Introduction

In recent years, researchers in energy, environment, and science have increasingly discussed the challenging implications of climate change for future human well-being, economic growth, and environmental sustainability in the context of global warming (Kirikkaleli and Adebayo 2021a). What we all know is that greenhouse gases (e.g., CO2, CH4 and N2O) are the main cause of global warming, which not only causes various environmental problems, but also reduces the carrying capacity of the earth and hinders sustainable development (Salem et al. 2021). In addition to the greenhouse effect, it is widely believed that this climate change risk is directly and closely related to the large-scale burning of fossil fuel energy, so it has attracted widespread attention from the perspective of economic and environmental sustainability (Martins et al. 2021). According to the 2020 Emissions Gap Report released by the United Nations Environment Programme, despite the reduction in CO2 emissions in 2020 due to the COVID-19 outbreak, the concentration of the main greenhouse gas (CO2) produced in the atmosphere continues to rise in both 2019 and 2020, leading to a global temperature rise of more than 3 °C (Hao, 2022a). For the global temperature continues to rise, it is likely to lead to catastrophic weather events, ozone depletion, and ecosystem degradation, which will pose a serious threat to human production and life. In response, Inger Andersen, Executive Director of the United Nations Environment Programme, also said that as economic globalization continues, the impact of CO2 emissions on the environment will continue to intensify and deal the heaviest blow to sustainable development in developing and less developed countries. Therefore, researchers and decision-makers from many nations have underlined the significance of lowering greenhouse gas (GHG) emissions in order to prevent the calamity that global warming brings to human society (Amran et al. 2014; Anwar et al. 2022a), but how to reduce greenhouse gas emissions has become a pressing issue for countries around the world.

Carbon dioxide emissions have had a key impact on ecological environment and sustainable economic development throughout the history of human development, yet there is still a great deal of debate regarding global climate change (Hotak et al., 2020). According to the United Nations Environment Programme’s 2020 Emissions Gap Report, although the Covid-19 pandemic has quickly reduced CO2 emissions, overall global carbon emissions are still at high levels and the concentration of carbon dioxide emissions continues to increase (Kang 2021). If left unattended, future catastrophic weather events, ozone depletion, and ecological degradation are all anticipated to result from this, posing grave risks to human life and production (Hasnisah et al., 2019). In this context, the international community is engaged in a heated debate on carbon neutrality in response to the climate crisis. Carbon neutrality ensures that net carbon emissions from economic activities are zero, so that the concentration of CO2 emissions no longer increases, also known as “net zero emissions.” In real life, carbon neutrality can be achieved through afforestation, energy conservation and emission reduction, and the use of negative emission technologies (e.g., biochar and bioenergy with CO2 capture and storage) to offset the carbon dioxide or greenhouse gas emissions generated by oneself, achieving positive and negative offsetting, and further achieving the goal of relative “zero emissions” (see Fig. 1).

Fig. 1
figure 1

Carbon neutrality in today’s climate (Sustainability Report 2015)

Full size image

Nevertheless, in the context of globalization, greenhouse gas emissions (e.g., carbon dioxide) are an inevitable consequence of a country’s economic activities, while energy consumption is the core of economic growth (Saidi and Hammami 2015; Tong et al. 2020; Kang 2021). According to the “2020 BP World Energy Statistics” and “Global Energy and Carbon Dioxide Status Report,” global carbon emissions mainly come from CO2 emitted from the combustion of fossil fuels in various economies around the world. Global energy consumption grew by 1.3% in 2019, almost double the average growth rate since 2010 (IEA 2020), but less than half the 2.8% increase in 2018, with China and India accounting for more than 80% of this growth. On the other hand, the global economy grew by 2.6% in 2019, compared with 3.7% in the previous year and an average growth of 3.2% in the previous 10 years, driving the increase in global energy demand, especially for fossil fuel energy. Fossil fuel energy consumption accounted for 78.9% of total global energy consumption in 2019, an increase of 0.7% compared to 2018. Among them, 7 emerging countries (Brazil, China, India, Indonesia, Mexico, Russia, and Turkey) account for more than 42% of fossil fuel consumption, far exceeding the G7 countries (Dong et al. 2020). In addition, according to the “Global CO2 Emissions Report 2019,” the economic growth rate of developed economies averaged 1.7% in 2019, but total energy-related CO2 emissions fell by 3.2%. However, the electric power sector has led the decline resulting in currently 36% of energy-related emissions across developed economies, but more than 47% of CO2 emissions in E7 countries (IEA 2020). As a result, the E7 countries are playing an increasing role in the world energy market and climate change, both in terms of CO2 emissions and energy consumption (see Figs. 2 and 3). Based on the above, the current challenge for E7 countries is to find reliable and affordable energy sources to replace fossil fuel energy sources while reducing greenhouse gas emissions (Jaforullah and King 2015).

