Anti-ELAB Movement, National Security Law, and ...

The law not only stipulates various actions taken by protesters during the Anti-ELAB Movement as national security infringements, ... Skiptomaincontent Accessibilityhelp Weusecookiestodistinguishyoufromotherusersandtoprovideyouwithabetterexperienceonourwebsites.Closethismessagetoacceptcookiesorfindouthowtomanageyourcookiesettings. Cancel Login × × Home OnlysearchcontentIhaveaccessto Hostname:page-component-684899dbb8-v9xhf Totalloadingtime:0.709 Renderdate:2022-05-21T05:51:57.296Z Hasdataissue:true FeatureFlags:{ "shouldUseShareProductTool":true, "shouldUseHypothesis":true, "isUnsiloEnabled":true, "useRatesEcommerce":false, "useNewApi":true } Home>Journals>JapaneseJournalofPoliticalScience>Volume22Issue4>Anti-ELABMovement,NationalSecurityLaw,andheterogeneous...English Français JapaneseJournalofPoliticalScienceArticlecontentsAbstractIntroduction:institutionaltrustinapost-repressionperiodTheoreticaldiscussionandhypothesesBackgroundontheHongKongcase:thebirthanddeathofamovementsocietyEmpiricalstrategiesEmpiricalfindingsandrobustnesschecksConcludingremarksFootnotesReferencesAnti-ELABMovement,NationalSecurityLaw,andheterogeneousinstitutionaltrustinHongKong PublishedonlinebyCambridgeUniversityPress:  29December2021HansH.Tung[Opensinanewwindow],Ming-JenLin[Opensinanewwindow] andYi-FanLinShowauthordetailsHansH.Tung*Affiliation:DepartmentofPoliticalScience,NationalTaiwanUniversity,Taipei,Taiwan CenterforResearchinEconometricTheoryandApplications,NationalTaiwanUniversity,Taipei,Taiwan Ming-JenLinAffiliation:CenterforResearchinEconometricTheoryandApplications,NationalTaiwanUniversity,Taipei,Taiwan DepartmentofEconomics,NationalTaiwanUniversity,Taipei,Taiwan Yi-FanLinAffiliation:DepartmentofEconomics,NationalTaiwanUniversity,Taipei,Taiwan * *Correspondingauthor.E-mail:[email protected] Figures Supplementarymaterials Metrics ArticlecontentsAbstractIntroduction:institutionaltrustinapost-repressionperiodTheoreticaldiscussionandhypothesesBackgroundontheHongKongcase:thebirthanddeathofamovementsocietyEmpiricalstrategiesEmpiricalfindingsandrobustnesschecksConcludingremarksFootnotesReferencesSavePDFSavePDF(0.74mb)ViewPDF[Opensinanewwindow]SavetoDropboxSavetoGoogleDriveSavetoKindleShareCiteRights&Permissions[Opensinanewwindow]AbstractHowdoesrepressiononoppositionprotestsaffectcitizens'institutionaltrustunderdictatorships?Therehasbeenaburgeoningliteratureinvestigatingempiricallybothlong-andshort-termimpactsofprotestsandtheirrepressiononcitizens'politicalpreferencesinbothdemocraticandnondemocraticcontexts.Yet,theliteraturetellsusrelativelylittleabouthowtheabovequestioncouldbeanswered.ThispapertriestoanswerthisquestionbytakingadvantageofarecentnaturalexperimentinHongKongwhenBeijingsuddenlyadoptedtheNationalSecurityLaw(NSL)inJune2020torepressdissidents'protestmobilization.Ourfindingsaretwofold.Firstofall,theNSLdroveawedgeintheHongKongsocietybymakingthepro-establishmentcampmoresatisfiedwiththepost-NSLinstitutionsontheonehand,whilealienatingthepro-democracycampwholosttremendoustrustinthemontheother.Second,ourstudyalsorevealsthatone'strustininstitutionsissignificantlyassociatedwiththeregimes'abilitytocurbprotesters'contentiousmobilization.TheHongKongerswhohadhigherconfidenceintheNSLtoreininprotestswouldalsohaveagreaterleveloftrustthanthosewhodidn't.Theeffect,however,issubstantiallysmalleramongpro-democracyHongKongersexceptfortheirtrustinmonitoringinstitutions.AsBeijingistransformingHongKong'scurrentinstitutionsfromwithinhopesofbringingaboutanewpoliticalequilibrium,ourstudyhelpsprovideatimelyassessmentofHongKong'sinstitutionallandscapeandshedslightonhowlikelythisstrategycanwork.KeywordsInstitutionaltrustAnti-ELABMovementcontentiouspoliticsrepressionauthoritairaninstitutionsNationalSecurityLaw Type ResearchArticle Information JapaneseJournalofPoliticalScience , Volume22 , Issue4,December2021,pp.287-311DOI:https://doi.org/10.1017/S1468109921000293[Opensinanewwindow] CreativeCommons ThisisanOpenAccessarticle,distributedunderthetermsoftheCreativeCommonsAttribution-NonCommercial-NoDerivativeslicence(http://creativecommons.org/licenses/by-nc-nd/4.0),whichpermitsnon-commercialre-use,distribution,andreproductioninanymedium,providedthatnoalterationsaremadeandtheoriginalarticleisproperlycited.ThewrittenpermissionofCambridgeUniversityPressmustbeobtainedpriortoanycommercialuseand/oradaptationofthearticle. Copyright Copyright©TheAuthor(s),2021.PublishedbyCambridgeUniversityPress 1.Introduction:institutionaltrustinapost-repressionperiod Howdoesrepressiononoppositionprotestsaffectcitizens'institutionaltrustunderdictatorships?Therehasbeenaburgeoningscholarlyinterestrecentlyininvestigatingempiricallybothlong-andshort-termimpactsofprotestsandmassmobilizationsoncitizens'politicalpreferencesinbothdemocraticandnondemocraticcontexts(Madestametal.,ReferenceMadestam,Shoag,VeugerandYanagizawa-Drott2013;SangnierandZylberberg,ReferenceSangnierandZylberberg2017;Mazumder,ReferenceMazumder2018;FryeandBorisova,ReferenceFryeandBorisova2019;El-Mallakh,ReferenceEl-Mallakh2020;TertytchnayaandLankina,ReferenceTertytchnayaandLankina2020).Moreover,thisgrowingliteraturealsowellextendstohowtheirrepressionsindifferentformsaffectthesamesetofoutcomes(Lawrence,ReferenceLawrence2016;RozenasandZhukov,ReferenceRozenasandZhukov2019;DinasandNorthmore-Ball,ReferenceDinasandNorthmore-Ball2020;NeundorfandPop-Eleches,ReferenceNeundorfandPop-Eleches2020;CurticeandBehlendorf,ReferenceCurticeandBehlendorf2021;Desposatoetal.,ReferenceDesposato,WangandWu2021;Wang,ReferenceWang2021).Yet,theliteraturetellsusrelativelylittleabouthowtheabovequestioncouldbeanswered. Whydoesitmattertoexpandtheliteraturetoincludetherelationshipbetweenprotestcrackdownandthepublic'sinstitutionaltrust?Foronething,thecomparativeauthoritarianismliteraturehastoldus,beyondtheirnominalfunctionalities,thepoliticalsignificanceofauthoritarianinstitutionsinbringingaboutbetterregimestabilityandperformancethroughpower-sharingandcooptation(Gandhi,ReferenceGandhi2008;KimandGandhi,ReferenceKimandGandhi2010;BoixandSvolik,ReferenceBoixandSvolik2013).Asaresult,intermsofunderstandingtheeffectsofrepressiononthesurvivalandprosperityofadictatorship,it'snotjustaboutknowinghowitmightgiverisetobacklashmovements(CurticeandBehlendorf,ReferenceCurticeandBehlendorf2021)orfosteranti-regimesentimentinthelongrun(Wang,ReferenceWang2021),butalsoaboutfiguringoutitseffectoncitizens'trustininstitutions. Foranother,whendictatorsusesubtlerformsofrepressionsuchasfurtherrestrictingcitizens'(alreadynotsomany)civillibertiestopreserveregimestabilitywithoutlosingtoomuchlegitimacy(Escribá-Folch,ReferenceEscribá-Folch2013),theimplementationofsuchasofterapproachmustbemoreembeddedintheexistinginstitutionalframeworkthanonewithmoreviolenceinvolved.Theeffectivenessandlegitimacyofsuchastrategy,however,criticallyhingesonthecredibilityoftheexistingauthoritarianinstitutionsamongthepublic.OnegoodexampleisThailand's2017constitutioninwhichanewelectoralsystemwasintroducedtogivethemilitaryadominantpowerinappointingmembersintheSenate,anunelectedbody,thatwouldworktogetherwiththepopularlyelectedHouseofRepresentativestopickaprimeminister.Footnote1ThefactthatsuchpoliticalmaneuveringwasapprovedpreviouslyinAugust2016byareferendumandthenlegitimizedwithKing'ssignatureclearlyshowsthatasoftrepressionlikethiscertainlystoodabetterchancetosucceedwhenThailand'sreferendumandmonarchyaspoliticalinstitutionsenjoyedenoughtrust. NowsinceBeijinghasadoptedaverysimilarinstitutionalistapproachtohandlingHongKong'ssituation,understandingtheeffectsoftherecentlypassedNationalSecurityLaw(hereafterNSL)onHongKongers'institutionaltrustthereforebecomesveryrelevant.WeexaminetheNSL'sshort-termeffectbystudyingHongKong'srecentcontentiousepisodeoftheanti-Extradition-Law-Amendment-Bill(ELAB)Movementbetween2019and2020.ThelawnotonlystipulatesvariousactionstakenbyprotestersduringtheAnti-ELABMovementasnationalsecurityinfringements,butalsoauthorizesthecreationofthecentrallydirectedOfficeforSafeguardingNationalSecuritycommandingawiderangeofpowersinlawenforcementandadjudication.AsitspassageinChina'sPeople'sCongresslastJune(2020)cameasashocktomostpeopleinHongKong,thetwosurveysweconductedrightbeforeandafteritallowustoidentityitseffectonHongKongers'trustinvariousrelatedpoliticalinstitutions. Wefindthat,foralltheinstitutionsunderscrutinyinthisstudy,Footnote2theNSLasasoftrepressionFootnote3didhaveaneffectonone'sinstitutionaltrustacrosstheboard.However,theeffectisfoundtobeheterogeneousbetweenpro-democracyandpro-establishmentrespondentswiththeformerhavingnegativewhilethelatterhavingpositiveNSL-inducedeffectoninstitutionaltrust.Basedontheinsightsfromthecomparativeauthoritarianismliterature,wearguethatthedifferencearisesfromone'ssenseofregimeinclusiveness.Whileit'shardforoursurveytointerviewrealregimeinsiders,thepro-establishmentrespondentswhoidentifiedthemselveswiththeregimecertainlywouldhavehigherinstitutionaltrustwhenrepressioncouldsuppressthesocialunrestinHongKongandhelpstrengthentheestablishedinstitutionsintheirfavor.Forpro-democracycounterparts,however,theirinstitutionaltrustplungedaspredictedsincerepressionnotonlyconsolidatedtheinstitutionsthatmightimpairHongKong'sdemocraticprospect,butalsosabotagedthosethathadhelpedsustainitsruleoflaw,widelyregardedasintegraltoHongKongers'identity. Moreover,ourresultsalsosuggestthattheeffectofrepressiononinstitutionaltrustisconditionedbyhowwelloneperceivesthestrategytobeworkinginreininginprotesters'mobilization.WefindtheoptimistsoftheNSLtobeassociatedwithahigherlevelofinstitutionaltrustthanthepessimists.Wearguethat,fromaninformationalperspective,comparedtothepessimists,theunexpectedNSLasashockupdatedboththepro-establishmentoptimistswiththenewlygainedstrengthfortheinstitutionsandthepro-democracyoneswithalowerlikelihoodforfuturepoliticalinterventionstofurtherdisrupttheinstitutionalqualitysincetherepressionhadworked.Footnote4 Finally,wealsofindthat,comparedtotheexecutivebranchoftheinstitutionsincludedinthisstudy,theNSLdidn'thurtthepro-democracyrespondents'institutionaltrustinHongKong'sCourtverymuch–lessthan10%lowerthanthepre-NSLlevel.WhilethisfindingmightsimplyattesttoHongKongers'conventioninhavinghighertrustforthejudicialbranch(ChanandChan,ReferenceChanandChan2006),ithoweverhasaprofoundimplicationforHongKong'spost-NSLpoliticaldevelopmentasBeijingdramaticallytightensitsgriponthisformerBritishcolony.Footnote5AsLhrmannandLindberg(ReferenceLhrmannandLindberg2019)pointout,theongoingthirdwaveofautocratizationoftentakesplaceunderalegalfaçadewherebyrestrictingcitizens'politicalrightsislegitimizedviatheproceduresofexistinginstitutionssuchaslegislaturesFootnote6orcourts.ThefindingsofourstudysuggestthatHongKongmightverywellbesuchacasewhereitsrelativelytrustedCourtcouldbeinstrumentaltolegalizingandlegitimizingtheweakeningofthepoliticaloppositionandfurthererosionofcitizens'politicalrightsandfreedoms.Footnote7 Thepaperisstructuredasfollows.Section2presentsourtheoreticalperspectivestobeempiricallytestedandthehypothesesderivedfromthem.WealsoprovideinSection3adiscussionofHongKong'spoliticalcontextrelevanttoourstudy.ItisthenfollowedbySection4wherethedatacollectionandempiricalstrategiesareexplicated.Section5presentsourempiricalfindingsandrobustnesschecks.Finally,Section6concludes. 2.Theoreticaldiscussionandhypotheses SinceDavidEaston(ReferenceEaston1975),politicalscientistshavemainlystudiedinstitutionaltrustthroughthelensoftwoconceptshehelpeddevelop:diffuseandspecificsupport.Thelatterdefinestrustasone'slevelofsatisfactionofaspecificinstitutionaccordingtoitsperformanceandthereforetendstobeshort-terminnaturegivenpossiblefluctuationsingovernmentpolicies.Incontrast,theformerdefinestrustasone'smoregeneralfaithinalargerpoliticalsystemandthereforeisbasicallyone'slong-termbelieflargelyshapedbythepoliticalsocializationheorshewentthroughpreviously.Inthispaper,weproposeseveralnewtheoreticalperspectivestoenrichthisframeworkandbetterexplaintheeffectsofprotestrepressiononinstitutionaltrustintimesofturmoil. 2.1NSL'sheterogeneouseffects Firstandforemost,aswehavementionedabove,therecentinstitutionalistturninthecomparativeauthoritarianismliteraturehasputtheinclusivenessofpoliticalinstitutionsunderdictatorshipsatcenterstage.Beyondtheirnominalfunctionalities,thekeyreasonforauthoritarianinstitutionstobethemainstayofregimestabilityistheirabilitytocrediblyincludeandcooptallthemajoractorswhosesupportisessentialtodictators'rule.Asthetheorygoes,thepeoplewhoareinstitutionallyincludedandsharedwithresourceswillbeincentivizedtosupportnotonlytheinstitutionsthatmakethecooptationpossible,butalsotheregimeitself(BoixandSvolik,ReferenceBoixandSvolik2013).Whilethetheoreticalframeworkisinitiallyformulatedforanalyzingtheintra-eliterelationship,theideaofinclusivenesscanbereadilyextendedtothepublicopinionsofrank-and-filecitizens.Neundorfetal.(ReferenceNeundorf,GerschewskiandOlar2020)adoptthisapproachbytappingintothecross-nationalvariationinthepre-transitionauthoritarianinclusiveness–definedas‘widerredistributionofsocioeconomicandpoliticalbenefits’–forexplainingcitizens'democraticsupportinpost-authoritariancountries.Theyfindthatthesupporttendstobeloweramongthecitizensincountrieswithamoreinclusiveregimebeforethetransitionsincetheirlivingstandardsmightnotnecessarilybesubstantiallyenhancedunderdemocracy. Alongthesimilarline,sinceourfocusiswithinandnotacrosscountries,wearguethatinstitutionaltrustcanalsovarybetweenordinarycitizenswhofeelthemselvesapartoftheregimeandthosewhodon't.Asfarastheeffectofrepressionisconcerned,thisargumentimpliesthatrepressionwillhaveheterogeneouseffectsbetweenthesetwogroups.Forthosewhoperceivethemselvestobeoutcasts,repressionscanfurtheralienatethemfromtheregimeandevenhavealong-termdampeningeffectontheirtrustinpoliticalinstitutionsandthegovernment(Lawrence,ReferenceLawrence2016;DinasandNorthmore-Ball,ReferenceDinasandNorthmore-Ball2020;Desposatoetal.,ReferenceDesposato,WangandWu2021;Wang,ReferenceWang2021).Forthoseself-perceivedregimeinsiders,however,theabilityoftheregimetoclampdownonprotestmobilizationsandrestoresocialordershouldhaveanoppositeeffectofboostingtheirconfidenceintheregimeanditsinstitutions.We,therefore,hypothesizethat: Inclusivenesshypothesis:TheNSLexertedanegativeeffectonpro-democracyHongKongers'institutionaltrust,butinducedapositiveoneonpro-establishmentcounterparts'institutionaltrust. 2.2Protestexpectations Second,whileprotestrepressioncertainlypresentsashocktoeveryone,itdoesn'tmeanitseffectwillnecessarilybeuniformtoall.This,therefore,createsanotherlayerofheterogeneityinrepression'seffectoninstitutionaltrust.Fromaninformationalperspective,protestrepressionasanexternalshockbringspeoplenewinformationabouttheirgovernments,whichcanalsohelpthem–assumingtheyareBayesian–updatetheirlevelsofinstitutionaltrustbyformingexpectationsaboutthefutureofpoliticalinstitutions.FryeandBorisova(ReferenceFryeandBorisova2019),forexample,provideasimilarreasoningthatunexpectedprotestsinRussiagaveitscitizenswhohadastrongprioraboutthegovernment'srepressionintentionnewinformationregardingtherulingelite'stoleranceofpublicopposition,whichupdatedtheirtrustinpoliticalinstitutionsupward.Inthecontextofrepression,wearguethatthisBayesiantheorygivesrisetodifferentpredictionsaccordingtoone'sassessmentofitseffects.Amongtheregimeinsiders,thosewhobelieverepressiontobeeffectiveattamingprotesterswillfindinstitutionsmoretrustworthyasthispieceofnewinformationreconfirmsthemtheresolveandabilityoftheregimetoprotecttheinstitutionsthatbenefitthem.Bycontrast,thosewhodon'tholdthesamebeliefwilllowertheirtrustforexpectingdecliningbenefitsfromtheirinstitutionalmemberships. Amongtheregimeoutsiders,despiteforadifferentreason,thedirectionofthepredictionremainsunchanged.Whilerepressionispredictedtoinduceanegativeeffectontheirinstitutionaltrustsincetheyderivenoutilityfrompoliticalinstitutions,thedegreeofthedownwardBayesianupdatewillbehigheramongthosewhoperceivetheeffectofarepressionshocktobelow.Thereasonisthat,ifthedictatorisunabletoreininprotestmobilizationsthroughrepression,italsoimpliesthattheremightbemoreprotestsandgreaterneedsforrepressionsthatmakeinstitutionsevenlessinclusiveandtrustworthyinthefuture.Bycontrast,ifrepressionworksfromtheoutset,thenitislesslikelyforsuchactionstobetaken,thusmakingpoliticalinstitutions‘less’untrustworthy.Combined,theBayesianperspectiveonrepressionasashockgivesusahypothesis: Informationalhypothesis:ForbothregimeinsidersandoutsidersinHongKong,thelowerone'sprotestexpectationwas–or,equivalently,themoreeffectiveoneexpectedthecrackdowntobe–,thehigherhisorherinstitutionaltrustwouldbe. ThissecondhypothesisalsoechoestherecentfindingsbyRozenasandZhukov(ReferenceRozenasandZhukov2019)thatpeople'sloyaltyunderrepressivedictatorshipiscontingentonthecredibilityofthedictator'sretributionthreattotheopposition. 