Unequal BequestsIZA DP No 8813Marco FrancesconiRobert A PollakDomenico Tabasso Unequal BequestsMarco FrancesconiUniversity of Essexand IZARobert A PollakWashington University in St Louis IZA Discu ID: 476289
Download Pdf The PPT/PDF document "Forschungsinstitut zur Zukunft der Arbei..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Forschungsinstitut zur Zukunft der ArbeitInstitute for the Study of Labor Unequal BequestsIZA DP No. 8813Marco FrancesconiRobert A. PollakDomenico Tabasso Unequal BequestsMarco FrancesconiUniversity of Essexand IZARobert A. PollakWashington University in St. Louis IZA Discussion Paper No. January 2015BSTRACTUnequal BequestsUsing data from the Health and Retirement Study (HRS), we make two contributions to the literature on endlife transfers. First, we show that unequal bequests are much more common than generally recognized, with onethird of parents with wills planning to divide their estates unequally among their children. These plans for unequal division are particularly concentrated in complex families, that is, families with stepchildren and families with genetic children with whom the parent has had no contact (e.g., children from previous marriages). We find that in complex families past and current contact between parents and children reduces or eliminates unequal bequests. Second, although the literature focuses on the bequest intentions of parents who have made wills, we find that many elderly Americans have not made wills. Although the probability of having awill increases with age, 30 percent of HRS respondents aged 70 and over have no wills. Of HRS respondents who died between 1995 and 2010, 38 percent died intestate (i.e., without wills). Thus, focusing exclusively on the bequest intentions of parents who have made wills provides an incomplete and misleading picture of endlife transfers.JEL Classification: D13, J12, K36Keywordsequests, intergenerational transfers, altruism, exchange, evolutionary motives, family structureCorresponding authoMarco FrancesconiDepartment of EconomicsUniversity of EssexWivenhoe ParkColchester CO4 3SQUnited Kingdommail:mfranc@essex.ac.uk We are grateful to Sonia Bhalotra and seminar participants at the University of Geneva, UCLA, the 2014 RES Conference, 2014 ESPE Conference, 2014 EEA Conference and 2014 EALE Conference for comments and suggestions. The text refers to supplementary online material, which is available at https://sites.google.com/site/domtabasso/appendixmaterialforunequalbequests 1.IntroductionInthispaperweinvestigatebequestbehaviorusingalargeandnationallyrepresentativeUSsampledrawnfromtheHealthandRetirementStudy(HRS)overtheperiod1995{2010.Adistinguishingfeatureofourworkisitsfocusoncomplexfamilies,inparticularonparentswithstepchildrenandparentswithgeneticchildrenwithwhomtheyhavelimitedornocontact(e.g.,childrenfrompreviousmarriages).WecomplementouranalysisofthebequestintentionsofparentswithwillsbyexaminingactualbequestsusingreportsaboutthedispositionoftheestatesofHRSrespondentswhodiedbetweenoneHRSwaveandthenext.Wendthatunequalbequests(bothintendedandactual)aremuchmoreprevalentthanpreviouslydocumented,withone-thirdofparentswithwillsplanningtodistributetheirestatesunequally.Unequalintendedbequestsaremostcommonincomplexfamilies,butcontactbetweenparentsandchildrenreducesoreliminatesunequalbequests.Finally,wendthatasubstantialfractionofelderlyparentsreporthavingnowillsandasubstantialfractionofHRSrespondentswhodiedhadnowills.Empiricalresearchhaslongestablishedthatasubstantialmajorityofparentsintendtodividetheirestatesequallyamongtheirchildren(Menchik1980;Wilhelm1996;McGarry1999;Cox2003;LightandMcGarry2004;BehrmanandRosenzweig2004).Despitethefactthatearliereconomicmodelspredictedunequalbequests(e.g.,Becker1974;Bern-heim,Shleifer,andSummers1985;Cox1987),morerecentstudieshavedevelopedtheoriesthatattempttorationalizetheprevalenceofequalbequests,sometimesalsoattemptingtoexplainwhyparentstreattheirchildrenunequallywithrespecttointervivostransfersbutequallywithrespecttobequests(e.g.,Andreoni1989;BernheimandSeverinov2003).TheproportionofAmericanparentsaged50andoverwhoreportedhavingwills,1inwhichtheirchildrenweretreatedunequallymorethandoubledbetween1995and2010,risingfrom16percenttoalmost35percent(Figure1).2Thisupwardtrend,whichholdsforbothmothersandfathers,isnotsimplydrivenbytheageingoftheHRSrespondents.3 1WefocusonthebequestintentionsofparentswithwillsbecausemostoftheinformationonbequestintentionscollectedbyHRSisfromparentswhoreportthattheyhavewills.Seethedatadescriptioninsubsection3.B.2Throughoutthepaper,stepchildrenarecountedas\children".Moreover,althoughthelegaldenitionofstepchildrenisnarrow(i.e.,astepchildisthechildofaspouse),weusethistermbroadly,toincludethechildrenofacohabitingpartneraswellasthechildrenofalegallymarriedspouse.3Between1998and2010theaverageageoftheHRSsampleincreasedonlybyfouryears.Thisisdue2 AsimilarincreasecanbeobservedacrossseveralcohortsofAmericansbornsince1890(Figure2).Infact,youngercohorts(i.e.,thosebornin1960orafter)aremorelikelytointendunequalbequests.Thesharpestincreasesoccurredamongindividualswhoarenolongermarried:widows,widowers,anddivorcedindividuals(Figure3).4Sincethemiddleofthe1990sthefractionofparentsreportingunequalbequestinten-tionshasconsistentlybeen30percentagepointshigheramongparentswithstepchildrenthanamongthosewithbiologicalchildrenonly(Figure4).Comparingparentswhohavehadnocontactwithatleastoneoftheirgeneticchildrenwiththosewhohavehadatleastsomecontactwithalloftheirgeneticchildren,wendtheno-contactparentsabout25percentagepointslesslikelytointendequalbequests.5Thetrendsinunequalbequestintentionsamongparentswithoutcontactwiththeirchildrenisthesameasthetrendobservedamongparentswithstepchildren.Ourdataindicatethatastheproportionofstepparentshasrisenconsiderablyinthelasttwentyyears,sohasthefractionofparentswhoplanunequalbequests.Whenindividualsdiewithoutavalidwill,theintestacylawsoftheirstateprovidethedefaultallocation,dividingthedecedent'sestatebetweenthesurvivingspouseandthechildren.6Afterprovidingasubstantialshareforthesurvivingspouse,intestacylawsdividetheremainderequallyamongthedecedent'sbiologicalandlegallyadoptedchildren.Unlikesurvivingspouses,underintestacylawssurvivingcohabitingpartnersinheritnothing.And,unlikebiologicalandlegallyadoptedchildren,underintestacylawsstepchildreninheritnothing(Fried1992;Brashier2004).Since1998theaveragefractionofHRSrespondentswithoutawillhasbeenaround42percent(Figure5).7Amongrespondentswithstepchildrentheaveragefractionwithoutwillsissomewhatgreater(49percent),anditisevengreateramongparentswhohave toanumberofreasons,includingattrition,deathofolderrespondentsand,since2004,theintroductionofnewcohortsofindividualsaged51to56.4HRSdatadonotallowustodistinguishbetweencohabitorsandmarriedindividuals.5In1995theHRSdidnotaskthequestiononparent-childcontact,whilein2010thequestionwasaskedonlytoasmallfractionofrespondents.ThesearethereasonswhyinFigure4thosetwoyearsarenotshown.6AsRosenbury(2005)shows,thereareimportantdi erencesinintestacylawsacrossstates,butwefocusonfeaturesthatarecommonacrossstates.7Thisgurestartsin1998becausethe1995HRSwaveoversampledolderpeople(withanaverageageof78years),whilethe1996waveoversampledyoungerindividuals,whoseaverageageis59years.From1998onwards,thestudycomprisedbothsubsamples,andinthatyearindividualswereonaverageaged67.3 nocontactwiththeirgeneticchildren(58percent).Parentsagedlessthan70aremuchlesslikelytohavewills,perhapsbecausewritingawillisnotyetsalientforthem,butastaggering30percentofparentsaged70andoverreportnothavingwills.Standardeconomicmodelsignorecomplexfamilies.The(usuallyimplicit)assumptionisthatallchildrenareborntoamarriedcouplewhoremainmarriedtoeachother.Whenonespousedies,thesurvivingspouseis(usuallyimplicitly)assumednottoremarry.Littleissaidaboutdivorce,remarriageorrepartneringandevenlessaboutmultiplepartnerfertility.8Byignoringdivorce,repartnering,orremarriage,canonicaleconomicmodelsfailtorecognizetheincreasedcomplexityofthefamily(BumpassandLu2000;StevensonandWolfers2007;LundbergandPollak2014).Inadditiontopresentingarepresentativepictureofcontemporaryend-of-lifetransfers,wemaketwocontributionstotheliterature.First,weshowthatunequalbequestsaremuchmorecommonthangenerallyrecognized.Unequalbequestsareconcentratedincomplexfamilies,thatis,familiesinwhichparentshavestepchildrenandfamilieswithgeneticchildrenwithwhomtheparenthashadnocontact(e.g.,childrenfrompreviousmarriages).Parentswithstepchildrenaremuchlesslikelytoincludeallchildrenintheirwillsthanparentswithoutstepchildren,andparentswithstepchildrenwhohavewillsaresubstantiallylesslikelytoplanequalbequests.Similarly,parentswhohavehadnocontactwithoneormoreoftheirgeneticchildrenarelesslikelytoincludealloftheirchildrenintheirwills.Whenallchildrenareincludedintheparent'swill,wendthatparentswithno-contactchildrenarelesslikelytoplanequalbequests.Thelikelihoodofunequalbequests,however,isreducedandoftenentirelyeliminatedbylongercoresidenceofstepparentsandstepchildren.Weinterpretthisndingasre ectingtheaccumulationoffamily-speciccapital(e.g.,trustanda ection)thattriggersnormsofequaltreatment.Oursecondcontributiontotheliteratureonend-of-lifetransfersshiftsthefocusfromindividualswithwillstothosewhodieintestate(i.e.,withoutwills).WendthatmanyelderlyAmericansdonothavewills.Morespecically,40percentofHRSrespondentsreportnothavingwillsand30percentofHRSrespondentsaged70andoverreportnothavingwills.OfHRSrespondentswhodied,38percentdiedintestate.Hence,theusual 8Inatraditionalnuclearfamilyallthechildreninthehouseholdarejointchildren,butoneorbothspousesmayhavechildrenfrompreviouspartnershipslivingelsewhere.4 focusonbequestintentionsprovidesanincompleteandmisleadingpictureofend-of-lifetransfers.Thepaperisorganizedasfollows.Section2reviewspreviousworkonbequestsanddescribesthelegalenvironmentinwhichindividualsmakeend-of-lifetransfers.ThedataaredescribedinSection3.Section4presentstheresultsofourempiricalanalysisofbequestintentionsandoftheactualdivisionofestates.Section5providesasimpleconceptualframeworkforinterpretingend-of-lifetransfers.Section6concludes.2.BackgroundandRelatedLiteratureAlmostalleconomicmodelspredictunequalbequests.9Forexample,thealtruistmodelassumesthatparentsequalizemarginalutilitiesacrosschildren(Barro1974;Becker1974;BeckerandTomes1979;Tomes1981).Thisassumption,togetherwithsomestrongassumptionsaboutpreferencesandintervivostransfers,impliesthatparentswillbequeathmoretotheirlesswell-o children.Exchangemodelsassumethatbequestsaremadetochildreninreturnfortheirservicessuchasattentionandcare(Bernheim,Shleifer,andSummers1985;Cox1987;CoxandRank1992).Becausechildrenfacedi erentopportunitycostsofprovidingtheseservices,exchangemodelspredictthatchildrenwillprovidedi erentamountsofservicesandwillreceiveunequalbequests.10Althoughboththealtruistmodelandexchangemodelshavesomeempiricalsupport(Tomes1988,CoxandRank1992;LaitnerandOhlsson2001),mostempiricalstudieschallengebothclassesofmodelsndingthatanoverwhelmingmajorityofparentsdividetheirestatesequallyamongtheirchildren(Menchik1980,1988;Wilhelm1996;McGarry1999;LightandMcGarry2004;BehrmanandRosenzweig2004).11Signicante orthasthenbeendevotedtorationalizingequalbequests.Forinstance,BernheimandSeverinov(2003)proposeamodelofintergenerationaltransfersbasedontheassumptionthateach 9ComprehensivereviewsoftheextensiveeconomicsliteratureonbequestsaregivenbyBehrman(1997),Laitner(1997),LaferrereandWol (2006),andMcGarry(2008,2013).Foranintroductiontothelegalliterature,seeFriedman(2009)andGrossmanandFriedman(2011).Foracomprehensivetreatmentofthelegalissues,seeDukeminierandSitko (2013).10Bargainingpowerandbargainingabilitywillalsoplayaroleintheabsenceofspecialassumptions(e.g.,theabilityofparentstomaketake-it-or-leave-ito erstotheirchildren.)11Wilhelm(1996)whichusefederalestatetaxdata,doesallowforadoptedchildrenbutnotforstepchil-drenandassumesthatparentshaveequal(symmetric)concernforalltheirchildren.Tothebestofourknowledge,allbequestmodelsdrivenbyaltruismorexchangemotivesignorestepchildren.5 child'sperceptionofparentala ectionin uenceshisorhersubjectivewell-being.Childrencannotdirectlyobserveparentalpreferences,butparentssignala