Riphahn, Regina T.; Feich maye , Jenni e
A icle — Published Ve sion
In e gene a ional T ansmission o Wel a e Bene i Receip :
E idence om Ge many
Re iew o Income and Weal h
P o ided in Coope a ion wi h:
John Wiley & Sons
Sugges ed Ci a ion: Riphahn, Regina T.; Feich maye , Jenni e (2024) : In e gene a ional T ansmission
o Wel a e Bene i Receip : E idence om Ge many, Re iew o Income and Weal h, ISSN 1475-4991,
Wiley Pe iodicals, Inc., Hoboken, NJ, Vol. 70, Iss. 4, pp. 1226-1251,
h ps://doi.o g/10.1111/ oiw.12680
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/313731
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by/4.0/
bs_bs_banne
Re iew o Income and Weal h
Se ies 70, Numbe 4, Decembe 2024
DOI: 10.1111/ oiw.12680
INTERGENERATIONAL TRANSMISSION OF WELFARE BENEFIT
RECEIPT: EVIDENCE FROM GERMANY
BY REGINA T. RIPHAHN*and JENNIFER FEICHTMAYER
F ied ich-Alexande -Uni e si y E langen-Nü nbe g
We s udy he in e gene a ional ansmission o wel a e bene i eceip in Ge many. We i s desc ibe
he co ela ion be ween wel a e eceip expe ienced in he pa en al household and subsequen own wel-
a e eceip o young adul s. In a second s ep, we in es iga e whe he he obse ed co ela ions e lec
causal e ec s o pas wel a e expe ience. We use amily ixed e ec s es ima ions and Go schalk’s (1996)
app oach and ake ad an age o he long- unning Ge man Socio-Economic Panel Su ey o con ibu e
o a spa se li e a u e. We ind s ong posi i e co ela ions be ween pa en al and own wel a e eceip .
These pa e ns do, howe e , no pe sis a e con olling o unobse ed he e ogenei ies. The e o e, ou
esul s sugges ha he s ong in e gene a ional co ela ion o wel a e bene i eceip is de e mined by
amily backg ound a he han by he expe ience o pa en al wel a e bene i eceip .
JEL Codes: I32, I38, J62, C36
Keywo ds: causal e ec , amily ixed e ec s, Go schalk es ima o , in e gene a ional mobili y, social
assis ance, wel a e
1. INTRODUCTION
I is well known ha pa en well-being a ec s child well-being. In e gen-
e a ional ansmission pa e ns a e s udied in ensely as he ansmission o
disad an age om pa en s o hei child en indica es inequali y o oppo uni ies.1
This pape in es iga es he in e gene a ional ansmission o pa icipa ion in
means- es ed minimum-income p o ec ion p og ams. The pu pose o such wel a e
p og ams is o li households om he mos p essing economic oubles and o
p o ec he nex gene a ion. I , howe e , wel a e p og ams cause wel a e eceip
o be passed om gene a ion o gene a ion hen hese p og ams do no wo k
p ope ly o he young. Ins ead, hey impose nega i e ex e nali ies and ha m he
nex gene a ion. He e, wel a e e o ms ha educe pa en al pa icipa ion can be
bene icial and pay o o he nex gene a ion, as well.
Va ious mechanisms may de e mine he in e gene a ional ansmission o
wel a e bene i eceip : a e expe iencing pa en al wel a e eceip you hs may be
be e in o med abou applica ion p ocedu es and ins i u ional ea u es; hey may
We hank Libe ad González, K is iina Hu unen, Kundu Anus up, Che-Yuan Liang, and pa ic-
ipan s o he 33 d Annual ( i ual) Con e ence o he Eu opean Associa ion o Labou Economis s
(EALE) 2021, he 77 h Annual Cong ess o he In e na ional Ins i u e o Public Finance 2021, and
he Annual Mee ing o he Ge man Economic Associa ion 2021 o help ul commen s and sugges ions.
Open Access unding enabled and o ganized by P ojek DEAL.
*Co espondence o: Regina T. Riphahn [email p o ec ed]
1Fo an ea ly su ey see, o example, Black and De e eux (2011). La e con ibu ions on he
ansmission o ea nings, educa ion, and place-based e ec s a e, o example, Ade mon e al. (2021),
Blanden (2013), and Che y and Hend en (2018a,2018b).
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion License, which
pe mi s use, dis ibu ion and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed.
1226
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
be a ec ed by pa en al ole models and be less subjec o s igma conce ns; hey may
know less abou he labo ma ke , and ecei e less pa en al suppo wi h espec o
human capi al in es men s o labo ma ke ne wo ks, compa ed o pee s who g ow
up wi hou wel a e. I he expe ience o wel a e eceip in he pa en al household
inc eases he nex gene a ion’s wel a e eceip by any such mechanism he wel a e
p og am has nega i e ex e nali ies. In e na ionally, mos s udies con i m posi i e
in e gene a ional co ela ions o wel a e bene i eceip , bu he e idence on causal
e ec s is mixed.
Mos o he li e a u e on he in e gene a ional ansmission o wel a e co e s
ei he he Uni ed S a es o Scandina ian coun ies such as Sweden and No way.
We a e he i s o o e e idence on he ecen in e gene a ional ansmission o
wel a e eceip o Ge many which p o ides an in e es ing labo a o y o s udy he
causal ansmission o wel a e bene i eceip and he exis ence o wha he li e a-
u e e med a “wel a e ap”. Ge many is a ele an in e media e case as i anges
be ween he Scandina ian coun ies and he U.S. wi h espec o income inequali y,
he p e alence o po e y, and he gene osi y o minimum income p o ec ion and
social spending (OECD, 2019). Ge many has much lowe po e y a es a e axes
and ans e s han he U.S. bu highe ones han Sweden (Imme oll e al., 2022).
The ins i u ions o he wel a e s a e a ec he e ec i eness o minimum income
p o ec ion including i s in e gene a ional e ec s; he e o e, a compa ison o in e -
gene a ional ansmission e ec s ac oss he di e en na ional wel a e sys ems (e.g.,
ollowing Esping-Ande sen, 1990, libe al s. conse a i e s. social-democ a ic sys-
ems) can be in o ma i e. While he Ge man wel a e sys em p o ec s he poo com-
pa a i ely well, we do no know ye whe he his ela i ely gene ous sys em also
succeeds in an in e gene a ional pe spec i e, ha is, by p o ec ing he nex gene a-
ion om inhe i ed dependence. This is ou esea ch ques ion.
We con ibu e o he in e na ional li e a u e on in e gene a ional wel a e
ansmission by o e ing e idence om mo e han h ee decades o su ey da a.
Long- unning longi udinal da a on pa en s and child en a e equi ed o examine
he ansmission o wel a e ac oss gene a ions. Fo mos coun ies, such da a is
no a ailable. The Ge man Socio-Economic Panel Su ey allows us o s udy he
ansmission o you h wel a e expe ience o indi iduals bo n 1969–1991. We can
conside pa en al wel a e eceip when he you h is 10–18 yea s old and in es iga e
i s associa ion wi h he young pe son’s own wel a e eceip a ages 25–29. The eby
ou analysis uses wide obse a ion windows han much o he p io li e a u e.2
In a i s s ep o ou analysis, we desc ibe he co ela ion be ween pa en
and child wel a e eceip . We s udy he co ela ion pa e ns be o e and a e a
majo wel a e e o m which is use ul o assess he sensi i i y o co ela ion pa e ns
o ins i u ional change. In con as o much o he li e a u e which ocuses on
mo he -daugh e pai s, we compa e ou comes o young men and women and
sepa a ely e alua e he ansmission om a he s and mo he s. Ou da a allow
2Appendix Table A.1 cha ac e izes he numbe o yea s o obse a ions used in p io s ud-
ies (see columns en i led “Exposu e ( 0)” and “Own wel a e ( 1)”). Fo pa en al wel a e eceip ,
Beaulieu e al. (2005) can use 10yea s o obse a ions. Howe e , se e al con ibu ions ha e ewe yea s:
An el (1992) and Le ine and Zimme man (1996) obse e pa en al wel a e eceip only o 1 yea , Edma k
and Hanspe s (2015) use only 3 yea s, and Boschman e al. (2019) 2yea s o child wel a e ou comes in
adul age.
