The relationship between energy consumption, economic growth and carbon dioxide emissions in Pakistan

Khan, Muhammad Kamran; Khan, Muhammad Imran; Rehan, Muhammad Article The relationship between energy consumption, economic growth and carbon dioxide emissions in Pakistan Financial Innovation Provided in Cooperation with: Springer Nature Suggested Citation: Khan, Muhammad Kamran; Khan, Muhammad Imran; Rehan, Muhammad (2020) : The relationship between energy consumption, economic growth and carbon dioxide emissions in Pakistan, Financial Innovation, ISSN 2199-4730, Springer, Heidelberg, Vol. 6, Iss. 1, pp. 1-13, https://doi.org/10.1186/s40854-019-0162-0 This Version is available at: https://hdl.handle.net/10419/237190 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ RESEARCH Open Access The relationship between energy consumption, economic growth and carbon dioxide emissions in Pakistan Muhammad Kamran Khan * , Muhammad Imran Khan and Muhammad Rehan * Correspondence: [email protected] School of Economics, Northeast Normal University, Changchun, Jilin, China Abstract Developing countries are facing the problem of environmental degradation. Environmental degradation is caused by the use of non-renewable energy consumptions for economic growth but the consequences of environmental degradation cannot be ignored. This primary purpose of this study is to investigate the nexus between energy consumption, economic growth and CO 2 emission in Pakistan by using annual time series data from 1965 to 2015. The estimated results of ARDL indicate that energy consumption and economic growth increase the CO2 emissions in Pakistan both in short run and long run. Based on the estimated results it is recommended that policy maker in Pakistan should adopt and promote such renewable energy sources that will help to meet the increased demand for energy by replacing old traditional energy sources such as coal, gas, and oil. Renewable energy sources are reusable that can reduce the CO2 emissions and also ensure sustainable economic development of Pakistan. Keywords: Energy consumption, Economic growth, CO2 emissions, ARDL Introduction Pakistan is a developing country in South Asian countries, economy of Pakistan is growing rapidly and it is expected that the economic growth of Pakistan will continue with same trend in the future. Pakistan’s economy depends on agriculture, and agriculture is the main dominant sector of the country, but due to repaid growth of industrial sector in Pakistan, the agriculture land is cutting. Besides this, rapid increase in population causes deforestation; Pakistan is top ranked country in Asian countries that faces the problem of deforestation. Increase in economic growth and industrial sectors use energy for growth that causes environmental degradation. Pakistan is facing high demand of energy for which traditional energy sources are used to meet its fast increasing demand for energy. Wolde-Rufael and Menyah (2010) stated that use of traditional energy resources causes to discharges carbon dioxide that helps to deteriorate environmental quality. Ahmed et al. (2015) stated that environmental degradation affect the environment and health of the human being in Pakistan. Yang and Li (2017) stated that environmental degradation is caused by vast amount of greenhouse gas emissions, including carbon dioxides, nitrous oxide, and methane. Shahbaz et al. (2013) stated that use of fossil fuels for daily life, massive smoke expulsion from the factories and consumption of wood as an energy © The Author(s). 