Avril 2015

 

 

Working paper 

CAN THE US SHALE REVOLUTION

BE DUPLICATED IN EUROPE?

 Aurélien Saussay 

OFCE‐SCIENCES PO 

2015‐10 

CantheUSshalerevolutionbeduplicatedinEurope?

by

AurélienSaussayFrenchEconomicObservatory

69,quaid’Orsay75007,Paris,France

Phone:+33144185472/Email:aurelien.saussay@sciencespo.frAbstractOver the past decade, the rapid increase in shale gas and shale oil production in theUnitedStateshasprofoundlychangedenergymarketsinNorthAmerica,andhasledtoasignificantdecreaseinAmericannaturalgasprices.Thepossibleexistenceoflargeshaledeposits in Europe, mainly in France, Poland and the United Kingdom, has fosteredspeculation onwhether the "shale revolution", and its accompanyingmacroeconomicimpacts, could be duplicated in Europe. However, a number of uncertainties, notablygeological, technological and regulatory, make this possibility unclear. We present atechno‐economic model, SHERPA (SHale Exploitation and Recovery Projection andAnalysis),toanalyzethemaindeterminantsoftheprofitabilityofshalewellsandplays.We calibrateourmodelusingproductiondata from the leadingAmerican shaleplays.We use SHERPA to estimate three shale gas production scenarios exploring differentsetsofgeologicalandtechnicalhypothesesforthelargestpotentialholderofshalegasdeposits in Europe, France. Even considering that the geology of the potential Frenchshale deposits is favorable to commercial extraction, we find that under assumptionscalibratedonU.S.productiondata,naturalgascouldbeproducedatahighbreakevenpriceof$8.6perMMBtu,andovera45yeartimeframehaveanetpresentvalueof$19.6billion– less than1%of2012FrenchGDP.However, thespecificitiesof theEuropeancontext, notably high deposit depth and stricter environmental regulations, couldincrease drilling costs and further decrease this lowprofitability.We find that a 40%premiumoverAmericandrilling costswouldmakeshalegasextractionuneconomical.Absent extremewell productivity, it appears very difficult for shale gas extraction tohave an impact on European energy markets comparable to the American shalerevolution.

1 IntroductionOverthepastdecade,therapidincreaseinoilandgasproductionfromshaledepositsintheUnitedStateshasprofoundlychangedenergymarketsinNorthAmerica.Intheearly2000s,acombinationof improvedhorizontaldrillingandhydraulic fracturingtechnologyallowedtoextractnaturalgasfromformerlyinaccessibleshaledeposits.Anenvironmentofincreasinggaspricesinthefirsthalfofthelastdecade,alongwithmodificationstotheenvironmentalregulatoryframeworkbroughtbytheEnergy Policy Act of 2005 (Pub.L. 109–58, 2005), have made these new natural gas reservescommerciallyexploitable.The impact on domestic natural gas production in the United States was large and swift: whileannualgrosswithdrawalsofnaturalgashadbeenoscillatingsincethemid‐1990sbetween23.7and24.5Tcfperyear,theygrewby29%from2005to2013toreach30.2Tcfin2013(EIA,2014b).Theapplicationofthesametechnologytotightoildeposits(sometimesalsoreferredtoas“shaleoil”)hasbeenarguablyevenmoredramatic.Breakingalong‐termdeclinetrendwhichhadseena44%

decline between 1985 and 2008, domestic U.S. crude oil production has increased by 55% since2008toreach7.4MMbbl/dayin2013.The rapid expansion of U.S. fossil fuel production has had a number ofmacroeconomic impacts,notably in the formof increasedactivity fromintensivedrilling, lowerednaturalgasprices,andareduction in fossil fuel imports. However, the magnitude of these impacts remains a matter ofcontroversy.Somereports,notablyIHS(2011)andIHS(2013),haveestimatedthatdrillingactivity,combined with the on shoring of some industries back in the United States – petrochemical inparticular–couldsupportupto870,000jobsby2015.These conclusions are challenged by studies such as EMF (2013) or Spencer, Sartor & Mathieu(2014). Recognizing that shale well drilling only has a highly localized impact on activity andemployment, EMF (2013) estimates that shale developmentwould only boost GDP by “amodest0.46%”, and downplays the importance of shale gas as a “game‐changer” for the U.S. economy.Similarly, Spencer, Sartor & Mathieu (2014) highlight that the small share of energy‐intensiveindustries in the U.S. economy and of natural gas expenditures in households’ budgets limit theoverallmacroeconomicimpactthatcanbeexpectedfromthesteepreductioninnaturalgasprices.Still, theexistenceofpotentially largeshaledeposits inEurope,notably inFrance,PolandandtheUnited Kingdom, has fostered speculation on whether the oft‐called “shale revolution” could beduplicated on the continent. This issue is particularly relevant for natural gas, as Europeandependency on foreign exports has important energy security and geopolitical ramifications,notablyvis‐à‐vistheRussianFederation(IEA,2012).Gény (2010) examines this issue and concludes that the large differences in terms of onshoredrilling industry maturity, ease of access to land, mineral ownership rights and environmentalregulationsmakeU.S.operationalandbusinessmodelforshalegasdevelopmentinapplicabletotheEuropean context. Absent a focus on geological “sweet spots, R&D, workforce training, and newtechnologydevelopments”, the shale revolutionappears impossible to replicate inEurope.Onthesamenote,Spenceretal.(2014)findsthat“[i]tisunlikelythattheEUwillrepeattheUSexperienceintermsofthescaleofunconventionaloilandgasproduction”,andthat“[s]haleproductionwouldnothavesignificantmacroeconomicorcompetitivenessimpactsforEuropeintheperiodto2030‐2035”.TheseanalysesarehamperedbythelackofdataonthegeologyofEuropeanshaledeposits,onshalegas wells productivity or on drilling costs in Europe. In the present paper, we propose tocompensate for thisbyusinghistoricalproductiondata from theU.S. case tocalibrateamodelofshalegasproduction–notablyregardingwellproductivity,drillingcostsandoperationalcosts.WethenadaptthoseassumptionstotheEuropeancontextthroughananalysisof itsspecificities.ThismodelcanthenbeusedtosimulateshalegasproductionscenariosinEuropeancountries.We choose to focus our scenarios on France, as it is the largest potential holder of shale gasresourcesincontinentalEurope.Besides,thebanonallexplorationandextractionofshaledepositspassedin2011,thenconfirmedbytheFrenchConstitutionalCouncilin20131,makesitagoodcasestudyoftheimpactoftheregulatoryenvironmentonshaledevelopment.The paper is structured as follows: we first present a techno‐economic model, SHERPA (SHaleExploitation and Recovery Projection and Analysis), to analyze the main determinants of theprofitabilityofshalewellsandplays.WethenperformadetailedanalysisofU.S.productiondatain

