Analysis and Projection of Transport Sector Demand for Energy and Carbon Emission: An Application of the Grey Model in Pakistan

Abstract

The incredible increase in carbon emissions is a major global concern. Thus, academicians and policymakers at COP26 are continuously urging to devise strategies to reduce carbon and other greenhouse gas emissions. The transportation sector is a major contributor to greenhouse gas emissions in developing countries. Therefore, this study projected an increase in fossil fuel demand for transportation and corresponding carbon dioxide emission in Pakistan from 2018 to 2030 by employing the Grey model and using annual data from 2010 to 2018. Furthermore, the determinant of fossil fuel demand is modeled using an environmental sustainability model such as stochastic regression IPAT that links environmental impact as a product of population, affluence, and technology on annual time series data spanning from 1990 to 2019. The projected values of oil demand and carbon emissions reveal an increasing trend, with average annual growth rates of 12.68% and 11.45%, respectively. The fully modified ordinary least squares (FM-OLS) findings confirmed the environmental Kuznets hypothesis. The increase in population growth emerged as the major driver for oil demand and carbon dioxide emissions, while technological advancement can reduce oil demand and corresponding carbon emissions. This study urges Pakistan to switch from oil to gas and other green energies by encouraging hybrid vehicles, as the number of vehicles on the road positively impacts the transport sector’s oil demand. Moreover, increasing economic growth and controlling the population growth rate by discouraging more children can be a valid policy for reducing oil demand and corresponding carbon emissions.

1. Introduction

Environmental degradation due to greenhouse gas emissions poses a grave threat to the ecology and climate of the Earth. Therefore, the policymakers at COPE 26 (COPE26 stands for Conference of the Parties 26 Summit) and academicians are advocating for sustainable policies to reduce dependence on fossil fuels. The transport sector is a major contributor to greenhouse gas emissions, especially in developing countries. Furthermore, rapid urbanization, an increase in per capita income, and the population growth rate have considerably increased the demand for transportation. Petroleum and other liquid fossil fuels are energy sources for the transportation sector. According to recent projections of the United States (US) Energy Information Administration, the transportation sector’s consumption of liquid fuel grows by 36 quadrillions Btu (British thermal unit), with a diesel consumption of 13 quadrillion Btu, a jet fuel consumption of 10 quadrillion Btu, and Moto gasoline consumption of 9 quadrillion Btu, Reference [1].

Furthermore, the total energy demand of road transport accounts for almost 75% of total transportation sector energy consumption [2,3]. Hence, the transportation sector is one major consumer of fossil fuel energy and corresponding carbon dioxide (CO₂) emissions. Global CO₂ emissions from fuel combustion reached 32.8 billion tons in 2017, while carbon emissions from the transportation sector witnessed an annual growth of 2% from 2000 to 2017. According to [1], carbon emission from road transportation has increased by +1.7 GtCO₂ (GtCO₂ stands for one billion tons of carbon dioxide). Hence, massive energy consumption with accelerated urban activities, without addressing environmental sustainability, puts pressure on energy security and generates a negative externality in climate change [2,3]. Therefore, the sustainable planning of transportation policy along with the development of transport infrastructure can be instrumental to achieving Sustainable Development Goals 11 (sustainable cities and communities), 13 (climate actions), and 7 (affordable and clean energy).

Pakistan’s transportation sector depends on non-renewable energy, which consumes 80% of liquid fuel. In recent years, especially after 2010, Pakistan has witnessed phenomenal growth in the demand for vehicles and oil energy. In contrast, gas energy consumption has fluctuated with declining trends (Figure 1). The total registered vehicles on the road were 11,490 thousand in 2010, which has increased to 25,000 thousand in 2017. Similarly, the oil energy demand for road transport has increased from 8.89 million tons in 2010 to 16.05 million tons in 2017. Whereas, in these years, the transportation sector consumed 70,455 mm cubic feet (mm cft) and 113,055 mm cft CNG (CNG stands for Compressed Natural Gas), respectively [4]. Road transportation is the most cost-efficient medium of transport, which has a positive impact on the process of economic growth and development (https://transportgeography.org/contents/chapter3/transportation-and-economic-development/ and https://www.finance.gov.pk/survey/chapters/14-Transport%20final08.pdf, accessed on 10 January 2023). Pakistan is facing high demand for road infrastructure to meet the transportation need of the increasing population and growing economy. The govt. of Pakistan has recently constructed forty-seven national highways, expressways, and motorways, with a cumulative length of road 12,743 km under the umbrella of the National Highway Authority, and launched thirty-eight new transport projects with an estimated cost of 1104 million dollars under the Pakistan Social Development Program in 2018. Further, the China–Pakistan Economic Corridor project of the Belt and Road Initiative, with an estimated cost of more than 28.5 billion dollars, has also boosted trade activities. These projects are expected to enhance road transportation drastically, increasing the demand for liquid energy.

