DP
RIETI Discussion Paper Series 13-E-035
The Effects of Birth Weight:
Does fetal origin really matter for long-run outcomes?
NAKAMURO Makiko
Keio University
UZUKI Yuka
National Institute for Educational Policy Research
INUI Tomohiko
RIETI
The Research Institute of Economy, Trade and Industry
http://www.rieti.go.jp/en/
RIETI Discussion Paper Series 13-E-035
April 2013
The Effects of Birth Weight:
Does fetal origin really matter for long-run outcomes? 1
NAKAMURO Makiko (Keio University) 2
UZUKI Yuka (National Institute for Educational Policy Research)
&
INUI Tomohiko (Nihon University / RIETI)
Abstract
This paper investigates whether birth weight itself causes individuals’ future life
chances. By using a sample of twins in Japan and controlling for the potential
effects of genes and family backgrounds, we examine the effect of birth weight on
later educational and economic outcomes. The most important finding is that birth
weight has a causal effect on academic achievement at around the age of 15, but
not on the highest years of schooling and earnings.
Keywords: Birth weight, Nutritional intake, Identical twins, Endogeneity
JEL classification codes: I10, I20
RIETI Discussion Papers Series aims at widely disseminating research results in
the form of professional papers, thereby stimulating lively discussion. The views
expressed in the papers are solely those of the author(s), and do not represent
those of the Research Institute of Economy, Trade and Industry.
1 This study was conducted as part of a project titled “Research on Measuring Productivity in
the Service Industries and Identifying the Driving Factors for Productivity Growth” of the
Research Institute of Economy, Trade, and Industry (RIETI). We gratefully acknowledge that
this research was financially supported by Grant-in-Aid for Scientific Research (A) titled “The
Assessments of the Quality and the Productivity of Non-marketable Services” (Research
Representative: Takeshi Hiromatsu, No. 3243044).
2 Corresponding author: Makiko Nakamuro, Faculty of Policy Management, Keio University,
e-mail: makikon@sfc.keio.ac.jp. The authors would like to thank Atsushi Nakajima, Masahisa
Fujita, Masayuki Morikawa, Keiichiro Oda, Yoshimichi Sato, Shinji Yamagata, Daiji Kawaguchi,
Andrew Griffen and other participants of Tokyo labor economics workshop for their insightful
comments and suggestions on the draft of this paper. All the remaining errors are ours.
1
Introduction
Itwouldbegenerallybettertohaveasmallbabyatbirthandthenraise
him/hertogrowbiglaterinlife–thishaslongbeenbelievedtobegoodpracticein
child‐bearinginJapan.BeforeC‐sectiondeliveryorotherobstetricprocedures
becamepopularthroughoutsociety,peopleperhapsaimedtoreducetheriskof
endangeringmothers’livesbygivingbirthtoasmallbaby.Thiswidespreadbelief
seemstobestilldominanttoday.However,itisunexpectedlylittleknownthat
thereisahiddenriskofhavingasmallbaby:recentresearchhasfoundthatlow
birthweightissignificantlyassociatedwithbothshort‐andlong‐runadult
outcomes,suchasinfantmortality,studentachievements,andadulthoodhealth
(Conley&Bennett(2000);Linnetetal(2006);Currie&Hyson(1999),etc.). Whyisthishappening?Lowbirthweightiscausedbypretermdeliveryor
lowfetalgrowththatmayreflectvariationinnutritionalintakeinthewomb. Low‐birthweightisthusrecognizedastheleadingindicatorofpoorhealthamong
infants,whichmaydelaybrainandsomaticdevelopmentandthenaffectawide
rangeofsubsequentoutcomeslaterinlife.Thismechanismhasbeenalsorapidly
revealedasanobjectofepigenetics(e.g.,Petronis,2010etc.).Liketheresults
drawnfromdatainU.S.,DenmarkandEngland,Kohara&Ohtake(2009)used
officialstatisticsfromtheVitalStatisticsandtheNationalAssessmentofAcademic
AbilityinJapanandfoundanegativecorrelationbetweenbirthweightand
academicachievementsmeasuredbystandardizedtestscoresinG6andG9atthe
prefecturelevel.Ifthisisthecase,cantherebeanydoubtthattheGovernmentof
Japanmustshapeapolicyagendatoincreasethebirthweightofnewbornbabies,
forexample,throughimprovementsinthehealthofpregnantmothers?
Unfortunately,however,thereisnosimpleanswertothisquestion. Whilemuchisknownaboutthecross‐sectionalcorrelationbetweenbirth
weightandadulthoodoutcomes,littleisknownregardingtheextenttowhatwould
havehappentoanindividualoutcomeifapersonwhowasactuallybornwitha
heavierbirthweighthadbeenbornwithalighterbirthweight.Inotherwords,itis
highlypossiblethatobserveddifferencesinbirthweightsamongnew‐borninfants
maysimplyreflectunobservedparentalcharacteristicswhicharealsocorrelated
withadulthoodoutcomesofanindividual:aselectionbiasariseswhenpartof
individualoutcomescanbeexplainedbyunobservedparentalcharacteristics.