1“France’sconstitutionalcouncilupholdsbanonfracking”,FinancialTimes,11October2013

leadingshaleplays,whichallowsustocalibrateSHERPA.Wethenexaminethespecificitiesof theEuropeancontextthathaveabearingonthetechnicalandeconomicassumptionsused.Wefinallypresent three scenarios of production based on different geological and technical hypotheses forFrance, the largestpotentialholderof shalegasdeposits inEurope; estimate themwithSHERPA;andconclude.

2 ModelingshaleproductionIn order to model shale production scenarios in Europe and identify the main parameters thatdetermine the cost of the production flow alongwith its volume,we develop a techno‐economicmodel named SHERPA (Shale Extraction and Recovery Projection and Analysis). This sectionpresentsthemodelandspecifiesitsequations.ProductionprofileofasinglewellOil and gas wells follow a well‐identified production profile during their life cycle (Arps, 1944).Theirproduction flowusuallyreaches itsmaximumearlyon,andthendecreasesatadeclineratethatcanvaryoverthewell’slifespan.ThisproductionprofilehasbeencharacterizedbyArps(1944).Initsmostgenericformulation,theproductionofawellcanbeexpressedasfollows:

1

1 (1)

where is the initial production, the initial decline rate, and (0 1) a parametercontrollingtheevolutionofthedeclinerateovertime.Theparameter notablydeterminesthetypeofdecline(seeFigure1):

exponential ( 0),whereproductiondecreasesover timewithaconstantdeclinerate. Ifthis decline rate is high, most of the production is front‐loaded over the first years ofexploitation;

hyperbolic (0 1),where thedecline ratedecreasesover time. If thisdecrease is fastenough,theimpactofhighinitialdeclineratesonthewell’sproductioncanbebalancedbyalongerwelllifespan;

harmonic( 1),whichisaspecialcaseofhyperbolicdecline.Itistheslowestofallthreetypesofdeclines,i.e.theonethatyieldsthelargestlate‐lifeproductionflows.

Figure1:Typesofwellproductiondecline( , %)

0

200

400

600

800

1000

0 5 10 15 20 25 30

Harmonic

Hyperbolic(b=0.5)

Exponential

Thisequationhighlightsthetwomostimportantparameterswhenestimatingtheexpectedoutputofawellover its entire lifespan: thewell’s initialproductionand thedynamicof thedecline rateoverthewell’slifecycle.Inthispaper,weshallconsideradiscretizedversionoftheArpsequationtoestimatethemonthlyproductionofthewellsthatwillbemodeled.Usingadeclinerateof inmonth ,theproductionformonth canbeexpressedas:

1 (2)Thetotalproductionoverthewell’slifetime, ,whichamountstoitsEstimatedUltimateRecovery(EUR),becomes:

1 (3)

Ifwethensupposethatdrillingcostsamountto ,themarginalcostperunitofproductionamountsto ,andthewholesalepriceamountsto ,theNetPresentValue(NPV)ofthisproductionis,foradiscountrateof :

11

(4)

Thebreakevenprice, ∗,correspondstothepriceforwhichthisNPViszero.Fromequation(4),wefind ∗,whichcanbesplitintoamarginalcomponentandafixedcostscomponentwhichamortizestheinitialdrillingcosts:

∗

∑ 0 110

(5)

ProductionofaplayandproductionplateauIn order to model the total production of a play, we consider the production profile of arepresentative well. Thus, we capture the diversity observed across American plays in wellproductivity and decline speed solely through this “average” well. This approach allows us toanalyze the aggregate production of the play across its entire lifespan. This hypothesis is asimplificationsinceitiswelldocumentedthattheproductivityofshalewellsvarygreatly,includingwithinasingleplay(EIA,2011).However,itisexpectedthatthisdiversitywouldmostlybearonthedynamicsof thedrillingeffort–with themostproductivespotsbeingdrilled firstonce identified.Characterizingtherepresentativeaverageissufficienttoestimatefield‐widevariablesofinteresttothe present study, such as aggregate production flow, expected total production, or averagebreakevenprice.

Letusconsideraplayforwhichtheproductionoftherepresentativewellisdescribedbyequation2.If wellsaredrilledinmonth ,thentheproductionoftheplayinmonth , ,isexpressedas:

1 (6)

Ifwecall thehighestdrillingrateovertheproductionperiodof theplay ,and thesmallestdeclinerateobservedinanygivenmonthacrossthelifespanoftherepresentativewell:

max

min (7)

Thenweobtainthefollowingupperboundontheproductionoftheplay:

(8)

This inequality illustrates an important phenomenon. The production of the play is at all timesboundedbyaproductionplateau,whosevalueonlydependsontheaverageinitialproduction,thelower boundof thedecline rate across the lifespanof the representativewell, and themaximumdrillingrate.Asthedrillingrateincreasestowardsitsmaximumvalue,andastheplaymatures,theproduction of the entire play actually converges towards a plateauwhich is strictly bounded byequation 7, with the speed of convergence depending on the steepness of the decline rate. Thisplateaucorrespondstothephase intheplay lifecyclewhendrillingnewwellscanonlyoffsetthedecliningproductionofoldwells,withoutincreasingtheaggregateproductionoftheplay.Yet,iftheaveragedeclinerate is low, thisplateaucanbesohighas toneverbeboundingwithin theplay’slifespan.Indeed, in the case of conventional oil and gas fields, the observed annual decline rates can beestimatedtobearound3to4%(Höök,Hirsch,&Aleklett,2009):inthatcase,theproductionofthefieldremainsbelow95%oftheplateau’svalueduringthefirst75yearsofextraction.However,forannualdecline rates closer to the50% levels observedon shaledeposits (seeTable3), the samethresholdcanbereachedwithinfouryearsonly.This phenomenon of production plateau, where new wells are only drilled to maintain existingproduction volume, is thus characteristic of shale plays. Once reached, breaching the productionplateau entails improving well productivity through better technology, or increasing the drillingrate.Geologicalconstraintscansethardlimitsonwellproductivity,althoughrecentimprovementindrilling and fracturing technology have achieved some improvements (EIA, 2014a); the simplestwaytomaintainproductiongrowthistosustainacontinuousincreaseinthedrillingrate.

