Investigating the Impact of Energy Consumption and Economic Activities on CO₂ Emissions from Transport in Saudi Arabia

Abstract

This study examines the relationships between CO₂ emissions, gross domestic product (GDP), financial development, energy export, sustainable power, unsustainable power depletion, and commercial growth in the Kingdom of Saudi Arabia (KSA) from 1990 to 2022 by using the auto-regressive distributed lag (ARDL) approach and the vector error correction model (VECM) approach. In the first step, we have used tests such as the augmented Dickey–Fuller (ADF) test and the Dickey–Fuller generalized least squares (DF-GLS) to capture the order of integration of the variables, and the results show that all the variables are stationary in regard to the first difference. In the second step, we have applied the examination of bounds in order to validate the presence of long-term cointegration relationships between the variables. The results of the ARDL approach show that financial development, sustainable energy, and commercial openness have a negative impact on CO₂ emissions. However, GDP, energy export, and unsustainable energy lead to an increase in environmental degradation. Finally, the Granger causality test shows mixed causality relationship among the variables. Accordingly, governments should encourage the development and use of sustainable energy alternatives, such as solar power, wind power, and hydroelectric power, through incentives and subsidies, in addition to conducting new research concerning the topic and starting new initiatives. Protecting and expanding green areas is crucial to mitigate CO₂ emissions, and strategies for transitioning to cleaner energy alternatives should be developed. Additionally, facilitating the transfer of sustainable energy technologies and promoting collaboration in research and development can accelerate the adoption of clean energy solutions. These policy actions can contribute to reducing CO₂ levels, as well as promoting sustainable energy practices in the country.

1. Introduction

The relentless growth of the global population translates directly into an ever-increasing demand for energy, an essential resource that powers homes, businesses, and industries. Affordable and reliable energy is a crucial driver of economic development, fostering productivity and enhancing a nation’s competitiveness in the global marketplace [1]. It supports the development of vital infrastructure, manufacturing capabilities, and efficient transportation networks, which are the cornerstones of a thriving economy [2]. Beyond economic benefits, access to energy significantly improves individual well-being by providing basic necessities like lighting, heating, and cooking. Conversely, inadequate access to energy can lead to poverty, reduced productivity, and economic stagnation, according to the International Energy Agency [3].

Despite the historical significance of non-renewable energy sources, such as oil, coal, and natural gas, their finite nature and environmental impact require a paradigm shift in our energy strategy. The combustion of these fuels contributes heavily to greenhouse gas emissions, a major factor in global warming and climate change [4]. The looming depletion of these resources, coupled with their environmental impact, requires a shift towards sustainable energy solutions. This pressing need has propelled the rise of renewable energy sources like wind, solar, and hydropower. Their environmentally friendly nature offers a compelling alternative. Transitioning to these sources holds the potential to significantly reduce our dependence on non-renewable resources, curtail carbon emissions, and mitigate the ongoing climate crisis, according to the Renewable 2020 Global Status Report [5]. However, this transformation is not without its challenges. The large-scale adoption of renewable energy requires substantial investments in infrastructure development and grid modernization, according to the International Renewable Energy Agency [6].

Navigating this complex energy landscape requires a nuanced approach that balances economic growth, environmental sustainability, and long-term energy security. This study delves into these critical issues within the context of the Kingdom of Saudi Arabia (KSA). As a major oil producer, the KSA stands at a crossroads. As the world transitions towards cleaner energy sources, the KSA must chart a course that ensures economic prosperity, while safeguarding the environment for future generations. This research investigates the factors influencing CO₂ emissions in the KSA and proposes strategies for achieving a sustainable energy future [1,4].

This study investigates the relationships between CO₂ emissions, gross domestic product (GDP), financial development, energy export, sustainable power, unsustainable power depletion, and commercial growth in the KSA from 1990 to 2022.

Using advanced econometric techniques, such as the auto-regressive distributed lag (ARDL) approach and the vector error correction model (VECM), this research aims to uncover the long-term and short-term relationships among these variables. Initial tests, including the augmented Dickey–Fuller (ADF) test and the Dickey–Fuller generalized least squares (DF-GLS) test, establish the stationarity of the variables, paving the way for a robust analysis of their interactions.

