ISSN(Online):23676957 ISSN(Print):23676361

Izvestiya Journal of Varna University of Economics 4 (2019)

I ZV E STI Y A Journal of Varna University of Economics http://journal.uevarna.bg

OIL PRICE SHOCKS AND ENERGY CONSUMPTION IN Patricia Iyore AJAYI 1, Adedayo Emmanuel LONGE 2, Oladayo Ayokunle OMITOGUN 3, Shehu MUHAMMAD 4

DepartmentofEconomics,UniversityofIbadan,Nigeria.Email:[email protected] 2,3 CentreforPetroleum,EnergyEconomicsandLaw(CPEEL),UniversityofIbadan,Nigeria. Email:[email protected],[email protected] 4 DepartmentofSocialScience,FederalPolytechnic,Bida,NigerState,Nigeria. Email:[email protected] JEL Q43, Q48 Abstract The study investigates the impact of oil price shocks on energy consumption in Nigeria using, BaiPerron Structural Breaks Test, NARDL approach and VAR model to analyse the data spanning from 1978through2016.Fromtheresults,inthelongandshortrun,positive andnegativechangesinbonnylightcrudeoilpricepositivelyimpacton energyconsu mptioninNigeria.Thestudyconcludesfromthesefindings thatirrespectiveofchangesinthepriceofoil,oilrelatedenergysources are still considered and this is because of the challenges faced in the development of the country’s energy sector in ge neral. Based on the findings,thestudyrecommendsthatstrategicinvestmentattentionshould Key words: bepaidtoalternativesourcesofenergysuchastherenewablesinorderto Oilprice,Energy reducelargedependencyonoilrelatedenergysuchasfossilfuels. Consumption,BaiPerron, NARDL,VARModel . © 2019 University of Economics – Varna Citation: AJAYI,P.I.,LONGE,A.E.,OMITOGUN,O.A.,Muhammad,S(2019).OilPriceShocks andEnergyConsumptioninNigeria.Izvestiya Journal of Varna University of Economics .63(4).p.275 293.

1. Introduction Intheliterature,theimpactofoilpriceshocksonmacroeconomicvariablesof oil importing and exporting countries has gained an interesting attention among researchersovertheyearsanddiversefindingshaveemergedasaresultofdifferent methodsandscopeofeachstudy.Anuncommondiscussionintheliteraturefrontisthe

275 Izvestiya 2019•Volume63•№4 relationshipbetweenoilpriceshocksandenergyconsumption.Seeingthattheprice of oil varies and is vulnerable to both internal and external shocks: shocks in oil prices predict the level of energy consumption and wealth of countries (most especially oil dependent developing countries such as Nigeria) (Odusami, 2010). Nigeriabeinganoilexportingcountryisfavouredonitsrevenuebypositivechanges inoilpriceandviceversa.Ontheotherhand,beenanimporterofrefinedoilproducts (premiummotorspirit,automotivegasoil,anddualpurposekerosene,amongothers) the economy is negatively affected when there is an increase in the price of oil,throughimporters’willingnesstosellathigherpricestocovertheadditionalcost incurredintheproductionprocess(Agrietal.,2016).Inmostdevelopingcountries, forexampleNigeria,toreducethecostburdenofrefinedoilproductstotheendusers, subsidy policy was introduced. The subsidy policy however balanced the link betweenthemovementinoilpriceandoilrelatedenergyconsumptionintheNigerian economy.ThemovementinthepriceofBonnyLightcrudeoilandoilrelatedenergy consumptionisillustratedinfigure1.

1000

500

0 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016

EnergyConsumptionPerCapital(OilEquivalent) BonnyLightCrudeOilPrice

Fig. 1. Relationship between Energy Consumption and Bonny Price Data Source: WDI and EIA (2017). From figure 1, it can be verified that the movement in oil price and energy consumption exhibits the same slope throughout the period, 1978 to 2016. This impliesthatthevariablescomplementseachother. As mentioned earlier, studies have focused on various impacts of oil price/energyconsumptionondifferenteconomies.AmongthesestudiesareOrhewere and Henry (2011), Saibu (2011), Nwosa and Akinbobola (2012),Onakoya et al., 276 P.I.Ajayi,A.E.Longe,O.A.Omitogun,S.Muhammad OilPriceShocksandEnergyConsumptioninNigeria

