EconomicDistressStimulatesReligiousFundamentalism
DanielL.Chen⇤January8,2020
Abstract
Whensocialconflictoccursalongethnic-religiouslinesinsteadofclasslinesisasubjectof muchdebate.Thispaperexploitsrelativepriceshocksinducedbythe1997Indonesianfinancial crisisandvariationinreligiousinstitutionsacrossIndonesiabeforethecrisistoidentifytheeffect ofeconomicdistressontherelationshipbetweenreligiousinstitutionsandsocialviolence.Six resultsemerge.Highreligiousintensityareasbeforethecrisishavemoresocialviolenceafterthe crisis.Strongermeasuresofreligiousintensityaremorestronglyassociatedwithsocialviolence. SocialviolenceincreasesfastestwhereparticipationinKoranstudyalsoincreasesthefastest. Instrumentingforeconomicdistressusingrelativepriceshockssuggestsacausalrelationship betweeneconomicdistressandtherelationshipbetweenreligiousintensityandsocialviolence. Creditavailabilitymitigatesthiseffect.Economicdistressalonedidnotstimulatesocialviolence butstimulatesitinthepresenceofreligiousinstitutions.Iexplainthesefindingsinamodel wherehighmarginalutilitiesduringeconomicdistressincreaseincentivesforgroupconflict wheregroupconflictincreasesthebudgetofinsurancegroups.Withvolatility,religionswith strongersanctionsorviolencearemorestableandsuccessful.Asvolatilitydeclines,benign groupsandreligionsbecomerelativelysuccessful.
Keywords:Religion,ClubGoods,Conflict,Fundamentalism
JELcodes:D71,D74,E21,G22,H41,O17,Z1,Z12
⇤ ToulouseInstituteforAdvancedStudies,daniel.li.chen@gmail.com.Iamextremelygratefultomyprincipal advisor,EstherDuflo,forherguidanceandintellectualgenerosity.IwouldalsoliketothankJoshuaAngrist,David Autor,AbhijitBanerjee,EliBerman,IvanFernandez-Val,ChrisHansen,SeemaJayachandran,BenOlken,VijayendraRao,ZakiWahhaj,MotohiroYogo,andespeciallyJeremyTobacman,forcomments.MaulinaCahyaningrum, PrimaFortunadewi,andJuliusKusumaprovidedoutstandingtranslationassistance.Workonthisprojectwas conductedwhileIreceivedfinancialsupportfromtheNationalScienceFoundationGraduateResearchFellowship, MacArthurFoundation,SocialScienceResearchCouncil,MITSchultzFund,NationalInstituteofChildHealthand HumanDevelopment,RussellSageFoundation,JohnM.OlinFoundation,theInstituteforHumaneStudies,andthe KauffmanFoundation.Allmistakesaremyown.
Whethersocialviolencerisesorfallswithreligiousintensityisasubjectofmuchdebate.Some creditreligion’sroleinreducingsocialviolenceasameansofkeepingpotentialviolentactorsoff ofstreets(BerrienandWinship2003,McRoberts2003,Freeman1986).Otherscitereligion’s roleinincreasingsocialviolenceasameansoforganizingindividualsintopotentialviolentactors (Glaeser2002,Berman2003).Whymightreligiousintensitybelinkedwithsocialviolenceinsome environmentsbutnotothers?Doesthelinkdependoneconomicconditions?Socialscientistshave longspeculatedontheconnectionbetweenreligiousintensityandsocialviolence;answerstothese questionsmaysuggestappropriatepoliciestoaddressideologicalextremism.
ThispaperexploitsdifferencesinreligiousintensityacrossIndonesiabeforeandduringthe financialcrisisandrelativepriceshocksinducedbythecrisistoidentifytheeffectofeconomic distressontherelationshipbetweenreligiousintensityandsocialviolence.Apriori,thereappears tobeprimafacieevidenceofariseinreligious-basedviolenceduringandafterthefinancialcrisis. Evenifviolencebeganfornon-religiousreasons,thelinesofdemarcationoftenbecamereligious.I
firstshowastrongrelationshipbetweenreligiousintensitybeforethecrisisandsocialviolenceafter thecrisis.Ithendocumentarelationshipbetweenchangesinreligiousintensityandchangesin socialviolenceduringthecrisis.Inextexploitthefactthatrelativepriceshocksinducedbythe crisisfavoredgrowersofstaples,namelyrice,andhurtstickywage-earners,particularlygovernment employeeswhosesalariesaresetbyfederallaw.Iinstrumentforeconomicdistressusingprecrisishectaresofwetlandownershipandgovernmentoccupationtoidentifytheimpactofeconomic distressontherelationshipbetweenreligiousintensityandsocialviolence.Bysodoing,Ipresent causalevidencetoaugmentpreviousdescriptiveanalysisoftheIndonesianfinancialcrisis(Chen 2007).
Inthefollowingsections,IpresentananalysisofdatafromtheDatabaseonSocialViolencein Indonesia1990-2001,collectedbytheUNSupportFacilityforIndonesianRecovery,anddatafrom theHundredVillagesSurvey,apanelof8,140households,conductedbytheIndonesianCensus Bureau.Section2presentsbackgroundanddataonreligionandsocialviolenceinIndonesia andamodelofthelinkbetweenreligiousintensityandsocialviolence.Totheextentreligious groupsprovidemutualinsurance(Berman2000,Chen2004),highmarginalutilitiesduringeconomic distressincreaseincentivesforgroupconflictasgroupconflictmayincreasethebudgetofmutual insurancegroups.Ifreligiousintensityisafunctionofthedegreetowhichsomeoneparticipates inmutualinsurance,increasingthenumberofparticipantsininsurancemattersmoreforthose whoareparticipatingathigherlevelsofreligiousintensity.Socialconflictofonegroupbyanother candirectlyappropriateresourcesorlowertheoutsideopportunitiesofclubmembersinorderto increasethereligiousintensityofthelessintense.
Between1997and1998,Indonesia’sRupiahfelldramaticallyfrom2400totheUSdollarto16000 totheUSdollarandtheCPIforfoodincreasedfrom100to261.Section3establishesthatreligious intensityandsocialviolenceareindeedrelatedduringthefinancialcrisis.OLSestimatesindicate thatinhighreligiousintensityareas,violenceismorelikelytoarise,whereviolenceismeasuredby
totalnumberofincidentsofsocialviolenceaswellasnumberofincidentswithminimumof1death. Theseresultsholdevenaftercontrollingforalargesetofvillageandenvironmentalcharacteristics.
Becausemostreligiousintensitymeasuresarecollectedbeforethecrisisandarerelativelytimeinvariantandbecausevillagesareunlikelytobuildschools,seminaries,orreligiousbuildingsin anticipationofsocialviolencethatmostlyoccurredafterthecrisis,reversecausalityisanunlikely confound.Infact,therelationshipbetweenpre-crisismeasuresofreligiousintensityandsocial violencelargelybeginsafterthecrisis.Inaddition,certainformsofreligiousintensity,suchas religiousschoolsandseminaries,aremorestronglyassociatedwithviolencethanareweakerforms, suchasKoranstudyandworshipbuildings.Multiplyingtheestimatedcoefficientsbythemeanof thereligiousintensitymeasuressumsuptothemeanoftheviolenceincidents,suggestingreligious intensitymayexplainpracticallyalltheviolencethatoccurredifthevectorofreligiousmeasuresare takenasexogenous.The R 2 ofthespecificationssuggestreligiousintensitymayexplainone-third ofthevarianceofviolencethatoccurred.
Socialviolenceisnegativelyassociatedwithothersocialactivities.Thissuggestsomitted variablesthatareassociatedwithbothKoranstudygroupsand“placebo”socialactivitiesarenot drivingtherelationshipbetweenreligiousintensityandsocialviolence.
AfundamentalissueintheinterpretationoftheOLSspecificationisthepresenceoffixed unobservablefactorsthatarecorrelatedwithreligiousintensityandsocialviolenceacrossprovinces. Toaddressthispotentialsourceofbias,IuselongitudinaldataonKoranstudy,whichistracked overtime.IfindthatKoranstudyremainsassociatedwithcommunalviolenceaftercontrolling forprovinceandtimefixedeffectsbutisunrelatedtostateorindustrialviolence.Thislastfinding lessenstheconcernthatomittedvariablesdrivechangesinbothKoranstudyandviolencesincethere issomethingspecificaboutcommunalviolenceratherthanviolenceingeneralthatisassociated withKoranstudyduringthefinancialcrisis.
ThevarietyofevidencepresentedinSection3suggestsastrongrelationshipbetweenreligious intensityandsocialviolenceduringthecrisis.Section4estimatesmodelswithinteractionterms toexplorewhyreligiousintensityandviolencearelinked.Pre-crisisreligiousintensity–schools, worshipbuildings,andseminaries–ismorestronglylinkedwithsocialviolenceinregionsthatare moreeconomicallydistressed.Economicdistressalonedoesnotstimulatesocialviolence.(Identitybasedviolencehastypicallybeenfoundtobeunrelatedtoeconomichardshipalone(Green,McFalls, andSmith2001,KruegerandPischke1997,Ruhm2000).)Iinstrumentforeconomicdistressusing pre-crisishectaresofwetlandownershipandgovernmentoccupationfollowingtheidentification strategyinChen(2010).
Importantly,alternativesocialinsurancemitigatesthiseffect.Theeffectofeconomicdistress onthelinkbetweenreligiousintensityandsocialviolenceisroughlyhalfasstrongwhencreditis availableintheformofbanks,microfinanceinstitutions,orBRIloanproducts.
Understandingtherelationshipbetweenreligiousintensityandsocialviolenceisnotlimited toviolenceprevention.Researchershavetraditionallyfocusedontheprivatereturntoreligious intensity(Iannaccone1998),butarenowbeginningtolookbeyondtheprivatereturns,forexample,
ontheimpactofreligiousintensityoneconomicgrowth,theattitudesthatunderlieeconomic growth,andgenderviolence(BarroandMcCleary2002,Guiso,Sapienza,andZingales2003,Chen 2005).Violenceisanegativeexternalitywithenormoussocialcosts(e.g.AbadieandGardeazabal (2003)andAlesina,et.al(1999)),sototheextentreligiousintensityandsocialviolencearerelated, thesocialreturnsmaybedifferentthantheprivatereturns.Theresultsinthispapersuggest economicvolatilitydecreasesthesocialreturnstoreligiousintensity:groupswithstrongersanctions orviolencearemorestablewitheconomicvolatility.Asvolatilitydeclines,benigngroupsorreligions becomesrelativelysuccessful.
