Do Voters Support Local Commitments for Climate Change Mitigation in Italy?

2018·0.72 MB·English

by  Simone Martelli

EcologicalEconomics144(2017)27–35
ContentslistsavailableatScienceDirect
Ecological Economics
journal homepage: www.elsevier.com/locate/ecolecon
Do Voters Support Local Commitments for Climate Change Mitigation
in Italy?
SimoneMartellia,b,*,GreetJanssens-Maenhoutb,PaoloParuolob,ThierryBrécheta,c,EricStrobld,Diego
Guizzardib,AlessandroK.Ceruttib,AndreeaIancub
aCORE,ChairLhoistBerghmans,UniversitéCatholiquedeLouvain,Louvain-La-Neuve,Belgium
bEuropeanCommission,JointResearchCentre(JRC),Ispra,Italy
cLouvainSchoolofManagement,UniversitéCatholiquedeLouvain,Louvain-La-Neuve,Belgium
dDepartmentofEconomicsandOeschgerCentreforClimateChangeResearch,UniversityofBern,Bern,Switzerland
A R T I C L E I N F O A B S T R A C T
Articlehistory: Thereisagrowinginterestinvoluntaryprogrammesforclimatechangemitigation,includinggreenhouse
Received29August2016 gas(GHG)emissionreductioncommitments.Thispapergaugesevidenceonthesupportofcitizensforcli-
Receivedinrevisedform9June2017 matechangemitigationprogrammesatthelocallevel,analysingvotingbehaviour.Aquasi-experimental
Accepted28June2017 set-upisofferedbytheEUCovenantofMayors(CoM)initiative,whichisthemainstreamEuropeanmove-
mentforlocalauthoritiesvoluntarilycommittingtomeetandexceedtheEuropeanUnion20%GHGemission
JELclassification:
reductiontargetby2020.TheelectoralimpactoftheparticipationofItalianmunicipalitiestotheCoMis
Q54
Q58 estimated,usinganinstrumentalvariable(IV)approach.Mayorscommittingtoreducegreenhousegasemis-
D72 sionsintheirmunicipalityappearnottoloseelectoralsupportatsubsequentelections;thisiscontraryto
whatwouldbeimpliedbyasimple(biased)ordinaryleastsquaresregression.Moreover,IVpointestimates
Keywords: arepositive,albeitinsignificantatstandardlevels;thiscouldbeduetothepossibilityofsomesupportof
Climatechangemitigation citizensforemissionreductioncommitments.Finally,strongheterogeneityinsocio-economicanddemo-
Emissionreduction graphiccharacteristicsisfound,withsupportoftheCoMbeingmorepronouncedinwealthierandyounger
Voting cities.
CovenantofMayors ©2017ElsevierB.V.Allrightsreserved.
1. Introduction Actionforclimatechangemitigationrequiresstrongandunwa-
veringpoliticalcommitment.Forthisreason,internationalinitiatives
Mountingscientificevidenceonthecausesofclimatechangeand aregrowingtoovercomecoordinationfailuresandstrengthenjoint
itspotentialconsequenceshasincreasedtherelevanceofenviron- efforts for emission reduction, including international coalitions,
mentalpoliciesforgreenhousegas(GHG)emissionreduction(see, voluntaryagreementsandpeer-reviewprocesses.
forinstance,therecent‘EnergyUnion’priorityoftheEuropeanCom- TheUNFrameworkConventionforClimateChangeisthewidest
mission(EC,2015)).However,nationalandlocalactionsforclimate coalitionforemissionreduction.Itinvolvesmorethan195countries
change mitigation generate mostly global and future benefits (dif- in the world, including the largest CO emitters, all with intended
2
ferentlyfromlocalpollutionattenuation),underminingthepolitical nationally determined contributions submitted to the conference
incentiveforit. of Parties in Paris (December 2015). The analysis of efficiency and
This paper tests one source of political incentives for climate stabilityoflargecoalitionhasbeenstudied,forexample,byBréchet
changemitigation:theexistenceofanelectoraldividendformay- etal.(2011).
orsthatcommittoreduceemissionsatthelocallevel.Inparticular, Local alliances supporting voluntary agreements at different
it estimates whether the political commitment to reduce munici- administrative levels have also been launched to involve sub-
pal emissions, undertaken within the EU Covenant of Mayors, has nationalandlocalactors,focusingonbottom-upandmultilevelgov-
a positive or negative impact on electoral support at subsequent ernance approaches. They generally rely on the voluntary engage-
electionsforincumbentmayorsrunningforasecondterm. mentoflocalauthoritieswithoutanylegallybindingcommitment.
Internationalexamplesoflocalalliancesforemissionreduction
havebeengrowingsincetheearlynineties.Thebasisofthisstudy
is the EU Covenant of Mayors (CoM). It was formed in 2008 and
*Correspondingauthor.
E-mail address:[email protected](S.Martelli). itisnowthemainstreamEuropeanmovementforlocalauthorities
http://dx.doi.org/10.1016/j.ecolecon.2017.06.035
0921-8009/©2017ElsevierB.V.Allrightsreserved.
28 S.Martellietal. /EcologicalEconomics144(2017)27–35
voluntarily committing to meet and exceed the European Union commitments. Citizens living in poorer cities seem not to support
20% GHG emission reduction target by 2020 (as laid down in the emission reduction commitments, contrary to richer cities. A sim-
Europe 2020 strategy), under their mandate (Cerutti et al., 2013). plehierarchyofneedsapproachpredictsenvironmentalconcernsto
OtherinternationalnetworksaretheICLEI-LocalGovernmentsfor benegativelyrelatedtoeconomicconditions,asKahnandKotchen
Sustainability, established in 1990 (ICLEI, 2014); the U.S. Confer- (2011)foundlookingatthefrequencyofrelatedsearchesontheweb.
enceofMayorsClimateProtectionAgreement,launchedin2005(The Therestofthepaperisorganisedasfollows.Thedetailsofthe
US Conference of Mayors, 2007); and the UN Compact of Mayors, Covenant of Mayors are reported in Section 2. Section 3 describes
anagreementbetweenexistingcitynetworkscreatedin2014(UN the dataset while Section 4 provides the empirical framework of
Headquarter,2014).Interestforlocalclimatepoliciesisalsogrowing analysis and the identification strategy. Potential sources of endo-
inChina,asanalysedbyZhengetal.(2014). geneityareassessedbasedontheliteratureonthepoliticaleconomy
Awideeconomicliteratureontheattractivenessandeffective- ofenvironmentalpolicies.ResultsarereportedinSection5.Section6
nessofenvironmentalvoluntaryagreementsbetweenpublicauthor- concludes.
itiesandtheprivatesectorhasbeendeveloped.CarraroandLvłque
(1999), Croci (2006), OECD (2003) are notable example develop-
2. TheCovenantofMayorsInitiativeandSimilarPolicies
ing an extensive assessment of the effectiveness, efficiency and
adequacyofpublic-privatevoluntaryagreementscomparedtotradi-
The Covenant of Mayors (CoM) is the mainstream European
tional“commandandcontrol”policies.Boththepolicytoolsusedby
movement involving local authorities voluntarily committing to
CoMcitiestoinducechangesinprivatesectoremissionsaswellas
meetandexceedtheEurope2020targetofGHGemissionreduction
theireffectivenessrestbeyondthescopeofthepresentwork.
(−20%by2020),undertheirmandate.2ItwaslaunchedbytheEuro-
Thescopeofthecurrentworkistolookatmultilevelgovernance
peanCommissionaftertheadoptionoftheEUClimateandEnergy
approaches to climate change mitigation within the public sector.