Climatic changes associated with rising average annual temperature and changing intensity and pattern of rainfall are posing a serious threat to agricultural productivity and food security [5,6,7,8,9]. Pakistan is witnessing an exponential increase in average annual temperature, depleting frozen water reservoirs at the Karakorum and Himalayan ranges, which urges long-term plans for energy needs and associated pollution. Many models have been used in the literature to address the factors responsible for energy demand and environmental degradation. The most important model that addresses the behavior of environmental degradation is the (IPAT) model of Reference [10]. According to the IPAT model, the increasing population, per capita fuel consumption, and the use of traditional fluid fuel-based automobiles are major determinants of CO₂ and other greenhouse gas emissions. The carbon emissions per capita scenario in Pakistan has exhibited an increasing trend in recent years as 0.72 metric tons of carbon per capita were emitted in 2010 which has increased to 0.85 metric tons per capita in 2019 (https://www.worldometers.info/co2-emissions/pakistan-co2-emissions/, accessed on 10 January 2023). The review of empirical studies on energy and environmental degradation has found a plethora of literature on the effect of climatic changes on economic aggregates and other social and health indicators. This study does not find any reliable empirical study on Pakistan that has modeled energy demand and carbon emissions from the transportation sector.

Road transport is a fossil fuel-intensive sector; improving and expanding the road network will increase non-renewable liquid energy demand, consequently deteriorating environmental quality. Hence, analyzing and projecting energy demand and carbon emissions from the transport sector is important for developing an environmentally sustainable energy policy for urban–rural transport planning. With the increase in population growth and per capita income, all modes of road transport have accelerated considerably, resulting in a dramatic surge in oil demand and unprecedented carbon emissions. There are three categories of the model to forecast future trends. First, statistical analysis that uses regression analysis such as econometric regression, logistic regression, Markov chain model, spatial-temporal model, and Kalman filter model to predict energy consumption, Reference [11]. The statistical prediction models require a large sample and the assumption of normality, constant variance, and stationarity. Most economic and energy data needs to maintain the assumptions of statistical analysis models. Second, computational intelligence models use machine learning and artificial neural network analysis. These models are better suited when the sample data are at least thirty observations. Third, the Grey model can provide efficient projected future values without imposing the above-discussed limitations of other models. Given the limitations of our sample size and the nature of the data, this study employs the Grey model to project future values of energy demand and carbon dioxide emission from 2018 to 2030 based on actual data from 2010 to 2018.

This study then explores projected values from 2019 to 2030, subjected to econometric analysis by incorporating explanatory variables suggested by the environmental Kuznets hypothesis and the STIRPAT model. In doing so, this study makes the following contribution to the literature. First, this study is a preliminary attempt to explore the projected future value of energy demand of the transportation sector in Pakistan along with corresponding carbon dioxide emissions. Second, the determinant of Pakistan’s transport sector energy demand is modeled to devise policies for sustainable development of the transportation sector.

The rest of this study is organized as follows: Section 2 reviews the literature; Section 3 discusses the methodological framework; Section 4 analyzes estimated results; and Section 5 concludes the study with policy implications.

2. Literature Review

Climate change seriously threatens humanity and biodiversity [12]. The major determinant of environmental degradation is burning fossil fuels for various purposes, including transportation. This section reviews some selected literature on energy demand and environmental degradation. Many studies have examined the effect of energy demand and other instrumental variables on environmental quality in developed and developing countries. Most studies used greenhouse gases, deforestation, air and water pollution, carbon emission, and ecological footprints as proxies for environmental quality. They have obtained different outcomes by employing different econometric models and datasets.

Concerning the projection of carbon emissions, Reference [13] projected the intensity of carbon emissions and estimated the influencing factor’s effect on China’s energy demand. The Grey and the STIRPAT models have been used for the projection and factors’ effects on energy, respectively. The result suggests that under current policies, China can reduce carbon intensity in 2020 and 2030. Economic growth, population, and time trend have a strong positive effect on energy consumption. Wu et al. (2015) used a novel multi-variable Grey model to predict carbon emissions for the BRICS nations (Brazil, Russia, India, China, and South Africa). The research confirms that between 2015 and 2020, South Africa and the Russian Federation will have the lowest CO₂ emission growth rates, while China will have the highest growth rates. Reference [14] addressed whether China will achieve its carbon emission peak. While using optimal control and the STIRPAT model. The findings suggest that the estimated peak in carbon emissions, expected to be at 117.70 MtCO2e, may occur in 2028. For China to meet the goals outlined in the Paris Agreement by 2030, based on empirical findings, the carbon price and investment in carbon mitigation must be higher than US$32.5/tCO₂e and US$57,370,000/year, respectively. Reference [15] used a novel multi-variable Grey model to predict carbon emissions for the BRICS nations (Brazil, Russia, India, China, and South Africa). The research confirms that between 2015 and 2020, South Africa and the Russian Federation will have the lowest CO2 emission growth rates, while China will have the highest growth rates.

