The industrial revolution in the 19th century transformed production processes and enabled the use of, energy in production. The industrialization of more and more countries has continued to increase rapidly in the world’s energy production and consumption, especially in the last few decades. This has put enormous pressure on the ecological balance while providing safe protection for the development of human society (Song et al., 2021). Result of this pressure, the world has been exposed to various environmental problems such as environmental pollution, climate change and global warming (Ozcan et al., 2019; Mujtaba et al., 2020). Especially the large amount of carbon dioxide formed as a result of the use of fossil energy sources entered the atmosphere, causing an increase in the greenhouse effect (Ge et al., 2022). This, in turn, has increased climate problems by causing glaciers to melt, sea level rise, and desertification, abnormal weather conditions (Le Xu et al., 2021). Non-renewable energy, sources such as natural gas, coal, and oil are seen as the most important cause of environmental pollution and global warming (Gyamfi et al., 2021; Usman et al., 2022; Zhang et al., 2023). Under the assumption that nothing has been done about this problem, the probability of a 4°C increase in global temperatures is 75%, and the probability of an 8°C increase is 21% (Waldhoff and Fawcett, 2011). Such a temperature rise could upset the ecological balance of the planet. Considering that the global carbon dioxide emission reached its highest level in 2022 (IEA, 2022), it is necessity to urgently address and implement policies to reduce carbon emissions at the global level.

Renewable energy (such as solar, geothermal, wind, and biomass) can play an important role in reducing carbon emissions (Charfeddine and Kahia, 2019; Chien et al., 2022). Renewable energy sources are a clean alternative to energy sources with high carbon content as they do not release carbon dioxide when converted into energy (Anwar et al., 2021). These clean energy sources are recommended for the realization of production processes with environmental sustainability (Khan et al., 2020). Renewable energy sources, including hydropower and biomass, contribute approximately 17% of the world’s overall energy demand while significantly reducing carbon dioxide emissions (Sun et al., 2022).

Grossman and Krueger (1991) is one of the first studies to examine the relationship between environmental pollution and economic growth. They concluded that carbon dioxide emissions increase with gross domestic product (GDP) per capita at lower national income levels, but environmental pollution decreases at higher income levels. Panayotou (1993); Cole et al. (1997) also confirmed this relationship. Thus, the relationship between environment and income is accepted as the Environmental Kuznets Curve (EKC) in the literature, based on the Kuznets Curve proposed by Kuznets (1955). According to the EKC hypothesis, there is an inverted U-shaped correlation between environmental quality and economic growth. With the increase in income, carbon emissions will increase up to a certain income “threshold” and then carbon emissions will tend to decrease. In this framework, the relationship between carbon emissions and economic growth under the EKC hypothesis has been the subject of many studies in the last 3 decades (Selden and Song, 1994; Lean and Smyth, 2010; Van Hoa and Limskul, 2013; Kais and Sami, 2016; Sinha et al., 2017; Balsalobre-Lorente et al., 2018; Nabavi-Pelesaraei et al., 2018; Kaab et al., 2019; Nabavi-Pelesaraei et al., 2019; Zafar et al., 2019; Cheikh et al., 2021; Wang et al., 2023a).

According to various environmentalists, excessive energy use is the biggest contributor to environmental pollution and environmental degradation, complicating the validity of the EKC hypothesis. For this reason, countries encourage the use of renewable energy sources instead of non-renewable energy sources (Munir, 2022). On the other hand, as the income level increases, citizens will give more importance to environmental pollution and will force the sectors to use clean, that is, renewable energy sources (Xie et al., 2020). Since the production and consumption of renewable energy sources are environmentally friendly, it is also one of the important determinants of breaking the strong link between fuel pollution, CO₂ emissions and the growth of an economy (Balsalobre-Lorente and Leitao, 2020; Akadiri and Adebayo, 2022; Shahbaz et al., 2021; Banga et al., 2022; Sun et al., 2022). Therefore, diversification of renewable energy sources and renewable energy supply plays, an important role in reducing environmental and energy problems (Mert et al., 2019; Zaidi et al., 2019). In addition, renewable energy contributes to sustainable economic development (Apergis and Payne, 2010; Chen et al., 2020; Dogan et al., 2020; Ivanovski et al., 2021) and helps to stabilize both inflation and exchange rates (Deka and Dube, 2021; Deka et al., 2022; Mukhtarov et al., 2022).

