Fi ms and he In e gene a ional T ansmission
o Labo Ma ke Ad an age
Pe Engzell, Na han Wilme s
Documen ype
Pos -p in : This manusc ip has passed pee e iew and been accep ed o publica ion by a jou nal.
I may include inal edi s by he au ho (s) bu no edi ing o o ma ing by he publishe .
Funding in o ma ion
This esea ch was unded by he Eu opean Resea ch Council, G an Ag eemen No. 101165962
Ma ke s and Mobili y: How Employe s S uc u e Economic Oppo uni y.
Sugges ed ci a ion
Engzell, Pe , and Na han Wilme s. (2025). Fi ms and he In e gene a ional T ansmission o Labo
Ma ke Ad an age. Ame ican Jou nal o Sociology, o hcoming.
Da e o eco d
Sep embe 10, 2025.
Te ms o use
This wo k is licensed unde a C ea i e Commons A ibu ion-NonComme cial-NoDe i a i es 4.0
In e na ional license: h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
Fi ms and he In e gene a ional T ansmission o
Labo Ma ke Ad an age
Pe Engzell
Uni e si y College London
Na han Wilme s
MIT Sloan
July 2025
Abs ac
Pay inequali y s ems bo h om diffe ences in wo ke s’ indi idual cha ac-
e is ics and om i m pay-se ing. Some i ms cap u e mo e alue o pass on
a la ge sha e o su plus o wo ke s. Ye , in e gene a ional mobili y esea ch
ocuses on ansmission o indi idual ai s and has neglec ed he ques ion
o how i ms shape he inhe i ance o inequali y. We use h ee decades o
Swedish popula ion egis e da a o decompose he in e gene a ional ea n-
ings co ela ion in o i m pay p emiums and wo ke effec s. One qua e o
he in e gene a ional ea nings co ela ion a midli e is explained by so ing be-
ween i ms wi h unequal pay. Employe inhe i ance accoun s o a small sha e
o his i m-based ea nings ansmission. Ins ead, high-educa ion and high-
occupa ion wo ke s disp opo iona ely wo k a high-paying i ms. Pa en al
e e al ne wo ks and he inhe i ance o indus y and labo ma ke con ex
play a supplemen a y ole. As wo ke s wi h high-educa ion o high-s a us
jobs inc easingly bene i om high paying i ms, i m p emiums cons i u e a
cen al mechanism h ough which collec i e p ocesses d i e in e gene a ional
ea nings ansmission.
Acknowledgemen s: This wo k ecei ed suppo om he Eu opean Resea ch Council g an no.
101165962 (MaMo), Russell Sage Founda ion g an no. R-2111-34876, Swedish Resea ch Council
o Heal h, Wo king Li e, and Wel a e (Fo e) g an no. 2016-07099, Swedish Resea ch Council
g an s no. 2019-02552 and 2022-02314, and he F ench Na ional Resea ch Agency (ANR) as pa
o he “In es issemen s d’A eni ” p og am LIEPP (ANR-11-LABX-0091, ANR-11-IDEX-0005-02)
and he Uni e si é Pa is Ci é IdEx (ANR-18-IDEX-0001). Richa d B een p o ided in aluable
help a an ea ly s age o he p ojec . O he many people who p o ided eedback, Ma in Häll-
s en, Max Thaning, and Wou e Zwysen dese e special men ion. Fo u he commen s, we
hank se en anonymous e iewe s and nume ous audience membe s a he RC28 Sp ing Mee -
ing in Tu ku, 2021, he Eu opean Conso ium o Sociological Resea ch Annual Con e ence in
Ams e dam, 2022, he Ame ican Sociological Associa ion Annual Mee ing in Los Angeles, 2022,
and semina s and wo kshops a Bocconi Uni e si y, Co nell Uni e si y, Eu opean Uni e si y In-
s i u e, Impe ial College London, Ins i u e o Fiscal S udies, Massachuse s Ins i u e o Tech-
nology, Ox o d Le e hulme Cen e o Demog aphic Science, Rockwool Founda ion, Sciences Po,
Swedish Ins i u e o Social Resea ch, Uni e si y o Kons anz, Uni e si y o Lausanne, Uni e -
si y o Wisconsin–Madison, and Queen Ma y Uni e si y o London. Di ec co espondence o
Pe Engzell, UCL Social Resea ch Ins i u e, 55-59 Go don Squa e, London WC1H 0NU. Email:
[email p o ec ed].
1
1In oduc ion
In capi alis economies, alue is c ea ed and alloca ed la gely h ough collec i e ac-
ion wi hin i ms. P oduc i e collabo a ion among co-wo ke s can aise e enue and
po en ial pay abo e wha wo ke s’ indi idual cha ac e is ics imply. Monopsonis ic
employe s, in con as , can supp ess pay below wo ke s’ ma ginal p oduc . Effec i e
collec i e ba gaining by wo ke s a one i m can boos hei sha e o su plus abo e
ha ecei ed by simila wo ke s in a diffe en i m. Fi ms a e hus a c i ical ehicle
h ough which collec i e p ocesses, i educible o indi idual ai s, ansla e in o
unequal ou comes.
Ye , a hal cen u y o esea ch on in e gene a ional mobili y has la gely ig-
no ed he ole o i ms and employe s. Ins ead, esea che s a gue ha social o igin
in luences oppo uni y ia las ing effec s on indi idual wo ke cha ac e is ics: Pa -
en s wi h g ea e esou ces a e able o in es in hei child en in nume ous ways, and
cul i a e ai s in hem such as skill, ag eeableness, o wo k o ien a ion, ha a e
alued in he labo ma ke (Black and De e eux, 2011; Bowles e al., 2009; Fa kas,
2003; Jæge and B een, 2016; Jencks e al., 1979; Mo gan e al., 2006; Sewell e al.,
1969). This s anda d app oach es s on he assump ion ha indi idual cha ac e -
is ics pay offin a compe i i e labo ma ke .
A g owing body o esea ch challenges his assump ion by s udying i ms as
asi eo eme gen inequali yi educible oma ke -p icedindi idual ai s.Wo k-
ing a a high- o low-paying i m affec s ea nings beyond wha would be expec ed
om indi idual wo ke cha ac e is ics (Abowd e al., 1999; G oshen, 1991). Recen
changes in he labo ma ke sugges ha hese i m effec s may play a p ominen
ole in in e gene a ional ea nings ansmission. Be ween- i m inequali y has g own
ac oss many de eloped coun ies (Tomasko ic-De ey e al., 2020). Mo eo e , oc-
cupa ion and wo ke cha ac e is ics a e inc easingly co ela ed wi h wo king a a
high-paying i m (Ca d e al., 2013; C iscuolo e al., 2020; Wilme s and Aeppli,
2021). This consolida ion o po en ially independen dimensions o inequali y—
i m and occupa ion o educa ion—means ha e en unde s anding how indi idual
ai s affec in e gene a ional ea nings ansmission equi es s udying i ms.
We examine he ole o i ms in s a i ica ion in se e al ways, beginning wi h a
undamen al ques ion: how much do i m diffe ences in pay con ibu e o in e gen-
e a ional ea nings pe sis ence? We hen go on o answe a se o ques ions on how
hese i m p emiums a ise and how hey bene i diffe en wo ke s. Do wo ke s om
diffe en social backg ounds bene i equally om wo king a high-paying i ms? Is
i m-based ansmission mos ly ele an in he ea ly ca ee o does i ca y las ing
ad an ages? Do p i ileged wo ke s so in o i ms wi h highe p o i s, i ms ha
pass on a g ea e sha e o employees, o bo h? How does s a i ica ion by i ms map
on o o he , mo e commonly s udied dimensions such as educa ion and occupa ion?
And las ly, wha helps wo ke s o p i ileged backg ound so in o high-paying i ms?
To answe hese ques ions, we b ing me hods om esea ch on labo ma ke
inequali y o he s udy o in e gene a ional mobili y (Abowd e al., 1999). Empi -
ically, we d aw on popula ion-wide linked employe -employee da a om Sweden,
which offe a sufficien ly long ime span o obse e bo h pa en s and child en a
p ime wo king age. By ollowing wo ke s ac oss employe s o e ime, we dis inguish
2
he componen s o ea nings a ibu able o wo sou ces: (a) he i m one wo ks a
and (b) du able indi idual wo ke ai s. The i m componen is he p emium asso-
cia ed wi h wo king a a gi en i m, condi ional on he composi ion o i s wo ke s.
The indi idual ea nings componen e lec s wo ke ai s ha a e consis en ly e-
wa ded ac oss i ms. We nes his app oach in a pa h model o p ecisely quan i y
he con ibu ion o i m pay p emiums o he in e gene a ional ea nings co ela ion.
In addi ional analyses, we es mechanisms by s udying how educa ion, occupa ion,
pa en al co-wo king ne wo ks, and s uc u al labo ma ke posi ions media e i m
so ing.
Ou app oach unco e s se e al new ac s abou he ole o employe s in in e -
gene a ional ea nings pe sis ence. Inequali y in i m p emiums by social backg ound
is e iden a labo ma ke en y and pe sis s h oughou he ca ee . The i m p e-
mium among child en om p i ileged backg ounds is ela i ely cons an in absolu e
e ms, bu accoun s o a sh inking ela i e pa o ea nings ansmission as indi-
idual ea nings an ou o e he ca ee . Ye , e en in ma u e ea nings, i m-based
so ing accoun s o a qua e o in e gene a ional pe sis ence. This i m p emium
ad an age enjoyed by high ea ning-backg ound wo ke s is a unc ion o wo king
bo h a i ms ha ha e a highe alue-added pe wo ke and a i ms ha pass
a la ge sha e o alue-added in o wo ke ea nings. Mo eo e , i is access o hese
i ms ha ma e s: he ea nings boos om wo king a a high-paying i m is simila
o wo ke s om high and low pa en al ea nings backg ounds.
Wha accoun s o hese pa e ns? We show ha educa ion and occupa ion
accoun o oughly wo- hi ds o i m ad an ages by social backg ound. Mos o he
emaining hi d is accoun ed o in equal pa s by e e al ne wo ks om a pa en
and inhe i ance o indus y and local labo ma ke . Fi ms a e an especially po en
explana ion o he pa o income inhe i ance no accoun ed o by educa ion o
occupa ion: hal o he di ec effec o pa en on child ea nings, and mos o he
class o igin gap in ea nings, is explained by so ing ac oss high- and low-paying
i ms. We conclude ha i m p emiums a e an impo an and unde s udied channel
in he in e gene a ional ansmission o ad an age.
2Fi m p emiums and s a i ica ion
Con empo a y employmen has bi u ca ed by i m. Supe s a s in echnology and
p o essional se ices offe s agge ing compensa ion, while ou sou ced con ac o s,
sh inking manu ac u e s, and la ge e aile s pay a less. I in an ea lie e a, Fo d
Mo o Company offe ed ela i ely high pay o p oduc ion wo ke s, now i is la gely
manage s and p o essionals, a companies like Google and Goldman Sachs, ha
bene i om employmen a high-paying employe s (Wilme s and Aeppli, 2021).
This bi u ca ion implies inequali y in i m pay p emiums: du able, i m-wide
diffe ences in pay ha affec wo ke s’ ea nings. Some i ms a e high-paying, and
consis en ly pay hei wo ke s mo e ac oss job i les (G oshen, 1991). Fi m pay
p emiums a e pe sis en o e ime (Engbom e al., 2023) and appea e en when
compa ing i ms wi h simila occupa ional composi ion (G oshen, 1991; Wilme s
and Aeppli, 2021) and employing wo ke s o simila skill le els (Song e al., 2019).
3
These p emiums a e quan i a i ely impo an o wo ke ea nings, accoun ing o
15% o 30% o ea nings inequali y in de eloped coun ies (C iscuolo e al., 2020;
Tomasko ic-De ey e al., 2020).
