Front. Environ. Sci. Frontiers in Environmental Science Front. Environ. Sci. 2296-665X Frontiers Media S.A. 1793775 10.3389/fenvs.2026.1793775 Original Research Exploring education’s moderating role IN CO2 emission and economic dynamics in Asia nations Le et al. 10.3389/fenvs.2026.1793775 Le Hang My Hanh 1 * Supervision Methodology Writing – original draft Conceptualization Ho Minh Trung 1 Data curation Formal Analysis Writing - original draft Tran Ngoc Bao Tram 1 Writing – original draft Investigation Formal Analysis Turała Maciej 2 Supervision Writing - review and editing Validation Ho Chi Minh City Campus, Foreign Trade University, Ho Chi Minh, Vietnam Faculty of Management, University of Lodz, Lodz, Poland Correspondence: Hang My Hanh Le, lehangmyhanh.cs2@ftu.edu.vn 04 03 2026 2026 14 1793775 22 01 2026 17 02 2026 17 02 2026 Copyright © 2026 Le, Ho, Tran and Turała. 2026 Le, Ho, Tran and Turała https://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

This study explores the interconnections between carbon dioxide emissions, economic growth, energy consumption, trade openness, and education across 36 Asian countries, focusing on education’s direct and moderating effects. Using the Feasible Generalized Least Squares, Canonical Cointegration Regression, Fully-modified Ordinary Least Square, and two-step Generalized Method of Moments, the study confirms the Environmental Kuznets Curve hypothesis applied for Asian panel data, reflecting the impact of economic growth on CO2 emissions. Additionally, we find that while education directly reduces environmental pollution, it also amplifies the negative effects of energy consumption on the environment in the short run and long run. The impact of trade openness on CO2 emissions varies both in the short run as well as the long run, with the presence of CO2 endogeneity and the moderating role of the high level of education. The study highlights the need for tailored policy approaches and further research, thus calling for context-specific strategies and further research on education’s role in environmental outcomes.

 Asia CO2 emission economic growth education environmental degradation The author(s) declared that financial support was not received for this work and/or its publication. section-at-acceptance Environmental Economics and Management Introduction

There is a broad consensus that the primary cause of climate crisis has risen around the release of greenhouse gases–carbon dioxide (CO2) emissions due to human activities. Supporting this, Aydin (2023) identifies that economic growth, intensive energy use, and foreign direct investment are the driving forces behind rising CO2 emissions in G8 countries. This atmospheric shift is compounded by the findings of Filonchyk et al. (2024), who demonstrate how the combined influence of CO2, CH4 and N20 accelerates global climate change through their collective radiative forcing. Furthermore, Aydin (2024) connects these emissions to specific regional environmental disasters, providing evidence that increased CO2 levels are a significant factor in the rising frequency and severity of droughts in the United States. Consequently, the majority of environmental and energy consumption reports focus on measuring the quantity of carbon dioxide released into the environment. IEA (2024), IEA (2025) analyzed worldwide CO2 emissions in 2023 and 2024, then found that renewable energy growth reduced the surge, with 2023 emissions rising by 1.1% and 2024 emissions rising by 0.8%. According to the report, carbon dioxide emissions set a new record in 2023, but sustainable energy efforts have slowed their rise.

To understand the causality of CO2 emissions, researchers globally have analyzed the interrelationship between CO2 emissions, economic growth, and energy usage. In recent decades, some scholars have expanded this analysis to include the role of education and human development indices (Salequzzaman and Davis, 2003) identified key challenges to achieving ecologically sustainable progress in densely populated regions with high population growth and limited natural resources. They emphasized the importance of having a comprehensive environmental education initiative and the cultivation of local expertise to foster substantial shifts in environmental behavior. Furthermore, leveraging the formal education system provides a chance to engage a significant section of the current population while also instilling in future generations an understanding of the need for environmental preservation. Zsóka et al. (2013) supported this view and implied that education increases environmental consciousness, cultivates a profound sense of accountability, guides individuals away from environmentally detrimental behaviors, and the components of environmentally friendly actions among students, including their knowledge, attitudes, and self-reported behaviors.

Given the proven benefit of education to environmental preservation, the question arises “Does education have the potential to reduce emissions?” Some studies affirmed the beneficial effect of education on CO2 emissions. By showing a significant and negative relationship between the share of public expenditures on education in GDP and CO2 emissions, Mongo et al. (2021) education a powerful tool for environmental policy. Similarly, Alkhateeb et al. (2020) argue that education is a powerful tool in Saudi Arabia, enhancing secondary education can improve the environment by reducing CO2 emissions since people with high awareness tend to use cleaner energy sources. However, the relationship between education and CO2 emissions is not universally positive. Osuntuyi and Lean (2022) conclude that the direct effects of education worsen environmental degradation across income groups by promoting energy-intensive behaviors and sustaining unsustainable lifestyles. Similarly, Ahmed et al. (2021) found that human capital contributes to environmental harm, arguing that education alone cannot mitigate environmental degradation without incorporating an environmentally focused curriculum. To enhance understanding of education’s role in reducing CO2 emissions, this research investigates the effects of various factors on CO2 emissions across Asian nations panel, with a focus on both the direct and moderating impacts of education. This approach aims to provide policymakers with quantitative insights into the factors influencing CO2 emissions, forming a basis for creating informed and appropriate macroeconomic policies. In addition to the introduction, the article’s structure includes the following main sections: Section 2 presents the literature review. Section 3 introduces the methodology. Section 4 presents the estimation results and discussion. Section 5 provides conclusions and policy implications.

 Literature review

Over the past few years, concerns about environmental sustainability have grown significantly, gaining attention from scholars and policymakers worldwide. This section highlights particular direct relationships between economic growth, energy consumption, trade openness, education, and the environment. The indirect role of education on the environment is also discussed, specifically its mediating effect on energy usage and trade openness.

Environmental Kuznets Curve (EKC) hypothesis

Since the mid-1950s, when Kuznets introduced the EKC hypothesis, it has been widely acknowledged as a theoretical framework for analyzing the intricate relationship between economic development and environmental factors. The EKC hypothesis suggests an inverted U-shaped relationship, beginning with low-income levels and minimal environmental degradation. As economies grow, there is an increased exploitation of natural resources, along with greater waste production and environmental toxicity, particularly during phases of rapid agricultural and industrial expansion.

The applicability of the EKC, although widely studied, is still a matter of debate. For some researchers, like Grossman and Krueger (1991), analyzing the effects of the North American Free Trade Agreement has lent support to the assumptions of the EKC. Other research, in contrast, for example, Seppälä et al. (2001), present an argument that the EKC does not apply to some countries and areas. More recent studies have added complexity to the picture. Wang et al. (2023) for example, suggest that income inequality could shift the EKC relationship and as such propose a N-shaped relationship. On the other hand, Friedl and Getzner (2003) have also reported N-shaped relationships between the growth of the economy and the emissions of CO2. Other studies done by Farhani and Ozturk (2015) as well as Odhiambo (2011) in particular contexts, showed a positive and monotonic relation between the emissions of CO2 and the economic growth in the said circumstances.