Package, in 2008, to encourage the implementation of sustainable
Theanalysisofrelevantdriverssupportingthepoliticalengagement
energy policies at the local level. Based on the subsidiarity princi-
ofmayorsforclimatemitigationisofkeyimportancegiventhecon-
ple, different institutional levels are invited to cooperate in order
sistent contribution of cities to global emissions and the limited
tolocallyaddresstheglobalchallengeofclimatechange.Inpartic-
ability of national and supranational authorities to act at the local
ular, mayors willing to formally commit to reduce emissions need
level,inlinewiththeprincipleofsubsidiarity(Collier,1997).Inpar-
to adopt and implement a Sustainable Energy Action Plan (SEAP)
ticular,theelectoralsupportforlocalcommitmentstoreduceGHG
withintheirmandates,whoseconsistencyisensuredbythetechni-
emissionsisunderscrutiny.
calassessmentoftheEuropeanCommission-JointResearchCentre
Electoralincentives(inadditiontotheactivityoflobbygroups)
(JRC).
seemtoexplainthestringencyofenvironmentalpolicies(Fredriks-
In2015thenewCovenantofMayorsforClimateandEnergywas
son et al., 2005) and the level of environmental spending (Bouton
announced by European Commission Directorate-General Energy.
et al., 2013; List and Sturm, 2006). Environmental preferences of
TheCoMmandatewasextendedtocoverbothmitigationandadap-
citizenshavealsobeenempiricallyproventobeadriverofnewemis-
tation,andatimehorizonto2030(inlinewiththeEU’s2030climate
sionreductionpoliciesatthelocallevel(KahnandMorris,2009)and
andenergypackage).
cityclimateplanning(Millard-Ball,2012).Onthecontrary,theyseem
InItalianmunicipalities,theCoMisthebestknownandrecog-
not to be a determinant of carbon emissions for firms (Cole et al.,
nized action primarily targeting the reduction of GHG emissions.
2013;Matisoff,2013).
ThisisevidentfromFig.1,reportingthestandardisedfrequencyof
Thepoliticalcommitmenttoreduceemissionsmayhaveaposi-
GooglesearchesinItaly(overallaverageandtimetrend)forthekey
tiveeffectonelectoralresultsifthereisademandforenvironmental
words“CovenantofMayors”(solidline);“PattodeiSindaci”,theItal-
policies(andifthecommitmentiscredible).Wheninterviewed,EU
iantranslationforCoM(dash-dottedline);any“Other”similarpolicy
citizensaskforstrongerenvironmentalprotection(EC,2007,2011).
discussedbelow(dashedline,whichisflatatzero).Stronginterest
Forinstance,inthe2011SpecialEurobarometersurvey,EC(2011),
intheCovenantofMayorsstartsinJanuary2010andanyalternative
63% of EU citizens stated that the European Union is not doing
initiativeisnegligiblecomparedtoit.
enough to use natural resources efficiently, despite the ambitious
The number of mayors participating in the CoM has increased
Europe2020targets.But,dostatedpreferencestranslateintovoting
overtime.Therewere5049signatoriesover47participatingcoun-
behaviouratthemunicipallevel?Areemissionreductioncommit-
triesbyMarch2013(4916signatoriesintheEU-28),corresponding
ments,undertakenwithinaglobalalliance,rewarded(orpunished)
toapopulationof187million(160millionintheEU-28),seeCerutti
byvotersatthelocallevel?
etal.(2013).Bytheendof2013,theygrewto6186signatoriescorre-
The CoM provides a unique quasi-experimental setup to assess
spondingto213millioninhabitantsandtoanoverallGHGreduction
the electoral impact of emission reduction commitments at the
commitmentof27%(7%higherthantheminimum).
local level. Self-selection of mayors into the CoM and the result-
ItalyistheEuropeancountrywiththehighestparticipationrate,
ingendogeneitybiasissolvedherethankstoanintention-to-treat
both in terms of number of signatories and inhabitants involved,
instrumental variable approach. An exogenous instrument for the
makingitthemostsuitedStatetoperformacounterfactualanalysis.
effort required to join the movement is available through the so-
3355ItalianmunicipalitiesjoinedtheCoMbytheendof2013,indi-
called‘CovenantTerritorialCoordinators’(CTCs),asdescribedinthe
viduallyorjointlywithothermunicipalities,correspondingto50%
followingsections.Thisallowstoestimatetheimpactofjoiningthe
of total signatories in Europe and 41% of Italian municipalities. As
CoMontheelectoralresultofanincumbentmayor1,i.e.toanswer
aresult,39millionItalianswerecoveredbytheCoMattheendof
thequestion:“DoestheCoMincreasetheapprovalofacandidate?”
2013,correspondingto65%ofthecountrypopulation.
The electoral effect of voluntary participation in the CoM, for
TheadhesiontotheCoMrequiresanoteworthyadministrative
theaveragecity,isfoundtobenull.However,socio-economicand
and technical effort for the municipal administration. Signatories
demographic heterogeneity are found to be key in explaining the
needtocomputetheirBaselineEmissionInventory(BEI)andsubmit
presence or absence of individual support for emission reduction
aSEAP,specifyingconcreteactionstoreduceGHGemissionsby2020,
1 TheimpactoftheCoMonactualGHGemissionreductionremainsbeyondthe
scopeofthepresentstudy. 2 www.covenantofmayors.eu;accessed01/09/2015.
S.Martellietal. /EcologicalEconomics144(2017)27–35 29
Fig.1. FrequencyofGooglesearchesrelatedtoenvironmentalpoliciesinItaly.StandardisedfrequencyofGooglesearchesasprovidedbyGoogleTrend(www.google.com/trends)
fortheperiod06/2008–12/2013.“Others”correspondstoanyofthealternativeinitiativeslistedinSection2.
togetherwithanimplementationstrategy.Theeffortrequiredtojoin emissionreduction.EvenifitisasignallingtoolforGHGemis-
theCoMhasbeenrapidlyconsideredtobearelevantbarriertoentry, sionreductionpolicies,itisnotwidelyknowninItaly,when
asreportedtotheauthorsofthispaper,inprivatecommunications, comparedtotheCoM.
byseveralmayorsaswellasexpertsworkingontheinitiative. • TheC40CitiesClimateLeadershipGroup(www.c40.org).This
Thus,theECsupportedtheorganizationof‘CovenantTerritorial istheonlyinitiativewiththesameaimastheCovenantofMay-
Coordinators’(CTCs)toprovidefinancial,technicalandadministra- ors.However,onlythreeItaliancitiesareparticipatinginthe
tive support to decrease the costs of adhesion to the CoM and to initiative(Milan,RomeandVenice)andnoparticularcommit-
furtherpromotetheinitiative.CTCsaresub-nationaldecentralised mentisrequiredforit.Moreover,C40wascreatedin2005,but
bodiestargetingthemunicipalitieswithintheiradministrativebor- therelatedaward,theuniquesignallingtoolintheinitiative,
ders(e.g.provinceorregion).TheyfrequentlydeveloptheBEI(witha wasestablishedin2013anditwasnotassignedtoItaliancities
uniformapproach)andoffertechnicalsupporttopreparetheSEAPs, inthefirstyear.Thus,itisnotasubstitutesignallingtoolfor
seeCovenantofMayorsOffice(2012b). mayorsinthepresentsample;
The active involvementof provincesand regionsthrough CTCs, • TheUnder2MOU(under2mou.org).Thisisagroupofstates
whichplayapivotalroleinenticinglocalauthoritiestojointheini- andregionscommittingtowardsemissionsreduction“tohelp
tiative,wasconsideredtobethereasonforthesuccessoftheCoMin galvanizeactionatCOP21”.Evenifitsignalsemissionreduction
Italy(CovenantofMayorsOffice,2012a).Onaverage,anItalianCTC commitments,citiescannotjoinit.
coordinates26municipalities,butfiguresvarysignificantlyassome
Thus,noneoftheaboveinitiativesisacompetitivepolicywith
provincescoordinatemorethan100municipalities,suchastheCTC
respecttotheCoM,asshowninFig.1.
oftheProvinceofBergamo(186)andtheProvinceofL’Aquila(109).
Moreover,citizens’awarenessofthemayor’ssigningoftheCoM
ThefirstCTCswerecreatedsoonafterthelaunchoftheCoM,inmid-
(andrelatedcommitments)iswitnessedbymanypaperandonline
2009.Attheendof2013,therewere99CTCsinItaly,spreadoverthe
records. Indeed, mayors appear to communicate extensively their
wholecountry.
adhesion to the CoM and to promote their engagement. The news
Alternative initiatives are available in Italy to signal the green
isregularlyreportedinlocalnewspapers,localnewscastandother
commitmentofmayors;however,theyarenotaspopularandthey
televisionprogrammes,aswellasinstitutionalvideoson theweb,
areexpectedtohaveanegligibleeffectonvotingchoices,whencom-
withnogeographicalbiasinthecountry.3
paredtotheimpactoftheCovenantofMayors.Moreover,theyare
eithernotspecificallytargetingGHGemissionreductionsortheydo
3. Data
notrequireaformalcommitmenttoimplementenvironmentalpoli-
cies.Inparticular,thefollowinginitiativesaretheclosesttotheCoM This section describes the cross-section dataset constructed for
targets: the counterfactual evaluation. The unit of analysis is the Italian
municipality,indicatedbytheindexi.Itisthesmallestadministra-
tivedivisioninItaly,governedbyamayorwithpoliticalpoweron
• European Energy Award (www.european-energy-award.org). themanagementoflocalinfrastructuresandenvironmentalissues.4
There were 8090 municipalities in Italy in December 2013 with
This is a quality management system and certification pro-
anaveragepopulationof7500,rangingfrom30inhabitantsto2.6
grammetosupportcommunitieswillingtocontributetosus-
million.Mayorsareelectedevery5yearsandtheycanserveonlytwo
tainable energy policy and urban development. Even if the
termsconsecutively.
certification is a signalling tool, it mainly focuses on techni-
The treatment variable is the decision to participate into the
calsupportanditisparticularlypopularinGermany,whileit
Covenant of Mayors initiative, identified with the signing date of
clearlydidnottakerootinItaly,apartfromfewcases;
• Res Champions League for Europe (www.res-league.eu) and thecommitmenttoreduceGHGemissions.Thisisindicatedassigni,
thecorrespondingItalian“CampionatoSolare”.Thisismainly
focusedon“spreadingtheuseofsolarenergy”and“showingthe
potentialofasustainableenergymodel”; 3 Thiscanbecheckedbysearchingfor“PattodeiSindaci”underthevariousonline
• SustainableEnergyEuropeAward(eusew.eu).Thishighlights channels,suchasYouTube,Twitter,GoogleNews.
4 Otherresponsibilitiesdelegatedtothemunicipaladministrationincludetheman-
publicandprivateinitiativesthatareactivelycontributingto
agementoftraffic-relatedissues;localpolice;commercial,cultural,sport,education
the EU’s 2020 energy and climate objectives, including CO2 andtrainingactivities.
30 S.Martellietal. /EcologicalEconomics144(2017)27–35
Table1 years),8 municipalities that were merged with neighbouring ones
Yearoffirstelectionandyearofcommitmenttoreduceemissionsinthesampleand and municipalities where only one candidate was present on the
Italy.
ballot.
Yearoffirstelection Totalsample TotalItaly Table2comparesmeanvaluesofrelevantvariablesinthesample
Signingyear 2004 2005 2006 2007 2008 municipalitieswiththeonesofthewholeofItaly.Thesemeanvalues
areverysimilar,raisingnomajorconcernsontherepresentativeness
Nosign 1001 91 250 161 125 1628 3868
ofthesamplewithrespecttotheItalianpopulationofmunicipalities.
2008 5 0 1 2 2 10 20
2009 81 7 19 12 12 131 309 Theinstrumentalvariableconsideredinthispaperisadummy
2010 195 13 43 39 31 321 785 variableforthepresenceofaCovenantTerritorialCoordinator(CTC),
2011 129 40 62 51 26 308 770 asintroducedbysomeprovincesandregionstohelpallmunicipal-
2012 146 19 34 26 14 239 630 itiesintheirareatojointheCoM.Thectcdummyequals1ifaCTC
2013 169 13 46 24 24 276 678
wasavailableforamunicipalitybeforeitsmostrecentelectionand
Totalsample 1726 183 455 315 234 2913 –
TotalItaly 3860 595 1213 828 564 – 7060 itis0otherwise.Asaresult,bothmunicipalitiesthatjoinedtheCoM
afterandbeforethecreationofaCTCwereclassifiedassupportedby
Numberofcitiesinthesamplebyyearoffirstelectionandyearofadhesiontothe
aCTC.Theexogeneityoftheinstrumentliesintheimpossibilityof
CoM.Marginaldistributionsofelectionyearandcommitmentyearinthesampleare
comparedwiththecountrydistribution. asinglemayortoinfluencethedecisiontocreateaCTCbyahigher
levelofgovernment.
which equals 1 for those municipalities whose mayor signed the The number of signatories in the sample (municipalities that
CoMbeforethemostrecentelectionandafterthepreviousone,5 joined the CoM) and the number of municipalities covered by a
yearsearlier.Itis0otherwise.Detailedadministrativedataonthe CTC are reported in Table 3, together with the totals for Italy. The
CovenantofMayorsmovementandparticipatingmunicipalitiesare sample is well balanced with respect to the Italian situation for
providedby theEuropean Commission - DG Joint Research Centre municipalitieshavinganelectionintheyears2009–2013.
andtheyarepubliclyavailableon-lineontheCovenantofMayors Thegeographicaldistributionofsamplesignatoriesandterrito-
Office’sdedicatedwebsite.5 rialcoordinatorsisreportedinFig.2,wheremunicipalitiesinwhite
The number of municipalities participating in the CoM move- areoutofthesample,yellowmunicipalitiesarethoseneithercov-
ment is reported in Table 1, showing the joint distribution of the eredbyaCTCnorparticipatingintheCoM;lightgreenmunicipalities
year of first election and the year of commitment to reduce GHG decidedtoparticipateintheCoMwithoutanyCTC;orangemunici-
emissionswithintheCoM.ThefirstmunicipalityjoinedtheCoMin palitiesdecidednottojointheCoMeveniftheyreceivedtheofferto
2008.Thenumberofnewsignatoriespeakedin2010and2011,with besupportedbyaCTC;darkgreenmunicipalitiesjoinedtheCoMin
morethan300mayorsjoiningthemovement.Theoverallpatternfor areassupportedbyaCTC.
Italy,includingallmunicipalitiesthatrunanelectionintheperiod Control variables aim at capturing economic conditions at the
2004–2008,parallelsthesampletrend. locallevelandelectoralcharacteristics.Averageper-capitaincome
The outcome variable ofinterestis theshare of votescollected atcitylevelwasincludedinlogarithm.Thedatasetisbasedontax-
bycandidatesrunningforre-electionbetween2009and2013;this ableincomeatmunicipallevel(excludingtax-deductibleexpenses
variable is indicated as vote. In particular, vote equals the frac- suchasprivatepensionschemes,lifeinsurancepremium,expenses
i i
tionofvalidvotesobtainedbytheincumbentcandidateinthefirst relatedtojobreasons),asreportedfortheItalianannualtaxdecla-
roundofre-election.6 Dataonelectoralresultswereobtainedfrom ration(IRPEF).DataispubliclyavailableonthewebsiteoftheItalian
thepubliclyavailableon-linedatasetcalled“ArchivioStoricodelle MinistryofEconomyandFinance(MEF,2015-09-30).
