Table Of Contentco Emissions, Energy Consumption,
2
Economic Growth and fdi in Vietnam
DinhHongLinh
ChungYuanChristianUniversity,Taiwan
[email protected]
Shih-MoLin
ChungYuanChristianUniversity,Taiwan
[email protected]
This study examines the dynamic relationships between co emissions,
2
energy consumption, fdi and economic growth for Vietnam in the pe-
riodfrom1980to2010basedonEnvironmentalKuznetsCurve(ekc)ap-
proach,cointegration,andGrangercausalitytests.Theempiricalresultsdo
notsupportthe ekc theoryinVietnam.However,thecointegrationand
Grangercausalitytestresultsindicateadynamicrelationshipamongco
2
emissions, energy consumption, fdi and economic growth. The short-
runbidirectionalrelationshipbetweenVietnam’sincomeand fdi inflows
impliesthattheincreaseinVietnam’sincomewillattractmorecapitalfrom
overseas. Inversely, fdi inflow is also driver of national income growth.
Theexistenceofbidirectionalrelationshipsinthelong-runprovidesim-
portantpolicyimplications.Werecommendimplementingadualstrategy
ofincreasinginvestmentinenergyinfrastructureandpromulgatingenergy
conservationpoliciestoincreaseenergyefficiencyandreducewastageof
energy.
KeyWords: ekc,Cointegration,GrangerCausality,pollutionheaven,
Vietnam’sco emissions,Vietnam’sEnergyconsumption,
2
fdi inflows,income
jel Classification: c33, o44, o53
Introduction
VietnamisoneofthefastestgrowingemergingeconomiesinAsia;aver-
agingaround7.1annual gdp growthfrom2006to2009,6.8in2010,
and6.0in2011.1However,rapideconomicgrowthisusuallyaccompa-
niedbyincreasedenergyconsumptionandmaycauseunexpectedeffects
on energy resources and the environment. Vietnam’s energy consump-
tion in 2010wasfour timeshigherthan itsconsumption in 1980, rising
from14.39thousandkilotonnes(kt)oilequivalence(1980)to58.91thou-
sandkt(2010).co emissionsgrewsignificantlyfasterthanthegrowthof
2
ManagingGlobalTransitions12(3):219–232
220 DinhHongLinhandShih-MoLin
energyconsumption,from16.82thousandktcarbondioxideemissionsto
150.23thousandkt;anincreaseofalmosttentimesforthesameperiod.2
There are many reasons that may explain the rapid growth of Viet-
nam’seconomy, fdi inflowsbeingoneofthemostimportantindicators.
It has increased from us$1.4 billion in 1980 to us$1.65 billion in 1990;
us$20.6 billion in 2000 and us$65.3 billion in 2010. The average fdi
annual growth rates observed for three periods are: 16 for 1980–1990;
29.3and12.4for1991–2000and2001–2010,respectively.Thisgrowth
in fdi can be attributed to Vietnam’s political, social, and macroeco-
nomic stability. Moreover, a country with 90 million people provides a
hugepotentialconsumermarket.Itslaborforceoffersyoung,skilled,and
relatively well-educated workers, with labor costs that are competitive
with other economies in the region. Its geographic location, abundant
naturalresources,andfavorablepoliciesareotherfactorsthatattractin-
vestments from many countries into Vietnam, propelling its economic
developmentforward.
Many previous papers have investigated the contribution of fdi to
economic development of different countries, such as those of Bende-
Nabendeetal.(2000),Alfaroetal.(2004),andWang(2009).Vu(2008)
and Anwar and Nguyen (2010) examined fdi’s influence on Vietnam’s
economicgrowth.Morerecentliteraturestudiedtherelationshipsamong
energy consumption, economic growth, and fdi. Acharyya (2009) and
PaoandTsai(2011)examinedthemultivariateGrangercausalityassoci-
ationamongco emissions,energyconsumption, fdi,and gdp forIn-
2
diaand bric,respectively.NguyenandAmin(2002)analyzedtheeffect
of fdi onenergydemandandenvironmentaldegradationinHanoi,the
capital city of Vietnam. However, the samples used by these studies are
limited, focusing only on one city, one country or on developing coun-
tries.Tothebestofourknowledge,noempiricalresearchhasbeenun-
dertakenyetabouttherelationshipsamongco emissions, energycon-
2
sumption,economicgrowthand fdi inVietnam.