3 DatacalibrationCalibrating the equationdescribing theproductionprofileof the representativewell (equation2)requires detailed knowledge of the geological characteristics of the play considered. GreatuncertaintiesremaininEuropeovertheactualvolumeofresourcesinplaceandoftechnicallyandcommerciallyrecoverablereserves(IFPEN,2013).Besides,sinceonlyaround50experimentalwells

havebeendrilledonthecontinentsofar(Spenceretal.,2014),productiondatahasyettobemadeavailablepublicly.It is therefore necessary to gather this calibration data from a different source. Ever since thecommercial extractionof shaledeposits beganduring the lastdecade, close to60 shalegasplayshave been drilled in the United States (Hughes, 2013). 30 out of these 60 plays have provedprofitable,withonlysixofthoseaccountingformorethan90%ofthetotalnaturalgasproductionfromshaledeposits in theUnitedStates (EIA,2013).Productiondata fromNorthAmericanplaysthuscoversawidevarietyofdistinctgeologicalconfigurations.Adetailedanalysisofthisdatacanprovideabasisforthecalibrationofourmodel.In a number of States, home to some of North America’s largest shale plays, production reportsprovidedbyoperatorsaremadeavailablepublicly.Thisdataprovidesaveryprecisedescriptionoftheshalewellproductionprofiles.Usingthisinformation,wecanestimaterealisticintervalsforthekeyparametersfequation2,initialproductionanddeclinerate.InitialproductionWehaveundertakenthisstatisticalanalysisintwomajornaturalgas‐producingplays,Haynesvilleand Fayetteville. The production data was obtained from the Louisiana Department of NaturalResourcesandfromtheArkansasOilandGasCommission.Wethereforerestrictouranalysistotheportionof theHaynesvilleplay located inLouisiana and theArkansas sectionofFayetteville.Ourdata includes initial production and drilling date for 2,432 wells in Haynesville, and monthlyproductionanddrillingdatefor4,882wellsinFayetteville.Wefirstestimatetheaverageinitialproductionofwellsineachplay,andexamineitsevolutionovertimebyyearofdrilling.TheresultsarepresentedinFigure2below.

Figure2:Well'sinitialproductionbydrillingyearintheHaynesvilleandFayetteville

Haynesville(Louisiana)

Source:Louisiana’sDepartmentofNaturalResources/Author’scalculations

Fayetteville(Arkansas)

Source:ArkansasOilandGasCommission/Author’scalculations

Wefindthattheevolutionof initialproductionexhibitsacommonpatterninbothplays. Ina firstperiod, ranging from2006 to2009 inHaynesvilleand from2005to2010 inFayetteville, averageinitialproductionsgraduallyincreasewithdrillingyear.Providedthatdeclineratesremainconstantacrossdrilling years, this indicates an improvement inwell productivity over time in eachof the

plays. Indeed, a simultaneous increase of decline rates over time could cancel out the impact ofimprovedinitialproductionsoverthewell’stotallifecycleproduction.This improvement can be driven by at least two causes: an improvement in extraction andfracturingtechnologies,which leadstoan increase inrecoveryratesofnaturalgas fromtheshaleresource(EIA,2014);andabetterknowledgeofthefield’sgeology,notablytheidentificationofso‐called“sweetspots”–regionsoftheplaywherewellproductivitytendstobeoptimal–whichoncefoundconcentratethedrillingactivity,therebyincreasingaveragewellproductivityintheplay(EIA,2011).Inbothcases,thisfirstperiodofincreasingwellproductivitycanbeunderstoodasalearningphase,eitherattheplaylevel–duringwhichoperatorsincreasetheirgeologicalknowledgeoftheshale play –, or at the industry level – whereby technologies used to extract shale deposits areimprovedsimultaneouslyacrossallplays.Furtherresearchwillbeneededtodistinguishtherelativecontributionsofeachofthesefactorsintheobservedoverallincreaseinwellproductivityovertime.Once this learning phase is over, average initial production reaches a stable level that has beenroughlymaintainedtothepresent,althoughthedistributionof initialproductionshasvariedovertime in each play. In the Louisiana section of Haynesville, average initial production stabilizesbetween11,930and12,880Mcf/day,whileintheArkansassectionofFayetteville, it iscomprisedbetween3,000and3,348Mcf/day.Tocomplementtheresultsofthisanalysis,weextendourestimateswiththoseofHughes(2013),whohasconductedasimilarstudyonthelargestshaleplaysintheUnitedStates.HisestimatesfortheaverageinitialproductionofawellineachoftheseplaysarecollectedinTable1.

Table1:AverageinitialproductionofawellinthesixlargestshalegasplaysintheU.S.

Haynesville Barnett Marcellus Fayetteville EagleFord Woodford

Initialproduction(Mcf/day) 8,201 1,619 1,947 2,069 1,920 2,292

Source:Hughes(2013)WiththeexceptionofHaynesville,theaverageinitialproductionofawell isremarkablysimilarinfiveofthesixmainAmericanshalegasplays,between1,600and2,300Mcf/day.ThediscrepancybetweenourestimatesandHughes’(2013)fortheHaynesvilleandtheFayettevillehastwocauses:first,hisanalysiswasconductedoverthewholeplayineachcase,whileweonlyhadaccesstotheportionof theplays located in States that releasepublicproductiondata fromoperators; second,Hughes’(2013)averagesarecalculatedovereverywellseverdrilledintheplay–eventhoughouranalysis shows that there has been a significant increase in initial productions over time. Thisimplies that estimates of average initial production made over the whole play’s lifespan arenecessarily lower than the post‐learning phase plateau we identified in our analysis, since theyincludewellsdrilledwhiletechnologicalandgeologicalknowledgewasstillimproving.DeclineratesThe data provided by the Arkansas Oil and Gas Commission for the Fayetteville play includesmonthlyproductionreportsbywell.Thisallowsustoestimateaveragedeclineratesovertheentireplay, along with their evolution by drilling year. Unfortunately, this was not possible with theHaynesvilledata,whichonlyprovidesinitialproductionforeachwell.Figure3showstheaverageratioofremainingproductiontoinitialproductionforaFayettevillewellafter one to five years of extraction, as a function of the well’s drilling year. The decline inproductionovertimeisverysteep:afterthreeyears,theproductionflowisreducedonaverageto

