does energy consumption contribute to environmental pollutants? evidence from saarc countries
TRANSCRIPT
RESEARCH ARTICLE
Does energy consumption contribute to environmentalpollutants? evidence from SAARC countries
Ghulam Akhmat & Khalid Zaman & Tan Shukui &Danish Irfan & Muhammad Mushtaq Khan
Received: 28 November 2013 /Accepted: 6 January 2014 /Published online: 23 January 2014# Springer-Verlag Berlin Heidelberg 2014
Abstract The objective of the study is to examine the causalrelationship between energy consumption and environmentalpollutants in selected South Asian Association for RegionalCooperation (SAARC) countries, namely, Bangladesh, India,Nepal, Pakistan, and Srilanka, over the period of 1975–2011.The results indicate that energy consumption acts as an im-portant driver to increase environmental pollutants in SAARCcountries. Granger causality runs from energy consumption toenvironmental pollutants, but not vice versa, except carbondioxide (CO2) emissions in Nepal where there exists a bidi-rectional causality between CO2 and energy consumption.Methane emissions in Bangladesh, Pakistan, and Srilankaand extreme temperature in India and Srilanka do not Grangercause energy consumption via both routes, which holds
neutrality hypothesis. Variance decomposition analysis showsthat among all the environmental indicators, CO2 in Bangla-desh and Nepal exerts the largest contribution to changes inelectric power consumption. Average precipitation in India,methane emissions in Pakistan, and extreme temperature inSrilanka exert the largest contribution.
Keywords Energy consumption . Environmental pollutants .
SAARC countries
Introduction
Climate change is a major challenge for humanity and hasemerged as an area of critical concern in the globalizedworld today. This problem is not amenable to nationalsolutions and, hence, regional collaborative efforts to mit-igate adverse impact of climate change are inevitable. Theregional organization in South Asia, i.e., South AsianAssociation for Regional Cooperation (SAARC), has tak-en several initiatives in this direction. Since 1987, theGovernments of SAARC at successive summits have re-iterated the need to strengthen and intensify regionalcooperation to preserve, protect, and manage the diverseand fragile ecosystems of the region including the need toaddress the challenges posed by climate change and nat-ural disasters (SAEO 2009).
Climate change is no longer an issue for the distant future.Climate change is already taking place, and the South Asiancountries, particularly the poorest people, are most at risk. Theimpacts of higher temperatures, more variable precipitation,more extreme weather events, and sea level rise are felt inSouth Asia and continue to intensify. These changes arealready having major impacts on the economic performanceof South Asian countries and on the lives and livelihoods ofmillions of poor people. The impacts result not only from
Responsible editor: Philippe Garrigues
G. Akhmat : T. ShukuiCollege of Public Administration, Huazhong University of Scienceand Technology (HUST), 1037Luoyu Road, Wuhan, People’sRepublic of China
G. Akhmate-mail: [email protected]
T. Shukuie-mail: [email protected]
K. Zaman (*)Department of Management Sciences, COMSATS Institute ofInformation Technology, Abbottabad, Pakistane-mail: [email protected]
D. IrfanDepartment of Computer Sciences, COMSATS Institute ofInformation Technology, Abbottabad, Pakistane-mail: [email protected]
M. M. KhanDepartment of Humanities, COMSATS Institute of InformationTechnology, Abbottabad, Pakistane-mail: [email protected]
Environ Sci Pollut Res (2014) 21:5940–5951DOI 10.1007/s11356-014-2528-1
gradual changes in temperature and sea level but also, inparticular, from increased climate variability and extremes,including more intense floods, droughts, and storms (WorldBank 2006). Sabouni et al. (2013) argued that global warmingand environmental issues have attracted the attention of manyresearchers and environmental scientists in the twenty-firstcentury due to the rapid increase of population and energyconsumption all over the world. Zwaan et al. (2002) investi-gates the impact on optimal CO2 abatement and carbon taxlevels of introducing endogenous technological change in amacroeconomic model of climate change. The results confirmthat including endogenous innovation implies earlier emissionreduction to meet atmospheric carbon concentration con-straints. Moreover, the development of nonfossil energy tech-nologies constitutes the most important opportunity for emis-sion reductions. Optimal carbon tax levels, reducing fossilenergy use, are lower than usually advocated. According toChapman (2007, p. 354),
“Pressure is growing on policy makers to tackle theissue of climate change with a view to providing sus-tainable transport. Although there is a tendency to focuson long-term technological solutions, short-term behav-ioral change is crucial if the benefits of new technologyare to be fully realized.”
Suri and Chapman (1998) econometrically quantify theeffect of environmental Kuznet curve (EKC) with respect tocommercial energy consumption and many environmentalvariables by using pooled cross-country and time series data.The results found that both industrializing and industrializedcountries have added to their energy requirements byexporting manufactured goods, and the growth has been sub-stantially higher in the former. At the same time, industrializedcountries have been able to reduce their energy requirementsby importing manufactured goods. Exports of manufacturedgoods by industrialized countries have thus been an importantfactor in generating the upward sloping portion of the EKC,and imports by industrialized countries have contributed to thedownward slope. Streets and Waldhoff (2000) present esti-mates of emissions for three of the major air pollutants inChina, i.e., sulfur dioxide (SO2), nitrogen oxides (NOx), andcarbon monoxide (CO). Emissions are estimated for each ofthe 29 regions of China including Hong Kong and Taiwanover the period of 1990–1995 along with two projections forthe year 2020 under alternative assumptions about levels ofenvironmental control. The results reveal that emissions of allthree species are concentrated in the populated and industri-alized areas of China, i.e., the Northeastern Plain, the eastcentral and southeastern provinces, and the Sichuan Basin.
Roca et al. (2001) empirically examine the EKC hypothesisfor different pollutants in different countries. This study fur-ther analyzes the trends of annual emission flux of six
atmospheric pollutants in Spain. The results show that thereis no correlation between higher-income level and smalleremissions, except for SO2 whose evolution might be compat-ible with the EKC hypothesis. Roca and Serrano (2007)further analyze the relationship between income growth andnine atmospheric pollutants in Spain by using structural de-composition analysis for the period of 1995–2000. The resultsreveal that economic growth increases atmospheric pollutantsin Spain. Tanczos and Torok (2007) examine the connectionbetween CO2 emission and climate change of the transportsector in Hungary. The results conclude that the effectivenessof transportation services must be increased, while the envi-ronmental pollution must be decreased or prevented. Martí-nez-Zarzoso et al. (2007) analyze the impact of populationgrowth on CO2 emissions in European Union (EU) countriesover the period of 1975–1999. The results show that theimpact of population growth on emissions is more than pro-portional for recent accession countries whereas for old EUmembers, the elasticity is lower than unity and nonsignificantwhen the properties of the time series and the dynamics arecorrectly specified.