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1227
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
us o desc ibe he ele ance o he age a which you hs a e exposed o pa en al
wel a e eceip and hus o de e mine he mos imp essionable yea s (K osnick
& Alwin, 1989). We look in o he po en ial media ion e ec o child educa ional
ou comes.
In a second s ep, we add ess he po en ial impac o unobse ed he e o-
genei ies ha ende pa en al wel a e eceip endogenous o he nex gene a ion’s
ou comes. In pa icula , pa en al cha ac e is ics, such as human capi al, a i udes
owa ds wo k and amily, heal h, addic ions, and emo ional well-being may a ec
bo h gene a ions’ wel a e eceip and hus can gene a e spu ious in e gene a ional
wel a e co ela ions. To accoun o his, we conside he empi ical s a egy de el-
oped by Go schalk (1996) and apply amily ixed e ec s es ima ion. The wo
app oaches iden i y di e en e ec s and apply di e en me hods o con ol o he
po en ial endogenei y o pa en al wel a e eceip . I he iden i ying assump ions
hold, applying bo h me hods allows us o ge close o answe ing he ques ion o
whe he pa en al wel a e eceip causally a ec s he wel a e eceip o he nex
gene a ion in Ge many.
We ind a s ong in e gene a ional co ela ion in wel a e ou comes o ou h ee
wel a e indica o s. The co ela ions a e la ge o emales han o males. We do
no ind impo an di e ences in wel a e ansmission om a he s e sus mo he s.
Exposu e o pa en al wel a e eceip a he ages o 10–12 and 16–18 yields s onge
co ela ion pa e ns han in he 13–15 age window. Compa ing he co ela ion pa -
e ns be o e and a e a majo wel a e e o m we ob ain inconclusi e esul s and
canno con i m ha in e gene a ional ansmission declined pos - e o m.
Bo h, he amily ixed e ec s and he Go schalk (1996) me hod iden i y causal
e ec s unde ce ain, ye di e en assump ions. In ou case, bo h s a egies ail o
ind e idence o a causal impac o pa en al wel a e eceip on child wel a e ou -
comes. Thus, in he Ge man ins i u ional amewo k, i does no appea o be he
expe ience o pa en al wel a e eceip ha d i es subsequen child wel a e eceip
bu he co ela ion o indi idual cha ac e is ics and ci cums ances in he child and
pa en household. Le ine and Zimme man (1996) call his si ua ion a “po e y
ap” as opposed o a “wel a e ap”.
These esul s ha e clea policy implica ions as hey show ha i is no he cha -
ac e o wel a e ins i u ions hemsel es ha lea es he o sp ing o wel a e ecipien s
a an ele a ed isk o wel a e eceip . The e o e, any ini ia i e o educe in e gene a-
ional co ela ion in wel a e eceip mus no ocus on he ins i u ions o he wel a e
sys em bu add ess cha ac e is ics a he indi idual and household le el and, o
example, imp o e human capi al, heal h, and labo ma ke in ol emen .
In he nex Sec ion, we summa ize he s a e o he li e a u e. Sec ion 3 hen
p o ides ins i u ional backg ound. We ou line ou empi ical app oach in Sec ion 4
and desc ibe ou da a in Sec ion 5. Nex , we p esen he esul s o ou desc ip i e
analyses o in e gene a ional ansmission pa e ns and o ou causal es ima es in
Sec ion 6. Finally, we d aw conclusions in Sec ion 7.
2. PRIOR LITERATURE
While a b oad in e na ional li e a u e desc ibes in e gene a ional co ela ion
in wel a e eceip , ewe s udies iden i y causal e ec s o minimum income p o-
g ams; Appendix Table A.1 o e s a b ie cha ac e iza ion o p io con ibu ions
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1228
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
and hei esul s. Ea ly con ibu ions applied s uc u al es ima ion app oaches
and U.S. su ey da a (An el, 1992; Le ine & Zimme man, 1996) wi h opposi e
esul s. While An el (1992) concludes ha ma e nal wel a e use causally a ec s
daugh e s’ eceip , Le ine and Zimme man (1996) ind only a co ela ion in
incomes. Go schalk (1996) s udies U.S. wel a e ansmission by applying e en
s udy me hods and con i ms a causal ela ionship be ween mo he s’ and daugh-
e s’ wel a e eceip . Peppe (2000) compa es al e na i e empi ical app oaches and
con i ms a causal ela ionship. Ha ley e al. (2022) use ins umen al a iables
and di e ence-in-di e ences s a egies based on egional he e ogenei ies and ind
ha mo he s’ wel a e eceip inc eases he p obabili y o hei daugh e s’ wel a e
pa icipa ion. Howe e , wel a e e o ms did a enua e he ansmission. Finally,
Mi nik (2010) s udies he in ensi e ma gin o wel a e eceip ; applying ma ching
and amily ixed e ec s es ima o s he does no ind causal e ec s.
The e a e only a ew s udies co e ing coun ies ou side he U.S. Beaulieu
e al. (2005) exploi adminis a i e da a on social assis ance eceip in Quebec,
Canada, and con i m causal in e gene a ional e ec s. Edma k and Hanspe s (2015)
apply amily ixed e ec s es ima ion o Swedish egis e da a and ind no causal
e ec s. In hei s udy using adminis a i e da a om No way, De Haan and
Sch eine (2018) apply bounds analyses wi h ins umen al a iables and con-
i m signi ican posi i e causal ansmission e ec s. Boschman e al. (2019)and
Cobb-Cla k e al. (2022) apply he Go schalk (1996) app oach o Du ch and
Aus alian adminis a i e da a, espec i ely, and ind no signi ican causal e ec s
o social assis ance bene i s. O e all, he e idence on he causal in e gene a ional
ansmission o wel a e eceip is mixed.3
So a , li le esea ch has add essed in e gene a ional wel a e ansmission
in Ge many. While he e a e a numbe o s udies on income and unemploymen
ansmission, esea ch on wel a e eceip is limi ed. Closes o ou analysis
is Siedle (2004): using ea ly da a om he Ge man Socioeconomic Panel
(1984–2002) he in es iga es in e gene a ional co ela ion in social assis ance
eceip . He ocuses on young adul s’ bene i eceip a age 22 o abo e, ha is, a
an age when almos 40 pe cen o he sample s ill li e in he pa en al household.
He applies egional cha ac e is ics as ins umen s as well as bounds analyses and
concludes ha pa en al bene i eceip is exogenous. The e o e, he co ela ion
pa e ns a e in e p e ed as causal e ec s.
3. INSTITUTIONAL BACKGROUND
The Ge man cons i u ion gua an ees each esiden he igh o a “digni ied
li e”: i an indi idual o household canno mus e he inancial means o a “digni-
ied li e”, he pe son o household can demand he suppo o he s a e. Di e en
3The li e a u es on he ansmission o disabili y and unemploymen bene i s apply simila me h-
ods wi h mixed esul s: Dahl e al. (2014), Dahl and Gielen (2021), and G übl e al. (2020) con i m
causal in e gene a ional ansmission while Ekhaugen (2009) and Maede e al. (2015) ejec i . B a be g
e al. (2015) and Muelle e al. (2017) ind causal ansmission pa e ns o some amily ela ionships bu
no o o he s.
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1229
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
Figu e 1. Unemploymen and wel a e ins i u ions o he wo king-age popula ion in Ge many be o e
and a e he 2005 e o m.