ectionthroughtheiractions,includingbequests.Altruisticparentsthenmustconsiderthepossibilitythatunequalbequestsmayleadtheirchildrentoinferthattheyarelovedeithermoreorlessthantheirsiblings.Theassumptionthatthedivisionofintervivosgiftsisnotobservedbyallthechildren,whereasthedivisionofbequests(orthedivisionimpliedbybequestintentions)is,remainsuntested.Norisitclearwhetherandhowparents'statedbequestintentionsa ectchildren'sactions(e.g.,caregiving)regardingtheparents.Equaldivisionisalsoconsistentwithparents'indi erenceoverhowtheirestatesaredividedamongtheirchildren.Butindi erenceisbothimplausibleandtheoreticallyunsatisfyingbecauseitiscompatiblewithallpossibledivisionpatterns.12Evolutionarypsychologysuggestsasuiteofhypothesesaboutend-of-lifetransfersthatarestilllargelyuntested(Cox2003,2007).Theunderlyingpremiseisthatparentsbehavesoastomaximizetheprobabilityofsurvivaloftheirgenesandthatchildrenwithgreaterwealtharemorelikelytopassontheirgenes.Oneimplicationiswhatwecallthe\genetic-childhypothesis"|thatis,parentswillmakeend-of-lifetransferstotheirgeneticchildrenratherthantotheirsocialchildren(i.e.,geneticallyunrelatedchildrensuchasstepchildrenwholiveinthesamehousehold).13Weconsidertheimplicationsofthishypothesisintwocases:stepchildrenandgeneticchildrenwithwhomthedecedenthashadnocontact(e.g.,childrenfromapreviousmarriagewhowereveryyoungwhentheparentsdivorced).Thegeneticchildhypothesismakesclearpredictionsinbothofthesecases.Decedentswillfavortheirgeneticchildren.14 12Pollak(1988)arguesthatthecredibilityoftheparents'threattodisinheritachildinthestrategicbequestmodelofBernheim,Shleifer,andSummers(1985)cruciallydependsontheassumptionthatparentsareindi erentoverhowtheirestatesaredividedamongtheirchildren.13Unlikemodelsbasedonaltruismorexchange,whichhingeonactionstakenindividuallybyparentsandchildren(suchasthechild'sneedforsupportorthefrequencyofvisitstoandothercontactswithelderlyparents),thegenetic-childhypothesisreliesononespecictrait{thegeneticlinkbetweendece-dentsandpotentialbeneciaries.Inthisrespect,thegenetic-childhypothesisissimilartomodelsofintrahouseholdallocationthatemphasizeasingle(exogenous)attribute,suchasbirthorder(BehrmanandTaubman1986;Chu1991)orthechild'ssex(Behrman,PollakandTaubman1986).Althoughtheseothersingle-attributemodelsmaybeusefulforunderstandingdi erentialintervivostransfersorbequestsrelatedtobirthorderorsex,theycannotexplainunequalbequestsorbequestintentionstowardgeneticandsocialchildren.14Evolutionaryreasoningalsospeakstothedistributionofbequestsamongbiologicalchildrenand,whenitdoes,itseldompredictsequalbequests.Forexample,achildlesspost-menopausaldaughterwouldnotbepredictedtoreceivebequests.Theseaccountshoweverareoutsidethescopeofourpaper.6 Theonlyempiricalworkthatattemptstoassesstherelativeimportanceoftheal-truism,exchange,andthegenetic-childhypothesesisLightandMcGarry(2004).Usingintendedbequestdataforasampleof45-to80-year-oldmothersdrawnfromtheNa-tionalLongitudinalSurveysofYoungWomenandMatureWomen,theyndthatthevastmajorityofmothers(morethan92percent)intendtoleaveequalbequests.Themotherswhosaidtheyintendedtoleaveunequalbequestswereaskedtoexplainwhy.Somerespondedwithexplanationsthatwereconsistentwithaltruism,otherswithex-changeand,amongmotherswithstepchildren,somewithexplanationsconsistentwiththegenetic-childhypothesis.LightandMcGarryndthatgreaterwithin-familyvariationinchildren'sincomes(aproxyforaltruism),poormaternalhealth(aproxyforexchange),andthepresenceofstepchildren(aproxyforthegenetic-childmotive)areassociatedwithhigherprobabilitiesofunequalintendedbequests.Morespecically,formotherswithatleastonegeneticchildandatleastonestepchild,theyndthattheprobabilityofunequalintendedbequestsincreasesfrom7.9to11.3percent,a43percentincrease.Otherstudiesexaminetheextenttowhichthedivisionofend-of-lifetransferscom-pensatesforcaregiving.Forexample,usingdatafromtherstwaveoftheAssetsandHealthDynamicsamongtheOldestOld,Brown(2006)ndsthatchildrenwhoarecur-rentlycaregiversare32percentagepointsmorelikelythantheirnoncaregivingsiblingstobeincludedintheirparents'lifeinsurancepolicies,whileexpectedcaregiversarethreepercentagepointsmorelikelytobeincludedintheirparents'willsand15percentagepointsmorelikelytobeincludedintheirparents'life-insurancepolicies.Unequaltransfersfromparentstochildrenandfromchildrentoparentshavealsobeendocumentedindivorcedfamilies.Analyzingthee ectsofparentalmaritaldisruptiononlate-lifeintervivostransfers,PezzinandSchone(1999)ndthatparents(especiallyelderlymen)engageinsubstantiallylowerlevelsoftransferswithstepchildrenthanwiththeirgeneticchildren.MaritaldisruptionisalsocentralinPezzin,Pollak,andSchone(2008).Thatstudy,however,concentrateson\upstreamtransfers"(i.e.,adultchildren'stimeandcashtransferstotheirunpartnereddisabledelderlyparents)ratherthanon\down-streamtransfers"(i.e.,transfersoftimeandcashfromparentstotheirchildren).Theyndunequal owsofservicestoparentsinfamiliesthatexperiencedivorce,withstepchil-7 drenprovidingsignicantlylowertransfersthangeneticchildren.OtherthanLightandMcGarry(2004),however,nopreviousstudyhasexaminedbequeststostepchildren.End-of-lifetransfersandbequestshavealsobeenanalyzedbylegalscholarsandcom-mentators.Unequaldivisionofestatesamonggeneticchildrentypicallygeneratesuneaseamongtrustandestateslawyersbecausetheyviewunequalbequestsasinvitationstolitigation(Collins2000;Blattmachr2008;AmericanBarAssociation2013).Stepchildren,however,belongtoadi erentcategorysincethelawtreatsstepchildrenasunrelatedstrangersratherthanasfamilymembers(SchanzenbachandSitko 2009).Legalscholarsalsowriteonintestacy,asubjectthusfarentirelyneglectedbyeconomists.Intestacystatutesdividetheestatesofmarrieddecedentswithchildrenbetweenthesur-vivingspouseandthedecedent'sgeneticandlegallyadoptedchildren.15Ifastepparentdieswithoutavalidwill,stepchildreninheritnothing.16Asanumberofscholarshavepointedout,stepchildrenhaveneverfaredwellunderintestacystatutes(Mahoney1989;Gary2000;Noble2002;Brashier2004;Cremer2011).Stepchildrenhowevermayinheritfromtheirabsentbiologicalparent,andsoitisunclearwhethertheyaretrulydisadvan-tagedbyintestacylaw.17ThisisanissuethatcannotbeexploredwiththeHRSdataandrequiresadditionalresearch.Somestepparentsmayintentionallyforegowritingawillpreciselybecausetheyknowthatintestacylawswillmandateanequaldivisionamongtheirbiologicalandadoptedchildrenandgivenothingtotheirstepchildren.Ontheotherhand,parentswithgeneticchildrenwithwhomtheyhavehadnocontactmightwritewillstoavoidgivingthesechildrenanequalshareoftheirestates.Someparentsarenodoubtawareofthedefaultdivisionimposedbyintestacylawwhileothersarenot.Unfortunately,HRSprovidesnoinformationaboutrespondents'knowledgeoforbeliefsaboutintestacylaw.Wecanexamine,however,forthersttime,whetherparentsarelesslikelytowriteawillifthey 15TheUniformProbateCodetreatsadoptedchildrenasiftheywerebiologicalchildrenofthedecedent(Noble2002;Cahn2005).16Californiaprovidesanarrowexceptiontothisgeneralization:astepchildmayinheritifitcanbeshownthatthestepparentwouldhaveadoptedthestepchildbutwaspreventedfromdoingsobyalegalbarrier.Thisexceptionisavailableonlytoastepchildwhosatisesthelegaldenitionofastepchild(i.e.,achildofthedecedent'sspouse,notofthedecedent'scohabitingpartner).SeeHanson(1995)andNoble(2002)formoredetails.17Atcommonlaw,infact,therelationshipofstepparentandstepchildgenerallyconfersnorightsandimposesnoduties(Wypyski1984).8 havestepchildrenoriftheyhavegeneticchildrenwithwhomtheyhavehadnocontact.3.DataA.SamplesOuranalysisusesdatacollectedbetween1995and2010bytheHealthandRetirementStudy(HRS),whichcontainsdetailedinformationaboutbequestintentionsandtheactualdistributionofestates.18TheHRSisalongitudinalsurveyofanationallyrepresentativesampleofmorethan26,000Americansovertheageof50whoareinterviewedeverytwoyears.19Ifarespondenthasaspouseorpartner,thespouseorpartnerisinvitedtobecomeanHRSrespondenttoo.Ineachsurveyyear,the\coreles"providedatafromstandardquestionnairesadministeredtoallrespondents.The\exitles"provideinformationabouttheactualdistributionoftheestatesofHRSrespondentswhodiedsincethepreviouswave;thisinformationiscollectedfromaproxyrespondent,suchasthesurvivingspouse,anadultchild,oranotherclosefamilymember.Fromthecorelesweselectrespondentswithatleastonechildandwithnonmissinginformationonintendedbequestsandotherbasicvariables.Thisleavesuswithanunbal-ancedpanelof23,984individuals,foratotalof117,189person-waveobservations.Whenrstobservedinthestudy,11,221individuals(47percentofthesample)reporthavingnowill,while12,763reporthavingawill.Asthesurveyprogresses,thepercentageofindividualswithoutawilldecreasesto42percent.Ofthe21,140parentsinoursamplewithmorethanonechild,11,170(53percent)reporttheirplanstodistributetheirestatesamongtheirchildren.Morethanonefthofthewholesample(5,082parents)reporthav-ingbothbiologicalchildrenandstepchildren.20Oftheseindividuals,2,342reporthavingawill(about18percentofthesampleofparentswithwillsorover46percentoftheparentswithstepchildren).TheHRSalsocollectsinformationonthefrequencyofcontactsbetweenparentsandchildren.Weusethisinformationforthe12,739individualswhohavegeneticchildren 18AsnotedintheIntroduction,HRSasksrespondentsabouttheirbequestintentionsonlyiftheyreporthavingwillsortrusts.19Seehttp://hrsonline.isr.umich.edu/&#x]TJ/;གྷ ; .96;& T; 7.;݉ ;� Td;&#x [00;formoreinformationaboutthedata.20TheHRSdoesnotdistinguishbetweengeneticchildrenandadoptedchildren.9 only.Justover17percentofthem(2,178parents)reporthavingnocontactinthepreviousyearwithatleastoneoftheirchildren.21Amongno-contactparents,theproportionwithoutawillexceeds57percent.Thisissubstantiallygreaterthanthatobservedamongparentswhohavemorefrequentcontactwithalltheirgeneticchildren(42percent).Theexitlesprovideinformationontheactualdispositionofestatesandotherbasicvariablesfor7,416individuals(almost85percent)ofthe8,800HRSrespondentswhodiedoverthesampleperiod.Thereare2,781parents(38percent)whodiedintestate,aslightlysmallerproportionthanthe42percentofHRSrespondentsinthecoreleswhoreporthavingnowills.Oftheremaining4,635whodiedwithawill,3,897hadmorethanonechildand630(18percentofthesampleofdecedentswithwillsandwithmorethanonechild)hadbothstepchildrenandgeneticchildren,representingmorethan90percentofdecedentswithstepchildren.22B.OutcomesTable1showsthemeansofourmaindependentvariablesbrokendownbythepresenceofstepchildreninboththecoreandtheexitlesandbyparents'maritalstatus.About42percentofthesampleinthecorelesdoesnothaveawill(column(a)).Therawdi erenceof6.9percentagepointsbetweenthosewithstepchildren(column(b))andthosewithgeneticchildrenonly(column(c))isstatisticallysignicant.Almosttwo-thirdsofdivorcedparentsdonothaveawill.Again,parentswithstepchildrenarelesslikelytohaveawillthanparentswithgeneticchildrenonly.Thesamepictureemergesfromtheexitles,eventhoughthefractionofallparentswithoutawillisalmost38percent,somewhatlessthaninthecoreles.Weshallreturntothisissueinthenextsection.AsmentionedaboveandintheIntroduction,wefocusonparentswithwillsbecausemostoftheinformationcollectedbyHRSisfromparentswhoreportthattheyhavea 21Amongparentswithbothstepchildrenandgeneticchildren,theproportionofparentswithnocontactwithatleastoneoftheirbiologicalchildrenishigher,beingaround20%.Sinceonly450parentsinoursamplebelongtothiscategory,however,adetailedanalysisoftheirbequestintentionsisnotpossible.22Theexitlesyieldasamplethatissmallerthanthatobtainedfromthecoreles.Moreover,atpresent,itdisproportionallyrepresentsindividualswithlowersocioeconomicstatusandhighermortalityrisks(Cutleretal.2011).10 will.23Conditionalonhavingawill,aboutthree-quartersoftheparentsfromthecorelesreporttheyincludeallchildrenintheirwills,whileonly59percentoftheestateswhosedistributionsarereportedintheexitlesweredividedinawaythatincludedallchildren.Intendedinclusionofallchildrenissubstantiallylesslikelyamongparentswithstepchildrenregardlessoftheirmaritalstatus,butthisdoesnotemergeintheexitles,wherenoneofthedi erencesarestatisticallysignicantatconventionallevels(column(d),panelB).Finally,inthecoreles,almostone-thirdofallparentswithawillreportthattheyplantodistributetheirestateunequallyamongtheirchildren(column(a),panelC).Intendedunequaldivisionsaremuchmorelikelyamongparentswithstepchildren(61percentforallparentsandastaggering75percentfordivorcedparents,column(b))thanamongparentswithgeneticchildrenonly(27and29percentrespectively,column(c)).Intheexitles,however,theproportionofestatesthataredividedunequallyissubstantiallygreater(53percent).Thedi