Kander and Lindmark (2004) examine the evolution ofenergy use and pollution emissions in Sweden over the past2 centuries. The analyses show that technical change in abroad sense has been crucial for explaining the long-termdecline in both energy intensity and pollutant intensity, whilethe transition to the service economy had negligible effects.Changed preferences affected the decline in emissions after1970. Menyah and Wolde-Rufael (2010) examine the long-run and the causal relationship between economic growth,pollutant emissions, and energy consumption for South Africafor the period of 1965–2006. The results of Granger causalitytest found a unidirectional causality running from pollutantemissions to economic growth, from energy consumption toeconomic growth, and from energy consumption to CO2
emissions all without a feedback. South Africa has to sacrificeeconomic growth or reduce its energy consumption per unit ofoutput or both in order to reduce pollutant emissions. Varotsoset al. (2013) investigate the levels of the local and regionaloxidant concentration at Athens, Greece, by analyzing theobservations obtained at an urban and a rural station, during2001–2011 and 2007–2011. A progressive increase of thedaytime and nighttime average of [NO2]/[Ox] versus [NOx]is observed which shows a larger proportion of Ox in the formof NO2 when the level of NOx increases. In addition, similarresults are observed when studying the variation of meanvalues of [NO2]/[NOx] versus [NOx].
Zhang et al. (2013) analyzed the interactions amongChina’s economic growth and its energy consumption, airemissions, and air environmental protection investment dur-ing 2000–2007. The results show that energy consumptionrapidly rises with China’s fast economic growth; however,energy efficiency and environmental loading intensity from
Environ Sci Pollut Res (2014) 21:5940–5951 5941
energy consumption are reduced simultaneously, and theirimprovements fall far behind their economic growth rate.Impact of air emissions on human health (especially dust) isdecreased. The performance of air environmental protectioninvestment is declined in the study period. Khan et al. (2013)investigate the long-run and the causal relationship betweengreenhouse gas emissions, economic growth per unit of ener-gy use, and energy consumption in Pakistan over a 36-yeartime period, i.e., between 1975 and 2011. The finding sug-gests that energy consumption acts as an important driver forthe increase in greenhouse gas emissions in Pakistan. Theresults indicate that on average, causality runs from energyconsumption to greenhouse gas emissions, but not vice versa.According to Tan et al. (2012, p. 151),
“Since few decades ago, air pollution has become a hottopic of environmental and atmospheric research due tothe impact of air pollution on human health. Ozone isone of the important chemical constituents of the atmo-sphere, which plays a key role in atmospheric energybudget and chemistry, air quality, and global change.”
Zaman et al. (2012a) investigate the casual relationshipbetween energy consumption and agricultural technology fac-tors in Pakistan over the period of 1975–2010. The resultsinfer that tractor and energy demand has bidirectional rela-tionship, while irrigated agricultural land, share of agricultureand industry value added, and subsides have supported theconventional view; i.e., agricultural technology causes energyconsumption in Pakistan. On the other hand, neither fertilizerconsumption and high-technology exports nor energy demandaffects each other. Ahmed et al. (2013) examine the relation-ship between electricity consumption per capita and real percapita income in the context of Pakistan, over a 34-year period(between 1975 and 2009). The results provide evidence ofbidirectional causality between the electricity consumptionper capita and real per capita income on one hand, and energyconsumption per capita and real per capita income on the otherhand as the direction of causality has significant policy impli-cations. Zaman et al. (2013) employed the bivariatecointegration technique to estimate the long-run relationshipbetween four energy consumption variables (i.e., oil/petroleum consumption, gas consumption, electricity con-sumption, and coal consumption) and four macroeconomicfactors, i.e., balance of payment (BOP) factors, fuel factors,agricultural crop yield per hectare, and industrial productionitems. The results confirm the long-run relationship betweentotal commercial energy consumption and macroeconomicfactors in Pakistan. The empirical results only moderatelysupport the conventional view that energy consumption hassignificant long-run casual effect on macroeconomic variablesin Pakistan. The study finds evidence of unidirectional
causality between the commercial energy consumption factorsand macroeconomic factors in Pakistan.
Mudakkar et al. (2013) investigate the causal relationshipamong four energy consumption variables (i.e., nuclear ener-gy consumption, electricity power consumption, and fossilfuel energy consumption), economic growth, industrializa-tion, environmental degradation, and resource depletion inPakistan over the period of 1975–2011. The results infer thatthere exists a unidirectional causality running from nuclearenergy to industrial GDP, nuclear energy to water resources,and nuclear energy to carbon dioxide emissions, but not viceversa. Similarly, electric power consumption Granger causesagriculture GDP, but not the other way around; further, there isa bidirectional causality running between electric power con-sumption to population density in Pakistan. Zaman et al.(2012b) reinvestigate the multivariate electricity consumptionfunction for Pakistan, particularly, economic growth, foreigndirect investment, and population growth over a 36-year timeperiod, i.e., between 1975 and 2010. The results reveal thatdeterminants of electricity consumption function arecointegrated, and influx of foreign direct investment, income,and population growth is positively related to electricity con-sumption in Pakistan. These results conclude that income,foreign direct investment, and population growth induce anincrease in electricity consumption in Pakistan. Dynamicshort-run causality test indicates that there has been unidirec-tional causality which is running from population growth toelectricity consumption in Pakistan. Abdallah et al. (2013)examine the causal mechanism between indicators for sustain-able energy development related to energy consumption fromTunisian road transport sector during the period of 1980–2010. The results show that road-transport-related energyconsumption, transport value added, transport CO2 emissions,and road infrastructure are mutually causal in the long run.These results do not support the hypothesis of neutralitybetween energy and income for the Tunisian road transportsector. Also, there is a unidirectional causality running fromfuel price to road-transport-related energy consumption withno feedback in both the short and long runs.
