Energy Consumption and Carbon Emissions: An Empirical Study of Saudi Arabia

Abstract

For several decades, Saudi Arabia has depended on fossil fuels for energy consumption in its sectors, which in turn has increased carbon dioxide emissions. Therefore, it is necessary to estimate the effect of energy consumption on the quality of the environment and explore the role of energy-efficient technological innovation. This study uses a structural time series model (STSM) to examine the efficiency of the energy technological innovation role in Saudi Arabia from 1980 to 2019 based on two models. Findings of long-run elasticities estimated in both models indicate that energy consumption impacts carbon dioxide emissions significantly. Also, the underlying energy demand trend (UEDT) evident in both models implies that over the study period, improvement in the efficiency of energy does not exist. But from 2016 onwards, the UEDT showed a downward slope, because the country became interested in energy efficiency and launched a few energy-efficiency policies and programs in the 2010s. Finally, this study highlights some important energy and environmental policies that can help to mitigate carbon emissions. Recognising the role of energy efficiency in environmental quality may help policymakers to act and apply energy efficiency in the industrial sector. Also, Saudi Arabia’s policymakers have to accelerate the enforcement of energy-efficiency programs with mandatory implementation.

1. Introduction

Climate change has become a considerable concern for humanity, and one reason behind rising environmental degradation is the rapid increases in global energy consumption driven by growth in economies and in the population. These have resulted in increased GHG emissions which threaten the planet and life on it. Thus, emissions are reduced when less energy is consumed [1]. Globally, energy efficiency plays a major role in improving energy security and affordability, as well as in speeding the transition to clean energy; policymakers around the world are now placing a great deal of emphasis on it. Because it offers some of the quickest and most affordable ways to mitigate CO₂ emissions while reducing energy costs and boosting energy security, energy efficiency is referred to as the first fuel in clean energy transitions [2].

The economic sector in Saudi Arabia still relies on oil as an energy source. Domestic energy consumption in the country has increased greatly, from 42 million metric tonnes of oil equivalent (MMtoe) in 1980 to 265.33 million (MMtoe) in 2019 [3]. Energy consumption of the industrial sector has increased from 12.4 million (MMtoe) in 1980 to 47.73 million (MMtoe) in 2019 [4]. As for petroleum and other liquid consumption, Saudi Arabia ranked fifth, with 3,649,000 barrels per day (p/d) in 2022 [3]. The environmental issues in Saudi Arabia lie in its economic growth, which is dependent on fossil fuels. Currently, over 99% of the country’s energy mix is made up of fossil fuels; also, the carbon intensity of the energy mix has hardly changed over the previous few decades [5]. Statistics show that emissions of carbon dioxide from combustion fossil fuel grew from 97.9 million tonnes (MT) in 1980 to 579.9 million tonnes (MT) in 2019 (see Figure 2), with a growth rate per annum of 3% during the period 2009 to 2019, compared to the world growth rate of 1.4% during the same period [6]. Industrial emissions in Saudi Arabia account for about half of its direct CO₂ emissions (46%) and are growing at a rapid rate. The emissions intensity of the sector increased by about 10% between 2011 and 2016. Thus, the importance of the study lies in the proposition that, with a “fair share” compatible with a global temperature of 1.5 °C, Saudi Arabia must reduce its emissions to below 389 metric tonnes of carbon dioxide equivalent (MtCO₂e) by 2030 and below 263 MtCO₂e in 2050 to remain within its emissions allowance [5] (p. 1). Saudi Arabia pays attention to environmental concerns, which is why it has ratified several climate agreements. In addition, a number of initiatives have been launched to improve energy efficiency in the country, such as the rationalizing of electricity consumption by the establishment of the Saudi Energy Efficiency Center. In addition to the Saudi Energy Efficiency Program, a national campaign has been launched to raise public awareness through lectures and seminars [7]. Furthermore, since 2011, the energy-efficiency framework has been applied to industrial plants in Saudi Arabia, with an overall target of around 9% from 2010 to 2019, or 1% per year, to improve energy intensity. Also, soft loans are provided to energy efficiency-related projects in industry from the Saudi Industrial Development Fund [5].

The Saudi Energy Efficiency Center in 2010 established a national program to improve and raise energy consumption efficiency in three main sectors (buildings, industry, and road transport). In 2018, the Saudi Energy Efficiency Center was expanded to include the full scope of energy efficiency, including industrial processes, electricity production, transmission, and water desalination. In the future, to ensure the sustainable development of industrial capacity, Saudi Arabia will apply comprehensive regulatory frameworks to more than 180 separate production facilities, moving towards greater energy efficiency in domestic industry [8]. For Saudi Arabia, following a suitable environmental policy aiming to preserve environmental quality and ensure economic growth is considered a great challenge.