Fig. 2
figure 2

Portion of E7 countries in total CO2 emissions (BP-Statistics 2021)

Full size image
Fig. 3
figure 3

Portion of E7 countries in total energy consumption (BP-Statistics 2021)

Full size image

There are several reasons why we focused our study on E7 countries. Firstly, since 2010, the economic development of emerging market economies (e.g., E7 countries) has promoted the rapid growth of energy consumption and CO2 emissions, becoming the main driving force for the future growth of global CO2 emissions (Anwar et al. 2022b; Ojekemi et al. 2022). However, E7 countries generally face the problems of different accounting calibers, imperfect scales, and inconsistent methodologies for CO2 emissions inventories. The lack of basic data has also become a major obstacle for studying the characteristics of CO2 emissions in E7 countries, which limits the research and policy discussion on the low-carbon development path of E7 countries to a certain extent. Secondly, the E7 countries are the seven developing countries in the G20 with relatively high economic growth rates, high economic openness, broad representation, and comparable economic and population size. In the past three decades, the E7 countries have been particularly prominent in terms of their share of global output, becoming the new engines of global economic growth and green innovation, but at the same time the fastest growing group in terms of global energy consumption and CO2 emissions. Thirdly, compared with developed and developing economies, these countries have developed faster in industrialization, and the excessive consumption of energy in the process of producing a large number of goods and services has led to increased environmental pollution. To reduce CO2 emissions, E7 countries have intensified their investments in green innovation and renewable energy in the last decades. In particular, green innovation and renewable energy investments in China and India have grown by more than 38% and 42%, while investments in all emerging countries have also grown by more than 16% and 30%. Finally, according to PwC’s forecast (The World in 2050, 2017), the average annual economic growth rate of the E7 countries by 2050 is more than twice that of the G7 countries. For example, the seven largest emerging market countries (E7), namely, China, Brazil, India, Indonesia, Mexico, Russia, and Turkey, will have an average annual economic growth rate of 3.5%, but the seven largest industrial countries (G7), namely, the USA, the UK, Canada, France, Germany, Italy, and Japan, will have an average annual economic growth rate of only 1.6%. In addition, the report also states that by 2050, E7 countries will account for half of the global economy in terms of GDP levels, while the share of G7 countries will shrink to 20%. Therefore, the study of the correlation between renewable energy consumption, green innovation, economic growth, and environmental pollution has significant theoretical and practical significance for promoting sustainable development of 7 emerging market economics, which will provide a certain reference value for helping other emerging market economics formulate emission reduction policies.

The rest of this paper is structured as follows. After this introductory section (“Introduction”), “Literature review” provides a critical review of previous studies on the renewable energy, green innovation, economic growth, and CO2 emissions. “Methodology and data” presents the data and discusses the methodology. The empirical results and discusses are set out in “Empirical results and discussions.” Finally, “Conclusions and policy implications” concludes by setting out the key findings and wider policy implications.

Literature review

In recent years, the problems caused by the continuous increase in global warming have seriously affected socioeconomics, human health, population migration, food security, and terrestrial and marine ecosystems. Therefore, this chapter is devoted to summarizing (reviewing) literature on the impact of renewable energy consumption, green innovation, and economic growth on CO2 emissions (see Fig. 4).

Fig. 4
figure 4

The conceptual framework

Full size image

Relationship between renewable energy consumption and CO2 emissions

Since the 1990s, the massive emissions of greenhouse gases, mainly carbon dioxide, have led to a frequent occurrence of extreme weather around the world. As a result, the debate among researchers around the world on the accelerated development of renewable energy and its impact on environmental quality has been intense and has not reached a consensus so far (Kahia et al. 2017; Bai et al. 2021; Salem et al. 2021; Djellouli et al. 2022). From the perspective of global climate change, the use of renewable energy is considered to be the most effective and direct means of reducing atmospheric CO2 concentrations and has a significant impact on environmental sustainability (Kirikkaleli and Adebayo 2021a, b; Sun et al. 2022). However, the empirical results are mixed due to the characteristics of adopted technology, key countries or regions, research cycle, and economic variables. For example, the consumption of renewable energy significantly reduces CO2 emissions, according to Omri and Nguyen (2014), who looked at the factors influencing this consumption in 64 nations between 1990 and 2011. To this end, Anwar et al. (2022a) and Yunzhao (2022) found a positive contribution to carbon emission reduction by examining the role of renewable energy in carbon reduction in E7 countries. This is consistent with the findings of Lei et al. (2022) for China and Ojekemi et al. (2022) for the BRICS countries, but they found that negative shocks to renewable energy consumption lead to increased pollution emissions in the long term. On the contrary, Padhan et al. (2020) used the Machado and Silva Panel quantile regression method to study the relationship between renewable energy consumption and CO2 emissions in OECD countries from 1970 to 2015. They have found a long-term association between renewable energy consumption and per capita carbon emissions, and that per capita carbon emissions had a positive effect on renewable energy consumption in these countries. Moreover, Djellouli et al. (2022) concluded that there is cross-independence between carbon emissions and renewable energy and that renewable energy has a significant negative impact on CO2 emissions. In addition, some researchers have also found a feedback causal relationship between renewable energy consumption and CO2 emissions. For example, Musah et al. (2020) explored the relationship between carbon emissions and renewable energy consumption in West African countries during 1990–2018 using the CCEMG and DCCEMG estimation methods. They have found a feedback causal relationship between renewable energy and CO2 emissions in West African countries. Çıtak et al. (2021) used a NARDL estimation method to explore the asymmetric impact of renewable energy on carbon dioxide emissions in selected ten most populous states in the USA over the period 1997–2017. They have found a long-term relationship between renewable energy and CO2 emissions in the eight states used in the research and support the feedback hypothesis. On the other hand, from a non-economic standpoint, the development of green and clean energy offers a number of economic and environmental benefits. These economic advantages include, but are not limited to, addressing a variety of issues, including energy security and portfolio diversification, as well as job creation, as renewable energy is more labor- and technology-intensive than the non-renewable energy sector (Blazejczak et al., 2014). Therefore, we propose the following research hypothesis:

  • Hypothesis 1. Renewable energy consumption significantly reduces CO2 emissions in E7 countries.