Inco po a ing pay diffe ences be ween i ms equi es upda ing heo ies o s a -
i ica ion, jus as p io mobili y esea ch has epea edly inco po a ed newly ascen-
dan sou ces o inequali y. Ma x a gued ha owne ship o he means o p oduc ion,
a he han inhe i ed i le, inc easingly de e mined he dis ibu ion o economic e-
sou ces. Blau and Duncan (1967) coun e ed ha he sepa a ion o owne ship om
con ol equi ed a en ion o g ada ional diffe ences be ween employees o diffe -
en au ho i y. Gold ho pe (1982) a gued ha he au onomy o much office wo k
os e ed a mo e du able “se ice ela ion” dis inc om he labo con ac , while
Weeden and G usky (2005) unde lined how co e class mechanisms such as closu e
and solida i y can ope a e a he le el o de ailed occupa ions. These shi s in s a -
i ica ion esea ch ha e hus acked he ise o capi alis s; he manage ial e olu ion;
he pos -indus ial ansi ion; and he subsequen ac u ing o class iden i ies. I
in con empo a y labo ma ke s, de ined by supe s a i ms and skill seg ega ion,
economic success equi es wo king a a high-paying i m, how does his affec in e -
gene a ional mobili y?
We heo ize how i m p emiums affec s a i ica ion in a wo-s age model.
Fi s , i ms accumula e esou ces, ei he h ough ma ke powe o h ough he e -
ec i e combina ion o ac o inpu s. Second, wo ke s and o he s akeholde s ba gain
o e he alloca ion o hese esou ces (Tomasko ic-De ey and A en -Hol , 2019). In
bo h o hese s ages, p i ileged wo ke s can disp opo iona ely eap he bene i s o
i m p emiums.
In he i s s age, i ms a e a dis inc basis o inequali y—compa ed o capi al
owne ship, educa ion le el, o occupa ional membe ship—in ha hey a e di ec
ehicles o collec i e ac ion, a he han an indi idual’s endowmen . A g oup o
doc o s may coo dina e o es ic he supply o medical g adua es, as a lobbying
effo ou side o hei usual wo k, o inc ease he alue o hei skills (Weeden, 2002).
Bu a g oup o co-wo ke s coo dina es in i ue o being co-wo ke s. Fi ms o m o
acili a e p oduc i e ac i i y ha imposes excessi ely high ansac ion cos s when
pe o med as a ma ke -media ed nexus o con ac s (Williamson, 1975). As such, in
a i s s age, i mscanmo eo lesssuccess ullyde elopama ke nicheo cul i a e
efficien p oduc ion p ac ices ha allow hem o accumula e esou ces.
Wo ke s’ socio-economic backg ound can ma e in se e al ways o his e-
sou ce accumula ion s age. Mos di ec ly, class solida i y can educe co-wo ke s’
emp a ion o shi k o o he wise unde mine a i m’s pe o mance. As Bu awoy and
W igh (1990, p. 265) a gue, “ i ms wi hin which wo ke s can effec i ely educe
ee iding will be he mos p o i able”. Indeed, his insigh mo i a es ex ensi e
esea ch on ways ha o ganiza ional ci izenship beha io boos s i m pe o mance
(Podsakoffand MacKenzie, 2014). Class coope a ion in moni o ing and disciplining
co-wo ke s can gi e owne s and manage s incen i e o ga he wo ke s o a common
class backg ound. This homogenei y dynamic can bene i high-ea nings backg ound
wo ke s, as when hedge und employees om p i ileged backg ounds ely on “ne -
wo ks o us and loyal y” o blun inancial isks (Neely, 2022, p. 21). Bu i
4
can also bene i less p i ileged wo ke s, as in blue colla union o ganizing (Wilme s,
2019) o e hnic encla es (Po es and Sensenb enne , 1993).
Beyond solida i y and moni o ing, posi i e pee effec s and human capi al
complemen a i ies can aise p oduc i i y o eams o high-educa ion and high-
s a us occupa ion wo ke s, o e and abo e wha indi idual wo ke s on hose eams
could p oduce in o he con ex s (K eme , 1993; Song e al., 2019; Wilme s and
Aeppli, 2021). Inso a as wo ke s o p i ileged backg ounds ob ain be e educa ion
and highe s a us occupa ions, hey a e mos likely o bene i om his addi ional
p oduc i i y boos . Finally, e en i a i m’s esou ce ga he ing ad an age is en i ely
exogenous o he cha ac e is ics o i s wo ke s—a monopolis ic posi ion in a p oduc
ma ke o example, o a luc a i e pa en (Kline e al., 2019)—p i ileged wo ke s
may use ne wo ks and s a egic knowledge o effec i ely ob ain posi ions wi hin
p o i able i ms. By s a egic knowledge, we mean awa eness o hese high- esou ce
i ms and hei selec ion and hi ing p ocesses (Ri e a, 2012).
All o hese i s -s age esou ce accumula ion p ocesses inc ease po en ial ea n-
ings o wo ke s. Howe e , he ewa ds o wo ke s a a highly p oduc i e i m a e
only po en ial (Fligs ein and Fe nandez, 1988). In a second s age, i ms hen allo-
ca e ewa ds ac oss s akeholde s. He e, solida i y again ma e s. High class wo ke s
may ba gain mo e effec i ely, in oking hei educa ional and occupa ional c eden-
ials o demand a la ge sha e o a i m’s esou ces. This would mean ha i ms
ha employ p i ileged wo ke s mus pay la ge i m p emiums. In con as , i ms
ha employ ulne able wo ke s could pay less, e en exe cising monopsonis ic powe
o push wages below wo ke s’ ma ginal p oduc i i y. These second s age p ocesses
look mo e like con en ional social closu e (Mu phy, 1988; Pa kin, 1979)—a ze o-
sum ba gaining se ing, in which wo ke s y o cap u e su plus om a p incipal,
o en by es ic ing labo supply.
While in he abs ac , hese wo s ages o i m p emium de elopmen need
no deli e ad an age disp opo iona ely o p i ileged wo ke s, ecen esea ch p o-
ides sugges i e e idence ha con empo a y i m p emiums a e bene i ing p i i-
leged wo ke s. In he mid- wen ie h cen u y, i m p emiums we e associa ed wi h
blue-colla employe s, o en la ge and unionized (Cobb and Lin, 2017; Wilme s,
2019). These ela i ely high-paying jobs o non-college wo ke s p o ided gene a-
ional secu i y o low-ea ning amilies: union ca ds and jobs in speci ic plan s we e
passed om a he s and uncles o sons and nephews (B yson and Da ies, 2019).
Simila ly, e hnic labo ma ke niches in la ge ci ies p o ec ed secu e public sec-
o o cons uc ion jobs o pa icula immig an g oups (Waldinge , 1999). The
monopoliza ion o hese i m p emiums can p o ide highe ea nings o ela i ely
low-educa ion and -s a us backg ound wo ke s.1
Since he 1970s howe e , high-paying blue colla employe s ha e aced in-
c easing cos p essu e ha has unde mined hei i m p emiums (Sø ensen, 2000;
Wilme s, 2018). Howe e , i m p emiums ha e no he eby disappea ed. On
1In he language o Pa kin (1979), his would amoun o “usu pa iona y” social closu e, which
in ol es he effo s o subo dina e g oups seeking o gain access o esou ces, oppo uni ies, o
posi ions p e iously monopolized by a dominan g oup. The p e ailing mode o social closu e is
“exclusiona y,” which in ol es he effo s o a dominan g oup o es ic access o esou ces and
main ain hei p i ileged s a us and powe .
5
he con a y, in ecen decades i m p emiums ha e ei he g own in impo ance
(C iscuolo e al., 2020) o emained s able (Song e al., 2019) ac oss mos de el-
oped coun ies. Bu , he e has been a ealloca ion o i m p emiums om blue-colla
wo ke s owa d whi e-colla and p o essional employees (C iscuolo e al., 2020; Song
e al., 2019; Wilme s and Aeppli, 2021). In he place o high-paying blue colla em-
ploye s, he ise o inance (K ippne , 2011), supe s a echnology i ms (Au o
e al., 2020), and high-paying sole-p op ie o ships (Smi h e al., 2019) ha e mean
aqui ediffe en indus y and occupa ional composi ion o high-paying i ms. These
new sou ces o i m p emiums a e o en ma ked by c eden ialism and social exclu-
sion, as in inance (Neely, 2022), o p o ide oppo uni y o di ec ansmission o
high-paying posi ions, as in amily inhe i ance o sole-p op ie o ships (Co ak and
Pi aino, 2011). Recen changes in he sou ces o i m p emiums ac oss indus ies
and occupa ions he e o e make hem pa icula ly likely o bene i high-income
backg ound wo ke s.
3Limi s o s a us a ainmen and human capi al
models
Theo izing hese dynamics linking i m p emiums o wo ke s’ class backg ound e-
sponds o longs anding calls o inco po a ing i ms and labo ma ke s in o mod-
els o s a i ica ion (Ba on and Bielby, 1980; Spile man, 1977; S olzenbe g, 1978).
While ea ly models o s a us a ainmen posi ed he di ec in ol emen o pa en s in
inding a job (Blau and Duncan, 1967), his ocus was soon eplaced by psycholog-
ical in luences on abili y, aspi a ions and, ul ima ely, schooling (Sewell e al., 1969;
Sewell and Hause , 1975). The esul ing model aced epea ed c i icism o i s one-
sided ocus on he indi idual (Ba on, 1984; Fea he man, 1981). Ye , a compelling
al e na i e ailed o eme ge and he s a us a ainmen model emains, implici ly o
explici ly, he p ima y backd op o s a i ica ion esea ch. The ield oday he e-
o e emains p eoccupied wi h educa ion and o he o ms o human capi al (Bloome
e al., 2018; B een and Mülle , 2020). This is no wi hou eason, as educa ion is
bo h he main condui o social ep oduc ion and a ehicle o in e gene a ional
mobili y (Hou , 2012; To che, 2015).
Howe e , ecen esea ch emphasizes ha human capi al is no he only sou ce
o ad an age o p i ileged child en (Be na di and Balla ino, 2016; Wi e een and
A ewell, 2017, 2020). Among UK manage s and p o essionals wi h simila edu-
ca ional backg ound, high-class child en end up highe -paid, in pa by so ing o
la ge and mo e cen ally loca ed i ms (Lau ison and F iedman, 2016). In he
US, while educa ion accoun s o he bulk o occupa ional s a us ansmission, i
only accoun s o abou hal o he in e gene a ional pe sis ence in amily income
(To che, 2016). E en he classical inding ha high-skilled labo ma ke s ope a e
mo e me i oc a ically is weake when income is s udied, and when accoun ing o
selec ion p ocesses in o educa ion (Fiel, 2020; Zhou, 2019).
These indings highligh he impo ance o unde s anding p ocesses ha ake
place o e and abo e educa ion in s udying in e gene a ional ea nings ansmis-
sion. Labo economics p o ides a ool o do so h ough he use o linked employe -
6
employee da a (Abowd e al., 1999; Song e al., 2019). Obse ing he same indi id-
uals wo king ac oss mul iple i ms, and he same i ms employing successi e indi-
iduals, e eals ha some i ms consis en ly p o ide highe pay o he same kind
o wo k. Con e sely, some wo ke s consis en ly ea n a high sala y no ma e whe e
hey wo k. These wo aspec s allow us o iden i y a pa o ea nings a ibu able
o i m pay-se ing, and one a ibu able o du able indi idual cha ac e is ics.
Applica ions o his model in economics ha e ended o offe an in e p e a ion
o hese wo componen s in e ms o i m p oduc i i y, on he one hand, and wo ke
human capi al o “abili y” on he o he (Ca d e al., 2018). We a e ca e ul o a oid
his in e p e a ion. As we ha e a gued, he i m-speci ic componen o pay e lec s a
complex combina ion o ma ke powe , complemen a i ies among ac o inpu s, and
con es ed p o i alloca ion decisions. Simila ly, he indi idual componen e lec s
any hing ha helps he wo ke gain a pe sis en ad an age on he co-wo ke s, in-
cluding human capi al bu also pe sonal connec ions, cul u al capi al, en i lemen ,
o s a us cha ac e is ics (Co ell e al., 2017; Khan, 2010).2
Fo hese easons, pay diffe ences wi hin and be ween i ms do no map on o
asimpledis inc ionbe weenme i oc a icandnon-me i oc a icpay.Ne e heless,
we show ha bo h he ole o educa ion, and he effec s o social backg ound o e
and abo e educa ion, can be cla i ied by a en ion o pay diffe ences be ween i ms.
