The nexus of carbon dioxide emissions, economic growth, and urbanization in Saudi Arabia

Saudi Arabia has implemented its ambitious and comprehensive national strategy, i.e., Saudi Vision 2030, to achieve major economic, social, and environmental objectives. The main aim of this paper is to study the Granger causality relationships between economic growth, environmental degradation, and urbanization in Saudi Arabia over the period from 1985 to 2019. At first, Augmented Dicky-Fuller (ADF) and Phillips-Perron (PP) tests were applied in order to check the stationarity of the panel time-series data. Since the data were of mixed order of integration I(0) and I(1), the Autoregressive Distributed Lag (ARDL) framework was employed to perform the statistical analysis. Then, the short- and long-run relationships were evaluated using the bounds test for cointegration applied on the Error Correction Models (ECMs) for GDP, CO2 emissions, and urbanization as the dependent variables. Furthermore, the direction and significance of causality were estimated in the ARDL/ECM framework. In addition, the Environmental Kuznets Curve (EKC) hypothesis was examined for the sample data. To assess the generalization capability of the findings in this study, robustness and diagnostic tests were applied. In the long-run, the empirical findings indicate that 1% increase in economic growth Granger caused 0.15% increase in CO2 emissions and 0.006% decrease in urbanization. Whereas 1% increase in urbanization Granger caused 2.5% increase in the economic growth. In the short-run, a unidirectional causal relationship existed from economic growth to both CO2 emissions and urbanization with 1% increase in GDP Granger caused 0.3% and 0.004% increases in CO2 emissions and urbanization, respectively. Finally, policy recommendations were presented in light of the Saudi Vision 2030.


Introduction
Climate change has been identified as a major environmental issue across the world. Since the early 1970s, there has been a rising understanding among individuals and governments about the importance of working together to address this environmental issue [1]. In 2015, the United Nations Framework Convention on Climate Change (UNFCCC) responded to the climate change risks by adopting the Paris Climate Change Agreement which aims to 'limit global warming to well below 2 degrees Celsius, preferably to 1.5 degrees Celsius, compared to preindustrial levels' [2]. Globally, countries are being urged to develop mitigation measures to aid in the reduction of greenhouse gas (GHG) emissions in order to protect and sustain the world's climate. Although GHGs are important to maintain the Earth's temperature at levels to preserve life, increasing amounts of GHG emissions may cause hazardous environment [3]. Understanding the interrelations between economic and social activities and climate change is central to adopt the right policies to achieve sustainable growth and preserve the environment.
Saudi Arabia has made significant social and economic growth over the previous four decades, but it has long been one of the world's largest CO 2 emitters per capita, owing mostly to oil and gas production. CO 2 constitutes approximately 76% of GHGs and most CO 2 emissions come from the combustion and processing of fossil fuels like crude oil and natural gas [4]. Saudi Arabia geography constitutes about 80% of the Arabian Peninsula with an area of 2.15 million square kilometers and a total population of around 35 million [5]. Over the last four decades, the rate of urbanization in Saudi Arabia has been steadily growing. Given the current rate of urbanization of above 84%, cities now account for a significant portion of Saudi Arabia's economic growth and CO 2 emissions. For economic growth, the per capita share of Gross Domestic Product (GDP) rose from 7,919 (current USD) in 1985 to 23,139 (current USD) in 2019.
The Saudi government's ambitious plan, Saudi Vision 2030, was launched in 2015 and serves as the foundation for all adopted economic, social, and environmental policies and projects [6]. The vision intends to diversify the economy and move away from reliance on the oil industry, as well as promote social development and environmental preservation. In order for Saudi Vision 2030 to attain its objectives, it is crucial to comprehend the causes of environmental deterioration and their links to economic expansion and urbanization. Researchers have paid close attention to the interrelationship between economic growth and the environment, particularly in the context of the EKC, which states that there is an inverted U-shaped relationship between some measures of environmental degradation and some measures of welfare and economic growth (such as the level of income and output). The findings of the Samargandi study [7] refute the presence of a hypothesis EKC in the case of Saudi Arabia.
There are many aspects in which this paper contributes to the existing literature. Firstly, this study adequately addressed the Granger causality relationship between GDP, CO 2 emissions and urbanization in Saudi Arabia with robust statistical and econometrics methods. Up to the authors' best knowledge, there is no similar study in the literature where the results can be generalized and utilized to support the policy making stakeholders in the public and private sectors. Secondly, the study evaluated the short-run and long-run relationships with causality analysis including direction and coefficient estimation. Thirdly, the EKC hypothesis was validated for Saudi Arabia during the study period. The rest of the paper is organized as follow. Section 2 presents a literature review about the topic of the study. Then, materials and methods are explained in section 3. Section 4 presents the results and discussions. Finally, we conclude our work in section 5.