Relationship between green innovation and CO2 emissions

The sustainable global economy is severely constrained by the increasing environmental pollution and greenhouse effect (Li et al. 2020). Therefore, under the background of “carbon neutrality,” governments around the world have formulated corresponding environmental regulations to reduce pollutant emissions to promote environmental technology innovation, industrial structure upgrade, and sustainable development of green economy (Hsu et al. 2021). However, Brunel (2019) suggested that in addition to measuring environmental policy, innovation needs to be measured. For example, Chen et al. (2006) explored the impact of green innovation on competitive advantage, which they consider to be equivalent to hardware or software innovation related to green products or green processes. Regardless of the way in which green innovation is defined, the general end purpose is related to technological progress with environmental effects, in the form of new products or processes that contribute to environmental protection and sustainable development (Liu et al. 2022a, b).

The available literature has extensively explored green product innovation from an environmental sustainability perspective (Luttropp and Lagerstedt 2006; Dangelico 2016). Mensah et al. (2019) used the ARDL method to examine the impact of innovation on CO2 emissions in OECD countries, and found that green innovation reduced CO2 emissions to a certain extent and supported the feedback hypothesis. However, Liu et al. (2022a, b) argue that the emission reduction effect of green innovation (environmental innovation) is only significant at higher levels of carbon emissions. Based on the above research, Li et al. (2022) used the NARDL method to examine the impact of green innovation on China’s CO2 emissions from 1991 to 2019, and found that green innovation plays an important role in creating jobs, improving green economic activities and enhancing environmental sustainability. Moreover, in the long term, the improvement of green innovation will reduce China’s CO2 emissions, whereas the decline in green innovation will increase China’s carbon dioxide emissions. Similarly, Doğan et al. (2022) examined the impact of green innovation on environmental quality in E7 and G7 countries using an augmented model, which found that green innovation reduced carbon emissions in G7 countries, but exacerbated environmental problems in E7 countries. This view is supported by the findings of Yuan et al. (2021), Habiba et al. (2022), Khattak et al. (2020), and You et al. (2022) for China, Top 12 Carbon Emitters, OECD countries, and the USA, respectively, which vary widely among countries. Furthermore, Ganda (2019) used the GMM estimation method to examine the impact of green innovation and technology investment on CO2 emissions in selected OECD economics from 2000 to 2014 and found that green innovation and technology investments in these countries have different impacts on CO2 emissions and still have the potential to reduce environmental quality. Khattak et al. (2020) investigated the complex interplay between green innovation and CO2 emissions in BRICS economies from 1980 to 2016 within the framework of the environmental Kuznets curve. They have found that green innovation activities in China, India, Russia, and South Africa had a suppressive effect on CO2 emissions, but had a positive impact on CO2 emissions in Brazil. Therefore, we propose the following research hypothesis:

  • Hypothesis 2. Green innovation significantly reduces CO2 emissions in E7 countries.

Relationship between economic growthand CO2 emissions

Reviewing some past research efforts, researchers in various countries have mainly focused on studying the relationship between CO2 emissions and economic growth in a single context, especially CO2 emissions due to energy use, but the empirical results are contradictory. For example, Xiong and Xu (2021) used the ARDL bounds test method to examine the relationship between energy use, economic growth, and environmental pollution in China from 1995 to 2015. The research found that industrial growth has a positive impact on CO2 emissions and environmental degradation is caused by economic growth, which are consistent with the findings of Wang et al. (2016) for China, but they argued that the impact of CO2 emissions shocks on economic growth or energy consumption is marginal. Not only that, some researchers have also found the “EKC” hypothesis between environmental pollution and economic growth, especially the inverted “U-shaped” relationship between CO2 emissions and income levels. For example, Danish et al. (2019) and El Menyari (2021) used the environmental Kuznets curve (EKC) model to examine the relationship between economic growth and CO2 emissions in the BRICS and four North African countries (namely, Morocco, Algeria, Tunisia), respectively. The study found an inverted “U-shaped” relationship between environmental quality and real income (income per capita), but the relationship was initially negative. However, the findings of Laverde-Rojas et al. (2021) for Colombia reject this view, who argue that for developing countries like Colombia, the “EKC” hypothesis does not exist and does not benefit from increased economic complexity. Moreover, some researchers also found that the balance between environmental protection and economic growth has been a great challenge for policy makers (Zheng et al. 2015). Liu et al. (2020) examined the relationship among economic growth, governance, and carbon dioxide emissions within the framework of the “EKC” in five high carbon dioxide–emitting countries from 1996 to 2017, and found that political, economic, and institutional governance can improve the quality of the environment. Espoir et al. (2022) used panel and time-series techniques to examine the heterogeneous effects of CO2 emissions and temperature on incomes in Africa from 1995 to 2016. It was found that environmental policies specifically designed to reduce CO2 emissions across Africa could have a significant impact on production in the long term. As such, they emphasize country-specific policies aimed at sustained reductions in CO2 emissions in Africa, rather than global climate policies. Therefore, we propose the following research hypothesis:

  • Hypothesis 3. Economic growth positively moderates green innovation’s reduction of CO2 emissions in E7 countries.