Doing so highligh s ha educa ion, a om simply con eying human capi al, also
se es as a powe ul so ing de ice in ma ching o employe s—in line wi h longs and-
ing sociological accoun s (Bills, 2003; Collins, 1979; Domina e al., 2017). Be o e
u ning o ou quan i ica ion o he ole o i m p emiums in in e gene a ional ea n-
ings ansmission, we dis inguish speci ic mechanisms h ough which i m p emiums
may be dis ibu ed unequally among wo ke s o diffe en backg ounds.
4Fi m p emium mechanisms o in e gene a ional
ansmission
The e a e se e al links ha ie high-ea ning amily backg ound wo ke s o high-
paying i ms. Fi s , p i ileged wo ke s can ob ain c eden ials and occupa ional
s anding ha ei he suppo i m p emiums o gain wo ke s access o high-paying
i ms. Second, high-p emium i ms and indus ies can be ob ained h ough pa en al
e e al and co-wo ke ne wo k ies, as a ype o social capi al. Las ly, child en may
wo k o simila employe s o hei pa en s, e en in he absence o ac i a ed ne wo k
ies, simply due o a common oppo uni y s uc u e. In he ollowing, we elabo a e
hese mechanisms, be o e es ing hem empi ically.
4.1 Educa ion- and occupa ion-d i en i m ad an ages
P io s a i ica ion esea ch has demons a ed ha child en o high-ea ning pa -
en s ha e mo e educa ion and wo k in highe -s a us occupa ions han child en o
2Swi (2005) a gues ha ai s ha a e “cons i u i e o he pe son” can se e as a legi ima e
basis o diffe en ial ewa ds in he labo ma ke . While ou decomposi ion is well-sui ed o iden i y
such ai s, we emain less op imis ic abou he e hical signi icance o his dis inc ion.
7
lowe -ea ning pa en s. Wha does his mean o i m p emiums? Concen a ed em-
ploymen o high-educa ion and high-s a us occupa ions can i sel ancho i m p e-
miums. In O-Ring models o p oduc ion, complemen a i ies among high-pe o ming
wo ke s can d i e highe p oduc i i y and pay han any indi idual wo ke would
ecei e in o he con ex s (Co nelissen e al., 2017; K eme , 1993). Beyond pu e p o-
duc i i y effec s, concen a ed employmen o posi ionally powe ul wo ke s can in-
c ease collec i e ba gaining powe (Tomasko ic-De ey and A en -Hol , 2019). No e
ha hese effec s, while a unc ion o educa ion, a e dis inc om an indi idual
wo ke ’s pay-off o human capi al. Eme gen p oduc i i y and ba gaining powe
he e bene i s g oups o highly educa ed and high-s a us wo ke s o e and abo e
hei ou side op ions as indi iduals compe ing on he labo ma ke .
Occupa ional and educa ional ad an ages can also so wo ke s o be e i ms
(Hi sch, 1977; Thu ow, 1975). One US analysis inds ha a qua e o he highe
educa ion pay p emium is a ibu able o so ing ac oss high-paying i ms (Eng-
bom and Mose , 2017). Indeed, a he highe end o he ins i u ional hie a chy, a
small numbe o uni e si ies se e as pipelines o eli e employe s (Binde e al.,
2016; Bühlmann e al., 2022; Ri e a, 2016; Weins ein, 2018). In he Scandina ian
con ex we s udy, ield o s udy can be a mo e powe ul sou ce o s a i ica ion han
ins i u ion (Bo gen and Mas ekaasa, 2018; Hälls en, 2010; Hälls en and Thaning,
2018). This pa hway can ou e high-ea ning backg ound wo ke s o high p emium
i ms.
Beyond le els o e en ields o educa ion, child en o high-ea ning pa en s
may cul i a e cul u al disposi ions ha gain he con idence o ga ekeepe s a high-
p emium i ms (A ms ong and Hamil on, 2013). As long as class-based disposi ions
a e domain gene al, like ease and en i lemen , hey can bo h bene i child en as in-
di idual ai s and in he i m so ing p ocess (Bou dieu, 1984; Khan, 2010; La eau,
2011). Indeed, ecen quali a i e s udies o eli e employe s all emphasize he im-
po ance o aci cul u al pe o mance o ge ing in and s aying employed. A s udy
o accoun ing, b oadcas ing and a chi ec u e i ms ound ha beha io al codes, o -
en lea ned in eli e educa ional en i onmen s, ga e high-class backg ound wo ke s
an ad an age (F iedman and Lau ison, 2020). In hedge unds, uppe -class men
succeed by embodying s e eo ypes abou high isk- ole ance and us wo hiness
(Neely, 2022).
Consis en wi h his mechanism, a con empo aneous s udy o Is ael inds ha
high-ea ning backg ound child en disp opo iona ely so o high-paying i ms and
ha his so ing is la gely due o hose child en’s indi idual ai s (Dobbin and
Zoha , 2023). Th ough bo h alue c ea ion and cap u e, p i ileged wo ke s’ edu-
ca ional and occupa ional ad an ages can deli e hem disp opo iona ely o high
p emium i ms.
4.2 Job e e al ne wo ks and biased so ing
Social ne wo ks a e a ubiqui ous sou ce o inequali y (Bou dieu, 2000; DiMaggio
and Ga ip, 2012; Lin, 2002; Small, 2009) and he job sea ch p ocess is no excep ion
(Cas illa e al., 2013; G ano e e , 1974; Ma sden and Go man, 2001; T imble and
Kmec, 2011). While much esea ch has ocused on “weak ies,” acquain ances wi h
8
6.2 Pa h model
Ha ing es ima ed he sepa a e i m and indi idual wo ke con ibu ions o ea nings,
we explo e ansmission in a pa h model. Does in e gene a ional ea nings ans-
mission wo k h ough he inhe i ance o indi idual ai s as assumed in adi ional
in e gene a ional mobili y esea ch? O is a signi ican pa o in e gene a ional
ad an age ins ead due o employmen a a high-paying i m? Figu e 1 displays his
model, whe e he supe sc ip ⇤deno es he pa en al gene a ion. We model ea nings
as a unc ion o i m and wo ke componen s in each gene a ion:
Y⇤
i=a1F⇤
i+b1W⇤
i+✏1i,(3)
Yi =a2 Fi +b2 Wi +✏2i ,(4)
whe e Y⇤
iis he a he ’s a e age ea nings om 1990–1999, Yi is he child’s
ea nings a age ,and he igh -handside a iablesa e heco esponding i m
(F⇤
i,F
i )andwo ke (W⇤
i,W
i ) componen s. We a e in e es ed in he ollowing
pa hways:
Fi m- o- i m: Y⇤
i F⇤
i!Fi !Yi
Wo ke - o- i m: Y⇤
i W⇤
i!Fi !Yi
Fi m- o-wo ke : Y⇤
i F⇤
i!Wi !Yi
Wo ke - o-wo ke : Y⇤
i W⇤
i!Wi !Yi
We es ima e he impo ance o each o hese pa hs o ansmission by aking
he p oduc o each in e gene a ional co ela ion o ea nings componen s (deno ed
by ⇢wi h i m and wwo ke subsc ip s) and he impo ance o each componen o
o e all ea nings wi hin each gene a ion (aand b om Equa ions 3 and 4) (Duncan,
1966).
Akey ea u eo ou modelis ha i mandwo ke cha ac e is icsmaybe
co ela ed bo h ac oss and wi hin gene a ions, as ep esen ed by he bidi ec ional
a ows in Figu e 1. This co ela ion is implici in ou use o bi a ia e co ela ions
as opposed o pa ial eg ession coefficien s o he in e gene a ional pa ame e s ⇢ff,
⇢ w,⇢w ,and⇢ww. We elabo a e on his poin in Appendix A8, whe e we display
an al e na i e decomposi ion ha sepa a es he pa ial eg ession coefficien s and
he co a iance be ween a he i m and wo ke componen s. This gi es iden ical
esul s o he o al magni ude o each pa hway, bu di ides hese effec s in o di ec
and indi ec pa hs. We also es o in e ac ions be ween amily backg ound and
i m p emiums in de e mining child en’s ea nings, o alida e he assump ions o
he pa h model.
While ou model allows o wi hin-gene a ion co ela ion be ween i m and
wo ke componen s, we ake no s ong s and on he na u e o his co ela ion. I may
s em om wo ke s in luencing i ms, i ms in luencing wo ke s, o a combina ion o
bo h, which is no well-cap u ed in ou pa h diag am. Explici ly inco po a ing hese
as causal pa hs would make o a iche model, bu a he expense o analy ical
ac abili y. To keep he g aph acyclical, imposing a pa h in one di ec ion (e.g.,
om wo ke o i m wi hin gene a ion) ules ou he possibili y o a pa h in he
opposi e di ec ion. The e o e, while ou model is couched as a causal diag am,
15
we in e p e i as p o iding a desc ip i e decomposi ion o he in e gene a ional
ea nings co ela ion.
Gi en a se o assump ions ha we desc ibe u he in Appendix A9, he
in e gene a ional ea nings co ela ion a child’s age can hen be decomposed as:
⇢ =co (Y⇤
i,Y
i )(5)
=⇢ff, ·a1·a2
| {z }
i m- o- i m
+⇢w , ·b1·a2
| {z }
wo ke - o- i m
+⇢ w, ·a1·b2
| {z }
i m- o-wo ke
+⇢ww, ·b1·b2
| {z }
wo ke - o-wo ke
.
Equa ion 5 shows ha he ela i e con ibu ion o each mechanism o he o e all
in e gene a ional co ela ion depends no only on he co ela ion be ween a he ’s
componen and child’s componen ( he ⇢ e ms), bu also on he impo ance o hose
componen s o each gene a ion’s ea nings ( he aand b e ms). Fo example, a high
co ela ion be ween a he ’s i m p emium and son’s i m p emium may accoun o
asmallsha eo heo e allin e gene a ionalco ela ioni he i mp emiumisa
small componen o o e all ea nings.
6.3 Fi m con ibu ions o mobili y by ca ee s age
So a , we ha e abs ac ed om he dimension o child en’s age. The 30-yea span
o ou da a le s us obse e a he s and child en o e se e al yea s nea mid-ca ee
ea nings. He e we desc ibe how we implemen his ocus on mid-ca ee ea nings,
and how we ex end his analysis o s udy ea nings mobili y h oughou he ea ly
ca ee .
In all ou analyses, we ea pa en al social backg ound as ime-in a ian .
We do so by collapsing a he ’s ea nings Y⇤
iac oss he i s 10 yea s in ou da a,
1990–1999. Fa he s a e in he age ange 30–59, which is a ime when male ea nings
a e ep esen a i e o li e ime income and he child en would ha e been in hei
o ma i e yea s. Taking mul i-yea a e ages cen e ed a p ime wo king ages o
ea nings is common p ac ice in he li e a u e on in e gene a ional mobili y and key
o a oid common biases o a enua ion and li e cycle bias (Fox e al., 2016). Because
he a e age o a uni o m a iable is gene ally no uni o m, we eimpose he ank
ans o m wi hin (child) gende –age cells o e ain he in e p e a ion o coefficien s
as ank-o de co ela ions. We c ea e measu es o a he s’ i m-le el componen F⇤
i
and indi idual wo ke -le el componen W⇤
iin he same way, by collapsing p edic ed
alues o e he yea s 1990–1999 and epea ing he ank ans o m.