The ongoing debate in the context of the EKC therefore brings out the twin nature of economic growth and environmental deterioration to the forefront. Panayotou (1993) came up with the benchmark, suggesting that in the United States, individual emissions began to decline when income levels ranged between $3,800 and $5,500 (in 1993 real dollars) per capita over the course of a year. Among the developed countries, there are also clear indications that the EKC exists in rich countries. In his complete research extending from 1980 to 2010, Ben Jebli et al. (2016) stated that the presumption of the EKC is applicable to OECD countries. In the study by Salari et al. (2021), the findings show an inverted-U shape relationship between CO2 emissions and GDP which provides enough evidence to validate the EKC hypothesis across US states. These results provide more support for the EKC theory. However, the situation is distinct for emerging nations. Study of Feriansyah et al. (2022) conducted on eight ASEAN countries spanning from 1994 to 2018, provides further evidence of a direct correlation between GDP and carbon emissions, suggesting a significant elasticity coefficient. In summary, we recognize that EKC hypothesis has a greater chance of validation in the general path of economic growth and affluent nations, but less chance in developing countries. In this study’s panel data, we collect data of 36 developed and developing countries in Asia, therefore we expect to validate the EKC existence in the general economic development path of these Asian countries.

 Energy consumption and CO2 emission linkages

The impact of energy consumption on the environment, particularly in relation to climate crisis, has been evident since the onset of the Industrial Revolution. Ozcan (2013) used energy usage as a control variable to explore the relationship between carbon emissions, energy consumption, and economic growth in Middle East countries and found that causality runs from energy consumption and growth to CO2 emissions. Saboori and Sulaiman (2013) also controlled energy consumption to examine the link between CO2 emissions, energy consumption, and economic growth in ASEAN countries and found a positive and statistically significant relationship between carbon emissions and energy consumption in both the short and long-run. To accurately assess the true influence of economic activity, we also incorporate the use of energy as a control element in our investigation, as it can help distinguish between the specific impact of policies on air quality and the overall rise in emissions brought on by increased energy use. Moreover, including this factor can mitigate its influence on both the independent and dependent variables, hence bolstering the validity of the study’s results. Sohag et al. (2021) conducted a study to isolate the influence of pilgrimage tourism on CO2 emissions by taking into account energy usage as a controlling factor. Acemoglu and Restrepo (2017) examined the influence of technical progress on GDP while taking into account energy use and other variables.

 Trade openness and CO2 emission linkages

Economists have also investigated trade openness as a control variable, with research producing diverse findings. For instance, Ozturk and Acaravci (2013) identified a negative correlation between trade openness and CO2 emissions in Turkey, suggesting that increased production and export-oriented industries may contribute to emissions reduction. In contrast, Koengkan et al. (2021) reported mixed results across several nations in Latin America and Caribe, indicating that the impact of trade openness on CO2 emissions depends on the specific characteristics and policies of individual countries. A study conducted by Wen et al. (2020) revealed that trade openness resulted in increased exports, consequently stimulating domestic production by expanding the scale of industries. Nevertheless, this growth also led to a surge in pollution levels in China. Sajeev and Kaur (2020) suggested that increased trade openness and foreign direct investment in emerging economies with insufficient environmental legislation are likely to result in elevated emissions. Due to the extensive utilization and significance of trade openness, the authors have chosen to incorporate this variable as a component in our models.

 Education and CO2 emission linkages

In addition to economic growth factors such as GDP per capita, trade openness, and energy consumption, the environmental impacts of education - both direct and indirect - have attracted increasing research interest in recent years. Eyuboglu and Uzar (2021) found that higher education levels help reduce CO2 emissions, contrasting with economic growth and energy consumption, which drive emissions in both short and long-term scenarios. Khan et al. (2023) demonstrated that the relationship between foreign direct investment (FDI) and CO2 emissions in developing economies is influenced by the level of education. As educational attainment rises from low to high, the impact of FDI on CO2 emissions transitions from positive to negative. Focusing on BRICS nations, Mahalik et al. (2021) identified primary education, non-renewable and total energy consumption, economic growth, and globalization as contributors to rising carbon emissions. Likewise, Osuntuyi & Lean (2023), examining primary education, concluded that education directly exacerbates environmental degradation across income groups. They found that education’s moderating role mitigates energy consumption’s environmental impact in high and upper-middle-income groups but intensifies it in lower-middle and low-income groups. Katircioglu et al. (2020) similarly concluded that in Africa, education exacerbates the adverse environmental effects of energy consumption, supporting the findings of Osuntuyi and Lean (2023). For this study, we use secondary education enrollment, following Mahalik et al. (2021), as an indicator of education level.

A review of the existing literature reveals that limited research has focused on the influence of education on the relationship between economic indicators and carbon emissions, particularly in Asian countries. Education is widely considered a catalyst for various economic dynamics, such as growth, technological innovation, structural transformation, foreign trade activities, and demographic changes, all of which can impact emission levels. Consequently, in the interplay between education, energy consumption, and trade openness, education exhibits two contrasting effects on energy consumption. On the one hand, Mahalik et al. (2021) suggest that secondary education enhances environmental awareness, encouraging individuals to adopt more sustainable behaviors. This shift in behavior can support environmental policies and improve energy efficiency. On the other hand, Osuntuyi and Lean (2023) argue that education may also lead to increased consumption of polluting technologies and environmentally harmful goods and services. This outcome is often linked to the higher income levels typically associated with more education, which can drive greater consumption and, consequently, higher levels of pollution. These contrasting effects illustrate both the direct and moderating roles of education, highlighting the potential transmission channels among the variables.

Regarding the impacts of education on trade openness, Bilal et al. (2021) discover the negative effect of technology innovation interacting with globalization on carbon emissions. This signifies that education can enhance eco-friendly and energy-efficient innovations to international investment and trade openness. On the contrary, economic integration and technological innovation foster investment in the transportation sector and in energy-intensive goods and services, which in turn trigger the level of emissions in these economies (Saint Akadiri et al., 2020).

 Research methodology Empirical model

This study utilizes the EKC hypothesis to investigate the effects of economic growth, energy consumption, and trade openness on environmental degradation, incorporating education as a moderating variable across 36 Asian countries. To bridge the gap in the existing literature and provide contemporary insights, the research explores the dual role of education - both direct and moderating - in shaping the relationships among economic growth, energy consumption, trade openness, and CO2 in Asian contexts. Furthermore, this study calculates the marginal effects of energy consumption and trade openness on environmental degradation using the analyzed panel data. It also investigates causal relationships by applying the Dumitrescu & Hurlin test, which accounts for cross-sectional dependence and heterogeneity–an aspect not considered by Katircioglu et al. (2020). Consequently, the research provides novel evidence of education’s direct and moderating impacts on environmental degradation from a unique perspective. The key findings offer fresh insights into the interplay between economic growth, energy consumption, trade openness, and environmental degradation under the influence of education, both as a direct factor and a moderator.