Observedcorrelationsusingcross‐sectionaldatainpreviousliteraturethusdidnot
2
provideafulldescriptionoftheeffectofbirthweightandresultinbiasedand
inconsistentestimates. Inthisresearch,wewouldliketoanswerthequestionsofwhetherbirth
weightitselfcausesindividuals’adulthoodoutcomeslateroninlife.Causalityis
thusobviouslythekey.Oneoftheinnovativemethodsthatsocialscientistshave
employedinrecentyearsaddressthecausalrelationshipbetweenbirthweightand
adulthoodoutcomesistouseasampleoftwins(orsometimessiblings).Infact,
manyeconomists,suchasBerhman&Rosenzweig(2004),Royer(2009),Almond
etal(2005)fortheU.S.,Milleretal(2005)forAustralia;Lawloretal(2006)for
Scotland;Oreopoulosetal(2006)forCanada,andBlacketal(2007)forNorway
(seeCurrie(2009)foramorecomprehensivesurvey)useadatasetcontaining
informationontwin‐pairsandattempttocopewiththeproblemofunobserved
differencesinabilityandfamilyenvironments.Theseconsiderableeffortshave
beendedicatedtouncoveringtheeffectofbirthweightonadulthoodoutcomes:
previousresearchreachedaconsensusthatbirthweightdoesmatterbothinthe
short‐andlong‐run. Wealsofollowthisapproachtodealwiththeaforementionedbias,
comparingthedifferencesbetweentwin‐pairstoisolatethepureeffectofbirth
weightontheadulthoodoutcomes,holdinginnateabilitiesandfamily
environmentsconstant.Anotheradvantageofusingasampleoftwinsisthat,
becausetwinpairshavethesamegestationlength,thedifferencesinbirthweight
betweentwinsareattributedsolelytodifferencesinfetalgrowthrates.Themain
researchquestionofinterestinthispaperisthus:doesnutritionintakeinutero
reallymatterforone’slifechance?Ifso,whichstageofone’slifeisthemost
affected?
Tothebestofourknowledge,thecaseofJapanisrelativelyunexploreddue
tothedatalimitation.Thisisunfortunategiventherecentvariablefindingsin
Japanthatlowbirthweightisassociatedwithparentalsocioeconomicfactors,such
asthemother’ssmokinghabitsandemploymentstatus(Tsukamotoetal,2007;
Kawaguchi&Noguchi,2012betc.).Theunderstandingofwhetheranindividual
inheritshis/herparentalsocioeconomicstatusatfetaloriginwouldcontributeto
furtherdiscussionoftheintergenerationaltransmissionmechanismofsocial
stratification,towhichpolicycirclesmaypayconsiderableattention.Inthisstudy,
wetakeadvantageoftheuniquetwins‐datasetsthattheauthorshavecollectedin
Japanthroughaweb‐basedsurvey. Toanswerourresearchquestion,wefollowtheprotocolofprevious
3
literatureandoutlineatwin‐fixedeffectstrategyusingasampleofmonozygotic
twins(hereafter,MZtwins)whoaregeneticallyidentical.However,interestingly,
thereisavariationinbirthweightbetweentwin‐pairsingeneral.Aspointedoutby
AshenfelterandRouse(1998),first‐borntwinsareusuallyheavierthantheir
second‐bornsiblingsatbirth.Thissettingallowsustocreateacounterfactual
situationofwhatwouldhavehappenedtoadulthoodoutcomesofapairoftwin
whowerebornwithalowerbirthweightifs/hehadbeenbornwithaheavier
birthweightinstead.Wethensetupfivemainoutcomestobeexamined:(i)
participationinprivate(ornational)middleschools;(ii)studentperformanceat
theageofaround15;(iii)rankingatthecollegeattended;(iv)yearsofschooling;
and(v)earnings.Thesignificantfindinginthispaperisthatbirthweightonly
causesacademicachievementundertheageof15.Unlikesomeoftheevidence
fromwesterncountries,thiseffectsubsequentlydisappears.Ourempiricalresults
showthatfetalgrowthmayaffectstudentperformanceinyoungchildren,butit
doesnotdirectlyaffecthis/heradulthoodoutcomesinlaterlife,suchas
educationalattainmentsandearnings.
Therestofthispaperisorganizedasfollows:thenextsectionreviews
relevantliteraturetosortoutinformationonwhatwestilldonotknowand
explainshowwetackledthemethodologicalproblemsinpreviousresearch.The
followingsectionsintroducetheempiricalspecificationstobeestimated,identify
thepotentialbiasemergingintheeconometricanalysis,anddeterminethe
analyticaltechniquestobeusedtoidentifythecausalimpactofbirthweighton
adulthoodoutcomelaterinlife.Theninthefinalsection,wedescribetheunique
twinsdatasetusedforempiricalanalysisandpresenttheempiricalresults.
RelevantLiterature
Evidencetoshowwhetherandtowhatextentincreasingthebirthweight
ofnewborninfantscanimprovetheirfuturelifechanceswouldbeusefulfor
framinganappropriatepolicydirectionregardingthenutritionalintakeof
expectantmothers.Agrowingbodyofresearchhasattemptedtoidentifythe
causaleffectsofbirthweightonnotonlyshort‐termbutalsolong‐termoutcomes
bytheuseoftwindata.Suchdataenablesresearcherstoruleoutthepotential
influencesofgeneticmakeupandfamilybackgroundsthataffectbothbirthweight
andlateroutcomesandtoobtainbetterestimatesofthecausaleffectsofbirth
weightthanthosederivedfromconventionalcross‐sectionalanalysis.Inthis
4
section,wereviewtherelevantliteratureinvestigatingthecausaleffectofbirth
weightoneducationalandeconomicoutcomes,inparticularbyusingtwindata.
