While the decarbonization of economic growth is an inevitable response to intensified environmental pressures and a fundamental requirement for achieving the Sustainable Development Goals (SDGs), it has also been known as one of the most important policy priorities for countries worldwide¹. This is especially valid in Asia, where rapid industrial development and population growth are driving up energy consumption. Asian countries are striving to achieve economic growth that is not only sustainable but also environmentally safe. However, this balance is struck by placing more reliance on renewable energy (RE) and technology innovation (TI) that have been recognized as crucial instruments in finding ways to reduce carbon emissions at the lowest economic cost. Consequently, limiting greenhouse gas (GHG) emissions has now become a major issue for economies worldwide that have committed to the Paris Climate Accord, including Japan. The SDGs call for urgent action to combat climate change, with one of the targets under SDG 7 being to promote RE use². CO₂ emissions represent a significant proportion of global GHG emissions and are therefore widely regarded as having a significant impact on the environment³. Being capable of absorbing and re-emitting heat, CO₂emissions are strongly linked with climate change and its adverse effects, rising sea levels, heavy precipitation, melting glaciers, and desertification⁴.

Rising emissions may hinder economic growth as environmental protection is not sustainable without long-term economic development⁵. Further, CO₂ emissions have social costs as they contribute to global warming and reduce overall societal welfare. Cooking on carbon-emitting fuels is thought to impact food security by reducing agricultural productivity⁶, as well as to have health impacts on people through the exposure to household air pollution (HAP)⁷. As such, global and national policymakers are actively seeking effective strategies to mitigate the continued rise in global CO₂ emission rates driven by these environmental and socioeconomic challenges.

Globalization also plays a key role in shaping CO₂ emissions trends. Numerous studies examined the relationship between globalization and energy use or human consumption. Trade, investment, and information flow across borders have increased dramatically in the last few decades⁸, as nations tore down trade barriers to stimulate economic growth. This antagonistic feature becomes even more problematic in the context of environmental crises associated with globalization, where increases in economic activity often entail corresponding rises in pollution. As a result, globalization may lead to less energy use and CO₂ emissions as technology is shared and exporting countries become more effective. Globalization may lower emissions over time by raising income and production efficiency. Nonetheless, mixed results are described in earlier studies, and it implies that the environmental impacts of globalization may be different on economies or regions⁹. A similar diversity characterizes the relationship between energy consumption and globalization in both the short and long term¹⁰. Thus, worldwide comparisons will rather indicate whether and to what extent Western countries are influencing the rest¹¹.

There are only a few studies on the contribution of RE to CO₂emission reduction^12,13,14,15. Most studies reveal that RE promotes environmental sustainability and emissions reduction; however, some studies question its effectiveness in improving environmental quality¹⁶. The impact of RE, economic growth, non-renewable energy, and Technological innovation on CO₂ emissions in large-scale regions, including the global level, has been studied, though studies at the Asian scale are still limited using traditional analysis approaches¹⁷.

Technological innovation can improve the efficiency of resources and the move towards economic and environmental sustainability, as raised by international policy debate¹⁸. The importance of TI to mitigate the environmental costs of production processes and achieve SDGs has been highlighted by the International Energy Agency¹⁹. Progress in energy-intensive sectors and energy-related technologies has contributed to lower emissions by improving energy efficiency and accelerating the transition toward low-carbon energy systems. Technological innovation also mitigates the environmental consequences of resource extraction²⁰, thus demonstrating that technological development is important in improving the environment. High-level authorities and decision-makers are increasingly concerned about environmental depletion and in search of new policies to manage the exploitation of resources and promote cleaner technologies.

Natural resource exploitation may stimulate economic and financial development; however, higher levels of economic activity often lead to environmental degradation. Consequently, countries are under pressure to develop policies that promote green growth while addressing both economic and environmental objectives¹⁸. Policymakers are called upon to redirect the trajectory of economic activity toward resource-efficient and environmentally sound goods and services. This study tests these relationships grounded in the Environmental Kuznets Curve (EKC) by using the CS-ARDL method to examine short-run and long-run relationships, capturing possible estimation issues due to cross-country variation.

Further efforts are still required to break the association between environmental degradation and the current patterns of energy consumption and economic growth. So, it is essential to understand how technological innovation, globalization, REC, and economic growth are interconnected and how they collectively affect CO₂ emissions. Hence, the present study examines the impacts of technological innovation, globalization, economic growth, renewable energy, and natural resource utilization on environmental quality in Asian nations.

The alternative text for this image may have been generated using AI.

Full size image

Key Trends.

However, evidence of the cost-effective and sustainable potential of RE technologies to meet energy demand and mitigate CO₂ emissions has accumulated since the mid-twentieth century. A comprehensive literature review providing empirical estimates of the combined impact of renewable energy and technological innovation on economic growth remains lacking, particularly in the Asian context. Despite the fact that many studies have explored the effect of renewable energy on economic growth, which are concentrated on single countries and regions, they pay less attention to the economic diversity of Asian economies. Second, inadequate research has focused on analyzing how technological progress in renewable energy stimulates economic development throughout the Asian region. In light of the dearth of research looking at the joint effects of renewable energy, technological innovation, and economic development in a large heterogeneous sample of Asian countries, our study tries to fill this gap.