Sou ce: Own illus a ion .
p og ams p o ide assis ance o g oups such as he unemployed, he elde ly, he
disabled, and he poo .4In ou analysis, we join ly conside hose b anches o he
wel a e s a e ha p o ide means- es ed minimum income suppo o indi iduals
below e i emen age ( o a simila s a egy see Boschman e al., 2019).
As he wel a e s a e unde wen a majo e o m in 2005, we dis inguish be ween
p e- and pos - e o m ins i u ions (see Figu e 1). We conside he eceip o social
assis ance (Sozialhil e) and unemploymen assis ance (A bei slosenhil e) be o e he
e o m and social assis ance, unemploymen bene i II (UB II), and social money
a e he e o m o cap u e means- es ed minimum income suppo . We label he
combined ins i u ions “wel a e” h oughou .
Be o e he e o m, indi iduals could claim means- es ed social assis ance
(Sozialhil e) i hei household income, ha is, he combina ion o ea nings o
o he income, unemploymen bene i s, o unemploymen assis ance, was oo low
o co e he o mally de ined inancial need o he household. Social assis ance
p o ided gene al income suppo o he employed, he unemployed, and hose ou
o he labo o ce. In addi ion, hose who had exhaus ed hei insu ance-based
unemploymen bene i s and hose who we e no (ye ) en i led o unemploymen
bene i s we e eligible o a second, ax- inanced and means- es ed unemploymen
assis ance (A bei slosenhil e). Unemploymen assis ance eplaced up o 57 pe cen
o p e ious ne labo ea nings and in mos cases was p o ided wi hou a ime limi ,
ha is, a mos un il e i emen .
On Decembe 24, 2003, he e o m law (Vie es Gese z ü mode ne Diens leis-
ungen am A bei sma k called “Ha z IV”) was passed which came in o e ec on
Janua y 1, 2005. I s objec i e was o educe ans e dependence and sho en he
4Po e y is es ablished in a means es : i s , he inancial need o a gi en household is o mally
de e mined. I consis s o adminis a i ely ixed amoun s o all household membe s plus housing expen-
di u es ( en and hea ing). I household income and weal h a e oo low o co e he hus calcula ed
inancial need he household can claim go e nmen suppo . While ins i u ional egula ions a e gende
neu al, emales a e mo e a ec ed by po e y han males.
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1230
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
ans e eceip pe iod. Excep o sho ened payou pe iods, he unemploymen
insu ance bene i was no a ec ed by he e o m; Riphahn and Sch ade (2020)
s udy he e ec o he educed payou pe iod. Figu e 1summa izes he ins i u ional
changes caused by he e o m: he o me unemploymen assis ance and social
assis ance p og ams we e combined in he new UB II p og am, a means- es ed
and ax- inanced bene i o hose able o wo k. Since he e o m, indi iduals who
exhaus hei unemploymen insu ance bene i (i.e., UB I) o whose UB I claim is
insu icien o co e he household’s inancial need may be eligible o UB II (pos-
sibly in addi ion o UB I). The UB II bene i co e s he legally de ined minimum
income (household inancial need). Gene ally, all indi iduals—including hose
who a e employed o ou o he labo o ce—can claim UB II i hei household
passes he means es and i hey a e physically able o wo k a leas 15 h pe week.
Thei child en o o he household membe s who a e no able o wo k can claim
a simila bene i called social money (Sozialgeld). Independen indi idual claims
agains he UB II sys em a e possible s a ing a age 25; hen, inancial means a e
compa ed o hei needs o he young pe son o he pe son’s own co e amily.
Since he e o m, he p e ious social assis ance (Sozialhil e) p og am is a ailable
only o hose who a e no able o wo k, o example, due o sickness, disabili y, o
ca e esponsibili ies, and who do no ha e an employable household membe .
The main change induced by he e o m was he aboli ion o he unemploy-
men assis ance p og am. Indi iduals wi h high p io labo ea nings who p e i-
ously ecei ed unemploymen assis ance aced cu s: hei bene i claims declined
and in addi ion, hey had o pass mo e s ingen means es s han be o e. Those
who ecei ed social assis ance be o e he e o m con inued o be eligible o UB II
as long as hey we e able o wo k; o de ails on he Ge man wel a e sys em see
BMAS (2019,2020).
Figu e 2desc ibes he u iliza ion o he wel a e p og ams o e ime. The
absolu e numbe o social assis ance ecipien s (dashed line) inc eased since 1980
om below 1million o almos 3million indi iduals in 2004.5Simila ly, he numbe
o unemploymen assis ance ecipien s (do ed line) inc eased subs an ially o e
ime—since 1991 co e ing Eas Ge many, as well. The hin do ed g ey line p esen s
he sum o social assis ance and unemploymen assis ance bene icia ies; as some
indi iduals may ha e bene i ed om bo h p og ams he addi ion gene a es an
o e coun . The unemploymen assis ance p og am disappea ed in 2005. Immedi-
a ely a e he e o m, he numbe o unemploymen bene i II (UB II) ecipien s
(bold black line) su passed 5million basically con inuing whe e he sum o he wo
p io bene i s le o . The numbe declined in subsequen yea s. The numbe o
social money ecipien s was cons an a abou 0.8 million and e lec s indi iduals in
he household o UB II ecipien s who canno wo k, ha is, mos ly child en. A e
he e o m, he social assis ance bene i was used only by indi iduals unable o wo k
a leas 3 h pe day and d opped. The igu e sugges s ha he join conside a ion
o he wo means- es ed p og ams o social and unemploymen assis ance be o e
5This co e s wel a e ecipien s who li e independen ly (Hil e zum Lebensun e hal auße halb on
Ein ich ungen). The g oup o disabled indi iduals was suppo ed by a di e en social assis ance p og am
(Hil e in besonde en Lebenslagen) and is no e lec ed in Figu e 2.
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1231
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
0
10,00,000
20,00,000
30,00,000
40,00,000
50,00,000
60,00,000
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
2020
Soc. Assis . Un. Assis . UB II Social Money SA + UA
Figu e 2. U iliza ion o wel a e p og ams o e ime.
Sou ce: Own depic ion based on in o ma ion om di e en sou ces. Un il 1990 only Wes
Ge many, s a ing 1991 Eas and Wes Ge many. Social Assis ance ( ecipien s as o 31.12. annu-
ally) om h ps://www.des a is.de/DE/Themen/ Gesellscha -Umwel /Soziales/Sozialhil e/Tabellen/
lis e-hil e-lebensun e hal -emp aenge -z .h ml [las accessed July 28, 2021]. Unemploymen assis-
ance (annual a e age numbe o ecipien s), BA (2020), A bei slosengeld und A bei slosenhil e
on 1991–2004 (Zei eihen Mona s- und Jah eszahlen); o ea lie yea s: annual publica ions o
Am liche Nach ich en de Bundesagen u ü A bei (ANBA). Unemploymen bene i II (UB II) and
Social Money ecipien s as o Decembe each yea : BA (2021), S uk u en de G undsiche ung SGB
II – Deu schland, Wes /Os , Lände und K eise (Zei eihe Mona s- und Jah eszahlen ab 2005), Table 1
(e we bs ähige und nich e we bs ähige Leis ungsbe ech ig e).
and he UB II p og am a e he e o m gene a es a plausible e lec ion o wel a e
eceip .6
4. EMPIRICAL MODEL AND METHODS
4.1. The Model
We a e in e es ed in whe he he wel a e eceip o young adul s, ha is, he
child gene a ion, is associa ed wi h and po en ially caused by expe iencing he wel-
a e eceip o hei pa en s. We ollow he p e ious li e a u e and model child i’s
6The e is subs an ial non- ake-up in he Ge man wel a e sys em o mo e han 40 pe cen o he
eligible popula ion. Howe e , his is mode a e by Eu opean compa ison (Eu o ound, 2015,Table1).