Economic income level, population and energy supply are critical in assessing environmental quality. Under the EKC hypothesis, limited attention has been paid to the role of energy supply in carbon pollution. Therefore, unlike previous studies, this study examines the asymmetric relationships between carbon emissions and energy resources on the one hand, and the EKC hypothesis on the other hand, for Organisation for Economic Co-operation and Development (OECD) countries, which consist of countries with significant renewable energy resources. OECD countries have a significant share of the world economy. In 2019, the total GDP of the world was 84.7 trillion dollars. OECD countries have about 61% of this GDP value ($51.5 trillion) (WDI, 2023). CO₂ emissions in the world were 20.3 million kilotons in 1990 and reached 35.5 kilotons in 2019. However, a similar increase was not observed in OECD countries. That is, CO₂ emissions, which were 11.3 million kilotons in total in the OECD countries in 1990, reached 13.3 million kilotons in 2007 and decreased to 11.6 kilotons in 2019 (OECD, 2023). In 2020, when the COVID-19 pandemicbegan, CO₂ emissions fell sharply. By the end of 2020, containment measures related to COVID-19 had reduced emissions by about 7% below 2019 levels. The largest share of this decrease was due to transportation emissions (Le Quéré et al., 2021). Mousazadeh et al. (2021) identified several benefits during the first months of COVID-19 on carbon emissions in the atmosphere. Since electricity generation is considered to be one of the main causes of greenhouse gas emissions, public policies such as mandatory lockdown events during COVID-19 have likely reduced final electricity consumption. Thus, a potential reduction of greenhouse gases is expected. Previous research has also found that government regulations or containment policies in response to the outbreak of COVID-19 can significantly reduce air pollution and reduce CO₂ emissions (Abbas et al., 2021; Dang and Trinh, 2021). On the other hand, renewable energy supply is constantly increasing in OECD countries. While the renewable energy supply was 6.13% of the primary energy supply in 1990, this rate was 10.97% in 2019. All these statistics are of great importance for the study of OECD countries (OECD, 2023).

However, energy supply shocks are a different issue that needs to be analysed. The rapid increase in oil prices in the 1970s adversely affected the output levels of countries. This crisis is one of the best examples of supply shock in the economic literature. Similar types of supply shocks were encountered during the COVID-19 pandemic. The literature has generally focused on the relationship between energy supply and output level or the effect of energy consumption on CO₂ emissions. In this study, unlike other studies, we tried to determine the effect of positive and negative shocks in energy supply on CO₂ emissions. Another feature that distinguishes this study from other studies is that we investigate the positive and negative effects of both renewable and non-renewable energy supply. For this purpose, we had to use a non-linear method to adjustment of the variables from short-run to long-run equilibrium. We also used the nonlinear ARDL approach of Shin et al. (2014) to identify the positive and negative effects of energy supply shocks. In order to make a comprehensive analysis, we included OECD countries in our analysis. Because OECD countries cover 61% of the world GDP and OECD countries have almost doubled the renewable energy supply in total energy supply in the last 30 years. In this context, annual data from 1990 to 2021 for OECD countries were used in the study, and the analyzes were performed with panel nonlinear ARDL regression, unlike previous studies. In addition, Driscoll-Kraay standard errors and Augmented Mean Group (AMG) estimator were used to check the robustness of the findings.