Abu-Madi and Rayyan (2013) quantified three main green-house gases emissions from household energy consumption inthe West Bank, Palestine. The results show that the contribu-tion of households’ energy consumption in the West Bank toglobal CO2 emission is about 0.016 %, while contribution oftotal energy consumption by all sectors is about 0.041 %. Theresults show that wood is the most polluting energy source interms of CO2 and NOx emission, while electricity is the mostpolluting source in terms of SO2. Liu et al. (2014) describe thedevelopment of energy-based urban dynamic model simulat-ing the observed resource consumption, economic growth,and environmental impact of Beijing from 1999 to 2039.The simulation revealed that water resources are the mostimportant limiting factor for the sustainable development of
5942 Environ Sci Pollut Res (2014) 21:5940–5951
Beijing. Zaman et al. (2012c) investigate the influence ofagricultural technologies on carbon emissions in Pakistan byusing annual data from 1975 to 2010. The results reveal thatagricultural technologies act as an important driver for theincrease in carbon emissions in Pakistan. Results indicate thatunidirectional causality runs from agriculture machinery tocarbon emissions, but not vice versa. Agricultural technolo-gies are closely associated with economic growth and carbonemissions in Pakistan.
Akhmat and Zaman (2013) investigate the causal rela-tionship among nuclear energy consumption, commercialenergy consumption (i.e., oil consumption, gas consumption,electricity consumption, and coal consumption), and eco-nomic growth in South Asian countries, namely, Afghani-stan, Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan,and Sri Lanka, over the period of 1975–2010. The resultsreveal that nuclear energy consumption Granger causes eco-nomic growth in Nepal and Pakistan, while commercialenergy consumption, i.e., oil consumption Granger causeseconomic growth in Bangladesh, Bhutan, Maldives, Nepal,and Srilanka; gas consumption Granger causes economicgrowth in Bangladesh, Bhutan, India, and Maldives; elec-tricity consumption Granger causes economic growth inIndia and Srilanka; finally, coal consumption Granger causeseconomic growth in Bangladesh, Bhutan, Nepal, andSrilanka. Shahbaz et al. (2013) investigate the relationshipbetween energy use and economic growth in the case ofChina over the period of 1971–2011. The results showedthat energy use, financial development, capital, exports,imports, and international trade have positive impact oneconomic growth. The Granger causality analysis revealeda unidirectional causal relationship running from energy useto economic growth. Financial development and energy useGranger cause each other. There is bidirectional causalitybetween international trade and energy use. The feedbackrelation exists between financial development and interna-tional trade. There is also bidirectional causality that existsbetween capital and energy demand, financial developmentand economic growth, and international trade and economicgrowth. The above discussion confirms the strong correla-tion between environmental pollutants and energy consump-tion. In the subsequent section, an action has been made tofind statistical relationship between energy and atmospherein the selected SAARC countries.
The objective of this study is to empirically investigate theinfluence of electric power consumption on environmentalpollutants in selected SAARC countries. The more specificobjectives are the following:
I. To estimate whether there is a long-run relationshipbetween environmental pollutants, GDP per unit useof energy, and electric power consumption inSAARC region.
II. To explore the influencing directions between technolog-ically led growth and rural poverty.
III. To compare the influencing magnitude of technology onrural poverty in Pakistan.
The study arranges in the following manners. After intro-duction which has been discussed in Sect. 1 above, method-ology is mentioned in the Sect. 2. Results are discussed inSect. 3. The final section concludes the study.
Data source and methodological framework
The annual time series data are employed for the selectedSAARC countries, i.e., Bangladesh, India, Nepal, Paki-stan, and Srilanka over the period of 1975–2011. The dataset of environmental variables includes average precipita-tion in millimeter (mm) per year, carbon dioxide emis-sions from transport in percentage of total fuel combus-tion, agriculture nitrous oxide emissions in percentage oftotal, methane emissions from livestock in thousand met-ric tons of CO2 equivalent, and droughts, floods, andextreme temperature in percentage of population; average,1980–2009, are taken from Essential Climate Variables(ECV 2012) data access matrix which is published byGlobal Climate Observing Systems (GCOS). The dataset for energy consumption comprises electric power con-sumption in Kilowatt-hour (kWh) per capita which is usedto shed light on the possible impact of energy consump-tion on environmental variables in selected SAARC coun-tries in the presence of GDP per unit of energy use in PPP$ per kilogram of oil equivalent. The data from WorldBank development indicators are published by WorldBank (2012). All these variables are expressed in naturallogarithm and, hence, their first differences approximatetheir growth rates.
In this study, the focus is on the relationship amongenergy consumption, GDP per unit use of energy, and inthe presence of diverse atmosphere variables, i.e., averageprecipitation (AVGPER), carbon dioxide emissions (CO2),agriculture nitrous oxide emissions (NIRO), methane emis-sion from livestock (METH), and droughts, floods, andextreme temperature (EXTREME). These variables are se-lected because of their vital importance to emerging econo-mies like SAARC countries. The following Fig. 1 highlightsin schematic fashion the methodological approach adoptedin the paper. According to this framework, energy consump-tion has been checked on GDP per unit use of energythrough environmental indicators.
The following five equations (panels A to E) are used toassess the impact of energy consumption on environmentalvariables in selected SAARC countries:
Environ Sci Pollut Res (2014) 21:5940–5951 5943
PanelA : EPC;AVGPER;GDPENRG ð1Þ
PanelB : EPC;CO2;GDPENRG ð2Þ
PanelC : EPC;NITRO;GDPENRG ð3Þ
PanelD : EPC;METH;GDPENRG ð4Þ
PanelE : EPC;EXTREME;GDPENRG ð5Þ
Where, EPC represents the electric power consumption inkWh per capita, GDPENRG represents the GDP per unit useof energy in PPP $ per kilogram of oil equivalent, AVGPERrepresents the average precipitation in millimeter per year,CO2 represents the carbon dioxide emissions from transportin percentage of total fuel combustion, NITRO represents theagriculture nitrous oxide emissions in percentage of total,METH represents the methane emissions from livestock inthousand metric tons of CO2 equivalent, and EXTREMErepresents the droughts, floods, and extreme temperature inpercentage of population. All the variables seen in Table 1were expected to have a positive impact on rural poverty.
Econometric framework of the study
The time series data often show the property of nonstationarityin levels, and the resulted estimates usually provide spuriousresults. Thus, the first step in any time series empirical anal-ysis was to test for presence of unit roots to remove the
problem of inaccurate estimates. The other important stepwas to check the order of integration of each variable in a dataseries in the model to establish whether the data underhandsuffer unit root, and how many times it needed to bedifferenced to gain stationarity.
The test for cointegration consists of two steps: first, theindividual series are tested for a common order of integration.If the series are integrated and are of the same order, it impliescointegration1. Dickey and Fuller (1979, 1981) devised aprocedure to formally test for nonstationarity. The augmentedDickey–Fuller (ADF) test is used to test the stationarity of theseries. The ADF test is a standard unit root test: it analyzes theorder of integration of the data series. These statistics arecalculated with a constant and a constant plus time trend,and these tests have a null hypothesis of nonstationarityagainst an alternative of stationarity.