This study aims to examine the efficiency of the energy technological innovation role in Saudi Arabia from 1980 to 2019. Therefore, the objectives of this paper are to quantify the impact of energy consumption in general, and the energy consumption of the industrial sector specifically, on environmental quality, in addition to other variables, from 1980 to 2019 in Saudi Arabia. Thus, our hypothesis is that evaluating energy efficiency will contribute to Saudi Arabia’s improved environmental quality. A structural time series model (STSM) will investigate the role of energy-efficient technological innovations in protecting environmental quality. This involves finding out the current level of energy efficiency in Saudi Arabia, to help the country know the level of efficiency needed to preserve environmental quality; this situation is the motivation behind the current study. Thus, the implications of the results of this study will be beneficial to Saudi policymakers in their attempts to reduce their carbon emissions without affecting the economic growth of the state. This research is thought to be the first to use a structural time series model approach to examine how the industrial sector affects carbon dioxide emissions. Especially in this nation, employing this methodology will fill a gap in the existing literature. Determining this nation’s level of energy efficiency will support global efforts to improve environmental quality by reducing CO₂ emissions.

The outline of the paper is as follows: Section 2 presents a literature review. The methodological framework and data used are presented in Section 3. Empirical results from the STSM methodology and related discussion are presented in Section 4. Section 5 contains conclusions and policy implications.

2. Literature Review

Over the past few decades, empirical research has examined the relationships between environmental degradation, energy consumption, industrial sector, GDP, and population. For Shandong Province, Xiaoqing and Jianlan [9] found a stable long-term relationship between industry structure and carbon dioxide emissions. They also concluded that changes in industry structure curb emissions and that technical efficiency is the main factor in reducing CO₂ emissions. Foster and Bedrosyan [10] stated that about 40% of global CO₂ emissions come from the energy sector and the contribution of the industrial sector is about 20% of global emissions. Also, Wang et al. [11] investigated the Granger causality between energy consumption, carbon dioxide emissions, and urbanization for China between 1995 and 2011. The results showed a long-run bi-directional Granger causality between urbanization, energy consumption, and carbon dioxide emissions, and that emissions would increase steadily in the country. Huisingh et al. [12] studied the roles of policy interventions and technical innovations in making progress in energy efficiency in several sectors, including the industrial sector, in initiatives to decrease the emissions of these sectors. They found that implementing low fossil–carbon renewable-energy widely and improving energy efficiency are successful methods which can be used to reduce CO₂ emissions.

Ajmi et al. [13] investigated the Granger causality for the G7 countries using a time-varying method. Their results reveal causal relationships between variables (energy consumption, emissions of carbon, and GDP). Begum et al. [14] used the ARDL technique for Malaysia from 1970 to 2009 to examine the impacts of energy consumption on CO₂ emissions, along with GDP growth and population growth. The results indicate that variables have positive long-run impacts on carbon dioxide emissions, which suggests that carbon dioxide emissions could be influenced by economic growth in the long run. Kasman and Duman [15] investigated a panel of EU countries from 1992 to 2010. Their findings show a long-running cointegrated relationship existing between emissions of carbon, energy consumption, real income, trade openness, and urbanization. Similarly, Wang et al. [16] illustrated a long-running relationship among energy consumption and carbon emissions in China, using data from 1990–2012. Balogh and Jámbor [17] employed a complex model comprising economic growth, industrial structure, and energy use, in addition to other variables, using GMM models on a panel dataset comprising 168 countries from 1990 to 2013. Their results confirmed that energy use increased environmental pollution through rising CO₂ emissions. Recent studies such as Olivier and Peters [18] have stated that in 2018, the demand for primary energy and the energy mix were increased, resulting in an increase in global carbon dioxide emissions. Javid and Khan [19] investigated the relationship between energy consumption and emissions of carbon, using the STSM approach, over the period of 1971 to 2016 for China, the USA, India, Germany, and Japan. Additionally, the estimated method allows underlying energy demand trends (UEDT), which cover technical changes in the capital stock, along with other exogenous factors, in order to capture the technical effects of the environmental regression. The results revealed a positive relationship among the variables for all five countries, and that technological advancements and energy-efficient appliances offer significant means of managing the environmental impact of the economic output.

Recently, Agbede et al. [20] used panel data from 1971 to 2017 for MINT countries, and their findings indicate a long-term relationship between energy consumption and environmental quality. Hernández and Fajardo [21] estimated air pollutant emissions under different assumptions on industrial energy matrices in Bogotá through 2050 under three scenarios. The results showed that under the business-as-usual scenario, carbon emissions would increase by 26.22% by 2050, compared to 2014. Under the carbon reduction scenarios, carbon emissions would decrease by 38.39% in 2050. The last scenario, the mitigation scenario, revealed a decrease in CO₂ emissions of 49.15% during the studied period. Also, Rahman et al. [22] investigated empirically the variables that affected carbon emissions for the BRICS region from 1989 to 2019, using a cointegration method. The results indicate a long-running association between their variables and energy consumption affecting emissions positively and significantly in the BRICS region.