3.BackgroundontheHongKongcase:thebirthanddeathofamovementsociety GiventhecentralityofsocialmovementsandpoliticalactivisminHongKong'spoliticallandscape(Cheng,ReferenceCheng2016),itsrecentAnti-ELABMovement–beganinJune2019andofficiallyendedwiththepassageoftheNSLinJune2020–providesusagreatcontextfortestingourtheoreticalargumentsandtheeffectsoftheNSL.Firstofall,sincethesovereigntyhandoverfromtheUKtothePeople'sRepublicofChinain1997,HongKonghasbecomeChina'sSpecialAdministrativeRegion(SAR)andwaspromisedtoenjoyself-governanceunderanad-hocpoliticalframeworkof‘OneCountry,TwoSystems’(OCTS)for50years.Asaresult,underitsBasicLaw,theframeworkhad,thus,createdahybridregimeinHongKongwhere,ontheonehand,newauthoritarianinstitutionssuchasthelegislature(i.e.,theLegCo)andtheindirectelectionfortheChiefExecutive(theheadofthegovernment)wereadopted,Footnote8and,exceptfortheuniversalsuffrage,severalliberaldemocraticcomponentssuchascivilliberties,theruleoflaw,andthejudicialindependence,onwhichHongKong'smarketeconomywascriticallypredicated,alsoco-existedontheother(Ma,ReferenceMa2007). BeforetheNSL,thetwocomponentsabovejointlydeterminedthelevelofinclusivenessoftheregimeforbothelitesandrank-and-filecitizensinHongKong.Ontheonehand,partiallyinheritingfromthecolonialperiod,theregimecooptedpowerfulelitesthroughvariousauthoritarianinstitutionssuchastheLegCoandindirectelectionsforkeygovernmentpositions(Fong,ReferenceFong2013).Footnote9Forexample,theChiefExecutivewaselectedindirectlybyasmall‘selectorate’–a1200-peopleElectoralCommittee–composedofpoliticians(representativesatalllevelsoflocallegislatures)andalimitednumberofelitesfromavarietyofprofessions(29intotal;e.g.,legalprofession,businesscommunity,medicalprofession,etc.).Furthermore,intheLegCo,halfoftheseatswerefilledbyrepresentativeselectedfromeachofthe‘functionalconstituencies’–thesamegroupsofprofessionsfortheChiefExecutive'sElectoralCommittee–,whiletheotherhalfwerepopularlyelectedfromfivedifferentdistricts.Theseinstitutionsweredesignedtomaketheinterestsofthecooptedsocialelitesalignedwiththoseoftheregime. Ontheotherhand,forordinarycitizens,thesenseofregimeinclusivenessàlaNeundorfetal.(ReferenceNeundorf,GerschewskiandOlar2020),however,couldonlybederivedfromthecivillibertiesgrantedtothemundertheOCTSframework,and,asHongKong'spoliticaldevelopmentsince1997hasshown,thissensewasgainedtoalargeextentbymeansofthemostcontentiousformofthecivilliberties:protests.AttheinceptivestageoftheAnti-ELABMovementwhenthecontentioussituationwasescalatingafterthe12Juneconfrontationin2019,Footnote10anauthoritativeandinternationallyknownbondcreditratingcompany,Moody'sInvestorsService,publishedanarticle,‘Moody'saffirmsHongKong'sAa2ratings,maintainsstableoutlook’on5July2019,Footnote11wherethecompanywrote: TheratingincludesMoody'sassessmentofpoliticalriskforHongKongthattakesintoaccountperiodicchallengestothegovernment'spoliciesinrecentyears,andparticularlyinlarge-scaleprotestsbythepopulation.SuchprotestsarepartofthechecksandbalancesinplaceinHongKong,thatsupportinstitutionalstrength.SignsthatchecksandbalancesweakenwouldbeanegativeforHongKong'screditprofile.[italicsadded] Moody'sstatementbearsstrongtestimonytothecentralityofprotestsastheessentialmeansonwhichordinaryHongKongersdependedforenjoyingtheliberaldemocraticpartofthehybridregimewhereinstitutionalchecksandbalanceswereabsent.Table1documentsallthemajorcontentiousepisodessince1997,includingthecontestedpolicesthattriggeredtheprotests,theinformationandturnoutoftheprotests,andthegovernment'sresponsestothem. Table1.HongKong'scontentiouspolitics:majorepisodessince1997 Itisclearfromthetablethat,beforethefinalblowoftheNSLwasimposedin2020,exceptfortheprotestoverthe‘NPCSC'sPowerofFinalInterpretation’in1999thatmightfailtooverridethegovernmentpolicyforthelackofenoughcriticalmassforthemovement,HongKongerswerealwaysabletopushbackagainstBeijing'sandtheSARgovernment'seffortstobringHongKongclosertothemainlandeconomically,politically,andsocially.Thisprotest-basedchecks-and-balancesmodel,however,wasnotsustainableforthelong-rungiventhetutelarynatureofthishybridregime.Inotherwords,whiletheSARgovernmentyieldedtoprotesters'massmobilization,itdidn'treallystoptheCentralGovernmentinBeijingfromforgingaheadwithitspoliticalagendainHongKongtotiltthebalancetoward‘OneCountry’over‘TwoSystems’withintheOCTSframework,especiallysinceXiJingpingbecameChina'ssupremeleaderin2012. Thewatershedmomentwastheyearof2014whenBeijingputoffagainitspromisetograntthefullversionoftheuniversalsuffragetoHongKongersregardingtheelectionsforboththeChiefExecutiveandtheLegCo.Beijing's2014decisionfirstgaverisetothe‘UmbrellaRevolution’wherestudentsandpoliticalactivistsoccupiedthesquareinfrontoftheAdmiraltyMRTstationandthegovernmentcompounds.Lateron,thepoliticalgrievancesagainstBeijinginHongKonggrewdramaticallyandtherewerevariousdemonstrationsorganizedtoprotestBeijing'spoliticalagendasuchasthepatriotismeducationandtheNSL.Inearly2019,thecontroversyroseagainwhentheSARgovernmenttriedtointroduceamilderversionoftheLaw,accordingtowhichsuspectswhowereaccusedofcommittingacrimecouldbeexpeditedtoChinafortrials.GiventhehugediscrepancyinjudicialsystemsbetweenChina(ContinentalCivilLaw)andHongKong(CommonLaw),thebillencounteredastrongreactionfromHongKong'sdemocratsandcivilsociety,whicheventuallyevolvedintotheAnti-ELABMovementthatpersisteduntilcoronavirusdisease2019(COVID-19)arrivedinearly2020andwasboughttoanendabruptlybytheNSL. 4.Empiricalstrategies 4.1Datacollection,sample,andvariableconstruction TotestthehypothesesspecifiedinSection2,thisstudyexploitstheNSLasanexternalshockbyconductingtwosurveyswithalmostthesamesetofquestionsinHongKongimmediatelybeforeandafteritspassage.On21May2020,theChinesegovernmentannouncedthatanewversionoftheNSLtailor-madeforHongKongwouldbedeliberatedandvotedoninthecomingsessionoftheNationalPeople'sCongressheldinBeijingbetween22and28May2020.ThenewsbroughtashocknotonlytoHongKong,butalsototherestoftheworldsincetheintensityoftheAnti-ELABMovementhadgraduallytaperedoffowingtotheCOVID-19outbreaksinceJanuary2020.Thedraftwaseventuallypassedon28May. Thefirstsurveybeganon15Mayandendedon21May,thedayoftheChinesegovernment'sinitialannouncementabouttheNSL.Werecruited1,424HongKongrespondentsfromtheonlinepanelmaintainedbytheRakutenInsight,aglobalsurveycompanyandthesurveyswerescriptedinQualtrics.Thesamplingstrategywasquotasampling,takingintoconsiderationHongKong'spopulationdistributionsofageandgender.Footnote12 TomakesurezerosuspicionsabouttheNSL'sadoptionandwideawarenessofitamongHongKongers,wewaiteduntil10June,nearly2weeksafterthelawhadbeenofficiallypassedattheNationalPeople'sCongressinBeijing,toadministerthesecondround,whichendedon26June.Asforthesampleattritionofthepost-NSLround,giventheshortspanbetweenthetwosurveys,wewereabletoretainalmost90%ofourpre-NSLrespondentsforourpost-NSLround(1,256).Footnote13 Weincludedinbothsurveysquestionsregardingrespondents'basicdemographics(e.g.,gender,age,andeducation),attitudesandbeliefsregardingHongKong'seconomicandpoliticalprospects,protestexpectations(tobedetailedinSection5.2andAppendixC),institutionaltrust(ourmainvariablesofinterest),andpoliticalstances.SeeTableB.1inAppendixBforhowourvariableswereconstructedfromthesequestions.WhileourempiricaldesigncertainlyenjoyedtheadvantageofhavingtheNSLasanaturalexperiment,whatshouldbenotedhereisthatsolicitingpeople's(includingbothprotestersandnon-protesters)politicalattitudesinacontentiouscontextwherethefreedomofexpressionwaspotentiallyunderthreatcouldbedifficult.Toaddresstheissue,wenotonlyavoidedaskingourrespondentsoverlysensitivequestionssuchastheirpreviousprotestparticipation,butalsoemployedseveralde-identificationmeasurestoprotectthem(seeAppendixAfordetails). 