2ReligionandSocialViolenceinIndonesia
Between1990and2001,socialviolenceledtomorethan6,208deathsinIndonesia,increasing sharplyafterthefinancialcrisisof1997(Tadjoeddin2002).Inoneyear,assetvaluesdroppedby 91%.Incontrast,ittookthreeyearsforassetvaluestodrop87%duringtheUSGreatDepression (Friend2003).Millionsofpeoplelostjobsorshiftedtotheinformalsector(Irawan,et.al.,2000). Thecrisisreachedapeakinearly1998andledtoriotsandlootingsineveryprovincebutone.It stimulatedmassfrustrationamongmarginalgroupsmanyofwhomlostjobsduetoeconomiccrisis andcompanybankruptcy.
TheempiricalanalysisdrawsfromtheUNSFIRDatabaseonSocialViolenceinIndonesia19902001(Tadjoeddin2002),whichcontainseveryincidentofsocialviolencereportedbythenational newsagency, Antara,andthenationaldaily, Kompas.Thedatabasetrackspropertydamageas wellasinterpersonalviolence.Socialviolencereferstophysicalactsofdestruction,killing,looting, attacks,burning,clashes,takinghostages,etc.,byagroupofpeople.Becausepresspoliciesdiffer beforeandafterthecrisis,theanalysisusescross-sectionalaswellaslongitudinaldatatoavoid relyingsolelyontime-seriesvariationofmediacoverage.
Communalviolenceaccountsfor77%ofthetotaldeathsduetosocialviolence;theothercategoriesarestate-communityandindustrialviolence.Communalviolenceisdefinedasviolence betweentwogroupsofcommunity,onegroupbeingattackedbytheother.State-communityviolenceisviolencedonebycommunitiesprotestingagainststateinstitutions,suchasthemilitary, theadministration,orsecurityofficials.Industrialviolenceisviolencethatarisesfromproblemsof industrialrelations.Communalviolencehasthewidestregionaldistribution.Itisfoundin116of 295district/citiesand22of26provinces.
Ethnic,religion,andmigration-relatedviolence,isthemostseveretypeofcommunalviolence, accountingfor68%oftotaldeathsduetocommunalviolence.Whilebothethnicandreligiousviolencearecodedtogether(ethnicgroupsareusuallyassociatedwithaparticularreligioninIndonesia), atleastsomeoftheseactsofcommunalviolencearedefinitelyreligiousinnature:descriptionsin Tadjoeddin(2002)include"killingbyevokingblackmagicshaman"(aformofvoodoo),"massrage assomeonerecognizedhimselfasGod’smessenger","churchruined","immorallocationruined", "mantakenhostagebyIslamholywarrior","gamblingandprostitutionlocationdestroyed",and
"burningofentertainmentplace".Violenceasaresultofdifferenceinpoliticalviewsaccounts foronly3.3%ofdeathsduetocommunalviolence(Table2).Thisisdefinedasviolencedueto conflictsbetweenandwithinpoliticalpartiesandtheirsupporters.Figure1indicatesthatthescale ofviolenceincreasedsharplyin1998.Tables1and2andFigure1arefromTadjoeddin(2002).
Sincethenationalmediaoftendonotrecordlocalizedconflicts,thedatamayunderreport levelsofconflict.UNSFIRcaptures1,093incidentsofconflictand6,208deathsover12years. Underabroaderdefinitionofconflict,thePODESdata(PotensialDesa/VillagePotentialStatistics) documentsalmost5000villagesasreportingconflictsinoneyearalone(Barron,et.al.2004).
2.1Theory
Religiousintensityasinsuranceprovidesatheoryforwhyviolenceandreligiousintensitymaybe linkedinsomeenvironmentsbutnotothers.FollowingthemodelinChen(2010),supposeagents receiveahigh(H )orlow(L)incomeshock.Letreligiousintensity, Q,representthedegreeto whichsomeoneparticipatesinmutualinsurance,i.e.thefractionofincomeshocksharedwiththe insurancegroup.Moreprecisely,lettheutilityofanagentreceiving x incomeshockbe, Ux = u[(1 Qx )x + Qx Q (µ)] V ( Q Qx ) C (Qx ),where (1 Q) representsthefractionoftheshock x agents keepforthemselves, Qx /Q istherelativereligiousintensity, Q istheaveragereligiousintensity, and µ isthebudgetofthereligiousgroup. V ( Q Q ) isthesocialsanctionfunctionandisincreasing inrelativelackofintensity, Q/Qx ,and C (Qx ) isthecostofdisplayingreligiousintensity.Inthe model,thegreaterisone’sreligiousintensitythemoremutualinsurancegivenandreceived.If Q =1,itisasiftheagentparticipatesinacommune,sharingallhisshockswiththegroup.If Q =0,theagentkeepsallhispositiveshocksandsuffersallhisnegativeshocks.
Chen(2010)exploitsrelativepriceshocksinducedbytheIndonesianfinancialcrisistodemonstrateacausalrelationshipbetweeneconomicdistressandreligiousintensityandfindssuggestive evidenceoftheroleofreligionasex-postsocialinsurance:creditavailabilityreducestheeffectof economicdistressonreligiousintensitybyroughly80%,religiousintensityalleviatesneedingalmsor credittomeetbasicneedsatthepeakofthecrisis,andreligiousinstitutionsfacilitateconsumption smoothingamongvillagers.
Nowconsidertheextremechoiceoffulldeviationornon-participationby M<N numberof agents.Let Eh (Q) denotetheexpectedvalueofsocialinsurancewhen h individualsparticipate at Q levelofinsurance.Then EN (Q) EN M (Q) isthevaluetoparticipantsofencouraging participationby M individuals.
Socialsanctionsincreasewithreligiousintensity, Q,because Q EN (Q) EN M (Q) > 0.Intuitively,as Q ! 0,thevalueofencouragingparticipationalsoapproaches0.Inotherwords,ina modelwherereligiousintensityrepresentsthedegreetowhichsomeoneparticipatesinmutualinsurance,increasingthenumberofparticipantsininsurancemattersforthosewhoareparticipating athigherlevelsofreligiousintensity.
Moreprecisely,athigherlevelsofreligiousintensity,groupssetahigher V 0 ( ),sosocialsanctions willbeincreasinginrelativelackofintensityif Q> 0,andwillincreasemoresharply,thehigher
thereligiousintensity.
Forconcreteness,supposethereare N M individualsreceivinglowshockswhodesireahigh degreeofmutualaid.Theysetahighreligiousintensity Q,thefractionofincomesharedorreceived fromthereligiousgroup,whilethe M individualsreceivingpositiveshockswhodesirealowdegree ofmutualaidchoosealowlevelofreligiousintensity.Sincereligiousintensitycanbechosenafter individualsreceivetheirshocks,toencouragethe M individualstoprovidemoremutualaid,the N M individualssetahighlevelofsocialsanctionagainstlowreligiousintensity Q.Individuals withhigh Q arewillingtoimposesanctions:empirically,FehrandGachter(2000)findthatpeople arewillingtopaytopunishinpublicgoodsgameseveninthelastperiodwhenthereisnopossible benefittothemselves.
Bothintergroupandintragroupconflictarecapturedbythemodel.Thepreviousexample suggestsintragroupconflict,butonecanalsointerpret V (.) asaformofdirectappropriationofthe H individualsbythe L individualsfromanotherreligiousgroup.Or,onecanthinkofaclubmodel, where V ( ) lowerstheoutsideopportunitiesofclubmembers:escalatingconflict(awaronperceived terror)andmutualantagonismincreasesthereligiousintensityofthelessintense.Socialconflictof onegroupbyanothercandirectlyappropriateresourcesorlowertheoutsideopportunitiesofclub membersinordertoincreasethebudgetofthereligiousgroup.
Anecdotalevidencesuggeststhismaybethecase.Accordingtopersonalinterviews,neighbors founditmoredifficulttotalkwithneighborsofadifferentreligionduringtimesofgroupconflict. YouthattendingIslamicschoolsaythattherearetensionsbetweenMuslimyouthwhoattend mosqueandthosewhodonot.Conflictwithnon-Muslimsmayactasaunifyingforceasindividuals donotwanttobemistakenforbeingorhelpinganon-Muslim.Conflictcanalsodirectlyappropriate resources,suchaslandorproperty.Inthiscase,encouragementorsanctionscantaketheformof intimidationorconversion.InterviewsindicatesomeindividualsconvertedtoIslamduringgroup conflict.Bothintragroupandintergroupconflictcanincreasethebudgetofmutualinsurance groups.
Akeypredictionofthemodelisthateconomicdistressincreasestheincentivetoenactsanctions becauseofhighmarginalutilitiesduringeconomicdistress.Alternativeformsofsocialinsurance shouldmitigatethiseffect.Chen(2010)showshowgroupswithoutstrongsanctionssuchasrotating savingsgroupstendtodeclineduringcrisis.Thissuggestsreligionswithharsherpunishmentor violencearemorestableandsuccessfulinanenvironmentwithhighvolatility.Asvolatilitydeclines, benigngroupsandreligionsbecomerelativelysuccessful.
2.2ReligionandEconomicData
HouseholdreligionandeconomicdatacomefromTheHundredVillagesSurvey,collectedbythe IndonesianCentralStatisticsOffice.Thepaneldatasetfollows8,140householdsfromMay1997 toAugust1999,beginningbeforethecrisisandcontinuinginfourwavesafterthecrisis(Figure 2).Religiousintensityatthehouseholdlevelismeasuredusingtheresponseto“Inthepast3 months,hasyourhouseholdincreased,decreased,stayedthesame,ornotparticipatedinthestudy
ofKoran(Pengajian)?”Thisquestionisaskedafterthecrisisandiscodedas1/0.Chen(2010) verifiesPengajianparticipationactuallymeasuresreligiousintensitybyexaminingitscorrelation withothermeasuresofreligiousintensity,suchasIslamicschoolattendance,Koranownership, worshipping,andmeasuresofbeliefsuchasanswering,“ItisuptoGod,”inresponseto“Whatis youridealnumberofsons?”aswellasreligiousoppositiontocontraceptionuse.1
Village-levelreligiositymeasuresofpercapitanumberofmosques,Islamicchapels,churches, Hindutemples,andBuddhisttemplesaretakenfromthe1997PODESdata(PotensialDesa/Village PotentialStatistics),whichasksfor1996information.Thereligiositymeasuresofpercapitanumber ofIslamicboardingschools,religiousschools,andseminariesaretakenfrom1998PODES.Sinceit isunlikelythatnewreligiousinstitutionswerebuiltduringthecrisis,Iinterprettheseaspre-crisis numbersanddividebythe1997PODESpopulationaccordingly(1998PODESpopulationnumbers wouldbeaffectedbycrisis-inducedmigration).