erencebetweenparentswithstepchildrenandparentswithgeneticchildrenonlyissmallerthanthatobservedinthecoreles,butisalwaysstatisticallysignicant.Thismayre ectachangeinparents'behaviorbetweenthetimetheyreportedtheirintentionsinthecorelesandtheactualdistributionoftheirestatesreportedintheexitles.Itmayalsobedrivenbyselection(parentsintheexitlesareolder)orre ectthedi erenceinmortalityratesbysocioeconomicstatus(parentsintheexitlesarelesseducatedandlesshealthy,andthesemightbethetypeofparentswhoaremorelikelytodistributeunequallytheirend-of-liferesources).AppendixTableA1mirrorsTable1focusingonno-contactparents.Asinthecaseofparentswithstepchildren,parentswhohavehadnocontactwiththeirgeneticchildreninthepastyeararemuchlesslikelytohaveawill.About57percentofno-contactparentshavenowill,asopposedto42percentofparentswhohaveregularcontactwithalltheirgeneticchildren(panelA).Amongdivorcedno-contactparentstheproportionofthosewithoutawillisnearly70percent.Lookingonlyatparentswithawill,almost85percentofthosewithregularcontactswithalltheirgeneticchildrenreporttheintention 23Theexceptionisinrelationtotheprobabilitythatrespondentsbequeathaspecicamountofmoney,whichisaskedtoeachhouseholdindependentlyofwhetherrespondentshaveawillornot.Attheendofsubsection4.Bwediscusstheresultsfoundusingthatinformation.11 toincludealloftheminthewill.Thisproportionis19percentagepointsgreaterthanthatobservedamongno-contactparents,andthedi erenceishighlysignicant(panelB).Finally,parentsinregularcontactwiththeirchildrenalsoreporttheyaremorelikelytodivideequallytheirestatethanno-contactparents(79and57percentrespectively).Thedi erencebetweenthesetwogroupsofparentsisevengreaterwhenweconsiderdivorcedandwidowed(panelC).Thesepatternsareconsistentwiththepredictionsimpliedbyexchangemodels.C.ExplanatoryVariablesTable2reportssummarystatisticsfortheexplanatoryvariablesweusetomodeltheprob-abilitiesofreportinghavingawillandreportingtheintentiontoleaveunequalbequests.Weshowguresforthesampleofelderlyparentswhoreporthavingwills(column(b))andforthebroadersampleof23,984individualsthatalsoincludesparentswhoreportnothavingwills(column(a)).Thetablealsopresentssummarystatisticsforthesubsam-pleofallparentswhohavebothgeneticchildrenandstepchildren(column(c))andthesubsampleof2,342parentswithstepchildrenwhoreporthavingwills(column(d)).24Ourcovariatesincludestandarddemographiccontrolsforparents'age,sex,race,mar-italstatus,andnumberofmarriages.Thesevariablescapturebasicheterogeneitywithinandacrosshouseholds.Wealsoincludemeasuresofannualfamilyincomeandtotalwealth(bothexpressedin1995prices),parents'education(highestgradecompleted)andemploymentstatus.Weusethesevariablesascontrolsforheterogeneityinparentalre-sources.HRSrespondentsarealsoaskedwhethertheygavemoneytoatleastonechildortoallchildrenequally;intervivostransfersareknowntodependmoredirectlythanbequestsonchildren'scurrentincomesandthustendtobedividedlessevenly(McGarry1999).Oneofourkeyexplanatoryvariablesisanindicatorvariableforthepresenceofatleastonegeneticchildandatleastonestepchild.Thegenetic-childhypothesispredictsthatparentswilltreatgeneticchildrenandstepchildrendi erentlyinallocatingresources.Becauseparents'abilityorwillingnesstomaketransfersmaydependonthetotalnumber 24Forthesakeofbrevity,summarystatisticsoftheindependentvariablesbrokendownbyparentalabsencestatusarenotreported.12 ofchildrenandstepchildren,weincludethesecharacteristicsaswell.Anotherkeyvariableisparentalabsence,proxiedbyparents'lackofcontactwiththeirgeneticchildren.Parentswhoreporthavinghadnocontactwithatleastoneoftheirgeneticchildrenoverthepreviousyeararedenedtobeno-contactparents.Insubsection4.Fwewillalsoconsiderparentswithinfrequentcontact.Thislattergroupcomprisesparentswhoreporthavinghadatleastonecontactwiththeirchildrenoverthelastyear,whileparentswithfrequentcontactsarethosewhohavecontactatleastonceamonth.Exchangemodelspredictthatchildrenwithmoreregularcontactswillbemorelikelytoreceivebequestsifthesechildrenarealsomorelikelytoprovidecareandsupporttotheirneedyelderlyparents(Bernheim,Shleifer,andSummers1985;Cox1987).Weuseparentalhealthstatus(\poororfair"asopposedto\goodorexcellent")asaproxyforaparent'sneedforchildren'sservicesandhence,willingnesstopayforthem(i.e.,theexchangemotive).Usingtheparents'reportsofthechildren'sincomeandwealthwouldsubstantiallyreducethenumberofobservations,soinsteadofdoingsowepredicteachchild'sincomeusingobservedcharacteristics.FollowingLightandMcGarry(2004),wepredictincomesusingestimatedparametersfromincomemodelsthatwettothedatafromtheCurrentPopulationSurvey(CPS)between1994and2006.OursampleconsistsofallCPSrespondentsinthesameagegroupastheparents/childreninourHRSsample.Weestimateseparatemodelsformenandwomenusingasregressorsaconstant,aquarticinage,vedummyvariablesindicatingthehighesteducationalattainment,andindicatorvariablesforrace,maritalstatus,numberofchildren,andhomeownership.Wethenusethispredictedincomevariabletoconstructameasureofincomedi erences,thecoecientofvariation(obtainedbydividingthestandarddeviationofestimatedincomebyitsmean)amongthechildrenofeachindividual.Foreachofourcovariates,weobservedi erencesbetweenparentswithawill,parentswithstepchildren,andparentswithstepchildrenandawill(columns(b){(d)).Perhapsunsurprisingly,themoststrikingdi erencesarebetweenparentsinthecoreles(column(a))andthoseintheexitles(column(e)).Comparedwiththeparentsinthecoreles,parentsintheexitlesarelesseducatedandmorelikelytoreportbeinginpoororfairhealthinthelastwaveinwhichtheyparticipated.Parentsintheexitlesareolder(and13 thusmorelikelytoberetired),andmorelikelytohavebeenwidowed(andhencetohavelowerhouseholdincome).4.EmpiricalResultsWepresentourempiricalresultsinsixsubsections,emphasizingtherolesofstepparentsandparentswithoutcontactwithatleastoneoftheirgeneticchildren.Werstdescribewhohasawillandwhodoesnot(subsectionA).Thenweanalyzewhetherparentsincludeallchildreninthewilland,iftheydo,whethertheyintendtoleaveequalbequeststoallchildren(subsectionB).25Wenextinvestigatehowhealth,wealth,andpreviousintervivostransfersa ecttheprobabilitythatHRSrespondentsintendtotreattheirchildrenunequally(subsectionC).InsubsectionDweexploitthelongitudinalaspectoftheHRSexaminingchangesinbequestintentions,whileinsubsectionEweturnfrombequestintentionstotheactualdivisionofestatesusingtheexitles.Finally,insubsectionFweexaminehowcontactsandinteractionsbetweenparentsandchildrenincomplexfamiliesarerelatedtoend-of-lifetransfers.A.WhoHasaWillandWhoDoesNotAswesawinTable1,thefractionofHRSrespondentswhohavechildrenbutdonothavewillsissubstantial:42percentofthoseinthecorelesreportnothavingawill.Acomplicationwithbequestintentiondataisrightcensoring:someparentswhowilleventuallywritewillshavenotdonesoatthetimetheyrespondtothesurvey.Olderandlesshealthyparents,however,mightbemorelikelythanotherrespondentstowritewills.Somightunpartneredwidowsorwidowers,whoarethelastonestohavetheresponsibilityofpassingonthefamilyestatetofuturegenerations.Toaccountforthesepossibilities,wecontrolforparentalageandhealthstatus,andestimateseparatelytheresponseofwidows,widowersanddivorcedparents.Toassesstheextentoftherightcensoringproblemmoredirectly,weusedatafromtheexitles.26 25Itisworthkeepinginmindthatstepchildrenarecountedaschildren.26AsmentionedinSection3,38percentofparentsintheexitlesdiedintestate.Thisslightlylowerproportionthanthatrecordedinthecorelesmayre ectanactualchangeinparents'behaviororselectiondrivenbydi erentialattritionbasedonage,health,andsocioeconomicstatus.Weleavethisinterestingissueforfutureresearch.14 Table3presentsrandome ectsprobitestimatesoftheimpactofthepresenceofstepchildrenontheprobabilitythatparentsreportnothavingawill.Thetable,basedondatafromthecoreles,showstheresultsfromvespecications.Inspecication(a)weincludeourbasicsetofcontrols(demographics,education,andemploymentstatus)aswellashealthstatusindicators.Specication(b)addsmeasuresofmoneytransferstochildren,whilespecication(c)dropsthesemeasuresbutincludescontrolsforparents'expectedincomeandwealth.Specication(d)includesallpreviousmeasures,whilespecication(e)alsoincludesthecoecientofvariationbetweenchildren'sincome,asanadditionalregressor.Besidestheresultsforallparents,thetablealsoshowsresultsseparatelyforfathersandformothers,andreportsthepooledandseparatee ectsforthesubsamplesofwidowedanddivorcedparents.Althoughthedescriptivestatisticsshowthatindividualswithbothgeneticchildrenandstepchildrenarelesslikelytoreporthavingwillsthanthosewithgeneticchildrenonly,theestimatesinTable3implythatourbasiccontrolvariableslargelyaccountforthisdi erence(column(a)),exceptforwidowsandfordivorcedfathers.Widowsareindeed7percentagepointslesslikelytohaveawill.Thisisconrmedwhenwecontrolforintervivostransfersandparentalincomeandwealth(columns(b){(d)),buttherelationshipbecomesweakerandlosesstatisticalsignicancewhenwefurthercontrolforthecoecientofvariationbetweenchildren'spredictedincome(column(e)).Fordivorcedfatherswithstepchildrenwendthattheprobabilityofhavingawillisnearly8percentagepointsgreaterthanthecorrespondingprobabilityfordivorcedfatherswithgeneticchildrenonly(column(a)).Controllingforintervivostransfersandparentalincomeandwealthleadstoevengreaterestimates,rangingfrom8to12percentagepoints(columns(b)to(d)).Whenwecontrolforwithin-familyincomedi erences,theprobabilityofnothavingawillgoesuptoalmost20percentagepoints.Inthenextsubsectionweexaminewhetherthedi erentialpropensitiestohaveawillre ectparentalpreferencestofavorownbiologicalchildrenorpreferencestoequalizethedistributionofestatesacrossallchildren,includingstepchildren.Table4showshowtheprobabilitythatparentshavenowillisa ectedbycontactbetweenparentsandchildren.Thevespecicationsinthistablearethosepresentedin15 Table3.Amongallparentswithgeneticchildrenonly,thosewhohavenocontactwithoneormoreoftheirchildrenarenearly15percentagepointsmorelikelytohavenowill.Themagnitudeofthisimpactvarieswiththeparent'sgenderandmaritalstatus,butthee ectisgenerallyrobustacrossspecicationsandsubsamples.Lackofcontact,therefore,isassociatedwithasignicantreductionintheprobabilityofhavingawill.B.UnequalIntendedBequestsConditionalonparentshavingawillandhavingmorethanonechild,wenowinvestigatetwoquestions:whetherparentsincludeallchildrenintheirwillsandwhethertheytreatallchildrenequallyintheirwills.Table5presentsrandome ectsprobitestimatesofthestepchildvariableontheproba-bilitythatindividualsincludeallchildrenintheirwills.Thevespecicationsincolumns(a){(e)andtherestoftheorganizationofthetablearethesameasinTable3.Table5revealsthatforparentswithstepchildrenthelikelihoodofincludingallchildreninthewillis28{39percentagepointslowerthanforparentswithoutstepchildren.Thisimpactisstrongerinabsolutevalueformothersthanforfathers,althoughcontrollingforincomedispersionbetweensiblingsreversesthisresult(column(e)).Lookingatspecication(d),motherswithstepchildrenarealmost38percentagepointslesslikelytoreportincludingallchildrenintheirwillswhilefatherswithstepchildrenareabout27percentagepointslesslikelytodoso.Thesenegativee ectsarestrongestfordivorcedparentswithstepchil-dren,withestimatedimpactsrangingbetween59and63percentagepointslowerthanfordivorcedparentswithoutstepchildren.Wealsondstrongnegativee ectsamongwidowsandwidowers.Thismighthappenbecausewidowsandwidowerswritenewwillsafterthedeathofthespouse.InsubsectionD,whereweanalyzetransitions,weshallreturntothispossibility.Werepeatedthesameanalysisbutfocusingonthebehaviorofparentswhohaveonlygeneticchildrenbutwithwhomtheyhavenocontact.TheresultsareinAppendixTableA2.TheseareremarkablysimilartothosereportedinTable5.Inparticular,parentswhohavenocontactwithatleastoneoftheirgeneticchildrenare20percentagepointslesslikelytoincludeallchildreninthewill.Divorcedorwidowedparentsshowevensmaller16 chancesofincludingalloftheirchildreninthewill.No-contactparentsthereforehaveasubstantiallygreaterpropensitytotreattheirgeneticchildrenunequally.Conditionalonhavingawill,werepeatedthesameanalysisforthelikelihoodofhavingawillinwhichallchildrenaretreatedequally.Table6reportstherandome ectsprobitestimatesforparentswithstepchildren,whileAppendixTableA3reportstheestimatesforno-contactparents.Table6indicatesthatthepresenceofstepchildrenisalwaysassociatedwithaconsiderablylowerprobabilityofequalintendedbequestsandthesedi erencesarealwaysstatisticallysignicantatthe1percentlevel.Forexample,havingbothgeneticchildrenandstepchildrenasopposedtohavinggeneticchildrenonlyreducestheprobabilityofawillinwhichallchildrenaretreatedequallyby31percentagepoints(rstrow,column(d)).Thisnegativee