The effect of GDP, renewable energy and total energy supply on carbon emissions in the EU-27: new evidence from panel GMM

The vast usages of sources of energy that pollute the environment have resulted in major problems of global warming in the world. Global warming and greenhouse effect causes droughts, hunger, and starvation among many other health problems. In this research, the effect of energy use, economic, growth, and renewable energy on carbon emissions, in the European Union region from 1990 to 2019, is examined. The current study differs from previous researches, in that it specifies “effective capital” which is the interaction between energy and capital (a product of energy and capital) in the model. Effective capital is inevitable in the production process because physical capital such as machinery, without power or energy to fuel it, is dysfunctional. The current research employs the Generalized Method of Moments which is strong over endogeneity and overcomes heteroskedasticity, serial, and autocorrelation problems. The findings of this research support past studies that renewable energy reduces carbon emissions and gross domestic product exacerbates carbon emissions. Effective capital and energy use are observed to promote carbon emissions, whereas capital and population size reduces carbon emissions in the European Union.


Introduction
Carbon emissions (CEs) have been observed as the major cause of greenhouse effect and global warming (Salim and Rafiq (2012); Banga et al. (2022); Becker and Fischer (2013)). CE comes from the use of sources of energy that are not environmentally friendly, such as non-renewable energy (NRE) or fossil fuels (Bakhsh et al. (2022); Boukhelkhal (2022); Akadiri and Adebayo (2021)). The EU region is one of the top leading areas of CE (Balsalobre-Lorente and Leitão 2020), among many other developed nations and emerging economies such as China and India (Qin et al. 2021). Therefore, vast use sources of energy that pollute the environment have caused major problems of global warming in the world. Global warming and greenhouse effect cause droughts, hunger, and starvation among many other health problems (Mukhtarov et al. (2022)). Thus, it is paramount and imminent to work towards reducing CE in the world. In a bid to reduce global warming problems, the United Nations came up with the carbon neutrality goal. Nations, worldwide, are encouraged to work towards the use of clean energy. By doing so, sustainable economic development and environmental quality improvement is achieved. Renewable energy (RE) is one of the best alternative source that is friendly to the environment (Banga et al. (2022); Balsalobre-Lorente and Leitão (2020); Akadiri and Adebayo (2021); Shahbaz et al. (2021)), improves economic development (Ivanovski et al. (2021); Apergis and Payne (2010); Chen et al. (2020); Dogan et al. (2020)), and stabilizes the rate of inflation and exchange rate , Mukhtarov et al. (2022); Deka and Dube (2021)).
Literature shows that economic growth and the emissions of carbon are significantly related (see, e.g., Asiedu et al. (2021), Balsalobre-Lorente and Leitão (2020), and Responsible Editor: Eyup Dogan Boukhelkhal (2022)). The findings of these studies show a strong significant link between CE and economic growth. This shows that CE affects economic growth and economic growth affects CE. In addition to that, Akadiri and Adebayo (2021); Abbasi et al. (2021); Bilgili et al. (2022); and Bouyghrissi et al. (2021) provide economic growth as one major factor that significantly exacerbates the emissions of carbon in the environment, whereas Ben Youssef (2020), Adebayo (2021), Bakhsh et al. (2021), and Ahmad et al. (2020a, b) provide that CEs significantly impact economic growth positively. The positive and significant relationship between CE and economic growth is due to the use of NRE, which emits CE, to enhance economic growth. The findings of past studies show that indeed NRE provides a significant positive effect on economic growth (Ntanos et al. (2018); Pao and Fu (2013); Afroz and Muhibbullah (2022); Ben Youssef (2020); Bhat (2018); Asif et al. (2021); Behera and Mishra (2020)). Economic growth is very crucial for the economy, at the same time its negative effects environment, due to the vast use of NRE which emits that CE is desirable to the economy. Thus, policies that raise economic growth of a country together with rising environmental degradation pose a dilemma to economists, environmentalists, and the government. Other studies provide for a negative and neutral effect of NRE on economic growth (Ahmed et al. 2019;Baz et al. 2021). The negative and neutral effect of NRE on economic growth is due to the shift to RE use which is friendly to the environment. NRE causes environmental degradation (Boukhelkhal 2022;Asif et al. 2021), RE reduces CE (Akadiri and Adebayo 2021;Mathiesen et al. (2011);Bhat 2018;Akram et al. 2022), and RE improves the quality of the environment (Balsalobre-Lorente and Leitão 2020). Nuclear energy is also observed to reduce carbon emissions and promote economic growth (Menyah and Wolde-Rufael (2010); Omri et al. (2015)). It is also observed that resource based and green nations perform better economically (Gatto and Drago (2020a, b)) and that proper energy policy making should be followed by governments (Gatto and Drago 2020a, b;Drago and Gatto 2022); this is so because some government backed policies, meant to achieve political Goals, have been observed to impede RE investment (Aguirre and Ibikunle 2014). Therefore, to promote both environmental quality and economic growth, it is encouraged to resort to the use of RE. By doing so, the carbon neutrality goal is achieved.
