The-Chinese-Saving-Puzzle-and-the-Life-Cycle-Hypot

JournalofEconomicLiteratureVol.XLII(March2004)pp.145–170TheChineseSavingPuzzleandtheLife-CycleHypothesisFRANCOMODIGLIANIandSHILARRYCAO11451.IntroductionWithtoday’sworldeconomyplaguedbyrecessionandsluggishperformance,Chinastandsoutwithitssustainedhighgrowth,perhapsduetomarket-orientedeconomicreformsthatbeganin1978.Thegrowthhasbeenaccompaniedbyanexplo-sioninChina’shouseholdsavingratio,whichhasreachedanimpressivelevelinrecentyears,justwhentherehasbeenaworldwidereductionintheprivatesavingratio.In1994,forexample,ourestimateofthehouseholdsaving-to-incomeratioinChinawascloseto34percent,rivalingtheJapaneseexperienceinthe1960s,eventhoughthelevelofChina’sper-capitaincomeremainedwellbelowthoseoftheindustrializednations(seefigure1).Moresurprisingly,lookingbackatthepostwarhistoryofChinesehouseholdsavings,onefindsthat,fromthe1950sthroughthemid-70s,the“thrifty”Chinesewerenotsothriftywithanaveragehouseholdsavingratiobelow5percent.Inthispaper,weattempttoexplaintheapparentparadoxofthesud-denspurtinthesavingratio’smagnitudebyusingtheframeworkofthelife-cyclehypothesis(LCH)developedbyFrancoModiglianiandR.Brumberg(1954).TheLCHwasinitiallythoughttoberele-vantfordevelopedmarketeconomiesonly.China’sper-capitaincomein2002isstillrankedbelow100thintheworld,accordingtotheWorldBank.2Inthemiddleofoursample,therewasasharpregimeshiftfromahighlyplannedeconomytoamarket-orientedeconomy.Therefore,testingtheexplanatorypoweroftheLCHforChinaisnotonlyrelevantinendeavoringtoofferaplausibleexplanationofthedrasticchangesinthepersonalsavingratioofthatcountry,butalsohastheoreticalimplicationsastotheapplicabilityoftheLCHmodeltoamoregeneralenvironment,includingdevelopingcountries.Sinceitiswidelybelievedthatthestan-dardKeynesiantheorymayplayanimpor-tantpartinexplainingbehaviorpertainingtosavinginatleastthelow-incomenations,we1Modigliani:InstituteProfessorEmeritus,MassachusettsInstituteofTechnology.ShiLarryCao:MunderCapitalManagement;formerlywithPeople’sBankofChinaandMIT.Editor’snote:InsubmittingthispapertotheJournalofEconomicLiterature,FrancoModiglianiwrote,“Thisisapaperwhichhasspecialmeaningforme,asIseeitasafit-tingconclusiontomylife’sworkonsaving.”ProfessorModiglianipassedawayonSeptember25,2003.NotebyLarryCao:IamdeeplysaddenedthatIhavelostamentorbuttakecomfortinknowingthathesawthecompletionofthisproject.Francoremainedactivetilltheend.OnSeptember20,whenwelastspoke,hementionedanadditionalchangehewantedtomake.WethankIreneY.ChanandFrancescoFrancoforhelpwithdataandedit-ing,andOlivierBlanchard,theeditor,andananonymousrefereeforhelpfulcomments.IalsothankChenYuanandLiRuoguforsupportingthisproject.2WorldDevelopmentIndicatorsdatabase,WorldBank,July2003.mr04_Article43/6/0410:24AMPage145146JournalofEconomicLiterature,Vol.XLII(March2004)-0.050.000.050.100.150.200.250.300.350.401953195819631968197319781983198819931998Yearsavingratio0100200300400500600700householdsavingratiopercapitaincomein1950YuanFigure1:China’sHouseholdSavingRatioandPer-CapitaIncome:1953–2000includeacomparisonoftheresultsoftheLCHandtheKeynesianmodel.2.BasicDataTable1summarizestheincomeandsavinginformationonChinafortheperiod1953to2000.Itisimportanttokeepinmindthatpersonalsaving(seedataappendix)isnotmeasuredas(disposable)incomeminuscon-sumptionasisconventionalinthenationalincomesaccount(NIA).Thisisduetotheunavailabilityoftherequireddata.Inthiscase,weuseanalternativeapproachthat,conceptually,shouldyieldanestimateofNIAsaving,exceptformeasurementerrors.Itconsistsofmeasuringthe(annual)increaseinpersonalwealththatresultsfrompersonalsaving.Ourestimatedcalculationstoshowtheincreaseinpersonalwealth,W(t)–W(t–1),arethesumoftwocomponents:1.Thefirstistheincreaseintheholdingsofalistofintangibleassets,A(t)–A(t–1).Wehaveendeavoredtomakethelistofrelevantassetsascomprehensiveaspossible,ataskfacilitatedbytheverylimitedchoicesavail-ableduringtheperiodwecover.Theassetsweincludeconsistbasicallyofnominalassets(money,deposits,andgovernmentbonds).2.Thesecondcomponentisanestimateoftheincreaseinthestockofsomemajortangibles(e.g.privateresidences).HerewemeasuretheutilizationofsavingbytheestimatedflowofinvestmentI∗.Thus,S=[A(t)–A(t–1)]+I∗=I+e,whereIistheaggregateinvestment.Thelastequalityismeanttoemphasizethatourmeasureofsavingislogicallyequiv-alenttothetraditionalone,butthatthetwoestimatescandifferbecauseoferrorsofmeasurementineitherorbothterms.Ourmeasureofincomeisobtainedbycombiningsumsforconsumption,forwhichofficialestimatesareavailable,withsumsforsavingasestimatedabove.Thisapproachmeasures“disposable”income,whichisactuallyveryclosetopersonalincome,sincemr04_Article43/6/0410:24AMPage146TABLE1CHINESEHOUSEHOLDINCOMEANDSAVING,1953–2000CPI,HouseholdHouseholdHouseholdHouseholdCPI,PreceedingRealPercapitaYearConsumptionSavingIncomeSavingRatio1950=100Year=100IncomeRealIncomeE/M1953529.2041.50570.700.0726121.4105.1470.179.951.00151954550.0031.80581.800.0546123.1101.4472.678.420.98761955602.0030.10632.100.0476123.5100.3511.883.270.97961956646.8051.00697.800.0731123.499.9565.590.010.97741957686.6031.80718.400.0442126.6102.6567.487.760.9