Regardingeducationaloutcomes,thereisevidencebasedontwinstudies
thatbirthweighthasalong‐termimpact.BehrmanandRosenzweig(2004),Black
etal.(2005)andOreopoulosetal.(2008),usingtwindatafromMinnesota,Norway
andManitoba,respectively,foundthatbirthweighthasapositiveeffectonhigh
schoolcompletion.LinandLiu(2009)analyzedTaiwanesetwindataandfound
thatbirthweightincreasedgradesatage15.ItisnoteworthythatBehrmanand
Rosenzweig(2004)andLinandLiu(2009)showedthattheOLScoefficientsfor
birthweightwithoutcontrollingforgenesandfamilybackgroundsare
underestimatedby50%,whileBlacketal.(2005)foundthatOLSestimatesand
twin‐fixedeffectsaresimilarinsize.BehrmanandRosenzweig(2004)andLinand
Liu(2009)arguethattheirfindingssuggestthatparentsmayinvestmoreinlighter
twinstomakeupfortheirdevelopmentaldisadvantage. However,somestudiessuggestthattheeffectofbirthweightonyearsof
schoolingisrathersmall(Royer,2009)orthatthereisnosignificantrelationship
betweenbirthweightandeducationalattainmentorcognitiveabilitymeasuredby
languagetestscores(Milleretal.,2005;Oreopoulosetal.,2008).Itremains
unclearwhetherthismixedevidenceisduetodifferencesinmeasuresregarding
educationaloutcomesordatasources. Thereisrelativelylessevidenceonthedirecteffectofbirthweighton
economicoutcomes.Milleretal.(2005)areoneoftheexceptionalgroupsof
researcherswhofoundapositive,directeffectofbirthweightonearnings,but
arguethatbirthweightplaysonlyaminorroleindeterminingearnings,witheach
additionalounceofbirthweightincreasingearningsby0.4%.Otherstudiesthat
examinetheeffectofbirthweightoneconomicoutcomesincludeRoyer(2009)and
Oreopoulosetal.(2008).However,Royer(2009)didnotfindevidencetoshowthat
birthweightisassociatedwithneighborhoodincomelevelsinadulthood. AlthoughOreopoulosetal.(2008)foundthatbirthweightaffectssocialassistance
takeupandlengthinadulthood,butgiventhattheyalsofoundtheeffectofbirth
weightonhighschoolcompletion,itisunclearwhetherthebirthweighteffecton
economicoutcomeswouldremainaftercontrollingforthemediatingeffectof
educationaloutcomes.
InJapan,noresearchhasusedtwindatatoinvestigatethecausaleffectof
birthweightoneducationalandeconomicoutcomesinadulthood.Themost
importantreferenceworkisanon‐twinstudyconductedbyKawaguchiand
5
Noguchi(2012b),analyzingearlychildhooddatafromtheLongitudinalSurveyof
Babiesinthe21stCentury.Theyfoundthatlowbirthweight,definedasaweight
below2,500grams,isassociatedwithadelayindevelopmentatagetwoandahalf,
butnotwithbehaviorsatagesixandahalf.Apossiblereasonfortheassociation
disappearingatanolderagemaybethefactthatJapaneseparentsinvestmorein
theirchildreniftheirdevelopmentisobservedtobeslowatanearlierstage.
Therefore,thisevidencedoesnotconfirmwhetherthebirthweighteffectremains
onlyforashortperiodoftimeinJapan.Furthermore,littleisknownaboutthe
effectofbirthweightonmuchlatereducationalandeconomicoutcomesinJapan. WeaimtofurthertheliteraturebyusingJapanesetwindatatobringout
newfindingsinthefollowingthreerespects.Firstlyandmostimportantly,wewill
estimatetheeffectsofbirthweightbyaddressing,forthefirsttimeinJapan,
potentialendogeneitybiasesduetotheeffectsonbirthweight,post‐natal
developmentandlateroutcomesofgenesandparentalbehaviorsusually
associatedwithafamily’ssocio‐economicstatus.Secondly,wewillinvestigate
longer‐termeducationalandeconomicoutcomesashasbeendoneinrelevant
previousresearchinJapan,whichmayleadtoinsightintohowlong‐lastingthe
birthweighteffectswouldbe.Lastly,wewillexamineeducationaloutcomes
measuredinseveralwaysforthesamesample,whichmaycontributetoclarifying
whicheducationaloutcomesshouldbehighlighted,giventhecurrentmixed
evidenceacrosstheworldontheeffectsofbirthweightoneducationaloutcomes. EmpiricalSettings
Inordertoaddressourresearchquestionsofhowbirthweightaffects
adulthoodoutcomes,webegintheanalysisusingaconventionalOLStoreportthe
cross‐sectionalcorrelationswiththeentiretwinsample,inwhichmanyprior
studieshavefoundthatthebirthweightisstronglyassociatedwithawiderangeof
adulthoodoutcomes.Followingthepreviousliterature,weoutlinethesimple
educationproductionfunctionillustratingtheinput‐outputrelationshipathomeor
schoolthatparticularlyhighlightstheroleofchildhealth.Themodelcanbe
formallyexpressedinthefollowingmathematicalequationwhereyisoutcomes
andisafunctionofthebirthweight(bw)andunobservables(A),suchasgenetic
makeupormaternal/pre‐natalcareincombinationwithothercharacteristics(X)
andrandomdisturbancewithmeanzeroandconstantvariance(e)asspecifiedin
theequation(1)and(2)below.Notethatthefirst‐borntwinisdenotedas1bya
6
subscriptandthesecond‐bornas2.
