空氣污染的影響及其對(duì)商業(yè)健康保險(xiǎn)需求的關(guān)注

TheImpactofAirPollutionanditsAttentionontheDemandofCommercialHealthInsuranceAbstractAirpollutionhasalwaysbeenahotissueinChinesesociety,anditsimpactonpersonalhealthandeconomicdevelopmentcannotbeignored.Usingtheperspectiveofbehavioralfinance,thisarticlecombinesrealitywithcity-levelpaneldatatostudytheimpactofairpollutionondemandofChinesecommercialhealthinsurance.Specifically,thisarticlestudiestheimpactoftheannualaverageofPM2.5concentrationonpremiumincomeofcommercialhealthinsurance.EmpiricalanalysisfoundthattheriseintheannualaverageofPM2.5concentrationwillsignificantlypromotethedemandforcommercialhealthinsurance.Atthesametime,byconstructingtheindiesbasedonAQI,thisarticlefindsthatwhenthenumberofdayswithanAQIexceeding150inacertainareaincreases,itwillsignificantlypromotethedemandforhealthinsurance.Inaddition,thisarticlealsofindsthatincreasingattentiontowardsPM2.5willalsosignificantlypromotethedemandforhealthinsurance.Thispapervalidatestheaboveconclusionsthroughheterogeneityandrobustnesstests.KeywordsAirpollution,Demandofcommercialhealthinsurance,PM2.5,AQI,BaiduIndexIntroductionAirpollutionanditsimpactonsociallifehavebeenalwaysahotissueinChina'senvironmentalprotectionandeconomicdevelopment.Itisbeenprovedbyresearchthatairpollutioncanimpactonseveralfieldsinsocietyseriously,notonlythephysicalconditionandthebehaviorsofindividuals,butalsotheeconomicdevelopment,whichcannotbeignored.Especiallyforindividuals,thehealthrisksthatairpollutionmaycausecannotbecompletelyavoidedthroughpensionsandcompulsorysocialhealthinsurance.Theessenceofcommercialhealthinsuranceisariskhedgingproductinpersonalassetallocation,whichcanplayagoodroleinevadingrisksforindividuals.However,therearefewpiecesofliteraturethatdirectlyfocusontherelationbetweendemandforcommercialhealthinsuranceandairpollution.Furthermore,fromtheframeworkofbehavioralfinance,bytheresearchfocusontheairpollutionanddemandofcommercialhealthinsurance,themechanismbeneaththefinancialbehaviorsofindividualscanalsobediscovered.Therefore,itisofgreatnecessitytostudytheimpactfromtheperspectiveofbehavioralfinanceandenvironmentaleconomics.Theremaybetwoperspectivesoftheinnovationandmaincontributionsforthisliterature:Firstly,existedstudieshaveselectedtransaction-levelsamplesorprovincialpanelsamplesforresearchduetothedifficultyofobtainingsamples,whichmayleadtobiasesinempiricalresearchresults.Forthisresearch,city-levelpanelsamplesareusedforempiricalanalysis,coveringmostoftheprefecture-levelcitiesinChina,makingtheconclusionsofthisarticlemoreconvincingandmoreconsistenttotheactualsituation.Inthisthesis,convincedheterogeneitytestsareusedtocheckthefinalconclusionsalso.Secondly,thisarticleusesaperspectiveofbehavioralfinancetobuildatheoreticalanalysisframeworkandempiricallyrevealsthatwhenthepublicfacesairpollution,fromthewayofinformationtransmissionorthewayofinformationacquisition,theindividual'sriskperceptioncangenerateavarietyofcognitivebiases,includingdifferentiation,ambiguity,andtimelag,thesebiaswillfinallyaffectthefinalfinancialdecisionoftheindividuals.Inthisthesis,wedrawtheconclusionbyempiricalresearch,thatis,airpollutioncanindeedpromotethedemandofcommercialhealthinsurancesignificantly,andtheimpactscanbedifferentbyregionsandtime.Meanwhile,bytheindiescreatedinthethesisandtheindependentvariablesintroduced,wealsorevealtheviewthatthereiscognitivebiasbetweenairpollutionandthefinancialbehaviorofindividuals,andpropersignalscanhelppublicpreventthecognitivebias.Theconclusionsabovepassedtheheterogeneityandrobustnesstests.Theremainingcontentofthisarticlewillproceedasfollows:thesectiontwoisaliteraturereviewfocusontheissues,furtherthethirdsectioniswillingtoshowthemodelconstruction,variabledescriptionandsummarystatistics,thefourthwillintroducetheempiricalanalysisofthedemandforcommercialhealthinsuranceconsumptionbyairpollution,andfinallyistheconclusion.LiteraturereviewInChina,theissueofairpollutionhasbeentakenseriouslyandwidelyconcernedbysociety.Thisisbecausepeoplehavefeltandrealizedtheadverseeffectsofairpollutiononworkanddailylife,notonlyinsubjectivelywaysbutalsoobjectivelyways.Atthesametime,lotsofresearchanddiscussionshavebeencarriedoutonrelatedissues.Theexistingliteraturehasstudiedthenegativeimpactsofairpollutionfromperspectivesincludingphysicalhealth,mentalhealth,economicexpenditure,socialbehaviour,andsoon.Fromtheangleofindividualhealth,thehugephysicaldamagecausedbyairpollutionissignificant.Intheshortterm,excessivePM2.5,SO2andotherharmfulsubstancescontainedinpollutedairmayincreasetheburdenofbreathingandcauseharmtolungfunction;airpollutioncanalsocausearrhythmiaandsuddenheartdisease;forsensitivepeople(suchaschildren,theelderly,etc.),dirtyairwillincreasetheirdiscomfort(EPA).Inthelongterm,airpollutioncancausechronicheartandlungdisease,respiratoryinfections,andgreatlyincreasetheriskofdevelopingmalignanttumours(EPA,2004).Existingempiricalstudieshaveshownthattheincidenceofneonatal-relateddiseaseswillincreasesignificantlyinareaswithlong-termairpollution(Chay&Greenstone,2003);meanwhile,thecontaminatedairwillincreasetheincidenceofasthma.(Neidell,2004)Andreducedlifeexpectancy(Chenetal.,2013).Airpollutioncanalsoaffectanindividual'smentalhealth,suchassignificantlyreducingtheirhappiness(Huangetal.,2013;Zhengetal.,2019).Airpollutionalsohasimpactsonsocialbehaviourandproductivity.Forexample,thestudyonChineseprofessionalfootballplayersshowsthatairpollutionwillaffectindividualproductivity,especiallytheproductivityofrelativelylowproductivityindividuals(Weietal.,2017);Fu(2018)pointedoutintheresearchthatairpollutionwillcauseadverseimpactsonpercapitavalueaddedtotheenterprises,whichinturnreducestheproductivityoftheenterprise;andresearchbyChenetal.(2017)andSun(2019)confirmedthattheincreaseinPM2.5contentintheairwillleadtopopulationmigrationandaffecttheattractivenessofregionalmigration,Whichinturnaffectsaseriesoffieldssuchashumanresourceallocation.Itisworthnotingthestudyfoundthatifpeopleoftenexposedtopollutedair,theiradrenalcortisollevelwillincreasesignificantly,whichwillcausemetabolicdisorders,reducethedesiretopursueriskybehavior,andevenchangethepublic'sriskpreferences(Nowakowicz-Debeketal.,2004).Itcanbeexpectedthatthecostofairpollutionisnotonlyreflectedinphysical,mentalandbehaviouralaspects,butitwillalsoleadtounnecessaryexpensesinthepreventionandtreatmentofrelateddiseases.Forexample,Deryuginaetal.(2016)studiedthecausalrelationshipbetweenparticulatematter(PM2.5)exposure,mortalityandhealthcareuseintheUnitedStates;Barwicketal.(2017)foundthatbothshort-term(withinaweek)andlong-term(withinthreemonths)healthexpenditureshaveasignificantimpactcausedbyPM2.5.Xieetal.