Literature review
The literature is divided based on the nexus relationship between the three variables of interest in this study: CO 2 emissions, economic growth and urbanization. At the end of this section, a summary of the existing studies on the relationships between the variables of interest is presented in table 1.

The nexus between CO 2 emissions and economic growth
Radmehr et al used Generalized Spatial Two-stage Least Squares (GS2SLS) method to study the relationships between economic growth, carbon emissions, and renewable energy consumption in European Union countries [8]. Their study concluded that a bidirectional positive relationship exists between economic growth and each of carbon emissions and renewable energy consumption, whereas renewable energy consumption leads to reductions in carbon emissions. The bidirectional positive relationship between economic growth and carbon emissions was also obtained for the top ten emitting countries by Mohmmed et al [9], who also found that population positively impacted carbon emissions. In another work, Wang et al studied the decoupling of carbon emissions from economic growth in the United States which exhibited positive relationship before 2007 and then negative relationship during the period from 2007 to 2016 [10]. Naz et al [11] looked at data from Pakistan for over 40 years and came to the conclusion that economic growth and foreign direct investment were the primary causes of CO 2 emissions, and that the EKC hypothesis was incorrect. In addition, renewable energy consumption had a negative relationship with CO 2 emissions. Bekun et al [12] found a negative relationship between economic growth and CO 2 emissions in South Africa. In contrary to most studies, Litavcova and Chovancova [13] obtained varying relationships between economic growth, carbon emissions, and energy consumption for 14 Danube region countries. Another study by Rahman et al [14] concluded that carbon emissions and population density had unidirectional positive effect on the economic growth in the long run, with this relation became bidirectional in the short run for five South Asian countries: Bangladesh, India, Nepal, Pakistan and Sri Lanka. In another study about European Union 5 (EU-5) countries (Germany, France, Italy, Spain, and the United Kingdom), Balsalobre-Lorente et al [15] found N-shaped relationship between carbon emissions and economic growth. Meanwhile, renewable energy and natural resource availability improved environmental quality. Salahuddin et al [16] discovered a positive relationship between carbon emissions and economic growth, foreign direct investment, financial development, and energy consumption in Kuwait. Salari et al [17] concluded that economic growth and carbon emissions relationship followed an inverted U-shaped pattern for USA and renewable energy consumption reduced the carbon emissions. A more comprehensive review about the relationship between economic growth and CO 2 emissions was carried out by Mardani et al [18].

The nexus between CO 2 emissions and urbanization
Zhou et al [19] found that increasing population urbanization lead to increasing CO 2 emissions in east China. In another study, Mahmood et al [20] found that urbanization and industrialization had positive relationship with CO 2 emissions in Saudi Arabia with asymmetrical effect of industrialization, i.e., CO 2 emissions promoted by 1% increase in industrialization were greater in absolute value than the decrease in CO 2 emissions caused by 1% decrease in industrialization. In a study on China by Chen et al [21], an inverted U-shaped between urbanization and carbon emissions during the study period. Moreover, urbanization had an impact on energy consumption structure by promoting more coal consumption at the expense of natural gas. In another study, Ali et al [22] found that urbanization caused increased CO 2 emissions in Pakistan. Similar findings were obtained by Abbasi et al [23] who found a positive and significant impact of urbanization and energy consumption on CO 2 emissions in eight Asian countries.