′
′
(1)
(2)
Thecoefficientofβreferstotheeffectofthebirthweightonoutcomevariables
holdingotherobservedcharacteristicsconstant.Aswediscussedearlier,the
cross‐sectionalestimateofβmaybebiasedandinconsistentbecause
unobservables,Aij,affectboththebirthweightandoutcomes.However,identical
twinsenableustosetupA1j=A2j,giventhattheysharegeneticmakeupandfamily
environments.Wewillthustakeatwinfixedeffectapproachtakingthedifference
betweenequation(1)and(2)toobtainawithin‐twinfixedeffectsestimateofβ,
yielding: ) (3)
Inequation(3),unobservable,Aj,iseliminated,relievingusoftheconcernthatthe
outcomesarepartlyexplainedbyindividualunobservedcharacteristics.Giventhe
assumptionthattheerrortermisanidiosyncratic,whichisindependentofall
othertermsintheequation,βisconsideredastheconsistentestimate.
Data
Thedatausedforourempiricalanalysiswascollectedthrougha
web‐basedsurveyinJapanbetweenthemonthsofFebruaryandMarch2012(see
Nakamuro&Inui(2012)formoredetailedinformationonthedatacollection
strategy).Weconductedthesurveythroughaweb‐basedsurveycompany,Rakuten
Research,withover2.2millionmonitors.Inordertoanalyzetheeffectofbirth
weightonadulthoodoutcomes,oursampletargetedtwinswhoarenon‐students
betweentheagesof20and60.Throughthisweb‐basedsurvey,onememberofa
twinpairisresponsibleforreportingregardinghim/herselfandhis/hertwin
siblingatonetime,andtheresultsaredesigneddifferentlyfromthoseoftheother
twinsurveyfilledoutbybothmembersofthetwinpair3.
3
Onemayquestionthat,inoursurvey,theremayexistsubstantialmeasurementerrorsin
self‐reportedbirthweightandotheroutcomesbyoneofthetwinpairs,insteadofboth.Itis
importanttonotethatwehave23twinpairs,eachmemberofwhichwasincludedinthis
survey.Whenwechecktheirresponses,wefindoutthattheirresponsesreportedbyeach
7
Oncethemonitor(s)filledoutthequestionnaires,theywouldbegivena
certainamountofcash‐equivalent“points”thatcouldbespentonRakutenonline
shopping.Inordertoexclude“fake”twins,whopretendtobetwinstocollectthe
cash‐equivalentpoints,wecarefullydevelopedthefollowingdatacollection
strategy:wedidnotinformrespondentsthatthepurposeofoursurveywasto
collectdatafromtwins.Furthermore,westartedwithfivequestionsonfamilyand
siblingsthatwerenotrelatedtotwinstatusandthen,inthesixthquestion,forthe
firsttime,askedwhetherornotarespondentwasatwin.Iftherespondent
answered“No”inthisquestion,s/hewouldbeautomaticallyexcludedfromthe
survey.Wediscovered23twinpairs,eachmemberofwhichwasincludedinthis
survey,thenthoroughlycheckedtheresponsesofbothtwins,andeliminatedoneof
thetwinsrandomlyfromoursample.Ourweb‐basedsurveyovercamethe
disadvantagesofthedatacollectioninpreviousliterature,suchassmallsample
sizeordataattrition.Consequently,wecollected2,360completepairsoftwins
(4,720individuals)while1,371twinpairs(2,742individuals)aremonozygotic
(seeTable1).Tothebestofourknowledge,thisisoneofthelargestdatabasesof
twinscompiledinJapannationwide,anditconveysawiderangeofsocioeconomic
information. Variables
Theindependentvariableofinterestis,ofcourse,birthweightforwhich
wesetupseveralvariantsinthefollowingway:theprimarymeasureisthebirth
weightwhichisself‐reportedbyoneofthetwin‐pair.Theresponsecategoryinthe
originalquestionnairerangedfrom1(=lessthan1,500grams)to7(=morethan
4,500grams)and8(=don’tknow).Wesettheminimumto1,500andthe
maximumto4,500grams.Thenwetookthemid‐valueforthecategoriesbetween
2(=1,750grams)and7(=4,250grams).Basedonthisvariable,wecreatetwo
variantsofthekeyindependentvariables:avariantisthenaturallogarithmofthe
otherarequiteaccurate:thecorrelationsbetweenself‐reportedandcross‐reportedbirth
weightis91.2%.Notonlythebirthweightbutalsootheroutcomesshowover90%of
correlations.Furthermore,wecheckwhetherthereexistsignificantdifferencesbetween
responsesonhim/herselfandonhis/hertwinsiblings;forexample,onemaybedoubtfulthat
respondentsarepronetopretendthattheirearningsoreducationarehigherthanthoseof
theirtwinsiblings.However,accordingtotheresultdrawnfromtwosamplet‐testsfor
differenceofthemeans,thereisnodifferencebetweenthem.Asafurtherrobustnesscheck,we
includearespondentdummyinallspecifications,butthedummiesarestatistically
insignificant.