ectisgreaterformothersthanforfathers(36versus25percentagepoints,respectively),andthedi erenceisstatisticallysignicantatconventionallevels.But,asinTable5,thisgenderdi erenceisnotrobusttotheinclusionofthecoecientofvariationofchildren'spredictedincome(column(e)).Unpartneredparents(eitherdivorcedorwidowed)aregenerallylesslikelytoplanequalbequestsiftheyhavegeneticchildrenandstepchildren.Forexample,divorcedfathersare52percentagepointslesslikelytotreatallchildrenequallythandivorcedfatherswithgeneticchildrenonly(column(d)).Similarresponsesarefoundfordivorcedmothersaswellasforwidowsandwidowers,andthegenderdi erencesareneverstatisticallysignicant.Withoutexceptions,theestimatesinAppendixTableA3revealthatparentswhohavenocontactwiththeirgeneticchildrenarealsolesslikelytoplananequaldivisionoftheirestates.Parentalabsenceisonaverageassociatedwithareductionintheprobabilityofequalbequestintentionsof25{30percentagepoints.Strongerreductionsareobservedamongallfathersandamongthosewhoaredivorcedorwidowed.27Insum,theestimatesinTables5and6andthoseinAppendixTablesA2andA3tellaconsistentstory.Parentsincomplexfamiliesarelesslikelytomentionalltheir 27TheresultsinTables6andA3werefoundonparentswithtwoormorechildrenandawill.Analternativesampleselectionistoincludeonlyparentswithtwoormorechildrenandawillthatmentionsallofthechildren.Thisselectionclearlyleadstosmallersamples.Butevenwhenthismorerestrictivedenitionisused,wende ectestimatesthatareinlinewith(albeitofsmallermagnitudethan)thoseshowninTables6andA3.TheyarereportedinAppendixTableA4.17 childreninthewill.Amongsuchparents,thosewhodomentionallchildrenaremorelikelytoplananunequaldivisionofend-of-lifetransfers.Thisevidencesuggeststhatstepchildrenandgeneticchildrenwithnocontactwiththeirparentsappeartofacesimilarchancesofinheritingfromtheirstepparentsandparents.Thisresultsitsatoddswiththegenetic-childhypothesis,accordingtowhichparentsfavortheirowngenetico springoverstepchildren,andseemsinsteadtobedrivenmainlybyothermotives.Inthenextsubsectionweexplorethispossibilityfurther.28C.Health,Wealth,andInterVivosTransfersInouranalysisweincludedvariablesthataremeanttoproxyaltruismandexchangeasbequestmotives.Wealsocontrolledforgiftstochildren,sinceparentsmightadjusttheirbequestintentionsiftheymadesubstantialintervivostransfers.Theadjustmentcouldgoineitherdirection.Parentsmightgivelesstochildrenwhoalreadyreceivedintervivostransferstoequalizelifetimetransfers,ortheymightgivemorebecauseprevioustransfersindicategreaterneedorcloserties.Wenowbrie ydiscusstheseresultsthatcanbefoundinthewebappendix.Infamilieswithstepparentsandinfamilieswithno-contactparents,wealthandincomehaveoppositee ectsontheprobabilityofhavingawill.Wealthierparentsaremorelikelytoreporthavingawill.Buthigherincomeparentsarelesslikelytoreporthavingawill,perhapsbecausesuchparentsarelikelytobeyoungerandstillinthelaborforce.Inbothtypesoffamilies,havingmadeintervivostransfersisassociatedwitha5percentagepointincreaseintheprobabilityofreportinghavingawill.Beinginfairorpoorhealth 28Itshouldbeemphasizedthatweperformedseveralrobustnesschecks.Theestimatesfromtheseexercisesbyandlargeupholdthepictureonbequestintentionsemergedsofar.Inoneoftheexercises,wedisaggregatedtheoveralle ectofthestepchild(ornocontact)indicatorvariablebythenumberofgeneticchildrenandstepchildrenanddistinguishedamongparentswithonechildineachcategory,thosewithtwogeneticchildrenandonestepchild,thosewithonegeneticchildandtwostepchildren,andthosewithtwoormorechildrenineachcategory.Virtuallyallourearlierresultsarerobusttothischange.Inanothercheck,wetookadvantageofthefactthat,regardlessofwhetherindividualshaveawill,theHRSasksonerespondentperhouseholdtoreporttheprobabilityofleavingabequestworthatleast$10,000,$100,000,and$500,000,excludinganyinheritancetobelefttothesurvivingpartnerifhe/sheisstillalive.Webandedallanswersintosixgroups,i.e.,wedistinguishedindividualswhoreportazeroprobabilityfromthosewithapositiveprobability,andtheseinturnwerebandedintovequintiles.Usingrandom-e ectsorderedprobitregressions,wethenre-analyzedthemodelsofhavingawill,inclusionofallchildreninthewill,andequalintendedbequests.Again,theresultsfromthisanalysisarequalitativelyverysimilartothosediscussedabove.Theseandthepreviousestimatesarereportedintheonlineappendix.18 isassociatedwitha7percentagepointreductionforstepparents,whiletheestimateddecreaseis9percentagepointsforno-contactparents.Lesshealthyparents,whomightbemoreinneedofcarefromtheirchildren,mayhaveanincentivetohaveawill.Finally,largerdi erencesinchildren'sincomesareassociatedwithalowerprobabilityofhavingawill,althoughthisrelationshipisnotstatisticallysignicant.Wealthierstepparentsandno-contactparentsaremorelikelytoreportincludingallchildrenintheirwillsaswellastheirintentiontodividetheirestatesequallyamongchildren.Forparentswithstepchildren,greaterparentalincome,whichmightbemorecommonamongyounger(stillworking)parents,isassociatedwithareductioninbothprobabilities.Forsuchparents,wealsoestimatethatparentswhohavealreadymadeintervivostransfershavea7percentagepointhigherprobabilityofincludingallchildrenintheirwillsand11percentagepointhigherprobabilityofintendingequalbequests.Forno-contactparentfamilies,thenegativerelationshipwithincomeisweakerandthee ectofintervivostransfersissmaller.Forinstance,havingalreadymadeintervivostransfersincreasestheprobabilityofincludingallchildreninthewillby3percentagepointsandtheprobabilityofequalbequestintentionsby5percentagepoints.Forbothtypesoffamilies,greaterincomedispersionbetweensiblingsisassociatedwithalowerprobabilitythatwillsincludeallchildrenandagreaterprobabilitythatparentsintendtodividetheirestatesunequally.Althoughthesecorrelationsarenotstatisticallysignicant,ifweassumethatparentsintendtogivemoretolow-incomechildren,thentheyareconsistentwithaltruisticbehavior.Parentsinpoorhealthare2to3percentagepointslesslikelytoincludeallchildreninthewilland,conditionalonmentioningallchildren,around4percentagepointslesslikelytoincludeallofthemequally.Ifpoorhealthre ectsparents'long-termneedforchildassistance,theseresultssuggestthatparentsmayusetheirintendedfuturetransfers,whichtheycouldmakeknowntotheirchildren,toelicitalong-term owofservices.ThisisconsistentwiththeexchangemotivepostulatedbyBernheim,Shleifer,andSummers(1985)anddocumentedbyLightandMcGarry(2004).19 D.TransitionsinBequestIntentionsExploitingthelongitudinalaspectoftheHRS,weanalyzechangesinbequestintentions.Inparticular,wefocusonthetransitionsfromnothavingawilltohavingawill,fromhavingawillthatdoesnotincludeallchildrentohavingawillthatincludesallchildren,andfromunequalintendedbequeststoequalintendedbequests.Thepictureemergingfromtheninetransitionsintheotherdirectionsisconsistentwiththatobtainedfromthethreetransitionsjustmentioned.Theyarethereforenotpresented,butareavailableinthewebappendix.Ofcourse,severalothertransitionsmightinteractwithbequestplansandthejointpresenceofgeneticchildrenandstepchildrenorthelackofcontactwithgeneticchildren,suchaschangesinparentalhealthandchangesinchildren'seconomicsituations.Inouranalysisweexplicitlyconsidertheinteractionofthepresenceofstepchildrenorparentalabsencewithchangesinparents'maritalstatus(e.g.,divorceanddeathofaspouse).Table7showstheresultsfromaspecicationsimilartospecication(d)inTables3{6,whichincludeschangesinourbasicsetofcontrols,healthstatus,moneytransferstochildren,andparentalincomeandwealth.Theestimatesofinterestarerobusttotheirexclusion.InpanelAwelookatstepparents,whileinpanelBwefocusonno-contactparents.Foreachtransition,wereporttwosetsofcoecients.Intherstcolumn,wepresenttheimpactofhavingstepchildrenorgeneticchildrenwithwhomtheparenthasnocontactonthetransitionunderstudy.Inthesecondwealsoshowtheinteractiontermsofthestepchildorno-contactparentvariablewithtwochangesinparentalmaritalstatus:frommarriagetodivorceandfrommarriagetowidowhood.Thisisimportantbecause,followingdivorceorthedeathofaspouse,anindividualmightwriteanewwill.Infact,inpanelsCandDweanalyzethispossibilitymoredirectlybyconsideringonlythesubsampleofwidowsandwidowers.Inthiscase,werstanalyzewhetherparentswriteanewwillorchangeanoldwillafterthedeathoftheirpartner(rstcolumn).Inthesecondcolumnwethenpresenttheestimatesofhowthepresenceofstepchildrenorno-contactchildrenchangesbequestintentionsfurther.29PanelAofTable7revealsthatthetransitionfromnothavingawilltohavingawill 29Otherinterestingchangesinmaritalstatus(e.g.,remarriageandrepartnering)cannotbeanalyzedseparatelyduetosamplesizelimitations.20 isa ectedinthesamewaybythejointpresenceofstepchildrenandgeneticchildrenasbythepresenceofgeneticchildrenonly.Thisisalsothecasewhenthepresenceofstepchildrenisinteractedwiththetwochangesinmaritalstatus.Thejointpresenceofgeneticchildrenandstepchildrensignicantlyreducestheprobabilityofatransitiontoawillinwhichallchildrenareincludedaswellasthetransitionfromunequaltoequalbequests.Havingstepchildrenreducestheformertransitionbyalmost9percentagepoints(column(c))andthelatterbyaround8percentagepoints(column(e)).Controllingforchangesinparents'maritalstatusdoesnotsignicantlyaltertheseresults(seecolumns(d)and(f)).Thepresenceofstepchildrenthereforeisnegativelycorrelatednotonlywiththeprob-abilityofequalintendedbequestsbutalsowiththeprobabilityofchangingthewillfromunequaltoequaltreatmentofchildren.Havinglostapartnerthroughdivorceordeathgenerallymakesthistransitionevenlesslikely.PanelBconrmsvirtuallyallthepreviousresultsforthecaseoffamilieswithno-contactparents.Generallytheestimatede ectsarelargerinabsolutevalueamongthesefamiliesthanamongfamilieswithstepparents.Theonlyexceptionsareincolumns(d)and(f)wherethereductionsintheprobabilitythatno-contactchildrenareincludedinthewillandarementionedequallyarenotstatisticallysignicantiftheno-contactparent'spartnerdies.Thiscouldbedrivenbythelowstatisticalpowerofoursmallsamples.Weassessthisissuemoredirectlyinthenexttwopanelsinwhichwefocusonthesubsampleofwidowsandwidowers.Forthissubsamplewedrawattentiontofourinterestingresults.First,theestimatesforwidowedstepparentsaresimilartothoseforno-contactparents.Second,widowhoodinallcomplexfamiliesincreasesthelikelihoodofchangingbequestintentions:itincreasesthetransitiontowritingawill,havingawillinwhichallchildrenarementioned,andhavingawillinwhichallchildrenareequallyincluded(columns(a),(c),and(e)).Third,stepchildrenorno-contactchildrenneitherincreasenordecreaseparents'greaterpropen-sityofwritingawill(column(b)),buto settheirparents'greaterpropensitiesofincludingallchildreninthewillsandincludingthemallequally(columns(d)and(f)).Fourth,thepresenceofstepchildrencombinedwithaspouselossfurtherreducesthelikelihoodof21 movingtoawillwithfullinclusionby13percentagepointsandthelikelihoodofmovingtoawillwithequaldivisionby9percentagepoints(panelC).Suchfurtherreductionsarenotobservedamongno-contactparents(panelD).Thisevidenceisinlinewithwhatwefoundinthetoptwopanels.E.ActualEnd-of-LifeTransfers:EvidencefromtheExitFilesTheHRSexitlesprovidedirectinformationaboutend-of-lifetransfers.TheexitlescontainreportsbythesurvivingspouseorpartnerorbyotherclosefamilymembersofthedeceasedHRSrespondentandtheyallowustoanalyzetheactualdivisionofestatesratherthanbequestintentions.Unlikethecoreles,theexitlesdonotsu erfromrightcensoringbuttheyaremuchsmallerthanthecorelesandtheHRSrespondentswhodieearlyarenotarandomsubsampleofHRSrespondents.Table8givesasummaryofourresults.Foreachoutcomeweshowprobitestimatesfromtwospecications.Therstincludescontrolsforstandarddemographics(column(a)),whilethesecondfurthercontrolsforyear-of-deathxede ects,anindicatorforwhetherthedeathwasexpected,earliertransferstochildren,andparentalwealth(column(b)).30Whenlookingattheprobabilityofintestatesuccession(inthersttwocolumnsofTable8),theestimatedcoecientofthestepchildvariableisalwaysstatisticallyindis-tinguishablefrom0.Thisresult,whichholdstrueacrossallfamilytypesandforbothfathersandmothers,conrmsthegeneralndingsreportedinTable3.Fromtheexitles,however,wedetectneitherthenegativee ectforwidowsnorthepositivee ectfordivorcedfatherswhichwerefoundusingthecoreles.Thismightre ectanactualchangeinparents'behavior.Butitcouldalsobeduetothelowstatisticalpowerimpliedbythesmallsamplesizeoftheexitles.Inthenexttwocolumns,weconsidertheprobabilityofincludingallchildreninthewill.ThisisestimatedonthesubsampleofHRSrespondentswhowroteawill,whilethosewhodiedintestatearenotincluded.Wegenerallyndthathavingstepchildren 