4.2Estimation Thepaper'smainobjectivesaretoestimate(1)theheterogeneouseffectofNSLoninstitutionaltrustaccordingtoone'ssenseofregimeinclusivenessand(2)thecorrelationbetweenone'spost-NSLprotestexpectationandhisorherinstitutionaltrust.Toestimatesucheffects,wefirsttappedintotheexogenousvariationintheinstitutionaltrustinducedbytheNSL.Foreachinstitutionj,individuali'sinstitutionaltrustYattimetcanbeempiricallymodeledas: (1)$$Y_{ijt}=\betaNSL_t+\lambdaC_{it}+\alpha_{it}+\gamma_{ij}+\delta_{\,jt}+\varepsilon_{ijt}$$ NSLisanindicatorvariablewith1denotingthepost-NSLperiodanditscoefficientβcapturestheaverageeffectoftheNSLasashocktoi'strustininstitutionj.Cisthevariablethatdenotesi'spost-NSLshockinhisorbeliefofHongKongsociety'scollectiveactionpotential,anditscoefficient,λ,istheparameterofinterestthatcapturesC'seffectsoni'sinstitutionaltrust.Ifsignificant,ourtheorypredictsλ'ssigntobepositive.Themodelalsoincludesseveralcontrolvariables.First,αitisasetofindividual-timefixedeffectsthatcapturesothertime-varyingindividual-levelattitudesandbeliefsthatmightalsoaffectone'sinstitutionaltrust.Incontrast,δjtisasetofinstitution-timefixedeffectsthataccountsfortheinfluencearisingfrominstitutionsthemselves.Forexample,aninstitutionalchangesuchastheongoingelectoralreforminHongKongthatdrasticallychangeshowtheChiefExecutiveandLegComembersareelectedwilldefinitelyhaveanimpactonpeople'strustinthem.Inaddition,γijisacollectionoftime-invariantvariablesatbothindividualandinstitutionallevelsthatmightalsoaffectinstitutionaltrust.Forinstance,dependingonone'sviewtowardthecolonialperiod,heorshemightplaceahigherorlowertrustfromtheverybeginningintheinstitutionsthatarepartofthecoloniallegaciescomparedtothosecreatedafterthehandoverin1997.Inotherwords,theinstitutionaloriginmightplayaroleindeterminingone'sinstitutionalpreferences.Alternatively,thedemographicvariablesthatdon'tchangeovertimecannotbeignoredeither.Thelastvariableɛijtistheerrorterm. Theestimationofβisstraightforward.WetookadvantageoftheNSLasanexogenousshocktothevastmajorityofHongKongersbetweenthetwosurveysbypoolingtheirresponsestogethertoobtainindividual-wiserepeatedobservationsoverthetwoperiods,i.e.,apaneldatastructure.SinceCwasonlymeasuredinthepost-NSLperiod,thepanelspecificationisgivenby: (2)$$Y_{ijt}=\betaNSL_t+\alpha_{it}+\gamma_{ij}+\delta_{\,jt}+\varepsilon_{ijt}$$ Moreover,toestimateλin(1),wefirstnoticethattheindividual-institutionfixedeffects,γij,canbereadilyeliminatedbyfirst-differencing(1)toobtain: (3)$$\DeltaY_{ij1}=\lambdaC_{i1}+\alpha_i^{\prime}+\delta_j^{\prime}+\varepsilon_{ij}^{\prime}$$ whereΔYij1 = Yij1 − Yij0indicatesthedifferencebetweeni'spre-NSLandpost-NSLtrustininstitutionj,and,giventheabsenceofanycollectiveactionshockbeforethelaw(Ci0 = 0),ΔCi1 = Ci1.Furthermore,$\alpha_i^{\prime}=\alpha_{i1}-\alpha_{i0}$accountsforotherNSL-inducedeffectsonone'sinstitutionaltrust,and$\delta_j^{\prime}=\delta_{j1}-\delta_{j0}$capturespost-NSLinstitutionalshocks.Finally,theerrortermisalsore-specifiedas$\varepsilon_{ij}^{\prime}=\varepsilon_{ij1}-\varepsilon_{ij0}$.Whatshouldbenotedhereisthat,giventheextremelyshortspanbetweenthetwosurveys(about2weeks),institutionalshockswereessentiallynon-existentforthisstudyandtherefore$\delta_j^{\prime}$shouldbesettozerotoyield: (4)$$\DeltaY_{ij1}=\lambdaC_{i1}+\alpha_i^{\prime}+\varepsilon_{ij}^{\prime}$$ Finally,asourtheoriespredict,theeffectsoftheNSLshouldbeheterogeneousbetweenrespondentswithdifferentperceptionsofregimeinclusiveness.Toincorporatetheeffectheterogeneities,theestimationequationisfurthergeneralizedtobeasfollows: (5)$$\DeltaY_{ij1}=\left\{{\matrix{{\lambda_{PD}C_{i1}+\alpha_i^{\prime}+\varepsilon_{ij}^{\prime}}&{{\rmPro}\hbox{-}{\rmDemocracy}}\cr{\lambda_{PE}C_{i1}+\alpha_i^{\prime}+\varepsilon_{ij}^{\prime}}&{{\rmPro}\hbox{-}{\rmEstablishment}}\cr}}\right.$$ 5.Empiricalfindingsandrobustnesschecks 5.1NationalSecurityLawandinstitutionaltrust Tomeasureinstitutionaltrust,weaskedrespondentsinbothpre-andpost-NSLsurveystoratetheirlevelsoftrustineightgovernmentinstitutionsusinga100-pointscalewhere0denotes‘notrustatall’and100denotes‘fulltrust.’Bothcentral-level–theCentralGovernment,theLiaisonOffice,thePeople'sLiberationArmy(PLA)–andlocalinstitutions–theChiefExecutive,theCourt,theLegCo,thePolice,andtheRegistrationandElectoralOffice–wereincluded.Wetookanaverageofone'slevelsoftrustinthethreeCentralGovernmentinstitutionsthatweredirectlyresponsiblefortheNSLlegislationandcrackdowntoconstructourmaindependentvariableofinterest,AverageCentralGovernmentTrust(ACGT). ToformallyestimatetheNSL'seffectsontheaveragegovernmenttrust,wetookadvantageofourrepeatedobservationsofthesamesetofrespondentsrightbeforeandaftertheLawwaspassedbypoolingtogetherthedatafromthetwosurveysforourregressionalanalyses.Moreover,totestourtheoreticalargumentabouttheNSL'sheterogeneouseffects,wefurtherdividedourrespondentsfurtherintothreegroupsofpoliticalstances:(1)pro-democracy(PRO_DEMO,self-perceivedregimeoutcast),(2)pro-establishment(PRO_ESTAB,self-perceivedregimeinsider),and(3)nostance(NO_STANCE)(seeTableB.1inAppendixBfordetails).Footnote14Givenourtripartitetypologyofpoliticalstances,wesettheno-stancetobethebaselinefortheanalyses.Footnote15 OurfindingsnotonlyconfirmtheNSL'seffectsonone'sinstitutionaltrust(Inclusivenesshypothesis),butalsopaintaverypolarizedHongKong.AsTable2(model(1)baseline)documents,theNSLinducedamildupwardshiftbyroughly6%forno-stancersintheiraverageinstitutionaltrustintheCentralGovernmentinstitutions.Furthermore,Figure1visualizesmoreclearlytheNSL'sheterogeneouseffectsvis-á-visisthebaselinebetweenpro-democracyandpro-establishmentHongKongers.Ontheonehand,theNSLsubstantiallyreducedtheformer'strustbylargemargins(−26%),whileinduceddiametricallyopposedeffects(17%)oninstitutionaltrustforthosewholeanedtowardtheestablishmentontheother. Table2.NSL'seffectsoninstitutionaltrust:ACGT Robuststandarderrorsinparentheses. ***P<0.01,**P<0.05,*P<0.1. Figure1.NSL'seffectsoninstitutionaltrustbypoliticalstances:ACGT. Moreover,theestimatesarealsoverystableandrobustwiththeadditionofdifferentsetsofcontrolsFootnote16:(1)model(2):one'sprobabilityassessmentofhowlikelytheelectedoppositionLegComembersweregoingtobedisqualifiedaftertheSeptemberelectionin2020Footnote17(DQ_OPM),(2)model(3):one'sexpectationsofHogKong'sfutureeconomy(HKECON)andsocialwelfare(WELEXP),(3)model(4):demographicvariables(age,gender,education,residentialdistrict,class,andoccupation),and(4)model(5):thefullspecificationthatincludesallthevariablesabovealongwithone'sstanceinthefirstsurvey(PRO_DEMO_PREandPRO_ESTAB_PRE).Footnote18 WhatisworthnotinghereisthatthefullspecificationalsogivesusseveraladditionalinsightsintoHongKongers'institutionaltrustindependentoftheNSL'seffect.AccordingtoTable2,one'sageisfoundconsistentlytobeasignificantpositivepredictorofhisorhertrustlevelintheCentralGovernmentinstitutionsonaverage(models(2)and(5)).WealsofindadditionalcorrelationalpatternsthattherespondentswhotrustedtheCentralGovernmentmoretendedtobethosewhohadlowerexpectationsofHongKong'seconomicprospect,socialwelfareprovision,andDQincidents(models(3)–(5)). TheresultsabovehelpfurtherenrichtheliteratureonhowprotestsandtheirrepressionaffectpublicopinionsbecausetheynotonlyextendittothecaseofHongKong,butalsopaintamorecompletepicturethroughtheeffectheterogeneitiesouranalysesidentify.Ontheonehand,ourresultsabouttheNSL'seffectsonone'sinstitutionaltrustpartiallyechoSangnierandZylberberg's(ReferenceSangnierandZylberberg2017)studyinAfricainfindingtheplungeintrustamongpro-democracyHongKongersduringthepost-repression(NSL)period.Ontheotherhand,despitethroughadifferenttheoreticalmechanism,thesurgeofinstitutionaltrustamongthepro-establishmentcampisconsistentwithFryeandBorisova's(ReferenceFryeandBorisova2019)findingthattheinformationaboutunexpectedprotestscanpositivelyupdateRussiancitizens'perceptionsofthegovernmenttrustworthiness. 