The gap between fast technological progress and continued environmental degradation is even more unsettling, as it points out that innovation trajectory and knowledge usage are not sufficiently coherent with where we would have sustainable energy deployment and emissions reductions. This is a major policy failure, as Fig. 1 above shows. Underinvestment in tech innovation for green energy and GHG mitigation can potentially undercut Asia’s push towards sustainable development and climate resilience.

Based on the above discussion, this study addresses the following research questions:

Q1. How do technological innovation and renewable energy consumption, both individually and interactively, influence environmental sustainability as measured by CO2 emissions across Asian economies?

Q2. Does the Environmental Kuznets Curve hypothesis hold for Asian countries when considering the roles of economic growth, globalization, and natural resource rents?

In addition, this work adds to the extant literature as it analyzes the perspectives of over 33 Asian countries from the year 2000 to 2022 and provides empirical evidence about technological innovation, globalization, renewable energy consumption, economic growth, natural resource use, and CO₂ emissions. The study makes the following contributions to the literature. Its re-tests the environmental Kuznets curve hypothesis, with a view on globalization in Asian countries. Second, it constructs an analytical framework to investigate how technological innovation moderates the enhancement of the environmental benefits of renewable energy. Third, it broadens the EKC literature by adding natural resources to economic growth. In this way, the study delivers comparable and policy-relevant empirical evidence on environmental sustainability in Asia.

The remainder of the paper is structured as follows. Section 2 provides an overview of the literature, Sect. 3 describes the methodology used, Sect. 4 reports and discusses the empirical results, and Sect. 5 summarizes the main findings and discusses policy implications.

This study addresses the impact of the GI, globalization (GLO), economic growth, natural resources, and CO₂ emissions by focusing on Asian countries. Second, we highlight the contribution of REC to reducing CO₂ emissions in Asian countries.

Technology innovation and environment

Studies^21,22,23,24 are among the limited body of research that provides evidence of the beneficial effects of TI on environmental quality at the global level. Using data spanning 27 years from 1980 to 2017, Ahmed²⁵ examined the relationship between non-renewable energy, RE, and technological innovation in Malaysia. Empirical results obtained through the ARDL approach confirm a long-run relationship between CO₂ emissions and non-renewable energy, RE, technological innovation, and economic growth. The findings indicate that CO₂ emissions are negatively associated with technological innovation and RE, while non-renewable energy consumption and economic growth are positively associated with emissions in the long run. Empirical evidence also supports the existence of the Environmental Kuznets Curve in Malaysia. Using a sample of 96 countries, Fu²⁶ examined the potential role of technological innovation in reducing CO₂ emissions by employing spatial econometric models based on World Development Indicators data. The study did not find a statistically significant relationship between technological progress and CO₂ emissions for the full sample. However, when countries were categorized by income level, technological innovation was found to play a more significant role in reducing CO₂ emissions in high-income and high-technology emitting countries. In addition, higher levels of globalization were associated with broader spillover effects of technological innovation across countries.

The primary source of economic activity is manufacturing concerns. The manufacturing sector is a key driver of economic progress as it satisfies market needs by converting inputs into outputs according to which inputs transform themselves into outputs based on the production function²⁷. It also creates job opportunities in the manufacturing industry and helps to minimize unemployment, its multiplier effect leads to a rise in GDP, standard of living as well as long-term economic development²⁸. Despite these benefits, pursuing profit while ignoring the environmental and social costs generally leads to large environmental and social externalities. Convergence mechanisms looking at the link between per capita income and the emission of pollutants like carbon dioxide and nitrogen oxides are generally based on Environmental Kuznets Curve theory facts. The EKC hypothesis posits that economic growth initially hurts the environment, but once average earnings reach some threshold level, continued economic expansion is likely to result in environmental improvement²⁹.

Shang³⁰ discovered the nonlinear relationship between per capita CO₂ emissions and per capita GDP in eleven economies in transition during 1993–2010. The authors of the analysis indicated that adding more control variables would improve the precision of any future empirical examination. Li³¹ found it impossible to validate the EKC in a set of eleven Central and Eastern European economies with a socialist legacy. Zhou³² also obtained solid empirical evidence of the EKC hypothesis in Slovakia, Romania and the Czech Republic; such a relationship has been referred to as beneficial environmental policies³³. Stern reported in a related paper that the empirical evidence for an inverted U-shaped (non-linear EKC) is only valid for some environmental parameter values, and it needs to be developed further. The study puts an emphasis on examining factors that do not contribute to the elimination of pollution³⁴. Alola³⁵ tested the EKC hypothesis with ecological footprint indicators and found that alternative models improved the degree of explanation.

A similar study was conducted by Gan³⁶ based on data obtained from 280 cities in China from 2014 to 2018; to analyze how technological efficiency is correlated with technological innovation. The results suggest that technical innovation has a positive impact on the environment. Wang³⁷ empirically analyzed the technology innovation and environmental degradation, characterized that technology efficiency promotes environmental quality as a result of an increase in R&D activity. The majority of empirical evidence suggests that technological progress leads to a mitigation of environmental degradation³⁸ by inducing the adoption of advanced technologies to demonstrate higher energy efficiency and lower utilization rate of fossil fuels³⁹. However, some arrive at the same evident output that technological innovation can only aggravate environmental destruction. Study of Sirivastava⁴⁰ also proposed investment in green technology results in trickle benefits to the economy, but needs resource mobilization (physical) to enjoy it⁴¹. In general, the impact of technological innovation on environmental degradation is ambiguous, and there are successful and unsuccessful examples from empirical studies. The technological innovation progress of belt and road economies has been illustrated in previous work⁴².