The nexus between economic growth and urbanization
Arguably, economic growth and urbanization 1 are inseparable, i.e., economic growth cannot take place without urbanization and vice versa. However, the causal relationship between these two variables is not clear [8].
Nguyen and Nguyen [24] found an inverted U-shaped relationship between economic growth and urbanization.
In another study, Narayan [25] found that a unidirectional positive effect of economic growth on urbanization in India. In China, Zheng and Walsh [26] indicated the existence of a U-shaped relationship between urbanization and economic growth. Another study was conducted on UAE by Sbia et al [27] and found that economic growth had positive impact on urbanization and vice versa.

The nexus between CO 2 emissions, economic growth and urbanization
Xie and Liu [28] studied the combined nonlinear effects of economic growth and urbanization on CO 2 emissions for China during 1997-2016 period and concluded that there was an inverted-U relationship between economic growth and carbon emissions. They also discovered heterogenous effects of urbanization on carbon emissions with three inflection points. Another study by Anwar et al [29] found that economic growth, urbanization, and trade openness enhanced carbon emissions in nine East Asian countries. Koengkan et al [30] concluded that economic growth, urbanization and fossil energy consumption promoted carbon emissions while renewable energy suppressed them in five South American countries including Brazil, Argentina, Venezuela, Paraguay, and Uruguay. They also found that carbon emissions, urbanization, and renewable energy had negative effect on the economic growth. In another study, Alotaibi and Alajlan [31] used quantile regression to analyze the relationship between carbon emissions and a group of social and economic indicators for the G20 countries. They concluded that the relationship between CO 2 emissions and each of economic growth, human development and prosperity exhibited an inverted U-shaped relationship. Whereas urbanization and trade openness had a negative relationship with CO 2 emissions. Ridzuan et al [32] employed the ARDL framework to study the drivers of CO 2 emissions in Malaysia. Their findings support the EKC hypothesis about the existence of an inverted U-shaped relationship between CO 2 emissions and economic growth. In addition, urbanization had positive relationship with CO 2 emissions whereas renewable energy and agriculture had negative relationship. Hashmi et al [33] investigated the spatial relationship between urbanization and CO 2 emissions in South and East Asian countries. An inverted U-shaped relationship was detected except for the largest cities where a U-shaped relationship existed. Economic growth and energy intensity positively impacted CO 2 emissions whereas trade openness had a negative relationship with CO 2 emissions. In another study, Wang et al [34] studied the relationships between economic growth, CO 2 emissions, urbanization, and energy consumption for 170 countries and found significant positive relationship existed between the variables. However, the relationship varied depending on the level of economic development among these countries. Odugbesan and Rjoub [35] study which covered MINT countries (Mexico, Indonesia, Nigeria, and Turkey) concluded that economic growth, CO 2 emissions, and energy consumption had a significant positive effect on urbanization. In another study, Adebayo et al [36] discovered an inverted U-shaped relationship between economic growth and CO 2 emissions, as well as the fact that urbanization contributes to environmental degradation.

Materials and methods
In this paper, econometric techniques were used to analyze the three-way relationship between CO 2 emissions, economic growth, and urbanization. At first, unit-root tests were applied to check the stationarity of the timeseries data. Then, time-series cointegration techniques were employed to analyze the presence of short-run and long-run relationships among the variables. Furthermore, Granger causality test was used to estimate the direction of causality between the variables of interest. Finally, the stability of the obtained results was analyzed.

Unit-root test
In regression modelling, non-stationary time series data can lead to spurious regressions between the time series variables. In fact, most econometric time series data are non-stationary. Therefore, unit-root tests were used to examine the stationarity of the time series data. In this paper, two widely-used tests in econometrics were used, namely, Augmented Dicky-Fuller (ADF) test [37] and Phillips-Perron (PP) test [38]. In both tests, the null hypothesis assumes existence of unit root and hence, no stationarity. Whereas the absence of the unit root means stationarity as the alternative hypothesis. In econometrics, unit root tests are used to determine the order of integration of time series denoted by I(k), where k is the number of difference operations required to bring the time series to stationarity.