28.3% of its initial value; after five years, the remaining flow is 15.5% of initial production onaverage.Figure3:Averageremainingproductionasapercentageofinitialproductionafter1to5years,bydrillingyear

Source:ArkansasOilandGasCommission/Author’scalculations

WiththeexceptionofthefirsttwoyearsforwhichwehavedataintheFayetteville,whichconcernsonly 1% of the sample, the average decrease in production after one to five years is remarkablystable over drilling years – and therefore, so are the associated decline rates. This indicates thatunlike initial production, the averagewell’s production profile does not exhibit a learning phaseafter which observed decline rates would be reduced. It is therefore reasonable to assume thataverage decline rates estimated over thewhole play’s lifespan are applicable to themost recentwells. Table2presents our estimates for the average year‐on‐yeardecline ratesof awell in theportionoftheFayettevilleplaylocatedinArkansas.

Table2:Averageyear‐on‐yeardeclineratesofawellinFayetteville(Arkansas)

Declineinyear1 Declineinyear2 Declineinyear3 Declineinyear4 Declineinyear5

Fayetteville 57% 34% 24% 21% 9%Source:ArkansasOilandGasCommission/Author’scalculations

TogetdeclineratesestimatesforotherU.S.shalegasplays,wecomplementouranalysiswiththatofHughes (2013).His estimates for the average year‐on‐year decline rates of awell inHaynesville,BarnettandMarcellusarepresentedinTable3Table1.

Table3:Averageyear‐on‐yeardeclineratesofawellinthethreemainshalegasplaysintheU.S.

Haynesville Barnett MarcellusDeclineinyear1 68% 61% 47%Declineinyear2 49% 32% 66%Declineinyear3 50% 24% 71%Declineinyear4 48% 18% 47%Declineinyear5 15%

Source:Hughes(2013)First,Table2 andTable3 illustrate the largediversityofdecline ratesobservedacross shale gasplays. Second, we find that in general, the production decline cannot be described as eitherexponential, since the annual decline rate varies over the well’s lifespan, nor hyperbolic, sincedecline ratesdonotdecreasemonotonicallyover time. We therefore calibrateequation2withamonthlydeclineratethatvariesforeachyearofproduction.Beyondyear5,annualdeclineratesareassumedconstant.The production profile of the representativewell is tied to the specific shale play considered, inparticulartoitsgeologicalcharacteristics.Still,inkeepingwithourapproachofusinghistoricalU.S.production data to calibrate SHERPA, we use a weighted average of the decline rates estimatedabove, using the 2012 annual production of each shale play examined asweight. The results arepresentedinTable4.

Table4:DeclineratesusedintheSHERPAmodel

Declineinyear1 Declineinyear2 Declineinyear3 Declineinyear4 Declineinyear5

59% 46% 44% 36% 13%

4 SpecificitiesoftheEuropeancontextGaspriceformationThe large drop in natural gas prices over the past decade, from a weekly average high of14.49$/MMBtuinDecember2005toalowof1.86$/MMBtuinApril20122(seeFigure4),hasbeenoneof themore significant consequencesof the large increase indomestic gas production in theUnitedStates.Unlike other energy commodities, crude oil in particular, the market for natural gas is stillfragmented into several regional markets. The price of natural gas is therefore different in theUnited States, Europe and East Asia (IEA, 2012). Hence, the decrease in gas prices illustrated inFigure4hasremainedlocalizedintheUnitedStates.Thisisdueinlargeparttothedifficultyoftransportingnaturalgas.Acrossoceans,wherepipelinescannotbeused,naturalgasmustbeliquefiedandtransportedinLNGtankers.Thisentailsbuildingveryexpensiveprocessingfacilitiestoliquefythegasondeparture,andgasifyitbackonarrival.Inaddition, processing plants used for liquefaction cannot be used for gasification without costlyretrofitting(IGU,2012).

2HenryHubNaturalGasSpotPrice,weeklyaverages.Source:U.S.EIA

Figure4:HenryHubnaturalgasweeklyaverageprice(1997‐2013)

Source:U.S.EIA

Besides, gasprice formationmechanismsaredistinct ineachof themajormarkets. In theUnitedStates, thepriceofnaturalgas is set throughgas‐on‐gascompetition.Naturalgas is tradedoveravarietyoftimeframes(e.g.daily,monthlyorannually)atanumberofphysicalhubs–Texas’sHenryHub being the largest –, and the interplay of supply anddemanddetermines theprice. In such amarket,changes in thebalancebetweensupplyanddemandhavean immediate impactonprices.TheUnitedStates,whichuntilthelate2000sexpecteddomesticnaturalgasproductiontodecline,had built LNG plants to import gas, but not to export it (EIA, 2011).When shale gas extractionrapidlygrew,thenewfounddomesticproductionofnaturalgaschangedthelocalbalanceofsupplyand demand immediately. From 2008 to 2012, domestic production grew at a rate of 3.6% perannum,outpacingconsumption,whichonlygrewat2.3%perannum3:thisledtothelargedropinprices.InEurope,gasprice formation followsadifferentmechanism.Traditionally,Europeannaturalgassupplieshavebeenpricedthroughamixoflong‐termcontractswithproducingcountriesandspotmarket pricing. Long‐term contracts are mostly priced using a mechanism known as oil priceescalation,wherebygaspricesarelinked,usuallythroughabasepriceandanescalationclause,tothepriceofcompetingfuels–typicallycrudeoil(IGU,2012).OilpriceescalationusedtodominatenaturalgaspriceformationinEurope.However,sincethelate2000s,oil indexationofnaturalgascontractshasbeendecreasing:asof2012,51%ofEuropeangasconsumptionwaspricedthroughanoilpriceescalationclause,downfrom59%in2010.Meanwhile,from2007to2012,spot‐pricednaturalgasvolumeshavedoubled,toreach44%ofconsumption(EC,2013).ThispricingstructuremakesEuropeanwholesalegaspriceslesselastictochangesinthebalanceofsupplyanddemand.Whileanincreaseindomesticproductioncouldimprovethebargainingpowerof European countrieswith their suppliers, the impact of introducing small volumes of domesticshalegasproductionintheEuropeansupplymixongaspricesisunclear.Wethereforechoosetoestimatetheprofitabilityofshalegasextractionatagivenwholesaleprice.For our main scenarios, we use the Russian Natural Gas border price in Germany, which is acommonbenchmarkforgaspricesincontinentalEurope,averagedovertheperiod2011‐2013.Weobtain a wholesale price hypothesis of $11.6/MMBtu. We then explore the sensitivity to thathypothesisinseveralvariants.