Given the previous studies of Osuntuyi and Lean (2023), Ehigiamusoe et al. (2020), Katircioglu et al. (2020) and Sahu et al. (2024), the following base model is specified for the direct impact of education:

Base model CO 2 it = α + β 1 GD P it + β 2 GDP 2 it + β 3 EN C it + β 4 OPE N it + β 5 ED U it + ε it where α is the constant term, β are the regression coefficients measuring impact of each independent variable on CO2 emissions, i represents the cross-sectional unit referring individual countries, t identifies the specific year of the observation, GDP is real Gross Domestic Product per capita (2015 constant prices) as a proxy for economic growth, GDP2 is GDP squared, ENC is energy consumption per capita (measured in million tonnes of oil equivalent), EDU is education measured as gross secondary school enrollment in percent, OPEN measures the total trade (exports + imports) as a percentage of GDP, and ε is the error term. CO2 is carbon emissions (in metric tons) as a proxy for environmental degradation. Natural logs (ln) are used for all variables.

Previous researches have used different proxies to measure education including the number of graduate and postgraduate students in the country (Balaguer and Cantavella, 2018), the percentage of the population that has attended secondary school (Zafar et al., 2020). This study used gross secondary school enrollment in percent as a proxy for education, consistent with Zafar et al. (2020). Since secondary education is not compulsory in many Asian countries, including Vietnam, Cambodia, Laos, and Indonesia, meaning enrollment levels reflect real differences in access, investment, and household choices. Even in countries where it is compulsory (e.g., China, Thailand, Japan, South Korea), enrollment never reaches 100% due to regional inequalities, resource constraints, and dropout patterns. Thus, it still captures variation in educational quantity and system performance.

To account for the moderating effect of education on environmental degradation, we extend Equation 1 by incorporating an interaction term between education and energy consumption (Osuntuyi and Lean, 2023; Katircioglu et al., 2020; Sahu et al., 2024), and interaction variable of education and trade openness as follows:

Hierarchical model (moderating effect) CO 2 it = α + β 1 GD P it + β 2 GDP 2 it + β 3 EN C it + β 4 OPE N it + β 5 ED U it + β 6 EN C it xED U it + ε it CO 2 it = α + β 1 GD P it + β 2 GDP 2 it + β 3 EN C it + β 4 OPE N it + β 5 ED U it + β 6 OPE N it xED U it + ε it where ENCxEDU is the interactive term of the logs between energy consumption and education, and OPENxEDU is the interactive term of the logs between trade openness and education.

In theory, to support the EKC hypothesis (inverted U-shaped curve), all models should have positive β1 and negative β2. Conversely, if β1 is negative and β2 is positive, the EKC hypothesis (U-shaped curve) is not supported. While education and its moderating variable are expected to alleviate environmental degradation, economic expansion and energy use are predicted to worsen it. Furthermore, to capture the impact of energy consumption on environmental degradation along with education’s moderating effect, we use the partial derivatives of Equation 2 and Equation 3 concerning energy consumption and trade openness. The marginal impacts model is specified in Equation 4 and Equation 5: ∂ C O 2 it / ∂ E N C 2 it = β 3 + β 5 ED U it ∂ C O 2 it / ∂ O P E N 2 it = β 4 + β 5 ED U it

Data

The authors have collected data on 36 countries in Asia over 30 years, from 1992 to 2021, from World Bank and U.S. Energy Information Administration. Based on the classification of income levels, these countries are grouped as follows.

+ High income: Bahrain, Brunei, Israel, Japan, Singapore, South Korea, United Arab Emirates.

+ Upper-middle Income: Armenia, Azerbaijan, China, Georgia, Indonesia, Iran, Iraq, Jordan, Kazakhstan, Lebanon, Malaysia, Maldives, Thailand, Turkey, Turkmenistan.

+ Low & Lower-middle Income: Bangladesh, Bhutan, Cambodia, India, Kyrgyzstan, Laos, Nepal, Pakistan, Philippines, Sri Lanka, Syria, Tajikistan, Vietnam, Yemen.

This study begins with a descriptive statistics and correlation analysis, followed by the application of the White (1980) test to identify potential heteroskedasticity issues in the panel data. Cross-sectional dependence is assessed using the Breusch-Pagan LM, Pesaran scaled and Pesaran (2004) CD tests. The results of the CD tests indicate the presence of cross-sectional dependency in the unbalanced panel data, prompting the use of secondary unit root tests. To determine co-integration among variables, the Pedroni & Westerlund co-integration tests are employed.

For short-run regression analysis, Feasible Generalized Least Squares (FGLS) is utilized after addressing significant panel data issues. Long-term associations between variables are examined using the Fully Modified Ordinary Least Squares (FMOLS) and Canonical Cointegration Regression (CCR). A robustness check is performed using the two-step Generalized Method of Moments (GMM) approach, which addresses causality and mitigates potential omitted variable bias contributing to endogeneity (Sahu et al., 2024).

The marginal effects of energy consumption and trade openeness on environmental degradation are calculated at secondary education’s minimum, mean and maximum levels based on FMOLS results. Lastly, the Dumitrescu & Hurlin panel causality test, which accounts for cross-sectional dependence and heterogeneity, is employed to analyze causal relationships among the variables.

 Empirical results Descriptive statistics and tests

Table 1 presents descriptive statistics and the correlation coefficient matrix for the variables used in the analysis. Based on a dataset of 1,080 observations, the mean CO2 emission level is approximately 0.864, with a standard deviation of 1.362. Among the independent variables, GDP exhibits a mean of 8.249, with minimum and maximum values of 5.869 and 11.115, respectively. The variable OPEN has a mean value of 4.367, a standard deviation of 0.567, and ranges from 2.519 to 6.08. Similarly, ENC has a mean of 9.459, with a minimum of 5.21 and a maximum of 12.203, reflecting significant variation likely due to differing levels of energy consumption across industrialized and less industrialized nations.

Summary statistics of data.

Statistic CO2 GDP OPEN ENC EDU
Mean 0.864 8.249 4.367 9.459 4.233
S.D 1.362 1.279 0.567 1.355 0.379
Minimum −2.82 5.869 2.519 5.21 2.878
Maximum 3.43 11.115 6.08 12.203 4.901
Observation 1,080 1,080 1,080 1,080 1,080
CO2 1.000 ​ ​ ​ ​
GDP 0.856*** 1.000 ​ ​ ​
OPEN 0.287*** 0.247*** 1.000 ​ ​
ENC 0.938*** 0.832*** 0.391*** 1.000 ​
EDU 0.599*** 0.586*** 0.225*** 0.635*** 1.000

Source: Author’s estimation.

For the education variable (EDU), the mean value is 4.233, with minimum and maximum values of 2.878 and 4.901, respectively. The correlation coefficients displayed in the lower panel of the table indicate that all variables are positively correlated with one another. Notably, GDP, ENC, and OPEN exhibit positive and statistically significant relationships with CO2 emissions. The correlation between ENC and CO2 shows the strongest association, which confirms that an increase in energy consumption will lead to a rise in CO2 emissions.

Heteroskedasticity problem

The results of White’s test presented in Table 2 indicate that the probability value of the chi-square statistic is less than 0.01. This suggests that the null hypothesis of constant variance can be rejected at the 1% level of significance, confirming the presence of heteroscedasticity in the residuals.

White (1980) homoskedasticity test.