(2013)studiedoneheavyhazepollutioneventthatoccurredatacityinChinainJanuary2013andfoundthatshort-termhigh-concentrationPM2.5pollutionposedahighrisktothehealthofpublicandhascausedrelatedhealthandeconomicharmthatcannotbeignored.Moreindividualsareawareoftheimpactofpollutiononthemselves,suchasairpollution,especiallywhentheyneedtofacewithhighmedicalexpenses,manypeoplefindthatitisimpossibletocompletelyavoidthepossiblefuturerisksonlybypensionsandcompulsorymedicalinsurance.Asarisk-avoidingfinancialproduct,commercialhealthinsuranceisaneffectivesupplementtotheallocationofindividualassets;combinedwiththepopularizationoffinancialknowledgeforthewholesociety,thepublic'senthusiasmforcommercialhealthinsurancecontinuestorise:in1999,China'scommercialhealthinsurancepremiumtherevenuewasonly3.654billionyuan,whichreached11.2167billionyuanin2013,withanaverageannualgrowthrateof27.53%(Peng,2017).Existingresearchstudiesthedevelopmentofcommercialhealthinsurancefrommultipleperspectivesofsupplyanddemand.Fromtheperspectiveofsupply,tobesurprised,China'ssocialhealthinsurancesystemdidnotcrowdoutcommercialhealthinsurancebutindeedpromotedthedevelopmentofcommercialinsurance(Peng,2017;Liuetal.,2012).Fromtheperspectiveofdemand,researchbyLiuetal.(2012)pointedoutthatpersonalriskappearanceandpurchasingpowerareimportantinfluencingfactorsforresidents'medicalinsurancepurchase.Besides,individuals'consumptionofcommercialhealthinsuranceisalsoaffectedbymanyfactors,suchasculture(Wanyan,2013),andtaxpolicy(Thomasson,2003).Atthesametime,thestudyfrombehavioralfinancehasshownthatwhenthepublicmakesdecisionsaboutpurchasingcommercialhealthinsurancewhenconsideringbenefits,theweightisgiventotheideasof"avoidanceofrisk"istwicethantheideaof"increasingprofit"(Yietal.,2006).Intermsofintuitiveimpact,becausecommercialhealthinsuranceisarisk-aversefinancialproduct,andwhentheenvironmentwillincreasetheriskofresidentsbeingharmed,residentstendtoshowariskaversion,soitwillinevitablyleadtoanincreaseindemandforhealthinsurance.Basedontheabovebackgroundandacademicliterature,therehavebeenstudiesontheimpactofairpollutionfromvariousperspectives,andanalysisofthefactorsaffectingthedevelopmentofcommercialhealthinsurancefromthelevelofsupplyanddemand,butthereislessstudydirectlyfocusontheimpactofairpollutiononthedemandforcommercialhealthinsurance,alsothereislessliteraturestudyingontheimpactofairpollutiononconsumers'assetallocationpreferencesthroughrelatedresearch.TheresearchbyChangetal.(2018)usestransaction-leveldatatoempiricallyconcludethatairpollutionhasapositiveeffectonthedemandforcommercialhealthinsurance;atthesametime,ithasalsobeenfoundthatunderthechangeoftheouterenvironment,individualdecisionsarepronetobias,whichwillimpacttheallocationsofcommercialinsurance.However,thestudyhasthefollowingshortcomings:①Thisarticleusesshort-termdatafromasinglecompanyandsmall-scalecities,andtheremaybeacertainsamplebias(especiallyfromtheperspectiveofthedifferenceininsuranceproducts),whichreducestheconvincingofthestudy.②Individualanddemographycharacteristicsareimportantfactorsinfluencingindividualdecision-making.However,thearticledoesnotcontroltheindividualcharacteristicsofresidents,andtheremaybeproblemswithmissingkeyvariables,whichweakensthereliabilityoftheresearchresults.