Cointegration analysis using ARDL/ECM framework
Cointegration presents means to deal with nonstationary data. That is, cointegrated variables can jointly exhibit stationary behavior and build more reliable models. Cointegration is intended to capture long-run equilibrium among econometric time series that converges over time. Cointegration techniques put assumptions regarding the order of integration of the time series variables under consideration. In this paper, the ARDL bounds test was used to test for cointegration [39], which has several advantages over classical cointegration tests such as Johansen and Juselius [40]. Firstly, ARDL modeling can accommodate integrated variables of different orders, I(0) and I(1). However, the ARDL cointegration test does not work on I(2) time series data. Secondly, ARDL modeling performs relatively well in case of small data samples. For time series Y , t the ARDL model is characterized by the parameters ( ) p q q , , ., r 1 ¼ as in the following equation: p is the number of lagged terms of the dependent variable, j i , b are coefficients of the exogeneous variables X j t , and their lagged terms X j t i , -for i 0, > r is the number of exogeneous variables, q j is the number of lagged terms of the exogeneous variable X , j and t e is zero-mean white noise term. The ADRL model is used to find the long-run solution that describes the equilibrium relationship as well as short-run dynamics. Equation (1) can be rewritten in differential form to capture the short-run dynamics of the dependent variable as follows: . In this paper, the critical values given by Narayan [41] were used since they are suitable for small samples. The null hypothesis is rejected if the F-value exceeds the upper bound F(1) and cannot be rejected if F-value falls below the lower bound F(0). The test outcome is inconclusive if F-value is between F(0) and F(1). If the outcome of the bounds test indicates failure to reject the null hypothesis, then there is no cointegration and only the short-run model can be specified as: Where Z t 1 -is called the Error Correction Term (ECT) and given by: The ECT links the influence of the deviation from long-run equilibrium to the short-run dynamics of the dependent variable. The coefficient AE is called the speed of adjustment since it controls the rate at which Y t returns to equilibrium after changes in the exogeneous variables X . j t , AE must be negative in order to achieve convergence.

Causality analysis
The determination of the causal relationships among the variables of interest depends mainly on the outcome of the cointegration analysis. If the variables are cointegrated, then there is at least a unidirectional causal relationship [43]. In this paper, Granger causality analysis is carried out using the ARDL/ECM framework explained in the previous subsection. To evaluate the directional causal relationship in case there is cointegration among the variables, each variable is considered as dependent variable in equation (4) and the remaining variables as independent variables as in the following [44]: are the lag lengths of the dependent and independent variables, respectively. The short-run Granger causality can be inferred based on the significance of the t-statistics of the s w¢ coefficients. Whereas the long-run causality can be estimated using the t-statistics of the s ¢ AE coefficients. If the ECT coefficient is negative and has significant t-statistic, then the long-run causal effect of each independent variable on the dependent variable can be estimated as in the following equations: The parameters in equations (6) to (11) can be estimated using the Ordinary Least Squares (OLS) method. In order to validate the EKC hypothesis, a term representing the square of lnGDP, i.e., lnGDP2, was included in equations (6) and (9) for the short-and long-run relationships, respectively, as in the following:  table 3. The skewness and kurtosis descriptors were computed based on thirdorder and fourth-order central moments, respectively. The skewness reflects the asymmetry of the distribution whereas kurtosis describes its steepness. For the three variables, the Jarque-Bera test [45] scores indicate failure to reject the null hypothesis which states that the variable follows the normal distribution. This indicates that the variables in this study were drawn from normal distributions. The logarithmic trends for the three variables are shown in figure 1. Evidently, during the period of the study, the trend was positive for all variables. The Saudi Arabian economy grew at its fastest rate between 1998 and 2012, while urbanization developed steadily during the study period.

Unit-root test results
In order to analyze the causal relationships between CO 2 emissions, economic growth and urbanization in Saudi Arabia, the first step is to check for the stationarity of the time-series data of these variables before performing the modelling and regression analysis. To this end, the ADF and PP tests were employed to detect the presence of any unit roots and identify the order of integration. Table 4 shows the results of the ADF and PP unit-root tests. For the variable lnCO2, both tests revealed the failure to reject the null hypothesis at level and the rejection of the null hypothesis with 1% significance for the first difference, which indicates that lnCO2 variable is stationary of order one, i.e., I(1). Meanwhile, the results of both tests for the variable lnURB clearly showed the rejection of the null hypothesis and no existence of unit roots at level, which means that lnURB is stationary of order zero, i.e., I(0). In addition, both tests resulted in weak rejection of the null hypothesis at level and strong rejection at the first difference for the variables lnGDP and lnGDP2. In conclusion, the variables of interest in this study were stationary of mixed orders at level I(0) and at first difference I(1), with none of them was of the second order I(2). Therefore, the results of the stationarity tests satisfied the ARDL assumptions and indicated the relevance to proceed to perform the cointegration and relationship analyses using the ARDL framework.