The study on the effect of GDP, RE and total energy supply on CE is of paramount importance. It helps understand how and which factors degrade the environment through the emission of carbon in the air. By doing so, proper policy making and implications are provided. The current research is done in the European Union (EU) region, which is among the largest emitters of CE, hence the major region that causes global warming. The current study differs from previous researches done in this region, in that it specifies in the model "effective capital," which is the interaction between energy and capital (a product of energy and capital). Effective capital is inevitable in the production process because physical capital such as machinery by itself without power or energy to fuel it is dysfunctional. Thus, this study adds effective capital on top of other series (GDP, energy, capital, population, RE) to see how they affect CE in the EU region. Moreover, the current research employs the Generalized Method of Moments (GMM) which is strong over endogeneity (Banga et al. (2022)) and overcomes heteroskedasticity, serial, and autocorrelation problems (Fraj et al. (2018)) for robust results. The findings of this research support past studies that RE reduces CE and GDP exacerbates CE. Effective capital and energy use are observed to promote CE, whereas capital and population size reduces CE in the EU.

Carbon emissions, non-renewable energy, and the environment
It is crucial to understand the relationship that exists between CE, NRE, and the environment among world economies. NRE is seen as the major cause of CE which in turn causes environmental degradation. Thus, for proper policy making and implementation, thorough research is required. According to Anser et al. (2021), in a study of selected South Asian countries, environmental quality is being worsened. This is due to the wide usage of sources of energy that are not friendly to the environment. Moreover, Balsalobre-Lorente and Leitão (2020) allude that greenhouse gas and climate change are directly correlated with growth in the EU-28, whereas Bakhsh et al. (2022) allude that sulphur dioxide (SO2) emissions are related significantly with energy consumption, in Pakistan. In the case of 35 African countries, Boukhelkhal (2022) provides that NRE causes environmental degradation. On the other hand, Adedoyin et al. (2020) provide that environmental degradation across the Central and Eastern Europe (CEE), Commonwealth of Independent States (CIS), and New Member States (NMS) blocks is different, whereas Ben Jebli and Ben Youssef (2015) provide that the Environmental Kuznets Curve hypothesis is not supported.
On the relationship between NRE and the CE, Akadiri and Adebayo (2021) provide that NRE has a positive effect on CE in India. These postulations support the findings of Bhat (2018) in the BRICS economies that NRE increases emissions. Therefore, excessive use of NRE results in environmental stress. Above that, Asif et al. (2021) support the assertion above, by providing that in the 99 world countries, NRE negatively affects the environment due to the emissions of carbon in the air. In addition to that, Abbasi et al. (2021) allude that energy use, industrial value added, and population growth, in the UK, boost the sustainability of the environment in the long run. Environmental quality and the consumption of energy in Pakistan are observed to exhibits for a symmetric feedback effect. Other studies also allude that consumption of energy is symmetrically caused by environmental quality (Baz et al. (2021). Therefore, the evidence provided for in past literature studies show that NRE increases the emissions of carbon and hence the degradation of the environment.

Non-renewable energy (NRE) and economic growth
Inasmuch as NRE causes CE and hence environmental degradation, it is observed to increase economic growth. Boukhelkhal (2022) in the 35 African countries from 1980 to 2016 observed a bidirectional relationship between economic growth and the consumption of NRE. Moreover, Afroz and Muhibbullah (2022) observed that in Malaysia during 1980-2018, in the short run and long run, the asymmetric effect of NRE consumption on economic growth is confirmed. Positive shocks of NRE in the short and long run are more than RE positive shocks (Afroz and Muhibbullah (2022)). The findings of Afroz and Muhibbullah (2021) also show hat, economic growth in Malaysia is more dependent on NRE than RE consumption. Moreover, according to Ben Youssef (2020) in research done in the USA during the period 1980-2016, a long run unidirectional causality from consumption of fossil fuel energy to economic growth exists. In addition to that, in the G7 countries short run causality between the consumption of NRE with economic growth exists (Behera and Mishra (2020)). In the BRICS countries NRE positively affects economic growth in the long run (Bhat (2018)). This supports the findings of Asif et al. (2021) in the 99 world countries that NRE has a significant positive effect on economic growth. Therefore, the evidence provided for in past studies is overwhelming to prove that NRE is a major driver towards economic growth.