Elezioni”,whichprovidesresultsforallItalianregionsexceptValle The number of eligible voters (in logarithm) and turnout rate
d’Aosta,TrentinoAltoAdige,FriuliVenezia-GiuliaandSicily7;these (in percentage points) are included as further controls computed
16 regions include 7060 municipalities with available data for the ondatafromtheItalian“ArchivioStoricodelleElezioni”(Ministero
periodconsidered. dell’Interno, 2014-10-31). Moreover, dummy variables identifying
Only a subset of these municipalities was considered in the municipalitieswith3candidates,4to9candidatesand10ormore
analysis,duetothefollowingselectioncriteria.Outof7060munic- candidatesontheballotareincludedtocontrolfornon-lineareffects
ipalitiesreportingelectoraldatatothenationaldatabase,only3195 ofthenumberofcandidates.Finally,theshareofvotescollectedby
presented mayors running for a second term. The sample reduced greenpartiesatnationalelectionsof2006isincludedtocontrolfor
further to 2913 municipalities due to the exclusion of cities with baselinegreenpreferences.Theshareofyoungvoters,proxiedbythe
abnormal duration of the electoral mandate (i.e. different from 5 shareofinhabitantsaged18–30,isalsoincluded.
4. ModelandInstrument
5 http://www.covenantofmayors.eu.
6 InItaly,theelectoralsystemtoelectmayorsandmembersofthecitycouncil Thissectionpresentstheidentificationstrategy.Itdescribesthe
dependsonthesizeofthecity.Atwo-roundelectoralsystemisusedinmunicipali- referencemodel,potentialsourcesofendogeneity,theinstrumental
tieswithmorethan15,000inhabitants.Thefirstroundisusedtoselectthefirsttwo
variableapproachandevidenceonthestrengthoftheinstrument.
candidates,whoarethenvotedintoofficeinthesecondround(aftertwoweeks).The
secondroundisnotexecutedifonecandidatewinsover50%ofthevoteinthefirst
round.Onthecontrary,directelectionsareheldinsmallermunicipalities,withupto 4.1. ReferenceModel
15,000inhabitants.Here,asecondroundisorganisedonlyincaseoftie.Inbothcases,
voterscanexpressadirectchoiceforthemayororanindirectchoicevotingforoneof The goal of the paper is to estimate the impact of joining the
thepartiesofthecandidate’scoalition.Mayorsordinarilyserveatermsoffiveyears.If CoMontheshareofvotescollectedbyincumbentmayorsatlocal
amayorresigns,diesorisoustedfromofficeaftermorethanhalfthemunicipalcoun-
elections.Thereferencemodelis:
cillorssteppeddown,anearlymunicipalelection(forthemayorandforallmunicipal
councillors)iscalled.Finally,allmayorswereallowedtoserveforamaximumoftwo
cvoenrysescmutailvlecitteiremss(winitohffi<c3e0u0n0tiiln2h0a1b3i.taAntths)irsdtacrotninsgecfurtoimvettheermmhuansicbiepeanleallelocwtioendsfoorf votei=h•signi+b(cid:2)Xi+q•vote prei+ei (1)
2014,andthusbeyondtheperiodofanalysisconsideredhere.
7 See http://elezionistorico.interno.it/. Valle d’Aosta, Trentino Alto Adige, Friuli
Venezia-GiuliaandSicilyoptedforthecreationoflocalelectoralofficesandthusthey 8 Thisdiscrepancycanbetheresultofadministrativechangesorcriminal-related
donotreportelectoralresultstothecentraladministration. events.
S.Martellietal. /EcologicalEconomics144(2017)27–35 31
Table2
SummaryofvariablesandmeancomparisonbetweensamplemunicipalitiesandItaly.
Sample Italy Description
Vote(%) 54.21 – Outcome:shareofvotescollectedbycandidatesrunningforasecondterm
(15.40) –
Sign 0.11 0.11 Treatment:dummyequalto1formunicipalitiesthatsignedtheCoM
(0.31) (0.32)
ctc 0.25 0.26 Instrument:dummyequalto1formunicipalitiescoveredbyaCTC
(0.43) (0.44)
Controls
Income(kEUR) 10.66 10.53 Averageper-capita(taxable)incomeatcitylevel
(3.08) (3.09)
Electorate 6.28 6.20 Number(thousand)ofeligiblevotersatsecondelection
(45.45) (35.85)
Turnout(%) 74.74 74.20 Turnoutrateatsecondelection
(9.48) (9.60)
Uniquecand 0.00 0.06 Dummyformunicipalitieswithonly1candidateontheballot(excludedfromthesample)
(0.00) (0.24)
2candidates 0.51 0.44 Dummyformunicipalitieswith2candidatesontheballot(droppedintheregression)
(0.50) (0.50)
3candidates 0.28 0.27 Dummyformunicipalitieswith3candidatesontheballot
(0.45) (0.44)
4–9candidates 0.20 0.22 Dummyformunicipalitieswith4to9candidatesontheballot
(0.40) (0.42)
10+candidates 0.0041 0.0045 Dummyformunicipalitieswith10ormorecandidatesontheballot
(0.0641) (0.0672)
Greenvoters(%) 1.90 1.82 Shareofvotescollectedinthemunicipalitybygreenpartiesatnationalelectionsof2006
(2.05) (1.93)
Age18–30 0.14 0.14 Shareofinhabitantsaged18–30
(0.02) (0.02)
Convergencearea 0.19 0.19 DummyformunicipalitiesinSouthernItaly
(0.39) (0.39)
Observations 2913 7060
Meanoftreatmentvariable,instrumentvariableandcontrolvariablesinthesampleandinItalyarereported,withstandarddeviationinparentheses.
wherevote istheshareofvotescollectedbytheincumbentmayor AssumingthatcitizenssupportGHGemissionreductionpolicies
i
atelections,sign isthetreatmentdummyvariable,X isavectorof andskilledmayorsaremorelikelytojointheCoM,preferencesfor
i i
controlvariablesandvote_pre isthepercentageofvotescollectedby the CoM will be overestimated in a simple ordinary least squares
i
theincumbentmayorinthepreviouselection(seeSection3).histhe
mainparameterofinterest,whichcharacterisestheimpactofsign
i
on the expected value of votei, the b(cid:2) vector resumes the effect of
controlvariablesandeiistheregressionerror.Thereferencemodel
isestimatedbothinitsstaticanddynamicspecifications,wherethe
staticmodelcorrespondstoqequalto0.
The identification of the causal effect of joining the CoM (sign)
onvotingchoices(vote)requirestodiscussthepossibleendogeneity
biasdeterminedbytheself-selectionofmayorsintotheprogramme.
Individualskills,personalpreferencesorstrategicbehaviourmight
be exploited in different policy areas (which cannot be controlled
for)andmayleadtotheparticipationintheCoMaswellastoother
outcomesappreciated(oropposed)byvoters.Thiswouldresultin
apositive(ornegative)biasoftheestimatedenvironmentalprefer-
ences.Forexample,askilledmayorthatdecidestojointheCoMmay
alsomanageothercityprojectswell.Asaresult,thechangeinvot-
ingchoicesbetweentwoelectionscapturesthecumulativeeffectof
supportingtheCoMandbeingaskilledmayor.