Johansen’s cointegration tests were applied on the se-ries of the same order of order of integration, i.e., I(1)series which determine the long-run relationship betweenthe variables. When series are cointegrated of order 1,trace test (Johansen’s approach) indicates a uniquecointegrating vector of order 1 and hence indicates thelong-run relationship. In the multivariate case, if the I(1)variables are linked by more than one cointegrating vec-tor, the Engle and Granger (1987) procedure is not appli-cable. The test for cointegration used here is the likeli-hood ratio put forward by Johansen and Juselius (1990),indicating that the maximum likelihood method is moreappropriate in a multivariate system. Therefore, this studyhas used this method to identify the number ofcointegrated vectors in the model. The Johansen andJuselius method has been developed in part by the litera-ture available in the field and reduced rank regression,and the cointegrating vector “r” is defined by Johansen asthe maximum eigenvalue and trace test or static; there is“r” or more cointegrating vectors. Johansen (1988) andJohansen and Juselius (1990) proposed that the multivar-iate cointegration methodology could be defined as:
EPCð Þt ¼ GDPENRG;AVGPER;CO2;NITRO;METH;EXTREMEð Þ
which is a vector of elements, considering the followingautoregressive representation:
EPCt ¼ π� þX
T ¼ 1
K
πi EPCð Þt−1 þ μt
1 If the series are integrated with the mixture of order of integration, i.e.,I(0) and I(1), it implies bond testing approach which was proposed byPesaran et al. (2001).
Electric PowerConsumption
(EPC)
Carbon dioxide emissionsfrom transport
(CO2)Average Precipitation
(AVGPER)
Methane emission fromlivestock (METH)
Agriculture nitrousoxide emissions
(NITRO)
GDP per unit Use of Energy(GDPENRG)
Droughts, floods &extreme temperature
(EXTREME)
Fig. 1 Research framework. Source: self extract
5944 Environ Sci Pollut Res (2014) 21:5940–5951
Johansen’s method involves the estimation of the aboveequation by the maximum likelihood technique and testingthe hypothesis Ho; (π=Ψξ) of “r” cointegrating relation-ships, where r is the rank or the matrix, π(0∠r∠Ρ),Ψ isthe matrix of weights with which the variable enterscointegrating relationships, and ξ is the matrix ofcoin tegra t ing vectors . The nul l hypothes is ofnoncointegration among variables is rejected when the es-
timated likelihood test statistic φi ¼ −n ∑t¼rþ1
p
ln 1−bλi
�� �exceeds its critical value. Given estimates of the eigenvaluebλi
� �,the eigenvector (ξi), and the weights (Ψi), we can
find out whether or not the variables in the vector (EPC)are cointegrated in one or more long-run relationships
among GDPENRG, AVGPER, CO2, NITRO, METH, andEXTREME.
This study investigates the influence of electric powerconsumption on environmental variables from two perspec-tives. One is to conduct the modified Granger causality andsecond, Johansen cointegration tests to explore the influencingdirections between electric power consumption and environ-mental variables, respectively; the other is to compare theinfluencing magnitude of different environmental variableson electric power consumption, based on the vector errorcorrection model (VECM) and variance decompositionapproach.
In order to undertake the modified version of Grangercausality for a VAR model with 3 lags (k=2 and dmax=1),we estimate the following system of equations:
EPCAVGPERGDPENRG
24
35 ¼ A0 þ A1
EPCNITROGDPENRG
24
35þ A2
EPCCO2GDPENRG
24
35þ A3
EPCMETHGDPENRG
24
35þ
A4
EPCEXTREMEGDPENRG
24
35þ ε1t
ε2tε3t
24
35
ð6Þ
Where A1 and A4 are the 3×3 matrices of coefficientswith A0 being a 3×1 identity matrix, and εt is the distur-bance term with zero mean and constant variance. FromEq. (6), we can test the hypothesis that electric powerconsumption does not Granger cause environmental vari-ables with the following hypothesis:
H10 ¼ a112 ¼ a212 ¼ 0
where a121 is the coefficient of the environmental variables in
the first equation of the system presented in Eq. (6). Besides,
we can test the opposite causality from environmental vari-ables to electric power consumption in the following hypoth-esis:
H20 ¼ a121 ¼ a221 ¼ 0
where a211 is the coefficient of the electric power consumption
in the second equation of the system presented in Eq. (6). Itshould be noted that we incorporate the variable GDPENRGin to Eq. (6) to avoid the omitted variable bias when weexamine the Granger causality bias and the Granger causality
Table 1 Variables in the equations and expected signs
Variables Measurement Expected signs Data source
Electric power consumption (EPC) kWh per capita World Bank (2012)
Intervening variable
GDP per unit use of energy (GDPENRG) PPP $ per kg of oil equivalent Positive World Bank (2012)
Atmospheric variables
Average precipitation (AVGPER) mm per year Positive ECV (2012)
Carbon dioxide emissions from transport (CO2) Percentage of total fuel combustion Positive ECV (2012)
Agriculture nitrous oxide emissions (NITRO) Percentage of total Positive ECV (2012)
Methane emissions from livestock (METH) Thousand metric tons of CO2 equivalent Positive ECV (2012)
Droughts, floods & extreme temperature (EXTREME) Percentage of population Positive ECV (2012)
Environ Sci Pollut Res (2014) 21:5940–5951 5945
between electric power consumption and environmentalvariables.
Results and discussions
Cointegration among environmental indicators and electricpower consumption
The present study conducts the ADF unit root tests for allvariables with regard to their stationary properties. The de-tailed results are shown in Table 2.
The results indicate that majority of the time series for fivedifferent countries are nonstationary, when the variables aredefined at levels with constant. While in case of GDPENRGfor Srilanka, AVFPER for Nepal and Pakistan, NITRO forSrilanka, CO2 for Pakistan, and EXTREME for India, the nullhypothesis of unit root defined at levels can be rejected at 5 %level of significance indicating the stationary time series, i.e.,I(0). However, the electric power consumption for all fivecountries becomes stationary when the series are differencedonce; the null hypothesis of unit root can be rejected after firstdifferencing at 5 % level of significance. This indicates thatthe variables are integrated of order 1, i.e., I(1).