Also, o he social policy p og ams in Ge many ea u e e en highe non- ake-up a es (B uckmeie &
Wieme s, 2018,Table3). These au ho s poin o a complica ed bene i s uc u e and a gue he expec ed
u ili y o he en i lemen s as well as in o ma ion cos s and s igma iza ion explain ake-up beha io . Fo
ecen e idence see, o example, B uckmeie e al. (2021), B uckmeie and Wieme s (2017), and o he
p e- e o m wel a e sys em Riphahn (2001).
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1232
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
wel a e eceip (WiC) in obse a ion pe iod 1as a unc ion o pa en al wel a e
eceip (WiP) in an ea lie obse a ion pe iod 0:
(1) WiC=WiP𝛽0+ε
0iC.
The es ima e o coe icien 𝛽0 e lec s he uncondi ional in e gene a ional co -
ela ion in wel a e eceip . As his co ela ion may be a ec ed by a ious ac o s, we
conside an ex ended speci ica ion ha con ols o a se o indi idual and house-
hold le el co a ia es (X) such as age, gende , mig a ion backg ound, and egion o
esidence:
(2) WiC=WiP𝛽1+Xiγ+ε
1iC,
whe e 𝛽and 𝛾a e coe icien s o be es ima ed. The es ima e o 𝛽1 e lec s he condi-
ional co ela ion o wel a e eceip ac oss gene a ions. While i may no p o ide he
causal ansmission e ec , i quan i ies he o e all associa ion be ween pa en and
child ou comes. I is in e es ing o compa e his associa ion o di e en subg oups
and o di e en ypes o exposu e.
Es ima es o 𝛽1can be in e p e ed as a causal e ec only i pa en al
wel a e eceip is exogenous, ha is, unco ela ed wi h he e o e m ε1iC.
Howe e , his is unlikely i pa en and child wel a e pa icipa ion a e a ec ed
by unobse ed he e ogenei ies (e.g., as es, p e e ences, biological ac o s, abil-
i ies, o unobse ed egional cha ac e is ics). Le pa en al wel a e eceip be
modeled by
(3) WiP=XiPδ+ε
iP.
Then, he e o e ms o child and pa en wel a e eceip may ollow
(4) εiC=𝛼iC+𝜇iC
(5) εiP=𝛼iP+𝜇iP,
whe e 𝜇iCand 𝜇iPa e unco ela ed andom e o componen s. I he e a e unob-
se ed amily cha ac e is ics we expec co (𝛼iC,𝛼iP)≠0. This co ela ion causes a
bias in he OLS es ima e o 𝛽in Equa ions (1) and (2): he coe icien es ima e mixes
he causal e ec o expe iencing pa en al wel a e eceip in pe iod 0and he e ec s
o sha ed amily unobse ables.
In he i s s ep o ou analysis, we es ima e he in e gene a ional co ela ion
o wel a e eceip using wo model speci ica ions. In a basic speci ica ion, we do
no conside a de ailed se o con ol a iables. In an ex ended speci ica ion, we
accoun o he e ogenei y along indi idual and pa en al backg ound dimensions.
In pa icula , we con ol o cha ac e is ics o he indi idual (yea o bi h, gende ,
immig a ion backg ound, and pa i y, i.e., he ank posi ion in he amily bi h
o de ), cha ac e is ics o pa en s (yea o bi h, pa en al educa ion) and household
cha ac e is ics a age 17 o he indi idual (household size, numbe o child en in
pa en al household, ede al s a e o esidence). Holding hese dimensions cons an
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1233
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
In panel B o Table 2, we p esen he es ima es o he ex ended speci ica ion.
The con ols accoun o some o he in e gene a ional co ela ion in wel a e
eceip : he coe icien s decline in magni ude bu emain highly s a is ically sig-
ni ican . Condi ional on indi idual, pa en , and household cha ac e is ics young
indi iduals a e abou 14 pe cen age poin s mo e likely o ecei e wel a e when hei
pa en s ecei ed wel a e du ing hei een yea s, a subs an ial di e ence; Table A.4
in he Appendix p esen s he ull se o es ima ion esul s. The es ima e in column 2
shows an inc easein he numbe o ownyea s o wel a e expe iencebyabou 0.16 o
each yea o pa en al wel a e eceip . Column 3 sugges s ha he sha e o obse ed
yea s on wel a e as an adul is associa ed wi h a signi ican inc ease o 21 pe cen age
poin s when pa en s we e on wel a e o he ull obse a ion pe iod. The educed
in e gene a ional co ela ions in Panel B compa ed o Panel A sugges ha he con-
ol a iables a e indeed co ela ed wi h he p opensi y o ecei e wel a e bene i s:
he in e gene a ional co ela ion o wel a e eceip is smalle wi hin demog aphic
g oups han on a e age. Below we inspec hese he e ogenei ies in g ea e de ail.
Nex , we in es iga e whe he he associa ion be ween child wel a e eceip and
he du a ion o pa en al eceip is indeed linea . We sepa a ely eg ess he ex ensi e
ma gin o child wel a e eceip (i.e., e e wel a e in pe iod 1) on ha ing expe ienced
a leas xnumbe o yea s o pa en al eceip , whe e x uns om 1 o 9. Figu e A.1 in
he appendix shows he esul s o bo h speci ica ions: he p opensi y o e e ecei e
wel a e inc eases wi h he numbe o yea s o pa en al wel a e eceip expe ienced,
howe e , con idence in e als a e wide. In Figu e A.2 we desc ibe he de elopmen
o co ela ion pa e ns as es ima ed by he basic eg ession speci ica ion sepa a ely
o subsequen bi h coho s. We use olling eg essions on h ee neighbo ing bi h
coho s. The pa e ns a e simila o all h ee ou comes wi h a peak in co ela ions
in he ea ly 1970s and a signi ican posi i e end o mo e ecen bi h coho s.
6.2. He e ogenei y by Child and Pa en Gende
Nex , we ollow he li e a u e and in es iga e whe he in e gene a ional
wel a e co ela ion di e s o young men and women; desc ip i e s a is ics yield
highe wel a e eceip among emales han males.9We apply di e en s a egies
o desc ibe he gende -speci ic pa e ns in ou da a. Fi s , we e-es ima ed he
ex ended speci ica ion desc ibed in Table 2and addi ionally in e ac ed pa en al
wel a e eceip wi h child gende . Panel A o Table 3shows ha he co ela ion
be ween pa en and child wel a e eceip is subs an ially bu mos ly insigni ican ly
highe o emales. Panels B and C o Table 3show sepa a e es ima ions o he
basic and ex ended speci ica ions by gende and con i m highe in e gene a ional
co ela ions o emales han males ac oss all wel a e indica o s. This ag ees wi h
he li e a u e (e.g., Dahl & Gielen, 2021; Hoynes e al., 2016). One mechanism
may be ha he single pa en hood isk is la ge o emales and can be ansmi ed
ac oss gene a ions (Musick & Ma e, 2004). Also, ole model expec a ions and
social no ms may con ibu e o gende di e ences in economic independence.
9In ou sample, 14 and 11 pe cen o emales and males e e ecei e wel a e, espec i ely. Fo
emales, we obse e on a e age 0.62 and o males 0.48yea s o wel a e eceip . The di e ences in he
pa en gene a ion a e small and insigni ican (see Table A.5 o desc ip i e s a is ics by gende ).