The remainder of this study is structured as follows. The relevant literature review is presented in Section 2. Section 3 describes the economic framework, econometric methodology, dataset and model used in the study. In section 4, we present the empirical results. Section 5 concludes the article by discussing the findings.

Green House Gases (GHGs) are major environmental degradation factors and these factors affect the population’s health directly. One of the reasons of environmental degradation is energy consumption and the main reason of energy consumption is production. Governments aim to develop their economy and increase their output, etc. Both renewable and non-renewable energy resources play a vital role in economic development. Whereas non-renewable energy use is one of the reasons of greenhouse gases, renewable energy use is a critical component of combating global climate change (Uğurlu, 2019; 2022), reducing greenhouse gas emissions.

Because of this relationship, there is a wide literature to investigate the relationship GHGs and renewable or non-renewable energy sources. In this paper we estimate a non-linear model to estimate the relationship, that is why we mainly focus on papers which use non-linear models. Following paragraph, the presented papers show that in the literature different kinds of data (cross-section, time series, panel data) and different kinds of models are used.

Renewable and non-renewable energies are not the only source of GHGs. There are many papers which investigate the relationship between deagriculturalization, financial development and financial development data, etc. (Lin et al., 2022) study the impact of agriculturalization on CO₂ emissions by testing symmetric and asymmetric impact for the selected Asian countries using data from 1985 to 2019 (Boufateh and Saadaoui, 2020). investigates asymmetric financial development shocks on CO₂ emissions; the paper uses a nonlinear panel ARDL–PMG model for a panel of 22 African countries over the period of 1980-2014 (Koondhar et al., 2021). investigate asymmetric causality connection between energy use in agriculture for Pakistan using data period of 1976-2018 (Li X. et al., 2022). estimate panel linear and nonlinear autoregressive distributed lag models with the PMG and MG estimators to assess the financial deepening index’s impact on carbon emissions of BRIC countries for the period of 1990-2019 (Xu et al., 2022). dealt with G7 countries to examine the non-linear and asymmetric relationship between reorientation in financial development and renewable energy generation process which ensures environmental sustainability between renewable energy and CO₂ emissions between 1986 and 2019.