Fo child en, we pe o m bo h a ime-in a ian and a ime- a ian analysis
o ea nings. Fi s , in ou main analyses, we ea ea nings as ime-in a ian by
a e aging ea nings anks ac oss ages 38–42, which p o ides a good p oxy o li e ime
ea nings (Nybom and S uhle , 2017). He e we exclude he 1978 coho who is only
ollowed un il age 41. We a e age he i m and indi idual wo ke ixed effec s
(Fi ,Wi )o e hesameages,38–42. Like o he a he s,we eimpose he ank
ans o m a e aking mul i-yea a e ages. Second, in a ime- a ian analysis, we
s udy annual ea nings o each yea be ween he ages 25 and 40. This means
ollowing child en om 2000–2015 o ou oldes coho (bo n 1975) and 2003–2018
o ou younges (bo n 1978). These age-speci ic es ima es allow us o ack he
impo ance o diffe en ea nings componen s o mobili y ac oss ca ee s ages.
16
6.4 Dissec ing he i m p emium
Why exac ly does pa en al backg ound link up o i m p emiums? Ou heo e ical
amewo k highligh s wo ways ha i ms can boos indi idual ea nings: i s ly,
p o i able i ms gene a e a g ea e esou ce pool o e all, and secondly, i ms make
diffe en decisions abou wha sha e o ha pool ge s passed on o employees.
Wo ke s om high social o igins could bene i om bo h: hey may ha e be e
access o high-p o i i ms and also be be e posi ioned o ba gain o compensa ion.
The Swedish company egis e con ains alue added pe wo ke , which offe s a
measu e o po en ial i m p emiums be o e p o i alloca ion decisions a e made. We
use his o c ea e wo e sions o ou i m anking: one o he pa o he p emium
p edic ed by alue added, and ano he o he esidual ep esen ing edis ibu ion
condi ional on i m p o i . We co ela e bo h wi h a he ’s ea nings ank o es
whe he p i ileged wo ke s bene i om ma ching o i ms wi h highe p o i , highe
pass- h ough, o bo h.
Nex , we ask wo ques ions abou how i m-based ad an ages o e lap wi h
educa ion and occupa ion. Fi s , how much o he economic e u ns o educa ion
and occupa ion a e media ed by i m so ing? We use “ e u ns” loosely o e e
o obse a ional ea nings diffe ences ac oss deg ees and occupa ions, a he han
he causal coefficien s ypical o e u ns- o-schooling es ima es. Second, how much
o esidual in luence o social backg ound on ea nings— ha is, no media ed by
educa ion o occupa ion—is explained by i m so ing? We look bo h a he di ec
associa ion be ween pa en and child ea nings, and diffe ences in pay by social class
backg ound. In all hese analyses, o child en we ocus on ma u e ea nings a ound
age 40, and he e o e make use o he la es a ailable in o ma ion on educa ion and
occupa ion in 2019.
Ou da a on educa ion consis o 44 disc e e le els and 358 de ailed ields,
which oge he make up 1,993 unique ca ego ies.3Occupa ional da a consis o
Swedish census codes a he 3 digi (150 ca ego ies) and 4 digi (431 ca ego ies)
le el. Educa ion da a a e o he mos pa epo ed di ec ly by he deg ee-g an ing
ins i u ion. Occupa ional da a come om a ious sou ces, including pay oll da a
om la ge o public-sec o employe s, as well as su eys o small en e p ises. Com-
ple e da a on occupa ion is a ailable in 1990, he yea Sweden pe o med i s las
Census. We use his in o ma ion o cons uc a he s’ social class as desc ibed be-
low. Pos -1990, he occupa ion egis e d aws on a ange o sou ces a ailable a
diffe en in e als. Mos impo an is he wage s uc u e s a is ics, which co e s
public employe s and la ge p i a e employe s (mo e han 500 wo ke s). Fo smalle
companies, S a is ics Sweden su eys a sample o 47,000 annually, wi h comple e
co e age achie ed o e a pe iod o 5 yea s (SCB, 2019).
To cap u e obse a ional e u ns o educa ion and occupa ion in a single co-
efficien , we i s eg ess he indi idual’s ea nings ank a ound age 40 on a se
o indica o a iables e lec ing he de ailed ca ego ies, and s o e p edic ed alues
om his eg ession. We hen calcula e he co ela ion be ween p edic ed and ac-
3In ou ea lie analysis o ield-dependen ea nings ajec o ies, we use a coa se ca ego iza ion
o 118 ields necessa y o es ima e hese ajec o ies eliably. When using ield o s udy as a con ol
a iable, we ace no such cons ain , and hus use he mos de ailed ca ego ies a ailable.
17
ual ea nings ank. The esul ing co ela ion coefficien can be in e p e ed as a
measu e o he p opo ion o ea nings a ibu able o educa ion and occupa ion.
The ea e we calcula e pa ial co ela ions ha accoun o he child’s i m com-
ponen o ea nings, o see how much o he e u ns o educa ion and occupa ion a e
media ed by i m so ing.
To s udy di ec effec s o social o igin, we i s ocus on he esidual associa ion
be ween a he and child ea nings ha emains when he in luence o he child’s
educa ion o occupa ion is pa i ioned ou . Mo eo e , we look a he class gap
in ea nings wi hin occupa ions (Lau ison and F iedman, 2016). Fo a he ’s class,
we ely on 1990 Census da a and he Swedish socioeconomic classi ica ion (SEI),
simila o he E ikson–Gold ho pe class schema (Ca lsson e al., 1974). We g oup
he middle and highe sala ia (46, 56) oge he wi h p o essionals (60), and con as
hem agains emaining ca ego ies: wo ke s (11–22), lowe sala ia (36), a me s
(79), and en ep eneu s (89). The class pay gap is de ined as he diffe ence in
ea nings pe cen iles o child en om high and low class, once de ailed educa ion o
occupa ion a e accoun ed o .
6.5 Explaining so ing in o high-paying i ms
In u he analyses, we go on o es a se o heo e ically ele an explana ions o
how wo ke s so o i ms by social o igin: so ing by human capi al, job e e al
ne wo ks, and s uc u al loca ions in he labo ma ke .
Fi s , we ask how much o he i m so ing ad an age is explained by so ing
on educa ional and occupa ional a ainmen . We eg ess he child’s i m ank Fi
on a he ’s ea nings ank Y⇤
i, and see how hei co ela ion changes wi h he same
de ailed educa ional and occupa ional con ols as abo e. We also es whe he i m
so ing is due o unobse ed indi idual wo ke cha ac e is ics. High-paying i ms
may be pa icula ly good a iden i ying and a ac ing high p oduc i i y wo ke s.
I his is he case, hen i m so ing ad an ages o p i ileged-backg ound wo ke s
may be due o hei indi idual cha ac e is ics. To es his, we con ol o he child’s
indi idual componen o ea nings Wi in eg essing he child’s i m componen on
a he ’s ea nings.
We ope a ionalize job e e al in wo ways. Fi s , we look a di ec e e al as
he incidence o wo king a he same wo kplace as he a he . Second, we iden i y
indi ec e e al as inding wo k wi h a senio colleague who a some poin ea lie
has wo ked wi h he a he . To cons uc hese ne wo ks, we make use o wo kplace
iden i ie s a he han i m iden i ie s as in ou main analysis, and es ic links
o wo kplaces o 1000 people and ewe . This is o make i mo e plausible ha a
social ie will ac ually exis , which is doub ul a he le el o i ms o o e y la ge
wo kplaces. To ensu e ha hese o e laps do no simply cap u e o he mechanisms
such as geog aphical p opinqui y we also iden i y a g oup o “placebo” hi es. In
hese, he e is a wo ke connec ing he a he ’s and child’s wo kplace bu no sha ed
o e lap in ime, he o e lap wi h he a he occu ed a e he o e lap wi h he
child, o he wo ke en e ed he child’s wo kplace a e he child was hi ed. Again,
we eg ess he child’s i m componen o ea nings on a he ’s ea nings and inspec
18
$JULFXOWXUHRWKHU
)LQDQFHDQGEXVLQHVV
(GXFDWLRQDQGUHVHDUFK
+HDOWKSXEOLFDGPLQVHUYLFHV
6XSHUVWDUILUPV
/DUJHILUPV
+LJKYDOXHDGGHG
+LJKRSHUDWLQJSURILW
0DMRULW LPPLJUDQW
0DQXIDFWXULQJ
&RQVWUXFWLRQ
(QHUJ
7UDGHDQGFRPPXQLFDWLRQ
0HDQILUPSUHPLXPUDQN
0HDQIDWKHUHDUQLQJVUDQN
Figu e 2: Types o i ms by i m p emium and social selec i i y.
No e: Mean i m p emium ank shown on he e ical axis, mean a he ea nings ank shown on
he ho izon al axis. Ci cles a e shown wi h size p opo ional o hei employmen sha e, and he
diagonal line shows he weigh ed eg ession line o bes i .
how he coefficien diffe s o g oups who expe ience job e e al e sus hose who
do no .
To add ess he ansmission o s uc u al loca ion in he labo ma ke , we
s udy geog aphic immobili y, indus y o e lap, and i m inhe i ance. In hese anal-
yses, we ocus on he a he ’s and child’s main employe , de ined as as whe e he
a he held he longes enu e h oughou he pe iod 1990–1999, o he son h ough-
ou 2000–2019. Hence, we do no di ec ly obse e whe he a a he e e ed a child
o hei i m. Especially wi h la ge employe s, o e lap could occu by chance, so his
is a b oade mechanism han he e e al ne wo ks iden i ied abo e. Geog aphic im-
mobili y occu s when he a he ’s and child’s main employe a e bo h loca ed in he
same municipali y, and sec o inhe i ance when hey a e loca ed in he same 1-digi
indus y.4He e, oo, we eg ess he child’s i m componen o ea nings on a he ’s
ea nings and inspec how he coefficien diffe s o g oups who sha e municipali y,
indus y, o employmen wi h hei a he wi h hose who do no .
4Sweden consis s o 290 municipali ies, he popula ion o which a e aged abou 30,000 in 2000.
The 1-digi le el o he Swedish indus y classi ica ion (SNI) encompasses 10 ca ego ies: ag i-
cul u e, cons uc ion, cul u e and se ices, educa ion and esea ch, ene gy, inance and business,
heal h ca e, manu ac u ing, public adminis a ion, ade and communica ion, as well as an “un-
speci ied” ca ego y.
19
7Resul s
To mo i a e ou analysis, Figu e 2 plo s a ious ypes o i ms by hei pay p emium
and he deg ee o social selec ion in o he i m. The e ical axis shows he a e age
i m pay p emium, and he ho izon al axis shows he a e age a he ea nings ank
among employees. Ci cles a e displayed wi h size p opo ional o hei employmen
sha e, and he diagonal shows a weigh ed eg ession line o bes i .
An impo an sou ce o diffe ences in pay is indus y. Ag icul u e and, o
alesse ex en ,cons uc ion, ec ui p edominan ly om hebo omhal o he
a he ea nings dis ibu ion. Bu while ag icul u e offe s a lowe i m p emium
han o he indus ies, cons uc ion is close o he a e age i m p emium ank. As
such, he cons uc ion indus y p o ides ela i ely high-paying jobs o child en
om he bo om hal o he a he ea nings dis ibu ion. A he o he ex eme,
inance and business ec ui om he op hal o he a he ea nings dis ibu ion
and offe pay abo e he mean i m p emium ank. Mos o he indus ies d aw om
a he ’s ea nings anks ha a e age ou o he middle o he dis ibu ion, bu diffe
in he pay ha hey offe wo ke s.5Fo example, wo ke s in educa ion and esea ch
as well as heal h, public adminis a ion, and se ice—in la ge pa public-sec o
jobs—ge paid less. Con e sely, i ms in ene gy and manu ac u ing pay a highe
p emium. These i ms, hen, a e likely o offe upwa d mobili y o child en om
he b oad middle classes. This is no able gi en ha manu ac u ing employmen
has declined in Sweden like elsewhe e (see Appendix A10).