Thispaperinvestigatesthecausalityrelationshipsamongenvironmen-
taldegradation–energyconsumption–economicgrowth– fdi inflows
in this country from 1980 to 2010. Investigating the causal nexuses be-
tweentheabovevariables,theestimatedresultsshowthatenvironmental
pollutantsareaffectedbytheusageofenergy,economicdevelopmentand
thechangesin fdi withinmorethan30years.Ourfindingswillprovide
valuablepolicyimplicationsforVietnamandotherdevelopingcountries.
Theresearchiswrittenasfollows:thesecondsectionexplainsliterature
ManagingGlobalTransitions
EconomicGrowthandEnvironmentinVietnam 221
reviewandhypotheses,thethirdsectiondemonstratesmethodologyand
empiricalresults,andthefourthsectionpresentstheconclusion.
LiteratureReviewandHypotheses
The causal relationship between energy consumption and economic
growth has attracted much attention from economists and scholars be-
causethisrelationshiphassignificantpolicyimplications.KraftandKraft
(1978)discovered unidirectionalcausalityfromincometoenergyusage
intheUnitedStatesbyusingsampledatafortheperiod1947–1974.This
findinghascontinuouslybeensupportedbyotherstudies.Forinstance,
Chen et al. (2007) has detected the existence of co-movement between
environmentalproductivityandincomeinChina.Lee(2005;2006)dis-
playedthattherehavebeenlong-runandshort-runcausalitiesfromen-
ergyconsumptionto gdp,butdidnotshowevidenceofviceversa.This
finding suggested that economic growth might have adverse effects on
energy conservation, which may be a transitory or permanent trend in
developing countries. The two-way directional causality has been rep-
resented in the case of the us, and one-way directional causality from
energyconsumptiontoincomewasfoundforotherdevelopedcountries.
However,thedetrimentaleffectsof economicgrowthtoenergyconser-
vationaredifferentiatedamongthesecountries.
The relationship between economic growth and environmental pol-
lutants has been analyzed by another pervasive approach known as the
EnvironmentalKuznetsCurvetheory.The ekc theoryclaimesthatthe
environmental pollutants has increased at the early stages of economic
growthbuttendstoreversebeyondacertainlevelofincomepercapita.
This suggests that there is an inverted U-shaped relationship between
environmental degradation and other economic variables. Grossman
andKrueger(1991)measuredincome’spotentialenvironmentalimpacts
withintheNorthAmericanfreetradeagreement(nafta).Managiand
Jena (2008) applied ekc hypothesis and found the appearance of rela-
tionship betweenenvironmentalproductivity andincomein India.Pao
andTsai(2011)employed ekc theoryandrecognizedthecausalityrela-
tionships among co emission, energy consumption and gdp in bric
2
countries, based on the time series dynamic characteristics of these in-
dicators.Thisfindingprovidespolicyimplicationsfordevelopingcoun-
tries,onhowtoqualifyforeigninvestmentssothattheycanmitigateits
harmful effects on the environment. In addition, the study of Sari and
Soyta (2007) found that energy consumption has a significant relation-
Volume12·Number3·Fall2014
222 DinhHongLinhandShih-MoLin
ship with economic development. Other authors such as Keppler and
Mansanet-Bataller(2010),NarayanandNarayan(2010)andPaoandTsai
(2010)statedthateconomicgrowthandenergyconsumptionareaccom-
paniedwithenvironmentaldegradationinbothdevelopedanddevelop-
ingcountries.ThesestudieshavegeneratedaninvertedU-shapedcurve
representingpollutantmagnitude,butthereisnoinevitabilityaboutthat.
CoondooandDinda(2008)andAkbostancietal.(2009)testedthe ekc
theory focusing on time series dynamics of income and co emissions.
2
The relationship between co and income is revealed in the long-run
2
basedontimeseriesanalysis,butitalsorevealedanN-shapedrelation-
shipfortwokindsofpollutantsinTurkey.