SincetheHundredVillagesSurveydoesnotcoverseparatistareassuchasAceh,noincident ofseparatistviolenceisincludedinthefollowinganalysis.TheHundredVillagesSurveyand theDatabaseonSocialViolenceoverlapforthefollowingeightprovinces:Bali,JawaBarat,Jawa Timur,KalimantanTimur,Lampung,NusaTenggaraTimur,Riau,andSulawesiTenggara.Since violencedataisrecordedattheprovincelevel,province-levelclustersareincludedinspecifications wherereligiousintensityismeasuredatthevillagelevel.
3EstimatingtheRelationshipbetweenReligiousIntensityandSocialViolenceDuringtheFinancialCrisis
IexaminetherelationshipbetweenreligiousintensityrecordedintheHundredVillagesSurveyand violencerecordedintheDatabaseonSocialViolenceinIndonesia1990-2001.
3.1Cross-SectionalVariation
Isviolencemorelikelytoariseinhighreligiousintensityareas?Arestrongermeasuresofreligious intensitymorestronglyassociatedwithsocialviolence?Table3reportsOLSestimatesofan equationlinkingsocialviolenceandreligiousintensity:
where Vjp representsallsocialviolenceincidentsfrom1990-2001orallsocialviolenceincidentswith minimum1deathinvillage j inprovince p, Rjp isavectorforvillage j inprovince p representing percentageofPengajianparticipationinavillage,religiousworshipbuildingsper1000population, religiousschoolsper1000population,andseminariesper1000populationand Xjp representsvillage,geographic,andfiscalcontrolvariables(urbandummy,population,area,numberofshops
1 Animportantlimitationtothisstudyisthatthispaperdoesnotanalyzebeliefs.However,itispossiblethatan increaseinIslamicschoolattendancetranslatessubsequentlyintostricterbeliefs.
per1000population,meanpre-crisismonthlyper-capitanon-foodexpenditures,dummiesforgeographiccharacteristicsflat,steep,beach,forest,valley,andriver,1996-1997INPRESfundsper1000 populationforeconomicactivity,buildingandfacilities,officesandinstitutions,humanresources, andIDT,anothervillageassistanceprogram).
Theestimatesshowastrongassociationbetweeneachmeasureofreligiousintensityandviolence.Thestrongassociationremainsaftercontrollingforvillage,geographic,andfiscalcharacteristics(Columns2and4).Infact,strongermeasuresofreligiousintensity–religiousschoolsand seminaries–aremuchmorestronglyassociatedwithviolence.Religiousschoolsper1000population andseminariesper1000populationareassociatedwithviolenceat1%statisticalsignificancein mostspecifications(Table3Columns1-4).PercentageofPengajianparticipationandworship buildingsper1000populationareassociatedwithviolenceat5%to10%statisticalsignificance inthesespecifications.TheseresultscorroborateBarron,et.al.(2004)whoalsofindintheir cross-sectionalanalysisofthe2003PODESdata,higherpresenceoffaithgroupsisassociatedwith higherlevelsofconflict.
Inmagnitudes,multiplyingthecoefficient bythemeanofthereligiousintensitymeasures sumsuptothemeanoftheviolenceincidents.Thus,ifwetake Rjp asexogenous,thissuggests religiousintensitymayexplainpracticallyalltheviolencethatoccurredinaverages.The R 2 ofthe regressiondisplayedinColumn1is0.34,suggestingreligiousintensitymayexplainone-thirdof thevarianceofviolencethatoccurred.The R 2 oftheregressiondisplayedinColumn3is0.32. Columns2and4have R 2 of0.49and0.48respectively.
3.2ReverseCausality
Apossibleexplanationforalinkbetweenreligiousintensityandviolenceistheresponseofreligious intensitytosocialviolenceinsteadofviceversa.Theempiricalsetupprecludesthispossibility becauseofthefactthatmostreligiousintensitymeasuresarerelativelytime-invariantandare pre-crisismeasures.Sinceitisunlikelythatnewreligiousinstitutionswerebuiltduringthecrisis, thesemeasurescanbeinterpretedaspre-crisisnumbers.Mostviolence(96%)occursafterthecrisis (Figure1).Itseemsunlikelyvillagesbuildschools,seminaries,orreligiousbuildingsinanticipation ofsocialviolence,soreversecausalityisanunlikelyconfound.
Separatelyregressingviolenceyear-by-yearonpre-crisisreligiousintensity,withthefollowing regression,
suggeststhestrongrelationshipbetweenpre-crisisreligiousintensityandsocialviolencebeginsafter thecrisis.EstimatesarereportedinTable4.Forexample,theestimatesof t in1993comparing with1998risefrom0.252to12.107forPengajianparticipationand1.449to23.659forseminaries (Columns1and3).Figure3PanelAdisplaystherelationshipbetweenAugust1998Pengajian participationandyear-by-yearsocialviolence.Figure3PanelsB-Ddisplaytherelationshipbetween
pre-crisisper-capitaworshipbuildings,religiousschools,andseminaries,respectively,andyear-byyearsocialviolence.Table4andFigure3havenoinformationdisplayedfor1990-1992and1994 becausetherearenoreportedincidentsofsocialviolencethatoverlapwiththeHundredVillages Surveyinthoseyears.
3.3OtherSocialActivities
Socialviolenceisnegativelyassociatedwithothersocialactivities.Table5displaysseparate partialcorrelationsbetweensocialviolenceandeachrecordedsocialactivity.WhilePengajianis positivelycorrelatedwithsocialviolenceandstatisticallysignificantatthe5%level,socialviolence isnotsignificantlyassociatedwithanyothersurveyedsocialactivity:sports(Olahraga),burial society(Kematian),clubforobtainingskills(KarangTaruna),familywelfaremovement(PKKand “occasionaltrainingforwomen”),and“10helpsforhousing”(Dasawisma).Theseresultssuggest omittedvariablesthatareassociatedwithbothreligiousandnon-religioussocialactivitiesarenot drivingtherelationshipbetweenreligiousintensityandsocialviolence.
Infact,theestimatessuggestparticipationinnon-religioussocialactivitiesisnegativelyassociatedwithsocialviolence.EachpercentagepointofPengajianparticipationisassociatedwith0.39 moreincidentsofsocialviolencewhereaseachpercentagepointinparticipationinwomen’straining, housinghelp,skilllearning,orburialsocietiesisassociatedwith0.30to0.50fewerincidentsofsocial violence.TheseresultsagaincorroborateBarronet.al(2004)whofindthatnetworksofengagement acrossethnic-religiouslinesreducelikelihoodofconflict.Theremainderofthestudymovesbeyond Barron,et.al(2004)tomorecarefullyisolatetherelationshipbetweenreligiousintensityandsocial violence,andtoestimatethecausalimpactofeconomicdistressonthisrelationship.
3.4PanelData
Thesignificantrelationshipbetweensomemeasuresofreligiousintensityandsocialviolencebeforethecrisis(Table4)suggestssomeunobservedenvironmentalvariablesmaybecorrelatedwith religiousintensity(forexample,ifethnic-religiousdiversityisgreaterwheretherearemorereligiousinstitutions,anddiversityiscorrelatedwithviolence,thismaybiastherelationshipbetween religiousintensityandsocialviolenceupwards).Toaddressthispossibility,Ialsoexaminethe relationshipbetweenPengajianparticipationandsocialviolencecontrollingforprovinceandtime fixedeffects.Pengajianparticipationistheonlymeasureofreligiousintensitythatistime-varying. Fixedeffectscontrolsforenvironmentalcharacteristicssuchasreligiousorethnicdiversityacross regions.
Toconstructthepanelofreligiousintensityandsocialviolence,recallthatinformationonPengajianparticipationiscollectedfor3-monthperiods.ImatchtheaveragePengajianparticipation rateofeachprovincefora3-monthperiodtothenumberofincidentsofsocialviolenceforthesame 3-monthperiod.SincetheHundredVillagesSurveycollectedPengajianparticipationat3different times,thisgivesme8provincesand3timeperiodsforatotalof24observationstoestimate:
Vpt = Rpt + p + ⌧t + "pt
where Vpt representsincidentsofsocialviolence, Rpt representspercentagePengajianparticipationin province p attime t, p areprovincefixedeffects,and ⌧t aretimefixedeffects.Ishowspecifications withandwithoutweightingbythenumberofhouseholdsintheHundredVillagesSurveyper province.
Theestimateofabout4.4inColumn4ofPanelAinTable6indicatesKoranstudyisassociated withcommunalviolenceincidentswithminimum1deathat10%statisticalsignificance.Theestimateof5.1inColumn2indicatesKoranstudyispositivelyassociatedwithincidentsofcommunal violence.Thecoefficient4.3inColumn4issmallerthanthecoefficient11inColumn3ofTable 3,onereasonforwhichisthatviolenceisrestrictedtoa3-monthperiodherewhereasinTable 3,violencewasaggregatedfor1990-2001.Theestimatesareroughlythesamewithandwithout populationweights(Columns1and3).
Evenwithcontrolsforfixedeffects,omittedvariablesmaybedrivingchangesinbothKoran studyandviolence.TotheextenttheeconomicdistressthatstimulatesKoranstudywouldstimulateanykindofsocialviolence,observingtherelationshipbetweenchangesinKoranstudyand changesinothertypesofsocialviolenceprovidesatestofthispossibility.Whendifferenttypesof violenceareconsideredinPanelsBandC,theassociationbetweenKoranstudyandothertypesof violence,state-communityandindustrial,isweaker,withcoefficientsof-0.11and0.91respectively. Thesumofstate-communityandindustrialviolencealsoisweaklyassociatedwithKoranstudy. Thereisnotenoughvariationinstate-communityandindustrialviolenceincidentswithminimum 1deathtorunfixedeffectsregressions.Thesefindingslessentheconcernthatomittedvariables drivechangesinbothKoranstudyandviolencesincethereissomethingspecificaboutcommunal violenceratherthanviolenceingeneralthatisassociatedwithKoranstudy.
