Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
www.TheCIE.com.au
F I N A L R E P O R T
ReviewofdemandforecastsforSAWater
PreparedforEssentialServicesCommissionofSouthAustralia
January2013
THE CENTRE FOR INTERNATIONAL ECONOMICS
www.TheCIE.com.au
COMMERCIAL IN CONFIDENCE
TheCentreforInternationalEconomicsisaprivateeconomicresearchagencythatprovidesprofessional,independentandtimelyanalysisofinternationalanddomesticeventsandpolicies.
TheCIE’sprofessionalstaffarrange,undertakeandpublishcommissionedeconomicresearchandanalysisforindustry,corporations,governments,internationalagenciesandindividuals.
©CentreforInternationalEconomics2013
Thisworkiscopyright.Individuals,agenciesandcorporationswishingtoreproducethismaterialshouldcontacttheCentreforInternationalEconomicsatoneofthefollowingaddresses.
C A N B E R R A CentreforInternationalEconomicsGroundFloor,11LancasterPlaceMajuraParkCanberraACT2609
GPOBox2203CanberraACTAustralia2601
Telephone 61262457800Facsimile 61262457888Email [email protected]
Website www.TheCIE.com.au
S Y D N E Y CentreforInternationalEconomicsSuite1,Level16,1YorkStreetSydney NSW2000
GPOBox397Sydney NSW Australia 2001
Telephone 61292500800Facsimile 61292500888Email [email protected]
Website www.TheCIE.com.au
DISCLAIMER
WhileTheCIEendeavourstoprovidereliableanalysisandbelievesthematerialitpresentsisaccurate,itwillnotbeliableforanypartyactingonsuchinformation.
Review of demand forecasts for SA Water 1
www.TheCIE.com.au
Contents
Executive summary 3
Demandforecastshouldsupportpriceregulation 3Technicalamendmentstoforecasts 4
1 Introduction 7
Theroleofdemandforecastsineconomicregulation 7RegulatoryapproachforSAWater 8Thisreport 9
2 Structure of demand forecasts 11
CurrentpricesforSAWaterservices 11StructureofSAWater’sforecasts 12SAWater’smodel 13
3 Historical analysis of demand drivers 15
Overviewofapproach 15Validityofhistoricaldata 16Reviewofmodelestimationandapplicationofmodelresults 19
4 Projecting forward demand 27
Forecastcustomernumbers 27Bounceback 33
5 Revised forecasts 38
A Forecast of annual water consumption billed 42
BOXES, CHARTS AND TABLES
1 Originalandamendedforecastofconsumptionbilledoverregulatoryperiod 62 Cumulativechangestototalconsumptionbilledoverregulatedperiod GL 61.1 OverviewofregulatoryapproachforSAWater 82.1 Forecastnumberofcustomersconnectedtopotablewaternetwork 132.2 Forecastpotablewaterusebycustomercategory 133.1 DemanddriversusedinACIL’sdemandforecasts 153.2 RevisedABShistoricalpopulationdata 163.3 Transitiontoquarterlybilling 173.4 Changestomodelresultswithamendmenttohistoricalwaterrestriction
data 173.5 Sensitivityofhistoricalwaterrestrictiondata 18
2 Review of demand forecasts for SA Water
www.TheCIE.com.au
3.6 Fittedresidentialmodelwithadjustmenttohistoricalwaterrestrictions 183.7 Twonon‐stationaryseries 203.8 AnnualgrowthinpopulationandSAWatercustomers 213.9 Historicalcustomergrowthlesspopulationgrowth 213.10 Differenceinaverageannualgrowthratesforpopulation,dwellingand
customergrowth 223.11 AnnualgrowthrateofcommercialcustomersandGSP 233.12 Commercialaverageconsumptionmodel 254.1 Forecastofresidentialcustomers 284.2 Forecastofcommercialcustomers 294.3 RevisedGrossStateProductionforecasts 304.4 Reviewofadjustmenttoresidentialpriceelasticity 324.5 Totalwaterconsumptionforecastwithadjustedandunadjustedprice
elasticities 324.6 Originalandrevisedpricecoefficientsfromannualregressionmodels 334.7 Deviationofforecastsfromactualconsumptiondatain2011‐12 344.8 Bouncebackof50percentover2years 354.9 Bouncebackof75percentover3years 364.10 Bouncebackof100percentover4years 365.1 Originalandamendedforecastofcustomernumbers 385.2 Originalandamendedforecastofconsumptionbilledoverregulatoryperiod 395.3 Cumulativechangestototalconsumptionbilledoverregulatedperiod GL 405.4 Cumulativechangestoresidentialconsumptionbilledoverregulatedperiod
GL 405.5 Cumulativechangestocommercialconsumptionbilledoverregulated
period GL 415.6 Cumulativechangestonon‐residentialconsumptionbilledoverregulated
period GL 41A.1 Forecastofannualwaterconsumptionbilled GL 42
Review of demand forecasts for SA Water 3
www.TheCIE.com.au
Executivesummary
TheEssentialServicesCommissionofSouthAustralia ESCoSA iscurrentlydevelopingtheregulatorypricingapproachforSAWater.ESCoSAwillpublishitsfirstdeterminationforSAWaterinearly2013,coveringthethreeyearperiodfrom2013‐14to2015‐16.
ACILTasmanwascommissionedbySAWatertoforecastwaterdemandfortheregulatoryperiodfortheresidential,commercialandnon‐residentialsectors.
Subsequently,theCIEwasappointedbyESCoSAtoreviewACILTasman’sdemandforecasts.Ourapproachtothisreviewinvolvesthefollowingbroadsteps:
■ reviewthehistoricalanalysisofdemanddrivers,includingthevalidityofthehistoricaldataandreviewofACIL’smodelandapplicationofmodellingresults;
■ reviewofACILTasman’sindependentprojectionsanddemandforecastingmethodology,includingareviewofthemethodtoforecastcustomernumbersandaverageconsumptionperconnection;
■ providerecommendationswherenecessary;and
■ provideareviseddemandforecastbasedonamendmentsdiscussed.
InundertakingourreviewwefoundACIL’smodelandmethodologypaperprovidedahighleveloftransparencyinthedataandmodellingunderpinningtheforecast.Theapproachtoforecastingdemandwasgenerallysoundandourrecommendationsarelargelytoimproveonparticulartechnicalareasoftheforecastingapproachusedandupdatestothedataused.
ForthepurposesofpriceregulationofSAWater,wefoundtwocategoriesofissueswerecommendtobeaddressedintheforecasts.
1 Thedemandforecastsneedtobeabletosupporttherequirementsofpriceregulation.
2 Technicalamendmentstotheforecastingapproachandmethodologythatwilllikelyimproveonthecurrentforecasts.
DemandforecastshouldsupportpriceregulationThedemandforecastssubmittedbySAWatertoESCoSAaretoassistESCoSAinmakingrequiredregulatorydecisions.Akeypartofthisisthatforecastsprovideinformationoneachaspectofdemandthatischargedataseparateprice.Thismeansthatforecastsshouldincludecustomernumberswherethereisafixedchargepercustomerandtier1,2and3demandprojectionswherethereisadifferentpricechargedforeachtier oranothermethodologytoarriveatthis .
4 Review of demand forecasts for SA Water
www.TheCIE.com.au
Intheircurrentform,thedemandprojectionsdonotprovidesufficientinformationforESCoSAtomultiplypricesbydemandtoarriveatrevenue.Elementswhicharecurrentlymissinginclude:
■ aforecastofthenumberofnon‐residentialcustomerswhichisneededtoestimatefixedchargesfromthesecustomers;
■ forecastsofthetotalnumberofbilledcustomers;
– weunderstandthatthecustomernumberprojectionsrelatetocustomersusingwaterbutthatthereisalargersetofcustomersthatischargedforwater i.e.facingfixedbutnotusagecharges ;
■ forecastsofcustomernumbers orpropertyvaluestowhichchargesareapplied forsewerageservices,
– whichweunderstandtobedifferenttowaterservices;and
■ estimationofwaterdemandateachwaterpricingtierforresidentialusageoranalternativemethodtoallowfortheregulatortocalculaterevenuefromusageprojections.
Further,SAWaterseparatelydevelopedforecastsofsewerageconnections,valueofpropertyagainstwhichseweragechargesareleviedandquantityofseweragetreatmentdemanded.Thefirstandsecondwouldbothpotentiallyberelevantforrevenueprojections,howevertheseforecastshavenotbeenincludedinSAWater’ssubmission.
TechnicalamendmentstoforecastsBasedonourreviewofthedemandforecastswerecommendamendmentstothecurrentmodelassetoutbelow.Theserecommendationsincludeupdatestohistoricaldataandindependentprojections whererevisionhasoccurredsubsequenttoACIL’sanalysis aswellasamendmentstomodellinghistoricalrelationshipsandtheapplicationofempiricalresultstoforecastcustomernumbersandaverageconsumptionperconnection.
WehaveincorporatedtheserecommendationsintoACIL’sdemandforecastmodelandprovidedupdateddemandestimates.Thisallowsfordifferentpossible‘bounceback’scenariosasdiscussedbelow.
Theconsumptionoutcomefor2011–12suggeststhattherehasbeensomebouncebackinconsumptionbutthatconsumptionremainslowerthantheconsumptionexpectedgivencurrentconditionsandthereplacementofwaterrestrictionswithwaterwiserules.Itisdifficulttoknowhowmuchconsumptionwillcontinuetoincreasetothelevelsexpectedgivencurrentprices,waterwiserulesandotherfactorsthatimpactonconsumption.Consumptionmaytakealongtimetoreachthelevelthatwouldbeexpectedifrestrictionshaveledtopeopleinvestinginwatertanksorlowwatergardens.Itmayalsotaketimeforpeopletorevertbacktousingwaterforactivitiesthattheydidpre‐restrictions.
Review of demand forecasts for SA Water 5
www.TheCIE.com.au
RECOMMENDATIONS
1 HistoricalpopulationdatabeupdatedwithinACIL’sdemandforecastingmodel. 16
2 Annualbillingdatain2009‐10shouldbeadjustedtoreflectthatalargepartofnon‐residentialwaterconsumptionwasbilledquarterlypriorto2009‐10. 17
3 Thedemandforecastingmodelbeadjustedtoreflectthatbothlevel1and3waterrestrictionsappliedin2010‐11. 19
4 Analternativeapproachbeusedfordevelopingforecastsofcustomernumbers.Optionsaresetoutinchapter4. 20
5 Therestrictionthattheconstantiszeroforthecommercialusagemodelberemoved. 25
6 Themethodologyusedtoforecastresidentialandcommercialcustomernumbersbeamendedtoapplygrowthtothemostrecentdataoncustomernumbersratherthanforecastingthelevelofcustomernumbers. 28
7 UpdatedGSPforecastsshouldbeincorporatedintothedemandforecastingmodelusedtogenerateaverageconsumptionpercommercialcustomerandtotalannualnon‐residentialconsumption. 30
8 Theunadjustedpriceelasticity,asestimatedintheresidentialconsumptionmodel,shouldbeusedtoforecastaverageresidentialwaterconsumptionperconnection. 33
9 Anadjustmentshouldbemadetotheprojectionstoallowforwaterdemandtorecovergraduallytoexpectedconsumptionwhichalsoreflectsbouncebackofwaterconsumptionexperiencedin2011‐12. 37
Thepossibleoptionsforbouncebackcovertheextenttowhichconsumptionrevertsbacktolevelsexpectedforwaterwiserules givencurrentpricesandotherfactors andthetimingofthischange.Giventheuncertaintysurroundingtheoccurrenceofbouncebackofwaterconsumption,wehavemodelledtherevisedforecastoftotalconsumptionbilledovertheforecastperiodforthreebouncebackscenarios table1 .