(2013),OsigweandArawomo(2015),&Shahbaz,Sarwar,WeiandMalik(2017)who established that there is a positive correlation between energy consumption and economic growth, while Chai, Zhou, Liang, Xing, and Lai, (2016) argued negative impact of energy consumption on economic growth. On the effect of oilprice, the results ofNkomo (2011), Shahbaz et al., (2017), Nwanna and Eyedayi (2016)& Omitogun,Longe,andMuhammad(2018)supporttheviewthat,unstableoilprices distortthegrowthofoilimportingcountriesespeciallywhenitispositive,aidsfiscal instabilityandvulnerabilityofbudgetimplementation,anditpositivelycontributesto thegrowthoftheoilexportingcountriessuchasNigeriaasrevenuerises(Nwanna andEyedayi,2016).Uniquely,VenandFouquet,(2016)claimedthatoilpriceshocks are onlyrelatedtoenergy markets ratherthan economic development. On oilprice relation with other macroeconomic indicators such as current account balances, ForignDirectInvestmentandExchangerate,asignificantrelationshiphasalsobeen established(seeLonge,AdelokunandOmitogun,2018;Omitogun,LongeandAjulo, 2018;andLonge,Muhammad,AjayiandOmitogun,2019) Fewstudies:Glausre&Lee(2002),Wang(2013),Ebrahim,Inderwildi,&King (2014),andJobling&Jamasb(2015)ontherelationshipbetweenoilpriceandenergy consumption argued that oil price shocks is mostly detrimental to oil product importingdevelopingcountriesasithamperstheirpersonalconsumptionexpenditure andretardtheirdemandforoilproductsaspricesoftheproductskyrocket.Itwasalso submitted that the effect of oil price shocks on energy consumption in developing countriesisalongrunphenomenon,asittakestimeforconsumerstoadjustorreact totheoccurrence(Saibu,2011). In a developing country such as Nigeria, that largely depend on oil as its mainstay, and currently experiencing continuous increase in its population, which callsformoreenergydemandtosustaintheirlivingstandard,fluctuationsinoilprice affectitseconomicplans.Also,thepowersectoroftheeconomywhichissupposed to provide for the increasing population and increased activities in the economy is facedwithseveredeficienciesintermsoftransmissionanddistributionofadequate supply.Thesedeficienciesleaveenergyconsumerstootheralternativesamongwhich istheoilrelatedmeanswhichiscommoninNigeria.Theoilrelatedmeansnegates the sustainable environment objective. Therefore, a question asked is does the unstable price nature of oil influence the consumption of oil related energy? This study therefore sets out to investigate the impact oil price shocks (positive and negativechangesinoilprice)onenergyconsumptioninNigeria. The rest of the study has four sections. Section two captures the literature review,sectionthreepresentsthedatasourceandmodelspecification,theanalytical

