人與機(jī)器:自動化監(jiān)管的技術(shù)承諾和政治限制 Man vs. Machine:Technological Promise and Political Limits of Automated Regulation Enforcement_第1頁
人與機(jī)器:自動化監(jiān)管的技術(shù)承諾和政治限制 Man vs. Machine:Technological Promise and Political Limits of Automated Regulation Enforcement_第2頁
人與機(jī)器:自動化監(jiān)管的技術(shù)承諾和政治限制 Man vs. Machine:Technological Promise and Political Limits of Automated Regulation Enforcement_第3頁
人與機(jī)器:自動化監(jiān)管的技術(shù)承諾和政治限制 Man vs. Machine:Technological Promise and Political Limits of Automated Regulation Enforcement_第4頁
人與機(jī)器:自動化監(jiān)管的技術(shù)承諾和政治限制 Man vs. Machine:Technological Promise and Political Limits of Automated Regulation Enforcement_第5頁
已閱讀5頁,還剩70頁未讀 繼續(xù)免費閱讀

下載本文檔

版權(quán)說明:本文檔由用戶提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請進(jìn)行舉報或認(rèn)領(lǐng)

文檔簡介

NBERWORKINGPAPERSERIESANVSMACHINETECHNOLOGICALPROMISEANDPOLITICALLIMITSOFAUTOMATEDREGULATIONENFORCEMENTrowneovicaGazzehaelGreenstonetapshovaWorkingPaper30816http//papers/w30816NATIONALBUREAUOFECONOMICRESEARCHCambridgeMA8January3WethankValentinaBosetti,StefanoCaria,TatyanaDeryugina,ClementImbert,KoichiroIto,ElianaLaFerrara,AshleyLanger,LeslieMartin,EricaMeyers,LouisPreonas,andparticipantsatTWEEDS,MidwestEnergyFest,AERE,AFE,ASSA-AERE,EEAESEM,BSEInsightsWorkshop,13thConferenceontheEconomicsofEnergyandClimate,andseminarsatBocconiUniversity,BolognaUniversity,UniversityofRomaTorVergata,andNYUWagnerSchoolofPublicService.IrisSong,JacksonReimer,AdityaJain,LeonBeaufils,MonicaKim,andTejomayGadgilprovidedexcellentresearchassistance.OurpartnersinFresnoprovidedendlesssupportinimplementingtheexperiment.WearegratefultotheLauraandJohnArnoldFoundationandtheSmithRichardsonFoundationfortheirgenerousresearchsupport.Thisexperimentwaspre-registeredonJune14th,2018ontheOpenScienceFrameworkPreregistrywithDOI:”10.17605/OSF.IO/XYFU3”andintheAEARCTRegistrywithDOI:”10.1257/rct.8290-1.0”.Allremainingerrorsareourown.WeobtainedIRBapprovalfromtheUniversityofChicago.TheCityofFresnohadtherighttoreviewthearticlebeforepublication,andprovidedminimalclarifyingcomments,whichweincorporated.TheviewsexpressedhereinarethoseoftheauthorsanddonotnecessarilyreflecttheviewsoftheCityofFresnoortheNationalBureauofEconomicResearch.Atleastoneco-authorhasdisclosedadditionalrelationshipsofpotentialrelevanceforthisresearch.Furtherinformationisavailableonlineat/papers/w30816NBERworkingpapersarecirculatedfordiscussionandcommentpurposes.Theyhavenotbeenpeer-reviewedorbeensubjecttothereviewbytheNBERBoardofDirectorsthataccompaniesofficialNBERpublications.?2023byOliverBrowne,LudovicaGazze,MichaelGreenstone,andOlgaRostapshova.Allrightsreserved.Shortsectionsoftext,nottoexceedtwoparagraphs,maybequotedwithoutexplicitpermissionprovidedthatfullcredit,including?notice,isgiventothesource.ManvsMachine:TechnologicalPromiseandPoliticalLimitsofAutomatedRegulationEnforcementOliverBrowneLudovicaGazzeMichaelGreenstoneandOlgaRostapshovaNBERWorkingPaperNo.30816January3JELNoK,Q25STRACTNewtechnologiesallowperfectdetectionofenvironmentalviolationsatnear-zeromarginalcost,buttake-upislow.WeconductedafieldexperimenttoevaluateenforcementofwaterconservationruleswithsmartmetersinFresno,CA.Householdswererandomlyassignedcombinationsofenforcementmethod(automatedorin-personinspections)andfines.Automatedenforcementincreasedhouseholds’punishmentratesfrom0.1to14%,decreasedsummerwateruseby3%,andreducedviolationsby17%,whilehigherfinelevelshadlittleeffect.However,automatedenforcementalsoincreasedcustomercomplaintsby1,102%,ultimatelycausingitscancellationandhighlightingthatpoliticalconsiderationslimittechnologicalsolutionstoenforcementchallenges.OliverBrowneMichaelGreenstoneBerkeleyResearchGroupUniversityofChicagoobrowne@DepartmentofEconomicsLudovicaGazzeChicago,IL60637DepartmentofEconomicsandNBERUniversityofWarwickmgreenst@UnitedKingdomLudovica.Gazze@warwick.ac.ukOlgaRostapshovaUniversityofChicagoEnergy&EnvironmentLaboro@21Introductionregulationsinenvironmental(AlmandShimshack,2014;Duflo,Greenstone,Pande,andRyan,2013;Gibson,2019;ReynaertandSalleeVollaardZoutaxcollectionSarinandSummers20),andotherdomains.—akeyparameterinBecker’s(1968)canonicalmodelofcrime;theyincludespeedcameras,biometriccards,automaticairpollutionmonitors,andsatellites,andhaveinhibitedspeeding,corruption,pollution,deforestation,andtaxevasion(DuˇsekandTraxler,2022;MuralidharanandSukhtankar,2021;Greenstone,He,Jia,andLiu,2020;Ferreira,2021;CasaburiandTroiano,2016).Agrowingliteratureemphasizesthatloweringreliablemonitoringcostsiskeytoimprovingcompliancewithenvironmentalandworkplaceregulations,aswellasbillpayment(Duflo,Greenstone,Pande,andRyan,2018;Banerjee,Duflo,andGlennerster,2008;Meeks,Omuraliev,Isaev,andWang,2020).Yet,policymakersfrequentlychoosenottoadoptthesetechnologiesandwhentheydo,theyusethem(e.g.,Brollo,Kaufmann,andLaFerrara,2019).ThispaperprovidesarareopportunitytostudyboththebenefitsofonesuchtechnologyandthepoliticalcostsorwateruseregulationsinFresno,CA.TheseregulationshelpcitiescopewithincreasingdroughtconditionsduetoclimatechangeDiffenbaughSwainandToumaaslawnirrigationisthesinglelargestenduseofresidentialwater(HanakandDavis,2006).Utilitiestypicallydonotpricewateratmarginalsocialcostforpoliticalandethicalreasonsandinsteadrelyonnon-pricemechanismstomanageconsumption(Brent,Cook,andOlsen,2015;BrentandWard,2019).Fresno,likemostcitieswithoutdoorwateringrestrictions,hasreliedon“watercops”tomonitorcompliancewiththeserestrictions,despitehavingsmartmeterssince2013.Yet,violationswererampantandpunishmentswererare:68%ofhouseholdsviolatedtheserestrictionsatleastonceinthesummerof2016,butonly0.4%ofviolationsweresanctioned(Table1).1Weimplementedarandomizedfieldexperimentwiththenearly100,000Fresnohouseholds,introducingauto-matedenforcementofoutdoorwateruserestrictionsviasmartmetersthatenabledperfectdetectionofviolations.Fresnowasoneofthefirstlargemunicipalwaterutilitieswithuniversalsmartmeteradoptionamongsingle-familyresidentialcustomers.Withthecity,wedesignedandimplementedanevaluationthatexperimentallyvariedbothrsorcontinuedthestatusquoofin-personinspectionsbywatercops(i.e.,detectionratesof100%versus0.4%)—andeexperiment1Wedefineviolationsaswateruseabove300gallons/hourduringprohibitedhours.Thisisthethresholdthecityeventuallyselectedforautomatedenforcement.3DepartmentofPublicUtilities(DPU).TheexperimentprovidedaperhapsunprecedentedopportunitytovarythekeyparametersprobabilityofdetectionandpenaltythatdeterminethecostofcommittingacrimeinBecker’sogramTherearethreeprimaryfindings.First,automatedenforcementgreatlyincreasedenforcementaswellascom-pliancewiththelaw.Whiletheshareofhouseholdsfinedfornon-compliancegrewfrom0.1to14.3%thankstothentsonJulyinresponsetoanotherdroughtThetreatmentalsoreducedwaterconsumptionafterthepilot,suggestingevenlargerpotentialconservationeffects.Third,automatedenforcementcreatedpoliticalbacklashthatultimatelyledtotheprogram’stermination,asthenumberofhouseholdscallingtheutilityincreasedby654%andcallsidentifiableascomplaintsanddisputesofenforcementactionsincreasedby.Thereareseveralotherfindings.Lowerfinesdidnotaffectthefrequencyofviolations,waterconsumption,orlyconstantacrossthedistributionsofbaselinewaterconsumptionandincome.However,heavywaterusersandthenderstoodthemechanicsofYet,despiteattemptstobecustomer-friendly(e.g.,graceperiodsimplicitinthefineschedule),theautomatedenforcementprogramdidnotsurvive.Publicbacklash,includingcustomercalls,ledthecitytoimplementafinemoratorium,weakentheconservationrules,andfinallyinstitutenewrulesthatessentiallyoutlawedautomateddeandRyanReynaertandSalleeGibsonArelatedliteratureregulations(Zou,2021;Fowlie,Rubin,andWalker,2019;Mu,Rubin,andZou,2022;BanerjeeandDuflo,2006;ingsalsohighlighttheimportanceofdesigningprocessesthatmaximizepublicsupportforadoptionofnewtechnology,inlinewithrelatedliterature(Blumenstock,Callen,Faikina,andFiorin,2022;Bossuroy,Delavallade,andPons,2019;MuralidharanandSukhtankar,2021;Atkinetal.,2017).Second,thepaperaddstoanextensivecrimeliteraturethatfindsthatpeoplerespondtoexpectedfuturepunishment(Bar-Ilan4andSacerdote,2004;Drago,Galbiati,andVertova,2009),theperceivedauditingprobability(Klevenetal.,2011),andpastpunishment(Kuziemko,2013;MaurinandOuss,2009;Haselhuhn,Pope,Schweitzer,andFishman,2011;peercomparisons,outdoorwateruserestrictions,andcombinationsofpolicyinstruments(Jessoe,Lade,Loge,andSpang,2021;Pratt,2019;Wichman,Taylor,andHaefen,2016;HalichandStephenson,2009;Kenney,Klein,andClark,2004;RenwickandGreen,2000;Michelsen,McGuckin,andStumpf,1999;Hahn,Metcalfe,Novgorodsky,andPrice,2016).22ExperimentandData2.1ExperimentThesesmartmetersmeasurewaterconsumptionandtransmitdataeveryfifteenminutesbutdonotallowforwaterusescheduling.Priortoourexperiment,Fresnowasusingsmartmetersforbillingandsidentsfiercelyresistedtheimplementationofsmartmetersinprivateociationfiledalawsuittopreventtheinstallationofresidentialwatermeters,whichwasdismissed.Fouryearslater,Fresno’scitycharterprohibitedtheuseofsmartmetersforbillingtysdeliveriesofCentralValleyProjectwaterionsoftheserestrictionsFresnohadfiveparttimewatercopswhoissuedfinesinTolimitfineburdenandpoliticaldiscontent,Fresnoonlysanctionedthefirstviolationinamonth.Forthefirst,second,andthirdmonthwithviolations,householdswerechargedfinesof$0,$50,$100,respectively(baselineschedule).AmonthlywaterwnsprinklertimerstocomplywiththewateringscheduleementarandomizedfieldexperimentbetweenJulyandSeptembertoinformthecitywideimplementationofautomatedenforcementandthenewfineschedule.Werandomlyassignedallsingle-familyhouseholdsinthecitytooneoftwelveexperimentalgroups,varyingalongtwocross-randomizedandthescheduleoffines2ThemostcomparablestudyisWest,Fairlie,Pratt,andRose(2021)whichexaminesasimilarexcessive-use-thresholdpolicythatwasannouncedbutneverwentintoplace.Withoutpriorannouncement,thecityidentifiedhouseholdsthathadabove-thresholdconsumptionduringoneearlierweekandnotifiedthemofthishypotheticalviolationandofthefineschedule.Thesehouseholds,aboutathirdofaccounts,reducedwateruseby31%thereafter.Incontrast,ourstudyevaluatesthecitywide,average,treatmenteffectsofanenactedpolicy,thatisthepolicy-relevantparameter.3Authors’calculationsbasedonmostrecentviolationdataavailable(Browne,Gazze,andGreenstone,2021).5tedThecityannouncedthepilotprograminJune2018throughamediacampaign.Householdsreceivedamailerexplainingthatintheupcomingthree-monthpilottheywereassignedanenforcementmechanismviaalotterydmethodandfinescheduleUponcedonthefollowingmonthswaterbills.ThefirstnotificationsweresentonJuly18,2018.ThecitydidnotsanctionviolationsbetweenAugust1andAugust12,2018,toenableitscustomerservicetocatchupwiththebacklogofcustomercallsreceived.FigureA3overlaystheexperimentaltimelinewithtrendsinwateruse(PanelA)andcalls(PanelB)intheautomatedandtechnologieslikesprinklertimersorlawn-to-turfconversions.Moreover,ifthepolicyhadbeeninplaceforlonger,itmighthavebecomemoreeffectiveovertime,asmorehouseholdslearnedaboutitandhadmoretimetoadjustyorthatisnotprogrammablemighthavelargerimmediateeffectsFinallywecannotexcludethatthefineamountswerelesssalientthanotherexperimentalfeaturesasfineseviolation2.2DatartingwithFresnospopulationofsinglefamilyresidentialhouseholdswerestricttheexperimentalsampleinthreeways.First,weincludeonlyhouseholdswithpositivewateruseunder216,000gallons/monthinApril2017,andexcludehouseholdsthathadtheirwatershutofforusedmorethan300gallons/houronaverage.Second,weexcludeaccountsthatcouldnotbematchedtoasingle-familyparcelintheassessorfiles,whichwefromtheAmericanCommunitySurveyyearsusingCensusblockgroupidentifiersbecausemedianhouseholdincomeattheCensusblockgroupisavariableofinterestforheterogeneityanalysis.54Outreachmaterialsonlystatedtheassignedschedulewithoutreferencingthebaseline.5TheaveragenumberofsamplehouseholdsperCensusblockgroupis287.6WeexcludeanadditionalhouseholdswithmissingwaterusedataOurfinalanalysissampleconsistsof8,904dayswiththeoccasionalsmartmetermalfunctionpreventingaTologcustomercallstotheDepartmentofPublicUtilities,wehiredrepresentativestostaffadedicatedphoneThesmartmeterdatarunsfromJanuary2017throughFebruary2019.ThecalldatacoversJune2018throughbruaryperiodoupPanelAEachoftheothereleventreatmentgroupsincludesofsamplehouseholdsPanelBroupThetreatmentgroupsappearbalancedintermsofbaselinewateruse,violation,andclearancerate.Columns1and2ofTable1reporttherandomizationoeffectinbyaboutgallonshourusethresholdhadbeeninforceintavailableenforcementdatafromStrikinglyofhouseholdsexceededgallonshouraboveorbelowthemedianof1)baselinewateruseduringApril2017and2)medianincomeoftheCensusblockgroup(Table1,Columns8and9).Weadditionallystratifiedtherandomizationbycitycouncildistrict.Finally,enforcementgroup,atbaselinefineand300gallon/hourthreshold.0.5%ofhouseholdsoptedout,andthesehadhigherbaselinewateruseandviolationrates(TableA2).Becausetreatmentgroupssawhigherdropoutrates,wepresentIntention-To-Treat(ITT)treatmenteffectsthatincludeopt-outsintheirrandomlyassignedgroups.3EmpiricalAnalysisWeassesstheimpactoftheexperimentaltreatmentsoncompliance,wateruse,andcustomercalls.First,we6Webasedtheallocationacrossautomatedandnon-automatedgroupsonthecity’scapacitytohandlecalls.73.1TheEffectsofAutomatedEnforcementyit=α+βAutomatedi+工VjIn-Person×Fines+εitj∈{25,50}(1)whereyitisanoutcomeforhouseholdiinmontht,Automatediisanindicatorforhouseholdi’sassignmenttotstobeequalDuetorandomsignmenttheOLSestimatorcapturesthecausaleffectofautomatedenforcementWeclusterstandarderrorsatthelevelofrandomization