Assessing the Sustainability of Southeast Asia’s Energy Transition: A Comparative Analysis

Abstract

The rapid economic growth in Southeast Asia has heightened concerns about its environmental sustainability, particularly in relation to CO₂ emissions. Despite the growing focus on climate change mitigation, the region faces significant challenges in balancing economic development, energy transitions, and environmental conservation. Existing studies often overlook the complex interplay between these factors, leaving a critical gap in understanding how tailored strategies can address country-specific dynamics. To bridge this gap, this study introduces the “Sustainable Energy-Environment Nexus” (SEEN) framework, which integrates economic growth, energy transitions, and environmental conservation as interconnected elements necessary for achieving carbon neutrality in both the short and long run. Using data from eight Southeast Asian countries (Indonesia, Malaysia, China, South Korea, Vietnam, Singapore, Thailand, and Japan) over the period 1990–2023, this study employs the Autoregressive Distributed Lag (ARDL) approach and the Vector Error Correction Model (VECM) technique to analyze the relationships between CO₂ emissions, GDP, financial development, forest areas, renewable energy, non-renewable energy consumption, and trade openness. The findings reveal that GDP and non-renewable energy consumption significantly drive CO₂ emissions in countries like Indonesia, Malaysia, Japan, and South Korea. Conversely, forest areas, financial development, renewable energy, and trade openness are effective in reducing emissions in countries such as Vietnam and China. This study highlights the critical role of renewable energy adoption while addressing challenges such as inadequate infrastructure and limited technology transfer. It also identifies opportunities for regional cooperation in innovation and policy harmonization. To support sustainable energy development, tailored policy recommendations include incentivizing investments in renewable energy, enhancing technology transfer, expanding forest conservation efforts, and aligning regional renewable energy targets across ASEAN. The SEEN framework provides a robust foundation for advancing research and policy design aimed at reducing CO₂ emissions and promoting environmental sustainability across Southeast Asia.

1. Introduction

The growing global population and expanding economies have significantly increased the demand for energy. Energy is fundamental not only for national economic growth but also for improving people’s quality of life. It powers industries, transportation, businesses, and households, making it an essential driver of productivity, competitiveness, and economic development. Reliable and affordable energy sources enable industrial expansion, infrastructure development, and technological advancement, which are all critical to a nation’s prosperity. Furthermore, access to energy enhances living standards by supporting essential activities such as cooking, heating, and lighting. Conversely, the lack of access to energy can lead to economic stagnation, low productivity, and entrenched poverty.

However, the overwhelming dependence on non-renewable energy sources such as coal, oil, and natural gas poses significant environmental challenges. These energy sources are finite and are the primary contributors to greenhouse gas emissions, exacerbating climate change and global warming. This alarming trend has intensified the need for a transition to renewable energy sources, such as hydropower, solar, and wind energy, which are more environmentally sustainable. While renewable energy adoption offers the potential to limit carbon emissions and reduce reliance on non-renewable sources, the transition is fraught with challenges, including high infrastructure costs, inadequate technological readiness, and policy misalignments across nations. Southeast Asia, a region characterized by rapid economic growth and industrialization, is at the forefront of these energy and environmental challenges. The region’s heavy reliance on non-renewable energy has contributed significantly to its rising CO₂ emissions, making it critical to identify factors influencing emissions and develop effective strategies for mitigating their impact. Despite a growing body of research on the environmental effects of energy consumption, there is limited understanding of how economic, environmental, and energy factors interact in Southeast Asia. Existing studies often focus on individual determinants or adopt a one-size-fits-all approach, overlooking the region’s diverse economic structures, environmental policies, and energy needs. This represents a significant research gap as there is a pressing need to explore country-specific dynamics and propose tailored policy solutions.

To address this gap, this study examines the interactions between renewable and non-renewable energy consumption, economic growth, financial development, forest cover, trade openness, and CO₂ emissions in eight Southeast Asian countries (Indonesia, Malaysia, China, South Korea, Vietnam, Singapore, Thailand, and Japan) from 1990 to 2023. By analyzing these relationships, the study provides insights into the drivers of CO₂ emissions in the region and highlights opportunities for promoting sustainable energy development.

The primary objective of this research is to investigate the determinants of CO₂ emissions in these countries, with a focus on the roles of economic growth, financial development, forest cover, renewable energy, non-renewable energy consumption, and trade openness. Specifically, this study aims to fill the research gap by addressing the following questions: How do economic growth, financial development, forest cover, renewable and non-renewable energy consumption, and trade openness influence CO₂ emissions in the selected Southeast Asian countries? Are there significant country-specific differences in the relationships between these factors and CO₂ emissions? What policy implications can be derived from the findings to promote sustainable energy development and reduce CO₂ emissions in the region?

2. Background of Analysis

The link between energy consumption and CO₂ emissions in North Africa from 1990 to 2015 was covered by Mohammed et al. in 2021. They discovered that energy consumption has a significant role in determining CO₂ emissions in these nations by using panel data and cross-section approaches. Additionally, they found that there are bidirectional causal linkages between economic growth and carbon emissions as well as between energy use and carbon emissions.

In a different study, Ref. [1] showed that renewable energy had a short-term detrimental impact on CO₂ emissions in Thailand using the ARDL technique. On the other hand, both the short- and long-term effects of the rise in the domestic product and the rate at which non-renewable energy is being reduced were favorable for CO₂ emissions.

According to [2] renewable energy resources lead the sequence, and CO₂ emissions are out of phase with an anti-cyclic effect. The estimation findings indicate a significant long-term correlation between CO₂ emissions and renewable energy sources. Using panel data for 33 OECD countries from 2000 to 2018, Ref. [3] employed a nonlinear panel smooth transition regression model and discovered that the use of renewable energy has a large nonlinear influence on CO₂ emissions. But as globalization increases, the CO₂ emissions that come with using renewable energy sources grow more significant.

Ref. [4] investigated the connection between CO₂ emissions and renewable energy in Turkey using the ARDL approach. They discovered a cointegration connection between both, indicating that the long-term reduction of CO₂ emissions depends heavily on renewable energy. Their Toda–Yamamoto causality test, however, revealed a unidirectional causal link between the use of renewable energy sources and CO₂ emissions. Ref. [5] discovered in another study that although the industrial sector and the gross domestic product had a weakly favorable influence on CO₂ emissions, non-renewable energy consumption had a positive impact on CO₂ emissions in seven chosen nations. After analyzing energy consumption data from 2010 to 2019 in Sichuan Province, China, Ref. [6] discovered that energy use assembly has a positive effect on CO₂ emissions.

For 193 UN members, Ref. [7] conducted an analytical analysis of the correlations between CO₂ emissions and non-renewable energy sources. The analytical findings indicate a positive correlation between the human progress indicator and total energy consumption and carbon dioxide emissions. The indicator of human inequality is related to whole energy usage.

Ref. [8] contended that the use of non-renewable energy and economic expansion positively affect the burning of carbon emissions in sixteen EU member states. When Ref. [9] looked at 16 Latin American nations’ non-renewable energy usage and environmental deterioration, they discovered a strong association between the two. On the other hand, using fossil fuels results in higher CO₂ emissions. In the G-7, a bidirectional causal link between renewable and non-renewable energy use was discovered by Ali et al. in 2022. According to [10] non-renewable energy prices significantly and favorably affect the amount of renewable energy used in the US.

Using data from 55 nations, Ref. [11] examined the link between renewable and non-renewable energy use. There is proof that a rise in the use of renewable energy sources is linked to a fall in the use of non-renewable energy sources, and vice versa. Green areas play a key role in lowering carbon dioxide emissions according to various publications. According to a [12] study from South Korea, urban green areas can cut carbon dioxide emissions by as much as 8.9% annually. Comparably, a Chinese study by [13] showed that adding more green space to cities can cut carbon dioxide emissions by as much as 15%. According to a 2015 Kardan et al. research conducted in Canada, there might be a 1% drop in air pollution for every 10% increase in green area. Research conducted in France by [14] found that urban green areas can store up to 27.7 metric tons of carbon dioxide per hectare, offering substantial advantages for carbon sequestration. These results imply that by lowering carbon dioxide emissions, the development and maintenance of green areas in urban settings might be extremely important in lessening the consequences of climate change.