5.2Protestexpectations AstworecentrelatedstudiesonEgypt(El-Mallakh,ReferenceEl-Mallakh2020)andRussia(TertytchnayaandLankina,ReferenceTertytchnayaandLankina2020)show,peoplebecamemoreinclinedtovalueorderandstabilityaftertheirexposuretoprotestsandrepression.Inthecontextofourcurrentstudy,thisimpliesthatone'sinstitutionaltrustmightactuallygoupifheorshehadastrongerbeliefintheNSL'sabilitytoreininprotestmobilizationanddampenthesocialsupportfortheMovement.Inotherwords,thisintroducesanotherlayerofeffectheterogeneitythroughwhichtheNSLaffectedone'sinstitutionaltrust.ThefollowingquestioninoursecondsurveyallowedustoinvestigatethisquestionandfurtherunpacktherelationshipbetweentheNSLandone'sinstitutionaltrustinHongKong: HowwouldyouratetheprobabilityfortheNSLtocalmtheAnti-ELABMovementthatbrokeoutinJunelastyear(2019)? Respondentswerethenpresentedwithan11-gradescaleonwhichalargernumberdenotedahigherprobability.Adummyvariableofprotestexpectation,CA,wasconstructedwithresponseshigherthanitsaverage(3.79)designatedaslowexpectations(CA = 0)andotherwiseashighones(CA = 1). BeforereportingtheestimationofCA'seffectsoninstitutionaltrust,weshowbelowhowCAasasummarymeasureofone'soverallprotestexpectationdoesco-varywithhisorherassessmentsforindividualprotestforms.Duringthesecondhalfof2019,wehadwitnessedseveraloftheHongKongprotestersusedtovent,eitherpeacefullyorviolently,theirangerandgrievances.Table3providesacomprehensivesummary(andthevariablenamesweusedintheregressions). Table3.Protestrepertoire Toseeifone'sprotestexpectationmeasuredbyCAwascorrelatedwiththefullgamutofprotestrepertoireaspredicted,weregressedtheformeronrespondents'likelihoodassessment(onan11-gradescale)foreachformofthelatteradoptedbyordinarycitizensandprotesters,respectively(seeAppendixCformoredetailsabouthowweconstructedthevariables),andtheresultsillustratedinFigure2showexactlythat. Figure2.Protestexpectationsandrepertoire. Thatis,despitedifferentextents,acrossalltheprotestformsandforbothcitizensandprotesters,thelessoneexpectedtheNSLtoreinintheintensityoftheMovement,themorelikelyheorshewouldexpectanyoneoftheprotestformstotakeplace.TheseresultsnotonlymakeusreassuredabouttheinternalvalidityofCAasameasureofone'soverallprotestexpectation,butalsorevealthattheprotestoptimistsexpectedmorepeacefulprotestformstotranspire–i.e.,Chanting,Help,and,Yellow. Therearetwomajorfindingsfromourregressionalanalyses–basedontheestimationframeworks(4)and(5)–thathighlightourkeycontributionstotheliterature.Firstofall,Figure3illustratesthelinearpredictionsoffirst-differencedinstitutionaltrustbasedonthestance-specificestimatesforaverageCentralGovernmentinstitutionsfromTable4(model(1)baseline).Thefindingsshowthat,regardlessofone'spoliticalstance,agreaterprotestexpectation,oraweakerbeliefintheNSL'sstabilizingeffect,wasassociatedwithagreaterdecreaseininstitutionaltrust.Counterintuitively,evenforpro-democracyrespondents,whiletheNSLhadsubstantiallyreducedtheirinstitutionaltrust,thedegreeofthereductionhoweverwassmalleramongthedemocratswhohadstrongerbeliefsintheNSL'sabilitytorestoresocialorderandlowerprotestexpectations.Whilethisconclusiondoesn'tcarryasmuchcasualsignificanceasthepreviousanalysesontheNSL,itconfirmsourinformationalhypothesisthatthelowertheprotestexpectationonehas,thesmallerthedecreaseinhisorherinstitutionaltrustwillbe.ItalsoechoesEl-Mallakh's(ReferenceEl-Mallakh2020)andTertytchnayaandLankina's(ReferenceTertytchnayaandLankina2020)argumentsaboutthepost-protestcraveforstabilityandweshowitherethatthisdesireisalsopositivelyassociatedwithinstitutionaltrust.Footnote20 Figure3.Stance-wisepredictedmarginsbyprotestexpectations:ACGT. Table4.Protestexpectationsandfirst-differencedinstitutionaltrust:ACGT Robuststandarderrorsinparentheses. ***P<0.01,**P<0.05,*P<0.1. Second,whilealowerprotestexpectationisassociatedwithagreaterincreaseininstitutionaltrustforallpoliticalstances,theeffectishoweverheterogeneousacrossthem.Firstofall,usingtheno-stancersasthebaseline,Figure4visualizesCA'smarginaleffects(thecoefficientsoftheinteractiontermsinTable4)onthedifferencebetweenone'spre-andpost-NSLaveragetrustforCentralGovernmentinstitutionsforpro-democracyandpro-establishmentcamps,respectively.Asitshows,thedifferencesintheincreaseininstitutionaltrustbetweenno-stancersandpro-establishmentariansarestatisticallyignorable.Incontrast,amongpro-democracyrespondents,astheirprotestexpectationsgetlower,thecorrespondingupwardshiftsintheirinstitutionaltrustsare,however,substantiallysmallerinmagnitudethanthoseofotherstances.Fromtheinformationalperspectiveproposedabove,thiseffectheterogeneityimpliesthatdemocratsasself-perceivedoutcastsofauthoritarianinstitutionsdidn'tupdatetheirinstitutionalbeliefswiththeNSL'sshockonprotesters'mobilizationasmuchasself-perceivedregimeinsiders.Footnote21 Figure4.Protestexpectationsandfirst-differencedinstitutionaltrust:ACGT. Finally,wealsoconductedaseriesofrobustnesstestsforCA(models(2)–(4)inTable4)–exceptthatnodemographicvariableswereaddedsincetheywereexactlythesamegivenaveryshortspanbetweenthetwosurveys.Theresultsabovewereprovenverystableandrobusttomorecontrolsofrespondents'otherpoliticalaswellaseconomicattitudes,pre-NSLpoliticalstances,andCA'soriginal11-gradescale.Footnote22 5.3HongKong'slocalinstitutions Asasupplementtothemainresultsabove,weinvestigateinthissectionwhethertheNSLandHongKongers'protestexpectationsalsoaffectedtheirtrustinHongKong'slocalinstitutionsincludedinthisstudy–i.e.,theChiefExecutive(CE),theCourt,theLegCo,thePolice(POL),theRegistration,andElectoralOffice(REO).Firstandforemost,accordingtoTable5andFigure5,althoughtosomewhatdifferentdegrees,theNSL'sheterogeneouseffectsamongpoliticalstanceswerealsopresentinHongKongers'trustinallofthemandthepolarizingtrendbetweendemocratsandpro-establishmentarianswasespeciallysalientamongthelocalexecutiveinstitutions.Whiletheformer'sinstitutionaltrustintheChiefExecutiveandthePolicewasslashedbytheNSLbyroughly20%,itnonethelesswentupbyasimilarmagnitudeamongthelatter. Table5.NSL'seffectsoninstitutionaltrust:HongKong'slocalinstitutions Robuststandarderrorsinparentheses. ***P<0.01,**P<0.05,*P<0.1. Figure5.NSL'seffectsoninstitutionaltrustbypoliticalstances:HongKong'slocalinstitutions. Incontrast,thetrendwaslessobviousforthelocalmonitoringinstitutionsàlaSangnierandZylberberg(ReferenceSangnierandZylberberg2017).Firstofall,whilethedirectionofeacheffectremainedunchangedforbothstances,themagnitudeofthedifferencebetweenthemshrankbyhalfto10%fortheRegistrationandElectoralOffice.Second,asfarastheCourtandtheLegCoareconcerned,theeffectsizeswerefurtherreducedtodigitnumbers.Inotherwords,theNSL'seffectvariednotonlyamongpoliticalstances,butalsoamonginstitutions.AsfarastheCourtisconcerned,thiscontrastisnotsurprising.DespitetheaforementionedcontroversyaboutNPCSC'spoweroffinalinterpretation,HongKong'sCourtisclearlythequintessentialelementofthecommonlawtraditionleftfromthecolonialperiodandacrucialpillaroftheruleoflaw,whichhasalsobecomeanintegralpartofHongKongers'politicalidentity(ChanandChan,ReferenceChanandChan2006).Finally,theresultspresentedabovearealsoveryrobusttodifferentmodelspecificationswhenweaddedmorecontrolsincludingrespondents'demographics,otherpoliticalandeconomicattitudes,andpre-NSLpoliticalstances(seeTablesD.11–D.14inAppendixD.3.1formoredetails). Asfortheeffectsofone'sprotestexpectation(CA),Figure6againvisualizesitsstance-wiselinearpredictionsoffirst-differencedinstitutionaltrustforthesamesetoflocalinstitutionsinHongKongbasedontheregressionestimatesinTable6.JustaswhatwehavefoundfortheCentralGovernmenttrustabove,theresultsarealsoconsistentwithourinformationalhypothesisthat,forallpoliticalstances,ahigherexpectationoffutureprotestsisassociatedwithalowerlevelofinstitutionaltrustforallthelocalinstitutionsacrosstheboard.Footnote23 Figure6.Stance-wisepredictedmarginsbyprotestexpectations:HongKong'slocalinstitutions. Table6.Protestexpectationsandfirst-differencedinstitutionaltrust:HongKong'slocalinstitutions Robuststandarderrorsinparentheses. ***P<0.01,**P<0.05,*P<0.1. Moreover,wealsogainednewinsightsfromdelvingintoCA'sstance-specificmarginaleffects.Firstofall,asFigure7demonstrates,wedidfindthesamekindofeffectheterogeneityamongpoliticalstancesforHongKong'slocalinstitutions.Comparedtono-stancersandpro-establishmentarians,CA'snegativeeffectonone'sinstitutionaltrustwasdampenedamongdemocrats.Moreover,amongpro-democracyrespondents,thereexistedanothereffectheterogeneitybetweenthemonitoring–i.e.,theCourt,theLegCo,andtheRegistrationandElectoralOffice–andtheexecutiveinstitutions–i.e.,theChiefExecutiveandthePolice.AsFigure7shows,comparedtono-stancers,the(dampening)effectademocrat'sprotestexpectationinducesissmallerfortheformervis-á-visisthelatter.Basedonthesameinformationalperspective,thisimpliesthattheCA-relatedbeliefforaparticularinstitutiontoremainrelativelyunaffectedinthepost-NSLerawassubstantiallystrongerregardingthemonitoringonesasopposedtotheothers,andthereforeallowedtheformertoretaintrustwhenpeopleexpectedfewerproteststotranspire.Finally,theseresultsarealsoveryrobusttodifferentmodelspecifications(seeTablesD.15–D.17inAppendixD.3.2formoredetails). Figure7.Protestexpectationsandfirst-differencedinstitutionaltrust:HongKong'slocalinstitutions. Combined,thetwoeffectheterogeneitiespresentedabovehelpshedlightonHongKong'spost-NSLpoliticallandscape.Ontheonehand,itechoesRozenasandZhukov(ReferenceRozenasandZhukov2019)thattheregimesupport–eitherinstitutionaltrustorloyalty–underdictatorshipsiscloselyrelatedtotheregime'scredibilityinsuppressingpoliticaldissent.Moreover,thiseffectisdampenedwhentheregimeinclusivenessislow.Ontheotherhand,whiletherewasclearlyanNSL-inducedbifurcationinthetrustintheexecutiveinstitutionsbetweenpro-democracyandpro-establishmentHongKongers,bothcamps,however,stillsharedincommonsimilarlevelsoftrustinmonitoringinstitutions. 