Themaincontributionofthispaperisitteststhe ekc hypothesisand
examinesthecausalityrelationshipsamongco emissions–energycon-
2
sumption–economicgrowth– fdi inVietnam.Weaccordinglyintro-
ducetwohypothesesasfollows:
h1 In Vietnam, co emissions increase in the early stages of economic
2
growth, but tend to decrease beyond a certain level of income per
capita. co emissions, energy consumption, and economic growth
2
havecausalityrelationshipswitheachotherandcanbeobserved.
Inrecentyears, fdi isconsideredasanimportantdrivingforceofeco-
nomic development. Rapid fdi inflows have raised questions whether
there is a relationship between fdi, energy consumption, and air pol-
lution deterioration. The issue of fdi, economic growth, and environ-
mentaldeteriorationhasbeenreceivingincreasedattentionsincethelast
decade. Several studies have applied time series dynamic with Granger
causality test and explored if there are significant nexuses among fdi,
economicgrowth,energyintensity,andco emissions.Zang(2001)and
2
Kim and Seo (2003) applied a vector autoregression model to present
the dynamic correlations between fdi, domestic investment, and out-
put.Theyfoundthateconomicgrowthhasastatisticallysignificantand
highly persistent influence on fdi inflows. Li and Liu (2005) found a
strong complementary connection between fdi and economic growth,
notonlyindevelopedbutalsoindevelopingcountries.Chakrabortyand
Nunnenkamp (2008) discovered the spillover effects between fdi and
India’s economic output both in the short-run and long-run. Sadorsky
(2010)showedevidencethatnet fdi hasastatisticallysignificantimpact
onenergydemandfromasampleof22emergingcountries.Moreover,the
relationship between fdi, energy usage, and pollutants has been men-
ManagingGlobalTransitions
EconomicGrowthandEnvironmentinVietnam 223
tioned in many studies. For examples, Mielnik and Goldemberg (2002)
examinedthelinkagebetween fdi,energyconsumption,andco emis-
2
sions from a sample of 20 developing countries, and found that energy
intensity declines as fdi increases. Pao and Tsai’s (2011) findings sup-
port the ekc hypothesis with evidence from bric. The results showed
strong bidirectional causality between emissions and fdi, and signifi-
cantunidirectionalrelationshipbetweenoutputand fdi.ForSoutheast
Asia,ChandranandTang(2013)suggestedthelong-runassociationbe-
tween fdi and co emissions within five asean countries (excluding
2
Vietnam).
Based on these studies’ findings, we predict the relationship among
fdi,economicgrowth,energyconsumptionandco emissionsinViet-
2
nam,andassumethat:
h2 fdi, economic growth, co emissions and energy consumption ex-
2
hibitcausalrelationshipswitheachotherinVietnam.
MethodologyandEmpiricalResults
data and variable form
This study used annual gdp per capita, annual fdi inflows and stocks
per capita measured by us Dollars at current prices and current ex-
changerates.Alldatasetswereobtainedfrom unctad statisticsdataba-
se.3 in represents gdp percapitaland fdi standsforinflowsandstocks
per capita. The energy consumption and co emissions were collected
2
fromtheWorldBankIndicatordatabase.4Theunitthatmeasuresenergy
consumption and co emissions is kt oil equivalence and kt co emis-
2 2
sions,respectively.Allvariablesaredatedfrom1980to2010,andallare
annualdata.
Thestandard ekc regressionmodelhasnaturallogarithmicformfor
both dependent and independent variables. The logarithmic quadratic
formisalsotakenforindependentvariables.Thenewvariablesinnatural
logarithmicformarelnin for in,lnfdi for fdi,lnco forco emissions
2 2
andlnen forenergyconsumption.
model
ekc theory implies that the environmental impact is an inverted U-
shaped function of income (in) and logarithm of the indicator is mod-
eled as a quadratic function of the logarithm of in. Based on the ekc
hypothesis,alinearlogarithmquadraticmodelisformedtoperformthe
Volume12·Number3·Fall2014
224 DinhHongLinhandShih-MoLin
table1 UnitRootTestResults
Test Statistic lnco lnen lnin lnin lnfdi
2
() t-statistic Level –. –.*** –. –. –.
stdif. –.*** –.** –.** –.** –.**
() t-statistic Level –. –. –.*** –. –.