■ Bouncebackscenario1—allowingfor50percentbouncebackovertwoyears.Thisleadstodemandbeing4.1percentlowerthanACIL’soriginalforecastovertheregulatoryperiod.
■ Bouncebackscenario2—allowingfor75percentbouncebackoverthreeyears.ThisgivessimilardemandprojectionsovertheregulatoryperiodasACIL’soriginalforecast.
■ Bouncebackscenario3—allowingfor100percentbouncebackoverfouryears.Underthisscenarioexpecteddemandis2.9percenthigherthanACIL’soriginalforecast.
6 Review of demand forecasts for SA Water
www.TheCIE.com.au
1 Original and amended forecast of consumption billed over regulatory period
Total consumption billed (ML) over regulatory period (2013-14 to 2015-16)
ACIL’s original forecast
CIE Amended forecast
Bounceback assumption 100% over 1 year
50% over 2 years
75% over 3 years
100% over 4 years
Residential 367 592 338 906 355 993 370 958
Commercial 28 716 28 495 29 676 30 703
Non residential 147 707 153 971 156 116 157 931
Total 544 014 521 373 541 785 559 592
Source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
Theimpactsofthechangesrecommendedandeachbouncebackscenarioonconsumptionforecastsareshowninchart2.
2 Cumulative changes to total consumption billed over regulated period (GL)
Data source: The CIE.
490
500
510
520
530
540
550
560
570
580
590
AC
IL's
orig
inal
fore
cast
Upd
ate
AB
SP
opul
atio
n an
dQ
ua
rte
rly B
illin
g
His
toric
al w
ater
rest
rictio
ns
Fo
reca
st c
usto
mer
num
ber
s
Inte
rcep
t in
com
mer
cial
mod
el
Una
dju
sted
resi
dent
ial p
rice
elas
ticity
GS
P fo
reca
st
Bou
nceb
ack
(50%
/2yr
s)
Bou
nceb
ack
(75%
/3yr
s)
Bou
nceb
ack
(100
%/4
yrs)
Tota
l con
sum
ptio
n bi
lled
(GL)
Review of demand forecasts for SA Water 7
www.TheCIE.com.au
1 Introduction
TheEssentialServicesCommissionofSouthAustralia ESCoSA hasrecentlybeenappointedastheeconomicregulatorofthewaterindustryinSouthAustralia.ESCoSAwillpublishitsfirstdeterminationforSAWaterinearly2013,coveringthethreeyearperiodfrom2013‐14to2015‐16.ESCoSAwillbemakingadeterminationofSAWater’sdrinkingwaterandseweragepricesfor2013‐14to2015‐16underPart3oftheESCAct,asauthorisedbysection35oftheWaterIndustryBill2011.Inordertodothis,ESCoSAisrequiredtoconsidertheexpecteddemandforSAWater’sservicesintothefuture.ThedemandforecastsincludeboththenumberofcustomersexpectedtobeconnectedtoSAWater’snetworkandtheexpectedlevelofusageovertheupcomingregulatoryperiod.
TheroleofdemandforecastsineconomicregulationDemandforecastsaretypicallyaprimaryinputintodecision‐makingbyregulatedutilities.Theycanhelptoinform:
■ pricingstructureswhichmaybeabletobechangedthroughouttheregulatoryperiod;
■ marketing—demandforecastingrequiresanunderstandingofthechoicescustomersandpotentialcustomersaremaking,whichisusefulinformationfortargetingofcustomers;
■ risksandriskmanagement—ifdemandforecastshaveastochasticcomponentratherthanbeingasingleforecast;and
■ capitalandoperatingexpenditureplanningdecisions.
Demandforecastsalsoformaprimaryinputintothedecisionsbyeconomicregulators.Demandforecasts,forexample,caninfluence:
■ thenotionalrevenueallowancethrough:
– operatingexpenditureprojections;and
– capitalexpenditureprojectionsandhencetheregulatoryassetbase,whichinturnimpactsondepreciationandthereturnoncapital.
■ prices underapricecapapproach aspricesaresetsothatdemandmultipliedbypricesisequaltothenotionalrevenueallowance.
Theroleofdemandforecastswilldependontheregulatoryapproachadopted.InAustralia,economicregulationtypicallytakestheformofpricecapsorrevenuecaps.
■ Underapricecapapproach,pricesaresetbytheeconomicregulatorbasedontheirassessmentofthedeterminedrevenueforthebusinessbyassessingeachofthecoststhatformthe‘buildingblock’.Inthisinstance,expecteddemandisusedto‘translate’
8 Review of demand forecasts for SA Water
www.TheCIE.com.au
therevenuerequirementintopricesthataredeterminedbytheregulator.Thebusinessbearsthedemandriskwhereactualdemanddiffersfromthatforecastedattheregulatorydetermination.Regulatorymechanismscanbedevelopedtominimiseabusiness’exposuretothedemandrisk.
■ Underarevenuecapapproachtoregulation,theregulatordeterminestherevenuerequirement.Inthisinstance,thebusinesshascontroloverprices.Forexample,businessescanraiseorlowerpricestoachievethedeterminedrevenuerequirement,takingintoaccountthedemandsidefactors.
RegulatoryapproachforSAWaterTheregulatoryapproachforSAWaterisstillbeingdevelopedandthereareelementsoftheapproachthatwillneedtobeexplicitlydeterminedbyESCoSA.Chart1.1providesanoverviewofthebroadregulatoryframeworkcurrentlybeingconsideredforSAWater.
1.1 Overview of regulatory approach for SA Water
Source: The CIE
Review of demand forecasts for SA Water 9
www.TheCIE.com.au
TheregulatoryapproachforSAWaterincludesthefollowing.
■ TheexpectedrevenueineachyearoftheregulatoryperiodthatSAWaterwillreceivewillbebasedon:
– pricessetbytheSouthAustralianGovernment.InitsMay2013RegulatoryStatement,theGovernmentwillsetthe‘pricepathforecast’for2013‐14to2015‐16.1
– ESCoSAwilldeterminetheexpectednumberofcustomersandwateruse.2
■ TheestimatedBuildingBlockRevenuerequirementineachyearoftheregulatoryperiodwillbebasedon:
– ESCoSA’sestimatesoftheefficientoperatingandappropriatecapitalallowance.Thelateritemwillrequireanassessmentoftheefficientlevelofcapitalexpenditure,theWeightedAverageCostofCapital,theappropriateassetdepreciationratesandusefullifeofassets.3
– theSouthAustralianGovernment’sestimateoftheopeningRegulatoryAssetBase.4TheopeningRABisexpectedtobe‘back‐solved’toensurethattheBuildingBlockRevenueoverthe3yearregulatoryperiodequatestotheexpectedrevenueoverthethreeyearperiod.
DemandforecastsfortheregulatoryperiodshouldbepresentedinawaythatallowsESCoSAtounderstandtherevenueimplicationsofparticularpricepathsorthepriceimplicationsofparticularrevenuepaths.ThisrequiresthesubmissionofseparatedemandforecastsforeachofthepriceelementsthatSAWaterlevies.
ThisreportThepurposeofthisprojectistoprovideESCoSAwithadviceonthereasonablenessofSAWater’sprojectionsandtoprovideanalternativedemandforecastwheregapshavebeenidentified.Thefocusoftheanalysisisonaveragedemandineachyearofthe
1 TheSouthAustralianGovernmentindicatedthatit“doesnotexpectwaterpricestogrow
significantlyfasterthaninflation.However,theactualchangeinwaterpriceswilldependontheinitialvalueoftheRegulatedAssetBase,whichtheTreasurerwillsetinMay2013”http://www.escosa.sa.gov.au/library/121015‐ClarifyWaterIssues‐MediaRelease.pdf
2 ThePricingOrderrequirestheCommissionto“includeamechanismwhichallowsfortheadjustmentofallowablerevenuetobederivedwheretheCommissiondeterminestheretobeamaterialvariationbetweenforecastandactualwaterconsumptionorsewerageconnections”.InitsregulatorysubmissionSAWaterhasproposeda‘bankingmechanism’whichwouldbecarriedforwardtoadjusttherevenuerequirementatthenextregulatorydetermination.
3 Theoperatingandcapitalexpenditurerequirementswillalsoneedtobeconsistentwiththeforecastcustomernumbersandwateruse,giventhatthesefactorsareoneofthedriversofexpenditure.
4 TheopeningRABwillthevalueasat1July2013.TheRABat1July2014andfutureyearswillbebasedonrollingforwardcapitalexpenditureincurredduringpreviousyearwithadjustmentssuchasdepreciation.
10 Review of demand forecasts for SA Water
www.TheCIE.com.au
regulatoryperiod,ratherthanpeakdemand.5Whiletheaveragedemandispresentedforeachyear,regulatorstypicallydonotseektocapturethepotentialyearonyeardemandvolatilityduetochangingrainfallandtemperaturepatterns.Inthissense,theannualpredictivecapacityofanydemandforecastingmodelislessimportantthanthemodelresultsbeingcorrectonaverageovertheregulatoryperiod.
Ourfinalreportprovidesourexpertviewonanumberofspecificaspectsincludingwhetherthedemandforecasts:
■ recogniseandreflectthekeydriversofdemand egdemographicchange,priceeffects,economicfactors,weatherrelatedfactors,policyfactors ;
■ arebasedonsoundassumptions,applybestpracticeprinciplesandutilisethebestavailableinformation;
■ areconsistentwithotheravailableforecastmethodologies;
■ arebasedonthemostrecentlyavailabledata;
■ arebasedonsoundandrobustaccountsofcurrentmarketconditionsandfutureprospects;
■ arebasedonvalidmethodologyandassumptions;
■ havebeensubjecttoappropriatetestingandvalidationexercises;
■ arefreefromstatisticalbias;and
■ containanyactualorpotentialweaknesses,uncertaintiesandrisks.
Thefindingshaverequired,wherepossible,testingtheassumptionsincludedinSAWater’smodellingbasedondatafromotherindependentsources.Ourefforthasbeenfocusedonthosefactorsthatmateriallyimpactontheresults.
Finally,thisreportfocusesontheforecastsrelatingtopotablewaterdemand.AtthisstagerecycledwaterisnotsubjecttoESCoSA’sregulatoryreviewandpriceswillcontinuetobesetbytheSouthAustralianGovernment.6
5 Theanalysisofpeakdemandismorerelevantforopex/capexreviews,giventhatpeak
demandisadriverofinvestmentdecisions.