2. Literature Review

Consequently, several attempts have been made to deal with the issue. For instance, ref. [7] discussed the relationship between energy consumption and CO₂ emissions in North Africa during the 1990–2015 period. They used panel data and cross-section techniques and found that energy consumption is a major determinant of CO₂ emissions in these countries. They also discovered bidirectional causality relationships between energy consumption and carbon emissions, and between economic growth and carbon emissions.

In another study, ref. [8] used the ARDL approach to prove that renewable energy has a negative effect on CO₂ emissions in Thailand in the short term. However, it should be taken into consideration that the reduction rate of non-renewable energy and the growth in the gross domestic product had a desirable effect on CO₂ volumes in the atmosphere.

Likewise, ref. [9] found that renewable energy resources and CO₂ emissions are out of sync and have an anti-cyclic influence, with renewable energies leading the sequence. The estimation results show that there is an important association between CO₂ emissions and renewable energies in the long term.

In addition, ref. [10] applied a nonlinear panel smooth transition regression model, using a panel database for 33 different OECD nations, throughout the period between 2000 and 2018, and found that renewable energy consumption has a significant nonlinear impact on CO₂ emissions. However, the relationship between CO₂ emissions and renewable energy use becomes important as the globalization level rises.

What is more, it is important to mention that [11] used the ARDL method in order to examine the interconnection between CO₂ emissions and sustainable energy in Turkey. They found a cointegration relationship between them and suggest that renewable energy is an essential factor in reducing CO₂ emissions in the long term. However, the Toda–Yamamoto causality test showed a unidirectional causality relationship between renewable energy consumption and CO₂ emissions.

On the other hand, in a different study, ref. [12] noted that using unsustainable sources of energy has a positive effect on CO₂ emissions in seven selected nations, while gross domestic product and the industrial sector had a weak positive impact on CO₂ emissions. However, ref. [13] analyzed energy use data in Sichuan Province, China, from 2010 to 2019 and argued that energy use assembly has a positive impact on CO₂ emissions.

In a similar way, ref. [14] analytically analyzed the relationship between the release of CO₂ and unsustainable energy use in 193 member states of the United Nations. The analytical results show that there is a great depletion of power, in addition to a dramatic increase in the proportion of CO₂ in the atmosphere. Full energy consumption is related to the index of human disparity.

To delve more deeply into the issue, it is crucial to note that [15] argued that unsustainable energy utilization and economic growth have a positive effect on the burning of carbon emissions in 16 countries in the European Union. Ref. [16] examined the relationship between non-renewable energy consumption and environmental degradation in 16 Latin American countries and showed a high correlation between the two variables. However, the consumption of fossil fuels increases CO₂ emissions.

In addition to that, ref. [17] noticed that a bidirectional causality interconnection exists between sustainable and non-unsustainable energy usage by the G-7 nations. Ref. [18] affirmed that non-renewable energy prices have a significant and positive effect on renewable energy consumption in the United States. Additionally, to tackle the issue [19] used a database of 55 countries. They found evidence that every single increase in the depletion of sustainable power is associated with an increase in non-renewable energy consumption, and vice versa.

It is becoming more evident that green spaces have a crucial role to play in reducing carbon dioxide emissions, according to several articles. A study conducted by [20] in South Korea, discovered that urban green spaces can significantly reduce carbon dioxide emissions by up to 8.9% per year. Similarly, a research paper by [20] conducted in China, demonstrated that increasing green spaces in cities can lead to a reduction in carbon dioxide emissions, by up to 15%. On the other hand, in Canada, a study by [21] mentioned that a 10% increase in green space can result in a 1% decrease in air pollution. Finally, a study by [22] in France, revealed that urban green spaces can provide significant carbon sequestration benefits, with up to 27.7 metric tons of carbon dioxide stored per hectare of green space. These findings suggest that the creation and preservation of green spaces in urban environments can play a critical role in mitigating the effects of climate change, by reducing carbon dioxide emissions.

Ref. [23] examine the impact of two green growth indicators on environmental sustainability. These indicators relate to domestic material consumption and renewable energy consumption. The study uses data from across 62 countries, from 1980 to 2020, analyzing the data both regionally and globally.

The authors employ various statistical methods to assess the relationships between these factors and environmental degradation, typically measured by the amount of CO₂ emissions. Their findings suggest that both increased renewable energy consumption and reduced domestic material consumption contribute to lowering CO₂ emissions, indicating a positive effect on environmental sustainability.

