Wind energy and CO2 emissions: AMG estimations for selected countries

This study analyzes the relationship between wind energy consumption, coal energy consumption, globalization, economic growth, and carbon emissions. Data from 37 countries for the period 2000–2019 are included in the analysis. To examine the long-term relationship between the variables, the AMG method, which considers the cross-section dependence and slope homogeneity, was used. According to the long-term coefficient estimates of the cointegrated variables, wind energy consumption has a statistically significant and negative effect on carbon emissions in the long run. For example, a 1% increase in wind energy consumption reduces carbon emissions by 0.018%. On the other hand, the variable of globalization has a statistically significant and positive effect on carbon emissions in the long run. A 1% increase in globalization increases carbon emissions by 0.107%. These findings show the importance of wind energy consumption in reducing carbon emissions.


Introduction
With the industrialization process experienced globally, the increase in the population of countries caused a rise in energy demand. The energy consumption rate increased by 44% from 1971 to 2014 (Eren et al. 2019). In this period, especially fossil fuel consumption rates reached 80% (Bilgili et al. 2017). Increasing industrialization activities increased dependence on fossil fuels. This dependency is mainly concentrated in the consumption of coal resources. All these processes have revealed the risk of decreasing non-renewable energy sources. Along with the problems experienced in resources, environmental issues have also started to emerge (Baek 2016). Topics such as global warming and climate change have occurred with fossil fuels (Ozturk and Acaravci 2010). Such increases in greenhouse gas and CO 2 emissions have become dangerous for human life and living life (Can et al. 2020). The share of CO 2 emissions in greenhouse gas emissions has reached 60%. In 2011, 34,459 million tons of CO 2 emissions were emitted, which increased the CO 2 emission rate within the greenhouse gas emission by 80%. British Petroleum (BP) Statistical Review of World Energy (BP 2020) CO 2 emission, which was 29.714 million tons in 2009, increased to 34.169 million tons in 2019 (REN21 2020). Advances in the economy and energy cause problems in environmental and social areas when the same level of precautions are not taken. Increases in economic costs and the emergence of environmental issues have led global actors to produce alternative policies that consider the economy and the environment-these policies primarily directed to the emergence of new resources in the energy fields. With the use of renewable energy sources, many benefits are provided in social and environmental areas. The world's rate of renewable energy usage is increasing day by day (Can et al. 2021). This ratio increased to 18.1% between 2008(REN21 2019. This rate is expected to increase to 60% in 2050, according to the International Renewable Energy Agency (IRENA) (Gielen et al. 2019). Wind energy is seen Responsible Editor: Roula Inglesi-Lotz as the most promising resource for the future in renewable energy. Wind energy, a clean energy source due to not using fossil fuels, is a valuable and preferred type of energy in many aspects. It is estimated that wind energy will meet 5% of the world's energy needs by 2025 (Poore 2008). The use of wind energy has many benefits. Its main advantages are (Aydin 2019): • In wind energy, greenhouse gas emissions are reduced, so the carbon dioxide generated during the consumption phase is compensated by the carbon dioxide held during the photosynthesis process. • With the use of wind energy, the supply problem in nonrenewable energy sources is eliminated. • Wind energy harms CO 2 emissions.
With the use of wind energy, global effects are observed in the short and long term. The use of wind energy is a type of energy that impacts all areas such as economic, political, social, and environmental. With the help of this type of energy, significant advantages are obtained in all sectors. Increasing production and energy demands in the world with globalization have an essential effect on increasing carbon emissions. Using this type of energy causes significant gains in the future and provides substantial advantages in all sectors (Hernández et al. 2019;Magazzino et al. 2021). Increasing production and energy demands in the world with globalization have an essential effect on increasing carbon emissions. In other words, globalization is an inevitable variable of environmental pollution today (Liu et al. 2020;Nguyen and Le 2020). When the increasing energy demand in the world is met by wind energy, it eliminates the problems caused by non-renewable energy sources. At the same time, it ensures that they are aware of renewable resources. Decisions made on energy issues around the world affect the economy and the environment, and so on. It also seems to have effects in such areas.