3Source:U.S.EIANaturalGasstatistics

0

2

4

6

8

10

12

14

16

$/MMBtu

DrillingcostsPublicdataondrillingcostsisscarce,whichmakestheircalibrationdifficult.AccordingtotheU.S.EIA, drilling costs per well in the leading shale plays of Marcellus, Bakken and Eagle Ford arecomprisedbetween$6.5and$9million,includingbothhorizontaldrillingandhydraulicfracturing(EIA,2012).However,theseestimatescannotbeuseddirectlyintheEuropeancontext.Notably,oneofthemaindriversofdrillingcostsisthedepthofthewellandthelengthofitslateral(Pulsipher,2007).Table5presentsaveragedrillingcostsandaveragedepthinthemainU.S.shaleplays.

Table5:AveragedrillingcostsanddepthofthemainU.S.shaleplays

Haynesville Barnett Marcellus Fayetteville Woodford GraniteWash

Averagedrillingcosts(millionUSD) 9.5 3.5 5.3 2.8 8.5 7.8

Averagedepth(ft)

12,100 7,900 6,200 3,600 13,100 13,100

Source:Nickelson(2013),BermanetPittinger(2011)Asshownintheabovetable,drillingcostsincreasewithaveragedepositdepth:themostexpensivewells are located in the deepest shale deposits, between 12,000 and 13,000 ft on average. MostEuropean deposits have been identified in geological strata located at comparable depth: themajority of theFrench resourceswouldbe foundbetween10,000and14,000 ft, between10,000and12,500ftinPoland,between11,500and14,500ftinGermany,between11,000and12,500ftintheNetherlands, andaround11,000 ft in Spain;onlyBritish shaledepositswouldbe locatedatashallowerdepthof8,000ft(EIA,2013).TheseelementsleadustoestimatethatdrillingcostswillonaveragebehigherinEuropethanintheUnitedStates.ThisisindeedtheconclusionofWoodMackenzie(2012)ontheeconomicpotentialofshalegasresourcesintheUnitedKingdom,whichestimatedthatshouldtheBritishshaleresourcesbedeveloped,theaveragedrillingcostswouldreach$17million.ThiswouldamounttomorethanoneandahalftimestheaveragewellcostintheHaynesville,wheredrillingcostsarethehighestofanyAmericanplay.OilservicescompanySchlumbergeralsoestimatedin2011thatdrillingcostsinPolandcouldturnouttobethreetimeshigherthanintheUnitedStates4.Finally, it shouldbenoted that theavailabilityofdrillingequipment ismuchhigher in theUnitedStatesthaninEurope.Inthefirstquarterof2014,morethan1,700drillingrigswerebeingoperatedin the United States, including both oil and gas plays of the conventional and unconventionalvarieties (EIA, 2014). This is to be contrastedwith less than a hundred rigs available across theentireEuropeancontinentin2013(Hsieh,2011).Further,onlyasmallfractionoftheserigscanbeusedtodrillshalegaswells:forexample,in2011,outof15drillingrigsavailableinPoland,only5were suitable to shale gas extraction. These capacity constraints could limit the drilling rate inEuropeancountries,atleastinthefirstyearsofproduction.RegulatoryenvironmentInadditiontotheimprovementsmadetohydraulicfracturingandhorizontaldrillingtechnologies,the expansion of commercial shale gas extraction was also enabled by changes made to the

4“Shale‐GasDrillingCostinPolandTripleU.S.,SchlumbergerSays”,Bloomberg,29November2011

regulatory framework governing oil and gas production – especially regarding environmentalregulations.Indeed,theEnergyPolicyActof2005(Pub.L.109–58,2005)broughtsomesignificantmodificationstotheenvironmentallegislationsregulatingoilandgasdrillingintheUnitedStates.Passedatatimewhenconventionalgasplayswereexhibitingsignsofdepletion(EIA,2005),theEnergyPolicyActdefined new core principles for American energy policy,with a particular emphasis on reducingfuturedependencyonfossilfuelimports.TheActincludedanumberofmeasuresaimingtoincreasedomestic fossil fuel production. Notably, two existing environmental laws were amended tofacilitatetheuseofhydraulicfracturing–andthustheextractionofoilandgasfromshaledeposits:

the Safe DrinkingWater Act (Pub. L. 93‐523, 1974), which regulates the public drinkingwatersupply,andensuresitsqualityandsuitabilityforhumanconsumption.Originally,thisActbannedanydrillinginthevicinityofundergroundwaterreservoirs.Section322oftheEnergyPolicyActof2005liftsthisbanforoilandgasdrilling.

theCleanWaterAct(Pub.L.92‐500,1972),whichgovernswaterpollution.ThisActnotablydefineswhat constitutes awater pollutant. Through an amendment to section 502 of theCleanWaterAct,theEnergyPolicyActof2005excludesfromthisdefinition“water,gas,orothermaterialwhichisinjectedintoawelltofacilitateproductionofoilorgas”.