Globalization and environment

Environmental sustainability and climate change are some of the key domains in which GLO has been found to have a major influence¹⁰. According to a conceptual model, globalization affects environmental pollution by three different channels: the scale effect, the composition effect, and the technology effect⁴³. The scale effects reflect growing economic activity related to the process of globalization, increased use of energy, and greater environmental degradation⁴⁴. Compositional effect This category accounts for the development and structural shifts of an economy’s industry structure, which might affect how globalization impacts the environment, given the relative importance of nonpolluting and polluting sectors⁴⁵. The technology impact includes all the ways that globalization affects greenhouse gas emissions, by promoting cleaner production practices as well as by intensifying industrial development. Through globalization, developed countries can share environmentally friendly technology with developing nations.

Some empirical evidence on the impact of globalization on the environment with special reference to India. Some studies, like Shahzad⁴⁶ report that globalization causes degradation of the environment. They also state that pollution-intensive industries are shifted to the host, less developed countries of Western MNEs with lower environmental standards. Applying a second-generation EKC model, Duong⁴⁷ examined the data of eight Arctic region countries from 1990 to 2017 and indicated that globalization led to environmental damage. It was also found that globalization fosters enduring economic growth. Likewise, another research analyzed the association between globalization and environmental degradation in 73 developing countries over a span of time ranging from 1990 to 2016 and revealed that globalization decreased environmental degradation only for African and Latin American countries⁴⁸.

Xia⁴⁹ also investigated the effect of globalization on environmental quality and found that globalization has a deleterious effect on the environment. Ahmad⁵⁰ presents empirical results showing a positive impact of globalization on carbon dioxide emissions. By contrast, Cao⁸ found that globalization can diminish environment harm in developing countries. Majeed⁵¹ proposed that the relationship between globalization and carbon emissions is not linear, but with a doubled threshold, which relates to human development at different levels. Other works underscore positive environmental aspects of globalization. For instance, Awosusi⁵² noted that environmentally friendly technology innovations that are enabled by globalization raise GDP and lower emissions and environmental degradation. The overall consensus in the empirical and theoretical arena is ambiguous on the impact of globalization on the environment.

Natural resources and carbon emission

Previous studies, such as^40,53 have attempted to examine the relationship between carbon emissions and natural resource depletion. In addition, Wang⁵⁴ analyzed data from OECD countries to examine the relationship between RE use and CO₂ emissions. The findings indicate that the use of RE sources results in lower CO₂ emissions compared with non-renewable energy sources. The study also provided evidence that urbanization increases CO₂ emissions in line with a Kuznets curve relationship. Previous studies like^3,55 have analyzed the relationship among CO₂ emissions, RE, non-renewable energy, and GDP using panel data from 42 industrialized countries over the period 2002 to 2011. The results show that RE consumption contributes positively to economic growth, whereas non-renewable energy consumption exerts a negative effect.

Liu⁵⁶ employed the ARDL bounds testing approach and the Toda Yamamoto Granger causality method to examine the effects of renewable and non-renewable energy consumption on CO₂ emissions in India from 1965 to 2018. The results indicate that hydroelectric energy consumption has a positive long-run effect on CO₂ emissions, although this relationship is not statistically significant. By contrast, nuclear energy consumption harms CO₂ emissions. The study also shows that CO₂ emissions are strongly influenced by the use of non-renewable energy sources.

With panel data of 102 countries from 1996 to 2012, Bei⁵⁷ re-investigated the nexus between the utilization of RE and economic growth and pollution. The influencing factors of renewable and non-renewable energy consumption were analyzed separately. The findings show that the use of non-renewable energy has a significant effect on emissions between all income groups in the countries being considered.

Data

The choice of the 33 Asian countries in this study was made to attain a broad and representative picture on socio-economic and geographical variations across the region. The selected countries and numbers are listed in Appendix A. Asia is the most populous continent in the world, which covers a wide spectrum of economic development from Japan and South Korea, highly industrialized economies, to India and Indonesia, accessible growth markets. Notably, China and India, with almost 40% of the world’s population, are large consumers of natural resources and significant CO₂ emitters. These two countries also account for a disproportionately large share of global emissions, with China ranking as the world’s largest emitter and India as the third largest⁵⁸. The large populations demand high amounts of energy, also, so they hold a central position in global decarbonization¹⁹.

Apart from its emissions, China and India also account for a large portion of the world’s gross domestic product. China is the world’s second-largest economy, and India is still growing fast⁵⁹. Both countries thus significantly contribute to the global shift towards RE and sustainability. Furthermore, by incorporating additional countries from places like Southeast Asia and the Middle East, where technological innovation and RE deployment also occur at varying rates between high-potential markets and nascent markets, this study captures a wide variety of regional outlooks on this topic.

