Model 2		Model 3		Model 4
	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS
Variables	-Coefficients-
lnGDPpc	0.5638 *	0.5537 *	0.7201 *	0.7506 *	0.5441 *	0.4496 *	0.51474 *	0.5085 *
lnKOF	−0.8123 *	−0.7632 *	−1.253 *	−1.2990 *	−0.8013 *	−0.5334 *	−1.1728 *	−1.1754 *
LnREC	−0.1172 *	−0.1666 *	−0.1193 *	−0.1439 *	−0.0919 *	−0.1573 *	−0.1094 *	−0.1031 *
lnFinDev			−0.5428 *	−0.3850 *
lnFDI					0.0718 *	0.0877 *
lnURB							0.4010 *	0.4477 *
R-squared	0.3348	0.7855	0.5221	0.8366	0.3475	0.8234	0.3680	0.373

Note: * p < 0.01. Source: authors’ computation based on EViews 12.0 software.

and Table 10 the columns Model 2, Model 3 and Model 4, respectively, include the estimation of regression parameters including these control variables. As observed, the correlation between the variables of interest (GDPpc, KOF and REC) and CO₂ emissions is maintained. Economic growth induced an increase in CO₂ emissions, while globalisation and renewable energy use caused their reduction in the panel of 26 countries. Financial development (FinDev) we found to be a reducing factor of carbon emissions, while urbanisation and FDI contributed to their increase.

The second component of the robustness analysis consisted of running the regression model within two types of sub-panels of the EU countries, based on their level of economic development and their level of institutional quality, respectively.

Table 7 and Table 8 depict the results of panel estimation for the developed/developing countries subpanels. The negative impact of globalisation on carbon dioxide emissions is stronger in the developed EU economies than in the overall panel, and also in the developing countries subpanel. The values of regression coefficients are 0.4–0.5 units higher than those in the EU panel. In the case of renewable energy use, the effect on carbon dioxide emissions is higher in the subpanel of developing countries than in the developed countries and the overall panel. The values of regression coefficients are above EU levels. The impact of economic growth on the increase in carbon dioxide emissions is higher in the developed countries compared to the overall EU panel and developing countries subpanel. Thus, the third hypothesis of our study is also confirmed. Financial development has a different impact on carbon emissions depending on the level of economic development. In the overall panel and developed countries, the impact is negative, while in developing countries, it induces pollution. Urbanisation and foreign direct investment have positive influences on carbon dioxide emissions, and the effect is lower in developed countries. Developing countries face higher levels of carbon emissions due to these variables compared to developed countries.

Regarding the role of institutional quality in differentiating the impact of globalisation, renewable energy, and economic growth on carbon dioxide emissions, we noticed that the results are in some way similar to those of the level of economic development. The results are displayed in Table 9 and Table 10.

Table 10. Regression estimation panel of low-institutional-quality EU countries (Greece, Italy, Spain, Bulgaria, Croatia, Czech Republic, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia). Dependent variable: lnCO₂.

	Model 1		Model 2		Model 3		Model 4
	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS
Variables	-Coefficients-
lnGDPpc	0.3761 *	0.4123 *	0.3622 *	0.2213 *	0.3973 *	0.4038 **	0.2350 *	0.3224 *
lnKOF	−0.2617 *	−0.3571 **	−0.2202 *	−0.1582 **	−0.3096 *	−0.3352 **	−0.0848 **	−0.2401 *
LnREC	−0.3512 *	−0.3401 *	−0.3518 *	−0.2749 *	−0.3593 *	−0.3413 *	−0.2926 *	−0.3469 *
lnFinDev			0.0352 *	0.0569 **
lnFDI					0.0165 *	0.0165
lnURB							0.9284 *	0.7195 ***
R-squared	0.5930	0.9306	0.5955	0.9137	0.5951	0.9821	0.9291	0.9929

Note: * p < 0.01; ** p < 0.05; *** p < 0.1. Source: authors’ computation based on EViews 12.0 software.

Table 10. Regression estimation panel of low-institutional-quality EU countries (Greece, Italy, Spain, Bulgaria, Croatia, Czech Republic, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia). Dependent variable: lnCO₂.

