The effects of ownership and benchmark competition: An ...

The effects of ownership and benchmark competition: An ...

Documents
20 pages
Lire
Le téléchargement nécessite un accès à la bibliothèque YouScribe
Tout savoir sur nos offres

Description

Available online at www.sciencedirect.com
International Journal of Industrial Organization
26 (2008) 186–205
www.elsevier.com/locate/econbase
The effects of ownership and benchmark competition:

An empirical analysis of U.S. water systems
⁎Scott Wallsten , Katrina Kosec
American Enterprise Institute and AEI–Brookings Joint Center, 1150 17th Street, NW Washington, DC 20036, USA
Received 16 March 2006; received in revised form 30 October 2006; accepted 3 November 2006
Available online 26 December 2006
Abstract
Should governments or private firms own water systems? Can yardstick, or benchmark, competition
effectively discipline utilities? Little empirical research sheds light on these issues. We use a panel dataset
that includes every community water system in the United States from 1997 to 2003 to test the effects of
ownership and benchmark competition on regulatory compliance. We find that ownership does not
generally affect compliance with the Safe Drinking Water Act (SDWA). Greater benchmark competition,
however, is associated with fewer SDWA violations. The results suggest that ownership, per se, may not
mattermuch,butthatcompetition–eveninitsweakerformofbenchmarkingbyconsumersorregulators–
can improve performance.
© 2006 Elsevier B.V. All rights reserved.
JEL classification: L00; L33; L50; L95
Keywords: Industrial organization; Privatization; Regulation; Competition; Water utilities
1. Introduction
Privatizing utilities remains controversial, though most studies ...

Sujets

Informations

Publié par
Nombre de visites sur la page 44
Langue English
Signaler un problème
Available online at www.sciencedirect.comInternationalJournalofIndustrialOrganization26(2008)186205www.elsevier.com/locate/econbaseTheeffectsofownershipandbenchmarkcompetition:AnempiricalanalysisofU.S.watersystemsScottWallsten,KatrinaKosecAmericanEnterpriseInstituteandAEIBrookingsJointCenter,115017thStreet,NWWashington,DC20036,USAReceived16March2006;receivedinrevisedform30October2006;accepted3November2006Availableonline26December2006AbstractShouldgovernmentsorprivatefirmsownwatersystems?Canyardstick,orbenchmark,competitioneffectivelydisciplineutilities?Littleempiricalresearchshedslightontheseissues.WeuseapaneldatasetthatincludeseverycommunitywatersystemintheUnitedStatesfrom1997to2003totesttheeffectsofownershipandbenchmarkcompetitiononregulatorycompliance.WefindthatownershipdoesnotgenerallyaffectcompliancewiththeSafeDrinkingWaterAct(SDWA).Greaterbenchmarkcompetition,however,isassociatedwithfewerSDWAviolations.Theresultssuggestthatownership,perse,maynotmattermuch,butthatcompetitioneveninitsweakerformofbenchmarkingbyconsumersorregulatorscanimproveperformance.©2006ElsevierB.V.Allrightsreserved.JELclassification:L00;L33;L50;L95Keywords:Industrialorganization;Privatization;Regulation;Competition;Waterutilities1.IntroductionPrivatizingutilitiesremainscontroversial,thoughmoststudiessuggestthatityieldspositiveoutcomes.Thosepositiveoutcomes,however,aremuchmorerobustinindustriesthatsupportrigorouscompetition,liketelecommunications.Whataboutindustriesinwhichcompetitionismoredifficulttointroduceorsustain?Isaprivatemonopolylikelytooutperformastate-ownedTheauthorsgratefullyacknowledgetheassistanceofCarolineCecot,LeeKyle,AbeSiegel,MollyWells,andthecommentsofRobertHahn,GeorgeClarke,andtwoanonymousreferees.Theauthors,however,aresolelyresponsibleforanyerrors.Correspondingauthor.Tel.:+12029692950.E-mailaddresses:scott@wallsten.net(S.Wallsten),kkosec@stanford.edu(K.Kosec).0167-7187/$-seefrontmatter©2006ElsevierB.V.Allrightsreserved.doi:10.1016/j.ijindorg.2006.11.001
S.Wallsten,K.Kosec/Int.J.ofInd.Organ.26(2008)186205781monopoly?Canyardstick,orbenchmark,competitiondisciplineprivateorpublicprovidersintheabsenceofhead-to-headcompetition?ThispaperattemptstoanswerthesequestionsinthecontextofdrinkingwatersystemsintheUnitedStates.Theidealformofownershipinnetworkindustriesremainsheavilydebated.WaterprivatizationintheUnitedKingdom,forexample,isstillahotissuenearlytwentyyearsafterMargaretThatchersoldtenregionalwaterauthoritiesandestablishedanewregulator.IntheUnitedStates,wherenearlyallutilitiesareownedoroperatedbyprivatefirms,around80%ofthepopulationreceiveswaterfrompublicly-ownedsystems.Proponentsofprivateownershiparguethatfirmscandeliverservicesmoreefficientlythancangovernments.Governmentshavediverseandoftencontradictoryobjectivesthatcanleadtoamisallocationofresources,andmayfaceincompatibleincentiveswhentheyactasbothownerandregulatorofautility.Inaddition,politiciansmayfeelpressuretoprovidejobsintheutilitytoconstituents,forexample,ratherthaninvestinginthewaternetwork.Opponentsofprivateownershipcounterthatprivatizationisdifficultandimpracticalgivenwaterandsewerageexternalitiesandthelimitedscopeforcompetition.Someobjecttotheveryideaofconsideringwateraneconomicgoodorallowingaprivatecompanytorunawatersystem.PublicCitizen(2004),anoutspokencriticofprivatization,highlightstheintensefeelingsregardingwaterprivatization,claimingthatperhapsthegreatesttheftofcommonresourcesfacinghumanityandtheplanetisthecorporatetakeoveroftheworld'swater.Whiledirectcompetitionamongpipedwatersystemsisessentiallynonexistent,benchmark,oryardstick,competitionmaybepossible.Shleifer(1985)formalizedamodelofregulationviayardstickcompetitioninwhicharegulatedfirm'spricesdependonotherfirms'costs.Benchmarkingmayalsoapplytoobservableinformationotherthanprices.Evenifconsumerscannotchoosewhosuppliestheirdrinkingwater,utilitiesmayfaceperformancepressureifconsumersorregulatorscancomparesystems.Consumersdissatisfiedwiththeirutilitiescouldexpresstheirdispleasurethroughtheirelectedrepresentatives,whileregulatorscouldusetheinformationfrombenchmarkcomparisonstodemandpriceorqualitychangeswithpotentiallydramaticeffectsonfirmsinthefuture.Despitethepassionstheissueraises,thereisscantempiricalevidenceonwhetherpublicorprivatewatersystemstendtoperformbetterandwhetherbenchmarkcompetitionactuallyimprovesoutcomes.Toourknowledge,nosystematic,empiricalanalysiscomparespublicandprivatewatersystemsintheUnitedStatesandwearenotawareofanyempiricalliteratureexploringthepotentialimpactofbenchmarkcompetitioninthewatersector.WeaddressthesegapsbyexploringU.S.waterqualityandhouseholdexpendituresonwaterbysystemownershiptypeandthedegreeofbenchmarkcompetitionfaced.Inparticular,weuseapaneldatasetthatincludeseverycommunitywatersysteminthecountryfrom1997to2003totesttheeffectsofownershipandbenchmarkcompetitiononviolationsoftheSafeDrinkingWaterAct(SDWA).Controllingforlocationfixedeffects,income,theshareofthepopulationthatisurban,ownershipconcentrationinthelocalwatermarket,systemsize,watersource,andyearfixedeffects,wefindlittledifferencebetweenpublicandprivateownershipwithregardtoSDWAviolations.Amongsmallerwatersystems,privateownershipisassociatedwithfewerviolationsofmaximumcontaminantlevelsbutmoremonitoringandreportingviolationsthanpublicsystems.Amonglargersystemsespeciallythoseservingmorethan100,000peopleprivateownershipisassociatedwithmorecontaminantbutfewermonitoringandreportingviolations.Theresultsneithersupportcriticsofprivateownershipnorsuggestthatprivateownershipissuperior.Overall,compliancewithSDWAregulationsdoesnotappeartodependmuchonsystemownership.Benchmarkcompetition,however,appearstohaveamorerobustimpact.Greater
818S.Wallsten,K.Kosec/Int.J.ofInd.Organ.26(2008)186205benchmarkcompetition,asmeasuredbytheconcentrationoftheshareofconnectionsheldbywatersystemsineachcounty,isassociatedwithbothfewercontaminantandfewermonitoringandreportingviolations.Thissuggeststhatevenaweakformofcompetitioncanhavepositiveeffectsonregulatorycompliance.Overall,ourresultsareconsistentwiththehypothesisthatpublicorprivateownershipmaynotmakemuchofadifferencewhentheoperatorislargelymonopolistic,butthateventhen,benchmarkcompetitioncanyieldtangiblebenefits.2.TheeconomicsofwatersystemsWaterandsewerageserviceprovisionentailsuniquechallenges.Fixedcostsarehigh,barrierstoentryarelarge,thepotentialforcompetitionislimited,andthesectorhasmanyandlargeexternalities.Thelong-livedfixedassetscreateincentivesforbothfirmsandgovernmentstodelayinvestmentanduserevenuesforotherpurposes.Governmentsoftenencounterpoliticalandlegalobstaclestoprivatization,andprivateinvestorsworrythatgovernmentswillexpropriatetheirassets.Whiledirectcompetitioninwaterprovisionisdifficult,indirectcompetitionismorefeasible.Suchcompetitioncouldtaketheformofprivatefirmscompetingforoperationandmanagementcontracts(competitionforthemarket)orbenchmarkcompetition,wheretheexistenceofalargenumberofwatersystemscouldallowconsumersandregulatorstocomparetheirsystemtoothersandaccordinglypressuretheirownutilitytoimproveperformance.Waterservicesmaybesubjecttoavarietyofmarketfailuresthatrequiregovernmentinvolvementorregulation.Externalities,especiallyhealth-related,areespeciallypronouncedinwaterandsewerage,complicatingprivatesectorinvolvement.1Externalitiesnecessitatemechanismsofensuringthatprices,investment,andotheroperationaldecisionsarenotdivorcedfromexternalitiesthataffectactorsotherthanfirms.Somefeaturesofwaterandsewerageinfrastructurecreateincentivesforbothgovernmentsandfirmstodelayinvestment.Muchofthesector'sassets,suchasthepipes,aresunk,havenoalternativeuses,andareverylong-lived.Becausethefixedcostscomprisealargeshareoftotalcostsandbecausethesystemcanoperateforalongtimewithoutmuchintervention,aself-financingutilitywillearnquasi-rents(Noll,2002).Bothpoliticiansandinvestorsmayfeelpressuretousethequasi-rentsforshort-termprioritiesratherthaninvestment.Consumersmaydemandlowertariffs,andgovernmentsthatownandoperatetheirwatersystemsmayfacepressuretouserevenuesforotherimmediateneeds.Privateoperators,meanwhile,mayworryaboutexpropriationandfaceinvestorpressuretoavoidreinvestingthequasi-rents.22.1.TheevidenceonprivateversuspublicownershipAnumberofstudieshaveevaluatedprivatizationinmanyindustriesaroundtheworld.MegginsonandNetter(2001)andShirleyandWalsh(2000)findincomprehensiveliteraturereviewsthatprivatizationhasgenerallyimprovedefficiency.Littleofthatresearch,however,hasfocusedon1Theextenttowhichhealthisactuallyanexternalitydependsonwhetherconsumersandtheutilitybearthefullcostsofanyhealthimpacts.Nonetheless,absentsometypeofoversightconsumershavelittlemeansofmonitoringthehealthqualityoftheirwater.2U.S.watersystemsinthe19thcenturywerelargelyprivatelyrun,thoughcitiesbegantomunicipalizethemintheearly20thcentury.TroeskenandGeddes(2001)arguethatmunicipalizationoccurredbecauseprivatefirmswereunder-investingintheirnetworks,andthatthisunderinvestmentoccurredbecausethemunicipalitiescouldnotcrediblycommittonotexpropriatingtheirresources.Theyconcludethatunderinvestmentwasarationalresponsetothefearofexpropriation.