Chi-square test 245.56***

***, **, and * stands for significance level at 1%, 5%, and 10% respectively. Source: Author’s estimation.

 Serial correlation problem

After addressing the heteroskedasticity issue, we proceed by testing for serial correlation in the idiosyncratic errors of the linear panel-data model, as outlined by Wooldridge Jeffrey (2002) for the base model. The results showed in Table 3 indicate strong evidence against the null hypothesis of no serial correlation, with the rejection occurring at the 1% significance level.

Wooldridge Jeffrey (2002) autocorrelation test.

F Test 75.037***

***, **, and * stands for significance level at 1%, 5%, and 10% respectively.

 Cross-sectional dependence tests

To determine which unit root test would be best, we must examine cross-sectional dependence in the next step. In order to achieve this, we use the Breusch-Pagan LM test, the Pesaran scaled LM for cross-section dependence in panel time-series data, and the Pesaran CD test. Table 4 reports the CD test results. The results reject the null hypothesis of cross-sectional independence.

Results of cross-sectional dependence tests results.

Breusch-pagan LM 3,074.347***
Pesaran scaled LM 68.101***
Pesaran CD 4.143***

***, **, and * stands for significance level at 1%, 5%, and 10% respectively.

Source: Author’s estimation.

 Unit root tests

Panel data with cross-sectional dependency problems must be dealt with Second Generation unit root tests. For this reason, we utilize the Perasan’CADF and the Karavias & Tzavalis unit root tests. As shown in Table 5, the result suggests that while the Karavias & Tzavalis tests indicate that EDU are non-stationary at zero-difference level, the CADF tests show that all of the variables are stationary at level. This result permits us to utilize the FMOLS and CCR because the methods were designed to estimate a co-integrating relationship with a mix of I (1) and I (0).

Results of Unit root tests.

Variable Pesaran’CADF test Karavias & Tzavalis test
​ At level 1st difference At level 1st difference
CO2 −2.894*** −10.065*** −11.659*** −44.982***
GDP −6.053*** −9.135*** −30.097*** −44.594***
ENC −2.148** −11.572*** −15.055*** −56.748***
OPEN −1.277* −13.380*** −11.970*** −51.687***
EDU −1.743** −7.930*** −3.402 −40.486***

***, **, and * stands for significance level at 1%, 5%, and 10% respectively.

 Panel cointegration

As a second step, we looked at the cointegration connection among the variables of interest after confirming that all variables were integrated into order one. To evaluate equilibrium relationships, two cointegration tests are used: the Westerlund and the Pedroni tests. The cointegration test results are shown in Table 6. At the 0.05 significance level, the three models indicate the presence of a long-term cointegration connection between CO2 emissions and its drivers in the Pedroni & Westurlund cointegration tests. Therefore, we use the FGLS tests to assess the short-term effects of trade openness, GDP per capita, education, and energy use on CO2 emissions in all models.

Results of Panel cointegration.

All countries Model 1 Model 2 Model 3
Pedroni modified Phillips-Perron t 3.340*** 4.231*** 4.099***
Pedroni Phillips-Perron t −2.905*** −3.755*** −2.393***
Pedroni Augmented Dickey-Fuller t −2.578*** −3.589*** −1.725**
Westurlund cointegration test −2.754*** −2.672*** −2.047**

***, **, and * stands for significance level at 1%, 5%, and 10% respectively. Source: Author’s estimation.

 Empirical result discussion Examining the EKC hypothesis using FGLS regression

In order to examine the EKC hypothesis, we use the FGLS to test short-run regression while accounting for the serial correlation and heteroskedasticity across panels. Besides estimating the models with interaction terms, we estimate the models without them. Table 7 demonstrates that for the three models, all tests are valid with prob > chi2 at 1%. Examining the overall findings, it is evident that, for all variables, there is an inverted U-shaped association between GDP and CO2 emissions in the three models, with a significance level of 1%, further validating the EKC hypothesis of the study.

Results of FGLS regression.

Variable Model 1 Model 2 Model 3
GDP 1.748*** 1.961*** 1.759***
GDP2 −0.087*** −0.100*** −0.088***
ENC 0.695*** 0.421*** 0.696***
OPEN −0.007 −0.004 −0.088
EDU −0.031 −0.628*** −0.112
ENCxEDU ​ 0.065*** ​
OPENxEDU ​ ​ 0.019
_Cons −13.792*** −12.208*** −13.524***
Prob > chi2 0.0000 0.0000 0.0000

***, **, and * stands for significance level at 1%, 5%, and 10% respectively.

Source: Author’s estimation.

More precisely, in the interaction-free model, trade openness and education do not correlate with environmental damage. However, the coefficients of energy usage in these countries are positive and statistically significant, indicating that increased energy consumption directly leads to higher CO2 emissions. This is observed in multiple nations, including China, India, USA and Africa, where energy consumption per capita significantly impacts CO2 emissions (Ahmed et al., 2023; Muhammad et al., 2025).

Meanwhile, the model incorporating interaction terms ENCxEDU indicates that the coefficients of all variables are significant, except that of trade openness, which shows no relation to environmental degradation. Education’s direct impact turns negative with the appearance of ENCxEDU, decelerating effects on CO2 emissions. The interactive component ENCxEDU, on the contrary, increases the emissions of carbon dioxide. This study suggests that while secondary education can directly reduce CO2 emissions, this education sector itself faces challenges in meeting carbon reduction targets due to significant emissions from energy usage. Higher education can play a crucial role in reducing CO2 emissions by fostering awareness and promoting sustainable practices among students and staff (Eyuboglu and Uzar, 2021). Meanwhile, in Africa Northern Cyprus, higher education levels have been linked to increased pollution due to the marginal effects of energy consumption, including transportation, electricity, and oil consumption (Katircioglu et al., 2020; Osuntuyi and Lean, 2023).

In addition, variables OPEN, EDU, and their interaction term (OPENxEDU) show no clear short-term impacts on the environment in this model (incorporating the interaction term). Nevertheless, energy usage still imposes positive impacts significantly on the dependent variable.

 Moderating effects of education

Table 8 shows the results of the base and hierarchical regression models that examine the moderating effect of education on the relationship between trade openness energy usage and CO2 emissions in the long run. Overall, the results of FMOLS and CCR are similar, and all models showsignificance in every variable, which is different from the short-run regression of FGLS.

Results of FMOLS and CCR regression.