	Model 1		Model 2		Model 3		Model 4
	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS	FMOLS	DOLS
Variables	-Coefficients-
lnGDPpc	0.3761 *	0.4123 *	0.3622 *	0.2213 *	0.3973 *	0.4038 **	0.2350 *	0.3224 *
lnKOF	−0.2617 *	−0.3571 **	−0.2202 *	−0.1582 **	−0.3096 *	−0.3352 **	−0.0848 **	−0.2401 *
LnREC	−0.3512 *	−0.3401 *	−0.3518 *	−0.2749 *	−0.3593 *	−0.3413 *	−0.2926 *	−0.3469 *
lnFinDev			0.0352 *	0.0569 **
lnFDI					0.0165 *	0.0165
lnURB							0.9284 *	0.7195 ***
R-squared	0.5930	0.9306	0.5955	0.9137	0.5951	0.9821	0.9291	0.9929

Note: * p < 0.01; ** p < 0.05; *** p < 0.1. Source: authors’ computation based on EViews 12.0 software.

Globalisation has a stronger negative effect on carbon dioxide emissions in high-institutional-quality countries than in the EU panel and the low-institutional-quality countries subpanel. In the case of renewable energy consumption, the subpanel of low-institutional-quality countries showed a greater negative impact on carbon dioxide emissions. The positive impact of economic growth on carbon dioxide emissions is greater in countries with high institutional quality than in the EU panel and in the subpanel of low-institutional-quality countries. These findings validate the fourth hypothesis on the differentiating role of institutional quality in the relationships between carbon emissions and globalisation, economic growth, and renewable energy. In the subpanel of high-institutional-quality countries, financial development has a larger negative impact on carbon dioxide emissions than in the EU panel, while in the subpanel of low-institutional-quality countries, this effect was not validated. Urbanisation induced more carbon dioxide emissions in the low-institutional-quality countries than in the panel of countries of high institutional quality and the overall panel.

4.6. Panel Causality Test

The results of the panel causality test are displayed in

Table 11. Results of the Dumitrescu and Durlin causality test.

Null Hypothesis (H0)	z-Bar	p-Value	Conclusion
LnCO2 does not Granger-cause lnKOF	7.2026	0.000	KOF→CO2
lnKOF does not Granger-cause lnCO2	1.5211	0.1282
LnCO2 does not Granger-cause lnGDPpc	10.2093	0.000	GDPpc→CO2
lnGDPpc does not Granger-cause lnCO2	6.3591	0.000	CO2→GDPpc
LnCO2 does not Granger-cause lnREC	5.1745	0.000	REC→CO2
lnREC does not Granger-cause lnCO2	17.8105	0.000	CO2→REC
lnCO2 does not Granger-cause lnFinDev	4.8424	0.000	FinDev→CO2
lnFinDev does not Granger-cause lnCO2	2.0653	0.0389	CO2→FinDev
lnCO2 does not Granger cause lnFDI	0.3700	0.7114
lnFDI does not Granger cause lnCO2	3.8457	0.0000	CO2→FDI
lnURB does not Granger-cause lnCO2	12.4023	0.000	CO2→URB
lnCO2 does not Granger-cause lnURB	34.2891	0.000	URB→CO2

Note: lag = 1; Source: authors’ computation based on Stata 15 software.

. Bidirectional causal relationships were identified between carbon emissions and GDP per capita, renewable energy consumption, financial development, and urbanisation. Unidirectional causalities running from globalisation to carbon emissions and from carbon emissions to FDI are also validated.

The identified causality from globalisation to carbon dioxide emissions is consistent with the findings of Zaidi et al. [26] for APEC countries, and Sheraz et al. [61] for BRI countries, as well as Wang et al. [45] for OECD countries. Ayeche et al. [18], Amin and Song [63], Bosah et al. [73], Habiba et al. [104], and Jiang and Yu [153] also revealed a bidirectional causality between economic growth and carbon dioxide emissions. Our conclusion regarding the bidirectional influence of renewable energy on carbon emissions is in line with the findings of Amin and Song [62] for South Asian countries and Bosah et al. [73] and Sheraz et al. [61] for a global sample of countries. The bidirectional causality between financial development and carbon dioxide emissions has also been highlighted by Azam et al. [50]. The result of the bidirectional causality between urbanisation and carbon emissions is consistent with the conclusions of Azam et al. [50] and Amin and Song (2023) [62].