S.Wallsten,K.Kosec/Int.J.ofInd.Organ.26(2008)186205981water.Becauseprivatizationhasbeencombinedwithcompetitioninsectorsliketelecommunica-tions,itislessclearwhethergovernmentorprivateownershipisinherentlysuperiorwhenthepotentialforcompetitionislimited.Withoutappropriateincentivesintheformofregulation,sometypeofoversight,oraviablethreatofentryinsomesphereofoperationsitisnotobviousthataprivatemonopolywouldbebetterthanapublicmonopoly.Existingresearchcomparingpublicversusprivateownershipofwaterprovisionconsistsprimarilyofcasestudies,fromwhichitisdifficulttogeneralize.Casestudiesareusuallynotchosenrandomly,butrathertendtofocusonexceptionalcases.3Forwaterservices,thosecasesoftenfocusonspectacularfailures,suchasthedisastrousprivatizationattemptinCochabamba,4Bolivia.However,atleast25,000U.S.watersystemsareprivatelyowned,andothercountrieslikeFrancerelyheavilyontheprivatesectorwithoutdisastrousconsequences,suggestingthattherelativelysmallnumberofspectacularfailuresmaynotberepresentative.5Casestudiesofwaterownershipinindustrializedcountriesareinconclusive.TheEconomist(2003)arguesthatwaterprivatizationintheUKhasbeensuccessfulwhencomparedtowatersystemsinScotland,whichwentfromprivatetopublichands.SaalandParker(2001),contrarily,findhigherpricesbutlittleproductivityimprovementfollowingtheUKprivatization.Likewise,Cowanetal.(2000)estimatethatprivatizationintheUKledtoanetwelfareloss.Mostofthefewempiricalstudiesonwaterprivatizationlackgood,consistentdataandyieldgenerallyinconclusiveresults.OneexceptionisastudybyGalianietal.(2003),whorigorouslyexploretheeffectofwaterprivatizationonchildmortalityinArgentina.Theyfindthatchildmortalityfelleightpercentinregionsthatprivatizedtheirwatersystems,andthattheeffectswerelargestinthepoorestareas.6Clarkeetal.(2004)findinahousehold-levelstudyofdevelopingcountriesthatconnectionratestopipedwaterincreasedfollowingprivatization,evenamongthepoor,butthatconnectionratesincreasedsimilarlyinareasthatdidnotprivatize.Empiricalresearchontheeffectsofprivatesectorparticipationinindustrializedcountriesfocuseslargelyonoperatingefficiency,andisinconclusive.Severalstudiesinthe1980susingfairlysmalldatasets(120observationsandfewer)foundnosignificantdifferencesbetweenpublicandprivatewatersystemefficiencyintheU.S.(Byrnesetal.,1986;FeigenbaumandTeeples,1983;TeeplesandGlyer,1987).Bhattacharyyaetal.(1995),however,foundinasampleof190publicand31privatewatersystemsthatprivate(public)operatorstendedtobemoreefficientamongsmall(large)systems.RenzettiandDupont(2003)concludedthattheliteraturefoundlittleevidenceofunambiguousimprovementsfromprivatization,butthatpublicprivatepartner-shipsmightbepromising.Whileexistingempiricalstudiesontheeffectsofownershipfocusprimarilyonoperatingefficiency,criticsalsoworrythatprivateoperatorsmaygivelessweighttopublicinterestthan3Shirley(2002)isanexception.Thiscollectionofsixwaterprivatizationcasestudieswasundertakenwithacoherentframeworkmeanttoensurethatthecases,incountrieswithvaryinginstitutionalandphysicalcharacteristics,wereevaluatedinaconsistentmanner.4Afocusonfailuresisnotsurprisingtheymayinvolvelargeamountsofmoneyandaffectlargepopulations.Atlanta,forexample,signeda20-year,$428millionmanagementcontractwithaprivateoperatorin1998,onlytocancelitafewyearslaterasboththecityandthefirmaccusedtheotherofviolatingthecontract(Brubaker,2003).5In2003privateoperatorsowned25,977communitywatersystems(EnvironmentalProtectionAgency,2003).AformerwateranalystforPublicCitizennotedthatContrarytothecritics'contentions,privateoperatorshavearespectablerecordofprovidingqualitywaterandcomplyingwithenvironmentalstandards(Tsybine,2002).6Theiranalysisisrigorous:theauthorshonetheirfindingsbyseparatingmortalitythatcanbecausedbywaterconditionsfrommortalityunrelatedtowater.Privatizationwasuncorrelatedwithmortalityfromnon-watercauses,butstronglycorrelatedwithmortalitycausedbywaterconditions.Moreover,themeasuredeffectmaybeunderestimatedsincetheirwateraccessdataunder-sampledthepoorestareas,whichappearedtoreceivethelargestbenefits.