Thesechanges,madetotwopillarsoftheenvironmentalregulatoryframeworkoftheUnitedStates,wereessentialtoenablethewidespreaduseofhydraulicfracturing,andthustheshalerevolution.Theformerbanondrillingclosetowaterreservoirswouldhavepreventedanumberofcommercialwellsfromeverbeingdrilled.Similarly,withoutthechemicaladditivesthatwereformerlylistedaspollutants by the Clean Water Act, hydraulic fracturing would be much less effective, and wellproductivitywouldbesignificantlylower.EnvironmentalregulationsinEuropearemuchstricter.Inparticular,theuseofchemicaladditivesinthefracturingfluid,thetransportationandstorageofflowbackmudfromwellfracturing,orthedrillingofwellswithinproximitytowaterreservoirsorinhabitationswouldallbeverydifficultoroutright forbidden under the current European environmental legislation, both at the Union andMember State level (Gény, 2010). Other measures targeting both safety and environmentalprotection,suchasstandardsofsafetyvalvesandthecompulsorinessofmultiplecasingsaroundthewell’sbody,wouldhaveadirectimpactondrillingcosts.Atthisstage,itisimpossibletoknowwhethertheEuropeanUnionorsomeofitsMemberStateswillamend their existing legislations to lift some of the restrictions currently limiting the use ofhydraulicfracturing.Fosteringshalegasproductionontheirterritorywouldentailrescindingpartof their environmental protection framework to favor domestic on shore drilling, as the EnergyPolicyActdid in theUnitedStates in2005.Currently,compliancewith the local legislationwouldleadtosignificantlyhigherdrillingcostsinEuropethanintheUnitedStates(Gény,2010).

5 ProductionscenariosforFranceScenariosBased on the calibration obtained from the statistical analysis of U.S. production data,we designproduction scenarios for France, which is the largest potential holder of shale gas deposits inEurope. We focus our analysis on natural gas as according to EIA (2013), estimated technicallyrecoverable resources of shale gas dwarf those of shale and tight oil in Europe. To simplify the

analysis,we consider the twopotential shale fields in France, theParis basin and the South‐Eastbasin,asonesinglefield,onwhichweformulateaggregatehypotheses.Further, our price hypotheses are based on the assumption that dry gas will be produced andmarketed. We do consider the possibility of more valuable associated liquids such as ethane orbutanebeingproducedinavariantofthefirstscenariothough.Weconsiderthreemainscenarios:

The «Central» scenario aims to estimatewhat a realistic production scenariowould be,shouldthegeologyoftheFrenchshaledepositsprovefavorabletocommercialextraction.Basedonthishypothesis,weassumethatwellproductivitywouldbecomparabletothatofthe main U.S. commercial shale gas plays: we therefore assume an average initialproductivityof2,200Mcf/day,attheupperendoftherangeidentifiedinTable1.Inouranalysisofdrillingcosts,weobserved thatFrenchshaledepositswere locatedatadepthcomparabletothatoftheHaynesvilleplay,andthatdrillingcostswereproportionaltothe deposit depth. We therefore assume average drilling costs of $10 million. This lasthypothesis is conservative, as we consider that the European context – environmentalregulations inparticular–wouldonlyimposeahalf‐milliondollarpremium(5%)perwellonaverageinadditiontothe$9.5milliondrillingcostaverageobservedintheHaynesville.

The«ZeroNPV»scenarioaimsatdetermining,ceterisparibus,whatpremiumonaveragedrillingcostswouldcancelouttheprofitabilityofshalegasextractioninFrance.We thereforemaintain thewell productivity assumption of the Central scenario, and findthataddinga40%premiumondrillingcostsoverthoseobservedinHaynesvilleleadstoanNPVovertheentiredurationofthescenarioofzero.

The«Extreme»scenarioperformsasensitivityanalysisonwellproductivity.Weconservethedrillingcostsassumptionof theCentralscenario,butassociate itwithanaveragewellproductivitycomparabletothatofthebestplayintheU.S.,Haynesville,withaninitialproductionof8,000Mcf/day.Thishypothesisliesatanextremeendoftheprobabilityspace,asitamountstoconsideringthatbothFrenchshalegasfieldswouldhaveanaveragewellproductivitycomparabletothatoftheverybestplayknowntodate.

Further,inallscenarios,weconsideradrillingrateof30wellspermonth,whichamountstomorethan10,000wellsdrilledoveraperiodof30years.AssumptionsforthesethreescenariosaregatheredinTable6.

Table6:Productionscenariosassumptions

Scenario Initialproduction(Mcf/day)

Drillingcost(millionsUSD)

Drillingrate(wells/months)

1)Central 2,200 10,0 30

2)ZeroNPV 2,200 13,3 30

3)Extreme 8,000 10,0 30

Inadditionstothesedifferentiatinghypotheses,wealsomakethefollowingassumptionsacrossallscenarios:

thedrillingperiodlastsfor30years; the drilling rate is increased progressively during the first three years of the scenario, in

ordertoaccountforthelearningphaseoftheindustry; operationalcosts,estimatedbyMoniz, Jacoby&Meggs(2011)between$0.5et$1perMcf,

arepeggedat$0.75perMcf;and technicallyrecoverableresources(TRR)areestimatedat137Tcf(EIA,2013).

Finally,weassume that shale gasextractionwill bemanagedbya state‐ownedcompany. Indeed,calls havebeenmade– notablyby theFrench IndustryMinister – for shale gasproduction tobecarriedoutbyapublicbody,shouldthebanonshalegasexplorationandextractionbelifted5.Thisleadsustouseadiscountrateof4%inthecalculationofnetpresentvalues,as it is thestandardrate used by the French Treasury to evaluate the profitability of government‐sponsored projects(DGTrésor,2005).ResultsWeusetheSHERPAmodeltoestimatetheaveragebreakevenpriceandestimatedultimaterecovery(EUR)perwell,theshareoftechnicallyrecoverableresourcesextracted,andthenetpresentvalueofthenaturalgasproduced.ResultsarepresentedinTable7.