Cointegration analysis results
After confirming the mixed order stationarity of the time-series data, the next step is to test for cointegration and the existence of long-run relationships among the variables. For this purpose, table 5 shows the results of the  ARDL bounds test for cointegration for the three models when considering each of lnCO2, lnGDP and lnURB as dependent variables in equation (1) and parameters were estimated using the OLS method. The obtained F-statistics values were 7.62, 4.82, and 6.12 for the dependent variables lnCO2, lnGDP and lnURB respectively. These values were greater than the upper bounds proposed by [41] at 1%, 10%, and 1% significance levels, respectively. This means that the null hypothesis of no cointegration was rejected for all three models. The Figure 1. Logarithmic plots of (a) CO 2 emissions per capita [46], (b) GDP per capita [5], and (c) urbanization [5] in Saudi Arabia for the period 1985 to 2019. Note: *** , ** , and * indicate statistical significance at 1%, 5%, and 10%, respectively. regressors were included with unrestricted constant and no trend, whereas the optimal lag lengths were selected based on the Akaike information criterion [47]. In brief, the empirical results indicated the existence of cointegration and long-run relationship between CO 2 emissions, economic growth and urbanization in Saudi Arabia during the period between 1985 and 2019. In the following, we proceed to estimate the Error Correction Model (ECM) for each dependent variable model and evaluate the short-and long-run Granger causal relationships.

Granger causality results
The existence of cointegration indicates the presence of a causal relationship between the variables at least in one direction. To compute the optimal lag lengths for the ECM models of equations (6)-(8), table 6 shows the lag selection for each model using four different lag selection criteria. Table 7 summarizes the Granger causality results for the three models. The results show the coefficients and p-values of the short-and long-run causality.
For the short-run relationships, economic growth Granger caused CO 2 emissions at 10% level of significance whereas urbanization had no significant effect on CO 2 emissions in the short-run. The increase of 1% in GDP caused 0.3% increase in CO 2 emissions in Saudi Arabia. Both CO 2 emissions and urbanization had no significant short-run effects on the economic growth. For urbanization, CO 2 emissions had no significant effect whereas economic growth Granger caused urbanization at 5% significance level. The growth of 1% in GDP caused urbanization to increase by 0.004% in the short-run. In brief, there were bidirectional causal relationship between economic growth and urbanization and unidirectional causal relationship from economic growth to CO 2 emissions in the short-run in Saudi Arabia.
In the long-run, the coefficients for the speed of adjustment of the error correction term Z t 1 -were negative at 1% significance level for all three models as shown in table 7. This implies that Granger causality relationship existed between CO 2 emissions, economic growth, and urbanization in Saudi Arabia in the long-run. However, each variable returns to the long-run equilibrium with different speed of adjustment after short-run shocks. Urbanization converged to the long-run equilibrium by 59% in one time period. In other words, urbanization took 1.7 years to return to the long-run equilibrium after a short-run shock. Similarly, CO 2 emissions and economic growth adjusted to the long-run equilibrium in 0.96 and 0.79 years, respectively. Table 8 shows the long-run causality relationships between the variables of this study. Firstly, the GDP Granger caused the CO 2 emissions at 5% level of significance with 1% increase in GDP causing 0.15% increase in CO 2 emissions. This also means that the positive effect of economic growth on CO 2 emissions in the short-run was double that effect in the long-run. Secondly, urbanization Granger caused the economic growth in the long-run in Saudi Arabia at 1% level of significance. An increase of 1% in urbanization caused 2.5% increase in the GDP. Note that urbanization had no significant effect of economic growth in the short-run. Thirdly, the economic growth Granger caused urbanization at 1% level of significance. An increase of 1% in GDP was associated with 0.006% decrease in urbanization in the long-run. This means that economic growth negatively and positively impacted urbanization in the long-run and short-run, respectively. In brief, in the long-run, there were unidirectional Granger causality relationship from economic growth to CO 2 emissions and bidirectional Granger causality Hannan-Quinn information criterion. Table 7. Coefficient and p-value results of the short-run causality.
relationship between economic growth and urbanization in Saudi Arabia during the study period. Figure 2 summarizes the short-and long-run Granger causality relationships between CO 2 emissions, economic growth and urbanization in Saudi Arabia during the period from 1985 to 2019. To validate the EKC hypothesis, the parameters of equations (12) and (13) were estimated using OLS regression to obtain equations (14) and (15) Where * and *** indicate statistical significance at 10% and 1%, respectively. In equation (14), the coefficients of ΔlnGDP and ΔlnGDP2 had positive and negative values, respectively, with both coefficients were statistically significant at 10% level. This outcome provides empirical evidence that the EKC hypothesis was valid in the short-run in Saudi Arabia. In the long-run, the coefficients of lnGDP and lnGDP2 were not statistically significant. Therefore, there is no evidence about the EKC hypothesis validity in the long-run in Saudi Arabia.