Similar to our paper (Zaidi et al., 2018), and (Anwar et al., 2021) investigate renewable and non-renewable energy; whereas we use energy supply they use energy consumption (Onuoha et al., 2022), investigate both renewable and non-renewable energy sources (they used fossil fuel data) in relation to the EKC hypothesis for 15 ECOWAS ¹ countries by estimating the PNARDL model (Zaidi et al., 2018). estimate ARDL model for Pakistan from 1970 to 2016 (Anwar et al., 2021), estimate Fully Modified Ordinary Least Square (FM-OLS), Dynamic-OLS and Fixed-Effect-OLS models for ASEAN economies from 1990 to 2018 (Toumi and Toumi, 2019). use the NARDL model by using renewable energy, carbon dioxide emissions, and real GDP variables of the Kingdom of Saudi Arabia between 1990 and 2014 (Munir and Riaz, 2019). examines relationship of energy consumption and environmental quality (CO₂ emissions) in three South Asian countries (Bangladesh, India, and Pakistan) from 1985 to 2017. The authors use the PARDL model and estimate the model separately for each energy source such as oil, coal, gas, and electricity consumption (Akram et al., 2020). assess asymmetric impacts of energy efficiency and renewable energy, and other factors on CO₂ emissions in BRICS (i.e., Brazil, Russia, India, China, and South Africa) countries from 1990 to 2014 (Mosikari and Eita, 2020). test the EKC for the period 2005–2019 from 29 selected African countries by using the panel smooth transition regression (PSTR) technique (Kartal et al., 2022). use monthly data to estimate the asymmetric effect of electricity consumption on carbon dioxide in the United States, by collecting monthly data from January 1973 to November 2021 (Akadiri and Adebayo, 2022). uses ARDL model to capture relation between CO₂ emissions and renewable energy consumption and non-renewable energy consumption in India employing a dataset from 1970 to 2018 (Munir, 2022). aim to show existence of nonlinear effect of energy use on CO₂ emissions by using NARDL model for 21 EU countries from 1990 to 2018 (Akram et al., 2022). select MINT countries (Mexio, Indonesia, Nigeria and Turkey) to investigate energy efficiency and renewable energy impact on CO₂ emissions in 1990–2014) (Mujtaba et al., 2022a). use data over the period 1971-2014 to see the impact of economic growth, energy consumption, and population (POP) on CO₂ emissions in the five regions ² (Majeed et al., 2022). address the effect of nuclear energy on CO₂ emissions for Pakistan from 1974 to 2009 by using ARDL and NARDL models (Qamruzzaman, 2022). considers 30 nations from low-income countries and 38 Lower-Middle-Income countries to investigate the renewable energy carbon emission relationship from a different perspective by using renewable energy used in agricultural activities from 1985 to 2019 (Zhang et al., 2022). select institutional factors, GDP per capita, energy consumption, and foreign direct investment of BRICS countries (Brazil, Russia, India, China, and South Africa) to test the asymmetric effect of the variables on CO₂ emissions by using the Panel ARDL-PMG model from 1996 to 2019 (Iqbal et al., 2022). states that some natural resources can add to the current stock of CO₂ emissions and hurt the environmental quality, therefore the authors use natural resources, and economic progress in their research model which aims to analyze the asymmetric relationship between renewable energy production and CO₂ emission in Pakistan from 1980 to 2019 (Caglar et al., 2022). collected data from BRICS countries to investigate the relation between CO₂ emissions and renewable energy with some other control variables which are foreign direct investments, and natural resources. In the empirical section, balanced data was used in the NARDL model; 1992-2018 data for Russia, and 1980-2018 data rest of the BRICS countries (Sun et al., 2020). compare OECD and Belt and Road (B&R) countries over the period 1992-2015 based on their energy consumption, trade openness, urbanization and CO₂ emission relationship by estimating panel common correlated effects mean group (CCEMG) and augmented Mean Group (AMG) estimators (Rahman et al., 2022). claim that although economic growth has increased in 22 well-developed countries the level of CO₂ emissions has decreased. Based on this idea the authors aim to examine the affected factors of CO₂ emission which are renewable energy, technological innovation and export quality. To do this examination they use 22 selected countries ³ and estimate the NARDL model for the period 1990-2018 (Saqib, 2022; Saqib et al., 2022). investigate asymmetric linkages between renewable energy, technological innovation, and carbon-dioxide emission and EKC hypothesis in the context of renewable energy; the first paper uses NARDL for 18 developed economies ⁴ and the second paper use cross-sectional augmented autoregressive distributive lag (CS-ARDL), Augmented Mean Group (AMG), and Dumitrescu Hurlin causality tests for E7 economies ⁵ (Adebayo et al., 2023). use cointegration test and NARDL and addition they use Dynamic Multiplier and Spectral Causality for Turkey.

(Li et al., 2021; Li et al., 2022a; Li et al., 2022b; Wang et al., 2023b) are the some of the lastest panel data studies about relationship between carbon emissions and economic factors (Wang et al., 2023a).is one of the panel data stuies and generally confirm the EKC; in 2018, 72 countries of 208 countries reached the EKC turning point for per capita income. In (R. Li et al., 2021) investigated 147 countries consist of different income groups and the results show that the lower middle income group countries have mostly bidirectional causality, while the lower middle income group coountries have unidirectional relationship between energy structure and the selected variables (Wang et al., 2022). focus on urbanization and claim that trade openness and natural resource rents increase environmental pressure, population aging and renewable energy decrease environmental pressure (Li et al., 2022b). presents new perspective the 3E model, and assert that negative effect of renewable energy on the ecological footprint first weakens and then increases with the increase of urbanization (Li et al., 2022a). is the another panel data paper but use provinces of China instead of world countries. The authors find that that the decoupling between income growth and CO2 in the Chinese transport sector is particularly poor.