Ref. [15] used panel data and cross-section methodologies to investigate the link between North Africa’s non-renewable energy use and CO₂ emissions from 1990 to 2015. According to the estimated results, one of the main factors influencing CO₂ emissions in these nations was the use of non-renewable energy. The findings also show that there is a reciprocal causal link between economic growth and carbon emissions and energy consumption and carbon emissions. According to [16] European nations ought to encourage the use of renewable energy. Actually, the empirical findings demonstrated that using renewable energy has a direct and beneficial impact on lowering CO₂ emissions. The detrimental impact of using renewable energy on carbon dioxide emissions was proven by [17]. This conclusion was reached by analyzing the nonlinear relationship between CO₂ emissions and the use of renewable and non-renewable energy. It is based on the assessment of panel data encompassing 97 nations between 1995 and 2015.

The global energy landscape has undergone significant transformations in recent years, particularly exacerbated by the ongoing energy crisis. This crisis, primarily triggered by geopolitical tensions and supply chain disruptions, has led to soaring energy prices, increased energy insecurity, and heightened concerns about environmental sustainability. Southeast Asian nations, while relatively less affected by the immediate impacts of the crisis compared to regions like Europe, are nevertheless vulnerable to its long-term consequences. The region’s growing energy demand, coupled with its heavy reliance on fossil fuels, makes it susceptible to price volatility and supply disruptions. To address these challenges and mitigate the environmental impact of energy consumption, Southeast Asian countries are increasingly turning to renewable energy sources. However, various factors, including technological limitations, financial constraints, and policy barriers, hinder the full potential of renewable energy development.

While previous studies have established a positive link between non-renewable energy consumption and CO₂ emissions in different countries, there is a lack of research that explores the nuances of this relationship in the eight Southeast Asian countries (Indonesia, Malaysia, China, South Korea, Vietnam, Singapore, Thailand, and Japan) in short and long term. Also, the existing literature offers mixed results on the impact of renewable energy on CO₂ emissions. Some studies show a positive effect, while others suggest a negative or non-linear one. There is a need for further investigation in the context of the eight Southeast Asian countries listed above. Furthermore, the interplay between factors like economic growth, trade openness, and forest areas, alongside energy consumption and CO₂ emissions, has not been extensively explored in the these countries.

This work can potentially bridge this gap by analyzing the link between energy consumption (both renewable and non-renewable) and CO₂ emissions in eight Southeast Asian countries. It could provide insights into the effectiveness of renewable energy adoption in mitigating CO₂ emissions in the region, taking into account potential challenges like technological limitations and policy barriers.

By investigating the influence of economic growth, trade openness, and forest areas, this study could offer a more comprehensive understanding of the factors driving CO₂ emissions in the eight Southeast Asian countries. This research might inform the development of targeted policy recommendations that promote sustainable energy development and address the specific challenges faced by the eight Southeast Asian countries in the current global energy landscape.

3. Methodology

This paper’s goal is to examine the connections between economic development, green spaces, renewable and non-renewable energy sources, and the environment in eight chosen Southeast Asian nations (Indonesia, Malaysia, China, South Korea, Vietnam, Singapore, Thailand, and Japan). Utilizing yearly data spanning from 1990 to 2023, we conducted our study through simultaneous equations. First, we establish econometric models. Next, we attempt to develop the Vector Error Correction Model (VECM) and Autoregressive Distributive Lag (ARDL). Step three presents the estimations and test results.

Both the VECM methodology and the ARDL approach are effective time series econometric methods that are appropriate for our study for the following reasons: Managing both short-term and long-term dynamics is the first explanation. In actuality, both ARDL and VECM are capable of concurrently estimating the short- and long-term relationships between variables. This is important in our situation since our research focuses on elements like CO₂ emissions, GDP, financial development, forest areas, renewable energy, non-renewable energy consumption, and trade openness in eight Southeast Asian Countries, i.e., the variables most likely have a long-term equilibrium connection, which is known as cointegration. Dealing with cointegration is the second factor. VECM explicitly models cointegration, but ARDL can effectively identify and manage it. This guarantees meaningful and statistically sound outcomes. The adapting of heterogeneous data is the third justification. Thus, a rather lengthy time series (1990–2022) for five distinct nations is included in the research. Heterogeneous data with possible structural breakdowns or variations in country-specific dynamics can be accommodated by both ARDL and VECM. This enables us to develop findings that are both generalizable and reflect the subtleties of unique countries. The fourth justification for using ARDL and VECM is to deal with possible endogeneity. Concerns about food security may really have an impact on certain factors, such as government assistance, leading to endogeneity problems. Using strategies like instrumental variables and lag limitations, VECM excels at managing endogeneity. In conclusion, each of the two methods offers unique benefits. For instance, the ARDL is more adaptable when managing mixed orders of integration (I(0) and I(1) variables), easier to use and understand for the sample size (ARDL often performs better with smaller samples), and appropriate for testing certain long-term hypotheses. Nevertheless, the long-term cointegrating interactions are explicitly modeled by the VECM; it is more adaptable to complicated dynamics involving several variables and strong in managing endogeneity problems.

Our study, while employing the widely used ARDL and VECM models, offers several unique contributions to the existing literature. By focusing on eight Southeast Asian countries, we provide region-specific insights into the challenges and opportunities related to climate change and energy transition. Our comprehensive analysis, encompassing economic, environmental, and energy-related factors, offers a holistic understanding of the complex relationships between these variables and CO₂ emissions. Additionally, our study provides practical policy implications for Southeast Asian countries, emphasizing the importance of renewable energy adoption, forest conservation, and climate-friendly policies. By employing robust econometric techniques, we provide empirical evidence to support our findings and policy recommendations.

The three general models of equations approved in this research are expressed as follows:

$$\mathrm{M}\mathrm{o}\mathrm{d}\mathrm{e}\mathrm{l}1:{\mathrm{F}}_{\mathrm{C}\mathrm{O}2\mathrm{E}}(\mathrm{G}\mathrm{D}\mathrm{P},\mathrm{F}\mathrm{D},\mathrm{F}\mathrm{A},\mathrm{N}\mathrm{R}\mathrm{E},\mathrm{R}\mathrm{E},\mathrm{T}\mathrm{O}).$$

In the first model, CO₂E was expressed as the dependent variable, while the remaining variables are exogenous.

$$\mathrm{M}\mathrm{o}\mathrm{d}\mathrm{e}\mathrm{l}2:{\mathrm{F}}_{\mathrm{R}\mathrm{E}}(\mathrm{G}\mathrm{D}\mathrm{P},\mathrm{C}\mathrm{O}2\mathrm{E},\mathrm{F}\mathrm{D},\mathrm{F}\mathrm{A},\mathrm{N}\mathrm{R}\mathrm{E},\mathrm{T}\mathrm{O}).$$

In the second model, RE was expressed as the dependent variable, while the remaining variables are exogenous.

$$\mathrm{M}\mathrm{o}\mathrm{d}\mathrm{e}\mathrm{l}3:{\mathrm{F}}_{\mathrm{N}\mathrm{R}\mathrm{E}}(\mathrm{G}\mathrm{D}\mathrm{P},\mathrm{C}\mathrm{O}2\mathrm{E},\mathrm{F}\mathrm{D},\mathrm{F}\mathrm{A},\mathrm{R}\mathrm{E},\mathrm{T}\mathrm{O})$$

In the third model, NRE were expressed as the dependent variable, while the remaining variables are exogenous.

CO2E, GDP, FD, FA, NRE, RE and TO designed, respectively, CO₂ emissions (expressed by CO₂ emissions from transport measured by the percentage of total fuel combustion), growth domestic product (annual rate), financial development, forest areas (expressed by the percentage of land area), non-renewable energy consumption (expressed by fossil fuel energy consumption measured by the percentage of total energy consumption), renewable energy (expressed by percentage of total final energy consumption), and trade openness. Every piece of information was taken from Word Data Bank [6].