However, other studies confirm the existence of a negative effect of NRE on economic growth. Ahmed et al. (2019) in a study done in Myanmar during the period 1990-2016 observed that NRE has a negative influence on economic growth. In the presence of technological inefficiency, NRE is counterproductive (Ahmed et al. (2019)). Moreover, Baz et al. (2021) in Pakistan observed a neutral effect in negative and positive shocks of fossil fuel and economic growth. Thus, according to these studies, NRE reduces economic growth or does not provide a significant impact in enhancing economic growth. As discussed earlier, NRE is also blamed for exacerbating environmental degradation through the emission of CO2. Thus, if it does not promote economic growth, a major indicator in the economy, then it should not be used since it provides double problems, both to the environment and to the economy.

Carbon emissions and economic growth
On the association between CE and economic growth, a significant positive relationship between economic growth and CE (Akadiri and Adebayo (2021)), is observed. A bidirectional causality is observed between gross regional product growth and SO2 emissions (Ahmad et al. (2020a, b)). Moreover, Asiedu et al. (2021) allude for the existence of a long-run relationship between economic growth and CE in the 26 European countries, whereas Balsalobre-Lorente and Leitao (2020) allude that economic growth is positively correlated with CE in the EU-28 region. In a study of 35 African countries, a bidirectional relationship between economic growth and CE exists (Boukhelkhal (2022)). Moreover, in Morocco during the period 1990-2014, Bouyghrissi et al. (2021) observed a significant effect from economic growth to CE. Economic growth stimulates the emissions of carbon dioxide (CO2) in the short run (Abbasi et al. (2021)). It is also observed that the growth of fossil fuel consumption is increased by GDP growth (Bilgili et al. (2022)). In addition to that, Ben Mberek et al. (2018) also add that in Tunisia, CO2 emissions are affected by economic growth in the long and short run.
On the other hand, Ben Youssef (2020) observed a long run unidirectional causality from CO2 emissions to economic growth in the USA. This is in support of the findings of Adebayo (2021) in a study of Japan, which alludes that CE triggers economic growth. Additional causality from CE to economic growth is also observed (Adebayo (2021)). Another indicator of CE, SO2 emissions significantly impact GDP in Pakistan, Bakhsh et al. (2022). In China, Ahmad et al. (2020b) allude that the effect of economic performance deterioration is induced by sulphur dioxide emissions and that economic growth has the effect of introducing the emissions curtailment effect. Moreover, in the G7 economies, Balcilar et al. (2020) propose that in order to reduce CE, Italy, the USA, Japan, and Canada should sacrifice economic growth by reducing fossil fuel energy. However, in the case of France, since the 1990s and for German in the whole period of analysis, including some few exceptions in UK, the above situation is invalid (Balcilar et al. (2020)).

Carbon emissions and renewable energy
According to Akadiri and Adebayo (2021), the consumption of RE reduces CE in India. This is in line with the postulations of Bhat (2018) who alludes that the consumption of RE tends to decrease CE in the BRICS economies. Balsalobre-Lorente and Leitao (2020) postulate that environmental quality is improved by RE use. This shows that, in order for nations to achieve the carbon neutrality goal, they should resort to the use of RE. Moreover, Akram et al. (2022) allude that, in Mexico, Indonesia, Nigeria, and Turkey (MINT) region during 1990-2014, RE and energy efficient significantly reduce CE in the region. Energy efficient reductions positively affect CE, while energy efficient positive shocks have a negative relationship with CE (Akram et al. (2022)). However, according to Anser et al. (2021) in Bangladesh, globalization index and RE positively impact CE, while Menyah and Wolde-Rufael (2010) provide for no significant link between RE and CE. Adedoyin et al. (2020) postulate that an increase in RE generation rises CO2 emissions in CIS and CEE countries, by 0.04 and 0.02% respectively. It however results in a decrease in CE of NMS countries. Therefore, it is crystal clear that RE use mitigates the emissions of carbon; hence, the quality of the environment is improved (Mathiesen et al. (2011)).