Table3
Signingmunicipalitiesandsupportfromterritorialcoordinatorsinthesampleand
Italy.
CTC(territorialcoordinator) Totalsample TotalItaly
Sign Absent(0) Present(1)
No(0) 2118 471 2589 6248
Yes(1) 77 247 324 812
Totalsample 2195 718 2913 –
TotalItaly 5219 1841 – 7060
NumberofcitiesbysigningstatusandCTCstatusinthesample.Totalsforthesample Fig.2. GeographicaldistributionofcitiesinthesamplebysigningstatusandCTC
andforItalyarereportedforcomparativespurposes. status.
32 S.Martellietal. /EcologicalEconomics144(2017)27–35
(OLS) regression using specification Eq.(1): mayors collect more
votesinsubsequentelectionspartiallyinrecognitionoftheirgood
performanceandpartiallyasaresultofthewillingnessofcitizens
toreduceemissions.Sinceitisnotpossibletomeasureandcontrol
fortheskillsofamayor,sign canbecorrelatedwithunobservable
i
characteristics collected in ei. In this case, OLS would yield biased
results.
Theinstrumentalvariabletechniqueallowssolvingendogeneity
biasfromself-selectionofmayors(relatedtounobservablecharac-
teristics, such as skills and abilities of mayors) by making use of
exogenous variations in the incentives to join the CoM movement.
This requires an instrument that induces mayors to join the CoM
(validity of the instrument), but that is independent of any other
unobservablevariablethatiscorrelatedwiththeelectoraloutcome
vote (exclusionrestriction).Inshort,theinstrumentmusthavean
i
effectonvote onlythroughsign.
i i
The support for the adhesion to the CoM, offered by some
provincesandregionstoallcitiesintheirarea(creatingaCovenant
Territorial Coordinator, ctci), is a good instrument because it is Fig.3. DistributionofthetimingofcityadhesiontotheCoMwithrespecttothedate
expectedtoverifybothconditions(seedefinitioninSection3). ofCTCcreationintheirarea.
An instrumental variable (IV) estimation is performed. The IV
approachisappropriateforthenon-dichotomousoutcomevariablein Noanticipatoryeffect(mayorsanticipatingtheiradhesiontothe
thesecondstageandgenerallypreferredtoothertechniques,which CoM because they foresee the imminent creation of a CTC) can
require stronger specification assumptions (Wooldridge, 2002). bias the results because municipalities joining the CoM before the
Resultsarereportedforerrorsclusteredattheprovincelevelbutthey creationofaCTCareconsideredascoveredbyit.
arerobusttoclusteringatregionallevel.Inlightofthedifferentgeo- Finally, omitted variables correlated with both the creation of
graphicallevelsinvolvedintheanalysis(CTCsarecreatedbyregions a CTC and the local electoral outcome need to be ruled out. For
andprovinces),bothspecificationoferrorsclusteringaretested. example, political parties may be the drivers of both environmen-
tal policies and other local policies, such as land use regulation
(Solé-OlléandViladecans-Marsal,2013).Ifpoliticiansinthetwolev-
4.2. IdentificationStrategy
els of government belong to the same party, then the decision to
ThecreationofaCTCisexpectedtoimpacttheprobabilitytosign createaCTC(atprovinceorregionallevel)mightbecorrelatedwith
upfortheCoMbydecreasinginformative(CTCsactivelypromoteto other policy outcomes relevant for the electoral result (municipal
mayorstheCoM),administrativeandfinancialcosts.Givenasetof level).However,mostcandidatesinthesampleareassociatedwitha
incentivestojointheCoM,includingofficemotivation,policymoti- “listacivica”,alistwithnoofficialconnectionwithanationalpolit-
vation and political income (Persson and Tabellini, 2002), mayors icalparty;thisrulesoutpossibleinfluencesviaassociationstothe
face incomplete information and a budget constraint for available samepoliticalparty.
funds,manpower,technicalcapabilities,andpersonalskills.Thecre- GeographicalcoverageofthesampleincludesItalianmunicipal-
ationoftheCTCrelievesthemunicipalbudgetconstraint,allowingall ities only, assuring high homogeneity in national institutions and
mayorstojointheCoMatalowercost,byexploitingtheservicesand electoralrules.
fundingprovidedbytheCTC.Mayorsthathavealreadysignedthe Inconclusion,ctciisexpectedtoexogenouslyinfluencethedeci-
CoMcanalsotakeadvantageofanewlycreatedCTC.Thus,thecre- sionofcitiestojointheCoMmovementandnottohaveanydirect
ationofaCTCthatfollowsuponthedecisiontosignisassumedtobe impactonelectoralresults.Theinstrumentalvariable(IV)approach
stillcorrelatedwiththedecisionitself,butnotwithe.Thestrength provides consistent estimates of the causal effect of the volun-
i
oftheinstrumentisdiscussedinsubsection5.1. tarycommitmenttoreduceemissionsonelectoralresults,without
The creation of a CTC is also expected to verify the exclusion analysingthepoliticalprocessleadingmayorstojointheCoM,which
restriction.Thiscannotbeexplicitlytested,butitcanonlybesup- remainsbeyondthescopeofthispaper.
portedbytheoreticalargumentsandempiricalevidence.Firstly,the
creationofaCTCisdecidedbyahigheradministrativelevel(province 5. Results
orregion)anditinvolvestechnicalsupportdirectedtothemunicipal
administrationonly.Ithasnodirectconnectionwiththepopulation Thissectionreportstheestimatedaveragetreatmenteffectand
intheCTCarea(e.g.noconnectionwithlocalpoliticalcampaigns). itsheterogeneitywithrespecttoeconomicanddemographicchar-
Thus,ithasnoimpactonlocalelectionsandontheshareofvotes acteristics. First stage results are reported first; the main result is
personallycollectedbyincumbentmayorsatmunicipalelections. reportedinSubsection5.2;heterogeneityofpreferencesisreported
Moreover,singlemayorscannotinfluencethehigherleveldeci- inSubsection5.3.
siontocreateaCTC,asreportedbyCoMprojectmanagers.Indeed,
Italianregions(provinces)arecomposedbyanaverageof400(70) 5.1. StrengthoftheInstrument
municipalities.Moreover,mayorscannotself-selectintotheCTCby
joiningtheCTCofdifferentregionsorprovinces. Inthissubsectionthestrengthofthectc instrumentismeasured.
i
Fig.3reportsthegraphofthenumberofmunicipalitiesjoining Table4showsthefirststageestimationresultswheretheendoge-
the CoM as a function of the date from the creation of a CTC. The nousexplanatoryvariablesign ofmodelEq.(1)isregressedonthe
i
numberofsignatoriessignificantlyincreasesafterthecreationofa instrument (specification A). Controls X are included in specifica-
it
CTCandnodisplacementeffectisobserved(mayorsdonotappear tionB.BothspecificationsAandBarestatic,whilespecificationCis
todelaytheiradhesiontotheCoMasaresultoftheexpectedcre- dynamic.
ationofaCTC).Thisisinlinewiththeideathatsinglemayorscannot Theinstrumentappearsrelevantinallmodels.Thecreationofa
influencethecreationofCTCs. CTCispositivelycorrelatedwiththedecisiontojointheCoMwith
S.Martellietal. /EcologicalEconomics144(2017)27–35 33
Table4 Mayorscommittingtoreducegreenhousegasemissionsintheir
SupportingauthoritiesandadhesiontotheCoM(firststage). municipalityappearnottoloseelectoralsupportatsubsequentelec-
Specificationregression (A)OLS (B)OLS (C)OLS tionsintheIVestimation;thisiscontrarytowhatwouldbeimplied
ctc 0.34∗∗∗ 0.33∗∗∗ 0.33∗∗∗ byasimple(biased)OLSregression.Moreover,IVpointestimatesof
(0.05) (0.05) (0.05) theeffectofsigningupfortheCoMarepositive.