6 Currentlythepriceofrecycledwaterforresidentialcustomersthatreceivedualreticulationservicesis75percentofthesecondtierresidentialpotablewaterusageprice.
Review of demand forecasts for SA Water 11
www.TheCIE.com.au
2 Structureofdemandforecasts
Inaregulatorysettingtheregulatorisrequiredtodetermineforecastsofallelementsthatarerequiredtobeset.Thatis,thereneedstobeaspecificquantitytoapplytothepriceinordertodeterminetherevenueimplications.
CurrentpricesforSAWaterservicesThepricescurrentlysetforSAWater’sservicesincludethefollowing7:
■ Forpotablewaterservices:
– Awatersupplychargeperpropertywithaccesstothepotablewaternetwork.8… Forresidentialcustomersafixedchargeapplies.In2012‐13thechargewas
$73.25perquarter.… Forcommercialcustomersthewatersupplychargeisbasedonthevalueof
theirproperty,subjecttoaminimumcharge $73.25perquarterin2012‐13 .9
… Forothernon‐residentialcustomersafixedchargeapplies $73.25perquarterin2012‐13 .
– Ausagechargebasedonthevolumeofwaterconsumedduringtheyear.Anincliningblockusagechargeappliesforbothresidentialandnon‐residentialcustomers.… Forresidentialcustomersdifferentpricesapplyto3blocksofusage 0to30
kL,30to130kLandabove130kLofquarterlyusage .Thelastblockonlyappliestosingledwellings.Formultipledwellings egblocksofflats onlythefirsttwotiersapply.
… Forcommercialandothernon‐residentialcustomersasingleusagechargeappliesirrespectiveofthelevelofusage.10
■ Forwastewaterservices,afixedsupplychargeappliestoeachpropertyconnectedtothewastewaternetwork.Thechargeisbasedonthecapitalvalueoftheproperty.Nousagechargecurrentlyapplies.
7 ThisinformationisdrawnfromSAWater’swebsite
http://sawater.com.au/SAWater/YourBusiness/YourAccountBillPaymentCharges/Pricing Information.htm
8 Thisincludesvacantlandthatisconnectedtothenetworkbutcurrentlydoesnotconsumewater.
9 ThecapitalvalueofthepropertyisassessedbytheValuer‐GeneralinSouthAustralia.
10In2011‐12andearlieryearsatwotieredusagechargewasapplied 0to30kL,above30kLofquarterlyusage forcommercialandothernon‐residentialcustomers.
12 Review of demand forecasts for SA Water
www.TheCIE.com.au
■ Fordualreticulationrecycledwatercustomers,ausagepriceof75percentofthesecondtierpriceforresidentialusageapplies.Noadditionalfixedsupplychargeapplies.
Thepricesfortheseservicesareapplieduniformlyonastate‐widebasis.ForthisdeterminationESCoSAisresponsibleforregulatingpotablewaterandwastewaterservicesdeliveredbySAWater.Asnotedearlier,recycledwaterservicesdeliveredbySAWaterwillcontinuetoberegulatedbytheSouthAustraliangovernment.
StructureofSAWater’sforecastsIndevelopingitsdemandforecastsforthisdeterminationSAWaterengagedtheservicesofACILTasmantodevelopthedemandforecastsforpotablewaterprices.11ACIL’sapproachinvolvedthreesteps.
■ Identifyingthekeydriversofcustomernumbersandtheaverageconsumptionpercustomer
■ EstimatingtheimpactofeachofthedriversusinghistoricalbillingdatafromtheRatingsAnalysisreportsextractedfromSAWater’squarterlybillingdatabase.12ACILalsousedmonthlybulkwatersupplyasacrosscheckoftheanalysis.Thebulkwatersupplydataalsoincludesleakageandunbilledcustomers.
■ Generatingprojectionsforcustomernumbersandaverageconsumptionpercustomer basedonprojectionsofthekeydriversandtheestimatedimpactsofeachofthedriversderivedfromthepreviousstep .
ACILTasman’sforecastsofcustomernumbersarepresentedintable2.1.Separateforecastsarepresentedforthetwocustomercategories,residentialandcommercial.Thisdivisionreflectstheviewthatthedriversofcustomernumbers andusage arelikelytodifferacrosstheresidentialandcommercialsectorsand,therefore,separateanalysisisrequired.ItshouldbenotedthatSAWater’sbillingsystemincludesafurtherdisaggregationofthecustomercategories.Forexample,theresidentialcategoryismadeupof10sub‐categorieswhichincludesingledwellings,homeunitsandflats.
11DemandprojectionsrelatingtowastewaterservicesweredevelopedseparatelybySAWater.
Thesedonotformpartofourreviewwhichisfocusedonpotablewaterdemand.
12SAWaterhasindicatedthatrecycledwaterusageandcustomernumbershavebeenremovedfromtheRatingsAnalysisreport.Billedrecycledwaterusagein2010‐11wasapproximately2.4GLwhichwasalmostdoubletheusagein2009‐10.Therewere7514billedrecycledwatercustomersin2010‐11.
Review of demand forecasts for SA Water 13
www.TheCIE.com.au
2.1 Forecast number of customers connected to potable water network
Customer type 2012-13 2013-14 2014-15 2015-16
‘000 customers ‘000 customers ‘000 customers ‘000 customers
Residential 655.7 665.2 (1.4)
674.7 (1.4)
684.2 (1.4)
Commercial 27.4 27.7 (1.1)
28.0 (1.1)
28.3 (1.1)
Note: Figures in parentheses present the growth compared to the prior year. We understand that customer numbers includes only customers that use water and that there are additional customers that are charged but that do not have any water usage. SA Water has not provided projections of customer numbers for the non-residential sector.
Source: SA Water Regulatory Business Proposal, table 5-7.ACIL Tasman (2012) tables 10 and 13.
SAWater’spotablewaterusageforecastsbycustomercategoryarepresentedintable2.2.Anadditionalcategoryofother‘non‐residential’customersisincluded.Theusageforecastsarepresentedasanaggregateforeachcustomercategory.
2.2 Forecast potable water use by customer category
Customer type 2012-13 2013-14 2014-15 2015-16
GL GL GL GL
Residential 119.0 120.8 (1.5)
122.5 (1.4)
124.3 (1.5)
Commercial 9.1 9.3 (2.2)
9.6 (3.2)
9.8 (2.1)
Other non-residential 48.2 48.7 (1.0)
49.3 (1.2)
49.7 (0.8)
Total 176.3 178.9 (1.5)
181.4 (1.4)
183.8 (1.3)
Note: Figures in parentheses present the growth compared to the prior year.
Source: SA Water Regulatory Business Proposal, table 5-7.
SAWater’smodelTocalculatethevalueofExpectedRevenueflowingfromregulatedprices,atariffmodelthatisabletolinkprices/tariffstochangingcustomernumbersandtotalwaterusageisusuallyemployed.CurrentlyadetailedtariffmodelisnotavailableforthecalculationofExpectedRevenueestimatesinSouthAustralia.
Atariffmodelwillberequiredinthefuturefollowing,ordevelopedinlinewith,thecorrectionofanumberofgapsinSAWater’sforecasts.Theinformationgapsare:
■ Residentialsector
– Theforecastshavebeenpreparedatanaggregatelevelandseparateforecastshavenotbeenprovidedforeachofthethreetiers.Itcouldbeexpectedthatwaterusageineachtierwouldgrowatadifferentrate.SAWaterhassubsequentlyprovidedanestimateofthemarginalpriceforallcustomers.Wehavenotverifiedwhetherthismarginalpriceisreasonable.
14 Review of demand forecasts for SA Water
www.TheCIE.com.au
– Customernumberforecastsarenotforchargedcustomersbutonlyforthosethatusewater.SAWateralsochargespropertieswithaccesstowaterbutwithnousage.Thesearenotincludedintheforecastsandtheexpectednumberofsuchcustomersisrequiredtocalculatethisrevenueitem.
■ Commercialsector
– Theaveragepropertyvalueshavenotbeenprovidedtoestimatetherevenuefromthewatersupplycharge;instead,onlycommercialcustomernumbershavebeenprojected.Thismaybeaproblemwherenewcommercialcustomershaveadifferentaveragepropertyvaluetoexistingcustomers.
■ Othernon‐residentialsector
– Forecastcustomernumbershavenotbeenprovided,whichmeansthatfixedsupplyrevenuefromthiscategorycannotbeprojected.
Noinformationhasbeenprovidedthatwouldallowwastewaterrevenuetobecalculated.
Review of demand forecasts for SA Water 15
www.TheCIE.com.au
3 Historicalanalysisofdemanddrivers
OverviewofapproachACILTasmanhasusedhistoricaldatatoidentifyrelationshipsbetweenkeydriversandbothcustomernumbersandaverageconsumptionperconnection.
Thehistoricaldataondriversofdemandintheresidential,commercialandnon‐residentialsectorsusedtoforecastcustomernumbersandtotalconsumptionofwaterbilledforeachsectoraredetailedintable3.1.
ACILTasmanexplainedinitsmethodologypaperthatotherpossibledemanddriverssuchasdemandmanagementactivities,wereexcludedbasedonalackofstatisticalsignificanceinthemodelling.
3.1 Demand drivers used in ACIL’s demand forecasts
Demand drivers Residential Commercial Non residential
Total Bulk water supply
Customer numbers
Average usage
Customer numbers
Average usage
Total usage (monthly)
Population (% annual growth)
Economic activity (Gross State Product)
Price of water ($/kL, second tier)
Temperature (CDD18)
Water restrictions (level)
Rainfall (mm)
Evaporation(mm)
Source: ACIL, 2012, SA Water’s demand forecasting: a report on the development of Demand Forecasting Methodology, Model and Supporting Documentation. Attachment E.1 Demand Forecasting Methodology.
WereviewedthedataandmethodologyusedinACILTasman’sdemandforecastandfoundthat:
■ thehistoricaldatausedisreasonablewiththeexceptionofthetreatmentofwaterrestrictionsin2010‐11andtheadjustmentforquarterlybillingin2009‐10;
■ thedatausedwasuptodateatthetimeofthedevelopmentoftheforecastsbutthereisnowimprovedinformationonhistoricalpopulationgrowthfromthe2011census;and
■ thebroadmethodofordinaryleastsquaresregressionissoundbutthattherearemethodologicalissuesregardingtheuseofthistechniqueinsomeinstances.
Theareaswhereweconsiderthatthereareissueswiththeforecastsaresetoutbelow.Notwithstandingthese,inlargepartthemethodologyusedisconsideredreasonable.
16 Review of demand forecasts for SA Water
www.TheCIE.com.au
Validityofhistoricaldata
Historicalpopulationdata
TheAustralianBureauofStatisticsreviseditspublishedEstimatedResidentPopulationdataforSouthAustraliasubsequenttoACIL’sanalysis.RevisionsweremadetotheSouthAustralianseriesoftheABSEstimatedResidentPopulationdataforeachyearbetween2006‐07and2010‐11 table3.2 .