Model specifications
The major drivers towards CE in the world are: NRE, energy use, economic growth, fossil fuel use and population growth among many others. According to Apeaning (2021), the emissions of CO2 can be expressed as a function of GDP per capita, emission factor, total population, fossil fuel mix, final energy intensity of GDP, and conversion efficiency. Amirnejad et al. (2021) also provide GDP per capita and its square root as the major determinants of CE. Moreover, Akram et al. (2022); Akadiri and Adebayo (2021); and Adedoyin et al. (2020) specified CE as a function of economic growth, energy efficiency, GDP, NRE, and RE. Therefore, due to the postulations of past studies, the current study specifies CE as a function of GDP, capital, effective capital, population size, primary energy supply, and RE. Thus, the current study differs from past studies in that it adds a new determinant, effective capital, of CE. The model specification employed in this research is given in Eq. (1): CE is the emissions of carbon dioxide, GDP is the gross domestic product, POP is the population size, RE is renewable energy, PES is primary energy supply, CAP is the capital stock, and EC is effective capital. The statistical representation of the model employed in this research is specified in the Eq. (2): β0 is the constant term; β1, β2, ……, β6 are the coefficient terms; and μ is the error term of the model.
The variables employed in this research are CE, GDP, effective capital, RE, primary energy supply, population size, and capital. CE is the dependent variable in this study, while the rest of the series are independent variables. Moreover, the data of all the variables is retrieved from https:// data. oecd. org except for the CE indicator which is retrieved from https:// data. world bank. org. To cater for missing data and hence the quality of data analysis, pairwise deletion method has been employed. Pairwise deletion is of the assumption that the data is missing completely at random (MCAR). With pairwise deletion, all cases that contain the data together with those cases with missing data are employed in the analysis. This approach is advantageous because it allows more of the data to be utilized by data scientists. The disadvantage of pairwise deletion technique is that it causes more variation on the resulting statistics since its basis is on different data sets. Moreover, duplication of outcomes may be difficult considering a complete data set. Panel data is used in this research for 27 EU countries for the period that ranges from 1990 to 2019. The GDP is total value of products produced in a nation during a certain period (Deka and Dube (2021)), and in this research it is measured in total US dollars per capita. Primary energy supply is a total of NRE and RE, that is, measured in tons per 1000 US dollars. CE indicator in this research is represented by the emissions of CO2 and is measured in metric tons per capita. Capital in this study is the physical assets in a nation that are used in the production process of goods and services and is measured as a percentage of gross fixed capital formation (GFCF). Effective capital considers the interaction between energy use and capital, the physical machinery. In this study, to find the value of effective capital, a product of capital and energy use is considered. Population size is the number of people living in a country permanently during a set period of time and is measured in million persons as a total. RE consumption employed in this study consists of clean sources of energy that are friendly to the environment ), and is measured as a percentage of primary energy supply.

Method
The current research follows two major steps in data analysis.
Step 1: preliminary tests The current research in the data analysis begin by running the descriptive statistics, unit root test, and the cointegration test, as preliminary tests before employing the major techniques of data analysis. Preliminary tests are crucial in data analysis because they show the behavior of the data and how the variables are related.
The descriptive statistics shows the characteristics of each and every variable in terms of their, mean, median, standard deviation, and sum, among others. On the other hand, the unit root test shows if the variable is stationary or not and hence ascertains the order of integration of each variable. Variables that are not stationary, if specified in an ordinary least square (OLS) model, result in spurious regressions. Thus, in an OLS model, all indicators should be stationary. Unit root test is also crucial in ascertaining the best method to employ in a study. For example, cointegration methodologies, like dynamic ordinary least square (DOLS), vector error correction (VEC), and fully modified ordinary least square (FMOLS) models, require all variables to be integrated of order 1. On the other hand, autoregressive distributive lag (ARDL) technique works with variables that are integrated of order 0 or 1 or a mixture of both (Pesaran et al. (1997;1999;2001); see also  and Deka and Dube (2021) for review). In this research, the Philips Peron (PP) together with the Augmented Dickey Fuller (ADF) techniques by Phillips and Perron (1988) and Dickey and Fuller (1979) respectively are used to test for unit root. The PP and ADF tests are the traditional models of unit root and hence provide robust outcomes. The current research also tests for cointegration as a preliminary test. Cointegration test helps to ascertain the existence of a long-run association among variables that are specified in a model (Granger 1986;Engle and Granger 1987). Variables that are cointegrated can be expressed in cointegration regression techniques such as the DOLS, VEC, and FMOLS, to specify the long-run association of the variables. It follows that, if these two major preliminary tests (i.e., if variables are integrated of order 1 and if the variables are cointegrated) are satisfied, then cointegration regressions can be used. The panel cointegration tests of Johansen Fisher and Kao tests are used.