As for Saudi Arabia, many studies have discussed the industrial sector and carbon emissions relationship within the country, such as Omri [23], who used panel data to analyse the relationship among energy consumption, economic activity, and emissions of carbon dioxide empirically for 14 MENA countries from 1990 to 2011. The findings showed that there is a causal relationship between the variables for MENA countries. Alshehry and Belloumi [24] investigated the causal relationships between energy prices, energy consumption, and economic activity. The results suggest a long-running relationship between these variables and carbon dioxide emissions in Saudi Arabia. Also, Alkhathlan and Jaivd [25] analysed the STSM method relative to the effect of oil consumption in Saudi Arabia on environmental quality over the period of 1971 to 2013. The findings showed that oil consumption impacts CO₂ emissions growth positively. Alarenan et al. [26] employed the UEDT approach for industrial energy consumption in Saudi Arabia over the period of 1986 to 2016. A UEDT suggesting improvements in energy efficiency will decrease some of the growth of energy consumption in the industrial sector.

For the Gulf Cooperation Council countries, an empirical study by Atalla and Hunt [27] explored the factors behind the demand of electricity in the residential sector. Their findings reveal that device efficiency has to be raised, along with people’s awareness, in order to decrease electricity consumption. For Saudi Arabia, Atalla et al. [28] examined gasoline demand, using the STSM approach. The results illustrated that because of gasoline price inelasticity, gasoline consumption growth cannot be limited by the government. Therefore, the country needs to raise energy awareness and increase energy efficiency. Alajmi [29] recently studied the impact of electricity generation on carbon emissions using annual data from 1980 to 2017, using the STSM method. The results revealed that in the long run, Saudi Arabia’s CO₂ emissions will continue to rise for decades if electricity generation continues using fossil fuel. In addition, Alajmi [30] investigated the factors influencing GHG emissions in nine sectors, including the industrial sector. The results showed that the energy effect in Saudi Arabia is the leading factor increasing GHG emissions. Another empirical paper found that GDP and energy consumption along with other variables impact on emissions of carbon positively [31]. Recent research on the environmental performance and energy efficiency of Persian Gulf nations from 2000 to 2014 was published by Nikbakht et al. [32]. The United Arab Emirates ranked second in terms of TFEE, while Oman had the lowest rating, according to their findings; Saudi Arabia had the highest TFEE. The Persian Gulf countries may be able to lower their energy use, according to these findings. Alajmi [33] evaluated the Gulf Cooperation Council nations’ energy efficiency from 2000 to 2019. The studies applied Malmquist and data envelopment analysis techniques to estimate the overall factor for the energy efficiency of energy. According to the data, the technological progress index was mostly responsible for the decline in the total factor of energy efficiency in those regions.

This paper is considered the first study to investigate the impact of the industrial sector on carbon dioxide emissions by using a structural time series model approach, particularly for this country. Thus, the assessment of the impact of industrial sector energy consumption on environmental quality and the capturing of the impact of exogenous factors in Saudi Arabia by using this methodology will fill a gap in the current literature. Thus, we assist international initiatives aimed at enhancing environmental quality by assisting policymakers in determining the level of energy efficiency. After that, they will be able to act to increase energy efficiency as a component of a long-term and effective plan for economic development.

3. Methodology

In this paper, structural time series models (STSM) will be used to examine the role of energy efficiency achieved by technological innovation in reducing carbon emissions. Harvey [34] developed an STSM method that accounts for other unobserved factors. The baseline energy demand trend (UEDT) proposed by Hunt et al. [35,36], must be modelled stochastically. The STSM method identifies structural changes at specific dates within the data and interventions that provide information on significant breaks, in addition to capturing the technical effects in the environmental regression equation [19]. STSM allows the UEDT to be modelled stochastically, covering an increase or decrease of energy efficiency relative to energy consumption, along with other exogenous factors [19,37]. In addition, this method gives a better fit and allows for a more accurate prediction. This approach uses explanatory variables with stochastic trends that reflect changes that cannot be measured explicitly [38]. Therefore, the CO₂ emissions determinants’ influence levels differ over time, so they could be captured, along with energy efficiency, by allowing the UEDT to be stochastic [19].

An STSM model can be estimated within the framework of an Autoregressive Distributed Lags Model (ARDL). The ARDL approach can estimate both the long-term and short-term elasticities. The literature shows that the ARDL method can provide more consistent estimates in small samples, compared to other alternative methods. The ARDL approach is valid regardless of whether the regressors are exogenous or endogenous.