5.4Robustnesschecks WereportmoretestsinthissectionFootnote24tomakesureourresultsarerobustnotonlytodifferentsetsofcontrols,butalsotootherpotentialissuesthatmightbiasourempiricalestimations. 5.4.1Attritionanddatacensoring Wefirstaddresstwosample-relatedissues.Tobeginwith,despiteourlowattritionrateofroughly10%,thefactthattheNSLcouldaffectthefirst-roundrespondents'decisionstoacceptorrejectoursecondsurveyinvitationmightintroduceaselectionbiasandmakethesamplenonrandomwithrespecttotheNSLasanexternalshock.Toaddressthisissue,weadoptedLee(ReferenceLee2009)boundsestimatorthatestimatesanintervalforthetruevalueofthetreatmenteffectinthepresenceofnonrandomsampleselection.AsFigure8shows,weobtainfairlynarrowestimatedtreatment-effectboundsfortheACGT–−9to − 24forpro-democracyand17–27forpro-establishmentHongKongers.Whiletheformerimpliesthatourbaselinemodel(model(1)inTable2)slightlyoverestimates(intheabsolutesense)themagnitudeoftheNSLeffectamongdemocrats,theLee-boundsestimatorstillconfirmstherobustnessofthedirectionsandstatisticalsignificanceofthehypothesizedNSLeffectsacrossallpoliticalstances.Footnote25 Figure8.Treatment-effectbounds:ACGT. Moreover,Figure9presentstheestimatedLeeboundsoftheNSLeffectsforHongKong'slocalinstitutionsrespectively.Ontheonehand,themajorityofthebaselineestimatesinTable5fallwithintheestimatedboundsexceptthosefordemocrats'institutionaltrustintheChiefExecutiveandthePolicethatslightlyexceedthelowerbounds. Figure9.Treatment-effectbounds:HongKong'slocalinstitutions. Inaddition,estimatingtheeffectsoftheNSLoninstitutionaltrustcriticallydependsoncomparingone'schoicesovertrustlevelsbetweenthetwosurveys.Naturally,whiletheywereindependentlyadministered,thesizeofthedifferenceforsomerespondentsmight,however,beartificiallylimitedbythechoicerangeprovidedtotheminthesecondround.Forexample,foronewhoalreadyratedagiveninstitutionverylow(high)duringthefirstround,thefactthattherangeofvaluestobechosenwasboundedby0(100)frombelow(above)hadartificiallyforcedthedifferencebetweenhisorherpre-andpost-NSLtrustlevelstobesmallerthanitcouldhavebeenifthelower(upper)boundwasabsent.Inotherwords,thewayinwhichourquestionnairewasdesignedcensoredourdependentvariablefrombothaboveandbelow,andthiscouldbiasourresults.Toaddressthissecondissueinoursample,were-estimatedthembyrunningTobitmodelswiththecutoffssetat0frombelowand100fromabove. Figures10aand10bshowthattheeffectsizesaresubstantiallymagnifiedfortheNSL'snegativeimpactsonpro-democracycamp'sACGTacrossallmodelspecificationsandtrustinHongKong'slocalexecutiveinstitutions–theChiefExecutiveandthePolice–,butthepatternsremainthesame.Inotherwords,ourpreviousfindingsaboutthebifurcatedeffectstheNSLhadonpro-democracyandpro-establishmentcamps'institutionaltrustsarerobusttothedatacensoringinourdesign.WealsoestimatedseveralaugmentedspecificationswithadditionalcontrolsforHongKong'slocalinstitutions,andtheresultsstillhold.SeeAppendixE.1fordetails. Figure10.CoefficientsplotsfortheNSL'seffectsoninstitutionaltrustsbypoliticalstances(Tobitestimates). 5.4.2Endogeneity Furthermore,sinceone'spoliticalstancechoicecouldbeendogenoustohisorherinstitutionaltrust,wetookamachine-learningapproachbyusingtheKNN(K-nearestneighbors)algorithmtofindatotallyexogenousproxyforone'spoliticalstance.Foreachrespondent'schoiceofstance,thealgorithmdeterminedwhichotheroneswere‘nearest’toitintheparameterspacecomposedofthevariablesofeconomicandpoliticalattitudesinthesurveysthatwereessentialtoidentifyingits‘neighbors,’i.e.,thosewhoheldverysimilarattitudeswiththerespondentinquestionalongthesedimensions.Thealgorithmthenassignedapredictedpoliticalstancebasedonthemostfrequentchoiceamongtheidentifiednearestneighbors.Thisapproachexploitsthefactthatindividualsurveyresponsesinoursurveyswereindependentofeachother,andthereforeone'snewlyassignedpoliticalstancebasedontheKNNalgorithmwouldbeentirelyexogenoustohisorherchoiceofinstitutionaltrust.Footnote26 AsFigure11ashows,whilethemagnitudeoftheNSLeffectsontheACGTacrossallspecificationsamongpro-democracycampisslightlylarger,theirsignsandstatisticalsignificanceremainthesameasthebaselineestimationsabove.Forthesamesetofinstitutions,theresultsinFigure11balsoshowthatourbaselineestimatesfortheeffectsofprotestexpectationsarerobust.RegardingHongKong'slocalinstitutions,wecanfindfromtheestimatesdocumentedinFigures12aand12bthat,despitesomedifferencesinmagnitude,forboththeeffectsoftheNSLandprotestexpectations,theyareveryconsistentwiththebaselineresults.Combined,alltheKNN-basedestimationsaboveconfirmtherobustnessofourpreviousempiricalfindingstopotentialendogeneity.Footnote27 Figure11.KNN-basedestimations:ACGT. Figure12.KNN-basedestimations:HongKong'slocalinstitutions. 5.4.3DifferentcutoffsforCA(protestexpectations) Inourthirdsetofrobustnesstests,wetriedtoseeifourmainresultsforthefirst-differencedinstitutionaltrustintheCentralGovernment(ACGT)remainedstableunderdifferentcutoffchoicesfordichotomizingCA.Footnote28Weadoptedtwoapproaches.Firstofall,sinceourinitialcutoffwas4(>3.79),wecompareditst-statisticwiththecounterpartsunderthecutoffsof3and5,respectively,forthebaselinemodel(1)inTable4–i.e.,thecoefficientoftheinteractiontermbetweenprotestexpectationsandthestancedummyofpro-democracycamp.AsFigure13a–theredbarrepresentsthet-statisticsbasedonourinitialchoice–shows,changingthecutofftoeitherdirectiondoesn'taffectthestatisticalsignificancefortheothercoefficients.ThisimpliesthatourbaselineresultisrobusttodifferentCAcutoffs. Figure13.DifferentcutoffsofCA. Alternatively,wealsoadoptedDeatonandCartwright's(ReferenceDeatonandCartwright2018)andCantonietal.'s(ReferenceCantoni,Yang,YuchtmanandZhang2019)approachtorandomlyassign(fictional)CAgroupingstooursamplesandestimatedtheireffectsfor10,000times.Next,thet-statisticsfromtheestimatedeffectsforallthefictionaltrialswerederivedandcomparedtothatundertheinitialCAcutoffforthesamecoefficient.Thismethodhelpsusconfirmthattheeffectwefoundisstatisticallyverydifferentfromthosebasedonrandomcutoffsandthereforeisrobust.Figure13b–theverticalredlineindicatesthet-statisticsfromtheinitialCAassignment–clearlyshowsthatourresultisfarfromrandom. 6.Concludingremarks InthewakeoftherecentmilitarycoupinMyanmar,wehavewitnessedagainhowdictatorsdealtwiththeiropponentsathome.Whileweoftenseeviolentrepressionsindictators'toolkit,theyarecertainlynottheonlymeansthroughwhichauthoritarianstabilityisrestoredoranewdictatorshipiscrafted.Infact,HongKongisexactlysuchacasewheretheauthoritariangovernmentistryingtolegitimizeitsnot-so-violentrepressiveactsthroughtheexistinglegalandpoliticalframework.Thisiswhy,evenwhenmostpeoplewouldunambiguouslyannouncethe‘OneCountry,TwoSystems’frameworkdead,Beijinghasinsistedonusingthenameandrefusedtoreplacethecurrentframeworkwithacompletelynewone. Thesuccessofsuchastrategy,however,criticallyhingesonhowtrustworthythepost-repressioninstitutionsareamongHongKongers.Ourstudy,therefore,providesacomprehensiveandtimelyassessmentofthecurrentstatusofHongKong'sinstitutionallandscapeandananswertothequestionthispaperbeganwith:‘Howdoesrepressiononoppositionprotestsaffectcitizens’institutionaltrustunderdictatorships?’Ourfindingsaretwofold.Firstofall,theNSLasasoftrepressiondroveawedgeintheHongKongsocietybymakingthepro-establishmentcamp(self-perceivedregimeinsiders)–andnostancerstoalesserdegree–moresatisfiedwiththepost-NSLinstitutionsontheonehand,whilealienatingthepro-democracycamp(self-perceivedregimeoutcasts)wholosttremendoustrustinthemontheother.Thisbifurcationconfirmsourregimeinclusivenesshypothesis.Fortheliteratureoninstitutionaltrust,thefindingsuggeststhattheeffectofrepressiononinstitutionaltrustunderdictatorshipsissubstantiallycorrelatedwithone'ssenseofregimeinclusiveness.Especially,duringthetimeofturmoil,long-termideologicalinclinationshavebecomelessrelevantthanshort-termpoliticalfluctuationsasadeterminant,ourtimelystudyshowsthatadictator'sability(orinability)tocooptcitizenscanbecriticalindetermininginstitutionaltrustinaninstitutionalenvironmentwithoutprocedurallegitimacy. Second,ourstudyalsorevealsthatone'strustininstitutionsissignificantlyassociatedwiththeregimes'abilitytocurbprotesters'contentiousmobilizationandrestoresocialstability.Morespecifically,asourinformationalperspectivepredicts,wefindthattheHongKongerswhohadhigherconfidenceintheNSLtoreininprotestswouldalsohaveagreaterleveloftrustthanthosewhodidn't.Moreover,whilethiscorrelationappliestoallthepoliticalstripes,theeffectwassubstantiallysmalleramongpro-democracyHongKongersexceptfortheirtrustinmonitoringinstitutions(i.e.,theCourt,theLegCo,andtheRegistrationandElectoralOffice).Thiseffectheterogeneityimpliesacriticalpoliticaldynamicsinpost-NSLHongKong.Onbalance,aslongastheNSLwasviewedasaneffectivetoolinrepressingprotests,ithelpedcommandpro-establishmentarians'higherinstitutionaltrustsacrosstheboard.Asfordemocrats,whiletheNSLdidwrecktheirtrustsinalltheexecutiveinstitutionsinvolvedinenforcingBeijing'spoliticalagenda,itdidn'tmakethemequallydisillusionedwiththemonitoringinstitutions. ThisfindingillustratesnotonlyEscribá-Folch's(ReferenceEscribá-Folch2013)generalargumentaboutthesoftrepressioninHongKong'spoliticalcontext,butalso,moreprofoundly,thesocialfoundationforapotentialroadmapaccordingtowhichanewdictatorshipcanbecrafted.Namely,asBeijingistransformingHongKong'spoliticalsystemfromwithininhopesofbringingaboutanewequilibriumtoitspoliticswithoutstrongresistance,exploitingthepro-democracycamp'sremainingtrustinthemonitoringinstitutionsseemstobeoptimal.SomeexampleswehavewitnessedsincethepassageoftheNSLincludeconvictingdissidentsandprotestersthroughtheCourt,Footnote29formingnewpro-Beijingpoliticalparties,Footnote30andmanipulatingtheelectoralrulestheRegistrationandElectoralOfficewillbeinchargeofadministeringfortheChiefExecutiveandtheLegCoelections.Footnote31Whilethesepoliticalmaneuveringsarecertainlyobjectionabletopro-democracyHongKongers,thewidesupporttheseinstitutionsenjoycanstillallowBeijing'sinstitutionalengineeringtobegraduallyunfoldedanderodeHongKong'sprotest-basedmodelofaccountability. Supplementarymaterial Thesupplementarymaterialforthisarticlecanbefoundathttps://doi.org/10.1017/S1468109921000293andhttps://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/ZCLECH.Conflictofinterest Theauthorsdeclarenone. Footnotes1JonathanHead.‘Thailand'sconstitution:Newera,newuncertainties.’BBC,7April2017.https://www.bbc.com/news/world-asia-39499485(accessed:20April2021).2Thisstudyincludedthefollowinginstitutions:TheCentralGovernment,theChiefExecutive,theCourt,theLegislativeCouncil(LEGCO),theLiasionOffice,thePeople'sLiberationArmy,thePolice,andtheRegistrationandElectoralOffice(REO).3WeviewtheNSLassoftonlyinrelativetermsvis-á-visaTiananmen-stylecrackdownoracompletetakeover.WhilethereiscertainlynodenyingofBeijing'sheavy-handedeffortstouseNSLasalegalmeanstoputalmostallmajorpoliticaldissidentsbehindbars,thereisstillaqualitativedifferencebetweentheNSLasarepressiveinstrumentandviolentquasi-militaryactions.4Whatshouldbenotedhereisthatthispartofourfindingsisonlycorrelational.WhiletheNSLasarepressivemeansdidcomeasanexternalshock,whichmadeitseffectoninstitutionaltrustcausal,one'sassessmentoftheNSL'seffectonprotestswashoweverself-reportedandnotmanipulatedinthesurvey.Asaresult,apartfromtheinformationalperspectiveweprovidedabove,thecausaldirectioncanalsogotheotherwayaround.Thatis,wecannotexcludethepossibilitythatthosewhohadlower(higher)institutionaltrustwouldalsounder(over)-estimatetheNSL'sabilitytosuppresstheprotests.5JohnSudworth.‘China'sparliamentremakesHongKonginitsownimage.’BBC11March2021.Availableat:https://www.bbc.com/news/world-asia-china-56364912(accessed:30March2021).6Forexample,notonlydidHitlercometopowerlegally,buttheEnablingActin1933heusedtonailthecoffinoftheWeimardemocracywasalsoprocedurallylegitimate(LhrmannandLindberg,ReferenceLhrmannandLindberg2019:1105).7NatalieWongandJeffieLam.‘HongKong'snationalsecuritylaweightmonthson:arrests,movestoensure“patriots”takecharge,universitycutsoffstudentunion.’SouthChinaMorningPost28February2021.Availableat:https://www.scmp.com/news/hong-kong/politics/article/3123427/hong-kongs-national-security-law-eight-months-arrests-moves(accessed:30March2021).8WebynomeanstrytosuggestherethatthecolonialperiodinHongKongwasnecessarilymoredemocraticandliberalthanitistoday.Theconclusionsofourstudydonotdependonthecomparisonineitherway.Whatwewouldliketoemphasizehere,however,isthat,comparedtotheBritishcolonizers,Beijingdidtrytosetupadifferentkindofhybridregimebyintroducingnewauthoritarianinstitutions.9Despitethecontinuity,Fong(ReferenceFong2013),however,alsofindsachangeinthepost-handoverstate-businessrelationshipwherebusinesspeopleweresubstantiallysidelined.10DamienGayle,KateLyons,VernaYu.Hong‘Kongprotest:policefireteargasatdemonstrators–asithappened.’13June2019.Availableat:https://www.theguardian.com/world/live/2019/jun/12/hong-kong-protest-demonstrators-and-police-face-off-over-extradition-bill-live(accessed:22April2021).11Thearticleisavailableathttps://www.moodys.com/research/Moodys-affirms-Hong-Kongs-Aa2-ratings-maintains-stable-outlook--PR_403359?fbclid=IwAR3rPvPQQDZ3PrpiY_HvEdEVhnZabB2B9W_Ix-iXRJBm5dGcx8L4qWsP2QA(accessed:22April2021).12SeeTableB.1inAppendixBforboththesampleandthepopulationdistributionsinage,education,gender,andresidentialdistrict.Whilebothdistributionsaresubstantiallyclose,asmostonlinesurveys,oursampleisstillabityoungerandmoreeducatedthantheactualpopulationinHongKong.TheissueisaddressedinAppendixE.4byre-estimatingourmodelsusingthesamplere-weightedaccordingtothepopulationdistributionsabove.13Ourtwo-roundsamplenotonlyhasalowattritionrate,but,asTableB.1documents,thepost-NSLsampledistributionisalsoveryclosetothepre-NSLone.Tomakesureourresultswererobusttodifferentattritionscenarios,wefurtherimplementedtheLeeboundsestimator(Lee,ReferenceLee2009)inSection5.4.Wegreatlyappreciateananonymousreviewerforthesuggestion.14Inbothsurveys,respondentswereaskedtochooseapoliticalstanceaccordingtoafive-wayclassification:Centrist,Democrat,Establishmentarian,Localist,andNo-Stance.Toavoidalargemeasurementerror,wefurtherbundledtheLocalistwiththeDemocratrespondents(‘Pro-Democracy’)andtheCentristwiththeEstablishmentarianones(‘Pro-Establishment’)respectivelytomakeitathree-wayclassificationforourfollowingregressionalanalyses.Moreover,whathastobenotedhereisthatsomepeopleactuallyswitchedfromtheirpre-NSLpoliticalstancestoothersinthepost-NSLsurvey,whichwereusedintheourstatisticalanalyses.ThechangescouldbeinducedbytheNSLandimpliedthatone'schoicesofpoliticalstanceandinstitutionaltrustcouldbecorrelatedwitheachother.Weaddressthisissueinthesectiononrobustnesscheckstomakesureourestimatesarerobusttosuchcorrelations.15Descriptively,thestance-wisecomparisonsinTableB.2(AppendixB)establishthatthereareverysubstantialdifferencesinaverageinstitutionaltrustsatthecentrallevelamongthethreepoliticalstancesatbothaggregateanddisaggregatelevels.Theaveragelevelsoftrustthepro-establishmentrespondentsplacedinCentralGovernmentinstitutionsduringbothsurveyswerearound50vis-á-visis9givenbythepro-democracycounterparts.Asfarasno-stancersareconcerned,theirinstitutionaltrustslayrightin-betweentheothertwostancesacrosstheboard.Moreover,consistentwithourpredictionoftheNSL'sheterogeneouseffectsacrosspoliticalstances,thestripplotsinappendedFiguresB.1andB.2(reddotsrepresentingthemeans)indicatethat,onaverage,theNSLcausedpro-democracyrespondentstolowertheirinstitutionaltrustsacrosstheboard,butexertedoppositeeffectsonpro-establishmentandno-stanceones,withtheformershowingagreatermagnitude.16SeeAppendixD.1fortheresultsofindividualCentralGovernmentinstitutions.Theempiricalpatternsremainunchangedandequallyrobust.17Afteroursurveyswereadministered,theelectionwasthenpostponedfirsttoSeptember2021duetothepandemic,andthentoDecember2021againtogainmoretimefortheelectoralreformsBeijinginitiated.18Sinceone'spre-NSLstancewascertainlyindependentoftheNSL,controllingforitintheregressionsmakestheresultsrobusttohowmuchone'sstancewasswayedbytheNSL.19InthecolorpoliticsoftheAnti-ELABMovement,‘yellow'wasusedtolabelthose(politicians,shopowners,opinionleaders,etc.)whosympathizedwiththeMovement,while‘blue’usedforthosewhodisapprovedofit.20Whatshouldbekeptinmindisthat,sincethefindingsareonlycorrelational,itisalsolikelythatthosewhohadhigh-institutionaltrustsshowedstrongerconfidenceintheNSL'sstabilizingeffects.Ourdatalimitationmakesitimpossibletomakeacausalinferenceonthispartoftheresults,butitissurprisingenoughtofindsuchacorrelation.21WhileitmightappearinFigure4thattheeffectsizesarelargeramongpro-democracyrespondents,itshouldbenotedthatCA'sstance-specificeffectforeachinstitutionisthesummationofthismarginaleffectandthebaselineeffect(i.e.,CA'scoefficientinTable4).22SeeTablesD.6–D.9andFigureD.1inAppendixD.2forthelinearpredictionplotsandregressionresultsforindividualCentralGovernmentinstitutions.Moreover,theregressionresultswithCAasacontinuousmeasurearealsodocumentedinTableD.10inthesamesection.23Whatisworthnotinghereisthattheeffectofpro-establishmentrespondents'protestexpectationsontheirfirst-differencedtrustintheCourtissubstantiallysmallerthanotherinstitutions.BasedontheestimatesinTable6,whilethedifferenceisstillstatisticallysignificant,ahigherexpectationofprotestisonlyassociatedroughlywitha1%reductioninthetrustintheCourtamongpro-establishmentarians.24Tosavespace,weleaveotherrobustnesschecksonissuesofsamplere-weightingandnon-responses(self-censorship)inAppendicesE.4andE.5.25SeeFigureE.1inAppendixE.1fortheboundsestimatesfortheindividualCentralGovernmentinstitutions.Theresultsalsoconfirmtherobustnessofourbaselineestimations.26SeeAppendixE.2.1forthelistofvariablesweusedforestimatingattitudinaldistancetofindingone'snearestneighbors,stance-wisepercentagesofmatchesbetweenoriginalandKNN-basedclassifications,andtheregressionresultsforalltheKNN-basedestimationswhosecoefficientplotsarepresentedbelow.27Alternatively,wealsoconductedanotherrobustnesstestbyusingone'spre-NSLpoliticalstanceinsteadforestimatingtheeffectofprotestexpectationsoninstitutionaltrust.SeeAppendixE.2.2fortheresults.Sinceone'sownpre-NSLpoliticalstanceisentirelyexogenoustohisorherpost-NSLinstitutionaltrust,thefactthattheregressionresultsremainconsistentandstableconfirmsagainthatourbaselineestimationsarerobustnesstoendogeneity.28Tosavespace,SeeFiguresE.2andE.3inAppendixE.3fortheresultsofotherinstitutions.29EmilyFeng.‘9HongKongPro-DemocracyActivistsSentencedFor2019Protests.’NPR16April2012.Availableat:https://www.npr.org/2021/04/16/987959427/9-hong-kong-pro-democracy-activists-sentenced-for-2019-protests(accessed:22April2021).30JeffPao.‘Newpro-BeijingpartytoseekHKheartsandminds.’AsiaTimes10March2021.Availableat:https://asiatimes.com/2021/03/new-pro-beijing-party-to-seek-hk-hearts-and-minds/(accessed:30March2021).31JohnSudworth.‘China'sparliamentremakesHongKonginitsownimage.’BBC11March2021.Availableat:https://www.bbc.com/news/world-asia-china-56364912(accessed:30March2021).References Boix,CandSvolik,MW(2013)Thefoundationsoflimitedauthoritariangovernment:institutions,commitment,andpower-sharingindictatorships.JournalofPolitics75,300–316.CrossRefGoogleScholar Cantoni,D,Yang,DY,Yuchtman,NandZhang,YJ(2019)Protestsasstrategicgames:experimentalevidencefromHongKong'santi-authoritarianmovement.QuarterlyJournalofEconomics134,1021–1077.CrossRefGoogleScholar Chan,JandChan,E(2006)ChartingthestateofsocialcohesioninHongKong.ChinaQuarterly187,635–658.CrossRefGoogleScholar Cheng,EW(2016)Streetpoliticsinahybridregime:Thediffusionofpoliticalactivisminpost-colonialHongKong.ChinaQuarterly226,383–406.CrossRefGoogleScholar Curtice,TBandBehlendorf,B(2021)Street-levelrepression:protest,policing,anddissentinUganda.JournalofConflictResolution65,166–194.CrossRefGoogleScholar Deaton,AandCartwright,N(2018)Understandingandmisunderstandingrandomizedcontrolledtrials.SocialScienceandMedicine210,2–21.CrossRefGoogleScholarPubMed Desposato,SW,Wang,GandWu,JY(2021)Thelong-termimpactofmobilizationandrepressiononpoliticaltrust.ComparativePoliticalStudiesforthcoming.CrossRefGoogleScholar Dinas,EandNorthmore-Ball,K(2020)Theideologicalshadowofauthoritarianism.ComparativePoliticalStudies53,1957–1991.CrossRefGoogleScholar Easton,D(1975)Are-assessmentoftheconceptofpoliticalsupport.BritishJournalofPoliticalScience5,435–457.CrossRefGoogleScholar El-Mallakh,N(2020)Howdoprotestsaffectelectoralchoices?EvidencefromEgypt.JournalofEconomicBehaviorandOrganization179,299–322.CrossRefGoogleScholar Escribá-Folch,A(2013)Repression,politicalthreats,andsurvivalunderautocracy.InternationalPoliticalScienceReview34,543–560.CrossRefGoogleScholar Fong,BCH(2013)State-societyconflictsunderHongKong'shybridregime:governingcoalitionbuildingandcivilsocietychallenges.AsianSurvey53,854–882.CrossRefGoogleScholar Frye,TandBorisova,E(2019)Elections,protest,andtrustingovernment:anaturalexperimentfromRussia.JournalofPolitics81,820–832.CrossRefGoogleScholar Gandhi,J(2008)PoliticalInstitutionsUnderDictatorship.NewYork,N.Y.:CambridgeUniversityPress.CrossRefGoogleScholar Kim,WandGandhi,J(2010)Cooptingworkersunderdictatorship.JournalofPolitics72,646–658.CrossRefGoogleScholar Kuran,T(1997)PrivateTruths,PublicLies:TheSocialConsequencesofPreferenceFalsification.Cambridge,MA:HarvardUniversityPress.GoogleScholar Lawrence,AK(2016)RepressionandactivismamongtheArabSpring'sfirstmovers:evidencefromMorocco'sFebruary20thmovement.BritishJournalofPoliticalScience47,699–718.CrossRefGoogleScholar Lee,DS(2009)Training,wages,andsampleselection:estimatingsharpboundsontreatmenteffects.ReviewofEconomicStudies76,1071–1102.CrossRefGoogleScholar Lhrmann,AandLindberg,SI(2019)Athirdwaveofautocratizationishere:whatisnewaboutit?Democratization26,1095–1113.CrossRefGoogleScholar Ma,N(2007)PoliticalDevelopmentinHongKong:State,PoliticalSociety,andCivilSociety.HongKong:HongKongUniversityPress.GoogleScholar Madestam,A,Shoag,D,Veuger,SandYanagizawa-Drott,D(2013)Dopoliticalprotestsmatter?EvidencefromtheTeaPartyMovement.QuarterlyJournalofEconomics128,1633–1685.CrossRefGoogleScholar Mazumder,S(2018)ThepersistenteffectofU.S.Civilrightsprotestsonpoliticalattitudes.AmericanJournalofPoliticalScience62,922–935.CrossRefGoogleScholar Neundorf,AandPop-Eleches,G(2020)Dictatorsandtheirsubjects:authoritarianattitudinaleffectsandlegacies.ComparativePoliticalStudies53,1839–1860.CrossRefGoogleScholar Neundorf,A,Gerschewski,JandOlar,R-G(2020)Howdoinclusionaryandexclusionaryautocraciesaffectordinarypeople?ComparativePoliticalStudies53,1890–1925.CrossRefGoogleScholar Rozenas,AandZhukov,YM(2019)Massrepressionandpoliticalloyalty:evidencefromStalin's‘TerrorbyHunger’.AmericanPoliticalScienceReview113,569–583.CrossRefGoogleScholar Sangnier,MandZylberberg,Y(2017)Protestsandtrustinthestate:evidencefromAfricancountries.JournalofPublicEconomics152,55–67.CrossRefGoogleScholar Tertytchnaya,KandLankina,T(2020)Electoralprotestsandpoliticalattitudesunderelectoralauthoritarianism.JournalofPolitics82,285–299.CrossRefGoogleScholar Wang,Y(2021)ThepoliticallegacyofviolenceduringChina'sculturalrevolution.BritishJournalofPoliticalScience51,463–487.CrossRefGoogleScholar Viewincontent Table1.HongKong'scontentiouspolitics:majorepisodessince1997 Viewincontent Table2.NSL'seffectsoninstitutionaltrust:ACGT Viewincontent Figure1.NSL'seffectsoninstitutionaltrustbypoliticalstances:ACGT. Viewincontent Table3.Protestrepertoire Viewincontent Figure2.Protestexpectationsandrepertoire. Viewincontent Figure3.Stance-wisepredictedmarginsbyprotestexpectations:ACGT. Viewincontent Table4.Protestexpectationsandfirst-differencedinstitutionaltrust:ACGT Viewincontent Figure4.Protestexpectationsandfirst-differencedinstitutionaltrust:ACGT. Viewincontent Table5.NSL'seffectsoninstitutionaltrust:HongKong'slocalinstitutions Viewincontent Figure5.NSL'seffectsoninstitutionaltrustbypoliticalstances:HongKong'slocalinstitutions. Viewincontent Figure6.Stance-wisepredictedmarginsbyprotestexpectations:HongKong'slocalinstitutions. Viewincontent Table6.Protestexpectationsandfirst-differencedinstitutionaltrust:HongKong'slocalinstitutions Viewincontent Figure7.Protestexpectationsandfirst-differencedinstitutionaltrust:HongKong'slocalinstitutions. Viewincontent Figure8.Treatment-effectbounds:ACGT. Viewincontent Figure9.Treatment-effectbounds:HongKong'slocalinstitutions. Viewincontent Figure10.CoefficientsplotsfortheNSL'seffectsoninstitutionaltrustsbypoliticalstances(Tobitestimates). Viewincontent Figure11.KNN-basedestimations:ACGT. Viewincontent Figure12.KNN-basedestimations:HongKong'slocalinstitutions. Viewincontent Figure13.DifferentcutoffsofCA.Tungetal.supplementarymaterialAppendix PDF 2MB Tungetal.DatasetDatasethttps://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/ZCLECH Link YouhaveAccessOpenaccess1CitedbyCitedbyLoading... CrossrefCitations Thisarticlehasbeencitedbythefollowingpublications.Thislistisgeneratedbasedondataprovidedby CrossRef. Tung,HansH. and Kasuya,Yuko 2021. Resistingautocratization:theprotest–repressionnexusinHongKong'sAnti-ELABMovement. JapaneseJournalofPoliticalScience, Vol.22, Issue.4, p. 193. CrossRef GoogleScholar GoogleScholarCitations ViewallGoogleScholarcitations forthisarticle. × Cancel Confirm ×