Dilanchiev et al. (2023) proves that relationship between renewable energy production and GDP per capita and is significant and negative in Romania, Azerbaijan, Russia, Turkey, Bulgaria, and Greece. Sun et al. (2020) investigate relationship between natural resource rent (NRR) management, green technology innovation (GTI), and GDP growth and the results show that NRR and GTI decrease carbon emission. Asif et al. (2023) use questionaire to collect data and estimate structural equation modelling for the analysis and the findings show that awareness of environmental factors and a positive attitude support to consumers adopting renewable energy.

Some of the research which are mentioned above are confirmative to each other results but some of them are contradictory (Toumi and Toumi, 2019). find that in the long run, the asymmetric causal connection between carbon dioxide emissions and renewable energy is neutral, and both positive and negative shocks on renewable energy consumption have an adverse impact on CO₂ emissions. The results of (Kartal et al., 2022) show the important asymmetric effect on CO₂ emission from different kinds of electricity consumption (residential, commercial, industrial and transportation) and they are very important factors to protect the environmental quality of the United States. (Adebayo et al., 2023) assert that positive (negative) shocks in renewable energy usage and a positive shock in structural change decrease emission levels in Turkey (Akram et al., 2020). find that positive shocks in energy efficiency and fluctuation in the positive component of renewable energy reduce CO₂ emissions (Saqib, 2022). recognize that renewable energy is a significant variable to reduce CO₂ emissions (Saqib et al., 2022). supports the result of renewable energy decreasing the effect on environmental degradation (Akadiri and Adebayo, 2022).’s estimations show a negative relationship between renewable energy consumption and carbon emissions in India. The PSTR technique, in the research of (Mosikari and Eita, 2020), indicates that in both regimes, energy consumption has an increasing effect on carbon emissions that is why African countries should reduce their non-renewable energy sources (Sun et al., 2020). and (Munir and Riaz, 2019) show that energy usage has a nonlinear connection with CO₂ emissions. Also (Munir, 2022), estimates a non-linear model and finds coal, oil, gas and electricity consumption increases CO₂ emissions (Caglar et al., 2022).’s results show contradictory results which show a positive insignificant impact on CO₂ emission from renewable energy consumption; the authors state that the reason for insignificant impact is renewable energy consumption has not reached the considered level in BRIC countries. Moreover (Rahman et al., 2022) find that technological innovation and renewable energy use are valid ways to reduce carbon emissions.

This study aims to examine the Environmental Kuznet Curve (EKC) hypothesis in OECD countries by adding asymmetric effects of energy supply components on CO₂ emissions. Our main objective is to address, for the first time in the literature, the effects of asymmetric components of energy supply on CO₂ emissions in OECD. For this purpose, we use a nonlinear empirical framework and build an asymmetric panel ARDL model, to identify the impact of energy supply components on CO₂ in OECD countries, based on the nonlinear ARDL approach of Shin et al. (2014). Moreover, the adaptation of the EKC hypothesis test with asymmetric effects for the OECD constitutes the empirical infrastructure of the study. In addition, population density is included in the model to eliminate the possible effects of country size.

We began analysing the empirical findings with the summary statistics of the variables and their correlation linkages. The tables below define the variables respectively.

Table 2 shows that the highest standard deviation value is the standard deviation of population density, followed by the series of increased and decreased RE. This ranking is not surprising because population density is affected by the country’s area and the total population. Therefore, it reflects the geographical scale of countries. On the other hand, renewable energy supply is sensitive to weather conditions as it includes climatic features such as solar and wind. The volatility and standard deviation of carbon emissions, GDP and NRE supply series are relatively stable.

The correlation coefficients between primary energy supply (renewable and non-renewable) and CO₂ emissions are statistically significant and range between 6% and 49%. GDP variables exhibit a highly positive and significant relationship between energy supply indicators and CO₂ emissions. Table 3 summarizes correlation relations.