277 Izvestiya 2019•Volume63•№4 framework is presented in section four, while section five contains summary and conclusion. 2. Literature Review Thetheoreticalbackgroundofmostoftheempiricalstudiesontheconsequences of oil price on energy consumption is guided by notable theories that explore the importanceofpriceandincomeonindividuals’consumption.Amongthesetheories are; general equilibrium theory (Leon Walras, 1874), relative income theory of consumption(Duesenberry,1948),lifecycletheoryofconsumption(Modiglianiand Ando, 1957), and permanent income theory of consumption (Milton Friedman, 1957).The general equilibrium theory explains the relationship between supply, demandandpricesinaneconomy.Thetheoryisoftheopinionthattheeconomyisat generalequilibriumiftheconsumers,firms,industriesandallfactorservicesarein equilibrium as they are determined by commodities and factor prices. The relative income hypothesis explains that, consumption of a product is determined by the consumers’incomeandnotthestandardofliving.Thelifecycletheoryarguedthat theconsumptionandsavingsofanindividual,firmorindustryareplannedovertheir lifecycle.Thistheysuggestthatithelpsreducetheriskoflossesorspendingoutof plan.Thepermanentincometheorystatesthatchangesinpermanentincomeiswhat drivesconsumptionratherthanthetemporaryincome.Theyareofthesuggestionthat consumptionismostlyplannedonlongtermaverageincomechangesratherthanthe presentincome.Theyalsoproposedthatchangesinconsumptionarenotpredictable astheydependonindividualexpectations.Inrelationofthesetheoriestothecurrent study, oil prices largely determine consumption decision of oil related energy. ConsideringthestatusoftheNigerianpowersector,asdiscussedearlierfacedwith several challenges, among which are inadequacy in meeting the current energy demand, the profitability ratio of investing in the economy is a major concern to investors. While savings and standard of living is to individual users of oil related energyastheyconsidertheirlevelofincomeindeterminingthequantityconsumedof theoilrelatedenergyproducts. Wang (2013) used logistic smooth transition model and noted that in the G7 countries, oilprice has agradual smooth transitioneffect onpersonal consumption expenditure,inwhich1%increaseinoilpriceresultsto0.7410%increaseinpersonal consumption expenditure and 0.0517% increase in adjusted personal consumption expenditure.InCanadaandUSA,Valadkhani(2014)fromhisfindingsusingmarkov regimeswitchingmodelandBaiPerronsequentialmethodestablishedthatthereisa significantupwardshiftinthemarginaleffectofoilpricesonconsumerenergyprices 278 P.I.Ajayi,A.E.Longe,O.A.Omitogun,S.Muhammad OilPriceShocksandEnergyConsumptioninNigeria since 1999. The implication of this is that a percentage rise in crudeoil prices increases the price of consumed energy. Edelstein and Kilian (2009) employed bivariatevectorautoregressivemodelsintheUSAandaffirmedthatanincreasein energy price causes changes in expenditure patterns which weakly affects employmentandgeneraloutput.Itwasfurtheradmittedthattheriseinenergyprice causesconsumerstolosetheirpurchasingpowerwhichleadstoasignificantadverse effect on a wide array of consumer expectations as well as deteriorating consumer confidence in energy consumed. Odusami (2010) used linear and threshold multivariate autoregressive models the United States. The study confirmed that fluctuationsin theprices of oilpredict the level of consumption (energy inclusive) andwealthinthefuture . Similarly,Heetal.,(2016)usedridgeregressionanalysisandstatespacemodel inChinaandconcludedfromtheirfindingthatthereisanegativelinkbetweenenergy prices and energy consumption. In the same country, Chai et al., (2016) adopted VARmodelwithaconclusionthatinternationaloilpricehasanegativesignificanton energy consumption. Their findings further revealed that the elasticity of internationaloilpriceinrelationtoenergyconsumptionpercapitaandcarbondioxide emission is 4.31% and 6.51% respectively. In Sweden, Altai (2015) noted using vectorautoregressive(VAR)modelthatoilpricesdonothaveasignificanteffecton householdenergyconsumption. GlausreandLee(2002)inKoreausedVECMandbivariatecausalanalysisand established that fluctuations in oil price causes unstable relationship between real GDPandenergyconsumptionandabidirectionalcausalrelationshipbetweenenergy consumptionandrealincome.InChina,Zhangetal.,(2014)arguedthatanincrease inoilpricedoesnotreallyinfluencehouseholdconsumption,significantlyinfluenceit intheshortrun .Deniz(2017)usedpaneldataanalysisforoilimportingandexporting countries and fouind out that, volatility in oil price effect on renewable energy consumption in the countries differ, therefore the approach to analyse it for the countries should differ. For 8 OECD countries, Kuper and van Soest (2006) confirmedthattheeffectofoilpriceonenergyuseisasymmetricinnature. In Nigeria, Saibu (2011) used VECM Granger causality model to detect a unidirectionalcausalrelationshipbetweeneconomicgrowthandenergyconsumption, bidirectionalcausalityrunningfromenergyconsumptiontoprivateinvestment,while there is no causal link running from oil price to energy consumption and private investmentintheshortrun.Thishoweverimpliesthatoilpriceonlyhasalongrun effect on energy consumption and real GDP in Nigeria. Using same method, Orhewere and Henry (2011) asserted that a unidirectional relationship between gas 279 Izvestiya 2019•Volume63•№4 consumption and GDP in the shortrun and bidirectional causal relationship in the longrun.OsigweandArawomo(2015)arguedthatthereisabidirectionalcausallink between economic growth and energy consumption, electricity consumption and economic growth, electricity consumption and electricity price and no identifiable causal link between economic growth and energy product prices. Nwosa and Akinbobola (2012) used VAR model and submitted that a bidirectional causal link existsbetweenenergycosnumptionandsectorialoutputinNigeria.Therefore,thereis need for sectorial energy policies in the economy. Onakoya et al. (2013) using OrdinaryLeastSquares(OLS)methodarguedthattheenergyconsumptionlinkwith economicgrowthinNigeriaispositivelysignificantandgovernmentshoulddiversify itsenergyportfolio.Adoptingastructuralanalysiseconometrictechnique,Babatunde (2015)noteddifferentresponsesofexchangeratetooilpriceshocksinNigeriawhich laterinfluencesthepurchasingpowerofrefinedoilproductsintheeconomy.Itwas observedfromthefindingsthattheeconomyneedsarehabilitationofitsrefineriesto curbthiseffect. Fromtheempiricalfront,itisobviousthatlackofconsensusontherelationship between oil price, energy consumption and macroeconomic variables is due to the scope,methodologyandobjectiveoftheexistingstudies.InthecontextofNigeria, thecausallinkbetweenoilpriceshocksandenergyconsumptionhasbeendiscussed, butalackingdiscussionintheliteraturefrontistheasymmetriclinkbetweenoilprice shocksandenergyconsumptioninNigeria.Also,theresponseofenergyconsumption toshocksoverperiodsisyettobediscussed.Thisstudyhowever,adoptstheNon linear Autoregressive Distributed Lag (NARDL) and Variance Impulse Response FactortoexaminethisphenomenoninthecontextofNigeria. 3. Methodology and Data ThisstudyfollowedthemodelofGlasureandLee(2002),&VenandFouquet (2016).Themodelisspecifiedinitslinearlinearformas:

, , (1) isenergyconsumption, isBonnyLightcrudeoilprice, isconsumer priceindex,and isofficialexchangerateofnairatodollar. Themodelisrespecifiedinaneconometricloglogformas; (2) , , implies log form of energy consumption, log formofbonnylightcrudeoilprice,logformofconsumerpriceindexandlogformof

280 P.I.Ajayi,A.E.Longe,O.A.Omitogun,S.Muhammad OilPriceShocksandEnergyConsumptioninNigeria officialexchangerateofnairatodollar, istheinterceptofthemodel, are thecoefficientsoftheestimatedparameters, istimeand istheerrorterm. ThestudyadoptsARDLboundstesttotestforlongruncointegrationamong the variables inline withPesaran and Shin (1999)because ofits merits over other methods.Themeritsinclude;itsapplicabilityforvariablesstationaryatI(0),I(1)or combination of both [I(0) and I(1)]. Secondly, the error correction model can be estimated through a simple linear transformation that integrates both the shortrun adjustment and the longrun estimate without omitting any longrun information. Also,Pahlavanietal.,(2005)andAdometal.,(2012)addedthatthemethodprovides unbiasedlongrunestimatesevenwhensomevariablesareendogenised. Tovalidatelongruncointegrationamongthevariables,thenullhypothesisis stated as against the alternative hypothesis of longrun co : 0 integrationexistence .WeusedtheFstatisticsasaguidefor : 0 ourdecision.IftheFstatisticsisgreaterthantheUpperbound,weacceptthealterna tivehypothesisthatalongruncointegrationexists,ifotherwise,wedonothaveany reasontorejectthenullhypothesisofnolongruncointegration.IftheFStatistics liesinbetween,thenourresultisinconclusive. Inordertoachievetheobjectiveofthisstudy,equation(2)istransformedintoa nonlinearformandtheNonLinearAutoregressiveDistributedLag(NARDL)long run and shortrun model were formulated where BL was decomposed intopositive andnegativeperiodchanges