In the next step, we take electric power consumption (EPC)as the dependent variable, and each environmental variablesand GDP per unit use of energy (GDPENRG) together as theindependent variables, and then the Johansen cointegrationamong them is tested. From the results in Table 3, we find thatexcept carbon dioxide emissions, all other environmentalvariables have at least one cointegration relationship withelectric power consumption in case of Bangladesh and Nepal;however, in case of Pakistan and Srilanka, except agriculturenitrous oxide emissions, remaining environmental indicatorshave at least one cointegration relationship with energy con-sumption. Therefore, we may say that, for the most part ofSAARC countries, environmental variables have significantlong-term equilibrium with electric power consumption.
Causality among environmental indicators and electric powerconsumption
Subsequently, we conduct the modified Granger causalitytests by Toda and Yamamoto (1995) for environmental indi-cators and electric power consumption. The GDPENRG isincorporated as an explanatory variable to avoid the omittedvariable bias. Results are shown in Table 4.
The results reveal that electric power consumption thatdoes not Granger cause average precipitation in Bangladesh,India, and Srilanka; nitrous oxide emissions in Bangladesh,India, Nepal, and Srilanka; carbon dioxide emissions in Ban-gladesh and Pakistan; methane emissions in India and Nepal;and extreme temperature in Bangladesh, Nepal, and Pakistanis rejected at 5 % level. It means that there are unidirectionalcausality runs from electric power consumption to environ-mental indicators. Methane emissions and extreme tempera-ture that both does not Granger cause electric power consump-tion are accepted via both route; therefore, we may concludethat both variables are causality independent in nature. Amongall environmental indicators, carbon dioxide emission is theonly variable which has a bidirectional causality relationshipwith the electric power consumption in Nepal. The overallresults reflect that environmental indicators are closely asso-ciated with electric power consumption and GDP per unit useof energy. In reality, the environmental indicators are closelyrelated to electric power consumption in selected SAARCcountries.
Variance decomposition analysis
In order to compare the contribution extents of SAARC’svarious environmental indicators to the change of electricpower consumption, the variance decomposition approach isadopted over the sample period. First, we take the electricpower consumption as the dependent variable, while environ-mental indicators and GDP per unit use of energy are theindependent variables, and conduct the Johansen’scointegration test among these variables.
Table 2 Order of integration by using ADF unit root test at 5 % level of significance
Countries EPC GDPENRG AVGPER NITRO CO2 METH EXTREME
Bangladesh I(I) I(I) I(1) I(I) I(I) I(I) I(I)
India I(1) I(1) I(1) I(1) I(1) I(I) I(0)
Nepal I(1) I(1) I(0) I(1) I(1) I(I) I(I)
Pakistan I(1) I(1) I(0) I(1) I(0) I(I) I(I)
Srilanka I(1) I(0) I(1) I(0) I(1) I(I) I(I)
The null hypothesis is that the series is nonstationary or contains a unit root. The rejection of the null hypothesis is based on MacKinnon (1996) criticalvalues; i.e., at constant, −3.639, −2.951, and −2.614 are significant at 1, 5, and 10% level, respectively. While at constant and trend, −4.252, −3.548, and−3.207 are significant at 1, 5, and 10 % level, respectively. The lag length is selected based on SIC criteria; this ranges from lag zero to lag one
5946 Environ Sci Pollut Res (2014) 21:5940–5951
The results indicate that there exists statistically significantcointegration among environmental indicators and electricpower consumption in the SAARC region. Next, we applythe variance decomposition approach based on the vectorerror correction model (VECM) to explore the influence ofelectric power consumption on environmental variables andcompare their contribution difference. Results are shown inFig. 2.
The results show that, among all environmental variables,carbon dioxide emissions from transport exhibit the largestinfluence, whose steady contribution level for changes withelectric power consumption approaches to 66.5 % in Bangla-desh and 57.7 % in Nepal. While in the case of India, averageprecipitation changes to 78.5 % due to changes in electricpower consumption. In Pakistan, methane emissions fromlivestock exert the largest contribution, i.e., 52.8 % due tousage of electric power consumption, whereas, extreme tem-perature, drought, and floods exert the steady contribution,i.e., 40.5 % in Srilanka.
The following conclusions have been emerged from thisstudy:
& Air quality: Air quality diminished in big cities followingthe industrial revolution and declined further as the popu-larity of the automobile increased during the last century.Air pollution at Earth’s surface is at least partially depen-dent upon weather conditions. National efforts to reduceair pollution initially centered on improving automobile
fuel efficiency standards or modifying gasoline composi-tion (Foy 2011).
& Air pollutant emissions: The most important drivingforces for air pollutant emissions are human emission-generating activities, mainly energy consumption, indus-trial activities, transportation, and agriculture. The level ofthese emission-generating activities determines theamount of emissions. However, apart from the activitylevel, the amount of emissions also depends on technolo-gies used and the applicable legal framework. For in-stance, the shift from coal and heavy fuel oil to cleanerfuels such as light fuel oil use and gas has lead to decreas-ing emissions (Loibl et al. 2010).
& Average precipitation: A higher yearly average precipita-tion would imply higher surface water flow (rivers) andmore hydraulic power available for hydroelectric powergeneration. An efficient use of Earth’s resources is byreliance on the renewable green energy (LebaneseEconomic Forum, 2012).
& Agriculture nitrous oxide emissions: Nitrous oxide is nat-urally present in the atmosphere as part of the Earth’snitrogen cycle and has a variety of natural sources. How-ever, human activities such as agriculture, fossil fuel com-bustion, wastewater management, and industrial processesare increasing the amount of N2O in the atmosphere.Nitrous oxide emissions are due in part to annual variationin agricultural soil emissions and an increase in emissionsfrom the electric power sector (EPA 2013).
