Modeling the impact of transport energy consumption on CO₂ emission in Pakistan: Evidence from ARDL approach

Abstract

The objective of this research is to examine the relationship between transport energy consumption, economic growth, and carbon dioxide emission (CO₂) from transport sector incorporating foreign direct investment and urbanization. This study is carried out in Pakistan by applying autoregressive distributive lag (ARDL) and vector error correction model (VECM) over 1990–2015. The empirical results indicate a strong significant impact of transport energy consumption on CO₂ emissions from the transportation sector. Furthermore, foreign direct investment also contributes to CO₂ emission. Interestingly, the impact of economic growth and urbanization on transport CO₂ emission is statistically insignificant. Overall, transport energy consumption and foreign direct investment are not environmentally friendly. The new empirical evidence from this study provides a complete picture of the determinants of emissions from the transport sector and these novel findings not only help to advance the existing literature but also can be of special interest to the country’s policymakers. So, we urge that government needs to focus on promoting the energy efficient means of transportation to improve environmental quality with less adverse influence on economic growth.

Introduction

Transport infrastructure is a major tool to promote the organization and economic growth in the region. Transport sector influences all aspects of human life, education, research, entertainment, trade to manufacturing, culture, and defense. Countries around the globe realize its strength in economic growth and transforming resources into knowledge and communication (Mohmand et al. 2016; Subhra and Nath 2017). Growth is the main factor, which influences CO₂ emission (Andreoni and Galmarini 2012; Sobrino and Monzon 2014). Transport CO₂ emissions are intertwined with economic growth, and nonlinear effect of economic growth on CO₂ emissions is consistent (Xu and Lin 2015).

Despite the importance of transport infrastructure in economic growth, it also contributes to lower environmental quality. Road traffic is the largest contributor to the carbon footprint of the transport sector, and the reduction of transport has become one of the main objectives of sustainable transport policy (Sobrino and Monzon 2014). For instance, Chandran and Tang (2013) probe that transport energy consumption constitutes carbon dioxide and analyze that CO₂ emission varies due to change in weather. Also, online shopping causes to reduce carbon dioxide (CO₂) emissions from passenger transport (Liu et al. 2016; Smidfelt Rosqvist and Hiselius 2016). Road transport also contributes to CO₂ emissions. Also, both traffic for cargo and passenger reduced 4.3% of CO₂ emission during the evolution of commercial air traffic (Solís and Sheinbaum 2013; Amizadeh et al. 2016). In the same way, Shahbaz et al. (2015) conclude that road transportation enhances CO₂ emission in Tunisia. Timilsina and Shrestha (2009) have concluded that growth, population, and energy intensity are the main elements for motivating CO₂ emission from the transport sector in Asia. For instance, Duan et al. (2015) argue that over the last few decades, transport sector actively participates in the growing carbon emissions. Lin and Xie (2014) argue that carbon dioxide emissions will reach 1024.24 million tons (Mt) in 2020 in China; although its reduction potential will be 304.59 and 422.99 Mt under modest emission-reduction scenario and advanced emission-reduction scenario, respectively. The residents travel demand increased significantly. The results show that motor vehicle emissions are one of the main sources of air pollution, thus affecting the health of residents (HE and QIU 2016). The transport sector is second the largest energy consumer in the world and highly contributes to the economy more specifically in developing countries. However, despite the role of the transport sector in economic growth, it is a threat to global warming and counted as second largest carbon emitter in the world. Recently, transport infrastructure gets immense attention in the literature. To contribute to this field of research, the current paper begins to understand the ambiguous relationship between transport energy consumption and environmental quality, that is, we investigate the impact of important transport-related factors such as economic growth, foreign direct investment (FDI), and urbanization on CO₂ emissions. This is the first attempt to add potential factor related to transport in the existing body of knowledge related to transport energy consumption and CO₂ emissions.