∆InE ϑ ∆ ∆ ∆ ∆InCPI ∆ 3 The study recognizes some breaks caused by some shocks through policy implementationandreformulatedtheNARDLwithstructuralbreaksidentified

281 Izvestiya 2019•Volume63•№4

∆InE ϑ ∆ ∆ ∆ ∆InCPI ∆

4

Fromequation(3and4), ∆denoteschangesinthevariablesintheshortrun, is theoptimallaglength, errortermattime.Theparameters ( arethe 1,2,3,4,5 correspondinglongrunmultiplier,andtheparameters 1,2,3,4,5aretheshort rundynamicoftheNARDLmodel. istheparameterestimateoftheerrorcorrec tion model. is a dummy variable for the break defined as for ,otherwise 0. t representsthetimeperiod; isthestructural 1 breakdatewhere r =1,2,3,……., k and isthecoefficientofthebreakdummy. The study used annual time series data on energy consumption per capita (oil equivalent per capita), official exchange rate (LCU per US$, period average) and Consumer Price Index from World Development Indicators (WDI) (2017) and Oil Prices(WTIandBrent)fromBPStatistics(2017)andBonnyLightcrudeoilpricefrom EnergyInformationAdministration(EIA)(2017).Thedataspansfrom1978to2016. 4. Results and Outcomes It is important taking the preestimation analysis of the variables into considerationbeforeestimatingtheparametersinthestudy.Thepreestimationtests forthisstudyincludethedescriptivestatistics,correlationtestthatshowsthelinear association among the variables under study, and the ARDL bounds test: used to estimatethelongruncointegratingexistenceamongthevariables. Descriptive Statistics Thedescriptivestatisticsshowsthebehaviouralpatternofthevariablesoverthe yearsunderstudy.Fromtheresultaspresentedintable1,thevariablesbehavewithin theirmaximumandminimumvalues,whichimpliesthatthereactionofthevariables to policy review within the period under study is not outrageous, nor deviate too muchfromtheexpectedoutcomeorequilibrium.Thestandarddeviationresultshows thattheleastvolatilevariableisenergyconsumption(InE),whilethemostvolatile

282 P.I.Ajayi,A.E.Longe,O.A.Omitogun,S.Muhammad OilPriceShocksandEnergyConsumptioninNigeria variabletoshocksisgrossdomesticproduct(InGDP).TheJaqueBeraStatisticsalso showthatthevariablesarenormallydistributed.Thisisconfirmedwithprobability valuegreaterthan10%significancelevel(SeeTable1). Table 1 Descriptive Statistics

Mean 2.852 1.536 1.037 1.301 Maximum 2.902 2.071 2.265 2.404 Minimum 2.810 1.134 0.478 0.262 Std.Dev. 0.023 0.288 0.926 0.928 Jarque Bera 1.621 3.266 3.997 4.297 Probability 0.445 0.195 0.136 0.117 Observations 39 39 39 39 Source: Authors Computation (2019).

Correlation Matrix Thecorrelationtestshowstheleveloflinearassociationbetweenthedependent andtheindependentvariables.Table2presentstheresults.Theresultconfirmsthat theindependentvariables(BonnyLightCrudeOilPrice,ConsumerPriceindexand Officialexchangerateofnairatodollar)haveastrongpositiveandsignificantlinear associationwiththedependentvariable(energyconsumption). Table 2 Correlation Test

1 0.804 1 (0.000)* 0.837 0.594 1 (0.000)* (0.000)* 0.822 0.511 0.975 1 (0.000)* (0.001)* (0.000)*

*, **, *** implies significance level at 1%, 5% and 10% respectively While the parenthesis ( ) denotes the Prob. values Source: Authors Computation (2019).

283 Izvestiya 2019•Volume63•№4

Unit root test The study adopts the Augmented Dickey Fuller (ADF) testand PhillipPerron (PP)test.Theunitroottestsshowifthevariableshaveaunitrootproblemandifthey aremeanrevertinginthelongrun.Fromtheresultsintable3,theADFconfirmed stationarityofthevariablesafterfirstdifferencing, whileforPP,onlyCPIwasnot stationary at level and first difference. The implication of these results is that the variablesarenotmeanrevertinginthelongrun,canbeconcludedthatthereisunit rootproblemandthereisneedtotestifthereisexistenceoflongruncointegration amongthevariablesinthestudy. Table 3 Unit Root Results

ADF PP

First Order of Variables Level Level First Difference Difference Integration 3.137 4.962* 2.564 4.853* I(1) 1.609 5.427* 1.790 5.427* I(1) 0.862 3.756** 0.804 2.878 I(1) 1.014 5.281* 1.139 5.259* I(1) 4.219 4.227 CV 1% 3.533 3.537 5% 3.198 3.200 10% *, **, *** implies significance level at 1%, 5% and 10% respectively, Source: Authors Computation (2019). ARDL Bounds Test TheARDLboundstestisprimarilytotestlongruncointegrationrelationship amongvariables.ThestudytestsforusingthelogarithmformofBonnyLightprice withoutdecomposingittochecktheperiodiceffectandthestructuralbreaksisalso notaccountedinthistest.Fromtheresultaspresentedintable4,thereisalongrun cointegrationamongthevariables.Thisimpliesthatthevariablescanbeestimated. Table 4 ARDL Bounds Test Results