8
birthweight.Theotherisdefinedasadichotomousvariablemeasuringthelow
birthweight,codedas1ifthebirthweightismorethan2,500gramsandzero
otherwise. ThedescriptivestatisticssummarizedinTable2showthattheaverage
twinsinoursampleweighed2,441gramsatbirthandmorethanhalfofthemwere
categorizedasinfantswithlowbirthweights.Assuggestedinpreviousliterature,
ourdataillustratesadisparityinbirthweightbetweenthefirstandsecondbornof
twins:theaveragebirthweightofthefirstbornis2,464gramswhilethatofthe
secondbornis2,423grams,andthisdifferenceisstatisticallysignificantata5%
level.Furthermore,ourdatashowsthat24.8%ofMZtwinpairsweredifferentin
weightatbirth,whichiscruciallyimportanttoensureanaccurateestimateofthe
twinfixedeffectsmodel.IfwerestrictthesampletothosewhoareMZtwins,we
havequitesimilarresults(seeTable1‐a).Werunseparateregressionsforeach
variantofthebirthweight:theexplanatorypowerassessedbyR2statisticsfrom
withintwinfixed‐effectsestimationshelpsustochoosethebestpossibleoption
amongthesethreevariantsofbirthweight,aspresentedinTable3. Wethencharacterizesixoutcomesrangingfromtheperiodofchildhoodto
adulthood.Thefirstoutcomevariableisatypeofmiddleschoolattended,which
aimstomeasurescholasticabilityinearlychildhood.Sometwelve‐year‐old
childrenenrollinJapanprivatejuniorhighschoolsinsteadofpublicschools
becausemostprivatejuniorhighschoolsrequirechildrentopassentrance
examinations,whichareoftencompetitiveandselective.AccordingtotheSchool
BasicSurveyadministeredbyMinistryofEducation,Culture,Sports,Scienceand
Technology,theenrollmentrateofprivatejuniorhighschoolswas8%4 in2012,
withaconsiderablegeographicalvariation.Whilethemajorityofchildrenand
parentsdonotconsiderchoosingprivateschoolsatall,anon‐negligibleproportion
ofchildrenandparentsinJapanhaverecognizedtheentranceexaminationsof
privatejuniorhighschoolsasthefirstscreeningprocessthrougheducational
institutions.Inoursurvey,weaskwhichtypeofmiddleschooltherespondentand
his/hertwinsiblingsattended.17.1%ofrespondentswerestudentsofprivateand
nationaljuniorhighschools,whichisslightlyhigherthanthepercentageshowed
byofficialstatisticsnationwide.Thismayinpartbeduetothecharacteristicofour
surveythatitismorelikelytogatherinformationfromresidentsinlarge
4
Thisnumberincludestheenrollmentrateofnationaljuniorhighschools.Nationaljunior
highschoolsaregovernment‐ownedjuniorhighschools,mostlyaffiliatedwithnational
universities.Theseschoolsalsorequireapplicantstopassentranceexaminationstoenroll.
9
metropolitanareas,suchasTokyoandOsaka.Moreover,10.9%ofMZtwins‐pairs
attendeddifferenttypesofschool:forexample,oneattendedaprivateschooland
theotherapublicschool.Thesecondoutcomevariableisstudentperformanceat
middleschool,whichismeasuredona5point‐scale(1=lower;2=belowaverage;
3=average;4=aboveaverage;5=upper)basedonasubjectiveevaluationof
respondents’andtheirtwin‐siblings’academicachievements. Thethirdoutcomeistherankingofthecollegeattendedwitharestricted
sampleofcollege‐educatedrespondents.Oursurveyasksthenameofthehigh
schoolwhererespondentsandtheirtwinsiblingsgraduated.Weconvertthis
informationintoameasureofdeviationvalue(“hensachi”inJapanese),which
representstherankingofeacheducationalinstitutionwithameanof50and
standarddeviationof10byusingaseriesofdeviationvalues.Wematchthename
ofcolleges/universitiesandthedeviationvalueofeachdepartmentofeach
college/universitybyusingthedataset,“KawaijukuCollegeRankings”,calculated
andreleasedin2011byKawaijuku,oneofthelargestcramschoolsinJapan.
CollegeadmissionsinJapanaredeterminedalmostentirelybyperformancein
writtenexaminationsadministeredbyeachinstitutionexceptforsomespecial
admissionprograms,suchasathleticscholarshipprograms. Thefourthoutcomeisyearsofschooling.Toavoidthepossibilityof
institutionalmisreporting,inouroriginalquestionnaire,welisteverytypeand
levelofeducationalinstitution(26categories,including“don’tknow”),andthen
askrespondentstoselectthehighestdegreeobtained.Thechoiceof“dropoutor
stopped”wasinsertedbetweenthequestionsoneachtypeandlevelofinstitution
inordertodisentanglecasesofleavingschoolwithoutadiploma.Incalculatingthe
yearsofeducation,weassumethatthosewho“droppedout”or“stopped”finished
halfoftheminimumrequiredyearstocompletetheeducationalinstitutionlast
attended(e.g.,dropoutfromhighschool=10.5yearsofschooling). Itisimportanttonotethatourdatashowsthatthewithin‐twinvariations
ineducationalexperiencesbecomelargerwiththepassageoftime:only5.9%or
10.9%oftwinpairsattendeddifferenttypesofschoolswhentheywereprimaryor
middleschoolstudents.Eventually,38.4%oftwinpairsinoursampleacquired
differentnumbersofyearsofeducation.Somemaywonderwhytwins,whoshare
thesamegeneticmakeupandfamilyenvironment,eventuallyendupsodifferently
intermsofeducationalbackgrounds.BoundandSolon(1999)andNeumark
(1999)pointedoutthatevenifthewithin‐twinestimateisabletoremovethe
effectofgeneticendowment,itdoesnotnecessarilymeanthatthewithin‐twin
10
estimatecompletelyeliminatesallaspectsofabilitybiastotheextentthatability
goesbeyondgenes.Thepotentialendogeneitycouldbecausedbyunobserved
differencesinabilitybetweentwin‐pairswhichmayresultfromdifferent
experiencesinearlierchildhood.However,accordingtooursurvey,parentsare
unlikelytotreattwinsdifferently,particularlyintermsofeducationalinvestment.