The debate on energy, transport, and sustainable development is related to the discussion of sustainable development priorities and climate change. The main concern is the adverse effects of oil consumption, including greenhouse gas (GHG) emissions and air pollution. The transport industry is the world’s second-largest energy consumer. It consumes more than 30% of the world’s total energy consumption. In developing countries, the transport sector has made a significant contribution to the economy. However, while the transport sector has played a role in economic growth, it also poses a threat to global warming and is listed as the world’s second-largest carbon emitter. Therefore, transport infrastructure gets immense attention in the literature in recent past. Only a few studies regarding the energy–CO₂ emission nexus have focused on energy consumption from the transport sector and its impact on CO₂ emissions. For instance, Chandran and Tang (2013) suggest that economic growth and transport energy consumption significantly influence CO₂ emissions in the long run, whereas FDI has no impact on CO₂ emission in ASEAN-5 countries. Liddle and Lung (2013) determine causality running between transport energy consumption and economic growth in long run. In a study of Tunisia, Shahbaz et al. (2015) found that both road infrastructure and transport energy consumption cause to increase pollution by emitting CO₂ emission. Furthermore, bidirectional causality is found between transport energy consumption and CO₂ emissions. The bidirectional causality is also found between transport value-added and CO₂ emissions. Talbi (2017) argues that road transport mainly based on fossil fuel which is the leading contributor to transport sector’s CO₂ emissions. Achour and Belloumi 2016) conclude that road transport-related energy consumption, transport of carbon dioxide emissions, and total capital formation cause road infrastructure. In a study for Saudi Arabia, Alshehry and Belloumi (2017) disagree with the existence of inverse U-shaped relationship between transport CO₂ emissions and economic growth.

To summarize, prior studies mainly focus on aggregate and disaggregate energy consumption and its impact on CO₂ emissions. However, little attention has given to transport energy consumption and its influence on CO₂ emissions. Also, the conclusion drawn from these studies is inconclusive. Using time series data of Pakistan, this study is a first attempt to analyze the impact of transport energy consumption, economic growth, and CO₂ emission incorporating urbanization and foreign direct investment (FDI) of Pakistan. Energy saving and carbon dioxide emission reduction get attention worldwide. Pakistan is currently in the stage of industrialization and urbanization, which is characterized by rapid growth of energy consumption. Pakistan’s transport industry is highly energy-consuming and polluting. The rapid growth of transport, no high-quality fuel, and triggered problem of air pollution resist the solution to the problem. The surge in demand for energy leads to increasing use of coal (an average of 13 per year). The increase in oil, crude oil, electricity, and natural gas has an adverse impact on climate change (Danish et al. 2017a). Pakistan is an emerging economy; the transport sector consumes much energy, such as oil consumption (Shahbaz et al. 2017). Pakistan climate change is ranked among the highest in the world regarding vulnerability. Energy and transport sectors are the major sources of emission and together making up half of the national total. Presently, CO₂ emissions increase at the rate of 6% per year and will reach 400 Mt of CO₂ equivalent (per year) by 2030 if the business-as-usual scenario remains intact (UNDP 2015). The transport sector, because of the growing industrialization and urbanization of Pakistan, is one of the fastest growing sectors regarding output, energy consumption, and CO₂ emission. Its current performance and future outlook have the potential to leave behind problems related to the global issues of energy security and environmental sustainability (Lin and Ahmad 2016).

The present work contributes to the existing literature in numerous ways. First, to best of our knowledge, this study is the first attempt to investigate the impact of transport energy consumption on CO₂ emission in Pakistan-specific scenario incorporating FDI, urbanization, and economic growth. We add FDI and urbanization as a control variable to the transport energy and emission model because of technology and investment transfer in shape of FDI may influance the envrionment. We address this issue as greenhouse effect is a crucial concern from a policy perspective, particularly about CO₂ emission from transpor sector. In the previous literature, the role of transport in determining carbon dioxide emissions was not practicable, although it is becoming increasingly important in Pakistan. The rapid growth of transport, no high-quality fuel, and triggered problem of air pollution resist the solution to the problem in the case of Pakistan. The surge in demand for energy leads to increasing use of coal (an average of 13 per year). The increasing use of petroleum, crude oil, electricity, and natural gas in the transport sector is the main cause of negative climate change (Danish et al. 2017a). Identifying key drivers of carbon dioxide emissions is critical to the development of effective environmental and emission reduction policies. Second, earlier studies regarding the long-run impact of transport energy consumption and CO₂ emission have used total CO₂ emission to measure environmental quality. However, in this study, we use CO₂ emission from transport sector to measure the environmental quality. Lastly, we have added some important transport-related factor such as urbanization and FDI in a multivariate framework to avoid specification bias. It solves the issue of the specification and produces reliable and consistent empirical result. It would also be helpful for policymakers to formulate comprehensive environmental policy for sustainable economic development and considerably add to better long-run environmental performance. We perform ARDL methods to ascertain long-run and short-run dynamics. VECM Granger causality approach is applied to the direction of causality.