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1240
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
TABLE 3
GENDER-SPECIFIC EFFECTS
Dependen a iables: Wel a e eceip 1
(1) (2) (3)
E e
(0/1) Numbe
(yea s) Sha e
(%)
Panel A: Ex ended speci ica ion wi h gende in e ac ion (N=2403)
Pa en wel a e, 00.100*** 0.083** 0.151***
(0.036) (0.038) (0.050)
Female ×Pa en wel a e, 00.074 0.145** 0.117
(0.050) (0.059) (0.039)
Panel B: Male sample (N=1205)
Pa en wel a e, 0, basic speci ica ion 0.161*** 0.134*** 0.201***
(0.036) (0.038) (0.051)
Pa en wel a e, 0, ex ended speci ica ion 0.109*** 0.089** 0.154***
(0.037) (0.039) (0.049)
Panel C: Female sample (N=1198)
Pa en wel a e, 0, basic speci ica ion 0.213*** 0.269*** 0.306***
(0.037) (0.047) (0.058)
Pa en wel a e, 0, ex ended speci ica ion 0.166*** 0.214*** 0.254***
(0.035) (0.047) (0.056)
Panel D: Ma e nal wel a e eceip —basic speci ica ion (N=2375)
Ma e nal wel a e, 00.195*** 0.208*** 0.243***
(0.028) (0.035) (0.041)
Panel E: Pa e nal wel a e eceip —basic speci ica ion (N=2284)
Pa e nal wel a e, 00.197*** 0.233*** 0.296***
(0.032) (0.048) (0.055)
Panel F: Ma e nal wel a e eceip —ex . speci ica ion wi h gende in e ac ion (N=2375)
Ma e nal wel a e, 00.158*** 0.126*** 0.178***
(0.039) (0.042) (0.054)
Female ×Ma e nal wel a e, 00.071 0.154** 0.122
(0.056) (0.066) (0.081)
Panel G: Pa e nal wel a e eceip —ex . speci ica ion wi h gende in e ac ion (N=2284)
Pa e nal wel a e, 00.163*** 0.165*** 0.264***
(0.045) (0.060) (0.084)
Female ×Pa e nal wel a e, 00.061 0.126 0.052
(0.063) (0.093) (0.110)
No es: Each cell en y ep esen s a sepa a e eg ession whe e he pa en al wel a e measu e ma ches
he dependen a iable as lis ed in he column heade s (see Table 2). Panels D–G addi ionally con ol
o an indica o e lec ing whe he an indi idual e e li ed wi h a single pa en (i.e., in a non-couple
household). Robus s anda d e o s a e epo ed in pa en heses. Fo de ails on he basic and ex ended
speci ica ion see no es o Table 2.∗∗∗p<0.01; ∗∗p<0.05.
Sou ce: SOEP (1984–2017), own calcula ions.
In Panels D and E o Table 3, we p esen sepa a e es ima es based on whe he
ma e nal o pa e nal wel a e eceip was obse ed du ing childhood.10 Since any
10As wel a e is p o ided a he household as opposed o he indi idual le el he wel a e ou come in
ou da a was iden ical o 88 pe cen o pa en couples. The gende -speci ic e ec s a e iden i ied om
sepa a ed couples o single pa en s whe e he child en li e wi h only one o he wo pa en s. In e y ew
cases (28 o mo he s and 119 o a he s) we ha e no in o ma ion on he pe son-speci ic wel a e his o y.
The e a e no majo di e ences by pa en gende in ex ended speci ica ion. Resul s a e a ailable upon
eques .
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1241
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
di e ences migh be due o li ing wi h only one pa en a he han o he pa en al
gende in pa icula , we addi ionally con ol o single pa en hood in hese spec-
i ica ions. Ou esul s yield only mino di e ences in pa en -speci ic co ela ion
pa e ns o he basic speci ica ion; in sepa a e es ima ions (no p esen ed o sa e
space) we obse e la ge in e gene a ional co ela ion coe icien s i ma e nal wel-
a e eceip was expe ienced in a single-pa en household. In sepa a e es ima ions,
we conside ed child gende in e ac ion e ms in he es ima ions o a he s’ and
mo he s’ wel a e ou comes (see panels F and G o Table 3). These esul s indica e
posi i e bu again mos ly insigni ican coe icien es ima es con i ming he s onge
co ela ions o emale child en bu no majo di e ences by pa en gende .
O e all, he indings con i m pa e ns ound in o he s udies: using da a o
Ge many, Muelle e al. (2017) ob ained s onge in e gene a ional unemploymen
co ela ions o daugh e s han o sons. Using Du ch da a, Boschman e al. (2019)
also ound he co ela ion pa e ns o ma e nal and pa e nal social assis ance
eceip o be simila .11
6.3. He e ogenei y by Age o Exposu e
Nume ous con ibu ions discuss he ele ance o a child’s age a exposu e
o in e gene a ional ansmission e ec s. B a be g e al. (2015) and Dahl and
Gielen (2021) s udy he ele ance o age a exposu e wi h espec o he ansmis-
sion o pa en al disabili y.12 Ca nei o e al. (2021) s udied he connec ion be ween
he iming o pa en al income shocks and he nex gene a ion’s human capi al
ou comes. Condi ional on household pe manen income hey ind ha child en
bene i mos om posi i e income shocks du ing age 0–5 and 12–17.
Edma k and Hanspe s (2015) and Ha ley e al. (2022) compa e he ele ance
o pa en al wel a e eceip ac oss child exposu e ages. The o me ind he s onges
in e gene a ional co ela ion i he young gene a ion was exposed a age 17–19 and
a gue ha his may e lec ole-model o ne wo k- ela ed e ec s ha a e s onges
in he o ma i e yea s o he la e eens. Ha ley e al. (2022) ind la ge co ela ions
o olde ages a exposu e, ha is, a ages 10–14 h ough 13–17. The au ho s
simila ly sugges ha lea ning e ec s inc ease when child en expe ience wel a e
eceip a olde ages.
In ou analysis, we conside exposu e o pa en al wel a e eceip a ages 10–12,
13–15, and 16–18. As ou su ey does no allow us o go back in ime o all indi-
iduals, we s a ou wi h age-g oup-speci ic es ima ions which a y in sample size.
Panel A o Table 4shows he esul s. Ac oss all wel a e indica o s, we ind s onge
co ela ions o he younges and oldes age g oups and he smalles co ela ions
o he middle age g oup o 13–15-yea -olds. In o de o compa e he age-speci ic
11The s udies on in e gene a ional ansmission o disabili y bene i s in No way disag ee on his
issue: while Dahl and Gielen (2021) ind la ge ansmissions om mo he s, B a be g e al. (2015) obse e
la ge e ec s o a he s. While Dahl and Gielen (2021) ind no he e ogenei y by child gende , B a be g
e al. (2015) obse e la ge e ec s o daugh e s han sons.
12B a be g e al. (2015) compa e child age ca ego ies om below 15 o up o 40 and do no ind clea
he e ogenei ies o exposu e a younge ages. Dahl and Gielen (2021) compa e e ec s o child en up o
age 14, up o age 18 o a age 19 plus. They ind la ge in e gene a ional spillo e e ec s i he younge
gene a ion is young a he ime o pa en al ea men .
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1242
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
TABLE 4
HETEROGENEITY OF WELFARE CORRELATION BY AGE OF EXPOSURE
Dependen a iables: Wel a e eceip 1
(1) (2) (3)
E e (0/1) Numbe (yea s) Sha e (%)
Panel A—Sepa a e es ima ions by age g oup—Basic speci ica ion
Age g oup 10–12, 0(N=1242) 0.225*** 0.559** 0.282***
(0.046) (0.117) (0.060)
Age g oup 13–15, 0(N=1835) 0.189*** 0.400*** 0.196***
(0.037) (0.081) (0.040)
Age g oup 16–18, 0(N=2403) 0.201*** 0.427*** 0.203***
(0.030) (0.065) (0.032)
Panel B—Join Es ima ions o all Age G oups (N=1216)—Basic Speci ica ion
Age g oup 10–12, 00.135** 0.303*0.147**
(0.055) (0.156) (0.071)
Age g oup 13–15, 00.051 0.129 0.013
(0.052) (0.150) (0.068)
Age g oup 16–18, 00.123** 0.325*** 0.156***
(0.048) (0.121) (0.056)
No es: Each cell en y ep esen s a sepa a e eg ession whe e he pa en al wel a e measu e ma ches
he dependen a iable as lis ed in he column heade s (see Table 2). Robus s anda d e o s a e epo ed
in pa en heses. Fo de ails on he basic speci ica ion see no es o Table 2.∗∗∗p<0.01; ∗∗p<0.05; ∗p<0.10.