4TheInteractionofEconomicDistressandReligiousIntensity
Theprecedingsectionsfindthatsocialviolenceismorelikelytoariseinareaswithgreaterreligious intensityandthatsocialviolenceincreasesmorequicklywherereligiousintensityalsoincreasesmore quickly.Toexaminewhy,Iestimatemodelswithinteractionterms.Akeypredictionofasocial insurancetheoryofreligiousintensityisthatreligiousintensityismorestronglylinkedwithsocial violenceinregionsthataremoreeconomicallydistressed.Iexamineplaceswherethefinancial crisishadmoreorlesseconomicimpactusingtheidentificationstrategyestablishedinChen(2010). Rapidinflationduringthefinancialcrisisfavoredgrowersofstaples,particularlyrice,anddisfavored stickywage-earners.Hectaresofwetlandownershipandgovernmentoccupationbeforethecrisis actasinstrumentsforeconomicdistresswhilehectaresofdrylandownershipandserviceoccupation actasplaceboinstruments.
Sincemostmeasuresofreligiousintensityarecollectedbeforethefinancialcrisis,theassociation
betweenreligiousintensityandsocialviolenceisunlikelytobeduetoeconomicdistressstimulating bothreligiousintensityandsocialviolence.Evenifeconomicdistressexplainstherelationship betweenchangesinreligiousintensityandchangesinsocialviolence,wemighthaveexpectedstate andindustrialviolencetoalsoincreasewithKoranstudy,butthisisnotthecase.Interestingly, thereisnostrongrelationshipbetweensocialviolenceandtheinstrumentsforeconomicdistressand thepointestimatesareoftheoppositesign(notshown),whichisconsistentwithidentity-based violencebeingtypicallyunrelatedtoeconomichardshipalone(Green,McFalls,andSmith2001, KruegerandPischke1997,Ruhm2000).Instead,theinteractionofpre-existingreligiousintensity witheconomicdistressincreasessocialviolence,suggestingtheremaybesomethingparticularabout religiousintensitythatallowseconomicdistresstotranslateintosocialviolence.
4.1ReducedFormEvidence
Toexaminewhetherreligiousintensityismorestronglylinkedwithsocialviolenceinregionsthatare moreeconomicallydistressed,Table7PanelAestimatesthefollowingreducedformspecification:
where Zjp representstheinstrumentsforvillage j inprovince p,wheretheinstrumentsarethe averagehectaresofwetlandownedinvillageandpercentageofhouseholdheadswhoworkingovernment. Rjp representspre-crisisreligiousintensityforvillage j inprovince p,definedasthesum ofstandardizedpercapitareligiousworshipbuildings,religiousschools,andseminaries.
Thecoefficient-14.2inColumn1indicatesthatwheretherewasmorepre-crisisreligiousintensity,wetland,whichcushionedavillagefromthecrisisbecausevillagerscangrowmorerice,is linkedtofewerincidentsofsocialviolence.Thecoefficient94.5inColumn1fortheinteraction ofreligiousintensityand%governmentworkersindicatevillageshitharderbythecrisis,because morevillagershadstickywages,andwithmorepre-crisisreligiousintensitysawmoreincidentsof socialviolence.Asimilarpatternholdsforincidentswithminimum1death(Column3).These relationshipsarerobusttoincludingvillage,geographic,andfiscalcontrols(Columns2and4).
Theaveragehectaresofwetlandownedis0.17(AppendixTableA)andthefractionofgovernmentworkersis0.06.Multiplying-14.2by0.17gives-2.41andmultiplying94.5by0.06gives 5.67.Thissuggeststheaveragehectaresofwetlandownedreducestheeffectofreligiousintensity onsocialviolencebyroughly30%(2.41dividedby8.2inRow1Column1ofTable7)whilethe averagefractionofgovernmentworkersincreasestheeffectofreligiousintensityonsocialviolence byroughly70%.ThesecalculatedrelativeeffectsarelargerforColumns2-4.
InPanelBofTable7,Ireplace Zjp withplaceboinstrumentsforvillage j inprovince p,the averagehectaresofdrylandownedinvillageandpercentageofhouseholdheadswhoworkinservice occupations.Comparing-4.2onthereligiousintensity*drylandinteractionwith-14.2forreligious intensity*wetlandinteractioninColumn1isconsistentwiththefindinginChapter1thatdryland hectaresprovideroughlyhalfthecushionthatwetlandhectaresprovide.Comparing53.1on
religiousintensity*serviceinteractionand94.5onreligiousintensity*governmentinteractionin Column1alsosuggeststheplaceboinstrumentsdisplayamuchsmallerandstatisticallyinsignificant effectascomparedtotheactualinstruments.
4.22SLSEstimates
Table8estimatestheanalogous2SLSspecification: Vjp = 0 Rjp + 1 (Rjp Ejp )+ 2 Ejp + ↵ 0 Xjp + "jp
where Ejp istheaverageeconomicdistressforvillage j inprovince p,instrumentedusing Zjp instruments,averagehectaresofwetlandandpercentageofgovernmentworkers. Rjp represents pre-crisisreligiousintensityforvillage j inprovince p,definedasthesumofstandardizedpercapita religiousworshipbuildings,religiousschools,andseminaries.Thecoefficient-5.9inColumn1 suggestssocialviolencerisingmostlythroughtheinteractionofpre-crisisreligiousinfrastructureand economicdistress.Theaveragehouseholdsuffereda$4.70shockinpercapitanonfoodexpenditures (AppendixTableA).Multipying-5.9by$4.7andadding-5.4,thecoefficientonpre-crisisreligious intensityinColumn1,indicatessocialviolenceispositivelyrelatedwithpre-crisisreligiousintensity attheaverageeconomicdistresslevel.
4.3SocialInsuranceandSocialViolence
Doessocialinsurancemitigatesocialviolence?Ifirstfindthatinthecross-section,socialviolence isnegativelyassociatedwithcreditavailability,definedashavingbanks,microfinanceinstitutions, orBRIloanproducts.Toaddressthepossibilitycreditavailabilityproxiesforgeneraleconomic development,Ialsoincludeanurbandummyandthenumberofshops.Incidentsofsocialviolence arepositivelyassociatedwiththosecharacteristics,suggestingthereissomethingparticularabout creditavailability,ratherthanurbannessorcommercialdevelopment,thatisnegativelyassociated withsocialviolence.Theseestimatesarenotstatisticallysignificant,buttheinterpretationis similarconsideringincidentswithminimum1deathandafterincludingcontrols,andcorroborate thefindingthatloanprovisionreducescrime(GarmaiseandMoskowitz2004).
InTable9,Irunthe2SLSspecificationseparatelyforvillageswithcreditavailability(n=32) andwithoutcreditavailability(n=61):
Inotherwords,Iestimate
where Cjp representscreditavailabilityinvillage j andprovince p andcreditavailabilityisdefined ashavingbanks,microfinanceinstitutions,orBRIloanproducts.TheBRIbanksystemisthe backboneoftheruralfinancialsysteminIndonesia(Holloh2001)andBRIloanproductsarea specialformofcreditwithlowcollateralrequirements.Accordingtopersonalinterviews,inone villageinthesurvey,BRIloanproductswereespeciallyavailableduringthecrisisforgovernment workersbecausetheywereconsideredtohave"fixed"jobs.AfeatureoftheBRIloanproductthat helpsloancollectionisthereductionofinterestfortimelyrepayments(Holloh2001).Microfinance programsarealsohelpfulforlaid-off workerstostartsmallbusinessesoftheirown(Akatiga1999).
Idisplayseparate2SLSspecificationsforeaseofinterpretation.Comparingthecoefficientson theinteractionofreligiousintensityandeconomicdistress,-4.2inColumn1with-8.4inColumn 2inPanelAsuggestcreditavailabilityreducestheimpactbyroughly50%.Thisratioremains roughlythesameforincidentswithminimum1death(Columns3and4)aswellasforestimates whenthefullsetofvillage,geographic,andfiscalcontrolsareincluded(PanelB).Thedifference betweentherespectivecoefficients,i.e. 4 ,isnotstatisticallysignificantatthe10%level,sothese resultsshouldbetakenasmerelysuggestive.
4.4AlternativeExplanations
Onealternativeexplanationfortheriseinsocialviolenceisthatinsteadofeconomicdistress,it isthepoliticalvacuumcreatedduringthecrisisthatallowedsocialviolencetoarise.However, between1990and2001,violenceasaresultofdifferenceinpoliticalviewsaccountsforonly3.3% ofdeathsduetocommunalviolence(Table2).Arelatedhypothesisisthatviolenceariseswhen civic/policinstitutionsweakenandthatreligiousinstitutionsarecorrelatedwiththepresenceof civic/policeinstitutions.Thisalsodoesnotappeartobethecase.
Anotherpossibilityisthatreligiousfragmentationisnecessaryforsocialviolence.However, greaterreligiousfragmentationofanarea,ascomputedbytheHerfindahlindexofreligiousworship buildings,isnotstronglyassociatedwithpre-crisisreligiousintensity.Moreover,themodelin Section2suggestssocialviolenceassociatedwithreligionneednotbelimitedtointergroupviolence. Arelatedhypothesisisthatlessviolenceoccursunderasinglereligiousregimeandthattheremay befewerincidentsofsocialviolencewhenthereisexactly1mosqueinthevillage.Thisalsodoes notappeartobethecase.
5Conclusion
Theseresultssuggestthatinaworldofcompetinginsurancegroups,asvolatilityincreases,religions withstrongersanctionsorviolencearemorestableandsuccessful,butasvolatilitydeclines,benign groupsandreligionsbecomerelativelysuccessful.