We also dis inguish i ms in he op decile o i m size, alue added, ope a ing
p o i , o i m p emiums (“supe s a i ms”). Wi h he excep ion o la ge i ms, hese
ca ego ies p edominan ly hi e p i ileged wo ke s and offe signi ican ly highe pay
han o he i ms. Ano he inding eme ges o immig an -domina ed i ms, de ined
as hose whe e he majo i y o wo ke s ha e an immig an backg ound ( i s o
second gene a ion). These i ms o e whelmingly ec ui om he lowe hal o he
a he ’s ea nings dis ibu ion, ye hey offe highe p emiums han expec ed gi en
hei wo k o ce composi ion—possibly due o e hnic solida i y acili a ing p o i -
sha ing wi hin he i m (c . Åslund e al., 2021).
In Appendix A11, we p o ide a mo e de ailed b eakdown o how i m p e-
miums depend on sec o , indus y, and occupa ion. Wo k in he public o p i a e
sec o alone explains 9% o he a ia ion in i m p emiums and 1-digi indus ies
explain 20%. Sec o and indus y oge he explain 22% o he a ia ion, o 23%
wi h an in e ac ion e m be ween he wo. Occupa ion explains 32–35% depending
on whe he a 3- o 4-digi classi ica ion is used. The combina ion o occupa ion,
sec o , and indus y oge he explain 40% o he a ia ion in i m p emiums.
5Tha he ypical wo ke is ec ui ed om he middle does no imply ha ec ui men is uni o m
ac oss he whole ange o a he ea nings. In Appendix Figu e A21, we inspec ec ui men om
bo h ails o he dis ibu ion, e ealing u he nuances. Fo example, manu ac u ing, as well as
educa ion and esea ch, unde ep esen child en om bo h poo and ich households.
20
&KLOGUDQN
)DWKHUHDUQLQJVUDQN
)LUP :RUNHU
(a) Sons
&KLOGUDQN
)DWKHUHDUQLQJVUDQN
)LUP :RUNHU
(b) Daugh e s
Figu e 3: Mean child i m and wo ke ank by a he ea nings ank.
No e: The igu e shows he child’s expec ed i m and wo ke ank by a he ea nings ank. Fa he
ea nings ank measu ed 1990–1999, child i m and wo ke ank measu ed a age 38–42. Fa he ’s
ea nings a e associa ed bo h wi h child en’s indi idual ea nings and he ela i e pay o he em-
ploye hey so o.
7.1 The con ibu ion o i ms o o e all in e gene a ional
ea nings pe sis ence
We ha e es ablished ha so ing in o diffe en ypes o i ms a ies by social back-
g ound, bu subs an ial so ing likely occu s wi hin hese b oad i m ca ego ies as
well. To p o ide a mo e de ailed iew, Figu e 3 p esen s he expec ed ank o
child en’s i m and indi idual wo ke componen s o ea nings ac oss pe cen iles o
a he ’s ea nings. He e we also sepa a e esul s by child gende .
Figu e 3 shows ha a he s’ ea nings ha e a subs an ial associa ion wi h bo h
he indi idual wo ke and i m componen s o child ea nings. Fo sons, he indi-
idual wo ke ank associa ion is 0.24 wi h a he s’ ea nings, compa able o he
ea nings co ela ion i sel , a ⇢=0.24 (see Appendix Table A13). yyy The i m-
le el associa ion is smalle bu ne e heless subs an ial, a 0.15.Fo daugh e s, he
wo ke and i m associa ions a e mo e simila : he indi idual wo ke ank associ-
a ion is 0.20 and he i m-le el associa ion 0.16,compa ed oano e allea nings
co ela ion o ⇢=0.20 (see Appendix Table A14). So, a he ’s ea nings p edic no
only he indi idual wo ke componen o child en’s ea nings, emphasized in p io
in e gene a ional mobili y esea ch, bu also he i m p emium componen .
Gi en hese s ong co ela ions, we nex ask how much i m pay diffe ences
con ibu e o he o e all pe sis ence o ea nings ac oss gene a ions. To do his, we
es ima e all o he sepa a e e ms in Equa ion 5 o ou decomposi ion amewo k
abo e. We b ing hese componen s oge he in Table 1, which de ails he p opo -
ional con ibu ion o he ea nings co ela ion o each o he ou pa hs: i m- o- i m,
wo ke - o- i m, i m- o-wo ke , wo ke - o-wo ke . Filling in he pa h model is c i i-
cal, as, consis en wi h ea lie wo k on he wo-way wo ke - i m ixed effec model,
he wo ke componen is subs an ially mo e impo an o explaining o e all ea n-
ings a ia ion (as cap u ed o sons in he i m coefficien a2=0.36 compa ed o
21
Table 1: Decomposi ion pa ame e s.
Es ima e (pe cen )
Pa ame e Sons Daugh e s
i m- o- i m = ⇢ff·a1·a20.016 (6.80%) 0.007 (3.91%)
wo ke - o- i m = ⇢w ·b1·a20.037 (16.4%) 0.036 (19.2%)
i m- o-wo ke = ⇢ w ·a1·b20.013 (5.66%) 0.008 (4.45%)
wo ke - o-wo ke = ⇢ww ·b1·b20.162 (71.1%) 0.137 (72.4%)
ff+ w + w + ww = co (Y⇤
i,Y
i)0.228 (100%) 0.189 (100%)
⇢ff=co (F⇤
i,F
i)0.120 0.070
⇢w =co (W⇤
i,F
i)0.126 0.150
⇢ w =co (F⇤
i,W
i)0.048 0.031
⇢ww =co (W⇤
i,W
i)0.263 0.218
a1=co (F⇤
i,Y⇤
i|W⇤
i)0.363 0.366
b1=co (W⇤
i,Y⇤
i|F⇤
i)0.837 0.835
a2=co (Fi,Y
i|Wi)0.355 0.289
b2=co (Wi,Y
i|Fi)0.738 0.751
No e: Pa h componen s o he in e gene a ional ea nings co ela ion as desc ibed in Figu e 1.
Fa he ea nings ank measu ed 1990–1999, child ea nings ank measu ed a age 38–42. The
mos impo an pa h is wo ke - o-wo ke , ollowed by wo ke - o- i m, i m- o- i m, and i m- o-
wo ke . E en in ma u e ea nings, i m-based mechanisms accoun o mo e han a qua e o
in e gene a ional ea nings pe sis ence.
he wo ke coefficien b2=0.74). Al hough a he ’s ea nings s ongly p edic son’s
i m p emium, he i m p emium, o bo h gene a ions, accoun s o less a ia ion
in ea nings han does he wo ke componen .
Taking hese diffe en co ela ions wi h o e all ea nings in o accoun , we ind
ha he wo ke - o-wo ke pa h accoun s o be ween 71% (sons) and 72% (daugh-
e s) o he o al in e gene a ional ea nings co ela ion. Fi m- ela ed pa hways
accoun o he emaining qua e o in e gene a ional mobili y. This decomposi-
ion shows ha while wo ke cha ac e is ics a e he dominan sou ce o in e gene a-
ional ea nings ansmission, i m p emiums play an impo an ole in s a i ica ion.
Ex ending his analysis o wo kplaces e eals ha as much as wo hi ds o in e -
gene a ional ea nings esemblance is d i en by wo kplace seg ega ion, educing he
di ec wo ke - o-wo ke link o 61% o sons and 59% o daugh e s (Appendix A3).
Table 1 also p o ides ou i s e idence in o he mechanisms h ough which
i m p emiums affec in e gene a ional mobili y. Al hough he unde lying i m- o-
i m (⇢ff)andwo ke - o- i m(⇢w )co ela ionsa esimila , hewo ke - o- i mpa h
is mo e impo an o he o e all in e gene a ional co ela ion. This is because a
a he ’s indi idual wo ke componen is a mo e in luen ial p edic o o ea nings han
is his i m p emium ank. Speci ically, he i m- o- i m and i m- o-wo ke pa hs
each accoun o abou 6–7% o he in e gene a ional ea nings co ela ion o sons,
and abou 4% o daugh e s. Ma kedly mo e impo an is he wo ke - o- i m pa h
which alone accoun s o mo e han hal o he sha e explained by i ms: 16% o
he o e all co ela ion o sons and 19% o daugh e s.
22
&KLOGHDUQLQJVUDQN
)DWKHUHDUQLQJVUDQN
&KLOGILUPTXDUWLOH 4 4 4 4
(a) Equal e u ns
'HQVLW
)DWKHUHDUQLQJVUDQN
&KLOGILUPTXDUWLOH 4 4 4 4
(b) Unequal so ing
Figu e 4: Fa he and son ea nings by qua iles o he i m p emium.
No e: The le panel shows expec ed son ea nings ank by a he ea nings ank a diffe en qua -
iles o he son’s i m p emium. The igh panel shows he densi y o a he ea nings ank a
diffe en qua iles o he son’s i m p emium. Fa he ea nings ank measu ed 1990–1999, son i m
and ea nings ank measu ed a age 38–42. Diffe en ial so ing, no diffe en ial ewa ds, d i es
inequali y. Daugh e s shown in Appendix Figu e A23.
This i s se o esul s shows ha child en o high-ea ning pa en s disp opo -
iona ely wo k a high p emium i ms. While his p ocess is no he main channel o
in e gene a ional ea nings pe sis ence, i accoun s o a ound a qua e —a quan i a-
i ely meaning ul sha e. We u he show ha he main sou ce o his i m- ela ed
in e gene a ional ansmission is he wo ke - o- i m pa h: pa en s wi h aluable
wo ke cha ac e is ics ha e child en who wo k a high-paying i ms.
7.2 Who bene i s om wo king a a high-paying i m?
Do wo ke s om high- and low-ea nings backg ounds bene i equally om wo king
a high-paying i ms? Tes ing his ma e s o wo easons. Fi s , any policy a -
emp o inc ease mobili y by equalizing access o employe s would ha e o p oceed
on his assump ion. Second, ou pa h model assumes linea , addi i e effec s, so
iola ions he eo would challenge i s in e p e a ion. Theo e ically, his assump ion
could b eak down in diffe en ways. Wo king a a high-paying i m could diminish
pa en al in luence on a child’s ea nings, akin o how ob aining a college deg ee is be-
lie ed o ac as a le elle (Ka lson, 2019; To che, 2011). Al e na i ely, he in luence
o pa en al backg ound could loom especially la ge in hese se ings, as sugges ed
by li e a u e on he “class ceiling” in eli e manage ial posi ions and p o essional
se ices (F iedman and Lau ison, 2020).
We es his in Figu e 4, le , which shows a sepa a e a he –son ea nings
associa ion a each qua ile o he i m p emium. This igu e ein o ces he impo -
ance o i ms o o e all ea nings: sons om he same a he ea nings pe cen ile
ea n 20 pe cen ile poin s mo e when wo king o a op-qua ile i m compa ed o
23
abo om-qua ile i m.
6Mo eo e , i appea s ha he linea , addi i e model is
a ema kablyaccu a e i ,as he eisli lede ec ablehe e ogenei yo heslope
ac oss diffe en le els o i m p emiums. Co esponding esul s o daugh e s a e
shown in Appendix A12. They e eal ha i m p emiums in he op qua ile appea
especially impo an o boos daugh e s’ ea nings bu he inding o la gely pa allel
a he –child associa ions o ea nings a diffe en i m p emiums holds o daugh e s
as well.
In sum, i m p emiums seem o bene i p i ileged child en mainly in ha hey
a e mo e likely o ind wo k wi h a high-paying i m, bu once inside he i m, high-
and low-ea nings backg ound wo ke s de i e a simila boos (o penal y) o hei
ea nings. Inequali ies in access, no in ou comes, d i e he i m-based ad an age o
wo ke s om p i ileged backg ounds. Figu e 4, igh , isualizes he e y unequal
p obabili ies wi h which wo ke s o diffe en pa en al ea nings backg ound en e
i ms wi h high o low pay p emiums. The igu e shows ha his so ing is no
limi ed o a speci ic pa o he i m p emiums dis ibu ion: ac oss he dis ibu ion
o i m, unequal so ing by a he ’s ea nings eme ges.