Among the renewable energy sources, the effect of wind energy in terms of benefit and environmental friendliness is significant. Therefore, this study examines the relationship between wind energy and CO 2 emissions in 37 countries using wind energy. First, all energy types under renewable energy have been analyzed together (Apergis and Payne 2011; Destek and Aslan 2020). For example, Magazzino et al. (2021), Destek and Aslan (2020), Hernández et al. (2019), Dogan and Seker (2016), and Ohler and Fetters (2014) can be counted among these studies. This study analyzed only the effect of wind energy on CO 2 emissions for 37 countries. Thus, it is emphasized that the impact of wind energy on CO 2 emissions is revealed and how important it is for the environment. Secondly, for the country group used in the study, the relationship between wind energy and CO 2 is examined first. The analysis results obtained for this group of countries are thus translated into policy recommendations. And for these countries, the role of wind energy in preventing environmental pollution comes to the fore. In this respect, this study is a first in the literature and makes significant contributions. Thirdly, globalization, whose effects on environmental pollution are being investigated more frequently today (Can et al. 2020), has also been included in the analysis with wind energy. In this way, the effect of globalization, whose impact on environmental pollution is discussed in the literature, is also investigated in 37 countries included in the analysis. Therefore, in countries that cannot avoid the impact of globalization, the relationship between environmental protection and globalization is clarified. Fourth, the augmented mean group (AMG) estimation method is used for the first time both for this country group and for variables. Thus, more accurate estimation is made econometrically, and the reliability of the results obtained is increased. It is estimated that wind power has a statistically significant and negative effect on CO 2 in this group of countries. Furthermore, there is a substantial and positive relationship between the globalization variable and CO 2 . Based on these findings, investment encourages wind energy production and should plan consumption for this group of countries. Regulations that will accelerate the development of wind energy technology and facilitate infrastructure investments should be implemented. Policies supporting wind energy will make a significant contribution to both economic and environmental fields. On the other hand, to reduce the impact of globalization on the environment, there is a need for larger-scale macro plans that require the participation of all countries.
The sections that are subject to this study are listed in order. After the introduction of the study, the second section includes the literature section. In the literature section, studies on wind energy, globalization, and CO 2 emissions, which are the main variables of the study, are mentioned. In the third part of the study, definitions are made on variables and country groups. The method part is included in the fourth part of the study, while the fifth part consists of the analysis estimates. The last section contains the results and policy recommendations found in the study.

Wind energy and CO 2 emissions
Many studies in the literature examine the impact of renewable energy on carbon emissions. However, very few studies examine the effect of wind energy, which is included in renewable energy, on carbon emissions. In the studies in the literature, decreases in energy resources, climate change, and environmental degradation caused by carbon emission are mentioned. Due to such results of non-renewable energy, it has been recommended to carry out studies to increase the use of renewable energy resources (Burg et al. 2018;He et al. 2018;Shao and Rao 2018). However, in the studies in the literature, the problems of cross-section dependence and slope homogeneity have been ignored. Differently in our research, we make analyses that take into account our variables' cross-section dependence and slope homogeneity problems. Magazzino et al. (2021) analyze the relationship between renewable energies such as solar and wind energy and CO 2 emissions in China, India, and the USA. Data covering the period 1986-2017 were included in the analysis with the machine learning method.