Forexample,thereislittledifferenceinexperiencesofshadoweducationbetween
twin‐pairs,whichmaybeacrucialpartofhouseholdexpendituresoneducationin
Japan5.Oursurveyshowsthatthewithin‐twinvariationsinshadoweducationis
quitesmall,asthewithin‐twinvariationsinformaleducationbecomelargerwith
thepassageoftime(seeTable1‐b).Therefore,itcanbesaidthatwithin‐twin
variationsdonotreflectanysystematicdifferenceintheparentaleducational
strategyforeachtwin,whichoftenleadstodifferentexperiencesinearlier
childhood. Thefifthoutcomeislabormarketperformancemeasuredbythenatural
logarithmofannualwagebeforetaxdeductionsduringthe2010fiscalyear6.The
responsecategoryintheoriginalquestionnairerangedfrom1(=noincome)
through16(=morethan15millionJPY).Wesettheminimum(1=noincomeand
2=lessthan0.5millionJPY)tozeroandmaximum(16=morethan15millionJPY)
to15millionJPY.Then,wetakethemidvalueforcategoriesbetween3(=0.5
million‐0.99millionJPY)and15(=10million‐14.99millionJPY). EmpiricalResults
Table4presentstheresultsestimatedbyaconventionalOLStoreplicate
thecorrelationalstudiesinpreviousliterature.Regardlessofthevariantof
dependentvariables,birthweightsarestronglyassociatedwithadulthood
outcomes,exceptforthetypeofmiddleschoolandrankingofcollegeattended. Theresultscoupledwiththepositivecoefficientssuggestthatbirthweightaffects
themajorityofeducationaloutcomeslaterinlife,suchasstudentperformanceat
theageofaround15andthehighestyearsofschooling,aswellasthelabormarket
5
AccordingtotheofficialstatisticsoftheMinistryofEducation,Culture,Sports,Scienceand
Technology,alargepartofhouseholdexpendituresoneducationhasbeenspentonshadow
educationinJapan.TheBenesseEducationalResearchandDevelopmentCentershowedin
2009thatapproximately20%ofelementaryandhighschoolstudentsand50%ofjuniorhigh
schoolstudentsaccessedshadoweducation,suchascramschoolsorprepschools.
6 Thissurveyaskedaboutearningsduringthefiscalyearof2010,insteadof2011,because
earningsduringthefiscalyearof2011couldhavebeenaffectedbytheGreatEastJapan
EarthquakethatoccurredonMarch11,2011.
11
outcomemeasuredbyannualearnings.Theeffectseemsquitelarge:forexample,
anextra100gramsinbirthweightraisesearningsinthelabormarketby5.3%on
average.Takenasawhole,ourresultsareconsistentwiththemainstreamof
previousliteratureshowingapositivecorrelationbetweenbirthweightand
adulthoodoutcomes. Thenweanalyzetheeffectofbirthweightonadulthoodoutcomes
concerningwhetherthedifferenceinweightatbirthissubstantialorreflectsthe
factthatinfantswithlighterbirthweightsfundamentallydifferfromthoseoftheir
counterpartswithheavierbirthweights.Asexplainedearlier,weemploythe
twin‐fixedeffectstocomparethedifferencesbetweenMZtwinstoisolatethepure
effectofbirthweightontheadulthoodoutcomes,holdinginnateabilitiesand
familyenvironmentsconstant.Table5showsaquitedifferentstoryfrom
conventionalOLSestimates.Theeffectofthebirthweightontheadulthood
outcomesinthelongerrun,suchasthehighestyearsofeducationandearnings,
appearstobestatisticallyinsignificantacrossmodels.However,theeffectstill
remainsineducation:birthweightcausesacademicachievementattheageof
around15.Itmayalsocausetheprobabilityofpassingaprivatejuniorhighschool
andtherankingofcollegeinwhichtheyparticipated,althoughtheevidenceto
supportthisisweak.Thetwin‐fixedeffectsestimatesaresubstantiallylargerthan
thecross‐sectionalones.
Thispaperhasthussucceededinansweringtheresearchquestion.Like
theevidencegeneratedfromwesterncountries,ourfindingalsosuggestthatbirth
weightaffectseducationaloutcomes,particularlystudentperformanceattheage
ofaround15.Ontheotherhand,theeffectisnotlong‐lasting,andbirthweight
doesnotaffectlonger‐runoutcomes,suchasthehighestyearsofschoolingand
earnings.Strictlyspeaking,inthissense,birthweightitselfisnottheculpritwith
respecttothemoredisadvantagedadulthoodoutcomes. Conclusion
Thispaperhasinvestigatedthequestionofwhethertheeffectsofbirth
weightonlatereducationalandeconomicoutcomes,ifany,arecausal.Byusing
datafromasampleoftwinsinJapan,wehaveprovidedbestavailableestimatesfor
thecausaleffectsofbirthweightonacademicachievementattheageofaround15.
Wewouldberathercautiousaboutconcludingthatbirthweightdoesnotaffect
longer‐runoutcomes,suchashighestyearsofschoolingandearnings.Future
12
researchshouldexaminewhetherthisisalsothecasewhereeducational
attainmentmeasuredbyeducationalqualificationsobtained,orwhereearningsare
measuredwithsmallermeasurementerrorsthanwithself‐reportedearnings.