The study also investigates the influence of green growth policies on these relationships. While the results suggest a slight improvement in reducing CO₂ emissions with the implementation of green growth policies, the impact varies depending on the region. The authors further explore the potential for causal relationships between these factors. They find evidence of two-way causality between CO₂ emissions and both domestic material consumption and renewable energy consumption, globally, and in some regions. In other regions, one-way directional and neutral causalities are observed.

Overall, the research highlights the potential of promoting green growth through reduced material consumption and increased renewable energy use to achieve environmental sustainability. The study also emphasizes the importance of considering regional specificities when designing and implementing green growth policies.

Ref. [24] investigate the relationship between energy intensity, renewable energy adoption, economic growth, and CO₂ emissions, across African countries. The study examines data from various regions and across income levels within Africa. The authors’ found a positive correlation between energy intensity and CO₂ emissions, a negative correlation between renewable energy and CO₂ emissions, and a positive correlation between economic growth and CO₂ emissions. The study suggests that economic development in Africa might come at the cost of environmentally sustainable development if not accompanied by a shift towards renewable energy sources and improved energy efficiency.

Ref. [25] studied the relationship between CO₂ emissions, renewable energy consumption, economic growth, and population growth in seven East African countries, during 1980–2016. The researchers proved that renewable energy consumption has a negative impact on CO₂ emissions. This implies that as countries in East Africa increase their use of renewable energy sources, their CO₂ emissions decrease. However, economic growth and population growth have a positive impact on CO₂ emissions, which suggests that as these countries experience economic development and population increases, their CO₂ emissions will also rise.

The study employs advanced statistical methods not only to examine these overall trends, but also to explore potential asymmetries and regional variations. They found that the relationships between the variables can be complex and may differ across individual countries.

While the existing literature covers the global energy transition extensively, specific research on Saudi Arabia’s unique context remains limited. Key studies, such as those by Refs. [26,27], highlight aspects of Saudi Arabia’s path to sustainability and the potential for the reduction of emissions through the use of hydrogen fuel. However, gaps remain in understanding the nuanced relationships between CO₂ emissions, GDP, financial development, energy export, sustainable power, unsustainable power depletion, and commercial growth in Saudi Arabia. Additionally, there is a lack of comprehensive studies that use advanced econometric techniques, such as the ARDL and VECM approaches, to analyze these relationships over an extended period.

Moreover, while several studies have explored the impact of renewable energy on CO₂ emissions in various regions, there is a need for more in-depth analysis focusing specifically on the KSA, considering its status as a major oil producer and its ambitious Vision 2030 plan. This includes examining the role of financial development and energy exports, which are crucial for Saudi Arabia’s economy, in the context of the sustainable energy transition and the aim to reduce CO₂ emissions. In conclusion, this study aims to fill these gaps by investigating the factors influencing CO₂ emissions.

3. Data and Methodology

The objective of this paper is to analyze the relationships between renewable and non-renewable energy sources in light of the fact that they are interconnected with each other, in addition to studying their effects on economic growth, green areas, and the environment in the Kingdom of Saudi Arabia (KSA). We use annual data from the period 1990–2022 to conduct our analysis, in the form of simultaneous equations. In the first step, we define the econometric models and, in the second step, we try to write the auto-regressive distributive lag (ARDL) and vector error correction model (VECM). All of the outcomes and findings are dealt with in step 3.

The general econometric model approved in this research is expressed as follows:

$${\mathrm{F}}_{\mathrm{CO}2\mathrm{E}}\left(\mathrm{GDP},\mathrm{FD},\mathrm{EE},\mathrm{NRE},\mathrm{RE},\mathrm{TO}\right)$$

(1)

where CO₂E, GDP, FD, EE, NRE, RE, and TO, are the CO₂ emissions (expressed as the CO₂ emissions from transport, measured by the percentage of total fuel combustion), gross domestic product (annual rate), financial development, energy export (expressed as the number of barrels of fuel), non-renewable energy consumption (expressed as the fossil fuel energy consumption, measured by the percentage of total energy consumption), renewable energy (expressed as the percentage of total final energy consumption), and trade openness, respectively, presented in

Table 1. Variable definitions and sources.

Abbreviations	Meaning	Sources
CO2E	CO2 emissions	WIDI; 2023
GDP	Gross domestic product	WIDI; 2023
FD	Financial development	WIDI; 2023
EE	Energy export	WIDI; 2023
NRE	Non-renewable energy consumption	WIDI; 2023
RE	Renewable energy	WIDI; 2023
TO	Trade openness	WIDI; 2023

. All data are extracted from the World Data Bank [27].