:household.Forourmainresults,wealsoreportWestfall-Youngstepdownadjustedp-valuesthatcorrectformultiplehypothesestesting(Jones,Molitor,andReif,2019).actionsandcompliancebehavior,Column2detailseffectsonwateruse,andColumns3a-3bdocumentthecostsofautomatedenforcementintermsofcustomercalls,whichthecityinterpretedasameasureofdiscontentwiththepolicy.Thein-personinspectiontreatmenteffectsarereportedinTableA3.PanelAofTableA4probesthetafromsummerandPanelBincludeshouseholdedinourpreanalysisplanomatedenforcementtreatmentincreasedcompliancewiththewaterconservationpolicythroughitsreliabledetectionofviolations.Theaveragenumberofviolationspermonthdecreasedfrom3.7to3.1(Column1a)sviolatedinanygivenmonthdramaticallyHouseholdsintheautomatedenforcementgroupsweremorelikelytoreceivewarningsandfinesby1,715%and14,100%respectively(Table2,Columns1c-1d)andpaid$2.35morepermonth(Column1e),3%ofanFurther,automatedenforcementdecreasedwaterconsumptionby2.9%andtheseestimatesarepreciseand7TableA5reportsestimatesandstandarderrorsthataccountforthecovariate-adaptivestratifiedrandomizationfollowingBugni,Canay,andShaikh(2019).8TableA6estimatesthedifferenceinthenumberofmonthswithviolationsacrossautomatedandcontrolhouseholds.Householdsintheautomatedgroupwere3%lesslikelytoeverviolateandalmost20%lesslikelytoviolateallthreepilotmonths.8robust(Table2,Column2).9Thisdeclinewaspartlydrivenbydecreasesinheavyconsumptionhours:automatedbyandrespectively(TableA8).10PanelAofFigure1reportstheresultsfromestimatingequation(1)butallowsthetreatmenteffecttovaryforeachmonthofthesample,includingthemonthsbefore(whenwewouldexpectnoeffect)andafterthepilot.ThensumptionreductionofinJulywhenfineshadnotbeenissuedyetto4%inSeptember,when17.1%ofhouseholdshadreceivedawarningorafine.Whentheoutcomevariableisanindicatorforwhetherahouseholdhadatleastoneviolationinamonth,thetreatmenteffectincreasedoverthecourseofthesummerfrom1.2%inJulyto5.9%inSeptember(PanelB).Thesefindingsareconsistentwiththefurther.11effectisstillevidentbutsmallerinDecemberthroughFebruary(whenourdataends):-50.1gallon/monthperentwasalsofoundinrecentworkaboutenergyconsumptioninJapanItoIdaandTanakaandinBrazil(CostaandGerard,2021).KeepinginmindannualwaterreductionsmandatedbythestateofCalifornia,ourresultssuggestthepolicy’sthatscalingupthepolicycitywidewouldsaveanestimated174milliongallonsofwaterduringthethreemonthstrictionsareinplaceweestimatesavingsofmilliongallonsofwaterpersummerMoreoverincludingpostpiloteffectsonconservationinwinterandassumingpersistencebeyondFebruaryatthesamelevels,theprogramannuallyincreasedthemonthlycountofcustomerswhocalledatleastonceby654%(Table2,Column3a),andincreasedthemonthlycountofcustomerswhocalledwithacomplaintortodisputeanenforcementactionatleastonceby1,102%(Table2,Column3b).Evenconditionalonreceivingaviolationwarning,householdsintheautomatedontrolhouseholdsFigure9TableA7investigatespeereffectsofautomatedenforcement.WefindnoevidencethathouseholdsinCensusblockswithahigherproportionofhouseholdsinautomatedenforcementdisproportionatelyreducedwateruse.10Finelevelsappeartohaveinconsistenteffectsinthecaseofnon-automatedenforcement.TableA3revealsthatthegroupassignedtothehalvedfinescheduledecreasedwaterconsumption,whilethegroupassignedtothe25%finescheduleincreasedwaterconsumptionbyastatisticallyinsignificantamount,potentiallyduetoaslightimbalanceintherandomization