Table 3 Results of Johansen’s cointegration tests
Bangladesh—hypothesizedno. of CE (s)
India—hypothesizedno. of CE (s)
Nepal—hypothesizedno. of CE (s)
Pakistan—hypothesizedno. of CE (s)
Srilanka—hypothesizedno. of CE (s)
Panel A Nonea Nonea Nonea Nonea Nonea
Series: EPC, AVGPER,GDPENRG
At most 1 At most 1 At most 1 At most 1 At most 1
At most 2 At most 2 At most 2 At most 2 At most 2
Panel B Nonea Nonea Nonea None None
Series: EPC, NITRO, GDPENRG At most 1a At most 1a At most 1 At most 1 At most 1
At most 2 At most 2 At most 2 At most 2 At most 2
Panel C None Nonea None Nonea Nonea
Series: EPC, CO2, GDPENRG At most 1 At most 1 At most 1 At most 1a At most 1a
At most 2 At most 2 At most 2 At most 2 At most 2
Panel D Nonea Nonea Nonea Nonea Nonea
Series: EPC, METH, GDPENRG At most 1a At most 1a At most 1a At most 1a At most 1a
At most 2 At most 2a At most 2 At most 2a At most 2
Panel E Nonea Nonea Nonea Nonea Nonea
Series: EPC, EXTREME,GDPENRG
At most 1a At most 1 At most 1a At most 1 At most 1
At most 2 At most 2 At most 2 At most 2 At most 2
Dependent variable in each Johansen’s cointegration test is HCAP
CE cointegration equationsa Denotes rejection of the hypothesis at the 5 % level
Environ Sci Pollut Res (2014) 21:5940–5951 5947
Table 4 Causality test resultsamong financial indicators andhuman capital
The modified Granger causalitytest approach used in the table isprovided by Toda and Yamamoto(1995). And the causality testsbetween financial indicators andhuman capital are based on thesignificance of chi-square statis-tics for Wald tests of VAR models
Null hypothesis Chi-square statistic Probability
Bangladesh
EPC does not Grange cause the changes in AVGPER 10.895 0.0123
AVGPER does not Granger cause the changes in EPC 1.852 0.174
EPC does not Grange cause the changes in NITRO 36.159 0.000
NITRO does not Granger cause the changes in EPC 0.858 0.108
EPC does not Grange cause the changes in CO2 8.4365 0.037
CO2 does not Granger cause the changes in EPC 1.889 0.184
EPC does not Grange cause the changes in METH 1.7541 0.4145
METH does not Granger cause the changes in EPC 3.2845 0.1021
EPC does not Grange cause the changes in EXTREME 9.039 0.028
EXTREME does not Granger cause the changes in EPC 1.629 0.342
India
EPC does not Grange cause the changes in AVGPER 12.373 0.006
AVGPER does not Granger cause the changes in EPC 2.017 0.201
EPC does not Grange cause the changes in NITRO 7.3596 0.061
NITRO does not Granger cause the changes in EPC 0.180 0.314
EPC does not Grange cause the changes in CO2 1.473 0.478
CO2 does not Granger cause the changes in EPC 3.411 0.181
EPC does not Grange cause the changes in METH 6.8765 0.075
METH does not Granger cause the changes in EPC 0.967 0.325
EPC does not Grange cause the changes in EXTREME 2.845 0.982
EXTREME does not Granger cause the changes in EPC 1.895 0.385
Nepal
EPC does not Grange cause the changes in AVGPER 2.985 0.945
AVGPER does not Granger cause the changes in EPC 1.996 0.412
EPC does not Grange cause the changes in NITRO 12.608 0.005
NITRO does not Granger cause the changes in EPC 0.542 0.214
EPC does not Grange cause the changes in CO2 6.589 0.004
CO2 does not Granger cause the changes in EPC 11.258 0.000
EPC does not Grange cause the changes in METH 7.214 0.000
METH does not Granger cause the changes in EPC 2.584 0.125
EPC does not Grange cause the changes in EXTREME 7.992 0.000
EXTREME does not Granger cause the changes in EPC 0.582 0.102
Pakistan
EPC does not Grange cause the changes in AVGPER 1.252 0.185
AVGPER does not Granger cause the changes in EPC 2.010 0.241
EPC does not Grange cause the changes in NITRO 4.958 0.028
NITRO does not Granger cause the changes in EPC 0.854 0.182
EPC does not Grange cause the changes in CO2 12.333 0.006
CO2 does not Granger cause the changes in EPC 2.001 0.201
EPC does not Grange cause the changes in METH 1.258 0.180
METH does not Granger cause the changes in EPC 2.225 0.189
EPC does not Grange cause the changes in EXTREME 11.142 0.011
EXTREME does not Granger cause the changes in EPC 2.221 0.112
Srilanka
EPC does not Grange cause the changes in AVGPER 5.7819 0.016
AVGPER does not Granger cause the changes in EPC 1.858 0.187
EPC does not Grange cause the changes in NITRO 2.335 0.128
NITRO does not Granger cause the changes in EPC 0.989 0.425
5948 Environ Sci Pollut Res (2014) 21:5940–5951
& Greenhouse gas (GHG) emission mitigation: Collectively,the agriculture sector can contribute to GHG emissionmitigation efforts in a number of ways, especially byincreasing soil carbon sinks, reducing emissions of nitrousoxide and methane, and providing biomass-based alterna-tives to fossil fuel use. Prominent GHG emission mitiga-tion strategies in agriculture include the following: im-proved agricultural land management to increase soil car-bon storage and enhanced livestock and manure manage-ment to reduce methane emissions (Siikamäki and Maher2012).
& Methane emissions: Natural gas systems and coal minesare the major sources of methane emissions in the energysector. Methane emissions are generated by industrialprocesses in the production of iron and steel and chemicals(EIA 2011).
& Climate and the electricity transmission system—the grid:By altering energy demand, climate and weather alsoaffect the electricity transmission system, known as thegrid. Severe weather, such as high winds and ice storms,
can knock down power lines, causing blackouts. Shifts inclimate can change when the peak electricity season oc-curs and cause problems for the grid. The transmission ofelectricity through large power grids is critical to meetvarying demands across the nation; when entire regionswarm, meeting the electricity demand is challenging(Earth 2009).
Summary and conclusion
The world is facing the challenge of global warming andclimate change issues. The anthropogenic driver of climatechange is the increasing concentration of GHG in the atmo-sphere (Fong et al. 2012). The objective of the study is toexamine the long-run relationship among environmental indi-cators, GDP per unit use of energy, and electric power con-sumption in five SAARC countries. This study explores theinfluencing directions and magnitude toward electric power
BangladeshIndia
AVGPER[57.2%]
CO2[66.5%]
NITRO[12.2%]
... ... ...
METH[17.5%]
EXTREME
[27.5%]
AVGPER[78.5%]
CO2[27.9%]
NITRO[5.9%]
... ... ...
METH[3.2%]
EXTREME
[11.7%]
Nepal
AVGPER[35.5%]
CO2[57.7%]
NITRO[11.3%]
... ... ...
METH[37.8%]
EXTREME
[27.1%]
Pakistan
AVGPER[11.2%]
CO2[37.5%]
NITRO[27.5%]
... ... ...
METH[52.8%]
EXTREME
[11.8%]
Srilanka
AVGPER[25.5%]
CO2[37.5%]
NITRO[37.5%]
... ... ...