③Usingthemicro-enterprisedatamayleadtothepossibilitythattheconclusionmaybeaccidental,anditmaynotbeapplicabletothegeneralsituation,thusfurtherresearchisneeded.TheresearchofWuetal.(2019)isbasedonpaneldatafrom30provincesacrossChina,usingthespatialDurbinmodeltoempiricallystudythedirectimpactofairpollutiononthedevelopmentofcommercialhealthinsuranceandtheeffectofspacespillovers.Thestudyfoundthatairpollutionhasnosignificantdirectimpactonthedevelopmentofcommercialhealthinsurance,butapositivespatialspillovereffectonthedevelopmentofcommercialhealthinsurance,andthespacespillovereffectintheeasternandcentralregionsissignificantlylargerthanotherregions.ThisstudyeffectivelyfillsinthegapsintheliteratureandaddressestheproblemofexcessiverelianceonindividualdataintheresearchofChangetal.(2017).However,theprovincialdataisnotaccurateenoughcomparedwiththecity-levelorcounty-leveldata,whichmaysignificantlyaffecttheresearchresults.Anditmaybeunrealizablewhenthearticledividestheregressionofmodelbythemomentof“PM2.5”entryappearsinBaiduEncyclopediaforthefirsttimeInfact,thefirsttime‘PM2.5’entryappearedwasonSeptember6,2010,andthefirsttimePM2.5wasrenamedinChinesebygovernmentwasonApril19,2013.TherelatedentrieswerewidelyrevisedstartedinMarch2013..Infact,thefirsttime‘PM2.5’entryappearedwasonSeptember6,2010,andthefirsttimePM2.5wasrenamedinChinesebygovernmentwasonApril19,2013.TherelatedentrieswerewidelyrevisedstartedinMarch2013.Modelconstruction,variabledescriptionandsummarystatistics3.1ModelConstructionBasedontheliteratureandabovetheoreticalanalysis,thisthesisusescity-leveldataforresearch.Thecorrespondingfixedeffectmodelissetasfollows:lForallvariablesinmodel,HealthInsuranceisthenaturallogarithmofpremiumincomeforcommercialhealthinsuranceinyeartatareai,PM2.5istheannualaverageconcentrationofPM2.5inyeartinareai.Forthecontrolvariables,FIrepresentsthefiscalincomeofgovernmentsinyeartatareai.,whichisinnaturallogarithm,variableSAListheaveragetotalincomeofresidentsinyeartatareai,variableFDIrepresentstheamountofdirectforeigninvestmentatareaiinyeart,HEistheproportionofcollegesanduniversitiesstudentsperthousandpeopleatareaIinyeart,andHRismedicalcareresourcesinareaiint,thecalculationmethodistheproportionofhospitalsperthousandpeople,HHIistheinsuranceindustryHHIindexfortyearsatareai;Weatherisameteorologicalcontrolvariable,dividedinto3variables,whichareSunshine,Rain,andTemperature,representingthetotalnumberofsunshine,precipitationandaveragetemperatureinyeartinareaiaresubjectedtonaturallogarithmictreatmentofthreevariablesinregression.αiFortheparametersinthemodel,β0istheinterceptterm;β1~β8Variabledescription(1)Explainedvariable.Differentfromsomeliterature,thisresearchdirectlyusesthepremiumincometomeasuretheconsumptionofvarioustypesofcommercialinsurance.Thisisbecausepremiumincomecanrepresentthefinancialbehaviorofconsumersmoredirectly;itcanalsopreventcollinearitywithsomeofthecontrolvariablesorexplainedvariable.Fortheconvenienceofprocessingandinterpretation,thevariableusedinthemodelduringregressionisthenaturallogarithmofpremiumincome.