With the selection of 33 countries, we make sure to have a balance in terms of energy demand, CO₂ output, and technological development for most crucial Asian economies. Data for these economies are uniformly reliable and comparable in all the variables used, such as CO₂ emissions, GDP, RE consumption, technological innovation (technology), natural resource rents (NRRENTS), and globalization for 2000 through 2022. Including economies with different income in Asia enhances the empirical evidence drawing experiences on development stage, energy structure, and innovativeness. This wide coverage guarantees that the findings are robust and policy-relevant implications can be drawn for the region as well. The explanation of the variables is presented in Table 1.

The alternative text for this image may have been generated using AI.

Full size image

Shows the relationship among variables.

All data were converted throughout the study so that the resulting series would be normally distributed. Table 1 presents some quick facts on CO2, GDP, GLO, RE, NRR, and TI. Figure 2 shows the relationships between the variables.

Theoretical underpinning and model development

This study examines how Asian countries’ use of RE, NRR, globalization, and TI affects their CO₂ emissions. Globalization and economic growth are two additional control variables included in the study model. The econometric model used in this study is based on economic theories and existing empirical studies. Theoretically, this study’s framework is based on the EKC hypothesis proposed by Grossman and Krueger⁶⁰. They suggest that there is an inverted U-shaped association between economic growth and environmental degradation, indicating that environmental quality worsens in the early stages of economic growth but recovers at higher income levels. This hypothesis has profoundly influenced the development of econometric models that analyze the impact of GDP on environmental outcomes (CO₂). The inclusion of the globalization (GLO) variable in the model is linked to the study of Khaliq⁶¹. They showed that globalization may exacerbate or alleviate environmental degradation, depending on the level of technological advancement and the nature of energy consumption. The model also includes the interaction terms (RE×TI) and (NRR×TI), reflecting the statement that technological advancement may enhance the effectiveness of RE in reducing CO₂, and that the association between natural resources (NRR) and CO₂ emissions may depend on technological progression. This approach is supported by Khaliq⁶², who found that technological advancement plays a critical role in decoupling environmental degradation from economic growth. In terms of econometric methodology, we adopted the CS-ARDL approach, which has been extensively used in empirical studies to analyze the long-term association between economic growth, energy use, and environmental outcomes. The CS-ARDL approach is particularly well-suited for this study because it considers the short- and long-term dynamics while considering the potential endogeneity among variables. This method has been successfully applied in studies by Dai⁶³ and Khezri⁶⁴ to investigate the relationships among energy consumption, economic growth, and CO₂. Therefore, the model development in this study is grounded in a solid foundation of economic theory and empirical studies that link GDP, globalization, RE, and technological advancement with CO₂. Following Wang⁶⁵, the functional form of the models is presented as follows:

In the above Eq. (1), CO₂ represents Carbon dioxide emission, which is utilized as the dependent variable. (:GDP,GLO,RE,TI) represent gross domestic product, globalization, renewable energy, and technology innovation, respectively, which are considered explanatory variables. While (:(textRtextEtext*textTtextI)) and (:left(textNtextRtextRtext*textTtextIright)) denotes the technology innovation impact on CO₂ emissions with the moderating role of RE and natural resource rent.

To analyze the effects of RE, GLO, GDP, and TI on CO₂ emissions, the first model is derived from Eq. (1), which is presented below:

In the Eq. (2),(:beta:) is the intercept or constant, which is the value of the dependent variable when all explanatory variables are zero, whereas i and t denote the countries and time, respectively.(:beta:_1-beta:_4) are the coefficients of the independent variables and (:mu:) is the error term.

In the second model, we examine how changes in the NRR, GLO, GDP, and TI affect CO2.

In Eq. (3),(:beta:) is the intercept or constant, which is the value of the dependent variable when all explanatory variables are zero, whereas i and t denote the countries and time, respectively.(:beta:_1-beta:_4) are the coefficients of the independent variables ((:GDP,GLO,RE,TI)) and (:mu:) is the error term.

We assess the impact of GDP, GLO, NRR, and TI, along with the moderating role of RE (RE*TI), on CO₂ emissions in the third model.

In Eq. (4), (:beta:) is the intercept or constant, which is the value of the dependent variable when all explanatory variables are zero, whereas i and t denote the countries and time, respectively. (:While:beta:_1-beta:_5) are the coefficients of the independent variables ((:GDP,GLO,NRE,TI,textRtextEtext*textTtextI)) and (:mu:) is the error term.

The fourth model assesses the impact of GDP, GLO, NRR, and TI, along with the moderating role of natural resource rent (NRR*TI), on CO₂ emissions.

In Eq. (5),(:beta:) is the intercept or constant, which is the value of the dependent variable (CO₂ emissions), when all explanatory variables are zero, whereas i and t denote the countries and time, respectively.(:beta:_1-beta:_5) are the coefficients of the independent variables ((:GDP,GLO,TI,textNtextRtextRtext*textTtextI)) and (:mu:) is the error term.