Model for Estimation F-Statistics Lower-Upper bound at 5% 4.62 3.47 4.45 / / / Source: Authors Computation (2019). 284 P.I.Ajayi,A.E.Longe,O.A.Omitogun,S.Muhammad OilPriceShocksandEnergyConsumptioninNigeria

Structural breaks test Thestructuralbreaktestshows2010asthebreakperiodafterdecomposingthe bonny light oil price intopositive and negative changes. This can be traced to the periodwheretheoilpricewasjustgainingitsformfromtheshocksexperiencedin 2009 (see figure 2 for detailed movement in the oil prices). Table 5 presents the structuralbreakresult.

150

100

50

0 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016

Bonny Light Brent WTI Fig. 2. Trend analysis of Bonny Light, Brent and WTI Crude oil prices Table 5 Bai-Perron Test

Country Break Period Break Range Nigeria 2010 20102016 Compiled by Authors (2019). ARDL with Breaks Estimation Theresultisinterpretedanddiscussedbasedonthelagselectedautomaticallyby SchwarzInformationCriterion(SIC)[4,0,4,4,4](Seefigure1).

285 Izvestiya 2019•Volume63•№4

Schwarz Criteria (top 20 models)

-7.25

-7.30

-7.35

-7.40

-7.45

-7.50

-7.55 RDL(4, 0, 4, 4, 4) 4, 4, 0, RDL(4, 4) 4, 4, 1, RDL(4, 4) 4, 4, 1, RDL(2, 4) 4, 2, 1, RDL(2, 4) 4, 4, 1, RDL(1, 4) 3, 4, 1, RDL(1, 4) 3, 4, 0, RDL(4, 4) 4, 4, 2, RDL(4, 4) 4, 2, 1, RDL(3, 4) 4, 3, 1, RDL(2, 4) 4, 4, 1, RDL(3, 4) 3, 4, 3, RDL(1, 4) 4, 4, 2, RDL(2, 4) 4, 4, 0, RDL(2, 4) 3, 4, 1, RDL(2, 4) 4, 2, 2, RDL(2, 4) 4, 2, 1, RDL(4, 4) 3, 2, 1, RDL(2, 4) 3, 4, 0, RDL(1, 4) 4, 2, 0, RDL(4,

A A A A A A A A A A A A A A A A A A A A The estimation is done accounting for the break. The break period (2010, identified through the BaiPerron test) included in the estimation as a dummy variable, shows a positive significant impact in the shortrun and long run. The results show that, in the shortrun, positive changes in oil price positively, but insignificantly influence energy consumption in Nigeria. This implies that as the bonny light price changes positively, energy consumption increase by 0.008% insignificantly. Negative changes in bonny light oil price also influence energy consumption significantly in a positive direction. The implication of this is that, negativeperiodsofchangesinbonnylightpricehaveasignificantincreasingimpact of 0.045%. Consumer price index negatively affects energy consumption at 10% significancelevel.Thisimpliesthat,averagepriceofgoodsandservicesinNigeria discouragetheconsumptionofoilrelatedenergyinNigeriaby0.032%.Theofficial exchangerateofnairatodollarinNigeriaalsoshowsanegativelinkwithoilrelated energy consumption inNigeria and significant at 5%.This implies that, the rate at which the naira is falling against dollars ($) is not in favour of oil related energy consumption in Nigeria, as it discourages consumption of energy significantly by 0.038%.Theerrorcorrectionresultisrightlysignedandsignificant.Theimplication oftheresultisthat,deviationsofenergyconsumptioninNigeriafromtheequilibrium intheshortruncouldbecorrectedbyapproximately85%inthelongrun.Thisalso implies that the independent variables (bonny light oil price, consumer price index andexchangerate)arestrongdeterminantsofenergyconsumptioninthelongrun. Inthelongrun,positiveandnegativechangesinthebonnylightpricepositively and insignificantly have an impact on energy consumption in Nigeria. Positive and

286 P.I.Ajayi,A.E.Longe,O.A.Omitogun,S.Muhammad OilPriceShocksandEnergyConsumptioninNigeria negativechangesinthebonnylightpriceincreaseenergyconsumptionby0.009%and 0.106%respectively.Thisimpliesthatirrespectiveofthechangesinthepriceofbonny lightinthelongrun,oilrelatedenergyisstillconsumed.Thisisbecause;energysources relatedtooilarethemainalternativesourceofgeneratingpowerforbothhouseholdsand industries in the Nigerian environment. This therefore leaves the consumers with no choiceofconsideringtheenergysource.Theaveragepriceofgoodsandservicesinthe long run has a negative significant impact on energy consumption. This means that averagepriceofgoodsandservicesinthelongruninNigeria,reducesthepurchasing powerofconsumerstoconsideroilrelatedenergyasanalternativeforconsumptionby 0.043%. Official exchange rate of naira to dollar in the long run shows a positive significantimpactonenergyconsumption.Thisimpliesthatasthevalueofnairatodollar depreciatesfurther,consumersarestillencouragedtowardsconsumingmoreofoilrelated energyby0.069%.Thisalsocanbetaggedtothenochoiceeffectofalternativeenergy sourcesthanoilrelatedsources,asotheralternativesareexpensivetoaccess. Notably,theNARDLwithbreaksboundstestresultshowsastrongerexistence oflongruncointegratingrelationshipamongthevariables.Therefore,accountingfor structural break is very significant in studying oil price and energy consumption nexus in Nigeria. In addition, the diagnostic tests (Ramsey RESET test and Heteroscedasticity test) reveal that the model is well specified and free of heteroscedasticityproblem.Thatis,thevariabilityofthedependentvariableusedin thestudyisunequalrelatingnottotheindependentvariablesthatpredictit. Table 6 NARDL with breaks Estimation Results