(2)Independentvariables.InordertosolvetheproblemofmissinghistoricaldataofPM2.5concentration,thispaperusesthemethodofShao(2016)andadoptsthegridofannualaveragevaluesofPM2.5concentrationinChinapublishedbytheAtmosphericCompositionAnalysisGroupinDalhousieUniversity(VanDonkelaaretal.,2015&2019).TheoriginaldatawereanalyzedbyArcGISsoftwaretothespecificannualaveragePM2.5concentration(μg/m3)atthecitylevelinChinafrom2009to2017.ThedataisbasicallyconsistentwiththeanalysisofairpollutionstatusbytheChineseMinistryofEcologyandEnvironment,andalsobasicallyinlinewiththetrendoftheAQIindexissuedbytheChineseMinistryofEcologyandEnvironment.Atthesametime,inordertoperformtherobustnesstest,thisarticlecollectedtheAQIdatafor2009-2017andcounteditasayear-levelvariable;inspiredbyYang(2019),TheBaiduindexdataobtainedbysearchingforthekeyword"PM2.5"ineachcityat2011-2017hasalsobeencollectedandcountedthedataasayear-levelvariable.(3)Controlvariables.Accordingtotheexistingliterature,thefollowingfactorsinfluencethewillingnesstopurchasecommercialhealthinsurance:Fiscalrevenue(FI)isusedasasubstitutevariableforthedevelopmentoflocalsocialhealthinsurance.enhancement.IntheliteratureofPengetal.(2017),whenthedevelopmentofsocialhealinsurancewaspromoted,itwillalsoimprovethedevelopmentofcommercialhealthinsurance.Naturallogarithmwasperformedduringregression.Residentincome(SAL),asknownbyWu(2018),isagoodindicatorofeconomicperformanceinareaswhenresidents'incomeishigher,theyhavetheabilityandwillingnesstoallocatecommercialinsuranceforpersonalassetsasasupplement.Foreigndirectinvestment(FDI),accordingtoYanetal.(2017),hasasignificantspatialpositivecorrelationbetweenfaze(PM2.5)pollutionandFDI.Inordertocontrolendogenousnessinthemodel,FDIisaddedasacontrolvariable.Theunitofthevariableinthemodelismillionyuan.Educationlevel(HE)isusedtoreflectthelevelofculturaleducationinacertaincity.AccordingtoSuoetal.(2015)andWu(2018),itcanbeknownthatthehigherlevelofeducation,theeasierthatresidentswouldliketoconsumeforhealthrisksandhealthinsurancetopreventrisks.HealthResourcesIndex(HR)isusedtoreflectthetensionoflocalmedicalresources.Pengetal.(2017)pointedoutthatwhenmedicalresourcesarescarcer,theburdenofmedicalexpenseswillbeheavier,therebypromotinglocalresidentstopurchasehealthinsurance.HHIistheinsuranceindustryHHIindexforyeartinareai,whichisusedtocontroltheconcentrationoftheinsuranceindustryinacertainarea.Suoetal.(2015)proposedthat,ingeneral,thehighertheindustryconcentrationinacertainplace,thehigherthepriceofthecorrespondingproductmaybepurchased,whichmayhinderthedemandfortheproduct.HHIiscalculatedasthesumofthesquaredmarketsharenumbersofallcompaniesinaparticularmarket.Atthesametime,consideringthatthelocalweatherconditionscansignificantlyaffectindividualdecisions,andalsoaffectthelocalenvironmentalfactorsrelatedtothepurchaseofcommercialhealthinsurance,suchasPM2.5concentration(Zheng,2019),etc.thevariablesabouttheweatherareintroduced:Weatherisameteorologicalconditionvariable,whichisdividedinto3variables,calledSunshine,Rain,andTemperature,representingthetotalnumberofsunshine(unit:hour),precipitation(unit:millimetre),andaveragetemperature(unit:Celsius)inyeartatareai.Intheregression,naturallogarithmprocessingwasperformedonthetotalsunshineandprecipitation.3.3DatasourcesanddescriptivestatisticsThesampleperiodofthisstudyisfrom2009to2017,andthereare278citiesin29provincesofChinainoursample.DuetothelackofdatainTibetandXinjiang,citiesintheseprovincesarenotincludedinthesample,nordoesitincludeHongKong,MacauandTaiwan.Thedatasourceforthepremiumincomeofcommercialhealthinsurance,lifeinsurance,andaccidentinsurancecomesfromChinaInsuranceYearbook;excepttothemeteorologicalcontrolvariables,thedataandoriginaldataofothercontrolvariablesarederivedfromChinaCityStatisticalYearbook.Themeteorologicaldataareobtainedthroughthestatisticalyearbookdataofvariousregions.TheAQIdatacomesfromtheofficialwebsiteoftheMinistryofEcologyandEnvironmentofthePeople'sRepublicofChinaandisobtainedthroughwebcrawlingandmanualcollation.ThedataofBaiduIndexcomesfromtheBaiduIndexofficialwebsite,whichisobtainedthroughwebcrawlersandmanualcollation.AstheChinaInsuranceYearbookpublishedbytheChinaInsuranceRegulatoryCommissiononlyhascommercialhealthinsurancepremiumincomedatabefore2017,allthesampledatainthisarticlespansto2017.DetailsofdescriptivestatisticsareshowninTable1.Table1:DescriptiveStatisticsNameObs.MeanSDMinMedianMaxHealth28784.9031.441-1.4274.84012.777Acci.28763.3781.128-2.9963.30711.415Life28797.5731.1160.0107.58515.487PM2.5287037.39616.6774.67634.80986.480AQI156079.66324.82627.47775.241384.000Lnindex19113.6481.1770.0003.8298.099HR28280.5690.7420.0000.42911.477HE2828175.653235.3350.00088.3451311.241Salary280510.6060.3788.50910.63311.810Fiscal282813.6551.13210.17413.59218.010HHI28390.2440.1230.0670.2151.000Sunshine27287.5470.2786.3947.5548.402Rain27286.7410.6423.9596.7828.078temperature272814.8334.9680.20015.70026.9004.EmpiricalAnalysis:theimpactofAirPollutionontheDemandforCommercialHealthInsuranceConsumption4.1BaseregressionFirstly,abaseregressionisnowperformed,anditestimatesthedirectimpactoftheindependentvariable(theperiodN-1valueofairpollutionmeasuredbyPM2.5concentration)onthedemandforcommercialhealthinsurancethroughthefixeffectmodel.Inordertoperformtheheterogeneityanalysis,thispaperalsoestimatesthedirectimpactoftheperiodN-1valueofPM2.5oncommerciallifeinsuranceandaccidentinsurancedemandonperiodN.TheresultsareshowninTable2.Table2:BaseRegression(1)(2)(3)HealthAcci.LifePM2.5(lag)0.006**-0.0030.005***(2.02)(-1.33)(2.81)FI0.201***0.219***-0.094**(3.56)(3.40)(-2.53)SAL0.067-0.0160.121(0.69)(-0.15)(1.36)FDI-0.0440.316**0.223*(-0.37)(2.52)(1.91)HR-0.0060.0120.004(-0.39)(1.08)(0.59)HE-0.001*0.0000.000(-1.75)(0.16)(0.43)HHI0.3800.767-0.924***(1.06)(1.59)(-5.47)Sunshine-0.106-0.0820.005(-1.32)(-1.09)(0.09)Rain0.0330.0490.004(0.74)(1.29)(0.18)Temperature0.029-0.006-0.004(1.50)(-0.38)(-0.41)Constant0.5990.4877.344***(0.43)(0.37)(7.03)YearFixEffectYesYesYesCityFixEffectYesYesYesN239523942395RSquared0.8220.6320.612RSquared(adjusted)0.820.630.61AscanbeseenfromTable2,forthesampleatthenationallevel,thedirectimpactofairpollutiononcommercialhealthinsurancedemandissignificantatalevelof5%,thatis,ifthePM2.5ofthecity'speriodN-1increasesby1μg/yr.,thedemandofcommercialhealthinsuranceatperiodNwillincreaseby0.006%.Atthesametime,theimpactofairpollutiononthedemandforcommercialaccidentinsuranceisnotsignificant,whichisinlinewithourbasicexpectations,becausemostaccidentalinjurieshavenothingtodowithairpollution.Thesignificantleveloftheimpactofairpollutiononthedemandforcommerciallifeinsuranceisat1%.IfthePM2.5ofthecityatperiodN-1isincreasedby1μg/yr,thedemandofcommerciallifeinsurancewillhaveanincreaseof0.005%.Itmaybecausecommerciallifeinsuranceisalsoarisk-avoidingfinancialtool,andairpollutionwillhavealong-termimpactonthehealthofindividuals.Buyingcommerciallifeinsurancehasalongtermandhasmoreadditionalbenefits.Itisworthmentioningthatiftheindependentvariableinthismodelisreplacedbythecurrent(periodN)valueofPM2.5concentration,thedemandforanykindofcommercialinsurancewillnotbeaffectedbypollution.Thismaybeduetothecognitivelagofthepublic.Althoughairpollutionexistsobjectively,thepublic'ssubjectivecognitionoftheimpactofairpollutiononhealthhasatimelag.Onlywhenthepublicfullyrealizesthatairpollutionaffectshealthandimprovestheaccuracyofriskperceptionwillitbepossibletopurchasecommercialhealthinsuranceintheend.Forthecontrolvariables,variablessuchasmedicalresources,highereducationpopulation,andaverageindividualincomehavenotsignificantlyaffectedthedemandforvarioustypesofcommercialinsuranceintheregression.Therichnessofmedicalresourcesmaynotbeakeyfactorforindividualstoconsiderfuturehealthrisks,butabasiccondition.Aslongasmedicalresourcesareabundant,itisdifficulttoinfluenceindividualriskdecisions.Althoughthenumberofpeopleinhighereducationatschoolcandescribetheculturalqualityofthepopulationinthecity,theregressionresultsmayreflecttousthatthereisstillacertainlackoffinancialliteracyamongthepublic,therebylimitingtheimpactofeducationonthedemandofcommercialinsurance(Yinetal.,2015).Inthesameway,theincreaseinaveragepersonalincome,intheory,willpromoteitsuseoffinancialinstrumentsto"benefitfromavoidingharm”butitfinallyfailedtoobtaincorrespondingresultsinthereturn.Thismayalsobeduetothelackofuniversalaccesstofinancialeducationandhealthriskeducation.Thefiscalincomeofgovernmentindirectlyreflectsthepopularityofsocialhealthinsuranceintheregion.Intheregression,theinfluenceofthisvariableissignificantatthe1%levelforallthreetypesofcommercialinsurance.Thesocialhealthinsurancesystempromotesthedemandforcommercialhealthinsuranceandcommercialaccidentinjuryinsurance,whichsupportstheconclusionsofthehistoricalliterature(Liuetal.,2012,Haoetal.,2017,Xie,2010).Atthesametime,thesocialhealthinsurancesystemhasweakenedthedemandforcommerciallifeinsurance,whichisalsoreasonable,because,inChina,socialmedicalinsuranceandcommerciallifeinsurancemayhavesimilarfunctions.ForHHIindex,wecannotseeasignificanteffectoncommercialhealthinsurance,buta1%levelofsignificancewhenwefocusonthecommerciallifeinsurance,andhighHHIindexwilltakeanadverseimpacttowardsthedemandoflifeinsurance.Thatismaybecause,fortheChineseinsurancecompanies,theincomepremiumshareofhealthinsuranceissmallerthanthelifeinsurance.ItisworthnotingthatintheprocessingrobustnesstestusingIV,wecanseethatHHIshowsanegativesignificantimpacttowardsthedemandofcommercialhealthinsurance,whichisinourexpection.4.2Heterogeneityanalysis4.2.1RegressionbyRegionSampleItcanbeassumedthatthedemandforcommercialhealthinsurancealsovariesregionally,becausetheconsumptionofcommercialhealthinsuranceisaffectedbytheeconomic,geographicalconditions,andcultureofdifferentregions.Intheheterogeneityanalysis,theregressiontestofthesamplebyregionisfirstperformed.Thedivisionoftheregionisbasedontheupperboundaryofwarmtemperatezone,wheretheannualaveragetemperatureis13°C.Chinaisdividedintotworegions,thenorthernChinaandsouthernChina,bytheboundaryofQinMountain-HuaiRiver,whichisalsotheboundarybetweenthewarmtemperatezoneandthesubtropicalzone.TheregressionresultsareshowninTable3.Table3:Regression:ByTemperature(1)(2)<13℃≥13℃NorthernSouthernPM2.5(lag)0.006**0.012**(2.26)(2.57)FI0.210**0.117(2.29)(1.42)SAL0.049-0.029(0.45