Log-income −0.02 −0.02 Indeed, IV estimates of specifications (E) and (F) show that the
(0.05) (0.05) commitmenttoreduceGHGemissionsby20%orbeyonddonothave
Log-electorate 0.02∗∗∗ 0.02∗∗∗
anegativeimpactontheshareofvotescollectedatsubsequentelec-
(0.01) (0.01)
Turnout(%) −0.00 −0.00 tions by incumbent mayors, despite the opposite indication given
(0.00) (0.00) byasimple(biased)OLSregression.Onthecontrary,IVpointesti-
3candidates −0.00 −0.00 matesarepositive,albeitinsignificant;thisiscompatiblewiththe
(0.01) (0.01) possibilityofsomesupportofcitizensforemissionreductioncom-
4–9candidates 0.02 0.02
mitments, even if no statistical significance is found at standard
(0.02) (0.02)
10+candidates 0.09 0.09 levels. The increase of standard errors in specifications (D) to (F)
(0.11) (0.11) is a common feature of the IV approach which contributes to the
Greenvoters(%) −0.00 −0.00 lowstatisticalsignificanceofpointestimates(marginallylowerthan
(0.00) (0.00) standardthresholds).Finally,thedynamicspecification(F)doesnot
Age18–30 0.00 0.00
differsignificantlyfromspecification(E).
(0.00) (0.00)
Convergencearea −0.09∗∗ −0.09∗∗ Summarizing, voters do not oppose the commitment taken by
(0.04) (0.04) incumbentmayorstomeetandexceedtheEurope2020targetfor
vote_pre −0.00* emissionreductionatlocallevel.Positivepointestimates,thoughnot
(0.00)
significantatstandardthresholds,couldbeduetosomepreference
Constant 0.03∗∗∗ 0.26 0.29
formayorsjoiningtheinitiative.
(0.01) (0.45) (0.44)
Observations 2913 2913 2913
R2 0.215 0.237 0.238 5.3. Heterogeneity
F 46.77 11.68 11.17
The IV approach yields the average causal effect of signing on
FirststageOLSregressionofsignonctcforthecrosssectionofcitiesinthesample. electoralresultsforthemayorsthatreacttotheinstrument.Inorder
Specification(A)isthesimpleregressionofsignonctc.Specification(B)includesall to test heterogeneity in environmental preferences for GHG emis-
availablecontrolsbuttheshareofvotecollectedbytheincumbentmayoratprevious
sion reduction, the same model is estimated for a partition of the
electionsvote_pre.Specification(C)includesalsothelatter.
Standarderrorsinparentheses.Errorsareclusteredatregionallevel. sample.Thisisbuiltaccordingtothesamplemedianofincomeper
∗ p<0.10. capita,incomegrowthandtheshareofyoungpopulation(aged18–
∗∗ p<0.05. 30)atmunicipallevel.Moreover,theunemploymentrateatthelocal
∗∗∗ p<0.01. laboursystemlevelisusedtocapturecharacteristicsrelatedtothe
labourmarket.Thelocallaboursystemisthelowestavailablelevel
strong statistical significance. Interpreting the relation as a linear ofdisaggregationforunemploymentinItaly.Thisisageographical
probability model, as a first approximation, the creation of a CTC groupingofneighbouringmunicipalitiescomputedaccordingtothe
increasestheprobabilitytojointheCoMbymorethan30points. highestfrequencyofhome-worktripsregisteredinthe2001Census
Moreover,the1ststageF-statisticfortheOLSregressionisalways inItaly.9Itresultsin686locallaboursystems,groupinganaverage
abovethereferencevalueof10,suggestingtheinstrumentnottobe of12municipalities(from1to124)withanaveragepopulationof
weak. almost90,000(from3300to3.7millioninhabitants)andanaverage
sizeof440 km2(from10to3700km2).
5.2. MainResults Table6showstheestimatedmodelinthesub-samplesforincome
andincomegrowth.Table7reportsresultsfortheunemployment
Table5containsasummaryoftreatmenteffectestimates.Ordi- rateandagestructuresub-samples.Totalobservations,thenumber
naryleastsquares(OLS)arecomparedtoinstrumentalvariable(IV) ofsignatoriesandCTCsarereportedinthetableandtheyresultto
estimates. bebalancedbetweensub-samples.Statisticalsignificanceat5%level
The first model estimates a pooled OLS regression on the cross isneverdetected,butacommonpatterncanbeseen.Adverseeco-
section of second elections. OLS estimates are included only for nomicconditions,suchaslowincome(sub-sampleYl),lowincome
comparison purposes, giventhatOLS is possiblybiased due tothe growth(sub-sampleYl)andhighunemployment(sub-sampleUh),
g
endogeneityofsigni.Infact,mayorsvoluntarilychoosingtotakethe are always associated with lower point estimates for the commit-
commitmenttodecreaseGHGemissionshaveunobservablecharac- menttoreduceGHGemission,withrespecttomunicipalitieswith
teristics(e.g.skills)thatarelikelytobedrivingboththeirchoiceto lowerunemployment(sub-sampleUl),higherincome(sub-sample
jointheCoMandsubsequentelectoralresults. Yh) and higher income growth (sub-sample Yh). Overall, economic
g
The second model introduces municipal controls and the third conditionsseemtoinfluencetheelectoralsupportforclimatechange
further includes the share of votes collected at previous elections mitigation:thericherthecity,thehigherthesupportofcitizensfor
by incumbent mayors. The OLS point estimate is always negative climatecommitments.
andsignificantbutincreasingtowardszeroattheincreaseofcon- Theagestructureofthepopulation(Table7)seemstoinfluence
trol variables included in the model. Indeed, the treatment may revealedenvironmentalpreferencesaswell.Ahighershareofyoung
stillbeendogenousdespitecontrollingforavailablesocio-economic voters(sub-sampleYoung)isassociatedwithapositive(thoughsta-
characteristicsatmunicipallevel. tistically not significant) effect of the CoM on voting. Conversely,
When applying the unbiased IV approach, the estimated effect a relatively older electorate (sub-sample Old) is associated with a
ofsigningontheshareofvotescollectedbyincumbentmayorsat nearlyzeroeffectoftheCoMonelectoralsupport.
elections is never statistically different form zero (Table 5). Point
estimatesforIVmodelsincludingcontrolvariablesarepositiveand
equaltoabout3%,seecolumns(E)and(F).Thedifferencebetween
OLSandIVestimatesdocumentstheendogeneitybias. 9 TheanalysisisperformedbyISTAT,theItalianInstituteofStatistics.
34 S.Martellietal. /EcologicalEconomics144(2017)27–35
Table5
Impactofsignonvoteshare.