30Thetablereportstheestimatesonthestepchildindicatorfoundamongfamilieswithstepparents.Theexitlesdonotallowustoperformthesameanalysisonparentswhodidnothavecontactswiththeirgeneticchildren.22 doesnota ectthelikelihoodthatallchildrenareincluded(column(a)).Thisresultemergesforthefullsampleaswellasforallthesubsamplesofmothersandfathersbymaritalstatus.ItcontradictswhatwefoundinTable5usingthecoreles.Indeed,intwooutoftheninecasesunderspecication(b),weevenndapositiveandsignicante ectofthestepparentindicator.Thisresultmightindicateastrategyaccordingtowhichstepparentsendupincludingalltheirchildren(includingstepchildren)intheirnalwill.Wouldallchildrenbeincludedequally?Thelasttwocolumns,wherewepresenttheestimatesfortheprobabilityofequalbequests,suggestthattheyarenot.31Infact,inlinewiththebequestintentionestimatesofTable6,theresultsinTable8indicatethatthepresenceofstepchildrenreducestheprobabilitythatactualbequestsareequalby9percentagepoints(rstline,specication(a)).Thisimpactisroughlyafourthofthemagnitudeofthecorrespondingimpactonintendedbequests.Asinthecaseofthecoreles,theestimatede ectofstepchildrenonequalactualbequestsisgreaterforfathersthanformothers,althoughtherelativelylargestandarderrorsmakethisdi erentialresponsebygenderstatisticallyindistinguishablefrom0.Wealsocannotdetectdi erentialimpactsforthesubpopulationsofparentswhoexperienceddivorceorthedeathofaspouseorpartner.Buttheprobabilityofequalbequestsisparticularlylowfordivorcedmothersandwidowers,withestimatede ectssimilartothosewefoundfortheintendedbequests.Someofthesee ectslosetheirstatisticalsignicancewhenweestimatespecication(b),butbyandlargethemainpatternofresultsisconrmed.Insum,thepresenceofstepchildrendoesnota ecttheprobabilityofwritingawill.Stepparentsmightexplicitlydecidetorelyonintestacystatuteswhichwouldleavenothingtostepchildren.Forparentswhodiewithawill,thereisevidencethattheyincludeallchildren.Buttheactualbequestsobservedintheexitlesareconsistentwiththebequestintentionsreportedinthecoreles.Unequalbequestsareadistinguishingfeatureofindividualswithbothstep-andgeneticchildrenand,intheactualdivisionofestates,geneticchildrenarefavoredoverstepchildren. 31Asbefore,thisanalysisisbasedonlyonthesubsampleofparentswhodiedwithawill.23 F.InteractionsbetweenParentsandChildrenWenowfocusonhowinteractionsbetweenparentsandchildrenarerelatedtotheprob-abilitythatthechildrenarementionedinthewill.32Werstlookatthestepparent{stepchildinteractionandinvestigatewhetherstepchil-drenarelesslikelytobeincludedinthewillofastepparentwhenthestepparentalsohasgeneticchildren.Table9presentsrandome ectsprobitestimatesoftheprobabilitythatastepchildismentionedinthestepparentwillforthewholesampleofstepchildren.Thetableshowstheresultsfromvespecications.Inspecication(a)weincludeabasicsetofcontrols(i.e.,parentandchild'sageandgender),theageatwhichthechildbecameastepchild,andthenumberofyearsspentwiththestepparent.Specication(b)alsocontrolsforwhetherthestepparentreportedprovidingcareforthestepchild'schild(ren),whilespecication(c)includesanindicatorofwhetherthestepchildwasthemainre-cipientofintervivostransfersfromthestepparent.Specication(d)addsameasureofthestepchild'spredictedincome,whichwasconstructedusingtheproceduredescribedinSection3.Finally,inspecication(e)weaddcontrolsforthewithin-familydi erencebetweenthestepchild'spredictedincomeandthegeneticchild'sincome(ortheincomemeanwhentherearetwoormoregeneticchildren).Wedistinguishtwosubgroups,oneinwhichthestepchild'sincomeismorethan50percentlowerthanthegeneticchild's,andtheotherinwhichthe(negative)di erenceisbetween1and50percent.33Table9revealsthatforachildwhosestepparentalsohasageneticchildtheprobabilityofbeingincludedinthewillis3{4percentagepointslower,anaverageimpactofabout15percent(columns(a){(e)).Thisisconsistentwiththegenetic-childhypothesisaccordingtowhichparentstendtofavorchildrenwhosharetheirgenes.Thisnegativerelationshipisentirelyeliminated,however,ifthestepchild'spredictedincomeislowerthanthegeneticchild's,andismorethanoutbalancediftheincomedi erenceislessthan50percent(specication(e)).Nearlytwo-fthsofstepchildreninthesamplehaverelativelylowerincomes.Thisndingisconsistentwithaltruismifparentsnotonlyaremorelikelyto 32Wecannotdirectlyfocusonthestepchild'sorno-contactchild'sperspectivesbecauseintheHRSwehaveinformationonlyonparentsandstepparents.33Wecheckedthesensitivityofthiscuto usingdi erentpartitions,andfoundresultsthatareessen-tiallyidenticaltothoseshowninTable9.Thesealternativeestimatesarethusnotreported.24 mentionlow-incomestepchildrenintheirwills(aswendhere),butalsogivethemmore.WecannottestthislastpointbecausetheHRSdoesnotcontainanyinformationabouttheamountthattherespondentsintendtobequeathtoeachchild.Acrossallspecications,theolderthestepparentthehigherthelikelihoodthatthestepchildisincludedinthewill:thestepchildpenaltyisfullyo setifthestepparentis10yearsolderthantheaveragestepparent.Thismayindicateagreaterneedforthestepchild'sassistance.Thesendingsmirrortherelationshipbetweenparentsandtheirgeneticchildren.Inparticular,theysuggestthatstepparentsmayusebequeststoelicitalong-term owofreciprocalservicesratherthanepisodicshort-termcare.Itispossible,infact,thatepisodicshort-termcarecouldbe\paidfor"byintervivostransfers.ThisbehaviorisconsistentwiththestrategicuseofbequestspostulatedbyBernheim,ShleiferandSummers(1985).Childrenwhosestepparentsareinpoor/fairhealth(anotherpossibleindicatorofneed)havehowevera5percentagepointdecreaseintheprobabilityofbeingincludedinthewill.Table9alsoshowsthat,regardlessoftheageatwhichachildacquiredthestepparent,themoreyearshe/shespentwiththestepparentthehigherthelikelihoodofbeingin-cludedinthewill:7{9yearsofstepchildhoodcompletelyeliminatethestepchildpenalty.Moreover,astepchild'sprobabilityofinclusioninthewillgoesupbyabout6{8percent-agepointsifthestepparentreportshavingprovidedcareforthestepchild'schild(ren)(specications(b){(e))andbyanother13{14percentagepointsifthestepchildisalsothemainrecipientofintervivostransfers(specications(c){(e)).Thismayre ecttrustandbonding,whicharestrengthenedbyrepeatedinteractionsoverlongertimeperiods.Stratifyingthesamplebygender(seeAppendixTablesA5andA6),wendthatthenegativeassociationbetweentheprobabilityofbeingmentionedinthewillandthepresenceofastep-sibling(i.e.,thevariable`stepparenthasowngeneticchildren')iscon-centratedamongmalestepchildren,whoexperienceareductioninthisprobabilityofabout5{7percentagepoints.Femalestepchildrenexperienceareductionofatmost3percentagepointsbutthisreductionisneverstatisticallysignicant.Forfemalestepchil-drentheprobabilityofbeingmentionedinthewillisalsouna ectedbydi erencesinchildren'spredictedincomes.Thesamepatternofresults,withthesamegenderdi er-25 ences,persistswhenwefocusonthesubsamplesofstepchildrenofdivorcedandwidowedparents.Theseresultsarereportedintheonlineappendix.Wenowturnourattentiontono-contactparent{geneticchildinteractionsandinves-tigatehowcontactsincomplexfamiliesa ectbequestintentions.InTable10wereporthowtheprobabilitythatachildofano-contactparentismentionedintheparent'swillvarieswiththefrequencyofcontactandwiththefactthattheparent'sspouseisnotgeneticallyrelatedtothechild.34Thersttwocolumnsofthetableshowthatparentswhohavenocontactwiththeirgeneticchildrenare21{24percentagepointslesslikelytomentionthemintheirwills.Thelackofparent{childcontactthereforemorethanhalvesthechild'soddsofinclusionintheparent'swill.Havinginfrequentratherthanfrequentcontactsalsoreducesthelikelihoodofinclusionby14{15percentagepoints,but,asindi-catedbythep-valuesatthebottomofthetable,thedi erencebetweenhavinginfrequentcontactsandnothavingthematallisalwaysstatisticallysignicant.Theseresultsdonotchangewhenweaddcontrolsforchild'sincomeandotherparent-childinteractionterms(e.g.,grandchild'scareandintervivostransfers).Noticehoweverthatthe\no-contactgeneticchildpenalty"isfullyo setifthechildisthemainrecipientofintervivostransfers.Thismightre ectearlier(morefrequent)interactions.Similarly,iftheno-contactchildisafemale,herpenaltyisreducedby60{90percent,perhapsre ectingotherservicesexpectedbytheparentbutnotobservedintheHRS.Incolumns(c)and(d)weexplorehowthechild'soddsofbeingmentionedinthewillchangenotonlywhenthechildhasinfrequentornocontactwiththeparentbutalsoifhe/sheisnotthegeneticchildoftheparent'spartner.Wendthat,iftheno-contactchildisalsogeneticallyunrelatedtotheparent'spartner,his/herprobabilityofbeingincludedinthewillisfurtherreducedby8percentagepoints(column(c)).Thismeansthatforno-contactstepchildrentheoddsofinclusionareessentiallyreducedtozero.Asbefore,beingthemainrecipientofintervivostransfersandbeingafemalewillreducethepenaltysubstantiallybutwillnoteliminateitentirely. 34Duetosmallsamplesizewecouldnotestimateaspecicationwhichincludedthewithin-familycoecientofvariationinchildren'sincome.WethuscannotusetheestimatesinTable10toassesstheroleplayedbyaltruism.26 Wealsoconductedthisanalysisseparatelyformaleandfemalechildren,theresultsofwhichareshowninwebappendix.Ingeneral,thee ectsaresimilarforbothmenandwomen,exceptthatfemalechildrenwhoarenotthegeneticchildrenoftheparent'spartnerfaceanadditionalpenaltyonlyiftheyhavehavenocontactswiththeirparents,butnofurtherpenaltyiftheyhaveinfrequentcontactswiththem.Insum,parentsaremorelikelytotransferresourcestochildrenwhosharetheirgenesthantostepchildrenwhoaregeneticallyunrelatedtothem.Butasimpleevolutionarystory,suchasthegenetic-childhypothesis,doesnotseemappropriatetoexplainwhatweobserveinthedataas,forexample,inthecaseofno-contactparents.Furthermore,bequeststostepchildrenarea ectedbyaltruisticandexchangemotives.Parentsarelikelytoleavebequeststotheirlow-incomestepchildrensuggestingaltruisticmotives.Andstepchildrenaremorelikelytobementionedintheirstepparents'willsifthestepparentsareoldersuggestingexchangemotives.Trustandbondingseemtobehighlyrelevanttoowithinallcomplexfamilies:thereisnostepchildpenaltyifstepparentshelpwithchildcareorhavealreadymadeothertransfers.Similarpatternsoccurbetweenparentsandtheirno-contactgeneticchildren.Therelationshipbetweenstepchildrenandparentsincomplexfamiliesrequiresdeeper,moresubtleexplanationsthanthoseprovidedbythegenetic-childhypothesis.5.TheoryHereweprovideasimpleframeworktointerprettheresultsoftheprevioussection.WeborrowourkeyinsightsfromtheworkbyBehrman,PollakandTaubman(1982,1986,1995).Considerafamilycomprisedofoneelderlyparent(e.g.,thesurvivingspouse)andtwochildren,childgwhoishergeneticchildandchildswhoisherstepchild.35Supposetheparenthasaseparablewelfarefunctiondenedoverexpectedbequestsforeachchild,W(Bg;Bs).TheparentsmaximizeWsubjecttothreeconstraints.Therstconstraintisthebudgetconstraintthatappliestoresourcesdevotedtobequests.SuchresourcesaredenotedbyR,whileinvestmentsaredenotedbyH.Assumeparentsfaceaxedpricefor 35Alternatively,childscouldbeseenasthegeneticchildwithwhomtheelderlyparenthasnocontact.27 theirinvestmentsandthepriceisidenticalforbothchildren.Normalizingtheinvestmentpriceto1,wecanwritetheconstraintasHg+HsR.Theothertwoconstrainsarethebequestproductionfunctions,oneforeachchild.Oneoftheargumentsofthesefunctionsisbequestrelatedinvestments,H,whichincludethefullhistoryofeachchild'shealthandhumancapitalinvestmentsaswellasmaterialhelpandmoneytransfersthatparentshavemadeovertheirchildren'slivesandthat,inturn,mightdependonchildattributes,suchasageandsex.Child-specicincomeandfamilycircumstancesarealsodeterminantsofsuchinvestments.Anotherargument,labeledP,comprisesbequest-relevantparentalcharacteristics,suchasage,sex,maritalstatus,health,income,andwealth.Analcategory,labeledT,includesvariablesthatdeterminethetrustrelationshipbetweeneachchildandtheparent.Thistakesaccountofphysicalproximity,contact,andbondingbetweenparentandchild,andissignaledbyandshapedbyrepeatedinteractionsandrelativebargainingpowers.Puttingtogethertheseargumentsleadsustoachild-specicbequestproductionfunctionoftheform,Bk=B(Hk;Pk;Tk),k=g;s.Providedtheconstrainedmaximizationfacedbytheparenthasaninteriorsolution,therst-orderconditionforparentalbequestrelatedinvestmentforchildkis@W @Bk@Bk @Hk�=0;whereistheLagrangemultiplierforthebudgetconstraintthatappliestoresourcesdevotedtobequestrelatedinvestmentsforthetwochildren.Expressingtherst-orderconditionsinratioformyields@W=@Bs @W=@Bg=@Bg=@Hg @Bs=@Hs:Thisexpression,whichdoesnotincludetheunobservable,correspondstothestandardtangencyconditionforaconstrainedmaximum,i.e.,theslopeoftheparentalwelfarefunctionequalstheslopeofthebequestpossibilityfrontierforchildsversuschildg.ThisisillustratedbypointAinFigure6.Inthisgure,weassumeequalconcernforbothchildren,sotheindi erencecurvesaresymmetricalaroundthe45degreelinefromthe28 origin.Wealsoassumeasymmetricbequestpossibilityfrontier.Takentogether,thesetwoassumptionsimplyequalbequestsinequilibrium,Bg=Bs.Becauseintheprevioussectionwefoundstrongevidenceofunequalbequestintentionsthatfavorbiologicalchildrenoverstepchildren,weconsiderchangesintheassumptionsunderlyingFigure6.Astraightforwardchangeistoallowforunequalconcern.ThiscaseisillustratedinFigure7,withparentalpreferencesfavoringthechildgeneticallylinkedtotheparent,thatischildg.Iftheproductionpossibilityfrontierremainssymmetric,aparentwithunequalconcernfavoringthegeneticchildwilldividetheestatesunequally,insuchawaythatBg�Bs.AnotherdeparturefromthebenchmarkofFigure6isthroughachangeintheproduc-tionpossibilityfrontier,whilekeepingtheassumptionofequalconcern.Parentsmighthaveskewedbequestrelatedinvestmentsinfavoroftheirgeneticchildren,orgeneticchil-drenmighthavegreaterbargainingpower(emotionalcapital)towardtheirparentsthanstepchildrendo.PointEinFigure8illustratesthiscase.Clearly,ifparentsalsohaveunequalconcernfavoringtheirgeneticchildren,thenthebequestdivisionbetweenthestepchildandthegeneticchildcouldbeevenmoreunequal,asillustratedbypointE0inthegure.Theresultsinsubsection4.Fsuggestthatthesituationislessunfavorabletostepchil-drenifthereisgreaterbondingandtrustbetweenthemandtheirstepparentsandifstepparentshavegreateraccesstothestepchildren'sfamily-speciccapital.Thisbondingandtrustissignaledandfosteredbythestepparentprovidingcareforthestepchild'schildrenormakingdirecttransferstothestepchild.Whenthishappens,thebequestpos-sibilityfrontiercanshiftandthehouseholdwillreachthenewequilibriumE00inFigure8(undertheassumptionofequalconcern).6.ConclusionInthelastthirtyyears,Americanfamilieshaveexperiencedmassivechanges:aretreatfrommarriage,increaseddivorceandremarriage,andgrowthincohabitationandnon-maritalchildbearing.Stepparentsandno-contactparentsincomplexfamiliesmaybelessmotivatedthanparentsintraditionalfamiliestoprovideresourcestochildrenwithwhom29 theydonotsharetheirgenesorhavenotsharedtheirhomes.Andchildrenincomplexfamiliesmaybelesswillingthanchildrenintraditionalfamiliestoassistdisabledelderlyparents,especiallythosewithwhomtheyhavenogeneticconnectionoronlybrie ysharedahome.Foryoungercohorts,nonmaritalfertilityofcohabitingcoupleswhobreakupandrepartnerwillsubstantiallyincreasetheprevalenceofcomplexfamilies.ThoseintheHRScohortareolder,andtheircomplexfamiliesaregenerallyaby-productofdivorceandremarriage.Theimplicationsforintergenerationaltransfersofthesechangesinfamilystructurearediculttopredictbecausecomplexfamiliescreatedbycohabitationandnonmaritalfertilityinyoungercohortsmaybehavedi erentlyfromthosecreatedbydivorceandremarriageinoldercohorts.Wehaveshownthatbequestsaremuchmoreunequalnowthanintherecentpastandmuchmoreunequalthangenerallyrecognized.Innoncomplexfamilies(i.e.,familieswithneitherstepchildrennorno-contactchildren),equalbequestsarethedominantpattern.Incomplexfamilies,however,wendsubstantialinequalityinbothbequestintentionsandactualbequests.Wecannotassesstherelativeimportanceofgenetictiesandcontactbystudyingnoncomplexfamiliesbecauseinsuchfamiliesallofthechildrenhavethesamegenetictieswiththeirparents.Butthebequestpatternswendincomplexfamiliesimplythatcontacttrumpsgeneticties.Theeconomicsliteratureonend-of-lifetransfersassumesthatindividuals,oratleastelderlyindividuals,makewills.Wendinsteadthatparentsoftenfailtowritewillsand,eitherbydesignordefault,relyonintestacylawtodeterminethedistributionoftheirestates.Forparentswithstepchildren,thee ectofrelyingonintestacylawistoleaveeverythingtogeneticandlegally-adoptedchildrenandnothingtostepchildren.Forno-contactparents,thee ectofrelyingonintestacylawistotreatcontactandno-contactgeneticchildrenequally.Ifparentsunderstandthemostbasicprovisionsofintestacylaw,thisndingispuzzling.Itimpliesthatparentswhohavehadnocontactwithsomeoftheirgeneticchildrenaremorelikelytotreatalloftheirgeneticchildrenequallythanareparentswhohavemaintainedcontactwithalloftheirgeneticchildren.Wesuspectthattheabsenceofwillsre ectsthedisutilityofmakingwills(andcontemplatingdeath)30 ratherthanpreferencesforthedistributionmandatedbyintestacylaw.Unfortunately,HRSprovidesnoevidencethatspeakstothisissue.Regardlessofparents'motivationsfornotmakingwills,wehaveshownthatfocusingexclusivelyonbequestintentions(i.e.,onparentswhomakewills)providesanincompleteandmisleadingpictureofend-of-lifetransfers.31 Figure1:UnequalBequestIntentions,byParent'sGender Figure2:UnequalBequestIntentions,byParent'sBirthCohort32 Figure3:UnequalBequestIntentions,byParent'sMaritalStatus Figure4:UnequalBequestIntentions,bytheFamilyType33 Figure5:TrendsinIntestacyRates,byParent'sAgeandFamilyType Bi Bj Bi A Bj Parentalutilityfunction Bequestpossibilityfrontier Figure6:EqualConcernandSymmetricBequestPossibilityFrontier34 Bi Bj Bi A0 Bj Figure7:UnequalConcernFavoringChildjandSymmetricBequestPossibilityFrontier Bi Bj E E0 E00 Equalconcerns Figure8:AsymmetricBequestPossibilityFrontiersWithandWithoutEqualConcern35 Table1:DescriptiveStatistics{DependentVariables (a)(b)(c)(d) ParentswithParentswithstep-andbiologicalDi erenceAllbiologicalchildren(b)-(c)childrenonly(t-value)Nn A.NowillCorelesAll0.4180.4750.4060.069***117,18923,984(18.152)Divorced0.6230.6630.6170.046***13,7033,754(3.784)Widowed0.3760.3950.3730.022**27,3087,679(2.580)ExitlesAll0.3750.4010.3710.030*7,4167,416(1.957)Divorced0.5700.5870.5670.020702702(0.356)Widowed0.3480.3700.3460.0243,2203,220(0.928)B.WillincludesallchildrenCorelesAll0.7460.4620.805-0.343***60,99414,275(76.840)Divorced0.7340.2930.808-0.515***4,3631,456(29.636)Widowed0.7610.3580.834-0.476***14,5974,583(52.965)ExitlesAll0.5870.6180.5810.037*4,5674,567(1.816)Divorced0.8010.7740.806-0.033337337(0.547)Widowed0.8560.8760.8530.0221,9561,956(0.878)C.EqualintendedandactualbequestCorelesAll0.6760.3900.735-0.345***60,99414,275(71.539)Divorced0.6410.2450.708-0.462***4,3631,456(23.747)Widowed0.6790.2910.749-0.459***14,5974,583(45.638)ExitlesAll0.4730.4210.484-0.063***3,8783,878(2.998)Divorced0.6040.4530.639-0.186**280280(2.511)Widowed0.7120.6080.729-0.120***1,6101,610(3.656) Note:N=numberofobservations;n=numberofindividuals.*p0:10,**p0:05,***p0:01.36 Table2:DescriptiveStatistics{ExplanatoryVariables ParentswithParentsParentsAllParentsstep-andwithstep-includedwithbiologicalchildrenintheawillchildrenandawillexitlesa (a)(b)(c)(d)(e) 1ifhasstep-andbiologicalchildren0.2230.1961.0001.0000.136DemographicsFemale0.5640.5570.5590.5730.529Age68.5970.7665.6468.6578.17(10.59)(10.32)(10.81)(10.66)(10.39)1ifwhite0.8290.9270.7970.9100.8331ifmarriedorpartnered0.6440.6720.6920.6990.4761ifseparated,divorcedornevermarried0.1230.0780.1230.0820.0961ifwidowed0.2330.2500.1850.2180.429Numberofmarriages1.3551.3121.9201.9511.347(0.668)(0.612)(0.815)(0.767)(0.664)Health1ifinpoor/fairhealth0.2970.2460.2860.2420.622Education1ifbelowhighschool0.3680.3170.3500.3140.4231ifhighschool0.3410.3500.3380.3490.3061ifcollegeormore0.2920.3330.3120.3370.271Employment1ifinthelaborforce0.4120.3660.4590.3920.0681ifdisabled0.0820.0460.0990.0560.1541ifretired0.5060.5880.4430.5510.778ChildvariablesNumberofchildren3.5223.2745.0044.8523.464(2.092)(1.860)(2.436)(2.272)(2.219)Numberofbiochildren3.0922.8893.0752.8873.129(1.791)(1.551)(1.878)(1.691)(1.993)Numberofstepchildren0.4300.3841.9291.9650.335(1.149)(1.077)(1.741)(1.682)(1.102)Coecientofwithin-familyvariation0.4110.3980.4490.4330.414forchildren'spredictedincomeb(0.356)(0.265)(0.382)(0.217)(0.230) (cont.)37 (cont.) (a)(b)(c)(d)(e) FinancialvariablesRealannual26,94028,27129,78832,4927,630householdincomec(205,093)(157,141)(44,663)(56,093)(18,500)Realwealthd267,702387,954231,881351,730238,948(1,268,856)(1,543,720)(1,084,812)(1,370,953)(1,521,039)1ifgavemoneytoatleastachilde0.3620.4110.3790.4240.2551ifgavemoneytoallchildrenequallye0.0790.1010.0320.4150.081N117,18968,24826,11113,3487,071n23,98412,7635,0822,3427,071 Note:Figuresaremeansandstandarddeviations(forcontinuousvariablesonly)areinparentheses.N=numberofobservations;n=numberofindividuals.aValuesarefromthelastyearofobservationinthecoreles.Forsomeofthevariables,Nandnaredi erentfromthevaluesgivenatthebottomofthetable.Theyareavailablefromtheauthors.bBasedon60,055observationsfrom8,776individuals.cIn1995values,andbasedon74,099observationsfrom13,323respondents.dIn1995values,andbasedon52,078observationsfrom10,830respondents;includesvaluesofnancialandrealestateproperties.eBasedon78,773observationsfrom15,493respondents.38 Table3:E ectofHavingStepchildrenontheProbabilityofNotHavingaWill Specication (a)(b)(c)(d)(e) AllparentsEstimate0.0060.003-0.017-0.014-0.028(s.e.)(0.020)(0.024)(0.020)(0.023)(0.030)N117,18978,77367,61551,76132,199n23,98421,17816,19815,0419,285MothersEstimate-0.009-0.017-0.051*-0.045-0.037(s.e.)(0.029)(0.033)(0.030)(0.032)(0.036)N66,09852,41936,76333,00125,084n13,49812,6629,0558,8246,854FathersEstimate0.0240.0320.0110.0190.011(s.e.)(0.026)(0.033)(0.025)(0.033)(0.056)N51,09126,35430,85218,7607,115n10,5098,5307,1526,2252,435WidowsandwidowersEstimate-0.049-0.052-0.026-0.028-0.007(s.e.)(0.032)(0.032)(0.020)(0.021)(0.024)N27,30826,98621,13520,97216,104n7,6797,6386,6626,6375,248WidowsEstimate-0.074**-0.069**-0.043**-0.041**-0.019(s.e.)(0.033)(0.034)(0.019)(0.020)(0.024)N22,41622,19617,17917,07313,152n5,9605,9355,1585,1424,085WidowersEstimate0.019-0.0070.0460.0220.044(s.e.)(0.072)(0.071)(0.066)(0.063)(0.076)N4,8924,7903,9563,8992,952n1,7251,7091,5091,5001,166DivorcedparentsEstimate0.0330.0480.106**0.105**0.116**(s.e.)(0.031)(0.034)(0.048)(0.051)(0.056)N13,70313,14010,0879,8537,006n3,7543,6533,0342,9992,118DivorcedmothersEstimate-0.0190.0200.1070.097-0.050(s.e.)(0.054)(0.051)(0.073)(0.078)(0.138)N8,9168,6906,4966,4254,715n2,3462,3111,8781,8721,372DivorcedfathersEstimate0.076*0.082*0.118*0.121*0.195***(s.e.)(0.044)(0.048)(0.062)(0.065)(0.060)N4,7874,4503,5913,4282,291n1,4111,3441,1561,127746 Note:Theguresaremarginale ectsoftheparentwithstepchildrenindicatorfromrandome ectspro-bitmodels.Thecomparisongroupisgivenbyparentswithbiologicalchildrenonly.Seethetextforanexplanationofspecications(a){(e).N=numberofobservations;n=numberofindividuals.p0:10,p0:05,p0:01.39 Table4:E ectofNoContactsontheProbabilityofNotHavingaWill Specication (a)(b)(c)(d)(e) AllparentsEstimate0.145***0.155***0.148***0.152***0.148***(s.e.)(0.033)(0.032)(0.036)(0.036)(0.039)N41,07534,12724,43722,45018,118n12,73911,8688,7678,5296,927MothersEstimate0.120***0.128***0.126***0.137***0.144***(s.e.)(0.039)(0.038)(0.043)(0.044)(0.047)N33,84427,89519,19617,62814,381n9,8479,2596,5576,4015,246FathersEstimate0.213***0.223***0.169***0.167***0.137**(s.e.)(0.060)(0.058)(0.061)(0.062)(0.069)N7,2316,2325,2414,8223,737n2,9002,6152,2112,1291,682WidowsandwidowersEstimate0.196***0.189***0.087***0.084***0.075**(s.e.)(0.044)(0.044)(0.032)(0.032)(0.033)N14,38414,29111,22411,1649,198n5,3525,3314,6534,6343,804WidowsEstimate0.190***0.183***0.085**0.082**0.076**(s.e.)(0.051)(0.051)(0.037)(0.037)(0.038)N11,88711,8159,1679,1217,514n4,2374,2203,6673,6522,992WidowersEstimate0.203**0.186**0.0880.0820.059(s.e.)(0.083)(0.080)(0.065)(0.063)(0.064)N2,4972,4762,0572,0431,684n1,1181,114987983813DivorcedparentsEstimate0.073***0.075***0.122***0.127***0.124**(s.e.)(0.027)(0.028)(0.045)(0.044)(0.050)N7,8977,8175,8935,8474,391n2,7922,7762,2802,2641,699DivorcedmothersEstimate0.04880.0525*0.123**0.130**0.141**(s.e.)(0.032)(0.032)(0.056)(0.055)(0.058)N5,4835,4414,0264,0053,073n1,8511,8481,5041,4991,150DivorcedfathersEstimate0.131**0.137**0.1200.130*0.103(s.e.)(0.054)(0.055)(0.074)(0.074)(0.088)N2,4142,3761,8671,8421,318n943930776765549 Note:Theguresaremarginale ectsoftheno-contactparentindicatorfromrandome ectsprobitmodels.SeethenotetoTable3forfurtherexplanations.N=numberofobservations;n=numberofindividuals.p0:10,p0:05,p0:01.40 Table5:E ectofHavingStepchildrenontheProbabilitythattheWillIncludesAllChildren Specication (a)(b)(c)(d)(e) AllparentsEstimate-0.279***-0.311***-0.305***-0.328***-0.385***(s.e.)