Giventhatthemostplausiblefactoraffectingbirthweightdifferences
betweenMZtwinsisintrauterinenutrientintake,ourfindingssuggestthat
improvingpregnantmothers’nutrientintakemayleadtotheimprovementof
educationaloutcomesatage15,therebyimprovingfuturelifechances. KawaguchiandNoguchi(2012a)arguedthatthedecreaseintheaveragebirth
weightbetween1990and2005couldbepartlyexplainedbymedicalinstructions
offeredtopregnantmothers.Thisimpliesthatincreasingbirthweightthrough
improvingpregnantmothers’nutrientintakewouldbepolicy‐relevant. Inordertobringoutmoredetailedandsolidimplicationstopolicy,future
researchneedstotestwhetherthefindingsofthispaperbasedonatwinsample
canbegeneralizedtonon‐twinpopulations.Itisalsonecessarytoinvestigatefor
whomincreasingbirthweightismosteffectiveinimprovingfuturelifechances,for
instance,bydetectingatwhichpointinthebirthweightdistributiontheeffectof
birthweightisstrongest. 13
Table1:SampleCollectedthroughWeb‐BasedSurvey
MZTwins
DZTwins
Don’tKnow
2,742
(1,371pairs)
1,764
(882pairs)
214
(107pairs)
(Source)Authors’calculations
Table1‐a:DifferencesinEducationalOutcomesbetweenTwin‐Pairs
Birthweights
Privateprimary
school
Privatemiddle
school
Highestyearsof
education
Total
MZTwins
DZTwins
24.8%
21.0%
32.1%
5.9%
5.6%
6.5%
7.5%
7.1%
7.7%
38.4%
32.0%
48.8%
Table1‐b:DifferencesinEducationalExpenditureonShadowEducationbetweenTwin‐Pairs
Preschool Primaryschool
Middleschool
Highschool
Total
MZTwins
DZTwins
2.4%
1.6%
3.2%
4.1%
3.1%
5.8%
6.2%
4.7%
9.0%
6.5%
4.6%
9.0%
(Note)1.Shadoweducationrepresentsaccesstoprepschools,privatetutoring,anddistancelearning.
2.Privateprimaryandmiddleschoolsareincludingnationalschools. (Source)Authors’calculations
14
Table2:DescriptiveStatistics
WholeSample
DependentVariables:
Birthweight
Log(birthweight)
Non‐LBW(1>2,500grams)
IndependentVariables:
Typeofjuniorhighschool(1=privateornational)
Studentperformanceattheageof15 Rankingofcollegeattended
Yearsofschooling
Log(earnings)
(Source)Authors’calculations
15
MZTwins
Mean
STDV
Mean
STDV
2,441.00
7.77
0.45
564.87
0.23
0.50
2413.76
7.76
0.50
551.51
0.23
0.50
0.171
2.57
52.00
14.38
5.96
0.376
1.10
9.84
2.33
0.74
0.16
2.60
51.72
14.42
5.98
0.37
1.08
9.76
2.30
0.74
Table3:GoodnessofFit
Independent
Dependent
Birthweight
Log(birthweight)
Non‐LBW
Private
Middle
School
Student
Performance
(Age15)
Ranking
College
Highest
Yearsof
Schooling
Earnings
0.0060
0.0054
0.0075
0.0052
0.0070
0.0014
0.0061
0.0057
0.0194
0.0002
0.0001
0.0008
0.0013
0.0016
0.0006
(Source)Authors’calculations
16
Table4:EmpiricalResults(OLS)
Independent
Private
Middle
School
Birthweight
(/100)
Log(birthweight)
Non‐LBW
Observations
0.011
(0.016)
0.016
(0.038)
‐0.001 (0.017)
1,998
Dependent
Student
Performance
(Age15)
0.063*
(0.032)
0.140* (0.079)
0.035 (0.035)
3,577
Ranking
College
Highest
Yearsof
Schooling
Earnings
0.091 (0.442)
0.178 (1.082)
0.542 (0.517)
1,470
0.291*** (0.066)
0.810***
(0.167)
0.206** (0.070)
3,631
0.053***
(0.017)
0.122*** (0.043)
0.035* (0.019)
2,785
(Note)1.Thenumbersinparenthesesareheteroskedasticity‐robuststandarderrors. 2.***,**,and*represent1%,5%,and10%significancelevel,respectively.
3.Otherindependentvariablesare(a)privatemiddleschool:gender,father’seducation;(b)studentperformance:gender,father’seducation,
livingstandardattheageofaround15;(c)rankingofcollege:thesamewith(b);(d)highestyearsofschooling:thesamewith(b);(e)log
(earnings):age,agesquared,gender,father’seducation,livingstandardattheageofaround15,highestyearsofschooling,maritalstatus,years
oftenureatthecurrentemployment,hoursofworkperday.Mostofthecontrolvariablesarestatisticallysignificant.Thefullresultsare
availableuponrequest. (Source)Authors’calculations
17
Table5:EmpiricalResults(Twin‐FixedEffects)
Independent
Private
Middle
School
Birthweight
(/100)
Log(birthweight)
Non‐LBW
Observations
#oftwinpairs
0.050* (0.029)
0.112* (0.066) 0.053* (0.030) 1,257
(641)
Dependent
Student
Performance
(Age15)
0.210* (0.108) 0.575**
(0.249) 0.099
(0.088) 2,206
(1,138)
Ranking
College
3.138 (3.090) 7.326 (7.149) 5.659* (3.055) 918
(736)
Highest
Yearsof
Schooling
‐0.077
(0.170)
‐0.137
(0.417)
‐0.146 (0.162)
2,234
(1,144)
(Note)1.Thenumbersinparenthesesareheteroskedasticity‐robuststandarderrorsandclusteringatthefamilylevel. 2.***,**,and*represent1%,5%,and10%significancelevel,respectively.