Specificationregression (A)OLS (B)OLS (C)OLS (D)IV (E)IV (F)IV
Sign −4.62∗∗∗ −2.11∗∗ −1.70* −1.85 2.91 3.58
(1.19) (0.96) (0.95) (2.26) (2.37) (2.27)
Log-income 2.68 1.98 2.84 2.13
(1.71) (1.67) (1.83) (1.78)
Log-electorate −1.30∗∗∗ −0.63* −1.42∗∗∗ −0.74*
(0.38) (0.37) (0.39) (0.38)
Turnout(%) −0.21∗∗∗ −0.20∗∗∗ −0.18∗∗∗ −0.17∗∗∗
(0.04) (0.04) (0.04) (0.04)
3candidates −8.82∗∗∗ −8.59∗∗∗ −8.88∗∗∗ −8.65∗∗∗
(0.58) (0.56) (0.59) (0.56)
4–9candidates −16.17∗∗∗ −16.06∗∗∗ −16.51∗∗∗ −16.41∗∗∗
(0.93) (0.88) (0.93) (0.87)
10+candidates −18.46∗∗∗ −18.95∗∗∗ −19.39∗∗∗ −19.94∗∗∗
(4.07) (4.25) (4.32) (4.52)
Greenvoters(%) 0.22∗∗ 0.23∗∗ 0.24∗∗ 0.26∗∗
(0.10) (0.10) (0.11) (0.11)
Age18–30 −0.26 −0.15 −0.27 −0.16
(0.19) (0.18) (0.19) (0.19)
Convergencearea −1.25 −1.92 −1.10 −1.78
(1.32) (1.29) (1.34) (1.30)
vote_pre 0.20∗∗∗ 0.20∗∗∗
(0.02) (0.02)
Constant 54.72∗∗∗ 55.76∗∗∗ 48.49∗∗∗ 54.42∗∗∗ 52.07∗∗∗ 44.40∗∗∗
(0.51) (15.92) (15.74) (0.56) (17.05) (16.70)
Observations 2913 2913 2913 2913 2913 2913
R2 0.009 0.243 0.271 0.005 0.234 0.260
EstimationoftheimpactofsigningtheCoMontheshareofvotescollectedatsubsequentelectionsforthecrosssectionofcitiesin
thesample.Specification(A)isthesimpleOLSregressionofvoteonsign.Specification(B)and(C)includeallavailablecontrolswith
andwithoutthelaggedvotesharevote_pre,forthesimpleOLSregression.Specification(D)isthetwostageregressionofvoteonsign,
instrumentedbyctc.Specification(E)includesadditionalcontrolsandspecification(F)includesalsovote_pre.
Standarderrorsinparentheses.Errorsareclusteredatregionallevel.
∗ p<0.10.
∗∗ p<0.05.
∗∗∗ p<0.01.
Table7
Table6 Heterogeneousimpactonvoteshare-unemploymentandagestructure.
Heterogeneousimpactonvoteshare-income.
Subsample Uh Ul Young Old
Subsample Yh Yl Yh Yl
g g Sign 1.78 4.86 4.36 0.19
Sign 8.58* 0.83 3.00 1.17 (2.55) (7.07) (3.02) (3.19)
(4.47) (2.61) (3.04) (3.24) Log-income 5.87∗∗∗ −0.68 6.41∗∗∗ 0.08
Log-income −0.60 5.51∗∗∗ 3.99∗∗ 2.74 (1.79) (2.50) (2.10) (2.33)
(4.11) (2.03) (1.92) (2.53) Log-electorate −2.06∗∗∗ −0.80 −0.82* −1.49∗∗∗
Log-electorate −1.15∗∗ −1.78∗∗∗ −0.72* −1.90∗∗∗ (0.56) (0.65) (0.49) (0.51)
(0.57) (0.54) (0.42) (0.57) Turnout(%) −0.18∗∗∗ −0.21∗∗∗ −0.16∗∗∗ −0.24∗∗∗
Turnout(%) −0.06 −0.22∗∗∗ −0.19∗∗∗ −0.21∗∗∗ (0.05) (0.07) (0.05) (0.06)
(0.08) (0.04) (0.04) (0.07) 3candidates −7.96∗∗∗ −9.82∗∗∗ −10.49∗∗∗ −7.31∗∗∗
3candidates −9.11∗∗∗ −8.70∗∗∗ −8.77∗∗∗ −9.00∗∗∗ (0.92) (0.78) (0.76) (0.85)
(0.81) (0.89) (0.85) (0.79) 4–9candidates −15.50∗∗∗ −17.53∗∗∗ −19.42∗∗∗ −14.33∗∗∗
4–9candidates −17.23∗∗∗ −16.08∗∗∗ −16.45∗∗∗ −16.09∗∗∗ (1.41) (1.21) (1.26) (1.06)
(1.16) (1.57) (1.40) (1.18) 10+candidates −17.88∗∗∗ −12.24∗∗∗ −10.53∗∗∗ −19.36∗∗∗
10+candidates −21.49∗∗∗ 0.00 0.00 −17.38∗∗∗ (5.01) (2.98) (1.89) (4.70)
(4.42) (.) (.) (4.41) Greenvoters(%) 0.21* 0.09 0.30 0.19
Greenvoters(%) 0.15* 0.34 0.57∗∗ 0.02 (0.12) (0.65) (0.20) (0.13)
(0.08) (0.21) (0.25) (0.12) Age18–30 0.12 −0.56* 0.73∗∗ −0.71∗∗
Age18–30 −0.70∗∗ 0.02 −0.01 −0.61∗∗ (0.23) (0.29) (0.30) (0.36)
(0.30) (0.19) (0.22) (0.27) Convergencearea −0.82 0.72 −0.92 −4.02*
Convergencearea 12.25∗∗∗ −1.42 −3.70∗∗ 3.12* (1.48) (2.20) (1.38) (2.11)
(3.87) (1.34) (1.50) (1.81) Constant 18.92 90.66∗∗∗ 1.98 86.86∗∗∗
Constant 80.25∗∗ 26.91 38.36∗∗ 60.03∗∗ (17.16) (22.67) (20.56) (21.94)
(39.76) (19.01) (18.03) (25.26) Observations 1465 1448 1457 1456
Observations 1457 1456 1457 1456 NumSign 243 65 167 141
NumSign 138 170 104 204 NumCTC 567 110 373 304
NumCTC 278 399 247 430
IVestimationoftheimpactofsigningtheCoMonvoteshareforsub-samplesofcities.
IVestimationoftheimpactofsigningtheCoMonvoteshareforsub-samplesofcities. Sub-samplesarecreatedaccordingtothevalueofselectedvariablesbeingabove(•h)
Sub-samplesarecreatedaccordingtothevalueofselectedvariablesbeingabove(•h) orbelow(•l)thesamplemedian.Specifications(Uh)and(Ul)refertounemploy-
orbelow(•l)thesamplemedian.Specifications(Yh)and(Yl)refertoincomepercapita mentrateontheyearoffirstelection.Specifications(Young)and(Old)includecities
ontheyearoffirstelection.(Yh)and(Yl)refertoincomepercapitagrowthbetween wheretheshareofpeopleaged30orbelowisaboveorbelowthesamplemedian,
g g
firstandsecondelections. respectively(atfirstelection).
Standarderrorsinparentheses.Errorsareclusteredatregionallevel. Standarderrorsinparentheses.Errorsareclusteredatregionallevel.
∗ p<0.10. ∗ p<0.10.
∗∗ p<0.05. ∗∗ p<0.05.
∗∗∗ p<0.01. ∗∗∗ p<0.01.