Thus, by incorporating our variables, the econometric model takes on the following form:

$$lnCO2E_{it}={\beta }_0+{\beta }_{1}lnGD{P}_{it}+{\beta }_{2}lnF{D}_{it}+{\beta }_{3}lnE{E}_{it}+{\beta }_{4}lnNR{E}_{it}+{\beta }_{5}lnR{E}_{it}+{\beta }_{6}lnT{O}_{it}+{\epsilon }_{it}$$

(2)

where:

${Ɛ}_t$ is the white-noise disturbance term.

lnCO₂E represents CO₂ emissions expressed as a logarithmic function, while lnGDP denotes GDP expressed as a logarithmic function, lnFD denotes FD expressed as a logarithmic function, lnEE denotes EE expressed as a logarithmic function, lnNRE denotes NRE expressed as a logarithmic function, lnRE denotes RE expressed as a logarithmic function, and lnTO denotes TO expressed as a logarithmic function.

The auto-regressive distributed lag (ARDL) models are expressed in the following form, when there is a state of long-term cointegration present:

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

(3)

The error correction dynamics are denoted by the symbols γ, δ, θ, ϑ, μ, ρ, and τ, while D represents the first-difference operator. The long-run relationships between the variables in the model are indicated by ${\beta }_1$ to ${\beta }_7$ and the constants are represented by ${\beta }_0$. The optimal lags are represented by p and q and ${Ɛ}_t$.

In addition to that, ARDL models are used (F-statistic) in order to recognize whether there is a long-term relationship between the related variables or not. Additionally, (F-statistic) the model includes enforcing confines on the assessed long-term constants with different types of model variables. However, the outcome of the F-statistic value refers to a significance at the 10% level, thus rejecting the null hypothesis of non-cointegration in the long run. The null hypothesis (H₀) and alternative hypothesis (H₁) are formulated as follows:

H₀: 

${\beta }_1={\beta }_2={\beta }_3={\beta }_4={\beta }_5={\beta }_6={\beta }_7=0$ (there is no long-term relationship)

H₁: 

${\beta }_1\ne {\beta }_2\ne {\beta }_3\ne {\beta }_4\ne {\beta }_5\ne {\beta }_6\ne {\beta }_7\ne 0$ (there is a long-term relationship present)

As Ref. [28] suggest, the (ARDL) model is employed for testing cointegration. Here, 5 unlimited corrections of errors can be approximated, by taking every single variable into account. All of those variables are dependent, and they work according to the confines of the ARDL model, which examines the method presented by Refs. [29,30], respectively. In comparison to conventional methods of cointegration (such as [31,32,33]), the ARDL approach yields better results for small sample data sets, as argued by [34]. Additionally, according to [35], the unlimited (ECM) model seems to capture the data generating process, with adequate lags, by following a structure which moves from general to specific.

Thus, it is important to notice that the ARDL method does not require any variable classification, such as I(1) or I(0), by developing groups of significant values, which recognize stationary or non-stationary procedures. As opposed to other cointegration practices, such as Johansen’s technique, the ARDL approach presents an obvious way to carefully study a long-term relationship, paying no attention to whether the primary variables are entirely I(0) or I(1), or whether they are slightly combined in one way or another. As a result, there is no further need for any preceding tests concerning the unit root of the variables. What is more, despite the fact that conventional cointegration approaches can address many different kinds of issues, the ARDL-style approach may differentiate between explanatory variables and dependent ones. Consequently, assessments resulting from the ARDL approach are totally balanced and effective, due to the fact that they have nothing to do with issues related to endogeneity or serial correlations.

It is worth mentioning that the ARDL approach allows unequal orders of lags, in a different way from Johansen’s VECM. On the other hand, as [31] indicate, suitable adjustments in the orders in the ARDL approach are sufficient to make a concurrent correction for serial correlations and the issue of endogenous variables.

Another essential point to talk about is that once the existence of cointegration among the variables is confirmed, the Granger causality method based on the vector error correction model (VECM) is employed to investigate the direction of causality between the variables. The VECM is a constrained version of the vector auto-regression (VAR) model. Yet, the error correction term (ECT) in the VECM represents the cointegration term, which allows the endogenous variables to converge into their long-term relationships, while adjusting for short-term dynamics. By employing the VECM, it becomes possible to approximate the short-term relationships among the variables, while the ECT provides evidence of the presence of long-term relationships.