Table7:Mainscenarioresults

ScenarioBreakevenprice

(USD/MMBtu)

EUR(bcf)

Peakproduction(bcf/year)

ShareofTRRproduced

NPV(billionUSD)

1)Central 8.6 1.4 491 10% 19.6

2)ZeroNPV 11.2 1.4 491 10% 0

3)Extreme 2.8 5.0 1,787 37% 228

Source:SHERPAmodelTheassumptionsof theCentralscenariobringthebreakevenpriceofnaturalgasclosetoourgaspricehypothesis.Therefore, theNPVofnatural gas extractedover the entire scenarioduration isfairlylow,at$19.6billion–whichrepresentslessthan1%of2012FrenchGDP.Atpeak,domesticnaturalgasproductionwouldcover31%ofFrance’s2012consumptionof1,560bcf.Thesecondscenarioillustratesthesensitivityofbreakevenpricetotheaveragedrillingcost.Underourgaspricehypothesis,anincreaseof33%indrillingcostsovertheCentralscenarioisenoughtobringthebreakevenpriceonparwithourwholesalepriceassumption:theNPVisthusbroughttonaught.Similarly,areductioninaveragewellproductivity–forexampleduetoadversegeologicalconditions–,whilemaintainingconstantdrillingcosts,wouldhavequicklyreducedtheNPVtozero.Finally, the third scenario highlights the importance of well productivity in determining thebreakeven price, and therefore the profitability of natural gas extraction. Unsurprisingly, shouldFrenchshalegasplaysexhibitthesamewellproductivityasthebestAmericanplayknowntodate,theNPVofthegasproducedwouldamounttoseveralpercentagepointsofFrenchGDP.Underthese

5“Franceplanstoinvestinstateminingventure”,FinancialTimes,21February2014

extremewellproductivityassumptions,Francecouldevenbecomeanetnaturalgasexporter,asthepeakannualdomesticproductionis200bcfhigherthanthe2012consumption.The production profile and cash flow of the Central scenario are presented in Figure 5. Forreference,thesamechartsareprovidedforallscenariosintheannex.

Figure5:Centralscenarioresults

Annualnaturalgasproduction Cashflow

Source:SHERPAmodelThe aggregate natural gas production increases over the first 15 years of the scenario, beforereachingaplateauuntilyear30.This illustrates thephenomenonofproductionplateaudescribedpreviously, and is due to the fact thatpast the first 3 years of ramp‐up, thedrilling rate remainsconstant throughoutthedrillingphase,at30wellspermonth.Sincewellproductivityhypothesesareassumedtoholdoverthewholescenario’stimeframe,productioncannotincreasebeyondthatplateauwithoutanincreaseinthedrillingrate.Assoonasthedrillingceasesinyear31,aggregateproductionundergoesasteepreduction,atan initialrateof25%to30%ayear,thensofteningtoaround20%afterthefirst5yearsofterminaldecline.The largeupfrontdrillingcosts lead tonegativecash flows for the first fiveyearsof thescenario.Thisnegativerunwayincreasesforthefirst3yearsofthescenarioasthedrillingrateisrampedupto 30wells permonth. The positive spike in cash‐flow in year 31 corresponds to the end of thedrillingphase:wellsthatweredrilled inearlierperiodskeepproducing for15years,per thewelllifeexpectancyhypothesis,withnofurtherexpensebutthe$0.75/Mcfoperationalexpense.WethenperformthreesensitivityanalysesontheCentralscenario.WefirstconsiderthepossibilitythatFrenchdepositsproduce“wet”gas,i.e.bothmethaneandassociatednaturalgasliquids(NGLs)suchasethaneorbutane.Wemodelthisvariantasa10%increaseinthewholesalepriceofnaturalgasoverourhypothesisof$11.6/MMBtu,sinceonaper‐Btubasis,NGLscommandahigherpricethannaturalgasonEuropeanmarkets6.OneshouldnotethattheexpansionofwetgasintheUnitedStateshasactuallybroughtdownthepriceofsomeliquids.Forexample,theprofitabilityofethaneisnow on‐par with natural gas on a per‐Btu basis: the increased profitability of wet gas couldthereforebetransitory7.

6Source:OPIS7“ChangesinLongitudes–EthaneExportstoEurope”,RBNEnergy,24March2014

Next, we evaluate a Central scenario where the production would be undertaken by a privatecompanyinsteadofastate‐ownedstructure.Inthisvariant,wethusconsideradiscountrateof10%insteadofthe4%usedinthestandardCentralscenario.ThisallowsustoestimatethesensitivityofthebreakevenpriceandtheNPVtothediscountrate.Finally,weconsideravariantoftheCentralscenariowherenaturalgaspricesdonotremainstable,but instead halve over the duration of the scenario. This could occur if, for example, severalEuropeancountrieshadshalegasdepositswithfavorablegeology,anddecidedtostartextractingtheirnaturalgassimultaneously.TheircombinednewfounddomesticproductionscouldleadtoanoversupplyofnaturalgasontheEuropeanmarket,or,equivalently,increasetheirbargainingpowerwithforeignsupplierstorenegotiatelong‐termcontractsandlowernaturalgaspricesinEurope.

Table8:Centralscenariovariants

Assumptions Results

VariantWholesaleprice(USD/MMBtu)

DiscountrateBreakevenprice(USD/MMBtu)

NPV(billionUSD)

1a)Wetgas 12.7 4% 8.6 28

1b)Private 11.6 10% 9.5 5.5

1c)Decreasinggasprice

11.6 4% 8.6 4.9

Source:SHERPAmodelWefindthatanincreaseinwholesalegaspricesof10%overourmainhypothesisof$11.6/MMBtuwould increase theNPVby43%,at$28billion. Thishighlights thehighsensitivityof theoverallprofitability of shale gas extraction to the wholesale gas price. Conversely, should gas pricesdecreasebyhalfover theproductionperiod, theNPVwouldbereducedby75%.As illustrated inFigure6,invariant1c,shalegasextractionevenbecomesbrieflycash‐flownegativeattheendofthedrillingperiod.

Figure6:Cashflowforvariant1c–Decreasinggasprice

Finally, increasing the discount rate to 10% also reduces the profitability of shale gas extractionsharply: theNPV is reducedby72%comparedwith a4%discount rate, and thebreakevenpriceincreasesby10%to$9.5/MMBtu.Thus,underourCentralhypotheses, commercialproductionofshalegasbyaprivateentityinFranceappearsdifficult.