Once our variables are included, the econometric models look like this:

$$ln{CO2E}_{it}={\beta }_0+{\beta }_1{lnGDP}_{it}+{\beta }_2{lnFD}_{it}+{\beta }_3{lnFA}_{it}+{\beta }_4{lnNRE}_{it}+{\beta }_5{lnRE}_{it}+{\beta }_6{lnTO}_{it}+{\epsilon }_{it}$$

(1)

$$ln{RE}_{it}={\beta }_0+{\beta }_1{lnGDP}_{it}+{\beta }_2{lnCO2E}_{it}+{\beta }_3{lnFD}_{it}+{\beta }_4{lnFA}_{it}+{\beta }_5{lnNRE}_{it}+{\beta }_6{lnTO}_{it}+{\epsilon }_{it}$$

(2)

$$ln{NRE}_{it}={\beta }_0+{\beta }_1{lnGDP}_{it}+{\beta }_2{lnCO2E}_{it}+{\beta }_3{lnFD}_{it}+{\beta }_4{lnFA}_{it}+{\beta }_5{lnRE}_{it}+{\beta }_6{lnTO}_{it}+{\epsilon }_{it}$$

(3)

where:

${\epsilon }_{it}$: White-noise disturbance term.

The logarithmic functions for CO₂ emissions are represented by lnCO2E, GDP by lnGDP, FD by lnFD, FA by lnFA, NRE by lnNRE, RE by lnRE, and TO by lnTO.

When a long-term cointegration state is present, the Auto-Regressive Distributed Lag (ARDL) models are stated as follows:

$$\begin{array}{ll}{DlnCO2E}_t={\beta }_0& +{\displaystyle \sum _{i=1}^{p_1}}{\gamma }_{1i}{DlnCO2E}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\delta }_{1i}{DlnGDP}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\theta }_{1i}{DlnFD}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\vartheta }_{1i}{DlnFA}_{t-i}\\ & +{\displaystyle \sum _{i=1}^{q_1}}{\mu }_{1i}{DlnNRE}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\rho }_{1i}{DlnRE}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\tau }_{1i}{DlnTO}_{t-i}+{\beta }_{11}{lnCO2E}_{t-1}+{\beta }_{12}{lnGDP}_{t-1}\\ & +{\beta }_{13}{lnFD}_{t-1}+{\beta }_{14}{lnFA}_{t-1}+{\beta }_{15}{lnNRE}_{t-1}+{\beta }_{16}{lnRE}_{t-1}+{\beta }_{17}{lnTO}_{t-1}+{\epsilon }_{1t}\end{array}$$

(4)

$$\begin{array}{ll}{DlnRE}_t={\beta }_0& +{\displaystyle \sum _{i=1}^{p_1}}{\gamma }_{1i}{DlnRE}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\delta }_{1i}{DlnGDP}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\theta }_{1i}{DlnCO2E}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\vartheta }_{1i}{DlnFD}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\mu }_{1i}{DlnFA}_{t-i}\\ & +{\displaystyle \sum _{i=1}^{q_1}}{\rho }_{1i}{DlnNRE}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\tau }_{1i}{DlnTO}_{t-i}+{\beta }_{11}{lnRE}_{t-1}+{\beta }_{12}{lnGDP}_{t-1}+{\beta }_{13}{lnCO2E}_{t-1}\\ & +{\beta }_{14}{lnFD}_{t-1}+{\beta }_{15}{lnFA}_{t-1}+{\beta }_{16}{lnNRE}_{t-1}+{\beta }_{17}{lnTO}_{t-1}+{\epsilon }_{1t}\end{array}$$

(5)

$$\begin{array}{ll}{DlnNRE}_t={\beta }_0& +{\displaystyle \sum _{i=1}^{p_1}}{\gamma }_{1i}{DlnNRE}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\delta }_{1i}{DlnGDP}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\theta }_{1i}{DlnCO2E}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\vartheta }_{1i}{DlnFD}_{t-i}\\ & +{\displaystyle \sum _{i=1}^{q_1}}{\mu }_{1i}{DlnFA}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\rho }_{1i}{DlnRE}_{t-i}+{\displaystyle \sum _{i=1}^{q_1}}{\tau }_{1i}{DlnTO}_{t-i}+{\beta }_{11}{lnNRE}_{t-1}+{\beta }_{12}{lnGDP}_{t-1}\\ & +{\beta }_{13}{lnCO2E}_{t-1}+{\beta }_{14}{lnFD}_{t-1}+{\beta }_{15}{lnFA}_{t-1}+{\beta }_{16}{lnRE}_{t-1}+{\beta }_{17}{lnTO}_{t-1}+{\epsilon }_{1t}\end{array}$$

(6)

D stands for the first-difference operator, and γ, δ, θ, ϑ, μ, ρ, and τ indicate the error correction dynamics. β₁ to β₇ show the long-run connection between the variables in the model, while β₀ represents the constants. The optimal lags and the ideal delays are represented by p and q, and ε_t is a white-noise disturbance term.

The Wald test (F-statistic) was used by the ARDL model to determine if the variables in question had a long-term association. In order to perform the Wald test, the estimated long-term neonatal death rate coefficients must be constrained using the different model variables. Rejecting the null hypothesis of non-cointegration over the long term, the resulting F-statistic value shows significance at the 10% level. The following is the formulation of the alternative hypothesis (H1) and null hypothesis (H0):

H0: 

β₁ = β₂ = β₃ = β₄ = β₅ = β₆ = β₇ = 0 (There is no long-term relationship).

H1: 

β₁ ≠ β₂ ≠ β₃ ≠ β₄ ≠ β₅ ≠ β₆ ≠ β₇ ≠ 0 (There is no long-term relationship).

According to [18] the autoregressive distributive lag (ARDL) model is utilized to assess cointegration. We estimate five unconstrained error corrections, taking each variable as a dependent variable, using the ARDL bounds testing technique to cointegration provided by [18,19]. Ref. [20] argues that the ARDL technique produces superior findings for small sample datasets when compared to classic approaches to cointegration, such as [21,22,23]. Furthermore, within a general-to-specific specification framework, the unrestrained model of the error correction model (ECM) appears to describe the data generation process with appropriate delays [24,25,26].

By identifying stationary or non-stationary processes through the development of bands of critical values, the ARDL technique obviates the requirement to classify variables as I(1) or I(0). The ARDL model offers a way to analyze a long-term connection regardless of whether the underlying variables are entirely I(0) or I(1), or even fractionally integrated, in contrast to previous cointegration approaches like Johansen’s process. As a result, there is no longer a need for the variables’ prior unit root testing. Moreover, the ARDL approach can discriminate between dependent and explanatory factors, whereas conventional cointegration approaches could have endogeneity issues. Because they eliminate issues that may come from serial correlation and endogeneity, estimates derived from the ARDL technique of cointegration analysis are therefore unbiased and efficient.

It is important to remember that unlike Johansen’s VECM, the ARDL technique allows for unequal lag orders. Ref. [27] however, assert that a suitable adjustment to the ARDL model’s ordering is sufficient to address both the endogenous variable issue and residual serial correlation at the same time.

The Granger causality technique, which is based on the Vector Error Correction Model (VECM), is used to examine the direction of causation between the variables after the presence of cointegration among them has been established. A restricted variant of the Vector Autoregression (VAR) model is called the VECM. The cointegration term that allows the endogenous variables to converge to their long-term connection while accounting for short-period dynamics is represented by the Error Correction Term (ECT) in the VECM. While the ECT offers proof of the existence of long-term relationships, the VECM allows one to estimate the short-term relationships between variables.

The Vector Error Correction Model will be used to perform the Granger-causality tests (VECM). The following is the formulation of the VECM equations:

$$\begin{array}{ll}{DlnCO2E}_t={\beta }_0& +{\displaystyle \sum _{i=1}^{\alpha 1}}{\alpha }_{1i}{DlnCO2E}_{t-i}+{\displaystyle \sum _{i=1}^{\gamma 1}}{\gamma }_{1i}{DlnGDP}_{t-i}+{\displaystyle \sum _{i=1}^{\delta 1}}{\delta }_{1i}{DlnFD}_{t-i}+{\displaystyle \sum _{i=1}^{\theta 1}}{\theta }_{1i}{DlnFA}_{t-i}\\ & +{\displaystyle \sum _{i=1}^{\vartheta 1}}{\vartheta }_{1i}{DlnNRE}_{t-i}+{\displaystyle \sum _{i=1}^{\mu 1}}{\mu }_{1i}{DlnRE}_{t-i}+{\displaystyle \sum _{i=1}^{\pi 1}}{\pi }_{1i}{DlnTO}_{t-i}+{\phi }_1{ECT}_{t-1}+{\epsilon }_{1t}\end{array}$$