stdif. –.*** –.*** –.*** –.*** –.***
notes Rowheadingsareasfollows:() adftest,()Dickey-Fuller glstest.**and
***denotesandlevelofsignificance,respectively.
relationships between co emissions, energy consumption, economic
2
growthand fdi asfollows:
lnco = β +β lnen +β lnin +β lnin2+β lnfdi +ν , (1)
2t 0 1 t 2 t 3 t 4 t t
wheret = 1,...,Tdenotesthetimeperiod,X isthevectorofexplanatory
t
variablesandν istheerrortermwhichisassumedtobeserialuncorrela-
t
tion.Accordingtothe ekc theory,thisstudyexpectsthesignsoflnen ,
t
lnin are positive because the higher ratio in energy consumption and
t
income,thegreaterco emissions.Incontrast,weexpectthatlnin2 will
2 t
haveanegativesign.
unit root test
Ascanbeseen,table1showstheAugmentedDickey-FullerandDickey-
Fuller gls unitroottestswhichareselectedtotestallserieswithtrend
andintercept.Thenullhypothesisassumedthatthedataserieshasaunit
root.Theresultsexhibitthatallvariablesarestationaryatthefirstdiffer-
enceat1and5significancelevels,thismeansthatthenullhypothesis
of a unit root is rejected.Hence,all data series are integrated of order 1
(I(1))andappropriateforfurthertesting.
Furthermore, assuming that vector Z includes lnco and all other
t 2t
variablesinmodel(2).Fromtheunitroottestresults,allcomponentsof
thevectorZ areI(1),orthefirstdifferenceΔZ = (1−L)Z isintegrated
t t t
oforderzero;whereListhelagoperatorofZ and(1−L)isthefirstdif-
t
ference. Thus, cointegration test should be applied to find the causality
relationshipsamongthesecomponents.
cointegration test
Johansen (1991) cointegration test is employed to examine whether the
seriesarecointegrated.Forexample,Gonzalo(1994)pointedoutthatJo-
ManagingGlobalTransitions
EconomicGrowthandEnvironmentinVietnam 225
table2 ResultsoftheJohansenCointegrationTest
Tracetest Maximumeigenvaluetest
() () () () () ()
r=0 r≥1 .*** r=0 r=1 .***
r≤1 r≥2 .*** r=1 r=2 .***
r≤2 r≥3 .** r=2 r=3 .***
r≤3 r≥4 . r=3 r=4 .
r≤4 r≥5 . r=4 r=5 .
notes Columnheadingsareasfollows:()nullhypothesis,()alternativehypothesis,
()tracestatistic.Traceandmax-eigenstatisticscalculatedatlevel;**and***de-
notesandlevelofsignificance,respectively.Probabilitiesarecomputedbyusing
asymptoticChi-squaredistribution,andristhenumberofcointegrationequations.sic
criteriaisusedtochoosethelaglength,maximumlaglengthsareseven.
hansen test is the optimal one when error terms are not normally dis-
tributed.MaddalaandWu(1999)suggestedtheimplicationofJohansen-
Fishertestwhichallowedsomerelationshipstobecointegrated.Because
all variables are integratedin the same order, this paper has applied Jo-
hansentestinthetermofvectorautoregressivemodel.Johansencointe-
gration test uses trace and maximum eigenvalue tests to determine the
number of cointegration relationships. Table 2 shows the results of Jo-
hansencointegrationtestwithnullandalternativehypotheses.
Thetraceandmaximumeigenvaluestatisticindicatethatthereisadif-
ferenceinsignificantlevelofeachcointegrationequation.Themaximum
eigenvaluetestiscarriedoutwithseparatetestsoneacheigenvalueand
hasthesharperalternativehypothesis.Itsresultsshouldbeusedinchoos-
ingthenumberofcointegratedrelationships.Basedonmax-eigenstatis-
tic, there are three significant cointegrating vectors at 1 and 5 levels.
Thefindingsindicateheexistenceoflong-runrelationshipamongvari-
ablesandthespuriousregressionisavoided.Theexistenceofcointegra-
tion among model (2) variables suggests that the ordinary least square
(ols)estimationisthebestestimatorinaccordancewiththefindingsof
AlvesandBueno(2003).