Estimates emphasize that the effect of wind energy on CO 2 emissions varies from country to country. Yousefi et al. (2019) explore the impact of wind power capacity in 2017 on CO 2 emissions worldwide. Yousefi et al. (2019) determined that wind resources prevented at least 600 million tons of CO 2 emissions in 2017. If the current wind energy development rate continues, it is estimated to avoid up to 3100 million tons of CO 2 emissions in 2030. Forbes and Zampelli (2019) analyze the relationship between wind energy and CO 2 emissions in Ireland. Time series analyses were carried out with the data for the 2015-2018 period. According to estimates, wind energy harms CO 2 . Hernández et al. (2019) reveal the projection of wind energy on CO 2 and other renewable energies. The forecast covering the period 2015-2050 has been developed for the European Union. According to the projection, wind energy causes a decrease in CO 2 emissions during the projection period. Destek and Aslan (2020) examined the relationship between renewable energy and environmental pollution for G7 countries. Wind, solar, and hydroelectric variables are discussed in renewable energy. Destek and Aslan (2020) analyzed the relationship between renewable energy variables and carbon emissions with the AMG estimator. According to estimates, there is a negative relationship between renewable energy variables and carbon emissions. In other words, the increase in renewable energy consumption causes a decrease in carbon emissions. Sari et al. (2008) examined the relationship between renewable energy (hydroelectric energy, solar, wind energy), industrial production, and employment for the USA. They have used a panel ARDL estimator with 6-month data for the years 2001-2005. According to the estimation results, increases in income and employment were found to positively affect renewable energy. Ohler and Fetters (2014) examined the causality relationship between economic growth and renewable energy sources (biomass, solar, wind energy) for OECD countries. Ohler and Fetters (2014) found a bidirectional causality relationship between renewable energy and economic growth from 1990 to 2008. Dogan and Seker (2016) examined the impact of trade openness, renewable and non-renewable energy variables on carbon emissions for European Union countries. There is a negative relationship between the data from 1980 to 2012 and renewable energy and CO 2 emissions. On the other hand, there is a positive relationship between commercial openness, nonrenewable energy sources, and CO 2 .

Globalization and CO 2 emissions
Studies examining the relationship between globalization and CO 2 emissions are available in the literature. There are many studies in the literature that have a positive effect on the CO 2 emissions of globalization. However, the relevant literature has not examined the cross-section dependence and slope homogeneity relations between the variables. In our study, we made analyses that take into account the cross-section dependence and slope homogeneity problems of variables. Antweiler et al. (2001) examined the relationship between globalization and environmental pollution in their studies in 44 countries. The study found a positive relationship between globalization and environmental quality. Choi et al. (2010) examined the relationship between globalization, economic growth, and CO 2 emissions for China, Korea, and Japan. As a result of the study, a positive relationship was found between CO 2 emission and globalization variables. Naranpanawa (2011) examined the relationship between the globalization variable and CO 2 emissions for Sri Lanka in the period 1960-2006. A short-term relationship was found between the obtained results and variables. Rahman (2013) examined the relationship between CO 2 emission and globalization for Bangladesh over the period 1972-2009. The study found that the increase in globalization has a positive effect on CO 2 emissions. Yıldırım (2013) examined the relationship between globalization, economic growth, and CO 2 emissions for 20 developed and developing countries for the period 1990-2009. The study found that globalization and economic growth positively affect CO 2 emissions. Therefore, increases in globalization and economic growth will cause increases in carbon emissions. Gu et al. (2013) examined the relationship between globalization and CO 2 emissions using time series analysis for the period 1981-2010 for the Chinese economy. As a result of the study, a long-term relationship was found between the two variables. A one-way connection from the globalization variable to CO 2 emission with the causality test was found. Zhang et al. (2017) examined the relationship between globalization, economic growth, energy consumption, and CO 2 emissions. The study was conducted for the years 1971-2013 in 17 industrialized countries. The findings obtained found a positive relationship between globalization and carbon emission. Liu et al. (2020) examined the relationship between globalization and carbon emissions for G7 countries. A positive correlation has been found between the results obtained and globalization and CO 2 emissions. Similarly, Nguyen and Le (2020) examined in which direction there is a relationship between globalization and CO 2 in Vietnam. ARDL test was used in the study covering the period 1990-2016. With the survey, it has been determined that globalization in Vietnam increases CO 2 emissions. Furthermore, studies on different country groups found a positive relationship between the globalization variable and CO 2 emissions.