3.2 Revised ABS historical population data
Year Original data used in ACIL’s modelling
Updated data Difference
2007 1 585 794 1 582 559 - 3 235
2008 1 603 985 1 597 343 - 6 642
2009 1 624 512 1 614 375 - 10 137
2010 1 644 152 1 629 434 - 14 718
2011 1 657 001 1 638 232 - 18 769
Source: ABS, Australia Demographic Statistics. Table 4: Estimated resident population, states and territories. Cat. No. 3101.0.
NotethattheuseoftheserevisedprojectionsmakesasubstantialdifferencetotheresidentialcustomernumbermodelusedbyACIL.13
RECOMMENDATION
1 Historical population data be updated within ACIL’s demand forecasting model.
Adjustmentforquarterlybillingdata
SAWaterchangeditsbillingmethodin2009‐10fromsixmonthlytoquarterlybilling.ACILadjustedtheannualconsumptiondatatoavoidthe2009‐10dataoverstatingthebilledwatersales.Thisadjustmentwasmadeacrossthethreecustomerclassesresidential,commercialandnon‐residential .However,asnotedinACIL’sreport,thelargenon‐residentialcustomerswerebilledquarterlypriorto2009‐10andthesecustomersconstitutethemajorityofwaterconsumptionforthissector.Therefore,ACIL’sadjustmentmethodologyleadstowaterconsumptionforresidentialandcommercialcustomerstobeoverstatedandnon‐residentialwaterconsumptiontobeunderstatedfor2009‐10.
Wehaveadjustedtheannualbillingsdatain2009‐10toreflectthatthemajorityofnon‐residentialwaterconsumptionwasalreadybilledquarterly table3.3 .Thismakesnoadjustmenttonon‐residentialdata.Ifthereisinformationontheshareofnon‐
13WeupdatedACIL’sresidentialcustomernumbersmodelwiththerevisedABShistorical
populationdata.Thecoefficientonthesingleexplanatoryvariableincreasedfrom0.59estimatedinACIL’soriginalmodel to0.65.Theestimateonthecoefficientimpliesthatforagivenincreaseinthepopulationof100people,SAWater’scustomerbasewillincreaseby65customers.
Review of demand forecasts for SA Water 17
www.TheCIE.com.au
residentialconsumptionthatmovedfromsixmonthlytoquarterlybillingthenthissharecouldinsteadbeused.
3.3 Transition to quarterly billing
Annual consumption billings ACIL’s amendment to quarterly billing data
CIE’s amendment to quarterly billing data
2009-10 (ML) (ML)
Residential 123 267 117 703
Commercial 9 908 9 461
Non residential 52 455 58 466
Total 185 630 185 630
Source: ACIL, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
RECOMMENDATION
2 Annual billing data in 2009-10 should be adjusted to reflect that a large part of non-residential water consumption was billed quarterly prior to 2009-10.
Aggregationofwaterrestrictions
Inthehistoricaldata,eachyearisallocatedtoeithernorestrictions,orlevel1,2or3restrictionsbasedonthelevelofrestrictionexperiencedforthemajorityofthefinancialyear.Thisapproachisproblematicwhenmultiplewaterrestrictionslevelsappliedinagivenfinancialyear.Forexample,in2010‐11level3waterrestrictionsappliedfromJulyuntilDecember,whichwererelaxedtolevel1waterrestrictionsfromDecemberuntilJune withtheexceptionoftheEyrePeninsula .
InitsmodellingACILassumedlevel1waterrestrictionsappliedfortheentiretyof2010‐11.ThisdoesnotcapturethefivemonthperiodbetweenJulyandDecemberwherecustomersfacedlevel3waterrestrictions.
WetestedthesensitivityofACIL’sassumptionandre‐estimatedthewaterconsumptionmodelsforresidential,commercialandnon‐residentialwiththeassumptionthatbothlevel1andlevel3waterrestrictionseachappliedforhalfof2010‐11 table3.4 14.Theparameterswereverysensitivetothisassumption.
3.4 Changes to model results with amendment to historical water restriction data
Demand drivers Residential Commercial Non-residential
Original Amended Original Amended Original Amended
Intercept 4.57 4.65 0.00 0.00 6.46 6.41
GSP na na 0.47 0.48 0.36 0.37
Cooling degree days (CDD 18) 0.15 0.14 0.12 0.11 0.10 0.09
Real price residential -0.38 -0.27 -0.37 -0.27 -0.32 -0.28
(Continued on next page)
14Resultsintable3.4arefromamendinghistoricalwaterrestrictiondatain2010‐11,holdingall
elseconstant.
18 Review of demand forecasts for SA Water
www.TheCIE.com.au
Demand drivers Residential Commercial Non-residential
Original Amended Original Amended Original Amended
Level 1 -0.11 -0.10 -0.11 -0.10 -0.12 -0.12
Level 2 -0.15 -0.14 -0.15 -0.15 -0.14 -0.14
Level 3 -0.27 -0.29 -0.25 -0.27 -0.19 -0.19
Note: na not applicable because demand driver not included in model
Source: ACIL, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
Theestimatesforbothsalesperresidentialconnectionandtotalresidentialconsumptionbilledincreasebyapproximately7.6percentineachforecastedyearwhenlevel3restrictionsareincludedinthehistoricaldataforhalfof2010‐11 table3.5 .
3.5 Sensitivity of historical water restriction data
Assumed level 1 restrictions in place in 2010-11a
Assumed level 1 and level 3 restrictions both applied in 2010-11b
Changes to estimated residential sales per connection
2013-14 181.57 195.32
2014-15 181.61 195.35
2015-16 181.65 195.39
Changes to estimated total residential consumption billed (ML)
2013-14 120 775 129 919
2014-15 122 530 131 801
2015-16 124 286 133 684
a Original estimates from ACIL’s modelling. b Estimates from ACIL’s model with an amendment to historical water restrictions in 2010-11.
Source: The CIE.
Themodelfittoactualhistoricaldataimprovesslightlywhenlevel1andlevel3waterrestrictionsareappliedequallyin2010‐11 chart3.6 .
3.6 Fitted residential model with adjustment to historical water restrictions
4.90
5.00
5.10
5.20
5.30
5.40
5.50
5.60
5.70
1996-97 1998-99 2000-01 2002-03 2004-05 2006-07 2008-09 2010-11
ln(k
L/co
nnec
tion)
Actual Fitted (Level 1 in 2010-11) Fitted (Level 1 and 3 in 2010-11)
Review of demand forecasts for SA Water 19
www.TheCIE.com.au
Data source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
ThisissuehasbeenraisedwithSAWaterandSAWatersupportsACIL’sassumption.SAWaterindicatedthattheSouthAustralianGovernmenthadsignalleditsintentiontomovetolevel1restrictionsapproximately3monthspriortoDecember2011.SAWaterbelievedthatitisreasonabletoexpectthatconsumerschangedtheirwaterusebehaviourfromthedatewhichtheGovernmentsignalleditsintention.SAWateralsoindicatedthatwaterrestrictionspredominantlyimpactwateruseinthemonthsofJanuaryandFebruary.Therefore,itdidnotconsiderwhetherlevel1orlevel3waterrestrictionswereappliedintheJulytoDecember2010wouldmateriallyimpacttheresults.SAWaterhasnot,however,providedanyempiricalanalysistosupportitsviews.
Giventhisandtheimprovementintheempiricalfitofthemodelweconsiderthatanadjustmentshouldbemadetoallowforlevel3restrictionstoapplyforpartof2010‐11.
RECOMMENDATION
3 The demand forecasting model be adjusted to reflect that both level 1 and 3 water restrictions applied in 2010-11.
Reviewofmodelestimationandapplicationofmodelresults
Estimatingcoefficientsforcustomernumbersmodel
Modellingnon‐stationaryseries
ACILestimatedtherelationshipbetweenhistoricalresidentialcustomernumbersandpopulationgrowth,andsimilarlytherelationshipbetweenhistoricalcommercialcustomernumbersandgrossstateproduct.
Inboththeresidentialandcommercialcustomernumbersmodels,thedependentvariable customernumbers andthesingleindependentvariable populationintheresidentialmodelandgrossstateproductinthecommercialmodel areclearlynon‐stationaryseries chart3.7 .15
Inthiscasetheseriesarenon‐stationarybecausethereisapersistentlongruntrendofthecustomernumbers,populationandgrossstateproductovertime. Notrendhasbeenallowedforintheregressionanalysis. Modellingtwonon‐stationaryseriesashasbeendonecanestimateaspuriousrelationshipbecausethetwoseriesaretrendingovertime.Wherethisisthecasetherelationshipandgoodness‐of‐fitresultsrepresentedbytheR‐squaredwillbemisleading.16
15SAWaterhasindicatedthelowpowerofunitroottests.Customernumberseries,population
andGSPareclearlynon‐stationaryintheabsenceofanyallowancefortrends,asusedinACIL’sregressions,andthereisnoneedtorefertounitroottests.
16 GrangerCWJandNewboldP 1974 ,Spuriousregressionsineconometrics,JournalofEconometrics,2,pp111‐120.
20 Review of demand forecasts for SA Water
www.TheCIE.com.au
3.7 Two non-stationary series
Data source: ABS, Australia Demographic Statistics. Table 4: Estimated resident population, states and territories. Cat. No. 3101.0 and SA Water.
RECOMMENDATION
4 An alternative approach be used for developing forecasts of customer numbers. Options are set out in chapter 4.
Residentialcustomernumberandpopulationgrowth
TheannualgrowthofSAWater’scustomerbasehasexceededannualpopulationgrowthoverthepastdecade chart3.8 .Overthepastdecade,theaverageannualgrowthrateincustomernumbersisalmostdoublethepopulationgrowthrate.However,thisdifferenceingrowthrateshasbeenslowingoverthepastdecade chart3.9 .
0
5
10
15
20
25
1997-98 1999-00 2001-02 2003-04 2005-06 2007-08 2009-10
Ann
ual g
row
th r
elat
ive
to b
ase
(%)
Residential customers Population
Review of demand forecasts for SA Water 21
www.TheCIE.com.au
3.8 Annual growth in population and SA Water customers
Note: Population series based on most recent ABS data for Estimated residential population.
Data source: ABS, Australia Demographic Statistics. Table 4: Estimated resident population, states and territories. Cat. No. 3101.0 and SA Water.
3.9 Historical customer growth less population growth
Data source: The CIE.
Itisimportanttounderstandthereasoningbehindthesedifferentgrowthpatternsfordemandforecasting.Possibleexplanationsinclude:
■ SAWaterhasexpandedintonewgrowthareas;
■ theaverageoccupancyrateperhouseholdhasdeclinedinSouthAustralia;and/or
■ watercustomerspreviouslynotwithSAWaterhavemigratedtotheSAWaternetwork.