The rest of the paper is organized as follows: “Model specification, data sources, and econometric methodology” describes the data source and methodology. “Results analysis and discussion” explains the results and discussion and the concluding remarks are given in “Conclusion and policy implications.”

Model specification, data sources, and econometric methodology

Data sources

The present research work has gathered data from world development indicator (World Bank) and economic survey of Pakistan over the time span 1990–2015. CO₂ emission from transport sector is used to measure environmental quality. Economic growth is measured as GDP per capita (constant 2010 US Dollar). Transport energy consumption is measured in tons of oil equivalent. FDI has measured the net inflows of investment percentage of GDP. Urbanization is calculated urban population in the percentage of total population. The data on CO₂ emission, GDP, FDI, and urbanization are extracted from the database of world development indicator. The data on transport energy consumption is extracted from the international energy agency. For detail about variable and description, please refer to Table 1 of the study. Figure 1 shows the trends in CO₂ emission from transport sector, transport energy consumption, FDI, economic growth, and urban population.

Table 1 Variable name, symbols, description, and data source

Fig. 1

Trends in CO₂ emission from transport sector, transport energy consumption, FDI, economic growth, and urban population of Pakistan

Model specification

This study focuses on the relationship between CO₂ emissions for the transport sector, consumption for the transportation sector, FDI, and per capita real GDP in specific scenario of Pakistan; so employed the following model:

$$ Log{CO}_{2\mathrm{t}}={\beta}_0+{\beta}_1\left( Log{GDP}_{\mathrm{t}}\right)+{\beta}_2\left( Log{TRE}_{\mathrm{t}}\right)+{\beta}_3\left( Log{FDI}_{\mathrm{t}}\right)+{\beta}_4\left( Log{URB}_{\mathrm{t}}\right)+{\omega}_o $$

(1)

where CO₂ is the emission of CO₂ from the transport sector, GDP is the economic growth; TRE shows the transport energy consumption, FDI indicates the foreign direct investment, and URB is the urbanization. ω₀ is the error term, and t is the period, t = 1, 2, … n.

Economic growth, transport-related energy consumption, and CO₂ emission are strongly linked. Transport activities are essential for economic and social development. Nevertheless, the transport sector has also shown the fastest growth in energy consumption (Pablo-Romero et al. 2017). The transport sector is the major energy consumer in Pakistan particularly fossil fuel consumption. Urban density is growing at an average annual growth rate of 1.69%. Urbanized kilometers increase with the growth of urban density and spatial distribution of households and activities behind the urban roads (Talbi 2017). That is why we use urbanization as a control variable in our study. FDI is another control variable we use in the study; FDI is found to have an easing impact on the environment. Trade liberalization measures as export to GDP influences the environment.