Sou ce: SOEP (1984–2017), own calcula ions.
co ela ions o a gi en ye smalle sample we pooled he h ee age-g oup-speci ic
measu es in panel B o Table 4and es ima ed he co ela ion pa e ns in one join
model. We con inue o ind he weakes co ela ion o he middle age g oup and
la ge impac s o he younges and he oldes g oup. Table 4shows esul s o he
basic speci ica ion only. The esul s a e simila when he ex ended speci ica ion is
es ima ed (a ailable upon eques ). The inding o la ge coe icien s o he oldes
g oup ag ees wi h he li e a u e. The s ong co ela ion o 10–12-yea -olds is some-
wha su p ising. Possibly i is ela ed o he Ge man seconda y schooling sys em
whe e a a ound age 10 impo an acking decisions a e aken. I hese decisions
a e nega i ely a ec ed by inancial p oblems in he pa en al household he e ec s
may educe a e age human capi al wi h long- un e ec s. Boschman e al. (2019)
conside he he e ogenei y o co ela ion pa e ns by ecency o pa en al wel a e
eceip as a po en ial indica o o he ele ance o in o ma ion ansmission. Tha
we ind co ela ions o simila magni ude o 10–12 and 16–18-yea -olds does no
suppo he idea o ecen in o ma ion as an impo an media o .
6.4. P e Ve sus Pos Re o m Pa e ns
Du ing ou obse a ion pe iod, he Ge man wel a e p og am unde wen an
impo an e o m in 2005 ha is in ensely deba ed o his day. The e o m aimed
o ac i a e wel a e ecipien s who a e able o wo k (see Sec ion 3). I inc eased
job sea ch moni o ing and i educed bene i s o some long- e m unemployed. We
desc ibe in e gene a ional co ela ion pa e ns be o e and a e he e o m; o anal-
yses o p e- and pos - e o m s a e dependence in wel a e eceip a he indi idual
le el o e ime see, o example, Riphahn and Wunde (2013,2016). We conside
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1243
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
indi iduals who eached age 29 be o e 2005 (bi h coho s 1969–1975) o be subjec
o he p e- e o m wel a e egime and hose who u ned 25 in 2005 and a e (bi h
coho s 1980–1991) o be a ec ed by he e o m. Appendix Table A.6 desc ibes he
wo g oups’ wel a e ou comes and in e gene a ional co ela ion pa e ns. While he
wel a e ou comes o he wo subsamples in 1 a e simila , su p isingly, we ind much
highe pa en al wel a e eceip o he pos e o m g oup (see Panel A). Panel B
addi ionally shows highe in e gene a ional co ela ions o he pos - e o m o , o
example, 0.24 e sus 0.15 o he “e e wel a e” ou come. This may e lec agg ega e
ends o highe wel a e use o e ime (see Figu e 2).
Table 5shows he es ima ion esul s o bo h subsamples wi h he basic and
ex ended speci ica ions in Panels A and B, espec i ely. Panels C and D o e es ima-
ion esul s on he pooled samples wi h an in e ac ion e m. In Panel A, es ima ion
TABLE 5
PRE-VERSUS POST-REFORM OUTCOMES
Dependen a iables: Wel a e eceip 1
(1) (2) (3)
E e (0/1) Numbe (yea s) Sha e (%)
Panel A—Basic speci ica ion
P e-Re o m (N=778)
Pa en wel a e, 00.202*** 0.216** 0.163**
(0.0624) (0.0895) (0.0746)
Pos -Re o m (N=1210)
Pa en wel a e, 00.196*** 0.218*** 0.308***
(0.0317) (0.0342) (0.0475)
Panel B—Ex ended speci ica ion
P e-Re o m (N=778)
Pa en wel a e, 00.184*** 0.205** 0.146**
(0.0620) (0.0904) (0.0742)
Pos -Re o m (N=1210)
Pa en wel a e, 00.143*** 0.176*** 0.264***
(0.0313) (0.0348) (0.0459)
Panel C—Basic speci ica ion wi h in e ac ion e ms
Full pe iod (N=1988)
Pa en wel a e, 00.191*** 0.207** 0.159**
(0.0619) (0.0880) (0.0725)
Pos ×Pa e nal wel a e, 00.006 0.013 0.147*
(0.0695) (0.0945) (0.0865)
Panel D—Ex ended speci ica ion wi h in e ac ion e ms
Full pe iod (N=1988)
Pa en wel a e, 00.170*** 0.199** 0.142**
(0.0603) (0.0877) (0.0724)
Pos ×Pa e nal wel a e, 0−0.031 −0.029 0.111
(0.0679) (0.0942) (0.0851)
No es: In Panels A and B each cell en y ep esen s a sepa a e eg ession whe e he pa en al wel a e
measu e ma ches he dependen a iable as lis ed in he column heade s (see Table 2). Robus s anda d
e o s a e epo ed in pa en heses. Fo de ails on he basic and ex ended speci ica ion see no es o Table 2.
In Panels C and D he p e- and pos - e o m obse a ions we e pooled and an in e ac ion e m o he
pa en al wel a e indica o wi h he pos - e o m indica o was added o he speci ica ion. ∗∗∗p<0.01;
∗∗p<0.05; ∗p<0.10.
Sou ce: SOEP (1984–2017), own calcula ions.
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1244
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
esul s o he i s wo wel a e ou comes yield ha he in e gene a ional co ela-
ion did no change subs an i ely a e he e o m. This pa e n is no suppo ed by
he ou comes epo ed in Panel B, whe e co ela ions declined in he pos - e o m
pe iod; howe e , he in e ac ion e m es ima es in Panels C and D a e imp ecise.
In con as , co ela ion pa e ns o he hi d ou come inc eased subs an ially a e
he e o m (see Panels A and B). This is con i med by he s a is ically signi ican
es ima e o he in e ac ion e m coe icien in Panel C. O e all, hese esul s a e
inconclusi e: we ind nei he s ong e idence o inc easing no o dec easing co e-
la ion pa e ns.
6.5. Rele ance o Media o Va iable: Child Educa ion
I is possible ha child educa ion ac s as a media o o he pa en –child con-
nec ion in wel a e eceip . I pa en al wel a e eceip nega i ely a ec s child educa-
ional a ainmen (e.g., ia ole-model e ec s, s igma iza ion in school, low pa en al
sel -es eem, o esiden ial ins abili y) hen low child human capi al, ha is, cogni i e
and possibly non-cogni i e skills, may limi labo ma ke oppo uni ies and e en-
ually economic independence. We can es whe he child educa ional a ainmen
is a media o by adding child educa ional ou comes as a con ol a iable in he
es ima ions shown in Table 2whe e hey had been omi ed so a o a oid endo-
genei y issues. I he in e gene a ional co ela ion declines once we condi ion on
child educa ion hen media ion e ec s a e likely which may poin o use ul policy
s a egies.
We conside ou indica o s o he child’s highes educa ional deg ee ob ained.
Table 6shows he es ima ed co ela ion pa e ns ha esul a e adding he child
educa ion con ols o he se o co a ia es in he basic and ex ended speci ica ions.
All coe icien es ima es con inue o be posi i e and highly s a is ically signi ican .