(7)

$$\begin{array}{ll}{DlnRE}_t={\beta }_0& +{\displaystyle \sum _{i=1}^{\alpha 1}}{\alpha }_{1i}{DlnRE}_{t-i}+{\displaystyle \sum _{i=1}^{\gamma 1}}{\gamma }_{1i}{DlnGDP}_{t-i}+{\displaystyle \sum _{i=1}^{\delta 1}}{\delta }_{1i}{DlnCO2E}_{t-i}+{\displaystyle \sum _{i=1}^{\theta 1}}{\theta }_{1i}{DlnFD}_{t-i}+{\displaystyle \sum _{i=1}^{\vartheta 1}}{\vartheta }_{1i}{DlnFA}_{t-i}\\ & +{\displaystyle \sum _{i=1}^{\mu 1}}{\mu }_{1i}{DlnNRE}_{t-i}+{\displaystyle \sum _{i=1}^{\pi 1}}{\pi }_{1i}{DlnTO}_{t-i}+{\phi }_1{ECT}_{t-1}+{\epsilon }_{1t}\end{array}$$

(8)

$$\begin{array}{ll}{DlnNRE}_t={\beta }_0& +{\displaystyle \sum _{i=1}^{\alpha 1}}{\alpha }_{1i}{DlnNRE}_{t-i}+{\displaystyle \sum _{i=1}^{\gamma 1}}{\gamma }_{1i}{DlnGDP}_{t-i}+{\displaystyle \sum _{i=1}^{\delta 1}}{\delta }_{1i}{DlnCO2E}_{t-i}+{\displaystyle \sum _{i=1}^{\theta 1}}{\theta }_{1i}{DlnFD}_{t-i}\\ & +{\displaystyle \sum _{i=1}^{\vartheta 1}}{\vartheta }_{1i}{DlnFA}_{t-i}+{\displaystyle \sum _{i=1}^{\mu 1}}{\mu }_{1i}{DlnRE}_{t-i}+{\displaystyle \sum _{i=1}^{\pi 1}}{\pi }_{1i}{DlnTO}_{t-i}+{\phi }_1{ECT}_{t-1}+{\epsilon }_{1t}\end{array}$$

(9)

The constant in the VECM equations is denoted by β₀, and the coefficients that need to be calculated are α, γ, δ, θ, ϑ, μ, and π. The long-term equilibrium connection between the variables is represented by the error correction term (ECT), whereas ${\epsilon }_t$ is the white noise term.

4. Results

Our study’s initial step is to use the ADF and DF-GLS unit root tests to determine each variable’s stationarity order before looking at the link between them. The null hypothesis (H0) presupposes the presence of a unit root. We may continue with our econometric study if H0 is rejected. As all of the variables in

Table 1. Unit root test results.

ADF Test									PP Test
	Indonisia	Malaysa	Singapore	Thailand	Vietnam	China	Japan	South Korea	Indonisia	Malaysa	Singapore	Thailand	Vietnam	China	Japan	South Korea
	Case 1: Model with only constant term (level form)
lnGDP	−6.831 **	−0.554	2.110	0.324	1.021	1.361	0.331	1.221	−6.489 **	−1.432	1.521	1.245	2.242	0.546	−1.112	−1.332
lnCO2E	−4.114 *	2.232	1.002	0.112	2.212	2.132	0.700	−0.121	−12.549 **	−0.741	1.354	−1.562	−1.892	2.278	4.933	−3.628
lnFD	3.372	3.121	2.774	0.908	2.103	−0.121	0.426	3.112	−2.208 **	−1.357	1.223	−0.092	2.827	−0.372	0.003	0.234
lnFA	1.905	−1.998	1.008	0.131	0.423	−0.781	2.111	2.333	1.730	0.257	0.453	−2.562	3.132	1.093	0.738	3.124
lnNRE	−0.376	2.332	2.080	2.212	1.001	1.225	1.441	1.002	1.135	2.866	3.198	3.112	0.920	−1.267	2.425	2.982
lnRE	−0.985	−3.242 ***	−2.221	1.415	0.760	2.010	2.332	0.215	0.039	−2.229 **	−1.512	1.452	0.332	2.662	3.460	1.324
lnTO	1.102	1.973	2.773	1.023	0.807	2.053	1.354	2.113	1.713	0.527	−2.211	1.183	1.256	4.483	2.989	0.019
	Case 2: Model with constant and trend terms (level form)
lnGDP	−0.442	−1.445	−1.334	0.254	0.909	−0.353	−0.784	1.883	−1.435	−0.664	7.667	0.982	4.452	0.563	−1.737	−1.998
lnCO2E	−1.673	−0.672	−0.667	2.552	−0.343	−1.663	3.353	−0.425	−0.839	−1.272	1.627	−0.353	−1.849	1.998	−2.647	−2.562
lnFD	−4.583	1.049	1.271	−2.762	−0.221	−0.773	2.627	−1.356	−0.132	0.372	2.526	−1.036	1.353	−2.838	1.368	1.483
lnFA	2.829	−0.547	1.908	−0.110	1.772	−0.991	2.009	−1.252	2.490	−1.536	−0.637	−3.123	2.883	1.879	3.039	1.253
lnNRE	0.476	0.672	3.132	1.342	2.563	−0.330	2.467	5.546	−1.467	−2.009	2.039	1.245	3.737	1.024	3.356	1.829
lnRE	0.772	0.950	4.233	1.229	−0.093	−0.463	−0.552	2.778	−0.443	4.992	−3.536	0.840	2.352	1.342	1.009	3.552
lnTO	0.119	2.990	0.870	0.118	−1.564	2.537	−0.110	−1.637	6.425	2.434	−1.098	0.808	4.799	2.245	0.562	0.417
	Case 3: Model with only constant term (first difference)
lnGDP	−3.635 **	−3.667 ***	−3.562 ***	−4.565 *	−3.922 **	−3.772 **	−5.552 ***	−4.982 *	−5.002 ***	−4.722 ***	−5.827 ***	−3.663 ***	−4.562 ***	−5.261 ***	−4.252 ***	−4.916 **
lnCO2E	−2.193 **	−4.009 ***	−2.662 ***	−2.262 **	−2.762 **	−4.820 ***	−4.929 ***	−4.002 **	−3.722 ***	−6.992 ***	−3.829 ***	−4.118 ***	−3.134 ***	−2.712 ***	−6.676 ***	−2.324 ***
lnFD	−2.882 ***	−5.920 ***	−2.882 ***	−3.121 ***	−5.662 **	−5.552 ***	−2.902 ***	−2.772 ***	−4.947 ***	−6.012 ***	−2.324 ***	−4.782 **	−2.010 ***	−3.908 ***	−4.552 **	−1.002 ***
lnFA	−5.928 ***	−2.908 ***	−7.920 ***	−4.881 ***	−2.098 ***	−3.982 ***	−4.772 **	−4.820 ***	−3.826 ***	−3.920 ***	−4.256 ***	−6.929 ***	−5.008 ***	−4.134 ***	−3.119 *	−2.414 ***
lnNRE	−4.021 ***	−3.221 ***	−2.020 ***	−2.189 **	−3.621 **	−4.023 ***	−3.282 ***	−3.152 ***	−7.021 **	−4.112 ***	−3.908 **	−3.920 ***	−4.562 ***	−3.191 ***	−3.672 ***	−4.908 ***
lnRE	−2.602 *	−4.241 ***	−3.332 ***	−5.901 ***	−4.332 **	−8.029 **	−5.002 ***	−2.932 ***	−6.932 ***	−4.223 ***	−2.992 ***	−2.131 ***	−3.190 ***	−4.892 **	−2.782 ***	−3.197 ***
lnTO	−1.991 ***	−6.881 **	−5.556 ***	−6.098 ***	−5.992 *	−4.728 ***	−6.025 ***	−6.114 ***	−4.832 *	−6.920 ***	−7.024 ***	−5.092 ***	−7.901 ***	−0.5007 ***	−4.441 **	−5.476 ***

*, **, and *** indicate the significance, respectively, at 10%, 5%, and 1%.

are stationary in first difference, the null hypothesis was not accepted in this instance.

The Bounds test is used in the second phase to find long-term cointegration between the variables. The Schwarz Criteria (SIC) and Akaike Information Criterion (AIC) are minimized in order to find the ideal lag, and the F-statistic is computed and compared with critical values. The F-statistic is over the 10% upper bound, according to the findings of the Bounds test (

Table 2. Bounds test results.