7.3 When do i m p emiums ma e ?
How does he impo ance o hese i m and wo ke ansmission componen s change
o e a ca ee pe iod? Some sou ces o pe sis ence—like i m inhe i ance and di ec
pa en al aid in job sea ch—should be s onges a he ea ly ca ee s age, be o e
ape ing ou by mid-ca ee .
Figu e 5 es s his by decomposing he in e gene a ional ea nings co ela ion
in o i s pa h componen s a each yea o a child’s ca ee . Fi m so ing is indeed
mos impo an ea ly in he ca ee and media es he bulk o ea nings ansmission
in he i s ew yea s o labo ma ke en y. This ades g adually o abou a qua e
o he o al co ela ion by age 40 ( o ela i e numbe s, see Appendix Figu e A9).
This decline ac oss he ca ee in he i m- ela ed sha e o ea nings ansmission
is due o he inc ease in impo ance o he wo ke - o-wo ke pa h, as he child’s
wo ke componen becomes a mo e impo an po ion o hei ea nings. A he
same ime, i m- o- i m ansmission aces a small decline in absolu e e ms and
ala ge ela i edecline. Ha ingapa en employeda ahigh-ea ning i mmainly
bene i s wo ke s when hey a e young.
In con as , he wo ke - o- i m ansmission pa h g ows in absolu e e ms and
e ains an impo an sha e o in e gene a ional ea nings pe sis ence up h ough
age 40. No e ha his pa hway, in which child en o high-ea ning pa en s end
up wo king a high-p emium i ms, could be impo an bo h a ea ly and la e
ca ee s ages. Di ec pa en al aid should ma e mainly ea ly on, when pa en s
a e s ill ac i e and engaged wi h hei child en’s labo ma ke a ainmen . Indi ec
human and social capi al so ing ad an ages, in con as , could pay off h oughou a
child’s ca ee . The con inued impo ance o he wo ke - o- i m pa hway up h ough
6This gap in ea nings esul s om diffe ences in i m pay, bu also so ing o wo ke s in o i ms,
which we add ess in a la e sec ion. S udying he e ogenous e u ns while accoun ing o selec ion
is a complex ask, as highligh ed by s udies on highe educa ion and mobili y (Fiel, 2020; Ka lson,
2019; Zhou, 2019).
24
Table 2: Associa ion be ween a he ea nings and son i m so ing, wi h con ols.
b/se b/se b/se b/se
Coefficien 0.149 0.046 0.030 0.014
(0.003) (0.003) (0.004) (0.004)
Educa ion – XXX
Occupa ion – XXX
Wo ke FE – XXX
Di ec e e al – – XX
Indi ec e e al – – XX
Geog aphic immobili y – – – X
Indus y inhe i ance – – – X
Fi m inhe i ance – – – X
N 112931 112931 112931 112931
No e: Rank co ela ion be ween a he ea nings and son i m p emium, con olling o mechanisms.
Fa he ea nings ank measu ed 1990–1999, son i m ank measu ed a age 38–42. Codes o
educa ion and occupa ion comp ise 1,993 and 431 ca ego ies, espec i ely. Te ms o e e al,
immobili y, and inhe i ance include an in e ac ion wi h a he ea nings ank. Human capi al
ad an ages accoun o oughly wo- hi ds o i m ad an ages by ea nings backg ound, wi h mos
o he emaining hi d accoun ed o by labo ma ke e e al and inhe i ance o oppo uni y
s uc u e. Daugh e s shown in Appendix Table A10.
wi h he excep ion ha indus y inhe i ance is less common and concen a ed o
he op o he a he ’s ea nings dis ibu ion (Appendix Figu e A28).
To es how hese o ms o s uc u al o e lap shape i m so ing, we eg ess he
son’s i m componen o ea nings a ound age 40 on he a he ’s ea nings ank, and
pa i ion he sample by geog aphic immobili y, indus y inhe i ance, and employe
inhe i ance. Fo sons who ind hei main job in he same municipali y, indus y o
i m as hei a he , he ank co ela ion be ween a he ea nings and he son’s i m
componen o ea nings is 27%, 61%, and 115% s onge , espec i ely. When hese
g oups a e excluded, he coefficien is educed by 12%, 17%, and 6% espec i ely,
om he baseline o 0.15. Thus, geog aphic immobili y and indus y inhe i ance
explain a subs an ial sha e o he so ing ad an age. By con as , i m inhe i ance
is ela i ely inconsequen ial as his g oup is small enough ha emo ing i om
es ima ion does no subs an ially al e he associa ion in he emaining popula ion.
7.8 Mul i a ia e esul s
In Table 2, we b ing he abo e explana ions o so ing oge he in one model.
Co esponding esul s o daugh e s appea in Appendix Table A10. As be o e, we
eg ess he sons’s i m componen o ea nings a ound age 40 on he a he ’s ea nings
ank. We hen add s epwise con ols s a ing wi h educa ion, occupa ion, and he
wo ke componen o ea nings. We use he mos de ailed a iables: 1,993 educa ion
codes and 431 occupa ion codes. Fo he subsequen a iables e lec ing e e al
ne wo ks, s uc u al loca ion, and i m inhe i ance we include bo h he main e m
and an in e ac ion wi h a he ea nings ank. In his way, he emaining coefficien
31
cap u es pe sis ence in he compa ison g oup who do no expe ience job e e al o
s uc u al pe sis ence.
Con olling o de ailed educa ion, occupa ion, and he wo ke componen o
ea nings, oge he explains wo hi ds o he i m so ing ad an age by social o i-
gin (69%). Adding con ols o di ec and indi ec e e al explains ano he en h
o he emaining associa ion (11%), as does adjus ing o s uc u al loca ion and
i m inhe i ance (11%). Only a small p opo ion o he i m so ing ad an age e-
mains unexplained (9%). The p opo ions explained o daugh e s a e simila , wi h
63%, 9%, and 12% in each s ep, lea ing a somewha la ge p opo ion o he i m
so ing ad an age unexplained (17%) (Appendix Table A10).8These esul s ein-
o ce hose o ou p e ious analysis: P i ileged wo ke s p ima ily secu e posi ions
a high-paying employe s by means o hei educa ion and occupa ion, wi h e e -
al ne wo ks and he inhe i ance o indus y and labo ma ke con ex playing a
supplemen a y ole.
7.9 Addi ional analyses and obus ness
Wo kplace p emiums. Ou da a con ain wo ypes o iden i ie s ela ed o he
indi idual’s employmen : one o i m/employe and one o wo kplaces. We use
bo h o dis inc pu poses. In ou ea nings decomposi ion, we ocus on employe s
o consis ency wi h p e ious li e a u e and as mos heo e ical mechanisms ope a e
a his le el. Con e sely, we use wo kplace iden i ie s o de ine pa en al social
ne wo ks, which a e mo e plausible a his le el. Fo comple eness, we eassess ou
main esul s using wo kplace iden i ie s o he ea nings decomposi ion in Appendix
A3. The esul s assign a la ge impo ance o wo kplace s a i ica ion han i m
s a i ica ion, as would be expec ed gi en he ine g anula i y o his measu e.
A ound 40% o he ma u e in e gene a ional co ela ion in ea nings is explained by
so ing be ween wo kplaces ha offe unequal pay, compa ed o 29% o i ms in
ou main analysis.
Mo he s’ ea nings. Th oughou ou main analysis, we ocus on a he s, ypi-
cally he main b eadwinne s in he pa en gene a ion. Using amily income would
complica e he link wi h employe s, as i in ol es mul iple employe s as well as
sou ces beyond employmen . Ins ead, we examine he ole o mo he s’ ea nings sep-
a a ely in Appendix A4. This leads o lowe in e gene a ional co ela ions o sons
bu no daugh e s, while he ela i e b eakdown o ea nings componen s emains
simila . Essen ially, bo h a he s and mo he s con ibu e o ea nings pe sis ence
o e all and speci ically o i m-based ansmission. Appendix Table A5 shows mod-
e a e spousal co ela ions be ween a he s’ and mo he s’ ea nings, compa able in
size o in e gene a ional co ela ions. This implies ha mo he s’ ea nings con ain
impo an addi ional in o ma ion no p esen in a he s’ ea nings, highligh ing he
impo ance o gende ed pa hs o ansmission as a opic o u u e esea ch.
8Pa ly, his migh be due o us ha ing ope a ionalized se e al key mechanisms wi h e e ence
o he a he , whe eas mo he s may be mo e in luen ial o daugh e s.
32
Wo ke - a ying i m p emiums. Ou main analysis assumes i m p emiums
a e uni o m ac oss wo ke s, unaffec ed by indi idual cha ac e is ics. We also ind
no e idence ha payoffs o i m p emiums diffe by social o igin. Howe e , i m p e-
miums may a y along o he dimensions, such as educa ion and sex. In Appendix
Figu e A13, we add ess his by spli ing each i m iden i ie in o wo sepa a e ca -
ego ies: one o wo ke s wi h e ia y educa ion and ano he o hose wi hou .
This app oach, which ea s high- and low-educa ed wo ke s wi hin he same i m
as belonging o diffe en i ms, mo e han doubles he i m componen o ea nings
ansmission. This sugges s ha i ms do diffe en ia e be ween wo ke ypes in he
p emiums hey offe . Howe e , in e p e ing hese componen s as pu ely i m-le el
effec s is difficul , as b oade labo ma ke pay diffe ences by educa ion become
embedded in he i m componen . By con as , when i ms a e spli by male and
emale wo ke s, he decomposi ion esul s emain la gely unchanged (see Appendix
Figu e A14).9
Time- a ying i m p emiums. Ou main analysis assumes ha i m p emiums
emain cons an h oughou he obse a ion pe iod, and a e he same o pa en s
and child en. This is a s ong assump ion gi en e idence o shi ing i m p emiums
o e ime (C iscuolo e al., 2020; Song e al., 2019; Wilme s and Aeppli, 2021). In
Appendix Figu e A15, we allow i m p emiums o a y by eassigning i m iden i ie s
e e y 5 yea s, ea ing each i m as a diffe en uni in 1990–1994, 1995–1999, and
so o h. The indings closely esemble hose in ou main esul s sec ion. One
explana ion migh be he high a e o i m u no e in ou da a, as e idenced in
Appendix Figu e A19. Tu no e occu s no only due o businesses going bank up
o closing down bu also because o me ge s, acquisi ions, changes in owne ship,
o o he ypes o es uc u ing. Fo hese easons, iola ions o he assump ion o
cons an i m p emiums appea mi iga ed by he limi ed li e-span o i m iden i ie s.
Ne wo k e e al and wo ke ank. We ha e ocused on mechanisms ha
can explain he ma ching o high pa en al ea nings wo ke s o high-paying i ms.
Howe e , access o aluable social capi al may in luence bo h he i m p emium
and an indi idual’s ank wi hin he i m. In Appendix A15, we he e o e es he
effec o job e e al ne wo ks on indi idual wo ke ank. Doing so e eals ha
e e al imp o es wo ke p ospec s h ough he ank achie ed wi hin a i m, and
no jus he a e age pay o he i m—albei wi h somewha less o ce, especially o
weak ies. Fo sons, di ec e e al explains 8% o indi idual ea nings ansmission,
as opposed o 10% o i m-based ansmission. Indi ec e e al explains 4% o
indi idual ea nings ansmission, as opposed o 14% o i m-based ansmission.
By con as , geog aphic immobili y, indus y inhe i ance, and employe inhe i ance
do li le o explain he indi idual wo ke ank. These esul s sugges ha gaining
9Since ea nings ank is cons uc ed wi hin sex, age, and ea nings yea , labo ma ke -wide pay
diffe ences by sex a e excluded by design, unlike he educa ion pay p emium.
33
access o a i m h ough pe sonal con ac s, especially one’s pa en , allows wo ke s
o as - ack hei ca ee s a he expense o co-wo ke s.10
8Conclusion
Much esea ch in social s a i ica ion has ocused on educa ion, occupa ion, and
upwa d mobili y inside i ms. Bu ising be ween- i m inequali y has di ided he
economy be ween high-paying supe s a i ms and low-wage lagga ds. These labo
ma ke changes ha e been accompanied by empi ical me hods ha iden i y unequal
i m p emiums by con olling o i ms’ wo ke composi ion. Mo eo e , ecen quali-
a i e esea ch emphasizes ha i m so ing can be in luenced by social backg ound,
whe eby ad an aged job applican s ob ain en y o eli e i ms.