(Source)Authors’calculations
18
Earnings
0.065
(0.071) 0.177
(0.161)
‐0.042 (0.072)
1,832
(1,032)
Reference
Almond,D.,Chay,K.Y.&Lee,D.S.(2005).Thecostsoflowbirthweight.Quarterly
JournalofEconomics,120(3),1031‐1083.
Black,S.E.,Devereux,P.J.&Salvanes,K.G.(2007).Fromthecradletothelabor
market?Theeffectofbirthweightonadultoutcomes.QuarterlyJournalof
Economics,122(1),409‐439. Behrman,J.R.&Rosenzweig,M.R.(2004).Returnstobirthweight.Reviewof
EconomicsandStatistics,86(2),586‐601.
Blickstein,I.&Kalish,R.B.(2003).Birthweightdiscordanceinmultiplepregnancy.
TwinResearch,6(6),526‐531. Bound,J.&Solon,G.(1999).Doubletrouble:onthevalueoftwins‐based
estimationofthereturntoschooling.EconomicsofEducationReview,18(2),
169‐182.
Conley,D.&Bennett,N.G.(2000).IsBiologydestiny?birthweightandlifechances.
AmericanSociologicalReview,65(3),458‐467.
Currie,J.(2009).Healthy,wealthy,andwise:socioeconomicstatus,poorhealthin
childhood,andhumancapitaldevelopment.JournalofEconomicLiterature,
47(1),87‐122.
Currie,J.&Hyson,R.(1999).Istheimpactofhealthshockscushionedby
socioeconomicstatus?thecaseoflowbirthweight.AmericanEconomic
Review,89(2),245‐250.
Kanjuku(2011).Zenkokukoukouchuugakuhensachisouran[Acomprehensivelist
ofthedeviationvaluesofJapan’shigh‐andjuniorhighschools].
Kawaijuku(2011).Kawaijukuhensachiranking[Therankingofcolleges].
RetrievedFeb15,2013from
http://daigaku.jyuken‐goukaku.com/nyuushi‐hensati‐ranking/kawaijyuku/ Kawaguchi,D.&Noguchi,H.(2012a).Shinseijinotaijuhanazegenshoshiteiru
noka[Whyhasbirthweightdecreased?].Ihori,T.,Kaneko,Y.&Noguchi,H.
eds.Aratanarisukutoshakaihosho:shogaiwotsujitashiensakunokochiku
[Newrisksandsocialsecurity],UniversityofTokyoPress,17‐33. Kawaguchi,D.&Noguchi,H.(2012b).Teitaijyushussei:genintokiketsu[Thelow
birthweight:thecauseandtheconsequence].GlobalCOEHi‐StatDiscussion
PaperSeries,265.
Kohara,M.&Ohtake,F.(2009).Kodomonokyouikuseikanoketteiyouin[The
determinantsofeducationaloutcomesofachild].TheJapaneseJournalof
19
LabourStudies,588,67‐84.
Lawlor,D.,Clark,H.,Smith,G.D.,&Leon,D.(2006).Intrauterinegrowthand
intelligencewithinsiblingpairs:findingsfromtheAberdeenchildrenofthe
1950sCohort.Pediatrics,117(5),894‐902.
Lin,M.&Liu,J.(2009).Dolowerbirthweightbabieshavelowergrades?:Twin
FixedeffectandinstrumentalvariablemethodevidencefromTaiwan.Social
Science&Medicine,68,1780‐1787.
Linnet,K.M.,Wisborg,K.,Agerbo,E.,Secher,N.J.,Thomsen,P.H.&Henriksen,T.B.
(2006).Gestationalage,birthweight,andtheriskofhyperkineticdisorder.
ArchivesofDiseaseinChildhood,91(8),655‐660.
Miller,P.,Mulvey,C.&Martin,N.(2005).Birthweightandschoolingandearnings:
estimatesfromasampleoftwins,EconomicsLetters,86,387‐392. Nakamuro,M.&Tomohiko,I.(2012).Estimatingthereturnstoeducationusingthe
sampleoftwins;thecaseofJapan.RIETIDiscussionPaperSeries,12‐E‐076.
Neumark,D.(1999).Biasesintwinestimatesofthereturntoschooling.Economics
ofEducationReview,18(2),143‐148.
Oreopoulos,P.,Stabile,M.,Walld,L.&Roos,L.L.(2008).Short‐,medium‐,and
long‐termconsequencesofpoorinfanthealth:ananalysisusingsiblingsand
twins.JournalofHumanResources,43(1),88‐138.
Petronis,A.(2010).Epigeneticsasaunifyingprincipleintheaetiologyofcomplex
traitsanddiseases.Nature,465,721‐777.
Tsukamoto,H.,Fukuoka,H.,Koyasu,M.,Nagai,Y.&Takimoto,H.(2007).Risk
factorsforsmallforgestationalage.PediatricsInternational,49(6),985‐990.
Royer,H.(2009).Separatedatgirth:U.S.twinestimatesoftheeffectsofbirth
weight,AppliedEconomics,1(1),49‐85.
20