Step 2: major methodologies The next step after the preliminary tests is to specify the appropriate methodologies. In this research, we use two forms of methodologies, one that specifies the short-run coefficients of the model and another that specifies the long-run coefficients of the model. Due to the fact that the data set used in this research contains the number of crosssectional dependence (N) that is greater than the number of time series observations (t), hence the most appropriate model to employ is the Generalized Method of Moments (GMM) . GMM model is preferred because it overcomes endogeneity problems, autocorrelation, and heteroskedasticity problems (Banga et al. (2022); Fraj et al. (2018); for review, see also ). GMM technique was founded through the work of Holtz-Eakin et al. (1988); Anderson and Hsiao (1982); Arellano and Bond (1991); Blundell and Bond (1998); and Arellano and Bover (1995). Holtz-Eakin et al. (1988), Anderson and Hsiao (1982), and Arellano and Bond (1991) pioneered the first-difference GMM which uses differencing method to correct for endogeneity. First-difference GMM has one short fall, that is, it subtracts the previous data from the contemporaneous one, thereby magnifying the gaps in unbalanced data set. Systems GMM (Blundell and Bond (1998); Arellano and Bover (1995)), unlike first-difference GMM, uses orthogonal deviations which subtracts the average of the variable's average of future observations available (Blundell and Bond 1998;Arellano and Bover 1995). This research uses both first difference and systems GMM, and the outcomes are compared. For the purpose of providing long-term coefficients of the model, the DOLS and FMOLS techniques are used. The DOLS and FMOLS techniques are cointegration regressions that are used in the event that all the variables in a model are stationary at first difference and that they are cointegrated. The DOLS and FMOLS models are crucial since they produce robust long-term results that are crucial for policy making.

Results and data analysis
In this section of the study, we provide the results of the research. Table 1 gives the results of the descriptive statistics of all the variables RE, GDP, primary energy supply, CE, capital, effective capital, and population size. The total observations for each and every variable employed in this study is 837 (see Table 1). For GDP, In this study, we also provide the results of the PP and the ADF tests of unit root test. The results provided for in Table 2 of the ADF unit root test show that the variables GDP, CE, primary energy supply, RE, and population size are not stationary at level but stationary at first difference. The variables effective capital and capital according to the ADF unit root test are observed to be stationary at level and stationary at first difference. However, the findings of the PP test of unit root in Table 2 shows that all the variables GDP, CE, primary energy supply, effective capital, RE, population size, and capital are not stationary at level but stationary at first difference. Therefore, in this research, we conclude that all the variables specified in this research are integrated order one.
We also provide the results of the cointegration test as per Johansen Fisher panel and the Kao residual tests of cointegration in Table 3. The results provided in Table 3 indicate that the series in this research study are cointegrated. This shows that these series have a significant long run association. As a result, since these series are integrated of order one, according to the PP and ADF test of unit root, and are cointegrated, then they can be specified in cointegration regression techniques.