190S.Wallsten,K.Kosec/Int.J.ofInd.Organ.26(2008)186205wouldagovernmentoperator,resultinginlowerqualitywater.Toourknowledgenolarge-scaleempiricalstudieshaveexploredtheeffectsofownershiponwaterqualityinindustrializedcountries.Inapubliclyavailablepresentation,GasteyerandVaswani(2004)usesomeoftheEnvironmentalProtectionAgency(EPA)datathatweuse,butonlywithcross-sectionaldataandcountyaverages.7TheyconcludethatprivateoperatorshaveslightlymorecontaminantviolationsoftheSDWA.Noresearchappearstohaveexploredcompetitioninwater,totheextentthatitmightexist.3.RegulationandcompetitionTheSafeDrinkingWaterAct(SDWA)of1974coversalloftheapproximately160,000publicwatersystemsintheUnitedStatesandallowstheEPAtosetmaximumacceptablecontaminantlevels,outlinewatertreatmentprocedures,andrequirewatersystemstofollowaprescribedwaterqualitymonitoringschedule.8TheEPAalsoregulatesthestates,localities,andwatersuppliersthatimplementthosestandards.9Waterratesappeartoberegulatedinconsistentlyacrossthecountry(Mann,1993).Unfortunately,toourknowledge,therearenorecentsurveysofstateandlocalwaterregulatorsorregulations.A1989survey,however,foundthatingeneralstatesregulateprivatebutnotstate-ownedutilities(BeecherandLaubach,1989).Mann(1993)notesthatmayors,citycouncils,orotherelectedorappointedrepresentativesgenerallyapproveratesofwatersystemsownedbythegovernment.Inotherwords,theevidencesuggeststhatwithsomeexceptions,bothpublicandprivatewatersystemsaresubjecttosomepricingoversightdirectlybyregulatorsinthecaseofprivatefirms,orbypoliticianswhomustapprovepricesinthecaseofgovernment-ownedutilities.Perhapsinpartbecauseofscantdata,thereislittleresearchonwaterrateregulation.Wolak(1994)modelstheinteractionsbetweenregulatorsandwaterutilitiesunderasymmetricinformationinordertouncovertheutility'scostfunction.Timmins(2002)estimatestheobjectivefunctionofregulatorsandestimatesthedeadweightlosscausedbytheconsistentunderpricingofwaterrelativetoitsmarginalcost.Inadditiontoregulatoryorpoliticalconstraintsonutilities'behavior,theymayalsofacepressurefrombenchmarkcompetition(alongthelinesofShleifer,1985).Whiletheredoesnotappeartobeanyempiricalanalysisofbenchmarkcompetition,basedlargelyoninterviews7Ouranalysisimprovesontheirsinanumberofwaysthatallowsustoexplorethedatamorerigorously.First,weuseapaneldatasetcovering19972003,whiletheyuseonlyacross-sectionfromyear2000.Second,ourcontaminantanalysisisatthelevelofthewatersystem,whiletheyusecountyaverages.Thatis,forthecontaminantanalysisweusetheownershipstatusofeachwatersystemratherthantheshareofwatersystemsinacountythatarepublicorprivate.Third,weuseaneconometricapproachsuitedtothistypeofdatathatallowsustocontrolformanyfactorsotherthanownershipthatmightaffectquality.Contrarily,GasteyerandVaswanishowonlyaregressionlinehighlightingthecorrelationbetweenownershipshareandaveragenumberofviolationsinthecounty.8Testingmustbecarriedoutbystate-orEPA-certifiedlaboratories.Anymembersofawaterutilityorlabfoundguiltyofdeliberatelyfalsifyingdatacanbeprosecutedandjailed.Certifiedlabsmustalsopassperiodicon-siteaudits.Giventhesesafeguards,theEPAbelievesthatitsdataarecollectedinauniformmannerandarereliable.9SeetheEPA'sSDWAwebsitefordetails:http://www.epa.gov/safewater/sdwa/index.html(lastaccessed3/3/05).The1986amendmentsexpandedtheSDWA'sscopetoprotectaquifersfrompollutantsanddevelopdrinkingwaterstandardsforpreviouslyunregulatedcontaminants.The1986revisionadded83additionalcontaminantstothelist(SapatandTeske,2004).In1996,severalnewamendmentsemphasizedprotectingsourcewater,mandatedoperatortrainingandcertification,andrequiredwatersystemstoprovideinformationonwaterqualitytoconsumers.(Seehttp://www.epa.gov/safewater/sdwa/30th/factsheets/understand.html(lastaccessed3/3/05)andhttp://www.epa.gov/safewater/sdwa/theme.html(lastaccessed3/4/05)).