2020 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Financia l Innovation Khan et al. Financial Innovation (2020) 6:1 https://doi.org/10.1186/s40854-019-0162-0 source boost the CO2 emissions. Carbon dioxide emissions have a destructive impact on the economy and other sectors such as agriculture and forestry. Chaudhry (2010); Pao and Tsai (2010); Siddiqui (2004) investigated the association between energy consumption, economic growth, and sustainable environment. Most researches have been done for developed countries such as European countries and American countries (Kasman and Selman 2015). Early researches on the same subject have generally concluded that the development of economy and energy consumption causes the CO 2 emissions. Several research studies have pointed out the relationship between the development of the economy, non-renewable energy and CO2 emissions, that is essential for understanding and refining the development pattern of developing countries like Pakistan. The societies which have gifted with plentiful natural resources can efficiently mitigate import of fossil sources and emissions of carbon dioxide. Balsalobre et al. (2018) stated that energy strategy implementation is validated to decrease dependence on the use of non-renewable energy sources. Non-renewable energy sources still profoundly influence the energy mix. This explicates the sustainability of both energy sources, i.e., renewable and non-renewable that may occur in the long-run. The main objective of this research study is to investigate the association between energy consumption, economic growth, and carbon dioxide emission in Pakistan. Different researchers identified that environmental degradation is caused by using non-renewable energy consumption and economic growth in developed countries. This research study will help to clear the gap between early researches by controlling the model for energy consumption, economic growth, and CO2 emissions. This research study used newly developed econometric techniques the Auto-Regressive Distributive Lag (ARDL) bound testing for cointegration. ARDL model have different advantages as compared to other cointegration method. ARDL model can be applied if the variables are stationary at level or first difference of both of them while other cointegration methods need same order of integration. Different lag can be used for dependent and independent variables (Pesaran et al. (2001). This study will provide a new vision for policymakers to design key policy instruments on balancing economic growth and environmental quality. Literature review Acar and Lindmark (2017) investigated the convergence of CO2 emissions in OECD countries by using (oil vs. coal) as energy source. The authors divided the study period into two sub-periods. The first period covers the oil price shocks of OPEC, where the OECD oil policy was to a great extent governed by energy security concerns and Cold War strategic considerations. The second period correspond rise climate policy in several OECD countries. Due to such contextual differences, oil and coal behave differently in the two sub-periods on economic growth. Asumadu-Sarkodie and Owusu (2017a,2017b) stated that long-run equilibrium associations exist between environmental degradation, electricity use, economic growth and industrialization. The examined results of variance decomposition indicated that use of electricity and economic growth increase the environmental degradation by 7% 20% respectively. They recommended that in future using clean energy can decrease environmental degradation in Sierra Leone. Khan et al. Financial Innovation (2020) 6:1 Page 2 of 13 Destek (2017) indicated that economic development is positively affected by nonrenewable biomass energy consumption countries. Fan and Lei (2017) examined the associations between environmental degradation transportation and economic development in Beijing by using time series data for econometric analysis from 1995 to 2014. The estimated results pointed out that transportation and CO2 emissions have a positive influence on economic growth. Işik et al. (2017) applied Autoregressive distributed lag (ARDL) model to check the association between the study variables. The estimated results revealed that economic growth, the growth of financial system, international trade