Concerned with factors affecting energy consumption, Reference [16] investigated the role of renewable and urbanization in transportation carbon emissions for European countries while applying panel data from 1980 to 2014. They obtained the validity of the environmental Kuznets hypothesis and the negative impact of renewable energy consumption application in transportation on carbon emissions. In addition, unidirectional causality is obtained from renewable energy consumption, economic growth, and urbanization to carbon emissions from transportation. Reference [17] examined the renewable energy assessment for Turkey while using energy modeling for the period 2014–2050. They found that for the achievement of a sustainable future in Turkey, the social cost of adoption of renewable energy in the reference scenario will be $88.75 billion, in the alternative scenario (I) ($76.73 billion), and in the alternative scenario (II), ($71.50 billion). In addition, prices of different energy choices will increase in Turkey. Reference [18] explored the causality among energy intensity, carbon emissions, and renewable energy on economic growth for Romania while using the Toda–Yamamoto model. They obtained feedback causality between energy intensity and economic growth while confirming uni-causality from renewable energy consumption to economic growth. In addition, economic growth has a positive relationship with renewable energy consumption and carbon emissions. Reference [19] revealed that energy demand is one of the factors responsible for carbon emissions by employing generalized and system-generalized movement methods. Reference [20] estimated the linkages between energy intensity and carbon emissions in countries with the Belt and Road Initiative. The findings confirmed a positive link between energy intensity and carbon emissions. Reference [2] explored the effect of public and private investment on transportation-related carbon emissions in the case of China. The findings support that green energy for transportation enhances environmental sustainability by replacing carbon-intensive energy with green energy. Other relevant studies, for example, Reference [21] for China; Reference [22] for India and China; Reference [23] for Saudi Arabia investigated the impact of human capital, affluence, population, technology, urbanization, number of registered vehicles, fuel price, transport intensity, energy intensity on transport sector energy consumption in China, India, Saudi Arabia, while employing IPAT, STIRPAT, VECM, and LMDI models. Using the best available research, none of the studies have examined Pakistan’s transportation sector’s energy demand using the STIRPAT model.

Reference [24] revealed the nexus of driving forces of carbon emissions while applying the Logarithmic Mean Divisia Index (LMDI) decomposition approach. The main driving force that leads to carbon emissions is energy demand, while a limited role is observed by the energy mix used in the case of Iran. On the other hand, Reference [25] examined the impact of carbon emissions and economic growth on energy consumption through the Gaussian mixture model (GMM), covering the period 1990–2012. Their results supported the positive impact of carbon emissions and economic growth on energy demand. Reference [26] estimated the short-run and the long-run relationship between carbon emissions and their driving forces: economic growth and energy consumption. The study investigated the Pakistan economy, employing the autoregressive distributed lag (ARDL) model from 1965 to 2015. The results confirmed that economic growth and energy demand have short-run and long-run relationships with the carbon emissions which call for renewable energy use.

Furthermore, Reference [27] estimated the dynamic causality between Pakistan’s carbon emissions, energy demand, and economic growth. The ARDL and Vector error correction model (VECM) approaches have been applied to investigate the short-run and the long-run causality. The bidirectional causality holds between economic growth, energy demand, and carbon emissions. The results call for renewable energy use to mitigate carbon emissions. The present literature supports the argument that energy demand deteriorates the environmental quality and supports the implementation of environmentally friendly energy use. Studies have also investigated the projection of carbon emissions and their intensity.

