居民家庭储蓄相关外文文献Chinese household saving and dependent children Theory and evidenceWord文件下载.docx

居民家庭储蓄相关外文文献Chinese household saving and dependent children Theory and evidenceWord文件下载.docx

《居民家庭储蓄相关外文文献Chinese household saving and dependent children Theory and evidenceWord文件下载.docx》由会员分享，可在线阅读，更多相关《居民家庭储蓄相关外文文献Chinese household saving and dependent children Theory and evidenceWord文件下载.docx（14页珍藏版）》请在冰点文库上搜索。

Abstract

Thispaperexaminestheimpactoffamilysizeonhouseholdsaving.Wefirststudyatheoretical 

life-cyclemodel 

thatincludesfinitelifetimesandsavingforretirementandinwhichparentscareabouttheconsumptionbytheirdependentchildren.Themodelimpliesanegativerelationshipbetweenthenumberofdependentchildreninthefamilyandthehousehold 

savingrate.Then,wetestthemodel'

simplicationsusingnewsurveydataonhouseholdfinancesinChina.Weusethedifferentialenforcementoftheone-childpolicyacrosscountiestoaddressthepossibleendogeneitybetweenhouseholdsavingandfertilitydecisionsinatwo-stageleastsquares 

Tobitregression.WefindthatChinesefamilieswithfewerdependentchildrenhavesignificantlyhighersavingrates.Thedatayieldsseveraladditionalinsightsonhouseholdsavingpatterns.Householdswithcollege-agechildrenhavelowersavingrates,andhouseholdsresidinginurbanareashavehighersavingratesandalowerratioofeducationexpenditurestoincome.However,havinganadditionalchildreducessavingratesmoreforhouseholdsinurbanareasthaninruralareas.Ourregressionsalsoindicatethatsavingratesvarywithageandtendtobehigherforhouseholdswithmoreworkers,highereducation,betterhealth,andmoreassets.

Keywords：

China，Householdsaving，Demographics，Overlappinggenerations

Introduction

ThispaperpresentsevidencesupportingthehypothesisthatthedecreaseindependentchildrenwithinChinesefamilieshasledtoanincreaseinhousehold 

savingrates.Weestimatetherelationshipbetweensavingandfamilysizebyapplyingstandardregressiontechniquestodatafromanewhousehold-levelsurvey（theChinaHouseholdFinanceSurvey）.Thestrongresponseof 

fertilityrates 

tofamily-planningpolicies（e.g.theone-childpolicy）inChinaallowsustoaddressthepotentialendogeneitybetweensavingandbirthdecisions.Specifically,weinstrumentforthenumberofdependentchildreninthehouseholdwiththecountylevelnumberofbirthsbecauseenforcementofthefamily-planningpolicieshasvariedacrossgeographicregions.Thus,weregresshouseholdsavingontheinstrumentednumberofchildren（andadditionalcontrolvariables）atthehouseholdlevelviaatwo-stage 

Tobitregression.Ourmainfindingisthatfamilieswithfewerdependentchildrensavesignificantlymore.

Uncoveringthedeterminantsofhouseholdsavingis,ofcourse,animportanttopicingeneral,andChinesesaving,inparticular,hasbeenreceivingconsiderableattention.China'

shouseholdsavingratehasexplodedinrecentyears,andthissavinghashelpedcreateinvestmentledgrowthinChina.ExcessChinesesavinghasflowedtowardssafeassetsindevelopedcountries.Hence,policymakersbothinChinaandabroadwouldliketounderstandthefactorsbehindthehighsavingrate.Lookingforward,accumulatedhouseholdassetsmayhelpChinacopewithitsrapidlyagingpopulation.

The 

life-cyclehypothesis 

isaleadingcandidateforexplainingChina'

shighhouseholdsavingrate.1 

ModiglianiandCao（2004） 

wasthefirstpaperempiricallyshowingthecorrelationbetweenChina'

sagestructureandsavingrateintheaggregatedtimeseriesdata. 

Curtis,Lugauer,andMark（2015） 

developastructural 

ofhouseholdsavingdecisionstoillustratethetheoreticalconnectionsbetweendemographicsandsavings,and,throughaseriesofmodelsimulations,theyshowthatthedemographiceffectonChina'

saggregatesavingrateisquantitativelylarge.Webuildoff 

Curtisetal.（2015） 

byexaminingtheimplicationsofalife-cyclebasedmodelforsavingbehaviorinthecrosssection（i.e.microdata）.Wefocusonthemainimplicationfromthemodel:

householdsavingdecreaseswithfamilysize.

In 

Section2,wepresentasimplifiedversionofthemodelfrom 

Curtisetal.（2015）.Thekeymodelingredientsincludefinitelifetimes,savingforretirement,andthatparentscareabouttheconsumptionoftheirdependentchildren.Theexplicitvaluationofchildren'

sconsumptionenterstheparentalutilityfunctionwiththefunctionalformfrom 

BarroandBecker（1989）.Thestructuralmodelhasstarkimplicationsfortherelationshipbetweenthenumberofdependentchildreninthefamilyandhouseholdsavingbehavior.Therefore,ourempiricalregressionsactasatestofthelife-cyclehypothesisofhouseholdsavingbehavior.