In summary, researchers in energy, environment, and science have done a lot of research on the relationship among renewable energy consumption, green innovation, economic growth, and environmental pollution, but there are still some deficiencies: (1) Existing studies typically use a single variable to examine the effects of renewable energy consumption, green innovation, and economic growth on CO2 emissions, but ignore the impact of macroeconomic variables on CO2 emissions. (2) Most studies only focus on the direct effects of renewable energy consumption, green innovation, and economic growth on CO2 emissions, and do not reveal the long-term and short-term dynamic effects of renewable energy consumption, green innovation, and economic growth on CO2 emissions under the framework of the EKC hypothesis. Based on this, an in-depth analysis of the relationship between green innovation and environmental sustainability in the economies of E7 countries in a two-carbon context is necessary, an issue that has previously been neglected in existing studies. Importantly, the current study captures the joint effects of green innovation, renewable energy, and economic growth on CO2 emissions. Therefore, this study aims to fill the literature gap on the impact of economic growth, green innovation, and renewable energy on CO2 emissions in a global context. From a macroeconomic perspective, the long-term interconnection between economic growth, green innovation, and renewable energy with CO2 emissions in E7 countries from 1990 to 2020 is examined respectively with OLS, FMOLS, DOLS, and CCR estimators. Likewise, it employs the ARDL bounds test to examine the effects of differentiating short-term and long-term at different frequencies, which will help policy makers formulate sound environment-related policies for the sustainable development of “energy-green economy innovation-environment” in individual countries. The conceptual framework in this study is shown in Fig. 4.

Methodology and data

Econometrics methodology

Unit-root test

To avoid the “pseudo-regression” problem, it is necessary to test the data for stationarity, and different methods of testing for stationarity may lead to different results (Hao, 2022a). For this purpose, four unit root tests (the null hypothesis of the KPSS test is that the sequence has a unit root, and the null hypothesis of the other three tests is that the sequence does not have a unit root) were used in this study, namely, Dickey and Fuller (1979, ADF), Phillips and Perron (1988, PP), Kwiatkowski et al. (1992, KPSS), and Elliott et al. (1992, DF-GLS), which can make the test results more convincing. Since the ADF test and PP test may be ineffective for small-sample data, the KPSS stationarity test is more effective for small samples when choosing lower lag truncation parameters (Sabuhoro and Larue 1997). Therefore, this study will take advantage of the KPSS statistic to judge the stationarity of the data, and the test equation is as follows.

$${y}_{t}=\alpha +\beta t+d\sum_{i=1}^{t}{u}_{i}+{\varepsilon }_{t} , t=\mathrm{1,2},\dots ,T$$
(1)

Bayer and Hanck combined cointegration

Intending to get rid of the defects of traditional cointegration test methods, Bayer and Hanck (2013) proposed a new method of combined cointegration test, namely, the Bayer and Hanck (2013) combined cointegration test method (Rjoub et al., 2021), which is an extended combination of the cointegration test methods proposed by Engle and Granger (1987), Johansen (1991), Boswijk (1994), and Banerjee et al. (1998). As stated by Kirikkaleli and Adebayo (2021a), the Bayer–Hanck combined cointegration test method is mainly to eliminate unnecessary multiple testing techniques, thereby giving effective estimates for the typical problems arising from other combined cointegration tests. Moreover, Bayer and Hanck (2013) combined the statistical significance levels of individual combined cointegration tests to enhance test reliability and accuracy by using Fisher’s formula when building a combined cointegration test model (Olanipekun and Usman 2019). Individual combined cointegration tests are written in the following form.

$$EG-JOH=-2\left[\mathrm{ln}\left(pEG\right)+\mathrm{ln}(pJOH)\right]$$
(2)
$$EG-JOH-BO-BDM=-2\left[\mathrm{ln}\left(pEG\right)+\mathrm{ln}\left(pJOH\right)+\mathrm{ln}\left(pBO\right)+\mathrm{ln}(pBDM)\right]$$
(3)

where pEG, pJOH, pBO, and pBDM are the p-values of the combined cointegration tests of Engle and Granger (1987), Johansen (1991), Boswijk (1994), and Banerjee et al. (1998), respectively. Bayer and Hanck (2013) believed that when the calculated Fisher statistic is greater than the B–H critical value, the null hypothesis is rejected, that is, there is no cointegration relationship.

Long-term relationship: combined cointegration vector FMOLS, DOLS, and CCR estimation

Aiming at testing whether there is a long-term combined cointegration relationship among variables, it is necessary to estimate the combined cointegration equation. The most commonly used combined cointegration estimation method is OLS, but if the explanatory variables are endogenous or the regression error term is serial correlation, the parameters estimated by OLS are biased (second-order bias consisting of endogenous bias and non-centrality bias). To solve the problem caused by the OLS parameter estimation, Phillips and Hansen (1990) modified the OLS estimator with a nonparametric method, which is called FMOLS (fully modified OLS, fully modified least squares). Phillips (1995) and Kitamura and Phillips (1997) further extended FMOLS, and Park (1992) proposed canonical cointegrating regression (CCR), which is like FMOLS in that it uses a nonparametric method to revise the OLS estimator and uses a different method to eliminate non-centrality bias. Phillips and Loretan (1991), Stock and Watson (1993) modified the parameters of the OLS estimator by using the lead and lag periods of the first-order differences of the I(1) variables as explanatory variables. On this basis, Saikkonen (1991) proved that FMOLS, DOLS, and CCR estimators are asymptotically efficient estimators.