.Theseestimatesarenotstatisticallysignificantaftercorrectingformultiplehypothesestesting.11Treatmenthouseholdsdonotappeartoshiftwaterconsumptionfrombannedtopermittedhours(TableA9).9weretoscaleupautomatedenforcementcitywide,weestimatethatitcouldleadto4,090additionalcallsoverthesummerperiod.However,thisislikelyanupperbound,assomecallsquestionedthepilotspecifically,forexample,llmostcomplaintsreflectedalackofunderstandingofthepolicy’smotivation.MostcallsstartedwithinquiriesaboutthereasonsfornoticesandjustificationsfortheirththefineautomationallegingmetermalfunctionsandnotwantingtheirdatatobeusedinthepilotsttoallowDPUtocatchupwiththecallbacklog.Moreover,ourpartnersreportedthatmanycustomersalsocalledCityCouncilmemberscomplainingaboutthenewsystem,whichultimatelyhaltedthescale-upoftheautomatedenPanelCofFigure1isconstructedlikePanelsAandB,butusinghouseholds’callprobabilityastheoutcome.Theautomatedenforcementtreatmentcausedasharpincreaseincallsduringthepilot.Thistreatmenteffecthad3.2HeterogeneousEffectsInthissection,weexaminewhetherresponsesvaryacrosshouseholdswithdifferingcharacteristicsandassessinelevelsandexcessivewaterusethresholdsinexplainingtheimpactsofautomatedstratificationvariables,incomeandbaselineuse,aswellasacrossbaselinepropensitytouseexcessivewater.Specifically,weestimateaversionofequation(1)thataddsinteractionsoftheautomatedindicatorwithindicatorsforeachdecileofthecharacteristicofinterest(exceptthe5th)andincludesdecilesfixedeffects.Thus,thecoefficientsontheseelativetothethThecausaleffectofautomatedenforcementonwaterconsumptiondidnotvaryproportionallytobaselineuse(Figure2,PanelA).Incontrast,theeffectonthepropensitytocallsharplyincreasedwithbaselineuse.PanelsBlevelsbutthewealthyandlikelyviolatorswereresponsibleforadisproportionateshareoftheincreaseincallsandSecond,weexploitexperimentalvariationinfinesandthresholdswithintheautomatedenforcementtreatmenttoexaminetheirinfluenceonwateruseandpropensitytocallDPUTodothis,weestimatethefollowingequation:yit=α+β1Automatedi+β2Automated500i+β3Automated700i+β4Automated×Fines50%i+β5Automated×Fines25%i+(2)工VjIn-Person×Fines+εitj∈{25,50}thatcharacterizesthenineautomatedenforcementtreatmentswithfiveindicators.Specifically,theAutomatedistheeffectofthecitysdefaultautomatedpolicywhichhadagallonhourthresholdandthestandardfineschedule.Thefourotherindicatorsmeasuretheeffectsofincreasingthethresholdto500or700gallon/hourorreducingthefineto50%or25%ofthestandardschedule.12WeexpectthathigherexcessiveusendtherebydisplayingtheeffectofdeviationsfromFresnoslymeaningfulwayOurpreviousworkusingdatafromFresnoestimatesthataincreaseinmarginalwaterratesleadstoadecreaseinwateruseof19%(Browne,Gazze,andGreenstone,2021).However,adirectcomparisonwiththisfindingischallengingandprobablyinappropriate,becausethefinesarefordiscreteevents(i.e.,exceedingthehourlywaterconsumptionthresholds),ratherthanatariffonallwaterconsumption.3.3Households’BehaviorafterWarningsandFinesInanunconstrainedsetting,wewouldhavealsoexperimentallyassignedenforcementactionsinresponsetoviolationsoftheexcessiveusethresholdtolearntheircontributiontothetreatmenteffectsforouroutcomesoftstudystyleanalysesoftheimpactofreceivingwarnings12TableA10reportsthecoefficientsonthefullyspecifiedmodel,thatis:yit=對z∈{300,500,700}對j∈{25,50,100}βzjAuto+對j∈{25,50}γjIn-Person×Fines+εit(3)Thetablealsoreportsthep-valueofatestofthedifferencebetweentherestrictedmodelinequation(2)andthefullmodelinequation(3).Inotherwords,wetestH0:β?300,100?