The forecasting of energy demand has been investigated through long-range energy alternative planning (LEAP) systems, autoregressive integrated moving average (ARIMA), and Holt-Winters (HW) models in Pakistan. Reference [28] forecasted the carbon emissions of energy demand in Pakistan and revealed that business activities will promote energy demand and deteriorate environmental quality. It was found that there is an increasing trend in carbon emissions. Under the scenario, there is greater attention toward environmental sustainability, which will reduce carbon emissions from energy demand in the projected period. Reference [29] predicted electricity demand for Pakistan using the ARIMA model. The estimated results revealed that electricity demand is increasing in different sectors. Reference [30] anticipated Pakistan’s energy demand and carbon emissions through the ARIMA model and found that an energy shortfall will be built in the projected years. Reference [31] employed the HW and ARIMA models to project the electricity consumption in Pakistan, employing the time series data from 1980 to 2011, and projected an increasing trend in electricity consumption and reported that households will have relatively more electricity consumption from other sectors. Reference [32] employed the LEAP system to forecast the energy demand of Pakistan and found an increasing trend till 2020, and that the imported energy bill will increase. From the relevant literature, Reference [33] investigated the impact of transport sector energy consumption on carbon emissions in the case of Pakistan. One of the contributions of this research is to fill the extant literature through the empirical investigation of the factors influencing Pakistan’s transport sector oil demand using the STIRPAT model.

Similarly, this study is motivated by the cited literature because there is no previous research in Pakistan to project transport section energy demand and associated carbon emissions while employing the Grey approach. The analysis and projection of energy demand and carbon emissions from the transportation sector are crucial for the creation of an environmentally friendly energy policy for the planning of urban and rural transportation, as well as for the accomplishment of the Sustainable Development Goals, making this the second contribution of this study to the existing literature. With the increase in population growth and per capita income, all modes of road transportation have accelerated considerably, resulting in a dramatic surge in oil demand and unprecedented carbon emissions. Moreover, the econometric inquiry of the environmental Kuznets’ curve, IPAT model, and SRIPAT model will check the validity of these propositions and assist policymakers in devising sustainable energy and environmental policies. The major research hypotheses of this study are as follows:

Hypothesis 1 (H1). 

Environmental Kuznets’ curve hypothesis prevails in the transport sector energy demand.

Hypothesis 2 (H2). 

The increase in population can contribute to the energy demand of the transport sector.

Hypothesis 3 (H3). 

An increase in GDP per capita can contribute to the energy demand of the transport sector.

Hypothesis 4 (H4). 

Technology can influence the energy demand of the transport sector.

3. Material and Methods

This section first discusses the estimation strategy used to explore projected energy demand and carbon emission values, followed by an econometric technique to validate the environmental Kuznets’ curve and STIRPAT model.

3.1. Grey Modeling Approach

The Grey modeling approach as introduced by the Reference [34] to control problems of grey systems, where the system is applicable for partially known and partially unknown datasets. This model has the advantage of obtaining valuable predictions. The name of the grey system was chosen because the projection of an event’s future is based on available data, which is known, while the projection of the event in the future is unknown. This model only uses the projection, while the STIRPAT model uses the factors affecting the dependent variable. Therefore, to achieve our two-fold objectives, projection and factors affecting transport sector energy demand, the Grey and STIRPAT Models are applied, respectively. The supporting argument for applying the Grey model is that the forecasting and projection models are divided into three main categories in the extant literature. First, there are statistical analysis models where regression analysis regression, econometric regression, logistic regression, Markov chain model, spatial-temporal model, and Kalman filter model are used to predict energy consumption. The statistical prediction models have limitations on the requirement of sufficient samples, the assumption of normality, constant variance, and assumption of stationarity. Most economic and energy data fail to maintain the assumptions of statistical analysis models. Second, computational intelligence models are employed for forecasting, with support vector machine regressions and artificial neural networks being the most notable models in this area. These projection models are applied whenever the sample data are not less than thirty observations. It implies that these models fail whenever the sample is less than thirty. In contrast, this assumption often fails to be maintained in the case of energy data because of the unavailability of a large sample size. Third, the Grey Modeling approach, proposed by Reference [34], is an important tool for exploring projected values of any variable, which is widely used in empirical research to estimate energy demand and carbon emissions in developed and developing countries, References [11,13,35,36,37,38] given the limitations on the fulfillment of some statistical distributions and the relatively small size of the sequence of energy consumption in the transportation sector. Therefore, statistical and computational intelligence models are inappropriate for forecasting energy consumption in the transportation sector. This approach is better suited to time series observations with either an increasing or decreasing trend. Let the series ${x^0}_{\left(1\right)},{x^0}_{\left(2\right)},{x^0}_{\left(3\right)},\dots \dots \dots \dots \dots \dots .,{x^0}_{\left(n\right)}$ be non-negative time series under the constraint. Furthermore, $X^0$ is a column vector of the non-negative time series, which can be expressed as follows:

$$X^0=\left[\begin{array}{c}\begin{array}{c}{x^0}_{\left(1\right)}\\ {x^0}_{\left(2\right)}\\ {x^0}_{\left(3\right)}\end{array}\\ .\\ .\\ .\\ .\\ {x^0}_{\left(n\right)}\end{array}\right]$$