S.Wallsten,K.Kosec/Int.J.ofInd.Organ.26(2008)186205911Sawkins(1995)concludesthatyardstickcompetitionimprovedwaterregulationinEnglandandWales.Benchmarkcompetitioncouldworkthroughanumberofmechanisms.IntheUnitedStates,theEPApostsdetailedsystem-levelwaterqualitydataonitswebsiteandrequireswatersystemstosubmitannualconsumerconfidencereportstoconsumersandregulators.ThesereportsincludealistingofallviolationsofEPAhealthregulations,thepotentialsourcesofcontamination,andthelikelyhealtheffects.Thepurposeofthesereports,accordingtotheEPA,istoenableAmericans01tomakepractical,knowledgeabledecisionsabouttheirhealthandtheirenvironment.Consumersandregulatorscancomparetheirwaterqualitytoindustrybenchmarksandpressuresystemsthatfallshort.Ourcomprehensivewatersystem-levelpaneldataset,derivedfromdatacollectedbytheEPAundertheSDWAanddatacollectedbytheU.S.Census,allowsustoshedlightontheeffectsofownershipandbenchmarkcompetitiononwaterquality(asmeasuredbycompliancewithwaterqualityregulations).Wediscussourdata,empiricalapproach,andresultsbelow.4.DataandanalysisThe1996SDWAamendmentrequirestheEPAtocollectdataandmakemuchofitpublic.11WeobtaineddataoneachcommunitywatersystemintheUnitedStatesfromtheEPA'sSafeDrinkingWaterInformationSystem,FederalVersion(SDWIS/FED).12,13TheSDWIS/FEDcontainsdataoneverypublicwatersystemintheUnitedStates,includingtribalareasandU.S.territories.Asdiscussedabove,watersystemsmustperiodicallytesttheirwaterandreportresultstothestate.14Thestate,inturn,determineswhetherwatersystemsarecomplyingwiththeregulationsandreportingviolationstotheEPAperiodicallyasprescribed.15TheEPArecordsthisinformationintheSDWIS/FED,whichthereforecontainsdataonviolationsofmaximumallowedcontaminantlevelsaswellastreatmentandreportingviolations.16TheEPAreportsviolationsasdiscreteeventsannually,sothedatasetreveals,forexample,howfrequentlywatersuppliedbyeachwatersystemexceededthemaximumallowedlevelsofcertaincontaminantsina10http://www.epa.gov/safewater/ccr/frequentquestions.html(lastaccessed6/11/06).11Recentdataareavailableonlineandolderrecordsareavailableviafreedomofinformationactrequest.12WearegratefultoLeeKyleoftheEPAOfficeofWaterforextractingthesedataforusandpatientlyansweringourmanyquestionsandrequestsforadditionaldata.Theauthors,however,aresolelyresponsibleforanymistakesmadeinusingorinterpretingthedata.13Acommunitywatersystemisapublicwatersystemthatservesatleast15serviceconnectionsusedbyyear-roundresidentsorregularlyservesatleast25year-roundresidents.14TheEPAestablisheshowoftensystemsmustmonitorwatersuppliesforchemicalsbasedonsystemcharacteristics.Forinstance,ifasystemhashadnoviolationsofacertaincontaminantandislocatedinanareawherethecontaminationriskislow,thesystemcanapplyforanexemptionfromtestingforthatcontaminantasfrequentlyaswouldotherwiseberequired.See:http://www.epa.gov/safewater/pws/pdfs/qrg_smonitoringframework.pdf(lastaccessed:12/14/04).15Accordingto40CFRpart142,Section15:EachStateshallsubmitquarterlyreportstotheAdministratoronascheduleandinaformatprescribedbythe[EPA]Administrator,consistingofthefollowinginformation:(1)NewviolationsbypublicwatersystemsintheStateduringthepreviousquarterofStateregulationsadoptedtoincorporatetherequirementsofnationalprimarydrinkingwaterregulations,includingviolationsofthepublicnotificationrequirements.Seehttp://www.epa.gov/safewater/regs.html#cfr(lastaccessed12/15/04).16Atreatmentviolationisafailuretoproperlytreatadrinkingwatersourceinordertoreducethelevelofaspecifiedcontaminant.Areportingviolationisafailuretocollecttherequirednumberofsamples(includingconfirmationsamples)inthespecifiedtimeframe,afailuretoensurethesamplesareanalyzedproperly,orafailuretosubmitrequiredmonitoringinformation.