(0.017)(0.020)(0.021)(0.024)(0.029)N60,99439,76137,44628,03119,435n14,27512,3289,7628,9566,168MothersEstimate-0.360***-0.359***-0.391***-0.377***-0.365***(s.e.)(0.022)(0.026)(0.030)(0.032)(0.034)N33,52925,86219,63917,53515,302n7,8277,2435,2675,1124,589FathersEstimate-0.190***-0.251***-0.215***-0.266***-0.435***(s.e.)(0.025)(0.032)(0.031)(0.038)(0.057)N27,46513,89917,80710,4964,133n6,4605,0914,5003,8481,581WidowsandwidowersEstimate-0.421***-0.398***-0.393***-0.374***-0.377***(s.e.)(0.040)(0.041)(0.042)(0.043)(0.045)N14,59714,40212,62012,51411,075n4,5834,5544,2074,1873,817WidowsEstimate-0.444***-0.433***-0.419***-0.408***-0.410***(s.e.)(0.044)(0.045)(0.046)(0.047)(0.048)N11,91111,77610,24710,1789,064n3,5483,5293,2473,2352,975WidowersEstimate-0.343***-0.258***-0.288***-0.225**-0.219**(s.e.)(0.101)(0.010)(0.104)(0.101)(0.107)N2,6862,6262,3732,3362,011n1,0391,029963955843DivorcedparentsEstimate-0.604***-0.629***-0.605***-0.630***-0.589***(s.e.)(0.068)(0.073)(0.078)(0.080)(0.097)N4,3634,1723,6643,5732,960n1,4561,4131,2571,237994DivorcedmothersEstimate-0.495***-0.477***-0.467***-0.469***-0.462***(s.e.)(0.117)(0.132)(0.147)(0.154)(0.166)N2,8182,7412,3852,3562,042n877863763759645DivorcedfathersEstimate-0.633***-0.678***-0.623***-0.660***-0.581***(s.e.)(0.086)(0.088)(0.095)(0.097)(0.137)N1,5451,4311,2791,217918n580550494478349 Note:SeethenotetoTable3.p0:10,p0:05,p0:01.41 Table6:E ectofHavingStepchildrenontheProbabilitythatStepparentsIntendtoDividetheirEstateEquallyAmongAllChildren Specication (a)(b)(c)(d)(e) AllparentsEstimate-0.285***-0.310***-0.288***-0.311***-0.375***(s.e.)(0.016)(0.020)(0.021)(0.024)(0.028)N60,99439,76137,44628,03119,435n14,27512,3289,7628,9566,168MothersEstimate-0.371***-0.365***-0.376***-0.358***-0.347***(s.e.)(0.021)(0.025)(0.028)(0.031)(0.033)N33,52925,86219,63917,53515,302n7,8277,2435,2675,1124,589FathersEstimate-0.185***-0.233***-0.195***-0.246***-0.427***(s.e.)(0.025)(0.032)(0.031)(0.038)(0.051)N27,46513,89917,80710,4964,133n6,4605,0914,5003,8481,581WidowsandwidowersEstimate-0.404***-0.386***-0.382***-0.362***-0.357***(s.e.)(0.038)(0.039)(0.041)(0.041)(0.043)N14,59714,40212,62012,51411,075n4,5834,5544,2074,1873,817WidowsEstimate-0.422***-0.413***-0.404***-0.392***-0.386***(s.e.)(0.041)(0.042)(0.044)(0.045)(0.047)N11,91111,77610,24710,1789,064n3,5483,5293,2473,2352,975WidowersEstimate-0.325***-0.265***-0.275***-0.225**-0.206*(s.e.)(0.096)(0.100)(0.102)(0.103)(0.111)N2,6862,6262,3732,3362,011n1,0391,029963955843DivorcedparentsEstimate-0.468***-0.502***-0.478***-0.497***-0.469***(s.e.)(0.060)(0.064)(0.069)(0.071)(0.085)N4,3634,1723,6643,5732,960n1,4561,4131,2571,237994DivorcedmothersEstimate-0.368***-0.370***-0.324**-0.303**-0.284*(s.e.)(0.108)(0.121)(0.132)(0.140)(0.152)N2,8182,7412,3852,3562,042n877863763759645DivorcedfathersEstimate-0.468***-0.510***-0.493***-0.523***-0.524***(s.e.)(0.077)(0.079)(0.084)(0.085)(0.114)N1,5451,4311,2791,217918n580550494478349 Note:SeethenotetoTable3.p0:10,p0:05,p0:01.42 Table7:ChangesinParents'BequestIntentions Transition From`notallchildrenFrom`notallchildreninthewill'to`allFrom`nowill'inthewill'to`allchildrenequallyto`will'childreninthewill'inthewill' (a)(b)(c)(d)(e)(f) PanelA:ParentswithStepchildrenParenthasstepchildren0.0010.016-0.085***-0.083***-0.078***-0.058**(0.009)(0.017)(0.016)(0.027)(0.015)(0.025)Parenthasstepchildren0.012-0.238***-0.169***frommarriedtodivorced(0.039)(0.032)(0.048)Parenthasstepchildren-0.012-0.150***-0.113***frommarriedtowidowed(0.022)(0.029)(0.029)N22,4558,1468,7594,24311,4705,279n7,9454,0764,2122,5225,2903,089PanelB:No-contactParentsNo-contactparent-0.007-0.013-0.114***-0.124**-0.112***-0.127***(0.007)(0.023)(0.019)(0.054)(0.016)(0.045)No-contactparent0.169-0.311***-0.205*frommarriedtodivorced(0.118)(0.076)(0.111)No-contactparent0.090-0.063-0.089frommarriedtowidowed(0.060)(0.098)(0.083)N10,3922,4022,5538673,7171,136n4,5291,4551,6046472,211840PanelC:WidowedParentsandParentswithStepchildrenWidowedparent0.099***0.114***0.083***0.147***0.050**0.085***(0.016)(0.020)(0.024)(0.032)(0.020)(0.025)Parenthasstepchildren0.039*-0.114***-0.091***(0.022)(0.035)(0.031)Widowedparent-0.032-0.127***-0.092***Parenthasstepchildren(0.021)(0.035)(0.033)N4,1914,1912,3362,3362,9172,917n2,1672,1671,4481,4481,7761,776PanelD:WidowedParentsandNo-contactParentsWidowedparent0.112***0.101***0.168***0.199***0.092***0.128***(0.020)(0.025)(0.035)(0.049)(0.029)(0.041)No-contactparent0.001-0.141***-0.133***(0.025)(0.054)(0.046)Widowedparent0.070-0.032-0.075no-contactparent(0.058)(0.103)(0.087)N2,8481,9791,2467861,7041,043n1,5071,1838265891,111774 Note:SeethenotestoTables3.Allguresarecomputedonthecoreles.p0:10,p0:05,p0:01.43 Table8:E ectofHavingStepchildrenonActualBequestDecisionsfromtheExitFiles Probabilityof:NowillInclusionofEqualallchildrendivisioninthewill (a)(b)(a)(b)(a)(b) AllparentsEstimate-0.0030.007-0.0090.075**-0.093***-0.043(s.e.)(0.023)(0.024)(0.031)(0.033)(0.031)(0.035)N7,6747,2213,9523,7533,9523,753MothersEstimate-0.015-0.0200.0030.033-0.045-0.029(s.e.)(0.038)(0.040)(0.050)(0.049)(0.053)(0.056)N4,0523,8331,9511,8611,9511,861FathersEstimate-0.0130.0220.0310.102**-0.077**-0.043(s.e.)(0.029)(0.030)(0.039)(0.044)(0.036)(0.041)N3,6223,3852,0011,8902,0011,890WidowedparentsEstimate0.0220.0190.0480.032-0.078-0.094*(s.e.)(0.039)(0.041)(0.032)(0.036)(0.053)(0.057)N3,2333,1341,5971,5481,5971,548WidowsEstimate0.0330.0180.0600.0540.0040.001(s.e.)(0.050)(0.052)(0.046)(0.046)(0.067)(0.068)N2,4282,3501,1691,1321,1691,132WidowersEstimate-0.035-0.002-0.031-0.068-0.265***-0.352***(s.e.)(0.067)(0.070)(0.056)(0.070)(0.091)(0.101)N805783428415428415DivorcedparentsEstimate0.0220.071-0.071-0.039-0.228-0.216(s.e.)(0.086)(0.097)(0.123)(0.121)(0.139)(0.145)N715684281273281273DivorcedmothersEstimate0.1950.249-0.138-0.137-0.418*-0.473**(s.e.)(0.164)(0.176)(0.231)(0.222)(0.222)(0.211)N387376144142144142DivorcedfathersEstimate-0.0320.018-0.0240.049-0.168-0.102(s.e.)(0.101)(0.115)(0.155)(0.153)(0.178)(0.191)N328307137131137131 Note:Seethetextforanexplanationofspecications(a){(b).p0:10,p0:05,p0:01.44 Table9:ProbabilitythataStepchildisExplicitlyMentionedinHis/HerStepparent'sWill (a)(b)(c)(d)(e) Meanofdependentvariable0.2540.2540.2390.2620.261Stepparenthasown-0.042***-0.042***-0.031**-0.035**-0.035***biologicalchildren(0.014)(0.014)(0.014)(0.017)(0.017)Yearsspentwithstepparent0.005***0.005***0.004***0.005***0.005***(0.001)(0.001)(0.001)(0.001)(0.001)Ageatmarriage:0-6a-0.006-0.0060.040*0.067**0.073**(0.016)(0.016)(0.023)(0.031)(0.031)Ageatmarriage:7-12a-0.015-0.015-0.0010.0140.018(0.013)(0.013)(0.017)(0.022)(0.022)Ageatmarriage:13-18a-0.004-0.0040.0100.0100.008(0.010)(0.010)(0.013)(0.016)(0.016)Stepchildisfemale0.0080.0080.0120.074***0.063***(0.006)(0.006)(0.007)(0.011)(0.012)Ageofstepchild-0.001-0.0010.000-0.001-0.000(0.001)(0.001)(0.001)(0.001)(0.001)Stepparentisfemale0.0010.0030.0160.002-0.003(0.010)(0.010)(0.011)(0.014)(0.014)Stepparentage0.004***0.004***0.003***0.003***0.002***(0.001)(0.001)(0.001)(0.001)(0.001)Stepparentisin-0.045***-0.045***-0.057***-0.059***-0.056***poor/fairhealth(0.007)(0.007)(0.008)(0.010)(0.010)Stepparenttakescare0.064***0.082***0.078**0.079**ofstepchild'schild(ren)(0.019)(0.029)(0.032)(0.032)Stepchildismainrecipient0.143***0.128***0.127***ofintervivostransfers(0.021)(0.024)(0.024)Logofstepchild'sincome0.103***0.128***(0.011)(0.013)Stepchild'spredictedincomeisbelowbiologicalchildren'sincomeby:1{49percent0.061***(0.018)50+percent0.037***(0.014)N26,98326,98313,90411,18711,106n13,28813,2888,7627,3057,268 Note:Figuresaremarginale ectsfromprobitregressions.Standarderrorsareinparentheses.N=numberofobservations;n=numberofstepchildren.aTheagereportedherereferstotheageofthestepchildatthetimeinwhichhis/herparentformedapartnershipwithhis/herstepparent(whowritesthewill).Thereferencecategoryis18yearsormore.p0:10,p0:05,p0:01.45 Table10:ProbabilitythataGeneticChildisExplicitlyMentionedinHis/HerNo-contactorInfrequent-contactParent'sWill (a)(b)(c)(d) Meanofdependentvariable0.4280.4100.4310.431Frequencyofcontacts(base=frequent)Infrequent-0.154***-0.140***-0.169***-0.131**(0.007)(0.009)(0.045)(0.053)Nocontact-0.236***-0.213***-0.312***-0.271***(0.006)(0.009)(0.018)(0.037)Parentisfemale0.033***0.016*0.034*0.031*(0.008)(0.010)(0.017)(0.018)Parentage0.020***0.018***0.020***0.020***(0.000)(0.001)(0.001)(0.001)Childisfemale0.0040.127***0.100***0.100***(0.005)(0.007)(0.011)(0.011)Childage0.000-0.001*-0.001-0.001(0.000)(0.001)(0.001)(0.001)Parentismarried0.021***-0.002-0.132***-0.129***(0.006)(0.007)(0.019)(0.019)Parentmarriedmorethanonce-0.091***-0.065***-0.065***-0.066***(0.007)(0.008)(0.013)(0.013)Parenttakescareofchild'schildren0.002-0.0070.000(0.009)(0.013)(0.013)Childismainrecipientofintervivostransfers0.221***0.189***0.189***(0.010)(0.014)(0.014)Logofchild'sincome0.205***0.186***0.186***(0.007)(0.010)(0.010)Childisnotspouse'sgeneticchild-0.084***-0.065***(0.015)(0.016)Childisnotspouse'sgeneticchildinfrequentcontacts-0.122***(0.044)Childisnotspouse'sgeneticchildnocontacts-0.263***(0.031)N157,17394,00339,30039,300n73,52254,42326,90226,902p-valuea0.00000.00000.00180.0375 Note:Figuresaremarginale ectsandstandarderrorsareinparentheses.aReferstothep-valueofthetestofequalitybetweenthecoecientsof\Infrequentcontacts"and\Nocontacts".N=numberofobservations;n=numberofchildren.p0:10,p0:05,p0:01.46 ReferencesAmericanBarAssociation.2013.GuidetoWillsandEstates,4thedition.NewYork:RandomHouse.Andreoni,James.1989.\GivingwithImpureAltruism:ApplicationstoCharityandRicardianEquivalence."JournalofPoliticalEconomy,97(6):1447{1458.Barro,RobertJ.1974.\AreGovernmentBondsNetWealth?"JournalofPoliticalEconomy,82(6):1095{1117.Becker,GaryS.1974.\ATheoryofSocialInteractions."JournalofPoliticalEconomy,82(6):1063{1093.Becker,GaryS.andNigelTomes.1979."AnEquilibriumTheoryoftheDistributionofIncomesandIntergenerationalMobility."JournalofPoliticalEconomy,87(6):1053-1189.Behrman,JereR.1997.\IntrahouseholdDistributionandtheFamily."InHandbookofPopulationandFamilyEconomics,editedbyMarkR.RosenzweigandOdedStark,vol.1A.Amsterdam:North-Holland,pp.125{187.Behrman,JereR.,RobertA.Pollak,andPaulTaubman.1982.\ParentalPreferencesandProvisionforProgeny."JournalofPoliticalEconomy,90(2):52{73.Behrman,JereR.,RobertA.Pollak,andPaulTaubman.1986.\DoParentsFavorBoys?"InternationalEconomicReview,27(1):31{52.Behrman,JereR.,RobertA.Pollak,andPaulTaubman.1995.FromParenttoChild:IntrahouseholdAllocationsandIntergenerationalRelationsintheUnitedStates.Chicago:UniversityofChicagoPress.Behrman,JereR.,andMarkR.Rosenzweig,2004.\ParentalAllocationstoChildren:NewEvidenceonBequestDi erencesamongSiblings."ReviewofEconomicsandStatis-tics,86(2):637{640.BehrmanJereR.,andPaulTaubman.1986.\BirthOrder,SchoolingandEarnings."JournalofLaborEconomics,4(3,pt.2):S121{S145.Bernheim,B.Douglas,andSergeiSeverinov.2003.\BequestsasSignals:AnExplana-tionfortheEqualDivisionPuzzle."JournalofPoliticalEconomy,111(4):733{764.Bernheim,B.Douglas,AndreiShleifer,andLawrenceH.Summers.1985.\TheStrategicBequestMotive."JournalofPoliticalEconomy,93(6):1045{1076.Blattmachr,JonathanG.2008.\ReducingTrustandEstateLitigationthroughDisclo-sure,InTerroremClauses,MediationandArbitration."CardozoJournalofCon ictResolution,9(2):237{266.Brashier,RalphC.2004.InheritanceLawandtheEvolvingFamily.Philadelphia,PA:TempleUniversityPress.Brown,Meta.2006.\InformalCareandtheDivisionofEnd-of-LifeTransfers."JournalofHumanResources,41(1):191{219.Bumpass,Larry,andHsien-HenLu.2000.\TrendsinCohabitationandImplicationsforChildren'sFamilyContextsintheUnitedStates."PopulationStudies,54(1):29{41.Cahn,NaomiR.2005.\Adoptees,Families,StepfamiliesandInheritance."AdoptionQuarterly,8(2):57{66.47 Chu,C.Y.Cyrus1991.\Primogeniture."JournalofPoliticalEconomy,99(1):78{99.Collins,DennisW.2000.\AvoidingaWillContest:TheImpossibleDream?"CreightonLawReview,34(1):7{45.Cox,Donald.1987.\MotivesforPrivateIncomeTransfer."JournalofPoliticalEconomy,95(3):508{546.Cox,Donald.2003.\PrivateTransferswithintheFamily:Mothers,Fathers,Sons,andDaughters."InDeathandDollars:TheRoleofGiftsandBequestsinAmerica,editedbyAliciaH.MunnellandAnnikaSunden.Washington,D.C.:BrookingsInstitutionPress,pp.168-197.Cox,Donald.2007.\BiologicalBasicsandtheEconomicsoftheFamily."JournalofEconomicPerspectives,21(2):91{108.Cox,Donald,andMarkR.Rank.1992.\Inter-VivosTransfersandIntergenerationalExchange."ReviewofEconomicsandStatistics,74(2):305{314.Cremer,TerinBarbas.2011.\ReformingIntestateInheritanceforStepchildrenandStepparents."CardozoJournalofLawandGender,18(1):89{108.Cutler,DavidM.,FabianLange,EllenMeara,SethRichards-Shubik,andChristopherJ.Ruhm.2011.\RisingEducationalGradientsinMortality:TheRoleofBehavioralRiskFactors."JournalofHealthEconomics,30(6):1174{1187.Dukeminier,JesseandRobertH.Sitko .2013.Wills,Trusts,andEstates.NewYork:WoltersKluwer,9thed.Fried,MartinL.1992.\TheUniformProbateCode:IntestateSuccessionandRelatedMatters."AlbanyLawReview,55(4):927{40.Friedman,LawrenceM.2009.DeadHands:ASocialHistoryofWills,Trusts,andInher-itanceLaw.Stanford,CA:StanfordLawBooks.Gary,SusanN.2000.\AdaptingIntestacyLawstoChangingFamilies."LawandIn-equality,18(1):1{82.Grossman,JoannaL.,andLawrenceM.Friedman.2011.InsidetheCastle:LawandFamilyin20thCenturyAmerica.Princeton:PrincetonUniversityPress.Hanson,ThomasM.1995.\IntestateSuccessionforStepchildren:CaliforniaLeadstheWay,butHasItGoneFarEnough?"HastingsLawJournal,47(1):257{286.Laferrere,Anne,andFrancois-CharlesWol .2006.\MicroeconomicModelsofFamilyTransfers."InHandbookoftheEconomicsofGiving,AltruismandReciprocity,editedbySerge-ChristopheKolmandJeanMercierYthier.Amsterdam:North-Holland,pp.889{969.Laitner,John.1997.\IntergenerationalandInterhouseholdEconomicLinks."InHand-bookofPopulationandFamilyEconomics,editedbyMarkR.RosenzweigandOdedStark,vol.1A.Amsterdam:North-Holland,pp.189{238.LaitnerJohn,andHenryOhlsson.2001.\BequestMotives:AComparisonofSwedenandtheUnitedStates."JournalofPublicEconomics79(1):205{236.Light,Audrey,andKathleenMcGarry.2004.\WhyParentsPlayFavorites:ExplanationsforUnequalBequests."AmericanEconomicReview,94(5):1669{1681.Lundberg,Shelly,andRobertA.Pollak.2014.\CohabitationandtheUnevenRetreat48 fromMarriageintheU.S.,1950{2010."InHumanCapitalinHistory:TheAmericanRecord,editedbyLeahP.Boustan,CarolaFrydman,andRobertA.Margo.Chicago:UniversityofChicagoPress,pp.241{272.Mahoney,MargaretM.1989.\StepfamiliesintheLawofIntestateSuccessionandWills."U.C.DavisLawReview,22(3):917{950.Menchik,Paul.1980.\Primogeniture,EqualSharing,andtheU.S.DistributionofWealth."QuarterlyJournalofEconomics,94(1):299{316.Menchik,Paul.1988.\UnequalEstateDivision:IsitAltruism,ReverseBequests,orSimplyNoise?"InModelingtheAccumulationandtheDistributionofWealth,editedbyDenisKesslerandAndreMasson.NewYork:OxfordUniversityPress,pp.105{116.McGarry,Kathleen.1999.\InterVivosTransfersandIntendedBequests."JournalofPublicEconomics,73(3):321{351.McGarry,Kathleen.2008.\InheritanceandBequests."InTheNewPalgraveDictionaryofEconomics,SecondEdition,editedbyStevenN.DurlaufandLawrenceE.Blume.London:PalgraveMacmillan.McGarry,Kathleen.2013."AnAnalysisofFamilyTransfers:TheoryandEvidence."February,mimeo.Noble,AndrewL.2002.\IntestateSuccessionforStepchildreninPennsylvania:APro-posalforReform."UniversityofPittsburghLawReview,64(4):835{853.Pezzin,LilianaE.,andBarbaraSteinbergSchone.1998.\ParentalMaritalDisruptionandIntergenerationalTransfers:AnAnalysisofLoneElderlyParentsandTheirChil-dren."Demography,36(3):287{297.Pezzin,LilianaE.,RobertA.PollakandBarbaraSteinbergSchone.2008.\ParentalMaritalDisruption,FamilyType,andTransferstoDisabledElderlyParents."JournalofGerontology,SeriesB:SocialSciences,63(6):S349{S358.Pollak,RobertA..1988.\TiedTransfersandPaternalisticPreferences,"AmericanEco-nomicReviewPapersandProceedings,78(2):240{244.Rosenbury,Laura.2005.\TwoWaystoEndaMarriage:DivorceorDeath."UtahLawReview,2005(4):1227{1290.Schanzenbach,MaxM.,andRobertH.Sitko .2009.\TheEqualBequestPuzzle:ALegalPerspective."HarvardLawSchool,mimeo.Stevenson,Betsey,andJustinWolfers.2007."MarriageandDivorce:ChangesandtheirDrivingForces,"JournalofEconomicPerspectives,21(2):27-52.Tomes,Nigel.1981.\TheFamily,InheritanceandtheIntergenerationalTransmissionofInequality."JournalofPoliticalEconomy,89(5):928{958.Tomes,Nigel.1988.\InheritanceandInequalityWithintheFamily:EqualDivisionAmongUnequals,orDothePoorGetMore?"InModellingtheAccumulationandtheDistributionofWealth,editedbyDenisKesslerandAndreMasson.NewYork:OxfordUniversityPress,pp.79{104.Wilhelm,MarkO.1996.\BequestBehaviorandtheE ectofHeirs'Earnings:TestingtheAltruisticModelofBequests."AmericanEconomicReview,86(4):874{892.Wypyski,EugeneM.1984.TheLawofInheritanceinAllFiftyStates.NewYork:Oceana.49 AppendixTableA1:DescriptiveStatistics{DependentVariables (a)(b)(c)(d) ParentsParentshavenocontacthavecontactsDi erenceAllw.atleastwithall(b)-(c)onechildchildren(t-value)Nn A.NowillAll0.4460.5720.4210.152***41,07512,739(23.282)Divorced0.6270.6980.6020.096***7,8972,468(7.759)Widowed0.3910.5480.3600.188***14,3844,073(17.377)B.WillincludesallchildrenAll0.8220.6570.849-0.191***19,8115,734(24.725)Divorced0.8200.6030.885-0.282***2,473657(16.104)Widowed0.8320.6050.868-0.263***7,4612,045(21.441)C.EqualintendedbequestAll0.7560.5700.786-0.216***19,8115,734(24.937)Divorced0.7420.5070.811-0.304***2,473657(15.131)Widowed0.7510.5080.789-0.281***7,4612,045(19.765) Note:N=numberofobservations;n=numberofindividuals.*p0:10,**p0:05,***p0:01.Allguresarecalculatedonthecoreles.50 TableA2:E ectofNoContactsontheProbabilitythattheWillIncludesAllChildren Specication (a)(b)(c)(d)(e) AllparentsEstimate-0.194***-0.204***-0.227***-0.226***-0.192***(s.e.)(0.015)(0.017)(0.020)(0.021)(0.020)N19,81115,94912,46011,35410,724n6,9206,3654,8424,6804,470MothersEstimate-0.167***-0.179***-0.201***-0.199***-0.175***(s.e.)(0.017)(0.019)(0.022)(0.023)(0.022)N16,44913,1169,8388,9868,572n5,4185,0133,6323,5293,404FathersEstimate-0.287***-0.290***-0.302***-0.303***-0.249***(s.e.)(0.038)(0.042)(0.045)(0.048)(0.046)N3,3622,8332,6222,3682,152n1,5031,3531,2101,1511,066WidowsandwidowersEstimate-0.258***-0.255***-0.236***-0.230***-0.194***(s.e.)(0.028)(0.028)(0.030)(0.029)(0.027)N7,4617,4016,5156,4726,221n3,0753,0582,8392,8252,745WidowsEstimate-0.230***-0.229***-0.209***-0.204***-0.168***(s.e.)(0.030)(0.030)(0.031)(0.031)(0.028)N6,1066,0585,3035,2705,081n2,4152,4002,2262,2152,157WidowersEstimate-0.345***-0.338***-0.331***-0.323***-0.295***(s.e.)(0.076)(0.076)(0.088)(0.089)(0.087)N1,3551,3431,2121,2021,140n661659613610588DivorcedparentsEstimate-0.243***-0.244***-0.240***-0.240***-0.198***(s.e.)(0.043)(0.043)(0.047)(0.047)(0.045)N2,4732,4412,1152,0931,856n1,0121,004899891783DivorcedmothersEstimate-0.213***-0.222***-0.217***-0.232***-0.217***(s.e.)(0.052)(0.053)(0.059)(0.060)(0.060)N1,6781,6611,4471,4371,313n658655591588533DivorcedfathersEstimate-0.329***-0.321***-0.305***-0.279***-0.169**(s.e.)(0.077)(0.078)(0.080)(0.077)(0.073)N795780668656543n354349308303250 Note:SeethenotestoTables3and4.p0:10,p0:05,p0:01.51 TableA3:E ectofNoContactsontheProbabilitythatParentsIntendtoDividetheirEstateEquallyAmongAllChildren Specication (a)(b)(c)(d)(e) AllparentsEstimate-0.251***-0.265***-0.298***-0.298***-0.273***(s.e.)(0.016)(0.018)(0.021)(0.022)(0.022)N19,81115,94912,46011,35410,724n6,9206,3654,8424,6804,470MothersEstimate-0.226***-0.242***-0.284***-0.285***-0.266***(s.e.)(0.018)(0.021)(0.024)(0.026)(0.026)N16,44913,1169,8388,9868,572n5,4185,0133,6323,5293,404FathersEstimate-0.334***-0.338***-0.327***-0.325***-0.293***(s.e.)(0.037)(0.040)(0.042)(0.044)(0.045)N3,3622,8332,6222,3682,152n1,5031,3531,2101,1511,066WidowsandwidowersEstimate-0.339***-0.338***-0.337***-0.331***-0.301***(s.e.)(0.029)(0.029)(0.031)(0.031)(0.031)N7,4617,4016,5156,4726,221n3,0753,0582,8392,8252,745WidowsEstimate-0.324***-0.325***-0.320***-0.316***-0.284***(s.e.)(0.032)(0.032)(0.035)(0.035)(0.034)N6,1066,0585,3035,2705,081n2,4152,4002,2262,2152,157WidowersEstimate-0.387***-0.382***-0.399***-0.389***-0.366***(s.e.)(0.066)(0.066)(0.071)(0.072)(0.072)N1,3551,3431,2121,2021,140n661659613610588DivorcedparentsEstimate-0.331***-0.333***-0.320***-0.326***-0.312***(s.e.)(0.042)(0.042)(0.047)(0.047)(0.049)N2,4732,4412,1152,0931,856n1,0121,004899891783DivorcedmothersEstimate-0.312***-0.321***-0.309***-0.333***-0.342***(s.e.)(0.054)(0.055)(0.061)(0.062)(0.064)N1,6781,6611,4471,4371,313n658655591588533DivorcedfathersEstimate-0.368***-0.363***-0.329***-0.310***-0.269***(s.e.)(0.069)(0.070)(0.074)(0.073)(0.077)N795780668656543n354349308303250 Note:SeethenotestoTables3and4.p0:10,p0:05,p0:01.52 TableA4:E ectofHavingStepchildrenandNoContactsontheProbabilitythatParentsIntendtoDividetheirEstateEquallyAmongAllChildren|AlternativeSampleSelection Specication (a)(b)(c)(d)(e) A.ParentswithStepchildrenAllparentsEstimate-0.298***-0.327***-0.290***-0.322***-0.396***(s.e.)(0.018)(0.022)(0.023)(0.026)(0.031)N53,10535,28833,06625,20717,868n13,50411,5289,2558,4185,870B.No-contactParentsAllparentsEstimate-0.212***-0.229***-0.260***-0.259***-0.236***(s.e.)(0.017)(0.019)(0.023)(0.024)(0.023)N17,79514,48211,48310,5189,956n6,5746,0034,6244,4674,281 Note:Theestimatingsampleincludesonlyparentswithtwoormorechildrenandawillthatmentionsallofthechildren.Forotherexplanations,seethenotestoTables3and4.p0:10,p0:05,p0:01.53 TableA5:ProbabilitythataMaleStepchildisExplicitlyMentionedinHis/HerSteppar-ent'sWill (a)(b)(c)(d)(e) Meanofdependentvariable0.2610.2330.2590.2590.258Stepparenthasown-0.060***-0.060***-0.048***-0.065***-0.073***biologicalchildren(0.018)(0.018)(0.018)(0.023)(0.024)Yearsspentwithstepparent0.004***0.004***0.003***0.004***0.004***(0.001)(0.001)(0.001)(0.001)(0.001)Ageatmarriage:0-6a-0.016-0.0160.0230.0400.047(0.021)(0.021)(0.029)(0.039)(0.040)Ageatmarriage:7-12a-0.000-0.0010.0400.068*0.073**(0.019)(0.019)(0.026)(0.035)(0.035)Ageatmarriage:13-18a-0.016-0.015-0.0000.0030.001(0.013)(0.013)(0.016)(0.021)(0.021)Ageofstepchild0.0000.0000.002*0.0010.001(0.001)(0.001)(0.001)(0.001)(0.001)Stepparentisfemale0.0030.0050.033**0.0150.007(0.013)(0.013)(0.014)(0.019)(0.019)Stepparentage0.003***0.003***0.002**0.0010.000(0.001)(0.001)(0.001)(0.001)(0.001)Stepparentisin-0.060***-0.059***-0.080***-0.098***-0.095***poor/fairhealth(0.008)(0.008)(0.010)(0.012)(0.012)Stepparenttakescare0.090***0.099**0.0730.074ofstepchild'schild(ren)(0.030)(0.045)(0.048)(0.048)Stepchildismainrecipient0.164***0.165***0.160***ofintervivostransfers(0.031)(0.036)(0.036)Logofstepchild'sincome0.095***0.131***(0.015)(0.018)Stepchild'spredictedincomeisbelowbiologicalchildren'sincomeby:1{49percent0.103***(0.036)50+percent0.046**(0.019)N13,59813,5986,9895,6165,562n8,1988,1985,0014,1124,080 Note:Figuresaremarginale ectsfromprobitregressions.Standarderrorsareinparentheses.N=numberofobservations;n=numberofstepchildren.aTheagereportedherereferstotheageofthestepchildatthetimeinwhichhisparentfor-medapartnershipwithhisstepparent(whowritesthewill).Thereferencecategoryis18yearsormore.p0:10,p0:05,p0:01.54 TableA6:ProbabilitythataFemaleStepchildisExplicitlyMentionedinHis/HerStep-parent'sWill (a)(b)(c)(d)(e) Meanofdependentvariable0.2720.2450.2650.2650.265Stepparenthasown-0.028-0.028-0.001-0.005-0.003biologicalchildren(0.018)(0.018)(0.018)(0.022)(0.022)Yearsspentwithstepparent0.006***0.006***0.004***0.006***0.005***(0.001)(0.001)(0.001)(0.001)(0.001)Ageatmarriage:0-6a-0.004-0.0030.0480.081*0.085*(0.024)(0.024)(0.034)(0.045)(0.046)Ageatmarriage:7-12a-0.025-0.024-0.029-0.025-0.021(0.018)(0.018)(0.021)(0.027)(0.028)Ageatmarriage:13-18a0.0010.0010.0160.0090.007(0.015)(0.015)(0.019)(0.023)(0.023)Ageofstepchild-0.002**-0.002*-0.001-0.002-0.002(0.001)(0.001)(0.001)(0.001)(0.001)Stepparentisfemale0.0050.0070.002-0.006-0.008(0.014)(0.014)(0.015)(0.019)(0.019)Stepparentage0.005***0.005***0.005***0.004***0.004***(0.001)(0.001)(0.001)(0.001)(0.001)Stepparentisin-0.038***-0.038***-0.034***-0.024-0.021poor/fairhealth(0.010)(0.010)(0.011)(0.015)(0.015)Stepparenttakescare0.060**0.062*0.081*0.081*ofstepchild'schild(ren)(0.026)(0.037)(0.045)(0.045)Stepchildismainrecipient0.156***0.122***0.122***ofintervivostransfers(0.030)(0.033)(0.033)Logofstepchild'sincome0.129***0.142***(0.016)(0.019)Stepchild'spredictedincomeisbelowbiologicalchildren'sincomeby:1{49percent0.031(0.022)50+percent0.019(0.021)N13,38513,3856,9155,5715,544n8,1988,1985,0824,1734,159 Note:Figuresaremarginale ectsfromprobitregressions.Standarderrorsareinparentheses.N=numberofobservations;n=numberofstepchildren.aTheagereportedherereferstotheageofthestepchildatthetimeinwhichherparentfor-medapartnershipwithherstepparent(whowritesthewill).Thereferencecategoryis18yearsormore.p0:10,p0:05,p0:01.55