(1)

Now, apply the accumulating generating operation (AGO) to the raw data with the characteristic of the chaotic for constructing new series with the characteristic of monotonically increasing. Then, the inverse accumulating generating operation (IAGO) to the new fitting data is applied to project the series. Algebraic notations are presented as follows:

$$X^1=\left\{{x^0}_{\left(1\right)},{x^0}_{\left(1\right)}+{x^0}_{\left(2\right)}+,\dots ,{x^0}_{\left(1\right)}+{x^0}_{\left(2\right)}+{x^0}_{\left(3\right)}+\cdots ,\dots ..+{x^0}_{\left(n\right)}\right\}$$

(2)

$$X^1=\left\{{x^1}_{\left(1\right)},{x^1}_{\left(2\right)},{x^1}_{\left(3\right)},\dots \dots \dots \dots \dots {x^1}_{\left(m\right)}\right\}$$

(3)

where ${x^1}_{\left(2\right)}=\sum _{i=1}^2{x^0}_{\left(i\right)},{x^1}_{\left(3\right)}=\sum _{i=1}^3{x^0}_{\left(i\right)},and {x^1}_{\left(m\right)}=\sum _{i=1}^n{x^0}_{\left(i\right)}$ is a series of the AGO.

The second step is constructing the Grey model GM (1, 1). The first 1 stands for the first order differential while the second 1 stands for the one variable Xu et al., (2017) [13]:

$$\frac{{dx}^{\left(0\right)}}{dt}+a{z^1}_{\left(k\right)}=b$$

(4)

where

$${z^1}_{\left(k\right)}=\lambda {x^1}_{\left(k\right)}+\left(1-\lambda \right){x^1}_{\left(k-1\right)}$$

(5)

In Equation (4), $k$ stands for the time that starts from $k=1,2,3,\dots .,n$. a is called the development coefficient, and b is called the deriving coefficient, Reference [39]. The $\lambda$ in Equation (5) represents the equal probability, i.e., $\lambda =0.5$. The values of the development and derivatives coefficients are estimated through the ordinary least square estimation technique as

$$\left[\begin{array}{c}a\\ b\end{array}\right]={\left(A^{t}A\right)}^{-1}{A}^{t}X$$

(6)

where

$$A=\left[\begin{array}{c}\begin{array}{c}-{z^1}_{\left(1\right)}\\ -{z^1}_{\left(2\right)}\\ .\end{array}\\ .\\ -{z^1}_{\left(n\right)}\end{array}\begin{array}{c}\begin{array}{c}1\\ 1\\ .\end{array}\\ .\\ 1\end{array}\right] and X=\left[\begin{array}{c}{x^0}_{\left(2\right)}\\ {x^0}_{\left(3\right)}\\ \begin{array}{c}.\\ .\\ {x^0}_{\left(n\right)}\end{array}\end{array}\right]$$

The projected accumulating data can be obtained by applying the following AGO estimated equation as

$${{\widehat{x}}^{\left(1\right)}}_{(k-1)}=\frac{b}{a}+\frac{\left[{x^0}_{\left(1\right)}-\frac{b}{a}\right]}{e^{ak}}$$

(7)

The projected data for each period can be extracted by applying the IAGO estimated technique as

$${{\widehat{x}}^{\left(0\right)}}_{\left(k\right)}={{\widehat{x}}^{\left(1\right)}}_{\left(k\right)}-{{\widehat{x}}^{\left(1\right)}}_{\left(k-1\right)}$$

(8)

The projection of oil demand of the transportion sector of Pakistan for the period 2019–2030 is based on the oil demand sample data from 2010 to 2018. The data were obtained from the Pakistan Economic Survey 2018–2019. The ${x^0}_{\left(1\right)}$ is oil consumption in the transport sector in 2010, and ${x^0}_{\left(m\right)}$ is the oil demand in 2018.

3.2. Pakistan’s Transport Energy Demand Scenario

The oil demand of the transport sector, as presented in Figure 1, is exhibiting an increasing trend, whereas the demand for gas is a decreasing trend from 2010 to 2018. The decreasing trend in demand for gas can be due to the shortage of gas in stations resulting in the switching vehicles from gas to oil. In the past eight years, the share of the transportation sector’s oil demand has increased from 47.08% to 65.02%; the gas demand share has reduced from 9.11% in 2010 to 4.84% in 2017. The Government of Pakistan has taken several initiatives to increase the share of renewable energy generation resources to improve environmental quality. The increasing trend of transport sector oil demand will put pressure on the oil energy supply and consequently will not improve environmental quality.