NARDL Estimate Variable Coefficient Prob. 0.008 0.693 : 0, 4, 4, 4 ∆ 0.065 0.000 ∆ 0.017 0.454 ∆ 0.030 0.264 ∆ 0.045 0.031 ∆ 0.009 0.702 ∆ 0.011 0.792 ∆ 0.030 0.400 ∆ 0.032 0.085 ∆ 0.009 0.214 ∆ 0.007 0.574 ∆ 0.019 0.058 ∆ 0.038 0.002 ∆ 0.033 0.000 ∆1 0.849 0.004 287 Izvestiya 2019•Volume63•№4

Long Run Coefficients

Variable Coefficient Prob.

0.009 0.691 0.106 0.166 0.043 0.001 0.069 0.019 0.038 0.018 1 23.182 ARDL with breaks Bounds Test 1.608 0.234 Ramsey RESET Test 1.387 0.248 Heteroscedasticity Test Source: Authors Computation (2019). VAR Model We also proceed to use the VAR model to examine the response of the dependentvariable(energyconsumption)toshocksinthecoreindependentvariables (positive andnegativebonny light crude oilpriceshocks). Common in studies, the Impulse Response Functions (IRF) is used, but it has been criticized based on its sensitivity to variables ordering (Babatunde, 2015). However, we adopted GeneralisedImpulseResponseFunctions(GIRF)forthispurpose.Wewereableto adopttheVARmodelbecauseofthevariablesstationarityatfirstdifferencing.The VARmodelinthisstudyisspecifiedas: ⋯ 1, 2, … , Τ arethe parameter matrices. is the optimal lag , , , , lengthdeterminedbySIC. isthecommonvectorerrors. isthetimerange. Figure3and4impliesthat,( note: BL1andBL2onthegraphimpliespositive and negative changes in oil price respectively) as the positive changes in price increase, energy consumption increases throughout the first 4 periods, it declines throughoutthe5 th and6 th periodandmaintainsaconstantflowthroughthe7 th period tothe10 th period.Negativechangesinbonnylightcrudeoilpriceinthefirst3years and 6 months initiated energy consumption to decline, before it adjust back to the steady state point 6 months after the 3 rd year. After this adjustment, energy consumption fall below the steady state through throughout the 4 th period to 6 th period.However,itadjustedbacktothesteadystateequilibrium6monthsafterthe 6th periodandincreaseinstantaneouslythrough7 th periodtothe10 th period.

288 P.I.Ajayi,A.E.Longe,O.A.Omitogun,S.Muhammad OilPriceShocksandEnergyConsumptioninNigeria

Response to Generalized One S.D. Innovations ± 2 S.E. Response of INE to BL1 Response of INE to BL2

.02 .02

.01 .01

.00 .00

-.01 -.01

-.02 -.02 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10

Fig. 3 and 4. Impulse Response of Positive and Negative Brent Oil Price

Discussion of Findings Comparing the study findings to some existing studies, the correlation result conformswiththestudyofAgrietal(2016)thatthereisastrongpositivecorelation between oil price and energy consumption considering Nigeria as a developing country. Against the study of Chai et al., (2016), oil price fluctuations positively impactonenergyconsumptioninNigeria.Thisisbecause,thenatureoftheeconomy (China)studiedbyChaietal.(2016)islargelydifferentfromthatofNigeria.Chinais apureoilimportingcountrywhichisonlyexposedtoexternalshocksinoilpriceand the economy is very close to the 5 th stage of the Rostow’s growth theory which is stage of mass consumption given the global level of its product acceptability. Therefore expecting a similar findings between Nigeria and such country is inappropriateasNigeria is foundonthebothside of thesector (export and import side). This however confirms the findings of Deniz (2017) that impact of oilprice shocksdifferforoilimportingandexportingcountries. The study also confirms a significant link between oil price and energy consumption inline with Altai (2015) and more in the shortrun (Zhang et al., 2014).SupportingOdusami(2010)andSaibu(2011),thestudyestablishedthatthere isashortrunandlongrunlinkbetweenoilpriceandenergyconsumptioninNigeria. However,itcanbededucedfromthefindingsthatsustainableenvironmentaim in a developing country such as Nigeria remains doubting given the attitude to consumptionofaconventionalenergy(oilrelatedenergy)irrespectiveoftheswingof thepriceofoil.Oilrelatedenergyisconsideredastheeasiesttoaccessandstilla cheapermeanscomparedtotherenewableenergysource.Therefore,thefindingsof