Inthispaper,IpresentananalysisofdatafromtheHundredVillagesSurveyanddatafrom theDatabaseonSocialViolenceinIndonesia1990-2001.OLSestimatesshowalargepositive relationshipbetweenreligiousintensityandsocialviolence.Becausemostreligiousintensitymea-
suresarerelativelytime-invariantandarepre-crisismeasuresandbecausevillagesareunlikelyto buildschools,seminaries,orreligiousbuildingsinanticipationofsocialviolence,reversecausality isunlikelytoexplainthisassociation.Infact,astrongrelationshipbetweenpre-crisismeasuresof religiousintensityandsocialviolencebeginsafterthecrisis.Inaddition,strongerformsofreligious intensityaremorestronglyassociatedwithviolence.Tocontrolforpotentialomittedvariablesbias, IuselongitudinaldataonKoranstudy,whichistrackedovertime.Koranstudyremainsassociated withcommunalviolenceafterincludingprovinceandtimefixedeffectsbutisunrelatedtostateor industrialviolence.
Sheddinglightonwhyreligiousintensityandsocialviolencearelinked,religiousintensityis linkedtomoresocialviolenceinregionsthataremoreeconomicallydistressed.Alternativesocial insurancemitigatesthiseffect.Totheextentgovernments,internationalorganizations,andNGOs areconcernedaboutideologicalextremism,inparticularbecauseitmayleadtoreligiousconflict andviolence,theresultshereandinChen(2010)suggestincreasingtheirroleinsocialinsurance maymitigatefundamentalisttendencies.Countriesthatinordinatelydependonnaturalresources maybesubjecttogreaterfluctuationsandmayfindreducingfundamentalisttendenciestobeyet anotherreasontodiversify.Ifitisthecaseglobalizationincreasestheriskindividualsface,providing insuranceagainstthatriskmaybeimportantinpreventingethnic-religiousconflicts.Miguel(2003) andMiguel,Satyanath,andSergenti(2003)findevidencethateconomicshocksincreaseconflicts.
TheGreatDepressionmayhaveaffectedgroupidentitydifferentlyintheUSandGermany becausetheprovisionofsocialinsurancewasaffordableintheformerbutnotinthelatter.The resultssuggestoneexplanationforwhysomereligionsandgroupidentitiesreplaceothers.In thelong-runcompetitionbetweensocialinsurancegroups,socialinsurancewithsanctionswillbe relativelysuccessful,especiallyinavolatileenvironment.Withvolatility,religionswithharsher sanctionsaremorestableandsuccessful.Asvolatilitydeclines,groupsorreligionswithreduced sanctionsorviolencebecomerelativelysuccessful.Theincreaseinincomevolatilityfrom1972-1998 (KruegerandPerri2003)isconsistentwiththecontemporaneousriseinreligiousintensityinUS popularandpublicdiscourseaswellastheshiftfromaninclusionary(socialgospelmovement)to exclusionary(religiousright)religiousintensity(Fogel2000).
Onecanevenimagineanevolutionarybranchingofsocialinsurances.Earlysocialinsurances stayedwithinthefamily,whichlaterbecameethnicitiesandreligions.Ethnicitiesandreligions intimebegatnationalities,politicalideologies,andevengenderidentitieseachseekingtodisplace earlierformsofgroupidentity,astechnologiesdevelopedtobetteraddressimperfectinformation (improvedabilitytoobserveeffortsandtypesnecessaryforsocialinsurancetobeself-enforcingand self-propagating).
6DataAppendix
TheempiricalanalysisdrawsfromTheHundredVillagesSurvey,collectedbytheIndonesianCentral StatisticsOffice.Thepaneldatasetfollows8,140householdsfromMay1997toAugust1999,
beginningbeforethecrisisandcontinuinginfourwavesafterthecrisis(Figure1).Inthepre-crisis period,thesurveyobserves120randomlyselectedhouseholdsineachof100communities.However between1997and1998,thenumberofvillageenumerationareasincreasesfrom2to3,necessitating areplacementofabout40randomlyselectedhouseholdspervillage.Thepartialreplacementof pre-crisishouseholdsiswhythepanelcontains8,140insteadof12,000households.Thesurveyalso collectsvillage-levelinformationinthefirstwaveof1997and1998.Amoredetaileddescriptionof thesurveyquestionsandvariableconstructionusedinthetablesisprovidedbelow.Thesurveyis inIndonesianandwastranslatedwiththehelpoftwotranslators.
Onemeasureofreligiousintensityistheresponseto“Inthepast3months,hasyourhousehold increased,decreased,stayedthesame,ornotparticipatedinthestudytheKoran(Pengajian)?” Moreprecisely,thephraseis“Pengajian/kegiatanagarmalainnya,”whichtranslatestoreligious activity,howevertranslatorssaythequestionwouldbeinterpretedbynativeIndonesiansasspecificallyreferringtoKoranstudy;non-Muslimsmayinterpretthequestionasreferringtotheequivalent intheirrespectivereligion.Thisquestionisaskedafterthecrisisandiscodedas1/0.
Thecontrols, Xij ,includepre-crisisMay1997valuesof:villagecharacteristics–urbandummy, population,area,numberofshopsper1000population;geographiccharacteristics–dummiesfor flat,steep(theexcludedtopographydummyisslightangle),beach,forest,valley,riverterrain(the excludedgeographydummyisother);andfiscalcharacteristics–INPRES(PresidentiallyInstructed ProgramforVillageAssistance,implementedduring1996-1997)fundingreceivednormalizedto$ per1000population,whichdividesintofundsusedforproductiveeconomiceffort,forbuildings andfacilities,forofficesandinstitutions,andforhumanresources,andtotalIDT(anothervillage assistanceprogram)fundsreceivedbythehouseholdbetween1994-1996.
Iusetheentiresampleof8,140households.AppendixTableApresentssomedescriptivestatistics.
7References
Abadie,AlbertoandJ.Gardeazabal(2003)“TheEconomicCostsofConflict:ACaseStudyofthe BasqueCountry,” AmericanEconomicReview,93(1),113-132.
Akatiga(1999)“FactoryUnemployment:GenderIssuesandImpacts,"CenterforSocialAnalysis ResearchReport.
Akerlof,GeorgeA.andWilliamT.Dickens(1982)“TheEconomicConsequencesofCognitive Dissonance,” AmericanEconomicReview,72(3),307-19.
Akerlof,GeorgeandRachelKranton(2000)“EconomicsandIdentity,” QuarterlyJournalof Economics,115(3),715-753.
Alesina,Alberto,RezaBaqir,andWilliamEasterly(1999)“PublicGoodsandEthnicDivisions,” QuarterlyJournalofEconomics,114(4),1243-84.
Alvarez,FernandoandUrbanJ.Jermann(2000)“Efficiency,Equilibrium,andAssetPricing withRiskofDefault,” Econometrica,68(4),775-797.
Azzi,Corry,andRonaldG.Ehrenberg(1975)“HouseholdAllocationofTimeandChurchAttendance,” JournalofPoliticalEconomy,83(1),27-56.
Barro,RobertJ.andRachelM.McCleary(2002)“ReligionandPoliticalEconomyinanInternationalPanel,”NBERWorkingPaperNo.8931.
Barron,Patrick,KaiKaiser,andMennoPradhan(2004)“LocalConflictinIndonesia:Measuring IncidenceandIdentifyingPatterns,”WorldBankmimeo.
Berman,Eli(2003)“Hamas,Taliban,andtheJewishUnderground:AnEconomist’sViewof RadicalReligiousMilitias,”UCLAmimeo.
Berman,Eli(2000)“Sect,Subsidy,andSacrifice:AnEconomist’sViewofUltra-OrthodoxJews,” QuarterlyJournalofEconomics,65(3),905-953.
Berman,EliandAraStepanyan(2003)“FertilityandEducationinRadicalIslamicSectsEvidencefromAsiaandAfrica,”RiceUniversitymimeo.
Berrian,JennyandChristopherWinship(2000)“ShouldWeHaveFaithintheChurches?The Ten-PointCoalition’sEffectonBoston’sYouthViolence,”ined.BernardE.Harcourt,Guns,Crime, andPunishmentinAmerica.NewYorkUniversityPress.
Bertrand,MarianneandSendhilMullainathan(2001)“DoPeopleMeanWhatTheySay?ImplicationsforSubjectiveSurveyData,” AmericanEconomicReviewPapersandProceedings,91(2), 67-72.
Bisin,AlbertoandThierryVerdier(2000)“‘BeyondtheMeltingPot’:CulturalTransmission, Marriage,andtheEvolutionofEthnicandReligiousTraits,” QuarterlyJournalofEconomics, 115(3),955-988.
Brown,MichaelE.,O.R.Cote,S.M.Lynn-Jones,andS.E.Millereds.(2001)Nationalismand EthnicConflict,MITPress.
Buchanan,JamesM.(1965)“AnEconomicTheoryofClubs,” Economica,32(1),1-14.
Chen,DanielL.(2010)“ClubGoodsandGroupIdentity:EvidencefromIslamicResurgence DuringtheIndonesianFinancialCrisis,”JournalofPoliticalEconomy,118(2),300-354.
Chen,DanielL.(2007)“IslamicResurgenceandSocialViolenceDuringtheIndonesianFinancial Crisis,”ined.M.GradsteinandK.Konrad,InstitutionsandNormsinEconomicDevelopment, 179-200,MITPress.
Chen,DanielL.(2005)"GenderViolenceandthePriceofVirginity:TheoryandEvidenceof IncompleteMarriageContracts,"mimeo.
Chetty,Raj(2002)“ConsumptionCommitments,UnemploymentDurations,andLocalRisk Aversion,”Harvardmimeo.
Coate,StephenandMartinRavallion(1993)“ReciprocitywithoutCommitment:CharacterizationandPerformanceofInformalInsuranceArrangements,” JournalofDevelopmentEconomics, 40(1),1-24.
DiPasquale,DeniseandEdwardL.Glaeser(1997)“TheL.A.RiotsandtheEconomicsofUrban Unrest,” JournalofUrbanEconomics,43(1),52-78.
Fehr,ErnstandSimonGachter(2000)“CooperationandPunishmentinPublicGoodsExperi-
ments,” AmericanEconomicReview,980-994.
Fogel,RobertW.(2000)TheFourthGreatAwakeningandtheFutureofEgalitarianism,UniversityofChicagoPress.