In our study, the STSM will be applied to two models. The first model is the model of energy consumption in general: STSM will estimate the long-term relationships between the variables, such as energy consumption, financial development, GDP, population, exogenous factors, and carbon emissions in Saudi Arabia. The second model is the industrial sector model: STSM will estimate the long-term relationships between the variables, namely, industrial sector energy consumption, trade openness, GDP, population, other exogenous factors, and emissions. After estimating the variables, we follow the general-to-specific method to obtain a preferred equation, so we remove insignificant variables from the model and include interventions. Finally, STSM provides an estimation of the UEDT graphically, which captures technological innovation stock along with other exogenous factors, such as environmental policy.

3.1. Energy Consumption Model

Carbon emissions are modelled as a function of GDP (Y) and energy consumption (EC) in seven sectors: industrial, transport, residential, commercial and utilities, agriculture, non-energy and unspecified sectors. Financial development (FD), population (POP) and (UEDT) reflect the influence of exogenous factors. In this paper, STSM will be used, in line with the literature of Javid and Khan [19] and Alajmi [29], to model carbon dioxide emissions as a function of the following equation:

$${CO}_2=f \left(Y, EC, FD, POP, UEDT\right)$$

(1)

where CO₂ is carbon emissions in million tonnes (MT); Y is GDP in billion Saudi Riyals (Rls); EC is energy consumption in million metric tonnes of oil equivalent (MMtoe); FD is financial development as % GDP; POP is represented population; and UEDT a stochastic term considered as an expression of exogenous factors.

We transformed Equation (1) into the dynamic autoregressive distributed lag (ARDL) model for empirical analysis. See, for example, Javid et al. [37], and Alkhathlan and Javid [25]. The dynamic autoregressive distributed lag model (ARDL) was used as follows:

$$\begin{gathered} {co}_t={\alpha }_1{co}_{t-1}+{\alpha }_2{co}_{t-2}+{\beta }_0{y}_t+{\beta }_1{y}_{t-1}+{\beta }_2{y}_{t-2 }+{\gamma }_0{ec}_t+{\gamma }_1{ec}_{t-1} \\ +{\gamma }_2{ec}_{t-2}+{\theta }_0{fd}_t+{\theta }_1{fd}_{t-1}+{\theta }_2{fd}_{t-2}+{\delta }_0{pop}_t \\ +{\delta }_1{pop}_{t-1}+{\delta }_2{pop}_{t-2}+{UEDT}_t+{ԑ}_t \end{gathered}$$

(2)

where ${co}_t$, $y_t$, ${ec}_t$, ${fd}_t$, and ${pop}_t$ are the natural logarithms of CO₂, Y, EC, FD, and POP in year t. The coefficients ${\beta }_0, { \gamma }_0, {\theta }_0, \mathrm{a}\mathrm{n}\mathrm{d} {\delta }_0$ represent the short-run elasticities of CO₂ emissions with respect to corresponding variables in Equation (2). The corresponding long-term elasticity of variables can be represented as $\beta =\frac{{\beta }_0+{\beta }_1+{\beta }_2}{1-{\alpha }_1-{\alpha }_2}$, $\gamma =\frac{{\gamma }_0+{\gamma }_1+{\gamma }_2}{1-{\alpha }_1-{\alpha }_2}$, $\theta =\frac{{\theta }_0+{\theta }_1+{\theta }_2}{1-{\alpha }_1-{\alpha }_2}$, and $\delta =\frac{{\gamma }_0+{\gamma }_1+{\gamma }_2}{1-{\alpha }_1-{\alpha }_2}$, respectively. Based on the Akaike information criterion (AIC), a two-year lag length is used after deleting the insignificant variables from the model, and this includes the interventions.

The UEDT (${\mu }_t$) consists of level and slope components. The ${\mu }_t$ is supposed to have the following process:

$${\mu }_t={\mu }_{t-1}+{\beta }_{t-1}+{\eta }_{t }; {\eta }_t~NID\left(0,{\sigma }_{\eta }^2\right)$$

(3)

$${\beta }_t={\beta }_{t-1}+{\xi }_t ; {\epsilon }_t~NID\left(0,{\sigma }_{\xi }^2\right)$$

(4)

where ${\mu }_t$ is the level of the UEDT and ${\beta }_t$ is the slope of the UEDT. The ${\eta }_t$ and ${\epsilon }_t$ are the uncorrelated mutual white noise disturbance terms, with zero means ${\sigma }_{\eta }^2$ and variances ${\sigma }_{\epsilon }^2$. Based on Harvey and Koopman [39], significant breaks and structural changes can be given by slope, irregular and level interventions can retain the condition of normality of the auxiliary residuals. The UEDT can be written as [19]:

$$\begin{gathered} UEDT={\mu }_t+irregular intervention+level interventions \\ +slope interventions \end{gathered}$$