Thelife-cyclemodelmotivatesourbasicresearchquestionastowhetherthenumberofchildreninahouseholdaffectssavingdecisions,butthedataallowustoexamineadditionalcontrolvariables,includingthecharacteristicsofthehouseholdhead（age,education, 

riskaversion,andhealth）,thenumberofelderlypeopleinthehome,andhouseholdassets.Ourempiricalstrategy,then,istoleveragethepolicydriven（andthereforeplausiblyexogenous）differencesinthenumberofchildrentoestimatetheeffectoffamilysizeonhouseholdsaving,whilecontrollingfortheseotherfactors.Thekeyidentificationassumptionisthattheenforcementoffamily-planningpolicies（asmeasuredbycounty-levelbirthrates）affectshouseholdsavingdecisionsonlythroughthefertilitychannel.

Theempiricalresultssupporttheimplicationsforsavingfromthestructuralmodel.Householdsavingisdecreasinginfamilysize,asmeasuredbythenumberofdependentchildren.Theestimatedcoefficientsarelargeandstatisticallysignificantatthe1%level.Thus,ourmainfindingisinlinewiththemodel'

spredictionthatfewerchildrenincreasessaving.Note,though,afewrelatedtheories（inparticular, 

Choukhmane,Coeurdacier,andJin（2017） 

and 

İmrohoroğluandZhao（2017a））existthatalsopredictthenegativeimpactofchildrenonhouseholdsavingrates.Wediscussthesealternativeexplanationsfurtherbelow.

Inourregressions,thecoefficientestimatesforthecontrolvariableshavetheexpectedsigns.Wefindthathouseholdswithhighereducationlevelssavemoreonaverage.Poorhealthisassociatedwithlowersavingrates.Familieswithmoreworkershavehighersavings,andhouseholdsavingvariesbyage.

Thedataallowustoexaminesavingbehavioralongseveraladditionaldimensions.First,weexaminetherelationshipwithnonlinearspecificationsintermsofthefirstborn,second,andthirdormore.Wefindthelargesteffectfromthefirst-bornchild.Wealsoexaminetheeducationexpendituredata.Householdswithcollegeagechildrenspendabouttwiceasmuchoneducationcomparedtohouseholdswithchildrenbelowcollegeage.Householdswithcollegeagechildrenhavesignificantlylowersavingrates.Inaddition,forthegroupofhouseholdswhospendmoreontheirchildren,havingonelesschildincreasestheirsavingbymorethanforhouseholdsspendinglessontheirchildren.Theseresultsarenoteworthy,sincethestructuralmodelalsoimpliesdifferentsavingbehaviorduetotheparent'

spreferencesoverprovidingfortheirchildren.

Thehouseholdsinurbanareasspendanaverageof1/3moreoneducation;

however,urbanhouseholdsstillhavehighersavingratesthanruralhouseholdsduetoalowerratioofeducationexpendituretoincome.Interestingly,havinganadditionalchildreducesthesavingratemoreforhouseholdsinurbanareasthaninruralareas.Wefindasimilarpatternwhencomparinghouseholdswithatleastoneparentemployedinastateownedenterpriseorgovernmentagencyversushouseholdswithnofamilymembersemployedinthepublicsector.

Alife-cyclemodelofhouseholdsaving

Thissectionpresentsastructural 

ofhouseholdsavingdecisions.Themodelrepresentsasimplificationoftheframeworkemployedin 

Curtis,Lugauer,andMark（2017） 

tostudytheeffectofdemographicchangesonaggregatehousehold 

savingrates 

overtimeacrossseveralcountries.ThemodelmotivatesourreducedformempiricalregressionsbasedonChinesehousehold-leveldata.Themaintake-awayfromthemodelisthathouseholdsavingdecreaseswithfamilysize.

Householdsmakeconsumptionandsavingdecisionstakinginterestratesandwagesasgiven. 

Laborsupplyisinelastic,andfamilysize（demographics）isexogenous.WethinkthisassumptionisreasonablegiventheevolutionoftheChineseeconomyinregardstofamily-planningpolicies.Plus,thisassumptionmapsintoourempiricalidentificationstrategy.

Generationsoverlap,buteachagentlivesforonly3periods.Intheinitialperiodoflife,however,agentsaredependentchildrenandmakenodecisions.Themaindeparturefromastandard2-periodutilitymaximizationproblemistheinclusionofchildren'

sconsumptionintheparentalutilityfunction（viaBarro-Beckerpreferences）inthemiddleperiodoflife.Agentsretireinthefinalperiodoflifeandnolongersupportchildren.

Dataanddescriptivestatistics

WebasetheempiricalanalysisondatafromtheChinaHouseholdFinanceSurvey（CHFS）conductedbytheSouthwesternUniversityofFinanceandEconomicsinChina.TheCHFScollectsdetailedinformationbienniallyonhouseholds'

demographiccharacteristics,assetsandliabilities,insuranceandsocialwelfare,andincomeandexpenditures.Thesurveyisnew,andweprimarilyuseinformationfrom2013.Thesurveywasalsoconductedin2011,butthe2011sampleisconsiderablysmaller.7 

Weusethe2011sampletoconstructapaneldataset（bymatchinghouseholdsacrossthetwosamples）inarobustnesscheckreportedbelow.

TheCHFSdatasetisparticularlysuitedtoourpurposesbecauseitcontainsinformationonalargesampleoffamilieswithyoungchildrenandcoversmostofChina,bothurbanandruralareas.Inaddi