Wehavereviewedavailabledatatobetterunderstandwhatisdrivingthehigherannualgrowthrateincustomersrelativetoannualpopulationgrowth chart3.10 .Fromthis,itappearsthatthegrowthincustomernumbersreflectsgrowthinunoccupieddwellingssuchasbeachhouses,secondhousesetc .Customernumbergrowthhasbeenfar
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
1997-98 1999-00 2001-02 2003-04 2005-06 2007-08 2009-10
Ann
ual g
row
th r
ate
(%)
Residential customersPopulationResidential customers (geometric mean)Population (geometric mean)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
1997-98 1999-00 2001-02 2003-04 2005-06 2007-08 2009-10
Ann
ual g
row
th r
ate
(%)
22 Review of demand forecasts for SA Water
www.TheCIE.com.au
higherthangrowthinoccupieddwellings.ThiswasthelikelyreasonputforwardbyACIL. Theadditionofseconddwellingsmayhaveimplicationsforpercustomerusagethatarenotfurtherconsideredhere.
Therecentnarrowingofthegapbetweencustomergrowthandpopulationgrowthsuggeststhatgrowthinunoccupieddwellingsmayhaveslowed.Whetherornotthiswillcontinueisbeyondthescopeofthiswork.However,itwouldappearreasonabletoexpectalongperiod,suchasthelastdecade,wouldprovidetoooptimisticaviewontheadditionalgrowthtocustomernumbersthatwouldoccurfromseconddwellingsoverthecomingregulatoryperiod.
3.10 Difference in average annual growth rates for population, dwelling and customer growth
Data source: ABS, 3101.0 Australian Demographic Statistics. Table 4: Estimated resident population, states and territories. ABS Australian Social Trends, cat. No. 4102.0, SA Department of Planning, Transport and Infrastructure. SA Water.
CommercialcustomernumbersandGSP
ThecommercialcustomernumbersmodelisbasedonaregressionofhistoricalcommercialcustomernumbersandGSP.Asnotedearlier,thereareconcernsregardingtheapproachtomodellinghistoricalgrowthincommercialcustomernumbersfortwotrendingseries.Chart3.11showsthatthereisnorealrelationshipbetweenannualgrowthofcommercialcustomersandGSP.Thatis,periodsofhigherGSPgrowthhaveonlyaveryweak andstatisticallyinsignificant relationshiptoperiodsofhighgrowthincommercialcustomernumbers.Forexample,in2007‐08GSPgrewatcloseto6percentcomparedtoagrowthinthenumberofcommercialcustomersof1percent.Inthesubsequentyear,thegrowthrateincustomernumberswasalsocloseto1.5percent,despiteGSPalsogrowingatonly1.5percent.
SAWaterhassuggestedthat:
Theadvantageofourapproachwouldbeseenif,forexample,asignificantchangeineconomicactivitywasforecast.Consideracasewherearecessionwasforecast oraboom .Ourmodelwouldprojectaslowingincustomernumbergrowth oranacceleration .Atrendbased
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Population growth(old)
Population growth(new)
Growth occupiedprivate dwellings
(ABS)
Dwelling growth (SAPlanning)
Residentialcustomer growth
Ann
ual g
row
th r
ate
(%)
2003-2011
Review of demand forecasts for SA Water 23
www.TheCIE.com.au
approachwouldmissthatslowingandthenhavegrowthslowaftertherecession oraccelerateaftertheboom.
InourviewthisispreciselythedisadvantageofSAWater’sapproach.Historysuggeststhatwhenthereisaboomorrecessionthisdoesnotimpactoncustomernumbergrowth.Hencetoallowforthisisforecastsisnotreasonable.
3.11 Annual growth rate of commercial customers and GSP
Data source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water.
Estimatingthecoefficientsforaverageconsumptionpercustomermodel
Climate
Theconsumptionmodelsforthethreesectorsestimatedtheimpactoftemperatureonwaterdemandusingcoolingdegreedays CCD18 sourcedfromtheBureauofMeteorology’sannualweatherdata.
ACILTasmanfoundastrongrelationshipbetweenwaterdemandandtemperature.However,therelationshipsbetweenwaterdemandandeitherrainfallorevaporationwerenotasstrong.
ACIL’sincorporationofCDD18asaclimatevariabletomodelaverageconsumptionpercustomerisvalid.
Priceelasticity
Theaverageconsumptionpercustomerisinfluencedbytheassumptionsregardinghowconsumersrespondtofutureprices.Thepriceelasticitiesusedtoforecastcommercialandnon‐residentialwaterusagewereestimatedasthecoefficientonthepricevariableintherespectiveannualregressionmodelsforeachsector,notedintheprevioussection.
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
1997-98 1999-00 2001-02 2003-04 2005-06 2007-08 2009-10
Ann
ual g
row
th (%
)
Commercial customers GSP
24 Review of demand forecasts for SA Water
www.TheCIE.com.au
Thepriceelasticityusedtoforecastresidentialwaterusagewasestimatedasthemidpointbetweenthecoefficientonthepricevariableestimatedintheregressionmodelandtheresidentialpriceelasticityintheliterature.ACILprovidedanexplanationforthisapproachinthemethodologyreport.Weconsiderthistobeareasonableapproachbecauseitreflectsthesensitivityofelasticitiestodifferentmodelassumptions.
Itisimportanttonotehowever,thatchangestotheunderlyingdatathathavebeenrecommendedinthisreportwouldsubstantiallychangetheestimatesofpriceelasticity.Thisisexpendedoninchapter4ofthisreport.
Waterrestrictions
ACILestimatedtherelationshipbetweenlevel1,2and3historicalwaterrestrictionsonwaterdemandinallthreesectors.Inallcasestherewasastrongrelationshipbetweenwaterdemandandwaterrestrictions.Theestimatedcoefficientsforwaterrestrictionswereusedtoestimateaverageconsumptionperconnectionforresidential,commercialandnon‐residentialcustomers.
Thisapproachtoestimatetheinfluenceofwaterrestrictionsonwaterdemandisvalid.WhilewedonotsuggestchangestoACIL’sapproach,wenotethatthestrongandsignificantimpactofwaterrestrictionsonconsumptionamongstcommercialandothernon‐residentialcustomersisunusual.Commercialandothernon‐residentialcustomersarenotexpectedtohaveasignificantcomponentofoutdoorwaterusage theexceptionbeingmunicipalparklands andwouldnotbeexpectedtobeimpactedbywaterrestrictionstothesameextentasresidentialcustomers.OvertheforthcomingregulatoryperiodwewouldrecommendthatSAWaterconductfurtheranalysistobetterunderstandtheusagepatternsofthesecustomerclasses.Thiswouldenablefurthertestingofthepotentialimpactofwaterrestrictionsonthesecustomers.
Interceptincommercialmodel
InestimatingtheaveragecommercialconsumptionmodelACILsettheintercepttozero.
ACILdonotprovideevidencetoexplainthisassumptioninthereport.Informationhassubsequentlybeenprovidedtoindicatethattheinterceptwasdroppedbecauseitwasnotstatisticallysignificant.
Thisisarestrictiveassumptionwhichisgenerallyusedonlyinrareoccasionsrequiringthefittedlinetopassthroughtheorigin.Someregressionstatistics suchasR2 canbeinterpretedonlywherethereisaconstant ornon‐zerointercept inthemodel.Giventhesubstantialimpactonparametervaluesfromrestrictingtheconstanttobezero,werecommendthatthisassumptionberelaxed.Incorporatingthisamendmentaltersthecoefficientsonthekeydriversofcommercialdemand GSP,priceofwater,temperatureandwaterrestrictions table3.12 .17
17Resultsintable3.12arefromamendmenttointerceptinthecommercialmodel,holdingall
elseconstant.
Review of demand forecasts for SA Water 25
www.TheCIE.com.au
Notethatwhenaninterceptisallowedforinthemodelandtheadjustmentismadeto2009‐10data,wefindthatanumberofthedrivervariablesbecomestatisticallyinsignificant.Thisisnotthecasewhentherestrictionsareallowedtobepartlylevel1andpartlylevel3for2010‐11.
3.12 Commercial average consumption model
Independent variables Commercial average consumption model
Intercept set to origin Intercept not set to origin
Intercept 0 2.48
GSP 0.47 0.26
Cooling degree days (CDD 18) 0.12 0.10
Real price commercial -0.37 -0.33
Level 1 -0.11 -0.08
Level 2 -0.15 -0.11
Level 3 -0.25 -0.20
Adjusted R Square 0.91 0.87
Source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
RECOMMENDATION
5 The restriction that the constant is zero for the commercial usage model be removed.
Excludedvariables
ThereareadditionaldriversthatmayinfluencewaterdemandthathavenotbeenincludedinACIL’sdemandforecast.Theseinclude:
■ income;
■ householdsizeandcomposition;
■ typeofdwelling detached,units,townhouses ;and
■ quantityanduseofrecycledwater.
Exclusionofsomedriverscanresultinomittedvariablebias.Thiscanleadtoestimatesofotherparameters,suchaspriceelasticities,beinghigherorlowerthantheirtruelevelbecauseofrelationshipsbetweenomittedvariablesandincludedvariables.
Wehavetestedwhetherincomewasasignificantdriverofusage,throughusinggrossstateproductasaproxyforincome.Thiswasnotsignificantinthemodelandmayhavebeenexcludedforthisreason.
Inclusionofotherdriversatthehouseholdlevelwouldgenerallyrequiremoredetailedanalysisthanwasundertaken.Thedriversmentionedabovemayleadtolongertermdeviationsinwateruseperdwelling.Assuch,themodelmayfinditdifficulttoaccommodatestructuralchangesrelatingtodwellingcomposition,waterrecyclingandhouseholdcomposition.ItisnotpossibletoknowtheimpactthatthiscouldhaveontheforecastsintheabsenceofanunderstandingoftheimportanceofthesefactorsinSouthAustralianwaterdemand.
26 Review of demand forecasts for SA Water
www.TheCIE.com.au
Itislikelythattheimpactsinthe3yearsoftheregulatoryperiodwillberelativelysmall,althoughtheycouldleadtopersistenterrorsinforecasting.Giventhis,weconsiderACIL’sapproachtobereasonable.Thatsaid,theremaybescopetoincludeadditionalvariablesinfuturedeterminationswhereitisfoundthattheseareimportant.
Themodellingconducteddoesnotincludeanyconsiderationoflagsinconsumptionandlagsforrestrictions.Weconsiderthistobereasonableinlightoftheavailabledata.However,inclusionoflagsislikelytobecomemoreimportantinfuturemodellinginordertoconsidertheextenttowhichbouncebackofdemandhasoccurred.Amodelthatdoesnotincludelagswillleadtobiasedestimatesoftheimpactofrestrictionsandwaterwiserulesifthereisasubstantialpartofthedatainthepost‐restrictionsperiod.
Review of demand forecasts for SA Water 27
www.TheCIE.com.au
4 Projectingforwarddemand
IntheprevioussectionwediscussedACIL’sapproachtousinghistoricaldatatoestimatethecoefficientsforeachoftheunderlyingdriversofcustomernumbersandaverageconsumptionpercustomer.Driversincludedprice,GSP,weatherandwaterrestrictions.Thecoefficientsestimatetherelationshipsbetweenkeydriversandbothcustomernumbersandaverageconsumptionperconnection.