6 ConclusionTo assess whether the American shale gas revolution can be duplicated in Europe, we havedeterminedthemaindriversofshalegasextractionprofitability.Tothisend,wehavepresentedatechno‐economicmodelofshalegasproduction,SHERPA,whichallowedustoidentifythefollowingkeyparameters:wellproductivity,asdescribedbyinitialproductionanddeclinerates,anddrillingcosts.The volume and geological characteristics of shale gas resources in Europe remain speculative.Besides,experimentaldrillinghasremainedveryscarce. It is thereforenotpossibletoassesswellproductivity in the potential European shale gas plays.At this stage,we cannot directly calibrateSHERPAonEuropeanproductiondata.Toremedythislackofdata,weperformadetailedstatisticalanalysisofexistingproductiondataintheleadingU.S.shaleplaystocalibraterealisticrangesforinitialproductionsanddeclinerates.Wethen analyze the specificities of the European context, notably in terms of gas price formation,drillingcostsandenvironmentalregulations,todefinehypothesesreflectingtheseparticularities.Using SHERPA, we then estimate production scenarios for France, which is the largest potentialholderofshalegasresourcesinEurope.AllscenariosstartfromthepremisethatthegeologyoftheFrenchdepositsprovesconducivetothecommercialextractionofshalegas.However,evenunderthathypothesis,assumingwellproductivitycomparabletothatoffiveofthesixlargestU.S.shaleplays,andconservativedrillingcostsof$10millionperwell,wefindthatthebreakevenpriceofshalegasextractionwouldbehighand theprofitability relatively low. Indeed,only under extreme well productivity hypotheses does the profitability of shale gas becomesignificant.Conversely,wefindthatincreaseddrillingcosts40%abovetheirAmericancounterparts,adiscountrateof10%compatiblewithproductionbyaprivateentity,oraprogressivehalvingofwholesalegaspricesover45yearswouldallmakeshalegasextractionclose toorentirelyuneconomical inFrance.Thus,itappearsthatevenifthegeologyofEuropeanshaledepositswasfavorable,anexpansionofshalegasproductiononascalecomparabletotheAmericanexperienceoverthepastdecadecannotbereproducedinEuropeatpresent.OnlyunderthemostfavorablegeologicalconfigurationscouldshalegasextractionprovehighlyprofitableinEurope.Absentextremewellproductivity,oratechnologicalimprovementthatwouldlowerdrillingcostsorincrease recovery rates while complying with local environmental regulations, it appears verydifficultforshalegasextractiontohaveasignificantimpactonEuropeanenergymarkets.

References

Arps,J.J.(1944).AnalysisofDeclineCurves.InAIME(1944)160(p.228‐47).

Berman,A.E.,&Pittinger,L.F.(2011).U.S.ShaleGas:LessAbundance,HigherCost.

DGTrésor.(2005).Tauxd’actualisationpublicetcalculéconomique(p.1‐8).

EC. (2013).QuarterlyReportonEuropeanGasMarkets.DGEnergy (Vol.6).EuropeanCommission ‐DGEnergy.

EIA. (2011). Review of Emerging Resources: U.S. Shale Gas and Shale Oil Plays. U.S. EnergyInformationAgency.

EIA.(2012).Paddrillingandrigmobilityleadtomoreefficientdrilling.TodayinEnergy.U.S.EnergyInformationAgency.

EIA. (2013).TechnicallyRecoverableShaleOilandShaleGasResources :AnAssessmentof137ShaleFormationsin41CountriesOutsidetheUnitedStates.U.S.EnergyInformationAgency.

EIA.(2014a).DrillingProductivityReport,March2014.U.S.EnergyInformationAdministration.

EIA.(2014b).U.S.NaturalGasGrossWithdrawals.

Energy Modeling Forum. (2013). Changing the game? Emissions andmarket implications of newnaturalgassupplies.

Gény, F. (2010).CanUnconventionalGasbeaGameChangerinEuropeanGasMarkets ? TheOxfordInstituteforEnergyStudies.

Höök,M.,Hirsch,R.,&Aleklett,K.(2009).Giantoilfielddeclineratesandtheirinfluenceonworldoilproduction.EnergyPolicy,37(6),2262‐2272.

Hsieh,L.(2011).Europeanshalegas:alongroadahead.DrillingContractorMagazine,(July).

Hughes,J.D.(2013).Drill,Baby,Drill.PostCarbonInstitute.

IEA.(2012).GoldenRulesforaGoldenAgeofGas.InternationalEnergyAgency.

IFPEN. (2013).Hydrocarburesderoche‐mère‐Etatdeslieux. Institut Français du Pétrole EnergiesNouvelles.

IGU. (2012).WholesaleGasPriceFormation2012 ‐Aglobal reviewofdriversand regional trends.InternationalGasUnion.

IHS.(2011).TheEconomicandEmploymentContributionsofShaleGasintheUnitedStates.

IHS. (2013). America’s New Energy Future: The Unconventional Oil & Gas Revolution and the USEconomy.

Moniz,E.J.,Jacoby,H.D.,&Meggs,A.J.M.(2011).TheFutureofNaturalGas‐Appendix2D.

Nickelson, R. (2013). The SevenMajor U.S. Shale Plays.BakkenOilBusinessJournal, (November /December2012),30‐31.

Pulsipher,A.G.(2007).EstimatingDrillingCosts‐Conclusion:Systemsapproachcombineshybriddrillingcostfunctions.OilandGasJournal,(August).

Spencer, T., Sartor, O., & Mathieu, M. (2014). Unconventionalwisdom:aneconomicanalysisofUSshalegasandimplicationsfortheEU.

UnitedStatesCongress.CleanWaterAct,Pub.L.92‐500(1972).

UnitedStatesCongress.SafeDrinkingWaterAct,Pub.L.93‐523(1974).

UnitedStatesCongress.EnergyPolicyAct,Pub.L.109–58(2005).

WoodMackenzie.(2012).UKShaleGas–fiscalincentivesunlikelytobeenough.

Annex

Figure7:Centralscenarioresults

Annualnaturalgasproduction Cashflow

Figure8:ZeroNPVscenarioresults

Annualnaturalgasproduction Cashflow

Figure9:Extremescenarioresults

Annualnaturalgasproduction Cashflow

Figure10:Centralscenario,Wetgasvariantresults

Annualnaturalgasproduction Cashflow

Figure11:Centralscenario,Privatevariantresults

Annualnaturalgasproduction Cashflow

Figure12:Centralscenario,Decreasinggaspriceresults

Annualnaturalgasproduction Cashflow