ARDL bounds test

Besides the Bayer and Hanck (2013) combined cointegration test, the ARDL bounds test developed by Pesaran et al. (2001) can be adopted to detect whether there is a long-term combined cointegration relationship among variables (Rjoub et al., 2021), which is similar to the traditional combined cointegration test. Compared with the traditional combined cointegration test, it has the following advantages. Firstly, when variables are integrated with different orders, it can be used. Secondly, it is more robust for small samples. Finally, it can be used for impartial assessment of long- and short-term frameworks. In addition, as stated by Hao (2021), if the F-statistics is greater than the upper limit of the critical value of the asymptotic distribution, the combined cointegration relationship or the long-term relationship exists; otherwise, there is no long-term combined cointegration relationship. The general form of the ARDL bounds test model is as follows:

$$\Delta {CE}_{t}={\alpha }_{0}+\sum_{i=1}^{p}{\beta }_{1i}{\Delta CE}_{t-i}+\sum_{i=0}^{q}{\beta }_{2i}\Delta {RE}_{t-i}+\sum_{i=0}^{q}{\beta }_{3i}\Delta {GDP}_{t-i}+\sum_{i=0}^{q}{\beta }_{4i}\Delta {GIN}_{t-i}+\sum_{i=0}^{q}{\beta }_{5i}\Delta {X}_{t-i}{+\gamma }_{1}{CE}_{t-1}+{\gamma }_{2}{RE}_{t-1}+{\gamma }_{3}{GDP}_{t-1}+{\gamma }_{4}{GIN}_{t-1}+{\gamma }_{5}{X}_{t-1}+{\varepsilon }_{t}$$
(4)

where X represents macroeconomic variables (for example, foreign direct investment (FDI), trade openness (TRO), financial development (FIN), inflation rate (INF), and government expenditure (GOE));\({\beta }_{i}\) represents the short-term coefficient; \({\gamma }_{i}\) represents the long-term coefficient; p and q represent the lag term; and \({\varepsilon }_{t}\) represents a white noise sequence that obeys a normal distribution. If there is a long-term relationship between renewable energy consumption, green innovation, and economic growth with CO2 emissions in E7 countries under macroeconomic conditions, the corresponding conditional error correction model is as follows:

$$\Delta {CE}_{t}={\alpha }_{0}+\sum_{i=1}^{p}{\beta }_{1i}{\Delta CE}_{t-i}+\sum_{i=0}^{q}{\beta }_{2i}\Delta {RE}_{t-i}+\sum_{i=0}^{q}{\beta }_{3i}\Delta {GDP}_{t-i}+\sum_{i=0}^{q}{\beta }_{4i}\Delta {GIN}_{t-i}+\sum_{i=0}^{q}{\beta }_{5i}\Delta {X}_{t-i}+\delta {ECM}_{t-1}+{\varepsilon }_{t}$$
(5)

where \(\delta\) represents the speed of model adjustment or the error mechanism of returning to long-term equilibrium; ECM represents the error correction term, the coefficients (\({\beta }_{i}\)) of the lag difference term reflects the short-term dynamic coefficient of the model converging to equilibrium. Under the ECM technology, if the value of the ECM coefficient \(\delta\) of the error correction term is negative and significant, it indicates that there is a long-term causality. On the other hand, if the coefficients of each variable in the ARDL-VECM model are significant, it indicates that there is a short-term causality.

Data

This study examines the impact of renewable energy consumption, green innovation, and economic growth on environmental sustainability in seven emerging market economics (Brazil, China, India, Indonesia, Mexico, Russia, and Turkey) using annual panel data for the period 1990–2020. The choice of time span is based on the basic principles of availability and comprehensiveness of data from different sources. These data include the ratio of total renewable energy consumption (RE) in million tons of oil equivalent (Mtoe) to total primary energy consumption using the substitution method, carbon dioxide emissions per capita (tons) (CE), GDP per capita (GDP) in constant dollars in 2010, green innovation (GIN), and other macroeconomic variables that affect environmental sustainability. Among them, data come from Global Carbon Project, BP Statistical Review of World Energy, World Bank, and OECD green growth indicators (2022) (see Table 1). Moreover, we logarithmically transform all variables so that the interconversion can help to obtain a better normal distribution of all data and improve the heteroskedasticity criticality problem, making the final result meaningful and easy to interpret. Table 2 shows that the interquartile range (IQR) of all variables has no outliers.

Table 1 Data source and description
Full size table
Table 2 Descriptive statistics
Full size table

Empirical results and discussions

Stationarity test of panel data

The four methods mentioned above perform unit root tests on renewable energy consumption, green innovation, economic growth, CO2 emissions, and other macro variables, for example, the logarithm of foreign direct investment (FDI), trade openness (TRO), financial innovation (FIN), inflation rate (INF), and government final consumption expenditure (GOE), and their original series or first-order difference series, that affect environmental sustainability in E7 countries. The test regression formula contains constant terms, and the regression formula includes both constant terms and trend terms. The test results are shown in Table 3. The results of the unit root tests for CE, RE, GDP, FIN, and GOE using ADF, DF-GLS, PP, and KPSS tests, respectively, at the 10% significance level indicate that the first-order differences of the series are smooth. However, the unit root test results for GIN, FDI, TRO, and INF show that the series is stationary at the primitive order level. Therefore, CE, RE, GDP, GIN, FDI, TRO, FIN, INF, and GOE are all single integral I(1) or I(0) processes over the period 1990–2020.