5. Discussion

Globalisation and renewable energy use were found to be reducing factors of carbon dioxide emissions in the overall panel of 26 EU countries, and also in the two sub-panels (developed/developing and high-/low-institutional-quality countries). This is in line with findings of Adebayo et al. [138] for BRICS countries and also of Addai et al. [11] for the relationship between globalisation and environmental quality in Germany. It should also be mentioned that the same relationship between renewable energy use and carbon emissions was identified by Ansari et al. [31], Alola and Joshua [34], Murshed et al. [41] Saidi and Omri [52], Jebli et al. [53], Khan et al. [54], Kirikkaleli and Adebayo [55], Adebayo et al. [57], Wu et al. [58], Amin and Song [62], Sun et al. [63], Li et al. [64], Salahuddin et al. [65], and Acheampong et al. [146]. Globalisation’s effect of decreasing carbon dioxide emissions is weaker in developing and low-institutional-quality countries than in the overall panel and in developed and high-institutional-quality countries. This suggests that is more specific efforts and effective measures (i.e., carbon taxation, strict environmental rules) need to be designed and applied by strong institutions in order to mitigate pollution in these countries. On the other hand, institutional policies should take into consideration that the globalisation process is not always targeted towards pollution mitigation; between countries, the risk of the transfer of pollution-intensive operations through business globalisation processes is real. We found renewable energy consumption’s effect of decreasing carbon dioxide emissions is stronger in developing countries than in developed ones. It was pointed out that renewable energy use could be beneficial for the environment in these countries, putting forward the idea that further efforts must be considered in order to continue the extension of renewable sources into the energy consumption mix. It should be noted that 6 of these 11 countries (Latvia, Estonia, Croatia, Lithuania, Romania and Slovenia) recorded higher levels of the share of renewable energy in their final energy consumption statistics (23–42%); these levels are above the European Union average, according to Eurostat data for 2004–2021 [154]. It is also true, that in the subpanel of Western developed countries, five of them (Austria, Denmark, Portugal, Finland and Sweden) recorded very high levels of renewable energy in their final consumption statistics (34–62%) from the same period of time. In countries with low institutional quality, the carbon emissions-reducing effect of renewable energy use is greater than in countries with high institutional quality, suggesting that institutional quality has not yet played a decisive role in the expansion of renewable energy use.

Economic growth induced an increase in carbon emissions in the panel of EU countries and all sub-panels. This is consistent with the majority of studies focusing on this nexus [11,42,44,47,50,55,57,63,67,69,70,71,129,147]). These findings suggest, once more, the need to decouple economic growth from pollution in the European Union. The structure of the economy must be clearly shifted from traditional to clean energy sources (solar, wind hydropower, biomass, and geothermal), and appropriate energy policies must be shaped in support of energy intensity, efficiency, and ongoing efforts to extend the funding of R&D activities in the field of clean technologies.

Financial development alleviates carbon dioxide emissions according to the overall panel of the EU countries, a result that is in line with the findings of Vatamanu and Zugravu [12], Khan and Ozturk [80], Kim et al. [142], and Khan and Rana [147]. This contradicts the results of Horobeț et al. [14] for EU countries, showing a positive impact of financial development on carbon dioxide emissions. In the subpanel of developing EU countries, financial development has a positive and significant impact on carbon dioxide emissions, as found by Jiang et al. [83] and Acheampong et al. [82]. The same effect of financial development on carbon dioxide emissions is identified in the subpanel of low-institutional-quality EU countries. The alleviating effect of financial development is stronger in the subpanel of developed and high-institutional-quality countries, suggesting that (1) a low level of income and financial development increases carbon dioxide emissions, while a high level of income decreases them, as it revealed by Ehigiamusoe and Lean [75]; and (2) institutional quality can moderate the effect of financial development on carbon dioxide emissions [79,142,148].

Foreign direct investment was found to be stimulating factor of carbon emissions in the overall panel and all subpanels. This result is consistent with previous relevant studies [15,27,86,90,147]. The detrimental effect of foreign direct investment on environmental quality is higher in the EU’s developing countries than in the overall panel and the subpanel of developed countries, as highlighted by Jahanger et al. [33]. This suggests the need for more effective European policies to stimulate the attraction of green foreign investment and investment in sectors with low-carbon energy sources, as well as more restrictive legal directives on environmental degradation. It is also worth mentioning that, as Wang et al. [90] noticed, low-income countries should promote economic development in order to gain the capacity to alleviate increasing pollution due to inward FDI flow, while strengthening environmental regulation.