3.3. The Environmental Cost of Oil Energy Demand

The IPAT model presented by Reference [10] is the first model that explores the effect of human activities on the environment. Following Reference [40], I is the resource consumption that depends on population (P), affluence (A), and time (T) period. In the literature, Gross Domestic Product (GDP) is usually used to indicate affluence. According to this model, the impact of human activities (I) is equal to the product of three factors: population (pop), affluence (A), and technology (T). The stochastic impact of regression (STIR) was incorporated by Reference [41] and transformed IPAT into the STIRPAT model. This model can be expressed as

$$I=\alpha P^{\beta }{A}^{\gamma }{T}^{\delta }{e}^u$$

(9)

In Equation (9), the variables I, P, A, and T have the same description as explained in the IPAT model; the $\alpha ,\beta ,\gamma \mathrm{a}\mathrm{n}\mathrm{d} \delta$ are the parameters, and $u$ is a random term. The logarithmic transformation of the STIRPAT model can be expressed as

$$lnI=a+\beta lnP+\gamma lnA+\delta lnT+u$$

(10)

This study employs the projected oil demand of the transport sector (oil) as a proxy for human influence on the environment (I). The empirical literature has used various proxy variables for affluence and technology. This study has used the gross domestic product (GDP) as a proxy for affluence (A) and time variable as a proxy for technology (T), Reference [13]. The basic rationale for incorporating the gross domestic product in the model is to explore the existence of the environmental Kuznets’ curve (EKC) hypothesis. The EKC hypothesis argues that the initial stage of economic growth will be detrimental to the environment, but it will reduce in the later stages of economic development, Reference [42]. The STIRPAT model has the advantage of estimating factors that influence oil consumption through a linear regression model that includes an error term. Following Reference [40], the specific STIRPAT model is employed in this study and is presented as follows:

$$ln{oil}_t=a+\beta {lnpop}_t+\gamma ln{gdp}_t+\delta {ln{gdp}_t}^2+\partial lnvehicle+\theta time+{\epsilon }_t$$

(11)

where oil is the transport sector oil demand obtained from the Pakistan Economic Survey, published by the References [43,44]. Oil consumption is an important factor that degrades the environment and other combustions. The objective is to explore the effect of selected instrumental variables pop and gdp: population, gross domestic product, and the number of vehicles on the road on the oil demand transport sector. The data on these variables are also taken from the Economic Survey of Pakistan, published by the References [43,44]. Following Reference [13], time is incorporated for the effect of technology. The projected population and gross domestic product data are obtained from the Economic Research Service. The $a,\beta ,\gamma ,\delta ,\partial \mathrm{a}\mathrm{n}\mathrm{d} \theta$ are the coefficients for lnpop, lngdp, lngdp², lnvehicle, and time, respectively. Since the STIRPAT is a logarithmic transformation, coefficients associated with explanatory variables show elasticity. The elasticity term describes how explanatory variables affect the oil demand of the transportation sector, where elasticity measures the responsiveness of oil demand to one of the explanatory variables/its determinants.

4. Results and Discussion

4.1. Oil Demand and Carbon Emissions Projection

The following Grey models of oil demand and carbon emissions can be applied to calculate its projected series for 2019–2030.

$${{\widehat{oil}}^{\left(0\right)}}_k=\left[87.21\times e^{-\left(-0.10\right)\times k}-75.32\right]-{{\widehat{oil}}^{\left(0\right)}}_{\left(k-1\right)}$$

(12)

$${{\widehat{carbon}}^{\left(0\right)}}_k=\left[173.85\times e^{-\left(-0.09\right)\times k}-155.49\right]-{{\widehat{carbon}}^{\left(0\right)}}_{(k-1)}$$

(13)

Based on the actual oil demand data and carbon emissions from 2010 to 2018, the estimated equations give relatively accurate outcomes and can be verified from

Table 1. Result of oil demand and carbon emissions projections.

Year	Actual Oil Demand (Million Tons)	Projected Oil Demand (Million Tons)	Actual Carbon Emissions (Million Tons)	Projected Carbon Emissions (Million Tons)
2010	8.89	8.89	18.36	18.36
2011	9.27	9.25	19.13	18.10
2012	9.82	9.68	20.27	19.98
2013	10.30	10.69	21.26	22.06
2014	11.37	11.80	23.48	24.36
2015	13.02	13.03	26.88	26.89
2016	14.58	14.38	30.10	29.69
2017	16.05	15.88	33.13	32.79
2018	14.63	17.54	30.20	36.20
2019	-	19.36	-	39.97
2020	-	24.07	-	44.13
2021	-	26.85	-	48.72
2022	-	29.96	-	57.08
2023	-	33.42	-	63.36
2024	-	37.28	-	70.33
2025	-	41.57	-	78.07
2026	-	46.36	-	86.66
2027	-	51.69	-	96.19
2028	-	57.64	-	106.76
2029	-	64.26	-	118.49
2030	-	71.63	-	131.51

Source: Authors’ calculation based on Grey model. Note: The data on carbon emissions and oil demand is taken from Reference [14]. The transport sector oil demand for 2018 is obtained from provisional data.