The long-run results of the DOLS and the FMOLS techniques in this research study are provided in Table 4. The findings of this study show that RE consumption in the EU-27 countries has a significant negative influence on CE in the long-run, considering both the results of the FMOLS and the DOLS models. The results shows that when RE consumption increases by 1 unit, then CE will decrease by 0.23 or 0.24 units in the long-run. These findings show that, in order for countries in the EU-27 block to suppress CE, they should shift to the use of RE that is friendly to the environment and does not emit carbon into the air. Moreover, the long-run findings of the DOLS and the FMOLS models in this study as provided for in Table 4 show that GDP has a significant positive effect on CE in the EU-27 countries. The findings show that an increase in GDP by 1% causes CE to increase by 1.7% or 3.5% in the long-run. Therefore, in the EU-27 countries GDP is observed to exacerbates CE. This is so because, when nations increase productivity and hence achieve high economic growth, they use more of NRE which emits CO2 in the air. Thus, a positive link between GDP and CE is observed. Primary energy supply is also observed to provide a positive effect on CE in the EU-27 countries. An increase in primary energy supply by 1  respectively. The positive influence of primary energy supply on CE is due to the fact that primary energy supply is a combination of NRE and RE energy. Therefore, it is the fraction of NRE that is positively linked with CE. Hence, in order for nations to reduce CE, they should shun the use of NRE and concentrate on the use of RE. Effective capital is observed to provide a positive effect on CE as per the FMOLS results, while the DOLS model shows that the relationship is insignificant. The positive effect of effective capital on CE emissions is due to NRE that is included in total primary energy which is used to obtain the value of effective capital. Therefore, we observe that NRE causes the interaction between machineries and energy to produce substances that pollute the environment. It is the machineries that produce carbon in the air during the production processes because they are powered with NRE which is friendly to the environment. The long-run findings also show that population size in the EU exhibits for a significant negative effect on carbon emissions. This shows that an increase in the population size of the EU nations have a tendency to reduce carbon emissions. Normally, population growth exacerbates carbon emissions as more people scramble on the scarce resources causing environmental degradation. Also, high population entails that energy will be consumed in large quantities, and if NRE is consumed more due to high population growth, then carbon emissions increase. Therefore, the existence of a negative association between population size and carbon emissions in the EU countries can be explained in two ways. Firstly, EU nations are developed nations that have developed policies to reduce population growth; thus, their population is not growing much; hence, the use of energy is not strained. As a result, a negative association of population growth and carbon emissions is observed. Secondly, EU nations have started to shift to the use of RE which is not harmful to the environment; hence, a rise in the population size simply raises RE use, and hence carbon emissions are not worsened. Capital, the control variable in this study according to the FMOLS results, provides for a significant negative effect on CE, while DOLS model shows that the association is not significant.
The results of the R 2 and the adjusted R 2 for both the DOLS and FMOLS techniques in Table 4 are very high, close to 100%, indicating the goodness of fit in the model. Table 5 of this research provides the residual diagnostic testing results. The diagnostic testing results provided for in Table 5, for both the DOLS and the FMOLS, show that these two models have problems of autocorrelation and partial autocorrelation and that the residuals are not normally distributed.
The results of the panel GMM model specified in this study as provided in Table 6 give the results of both the first difference GMM and the systems GMM model. The findings of the GMM technique in Table 6 show that there is a significant positive effect of the first lag of CE on current CE in the EU-27 countries. The findings of the first difference and the systems GMM model show that an increase in the first lag of CE by 1 unit results in an increase in current CE by 0.82 units in the short run. Therefore, past positive CE increases CE in the future. The results also shows that GDP is positively linked with CE in the EU-27 countries. An increase in GDP by 1% causes CE to increase by 0.56 or 0.5% in the short run. This is so because nations strive to achieve economic growth through the use more of NRE which emits carbon gases into the air. However, RE in the EU-27 countries is observed to have a significant negative effect on CE. This shows that the use of RE in the EU-27 economies reduces CE in the air. Thus, in the EU-27 economies, the use of NRE should be shunned and more of RE energy should be used for the purpose of reducing CE. Moreover, primary energy supply is observed to provide The findings of GMM model, just like the long-run findings of FMOLS, show that an increase in the population size of the EU nations has a tendency to reduce carbon emissions. In line with the explanation given under the FMOLS findings, we provide that under normal circumstances and in line with postulations of past studies, population growth exacerbates carbon emissions. Also, high population entails that energy will be consumed in large quantities, and if NRE is consumed more due to high population growth, then carbon emissions increases. Therefore, the existence of a negative association between population size and carbon emissions in the EU countries can be explained in two ways. Firstly, EU nations are developed nations that have developed policies to reduce population growth; thus, their population is not growing much; hence, the use of energy is not strained. As a result, a negative association of population growth and carbon emissions is observed. Secondly, EU nations have started to shift to the use of RE which is not harmful to the environment; hence, a rise in population size simply raises RE use, and hence carbon emissions are not worsened. The results of the first difference and the systems GMM in Table 6 are robust because the J-statistic value indicates that the model correctly specifies the data. The J-statistic value is very low, and its P value is greater than 0.1 showing that, indeed, the model is correctly specified. Therefore, the findings of the GMM model are the most reliable results in this study and can be used for policy making.