291S.Wallsten,K.Kosec/Int.J.ofInd.Organ.26(2008)186205Table1Numberofactivesystemsin2003OwnershipSystemsizeVerysmallSmallMediumLargeVerylargePrivate21,774375163045850Localgovt.6469935937162879309FSteadteeraglogvto.vt.1230751204818775637001PNuatbilvice/pArimveartiecan4751211661592163563660Unknown54713224130Total30,29514,375469335163663N3o0t0e:,SMizeedicuatme:g3or3i0e1ses1t0a,b0li0s0h,eLdabrgye:US10E,0P0A1ba1s0e0d,0o0n0a,vVeerraygeladragiel:y1p0o0p,u0l0at1i+o.nserved:Verysmall:25500,Small:501givenyear,butnotbyhowmuch.17Thedataalsoincludeself-reportedinformationonsystemownership,numberofserviceconnections,populationserved,primarywatersource,andthegeographiclocationofeachwatersystem.U.S.watersystemsvaryacrossmanydimensions,makingthemidealfortestingtheceterisparibuseffectsofsuchcharacteristicsasownershiporwatersystemsize.Systemsrangefromsmallpublicutilitiesservingasfewas25peopleandrelyingongroundwatertolargeprivateutilitiesprovidingtreatedsurfacewatertomillionsofcustomers.Ownershipcategoriesincludeprivate,federal,state,local(municipal),tribal,public/private,andunknown.Privatesectorinvolvementcanrangefromcompleteownershipandoperationtosmallcontractoperationswork.Wehadhopedthatthepublic/privatecategorywouldcapturethesecontractoperations.Unfortunately,accordingtotheEPA,Public/Privatehasnoexplicit81definitionandthecategoryisnotusedconsistently.FewerthanfourpercentofwatersystemsclassifiedthemselvesasPublic/Private.Primarywatersourcecategoriesincludegroundwater,surfacewater,purchasedgroundwater,purchasedsurfacewater,andgroundwaterinfluencedbysurfacewater.Differenttypesofwatermaybemorepronetocontaminationortohigherfinalwaterprices,anditisthusimportanttocontrolfortheprimarywatertype.Theresultingpaneldatasetcovers19972003andincludes377,629activewatersystem-years.19Whiletheprecisenumberofwatersystemschangeseachyear,for2003thedataset02contains90,421communitywatersystems,ofwhich53,245areactive.Table1showsthenumberofwatersystemsbyownershipstatusandsizecategory,andTable2showsthetotalnumberofserviceconnectionsbyownershipandsize.Thetablesrevealthatmostwatersystemsareprivate,butthatthosesystemstendtobeverysmall.Acasualinspectionofthenamesofverysmallsystemssuggeststhattheyarelargelymobilehomeparks,apartmentandcondominiumcomplexes,andresortsandhotels.Nonetheless,asubstantialnumberofmediumtolargesystemsareprivate,althoughtheverylargest(thosethat71Seehttp://www.epa.gov/safewater/mcl.htmlforalistofcontaminants;http://www.epa.gov/OGWDW/mdbp/ieswtr.htmlfortheFederalRegisternoticeonwatertreatmenttechniques,andhttp://www.epa.gov/safewater/mdbp/qrg_st1.pdfforabrochureonwatertreatment.81ConversationswithLeeKyle,EPAOfficeofWater,6/17/04and12/3/04.19Theviolationsdataextendbackto1976,butarecordofownershipbeginsonlyin1997.20InactivewatersystemsarethosereportedtotheEPAasbeinginactiveaswellasthosethatreportedinoneyearandnotatallinthefollowingyear.
S.Wallsten,K.Kosec/Int.J.ofInd.Organ.26(2008)186205391Table2Totalpopulationservedin2003OwnershipSystemsizeVerysmallSmallMediumLargeVerylargePrivate3,082,8964,564,5023,608,74713,691,53419,550,550Localgovt.1,547,28413,854,54921,707,49381,009,849102,691,886FSteadteeraglogvto.vt.2464,,580524138783,,217735541290,,02745818,0550,77,233520103,414PNuabtilivce/pArimvaetreican1842,67,96540828203,,521505618441,,04370714,85,1876,078810,275,450Unknown79,223184,846148,678307,0920Total5,010,47320,269,11027,229,74898,888,198123,621,3003N3o0t0e:,SMizeedicuatme:g3or3i0e1ses1t0a,b0li0s0h,eLdabryge:US10E,0P0A1ba1s0e0d,0o0n0a,vVeerraygeladragiel:y1p0o0p,u0l0at1i+o.nserved:Verysmall:25500,Small:501serveover1,000,000notidentifiedseparatelyinthetable)areownedonlybylocalgovernments.Privateandlocalsystemsaccountformorethan90%ofallcommunitywatersystems.Watersystemsownedbystatesappeartoserveprimarilyprisonsanduniversities.Federally-ownedsystemsservemilitarybases,nationalparks,andsomeprisons.Finally,NativeAmericantribesoftenownandoperatetheirownwatersystems.Asdescribedabove,theSDWIS/FEDcontainsdataonviolationsofthemaximumallowedlevelofanumberofhealth-relatedcontaminants,treatmenttechniques,andmonitoringandreporting.21Table3showstheaveragenumberoftheseviolationsbysystemownershipandsizefrom1997to2003.Theaveragenumberofhealth-basedcontaminantviolationsissmallgenerallylessthan0.1persystem,peryear.Theaverage,however,masksvariationacrosssystemtypes.WatersystemsownedbyNativeAmericantribestendtoexperiencethemostfrequentcontaminantviolations,followedbyprivatesystemsandthenlocalsystems.Federally-ownedsystemsreportthefewestviolationsofhealth-basedmaximumcontaminantlevels.Violationsbysizegiveasomewhatdifferentview.Smallersystemstendtohavethelargestnumberofviolationsofalltypes,althoughlargewatersystemsappear,onaverage,tohavealmostasmanycontaminantviolationsasverysmallsystems.Differencesacrosssizemeanthatsimplecomparisonsacrossownershiptypesmaybedeceptive,ascertaintypesofsystemsaremorelikelytobeofacertainsize.Typically,largersystemshavemoresamplingpoints,andaviolationatanysamplingpointcountstowardsthetotalnumberofviolations.Onemightthereforeexpectmoreviolationsinlargersystemsforreasonsthathavelittletodowithinherentwaterquality.Thisobservationalsomeansthatalargeraveragenumberofviolationsinsmallsystemsmaybeevenworsethanthefigureimpliessinceallelseequal,systemswithfewersamplingpointsshouldhavefewermeasuredviolations.21TheEPAcompilesdataonthenumberofviolationsperwatersystemsperyear.Itisnotobvioushowbesttousethisvariable.Onemightnormalizeorweightitbypopulationservedonthegroundsthataviolationinalargesystempotentiallyaffectsmorepeoplethanaviolationinasmallsystem.However,weusedtheunweightednumberofviolationspersystembecauselargerwatersystemstypicallysampletheirwateratmorelocationsthansmallersystems.Largesystemscanthusreceivemoreviolationsthansmallersystemssinceaviolationatanysamplingpointcountsasaseparateviolation.(conversationwithAbeSiegel,EPAOfficeofWater,2/2/05).Assuch,weconsideredthevariableviolationspersystemperyeartobealreadynormalized,albeitimperfectly.Thesefacts,however,makeitespeciallyimportanttocontrolforthesizeofthewatersystemsincesizewillbecorrelatedwiththenumberofviolations.