The Granger causality test will be conducted using the vector error correction model (VECM). The VECM equations are formulated as follows:

$$\begin{array}{cccc}DlnCO2E_t& =& {\beta }_0& +{\displaystyle {\displaystyle \sum }_{i=1}^{\alpha 1}}{\alpha }_{1i}DlnCO2E_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^{\gamma 1}}{\gamma }_{1i}DlnGD{P}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^{\delta 1}}{\delta }_{1i}DlnF{D}_{t-i}\hfill \\ & & & +{\displaystyle {\displaystyle \sum }_{i=1}^{\theta 1}}{\theta }_{1i}DlnE{E}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^{\vartheta 1}}{\vartheta }_{1i}DlnNR{E}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^{\mu 1}}{\mu }_{1i}DlnR{E}_{t-i}\hfill \\ & & & +{\displaystyle {\displaystyle \sum }_{i=1}^{\pi 1}}{\pi }_{1i}DlnT{O}_{t-i}+{\phi }_{1}EC{T}_{t-1}+{\epsilon }_{1t}\hfill \end{array}$$

(4)

In the VECM equations, β₀ represents the constant, while the symbols α, γ, δ, θ, ϑ, μ, and π are the coefficients that are required to be assessed. The error term Ɛ is a white noise term and the error correction term (ECT) represents the long-term equilibrium relationship between the variables.

4. Results and Discussion

Before investigating the relationship between the variables, the first step in our study is to check the stationarity order of each variable using ADF and DF-GLS unit root tests. The null hypothesis (H₀) assumes the presence of a unit root. If H₀ is rejected, we can proceed with our econometric analysis. Our results, as shown in

Table 2. Unit root test results.

	Tests	ADF Test		DF-GLS Test
Variables
		Integration Order		Integration Order
		At level	First difference	At level	First difference
CO2E		−5.83 **	−3.773 **	−5.889 **	−3.−4.821 ***
GDP		−4.004 *	−2.773 **	−11.579 **	−4.−3.115 ***
FD		3.112	−2.334 ***	−2.222 **	−8.−4.122 ***
EE		1.665	−5.234 ***	2.132	−3.−3.901 ***
NRE		−0.576	−4.893 ***	3.115	−2.−7.980 **
RE		−0.125	−2.830 *	0.132	−3.−6.665 ***
TO		2.002	−1.882 ***	1.241	−4.767 *

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

, indicate that CO₂E and GDP are stationary at the set level (according to the ADF and DF-GLS tests); in addition, the FD is stationary at the set level, according to the DF-GLS test. However, in regard to the first difference, all the variables are stationary, implying that the null hypothesis is declined in the current case.

In the second step, the bounds test is used in order to detect the long-term cointegration between the variables, whereas the F-statistic is calculated and compared with critical values, and the optimal lag is selected by minimizing the Akaike information criterion (AIC) and the Schwarz information criterion (SIC). The bounds test results (

Table 3. Bounds test results.

Dependent Variable	F-Statistic
${\mathrm{F}}_{\mathrm{C}\mathrm{O}2\mathrm{E}}\left(\mathrm{G}\mathrm{D}\mathrm{P},\mathrm{F}\mathrm{D},\mathrm{E}\mathrm{E},\mathrm{NRE},\mathrm{RE},\mathrm{TO}\right)$	6.09
Critical Value Bounds
Significance level	I(0)	I(1)
10%	2.34	3.16
5%	3.51	4.32
1%	4.76	4.98

) show that the F-statistic is above the 10% upper bound, confirming the presence of long-term cointegration between the variables.

We conducted the Breusch–Godfrey serial correlation LM test to detect residual correlation, and the findings are presented in

Table 4. Diagnostic test.

Country	Dependent Variable	LM Test	ARCH Test	Reset Test	JB Test
Saudi Arabia	${\mathrm{F}}_{\mathrm{CO}2\mathrm{E}}\left(\mathrm{GDP},\mathrm{FD},\mathrm{EE},\mathrm{NRE},\mathrm{RE},\mathrm{TO}\right)$	0.031	0.214	0.018	0.603

. Both models showed no evidence of serial correlation. Additionally, we used the ARCH test to test for heteroscedasticity and found that all four models are homoscedastic, with a normal distribution of error terms.