Urbanisation increased carbon dioxide emissions in the overall EU panel and all sub-panels, as was found by Al-Mulali et al. [19], Islam et al. [23], Azam et al. [50], Raihan and Tuspekova [74], and Sun et al. [63]. This effect is stronger in the subpanel of low-institutional-quality countries and lower in high-income countries. As was suggested by Sun et al. [73], low institutional quality may cause inadequate planning of buildings and constructions, a lack of restrictive environmental rules, informal settlements, or the improper planning of the urbanisation process. Therefore, the root causes of environmental pollution induced by urbanisation should be investigated, alongside more restrictive environmental regulation and green projects for sustainable urbanisation. Wang et al. [155] proved that high-income OECD countries have been already able to decouple urbanisation from carbon emissions, as confirmed by the inverted U-shaped relationship between these two variables. In the initial stages, urbanisation may increase carbon emissions through economic growth and residential energy consumption, but when urbanisation exceeds the ”consumption effect”, through improving energy efficiency and restraining industrial energy consumption, the ”agglomeration effect” is produced. Based on these findings, it is suggested that promoting the urbanisation process in EU countries, improving energy efficiency, optimising the energy consumption structure, upgrading industry, and reducing carbon intensity will induce the scale effect, which is conducive to a reduction in carbon emissions.

6. Conclusions and Policy Recommendations

The present study intended to determine the links between globalisation, renewable energy consumption, economic growth, and carbon dioxide emissions for 26 EU countries using panel data from the period of 1990–2020. A set of control variables (financial development, foreign direct investment, and urbanisation rate) were also included in the empirical analysis. We applied a cross-sectional approach to examine the dependence among the variables, then the stationarity properties were examined using second-generation unit root tests (CIPS and CADF). The existence of a cointegration relationship between the considered variables was identified through the Westerlund test, and the regression coefficients are estimated using FMOLS and DOLS models. We conducted separate regressions for the overall panel of 26 EU countries and also for subpanels of developed/developing and high-/low-institutional-quality countries. Finally, causality between variables was identified with the Dumitrescu–Hurlin test.

We found that the carbon emission-reducing effect of globalisation was stronger in the panel of developed and high-institutional-quality countries than in the panel of developing and low-institutional-quality countries. Renewable energy consumption (including solar, wind, hydropower, geothermal, bioenergy, and marine energy) induced a decrease in the levels of carbon dioxide emissions in the overall panel and in all subpanels, but the negative impact was more intense in the developing and low-institutional-quality countries. GDP per capita was found to be a contributor to carbon dioxide emissions’ growth in the overall panel and also in all subpanels, a wider effect being revealed in the developed countries. Financial development induced a decrease in carbon dioxide emissions, while foreign direct investment and urbanisation caused their increase in the overall panel of the 26 EU countries. Financial development’s effect of decreasing carbon dioxide emissions was maintained in developed and high-institutional-quality countries, while in developing and low-institutional-quality countries, there was a reverse effect. An increase in carbon dioxide emissions determined by foreign direct investment was revealed in the overall panel, and also in the two subpanels. The impact was stronger in developing countries, while in the low-institutional-quality countries, it was not statistically significant. Urbanisation caused an increase in carbon dioxide emissions in the EU panel and also in the two subpanels, the increasing effect being stronger in the developing and low-institutional-quality countries than in the developed and high-institutional-quality countries. To sum up, the relationship between the variables under consideration (globalisation, renewable energy, economic growth, financial development, foreign direct investment, and urbanisation) and carbon dioxide emissions depends to a large extent on the respective levels of economic development and institutional quality. Economic development and high institutional quality are mainly associated with a higher positive impact of globalisation and financial development, and a lower negative impact of foreign direct investment and urbanisation on reducing carbon dioxide emissions. Renewable energy consumption’s effect of reducing carbon dioxide emissions is higher in developing and low-institutional-quality countries. Unidirectional causality relationships were identified between globalisation and carbon emissions and between carbon emissions and foreign direct investment; there were bidirectional relationships between economic growth, renewable energy consumption, financial development, and carbon dioxide emissions. All four hypotheses of our study are thus confirmed.