194S.Wallsten,K.Kosec/Int.J.ofInd.Organ.26(2008)186205Table3SDWAviolationsbysystemownership,size,andprimarywatersourceMaximumcontaminantTreatmenttechniqueMonitoringandreportinglevelviolationsviolationsviolationsOwnershipFederalgovt.0.0490.0411.312Localgovt.0.0950.0650.691NativeAmerican0.1220.0914.983Private0.1030.0311.449Stategovt.0.0750.0251.016eziSVerysmall0.1060.0371.48Small0.0880.0540.653Medium0.0830.0550.619Large0.1010.0870.738Verylarge0.0450.1411.061PrimarywatersourceGroundwaterundertheinfluence0.0750.3533.029ofsurfacewater(GU)Purchasedgroundwaterunderthe0.0290.1741.377influenceofsurfacewater(GUP)Groundwater(GW)0.1050.0181.166Purchasedgroundwater(GWP)0.0590.0130.299Surfacewater(SW)0.1610.2921.65Purchasedsurfacewater(SWP)0.0370.0260.29Note:verysmallsystemsserve25500people,smallserve5013300,mediumserve330110,000,largeserve10,001100,000,andverylargeservemorethan100,000.Monitoringandreportingviolationsarethemostcommonviolationtype,withallbutlocalsystemsinviolationatleastonceperyear,onaverage.Asforcontaminantviolations,tribalsystemsexperiencethemostfrequentmonitoringandreportingviolations.Locally-ownedsystemshave,onaverage,thefewestsuchviolations.Verysmallsystemstendtohavehighernumbersofmonitoringandreportingviolationsthanlargersystems.Theaveragenumberdecreasesthroughmediumsystemsandthenincreasesforlargeandverylargesystems.Treatmenttechniqueviolationsaregenerallytheleastcommonviolationtype.Again,NativeAmericansystemsarethemostfrequentviolators,followedbylocalsystemsandthenfederal.Privateandstate-ownedsystemsreportthefewesttreatmenttechniqueviolations.Thetablealsorevealsthatviolationfrequencyvariesacrosswatersource.Systemsthatrelyprimarilyonsurfacewaterhavethemostcontaminantviolations,andonaveragehaveabout50%moreviolationsthansystemsthatprimarilyusegroundwater,whichhasthesecond-highestnumberofviolations.Groundwaterundertheinfluenceofsurfacewater,meanwhile,generatesthemostmonitoringandreportingandtreatmentviolations.Thesetabulationsemphasizetheimportanceofcontrollingforwatertypeinanyanalysis.4.1.EconometricanalysisAsdiscussedabove,thedataincludethenumberofviolationseachyear,andwedonothavedataontheseverityofeachindividualviolation.Asaresult,thedependentvariable(violations)is
S.Wallsten,K.Kosec/Int.J.ofInd.Organ.26(2008)186205195countdata,andshouldbemodeledbyanappropriatediscretedistributionthataccountsforthelargenumberofzeroesandsmallvalues.InthecaseofviolationsofSDWArules,itseemsappropriatetoexpectthat,forsmalltimeintervals,theprobabilityofaviolationisproportionaltothelengthofthetimeinterval.ThissuggeststhatthedependentvariableisdrawnfromaPoissondistributionwithparameterλit:22ekitkyitPrðYit¼yitÞ¼it!ytiHere,yitisthenumberofviolationssystemireportsinyeart,lnkit¼bVxit;andxitisavectorofindependentvariablesItthereforefollowsthat;E½yitjxitVar½yitjxitkit¼ebxVit:ððð123ÞÞÞThePoissondistribution,however,assumesthatthemeanofthevariableequalsitsvariance.Inourcase,thevarianceofeachdependentvariableofinterestfarexceedsthemean,suggestingoverdispersioninthedataandthatanegativebinomialdistribution(forwhichthevarianceisaquadraticfunctionofthemean)ismoreappropriate(Greene,1993;LongandFreese,2003).23ThenegativebinomialregressionissimilartothePoissonregressionexceptthat42kit¼expðbVxitþeitÞandEðeitÞ¼0:ð4ÞAsnotedabove,yitisthenumberofviolationsinsystemiinyeart.Wefocusoncontaminantlevelandmonitoringandreportingviolations,asthenumberoftreatmenttechniqueviolationsistoosmalltogenerateconsistentresults.Thevectorxitincludessystemsizeanddummyvariablesindicatingownershiptype.Becauseofthedifferencesbysysteminsizeandownership,wealsointeracttheownershipvariableswithfivesizecategorydummies.Becausewaterqualityandtheeconomicsofdrinkingwaterprovisioncanbeaffectedbythephysicalandtopographicalcharacteristicsofthewatersystem'slocation(Noll,2002),wealsocontrolforlocationfixedeffects-dummyvariablesindicatinginwhichcore-basedstatisticalarea(CBSA)thewatersystemislocated.25AllthefixedeffectdummiesequalzeroforwatersystemsnotlocatedinanyCBSA.Wealsoincludedummyvariablesindicatingthewatersource(e.g