To ensure the stability of the ARDL model, we employed the CUSUM (cumulative sum) and CUSUMSQ (cumulative sum of the squares) tests, added to the findings illustrated in Figure 1.

The CUSUM and CUSUMSQ tests are statistical tools used to monitor the stability of econometric models over time, particularly regression models. They focus on identifying potential changes in the relationships between the variables.

Imagine tracking the sum of the model’s residuals (errors) over time, this is essentially what the CUSUM does. If this cumulative sum consistently deviates beyond a certain threshold (typically visualized as red lines), it suggests the model’s behavior might be unstable. In simpler terms, the relationship between the variables may be shifting in a way that the model is not capturing.

The results of the ARDL estimations for the three different models and for eight Southeast Asian countries are indicated in

Table 5. Long-term ARDL coefficients.

Saudi Arabia	Independent Variable (CO2E)
Dependent variables	LnGDP	1.03 (0.000) ***
	LnFD	−0.53 (0.000) ***
	LnEE	3.98 (0.000) ***
	LnNRE	2.04 (0.587)
	LnRE	−0.29 (0.000) ***
	LnTO	−0.76 (0.421)
	Constant	141.067 (0.000) ***
	CUSUM	Stable
	CUSUMSQ	Stable

*** indicate the significance at 1%, respectively.

.

According to the ARDL estimation, it appears that GDP, EE, and NRE have an important role in the increases in environmental degradation. In effect, the increase in GDP by one unit leads to an increase in CO₂E by 1.03 units. Economic development frequently relies on increased energy use, principally fossil fuels. These fuels release a significant amount of carbon dioxide and other greenhouse gases, openly contributing to the increase in CO₂E. EE strongly leads to the augmentation of CO₂E, by 3.98 units. When energy is extracted, it releases greenhouse gases like CO₂ into the atmosphere. On the other hand, the exportation of this energy (fuel, natural gas) provides the possibility to burn and use this energy in another location in the world, which results in the exportation of the carbon dioxide emissions associated with its production. Needless to say, NRE has a major impact on the CO₂E reaction. In effect, when NRE increases by one unit, CO₂ emissions increase by 2.04 units, which means that the use of natural non-renewable energy resources has an important role in environmental degradation. On the other hand, FD, RE, and TO cause a reduction in carbon dioxide emissions in the Kingdom of Arabia Saudi. Basically, financial development can be considered as an alternative solution to facilitate the increase in green technology assets, promote environmental knowledge and practices, and to maintain sustainable economic growth. In addition, it should be taken into account that renewable energy represents a solution to increase energy efficiency and it is considered to be a key in replacing the use of fossil fuels, which means no carbon dioxide emissions during future generations. Finally, trade openness provides the opportunity to transfer technology between nations, which encourages competition and the implementation of environmental regulations.

Here, it should be emphasized that the Granger causality analysis, the results of which are shown in

Table 6. Granger causality test results.

Causality Directions
	Short Term								Long Term
Dependent Variables		DLnGDP	DLnCO2E	DLnFD	DLnEE	DLnNRE	DLnRE	DLnTO	ECT
Dependent variables	LnGDP	----	1.143 (0.032) **	0.121 (0.886)	18.539 (0.623)	0.140 (0.069) *	1.563 (0.026) *	1.769 (0.087) *	−0.368 (0.094) *
	LnCO2E	3.061 (0.061) *	----	3.646 (0.038) **	3.323 (0.049) **	1.843 (0.075) *	0.121 (0.886)	1.669 (0.205)	0.036 (0.019)
	LnFD	0.229 (0.796)	0.949 (0.398)	----	1.048 (0.363)	1.123 (0.338)	0.395 (0.076) *	0.465 (0.632)	0.102 (0.070)
	LnEE	2.985 (0.165)	0.168 (0.845)	0.336 (0.716)	----	0.774 (0.470)	0.035 (0.965)	2.256 (0.122)	−17.071 (3.541)
	LnNRE	2.433 (0.004) ***	2.398 (0.008) ***	0.528 (0.594)	2.957 (0.067) *	----	0.490 (0.617)	1.056 (0.060) *	−0.009 (0.060) *
	LnRE	4.352 (0.021) ***	0.339 (0.714)	2.547 (0.095) *	2.475 (0.101)	1.083 (0.351)	----	2.024 (0.149)	6.309 (3.986)
	LnTO	0.054 (0.947)	1.622 (0.214)	3.728 (0.035) **	2.125 (0.137)	0.124 (0.083) *	1.930 (0.162)	----	−0.764 (0.299)