Table 3 Test result of unit root tests
Full size table

Long-term cointegration relationship

To further verify the long-term cointegration characteristics among the used variables, we used the Bayer–Hanck combined cointegration test. Table 4 presents the results of the Bayer and Hanck (2013) combined cointegration test; the Fisher statistics of EG-JOH and EG-JOH-BAN-BOS are much larger than the 5% critical values of 3.814 and 9.165, respectively. Therefore, there is a long-term combined cointegration relationship among CE, RE, GDP, GIN, FDI, TRO, FIN, INF, and GOE at the 5% significance level. Table 5 shows the ARDL Wald test results which also further confirms the long-term combined cointegration relationship among these variables, since the F-statistic is 8.414, which is greater than 3.77 at 1%.

Table 4 Bayer–Hanck cointegration test
Full size table
Table 5 ARDL Wald test results (F-value) for long-term cointegration
Full size table

The long-term effects of renewable energy, green innovation, and economic growth on CO2 emissions in the E7 countries are further examined in this study after determining the existence of cointegration or long-run relationships among the variables under macroeconomic (e.g., trade openness, FDI, financial innovation, inflation, and government spending) conditions. Firstly, as clearly pointed out in the overview section of this study, GIN is an important factor in reducing CO2 emissions (Mensah et al. 2019; Khattak et al. 2020; Li et al. 2022). However, surprisingly, within the framework of macroeconomic variables (FDI, TRO, FIN, INF, and GOE), there is no significant negative effect of GIN on CE at the 10%, 5%, and 1% significance levels.

Secondly, RE has a negative impact on CE, as expected. The long-term estimates of FMOLS, DOLS, and CCR show that for every 1% increase in RE, CE will decrease by 0.357%, 0.428%, and 0.348%, respectively, which indicates that RE has a favorable impact on the improvement of environmental quality in E7 countries. The reason for this may be that in the context of carbon neutrality, E7 countries are trying to promote high-quality green transformation of energy-intensive industries, build a green, low-carbon, circular development economic system, and vigorously develop safe, clean energy, and renewable energy to meet the requirements of the present and the future. Meanwhile, replacing non-renewable energy sources, such as fossil energy, with renewable energy, which is considered a green and clean form of energy, in real life and production processes can reduce the environmental consequences of energy use, such as CO2 emissions. This finding is supported by the studies of Omri and Nguyen (2014), Kirikkaleli and Adebayo (2021a, b), Djellouli et al. (2022), and Hao (2022b). However, this result is contrary to the findings of Musah et al. (2020) for West Africa, Padhan et al. (2020) for OECD countries and Çıtak et al. (2021) for USA, who concluded that RE has a significant positive effect on CO2 emissions, which is detrimental to these countries’ environmental quality improvement. This is mainly because these countries have not yet reached the threshold point for reducing CO2 emissions by renewable energy usage, which is consistent with the findings of Amer’s (2020) research of global economies, which concluded that environmental improvement is facilitated when renewable energy consumption accounts for about 8.39% of the country’s total energy consumption. Furthermore, as we expected, GDP has a positive impact on CO2 emissions in developing countries, especially in E7 countries. When other macroeconomic conditions (influencing factors) are constant (e.g., FDI, TRO, FIN, INF, and GOE), the results according to FMOLS, DOLS, and CCR show that a 1% increase in GDP per capita will increase CO2 emissions by 0.881%, 0.946%, and 0.875%, respectively, which indicates that the current economic growth has not reached the inflection point (inverted U-shaped relationship) for energy saving and emission reduction, i.e., the EKC assumption is valid for E7 countries (Doğan et al., 2022). Meanwhile, this also means that the rapid economic development of the E7 countries in the context of economic globalization has led to an increase in energy demand and thus to environmental degradation. The findings of Xiong and Xu (2021), Kirikkaleli and Adebayo (2021a), and Espoir et al. (2022), who found a long-term positive association between GDP and CO2 emissions, are supported by this one.

Finally, in terms of macro variables, FIN has a significant negative impact on CE. According to FMOLS, DOLS, and CCR, a 1% increase in FIN would result in a reduction in CO2 emissions of 0.301%, 0.336%, and 0.306%, respectively. This is similar to the findings of Kirikkaleli and Adebayo (2021a), Hung et al. (2022). and Emenekwe et al. (2022), who found that FIN improves environmental quality. However, this result is contrary to the case of Rjoub et al. (2021) for Turkey, Yao and Zhang (2021) and Khan et al. (2022) for the global case, and Ling et al. (2022) for China, who argue that the development of financial innovation has an adverse impact on the improvement of the environment, which further indicates that the development of financial innovation in these countries is mainly a model of economic development at the expense of the environment. Therefore, to achieve carbon emission reduction targets, financial innovation should be considered as a tool for a country’s financial development or economic development, which can be implemented to keep the environment clean by implementing financial regulations to promote sustainable financial and environmental development. Table 6 also proves that CO2 emissions will increase by 0.015%, 0.043%, and 0.022% for every 1% increase in INF according to FMOLS, DOLS, and CCR estimates, respectively. It indicates that INF has adverse effects on the environment; the main reason for this may be that the uncertainty of INF under macroeconomic conditions has greatly weakened the regulations of E7 countries related to activities that emit more carbon pollution, and indirectly led to the reduction of environmental-related economic investment, such as clean energy, renewable energy investment, and pollution control investment (Musarat et al. 2021; Pardo 2021). However, this finding is contrary to the studies of Ullah et al. (2020) for Pakistan and Ahmad et al. (2021) for Asian economies, who concluded that INF to a certain extent strongly encourages businesses and governments to invest in environmentally friendly type technology, so INF contributes to improving the environment (reduction of CO2 emissions). Furthermore, it is also unexpected to find that GOE has a positive effect on CE only under the DOLS estimates for GOE, but the significant effect at 10% is relatively weak and negligible. In the context of economic globalization, the governments of developing countries are facing huge fiscal revenue pressure, especially E7 countries, which leads local government officials to pay too much attention to economic growth indicators and weakens the country or regional environmental quality standards. It also indirectly condones the emission of CO2 by enterprises. This finding is consistent with the findings of He et al. (2017), Yang et al. (2018), and Wang et al. (2021), but contradicts the findings of Wang and Li (2019) and Lingyan et al. (2021), who argue that government expenditure is conducive to reducing CO2 emissions.