As an overall conclusion, globalisation and renewable energy contributed to decreasing levels of carbon dioxide emissions from 1990 to 2020 in European Union countries. Within the general decreasing trend of carbon dioxide emissions (based on World Bank data [8,107]), decreasing yearly rates in the EU have been highly accelerated in recent years (after 2009). Carbon dioxide emissions decreased in the EU by 30.91% (in kilotons) and 35.13% (in metric tons per capita) in 2020 compared to 1990 (computation based on World Bank data [8,107]). This suggests that globalisation and renewable energy, in tandem with other factors, could make a beneficial contribution to the achievement of carbon neutrality in European Union countries by 2050.

The study’s results suggest some directions for European policies in the context of the European Green Deal and the target of climate neutrality by 2050. Considering our findings, we recommend the following. (i) Globalisation must be promoted through the improvement of banking systems, thereby stimulating financial development and encouraging the development of green innovation and the expansion of a cleaner energy sector; this is also needed to enhance institutional quality by reducing corruption and ensuring property rights and business freedom. (ii) More effective policy measures to extend investment in renewable energy sources (provision of financial instruments for energy-saving projects, incentive through fiscal policies, grants to households) are needed. (iii) The increasing energy demands of higher levels of economic growth should be covered by expanded financial incentives for investment in renewable energy sources, updated infrastructure, and a larger R&D budget for eco-friendly power sources in developing countries, alongside carbon pricing, tariffs, and the advancement of industry 4.0. (iv) Financial development could be seen as a means of decreasing carbon dioxide emissions in EU countries (as was suggested by Horobeț et al. [14]). For developed countries, further efforts to enhance the restraining effect on carbon emissions are needed. Strengthening the financial system and the construction of capital markets would create innovative financial instruments meant to support investments in green business and energy-saving business, investments in research, and the development and restructuring of industry with the aim of decreasing its energy intensity. In developing countries, there is a need for effective regulation of the activities of financial institutions to prevent the financing of pollutant activities; the development of capital markets; improved efficiency in the allocation of financial capital; the promotion of technological innovation through financial capital; and subsidies for entrepreneurial activities to introduce environmentally friendly technologies. In all countries, financial resources should be directed towards investments in green transportation, green energy, and green industry, thus building the green economy. (v) European policies should be focused on tax and trade policies for foreign investors that are related to green/clean technology investment and green R&D activities. The perspectives of developing EU countries should be integrated, wherein more precautions and more strict rules and regulations are needed regarding the inflow of FDI into country in order to prevent harm to environment quality. Additionally, in developing countries, efforts should be made to promote carbon-reducing businesses in sectors with high energy consumption, and to monitor foreign direct investment in clean industries. Such businesses should be assisted in advancing their production technologies and reducing their carbon emissions, as suggested by Javed et al. [71]. Moreover, as Wang et al. [90] pointed out, developing economies should gain the capacity to master pollution and remediate excessive pollution. (vi) Governments should invest more in green technologies meant to reduce carbon emissions in urban areas and promote fiscal measures and financial incentives in order to encourage people to switch to cleaner energy sources. Financial institutions might give funding support to building developers in their green building projects, and also to the population for home renovations in order to meet green criteria (as suggested by Raihan and Tuspekova [74]). (vii) Institutional quality is essential to better environmental quality in the long run; therefore, developing countries need to further strengthen the quality of their institutions (i.e., control corruption, reduce the level of bureaucracy, improve political stability, and implement more restrictive environmental rules) in order to extend their clean energy sources and achieve sustainable development.

For a deeper analysis and more detailed results, future research may examine (i) the influence of additional variables on environmental quality, such as migration between countries; and (ii) the effect of all globalisation dimensions (economic, financial, political) and types (de jure/de facto) on carbon emissions in EU countries, taking into consideration other proxies for environmental quality (i.e., carbon footprint and ecological footprint). It would also be useful to explore the effect of a non-polluting energy consumption mix (i.e., renewable and nuclear energy, as suggested by Saidi and Omri [52] in their study on OECD countries) on carbon emissions.