Fox,Jonathan(2002)“LessonsonReligionandConflictResolutionBasedonEmpiricalStudies ofReligionandConflict,”inReligionandConflictResolution,Bar-IlanUniversity.
Frankenberg,E.,D.Thomas,andK.Beegle(1999)“TheRealCostsofIndonesia’sEconomic Crisis:PreliminaryFindingsfromtheIndonesiaFamilyLifeSurveys,”UCLAmimeo.
Freeman,RichardB.(1986)“WhoEscapes?TheRelationofChurch-GoingandOtherBackgroundFactorstotheSocio-EconomicPerformanceofBlackMaleYouthsFromInner-CityPoverty Tracts,”inTheBlackYouthEmploymentCrisis,eds.RichardFreemanandHarryJ.Holzer,UniversityofChicagoPress.
Friend,Theodore.(2003)IndonesianDestinies.HarvardUniversityPress.
Fryer,RolandG.(2003)“AnEconomicApproachtoCulturalCapital,”UniversityofChicago mimeo.
Garmaise,MarkJ.andTobiasJ.Moskowitz(2004)"BankMergersandCrime:TheRealand SocialEffectsofCreditMarketCompetition," JournalofFinance forthcoming.
Geertz,Clifford(1960)TheReligionofJava,FreePress.
Genicot,GaranceandDebrajRay(2003)“GroupFormationinRisk-SharingArrangements,” ReviewofEconomicStudies,70(1),87-113.
Glaeser,EdwardL.andBruceI.Sacerdote(2001)“EducationandReligion,”NBERWorking PaperNo.8080.
Glaeser,EdwardL.andScheinkman,JoséA.(1998).“NeitheraBorrowerNoraLenderBe:An EconomicAnalysisofInterestRestrictionsandUsuryLaws,” JournalofLawandEconomics,vol. 41,pp.1-36.
Green,DonaldP.,LaurenceH.McFalls,andJenniferK.Smith(2001)"HateCrime:AnEmergentResearchAgenda," AnnualReviewofSociology.27:479-504.
Greenwald,A.G.,Banaji,M.R.(1995)“ImplicitSocialCognition:Attitudes,Self-Esteem,and Stereotypes, PsychologicalReview,102(1),pp.4-27.
Greif,Avner(1994)“CulturalBeliefsandtheOrganizationofSociety:AHistoricalandTheoreticalReflectiononCollectivistandIndividualistSocieties,” JournalofPoliticalEconomy,102(3), pp.912-950.
Gruber,Jon(2003a)“PayorPray:TheImpactofCharitableSubsidiesonReligiousParticipation,”MITmimeo.
Gruber,Jon(2003b)“ReligiousMarketStructure,ReligiousParticipationandOutcomes:Is ReligionGoodForYou?”MITmimeo.
Guiso,L.,Sapienza,P.andL.Zingales(2003)“People’sOpium?ReligionandEconomicAttitudes, JournalofMonetaryEconomics,50(1),225-282.
Holloh,Detlev(2001)"ProFIMicrofinanceInstitutionsStudy,"BankofIndonesia.
Iannaccone,LaurenceR.(1992)“SacrificeandStigma:ReducingFree-ridinginCults,Communes,andOtherCollectives,” JournalofPoliticalEconomy,100(2),271-291.
Iannaccone,LaurenceR.(1998)”IntroductiontotheEconomicsofReligion,” JournalofEconomicLiterature,36(3),1465-95.
Irawan,PuguhB.,IftikharAhmedandIyanatulIslam(2002),LaborMarketDynamicsinIndonesia:Analysisof18KeyIndicatorsoftheLaborMarket(KILM)1986-1999,ILOOffice,Jakarta.
Jay,RobertR.(1969)JavaneseVillagers:SocialRelationsinRuralModjokuto,MITPress.
Jost,JohnT.,J.Glaser,A.W.Kruglanski,andF.J.Sulloway(2003),“PoliticalConservatism asMotivatedSocialCognition,” PsychologicalBulletin,129(3),pp.339-375.
Knowles,JamesC.,ErnestoM.Pernia,andMaryRacelis(1999),“SocialConsequencesofthe FinancialCrisisinAsia,TheDeeperCrisis,”WorldBankeconomicstaff paper.
Kocherlakota,NarayanaR.(1996),“ImplicationsofEfficientRiskSharingwithoutCommitment,” TheReviewofEconomicStudies,63(4),pp.595-609.
Krueger,AlanB.andJitkaMaleckova(2002)“Education,Poverty,PoliticalViolenceandTerrorism:IsThereACausalConnection?”NBERWorkingPaperNo.9074.
Krueger,AlanB.andJorn-SteffenPischke(1997)"AStatisticalAnalysisofCrimeAgainst ForeignersinUnifiedGermany," JournalofHumanResources, Winter1997,32(1),182-209.
Krueger,DirkandFabrizioPerri(2002)“DoesIncomeInequalityLeadtoConsumptionInequality?EvidenceandTheory”NBERWorkingPaperNo.9202.
Landau,David(1993)PietyandPower,theWorldofJewishFundamentalism.NewYork:Hill andWang.
Levinsohn,James,StevenBerry,andJedFriedman(2003)“ImpactsoftheIndonesianEconomic Crisis:PriceChangesandthePoor,”Chapter12inManagingCurrencyCrisesinEmergingMarkets, MichaelDooleyandJeffreyFrankel,ed.,TheUniversityofChicagoPress/NBER.
LOC(TheFederalResearchDivisionoftheLibraryofCongress,CountryStudies)(2003), http://lcweb2.loc.gov/frd/cs/.
Marty,MartinE.andR.ScottAppleby,ed.(1991-95)TheFundamentalismProject,vol.1-vol. 5,U.ChicagoPress.
McRoberts,OmarM.(2003)ChurchandCommunityinaBlackUrbanNeighborhood,U. ChicagoPress.
Miguel,Edward(2003)“PovertyandWitchKilling,”UCBerkeleymimeo.
Miguel,Edward,ShankerSatyanathandErnestSergenti(2003)“EconomicShocksandCivil Conflict:AnInstrumentalVariablesApproach,”UCBerkeleymimeo.
Montalvo,JoseG.andMartaReynal-Querol“TheEffectofEthnicandReligiousConflicton Growth,”forthcoming JournalofDevelopmentEconomics.
Montgomery,JamesD.(1996)“ContemplationsontheEconomicApproachtoReligiousBehavior,” AmericanEconomicReview,86(2),443-447.
Ray,Debraj(1997)DevelopmentEconomics,PrincetonUniversityPress.
Ruhm,Christopher(2000)"AreRecessionsGoodforYourHealth?" QuarterlyJournalofEconomics,115(2),May,617-50.
Scotchmer,Suzanne(2002)“LocalPublicGoodsandClubs,”ch.29,HandbookofPublicEconomics,VolIV,AlanAuerbachandMartinFeldstein,eds,North-HollandPress.
Smith,JamesP.,D.Thomas,E.Frankenberg,K.Beegle,andG.Teruel(2002)“Wages,Employment,andEconomicShocks:EvidencefromIndonesia,” JournalofPopulationEconomics,15(1), 161-193.
Tadjoeddin,MohammadZ.(2002)“AnatomyofSocialViolenceintheContextofTransition: TheCaseofIndonesia1990-2001,”UNSFIRWorkingPaper02/01-E.
Tadjoeddin,MohammadZ.(2002)“DatabaseonSocialViolenceinIndonesia1990-2001,”UNSFIR WorkingPaper02/01.
Townsend,RobertM.(1994)“RiskandInsuranceinVillageIndia,” Econometrica,62(3),539591.
Wagner,Steven.(1999)“SummaryofPublicOpinionPrecedingtheParliamentaryElectionsin Indonesia-1999,”TheInternationalFoundationsforElectronicSystemsReport.
Wolfensohn,JamesD.(2003)“FightingTerrorismandPoverty,”Op-Ed,www.worldbank.org.cn/ English/content/964m6286227.shtml.
3/1/1997 4/1/1997 5/1/1997 6/1/1997 7/1/1997 8/1/1997 9/1/1997 10/1/1997 11/1/1997 12/1/1997 1/1/1998 2/1/1998 3/1/1998 4/1/1998 5/1/1998 6/1/1998 7/1/1998 8/1/1998 9/1/1998 10/1/1998 11/1/1998 12/1/1998 1/1/1999 2/1/1999 3/1/1999 4/1/1999 5/1/1999 6/1/1999 7/1/1999 8/1/1999 9/1/1999 10/1/1999
Figure 2A: August 1998 Pengajian Participation and Social Violence, 19902001
Relationship Between August 1998 % Pengajian Participation and Year-by-Year Social Violence
Figure 2B: Pre-Crisis Worship Buildings Per 1000 Population and Social Violence, 1990-2001
Relationship Between 1997 Worship Buildings Per 1000 Pop and Year-By-Year Social Violence
Figure 2C: Pre-Crisis Religious Schools Per 1000 Population and Social Violence, 1990-2001
Relationship Between 1997 Religious Schools Per 1000 Pop and Year-By-Year Social Violence
Figure 2D: Pre-Crisis Seminaries per 1000 Population and Social Violence, 1990-2001
Relationship Between 1997 Seminaries Per 1000 Pop and YearBy-Year Social Violence
Table1--SocialViolencebyCategory,1990-2001
NumberofNumberofIncidentsNumberofDeaths%Deathto CategoryIncidentswithMin1Death(minimumvalue)TotalDeath (1)(2)(3)(4)
CommunalViolence465262477176.9
SeparatistViolence502369137022.1
State-CommunityViolence8819591.0
IndustrialRelationsViolence38480.1 Total10936546208100
SocialViolencedatafromUNSupportFacilityforIndonesianRecovery,"DatabaseonSocialViolenceinIndonesia1990-2001". Anincidentofsocialviolenceisrecordedifthenationalnewsagency, Antara ,orthenationaldaily, Kompas ,reportedanincident withatleastonevictim,beithuman(casualtiesorinjuries)ormaterial(suchashouses,buildings,orvehiclesdamagedorburned).
96%oftheincidentsoccurbetween1998-2001;mostarecommunalviolence,definedassocialviolencebetweentwogroups ofcommunity,onegroupbeingattackedbytheother.
CommunalViolence:socialviolencebetweentwogroupsofcommunity,onegroupbeingattackedbytheother.