ACILapplytheestimatedcoefficientstoprojectionsofthekeydrivers e.g.GSPandclimatedata toforecastcustomernumbersandaverageconsumptionpercustomer.InthissectionwereviewACIL’sapproachtoapplyingthisinformationtodevelopfuturedemandforecasts.
Forecastcustomernumbers
Forecastappliedaschangeratherthanappliedtothelevel
Therearetwoalternativemethodstoapplyestimatedcoefficientstoprojecteddriverstoforecastcustomernumbersandaverageconsumptionperconnection.Thefirstistoapplytheestimatedcoefficientstothe‘level’oftherelevantvariables,forexample,therealpriceofwaterinagivenyear,togiveanexpectedlevelofwaterconsumption.Thesecondistoapplytheestimatedcoefficientstothe‘change’oftherelevantvariable,forexample,thechangeintherealpriceofwaterrelativetothepreviousyear,toderiveanexpectedchangeinwaterconsumption.Alevelapproachwillperformbetterwheredatatendstorevertquicklytosomemeanlevel intechnicaltermsitisstationary andwherethereislittlelikelihoodofstructuralchange.Achangeapproachwillperformbetterwheredatadoesnotreverttosomemeanlevel isnon‐stationary andwherestructuralchangeismorelikely.
ACILusesalevelapproachforallitsanalysis.Thisisconsideredtobeareasonableapproachforforecastsoftheusagepercustomer,althoughthereisariskthatthereisstructuralchangenotbeingfactoredin suchasrelatedtopostwater‐restrictionconsumptionbehaviour .Forcustomernumbersthisapproachisnotconsideredtobereasonable,asdiscussedbelow.
Alevelsapproachtocustomernumberscanbebestunderstoodthroughasimpleexample,asfollows.
■ Thereare100customersin2012
■ Amodelindicatesthatonaveragea1percenthigherlevelofpopulationisassociatedwith1percenthighernumberofcustomers.
28 Review of demand forecasts for SA Water
www.TheCIE.com.au
■ Thelevelmodelhasaforecastof105customersfor2012,whichishigherthanactual.
■ Populationgrowthin2013isexpectedtobe1percent.
Undertheaboveexample,alevelmodelwouldprojectthatcustomernumbersin2013wouldbe105*1.01 106customers.Thatis,itwillbebasedofftheprojectionfor2012.Incomparison,applyingasachangewouldsuggestcustomernumbersof100*1.01 101customersfor2013andisbasedoffactualcustomernumbersfor2012.
ThisisexactlywhathappensinACIL’sprojections chart4.1andchart4.2 .
Intechnicalterms,thelevelmodelincludestheerrorinthemostrecentforecastintoprojections.Thisisokayforstationaryseriesbecausetheerrorinoneyeardoesnotimpact orimpactsless onfutureyears.Itdoesnotmakesensefornon‐stationaryseriesbecauseerrorsarepermanent.
Thetypicalapproachtoforecastingfornon‐stationaryseriesistoapplygrowthprojectionstothemostrecentyear. Thismethodisalsoillustratedinchart4.1and4.2asacontinuationofthehistoricalseries. ThisistheapproachthatwerecommendapplyingforSAWater’scustomernumbers.
RECOMMENDATION
6 The methodology used to forecast residential and commercial customer numbers be amended to apply growth to the most recent data on customer numbers rather than forecasting the level of customer numbers.
4.1 Forecast of residential customers
Data source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
620
625
630
635
640
645
650
2009-10 2010-11 2011-12
Cus
tom
ers
(tho
usan
ds)
Historical
ACIL
ACIL applied as achange
Review of demand forecasts for SA Water 29
www.TheCIE.com.au
4.2 Forecast of commercial customers
Data source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
Alternativeapproaches
Asnotedintheprevioussectionwehaveconcernsregardingtheapproachtoestimatingthekeydriversofcustomernumbersbyregressingtwonon‐stationaryseries.Giventhis,webelievethatanalternativemethodforgeneratingestimatesoffuturecustomernumbers,residentialandcommercial,shouldbeused.
Itisbeyondthescopeofthisprojecttoconsideralloptionsforalternativeforecastingapproaches.Weconsiderthatapproachesthatwouldbereasonablecouldinclude:
■ applyingatrendinhistoricalgrowthincustomernumbers,sayoverthepast5years;or
■ applyingABS’sforecastedpopulationgrowthwithanadjustmentfactortoaccountforthedifferencebetweenannualcustomergrowthandpopulationgrowthoverthepastfiveyears.
Ofthesethesecondisconsideredpreferablebecauseitincorporatesinformationonpopulationprojections.Weprovideestimatesofcustomernumbersanddemandusingthisapproach.
Inthecaseofcommercialcustomers,asnotedearliertheredoesnotappeartobearelationshipbetweenGSPgrowthandgrowthincommercialcustomernumbers.Oneapproachthatwewouldconsidertobereasonablewouldbetouseanaveragegrowthforanhistoricalperiod,suchasthelast5years.Weshowprojectionsusingthisapproachinchapter5.
GrossStateProductforecasts
SubsequenttoACIL’sanalysis,theGSPforecastsproducedbytheSouthAustraliangovernmentwererevisedfortheperiodbetween2011‐12and2015‐16 table4.3 .
26.4
26.5
26.6
26.7
26.8
26.9
27.0
27.1
27.2
27.3
27.4
2009-10 2010-11 2011-12
Cus
tom
ers
(tho
usan
ds)
Historical
ACIL
ACIL applied as achange
30 Review of demand forecasts for SA Water
www.TheCIE.com.au
4.3 Revised Gross State Production forecasts
Financial year Original GSP forecast used in ACIL’s modelling a
Revised GSP forecast b
% %
2011-12 2.25 1.50
2012-13 3.00 2.75
2013-14 3.25 3.00
2014-15 3.00 3.00
2015-16 2.45 3.25
a Sourced from South Australian Government 2011-12 Mid-Year Budget Review. b Sourced from South Australian Government, 2012, 2012-13 Budget Statement, http://www.statebudget.sa.gov.au/budget_papers.html. c Forecast for 2015-16 was not included in original GSP forecast, ACIL estimated 2.45 based on the historical average.
ACIL’sanalysisusestheGSPforecastsforforecastingcommercialcustomernumbersandforecastingaveragecommercialconsumptionpercustomerandannualnon‐residentialconsumption.Asnotedearlier,theCIEdoesnotsupportACIL’smodellingtoforecastcommercialcustomernumbers.
TheupdatedGSPforecastsshouldbeincorporatedintoACIL’smodeltoforecastaverageconsumptionpercommercialcustomerandtotalannualnon‐residentialconsumption.IncorporatingtheupdatedGSPforecastsdecreasesthetotalcommercialandnon‐residentialconsumptionbilledovertheforecastingperiodbyapproximately0.4percentrelativetoACIL’sforecast.
RECOMMENDATION
7 Updated GSP forecasts should be incorporated into the demand forecasting model used to generate average consumption per commercial customer and total annual non-residential consumption.
Climate
ACILforecastedcoolingdegreedays CDD18 basedontheannualmedianCDD18overtheperiod1977‐78to2010‐11.TheforecastedannualCDD18was682daysperyear.TheforecastedCDD18wasappliedintheresidentialandcommercialaverageconsumptionperconnectionmodelsandthenon‐residentialtotalannualconsumptionmodel.
Intheabsenceofotherclimateforecasts,ACIL’sapproachtoforecastCCD18isvalid.Weconsiderthatthemeanshouldbeusedinsteadofthemedian,astheforecastsareseekingtomeasureexpecteddemand,notmediandemand.Thisisbecausearegulatorisseekingtoensurethattheexpectedrevenuesofabusinessmatchitsexpectedcosts,ratherthanthatitsmedianrevenuesmatchitsexpectedcost.However,thedifferencebetweenthemedianandmeanforCDD18wasnotmaterialenoughtowarrantamendment.
Review of demand forecasts for SA Water 31
www.TheCIE.com.au
Itisinterestingandsurprisingthatrainfalldidnotappearassignificantintheforecastingmodel.Itmaybepossiblethatvariationsinclimaticvariablesortheirinteractionmightshowaninfluencefromrainfall.
Timeinwaterrestrictions
Thefutureprojectionsofaverageconsumptionpercustomerassumethatlevel1waterrestrictionsremainedinplaceduringtheforecastperiod.Itisdifficulttoforecastthelikelihoodofwaterrestrictions,however,itisimportanttoassessthesensitivityoftheforecasttokeyassumptions.Forinstance,ifwaterrestrictionsdonotapplyintheforecastperiodtheforecastedtotalwaterconsumptionwouldincreasebyapproximately12percent.Conversely,iflevel3waterrestrictionsapplytheforecastedtotalwaterconsumptionwoulddecreasebyapproximately12.5percent.
Notethatdemanddoesnotbouncebacktotheexpectedlevelofconsumptionatlevel1restrictionspriortolevel3restrictionsbeinginplacethentheimpactsofthetimeindifferentrestrictionperiodswouldbesmaller.
TheCIEbelievesthatACIL’sassumptionregardingthetimeinrestrictionsisreasonable,giventherelativelyhighstoragelevels,theoperationoftheAdelaideDesalinationPlantandtherecentinvestmentsinrecycledwaterfacilities.
Priceelasticities
Adjustingmodelledresidentialpriceelasticity
ACILhasassumedthatthepriceelasticityusedtoforecastresidentialwaterusagewasthemidpointbetween‐0.18 estimatedfromtheliterature and‐0.38 estimatedcoefficientfromACIL’sresidentialannualregressionmodel .
TherecanbegoodreasonstochangemodelparametersasACILhasdone,asstatisticalanalysiswillalwaysrequireadegreeofjudgement.Adjustmentshavetobedonecarefullyinordertoensurethatthemodelremainsrobust.Weconsiderthatadjustmentstoparameterscouldbemade,iftherearegoodreasonsandthemodelisseekingtoapplygrowthratestoexistingusagefigures.InthecaseofACIL’sforecastingmethod,whichmodelsthelevelofwaterusage,weconsiderthatmodelparametersshouldnotbeadjusted.
Adjustmentstothemodel’sparametersinalevelmodelchangethefitofthemodelhistoricallyandmaylockinadifferentlevelofconsumption.Forexample,inchart4.4weshowtheperformanceoftheACILmodelforresidentialconsumptionpercustomeranditsperformanceoncetheadjustmenthasbeenmadetothepriceelasticity.Byadjustingthepriceelasticityfigure,themodelsignificantlyoverestimatesrecentconsumption.Weunderstandthatthisisnottheintentionoftheadjustmentmade,withACILnotingthattherearefactorsnotincludedinthemodelthathavelikelymeantthatthehistoricalrelationshipbetweenpriceandconsumptionisoverstated.ACIL’sintentionistoapplytheadjustmentonlytofuturepricechangesratherthantoaltertheimpactofpastpricechanges.Thiswouldhavetobedoneusingtwoelasticity
32 Review of demand forecasts for SA Water
www.TheCIE.com.au
parametersonetotheexistingpriceandonetosubsequentpricechangesorthroughusingamodelofthechangeinconsumption.