.,groundwateror22SeeGreene(1993),pp.676679.23Themeansandvariances(mean,variance)oftheviolationsare:contaminant(0.10,0.27),treatment(0.05,0.23),monitoring(1.14,54.69).24SeeLongandFreese(2003)foradescriptionofthenegativebinomialregressionmodel.25County-levelorevensystem-levelfixedeffectswouldbesuperiortoourCBSAfixedeffects.Unfortunately,thoseprovedimpossiblegivenourresources.First,AllisonandWaterman(2002)pointoutthattheSTATAprocedureforestimatingthefixedeffectsnegativebinomialregressiondoesnotactuallycontrolproperlyforthefixedeffects.Second,giventhatprobleminSTATA,wedidnothaveenoughcomputerresourcestoincludeall3141countyfixedeffectsintheequation,letalone50,000+watersystemdummies.However,includingfixedeffectsmanuallyratherthanbyusingSTATA'sxtcommandshastheadvantageofallowingustouseproceduresdevelopedbyLongandFreese(2003)tosimulateexpectednumberofviolationsforeachtypeofsystembasedontheregressionresults.
916S.Wallsten,K.Kosec/Int.J.ofInd.Organ.26(2008)186205surfacewater),asthisaffectsinherentwaterquality,andyearfixedeffectstoaccountforchangesinregulatoryoversightorcompliancethatmightchangeovertime,suchasutilitiesgainingexperiencewiththeSDWAovertime.26Wealsocontrolforcounty-levelvariables,includingaveragecountyhouseholdincomeandtheshareofthepopulationthatlivesinurbanareas.Thesevariablescomefromthe2000U.S.Census,whichmeansthattheyunfortunatelyvaryonlyovercounty,andnotovertime.Incomeandurbanizationdonotchangeradicallyinashortperiodoftimeanddifferencesacrosscountiesprobablychangeevenlesssothesearereasonableproxiesforincomeandurbanizationeveninyearsotherthan2000.Finally,whileweknowofnopipedwatersystemsdirectlycompetingwithoneanotherforgivenhouseholds,theexistenceofalargenumberofwatersystemsinagivenregionmayfacilitatebenchmarkcompetition.Weconjecturethatutilitiesfacemorepressuretoperformwellincountieswithnumerousutilitiesbecauseitiseasierforconsumersandregulatorstocompareutilitiesingeographicallysimilarareasandmoredifficultforutilitiestoblamepoorqualityonfactorsbeyondtheircontrol(e.g.,geographicfactors).Tocapturethispossibility,wecreateaHerfindahlHirschmanIndex(HHI)tomeasureconcentrationinthewatersystemindustry.Wedefineacountyasamarketandtheshareofthetotalconnectionsinacountythatawatersystemservesasitsmarketshare.AcompletelymonopolisticmarketwouldyieldanHHIof10,000,andtheU.S.JusticeDepartmentdeemsanindustrytobeheavilyconcentratediftheHHIexceeds1800.Becausepipedwatersystemsdonotcompetewithinthesamegeographicspaceandcustomersdonothaveachoiceofwaterprovidershortofmovingtoanewwaterdistrict,amoretypicalHHIcalculationwouldimplyacompletelymonopolisticmarket.However,weconsiderourmeasureofHHIappropriateforexploringbenchmark,ratherthanhead-to-head,competition.Evenourmeasureofbenchmarkcompetition,inwhichamarketisacounty,suggestsahighlyconcentratedmarket.TheaverageHHIacrossallU.S.countiesis4224morethantwicetheJusticeDepartment'sthresholdforaheavilyconcentratedindustry.Still,nearlyone-thirdofthepopulationlivesincountieswithatleastmoderatebenchmarkcompetition(HHI1800).4.2.ResultsnointeractionsTable4showsboththecoefficientestimatesandtheincidentrateratios(IRRs)resultingfromestimatingthisequationwithouttheinteractionvariablesandwithrobuststandarderrors.27Thetableshowsthatcontrollingforsize,watersource,location,year,income,andurbanization,privately-ownedfirmshavefewercontaminantviolations,butmoremonitoringandreportingviolations,thandomunicipally-ownedfirms.TheHHIispositivelycorrelatedwithviolations,suggestingthatthenumberofviolationsdecreaseswhenthereismorecompetition,consistentwiththehypothesisthatcompetitioneveninitsweakbenchmarkformimprovesoutcomes.Thetablealsoshowsthat,ingeneral,thenumberofviolationsofeithertypegenerallydecreasesassystemsizeincreases.Thiscorrelationmayreflectthegreaterresourcesavailableto26Wedropsystemswhoseownershipstatusisunknownand,unfortunately,systemswhosestatusispublic/privatebecausethoseareeffectivelyunknownaswell,asdiscussedabove.EPAofficialshadnopriorsastowhichownershiptypethesewerelikelytobe.72Weadjustedthestandarderrorsforintra-CBSAcorrelation.