(5)

3.2. Industrial Sector Model

Carbon emissions for the industrial sector are modelled as a function of

$${CO}_2=f \left(Y, ECi, TO, POP, UEDT\right)$$

(6)

where CO₂ is carbon emissions in million tonnes (MT); Y is gross domestic product (GDP) in billion Saudi Riyals (Rls); ECi is energy consumption by the industrial sector in millions of tonnes of oil equivalent (Mtoe); and TO is trade openness as % GDP. (Trade openness is related to the industrial sector and CO₂ emissions; thus, we added a trade openness variable to Equation (6). Trade openness can improve or destroy the quality of the environment of any country, because that trade openness can facilitate foreign direct investment which can influence the economy and environment [19]. POP is population, and UEDT a stochastic term estimated by the STSM method and expressing exogenous factors.

Equation (7) is estimated using the dynamic autoregressive distributed lag model (ARDL) specification:

$$\begin{gathered} {co}_t={\alpha }_1{co}_{t-1}+{\alpha }_2{co}_{t-2}+{\beta }_0{y}_t+{\beta }_1{y}_{t-1}+{\beta }_2{y}_{t-2 }+{\gamma }_0{eci}_t+{\gamma }_1{eci}_{t-1} \\ +{\gamma }_2{eci}_{t-2}+{\theta }_0{to}_t+{\theta }_1{to}_{t-1}+{\theta }_2{to}_{t-2}+{\delta }_0{pop}_t \\ +{\delta }_1{pop}_{t-1}+{\delta }_2{pop}_{t-2}+{UEDT}_t+{ԑ}_t \end{gathered}$$

(7)

where ${co}_t$, $y_t$, ${ec}_t$, ${to}_t$, and ${pop}_t$ are the natural logarithms of CO₂, Y, ECi, TO, and POP in year t. The coefficients ${\beta }_0, {\gamma }_0, {\theta }_0, \mathrm{a}\mathrm{n}\mathrm{d} {\delta }_0$ represent the short-run elasticities of CO₂ emissions with respect to the corresponding variables in Equation (7). A two-year lag length is chased based on the AIC after dropping the insignificant variables from the model, and includes the interventions. The software package STAMP 8.30 is used to estimate both models.

3.3. The Data

Annual time series data from 1980 to 2019 for Saudi Arabia were used to analyse previous models to estimate the impact of energy consumption on environmental quality. Energy consumption data were obtained from the U.S. Energy Information Administration. Carbon dioxide (CO₂) emissions data were obtained from BP World Energy. Data on GDP, private-sector domestic credit as a percentage of GDP, trade openness as a percentage of GDP, and population were obtained from World Development Indicators. Finally, data on energy consumption in the industrial sector were obtained from the International Energy Agency. Energy consumption in the industrial sector was calculated as the sum of oil, petroleum products, natural gas, and electricity. A moving average method was used to forecast the value of energy consumption and energy consumption of the industrial sector for 2019, as well as financial development for 2018 and 2019.

Figure 1 shows energy consumption in the industrial sector by fuel type from 1980 to 2019. This figure illustrates how natural gas and petroleum product consumption in the industrial sector has changed over the past few decades, highlighting the rapid growth in energy consumption. Figure 2 shows a continuous upward trend in carbon emissions from 1980 to 2019, rising from 97.91 million tonnes in 1980 to 579.9 million tonnes in 2019. These graphs convey that increased energy consumption from conventional fuels will lead to higher emissions, which will negatively affect the quality of the environment. Figure 3 shows population trends developing over the period under consideration, and reaching 34 million [40]. Also, this shows that the GDP trend continuously rises, despite the decline in the oil sector, because of the contribution from non-oil sectors to GDP, which recorded a growth of 3.31% in 2019 [41].

4. Empirical Results and Discussions

4.1. The Energy Consumption Model

Given the 40-year time horizon, our empirical model uses two lags based on Akaike information criteria (AIC). We follow the general-to-specific method to obtain a preferred equation, so we remove insignificant variables from the model and include interventions.

Table 1. Energy consumption models’ estimated coefficients.