Howe e , in compa ison o he esul s in Table 2 hey a e smalle in magni ude by
abou 20 pe cen . Thus, a conside able pa o he in e gene a ional co ela ion
may ope a e ia a enua ed educa ional a ainmen o child en in wel a e- ecei ing
households. This ag ees well wi h he li e a u e (see Boschman e al., 2019 o
Bubonya & Cobb-Cla k, 2021).13
6.6. Causal Es ima ion App oaches—Family Fixed E ec s
The co ela ions in es iga ed so a canno gene ally unco e causal e ec s. To
ge close o causal e ec es ima ion we ake ad an age o siblings om he same
amily in he amily ixed e ec s model. This allows us o accoun o ime-cons an
amily unobse ables. I hese a e he only biasing ac o s hen he amily ixed
e ec s models p o ide causal e ec s (see he discussion in Sec ion 4).
Ou amily ixed e ec s sample o e s in o ma ion on 1161 siblings om 514
di e en amilies. Panel A o Table 7p esen s he baseline co ela ion es ima es
o he basic and ex ended speci ica ions o his pa icula subsample. The esul s
13Child educa ion di e s signi ican ly o he g oups wi h and wi hou pa en al wel a e eceip .
Those wi h pa en al wel a e eceip a e mo e han wice as likely o be in he lowes (sha es o 25 s. 12
pe cen ) and less han hal as likely o be in he highes ca ego y (12 s. 28 pe cen ).
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1245
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
TABLE 6
CONTROLLING FOR CHILD EDUCATION AS A POTENTIAL MEDIATOR
Dependen a iables: Wel a e eceip 1
(1) (2) (3)
E e (0/1) Numbe (y s) Sha e (%)
Pa en wel a e, 0(basic speci ica ion) 0.148*** 0.161*** 0.212***
(0.025) (0.031) (0.039)
Pa en wel a e, 0(ex ended speci ica ion) 0.112*** 0.129*** 0.180***
(0.025) (0.031) (0.038)
No es: Es ima ions used 2403 obse a ions. Each cell en y ep esen s a sepa a e eg ession whe e
he pa en al wel a e measu e ma ches he dependen a iable as lis ed in he columnheade s (seeTable 2).
Robus s anda d e o s a e epo ed in pa en heses. Fo de ails on he basic and ex ended speci ica ion
see no es o Table 2; all es ima ions addi ionally con ol o h ee indica o s o child educa ional a ain-
men . ∗∗∗p<0.01.
Sou ce: SOEP (1984–2017), own calcula ions.
TABLE 7
FAMILY FIXED EFFECTS ESTIMATION
Dependen a iables: Wel a e eceip 1
(1) (2) (3)
E e (0/1) Numbe (yea s) Sha e (%)
Panel A: OLS esul s o he FE Sample (N=1161)
Pa en wel a e (basic) 0.199*** 0.233*** 0.290***
(0.037) (0.044) (0.058)
Pa en wel a e (ex ended) 0.132*** 0.162*** 0.211***
(0.036) (0.043) (0.054)
Panel B: FE Reg essions (N=1161)
Pa en wel a e (basic) −0.007 −0.028 −0.150
(0.076) (0.087) (0.104)
Pa en wel a e (ex ended) 0.009 −0.031 −0.163
(0.78) (0.089) (0.115)
Panel C: FE Reg essions w/o wel a e expe ience o oldes child (N=1150)
Pa en wel a e (basic) 0.003 0.002 −0.155
(0.081) (0.086) (0.112)
Pa en wel a e (ex ended) 0.009 −0.031 −0.163
(0.078) (0.089) (0.115)
No es: Each cell en y ep esen s a sepa a e eg ession whe e he pa en al wel a e measu e ma ches
he dependen a iable as lis ed in he column heade s (see Table 2). Robus s anda d e o s a e epo ed
in pa en heses. Fo de ails on he basic and ex ended speci ica ion see no es o Table 2.∗∗∗p<0.01.
Sou ce: SOEP (1984–2017), own calcula ions.
a e a he simila o hose o he ull sample in Table 2.PanelBo Table7shows
he coe icien es ima es on pa en al wel a e when we apply he amily ixed e ec s
es ima o o bo h he basic and ex ended speci ica ions: he posi i e signi ican co -
ela ion coe icien s do no hold up o ixed e ec s con ols. This esul does no
suppo he exis ence o causal in e gene a ional ea men e ec s.
As a obus ness es , we show he ixed e ec s esul s when hose amilies a e
omi ed om he sibling sample whe e only he olde sibling expe ienced pa en al
wel a e eceip . In hese cases, he mechanisms ha gene a e in e gene a ional
ansmission such as educed s igma, a ailabili y o ins i u ional in o ma ion
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1246
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
migh pe sis in he amily e en hough a wel a e eceip is no obse ed o he
younge sibling. Panel C in Table 7yields ha he esul s ob ained so a , ha is,
no signi ican posi i e e ec s, a e obus o his addi ional es .14
The inding o no causal e ec s ag ees wi h he amily ixed e ec s es ima ions
o unemploymen bene i ansmissionin Ekhaugen(2009) andMuelle e al.(2017),
and o he ansmission o ma e nal (no pa e nal) bene i ansmission in B a -
be g e al. (2015). In hei ixed e ec s analyses, Edma k and Hanspe s (2015)
e en ob ained nega i e in e gene a ional ansmission esul s o wel a e eceip in
Sweden. The au ho s a gue ha ei he child en o wel a e ecipien s a e pa icula ly
eage o a oid wel a e o he coe icien s on pa en al wel a e eceip cap u e o he
di e ences be ween siblings ha co ela e wi h he wel a e expe ience.
6.7. Causal Es ima ion App oaches—Go schalk Es ima ion
As ou second app oach o app oxima e causal wel a e ansmission e ec s, we
apply he p ocedu e de eloped by Go schalk (1996) as cha ac e ized in Sec ion 4.3
abo e. He e, we accoun o amily-speci ic unobse ables ha migh o he wise
bias causal e ec es ima ion by con olling o pa en al wel a e eceip in he pe iod
a e obse ing he second gene a ion’s wel a e eceip . As we do no obse e hese
ou comes o all amilies he es ima ion wi h an addi ional pa en al wel a e con ol
can only be pe o med on a subsample. Appendix Table A.7 shows desc ip i e
s a is ics o he ull sample and he Go schalk subsample o whom in o ma ion
on la e pa en al wel a e eceip is a ailable. No su p isingly, indi iduals in he
Go schalk subsample and hei pa en s a e on a e age 2 yea s olde han he main
sample. The child gene a ion is insigni ican ly mo e likely o use wel a e (incidence
in 1o 14.3 s. 12.8 pe cen ) han he main sample whe eas he pa en s a e less
likely o use wel a e in 0. O e all, he subsample cha ac e is ics do no appea o
di e in impo an ways (signi ican age di e ences a e by cons uc ion). Nex , we
in es iga e whe he he co ela ion pa e ns in he Go schalk subsample e lec ou
esul s om Table 2.PanelAo Table8shows he basic and ex ended speci ica ion
es ima es o he Go schalk subsample. The coe icien es ima es do no di e in
impo an ways om p io esul s.
Panels B and C o Table 8show he es ima ion esul s o he ac ual Go schalk
es ima ion app oach o he basic and ex ended speci ica ions. Each indi idual
pa en al wel a e eceip indica o yields posi i e and signi ican coe icien es i-
ma es. The ow labeled “Go schalk e ec ” in each panel p esen s he di e ence
be ween he wo pa en al e ec s as es ima ed based on Equa ion (6). In no case
do we ob ain signi ican ly posi i e di e ences which would be indica i e o causal
14We pu sued wo s a egies o he obus ness es a e we de e mined he se o amilies whe e
pa en al “e e -wel a e” ou comes a ied ac oss siblings; ou o 1161 child en in 514 amilies, only 103
child en in 38 amilies had a ying pa en al ou comes ac oss siblings. Ou i s s a egy omi ed 49 obse -
a ions om amilies whe e al eady he i s -bo n child expe ienced pa en al wel a e eceip (es ima ing
wi h N=1112 obse a ions) because he one- ime expe ience may a ec he amily cha ac e is ics pe -
manen ly. In ou second s a egy, we omi ed obse a ions om amilies whe e only he i s -bo n child
expe ienced wel a e eceip (see Panel C o Table 7). The esul s ha dly di e ed be ween he s a egies.