and tourism expenditures positively impact the Greece’s CO2 emissions. They stated that tourism, as a leading sector in the Greek economy, has severe negative environmental impacts for Greece in the long run. Therefore, they suggested that Greece should actively take into consideration this threat from the tourism sector as this one sector dominates the whole Greek economy. Isik et al. (2019) investigated the impact of real GDP, population, and renewable energy and fossil energy consumptions on CO2 emissions in ten US states from 1980 to 2015. The examined results indicated that the EKC hypothesis is valid for the following states Florida, Illinois, Michigan, New York, and Ohio. The results indicated that fossil energy consumption have negative impacts on CO2 emission levels in Texas while energy consumption having positively influence on CO2 emissions in Florida but this impact is lower as compared to other states of US. Azam et al. (2016) inspected the influences of CO2 emissions, energy use, trade, and human capital on economy growth from 1971 and 2013 for China, the USA, India, and Japan by utilizing panel fully modified ordinary least squares (FMOLS) for checking the association among the study variables. The examined results pointed out that CO2 emissions and energy consumption negatively and significantly influences the economic growth while trade and human capital positively and significantly influences the economic growth. Hanif (2018) studied the influences of economic growth; urban expansion; and consumption of fossil fuels, solid fuels, and renewable energy on CO2 emission in SubSaharan Africa economies from 1995 to 2015 by utilizing the GMM model for examination of the association among the study variables. The examined results indicated that consumption of fossil and solid fuels positively impact the CO2 emissions while renewable energy helps to decrease the CO2 emissions. Saboori et al. (2017) examined the association of oil consumption, economic growth and with environmental degradations in three Asian countries from 1980 to 2013 by applying Johansen cointegration test for checking the relationship among the study variables. The examined results indicated that uni-directional causality running from oil consumption to economic growth in China and Japan, while oil consumption to CO2 emissions in South Korea. Bhat (2018) studied the impact of energy consumption, economic growth on carbon dioxide from 1992 to 2016 by utilizing Panel ARDL model for checking the association among the study variables. The examined results indicated that capital, labor, population, per-capita income, and non-renewable energy consumption positively impact the CO2 emissions. Sulaiman and Abdul-Rahim (2017) investigated the association of CO2 emission, energy consumption and economic in Malaysia from 1975 to 2015 by utilizing ARDL Khan et al. Financial Innovation (2020) 6:1 Page 3 of 13 model. The examined results indicated that economic growth is not impacted by by energy consumption and CO2 emission while energy consumption and economic growth positively influence the CO2 emission. Tamba et al. (2017) inspected the impact of gasoline energy consumption on economic growth in Cameroon by utilizing autoregressive vector (VAR) model and Wald test for testing causality. The estimated results showed that no long-term relationship exists among the study variables. Bidirectional causality relationship between gasoline consumption and economic growth exists in Cameroon. The estimated results showed that reducing gasoline consumption without appropriate and established energy policies are not a possible solution to maintain Cameroon’s economic growth. Apergis et al. (2010) and Zoundi (2017) stated that renewable energy exploitation was restricted by different conditions and the level of economic growth in the low-income countries. Sinha and Shahbaz (2018) stated that the high cost of initial stage of renewable