Research methodology

Panel heterogeneity and cross-sectional test

Since knowing whether or not slopes are heterogeneous can affect regression analysis and the validity of hypothesis testing, the authors employed a test developed by⁶⁶ to answer these questions. Cross-sectional dependencies (CDs)⁶⁷ are often believed to exist between indicators in basic panel data, which might skew results and introduce bias. It has been proposed that panel data models should have a high CD in light of the available literature⁶⁸. Given the extensive financial and economic ties between countries, assessing cross-national dependencies among the selected countries is crucial. As such, it is clear that cross-sectional units are dependent upon one another. In addition, these methods offer helpful unit root tests for deeper investigation.

Co-integration test

Once stationarity of the parameters is established at the first difference, the cointegration process can begin. Using this method, we may determine if the parameters are correlated throughout time or whether they change independently of one another. Long-run correlations between variables can be analyzed using panel cointegration. Therefore, we relied on the cointegration methods described by Westerlund⁶⁹. Compared to first-generation testing⁶⁹, approach is superior because it considers CD and SH.

CS-ARDL model

To scrutinize the short- and long-term associations between the regressors and response variables, this study applied the CS-ARDL approach. Pesaran⁷⁰ was the first to propose this approach based on the common correlated effect (CCE), and it was later extended by Chudik and Pesaran⁷¹. This approach is preferred over other panel data techniques due to its efficiency in undertaking particular issues commonly raised in panel data, including CD and slope heterogeneity⁷². The CD problem occurs when units in the panel (such as countries, entities, or firms) are connected due to common shocks or global influences. By incorporating this issue into the model through the CS-ARDL approach, the accuracy of the estimates is enhanced^73,74. Furthermore, the CS-ARDL method accounts for slope heterogeneity, which means that the associations between variables can vary across units in the panel. This flexibility is significant in practical applications because dissimilar units may respond differently to the same economic factors. Notwithstanding these variances in slope coefficients, the model can still estimate meaningful group averages^75,72. The CS-ARDL model also effectively addresses endogeneity issues, which are common in dynamic panel data models in which explanatory variables are correlated with the error term. This issue can also lead to biased results, but incorporating lagged cross-sectional means into the model helps lessen this issue, ensuring the results are more reliable and consistent⁷¹. Finally, the CS-ARDL model can handle mixed orders of integration among parameters, which is particularly valuable when dealing with variables that exhibit dissimilar integration orders^76,77. The CS-ARDL is estimated using the following model.

In Eq. (5),(:y_it) shows the dependent variable (CO₂) for individual cross-section(i), and time (t);(:alpha:_i) is the Fixed effect for cross-section (i), representing any time-invariant features of that cross-section.(:sum:_l=1^p_ylambda:_l,iy_i,t-l) is the autoregressive part of the model, where the (:y_i,t-l) (CO₂) will be regressed with its own past value. The coefficient ((:lambda:_l,i)) represent the influence of these past values. The term (:sum:_l=0^p_xbeta:_l,ix_i,t-l:) includes the lagged values for independent variables ((:GDP,GLO,RE,TI,textRtextEtext*textTtextI,textNtextRtextRtext*textTtextI)) for each cross-section. The coefficient (:beta:_l,i) shows the impact of these lagged variables on the CO₂ emission. Whereas (:sum:_l=0^p_phi:phi:_i,l^prime:stackrel-z_i,t-l) indicate the additional variables that also affect the response variable. Finally,(:epsilon:_it) is the error term, representing unobserved factors affecting (:y_it) (dependent variable).

The long-run coefficients of the mean group estimates represent the lagged cross-sectional average as follows:

In Eq. (6),(:widehattheta:_CS-ARDL,i) shows the long-run parameter estimates for cross-section (i), calculated by the ratio of the sum of the estimated coefficients ((:widehatbeta:_l,i)) to the normalization factor, including the autoregressive coefficient (:left(widehatlambda:_l,iright)).(:widehattheta:_MG) is the mean group estimate, representing an average long-run estimate across all cross-sections (denoted as N). It is calculated by averaging the individual estimates ((:widehattheta:_i)) from each cross-section. The corrective form of the CS-ARDL method is as follows.

Equation (7) delivers a corrective form of the CS-ARDL estimate that explicitly accounts for short-term dynamics by using changes in the variables (Δ) over time. The term (:Delta:y_it) representing change in the CO₂ emission for the cross-section (i) and time (t), while (:varnothing:_ileft[y_i,t-l-widehattheta:_ix_i,tright]) denotes the error correction mechanism, where the model corrects the deviation from the long-run equilibrium. (:alpha:_i) shows the fixed effect for cross-section (i), accounting for individual time-invariant characteristics. The terms (:sum:_l=1^p_y-1lambda:_l,iDelta:_ly_i,t-l),(:sum:_l=0^p_xbeta:_l,iDelta:_lx_i,t-l:andsum:_l=0^p_phi:phi:_i,l^prime:Delta:_lstackrel-z_i,t-l) represent the lagged change in dependent, independent, and external variables, respectively. Finally,(:u_it) is the error term for the corrected equation.

Regarding sample size, statistical power, and bias, CCE mean group estimators with delayed augmentation perform adequately when ∅_i is a suitable error correction adjustment speed. However, a negative bias emerged when researchers utilized the cutoff of t 50. On this basis, the split-panel jackknife method has been acknowledged by⁷⁸, to mitigate the effects of a relatively small sample size. The following equation describes the jackknife strategy:

In Eq. (8), (:stackrelsimpi:_MG) represents the mean group technique, which is the average long-run parameter estimation across all cross-sections. While, (:frac2stackrelsimpi:_MG-^12left(widehatpi:_MG^a+widehatpi:_MG^bright)) delivers the adjusted estimator, dealing with bias due to small sample sizes.