Dependent Variable	Indonesia		Malaysia		Singapore		Thailand		Vietnam		China		Japan		Korea
$F_{CO2E}(GDP,FD,FA,NRE,RE,TO)$	4.82		5.78		8.67		5.01		7.71		3.88		4.32		5.33
$F_{RE}(GDP,CO2E,FD,FA,NRE,TO)$	5.89		4.87		4.09		9.44		5.52		6.77		6.23		4.42
$F_{NRE}(GDP,CO2E,FD,FA,RE,TO)$	7.03		5.67		5.11		8.04		6.47		7.68		4.56		6.91
Significance Level	Critical Value Bounds
	I(0)	I(1)	I(0)	I(1)	I(0)	I(1)	I(0)	I(1)	I(0)	I(1)	I(0)	I(1)	I(0)	I(1)	I(0)	I(1)
10%	2.01	3.28	1.39	2.37	3.01	3.66	2.21	3.65	2.53	3.26	2.15	3.12	3.04	3.64	2.23	3.22
5%	3.21	4.22	2.35	3.81	3.83	4.73	3.92	4.93	3.88	4.07	3.46	4.17	3.82	4.13	3.15	3.87
1%	4.32	4.73	3.95	4.96	4.98	5.69	5.33	6.81	4.53	5.23	4.98	5.25	4.43	5.11	4.08	5.91

), indicating the existence of long-term cointegration between the variables.

To find residual correlation, we ran the Breusch–Godfrey serial correlation LM test, and the results are shown in

Table 3. Diagnostic test.

Country	Dependent Variable	LM Test	ARCH Test	Reset Test	JB Test
Indonesia	$F_{CO2E}(GDP,FD,FA,NRE,RE,TO)$	0.026	0.133	0.026	0.703
	$F_{RE}(GDP,CO2E,FD,FA,NRE,TO)$	0.048	0.764	0.547	0.625
	$F_{NRE}(GDP,CO2E,FD,FA,RE,TO)$	0.062	0.147	0.535	0.176
Malaysia	$F_{CO2E}(GDP,FD,FA,NRE,RE,TO)$	0.142	0.526	0.660	0.215
	$F_{RE}(GDP,CO2E,FD,FA,NRE,TO)$	0.296	0.331	0.023	0.252
	$F_{NRE}(GDP,CO2E,FD,FA,RE,TO)$	0.424	0.516	0.240	0.435
Singapore	$F_{CO2E}(GDP,FD,FA,NRE,RE,TO)$	0.143	0.532	0.532	0.562
	$F_{PS}(GDP,REC,NREC,FD,T,TR)$	0.042	0.654	0.526	0.452
	$F_{NRE}(GDP,CO2E,FD,FA,RE,TO)$	0.234	0.153	0.982	0.562
Thailand	$F_{CO2E}(GDP,FD,FA,NRE,RE,TO)$	0.213	0.212	0.102	0.010
	$F_{RE}(GDP,CO2E,FD,FA,NRE,TO)$	0.410	0.057	0.634	0.375
	$F_{NRE}(GDP,CO2E,FD,FA,RE,TO)$	0.343	0.972	0.849	0.942
Vietnam	$F_{CO2E}(GDP,FD,FA,NRE,RE,TO)$	0.802	0.834	0.028	0.843
	$F_{RE}(GDP,CO2E,FD,FA,NRE,TO)$	0.356	0.759	0.394	0.437
	$F_{NRE}(GDP,CO2E,FD,FA,RE,TO)$	0.994	0.202	0.870	0.798
China	$F_{CO2E}(GDP,FD,FA,NRE,RE,TO)$	0.545	0.981	0.058	0.902
	$F_{RE}(GDP,CO2E,FD,FA,NRE,TO)$	0.254	0.825	0.232	0.120
	$F_{NRE}(GDP,CO2E,FD,FA,RE,TO)$	0.950	0.209	0.600	0.073
Japan	$F_{CO2E}(GDP,FD,FA,NRE,RE,TO)$	0.654	0.420	0.978	0.608
	$F_{RE}(GDP,CO2E,FD,FA,NRE,TO)$	0.492	0.072	0.434	0.052
	$F_{NRE}(GDP,CO2E,FD,FA,RE,TO)$	0.067	0.749	0.040	0.699
South Korea	$F_{CO2E}(GDP,FD,FA,NRE,RE,TO)$	0.130	0.327	0.740	0.904
	$F_{RE}(GDP,CO2E,FD,FA,NRE,TO)$	0.023	0.705	0.352	0.637
	$F_{NRE}(GDP,CO2E,FD,FA,RE,TO)$	0.747	0.422	0.930	0.525

. Both models revealed no indication of serial association. Furthermore, we tested for heteroscedasticity using the ARCH test, and the results showed that all four models had homoscedastic error terms with a normal distribution.

We used the CUSUM and CUSUMSQ tests to make sure the ARDL model was stable. The outcomes are displayed in Figure 1.

5. Discussion

Table 4. Long-run ARDL coefficients.

Country	Independent Variables	CO2E	RE	NREC
Indonesia	LnGDP	0.028 (0.000) ***	0.110 (0.347)	0.017 (0.337)
	LnCO2E	…	27.993 (0.000) ***	0.773 (0.098) *
	LnFD	−0.332 (0.000) ***	−7.436 (0.016) **	0.249 (0.299)
	LnFA	−2.201 (0.000) ***	4.711 (0.757)	−3.247 (0.067) *
	LnNRE	0.191 (0.587)	16.70 (0.039) **	…
	LnRE	−0.19 (0.000) ***	…	−0.024 (0.175)
	LnTO	−0.055 (0.421)	10.832 (0.000) ***	−1.018 (0.020) **
	Constant	22.757 (0.000) ***	120.867 (0.856)	5.005 (0.266)
	CUSUM	Stable	Stable	Stable
	CUSUMSQ	Stable	Unstable	Unstable
Malaysa	LnGDP	0.429 (0.013) **	0.068 (0.364)	0.788 (0.539)
	LnCO2E	…	11.156 (0.100)	0.104 (0.000)
	LnFD	−0.053 (0.161)	0.182 (0.857)	0.002 (0.877)
	LnFA	−2.060 (0.061) *	94.908 (0.007) ***	0.814 (0.000)
	LnNRE	6.768 (0.000) ***	70.847 (0.062) *	…
	LnRE	−0.277 (0.004) ***	…	−6.540 (0.980)
	LnTO	1.236 (0.000) ***	−16.506 (0.023) **	−0.006 (0.718)
	Constant	78.946 (0.007) ***	−1196.400 (0.002) ***	−0.735 (0.609)
	CUSUM	Stable	Stable	Stable
	CUSUMSQ	Unstable	Unstable	Unstable
Singapore	LnGDP	0.039 (0.1038)	0.023 (0.008) ***	−0.001 (0.047) *
	LnCO2E	…	0.565 (0.080) *	0.005 (0.888)
	LnFD	0.961 (0.003) ***	−0.772 (0.038) *	−0.146 (0.000) ***
	LnFA	0.793 (0.281)	0.584 (0.463)	0.234 (0.000) ***
	LnNRE	1.606 (0.031) *	4.707 (0.042) *	…
	LnRE	−0.331 (0.039) *	…	−0.041 (0.118)
	LnTO	−0.692 (0.298)	−0.365 (0.258)	−0.077 (0.012) *
	Constant	4.070 (0.242)	20.075 (0.106)	4.930 (0.000) ***
	CUSUM	Unstable	Stable	Stable
	CUSUMSQ	Unstable	Unstable	Unstable
Thailand	LnGDP	0.008 (0.158)	0.0412 (0.576)	0.0037 (0.007) ***
	LnCO2E	…	14.148 (0.000) ***	0.0270 (0.638)
	LnFD	0.357 (0.077) *	13.489 (0.000) ***	−0.431 (0.938)
	LnFA	−8.339 (0.000) ***	132.423 (0.000) ***	1.745 (0.001) ***
	LnNRE	0.370 (0.396)	−12.009 (0.124)	…
	LnRE	−0.044 (0.000) ***	…	−0.010 (0.001) ***
	LnTO	0.0676 (0.607)	1.0249 (0.658)	−0.112 (0.000) ***
	Constant	−10.366 (0.000) ***	−303.959 (0.000) ***	10.268 (0.000) ***
	CUSUM	Stable	Stable	Stable
	CUSUMSQ	Unstable	Unstable	Unstable
Vietnam	LnGDP	0.016 (0.083) *	1.0744 (0.001) ***	0.022 (0.149)
	LnCO2E	…	20.506 (0.000) ***	0.642 (0.048) **
	LnFD	−0.052 (0.537)	−5.908 (0.007) ***	−0.095 (0.491)
	LnFA	−0.113 (0.862)	24.578 (0.159)	2.720 (0.014) *
	LnNRE	0.496 (0.011) **	3.000 (0.561)	…
	LnRE	−0.018 (0.004) ***	…	−0.003 (0.717)
	LnTO	−0.064 (0.370)	−1.023 (0.592)	−0.214 (0.051) *
	Constant	7.269 (0.001) ***	214.540 (0.000) ***	−4.187 (0.018) **
	CUSUM	Stable	Stable	Unstable
	CUSUMSQ	Unstable	Unstable	Unstable
China	LnGDP	9.924 (0.043) **	1.783 (0.016) **	2.245 (0.088) *
	LnCO2E	…	3.003 (0.029) **	0.064 (0.043) **
	LnFD	0.740 (0.317)	0.042 (0.353)	0.686 (0.080) *
	LnFA	−3.522 (0.046) **	0.883 (0.740)	−3.080 (0.000) ***
	LnNRE	39.627 (0.027) **	−0.223 (0.048) **	…
	LnRE	−0.047 (0.009) ***	…	−0.484 (0.399)
	LnTO	3.320 (0.015) **	2.077 (0.059) *	0.384 (0.001) ***
	Constant	0.963 (0.581)	22.038 (0.865)	10.751 (0.000) ***
	CUSUM	Stable	Stable	Stable
	CUSUMSQ	Unstable	Unstable	Unstable
Japan	LnGDP	7.841 (0.022) **	0.136(0.000)	0.251 (0.059) *
	LnCO2E	…	−14.052(0.000)	0.731 (0.090) *
	LnFD	−0.973 (0.054) *	−8.270(0.000)	0.274 (0.291)
	LnFA	−70.545 (0.031) **	−6.884(0.000)	51.190 (0.008) ***
	LnNRE	0.263 (0.002) ***	1.164(0.591)	…
	LnRE	−0.720 (0.078) *	…	−0.018 (0.013) **
	LnTO	0.461 (0.657)	0.881(0.187)	0.0419 (0.688)
	Constant	316.086(0.841)	2361.228(0.000)	−86.778 (0.019) **
	CUSUM	Stable	Stable	Stable
	CUSUMSQ	Unstable	Unstable	Unstable
South Korea	LnGDP	0.048 (0.195)	0.514 (0.549)	0.234 (0.358)
	LnCO2E	…	−0.627 (0.649)	0.290 (0.014) **
	LnFD	−0.129 (0.423)	−0.654 (0.016) **	0.027 (0.611)
	LnFA	−28.346 (0.000) ***	−277.650 (0.000) ***	−8.405 (0.011) **
	LnNRE	2.641 (0.082) *	6.882 (0.000) ***	…
	LnRE	−0.248 (0.001) ***	…	−0.057 (0.042) **
	LnTO	0.614 (0.240)	2.941 (0.004) ***	−0.085 (0.019) **
	Constant	26.034 (0.032) **	645.972 (0.000) ***	−18.581 (0.112)
	CUSUM	Stable	Stable	Stable
	CUSUMSQ	Unstable	Unstable	Unstable