4.4 Review of adjustment to residential price elasticity
Data source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model and actual by the CIE.
ACIL’soriginalforecastfortotalwaterconsumptionin2011‐12,usinganadjustedpriceelasticityof‐0.28andadjustedtoreflectactualweatheroutcomesforthisyear,was188.0GL.Thisis6.2percentaboveactualwaterconsumedin2011‐12.TheforecastusinganunadjustedresidentialpriceelasticityinACIL’soriginalmodel,providesacloserfit,2.4percentaboveactualwaterconsumedin2011‐12 table4.5 .Whentheadjustmenttotheelasticityisremovedalargerreductioninwaterusageisimpliedfor2012‐13,reflectingthepriceincreaseinthatyear.
4.5 Total water consumption forecast with adjusted and unadjusted price elasticities
Year ACIL’s original model
Adjusted price elasticitya Unadjusted price elasticity
2011-12 actual 177.0 177.0
2011-12 forecast 188.0 181.2
deviation from actual in 2011-12 6.2% 2.4%
2012-13 176.3 167.7
2013-14 178.9 170.2
2014-15 181.4 172.6
2015-16 183.8 174.9
a ACIL’s original forecast for total water consumption
Note: ACIL’s original forecast for 2011 12 has been adjusted for weather.
Source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
RECOMMENDATION
4.9
5.0
5.1
5.2
5.3
5.4
5.5
5.6
5.7
1996-97 1998-99 2000-01 2002-03 2004-05 2006-07 2008-09 2010-11
Res
iden
tial w
ater
con
sum
ptio
n (n
atur
al lo
g, k
L/cu
stom
er)
Actual
Fitted (unadjusted)
Fitted (adjusted)
Review of demand forecasts for SA Water 33
www.TheCIE.com.au
8 The unadjusted price elasticity, as estimated in the residential consumption model, should be used to forecast average residential water consumption per connection.
Revisiontopriceelasticities
Thechangestohistoricaldataandmethodimpactonthemeasuredpriceelasticities.Thehistoricalchangestodataandmethodrecommendedare:
■ updateofABShistoricalpopulationdataandamendmentofACIL’sadjustmentforquarterlybillingin2009‐10;
■ inclusionofbothlevel1and3waterrestrictionsinthehistoricaldatafor2010‐11;and
■ allowingthecommercialaverageconsumptionmodeltoestimatetheinterceptratherthanapplyingACIL’sassumptionofsettingtheintercepttozero.
Thechangetothepriceelasticitiesresultingfromtheseamendmentsaredetailedintable4.6.Thechangesinestimatedelasticitiesaresubstantialandshouldbereflectedindemandprojections.WenotethattheremovaloftheadjustmenttotheresidentialpriceelasticityandtheotheradjustmentsabovegenerateapriceelasticityclosetothatoriginallyusedbyACIL.
4.6 Original and revised price coefficients from annual regression models
Sector Estimated coefficient on price
Original estimates in
ACIL’s model
Amendment to ABS population
data and quarterly billing
adjustment
Amendment to water restriction data in 2010-11
Amendment to commercial
model to include
intercept
Residential (unadjusted price coefficient) -0.38 -0.40 -0.29 -0.29
Residential (adjusted price coefficient) -0.28 -0.29 -0.23 -0.23
Commercial -0.37 -0.39 -0.29 -0.28 a
Non-residential -0.32 -0.26 -0.23 -0.23
a Amendment to commercial model to include an intercept only changes price coefficient in commercial sector.
Source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
BouncebackACILTasmanhasassumed100percentbouncebackfromlevel3waterrestrictionsovertheforecastperiodtolevel1waterrestrictions,whichareconsideredtobesimilartowaterwiserestrictionsnowinplace.However,indiscussionswithSAWateritindicatedthat100percentbouncebackdoesnotappeartohaveoccurredsincelevel3waterrestrictionswerereplacedwithwaterwisemeasuresandsuggestedthatmanyofthechangesinwaterconsumptionarisingfromlevel3restrictionshavebeenlockedintosomeextent.
Itisdifficulttodirectlyobservetheextentofany‘bounceback’inconsumptionasclimateconditionsandotherexternalfactorsalsovaryandthetimeperiodsincelevel3restrictionswereremovedhasbeenshort.Deloittereviewedexperiencesofbounceback
34 Review of demand forecasts for SA Water
www.TheCIE.com.au
inwaterusageacrossAustralianjurisdictions,18howeverthelimitedtimeseriesdataavailabledidnotprovidecleartrendsofbouncebackandtheanalysisconducteddidnotallowforconsumptiontograduallychangeaftertheremovalofrestrictions.Theextentofbouncebackinotherjurisdictionshasbeen:
■ SydneyWater—recentsubmissiontotheProductivityCommissionSydneyWaternotedtherewasnoobservedbouncebackoverthefirstthreemonthsof2010‐11,followingreplacementoflevel3restrictionswithwaterwiserulesinJune2009.19
■ Victorianwaterutilities—Victorianurbanwaterutilitieshaveassumedsomelevelofbouncebackbetween70percentand90percentofpre‐restrictionlevels.20Atthisstage,wearenotawareofanyanalysisthathasbeenundertakentoconfirmtherobustnessoftheVictorianwaterutilities’assumptions.
■ ActewAGL—soughttotesttheextentofanybounceback,althoughithasbeenlimitedbyinsufficientobservedhistoricaldatafollowingtheliftingofwaterrestrictions.
BasedondiscussionswithSAWaterregardingbouncebackandconsumptionoutcomesin2011‐12,weconsiderthattheassumptionofimmediatebouncebacktousageatlevel1restrictionsisunreasonableandthatthisadjustmentwilltakesometime.
ACIL’stotalconsumptionforecastassumedfullbouncebacktowaterwiseconsumptionlevelsinthefirstyear.Usingthisbouncebackassumption,ACIL’sforecastwas188GL,or6.2percentaboveactualconsumptiondatain201112,afteradjustingforweather.Giventheconsumptionoutcomefor2011‐12,itappearslikelythatconsumptionbouncebackhasbeenabout25percenttowardsthelevelsexpectedforwaterwiserulestable4.7 .
4.7 Deviation of forecasts from actual consumption data in 2011-12
2011-12 Total water consumption Deviation to actual
GL Per cent
Actual 177.0
ACIL forecast 188.0 6.2
25 per cent bounceback 175.9 -0.6
Note: Total water consumption forecasts in 2011 12 were adjusted for weather.
Source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
18 Deloitte,2012,Measuring’bounceback’inwaterusage.PresentationtotheESCDemandForecastingWorkshop.http://www.esc.vic.gov.au/getattachment/8d0e817a‐ee72‐4c79‐829d‐d20b4491c95f/Deloitte‐Presentation‐Demand‐forecasts‐workshop‐Wa.pdf
19SydneyWater2010,Australia’sUrbanWatersector,AsubmissiontotheProductivityCommission.November,p.34.
20PricewaterhouseCoopers PwC 2009,WaterPriceReview2009:Demand,ReportforEssentialServicesCommission,p.13http://www.esc.vic.gov.au/NR/rdonlyres/E19898BE‐3012‐4619‐999E‐880367FD0615/0/RPTPWCWaterBusinessesdemandreport120090417.pdfAccessed2011.
Review of demand forecasts for SA Water 35
www.TheCIE.com.au
Wehavemodelledthreebouncebackscenarioswhichvarytheextentoftotalbouncebackandthelengthofthebouncebackperiod.The2011‐12totalwaterconsumptiondatarecentlyreleasedwasusedasabenchmarktoderivethethreebouncebackscenarios,settinganannualbouncebackof25percentineachyearofthebouncebackperiod.Thethreebouncebackscenariosare:
■ 50percentover2years chart4.8 :duringtheregulatoryperiod,CIE’sforecastoftotalwaterconsumptionisapproximately4percentlowerineachyearthanACIL’soriginalforecast.
■ 75percentover3years chart4.9 :duringtheregulatoryperiod,CIE’sforecastoftotalwaterconsumptionisapproximately0.4percentlowerineachyearthanACIL’soriginalforecast.
■ 100percentover4years chart4.10 :duringtheregulatoryperiod,CIE’sforecastoftotalwaterconsumptionisbetween1.5percenthigherinthefirstyear,increasingto3.6percenthigherattheendoftheregulatoryperiodcomparedtoACIL’soriginalforecast.
4.8 Bounceback of 50 per cent over 2 years
Data source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
150
160
170
180
190
200
210
220
230
240
250
1996-97 1999-00 2002-03 2005-06 2008-09 2011-12 2014-15
Tota
l wat
er c
onsu
mpt
ion
(GL)
Historical
ACIL forecast
CIE forecast
36 Review of demand forecasts for SA Water
www.TheCIE.com.au
4.9 Bounceback of 75 per cent over 3 years
Data source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
4.10 Bounceback of 100 per cent over 4 years
Data source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
Intheabsenceofadditionaldataonbounceback,thethreebouncebackscenariosprovidearangeforforecastsoftotalwaterconsumption.
Choosingbetweenbouncebackscenariosisdifficult,giventhattherehasnotbeensubstantialexperiencewithbouncebackinSouthAustraliatodrawon.Itmaybepossibletobetterunderstandbouncebackthroughmodellingofmonthlydatatoseeiftherehasbeenanyconvergencebetweentheactualfiguresandamodelwith100percentbouncebackoverthetwoyearssincelevel3restrictionswereremoved.Itmayalsobepossibletouseotherjurisdictionsasanexample,wherethesejurisdictionshavehadalongertimeoutofrestrictions.HistoricaldataforHunterWatermightbeoneoptionforthis.However,itisnotclearthatalljurisdictionswillbethesame.Theamountofbouncebackmayalsodependonotherfactorssuchasthedurationofrestrictions.
150
160
170
180
190
200
210
220
230
240
250
1996-97 1999-00 2002-03 2005-06 2008-09 2011-12 2014-15
Tota
l wat
er c
onsu
mpt
ion
(GL)
Historical
ACIL forecast
CIE forecast
150
160
170
180
190
200
210
220
230
240
250
1996-97 1999-00 2002-03 2005-06 2008-09 2011-12 2014-15
Tota
l wat
er c
onsu
mpt
ion
(GL)
Historical
ACIL forecast
CIE forecast
Review of demand forecasts for SA Water 37
www.TheCIE.com.au
Consumptiondataavailableforthemostrecenttwoyearssincewaterrestrictionswerereplacedsuggeststhattherehasbeensomelevelofbounceback,likelyaround25percentperyear.Itisnotclearwhetherconsumptionwillbouncebackinfulltothelevelexpectedbymodellinggiventoday’spricesandlevel1restrictions.