Econometric strategy

For the analysis of time series data, several econometric approaches have been used to estimate long-run and short-run dynamics. We use ARDL-bound testing approach proposed by Pesaran et al. (2001) to ascertain the impact of transport energy consumption and economic growth and transport CO₂ emission in the presence of FDI and urbanization. ARDL approach is preferred due to some of its advantages. First, ARDL is the best estimation technique in the case of a sample of data. ARDL specification is equivalent to standard error correction model from a statistical point of view. However, standard errors it produces are likely to be different. Therefore, the estimates calculate through ARDL are unbiased. ARDL approach calculates both long-run and short-run estimates at once through linear transformation technique. ARDL is considered to be the most suitable approach in the case of stationary level independent of I(0) or I(1). So in our case, data sample is small, so ARDL is the best choice to estimate long-run and short-run elasticities. The unrestricted error correction model (UECM) is followed, and the empirical equation for Eq. (1) is given below:

$$ \Delta Log{CO}_{2\mathrm{t}}={\varphi}_0+{\theta}_1 Log{TEC}_{\mathrm{t}-1}+{\theta}_2 Log{GDP}_{\mathrm{t}-1}+{\theta}_3 Log{FDI}_{\mathrm{t}-1}+{\theta}_4 Log{URB}_{\mathrm{t}-1}+\sum \limits_{i=1}^p{\pi}_1\Delta {\left( Log{CO}_2\right)}_{\mathrm{t}-\mathrm{i}}+\sum \limits_{j=0}^p{\pi}_2\Delta Log{TEC}_{\mathrm{t}-\mathrm{i}}+\sum \limits_{j=0}^p{\pi}_3\Delta Log{GDP}_{\mathrm{t}-\mathrm{i}}+\sum \limits_{j=0}^p{\pi}_4\Delta Log{FDI}_{\mathrm{t}-\mathrm{i}}+\sum \limits_{j=0}^p{\pi}_5\Delta Log{URB}_{\mathrm{t}-\mathrm{i}}+{\omega}_{2\mathrm{t}}. $$

(2)

∆ is the lag difference operator of the variable and P is the present lag length. The first step before applying ARDL is an investigation of cointegration among underlying variables. The null of the hypothesis of which is to be tested is (H₀ ≠ θ₁ ≠ θ₂ ≠ θ₃ ≠ θ₄ ≠ 0) against the alternative hypothesis which is (H₀ = θ₁ = θ₂ = θ₃ = θ₄ = 0). To test the hypothesis, we calculate F value independent of the order of integration. The F value exceeds the upper bound limit directed toward rejection of the null hypothesis of no cointegration and F value below the lower limit indication of no-cointegration. Further results will be considered inconclusive if F value lies between upper and lower limits. For upper lower and bound limits, we rely on the bound testing values (Narayan 2005). The short-run coefficients are obtained by the estimates of Ʃ. The coefficients of θ₁, θ₂, θ₃, and θ₄ estimate the long-run dynamics of Eq. (2). Based on the methodology of Danish et al. (2017a), to check the robustness of the ARDL model, we have used alternative long-run estimation technique such as fully modified least square (FMOLS) method, dynamic ordinary least square method (DOLS), and canonical cointegration regression (CCR). Further, for reliability and validity of estimate produced by ARDL approach, we have used several diagnostic tests such as LM and Durban Watson (DW) tests for autocorrelation and autoregressive conditional heteroskedasticity (ARCH) test for heteroskedasticity.

Results analysis and discussion

Unit root analysis

To estimate long-run and short-run dynamics between economic variable, it is imperative to check the level of stationary. Otherwise, the results would be biased and unreliable in the presence of stationarity. Another reason is that any variable integrated with order 2 restrict us to use ARDL-bound testing approach. For this purpose, the present study relies on Augmented Ducky Fuller, Phillips Pesaran, and DF-GLD test statistic. The reason behind using three tests is to confirm the level of stationary at different levels. The results of unit root test are reported in Table 2. After analyzing the results, we infer that the null hypothesis cannot be rejected at level but a first difference. So, it suggests that we can pursue cointegration approach in this study.

Table 2 Results of unit root tests

Bound testing results

Once we confirm that none of the variables are integrated at the second difference. Next, we move to find the cointegration. For this, we estimate F statistic by applying bound testing approach; however, before bounding testing approach, it is essential to find the lag length. We choose lag length 2 for the study under VAR lag length selection criteria. After confirmation of the optimal lag length, next we confirm the cointegration among variables of interest. For the purpose, Wald statistic is used. For lag length, we choose Akaike information criteria (AIC) most suitable for the study. Results in Table 3 show that we reject the null of the hypothesis of cointegration for the all the equations.