METH[12.5%]
EXTREME
[40.5%]
Fig. 2 Variance decompositionanalysis. Source: authorsestimation. Note: Large bracketshows the magnitude of thecoefficients obtained fromvariance decomposition analysis
Table 4 (continued)Null hypothesis Chi-square statistic Probability
EPC does not Grange cause the changes in CO2 11.990 0.007
CO2 does not Granger cause the changes in EPC 20.220 0.000
EPC does not Grange cause the changes in METH 2.852 0.189
METH does not Granger cause the changes in EPC 3.012 0.098
EPC does not Grange cause the changes in EXTREME 1.852 0.154
EXTREME does not Granger cause the changes in EPC 2.012 0.132
Environ Sci Pollut Res (2014) 21:5940–5951 5949
consumption to environmental pollutants over the period of1975–2011. These objectives have been achieved with thesophisticated econometric techniques, i.e., cointegration tests,Granger causality, and variance decomposition. The resultsreveal that environmental indicators have significant long-term equilibrium with electric power consumption in selectedSAARC countries. Granger causality runs toward electricpower consumption to environmental pollutants, but not viceversa, which shows unidirectional causality runs betweenthem. The result of variance decomposition analysis showsthat the CO2 exerts the largest share to influence changes withelectric power consumption in Bangladesh and Nepal, whileaverage precipitation in India, methane emissions in Pakistan,and extreme temperature in Srilanka exert the largest contri-bution in the SAARC region.
The increased rate of deforestation in the SAARC regionis also attributed to no increased use of environment-friendlycompatible clean sustainable renewable form of energy likehydropower, thereby resulting to serious environmental deg-radation and unsustainable practices (Pradhan 2013). Thehead of the states of SAARC countries should have toapprove energy conservation or the power utility industry;the appropriate national policy for entering into the practicalaction would enhance economic development in a sustain-able basis (Dhungel 2009). Regional cooperation among thecountries of South Asia as well as with surrounding countriesis essential for meeting the region’s future energy needs(Siddiqi 2007). Environmental degradation and resource de-pletion are the two main factors that contribute to populationmovement and subsequent conflict. Other key factors includerapidly growing populations and inequitable distribution ofincome and resources. The poor have been affected the mostas a result of environmental problems. This implies thatpolicy recommendations focus on sustainable resource utili-zation, consider the factors that underlie the populationgrowth rates, and address the inequitable distribution ofincome and access to resources within and between coun-tries. Greater effort is also required for improving environ-mental awareness at all levels, involving local communitiesin environmental programs, and increasing support forNGOs to assist the government agencies in environmentalprotection (Vasudeva 2001).
Being increasingly aware of global warming, climatechange, and environmental challenges facing the South Asianregion, which mainly include sea level rise, deforestation, soilerosion, salutation, droughts, storms, cyclones, floods, glaciermelt and resultant glacial lake outburst floods, and urbanpollution. The heads of state or government in the SAARCregion reiterated the need to intensify cooperation within anexpanded regional environmental protection framework, todeal in particular with climate change issues. They were ofthe view that SAARC should contribute to restoring harmonywith nature, drawing on the ancient South Asian cultural
values and traditions of environmental responsibility and sus-tainability (SAARC 2008).
Acknowledgments This work was financially supported by the Na-tional 985 Project of Non-traditional Security at Huazhong University ofScience and Technology, People’s Republic of China. The authors arethankful to the anonymous reviewers for their comments and suggestions.Any remaining errors are the authors’ own responsibility.
References
Abdallah KB, Belloumi M, Wolf DD (2013) Indicators for sustainableenergy development: a multivariate cointegration and causalityanalysis from Tunisian road transport sector. Renew SustainEnergy Rev 25:34–43
Abu-Madi M, Rayyan MA (2013) Estimation of main greenhouse gasesemission from household energy consumption in the West Bank,Palestine. Environ Pollut 179:250–257
Ahmed W, Zaman K, Taj S, Rustam R, Waseem M, Shabir M (2013)Economic growth and energy consumption Nexus in Pakistan.South Asian J Glob Bus Res 2(2):25, Early Cite
Akhmat G, Zaman K (2013) Nuclear energy consumption, commercialenergy consumption, and economic growth in South Asia: bootstrappanel causality test. Renew Sustain Energy Rev 25:552–559
Chapman L (2007) Transport and climate change: a review. J TranspGeogr 15(5):354–367
Dhungel KR (2009) Does economic growth in Nepal cause electricityconsumption. Hydrol Nepal: J Water Energy Environ 5:37–47
Dickey D, FullerW (1979) Distribution of the estimators for autoregressivetime series with a unit root. J Am Stat Assoc 74(2):427–431
Dickey D, Fuller W (1981) Likelihood ratio statistics for autoregressivetime series with a unit root. Econometrica 49(1):1057–1072
Earth Gauge (2009) Climate, Weather and Energy Consumption. EarthGauge – A national environmental educational foundation program.Online available at: http://www.earthgauge.net/wp-content/CF_Weather_and_Energy.pdf (accessed 1 February, 2013)
ECV (2012) Essential Climate Variables (ECV) Data Access Matrix.Global Observing Systems Data and Information (GOSIC),Asheville, USA
EIA (2011) Emissions of Greenhouse Gases in the U. S. United StatesEnergy information administration, Washington, D.C. Online avail-able at: http://www.eia.gov/environment/emissions/ghg_report/ghg_methane.cfm (accessed 19 June, 2013)
Engle RF, Granger CWJ (1987) Cointegration and error-correction: rep-resentation, estimation, and testing. Econometrica 55(2):251–276
EPA (2013) Overview of Greenhouse Gases. United States EnvironmentalProtection Agency. Online available at: http://epa.gov/climatechange/ghgemissions/gases/n2o.html (accessed 18 June, 2013)