energy development results demotivate in developing countries to invest in renewable energy sources. It appears that promoting renewable energy in some low-income countries may lead to restrain their economic progress in the short run. Inglesi-Lotz and Dogan (2018) suggested that shifting energy consumption away from fossil fuels to renewable energy sources is a challenge for developing countries. Different energy structures between the developing and developed countries are different because of technological and economic conditions. Methodology Unit root test Stationarity in time series data is a common problem. It is necessary to check the stationarity of the variables before using ARDL model. Traditional methods for assessment in applied econometrics are based on the supposition of normality saying mean, and variance does not change over time. However, the mean and variance of many economic elements do not remain constant and such type of variables is known as unit root variables. Traditional approach (i.e., ordinary least square, OLS) produces biased and unreliable estimates in the presence of stationary data. In this research, we used a great unit root test, such as PP (Phillips and Perron 1988), ADF (Dickey et al. 1979). Our empirical analysis checks stationarity of each variable by applying PP and ADF unit root test. Autoregressive distributive lag Time series data for this study was collected from world development indicator World Bank, from 1965 to 2015. The selection choice of the period was based on the data availability. Carbon dioxide emissions is used as a measure for environmental degradation, and it is measured as CO 2 emission per capita, PCI is used as proxy for economic growth, measured as percentage of gross domestic product (constant 2010 USD), CLCNM is coal consumption, OLCNM is oil consumption, NTGCNM is the natural gas consumption are explanatory variables. Following is the main regression equation for our variables. CO2emt¼β0þβ1PCItþβ2CLCNMtþβ3OLCNMtþβ4NTGCNMtþεt Following previous researchers Jebli and Youssef (2015a,2015b), Jebli and Youssef (2015a,2015b) and Alshehry and Belloumi (2015) and Khan et al. (2017,2019a,b)in this study we use the bound testing approach proposed by Pesaran et al. (2001)to Khan et al. Financial Innovation (2020) 6:1 Page 4 of 13 estimate the long run estimates between CO 2 emission, economic growth, energy consumption. Prior studies in energy economics suggest several econometric approaches to check the existence of the cointegration. However, the bound testing approach was preferred due to several reasons. The ARDL bound testing method is most suitable when the variables are integrated at the order of 1(0) or 1(1). Besides, it is useful when data size is small. The lag modification in the ARDL model gives fair estimations of the long run and effective t-statistic value even in the presence of endogeneity (Pesaran et al. 2001). Therefore, this study applied the ARDL method to investigate cointegration among energy consumption, economic growth, and CO 2 emission. The ARDL Bound testing approach is given by the following equations: ΔCO2emt¼β0þX q1 i¼1 β1iΔCO2emt−iþX q2 i¼1 β2iΔPCIt−iþX q3 i¼1 β3iΔCLCNMt−i þX q4 i¼1 β4iΔOLCNMt−iþX q5 i¼1 β5iΔNTGCNMt−iþδ0CO2emt−iþδ1PCIt−i þδ2CLCNMt−iþδ3OLCNMt−iþδ4NTGCNMt−iþμt ΔPCIt¼β0þX q1 i¼1 β1iΔPCIt−iþX q2 i¼1 β2iΔCO2emt−iþX q3 i¼1 β3iΔCLCNMt−i þX q4 i¼1 β4iΔOLCNMt−iþX q5 i¼1 β5iΔNTGCNMt−iþδ0PCIt−iþδ1CO2emt−i þδ2CLCNMt−iþδ3OLCNMt−iþδ4NTGCNMt−iþμt ΔCLCNMt¼β0þX q1 i¼1 β1iΔCLCNMt−iþX q2 i¼1 β2iΔCO2emt−iþX q3 i¼1 β3iΔPCIt−i þX q4 i¼1 β4iΔOLCNMt−iþX q5 i¼1 β5iΔNTGCNMt−iþδ0CLCNMt−i þδ1CO2emt−iþδ2PCIt−iþδ3OLCNMt−iþδ4NTGCNMt−iþμt ΔOLCNMt¼β0þX q1 i¼1 β1iΔOLCNMt−iþX q2 i¼1 β2iΔCO2emt−iþX q3 i¼1 β3iΔPCIt−i þX q4 i¼1 β4iΔCLCNMt−iþX q5 i¼1 β5iΔNTGCNMt−iþδ0OLCNMt−i þδ1CO2emt−iþδ2PCIt−iþδ3CLCNMt−iþδ4NTGCNMt−iþμt ΔNTGCNMt¼β0þX q1 i¼1 β1iΔNTGCNMt−iþX q2 i¼1 β2iΔCO2emt−iþX q3 i¼1 β3iΔPCIt−i þX q4 i¼1 β4iΔCLCNMt−iþX