Data
This study covers annual time series data from 2000 to 2019 for 37 countries: Argentina, Australia, Austria, Belgium, Brazil, Canada, China, Czech Republic, Denmark, Egypt, Finland, France, Germany, Greece, Hungary, India, Iran, Ireland, Italy, Japan, Luxembourg, Mexico, Netherlands, New Zealand, Norway, Poland, Portugal, Russian Federation, South Korea, Spain, Sri Lanka, Sweden, Switzerland, Turkey, Ukraine, USA, and the UK. We selected these countries because their data was available. We selected the following variables: wind energy consumption (million tones of oil equivalent per capita), coal energy consumption (million tones of oil equivalent per capita), gross domestic product (current USD per capita), CO 2 emissions (million tones per capita), and total globalization index (KOF index from 0 to 100). Wind energy consumption, coal energy consumption are obtained from British Petroleum Statistic (2019), the gross domestic product from World Development Indicators (2019), and the total globalization index from the KOF Index of Globalization (2019). Wind energy can be calculated in terms of exajoules and input-equivalent. The data used in this study are in exajoules (a unit of electrical energy equal to the work done by passing a current of one ampere through a resistor of one ohm for one second). Input-equivalent energy is the amount of fuel that thermal power plants need to produce the reported electrical output. Figure 1 shows the wind energy consumption of 37 countries in the period 2000-2019. According to the figure, wind energy consumption is increasing in all countries. Although this amount of increase is low, there is a steady increase compared to the figure.
The carbon emissions used in this study only cover activities related to combustion. For example, natural gas combustion (from 1975) reflects explosion through oil, gas, and coal consumption. These data are based on the "Default CO 2 Emission Factors for Combustion" listed in the IPCC guidelines. Figure 2 shows the CO 2 emissions in the countries. According to the figure, some decline was observed between 2005 and 2015 in Canada, Luxembourg, and Ukraine. In China, India, and Sri Lanka, CO 2 emissions seem to be on an increasing trend. In other countries, CO 2 emissions are almost unchanged.
In Fig. 3, CO 2 emissions and wind energy consumption of 37 countries are given together. CO 2 emissions and wind energy graphs by country are presented in Fig. 4. According to Fig. 3, the regression line with a constant slope has a negative slope. Therefore, the regression line shows that there is a negative relationship between the two variables. Table 1 shows the names and sources of the variables used, while Table 2 shows descriptive statistics. According to Table 2, the volatility in wind energy (WI) is very high. However, as standard deviations show, CO 2 emissions (CO 2 ) Fig. 1 Wind energy consumption of countries are lower than explosive wind energy. The lowest volatility is seen in the globalization (TGI) variable.

Emprical model
The relationship of CO 2 emissions with wind energy consumption, coal energy consumption, total globalization, and economic growth is written as: where CO 2 is per capita CO 2 emissions, WI is wind energy consumption per capita, CE coal energy consumption per capita, TGI is total globalization, and GDP is gross domestic product per capita. The natural logarithm of all variables is taken. The model used in the analysis is shown in Eq. (1), (1) CO 2 = f (WI, CE, TGI, GDP) where i represents the number of cross-sectional (i.e., 1, 2, 3, 4…N) and T indicates the period (2000-2019). lnCO 2it represents the dependent variable carbon dioxide emission; β 0 represents the slope-intercept, β 1 is the coefficient of wind energy consumption per capita, β 2 is the coefficient coal energy consumption per capita, β 3 is the coefficient of total globalization, β 4 is the coefficient of economic growth, and ε it expresses the error correction term. lnCO 2 represents the natural logarithm of CO 2 . lnWI shows the natural logarithm of wind energy. lnCE represents the natural logarithm of coal energy, lnTGI represents the natural logarithm of globalization, and lnGDP represents the natural logarithm of per capita GDP. In studies that analyze the relationship between renewable energy and CO 2 emissions in the literature, the non-renewable energy variable and GDP are often preferred in addition to the renewable energy explanatory variable (Fatima et al. 2021;Mehmood 2021). In this study, the relevant literature is followed. The effect of these variables on CO 2 emissions is expected similar to the literature. For example, wind energy is expected to reduce CO 2 emissions. On the other hand, non-renewable energy and GDP are expected to increase CO 2 emissions. On the other hand, studies such as Zhang et al. (2017), Gözgör and Can (2017), Liu et al. (2020), and Nguyen and Le (2020) emphasize how important globalization is in terms of CO 2 . Therefore, the globalization variable is included in the models as an explanatory variable by these researchers. Therefore, in this study, the globalization variable is included in the model. Furthermore, according to the literature, the effect of the globalization variable on CO 2 emissions is primarily positive. Therefore, in this study, it is expected that globalization will have a positive impact.