Table 3 Results of bound F statistic and diagnostic tests

ARDL long-run and short-run dynamics

The cointegration results suggest a long-run relationship between variables of interest. The results of a long run and short run are reported in Table 4. The results indicate that the coefficient of total transport energy consumption is positive and significantly elastic in the central point of the sample. The outcome shadows our expectation, and the transport energy consumption causes to pollute the environmental quality in the country. The result is similar to Chandran and Tang (2013) and Shahbaz et al. (2015), and they conclude the same result. Further, the similar results are acquired by (Saboori et al. 2014) for single country analysis for 27 OECD country. Also, Alshehry and Belloumi (2017) estimate the similar result for Saudi Arabia. Comparatively, results of our study are more interesting such as we calculate environmental quality by CO₂ emission from transport sector instead of using overall CO₂ emission. The plausible reason is that expansion in number of vehicles demand cuases to energy consumption. Rapidly growing modes of transport, no high-quality fuels, in turn, have raised air pollution problems and resisted climate change (Danish et al. 2017a). Fossil fuels power the vast majority of road vehicles in Pakistan. Controlling the transport energy consumption from fossil fuel is the most immediate and proper solution to mitigate CO₂ emission. The road infrastructure, the number of vehicles on the road, and the transport energy consumption are interrelated. For instance, Pakistan’s quest for rapid economic growth compromises the sustainability of the environment and it will be an alarming situation for the economic, health, and social life of the people of the country.

Table 4 Results of long-run and short-run estimates

Turn into the estimation of the effect of economic growth on CO₂ emission, the long-run results in Table 4 infer that economic growth has an insignificant impact on economic growth. This finding deviates from the existing studies due to several reasons. One plausible reason may be that in Pakistan, government focuses only toward the economic growth not transport energy consumption. The other reason may be that unlike other studies, we have used the proxy of CO₂ emission from transport sector instead of using total carbon emission. It implies that economic growth does not contribute to carbon emission from transport sector but not the total CO₂ emission. For instance, Danish et al. (2017a, b) have tested economic growth with total CO₂ emission by testing environmental Kuznets curve (EKC) in Pakistan and the study infers the positive relationship between economic growth and total CO₂ emission. So, we infer that economic growth in Pakistan is only favorable for the transportation sector. Another possible reason may be that mostly the national transport sector like metro bus projects is fuel-subsidized and the green bus projects are environmental friendly. This suggests the promotion of mega bus projects like metro bus all around the country. At the same time result is insignificant with (Danish and Baloch 2017) as they find neagtive impact of economic growth on environmental degardation in case of Pakistan.

About the control variable FDI, we infer that the influence of FDI on transport CO₂ emission is positive and significant. It implies that FDI speeds up the rate of emission of CO₂ in the transport sector to pollute the environment. For the upgrading of road transport, Pakistan mostly relies on FDI, and the majority of the FDI inflows are targeted at road infrastructure. So, it may be the reason that technology transfer for the purpose to expand and upgrade transport network more particularly for road network is not environmental friendly. The results are similar to the findings of Chandran and Tang (2013), who find the positive impact of FDI on CO₂ emission in Malaysia, Singapore, and Thailand.

Moving on to urbanization, the impact of urbanization on CO₂ emission is insignificant. In the process of urbanization transfer, the labor force from agriculture to industrial and service sector in urban areas and the shift in production to commercial goods from in-house production accelerate the transportation activities. The increase in transportation activities consumes more energy and consequently emission of CO₂ increases. Another reason is that it speeded the development of public and private transport, which consumes additional energy and results in decreasing of environmental quality. An increase in urban population may cause more transportation of products and raw materials from rural areas to urban areas which may boost the scale of production to meet the need and demand of the migrated population. Similarly, with the increase in intercity transport along with city sprawl and commuter traffic, the private transports are increasing from day to day globally (Zhu et al. 2017). This increase in transport can improve infrastructure in the form of construction facilities such as electricity and road network enhancements, sewage disposal and uncontaminated water supply, and sanitation and drainage system. The maintenance and construction of infrastructure may consume more energy which also increases CO₂ emission (Arvin et al. 2015). Thus, the rapid increase in population requires more production of goods and services which need more energy consumption and CO₂ emission rise. This can be attributed to technology innovation, market integration, as well as to human capital. Due to all of these factors, urbanization may cause degrade environmental quality in the future; so, policymaker in Pakistan should keep in view the negative role of urbanization, as urbanization is a long-run process and takes time to appear its influence on environmental quality.