Table 6 Long-term cointegration estimators for FMOLS, DOLS, and CCR
Full size table

ARDL bounds test

As demonstrated by the ARDL bounds test and Bayer–Hanck combined cointegration, there are long-term cointegration correlations between CE, RE, GDP, GIN, FDI, TRO, FIN, INF, and GOE. Therefore, we need to use the ARDL (1, 1, 1, 1, 0, 1, 1, 1) model to estimate the long-term and short-term coefficients of each variable proxy, and further analysis of the impact of RE, GDP, and GIN on CE under macroeconomic conditions (see Table 7). Firstly, the long-term impact paths of RE, GDP, and GIN on CE are basically consistent with the previous estimates of FMOLS, DOLS, and CCR. As previously expected, per capita GDP and inflation are the main factors affecting the increase in CO2 emissions in E7 countries (Musarat et al., 2021; Xiong and Xu 2021), while renewable energy consumption is beneficial to the environment in these countries’ quality improvement (Anwar et al. 2022a; Hao 2022a; Yunzhao 2022). However, unlike the previous long-term estimates of FMOLS, DOLS, and CCR, there is no significant negative effect of FIN on CE. While it is supported by the studies of Hung et al. (2022) and Jiang et al. (2022), they concluded that innovation has strong uncertainties in reducing CO2 emissions. Secondly, the short-term impact paths of RE, GDP, GIN, FDI, TRO, and INF on CE are almost consistent with the long-term estimates (symbols) of FMOLS, DOLS, CCR, and ARDL. However, in stark contrast to long-term estimates, GIN has a negative effect on CE at the 5% significance level, suggesting that GIN is beneficial for improving environmental quality. GIN aims to reduce the adverse impact on the environment, that is, by introducing new ideas, behaviors, products, and processes to reduce the environmental burden of enterprises or to achieve specific ecological sustainability goals (Brunel 2019; Hsu et al. 2021; Li et al. 2022; Liu et al. 2022a, b). Finally, FIN positively affects CE at the 1% level of significance, which indicates that FIN in these countries exacerbates environmental pollution and is not conducive to the improvement of environmental quality (Shen et al. 2021). The main reason is that under the background of economic globalization, developing countries pay too much attention to economic development and financial innovation and development, while neglecting environmental protection, especially E7 countries. The development of the financial industry in E7 countries indirectly optimizes resource allocation and reduces financing costs while promoting economic development, which expands the scale of production to a certain extent and leads to an increase in energy consumption and pollutant emissions (Xia et al. 2022). Finally, it is also found that GOE has a significant negative effect on CE, but the effect is relatively weak and negligible at the 10% significance level, which suggests that GOE in the short term can help reduce CO2 emissions in these countries. Moreover, this result is supported by the research of Wang and Li (2019) and Lingyan et al. (2021), but it is diametrically opposite to the previous long-term estimates of FMOLS, DOLS and CCR.

Table 7 ARDL estimation results for long and short terms
Full size table

In addition, the short-term estimates of the ARDL in Table 7 also show that the error correction term (ECM(-1)) for the specified (CE) model is negative and statistically significant at the 1% significance level, indicating a good rate of adjustment in the post-shock relationship process, as any periodic shocks in the model adjust to long-term equilibrium at a rate of 45%. Ahmed et al. (2013) confirmed this result by arguing that a highly significant error correction term provides further evidence for the existence of a stable long-term relationship. Moreover, once a shock has occurred, this also means that the long term would be shortly corrected back by 2.2 years for the CE (CO2 emissions) models.

To avoid model unreliability due to parameter instability, this thesis uses the cumulative sum of recursive residuals (CUSUM) and the cumulative sum of squares of recursive residuals (CUSUMSQ) to test the stability of the long- and short-term parameters estimated by the model. The test results are shown in Fig. 5. According to Fig. 5, the fluctuations of both CUSUM and CUSUMSQ do not exceed the boundary region, so all the fitted models are stable and accurate because the black line is within the red bandwidth, which is favorable for the policy impact. According to Fig. 3, the fluctuations of both CUSUM and CUSUMSQ do not exceed the boundary region, so all the fitted models are stable and accurate because the black line is within the red bandwidth, which is favorable for the policy impact.

Fig. 5
figure 5

Plots of recursive cumulative sum of residuals (CUSUM) and recursive cumulative sum of squared residuals (CUSUMQ)

Full size image

Conclusions and policy implications

This study explores the long-term and short-term impacts of renewable energy consumption, green innovation, and economic growth on environmental sustainability in E7 countries over the period 1990–2020 period within a macroeconomic framework. This study attempts to fill this gap in the environmental literature using the FMOLS, DOLS, and CCR, Bayer–Hanck cointegration, and ARDL bounds test. Firstly, the results of the Bayer–Hanck cointegration test and ARDL bounds test showed a long-term association between the environment and its possible causes (e.g., renewable energy, economic growth, green innovation, FDI, trade openness, financial innovation, inflation, and government spending).