289 Izvestiya 2019•Volume63•№4 this study give a better undertstanding of energy interest in terms of consumption amongstconsumers. 5. Summary and Conclusion Albeit, studies have examined the relationship between oil price and energy consumptioninNigeria,butmajorityhasonlyfocusedonthecausallinkbetweenthe variables.Thisstudyhowever,differentfromotherstudiesconsiderstwomajorissues: theperiodiceffectofpositiveandnegativeoilpriceshocksonenergyconsumptionin Nigeriaaccountingforstructuralbreaksandtheasymmetriceffectofthechangeson energyconsumptionfrom1978to2016.Toanalysethedata,BaiPerronLeastSquares Breakpoints test was used to account for the structural break, NonLinear Auto regressive Distributed lag was used to analyse the periodic effect of positive and negativeoilpriceshocksonenergyconsumptionandVARmodelwasusedtoaccount forenergyconsumptionresponsetopositiveandnegativeoilpriceshocks. The results indicate that, there was a significant break in 2010.Positive and negative oil price shocks positively influence energy consumption in Nigeria in the shortrunandlongrun,butnegativeoilpriceshocksaremoresignificantintheshort run. Consumer price index has a negative significant impact on energy consumption both in the shortrun and long run. Exchange rate negatively influences energy consumption in the shortrun, but positively in the longrun. The Impulse Response resultsshowedthatpositiveoilpriceshocksnevercausedenergyconsumptiontosteady state throughout the periods in the study, while negative oil price shocks show two periods of adjustment to steady state within the periods. It was also noted that, the effects of bonny light and Brent oil prices are similar. This can also be confirmed throughthestrongcorrelationinthemovementofthepricesdepictedin fig. 2 . Fromthefindings,thestudyconcludesthatoilpriceshocksdifferconsidering differentialeffectsandthe country understudy. GivenNigeriaasa net exporterof crudeoil,thecountryisalsoanimporterofrefinedoilproducts,whichexposesits investorsneeding moredollarstomeetthecostlierproducts.Thedemandforthese dollars however reduces the exchange rate power of naira to dollar, which in turn influences the consumer price index. In the process of trying to cover the cost incurredintherefinedproductsbuyingprocess,theimportersinflatethepriceofthe refinedproductsandleavetheconsumerswithnochoice,astheproductisthemost availableandaccessibleamongotheralternativeenergysourcesformeetingenergy needs. With this, development of the abandoned refineries is needed to reduce the costincurredthroughimportingrefinedproducts.Rehabilitatingtherefinerieswould also help the country feel the positive changes in oil price shocks the more in its 290 P.I.Ajayi,A.E.Longe,O.A.Omitogun,S.Muhammad OilPriceShocksandEnergyConsumptioninNigeria growthanddevelopment,asmoremoneywillbeavailabletodevelopotherstrategic sectors(suchasrenewableenergysectors,agriculture,infrastructural)oftheeconomy ratherthanspendingheavilyonsubsidy. References 1.Adom,P.,Bekoe,W.,&Akoena,S.K.(2012).Modelingaggregatedomestic electricitydemandinGhana:Anautoregressivedistributedlagboundscointegration approach. Energy Policy, 42 ,530537. 2.Agri,E.M.,Inusa,M.L.D.,&Kennedy,N.D.(2016).ImpactofOilPrice Volatility on Macroeconomic Variables and Sustainable Development in Nigeria. International Journal of Economics and Financial Research, 2 (2),3340. 3.Altai,M.(2015).Anestimationofoilpriceseffectonhouseholdconsump tion in Sweden. School of Economics and Management, Department of Economics, Lund University 139. 4.Babatunde,M.(2015).OilpriceshocksandexchangerateinNigeria. Interna- tional Journal of Energy Sector Management, 9 (1), 219. doi:DOI 10.1108/IJESM 1220130001 5.Chai,J.,Zhou,Y.,Liang,T.,Xing,L.,&Lai,K.K.(2016).ImpactofInterna tionalOilPriceonEnergyConservationandEmissionReductioninChina. Sustaina- bility, 8 (508),117. 6.Deniz,P.(2017).OilPriceandRenewableEnergy:OilDependentCountries. https://editorialexpress.com/cgi bin/conference/download.cgi?db_name=MEEA18&paper_id=49. 7.Duesenberry,J.S.,“IncomeConsumptionRelationsandTheirImplications,” in Lloyd Metzler et al., Income, Employment and Public Policy, New York: W.W.Norton&Company,Inc.,1948. 8.Ebrahim, Z., Inderwildi, O. R., & King, D. A. (2014). Macroeconomic im pactsofoilpricevolatility:mitigationandresilience. Front. Energy ,116. 9.Edelstein,P.,&Kilian,L.(2009).HowSensitiveareConsumerExpenditures toRetailEnergyPrices? Journal o fMonetary Economics, 56 ,766779. 10. Friedman,M.(1957).ThePermanentIncomeHypothesis:Comment, Amer- ican Economic Review ,48,pp.99091 11. Glausre,Y.U.,&Lee,A.R.(2002).TheImpactofOilPricesonIncome andEnergy. International Advances in Economic Research, 8 (2),148154. 12. He,L.,Ding,Z.,Yin,F.,&Wu,M.(2016).Theimpactofrelativeenergy pricesonindustrialenergyconsumptioninChina:aconsiderationofinflationcosts. SpringerPlus ,121. 291 Izvestiya 2019•Volume63•№4