Robustness tests

We use two econometric co-integration regression methods, namely, FMOLS and DOLS. FMOLS, designed by Phillips⁷⁹, is one of the semi-parametric methods that examine long-run parameters. This method yields strong results, even for small samples, and solves problems of endogeneity, serial correlation, omitted variable bias, and measurement error⁷⁹. FMOLS can accommodate heterogeneity of long-run parameters and perform a single cointegrating association for a combination of I (1) variables⁷⁹. This method is aimed at modifying both data and parameters, which is an improvement on outmoded EG cointegration techniques⁸⁰. In addition, the FMOLS estimator has been found to increase the inference in empirical analysis⁸⁰. DOLS, which is another co-integration regression method used in the current study, was designed by Stock and Watson⁸¹. It also takes care of the problems of endogeneity and serial correlation, which gives legitimate estimates for long-run parameters⁸¹. DOLS aims at controlling the bias in the coefficient estimations due to the existence of endogeneity⁸¹.

The summary statistic of the studied variables provides at Table 2. First, CO₂ has the widest variation of any metric, falling between 198.64 and 9899.34 on average; TI falls between 3,140 and 1,542,002 on average; RE falls between 3.18 and 58.65 on average; NRR falls between 0.95 and 22 on average; GDP drops between 14.61 and 13.64 on average; and GLO rises from 0.06 to 6.19 on average. All the variables are positively skewed except GDP. Even more so, whereas RE, TI, NRR, CO₂, and GDP are all leptokurtic, GLO is the only platykurtic.

Estimation for slope heterogeneity (SH) and CSD is the most pressing concern in panel data analysis. The economic interconnectedness of countries contributes to the CD issue, as do standard shocks and unnoticed components in the panel. Before performing the final estimation, we checked for possible heterogeneity and CD in the panel using slope heterogeneity based on the study of Pesaran⁸² and CSD tests; the results are shown in Tables 3 and 4.

Table 4 shows that the hypothesis that all four models have the same slope distribution was rejected. Also, the CSD is supported by Table 4’s results, which show that the null hypothesis is rejected at the 1% significance level. It indicates that cross-sections are interdependent, leading us to conclude that a shock in any aspect of one country will also influence the other nations.

In the presence of CD, many recent study has shown that first-generation unit root testing methods are unsuitable for determining the stationarity level of the variables and have instead recommended second-generation unit root testing⁸³. Table 5 shows the results of our CADF and CIPS checks on the stationarity of the variables under consideration.

Table 6 shows the findings of a test for unit roots, which shows that the hypothesis of non-stationarity after the I(1) was rejected. All the variables in our investigation were assumed to be stationary at the I(1) and to have an integration at the first difference⁸⁴, supported by our findings.

In addition, from a policy perspective, it’s critical to pinpoint the cointegration among factors. The study used Westerlund cointegration⁸⁵ test to observe the long-term association between variables, as it yields reliable estimates while considering the problem of CD. Table 6 shows the outcomes of the cointegration test.

The Westerlund cointegration test verifies the existence of cointegration between the analyzed variables. It rejects the null hypothesis due to no co-integration at the 5% and 1% significance levels across all four models.

Finally, Table 7 shows the results of a CS-ARDL analysis of the independent variables’ short- and long-term dynamics. All four models employ the same control variables, namely GLO and economic growth proxy by per capita GDP, in line with⁸⁶.

The obtained findings show that the GDP has a positive and statistically significant effect on CO₂emissions across all four models. These results addressing the impact of GDP reveal critical evidence that the BRICS economic boom is damaging the quality of the environment^87,88 are only a few studies that have identified a relationship between economic growth and CO₂ emissions. These studies show that rising GDP leads to more economic activity, increasing energy consumption and CO₂ emissions. In contrast, it is estimated that GLO positively influences the environment in the panel countries. The results show that, except for the first short-run model, GLO hurt CO₂ emissions in Asia at both the 5% and 10% levels of significance. GLO impact suggests that a large portion of BRICS investment is going toward green manufacturing and energy-saving technologies that reduce CO2 emissions. In China, Zeng⁸⁹ and in Khan⁹⁰ find that GLO reduces CO₂ emissions.

One of the key independent variables utilized across all four models is TI. According to Table 8, all four models show a negative and significant effect of TI on CO₂ emissions in the long run, except for model 4, where the effect of TI on CO₂ emissions in the short run is negligible. Several empirical investigations^24,91 corroborate our findings about the role of technology in lowering CO₂ emissions. Pata⁹² stated that TI can help Asian countries cut carbon emissions by increasing traditional energy sources and decreasing pollution from manufacturing. Additionally, projects in energy technology in China and Brazil are leading to a dramatic decline in carbon emissions. Also, Chen⁹³ stated that TI can help Asian countries cut down on CO₂emissions by raising the competence of traditional energy sources and decreasing pollution from manufacturing. In addition⁹⁴, found that Asian countries are on pace to cut CO₂ emissions through energy technology and R&D investment over the next few years. One typical justification for this impact is the belief that technological advancements are necessary to reduce damaging environmental impacts and increase resource productivity. One-way environmental technology can directly maintain environmental values is by reducing trash disposal in the natural environment. The energy transition is another area in which technology may help the environment. Since TI promotes energy conversion, it might also help raise RE production, which would benefit the environment.