Variable FMOLS CCR
Model 1 Model 2 Model 3 Model 1 Model 2 Model 3
GDP 2.694*** 2.614*** 1.523*** 2.268*** 2.615*** 1.523***
GDP2 −0.142*** −0.145*** −0.075** −0.148*** −0.145*** −0.075**
ENC 0.778*** −1.118*** 0.814*** 0.779*** −1.118*** 0.815***
OPEN −0.173*** −0.247*** −3.212*** −0.172*** −0.249*** −3.214***
EDU −0.237** −4.231*** −3.041*** −0.237** −4.232*** −3.041***
ENCxEDU ​ 0.471*** ​ ​ 0.471*** ​
OPENxEDU ​ ​ 0.688*** ​ ​ 0.688***
R2 0.913 0.751 0.615 0.911 0.749 0.624
Adjusted R2 0.912 0.750 0.613 0.910 0.748 0.622
S.e. 0.407 0.739 1.005 0.407 0.740 1.003
Long run S.e. 1.113 1.500 1.817 1.113 1.500 1.817

***, **, and * stands for significance level at 1%, 5%, and 10% respectively. Source: Author’s estimation.

In the base model, the analysis reveals the influence of various control variables, namely, GDP, OPEN, ENC, and EDU on CO2. The results indicate a notable EKC linkage between GDP and CO2, with positive GDP and negative GDP2 (p-value <0.01), suggesting that economic growth is associated with increased CO2 emissions to a certain point and then decreased CO2 emissions afterward. These findings corroborate prior research that established the EKC hypothesis (Mahmood et al., 2019). The EKC hypothesis confirms that Asia will overcome environmental problems through higher economic growth in the long run. In addition, ENC demonstrated a significantly positive impact on CO2, indicating that an increase in energy consumption may raise CO2 emissions. Conversely, EDU and OPEN both negatively impact CO2 in the long run–this reduction can be attributed to shifts in industrial trade activities and environmental awareness, enhancements in energy efficiency during production processes, and alterations in value chains. Our results are consistent with the result of Eyuboglu and Uzar (2021), who also show that an increase in higher education negatively affects CO2 emissions, while economic growth and energy consumption positively affect CO2 emissions both in the long run and short run. Similarly, regarding the mitigated impact of OPEN on CO2, Zhang et al. (2017) discovers that in ten newly industrialized countries, trade openness has been found to significantly reduce CO2 emissions, suggesting that trade can help restrain emissions while boosting economic growth.

Upon introducing the interaction term ENCxEDU in the second model, the model exhibited a decreased R-squared value of 75% compared with the initial value of 91% in Step 1. The inclusion of interactive terms had a statistically significant positive effect on CO2 (p-value <0.01), signifying that higher levels of education are associated with a corresponding increase in CO2. This suggests that education adversely aggravates energy consumption’s impact on environmental pollution in these Asian countries in general. These results differ from that of Tang et al. (2021), whose research indicates that educated individuals are more likely to steer away from environmentally destructive behavior toward more efficient energy resources. On the contrary, education itself, energy consumption, and trade openness all moderate environmental degradation, with negative coefficients significant at 0.01 level.

Model 3 shows the effect of the moderating variable OPENxEDU on CO2 emissions. EDU and OPEN have a significantly negative impact on CO2, implying that education and trade activities can improve the environment by decreasing CO2 emissions in Asian countries. Meanwhile, ENC and OPENxEDU have positive effects on CO2, signifying that higher levels of energy usage and higher levels of highly educated trade openness are associated with a corresponding increase in CO2. Highly educated people are more supportive of international trade, partly due to their realization that trade can benefit them economically (Stiller et al., 2022), and improvement in trade openness has a significantly positive effect on CO2 emissions (Chen et al., 2021). The Squared R of this model is also the lowest among the three models, about 61% only, highlighting the necessity of further research about this model.

 Robustness check

Some econometric techniques are employed to achieve the objectives of this study. To enhance the robustness of our findings, we use the two-step Generalized Method of Moments (GMM), specifically the system GMM approach, which ensures the reliability of our results. There are several reasons for adopting the two-step GMM methodology. First, the study spans 30 years, from 1992 to 2021, and includes data from 36 countries, indicating that the number of countries (N) exceeds the number of time periods (T) (Bakhsh et al., 2021). Additionally, the dynamic nature of panel data makes this approach particularly efficient, as emphasized by Baltagi (2021). The system two-step GMM method is also well-suited to address potential issues of endogeneity and heterogeneity.

Table 9 shows the results of GMM, which closely resemble the results of the FMOLS and CCR analysis, except for the results for Model 3 and Energy usage in Model 2. Two-step GMM shows that with the endogeneity of CO2 emission, in Model 3 education seems to have no direct impact on the environment, but rather through moderate effect with trade activities. Also in this model, the OPENxEDU is positive in previous analysis, but show negative impact in GMM with the presence of lagged CO2. As suggested by Tejedor et al. (2019), trade openness can decrease harmful impacts on environment through five active learning strategies (service learning, problem-based learning, project-oriented learning, simulation games, and case studies) practicing in college students. Moreover, the coefficient of energy usage in Model 2 is positive in GMM, while it is negative in FMOLS and CCR.

Two-step GMM.

Variable Model 1 Model 2 Model 3
L1. CO2 0.717*** 0.758*** 0.704***
GDP 0.704*** 0.674*** 0.728***
GDP2 −0.036*** −0.035*** −0.038***
ENC 0.209*** 0.116*** 0.223***
OPEN −0.035*** −0.038*** 0.098*
EDU −0.065*** −0.177*** 0.058
ENCxEDU ​ 0.014*** ​
OPENxEDU ​ ​ −0.032***
Constant −4.544*** −3.663*** −5.280***
AR(1) p-value 0.000 0.000 0.000
AR(2) p-value 0.622 0.671 0.592
Prob > F 0.000 0.000 0.000
Instruments 35 36 36
Countries 36 36 36

***, **, and * stands for significance level at 1%, 5%, and 10% respectively.

Source: Author’s estimation.

The results of AR(1) are significant across all models, indicating the presence of first-order autocorrelation, as expected in dynamic panel data models. Across all models, the results of AR(2) are insignificant, indicating the absence of second-order autocorrelation. Hence, based on these indicators, we can conclude that our results are robust and valid. However, with the inconsistency between Two-step GMM and FMOLS and CCR results, further investigations need to be carried out.

 Marginal effects results

The marginal effect, assessed at the minimum, medium, and highest education values, is displayed for the two model (4) and (5) in Table 10. The FMOLS results serve as the basis for the computation. The results show that for each of these models, the marginal effects are statistically significant at the Min and Mean value of education. We also draw a scatter plot graph as shown in Figures 1, 2 to capture the relationship between the interactive terms on CO2 emissions. The graph depicts a linear path between those factors.

Marginal effects through different level of education.

Variable Min Mean Max ​ Min Mean Max
∂ CO 2 it ∂ ENC it 0.236** 0.873*** 1.188*** ∂ CO 2 it ∂ O P E N i t −1.232*** −0.298*** 0.161

***, **, and * stands for significance level at 1%, 5%, and 10% respectively.

Interaction effect of Education and Energy Usage on CO2 emission.

Line graph showing average marginal effects of ENC on linear prediction with ninety-five percent confidence intervals, plotted against EDU values 2.878, 4.233, and 4.901, displaying a positive linear relationship.

Interaction effect of Education and Trade Openness on CO2 emission.

Line graph showing average marginal effects of OPEN on linear prediction with ninety-five percent confidence interval, plotted against EDU values 2.878, 4.233, and 4.901, displaying a positive linear relationship.

The results reveal that the marginal impacts of energy consumption are contingent upon education and increase environmental degradation at minimum, mean, and maximum levels. Figure 1 illustrates the upward interaction impact of education and energy usage on CO2 emissions, using secondary education’s minimum, mean and maximum levels. This result again validates the positive coefficient of ENCxEDU in both FMOLS, CCR, and two-step GMM, implying the magnitude of the unfavorable impacts increases with more enrollments. The findings are consistent with those of Osuntuyi and Lean (2023), who discovered that marginal effects grow with education levels. The results are different from Tang et al. (2021), who claim the beneficial impacts of education on reducing air pollutants. Our results suggest that as individuals get more educated, the negative effects of energy use on environmental deterioration decrease.