*, **, and *** indicate the significance, respectively, at 10%, 5%, and 1%.

displays the outcomes of the ARDL estimations for the three distinct models and for eight nations in Southeast Asia.

A comparison analysis was conducted for the first model, which included CO2E as the dependent variable, in order to look at the variables impacting CO₂ emissions in various nations. The findings indicated that different factors affected CO₂ emissions differently in each nation. In Indonesia, CO₂ emissions were positively impacted by GDP and NRE and negatively by FD, FA, RE, and TO. Similar outcomes were shown in Malaysia, where there were negative impacts from FA and RE and favorable effects from GDP, NRE, and TO. GDP, FD, FA, and NRE all had positive effects on CO₂ emissions in Singapore; however, RE and TO had negative effects. GDP, NRE, FD, and TO had positive effects on CO₂ emissions in Thailand; FA and RE had negative effects. In Vietnam, CO₂ emissions were positively impacted by GDP and NRE and negatively by FD, FA, RE, and To. China’s CO₂ emissions were positively impacted by GDP, FD, NRE, and TO but negatively impacted by FA and RE. GDP, NRE, and TO had a favorable impact on Japan’s CO₂ emissions; FD, FA, and RE had a negative impact. South Korea exhibited positive influences of GDP, NRE, and TO and negative impacts of FD, FA, and RE on CO₂ emissions.

The findings for the second model, which uses renewable energy (RE) as the dependent variable, show the potential and difficulties that various nations have in the growth of RE. Indonesia has a sufficiently high GDP, CO₂, FA, NRE, and TO to boost RE; nevertheless, FD reduces the potential for RE. Malaysia has potential for RE development, as seen by its high GDP, CO₂, FD, FA, and NRE; nevertheless, the declining TO may impede progress. Although Singapore’s GDP, CO₂, FA, and NRE are all rising, its FD and TO may make the growth of RE more difficult. While GDP, CO₂, FA, and TO are rising in Thailand and Vietnam and offer potential for RE development, declining NRE and FD may impede advancement. China’s growing GDP, CO₂, FD, FA, and TO indicate the country’s potential for RE development; nevertheless, it is imperative to address the declining NRE. Japan’s GDP, NRE, and TO are all rising in terms of RE development; nevertheless, declining CO₂, FA, and FD may be a problem. While South Korea’s GDP and NRE are rising in support of RE development, declining CO₂, FD, and FA may be a problem. To create plans that would optimize the potential of renewable energy, policymakers must take all relevant elements into account.

For the third model (NRE as the dependent variable), the findings demonstrated the variables affecting the growth of renewable energy (RE) in different nations. While Indonesia’s GDP, CO₂, and FD are rising and indicate possibilities for RE development, the country’s declining FA, TO, and RE may impede development. Malaysia has growing GDP, CO₂, FD, and FA, which may provide opportunities for RE development; nevertheless, evolving RE and TO may provide difficulties. Although Singapore’s declining GDP, FD, RE, and TO can be problematic, the country’s rising CO₂ and FA may also present opportunities for RE development. Growing GDP, CO₂, and FA in Thailand and Vietnam offer potential for RE development. Progress may, however, be hampered by the declining FD, RE, and TO. China’s growing GDP, CO₂, FD, and TO may present opportunities for the growth of RE, but the country’s declining FA and RE may present difficulties. Although Japan’s GDP, CO₂, DF, FA, and TO are all rising and indicate potential for RE development, declining RE may provide a problem. Finally, there are prospects for RE development in South Korea because to its growing GDP, CO₂, and FD. The declining FA, RE, and TO can be problematic, though. In conclusion, developing requires a grasp of the variables influencing CO₂ emissions and RE development.

In

Table 5. Granger causality test results.