Secondly, long-term estimates of FMOLS, DOLS, and CCR indicate that financial innovations in renewable energy consumption have a negative impact on CO2 emissions. As expected with the development of financial innovation, the use of renewable energy has improved the environmental quality of the E7 countries. Furthermore, economic growth in the E7 countries exacerbates the deterioration of environmental quality, which may be the result of these countries’ attempts to expand their economies without considering the environmental sustainability impacts of their actions. In the context of economic globalization and consumption reduction and emission reduction, E7 countries are striving to promote high-quality green transformation of energy-intensive industries, build a green, low-carbon, and circular development economic system, and vigorously develop safe, clean energy, and renewable energy, so as to meet the requirements of current and future economic and environmental sustainability. Furthermore, inflation has a negative impact on the environment (CO2 emissions). Uncertainty about inflation in macroeconomic conditions has weakened environmental regulations in these countries related to activities that emit more carbon pollution. Moreover, as previously analyzed, the increase in government expenditure in the E7 countries exacerbates environmental pollution and is not conducive to environmental improvement. Under the background of economic globalization for a long time, the governments of these countries are likely to face huge fiscal revenue pressures, leading local government officials to pay too much attention to economic growth indicators, thus weakening the environmental quality standards of the country or region. This also indirectly indulges companies to emit CO2.

Finally, the ARDL long-term estimates are basically consistent with the FMOLS, DOLS, and CCR long-term estimates, further indicating that economic growth and inflation are the main reasons for the increase in CO2 emissions in E7 countries. Economic growth increases the burning of carbon emissions, and the use of renewable energy has improved the environmental quality of these countries and regions. Moreover, the short-term impact paths of renewable energy consumption, economic growth, green innovation, FDI, trade openness, and inflation on CO2 emissions are almost consistent with the long-term estimates of FMOLS, DOLS, CCR, and ARDL. However, the impact of green innovation on CO2 emissions is diametrically opposed to long-term estimates. It is shown that green innovation development leads to environmental sustainability in E7 countries, that is, by introducing new ideas, behaviors, products, and processes to reduce the burden of enterprises on the environment or to achieve specific ecological sustainable development goals. Besides, ARDL’s short-term estimation results confirm that financial innovation and government expenditure has exacerbated environmental pollution in E7 countries, which is not conducive to sustainable economic and environmental development. Under the background of economic globalization, government officials are overly concerned with economic development and financial innovation development at the expense of environmental protection in developing countries (e.g., E7 countries). These results therefore make the following recommendations for policy makers in countries and regions around the world, especially developing countries.

  1. (1)

    Governments and enterprises around the world should actively promote and implement green innovation strategies. On the one hand, it is necessary to further improve the industry green standard system, continue to carry out the identification of green processes, green factories, green products, green parks, and green supply chains, and build a green manufacturing system for the entire life cycle of the industry. On the other hand, it is necessary to strictly implement the intellectual property protection system, focus on the major needs of energy conservation and carbon emission reduction and environmental pollution control and governance, and provide targeted financial and tax assistance to promote green innovation.

  2. (2)

    Governments should establish a list of smart carbon reduction technology proposals through top-level planning, and at the same time establish technical value evaluation criteria. It is also necessary to help enterprises make technology choices and strategic decisions based on the characteristics and applicable scenarios of specific technologies, as well as the emission reduction potential and potential profit margins of smart carbon reduction technologies.

  3. (3)

    Government agencies, energy enterprises, and industry associations cooperate to build industry-level carbon emission monitoring platforms. The platform can monitor the carbon emissions of high-energy-consuming enterprises in real time, and conduct data analysis on the application effect of smart carbon reduction technology, providing data support for the popularization of technology and the formulation of relevant national policies.

  4. (4)

    A green supply chain with key enterprises as the core should be actively built. Key enterprises should be encouraged and supported to implement green procurement in the whole life cycle of product production and services, implement ecological design, develop green products, promote green production, guide green consumption, promote the green manufacturing industry’s leading enterprises to actively build a green supply chain, take the lead, and drive the industry chain upstream and downstream enterprises to carry out energy-saving and environmental protection transformation as well as greening into the industrial development process, to achieve green and low-carbon development of the entire industrial system.

  5. (5)

    Enterprises should be encouraged to develop long-term low-carbon development strategies and carbon neutral roadmaps. According to the carbon peak and carbon neutral targets of each country, we will develop a carbon peak roadmap and carbon neutral action plan for enterprises and set reasonable medium- and long-term and phased targets. Through planning design and roadmap advancement, we ensure that enterprises complete their carbon reduction tasks within the established timeline, thereby supporting the achievement of industry and national carbon neutrality targets. Governments should also strengthen support and guidance, promote within the key enterprise industry, and strengthen the depth of cooperation between enterprises to give full play to the leading role of key enterprises.

In addition, there are certain limitations to this study which give directions for future research. Firstly, based on the availability of data, the institutional indicator data used in this study is only for 7 emerging market economics, not for countries or specific regions in the world, so it may produce errors of regional heterogeneity. Secondly, in examining the factors affecting environmental sustainability, this study only adopts five macroeconomic indicators and ignores other different determinants that affect environmental sustainability, such as resources, population, industrialization, urbanization, globalization, institutions, resource rent, and tax on environment or governance. Finally, this study used CO2 emissions as a predictor of environmental degradation, so future investigations should use other proxies of environmental degradation.