On the other hand, trade openness has marginal impacts that decrease environmental deterioration at the lowest and average levels, but the max level shows no significance, meaning it lacks the evidence to conclude that highly-educated nations decrease or increase carbon dioxide through international trade. However, we notice that the slope of OPEN’s marginal effect is upward, moving from a negative to a positive coefficient. Figure 2 illustrates the upward interaction impact of education and trade openness on CO2 emissions, using secondary education’s minimum, mean and maximum levels. This result partly explains the differences of OPENxEDU in FMOLS, CCR (showing positive OPENxEDU on CO2), and two-step GMM (showing negative OPENxEDU on CO2), as in the very long run when education level increases, we may witness detrimental effects occurring when education expansion accelerates trading activities and thus increases CO2 emissions. The two-step GMM, which produces only short-run results but considers the dependent variable’s endogenity, thus gives the result of negative interaction similar to the negative effect of education’s Min and Mean values. Our results further suggest that as more individuals get educated, they tend to engage in more trading resources and activities that produce harmful impacts on nature.

 Panel causality tests

In this section, we present a causality testing procedure developed by Dumitrescu and Hurlin (2012) (the DH test) to test for Granger causality in panel datasets. The causality test results are provided in Table 11, which is essential for identifying the causal relationships among the variables, particularly for informing policy decisions.

Results of Granger test.

GDP ≠> CO2 22.293*** ENC ≠> GDP 24.434***
CO2 ≠> GDP 16.859*** GDP ≠> ENC 18.676***
OPEN ≠> CO2 10.8948*** EDU ≠> GDP 13.207***
CO2 ≠> OPEN 4.840*** GDP ≠> EDU 14.105***
ENC ≠> CO2 11.255*** OPEN ≠> ENC 6.167***
CO2 ≠> ENC 3.781*** ENC ≠> OPEN 8.770***
EDU ≠> CO2 7.255*** EDU ≠> ENC 9.133***
CO2 ≠> EDU 7.982*** ENC ≠> EDU 11.390***
OPEN ≠> GDP 8.528*** EDU ≠> OPEN 6.554***
GDP ≠> OPEN 13.981*** OPEN ≠> EDU 5.885***

***, **, and * stands for significance level at 1%, 5%, and 10% respectively.

Source: Author’s estimation.

Table 11 shows significant Granger causality across most of the variables. Notably, the findings indicate bidirectional causality between economic growth and environmental pollution, as well as between economic growth and education enrollment. This result indicates the strong interconnections between these variables, and that each economic attribute in the study’s model can be used to predict the other one. Bidirectional causality between CO2 emissions and GDP per capita is consistent with Chaabouni and Saidi (2017), bidirectional causality between trade openness and CO2 emissions is similar to the result of Musah et al. (2021), two-way causality between energy consumption and CO2 emissions also appear in research on the United States and France (Ajmi et al., 2015). Regarding the education’s bidirectional relationship with energy consumption and CO2 emission, these results are inconsistent with Zafar et al. (2020), who found a unidirectional causality running from education to emissions and energy use.

 Conclusion

The study illustrates the impact of economic growth, energy usage, and trade openness on CO2 emissions, along with the moderating effect of education in 36 Asian countries over a 30-year period. The findings confirm the EKC hypothesis in the general path of economic development of these nations, which is consistent with ecological modernization theory. This theory posits that once countries reach a developed stage, their negative environmental impacts begin to decline. While trade openness does not have a significant short-term effect on CO2 emissions, it shows a long-term negative impact, which contradicts the pollution haven hypothesis and aligns more with the pollution halo effect. The pollution halo effect suggests that trade openness can lead to positive environmental outcomes, whereas the pollution haven hypothesis indicates adverse environmental consequences.

Furthermore, hierarchical analyses reveal that GDP has a positive and significant effect on CO2 emissions, but once a certain threshold is reached, it contributes to a reduction in emissions. This pattern indicates that economic growth in Asian countries eventually leads to environmental improvements. While energy consumption has a positive and significant effect on CO2 emissions, it is widely recognized that industrial growth often results in increased CO2 emissions. Therefore, it is recommended that Asian countries promote the adoption of clean energy to help mitigate global warming.

Additionally, the education variable (EDU) has a significant negative direct effect on CO2 emissions in the long-term, suggesting that greater investments in education lead to lower CO2 emissions. Higher education levels can enhance environmental quality by promoting behaviors such as reducing plastic use, increasing tree planting, and proper waste disposal, all of which can contribute to a decrease in carbon emissions per capita.

The results of the second hierarchical regression indicate that the interaction between education and energy consumption has a significantly positive impact on CO2 emissions, implying that education exacerbates the positive effect of energy consumption on CO2 emissions. Access to energy-intensive technologies and the use of non-renewable resources are expected to rise with education. This finding can be explained by the fact that there are not many countries with educational initiatives that emphasize environmental sustainability. Therefore, instruction lacking in energy-saving material and specific environmental awareness initiatives may harm the environment. This is particularly relevant in rapidly developing Asian nations where industrialization and urbanization are driven by an increasingly educated workforce.

Nevertheless, our results do not provide strong evidence of education’s moderating role in the association between trade openness and CO2 emissions. The results show a positive interactive term between EDU and OPEN, but a negative one in short run and marginal calculation. We hypothesize that in the initial stages, trade openness may reduce CO2 emissions by facilitating the transfer of environmentally friendly technologies, efficient production methods, and stricter environmental standards across borders. For example, developing countries in Asia such as Vietnam and Thailand may adopt cleaner technologies from advanced economies, thereby reducing their carbon footprint. However, as people become more educated, trade openness can lead to increased CO2 emissions. Educated populations often drive higher consumption of goods and services due to improved income levels, which amplifies production and energy demand. Additionally, education can boost innovation and industrial expansion, sometimes prioritizing economic growth over environmental considerations. This dual effect underscores the complexity of balancing trade policies with sustainable development strategies.

In conclusion, this study suggests that Asian countries should strengthen their educational policies and expand the focus on climate science within educational curricula to promote environmental conservation. As all countries face the challenges of global warming due to rising carbon dioxide emissions, it appears that even though education is found to have a direct negative impact on CO2 emissions, increased investment in education may amplify the emissions effect of energy use and thus lead to higher CO2 emissions overall, which may be attributed partly to insufficient environmental education in general curricula, and partly to the improvement of the expected living standard and the higher energy consumption that follows. Addressing this gap could help mitigate the environmental impact associated with energy consumption. To solve this situation, we suggest that the governments should push environmental preservation to be a responsibility rather than an optional action. We propose that those stakeholders should integrate environmental studies and nature appreciation as mandatory subjects into the existing curriculum such as biodiversity, ecology, sustainability, etc. These classes should be added to study performance assessment as an evaluation criterion for students to be qualified for the next grade. The initial moderating effect of education on trade openness to decrease CO2 emissions should be encouraged as trading activities increase along with the level of education. We suggest pushing community engagement in environmental protection can raise environmental awareness widely. People and firms with creative protective actions for nature need to be valued through daily sales and trades. This could involve regular features of outstanding individuals or businesses who have made positive contributions to environmental preservation and sustainability to encourage active awareness and participation among citizens. These findings underscore the importance of proactive measures to address the environmental challenges linked to energy consumption and trade openness, alongside efforts to boost productivity through the adoption of more environmentally sustainable energy sources.