Causality Directions
	Short Term								Long Term
	Dep.Var.	DLnGDP	DLnCO2E	DLnFD	DLnFA	DLnNRE	DLnRE	DLnTO	ECT
Indonesia	LnGDP	----	1.833 (0.022) **	0.172 (0.930)	1.339 (0.736)	0.829 (0.078) *	1.738 (0.033) *	1.883 (0.079) *	−0.678 (0.074) *
	LnCO2E	3.022 (0.051) *	----	3.811 (0.039) **	3.812 (0.048) **	2.922 (0.069) *	0.236 (0.673)	1.014 (0.202)	0.147 (0.002)
	LnFD	0.736 (0.686)	0.993 (0.278)	----	1.928 (0.890)	1.028 (0.832)	0.952 (0.069) *	0.810 (0.919)	0.213 (0.098)
	LnFA	2.102 (0.054)	0.291 (0.726)	0.309 (0.830)	----	0.945 (0.927)	0.027 (0.840)	2.267 (0.156)	−12.071 (0.836)
	LnNRE	2.544 (0.003) ***	2.035 (0.007) ***	0.592 (0.542)	2.005 (0.057) *	----	0.518 (0.783)	1.167 (0.060) *	−0.039 (0.070) *
	LnRE	4.456 (0.032) ***	0.122 (0.603)	2.464 (0.088) *	2.618 (0.291)	1.031 (0.241)	----	2.135 (0.783)	5.892 (0.937)
	LnTO	0.036 (0.893)	1.697 (0.352)	3.902 (0.046) **	2.419 (0.797)	0.783 (0.091) *	1.940 (0.299)	----	−0.762 (0.132)
Malaysia	LnGDP	----	0.945 (0.099) *	1.442 (0.252)	0.335 (0.717)	2.226 (0.125)	1.665 (0.206)	0.161 (0.051) *	−0.075 (0.036) **
	LnCO2E	4.646 (0.017) **	----	1.139 (0.333)	3.248 (0.052) *	1.886 (0.069) *	1.358 (0.272)	3.835 (0.032) **	5.961 (0.002)
	LnFD	3.552 (0.041) **	0.405 (0.670)	----	0.625 (0.541)	0.686 (0.511)	0.546 (0.584)	0.697 (0.505)	0.179 (0.088)
	LnFA	3.341 (0.049) **	0.840 (0.041) *	0.282 (0.755)	----	0.613 (0.547)	5.020 (0.013) **	4.207 (0.024) **	−9.313 (2.453)
	LnNRE	3.961 (0.029) **	0.535 (0.590)	0.837 (0.442)	5.506 (0.009) ***	----	3.315 (0.050) **	1.490 (0.241)	0.003 (0.005)
	LnRE	4.821 (0.015) **	0.035 (0.965)	0.111 (0.894)	4.328 (0.022) **	0.049 (0.951)	----	2.413 (0.106)	−1.878 (0.050) **
	LnTO	2.455 (0.102)	0.154 (0.857)	0.353 (0.705)	0.351 (0.706)	0.273 (0.762)	0.830 (0.445)	----	−0.050 (0.038) **
Singapore	LnGDP	----	1.170 (0.324)	3.688 (0.037) **	0.142 (0.867)	0.093 (0.911)	0.136 (0.872)	2.415 (0.056) *	0.438 (0.298)
	LnCO2E	7.234 (0.002) ***	----	7.239 (0.002) ***	1.130 (0.336)	1.145 (0.331)	2.750 (0.080) *	0.450 (0.041) **	0.002 (0.003)
	LnFD	7.801 (0.001) ***	0.312 (0.031) **	----	1.257 (0.299)	1.720 (0.196)	1.619 (0.214)	1.640 (0.410) **	−0.015 (0.003) ***
	LnFA	3.419 (0.045) **	0.068 (0.033) **	0.780 (0.467)	----	3.899 (0.031) **	1.252 (0.300)	0.925 (0.407)	0.002 (0.001)
	LnNRE	2.530 (0.096) *	0.630 (0.539)	0.188 (0.829)	0.138 (0.871)	----	2.107 (0.139)	1.180 (0.321)	0.002 (0.000)
	LnRE	1.993 (0.153)	0.765 (0.473)	0.994 (0.381)	0.160 (0.852)	0.213 (0.034) **	----	1.034 (0.037) **	−0.009 (0.005) ***
	LnTO	0.549 (0.583)	1.157 (0.007) ***	0.135 (0.873)	0.011 (0.988)	0.673 (0.088) **	1.832 (0.323)	----	−0.004 (0.014) **
Thailand	LnGDP	----	0.210 (0.811)	2.208 (0.127)	8.831 (0.000) ***	0.812 (0.053)	0.172 (0.842)	0.553 (0.580)	0.030 (0.285)
	LnCO2E	1.032 (0.000) ***	----	0.217 (0.805)	1.004 (0.378)	0.120 (0.086) *	0.322 (0.882)	1.481 (0.894)	−0.000 (0.004) ***
	LnFD	0.882 (0.424)	0.981 (0.386)	----	8.645 (0.001) ***	4.492 (0.019) *	0.498 (0.612)	3.633 (0.038) **	0.012 (0.040)
	LnFA	0.586 (0.562)	0.845 (0.439)	5.499 (0.009) ***	----	2.389 (0.108)	1.740 (0.192)	0.890 (0.421)	−0.005 (0.000) ***
	LnNRE	1.213 (0.000) ***	0.066 (0.935)	0.045 (0.955)	2.060 (0.145)	----	0.003 (0.996)	0.415 (0.663)	0.006 (0.002)
	LnRE	4.845 (0.015) **	0.322 (0.726)	3.156 (0.057) *	1.738 (0.193)	1.065 (0.357)	----	1.222 (0.308)	0.030 (0.114)
	LnTO	5.397 (0.009) ***	1.481 (0.243)	1.975 (0.156)	0.985 (0.384)	0.548 (0.583)	0.672 (0.518)	----	0.002 (0.006)
Vietnam	LnGDP	----	1.737 (0.193)	0.469 (0.629)	4.585 (0.018) **	2.400 (0.107)	0.823 (0.448)	0.945 (0.399)	−0.3452 (0.376)
	LnCO2E	1.801 (0.182)	----	0.798 (0.459)	2.677 (0.085) *	4.950 (0.013) **	1.355 (0.273)	2.455 (0.102)	0.023 (0.019)
	LnFD	0.950 (0.398)	0.422 (0.659)	----	2.327 (0.0851) *	4.950 (0.0139) **	1.355 (0.273)	2.455 (0.102)	0.025 (0.038)
	LnFA	0.945 (0.399)	1.411 (0.259)	2.225 (0.125)	----	8.425 (0.001) ***	1.784 (0.185)	2.740 (0.080) *	−0.002 (0.004) ***
	LnNRE	2.629 (0.088) *	1.922 (0.163)	0.953 (0.396)	0.220 (0.803)	----	0.169 (0.845)	2.625 (0.088) *	0.038 (0.012)
	LnRE	1.349 (0.274)	4.451 (0.020) **	0.127 (0.880)	1.093 (0.348)	3.493 (0.043) **	----	1.138 (0.333)	−0.696 (0.688)
	LnTO	2.354 (0.112)	1.514 (0.236)	0.159 (0.853)	0.209 (0.812)	7.897 (0.001) ***	0.131 (0.877)	----	0.044 (0.056)
China	LnGDP	----	0.364 (0.007) ***	2.156 (0.133)	1.362 (0.271)	0.050 (0.050) **	0.216 (0.006) ***	0.333 (0.019) *	−0.043 (0.051) *
	LnCO2E	2.385 (0.019) **	----	6.943 (0.113)	2.688 (0.084) *	0.305 (0.038) **	3.508 (0.042) **	2.148 (0.134)	0.285 (0.000)
	LnFD	2.013 (0.151)	1.108 (0.343)	----	1.605 (0.217)	2.700 (0.083) *	1.664 (0.206)	4.729 (0.016) **	0.001 (0.001)
	LnFA	3.201 (0.024) *	9.288 (0.000) ***	4.832 (0.015) **	----	6.572 (0.004) ***	4.871 (0.014) **	2.4816 (0.00) ***	−5.681 (1.605)
	LnNRE	0.924 (0.054) *	0.983 (0.085) *	2.942 (0.068) *	0.312 (0.734)	----	3.227 (0.058) *	1.034 (0.367)	0.062 (0.000)
	LnRE	4.944 (0.013) **	0.066 (0.936)	7.228 (0.002) ***	30.24 (6.308)	0.231 (0.095) *	----	2.725 (0.081) *	−0.047 (0.049)
	LnTO	0.085 (0.918)	3.643 (0.038) **	3.723 (0.035) **	0.057 (0.944)	0.096 (0.908)	1.296 (0.288)	----	−0.006 (0.002) ***
Japan	LnGDP	----	0.789 (0.084) *	4.491 (0.019) **	2.036 (0.148)	3.385 (0.047) **	0.436 (0.650)	1.209 (0.012) **	0.334 (0.221)
	LnCO2E	3.325 (0.049) **	----	0.792 (0.461)	1.513 (0.236)	1.234 (0.005) ***	8.009 (0.001) ***	2.557 (0.094) *	0.001 (0.002)
	LnFD	1.935 (0.162)	1.482 (0.243)	----	1.378 (0.267)	1.586 (0.221)	3.199 (0.055) *	0.065 (0.936)	−0.004 (0.004) ***
	LnFA	0.998 (0.380)	0.446 (0.644)	1.453 (0.249)	----	2.508 (0.098) *	0.738 (0.486)	0.830 (0.445)	−1.405 (0.005)
	LnNRE	0.043 (0.057) *	1.864 (0.172)	0.318 (0.730)	1.057 (0.359)	----	4.661 (0.017) **	0.023 (0.976)	−0.123 (0.001) ***
	LnRE	1.651 (0.208)	6.472 (0.004) ***	1.502 (0.238)	3.388 (0.047) **	1.118 (0.040) **	----	0.554 (0.580)	0.040 (0.042)
	LnTO	6.156 (0.005) ***	3.823 (0.033) **	2.261 (0.121)	2.712 (0.082) *	1.523 (0.034) **	4.715 (0.016) **	----	0.022 (0.007)
South Korea	LnGDP	----	3.978 (0.029) **	0.458 (0.636)	0.029 (0.071) *	0.491 (0.616)	2.031 (0.148)	3.951 (0.030) *	−0.474 (0.044) **
	LnCO2E	16.371 (2.126)	----	3.904 (0.031) **	1.585 (0.001) ***	0.962 (0.093) *	7.409 (0.002) ***	1.288 (0.290)	−29.361 (5.663)
	LnFD	6.537 (0.004) ***	0.359 (0.701)	----	0.581 (0.565)	1.000 (0.079) *	3.732 (0.035) **	3.671 (0.037) *	−0.335 (0.014) **
	LnFA	3.161 (0.056) *	0.151 (0.059) *	0.311 (0.734)	----	1.391 (0.2642	8.529 (0.0012) ***	1.041 (0.3653	−0.004 (0.003) ***
	LnNRE	5.463 (0.009) ***	0.341 (0.713)	1.202 (0.314)	0.016 (0.983)	----	5.606 (0.008) ***	0.415 (0.663)	0.028 (0.138)
	LnRE	0.656 (0.526)	0.482 (0.021) **	0.237 (0.790)	0.539 (0.588)	1.761 (0.089) *	----	1.507 (0.237)	−0.769 (0.015) **
	LnTO	2.762 (0.079) *	1.476 (0.244)	0.153 (0.858)	1.848 (0.175)	0.344 (0.711)	5.396 (0.010) ***	----	0.302 (0.170)