Variables	ARDL	t-Test
Estimated coefficients
${\mathit{\beta }}_{\mathbf{0}}$	0.04985	1.88469
${\mathit{\beta }}_{\mathbf{1}}$	-	-
${\mathit{\beta }}_{\mathbf{2}}$	0.07408	2.99736
${\mathit{\gamma }}_{\mathbf{0}}$	0.63176	9.03924
${\mathit{\gamma }}_{\mathbf{1}}$	-	-
${\mathit{\gamma }}_{\mathbf{2}}$	-	-
${\mathit{\theta }}_{\mathbf{0}}$	0.00349	3.98091
${\mathit{\theta }}_{\mathbf{1}}$	-	-
${\mathit{\theta }}_{\mathbf{2}}$	0.00266	3.35057
${\mathit{\delta }}_{\mathbf{0}}$	5.26485	4.20431
${\mathit{\delta }}_{\mathbf{1}}$	−3.99211	−3.39407
${\mathit{\delta }}_{\mathbf{2}}$	-	-
Estimated long-run coefficients
Y	0.12393
EC	0.63176
FD	0.00615
POP	1.27274
Hyper-parameters
Level	1.06743 × 10−5
Slope	5.91568 × 10−6
Irregular	9.19565 × 10−5
Interventions
Level 1994	0.14539	10.83037
Level 1990	0.16136	10.75028
Irregular 1984	0.04833	4.17086
Irregular 1986	−0.06584	−5.28379

shows the results of the final estimation of the model coefficients for carbon. Diagnostic tests of the residuals and auxiliary residuals are presented in

Table 2. Diagnostic tests for the energy consumption model.

Goodness-of-Fit
P.E.V	0.00017517
AIC	−7.9497
${\mathit{R}}^{\mathbf{2}}$	0.99949
${\mathit{R}}_{\mathit{d}}^{\mathbf{2}}$	0.94978
LR test	104.889
Residuals diagnostic
Std Error	0.013235
Normality	1.0058
$\mathit{H}\left(\mathit{h}\right)$	H(8) = 1.1633
r(1)	−0.44086
r(6)	−0.032474
DW	2.5636
$\mathit{Q}(\mathbf{6}\mathbf{,}\mathbf{6}-\mathbf{2})$	9.5187
Auxiliary residuals
Normality–Irregular	0.95349 [0.6208]
Normality–Level	0.030408 [0.9849]
Normality–Slope	3.7725 [0.1516]
Prediction Failure Chi2 (8)	9.5875 [0.2952 ]
CUSUM t(8)	1.0622 [0.3192]

Numbers between brackets are the p values.

. In addition, the resi duals of the estimated equation are checked for non-normality, serial correlation, and heteroscedasticity. In addition, the correctness of the fit is checked, and all diagnostic tests are passed. The stochastic specification of the underlying emissions trend in the data is not rejected (see the LR test in Table 2).

In addition, Table 1 presents the estimated long-term elasticities. The elasticity of energy consumption is (0.63), indicating a positive and statistically significant relationship between carbon emissions and energy consumption. In other words, in the long run, a 1% increase in energy consumption increases Saudi Arabia’s carbon emissions by 0.631. This result indicates that the level of energy consumption can affect carbon emissions. The reason for this result may be the lack of energy efficiency. Also, when comparing electricity prices, we found that the price of electricity in Saudi Arabia is low, compared to other countries [42].

For the long-run elasticities the results of both GDP and financial development indicate that in the long run, they affect carbon emissions in Saudi Arabia positively. When GDP increases by one percent, CO₂ emissions increase by 0.123; also, when financial development increases by one percent, CO₂ emissions increase by 0.006. For population, the findings imply a long-running relationship between emissions of carbon and population, because the elasticity is positive and significant. Thus, carbon emissions increase by 1.27 when the population of Saudi Arabia increases by one percent. This indicates an awareness of environmental issues inadequate to ensure a significant decrease in emissions in Saudi Arabian energy consumption. These results are in line with research concluding that GDP, population, and energy consumption influence carbon emissions, such as that of Alkhathlan and Javid [25], Agbede et al. [20], Alajmi [29], Olivier and Peters [18], and Javid and Khan [19].

An estimated UEDT which captures technological innovation stock, along with other exogenous factors such as environmental policy, is provided by the estimated STSM model. Also, the STSM method identifies interventions that give information on structural changes at certain dates in the series data and as to significant breaks.

Figure 4 shows that the UEDT increases from 1980 to 2016, indicating the need to protect environmental quality by improving energy efficiency or reforming energy-efficiency regulations with an accelerated transition to a low-carbon circular economy. Since 2016, however, the UEDT has been trending downward, which may be due to the Saudi government showing more interest in energy efficiency in the 2010s. In the 2010s, Saudi Arabia implemented several energy-efficiency policies and launched several programs of the Saudi Energy Efficiency Center.

4.2. Industrial Sector Model

Table 3. Industrial sector models’ estimated coefficients.