Ne e heless, we canno exclude he possibili y ha amilies wi h ime- a ying bene i eceip di e in
unobse able cha ac e is ics om o he amilies.
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1247
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
TABLE 8
GOTTSCHALK ESTIMATION
Dependen a iables: Wel a e eceip 1
(1) (2) (3)
E e (0/1) Numbe (yea s) Sha e (%)
Panel A: OLS esul s o Go schalk es ima ion sample (N=1221)
Pa en wel a e, 0, basic speci ica ion 0.178*** 0.249*** 0.260***
(0.039) (0.058) (0.062)
Pa en wel a e, 0, ex ended speci ica ion 0.131*** 0.192*** 0.217***
(0.038) (0.057) (0.060)
Panel B: Go schalk app oach es ima ion esul s, basic speci ica ion (N=1221)
Pa en wel a e, 00.130*** 0.191*** 0.201***
(0.039) (0.060) (0.063)
Pa en wel a e, 20.318*** 0.350*** 0.188***
(0.072) (0.076) (0.046)
Go schalk e ec −0.188** −0.160 0.014
(0.089) (0.104) (0.084)
Panel C: Go schalk app oach es ima ion esul s, ex ended speci ica ion (N=1221)
Pa en wel a e, 00.097*** 0.149*** 0.172***
(0.039) (0.059) (0.061)
Pa en wel a e, 20.254*** 0.271*** 0.156***
(0.073) (0.080) (0.046)
Go schalk e ec −0.157** −0.122 0.015
(0.089) (0.108) (0.083)
No es: In Panel A, each cell en y ep esen s a sepa a e eg ession whe e he pa en al wel a e mea-
su e ma ches he dependen a iable as lis ed in he column heade s (see Table 2). In Panels B and C
he pa en indica o s o pe iods 0and 2a e con olled join ly in he same eg ession model. The ows
labelled “Go schalk e ec ” p esen he di e ence be ween he wo pe iod-speci ic coe icien es ima es.
Robus s anda d e o s a e epo ed in pa en heses. Fo de ails on he basic and ex ended speci ica ion
see no es o Table 2.∗∗∗p<0.01; ∗∗p<0.05.
Sou ce: SOEP (1984–2017), own calcula ions.
in e gene a ional ansmission e ec s. The e o e, he inding o a lack o causal
ansmission om he ixed e ec s es ima ions is con i med wi h he Go schalk
app oach. In ac , he o e all e ec s on he incidence o wel a e eceip e en u n
ou signi ican ly nega i e.
While he nega i e o al e ec is su p ising, i e lec s he indings o o he
au ho s who s udied possibly he e ogeneous policy p og ams in di e en coun ies:
Ekhaugen (2009) and Muelle e al. (2017) simila ly ob ained nega i e es ima es o
he ansmission o unemploymen bene i s. Also, Boschman e al. (2019) ind nega-
i e e ec s o social assis ance and disabili y p og ams. These au ho s a gue ha i
is no he expe ience o he wel a e p og am i sel ha causes he nex gene a ion’s
pa icipa ion. Ins ead, amily-speci ic cha ac e is ics such as no ms and a i udes
ha a eno a ached o ac ually ecei ing he bene i may d i e hein e gene a ional
co ela ions. The same pa e ns appea o hold o ou da a.
7. CONCLUSIONS
The in e na ional li e a u e discusses whe he expe iencing pa en al wel a e
eceip in childhood o adolescence is co ela ed wi h and causally de e mines own
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1248
Re iew o Income and Weal h, Se ies 70, Numbe 4, Decembe 2024
wel a e eceip la e in li e. This is an impo an policy ques ion because in e gen-
e a ional ansmission o wel a e eceip indica es a ailu e o wel a e p og ams:
go e nmen suppo does no succeed in li ing amilies ou o po e y and may
e en impose nega i e ex e nali ies on he nex gene a ion.
We ake ad an age o a long- unning household panel (SOEP) su ey o s udy
he in e gene a ional ansmission o wel a e eceip o he case o Ge many. Com-
pa a i e esea ch sugges s ha Ge many o e s a ela i ely gene ous wel a e sys em;
howe e , i s in e gene a ional cha ac e is ics ha e no been in es iga ed be o e. The
ichness o ou da a allows us o add in o ma i e analyses o in e gene a ional co -
ela ion pa e ns o he li e a u e.
We conside h ee wel a e indica o s and ind s ong in e gene a ional co ela-
ion pa e ns. The co ela ions a e la ge o ecen han o olde bi h coho s and
o emales han o males. We do no ind impo an di e ences in he ansmission
o wel a e om a he s s. mo he s. Exposu e o pa en al wel a e eceip a he ages
o 10–12 and 16–18 yields s onge co ela ion pa e ns han exposu e in he 13–15
age window. Child educa ional a ainmen appea s o be a media o be ween pa en
and child wel a e eceip which may o e an oppo uni y o policy in e en ions.
We use amily ixed e ec s and he Go schalk (1996) me hod o go beyond
co ela ion analyses and o iden i y causal e ec s o pa en al wel a e eceip . Bo h
s a egies iden i y causal e ec s unde speci ic, ye di e en assump ions and he e-
o e complemen each o he . In e es ingly, bo h s a egies ail o ind e idence o a
causal impac o pa en al wel a e eceip on child wel a e ou comes. Thus, we do no
ind e idence ha i is he expe ience o pa en al wel a e eceip i sel and a “wel a e
cul u e” (Dahl e al., 2014) ha d i es subsequen child wel a e eceip . Ins ead, he
co ela ion o indi idual cha ac e is ics and ci cums ances in he child and pa en
household seems o de e mine ansmission pa e ns. This sugges s ha i is no
he cha ac e o wel a e ins i u ions hemsel es ha lea es he o sp ing o wel a e
ecipien s a an ele a ed isk o wel a e eceip . This is a highly policy- ele an ind-
ing. I cla i ies ha any ini ia i e o educe in e gene a ional co ela ion in wel a e
eceip mus no ocus on he ins i u ions o he wel a e sys em bu mo e plausibly
on cha ac e is ics a he indi idual and household le el and, o example, imp o e
human capi al, heal h, and labo ma ke engagemen .
Ou conclusions a e subjec o s ong iden i ying assump ions and should
be ein es iga ed when la ge samples a e a ailable. I seems wo hwhile o di ec
u u e esea ch o s udy he de e minan s and ele ance o you h educa ional
a ainmen , which migh be malleable by public policy. Also, i is impo an o
e-analyze any changes in in e gene a ional co ela ion a e he e o m o he
wel a e sys em. Finally, we ag ee wi h Ha ley e al. (2022), who poin ou ha in
a si ua ion o low bene i ake-up, in e gene a ional spillo e s and co ela ions can
be a good hing i hey educe non- ake-up.
REFERENCES
Ade mon, A., Lindahl, M., & Palme, M. (2021). Dynas ic human capi al, inequali y, and in e gene a-
ional mobili y. Ame ican Economic Re iew,111, 1523–48.
Angelini, V., Be oni, M., & Co azzini, L. (2018). Does pa e nal unemploymen a ec young adul
o sp ing’s pe sonali y? Jou nal o Human Capi al,12, 542–67.
© 2024 The Au ho (s). Re iew o Income and Weal h published by John Wiley & Sons L d on behal o
In e na ional Associa ion o Resea ch in Income and Weal h.
1249