The outcomes of the correlation analysis in Table 7 show that carbon emissions and GDP, carbon emissions and total energy supply, carbon emissions and effective capital, GDP and RE, GDP and effective capital, GDP and capital, GDP and population size, total energy supply and effective capital, effective capital and capital, effective capital and population size, and capital and population size exhibit for a significant positive association among each other. The existence of a significant positive correlation among these variables indicates that a rise in one variable should be followed by a rise in another. Also, Table 7 shows that GDP and total energy supply, RE and total energy supply, RE and population size, total energy supply and capital, and total energy supply and population size exhibit for a significant negative association among each other. The presence of a significant negative correlation among these variables shows that a rise in one variable will be followed by a decrease in another. Hence, in order to promote one variable, then another respective variable should be discouraged. We also show in Table 7 that CE and population size, RE and effective capital, and RE and capital are not significantly related. Therefore, these variables are not connected in a significant association, and a change in one variable will not cause any effect on another.

Discussion
In this research, we observe that the findings of the long-run FMOLS and DOLS models and that of the short-run systems GMM and first difference GMM models are similar. The current study based its policy recommendations on panel GMM model which has robust results and also because GMM technique is strong over endogeneity (Banga et al. (2022)) and overcomes heteroskedasticity, serial, and autocorrelation problems (Fraj et al. (2018)). In this research, we support the findings provided for by past researches that the use of RE has the significant effect of reducing CE. Past studies allude that RE significantly reduces CE and hence should be used to reduce environmental degradation (Ahmad et al. 2020a, b;Bhat 2018;Akadari and Adebayo, 2021). This shows that in order to reduce the effects of global warming, caused by NRE use, nations should shift to using RE. This is in support with the postulations of Balsalobre-Lorente and Leitao (2020) who allude that RE improves the quality of the environment. Therefore, in this research, we encourage the use of RE in the EU region. NRE should be shunned since it is the major driver of CE. Moreover, this research shows that GDP in the EU region promotes CE. The positive effect of GDP on CE is due to vast use of NRE in the EU in a bid to achieve high economic growth. These findings are supported by past studies that provides for the existence of a significant positive effect of economic growth on CE (see Bilgili et al. (2022); Abbasi et al. (2021); Bouyghrissi et al. (2021); and Akadiri and Adebayo (2021)). As a result, nations should shun the use of NRE to achieve high economic growth and resort to the use of RE which also promotes economic growth without polluting the environment (Eren et al. 2019;Ivanovski et al. 2021;Rahman and Velayutham 2020). This study also observes that effective capital and energy use increases CE. Effective capital is the interaction between total energy use and capital. Thus, we provide that the positive effect of effective capital on CE is due to the presents of NRE in total energy use variable. As a result, if nations shift to the use of RE, effective capital as a product of RE and capital may tend to reduce CE. Capital by itself reduces CE in the EU region, while population size by itself is observed to reduce CE as well. The negative effect of population size on CE is due to the policies adopted in this region that are meant to reduce population growth. High population growth may exacerbate CE since more people will strive over scarce resources; hence, more of NRE is used in the production process. Therefore, the existence of a negative association between population size and carbon emissions in the EU countries can be explained in two ways. Firstly, EU nations are developed nations that have developed policies to reduce population growth; thus, their population is not growing much; hence, the use of energy is not strained. As a result, a negative association of population growth and carbon emissions is observed. Secondly, EU nations have started to shift to the use of RE which is not harmful to the environment; hence, a rise on their population simply raises RE use, and hence carbon emissions are not worsened. Thus, all world countries are encouraged to set population growth policies to reduce CE.

Conclusion
In short, this study shows that RE is the best source of energy to reduce CE; hence, nations should shift from NRE to RE use. We also observe that high economic growth obtained from vast use of NRE increases CE. As a result, we encourage the use of RE to achieve economic growth. The significant positive effect of energy use and effective capital on CE can be explained by the existence of NRE in both Table 7 Correlation matrix *; **; *** represents 10%; 5%; 1% level of significant variables. The current research also encourages countries to adopt some policies to reduce population growth. By doing so, CE is reduced as well as global warming effects. The limitations of the study are that it does not directly include NRE in the model. However, indirectly NRE is included in total energy use. Thus, the impact on CE is indirectly observed from total energy use. For the purpose of future studies, the interaction between RE and capital should be specified to proxy effective capital. If only RE is considered in calculating effective capital, then effective capital may reduce CE.