Variables	ADEL	t-Test
Estimated coefficients
${\mathit{\beta }}_{\mathbf{0}}$	0.18393	3.85297
${\mathit{\beta }}_{\mathbf{1}}$	−0.11726	−2.64462
${\mathit{\beta }}_{\mathbf{2}}$	-	-
${\mathit{\gamma }}_{\mathbf{0}}$	-	-
${\mathit{\gamma }}_{\mathbf{1}}$	0.05686	3.35361
${\mathit{\gamma }}_{\mathbf{2}}$	-	-
${\mathit{\theta }}_{\mathbf{0}}$	−0.00375	−4.04908
${\mathit{\theta }}_{\mathbf{1}}$	0.00233	2.21297
${\mathit{\theta }}_{\mathbf{2}}$	0.00219	3.29467
${\mathit{\delta }}_{\mathbf{0}}$	3.89899	6.52494
${\mathit{\delta }}_{\mathbf{1}}$	-	-
${\mathit{\delta }}_{\mathbf{2}}$	−1.74207	−3.75680
Estimated long-run coefficients
Y	0.06667
EC	0.05686
TO	0.00077
POP	2.15692
Hyper-parameters
Level	1.97550 × 10−5
Slope	0.000000
Irregular	0.000276479
Interventions
Irregular 1989	−0.09751	−5.06129
Irregular 1993	−0.10652	−5.49399
Irregular 1992	−0.06809	−3.52110

presents the final estimated results of model coefficients in detail with two lags, and diagnostic tests for residuals and auxiliary residuals are presented in

Table 4. Diagnostic tests for industrial sector model.

Goodness-of-Fit
P.E.V	0.00027884
AIC	−7.4849
${\mathit{R}}^{\mathbf{2}}$	0.99919
${\mathit{R}}_{\mathit{d}}^{\mathbf{2}}$	0.92006
LR test	88.8296
Residuals diagnostic
Std Error	0.016699
Normality	1.0550
$\mathit{H}\left(\mathit{h}\right)$	H(8) = 0.54340
r(1)	0.048811
r(6)	0.23280
DW	1.8275
$\mathit{Q}(\mathbf{6}\mathbf{,}\mathbf{6}-\mathbf{2})$	8.6298
Auxiliary Residuals
Normality–Irregular	0.57613 [0.7497]
Normality–Level	0.27699 [0.8707]
Normality–Slope	6.113 [0.0471]
Prediction Failure Chi2 (8)	6.9708 [0.5398]
CUSUM t(8)	−0.9775 [1.6431]

Numbers between brackets are the p values.

.

Table 3 reports estimated long-run elasticities for second model variables for Saudi Arabia. The energy consumption of industrial sector’s elasticity is (0.056), which indicates a positive and statistically significant relationship between emissions of carbon and the energy consumption of the industrial sector in the country. In other words, in the long run, with a 1% increase in the energy consumption of the industrial sector, emissions of carbon increase by 0.056. Also, the long-run elasticities of both GDP and trade openness are found to be positive and statistically significant, which means there is a long-running relationship between these variables and CO₂ emissions. The long-run estimated elasticity values of GDP and trade openness in Saudi Arabia are 0.066 and 0.0007, respectively. Population long-run elasticity is positive and significant, which indicates the existence of a relationship between carbon emissions and population. In other words, carbon emissions increase by 2.156 when the population of Saudi Arabia increases by one percent. The study’s findings are consistent with works in the literature which have found relationships between carbon emissions, GDP, population, and energy consumption, such as Alkhathlan and Javid [25], Olivier and Peters [18], Javid and Khan [19], Alajmi [29], and Agbede et al. [20]. Also, the model passes all the diagnostic tests (see Table 4).

The UEDT values for these two models, a determination which reflects how energy efficiency and exogenous factors such as environmental policies affect CO₂ emissions, were upward until 2016. But since 2016, the UEDT has had a downward slope (see Figure 5). This may be due to the fact that the Saudi government has implemented a few energy-efficiency policies and energy awareness campaigns.

For both models, an upward or downward trend of the UEDT would indicate that if both models’ variables (GDP, trade openness, financial development, and population) were fixed between 1980 and 2019, then CO₂ emissions would rise or decline because of the existence of exogenous factors. Using the findings of both models, policymakers could reform energy regulations and consider environmental protocols, in addition to demanding strict environmental regulation related to energy efficiency, along with efforts accelerating the transition to a low-carbon circular economy [30].

4.3. Discussion of Findings

The economic growth of Saudi Arabia increased rapidly in the long run during 1980–2019. This growth required more energy consumption, which caused dramatically increased carbon emissions year on year.