SeparatistViolence:socialviolencebetweenthestateandthepeopleofacertainareabecauseofregionalseparatism.
State-CommunityViolence:violencebetweenthestateandthecommunitywhoareexpressingprotestsagainststateinstitutions.
IndustrialViolence:violencethatarisesfromproblemsofindustrialrelations.
Table2--CommunalViolencebySub-Category,1990-2001
DeathsIncidents
Sub-CategoryNumber%ofTotalNumberCity/District (1)(2)(3)(4)
Ethnic,Religion,andMigration323067.723339
TheMay98Riots120225.2610
DifferencesinPoliticalViews1563.37954
CivilCommotion(Tawuran)871.87028
Issueof'DukunSantet'651.42817
CompetingResources160.31610
FoodRiots50.12322
Other100.2109
Total4771100465116
SocialViolencedatafromUNSupportFacilityforIndonesianRecovery,"DatabaseonSocialViolenceinIndonesia1990-2001".
Anincidentofsocialviolenceisrecordedifthenationalnewsagency, Antara ,orthenationaldaily, Kompas ,reportedanincident withatleastonevictim,beithuman(casualtiesorinjuries)ormaterial(suchashouses,buildings,orvehiclesdamagedorburned).
96%oftheincidentsoccurbetween1998-2001;mostarecommunalviolence,definedassocialviolencebetweentwogroups ofcommunity,onegroupbeingattackedbytheother.
CommunalViolence:socialviolencebetweentwogroupsofcommunity,onegroupbeingattackedbytheother.
Ethnic,Religion,andMigration:religionpropagationrelatedtoparticularregionsandethnicgroups
TheMay98Riots:riotsinbigcitiesprecedingfallofPresidentSuhartoinMay1998
DifferencesinPoliticalViews:conflictsbetweenandwithinpoliticalpartiesandtheirsupporters
CivilCommotion(Tawuran):clashesbetweenvillages,neighborhoods,orgroups
Issueof'DukunSantet':killingsofpeopleaccusedofevilmagicandwitchcraft
CompetingResources:disputesbetweencommunitygroupscompetingforeconomicresources
FoodRiots:massriotsandlootingsforstaplefoodsbetweenJanuarytoMarch1998
AllViolence(OLS)
Table3--RelationshipbetweenReligiousIntensityandSocialViolence
IncidentsofIncidentswith SocialViolence
Minimum1Death
(1)(2)(3)(4)
%PengajianParticipationinVillage,August199835*36*11*11* (17)(17)(5)(6)
ReligiousWorshipBuildingsPer1000Pop4**3*1**1* (2)(2)(0)(0)
ReligiousSchoolsPer1000Pop16**14***5*5*** (7)(4)(2)(1)
SeminariesPer1000Pop115***101***36***32*** (18)(25)(6)(8)
R20.340.490.320.48 N93939393 ControlsNYNY
RegressionsareOLSregressionsof93villagesandincludeprovince-levelclusters.
SocialViolencedatafromUNSupportFacilityforIndonesianRecovery,"DatabaseonSocialViolenceinIndonesia1990-2001". Anincidentofsocialviolenceisrecordedifthenationalnewsagency, Antara ,orthenationaldaily, Kompas ,reportedanincident withatleastonevictim,beithuman(casualtiesorinjuries)ormaterial(suchashouses,buildings,orvehiclesdamagedorburned).
96%oftheincidentsoccurbetween1998-2001;mostarecommunalviolence,definedassocialviolencebetweentwogroups ofcommunity,onegroupbeingattackedbytheother.
ControlvariablesareVillage,Geography,andFiscalCharacteristicsarelistedbelow.
VillageCharacteristics--Urban,Population,Size,NumberofShopsPer1000Pop,MeanPre-CrisisPer-CapitaNon-FoodExpenditures GeographyCharacteristics--Flat,Steep,Beach,Forest,Valley,River
FiscalCharacteristics--1996-1997INPRESFundsPer1000PopforEconomicActivity,BuildingandFacilities,OfficesandInstitutions, HumanResources,andIDTfunds
Table4--RelationshipbetweenReligiousIntensityandYear-by-YearSocialViolence
DependentVariable:PengajianWorshipReligiousSeminaries IncidentsofSocialViolenceParticipationBuildingsSchools
(1)(2)(3)(4)
19930.2520.0040.339***1.449*** (0.211)(0.027)(0.085)(0.304)
1995-0.251-0.0140.032-0.668 (0.244)(0.016)(0.058)(0.748)
19960.2520.0040.339***1.449*** (0.211)(0.027)(0.085)(0.304)
19971.5090.0272.033***8.695*** (1.264)(0.160)(0.509)(1.824)
199812.107*1.504***1.38823.659** (5.352)(0.275)(1.447)(6.880)
19997.605*0.682**2.050**18.988** (3.704)(0.254)(0.706)(5.761)
200010.5210.9835.598***34.414*** (5.622)(0.613)(1.508)(8.141)
20014.456*0.2202.214**12.534*** (1.935)(0.249)(0.722)(3.243)
ControlsYYYY
RegressionsareOLSregressionsof93villagesandincludeprovince-levelclusters.
SocialViolencedatafromUNSupportFacilityforIndonesianRecovery,"DatabaseonSocialViolenceinIndonesia1990-2001". Anincidentofsocialviolenceisrecordedifthenationalnewsagency, Antara ,orthenationaldaily, Kompas ,reportedanincident withatleastonevictim,beithuman(casualtiesorinjuries)ormaterial(suchashouses,buildings,orvehiclesdamagedorburned). 96%oftheincidentsoccurbetween1998-2001;mostarecommunalviolence,definedassocialviolencebetweentwogroups ofcommunity,onegroupbeingattackedbytheother.
ControlvariablesareVillage,Geography,andFiscalCharacteristicsarelistedbelow.
VillageCharacteristics--Urban,Population,Size,NumberofShopsPer1000Pop,MeanPre-CrisisPer-CapitaNon-FoodExpenditures GeographyCharacteristics--Flat,Steep,Beach,Forest,Valley,River
FiscalCharacteristics--1996-1997INPRESFundsPer1000PopforEconomicActivity,BuildingandFacilities,OfficesandInstitutions, HumanResources,andIDTfunds
AllViolence(OLS)
Table5--RelationshipbetweenSocialActivitiesandSocialViolence
IncidentsofIncidentswith SocialViolenceMinimum1Death (1)(2)
%PengajianParticipationinVillage,August199839**12** (12)(4)
%TrainingforWomenParticipation,August1998-33-9 (30)(10)
%10HelpsforHousingParticipation,August1998-50-15 (30)(10)
%ClubforSkillLearningParticipation,August1998-32-10 (25)(8)
%BurialSocietyParticipation,August1998-30-9 (18)(6)
%SportsClubParticipation,August19983-0 (10)(4)
%SavingsClubParticipation,August1998-1-1 (22)(8)
ControlsYY
EachcoefficientrepresentsaseparateOLSregressionof93villages,conditionaloncontrols,andincludeprovince-levelclusters.
SocialViolencedatafromUNSupportFacilityforIndonesianRecovery,"DatabaseonSocialViolenceinIndonesia1990-2001". Anincidentofsocialviolenceisrecordedifthenationalnewsagency, Antara ,orthenationaldaily, Kompas ,reportedanincident withatleastonevictim,beithuman(casualtiesorinjuries)ormaterial(suchashouses,buildings,orvehiclesdamagedorburned).
96%oftheincidentsoccurbetween1998-2001;mostarecommunalviolence,definedassocialviolencebetweentwogroups ofcommunity,onegroupbeingattackedbytheother.
ControlvariablesareVillage,Geography,andFiscalCharacteristicsarelistedbelow.
VillageCharacteristics--Urban,Population,Size,NumberofShopsPer1000Pop,MeanPre-CrisisPer-CapitaNon-FoodExpenditures GeographyCharacteristics--Flat,Steep,Beach,Forest,Valley,River
FiscalCharacteristics--1996-1997INPRESFundsPer1000PopforEconomicActivity,BuildingandFacilities,OfficesandInstitutions, HumanResources,andIDTfunds
Table6--RelationshipbetweenReligiousIntensityandSocialViolence(Panel)
IncidentsofIncidentswith SocialViolence
Minimum1Death
(1)(2)(3)(4)
Panel A: CommunalViolence(FixedEffects)
%PengajianParticipationinProvince4.3405.1073.989*4.348* (4.598)(5.850)(2.209)(2.335)
Panel B: State-CommunityViolence
%PengajianParticipationinProvince-0.034-0.108n/an/a (1.090)(1.469)
Panel C: IndustrialViolence
%PengajianParticipationinProvince1.1900.909n/an/a (1.215)(1.311)
PopulationWeightedNYNY
FixedEffectsProvince,TimeProvince,TimeProvince,TimeProvince,Time
RegressionsareFixedEffectsregressionsof8provincesineachof3timeperiods,atotalof24observations, withprovinceandtimefixedeffects.Populationweightsarethenumberofhouseholdsperprovinceinthesample.
EachcoefficientrepresentsaseparateOLSregressionofPengajianParticipationRatesfor3-monthperiodonViolence.
SocialViolencedatafromUNSupportFacilityforIndonesianRecovery,"DatabaseonSocialViolenceinIndonesia1990-2001".
Anincidentofsocialviolenceisrecordedifthenationalnewsagency, Antara ,orthenationaldaily, Kompas ,reportedanincident withatleastonevictim,beithuman(casualtiesorinjuries)ormaterial(suchashouses,buildings,orvehiclesdamagedorburned).
96%oftheincidentsoccurbetween1998-2001;mostarecommunalviolence,definedassocialviolencebetweentwogroups ofcommunity,onegroupbeingattackedbytheother.
CommunalViolence:socialviolencebetweentwogroupsofcommunity,onegroupbeingattackedbytheother.
State-CommunityViolence:violencebetweenthestateandthecommunitywhoareexpressingprotestsagainststateinstitutions.
IndustrialViolence:violencethatarisesfromproblemsofindustrialrelations.
n/a:Toofewstate-communityandindustrialviolenceincidentswithminimum1deathtorunfixedeffectsregressions.