Emprical methodolgy
Before starting the analysis of the variables, it is necessary to decide which unit root test will be used first. In cases where there is no cross-section dependence between   series, first-generation unit root tests are used. Secondgeneration unit root tests are used when there is crosssection dependence. In the study, Pesaran scaled LM and Pesaran CD tests suggested by Pesaran (2004) are used to detect cross-section dependence (Pesaran 2004). The Pesaran scaled LM test was obtained by studying Breusch and Pagan (1980) LM test (Breusch and Pagan 1980). Breusch and Pagan's (1980) LM test becomes more suitable for panel studies with N > T property. Pesaran (2004) proposes an alternative Pesaran CD test in Eq.
(3) with the model residues used in the study.
In this test with model residues, more consistent results are obtained for panel cross-section dependence. In Eq. (3), ̂ 2 ij shows the correlation coefficient obtained with the model residues. Second-generation unit root tests proposed by Pesaran (2007) were used for variables that include crosssection dependence (Pesaran 2007). The most practical test among unit root tests for cross-section dependence is the second generation unit root test. Cross-section Generalized Dickey-Fuller (CADF) second-generation unit root test was used to stationarity the series. This test developed by Pesaran (2007) is called the cross-sectional Im et al. (2003) panel (CIPS) test (Im et al. 2003). Pesaran (2007) aims to eliminate cross-section dependence asymptotically in panel analysis with this test. Dickey-Fuller (CADF) regressions are used to eradicate cross-section dependence. The CADF is given in Eq. (4). In Eq. (4), y t−j is the cross-sectional mean of the lagged levels, and Δy t−j is the cross-sectional mean of the first differences of the individual series. CIPS statistics can be estimated after CADF analysis. For this purpose, the lagged variables' t-statistical means (CADF i ) is calculated by Eq. (5). The estimator suggested by Pesaran (2007) is given in Eqs. (4) and (5).
A cointegration test can be performed for variables that become stationary by taking the difference. However, slope parameters and cross-section dependence tests should be performed before proceeding with the cointegration test. Since the data in the panel has the property of N > T, the estimator suggested by Pesaran and Yamagata (2008) can be used. This estimator determines the slope homogeneity using the weighted fixed effect pooled estimator (WFE), OLS, and deviations from the mean (Pesaran and Yamagata 2008). The estimator suggested by Pesaran and Yamagata (2008) is given in Eqs. (6) and (7).
In Eq. (6), i is obtained from the OLS estimate. WFE is the coefficients obtained from the WFE estimation. M shows the identity matrix. x i indicates the processor that is sensitive to deviation from the mean containing explanatory variables. k is the number of regressors, and 2 i is the estimate of i . Cointegration estimation for variables is used estimators proposed by Pedroni (2001) and Kao (1999). However, these methods do not take into account the crosssection dependence and slope homogeneity. The technique developed by Westerlund (2005) takes into account slope homogeneity and cross-section dependence. Equation (8) includes the Westerlund (2005) estimator.