Turn into the short-run dynamics is shown in Table 5. The estimate of short run suggests different results than long-run estimates. In the short run, economic growth has a significant positive impact on transport CO₂ emission, while transport energy consumption, FDI, and urbanization have an insignificant impact on transport CO₂ emission. The variation of long-run estimates from short-run estimates suggests different policy suggestions for the long-term plan as vision 2025–2050 and short plan such as vision 2025.

Table 5 Results of FMOLS, DOLS, and CCR analysis (robust check)

Robust analysis

In this study, we also check robustness by applying three alternative techniques such as FMOLS, DOLS, and CCR to estimate Eq. (2) following Baek (2016) and Danish et al. (2017a). Table 5 presents that impact of GDP on CO₂ emission is insignificant. Additionally, findings approve that transport energy consumption accelerates CO₂ emission. Also, FDI and urbanization have a significant and positive impact on CO₂ emission. The results from all the three different methods confirm our results estimated by ARDL-bound testing approach. Furthermore, to see endogeneity, lack of structural invariance, and for the stability of model, we perform CUSUM of residual and CUMSUMsq of residual tests and results drawn are shown in Figs. 1 and 2, respectively. The results suggest that the blue line lies well between red lines which indicate the significance level which directs that results are stable (Fig. 3). Further, we use χ² − Reset, χ² – LM, and χ² − ARCH to see the presence of autocorrelation and heteroscedasticity and the results are noted at the bottom of Table 4. The results show that transport energy consumption, FDI, economic growth, and urbanization explain variation in transport CO₂ emission in the absence of autocorrelation and heteroscedasticity. Therefore, results of the study are reliable and can be used for policy implication.

Fig. 2

Results of CUMSUM at 5% level of significance. The area between red lines is the area for rejection of null hypothesis

Fig. 3

Results of CUMSUMsq at 5% level of significance. The area between red lines is the area for rejection of null hypothesis

The existence of long-run estimates is an indication for at least one-way causal relationship among variable of interest. To check causal relationship in the present study, we rely on VECM Granger causality approach. The estimation of VECM Granger causality is a two-step procedure. First, we to find error correction term by running the single regression. Next, we estimate error correction model in order to get long-run causality. In the second step, we calculate Wald statistic for estimation of the short-run causal relationship among core variable of the study. In the model among a single-period lagged error correction term (ECT_t − 1), because of Engle and Granger (1987) caution that if the series are integrated of order one, the cointegration VAR estimation in first difference will be misleading. Thus, the augmented Granger causality test with ECM is formulated below:

$$ \left(\begin{array}{l}\log {CO}_2\\ {}\log GDP\\ {}\log TEC\\ {}\log FDI\\ {}\log URB\end{array}\right)=\left(\begin{array}{l}{\beta}_1\\ {}{\beta}_2\\ {}{\beta}_3\\ {}{\beta}_4\\ {}{\beta}_5\end{array}\right)+\left(\begin{array}{l}{\beta}_{11\mathrm{K}}\kern0.75em {\beta}_{12\mathrm{K}}\kern1.75em {\beta}_{13\mathrm{K}}\kern1.25em {\beta}_{14\mathrm{K}}\kern1.25em {\beta}_{15\mathrm{K}}\kern1.25em {\beta}_{16\mathrm{K}}\kern0.75em \\ {}{\beta}_{21\mathrm{K}}\kern0.75em {\beta}_{22\mathrm{K}}\kern1.75em {\beta}_{23\mathrm{K}}\kern1.25em {\beta}_{24\mathrm{K}}\kern1.25em {\beta}_{25\mathrm{K}}\kern1.25em {\beta}_{26\mathrm{K}}\\ {}{\beta}_{31\mathrm{K}}\kern0.75em {\beta}_{32\mathrm{K}}\kern1.75em {\beta}_{33\mathrm{K}}\kern1.25em {\beta}_{34\mathrm{K}}\kern1.25em {\beta}_{35\mathrm{K}}\kern1.25em {\beta}_{36\mathrm{K}}\kern1em \\ {}{\beta}_{41\mathrm{K}}\kern0.75em {\beta}_{42\mathrm{K}}\kern1.75em {\beta}_{43\mathrm{K}}\kern1.25em {\beta}_{44\mathrm{K}}\kern1.25em {\beta}_{45\mathrm{K}}\kern1.25em {\beta}_{46\mathrm{K}}\kern1.25em \\ {}{\beta}_{51\mathrm{K}}\kern0.75em {\beta}_{52\mathrm{K}}\kern1.75em {\beta}_{53\mathrm{K}}\kern1.25em {\beta}_{54\mathrm{K}}\kern1.25em {\beta}_{55\mathrm{K}}\kern1.25em {\beta}_{56\mathrm{K}}\end{array}\right)\left(\begin{array}{l}\varDelta \log {CO}_{2\mathrm{it}}\\ {}\varDelta \log {GDP}_{\mathrm{it}}\\ {}\varDelta \log {TEC}_{\mathrm{it}}\\ {}\varDelta \log {FDI}_{\mathrm{it}}\\ {}\varDelta \log {URB}_{\mathrm{it}}\end{array}\right)+\left(\begin{array}{l}{\rho}_1\\ {}{\rho}_2\\ {}{\rho}_3\\ {}{\rho}_4\\ {}{\rho}_5\end{array}\right){ecm}_{\mathrm{it}-1}+\left(\begin{array}{l}{\vartheta}_1\\ {}{\vartheta}_2\\ {}{\vartheta}_3\\ {}{\vartheta}_4\\ {}{\vartheta}_5\end{array}\right). $$

(3)

t shows the period (1990–2015), i denotes i = 1, 2, 3 … 25, ecm_it - 1 is the lagged error correction term and the stochastic error term, and ω_it indicates the residual term.

The results of causality analysis are reported in Table 6. The long-run causal relationship indicates that bidirectional causality only exists between FDI and economic growth. In the feedback, the hypothesis is confirmed between transport energy consumption and CO₂ emission from the transport sector. The results are consistent with the findings of Saboori et al. (2014) for OECD countries and Alshehry and Belloumi (2017) detect bidirectional causality between transport energy consumption and CO₂ emission. Moreover, unidirectional causality is running from CO₂ emission to economic growth. Furthermore, bidirectional causality is detected between urbanization and CO₂ and between FDI and economic growth.

Table 6 Results of long-run and short-run causality analysis

Variance decomposition analysis

From Table 7, it is indicated that 51.79% of carbon emission is explained by an outer factor beyond the model of this study. The share of GDP, FDI, and transport energy consumption is 12.87, 15.96, and 15% and the share of urbanization is negligible. Similarly, the share of a GDP due to its shock is 47%. The contribution of CO₂ emission in a GDP is 26.62%, whereas variation due to FDI and urbanization is negligible; however, the share of transport energy consumption is 19.43%. The share of energy consumption due to its own shock is 32.20%. The share of CO₂ emission in transport energy consumption is 7.39%. The percentage of GDP, FDI, and urbanization is 33.54, 71.2, and 9.65%, respectively. The contribution of FDI due to its own shock is 24.29. The share of CO₂, transport energy consumption, and GDP is 28.24, 14.74, and 30.38, respectively. Furthermore, the share of urbanization is negligible. Lastly, the share of urbanization due to its own shock is 6.1%. The contribution of CO₂ emission, transport energy consumption, and GDP is 17.21, 25.97, and 46.71, respectively. However, the share of FDI in urbanization is negligible.