13. Jobling,A.,&Jamasb,T.(2015).PriceVolatilityandDemandforOil:A Comparative Analysis of Developed and Developing Countries. Energy Policy Re- search Group ,135. 14. Kuper,G.,&vanSoest,D.(2006).Doesoilpriceuncertaintyaffectenergy use? Energy Journal, 27 (1),5578. 15. Longe,A.E.,Adelokun,O.O.,andOmitogun,O.(2018).TheCurrentAccount andOilPriceFluctuationsNexusinNigeria. Journal of Competitiveness, 10(2),118131. 16. Longe, A. E., Muhammad, S., Ajayi, P. I., and Omitogun, O. (2019). Oil price,tradeopenness,currentaccountbalancesandofficialexchangerateinNigeria. Accepted Manuscript. In Early View .OPECEnergyReview. 17. Modigliani,FandA.Ando,“TestsoftheLifeCycleHypothesisofSavings: Comments.,andSuggestions,”BulletinoftheOxfordUniversityInstituteofStatis tics,1957. 18. Nkomo,J.C.(2011).Crudeoilpricehikesandissuesforenergysecurityfor SouthernAfrica. International Development Research Centre,, 21 (2),1216. 19. Nwanna,I.O.,&Eyedayi,A.M.(2016).ImpactofCrudeOilPriceVolatil ity on Economic Growth in Nigeria (1980 2014). IOSR Journal of Business and Management (IOSR-JBM), 18 (6),1019. 20. Nwosa,P.,&Akinbobola,T.(2012).AggregateEnergyConsumptionand Sectoral Output in Nigeria. An International Multidisciplinary Journal, Ethiopia, 6(4),206215.doi:http://dx.doi.org/10.4314/afrrev.v6i4.14 21. Odusami, B. O. (2010). To consume or not: How oil prices affect the comovementofconsumptionandaggregatewealth. Energy Economics, 32 ,857867. 22. Omitogun,O.,Longe,A.E.,andAjulo,K.(2018).ForeignDirectInvest ment Key Factor for the Economic Development of Emerging Countries. OECO- NOMICA. 14(2),520. 23. Omitogun,O.,Longe,A.E.,andMuhammad,S.(2018).Theimpactofoil priceandrevenuevariationsoneconomicgrowthinNigeria . OPEC Energy Review. 42(4),387402.https://doi.org/10.1111/opec.12139. 24. Onakoya,A.B.,Onakoya,A.O.,JimiSalami,O.A.,&Odedairo,B.O. (2013).EnergyConsumptionandNigerianEconomicGrowth:AnEmpiricalAnaly sis. European Scientific Journal, 9 (4),2540. 25. Orhewere, B., & Henry, M. (2011). Energy Consumption and Economic GrowthinNigeria. JORIND ,153165. 26. Osigwe, A. C., & Arawomo, D. F. (2015). Energy Consumption, Energy PricesandEconomicGrowth:CausalRelationshipsBasedonErrorCorrectionMod el. International Journal of Energy Economics and Policy, 5 (2),408414. 292 P.I.Ajayi,A.E.Longe,O.A.Omitogun,S.Muhammad OilPriceShocksandEnergyConsumptioninNigeria

27. Pahlavani,M.,Wilson,E.,&Worthington,A.C.(2005).TradeGDPnexus inIran:Anapplicationoftheautoregressivedistributedlag(ARDL)model. American Journal of Applied Sciences, 2 (7),11581165. 28. Pesaran, M. H., & Shin, Y. C. (1999). An autoregressive distributed lag modeling approach to cointegration analysis. In Strom, S. (Ed) Econometric and economic theory in the 20th Century: The Ragner Frisch Centennnal Symposium. Cambridge:CambridgeUniversityPress. 29. Saibu,F.(2011).OilPrice,EnergyConsumptionandMacroeconomicPer formance:FurtherEvidencefromNigeria(19702009). International Association for Energy ,125. 30. Shahbaz,M.,Sarwar,S.,Chen,W.,&Malik,M.N.(2017).Dynamicsof electricityconsumption,oilpriceandeconomicgrowth:GlobalPerspective. Energy Policy, 108 (2017),256270. 31. Solow,R.(1956).Acontributiontothetheoryofeconomicgrowth. Quarter Journal of Economics ,6594. 32. Valadkhani,A.(2014).Dynamiceffectsofrisingoilpricesonconsumeren ergypricesinCanadaandtheUnitedStates:Evidencefromthelasthalfacentury. Energy Economics, 45 ,3344. 33. Ven,D.J.,&Fouquet,R.(2016).HistoricalEnergyPriceShocksandtheir ChangingEffectsontheEconomy. Energy Economics ,141. 34. WalrasreadthefirstvolumeoftheCapital ,butprobablynotthesecondor thirdvolume,forwhichtherewasnoFrenchtranslationatthetime(Walras,L.,1990, 448).TheeditorofRevue Socialiste wasoneofWalras’bestfriends,GeorgeRenard (18481930). 35. Wang, Y. S. (2013). Oil price effects on personal consumption expendi tures. Energy Economics, 36 ,198204. 36. Zhang, D., Broadstock, D. C., & Cao, H. (2014). International oil shocks andhouseholdconsumptioninChina. Energy Policy ,111.

293