The first step of the empirical investigation started with the cross-sectional dependency (CSD) analysis, as it should be considered when conducting other analyses. Pesaran (2004) CD test is used in this study to examine the cross-sectional correlation between variables. CSD test statistics are presented in Table 4.

Autocorrelation and heteroskedasticity tests were applied to ensure the reliability of the results. The statistics from the residual diagnostic tests are presented in Table 5. The findings suggest that the null hypotheses should be rejected for both tests.

The CADF unit root test results are presented in Table 6 and show that all series for OECD are non-stationary. When the first differences are taken, they become stationary with a 5% significance level. The findings show that the series are I (1).

Table 7 shows the Pedroni (1999); Pedroni (2004) and Westerlund’s (2005) cointegration test statistics. The result indicates that lung run relationships exist among variables. Following Levin et al. (2002) algorithm, cross-sectional means were removed from the series to mitigate the impact of cross-sectional dependence. The Pedroni test statistics presented in the table reject the null hypotheses at the 1% significance level and indicate cointegration for the long run, and Westurlund test statistics indicate at 10%.

Causality relationships were investigated with the Juodis, Karavias and Sarafidis (2021). In general, this novel methodology developed by Juodis et al. (2021) offers several advantages, such as overcoming the “Nickell” bias, allowing cross-sectional dependence, and being suitable for homogeneous and heterogeneous panels. Granger non-causality test and the statistics are presented in Table 8.

The test statistics for all models indicate that the null hypothesis cannot be rejected and, therefore there exists a Granger causality relationship among the variables.

To assess the asymmetric effects of RE-NRE on CO₂ and to test the EKC hypothesis (see Eq. 1), we estimate panel nonlinear autoregressive distributed lag (PNARL) coefficients using the algorithm of proposed Shin et al. (2014). The findings are presented in Table 9.

Since one of the most important objectives of the study is to examine long-run asymmetries, Table 10 presents the results of the Wald test for the validity of these relationships. According to the Wald statistic for asymmetry, the null hypothesis is rejected for the long-run coefficient of non-renewable energy in model 5. The findings show that there is a valid asymmetric effect for non-renewable energy in the long run. In the short run, no such relationship is detected.

We will first interpret the findings of Model 5 and the other models can be interpreted by readers using the same logic. The error correction term (ECT), which indicates the cointegration relationship between the variables and the speed at which the system reaches equilibrium in the long run, is negative and significant. ECT is the coefficient that indicates how fast the variables in the model adjust to the long-run equilibrium. A negative sign of ECT represents the convergence in long-run equilibrium. The ECT term magnitude implies that the system returns to long-run equilibrium after a shock in approximately 4.2 years (50 months) with an annual correction of 23.7%.

The long-run coefficients in the model are statistically significant and have the theoretically expected signs. The findings showed that a positive and negative shift in NRE supply contributes to increased carbon emission and consequently reduces environmental quality. In contrast, however, a positive and negative shift in RE supply reduces carbon emissions and hence strengthens environmental quality. This finding points to the asymmetry between the two main components of energy supply and carbon emissions. According to the long-run estimates, a positive shock in RE induces a positive and statistically significant effect of 0.0539 on carbon emissions. A negative shock of RE causes a negative impact of 0.0538 on carbon emissions. The findings demonstrate that a positive shock to RE has a slightly higher impact on carbon emissions. It is clear from the magnitude of the asymmetric coefficients for NRE that its impact on CO₂ is quite high. These findings suggest that the long-run impact of positive shocks in both renewable and non-renewable energy supplies on carbon dioxide emissions is not similar to that of a negative shock, and there are asymmetric effects in the long run.

Moreover, carbon emissions positively respond to increases in GDP; a 1% increase in GDP leads to a positive and statistically significant increase in carbon emissions by 0.967 percent.