Then, in model 1, we account for GDP, GLO, and TI to see how the use of RE affects carbon emissions. Empirical studies found strong evidence for a negative correlation between RE and CO₂ output. CO₂ emissions are reduced by 0.24% and 0.35% for every percentage point increase in RE. Similar estimates were found in the investigations of^56,95. Since RE is created from non-fossil resources, these studies suggested it contributes to reducing CO₂ emissions because it produces no or very few emissions. In addition, Zhou⁹⁶ emphasized the importance of RE in reducing CO₂ emissions and supporting sustainable development, and they mandated that its use reduce CO₂ emissions in the Asian countries. An increase in RE, according to Zheng⁹⁷ emphasized the importance of RE in reducing CO₂ emissions and supporting sustainable development. They mandated that its use reduce CO₂ emissions in Asian countries. According to Hailiang⁹⁸, an increase in RE has a significant impact on lowering technological hurdles such as scarce energy sources and inefficient infrastructure in Asian countries. The combination of RE’s environmental benefits and the rising global need for energy means it will overtake all other forms of power shortly. The IRENA⁹⁹ report claims that 90% of the required decrease in CO₂ emissions connected to energy use may be provided by RE technologies at the lowest possible cost.

Model 2 also focuses on natural resources (NRR) as a primary independent variable, with the same control factors as Model 1. Results from Model 2 show that NRR also harms CO₂ emissions. CO₂ emissions are reduced by 0.01% over the long term and by 0.02% over the short term for 1% rise in NRR. While many studies have found that increasing the use of RE sources also increases CO₂ emissions, the study results are consistent with those of^100,101, who found that RE has a negative impact on CO₂ emissions. Despite its adverse effects, RE sources abundance paves the way for a more carbon-conscious economy by decreasing reliance on imported fossil fuels like natural gas and oil. These results are linked to the fact that Asian countries are increasingly relying on domestic energy sources like renewables and natural gas, which produce fewer greenhouse gas emissions than the fossil fuels those countries import. The Asian countries depend on NRR; hence this topic is essential for tackling CO₂ mitigation¹⁰².

TI, according to^91,103,104, reduces CO₂ emissions by increasing energy efficiency and incorporating carbon-free technology into the manufacturing process. In light of these considerations, models 3 and 4 present long- and short-term outcomes from an analysis of how TI interacts with RE and NRR to affect CO₂ emissions. The long-term effect of a 1% rise in TI*RE on CO₂ emissions is a reduction of 0.025%, but the short-term effect is negligible. However, model-4 results show that the combined impact of TI with NRR is negative and statistically significant over both the short and long term. In the long run, a 1% rise in TI*NRR reduces CO₂ emissions by 0.007% and in the short run, by 0.091%. In addition, the coefficient of ECM demonstrates annual adjustment speeds of 41%, 47%, 74%, and 46% to establish long-run symmetry and is empirically significant at the 1% level in all four models.

Robustness tests

For robustness purposes, two additional estimations, namely FMOLS and DOLS, are conducted. Table 8 reports the relationship between GDP, globalization, RE consumption, natural resource rents, technological innovation, and the interaction terms TI*RE and TI*NRR with CO₂ emissions in the selected countries. The coefficient of GDP shows a positive and statistically significant effect across all three estimation techniques, including CS-ARDL, DOLS, and FMOLS. Similarly, the direction and statistical significance of the globalization coefficient remain consistent across all methods. However, the magnitude of the effect differs, being strongest in the FMOLS estimates, moderately negative in the DOLS estimates, and weakest in the CS-ARDL results, while remaining statistically significant in all cases.

RE consumption also exhibits a consistently negative and significant relationship with CO₂ emissions across all estimation techniques. The magnitude of this negative effect is stronger in the FMOLS and DOLS models than in the CS-ARDL estimates, although the direction and significance remain consistent. In contrast, the results for natural resource rents show variation across methods. Both FMOLS and DOLS indicate a positive and significant association with CO₂ emissions, whereas the effect is weaker or statistically insignificant in the CS-ARDL estimates.

Technological innovation displays a negative and statistically significant effect on CO₂ emissions in the CS-ARDL and FMOLS models, while it shifts to a positive and significant coefficient in the DOLS estimation. The moderating effects of TI through interaction terms with RE and natural resource rents also vary across estimation techniques. In the CS-ARDL results, the interaction term TI*RE is negative and statistically significant, whereas it becomes positive and significant in the FMOLS model and remains negative in the DOLS estimation. Similarly, the interaction term TI*NRR shows a negative and significant effect in the CS-ARDL and DOLS results but a positive association in the FMOLS estimates.

Finally, CO₂ emissions and all independent variables are examined using the Dumitrescu–Hurlin panel causality test. The results reported in Table 9 indicate the presence of bidirectional causality between GDP and CO₂ emissions, as well as between RE consumption and CO₂ emissions. In contrast, unidirectional causality is observed running from globalization to CO₂ emissions, from natural resource rents to CO₂ emissions, from technological innovation to CO₂ emissions, and from the interaction terms NRR*TI and RE*TI to CO₂ emissions.