In his pape , we cla i y how inequali y in i m pay p emiums affec s in e -
gene a ional mobili y. Empi ically, we ind ha he endowmen o indi idual ai s
accoun s o he majo i y o ea nings ansmission, consis en wi h he main h us
o s a i ica ion esea ch. Ye , mo e han a qua e o ea nings ansmission oc-
cu s ia inequali y in i m p emiums. E en by age 40, child en om p i ileged
o igins con inue o wo k in highe -paying i ms han hei less o una e pee s. This
i m ad an age e lec s high-ea ning backg ound wo ke s enjoying employmen a
i ms ha bo h ha e highe alue-added and also ha sha e a la ge po ion o his
alue-added wi h wo ke s. The main way ha p i ileged wo ke s so o high-paying
employe s is h ough educa ion and occupa ion. Ne wo k e e al and inhe i ance
o s uc u al posi ion in a local labo ma ke o indus y play supplemen a y, bu
smalle oles.
The ela i e impo ance o hese mechanisms may be speci ic o he Swedish
con ex . Ac i e labo ma ke policies and obus labo unions con ibu e o a sys em
ha p omo es o mal job sea ch me hods o e in o mal connec ions, dis inguishing
i om many o he de eloped na ions (Bachmann and Baumga en, 2013). This
makes us hink ha , i any hing, Sweden is a “leas -likely” case ha p o ides a lowe
bound on he impo ance o con ac s. While Sweden has an egali a ian sys em o
educa ion unding, ca ee s a e hea ily s a i ied by ield o s udy, some hing ha
begins in seconda y school when amily in luence s ill looms la ge (Dahl e al., 2023).
This ea ly emphasis on specializa ion may in luence he signi icance o educa ional
ield o s udy in employe so ing p ocesses. In coun ies mo e hea ily s a i ied by
ins i u ion, school p es ige may gain in impo ance and u u e wo k should s udy
his issue wi h compa a i e da a.
I is impo an o cla i y ha he la ge ea nings ansmission componen a -
ibu able o ime-in a ian indi idual ai s does no necessa ily e lec me i oc acy
o human capi al. High ea nings pay-off ai s could be indi idual skills, in elligence,
pe sonali y o abili y. Bu hey also include consis en disc imina ion by gende o
e hnici y and cul u al capi al ha ga ne s de e ence ac oss mul iple domains. Like-
wise, i m p emiums can e lec ma ginal p oduc i i y, inso a as p oduc i i y is a
10Al e na i ely, hey may indica e ha employe s use ne wo ks o iden i y high-quali y candi-
da es who would ha e ea ned simila ly high wages in he absence o a e e al. Fu he longi udinal
esea ch should dis inguish hese explana ions.
34
unc ion o pee effec s o collabo a ion. Ou ocus is on dis inguishing he ole o
i m p emiums om he b oad se o indi idual endowmen s and a ibu es, a he
han dis inguishing me i pe se om unea ned p i ilege.
No ably, child en om p i ileged o igins enjoy a i m p emium ad an age
ega dless o he sou ces o pa en s’ ea nings ad an age. Pa en s wi h aluable
indi idual cha ac e is ics suppo and cul i a e hei child en o ob ain jobs a
high-paying i ms, e en when he pa en did no wo k a such a i m. These esul s
show how agile “achie ed” ewa ds in one gene a ion can ha den in o “asc ibed” o
class-based ad an ages in he nex . Pa en s who achie e a high indi idual s anding
by din o luck o skill ace a p oblem in how o pass on hei ad an age. Fo
example, he skills ha gained he pa en hei posi ion may be cos ly o ans e ,
o u n ou o be obsole e. S a egic knowledge abou he educa ion sys em o labo
ma ke , along wi h he mobiliza ion o ne wo k ies, may be less cos ly ways o pass
on his s a us.
Educa ional a ainmen explains an impo an pa , up o hal , o i m so ing
by social backg ound. Meanwhile, i m p emiums explain only abou a qua e o
he e u ns o educa ion. Howe e , i ms explain a much la ge po ion o he in-
e gene a ional ea nings ad an age no media ed by educa ion o occupa ion. This
esidual is o g owing conce n: as access o educa ion and whi e-colla jobs eaches
an inc easing sha e o he popula ion, p i ileged pa en s will be p essed o new
means o ep oduce hei success. We show ha i m so ing explains hal o mo e
o he di ec effec o pa en on child ea nings a e educa ion and occupa ion is pa -
i ioned ou . Fi m p emiums also explain mos o he class o igin gap in ea nings
ha much ecen schola ship has ocused on (Lau ison and F iedman, 2016).
The i m-based app oach ha we pu sue holds p omise o unde s anding he
ole o ne wo ks in he ep oduc ion o economic success. Abou a hi d o sons
om homes abo e median income ha e sha ed a wo kplace wi h hei a he , and
a simila p opo ion wi h a o me colleague o he a he . These e e als yield a
payoff,no h oughpe manen employmen bu a he ass epping-s onejobson he
pa h o a luc a i e ca ee . Inhe i ance o a main employe is a e and su p isingly
inconsequen ial. While pa en al e e al is ansien , child en’s bene i s in e ms
o i m so ing a e las ing. These esul s a e consis en wi h exis ing wo k on he
e u ns o social capi al o e he li e cou se (McDonald and Elde J , 2006).
The ole o i m p emiums as a c i ical componen o he appa en educa ion
p emium should change how we hink abou educa ion in he labo ma ke . Educa-
ional ad an age may be a ibu able in pa o skills, bu also o an ad an age in
i m so ing ia ne wo ks, p oduc i i y complemen a i ies and s a egic knowledge.
Aweal ho quali a i ewo kal eady es i ies o heimpo anceo hesemechanisms
(Ashley, 2022; Ho, 2009; Ri e a, 2016), ye quan i a i e s a i ica ion esea ch lags
behind wi h i s ocus on indi iduals o e ma ching p ocesses. A i m-based ap-
p oach is a p omising way o wa d. The use o new da a sou ces and ools could
iden i y pipelines be ween eli e educa ional ins i u ions and eli e employe s.
Ano he implica ion o ou wo k conce ns how he dis ibu ion o i m p emia
shape in e gene a ional ad an age ac oss diffe en segmen s o he economy. We
ha e p o ided ini ial analysis down hese lines, by showing ha mos indus ies
ec ui ai ly e enly ac oss social o igins and inhe i ed ad an age appea s consis en
35
ac oss he i m p emium dis ibu ion. Ne e heless, u u e esea ch may iden i y
labo ma ke s in which i m-media ed ad an age is weake o in e up ed al oge he .
Such esea ch could also help dis inguish he kinds o i ms and he sou ces o
i m pay p emiums ha p o ide mobili y oppo uni ies om he kinds ha lock in
ad an ages o al eady p i ileged wo ke s. While ends such as skill seg ega ion and
supe s a i ms will ampli y he social backg ound e u ns o his p ocess, policies
o cen alized ba gaining and minimum wages could dampen hem.
Fu u e esea ch should also assess how he in luence o i ms on in e gene a-
ional mobili y a ies wi h he o e all impo ance o i ms in de e mining wo ke s’
pay. Sweden is a conse a i e case he e, as i m p emiums accoun o a ela i ely
small sha e o ea nings inequali y compa ed o o he de eloped coun ies (C iscuolo
e al., 2020). Bu , mo e b oadly, ac oss mos ich coun ies, ea nings inequali y is
inc easingly be ween- i ms (Tomasko ic-De ey e al., 2020), due o a combina ion
o inc eased posi i e wo ke so ing ac oss i ms and pe sis en ly impo an i m
pay p emiums (Ca d e al., 2013; Song e al., 2019). In economies in which i ms
subs an ially de e mine wo ke ea nings, such as he US, we expec i ms’ ole in
in e gene a ional mobili y o be ampli ied.
This poin s o a esea ch agenda on how compa a i e wo k and labo o ga-
niza ion migh shape he inhe i ance o ad an age. He e ou analysis connec s o
Rela ional Inequali y Theo y (Rosen eld, 2021; Tomasko ic-De ey and A en -Hol ,
2019), which emphasizes he o ganiza ional and ins i u ional de e mina ion o ea n-
ings inequali y. While ha wo k has unco e ed de e minan s o pay-se ing and o
i m p emiums ha go beyond compe i i e ma ke p ice de e mina ion, i has o-
cused on explaining inequali y associa ed wi h asc ip i e cha ac e is ics like ace,
sex, and immig a ion s a us (Tomasko ic-De ey e al., 2015). I sugges s ha so-
cial closu e and esou ce pooling ac oss i ms can be an impo an sou ce o wage
inequali y. We add ha i m-speci ic o ganiza ional and ins i u ional de ia ions
om compe i i e ma ke wages can con ibu e o in e gene a ional ea nings ans-
mission. Beyond con ibu ing o inequali y, i m p emiums affec in e gene a ional
s a i ica ion.
Since i s o igins in he 1960s, s a us a ainmen esea ch has consis en ly o-
cused on ways ha high-s a us pa en s endow child en wi h educa ion and o he
indi idual esou ces alued in he labo ma ke . Al hough his esea ch p og am has
p o en ex ao dina ily p oduc i e, i has also obscu ed ways ha social ep oduc-
ion is in luenced by s uc u al ea u es o he labo ma ke . A s ong unde cu en
o s a i ica ion esea ch has ques ioned his ocus on he indi idual, hypo hesizing
labo ma ke segmen a ion as a d i ing o ce behind in e gene a ional s a i ica-
ion (Ba on, 1984; Ba on and Bielby, 1980; Fea he man, 1981; Spile man, 1977;
S olzenbe g, 1978). Recen a ailabili y o la ge-scale adminis a i e mic oda a and
new empi ical me hods make i possible o he i s ime o es his hypo hesis.
Ou indings demons a e he p omise o inco po a ing i ms and wo kplaces in o
an analysis o in e gene a ional mobili y and should inspi e u he wo k owa d
his aim.
36
Re e ences
Abowd, John M., F ancis K ama z, and Da id N. Ma golis. 1999. “High Wage
Wo ke s and High Wage Fi ms.” Econome ica 67:251–333.
And ews, Ma yn J, Len Gill, Tho s en Schank, and Richa d Upwa d. 2008. “High
wage wo ke s and low wage i ms: nega i e asso a i e ma ching o limi ed mobil-
i y bias?” Jou nal o he Royal S a is ical Socie y: Se ies A (S a is ics in Socie y)
171:673–697.
A ms ong, Elizabe h A and Lau a T Hamil on. 2013. Paying o he Pa y.Ha -
a d Uni e si y P ess.
Ashley, Louise. 2022. Highly Disc imina ing: Why he Ci y Isn’ Fai and Di e si y
Doesn’ Wo k. Policy P ess.
Åslund, Olo , C is ina B a u, S e ano Lomba di, and Anna Tho esson. 2021. “Fi m
p oduc i i y and immig an -na i e ea nings dispa i ies.” IZA Discussion Pape
No. 14960.
Au o , Da id, Da id Do n, Law ence F Ka z, Ch is ina Pa e son, and John
Van Reenen. 2020. “The all o he labo sha e and he ise o supe s a i ms.”
Qua e ly Jou nal o Economics 135:645–709.
Bachmann, Ronald and Daniel Baumga en. 2013. “How do he unemployed sea ch
o a job? – E idence om he EU Labou Fo ce Su ey.” IZA Jou nal o Eu opean
Labo S udies 2:22.
Ba on, James N. 1984. “O ganiza ional pe spec i es on s a i ica ion.” Annual
Re iew o Sociology 10:37–69.