The i term in Eq. (8) is the adjustment used to express the rate variables return to long-term equilibrium. Based on OLS estimates, Westerlund (2005) suggests two-panel statistics based on error correction and panel cointegration statistics with two groups of mean statistics. The fully modified ordinary least squares (FMOLS) method developed by OLS and (Pedroni 2001) can be used to estimate the long-term coefficients of cointegrated variables. The FMOLS method ignores the dependence and slope homogeneity between panel sections while calculating. Cointegration estimation can be made by taking the average coefficients estimated for each section in the panel data. Estimates made by ignoring the dependence between sections may cause erroneous and inconsistent results (Pesaran and Smith 1995) stated. It is an estimator created by the common correlated effect mean group (CCEMG), which estimates cross-section dependence and slope homogeneity. CCEMG, which makes essential predictions in the presence of cross-section dependence and slope homogeneity, includes linear combinations of crosssectional means of observed common effects and variables. In Eq. (9), y it and x it represent the visual elements. f t is the heterogeneous factor with the unobserved common factor, b i is the country coefficient estimates, e it is the error term, and i is the cut-off term. The estimator suggested by Pesaran (2006) is given in Eq. (9).
There is augmented mean group (AMG) prediction analysis developed by Bond and Eberhardt (2013) and Eberhardt and Teal (2010) and similar to CCEMG. Both estimates use cross-section averages for all variables. Unlike AMG, it uses dynamic processes for common factors that are unobservable for different reasons. Long-term coefficients CCEMG and AMG are expected to be cointegrated with the predicted variables.

Estimation results
Before testing the existence of a long-term relationship between variables, cross-section dependence tests of variables should be performed. Table 3 contains the cross-section dependence test results of the variables. Since the probe value (0.0000) is less than 0.05 for all tests in the table, "no cross-section dependence," included as the H0 hypothesis, is rejected. Thus, it is understood from the test results that there is a cross-section dependence between variables.
After cross-section dependency between variables, the second-generation unit root test was applied, which takes into account the cross-section dependency. Finally, Table 4 includes the stationarity test estimates of the variables at the level and the first difference. When the differences of nonstationary variables are taken in level values, they become stationary.
The data in Table 5 shows the homogeneity coefficients of the variables. According to the test results for all variables, the slopes of the variables are heterogeneous since the significance levels are less than 0.05.
Cointegration tests can be carried out for the variables that become stationary. Pedroni and Kao cointegration tests were applied to determine the cointegration between the variables. The results in Tables 6 and 7 show the Pedroni and Kao cointegration statistics values. Since most placement tests are significant in the Pedroni cointegration test, there is a cointegration relationship between the variables. Since the significance level of Kao's (1999) estimation is less than 0.05, the variables are cointegrated (Table 8). In this study, the Westerlund (2007) test, which includes more precise measurements, was also used since cross-sectional dependence and slope homogeneity were considered. Table 6 contains the statistical values of Westerlund cointegration. With the results obtained, a cointegration relationship between the dependent variable and the explanatory variables was determined.
We estimated the long-term relationships and coefficient values of the variables found to be cointegrated. Table 9 contains the results of 3 different tests showing the longterm relationship between variables. These tests are AMG, FMOLS, and OLS tests.
The second column of Table 9 contains the FMOLS test results. According to FMOLS results, wind energy  Table 9 contains the OLS test results. According to OLS test results, there is a negative relationship between wind energy and carbon emissions. A 1% increase in wind energy use reduces carbon emissions by 0.074%. There is a significant relationship between CO 2 emission and CO 2 emission of all variables except the globalization variable. Increasing use of coal, globalization, and growth variables cause increases in carbon emissions. In this study, AMG was also preferred for the estimation of long-term coefficients. The AMG test takes into account the problems of cross-section dependence and slope homogeneity. In panel data analysis, units such as countries are often heterogeneous. The studies with panel data aim to use the information obtained from the sample observations in the parameter estimations in the best way. For this purpose, different models have been created that reflect individual differences or time-dependent differences. The most important feature of these models is that they can predict the unit-specific or time-specific properties of explanatory variables that cannot be observed or measured. Ignoring this heterogeneity causes inconsistent estimates of the parameters of interest. Therefore, the AMG test, which gives more consistent results than FMOLS and OLS results, was also used in the study. According to AMG test results, there is a statistically significant and long-term relationship between the dependent variable CO 2 emission and wind energy consumption, coal energy consumption, total globalization, and growth. Therefore, coal and growth variables affect carbon emissions significantly and positively. Wind energy consumption has a negative and significant effect on carbon emissions. A 1% increase in wind energy use reduces carbon emissions by 0.018%. As shown in Table 9, the coefficient of influence of wind energy on CO 2 emissions according to AMG estimates is slightly lower than FMOLS and OLS estimates.  (2020). There is a significant and positive relationship between globalization, another independent, and carbon release. A 1% increase in the globalization variable increases the carbon emission by 0.10%.