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

, reveals the presence of mixed relationships, both unidirectional and bidirectional, among the variables. The VECM Granger causality test also yields mixed results regarding the relationships between the variables. The error correction term (ECT) is used to determine the presence of long-term causal relationships. When the coefficient is significant and negative, it refers to the presence of at least one long-term relationship between the variables, with the endogenous variable playing a crucial role as an adjustment element when the model deviates from the equilibrium. In other words, the coefficients of the ECT, as shown in Table 6, confirm the existence of long-term relationships among the variables, as detected by the cointegration tests. Additionally, the results demonstrate co-movement of the variables.

The results of the test on the short-term causality relationships show that the CO₂E variable has two bidirectional causality relationships with the NRE and GDP variables. In addition, a bidirectional causality relationship exists between the NRE and TO variables. Finally, the results show the existence of a bidirectional causality relationship between the NRE and GDP variables.

However, the results of the test on the long-term causality relationships confirm the presence of one bidirectional causality relationship between the GDP and NRE variables.

A summary of the causality relationships between the variables is presented in Figure 2.

5. Conclusions and Policy Implications

This study investigates the relationships among the CO₂ emissions, gross domestic product (GDP), financial development, energy export, non-renewable energy consumption, and trade openness in Saudi Arabia, from 1990 to 2022. Employing the auto-regressive distributed lag (ARDL) approach and the vector error correction model (VECM), we analyze the dynamic interactions between these variables. The augmented Dickey–Fuller (ADF) and the Dickey–Fuller generalized least squares (DF-GLS) tests reveal that all the variables are stationary in regard to the first difference. The bounds test confirms the long-term cointegration relationships among the variables, and the CUSUM and CUSUMSQ tests indicate the stability of the economic model over time.

The ARDL results show that GDP, energy exports, and non-renewable energy consumption positively impact CO₂ emissions, contributing to environmental degradation. In contrast, financial development, renewable energy, and trade openness negatively impact CO₂ emissions, suggesting these factors are crucial for preserving the environment in Saudi Arabia. The Granger causality analysis reveals mixed unidirectional and bidirectional relationships between the variables, with the error correction term (ECT) coefficients confirming the long-term relationships identified by the cointegration tests.

Our findings align with previous studies by Refs. [8,15], which highlight the significant impact of energy consumption on CO₂ emissions and the beneficial effects of renewable energy on reducing emissions. Similarly, [9,11] emphasize the long-term negative relationship between renewable energy and CO₂ emissions, while [17,36] underscore the positive impact of unsustainable energy on CO₂ levels.

Given these insights, it is crucial for policymakers to balance economic growth with environmental sustainability. Promoting financial development to diversify the economy, setting ambitious renewable energy targets, and encouraging trade openness alongside environmental considerations, are essential strategies. Additionally, investments in green infrastructure and trade agreements that include environmental requirements can further support these goals. The creation and preservation of green spaces, as highlighted by [20,36], can also play a critical role in mitigating CO₂ emissions.

This study highlights the critical challenge of balancing economic growth with environmental sustainability in Saudi Arabia. Analyzing CO₂ emissions, economic factors, and energy consumption, reveals several key policy recommendations:

First, promoting financial development is essential. Diversifying the Saudi economy beyond its dependence on oil through policies that encourage financial innovation and attract foreign investment can help reduce the reliance on high-emitting energy sources.

Second, increasing the adoption of renewable energy is vital. Investing in renewable energy infrastructure and setting ambitious targets for its use will significantly contribute to lowering CO₂ emissions, paving the way for a sustainable future.

Third, encouraging trade openness along with environmental considerations can make a significant impact. Promoting “green trade” through investments in green infrastructure and establishing trade agreements that incorporate environmental requirements will incentivize environmentally friendly practices across all sectors.

Finally, expanding green spaces is crucial. Studies have shown that green spaces play a vital role in mitigating CO₂ emissions. Saudi Arabia can explore urban greening initiatives, such as creating parks and green spaces in cities, to further reduce emissions and improve overall environmental health.