Table7--ReducedFormRelationshipBetweenSocialViolenceandPre-CrisisReligiousIntensityInteractedwithInstruments
SocialViolence IncidentsofIncidentswith Minimum1Death
(1)(2)(3)(4)
Panel A: Main Experiment
Pre-CrisisReligiousIntensity8.2**7.0**2.5**2.1** (2.8)(2.8)(0.9)(0.9)
Pre-CrisisReligiousIntensity*Wetland-14.2*-15.5*-4.2*-4.4* (6.2)(7.4)(1.8)(2.1)
Pre-CrisisReligiousIntensity*%Govt94.5**85.8**29.2**28.3** (35.3)(28.2)(11.4)(9.7)
Wetland-17.1-21.1**-5.3-6.5** (13.2)(8.3)(4.2)(2.4)
%Govt48.80.614.33.5 (26.4)(56.0)(8.2)(18.2)
Panel B: Placebo Experiment
Pre-CrisisReligiousIntensity7.89.2*2.12.5 (5.0)(4.5)(1.5)(1.3)
Pre-CrisisReligiousIntensity*Dryland-4.2-6.7-1.1-1.8 (4.3)(3.9)(1.3)(1.2)
Pre-CrisisReligiousIntensity*%Service53.155.717.618.5 (33.0)(31.8)(10.4)(10.7)
Dryland-20.8**-18.7*-7.2**-6.5** (8.8)(8.4)(2.6)(2.6)
%Service-11.5-29.7-6.0-8.5 (30.3)(39.2)(8.6)(11.4)
ControlsNYNY
Pre-CrisisReligiousIntensityisthesumofstandardizedReligiousWorshipBuildingsPer1000Pop,ReligiousSchoolsPer1000 Pop,andSeminariesPer1000Pop.WetlandandDrylandaretheaveragehectaresownedinvillage.GovernmentandServicearethe %ofhouseholdheadswhoworkinthatoccupation.
RegressionsareOLSregressionsof93villagesandincludeprovince-levelclusters.
SocialViolencedatafromUNSupportFacilityforIndonesianRecovery,"DatabaseonSocialViolenceinIndonesia1990-2001".
Anincidentofsocialviolenceisrecordedifthenationalnewsagency, Antara ,orthenationaldaily, Kompas ,reportedanincident withatleastonevictim,beithuman(casualtiesorinjuries)ormaterial(suchashouses,buildings,orvehiclesdamagedorburned).
96%oftheincidentsoccurbetween1998-2001;mostarecommunalviolence,definedassocialviolencebetweentwogroups ofcommunity,onegroupbeingattackedbytheother.
ControlvariablesareVillage,Geography,andFiscalCharacteristicsarelistedbelow.
VillageCharacteristics--Urban,Population,Size,NumberofShopsPer1000Pop,MeanPre-CrisisPer-CapitaNon-FoodExpenditures GeographyCharacteristics--Flat,Steep,Beach,Forest,Valley,River
FiscalCharacteristics--1996-1997INPRESFundsPer1000PopforEconomicActivity,BuildingandFacilities,OfficesandInstitutions, HumanResources,andIDTfunds
Table8--RelationshipBetweenSocialViolenceandPre-CrisisReligiousIntensityInteractedwithEconomicDistress
Incidentswith SocialMinimum1Death Violence Incidentsof
(1)(2)(3)(4)
Pre-CrisisReligiousIntensity-5.4*-12.4-1.6*-3.7 (2.5)(14.5)(0.8)(4.6)
Pre-CrisisReligiousIntensity-5.9***-6.9**-1.8***-2.2**
*ChangeinPerCapitaNonfoodExpenditure(IV)(1.2)(2.4)(0.4)(0.7)
ChangeinPerCapitaNonfoodExpenditure(IV)-6.6-8.4-1.9-2.7 (4.0)(6.6)(1.3)(2.0)
Pre-CrisisReligiousIntensityisthesumofstandardizedReligiousWorshipBuildingsPer1000Pop,ReligiousSchoolsPer1000 Pop,andSeminariesPer1000Pop.
ChangeinPerCapitaNonfoodExpenditureisthedifferencecomputedbetweenAugust1998andMay1997.
RegressionsareIVregressionsof93villagesandincludeprovince-levelclusters.
Theexcludedinstrumentsarewetland,government,andtheirinteractionswithpre-crisisreligiousintensity.
SocialViolencedatafromUNSupportFacilityforIndonesianRecovery,"DatabaseonSocialViolenceinIndonesia1990-2001". Anincidentofsocialviolenceisrecordedifthenationalnewsagency, Antara ,orthenationaldaily, Kompas ,reportedanincident withatleastonevictim,beithuman(casualtiesorinjuries)ormaterial(suchashouses,buildings,orvehiclesdamagedorburned).
96%oftheincidentsoccurbetween1998-2001;mostarecommunalviolence,definedassocialviolencebetweentwogroups ofcommunity,onegroupbeingattackedbytheother.
ControlvariablesareVillage,Geography,andFiscalCharacteristicsarelistedbelow.
VillageCharacteristics--Urban,Population,Size,NumberofShopsPer1000Pop,MeanPre-CrisisPer-CapitaNon-FoodExpenditures GeographyCharacteristics--Flat,Steep,Beach,Forest,Valley,River
FiscalCharacteristics--1996-1997INPRESFundsPer1000PopforEconomicActivity,BuildingandFacilities,OfficesandInstitutions, HumanResources,andIDTfunds
Table9--EvidenceSuggestingSocialInsuranceMitigatesSocialViolence
IncidentsofSocialViolenceIncidentswithMinimum1Death
Panel A: No Controls
Pre-CrisisReligiousIntensity-4.2-3.8-1.3-0.7 (2.5)(5.6)(1.0)(1.7)
Pre-CrisisReligiousIntensity-4.2**-8.4**-1.3**-2.8**
*PerCapitaNonfoodExpenditureChange(IV)(1.5)(2.9)(0.5)(1.0)
PerCapitaNonfoodExpenditureChange(IV)-5.8-11.7-1.8-4.0 (5.1)(8.4)(1.6)(2.7)
Panel B: With Controls
Pre-CrisisReligiousIntensity-5.8-38.5-1.9-10.7 (4.7)(94.5)(1.3)(26.7)
Pre-CrisisReligiousIntensity-4.2**-6.7-1.5**-2.3
*PerCapitaNonfoodExpenditureChange(IV)(1.7)(10.7)(0.6)(3.2)
PerCapitaNonfoodExpenditureChange(IV)-1.7-4.8-1.4-1.9 (5.1)(17.4)(1.5)(5.5)
N32613261
Pre-CrisisReligiousIntensityisthesumofstandardizedReligiousWorshipBuildingsPer1000Pop,ReligiousSchoolsPer1000 Pop,andSeminariesPer1000Pop.
CreditAvailabilityisdefinedashavingaBank,Microfinance,orBRILoanProductavailableinvillage.
RegressionsareIVregressionsof93villagesandincludeprovince-levelclusters.
Theexcludedinstrumentsarewetland,government,interactionswithpre-crisisreligiousintensity.
SocialViolencedatafromUNSupportFacilityforIndonesianRecovery,"DatabaseonSocialViolenceinIndonesia1990-2001".
Anincidentofsocialviolenceisrecordedifthenationalnewsagency,Antara ,orthenationaldaily, Kompas ,reportedanincident withatleastonevictim,beithuman(casualtiesorinjuries)ormaterial(suchashouses,buildings,orvehiclesdamagedorburned).
96%oftheincidentsoccurbetween1998-2001;mostarecommunalviolence,definedassocialviolencebetweentwogroups ofcommunity,onegroupbeingattackedbytheother.
ControlvariablesareVillage,Geography,andFiscalCharacteristicsarelistedbelow.
VillageCharacteristics--Urban,Population,Size,NumberofShopsPer1000Pop,MeanPre-CrisisPer-CapitaNon-FoodExpenditures GeographyCharacteristics--Flat,Steep,Beach,Forest,Valley,River
FiscalCharacteristics--1996-1997INPRESFundsPer1000PopforEconomicActivity,BuildingandFacilities,OfficesandInstitutions, HumanResources,andIDTfunds
PercentageOwnWetland31%StandardDeviationofVillageConsumptionShock11.42
PercentageOwnDryland66%duringCrisis(Aug1998-May1997)(1.56)
PercentageinFarming66%StandardDeviationofVillageConsumptionShock9.22
WetlandOwnership(Hectares)0.17Non-Crisis(May1999-Dec1998)(2.16) (0.01)TotalWorshipBuildingsPer1000Pop3.83
DrylandOwnership(Hectares)0.72(0.28) (0.01)ReligiousSchoolsper1000Pop0.12
SurnameIndicatesHajPilgrimage1.0%(0.04)
NumberofChildrenattendingIslamicSchool0.15Seminariesper1000Pop0.01 (0.01)(0.01)
MonthlyPer-CapitaFoodExpenditure,May199714.6%PengajianParticipationinVillage,August19980.61 (0.1)(0.03)
MonthlyPer-CapitaNon-FoodExpenditure,May19977.3CreditAvailable0.34 (0.2)(0.05)
HouseholdSize4.16NumberShopsPer1000Pop0.07 (0.02)(0.03)
Governmentworker6%Urban0.20
ServiceWorker10%(0.04)
N81401996-1997INPRESFundsin$/1000Pop0.91 (0.09)
N99
1998Aug1998Dec1999May1999Aug
MonthlyPer-CapitaNon-FoodExpenditure,Change-4.71.1-0.10.2 (0.2)(0.2)(0.2)(0.2)
PengajianParticipationRate61%unavail.67%71%
PengajianIncreaseinLast3Months9%unavail.7%7%
PengajianDecreaseinLast3Months9%unavail.10%11%
IncidentsofSocialViolence34.65 (3.20)
IncidentsofSocialViolencewithMinimum1Death11.26 (1.02)
IncidentsofCommunalViolence0.83 (3monthperiod)(0.29)
IncidentsofCommunalViolencewithMinimum1Death0.33 (3monthperiod)(0.16)
IncidentsofState-CommunityViolence0.08 (3monthperiod)(0.06)
IncidentsofIndustrialViolence0.17 (3monthperiod)(0.08)
%PengajianParticipationinVillage,August19980.66 (3monthperiod)(0.03)
NumberofProvinces8
Violence Summary Statistics Crisis Summary Statistics AppendixTableA--DescriptiveStatistics Household Summary StatisticsVillage Summary Statistics