Table 7 Variance decomposition analysis

Impulse response function

Turn into impulse response function which is the substitute for variance decomposition analysis. The impulse response function explains the reaction of independent variables (see Fig. 4). The findings of impulse response function reveal that CO₂ emission rises due to forecast error stemming from energy crises. This shows that carbon emission rises due to rising in transport energy consumption. The forecast error in economic growth enhances carbon emission. The response of FDI in CO₂ emission arises with the passage of time. The forecast error in FDI due to CO₂ emission, urbanization, economic growth, and energy consumption remains constant as shown in Fig. 4. The response of urbanization due to CO₂ emission is increasing whereas due to FDI, economic growth and transport energy consumption are increasing. The response of CO₂ emission in transport energy consumption is decreasing. The response of the urbanization and economic growth in transport energy consumption is increasing.

Fig. 4

Impulse response function

Conclusion and policy implications

This research investigates the relationship between transport energy consumption, economic growth, and carbon dioxide emissions to account FDI and urbanization in Pakistan. For the purpose, we perform a family of econometric series over the time 1990–2015. The robustness of the model is checked by using three alternative techniques, i.e., FMOLS, DOLS, and CCR approaches. For detection of causality, we rely on VECM Granger causality followed by innovative accounting approach. From estimation, we observe some interesting findings. First, transport energy consumption causes to increase in emission of CO₂ and degrades environmental quality. Additionally, FDI significantly accelerates the rate of CO₂ emission from the transport sector, which degrades environmental quality. Second and more interestingly, economic growth and urbanization both have an insignificant impact on CO₂ emission from transport as against our expectations.

The empirical findings suggest a linear relationship between transport energy consumption and environmental degradation, as we expected. Mass transit busses and trucks are the major vehicles on roads, which consume most of the fossil fuels and over half of the oil is consumed in the transport sector of Pakistan (UNDP 2015). Fossil fuel has adverse impact on environmental quality. Excessive burning of fossil fuel in the major sectors of the economy, e.g., transport sector, depletes the environmental quality. So, we suggest based on the estimation of present work that government needs to cut the number of vehicles which is threatening the environmental sustainability. Mass bus transit scheme should be converted into green transportation more particularly in big cities. As other studies also suggests that the government desires to endorse CNG-loaded transport vehicles to lessen fossil fuel consumption to support the country’s human-friendly environment (Danish et al. 2017a), we also urge that government should launch mass green bus transit scheme in all big cities of the country to reduce the dependency on private vehicles, which mostly run on diesel and oil products. Also, the externality effect may be responsible for high levels of transport energy consumption in Pakistan. This means that third parties can positively affect road transport energy consumption. In the case of Pakistan, the external effects of rail transport can be taken into account, since rail transport is an energy-efficient means of transport. The situation of rail transport is worsening due to the ignorance of the government. Pakistan railway performance is poor and the people of the country prefer to travel by roads. Poor performance of the railway is the main problem of long-term decline in profits, rising operating costs, investment, and system dichotomy and governance issues. As the population continues to grow, the railroad transport burden continues to develop and as a result, Pakistan’s road transport energy increased. As we discussed above, road transport and energy consumption are not effective and environmentally friendly. Therefore, we urge the government of Pakistan to focus on rail transportation to reduce the burden and dependency on road transport. At first sight, one would anticipate that the reduction in the consumption of fossil fuels can be achieved through improving the infrastructure for the public transport sector, such as the ride on public transport which can be made more convenient, luxurious, and attractive for all class of users and similar to the use of private cars powered by fossil fuels.

Additionally, the government needs to focus on FDI particularly in the transport section. As indicated by our econometric estimation, most of the FDI is worsening the environmental quality. The government needs to restrict the transfer of technology which pollutes the environment by imposing dumping duties. Another way to control transferring of polluting technology is the time of agreement government needs to include the condition that guest country or company would shift clean energy technology that may help to reduce pollution.