*, **, and *** indicate the significance, respectively, at 10%, 5%, and 1%.

, the Granger causality analysis indicates that there are mixed correlations between the variables, both unidirectional and bidirectional. Regarding the links between the variables, inconsistent findings are also obtained via the VECM Granger causality test. The errors correction term (ECT) is used to determine the presence of long-term causal linkages. At least one long-term link between the variables is implied by a substantial and negative coefficient, with the endogenous variable serving as an essential correction factor when the model deviates from equilibrium. The cointegration tests indicated the presence of long-term correlations between the variables, which are confirmed by the coefficients of the ECT in Table 5. Furthermore, the outcomes show that the variables move together.

Figure 2 presents an overview of the causal linkages between factors for the eight instances from Southeast Asia. The diagrams are intended to illustrate the bidirectional links meaning mutual causality where factors influence each other, and the unidirectional links indicates a one-way causal relationship.

6. Policy Implication

The policy recommendations derived from this study are grounded in the specific findings regarding the drivers and mitigators of CO₂ emissions in the eight Southeast Asian countries analyzed. The alignment of these policies with the results ensures their relevance and practical applicability:

6.1. Promote Renewable Energy Investments Based on Emission Drivers

The findings highlight that GDP and non-renewable energy (NRE) consumption are key contributors to CO₂ emissions in countries like Japan, South Korea, and Malaysia. To address this, these nations should prioritize investments in renewable energy sources such as solar, wind, and hydropower. Policies such as tax incentives, subsidies, and public–private partnerships should be introduced to attract investments in renewable energy infrastructure. For Vietnam and Thailand, which also exhibit rising GDP and CO₂ emissions, strengthening regional collaboration through ASEAN to secure financing and facilitate clean energy technology transfer is crucial. This is particularly relevant given their growing potential for renewable energy expansion but existing financial and infrastructural challenges.

6.2. Expand Forest Conservation Efforts in High-Deforestation Nations

This study shows that forest areas significantly mitigate CO₂ emissions in countries such as Vietnam and Singapore, while high deforestation rates in Indonesia and Malaysia exacerbate emissions. To counteract this, strict land-use policies should be implemented in Indonesia and Malaysia to prevent illegal logging and land conversion. These measures should be supported by leveraging carbon credit markets and international funding mechanisms for reforestation and afforestation programs. Expanding forest cover in these nations will not only enhance carbon sequestration but will also contribute to biodiversity preservation.

6.3. Leverage Financial Development for Green Growth in Emerging Economies

The mixed impact of financial development on CO₂ emissions observed across countries suggests the need for targeted financial mechanisms. In Vietnam and China, where renewable energy is instrumental in reducing emissions, policymakers should introduce green financial instruments such as green bonds, sustainability-linked loans, and investment funds to finance clean energy projects. Strengthening financial regulations can encourage private sector participation, reducing risks associated with renewable energy investments. Aligning financial incentives with environmental outcomes, such as lower interest rates for projects with high emission reduction potential, can further support green growth.

6.4. Enhance Trade Openness to Facilitate Clean Technology Transfer

This study underscores the role of trade openness in reducing emissions in countries like Singapore, Vietnam, and Thailand, where access to advanced clean technologies plays a critical role. Governments should leverage trade openness as a channel to import renewable energy technologies. Negotiating trade agreements that embed environmental standards and facilitate the transfer of advanced clean energy technologies should be a priority for countries such as Japan, South Korea, and Singapore. These agreements can enable the region to overcome technological and infrastructural barriers in renewable energy adoption.

6.5. Strengthening Regional Collaboration for Unified Climate Action

The varying effects of renewable and non-renewable energy across the studied countries highlight the need for a coordinated regional approach. ASEAN should play an active role in aligning renewable energy targets, carbon reduction goals, and policy frameworks among its member states. Shared funding mechanisms for clean energy projects, joint research and development programs, and harmonized standards for renewable energy deployment can facilitate knowledge sharing and resource optimization. For example, countries like Indonesia and Malaysia can benefit from regional initiatives to address deforestation, while Vietnam and Thailand can leverage shared platforms to enhance renewable energy adoption.

By explicitly linking each policy recommendation to the findings, these tailored strategies aim to address the region’s specific challenges and opportunities in achieving sustainable energy development and reducing CO₂ emissions.

7. Conclusions

This study explored the dynamic relationships between CO₂ emissions, GDP, financial development, forest areas, renewable energy (RE), non-renewable energy (NRE) consumption, and trade openness in eight Southeast Asian countries (Indonesia, Malaysia, China, South Korea, Vietnam, Singapore, Thailand, and Japan) over the period 1990–2023. Using econometric techniques such as Granger causality and cointegration tests, the findings highlight that GDP and NRE consumption are primary drivers of CO₂ emissions in countries like Indonesia, Malaysia, China, Japan, and South Korea. In contrast, factors such as forest areas, renewable energy, and trade openness play significant roles in reducing emissions in Vietnam, Thailand, and Singapore. The impact of financial development on emissions remains mixed across countries, reflecting varying stages of economic development and financial integration. The results demonstrate that Southeast Asia faces both challenges and opportunities in transitioning to a sustainable energy framework. High-GDP countries with rising emissions, such as Indonesia and Malaysia, have significant potential for renewable energy expansion but are constrained by financial barriers and infrastructure gaps. Similarly, emerging economies like Vietnam and Thailand show promise due to their expanding GDP and forest areas but must address their dependence on NRE. Countries such as Japan, South Korea, and Singapore face more complex environmental and economic dynamics, requiring tailored interventions to support the shift to renewable energy and reduce emissions.

The accuracy and reliability of data, particularly for historical periods, can significantly impact the robustness of research findings. Inconsistent or missing data can introduce biases into the analysis. Ensuring the availability of high-quality and consistent data across all countries and time periods is crucial for rigorous analysis.

While ARDL and VECM models are powerful tools, they rely on specific assumptions like stationarity and cointegration. Violations of these assumptions can lead to biased and inconsistent estimates. Sensitivity analysis and robustness checks can help assess the reliability of the results. Establishing causality between variables can be challenging due to potential endogeneity issues, such as reverse causality or omitted variable bias. Advanced econometric techniques like instrumental variable or generalized method of moments (GMM) estimation can help address these challenges.

Overall, this study underscores the importance of adopting integrated, country-specific approaches to balance economic growth and environmental sustainability while addressing Southeast Asia’s diverse energy transition challenges.