Ba on, James N. and William T. Bielby. 1980. “B inging he Fi ms Back in: S a -
i ica ion, Segmen a ion, and he O ganiza ion o Wo k.” Ame ican Sociological
Re iew 45:737–765.
Beggs, John J and Wayne J Villemez. 2001. “Regional labo ma ke s.” In Sou ce-
book o Labo Ma ke s, edi ed by I a Be g and A ne L Kallebe g, pp. 503–529.
Sp inge .
Be na di, Fab izio and Gab iele Balla ino. 2016. Educa ion, Occupa ion and Social
O igin: A compa a i e analysis o he ansmission o socio-economic inequali ies.
Edwa d Elga Publishing.
Bibb, Robe and William H Fo m. 1977. “The effec s o indus ial, occupa ional,
and sex s a i ica ion on wages in blue-colla ma ke s.” Social Fo ces 55:974–996.
Bihagen, E ik. 2008. “Does Class Ma e Equally o Men and Women? A S udy o
he Impac o Class on Wage G ow h in Sweden 1999–2003.” Sociology 42:522–
540.
37
Bills, Da id B. 2003. “C eden ials, signals, and sc eens: Explaining he ela ionship
be ween schooling and job assignmen .” Re iew o educa ional esea ch 73:441–
469.
Binde , Amy J, Daniel B Da is, and Nick Bloom. 2016. “Ca ee unneling: How
eli e s uden s lea n o de ine and desi e ‘p es igious’ jobs.” Sociology o Educa ion
89:20–39.
Bjö klund, Ande s. 2000. “Going Diffe en Ways: Labou Ma ke Policy in Denma k
and Sweden.” In Why De egula e Labou Ma ke s?, edi ed by Gøs a Esping-
Ande sen and Ma ino Regini, pp. 148–180. New Yo k: Ox o d Uni e si y P ess.
Black, Sand a and Paul De e eux. 2011. “Recen De elopmen s in In e gene a ional
Mobili y.” In Handbook o Labo Economics,edi edbyO leyAshen el e and
Da id Ca d, olume 4, pp. 1487–1541. No h Holland P ess, Else ie .
Blau, Pe e M and O is Dudley Duncan. 1967. The Ame ican occupa ional s uc u e.
New Yo k: John Wiley & Sons.
Bloome, Dei d e, Shauna Dye , and Xiang Zhou. 2018. “Educa ional inequali y,
educa ional expansion, and in e gene a ional income pe sis ence in he Uni ed
S a es.” Ame ican Sociological Re iew 83:1215–1253.
Bo gen, Nicolai T and A ne Mas ekaasa. 2018. “Ho izon al s a i ica ion o highe
educa ion: he ela i e impo ance o ield o s udy, ins i u ion, and depa men
o candida es’ wages.” Social Fo ces 97:531–558.
Bou dieu, Pie e. 1984. Dis inc ion: A social c i ique o he judgemen o as e.
Ha a d Uni e si y P ess.
Bou dieu, Pie e. 2000. The Social S uc u es o he Economy.London:Poli y.
Bowles, Samuel, He be Gin is, and Melissa Osbo ne G o es. 2009. Unequal
chances: Family backg ound and economic success. P ince on Uni e si y P ess.
B a be g, Espen, Jona han Da is, Bhashka Mazumde , Ma in Nybom, Daniel D
Schni zlein, and Kjell Vaage. 2017. “A compa ison o in e gene a ional mobili y
cu es in Ge many, No way, Sweden, and he US.” The Scandina ian Jou nal o
Economics 119:72–101.
B een, Richa d, Ca ina Mood, and Jan O Jonsson. 2016. “How much scope o a mo-
bili y pa adox? The ela ionship be ween social and income mobili y in Sweden.”
Sociological Science 3:39.
B een, Richa d and Wal e Mülle . 2020. Educa ion and In e gene a ional Social
Mobili y in Eu ope and he Uni ed S a es. S an o d Uni e si y P ess.
B yson, Alex and Rhys Da ies. 2019. “Family, place and he in e gene a ional ans-
mission o union membe ship.” B i ish Jou nal o Indus ial Rela ions 57:624–
650.
38
Bühlmann, Felix, F ançois Schoenbe ge , Lena Ajdacic, and Fabien Fou eaul . 2022.
“Eli e ec ui men in US inance: How uni e si y p es ige is used o secu e op
execu i e posi ions.” B i ish Jou nal o Sociology 73:667–684.
Bu awoy, Michael and E ik Olin W igh . 1990. “Coe cion and consen in con es ed
exchange.” Poli ics & Socie y 18:251–266.
Ca d, Da id, Ana Ru e Ca doso, Joe g Heining, and Pa ick Kline. 2018. “Fi ms and
labo ma ke inequali y: E idence and some heo y.” Jou nal o Labo Economics
36:S13–S70.
Ca d, Da id, Jö g Heining, and Pa ick Kline. 2013. “Wo kplace He e ogenei y and
he Rise o Wes Ge man Wage Inequali y.” Qua e ly Jou nal o Economics
128:967–1015.
Ca lsson, Gos a, Robe E ikson, Ca l Lö wall, and Bo Wä ne yd. 1974. “Socio-
ekonomiska g uppe inga .” S a is isk idsk i 5:381.
Ca nie o, P, Sa ah Ca an, Lo aine Dea den, Lau a an de E e, Sonya K u iko a,
and Lindsey Macmillan. 2020. “The long shadow o dep i a ion: Diffe ences in
oppo uni ies ac oss England.” Social Mobili y Commission Resea ch Repo .
Cas illa, Emilio J, Geo ge J Lan, and Ben A Rissing. 2013. “Social ne wo ks and
employmen .” Sociology Compass 7:999–1026.
Cheng, Siwei and Xi Song. 2019. “Linked Li es, Linked T ajec o ies: In e gene a-
ional Associa ion o In agene a ional Income Mobili y.” Ame ican Sociological
Re iew 84:1037–1068.
Cobb, J Adam and Ken-Hou Lin. 2017. “G owing apa : The changing i m-size
wage p emium and i s inequali y consequences.” O ganiza ion Science 28:429–
446.
Collins, Randall. 1979. The c eden ial socie y: An his o ical sociology o educa ion
and s a i ica ion. Columbia Uni e si y P ess.
Co ak, Miles. 2013. “Income inequali y, equali y o oppo uni y, and in e gene a-
ional mobili y.” Jou nal o Economic Pe spec i es 27:79–102.
Co ak, Miles and Pa izio Pi aino. 2011. “The In e gene a ional T ansmission o
Employe s.” Jou nal o Labo Economics 29:37–68.
Co nelissen, Thomas, Ch is ian Dus mann, and U a Schönbe g. 2017. “Pee effec s
in he wo kplace.” Ame ican Economic Re iew 107:425–56.
Co eia, Se gio. 2019. “REGHDFE: S a a module o pe o m linea o ins umen al-
a iable eg ession abso bing any numbe o high-dimensional ixed effec s.”
Co ell, Shelley J, Cecilia L Ridgeway, Ez a W Zucke man, Sha on Jank, Sa a
Jo dan-Bloch, and Sand a Nakagawa. 2017. “I ’s he con en ional hough ha
coun s: How hi d-o de in e ence p oduces s a us ad an age.” Ame ican Socio-
logical Re iew 82:297–327.
39
C iscuolo, Chia a, Alexande Hijzen, Cy ille Schwellnus, E ling Ba h, Wen-Hao
Chen, Richa d Fabling, P iscilla Fialho, Balazs S adle , Richa d Upwa d, Wou e
Zwysen, Ka a zyna G abska-Romagosa, Ryo Kambayashi, Timo Leidecke , Os-
ka No ds öm Skans, Capucine Riom, and Duncan Ro h. 2020. “Wo k o ce com-
posi ion, p oduc i i y and pay: he ole o i ms in wage inequali y.” OECD
Economics Depa men Wo king Pape s.
Dahl, Go don B, Dan-Olo Roo h, and Ande s S enbe g. 2023. “High school majo s
and u u e ea nings.” Ame ican Economic Jou nal: Applied Economics 15:351–
382.
Dahl, Molly W and Thomas DeLei e. 2008. “The associa ion be ween child en’s
ea nings and a he s’ li e ime ea nings: es ima es using adminis a i e da a.” Uni-
e si y o Wisconsin-Madison, Ins i u e o Resea ch on Po e y.
DiMaggio, Paul and Filiz Ga ip. 2012. “Ne wo k effec s and social inequali y.”
Annual e iew o sociology 38:93–118.
Dobbin, Caue and Tom Zoha . 2023. “Quan i ying he Role o Fi ms in In e gene -
a ional Mobili y.” CESi o Wo king Pape No. 10758.
Domina, Thu s on, And ew Penne , and Emily Penne . 2017. “Ca ego ical inequal-
i y: Schools as so ing machines.” Annual Re iew o Sociology 43:311–330.
Duncan, O is Dudley. 1966. “Pa h analysis: Sociological examples.” Ame ican
Jou nal o Sociology 72:1–16.
Engbom, Niklas and Ch is ian Mose . 2017. “Re u ns o educa ion h ough access
o highe -paying i ms: E idence om US ma ched employe -employee da a.”
Ame ican Economic Re iew 107:374–78.
Engbom, Niklas, Ch is ian Mose , and Jan Saue mann. 2023. “Fi m pay dynamics.”
Jou nal o Econome ics 233:396–423.
Engzell, Pe and Ca ina Mood. 2023. “Unde s anding Pa e ns and T ends in
Income Mobili y h ough Mul i e se Analysis.” Ame ican Sociological Re iew
88:600–626.
E ikson, Robe and John H Gold ho pe. 1992. The cons an lux: A s udy o class
mobili y in indus ial socie ies. Ox o d Uni e si y P ess.
E ikson, Robe and Jan O. Jonsson. 1996. Can Educa ion Be Equalized? The
Swedish Case in Compa a i e Pe spec i e. Boulde , CO.: Wes iew P ess.
Fa kas, Geo ge. 2003. “Cogni i e skills and noncogni i e ai s and beha io s in
s a i ica ion p ocesses.” Annual Re iew o Sociology 29:541–562.
Fea he man, Da id L. 1981. “Social s a i ica ion and mobili y: wo decades o
cumula i e social science.” Ame ican Beha io al Scien is 24:364–385.
40
Williamson, Oli e E. 1975. Ma ke s and Hie a chies: analysis and an i us impli-
ca ions: a s udy in he economics o in e nal o ganiza ion. The F ee P ess.
Wilme s, Na han. 2018. “Wage S agna ion and Buye Powe : How Buye -Supplie
Rela ions Affec U.S. Wo ke s’ Wages, 1978 o 2014.” Ame ican Sociological
Re iew 83:213–242.
Wilme s, Na han. 2019. “Solida i y wi hin and ac oss wo kplaces: How c oss-
wo kplace coo dina ion affec s ea nings inequali y.” RSF: The Russell Sage Foun-
da ion Jou nal o he Social Sciences 5:190–215.
Wilme s, Na han and Clem Aeppli. 2021. “Consolida ed ad an age: new o ganiza-
ional dynamics o wage inequali y.” Ame ican Sociological Re iew 86:1100–1130.
Wi e een, Di k and Paul A ewell. 2017. “Family backg ound and ea nings in-
equali y among college g adua es.” Social Fo ces 95:1539–1576.
Wi e een, Di k and Paul A ewell. 2020. “Reconside ing he ‘me i oc a ic powe
o a college deg ee’.” Resea ch in Social S a i ica ion and Mobili y 66:100479.
Zhou, Xiang. 2019. “Equaliza ion o selec ion? Reassessing he “me i oc a ic powe ”
o a college deg ee in in e gene a ional income mobili y.” Ame ican Sociological
Re iew 84:459–485.
Zimme man, Se h D. 2019. “Eli e colleges and upwa d mobili y o op jobs and op
incomes.” Ame ican Economic Re iew 109:1–47.
Zwysen, Wou e . 2016. “C owding ou o disad an aged young adul s in Ge many:
backg ound ma e s depending on local labou ma ke .” Eu opean Sociological
Re iew 32:662–674.
47