β?300,50=β?500,100?β?500,50=β?700,100?β?700,50andβ?300,100?β?300,25=β?500,100?β?500,25=β?700,100?β?700,25bycomputingF=(SSred/SSfull)forhousehold-monthlevelregressions.Foralevelofsignificanceα,werejectH0ifFislargerthantheupper1?αpercentileintheF(Nclusters?1,Nclusters?1)distribution.SSredandSSfullaretheresidualsumsofsquaresfromtheparsimoniousandthefullspecification,respectivelyandNclustersisthenumberofclusters.ForColumns1c-1d,weusetheformula:F=(SSredfull/s)/(SSfull/dffull)wheres=dfredfull,dfredanddffullarethedegreesoffreedomfromtheparsimoniousandfullmodel,andweusetheF(s,dffull)distribution.Basedonthistestandα=0.1,wecannotrejectthenullhypothesesthatthefine-thresholdinteractionsdonotmatterbutforColumns1c-1d.13Albeitsmall,thecoefficientontheeffectofa25%fineonthecallprobabilityisstatisticallysignificantatconventionallevels.applicationofpunishmentsiscalledthespecificdeterrence”effectandconceptuallycontrastswiththeoveralleffectofachangeinenforcementactionsthatalsobabilityofaviolationiegeneraldeterrenceGlueckAlargeliteraturefindsimportantspecificdeterrenceeffectsinothersettings(Kuziemko,2013;MaurinandOuss,2009;Haselhuhn,Pope,Schweitzer,andFishman,2011;DuˇsekandTraxler,2022).Weexploitacross-householdvariationinthetimingofviolationsinanevent-studydesignwithhouseholdandweekfixedeffectstoestimatetheeffectofenforcementactionsonwateruseandprobabilityofcallingDPU.Thistion12yit=α+工工βIit(jWeeksPostViolationa)+γi+γt+εit(4)j=?12a∈{1,2,3}whereyitisanoutcomeforhouseholdiinweekt,Iit(jWeeksPostViolationa)isanindicatorforweektbeingjweeksbeforeorafterhouseholdireceivedanenforcementaction(withtheindicatorsforweeksj=?12andj=12equaling1forweeks-12andearlierand12andlater,respectively),andγiandγtarehouseholdandweekfixedeffectsrespectively.Iit(jWeeksPostViolationa)isconstructedforWarning,Fine1andFine2andequalszeroforhouseholdsthatdidnotreceivethatenforcementaction.Wehaveatmosttenweeksafterthesecondfine.Thetimeisindexedrelativetotheviolation.IncontrasttothetreatmenteffectsinSections3.1and3.2,theseestimateselAofFigurereportsestimatesofequationwheretheoutcomevariableisthelogarithmofdailywateriolationInmmediatelyprecedingndduetoaspecialeventoraleakunderscoringthecificdeterrencefromenforcementactionsbeyondthefirstfewweeks,becausewhetherandhowquicklysuchreductionswouldhaveoccurredintherPanelBofFigure3reportsestimatesofequation(4)wheretheoutcomeisanindicatorthatequalsoneifthesAswithwaterconsumptionthereisevidenceof14Thesharpdecreasesinwaterusewhennoticesofviolationaresenttohouseholdsforsecondandthirdviolationsbutbeforefinesaccruesuggestatleastsomeofthebehaviorchangeisdrivenbythenoticesratherthanthe

溫馨提示

  • 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
  • 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
  • 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁內(nèi)容里面會有圖紙預(yù)覽,若沒有圖紙預(yù)覽就沒有圖紙。
  • 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
  • 5. 人人文庫網(wǎng)僅提供信息存儲空間,僅對用戶上傳內(nèi)容的表現(xiàn)方式做保護(hù)處理,對用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對任何下載內(nèi)容負(fù)責(zé)。
  • 6. 下載文件中如有侵權(quán)或不適當(dāng)內(nèi)容,請與我們聯(lián)系,我們立即糾正。
  • 7. 本站不保證下載資源的準(zhǔn)確性、安全性和完整性, 同時也不承擔(dān)用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。

評論

0/150

提交評論