Theestimatedequationofmodel2using ols ispresentedintable3.
The results from equation (3) illustrate that the estimated coefficient of
lnen present the same sign as expected at 1 level of significance. The
estimated signs of lnin and lnin2 follow the expectation of hypothesis
1, which indicate that co emissions increase in the early stages of eco-
2
nomic growth, but tend to reverse beyond certain levels of income per
Volume12·Number3·Fall2014
226 DinhHongLinhandShih-MoLin
table3 EstimatedEquationofModel2Usingols
lnco = –. +.lnen +.lnin –.lnin –.lnfdi ()
2 2
se (.) (.) (.) (.) (.)
t-stat –.** .*** . –. –.
notes **and***denoteandlevelofsignificance,respectively.
capita.However,theestimatedcoefficientsoflnin andlnin2arestatisti-
callyinsignificant.Asaresult,wecannotconcludethat ekc hypothesis
is supported by the correlation between co emissions and income in
2
Vietnamduringtheperiod1980to2010.Furthermore,theestimatedre-
sults emphasize that co emissions is elastic with energy consumption,
2
in which a 1 increase in energy consumption will increase co emis-
2
sions by 1.954, or expressed differently, a one-unit increase in energy
consumptionincreasesco emissions by1.954percentagepoints.The
2
estimatedcoefficients magnitude of lnfdi hasa negative (–0.008) sign.
Althoughthismagnitudeisverysmallbutthenegativesignsuggeststhat
fdi canstilleffecttoreduceenvironmentdegradation.However,thises-
timatedcoefficientoflnfdi isinsignificant,meaningthatthereisnoev-
idencewhichshows fdi isinelasticinreducingco emissions.
2
granger causality test
The cointegration test has performed the existence of long–run equi-
librium relationships among co emissions, energy consumption, eco-
2
nomicgrowthand fdi.Grangercausalitytestinthetermofvectorerror-
correctionmodel(vecm)willrevealwhetherhistoricalvalueofonevari-
ablemightaffectthecurrentvalueofothervariables.Theseresultsdetect
thedirectionsofcausalrelationshipsamongvariablesinmodel(2).The
Granger causality test in the term of vecm framework is described as
follows:
ΔYt = α10+α11(Yt−1−Xt−1)+σ11ΔYt−p+σ12ΔXt−p
+β1Δzt−p+εt
ΔXt = α20+α21(Yt−1−Xt−1)+σ21ΔYt−p+σ22ΔXt−p
+β2Δzt−p+νt (2)
wheret = 1,...,Tdenotesthetimeperiod,Δrepresentschangeoperator,
Y andX isapairofendogenousvariables,zisthevectorofothervari-
t t
ables;β1andβ2arevectorsofitsparametersineachequation;ε ,ν are
t t
twoerrorterms;and(Yt−1−Xt−1)istheerrorcorrectionterm(ect).α11
ManagingGlobalTransitions
EconomicGrowthandEnvironmentinVietnam 227
andα aretheparametersthatshowthespeedofadjustmenttothelong-
21
runequilibriumwhichmightconfirmthelong-runrelationshipamong
variables.
Inthisarticle,thepairsof(X ,Y )include(lnco ,lnen),(lnco ,lnin
t t 2 2
andlnin2),(lnco ,lnfdi),andotherpairsarecombinationsofeachvari-
2
ablewithoneortwoothervariablessuchaslnen withlnin andlnin2or
withlnfdi andsoforth.
Table4representsshort-runGrangercausalityresultswiththenullhy-
pothesisthatthereisnocausalrelationshipineachpairofvariables.The
results support hypothesis 2, indicating the existence of short-run rela-
tionshipsamongvariables.Therearetwobidirectionalcausalityrelation-
shipsbetweenVietnam’sincomeand fdi inflowsandbetweenVietnam’s
energy consumption and fdi inflows, which can be interpreted as fol-
lows:therapiddevelopmentofVietnameseeconomyoverthelastthree
decades has been driven by the increase in fdi inflows. The higher in-
comeinturnattractsmoreforeigninvestors.Moreover,increasingeco-
nomicactivitybroughtaboutby fdi inflowsrequiresmoreenergy,and
the sufficient energy supply in turn attracts more foreign investment.