. The reported series of actual oil demand is steadily increasing yearly, and the projected series supports demand-side pressure. The oil demand by 2019 will increase by 10.47 million tons (19.39–8.89), jumping about 117% compared to 2010. The annual average growth rate (AAGR) was 6.65% between 2010 and 2018 and 12.86% between 2019 and 2030. Thus, 6.16 (i.e., 12.86–6.65)% will be Pakistan’s transport sector’s additional projected oil demand during 2019–2030.

The reported result of carbon emissions also holds an increasing trend. It is expected to increase by 21.61 tons in 2019, jumping about 118% compared to 2010. The AAGR was 6.24% during the actual period and will be 11.45% between 2019 and 2030. Additional projected carbon that emits from oil demand by the transport sector to the atmosphere will be 5.20 (i.e., 11.45–6.24)% during 2019–2030. Thus, the AAGR of oil demand in the projected period must be lower than 6.65% if Pakistan is willing to put its positive share toward environmental sustainability. The same argument holds for the AAGR of carbon emissions. The policy measures to mitigate the transport’s carbon emissions are to switch from oil to gas fuel and place more dependence on public and hybrid vehicles.

4.2. Determinants of Projected Oil Demand

The initial step before the empirical investigation is to check whether variables are stationary or non-stationary. The non-stationary time series demands special treatment, otherwise it can lead to spurious estimates. Therefore, a unit root of the variables is tested, and the reports are displayed in

Table 2. Unit root test of variables.

	Level		First Difference
	t-Statistic	Prob.	t-Statistic	Prob.
lnoil	−2.208	0.2079	−3.506 **	0.0157
lngdp	−0.919	0.7673	−4.785 *	0.0007
lngdp2	−0.816	0.7995	−4.608 *	0.0010
lnpop	−0.5446	0.8666	−5.5961 *	0.0002
lnvehicle	1.0134	0.9956	−5.07967 *	0.0003

Source: Authors’ estimation. Note: *, ** show the rejection of the null hypothesis of a unit root at the 1%, and 5% levels of significance, respectively. The probability values are based one-sided p-values. The results of the unit root are based on the ADF test.

. The results confirm that all variables have a unit root, implying that variables are non-stationary at the level. A unit root is applied to the first difference of each variable, and the results of the first difference strongly reject a unit root while confirming that all variables are stationary at the first difference.

If all variables are stationary at the first difference, the common method for the relationship between variables is Fully Modified Ordinary Least Square FM-OLS, Reference [45]. The reported results of FM-OLS in

Table 3. Result of EKC and STIRPAT model (FM-OLS Method).

EKC Model			STIRPAT Model
Variables	Coefficient	t-Statistic	Coefficient	t-Statistic
lngdp	0.462 ** (0.247)	1.863	0.093 ** (0.045)	2.048
lngdp2	−0.036 * (0.01)	−1.78	−0.019 *** (0.006)	−2.798
lnvehicle	-	-	0.194 * (0.097)	1.986
time	-	-	−0.114 * (0.062)	−1.821
lnpop	-	-	0.525 * (0.205)	2.556
constant	8.484 *** (3.238)	2.62	3.063(5.411)	0.566
Sensitivity Analysis
R-squared	0.67			0.83
Adjusted R-squared	0.64			0.80
SE of regression	0.15			0.12
Long-run variance	0.05			0.00

Note: standard errors are in parentheses; *, **, and *** represent 10%, 5%, and 1% significance level, respectively. The EKC and STIRPAT stand for environmental Kuznets curve, and stochastic impacts by regression of population, affluence, and technology on the environment, respectively.

cover the modeling of EKC and STIRPAT. In the case of EKC modeling, the coefficient associated with economic growth is positive and confirms that income growth will increase the demand for projected oil in the transport sector. The negative coefficient associated with the square term of growth income confirms that an increase in further growth income will reduce projected oil demand. Thus, the positive sign of growth income and the negative sign of its square term support the environmental Kuznets’ hypothesis. The estimated income elasticity of oil demand is positive, and less than one confirms that transport sector oil demand is necessary for economic growth. The coefficient associated with economic growth is 0.462, which indicates that a 1 percent increase in economic growth will produce a 0.462 percent increase in transport sector oil demand. The elasticity concerning the square term of growth income is negative and statistically significant, confirming that oil demand is an inferior factor for further economic growth. The square term of the economic growth coefficient indicates that a 1 percent increase in the square term of economic growth will produce a 0.036 reduction in transport sector oil demand. The negative relationship between oil demand and the square term of growth income confirms that oil demand is an inferior factor. Further growth in income will reduce oil demand by transforming the transportation sector to clean energy.