S.Martellietal. /EcologicalEconomics144(2017)27–35 35
Summarizing, voters prove not to oppose further reduction of didnotreceiveanyspecificgrantfromfundingagenciesinthepublic,
GHG emissions and their voting behaviour seems to be consistent commercial,ornot-for-profitsectors.
with preferences stated in the Eurobarometer survey. The formal
commitmenttoreduceGHGemissionby20%orbeyondhasacausal References
averagepositiveimpactonthelikeabilityofcandidates.Thisisesti-
Bouton,L.,Conconi,P.,Pino,F.,Zanardi,M.,2013.Gunsandvotes.WorkingPapers
mated to be about 3 percentage points of vote share collected at
ECARES2013-39,ULB-UniversitéLibredeBruxelles.Nov.
electionsbyincumbentmayors.However,nostatisticalsignificance Bréchet,T.,Gerard,F.,Tulkens,H.,2011.Efficiencyvs.stabilityinclimatecoalitions:a
isfoundatthe10%level.Moreover,preferencesareheterogeneous conceptualandcomputationalappraisal.TheEnergyJournal32(1),49–76.
with respect to economic and demographic characteristics. Better Carraro,C.,Lvłque,F.,1999.Voluntaryapproachesinenvironmentalpolicy. 1sted.,
FondazioneEniEnricoMattei(FEEM)SeriesonEconomics,EnergyandEnviron-
economic conditions and a high share of young individuals in the ment14,SpringerNetherlands.ISBN978-90-481-5156-1,978-94-015-9311-3.
populationareassociatedwithhigherpositiveeffectsofjoiningthe Cerutti,A.K.,Iancu,A.,Janssens-Maenhout,G.,Melica,G.,Paina,F.,Bertoldi,P.,2013.
CoM. TheCovenantofMayorsinfigures5-yearassessment.TechnicalReport.European
Commission-JointResearchCentre.
Cole,M.A.,Elliott,R.J.,Okubo,T.,Zhou,Y.,2013.Thecarbondioxideemissionsoffirms:
6. Conclusions aspatialanalysis.JournalofEnvironmentalEconomicsandManagement65(2),
290–309.
Collier,U.,1997.LocalauthoritiesandclimateprotectionintheEuropeanUnion:
Thisstudypresentsevidenceinlinewiththehypothesisthatthe puttingsubsidiarityintopractice?LocalEnvironment2(1),39–57.http://dx.doi.
Covenant of Mayors (CoM) can be a self-sustainable approach to org/10.1080/13549839708725511.
CovenantofMayorsOffice,2012a.TogetherforEnergy-efficientBuildings.LaSpezia
strengthen efforts for climate change mitigation at the local level.
Province,Italy.TechnicalReport,EuropeanCommission..September.
The CoM is the mainstream European movement involving local CovenantofMayorsOffice,2012b.2012CovenantCoordinatorsMonitoringReport.
authorities voluntarily committing to reduce emissions by 20% (or TechnicalReportCovenantofMayorsOffice.
beyond) by 2020 under their mandate. The causal effect of the Croci,E.,2006.TheHandbookofEnvironmentalVoluntaryAgreements:Design,Imple-
mentationandEvaluationIssues.vol.43.SpringerScience&BusinessMedia.
adhesion to the CoM on electoral results of incumbent mayors is EC,2007.SpecialEurobarometer195-attitudesofEuropeancitizenstowardsthe
estimatedforasampleofItalianmunicipalities,viaaninstrumental environment:EB68.2.TechnicalReport,TheEuropeanCommission-Directorate–
variableapproach,usinganumberofcontrolvariables. GeneralCommunication,Brussels.
EC,2011.SpecialEurobarometer365-attitudesofEuropeancitizenstowardsthe
Thecentralfindingisthatcitizensdonotopposethecommitment
environment:EB75.2.TechnicalReport,TheEuropeanCommission-Directorate–
toreduceGHGemissionsby20%andbeyondby2020,asrequired GeneralCommunication,Brussels.
toparticipateintheCoM.Indeed,takingthecommitmenttoreduce EC,2015.EnergyUnionPackage-aframeworkstrategyforaresilientEnergyUnion
withaforward-lookingclimatechangepolicy.TechnicalReport,TheEuropean
emissionsatthemunicipalleveldoesnotcausealossofvotesfor
Commission,Brussels.
mayorsatsubsequentelections.Pointestimatesofthegaininelec- Fredriksson, P.G., Neumayer, E., Damania, R., Gates, S., 2005. Environmentalism,
toralvotesarepositivebutstatisticallyinsignificantatthe10%level, democracy,andpollutioncontrol.JournalofEnvironmentalEconomicsandMan-
agement49(2),343–365.
possibly because of the higher standard errors, typical of the IV
ICLEI,2014.Icleicorporatereport2014.TechnicalReport,ICLEILocalGovernmentsfor
approach. Sustainability.
Heterogeneityisfoundacrosseconomicanddemographicchar- Kahn, M.E., Kotchen, M.J., 2011. Business cycle effects on concern about climate
change: the chilling effect of recession. Climate Change Economics 2 (03),
acteristics.Municipalitieswithhigherincome,lowerunemployment
257–273.
andahighershareofyoungpeopleshowahigherpositiveeffectof Kahn,M.E.,Morris,E.A.,2009.Walkingthewalk:theassociationbetweencommunity
thecommitmentonthevoteshareofincumbentmayors. environmentalismandgreentravelbehavior.JournaloftheAmericanPlanning
These findings are obtained using some control variables, and Association75(4),389–405.
List,J.A.,Sturm,D.M.,2006.Howelectionsmatter:theoryandevidencefromenviron-
under the exogeneity assumption for the instrument; the latter mentalpolicy.TheQuarterlyJournalofEconomics121(4),1249–1281.
amountstotheassumptionthattheinstitutionofaCovenantTerri- Matisoff,D.C.,2013.Differentraysofsunlight:understandinginformationdisclosure
torialCoordinator(CTC)isexogenoustothelocalmayoralelections. andcarbontransparency.EnergyPolicy55(0),579–592.
MEF,Datisubasecomunale:redditieprincipalivariabiliirpef.http://www1.finanze.
As always, availability of more control variables and/or of more
gov.it/finanze2/pagina_dichiarazioni/dichiarazioni.php.2015-09-30.
instrumentswouldallowformorepreciseinferentialresults;further Millard-Ball, A., 2012. Do city climate plans reduce emissions? Journal of urban
researchmayimproveonthissituation. economics71(3),289–311.
Ministerodell’Interno,Archiviostoricodelleelezioni.http://elezionistorico.interno.it/.
Thepresentresultsareconsistentwithboththewidelypledged
2014-10-31.
supportstatedinopinionsurveys,suchastheSpecialEurobarom- OECD,2003.VoluntaryApproachesforEnvironmentalPolicy:Effectiveness,Efficiency
eter 365, and with the global and intergenerational dimension of and Usage in Policy Mixes. OECD Publishing, Paris. http://dx.doi.org/10.1787/
9789264101784-en.
climatechange.Finally,thecurrentstudyfocusedonemissionreduc-
Persson,T.,Tabellini,G.E.,2002.PoliticalEconomics:ExplainingEconomicPolicy.MIT
tion commitments. Implementation of GHG reduction measures is press.
on-goingandthemonitoringreportswillrevealtheireffectiveness Solé-Ollé,A.,Viladecans-Marsal,E.,2013.Dopoliticalpartiesmatterforlocallanduse
policies?JournalofUrbanEconomics78, 42–56.
inthenearfuture.
TheUSConferenceofMayors,600Mayorsinall50StatesandPuertoRicoTakeAction
toReduceGlobalWarming.pressrelease,072007.
Headquarter,U.N.,2014.ThecompactofMayorsActionStatement.TechnicalReport,
Acknowledgments
ClimateSummit2014.
Wooldridge,J.M.,2002.EconometricAnalysisofCrossSectionandPanelData.MIT
TheauthorsaregratefultoCatherineSimonetforinsightfulcom- press.
ments;theparticipantstotheseminaron‘Citizens’supportforGHG Zheng,S.,Kahn,M.E.,Sun,W.,Luo,D.,2014.IncentivesforChina’surbanmayorsto
mitigatepollutionexternalities:theroleofthecentralgovernmentandpublic
emission reduction programmes’, European Commission - DG JRC,
environmentalism.RegionalScienceandUrbanEconomics47(0),61–71.
Ispra (July 3rd, 2014); and two anonymous referees. This research