Conclusion and policy implications
Increasing production and consumption levels in the globalizing world in the twenty-first century have led countries to search for new in many areas. In these pursuits, steps have been taken towards meeting the energy demand, which is the basis of the production sector. Countries have met the energy needs of the current generation system from non-renewable energy sources until a specific period. The increasing population in the world and the increasing demands accordingly have led to the search for renewable energy. Renewable energy sources, which have many advantages both economically and environmentally, have attracted the attention of international institutions and organizations. The low cost of renewable energy resources, its minimum impact on environmental pollution, and the reduction of dependence on energy imports are why countries prefer this type. Wind energy is an essential renewable energy source among renewable energy sources regarding energy cost and environmental impact. Wind energy is more advantageous than other renewable energy sources both in terms of raw material and economical. It is expected that wind energy will become even more beneficial in terms of the future. By 2025, wind energy is expected to meet 5% of the world's energy needs. Wind energy has a significant reduction in environmental air pollution and carbon emissions. The benefit of wind energy indicates that it will have more demand in production and consumption areas on a global scale. The study examines the relationship between wind energy, globalization, fossil fuel, and economic growth variables, and CO 2 emissions in the group of countries using wind energy. The AMG estimation method, which solves the cross-section dependence and slopes homogeneity problems, was used for the long-term coefficient estimation between variables-for the analysis, using data from 2000 to 2019 from 37 country groups. Cross-section dependence of all variables is specified. The second-generation unit root test, the CIPS test, is used for unit root testing. The Westerlund cointegration test determined that the variables are cointegrated. The long-term coefficients of the cointegrated variables were estimated using the AMG method. According to the estimates, wind energy has a statistically significant and negative effect on CO 2 emissions in 37 country groups. A 1% increase in wind energy reduces CO 2 emission by 0.018%. On the other hand, a significant and positive relationship has been found between globalization and CO 2 emissions. A 1% increase in the globalization variable increases the carbon emission by 0.10%. Coal and growth variables affect carbon emissions significantly and positively. These results show that the group of countries using wind energy has reduced carbon emissions. The results obtained show how vital wind energy, a renewable energy source, is both economically and environmentally. Governments should prioritize increasing the share of renewable energy sources such as wind energy in energy production. Policymakers should prioritize sustainable energy sources such as wind while designing comprehensive environmental, growth, and energy policies because wind energy is clean and sustainable energy that will meet the targets of decarbonizing the environment. Such renewable energy sources can help make energy policies that can be strategic for countries by using different technologies. The distribution of energy from non-renewable sources in energy production can be made more costly. Enabling renewable energy sources such as wind power to change hands at a lower cost can help achieve a sustainable economic growth path. Wind energy, which is low carbon and sustainable, can be used for balanced development. For this reason, it may be beneficial to identify the sectors that will lead to more efficient and effective use of wind energy. Industry and services sectors to be determined for this purpose can be a driving force in economic growth and development. Public supports that will facilitate research that will make wind energy more efficient and cost-effective will thus contribute to the development of the country's economy. New studies to be done may differentiate the number of countries or the estimation method. In this case, the effect of wind energy on CO 2 emissions may differ. Therefore, the potential for expanding the literature on the impact of wind energy on CO 2 emissions is relatively high.
Author contribution TG and EU analyzed and interpreted the data, and contributed equally to the writing of the article. The authors read and approved the final manuscript.
Availability of data and material All data generated or analyzed during this study are included in this published article (and its supplementary information files).

Consent for publication Not applicable
Competing interests The authors declare no competing interests.