This study examined the roles of economic growth (GDP), technological innovation (TI), globalization (GLO), and renewable energy (RE) in determining CO₂ emissions across 33 Asian economies over the period from 2000 to 2022. Using the CS-ARDL approach, supported by robustness checks based on the FMOLS and DOLS estimators, the study provides strong empirical evidence supporting the Environmental Kuznets Curve (EKC) hypothesis in the Asian context. The results indicate that economic growth initially increases CO₂ emissions; however, beyond a certain income level, further economic development contributes to a reduction in environmental degradation, confirming an inverted U-shaped relationship. This pattern is particularly evident in countries such as Pakistan. Furthermore, technological innovation is found to significantly reduce CO₂ emissions in both the short and long run, highlighting its critical role in promoting green development. These findings support the argument that economic growth can be decoupled from environmental degradation through sustained investment in innovation and clean technologies.

The study also finds a negative and statistically significant link between renewable energy consumption and CO₂ emissions, which highlights the necessity to promote the use of renewable energy sources in fast-emerging Asian economies (like China, India, and Russia). While having different impacts, globalization favors exchange in goods and technology, but is likely to bring increased environmental pressure if not guided by a sustainability framework. These results provide policy lessons. For sustainable development in the ambitious Asian economies, green innovation is of high priority, renewable energy infrastructure needs to be scaled up, and the policy measures regulating it are needed for controlling environmental footprint by globalization. With global demand for resources fast approaching the limit of what the planet can provide, we are now witnessing such diverse strategies as the circular economy or energy efficiency, and green building standards being deployed nearly overnight to fill a growing gap.

Policy implications

The implications of the study are that it provides policy recommendations to east Asian economies in relation to how they should mix their economic growth and environmental sustainability. The inverse association found between technological innovation, green energy use, and CO₂ emissions confirms the EKC effect hypothesis and emphasizes the sustaining role of environmentally sound technologies. As they seek economic expansion, leaders should be aware of the dangers of unregulated growth. Strategic investments in innovation and RE technology can become effective instruments for decoupling economic growth from environmental harm. Advances in technology increase the efficiency of energy utilization and provide cost-effective, lower-emission options in multiple sectors.

The study also reveals the paradoxical effects of globalization itself, revealing that with no stringent environmental governance, FORs flow and industrialization can accelerate ecological degradation. To avoid severe environmental consequences in the future, policymakers will need a two-pronged approach that puts innovation, clean energy, and sustainability at the heart of national development strategies both now and in the future.

Key policy recommendations include:

• Increasing investment in technological innovation to enhance energy efficiency and reduce emissions, particularly in high-impact industries.

• Expanding renewable energy infrastructure, including wind, solar, and biomass, to reduce reliance on fossil fuels and support long-term decarbonization.

• Promoting green urban planning to develop low-carbon and resource-efficient cities with sustainable transport, energy systems, and housing.

• Strengthening education and public awareness programs that emphasize environmental responsibility and sustainable consumption.

• Enhancing regulatory frameworks to limit the import and use of environmentally harmful technologies and encourage green trade practices.

• Supporting research and development institutions focused on clean technologies and environmental innovation to promote inclusive green economic growth.

These policy measures are essential for aligning Asia’s economic ambitions with the Sustainable Development Goals and global climate commitments.

Limitations and Future Directions

Despite such findings, there exist some limitations. Firstly, it adopts the patent quantity as the surrogate for technology innovation. While it is true that the empirical literature uses mostly patents in research and standardize at an international level, this yet does not represent the performance, environmental relevancy or diffusion within a nation of innovations. In future research, authors could improve their accuracy by (1) using other indicators from innovation theory, like research and development expenditures or the diffusion of green technology, or (2) adopting particular outputs of innovation with respect to clean energies.

Second, the 33 Asian economies were chosen based on data availability and completeness over the time spanned 2000–2022. Consequently, certain countries and factors had to be removed from the analysis due to missing or conflicting data, which might affect the generalizability of our inferences. Further research could increase the number of samples, correct for missing data, or use more recent sources to improve coverage across regions and over time.

Third, while the CS-ARDL approach allows for robust estimation with respect to cross-sectional dependence and slope heterogeneity, it imposes linearities across the variables. This simplifying assumption may fail in the presence of intricate environmental and economic links. Subsequent research might use nonlinear model techniques such as threshold regressions and panel quantile regressions to address the possibility of asymmetries and differences in effects among income or development groups.

Moreover, no other possible relevant factors such as institutional quality, environmental regulation degree, population evolution or energy prices are considered in the model. These and other such factors might mediate or moderate the effects of economic growth, technology innovation, energy policy, and environmental outcomes. Including such elements in future models would give a deeper explanation for the decarbonization process. This study is only for Asian economies. Finally, this paper covers only Asian countries. Although this encourages local policy relevance, the generalizability of the results to other global settings is constrained. It remains an open question whether these patterns and policy implications are generalizable to Africa, Latin America, or Europe, and future research could expand this analysis in the form of comparative cross-regional or global panel studies.