The estimated result of the STIRPAT model reveals that an increase in affluence (GDP per capita) can increase the oil demand of the transport sector. This finding is consistent with the proposition of the STIRPAT model, as a 1 percent increase in GDP per capita can enhance oil demand by 0.09 percent. At the same time, the square of GDP per capita shows a significant negative impact, which confirms an inverted U-shaped relationship between oil demand and growth in income. The findings of GDP per capita and square of GDP per capita thus validate the robustness of the environmental Kuznets’ curve proposition. The number of vehicles reveals a significant positive impact on oil demand, as a one percent increase in vehicles can enhance the oil demand of the transport sector by 0.194 percent. This result indicates that most of the vehicles are fossil fuel dependent. Therefore, attempts should be taken to encourage the import and use of electric vehicles.

Additionally, the estimated elasticities of transport sector oil demand concerning the number of vehicles on the road and population are positive and statistically significant, implying that these drivers’ effect is different from zero on transport sector oil demand. An elasticity of 0.194 oil demand concerning the number of vehicles implies that a 1 percent increase in the number of vehicles on the road increases 0.194% oil demand in the transport sector. As represented by time, the increase in technology reveals a significant negative impact on oil demand. The findings reveal that an increase in technology can reduce oil demand by −0.44%. These findings make sense; in recent years, the number of hybrid vehicles has increased due to technological advancements in the transportation sector. The major influence of oil demand is the increase in population growth rate. Findings reveal that an increase in population can increase oil demand for the transportation sector by 0.53 percent. The population growth rate of Pakistan is higher than in the region. The increase in population can increase the demand for transportation and oil demand. Given the limited resources, Pakistan needs to control its population outburst. The authorities should adopt various policies to discourage the population growth rate.

5. Conclusions and Recommendations

The transportation sector of Pakistan is a major consumer of liquid fuel and is a major source of carbon dioxide emissions. This study examines the transportation sector’s projected oil demand and corresponding CO₂ emission from 2018 to 2030 by using the Grey model on actual data from 2010 to 2018, and exploring the environmental Kuznets and STIPAT hypotheses through an econometric inquiry from 1990 to 2019. The projected value of oil demand and carbon emission explored using the Grey model based on actual data series from 2010 to 2018 revealed an increasing trend. It is projected that the demand for liquid fuel will increase from 19.36 million tons in 2018 to 71.63 million tons during 2030; similarly, carbon emission will increase from 39.93 million tons in 2018 to 131.51 million tons in 2030. The model reveals a phenomenal increase in liquid fuel and carbon emissions during the projected period. It urges Pakistan to switch from oil to gas fuel and more renewable sources, encouraging hybrid vehicles to reduce transport sector oil demand and environmental degradation.

Similarly, the average annual growth rate (AAGR) of oil demand and carbon emissions in the actual period is 6.65% and 6.24%, respectively. The projected values of the AAGR of oil demand and carbon emissions should be lower than their actual values of AAGR to reduce oil demand and carbon emissions in the transport sector. The econometric result of EKC modeling confirms that the initial income growth can significantly increase oil demand, while further growth income can reduce oil demand. The statistically significant effect of income level supports the environmental Kuznets’ hypothesis. However, the STIRPAT model’s findings confirm an inverted U-relationship between oil demand and growth income. Population growth has increased as a major driver of oil demand and corresponding carbon dioxide emissions. The common conclusion which can be drawn from the study results is that Pakistan should control its population growth rate by enacting laws to restrain early/childhood marriages, decentralization of family planning, encouraging family planning programs, and discouraging more than one to two children. Furthermore, switching from oil to gas and encouraging hybrid and electric vehicles can be a valid policy option for reducing the oil demand growth rate.

The scope of this study is limited to road transport due to the unavailability of disaggregate data of various modes of transportation sectors such as air, water, and railways. Second, the period of the analysis is also a limiting factor. Therefore, future research can expand the scope of the current study by using disaggregated transportation and by expanding the sample size to other developed and developing countries. Furthermore, future studies can use innovations as a proxy for technological development instead of using time as a proxy for technological development.