This study investigates the short- and long-term asymmetries between CO₂ emissions, RE and NRE under positive and negative shocks for the panel data series of 38 OECD countries between 1990 and 2021, as well as the validity of the Environmental Kuznets Hypothesis with panel data econometrics. While many studies in the literature address asymmetric effects and EKC as energy consumption or utilization, the novelty of this study lies in approach the issue in terms of energy supply along with asymmetric effects for RE and NRE.

Negative and positive shocks in RE and NRE energy supply significantly affect CO₂ emissions in the long run. However, only NRE is effective on emissions in the short run. While NRE has a powerful effect on increasing emission, RE’s CO₂ emission-reducing impact is limited. Moreover, Wald test results indicate the existence of an asymmetric link between NRE and emissions in the long run. This finding raises questions about the effectiveness of measures taken against environmental degradation, even in OECD countries known for their high environmental awareness. The findings coincide with the findings of Munir (2022) in the NRE subgroups in EU countries regarding positive shocks but differ in terms of negative shocks. Moreover, the findings support the NRE and RE consumption coefficients of Mujtaba et al. (2022b) regarding symmetric and asymmetric effects for the OECD. However, while our results are quite consistent for RE, our study’s magnitude of NRE effects is lower.

The coefficient of GDP is 0.967, and the coefficient of the quadratic term is −0.058, and both coefficients are significant at the 1% level. The positive coefficient of the primary term and the negative coefficient of the quadratic term indicate the existence of an inverted U-shaped curve between economic growth and environmental pollution, supporting the validity of the EKC hypothesis for the study period in the OECD countries. The turning point of the Environmental Kuznets Curve is the calculated per capita income level of $4085.77. The findings on EKC are consistent with Sun et al. (2020); Fakher and Inglesi-Lotz (2022).

However, according to 2021 data, only 11.2 percent of the total energy supply in OECD countries is comes from renewable sources. This situation indicates the insufficiency of renewable resources in reducing emissions and the need for more technology, R&D and fixed capital investments in this field. Moreover, to reduce emissions and change the composition of the energy supply in favour of renewable resources, economic policies that encourage the use of renewable energy resources instead of traditional ones should be pursued.

Across the OECD, approximately 11.5 per cent of energy is provided from renewable sources. When analysed on a country basis, it is seen that this ratio rises to 88% in countries rich in renewable natural resources (e.g., Iceland). In comparison, it falls to about 2% in countries poor in natural resources (e.g., South Korea and Israel). These countries have the technical know-how and economic infrastructure to afford renewable energy investments. Therefore, the insufficiency of renewable energy supply cannot be explained only by the insufficiency of natural resources. This situation can be explained by similar countries’ environmental insensitivity and high dependence on non-renewable resources. The best policy to overcome this dependence may be to identify the renewable energy resources in which the country is relatively superior and to concentrate new energy investments in these areas.

Governments should develop incentives at both macro and micro levels to increase investments in renewable energy products. The government could provide tax exemptions to firms that adopt a certain percentage of renewable energy sources in their production and operation processes. Taxes and restrictions on the trade of renewable energy-intensive products could be lifted to increase the accessibility of these products to the final consumer. In addition, tariffs on carbon-emitting products could be expanded to impose a carbon tax on these products. This would shift production towards renewable energy-based areas. The contribution of transport services to carbon emissions is significant. Policies to encourage public transport should be followed to minimise emissions in this area. At the same time, carbon emissions in transport can be minimised by implementing green and electrical transport systems.

As in any empirical analysis, the study’s limits are directly proportional to the data’s accuracy and the model’s consistency. We restrict the period to a specific year and countries to the OECD to minimise these limitations. We also subject the results to robustness checks (see Tables A1, A2).

It is possible to improve our findings with further studies. Further studies in subgroups and periods will be helpful to confirm these findings. Future studies could also focus on asymmetric effects according to energy supply sources.