Future research could explore the relationship between higher education and the environment using alternative environmental indicators, such as the ecological footprint. Additionally, future studies should examine the influence of other factors, including educational policies, the quality of education, and their moderating effects, to provide a more comprehensive understanding of this relationship. Moreover, investigating the linkages between education and environmental degradation through more specific variables, such as non-renewable energy usage and renewable energy usage, or changing a different aspect of environmental pollution, from CO2 emissions to a more general concept, such as emissions of greenhouse gases (GHG) and other air pollution indicators, will enrich the theoretical and empirical literature on the subject.

 Data availability statement

Publicly available datasets were analyzed in this study. This data can be found here: The datasets analyzed in this study are publicly available and can be found at https://data.worldbank.org/; https://ourworldindata.org and https://www.eia.gov/.

 Author contributions

HL: Supervision, Methodology, Writing – original draft, Conceptualization. MH: Data curation, Formal Analysis, Writing – original draft. NT: Writing – original draft, Investigation, Formal Analysis. MT: Supervision, Writing – review and editing, Validation.

 Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

 Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

 Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

 References Acemoglu D. Restrepo P. (2017). Secular stagnation? The effect of aging on economic growth in the age of automation. Am. Econ. Rev. 107 (5), 174–179. 10.1257/aer.p20171101 Ahmed Z. Nathaniel S. P. Shahbaz M. (2021). The criticality of information and communication technology and human capital in environmental sustainability: evidence from Latin American and Caribbean countries. J. Clean. Prod. 286, 125529. 10.1016/j.jclepro.2020.125529 Ahmed M. Huan W. Ali N. Shafi A. Ehsan M. Abdelrahman K. (2023). The effect of energy consumption, income, and population growth on CO2 emissions: evidence from NARDL and machine learning models. Sustainability 15, 11956. 10.3390/su151511956 Ajmi A. N. Hammoudeh S. Nguyen D. K. Sato J. R. (2015). On the relationships between CO 2 emissions, energy consumption and income: the importance of time variation. Energy Econ. 49, 629–638. 10.1016/j.eneco.2015.02.007 Alkhateeb T. T. Y. Mahmood H. Altamimi N. N. Furqan M. (2020). Role of education and economic growth on the CO2 emissions in Saudi Arabia. JESI 8 (12), 195–209. 10.9770/jesi.2020.8.2 Aydin F. F. (2023). Effect of economic growth, energy use, trade openness and foreign direct investments on carbon dioxide emissions. Evidence from G8 countries. Environ. Sci. Pollut. Res. Int. 30 (53), 113538–113552. 10.1007/s11356-023-29827-5 37853215 Aydin M. (2024). Are CO2 emissions a reason for drought in the United States? J. Water Clim. Change 15 (12), 5973–5997. 10.2166/wcc.2024.536 Bakhsh S. Yin H. Shabir M. (2021). Foreign investment and CO2 emissions: do technological innovation and institutional quality matter? Evidence from system GMM approach. Environ. Sci. Pollut. Res. 28, 19424–19438. 10.1007/s11356-020-12237-2 33392988 Balaguer J. Cantavella M. (2018). The role of education in the environmental kuznets curve. Evidence from Australian data. Energy Econ. 70, 289–296. 10.1016/j.eneco.2018.01.021 Baltagi B. H. (2021). “Econometric analysis of panel data,” in Springer texts in business and economics. Cham: Springer International Publishing. 10.1007/978-3-030-53953-5 Ben Jebli M. Ben Youssef S. Ozturk I. (2016). Testing environmental kuznets curve hypothesis: the role of renewable and non-renewable energy consumption and trade in OECD countries. Ecol. Indic. 60, 824–831. 10.1016/j.ecolind.2015.08.031 Bilal A. Li X. Zhu N. Sharma R. Jahanger A. (2021). Green technology innovation, globalization, and CO2 emissions: recent insights from the OBOR economies. Sustainability 14, 236. 10.3390/su14010236 Chaabouni S. Saidi K. (2017). The dynamic links between carbon dioxide (CO2) emissions, health spending and GDP growth: a case study for 51 countries. Environ. Res. 158, 137–144. 10.1016/j.envres.2017.05.041 28624630 Chen F. Jiang G. Kitila G. M. (2021). Trade openness and CO2 emissions: the heterogeneous and mediating effects for the belt and road countries. Sustainability 13, 1958. 10.3390/su13041958 Dumitrescu E.-I. Hurlin C. (2012). Testing for granger non-causality in heterogeneous panels. Econ. Model. 29, 1450–1460. 10.1016/j.econmod.2012.02.014 Ehigiamusoe K. U. Lean H. H. Smyth R. (2020). The moderating role of energy consumption in the carbon emissions-income nexus in middle-income countries. Appl. Energy 261, 114215. 10.1016/j.apenergy.2019.114215 Eyuboglu K. Uzar U. (2021). A new perspective to environmental degradation: the linkages between higher education and CO2 emissions. Environ. Sci. Pollut. Res. 28, 482–493. 10.1007/s11356-020-09414-8 32815009 Farhani S. Ozturk I. (2015). Causal relationship between CO2 emissions, real GDP, energy consumption, financial development, trade openness, and urbanization in Tunisia. Environ. Sci. Pollut. Res. 22, 15663–15676. 10.1007/s11356-015-4767-1 26018289 Feriansyah F. Nugroho H. Larre A. A. Septiavin Q. Nisa C. K. (2022). Economic growth and CO2 emission in ASEAN: panel-ARDL approach. Econ. Finance Indonesia 68, 4. 10.7454/efi.v68i2.1558 Filonchyk M. Peterson M. P. Zhang L. Hurynovich V. He Y. (2024). Greenhouse gases emissions and global climate change: examining the influence of CO2, CH4, and N2O. Sci. Total Environ. 935, 173359. 10.1016/j.scitotenv.2024.173359 38768722 Friedl B. Getzner M. (2003). Determinants of CO2 emissions in a small open economy. Ecol. Econ. 45, 133–148. 10.1016/S0921-8009(03)00008-9 Grossman G. M. Krueger A. B. (1991). “Environmental impacts of a North American free trade agreement,” NBER Working Papers 3914, Cambridge, MA: National Bureau of Economic Research, Inc. IEA (2024). World Energy Outlook 2024. Paris: IEA. Available online at: https://www.iea.org/reports/world-energy-outlook-2024. IEA (2025). World Energy Outlook 2025. Paris: IEA. Available online at: https://www.iea.org/reports/world-energy-outlook-2025. Katircioglu S. Katircioĝlu S. Saqib N. (2020). Does higher education system moderate energy consumption and climate change nexus? Evidence from a small island. Air Qual. Atmos. Health 13, 153–160. 