Fong W, Matsumoto H, Ho C, Lun Y (2012) Energy consumption andcarbon dioxide emission considerations in the urban planning pro-cess in Malaysia. Online available at: http://eprints.utm.my/6626/1/Wee_Hiroshi_Chin_Yu_EnergyConsumptionAndCarbonDioxide.pdf (accessed 17 March, 2013)
Foy D (2011) Energy & Air Pollution. Online available at: http://www.kean.edu/~csmart/Observing/18.%20Energy%20and%20air%20pollution.pdf
Johansen S (1988) Statistical analysis of cointegrating vectors. J EconDyn Control 12:231–254
Johansen S, Juselius K (1990) Maximum likelihood estimation andinference on cointegration with applications to the demand formoney. Oxf Bull Econ Stat 52(1):169–210
5950 Environ Sci Pollut Res (2014) 21:5940–5951
Kander A, LindmarkM (2004) Energy consumption, pollutant emissions,and growth in the long run: Sweden through 200 years. Eur RevEcon Hist 8(3):297–335
Khan MA, Khan MZ, Zaman K, Khan MM, Zahoor H (2013) Causallinks between greenhouse gas emissions, economic growth andenergy consumption in Pakistan: a fatal disorder of society. RenewSustain Energy Rev 25:166–176
Lebanese Economic Forum (2012) Paraguay and Iceland, a role model inrenewable energy usage and sustainability. Online available at:http://lebanese-economy-forum.com/2012/paraguay-and-iceland-a-role-model-in-renewable-energy-usage-and-sustainability/(accessed 15 March, 2013)
Liu G, Yang Z, Chen B, Ulgiati S (2014) Emergy-based dynamic mech-anisms of urban development, resource consumption, and environ-mental impacts. Ecol Model 271:90–102
Loibl W, Orthofer R, Köstl M (2010) Response functions for energyconsumption and air pollution. Online available at: http://www.plurel.net/images/D238.pdf (accessed 19 May, 2013)
MacKinnon JG (1996) Numerical distribution functions for unit root andcointegration tests. J Appl Econ 11(6):601–618
Martínez-Zarzoso I, Bengochea-Morancho A, Morales-Lage R (2007)The impact of population on CO2 emissions: evidence fromEuropean countries. Environ Res Econ 38:497–512
Menyah K, Wolde-Rufael Y (2010) Energy consumption, pollutant emis-sions, and economic growth in South Africa. Energy Econ 32(6):1374–1382
Mudakkar SR, Zaman K, Khan MM, Ahmad M (2013) Energy foreconomic growth, industrialization, environment, and natural re-sources: living with just enough. Renew Sustain Energy Rev 25:580–595
Pradhan GL (2013) Food, Water & Energy security through Hydropowerdevelopment. SAARC CCI Council on Climate Change, Energy &water Resources, Nepal electricity authority. Online available at:http://nea.org.np/neademo/nea.php?obj=gyanendra_paper(accessed 15 June, 2013)
Roca J, Serrano M (2007) Income growth and atmospheric pollution inSpain: an input–output approach. Ecol Econ 63(1):230–242
Roca J, Padilla E, Farré M, Galletto V (2001) Economic growth andatmospheric pollution in Spain: discussing the environmentalKuznets curve hypothesis. Ecol Econ 39(1):85–99
SAARC (2008) SAARC Action Plan on Climate Change. SAARCWorkshop Climate Change and Disasters: Emerging Trends andFuture Strategies 21–22 August 2008 Kathmandu, Nepal
Sabouni R, Kazemian H, Rohani S (2013) Carbon dioxide capturingtechnologies: a review focusing on metal organic framework mate-rials (MOFs). Environ Sci Pollut Res. doi:10.1007/s11356-013-2406-2
SAEO (2009) South Asia Environment Outlook (SAEO) 2009, NewDelhi,India. Online available at: http://saarc-sec.org/areaofcooperation/detail.php?activity_id=28 (accessed 19 June, 2013)
Shahbaz M, Khan S, Tahir MI (2013) The dynamic links between energyconsumption, economic growth, financial development and trade inChina: fresh evidence frommultivariate framework analysis. EnergyEcon 40:8–21
Siddiqi TA (2007) Viable and environment-friendly sources for meetingSouth Asia’s growing energy needs. Asia Pacific issues 83. Onlineavailable at: http://scholarspace.manoa.hawaii.edu/bitstream/handle/10125/3826/api083.pdf?sequence=1 (accessed 8 July, 2013)
Siikamäki J, Maher J (2012) Climate change and U.S. agriculture, IssueBrief 13. Online available at: http://www.rff.org/rff/Publications/upload/31814_1.pdf (accessed 13 April, 2013)
Streets DG, Waldhoff ST (2000) Present and future emissions of airpollutants in China: SO2, NOx, and CO. Atmos Environ 34(3):363–374
Suri V, Chapman D (1998) Economic growth, trade and energy: impli-cations for the environmental Kuznets curve. Ecol Econ 25(2):195–208
Tan KC, Li HS, Jafri MZM (2012) Relationship between ozone and theair pollutants in Peninsular Malaysia for 2003 retrieved fromSCIAMACHY. AIP Conf Proc 1528:51–156. doi:10.1063/1.4803586
Tanczos K, Torok A (2007) The linkage between climate change andenergy consumption of hungary in the road transportation sector.Transport 22(2):134–138
Toda HY, Yamamoto T (1995) Statistical inferences in vectorautoregressions with possibly integrated processes. J Econ 66(1–2):225–250
Varotsos CA, Ondov JM, Efstathiou MN, Cracknell AP (2013) The localand regional atmospheric oxidants at Athens (Greece). Environ SciPollut Res. doi:10.1007/s11356-013-2387-1
Vasudeva G (2001) Environmental Security: A South Asian Perspective.Online available at: http://unpan1.un.org/intradoc/groups/public/documents/apcity/unpan015801.pdf (accessed 7 July, 2013)
World Bank (2006) Managing Climate Risk: Integrating Adaptation intoWorld Bank Group Operations. The International Bank forReconstruction and Development/The World Bank, Washington,D.C.
World Bank (2012) World Development Indicators – 2012. World Bank,Washington D.C
Zaman K, Khan MM, Ahmad M, Rustam R (2012a) The Relationshipbetween Agriculture Technology and Energy Demand in Pakistan.Energy Policy 44(5):268–279
Zaman K, Khan MM, Ahmad M, Rustam R (2012b) Determinants ofElectricity Consumption Function in Pakistan: Old Wine in a NewBottle. Energy Policy 50(1):623–634
Zaman K, Khan MM, Ahmad M, Khilji BA (2012c) The relationshipbetween agricultural technologies and carbon emissions in Pakistan:Peril and Promise. Econ Model 29(5):1632–1639
Zaman K, Khan MM, Ahmad M (2013) Factors affecting commercialenergy consumption in Pakistan: Progress in energy. Renew SustainEnergy Rev 19:107–135
Zhang Y, Wu J, Li Y, Lin L, Li L,Wang Y,Wang L (2013) Evaluating therelationships among economic growth, energy consumption, airemissions and air environmental protection investment in China.Renew Sustain Energy Rev 18:259–270
Zwaan BCC, Gerlagh R, Klaassen G, Schrattenholzer L (2002)Endogenous technological change in climate change modeling.Energy Econ 24(1):1–19
Environ Sci Pollut Res (2014) 21:5940–5951 5951