RECOMMENDATION
9 An adjustment should be made to the projections to allow for water demand to recover gradually to expected consumption which also reflects bounceback of water consumption experienced in 2011-12.
38 Review of demand forecasts for SA Water
www.TheCIE.com.au
5 Revisedforecasts
InpreviouschapterswerecommendedupdatesandamendmentstothedataandmodellingspecificationusedinACIL’sdemandforecasts,includingdiscussionofthechangetotheforecastofamendingeachinturn,holdingallelseconstant.
InthischapterwestepthroughthesuggestedupdatesandamendmentstodataandmodellingandreportthecumulativechangetothetotalforecastedwaterusagerelativetoACIL’soriginalforecastoftotalwaterconsumptionbilledfortheforecastperiodbetween2013‐14and2015‐16.
Ourrevisedforecastsforcustomernumbersandconsumptionbilledovertheregulatoryperiodareprovidedintable5.1and5.2.Giventheuncertaintysurroundingtheoccurrenceofbouncebackofwaterconsumption,wehavemodelledtherevisedforecastoftotalconsumptionbilledovertheforecastperiodforthreebouncebackscenarios.Forthethreescenarios,theCIE’srevisedforecastis:
■ Bouncebackscenario1:demandapproximately4.1percentlowerthanACIL’soriginalforecastwith50percentbouncebackover2years.
■ Bouncebackscenario2:demandapproximately0.4percentlowerthanACIL’soriginalforecastwith75percentbouncebackover3years;and
■ Bouncebackscenario3:demandapproximately2.9percenthigherthanACIL’soriginalforecastwith100percentbouncebackover4years table5.2 .
Annualforecastsoftotal,residential,non‐residentialandcommercialwaterconsumptionbilledforthethreebouncebackscenariosbetween2011‐12and2015‐16areprovidedisAppendixA.1.
5.1 Original and amended forecast of customer numbers
Year Residential customers Commercial customers
ACIL CIE ACIL CIE
2011-12 646 143 641 149 27 052 27 364
2012-13 655 659 650 660 27 365 27 675
2013-14 665 173 660 252 27 715 27 990
2014-15 674 689 669 926 28 048 28 309
2015-16 684 207 679 683 28 329 28 631
Source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
Review of demand forecasts for SA Water 39
www.TheCIE.com.au
5.2 Original and amended forecast of consumption billed over regulatory period
Total consumption billed (ML) over regulatory period
ACIL’s original forecast
CIE Amended forecast
Bounceback assumption 100% over 1 year 50% over 2 years 75% over 3 years 100% over 4 years
Residential 367 592 338 906 355 993 370 958
Commercial 28 716 28 495 29 676 30 703
Non residential 147 707 153 971 156 116 157 931
Total 544 014 521 373 541 785 559 592
Source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
Theextentoftheamendmentstoforecastedtotal,residential,commercialandnon‐residentialwaterconsumptionbilledisillustratedinchart5.3,5.4,5.5and5.6.
■ Populationdataandquarterlybilling—updatetheABShistoricalpopulationdataandamendtheestimationofwaterconsumptionbilledin2009‐10.
■ Disaggregationofhistoricalwaterrestrictions—revisethehistoricalwaterrestrictiondatatoincludelevel1andlevel3waterrestrictionsin2010‐11.
■ Forecastcustomernumbers—theprojectionofresidentialcustomernumbersisbasedontheABSpopulationforecastadaptedtoincludethefiveyearaveragehistoricaldifferenceingrowthratesofcustomersandpopulation chart3.8 .Commercialcustomernumbersareprojectedbasedonanaveragehistoricalgrowthincustomernumbersoverthemostrecentfiveyears.
■ Incorporateinterceptincommercialmodel—theassumptionthattheinterceptissettozerointhecommercialmodelwasrelaxed.
■ Unadjustedresidentialpriceelasticity—themodelledresidentialpriceelasticityisusedtoforecastaverageconsumptionperconnectioninsteadofanadjustedpriceelasticityasusedbyACIL.
■ UpdateGSPindependentforecast—updatedGSPindependentforecastswereincorporatedintomodel.
■ Bounceback—wehavemodelledthreebouncebackscenarios,50percentover2years,75percentover3yearsand100percentover4years.
Adjustingthehistoricalwaterrestrictiondatatoincorporatebothlevel1and3waterrestrictionsin2010‐11causedthelargestchangetotheforecast.Thisisbecausethisamendmentsubstantiallychangedtheestimatedcoefficientonthepricevariableforresidential,commercialandnon‐residentialdemand.Forexample,thecoefficientonthepricevariableintheresidentialmodelchangedfrom‐0.38to‐0.27 table3.4 ,whichresultedinthepriceelasticityusedintheresidentialmodelchangingfrom‐0.28to‐0.23.21Theamendmentmakesdemandrelativelymoreinelasticandthereforelessresponsivetopriceincreasesovertheforecastperiod.
21BecauseoftheACIL’sapproachtoestimatetheresidentialpriceelasticityasthemidpoint
between‐0.18andtheestimatedcoefficientonthepricevariable,theadjustedpriceelasticitywhenthehistoricalwaterrestrictiondatawasamendedwas‐0.23.
40 Review of demand forecasts for SA Water
www.TheCIE.com.au
Residentialwaterconsumptiondecreasedwhentheunadjustedresidentialpriceelasticitywasappliedintothemodel.TheunadjustedpriceelasticityismoreelasticthantheadjustedpriceelasticityusedbyACILwhichmakesdemandmoreresponsivetotheforecastedincreaseinpricein2011‐12and2012‐13.
ForecastedwaterconsumptiondecreasesacrossallthreesectorswhenwelimitbouncebackrelativetoACIL’sassumptionoffullbounceback.Thescaleofdeclineinconsumptionisrelativetothebouncebackassumption.
5.3 Cumulative changes to total consumption billed over regulated period (GL)
Note: Cumulative change presented for the three bounceback scenarios
Data source: The CIE.
5.4 Cumulative changes to residential consumption billed over regulated period (GL)
Note: Cumulative change presented for the three bounceback scenarios
Data source: The CIE.
490
500
510
520
530
540
550
560
570
580
590
AC
IL's
orig
inal
fore
cast
Upd
ate
AB
SP
opul
atio
n an
dQ
ua
rte
rly B
illin
g
His
toric
al w
ater
rest
rictio
ns
Fo
reca
stcu
stom
ernu
mb
ers
Inte
rcep
t in
com
mer
cial
mod
el
Una
dju
sted
resi
dent
ial p
rice
elas
ticity
GS
P fo
reca
st
Bou
nceb
ack
(50%
/2yr
s)
Bou
nceb
ack
(75%
/3yr
s)
Bou
nceb
ack
(100
%/4
yrs)
Tota
l con
sum
ptio
n bi
lled
(GL)
310
320
330
340
350
360
370
380
390
400
AC
IL's
orig
inal
fore
cast
Upd
ate
AB
SP
opul
atio
n an
dQ
ua
rte
rly B
illin
g
His
toric
al w
ater
rest
rictio
ns
Fo
reca
stcu
stom
ernu
mb
ers
Inte
rcep
t in
com
mer
cial
mod
el
Una
dju
sted
resi
dent
ial p
rice
elas
ticity
GS
P fo
reca
st
Bou
nceb
ack
(50%
/2yr
s)
Bou
nceb
ack
(75%
/3yr
s)
Bou
nceb
ack
(100
%/4
yrs)
Tota
l con
sum
ptio
n bi
lled
(GL)
Review of demand forecasts for SA Water 41
www.TheCIE.com.au
5.5 Cumulative changes to commercial consumption billed over regulated period (GL)
Note: Cumulative change presented for the three bounceback scenarios
Data source: The CIE.
5.6 Cumulative changes to non-residential consumption billed over regulated period (GL)
Note: Cumulative change presented for the three bounceback scenarios
Data source: The CIE.
26
27
28
29
30
31
32
AC
IL's
orig
inal
fore
cast
Upd
ate
AB
SP
opul
atio
n an
dQ
ua
rte
rly B
illin
g
His
toric
al w
ater
rest
rictio
ns
Fo
reca
st c
usto
mer
num
ber
s
Inte
rcep
t in
com
mer
cial
mod
el
Una
dju
sted
resi
dent
ial p
rice
elas
ticity
GS
P fo
reca
st
Bou
nceb
ack
(50%
/2yr
s)
Bou
nceb
ack
(75%
/3yr
s)
Bou
nceb
ack
(100
%/4
yrs)
Tota
l con
sum
ptio
n bi
lled
(GL)
142
144
146
148
150
152
154
156
158
160
AC
IL's
orig
inal
fore
cast
Upd
ate
AB
SP
opul
atio
n an
dQ
ua
rte
rly B
illin
g
His
toric
al w
ater
rest
rictio
ns
Fo
reca
st c
ust
om
ernu
mb
ers
Inte
rcep
t in
com
mer
cial
mod
el
Una
dju
sted
resi
dent
ial p
rice
elas
ticity
GS
P fo
reca
st
Bou
nceb
ack
(50%
/2yr
s)
Bou
nceb
ack
(75%
/3yr
s)
Bou
nceb
ack
(100
%/4
yrs)
Tota
l con
sum
ptio
n bi
lled
(GL)
42 Review of demand forecasts for SA Water
www.TheCIE.com.au
A Forecastofannualwaterconsumptionbilled
A.1 Forecast of annual water consumption billed (GL)
Year ACIL’s original a CIE
Bounceback scenario 100% over 1 year 50% over 2 years 75% over 3 years 100% over 4 years
Total water consumption billed (GL)
2011-12 188.0 175.9 175.9 175.9 2012-13 176.3 169.1 172.3 172.3 2013-14 178.9 171.4 178.1 181.6 2014-15 181.4 173.8 180.6 187.7 2015-16 183.8 176.2 183.1 190.3 Residential water consumption (GL)
2011-12 126.8 113.7 113.7 113.7 2012-13 119.0 109.7 112.4 112.4 2013-14 120.8 111.3 116.9 119.8 2014-15 122.5 113.0 118.7 124.6 2015-16 124.3 114.6 120.4 126.5 Non-residential water consumption (GL)
2011-12 51.5 52.9 52.9 52.9 2012-13 48.2 50.3 50.7 50.7 2013-14 48.7 50.8 51.5 51.9 2014-15 49.3 51.3 52.0 52.7 2015-16 49.7 51.9 52.6 53.3 Commercial water consumption (GL)
2011-12 9.7 9.3 9.3 9.3 2012-13 9.1 9.1 9.3 9.3 2013-14 9.3 9.3 9.7 9.9 2014-15 9.6 9.5 9.9 10.3 2015-16 9.8 9.7 10.1 10.5
a ACIL’s forecast in 2011-12 has been adjusted for weather.
Source: ACIL Tasman, 2012, SA Water’s Demand Forecasting. Prepared for SA Water. With amendments to model by the CIE.
44 Review of demand forecasts for SA Water
www.TheCIE.com.au
THE CENTRE FOR INTERNATIONAL ECONOMICS
www.TheCIE.com.au