10.1007/s11869-019-00778-6 Khan M. Rana A. T. Ghardallou W. (2023). FDI and CO2 emissions in developing countries: the role of human capital. Nat. Hazards 117, 1125–1155. 10.1007/s11069-023-05949-4 37125025 Koengkan M. Fuinhas J. A. Vieira I. (2021). Measuring the effect of trade liberalisation on the consumption of non-renewable energy sources in Latin America and the Caribbean countries. Econ. Bus. Lett. 10, 349–358. 10.17811/ebl.10.4.2021.349-358 Mahalik M. K. Mallick H. Padhan H. (2021). Do educational levels influence the environmental quality? The role of renewable and non-renewable energy demand in selected BRICS countries with a new policy perspective. Renew. Energy 164, 419–432. 10.1016/j.renene.2020.09.090 Mahmood N. Wang Z. Hassan S. T. (2019). Renewable energy, economic growth, human capital, and CO2 emission: an empirical analysis. Environ. Sci. Pollut. Res. 26, 20619–20630. 10.1007/s11356-019-05387-5 31104230 Mongo M. Belaïd F. Ramdani B. (2021). The effects of environmental innovations on CO2 emissions: empirical evidence from Europe. Environ. Sci. & Policy 118, 1–9. 10.1016/j.envsci.2020.12.004 Muhammad M. Hasudungan H. W. V. Abubakar A. B. (2025). Per capita carbon footprint convergence in developing countries: the role of energy factors. Clim. Dev. 0, 1–12. 10.1080/17565529.2025.2511900 Musah M. Kong Y. Mensah I. A. Antwi S. K. Osei A. A. Donkor M. (2021). Modelling the connection between energy consumption and carbon emissions in North Africa: evidence from panel models robust to cross-sectional dependence and slope heterogeneity. Environ. Dev. Sustain 23, 15225–15239. 10.1007/s10668-021-01294-3 Odhiambo N. M. (2011). Economic growth and carbon emissions in South Africa: an empirical investigation. Int. Bus. & Econ. Res. J. (IBER) 10, 75–84. 10.19030/jabr.v28i1.6666 Osuntuyi B. V. Lean H. H. (2022). Economic growth, energy consumption and environmental degradation nexus in heterogeneous countries: does education matter? Environ. Sci. Eur. 34, 48. 10.1186/s12302-022-00619-3 Osuntuyi B. V. Lean H. H. (2023). Environmental degradation, economic growth, and energy consumption: the role of education. Sustain. Dev. 31, 1166–1177. 10.1002/sd.2480 Ozcan B. (2013). The nexus between carbon emissions, energy consumption and economic growth in Middle East countries: a panel data analysis. Energy Policy 62, 1138–1147. 10.1016/j.enpol.2013.07.039 Ozturk I. Acaravci A. (2013). The long-run and causal analysis of energy, growth, openness and financial development on carbon emissions in Turkey. Energy Econ. 36, 262–267. 10.1016/j.eneco.2012.08.025 Panayotou T. (1993). Empirical tests and policy analysis of environmental degradation at different stages of economic development. Saboori B. Sulaiman J. (2013). CO2 emissions, energy consumption and economic growth in association of southeast asian nations (ASEAN) countries: a cointegration approach. Energy 55, 813–822. 10.1016/j.energy.2013.04.038 Sahu M. Prusty T. Alahdal W. M. Ariff A. M. Almaqtari F. A. Hashim H. A. (2024). The role of education in moderating the impact of development on environmental sustainability in OECD countries. Discov. Sustain 5, 237. 10.1007/s43621-024-00450-9 Saint Akadiri S. Alola A. A. Olasehinde-Williams G. Etokakpan M. U. (2020). The role of electricity consumption, globalization and economic growth in carbon dioxide emissions and its implications for environmental sustainability targets. Sci. Total Environ. 708, 134653. 10.1016/j.scitotenv.2019.134653 31806296 Salari M. Javid R. J. Noghanibehambari H. (2021). The nexus between CO2 emissions, energy consumption, and economic growth in the U.S. Econ. Analysis Policy 69, 182–194. 10.1016/j.eap.2020.12.007 Salequzzaman M. Davis J. (2003). Environmental education and environmental management in Bangladesh and their sustainability, environmental informatics archives. Environ. Inf. Arch. 1, 70–82. Sajeev A. Kaur S. (2020). Environmental sustainability, trade and economic growth in India: implications for public policy. Int. Trade, Polit. Dev. 4 (2), 141–160. 10.1108/ITPD-09-2020-0079 Seppälä T. Haukioja T. Kaivo-oja J. (2001). The EKC hypothesis does not hold for direct material flows: environmental kuznets curve hypothesis tests for direct material flows in five industrial countries. Popul. Environ. 23, 217–238. 10.1023/A:1012831804794 Sohag K. Chukavina K. Samargandi N. (2021). Renewable energy and total factor productivity in OECD member countries. J. Clean. Prod. 296, 126499. 10.1016/j.jclepro.2021.126499 Stiller Y. Dür A. Huber R. A. (2022). Education and trade attitudes: revisiting the role of economic interest. World Trade Rev. 21, 497–511. 10.1017/S1474745621000562 Tang C. F. Abosedra S. Naghavi N. (2021). Does the quality of institutions and education strengthen the quality of the environment? Evidence from a global perspective. Energy 218, 119303. 10.1016/j.energy.2020.119303 Tejedor G. Segalàs J. Barrón Á. Fernández-Morilla M. Fuertes M. T. Ruiz-Morales J. (2019). Didactic strategies to promote competencies in sustainability. Sustainability 11, 2086. 10.3390/su11072086 Wang Q. Yang T. Li R. (2023). Does income inequality reshape the environmental kuznets curve (EKC) hypothesis? A nonlinear panel data analysis. Environ. Res. 216, 114575. 10.1016/j.envres.2022.114575 36252836 Wen J. U. N. Mahmood H. Zakaria M. (2020). Impact of trade openness on environment in China. J. Bus. Econ. Manag. (JBEM) 21 (4), 1185–1202. 10.3846/jbem.2020.12050 White H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econ. Journal Econ. Soc. 48 (4), 817–838. 10.2307/1912934 Wooldridge Jeffrey M. (2002). Econometric analysis of cross section and panel data. Zafar M. W. Qin Q. Malik M. N. Zaidi S. A. H. (2020). Foreign direct investment and education as determinants of environmental quality: the importance of post paris agreement (COP21). J. Environ. Manag. 270, 110827. 10.1016/j.jenvman.2020.110827 32721301 Zhang Q. Zhang S. Ding Z. Hao Y. (2017). Does government expenditure affect environmental quality? Empirical evidence using Chinese city-level data. J. Clean. Prod. 161, 143–152. 10.1016/j.jclepro.2017.05.096 Zsóka Á. Szerényi Z. M. Széchy A. Kocsis T. (2013). Greening due to environmental education? Environmental knowledge, attitudes, consumer behavior and everyday pro-environmental activities of Hungarian high school and university students. J. Clean. Prod. 48, 126–138. 10.1016/j.jclepro.2012.11.030

Edited by: Shigeyuki Hamori, Yamato University, Japan

Reviewed by: Esther M. Folarin, Covenant University, Nigeria

Fatma Fehime Aydın, Van Yuzuncu Yil University, Türkiye