q5 i¼1 β5iΔOLCNMt−iþδ0NYGCNMt−i þδ1CO2emt−iþδ2PCIt−iþδ3CLCNMt−iþδ4OLCNMt−iþμt In the above equations of the bound testing approach, first difference operation is indicated by Δ,andμ t is the residual term. The null hypothesis to be tested is H o :δ 0 = Khan et al. Financial Innovation (2020) 6:1 Page 5 of 13 δ 1 =δ 2 =δ 3 =δ 4 = 0 and alternative hypothesis H 0 :δ 0 ≠δ 1 ≠δ 2 ≠δ 3 ≠δ 4 ≠0 indicating the long run association between the study variables. Cointegration is based on the results of the F value in the bound testing approach. If the calculated F-value exceeds the upper bound, the null hypothesis of no cointegration is rejected, but the result is considered un-decidable when the F-value lies between upper and lower bound values. The error correction model for the estimation of the short-run relationships is specified as: ΔCO2emt¼β0þX q1 i¼1 β1iΔCO2emt−iþX q2 i¼0 β2iΔPCIt−iþX q3 i¼0 β3iΔCLCNMt−i þX q4 i¼0 β4iΔOLCNMt−iþX q5 i¼0 β5iΔNTGCNMt−iþη1ECTt−iþμt ΔPCIt¼β0þX q1 i¼1 β1iΔPCIt−iþX q2 i¼0 β2iΔCO2emt−iþX q3 i¼0 β3iΔCLCNMt−i þX q4 i¼0 β4iΔOLCNMt−1þX q5 i¼0 β5iΔNTGCNMt−iþη2ECTt−iþμt ΔCLCNMt¼β0þX q1 i¼1 β1iΔCLCNMt−iþX q2 i¼0 β2iΔCO2emt−iþX q3 i¼0 β3iΔPCIt−i þX q4 i¼0 β4iΔOLCNMt−iþX q5 i¼0 β5iΔNTGCNMt−iþη3ECTt−iþμt ΔOLCNMt¼β0þX q1 i¼1 β1iΔOLCNMt−iþX q2 i¼0 β2iΔCO2emt−iþX q3 i¼0 β3iΔPCIt−i þX q4 i¼0 β4iΔCLCNMt−iþX q5 i¼0 β5iΔNTGCNMt−iþη4ECTt−iþμt ΔNTGCNMt¼β0þX q1 i¼1 β1iΔNTGCNMt−iþX q2 i¼0 β2iΔCO2emt−iþX q3 i¼0 β3iΔPCIt−i þX q4 i¼0 β4iΔCLCNMt−iþX q5 i¼0 β5iΔOLCNMt−iþη5ECTt−iþμt In the above equations η 1 to η 5 indicating the speed of adjustment, ECT t−i is the lagged error correction term. ECT t−i is expected to be negative and significant. The CUSUM and CUSMSQ are also used for model stability check (Brown et al. 1975). Breusch–Godfrey is used for checking the serial correlation, and Breusch-PaganGodfrey was used for checking Heteroskedasticity. Results and discussions (Table 1) Before considering the long run ARDL model, we checked the stationary level of each variable. The reported results claim to reject the null hypothesis of the unit root; results indicate that CO2 emissions, Per capita income, natural gas and oil consumption are stationary at level and at first difference. The estimated results indicate that ARDL Khan et al. Financial Innovation (2020) 6:1 Page 6 of 13 model can be applied for checking the short and long run association among the study variables. Table 2shows the results of Lag length selection criteria for Co-integration. Usually, the researcher uses AIC and SC criteria for lag length selection because these two are superior for the small sample. We have selected AIC lag length from the above table. According to AIC lag length selection criteria lag 2 is the best option for lag length and is appropriate for ARDL approach. Table 3shows the results of Heteroskedasticity and serial correlation LM test. The result of Breusch Pagan Godfrey chi-square value is 0.7259 which higher than 5%, which indicates that there is no Heteroskedasticity problem in our data. The result of the LM test shows that probability chi-square is more significant than 5% that indicates that our data is free of the problem of serial correlation. Table 4indicates the results of Bound test; F statistic applied for investigation of cointegration relationship. The above-estimated results are according to Narayan (2005). The empirical results show that the estimated F-statistics for FCO 2 (CO2|PCI, COLCNM, NTGCNM, OLCNM) value is 5.2017 that is higher than the upper bound value at 5%, based on the estimated result of first equation co-integration exists among the variables. In the second equation, we changed the dependent variable from CO2 emission to per capita income F PCI (PCI|CO2, COLCNM, NTGCNM, OLCNM). Fstatistics value of the second equation indicates that cointegration exists at 5% because Table 1 Unit Root Test ADF Phillips-Perron Level Variables Intercept Trend & Intercept Intercept Trend & Intercept Co2 Per Capita −0.0105 −3.9313 ** −0.1172 −2.3169 Per Capita % Of GDP −6.6133 * −6.6682 * −6.6134 * −6.6690 * Coal Consumption −0.4811 −1.8007 −0.2819 −2.5639 Natural Gas Consumption 0.2805 −1.9188 0.7109 −1.6321 Oil Consumption 1.0150 −2.0356 1.2493 −1.6065 First Difference Co2 Per Capita −6.6240 * −6.5945 * −6.6824 * −6.6536 * Per Capita % Of GDP −6.6240 * −11.3836 * −6.6824 * −26.4811 * Coal Consumption −3.9032 ** −4.4879 ** −7.0700 * −7.1254 * Natural Gas Consumption −3.5880 ** −3.7317 ** −3.5408 * −3.6997 ** Oil Consumption −3.5999 ** −4.0415 ** −4.0651 ** −4.7010 ** 1, 5 and 10% levels is indicated respectively by *, ** and *** Table 2 Lag length criteria for Co-integration Lag LogL LR FPE AIC SC HQ 0−381.6793 NA 9.6319 16.4544 16.6513 16.5285 1−140.3190 421.0967 0.0009 7.2476 8.4286 7.6920 2−76.9002 97.1521 * 0.0002 * 5.6128 * 7.7778 * 6.4275 * 3−59.8911 22.4376 0.0003 5.9528 9.1019 7.1379 4−30.7264 32.2673 0.0003 5.7756 9.9089 7.3309 1, 5 and 10% levels is indicated respectively by *, ** and *** Khan et al. Financial Innovation (2020) 6:1 Page 7 of 13 the F-statistics value is 11.2702 that is higher than the upper bound. In third equation F COLCNM (COLCNM|CO2, PCI, NTGCNM, OLCNM) we used coal consumption as the dependent variable, results show that F-statistics value is 14.0254 that is higher than upper bound, results of F-statistics indicates the long run relationship of the third equation. In the fourth equation of bound test, we used F NTGCNM (NTGCNM|CO2, PCI, COLCNM, OLCNM) natural gas consumption as the dependent variable. Fstatistics result indicates a long-run relationship between the variables at 5%. In the last equation we used oil consumption as dependent variable F OLCNM (OLCNM|CO2, PCI, COLCNM, NTGCNM), F-statistics indicate no-long run association among variables because the F-statistics value is between the lower and upper bound at 5%. Table 5shows the results of long-run coefficients of ARDL model; CO 2 emission per capita is the dependent variable. Three proxies were used for measuring energy consumption, i.e., coal consumption, natural gas consumption, and oil consumption and per capita income is used to measure the economic growth in Pakistan. Non-renewable energy resources have a positive and significant effect on environmental degradation. Coal consumption indicates positively effect on CO2 emissions in Pakistan. 1% increase in consumption of coal for energy use increase CO 2 emissions up to 6.70% in Pakistan. Another source of non-renewable energy is natural gas that is mostly used for energy consumption in Pakistan; the coefficient of natural gas indicates the positive and statistically non-significant effect on CO2 emissions in Pakistan. The coefficient of natural gas indicates that 1% increase in the use of natural gas for energy consumption boosts environmental degradation 3.05% in Pakistan. Oil consumption is the primary source of energy consumption in Pakistan, and the coefficient indicates a Table 3 Serial Correlation and Heteroskedasticity Test Breusch-Godfrey Serial Correlation LM Test F-statistic 1.8491 Prob. F(2,23) 0.1724 Obs * R-squared 4.6827 Prob. Chi-Square(2) 0.0962 Heteroskedasticity Test: Breusch-Pagan-Godfrey F-statistic 0.6425 Prob. F(25,6) 0.7808 Obs * R-squared 7.8589 Prob. Chi Square(12) 0.7259 1, 5 and 10% levels is indicated respectively by *, **and *** Table 4 Bounds test for the existence of Co-Integration Dependent variables F-Statistics Outcomes F CO2 (CO2|PCI, COLCNM, NTGCNM, OLCNM) 5.2017 Co-integration F PCI (PCI|CO2, COLCNM, NTGCNM, OLCNM) 11.2702 Co-integration F COLCNM (COLCNM|CO2, PCI, NTGCNM, OLCNM) 14.0254 Co-integration F NTGCNM (NTGCNM|CO2, PCI, COLCNM, OLCNM) 6.6677 Co-integration F OLCNM (OLCNM|CO2, PCI, COLCNM, NTGCNM) 3.6615 Inconclusive Critical Bounds Values Significance I(0) Bound I(1) Bound 10% 2.45 3.52 5% 2.86 4.01 2.5% 3.25 4.49 1% 3.74 5.06 Khan et al. Financial Innovation (2020) 6:1 Page 8 of 13