Robustness and diagnostics
In this paper, diagnostic tests were conducted to validate the robustness and generalization capability of the cointegration test results. To test for the presence of serial correlation, Breusch-Godfrey Serial Correlation Lagrange Multiplier test was employed [48]. Table 9 shows that the null hypothesis of no serial correlation cannot be rejected for all three ARDL models, which supports the absence of serial correlation. In addition, the residuals of the regression models were tested for heteroskedasticity using Breusch-Pagan-Godfrey Heteroskedasticity test [49] as shown in table 10. Clearly, the results of the test indicate the absence of heteroscedasticity in the models. Lastly, the stability of the models was analyzed using Cumulative Sum (CUSUM) and CUSUM square (CUSUMSQ) tests proposed by Brown et al [50], as shown in figure 3. The plots reveal the stability of the models.

Discussions
The results of this study reveal several findings in Saudi Arabia during the study period. The first finding is that a positive, unidirectional causality existed from economic growth to CO 2 emissions in both the short and long term. An increase of 1% in GDP Granger caused 0.33% increase in CO 2 emissions in the short term and 0.15% in the long term. This indicates that the short-term impact of economic growth on the environment is twice as Figure 2. Summary of the short-run and long-run causality relationships.  Note: *** and ** indicate statistical significance at 1% and 5%, respectively.
critical as the long-term impact. Saudi Arabia is encouraged to take on more sustainable pathways for economic growth. The government's plan to boost the renewable energy contribution in the nation's energy mix from less than 1% to 50% by 2030 is expected to enhance the growth sustainability. This finding is consistent with recent research on Pakistan [11], Kuwait [16], and nine East Asian countries [29]. Other research in 21 EU nations [8], the top 10 emitting countries [9], and 170 countries [34] found a positive, bidirectional association between economic growth and CO 2 emissions. Our findings, however, contradict those of another study, which found a negative correlation between economic development and CO 2 emissions in South Africa [12]. The second finding pertains to the EKC framework which was used to investigate the nonlinear relationship between economic growth and CO 2 emissions. Due to the fact that the coefficients of ΔlnGDP and ΔlnGDP2 terms in equation (14) were statistically significant and had positive and negative values, respectively, the empirical data indicates the existence of an inverted U-shaped relationship in the short term. On the other hand, this relationship could not be verified in the long term as the coefficients of lnGDP and lnGDP2 were not statistically significant. This finding might explain the government policies and initiatives, such as energy subsidy reforms [51] and the Saudi green initiative [52], that have been in effect since 2016 as part of Saudi Vision 2030 [6], with the aim of boosting economic growth while concurrently improving the environment. The third finding describes the relationship between economic growth and urbanization. In the long term, the empirical results revealed an asymmetric bidirectional causal relationship between economic growth and urbanization in Saudi Arabia during the study period. That is, as shown in table 8, growing urbanization by 1% contributed to increased economic growth by 2.5%, while increased economic growth by 1% resulted in lower urbanization by 0.006%. We think that rising industrial production, particularly in the oil sector, was lured by urbanization and that this improved economic growth. Meanwhile, the government was able to build a large network of roads, build schools, and extend the electric grid and healthcare to rural areas thanks to growing economic income, which also helped to slow down the urbanization of the country. This result contrasts with studies in India [25], where economic growth had a unidirectional positive influence on urbanization, and in the UAE [27], where economic growth and urbanization had a bidirectional positive association. In another study involving five South American countries [30], higher urbanization caused slower economic growth, but higher economic growth caused higher urbanization levels. In the short term, economic growth had positive effect on urbanization and there was no statistically significant effect of urbanization of economic growth. Other research have discovered a U-shaped association between economic growth and urbanization in China [26], as well as an inverted U-shaped relationship in ASEAN countries [24]. Finally, there was no statistically significant relationship between CO 2 emissions and urbanization in Saudi Arabia over the study period.
Three important Saudi Arabian sectors are linked in this study. In 1985, there were 13.1 million people, 72.1% of whom lived in cities; in 2019, there were 34.3 million people, 84.1% of whom were urban. The increase in urbanization had contributed positively in increasing the per capita income from 7920 USD in 1985 to 23140 USD in 2019. This remarkable economic growth caused increased CO 2 emissions from 13.1 to 17 metric ton per capita during the same period. The Ministry of Economy and Planning, which aims to achieve sustainable development, need to consider the outcomes of this study to better coordinate the national policies and initiatives with the Ministry of Environment, Water and Agriculture and other concerned authorities in the  urban planning, i.e., the Ministry of Rural Affairs and Housing and the Ministry of Transport and Logistic Services, and in the sectors of economy such as the Ministry of Energy and the Ministry of Industry and Mineral Resources.

Conclusions and policy implications
Saudi Arabia has experienced significant economic growth during the last four decades. This has been accompanied by a steady increase in urbanization. On the other hand, Saudi Arabis faces the difficulty of rising CO 2 emissions, which represents a threat to the environment. In this study, the relationships between CO 2 emissions, economic growth, and urbanization in Saudi Arabia were empirically investigated during the period from 1985 to 2019. This study used a number of econometric techniques to evaluate the causal relationships between these variables. The ARDL bounds test was employed check the existence of cointegration among the variables of interest. The short-and long-run associations were then estimated using ECM and regression analysis.
The empirical findings in this study revealed considerable disparities between long and short-term relationships. In the long term, economic growth positively drives CO 2 emissions and negatively drives urbanization in Saudi Arabia. Meanwhile, urbanization has positive effect on economic growth. Whereas in the short term, economic growth has positive impact on CO 2 emissions and urbanization. Furthermore, the EKC hypothesis is valid in the short term only in Saudi Arabia, i.e., there is an inverted U-shaped causation relationship from economic growth to CO 2 emissions.
Insightful policy recommendations are developed based on the findings of this study. Growing urbanization leads to economic growth in Saudi Arabia. This is in line with Saudi Vision 2030's goals of attracting more people to large cities, particularly Riyadh, and creating environmentally sustainable new cities such as NEOM and Amaala [6]. Furthermore, the oil industry has been the primary driver of economic expansion in Saudi Arabia. Saudi Vision 2030 intends to diversify the economy away from fossil fuel dependence in this regard. Although economic growth Granger causes an increase in CO 2 emissions, the EKC hypothesis's validity in the short term suggests that policies such as fossil fuel subsidy reforms, which were implemented in accordance with the Saudi Vision 2030, have reduced the impact of economic growth on the environment in recent years. The announced initiatives and investments to boost renewable energy production, such as green hydrogen generation, are projected to contribute positively towards achieving economic growth and yet reducing CO 2 emissions. This research also empirically found that urbanization has been steadily rising in Saudi Arabia with no discernible impact on CO 2 emissions. Policymakers are advised to concentrate their efforts on improving the quality of urban expansion by implementing appropriate energy management, transportation, and technology use regulations. Understanding the relationship between the factors of concern is critical to achieving Saudi Vision 2030's objectives. We believe that addressing environmental concerns while maintaining long-term economic and societal growth should be accomplished by tailoring policies to each country's unique conditions.

Data availability statement
The data that support the findings of this study are openly available at the following URL/DOI: https://doi.org/ https://databank.worldbank.org/source/world-development-indicators.