Thus,theeffectsof fdi varywidelyacrosssectors. fdi hasrisenupcon-
siderably since recent decades, and has close relationships with nation’s
incomeandenergyconsumption.Thefindingsareinlinewiththeprevi-
ousstudiesofZang(2001)andKimandSeo(2003).Theshort-runrela-
tionshipsbetween fdi andoutputarealsoimpliedintheresearchofLi
andLiu(2005)fordevelopedanddevelopingcountriesandChakraborty
andNunnenkamp(2008)instudyofIndianeconomy.Ontheotherhand,
the unidirectional causations are found from co emissions to fdi; in-
2
cometoco emissions andincometoenergyconsumption.Thesesup-
2
porttheinvestigationsofSadorsky(2010)andMielnikandGoldemberg
(2002),whichstatedthatenvironmentalpollutanthasunidirectionalef-
fectsonpromoting fdi inflows.
Thesignificanceoftheestimatedcoefficientsof ectsfrommodel(4)
expresseslong-run causal nexusamongdata series, whichsupportshy-
potheses 1 and 2. Table 5 has exploited four bidirectional causality rela-
tionshipsbetween:co emissionsandincome;energyconsumptionand
2
income;energyconsumptionand fdi;incomeand fdi.Thesefindings
areconsistentwiththoseofChenetal.(2007),whichfoundanassocia-
tionbetweenenvironmentandincomeinChina,andPaoandTsai(2011)
in their study of the same phenomenon in bric. The relationship be-
tween energy consumption and income is also in accordance with the
Volume12·Number3·Fall2014
228 DinhHongLinhandShih-MoLin
table4 ResultsofShort-RunGrangerCausalityTest
d(lnco ) d(lnen) d(lnin)† d(lnfdi)
2
d(lnco )→ . . .***
2
d(lnen)→ . . .**
d(lnin)→ .* .** .**
d(lnin)→ .* .*** .**
d(lnfdi)→ . .* .**
notes †and d(lnin);*,**and***denote,andlevelofsignificance,re-
spectively;→presentscausalitydirectionfromXtoY;↔detectsbidirectionalrelation-
shipbetweenXandY.
findingsofKepplerandMansanet-Bataller(2010),NarayanandNarayan
(2010) and Pao and Tsai (2010). On the other hand, the results indicate
oneunidirectionalcausalityrelationshipfromco emissionsto fdi in-
2
flows,whichimpliesthattherelativelylowerenvironmentalstandardhas
attracted fdi.Thebidirectionalcausalitybetweeneconomicgrowthand
energyusageimplies thatthesevariablesarejointly determinedandaf-
fect each other simultaneously. Furthermore, the bidirectional causali-
tiesbetweenco emissionsandenergyconsumptionwithincomeimply
2
thatVietnamhasbeendevelopingitseconomythroughincreasingitsen-
ergyconsumption.Withtheweaknessinenvironmentalprotectionreg-
ulations,theentryofinefficientenergytechnologiesmightleadtoenergy
wastageandenvironmentalpollution.
Conclusion
Thispaperteststhe ekc theoryinVietnam’seconomy.Basedontheem-
pirical results, we find that when income per capita is at 0.8 (in loga-
rithms) or 2,226 us dollars, co emissions begin to decline. However,
2
thisestimatedresultisstatisticallyinsignificant.Thismeansthatthedata
does not provide enough evidence to conclude that ekc hypothesis is
confirmedinVietnam’seconomy.
Second,thisstudyinvestigatesthedynamicrelationshipbetweenco
2
emissions, energy consumption, fdi and economic growth in Vietnam
fortheperiod1980to2010.ByusingtheGrangercausalitytestinthecon-
textof vecm,thispaperfoundtwoshort-runbidirectionalrelationships
between fdi inflows with Vietnam’s income and energy consumption.
Wealsofoundfourbidirectionalcausalityrelationshipsinthelong-run
between:co andincome;energyconsumptionandincome;energycon-
2
ManagingGlobalTransitions
Description:emissions, energy consumption, fdi and economic growth. The short- run bidirectional relationship between Vietnam's income and fdi inflows.