The results of this study show that there are long-run relationships between the variables in both models, which means that they contribute to carbon emissions. Naturally, the contribution of energy consumption to emissions is positive due to the fact that (1) Saudi Arabia’s energy mix is dominated by oil, which will remain unchanged for a long time, and (2) rapid industrial development and urbanization have led to high energy consumption, which in turn increases carbon emissions. Therefore, improving energy efficiency is an important policy for the reduction of emissions. Saudi initiatives will play a role in improving energy efficiency, and a national environmental strategy will help to reduce the growth of carbon emissions in the long term [30]. The empirical results also showed that GDP, as well as financial development and trade openness, play significant roles in increasing carbon emissions. Despite the national strategy to diversify the revenue base, GDP growth between 1990 and 2019 was driven by the oil sector, reflecting Saudi Arabia’s heavy reliance on oil revenues.

The results also show that, in the long run, population size affects carbon emissions, which means that population intensity is also an important factor, because increasing population size increases energy consumption. Thus, Saudi policymakers should pay more attention to increasing population awareness and energy efficiency in these two sectors. Since cities do not yet have public transportation, people use cars, which means increased energy consumption. In addition, Saudi Arabia has an abundance of fossil fuels, so energy companies supply energy at low prices, which explains why the price of electricity in the country is lower than in many other countries [42].

Therefore, improving the efficiency of energy consumption is important to reduce Saudi Arabia’s rapid growth in carbon emissions, especially in the energy-intensive industrial sector. In order to reach the path to low-carbon growth, it is necessary to effectively address the effects of emissions by developing and implementing a long-term energy policy [16]. However, determining each sector’s responsibility for emissions growth and allocating carbon dioxide reductions across sectors are seen as challenges facing Saudi Arabia. Furthermore, it is important for Saudi Arabia to focus on renewable energy sources, along with improving current energy efficiency. Thus, in order to make the climate carbon-free and combat carbon dioxide emissions, it is necessary to switch from non-renewable energy sources to renewable energy sources, along with afforestation [43].

5. Conclusions and Policy Implications

The purpose of this study is to evaluate the energy efficiency in Saudi Arabia, using annual data from 1980 to 2019. Thus, this paper estimates the impact of energy consumption on environmental quality by employing a structural time series model (STSM). This methodology allows stochastic modelling of the underlying energy demand trend (UEDT) in order to capture energy-efficient technological innovations and exogenous factors such as environmental policies.

To achieve the objectives of this paper, the work sought to quantify the impact of energy consumption in general, and the energy consumption of the industrial sector specifically, on environmental quality. A structural time series model (STSM) was employed to explore the role of energy-efficient technological innovations in the protection of environmental quality. The findings reveal that the long-run estimated elasticities of both models suggest that carbon emissions from energy consumption will continue to rise in the coming decades in Saudi Arabia. Also, the UEDT result, which would indicate an upward or downward trend, means that CO₂ emissions would rise or decline because of the existence of exogenous factors.

The Saudi government has started energy programs addressing the efficiency of energy, such as hydrocarbon demand sustainability and circular carbon economy programs [8], which could reduce carbon dioxide emissions in the future. Based on our results, the country needs to follow certain policies to raise energy efficiency: (1) Raising energy efficiency in energy consumption is a reasonable solution for Saudi Arabia. Thus, for example, policymakers should adjust the structure of energy consumption, reduce unnecessary energy waste, and increase the share of cleaner energy consumption to reduce CO₂ emissions [13]. (2) Also, investment in R&D for energy efficiency should be expanded to explore methods aimed at reducing the demand for energy, and its associated emissions [44,45]. Saudi Arabia needs to invest in clean energy, such as solar and wind power, and adopt energy-efficiency measures that will ensure the sustainable growth of the country’s economy [24]. (3) Also, upgrading the industrial structure can reduce carbon emissions significantly [46]. Finally, these findings are beneficial to Saudi policymakers and for oil-based economies, such as the GCC countries, that aim to reduce their carbon emissions. In sum, energy-efficiency programs could reduce the effects of energy consumption on environmental quality and mitigate climate change, and thereby enhance economic sustainability.

Even though this paper adds contributions to the existing literature, there are limitations, such as the fact that there are no available reliable data for periods before 1980. Thus, the sample size used in this study is limited by the number of observations available. Also, this study only focuses on energy consumption, GDP, private-sector domestic credit, trade openness, and population, as to their impacts on CO₂ emissions. Thus, there are other independent variables that could be added to the model and that might impact the results. Further studies could estimate the impact of environmental policies on environmental quality by employing different scenarios to predict the effectiveness of environmental policies in reducing carbon dioxide emissions. For example, future researchers could evaluate the effects of a wide range of energy-consumption abatement policies on economic growth and CO₂ emissions, as part of initiatives consistent with the Paris Agreement. Also, future research efforts might want to investigate our model in different countries, such as those in GCC regions, or estimate the model with new independent variables.