Unraveling the gender wage gap: Exploring early career patterns among university graduates

Sandne , Mal e; Yükselen, Ipek
A icle — Published Ve sion
Un a eling he gende wage gap: Explo ing ea ly ca ee
pa e ns among uni e si y g adua es
Sco ish Jou nal o Poli ical Economy
P o ided in Coope a ion wi h:
John Wiley & Sons
Sugges ed Ci a ion: Sandne , Mal e; Yükselen, Ipek (2024) : Un a eling he gende wage gap:
Explo ing ea ly ca ee pa e ns among uni e si y g adua es, Sco ish Jou nal o Poli ical Economy,
ISSN 1467-9485, Wiley, Hoboken, NJ, Vol. 72, Iss. 2,
h ps://doi.o g/10.1111/sjpe.12405
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/323718
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/
Sco J Poli Econ. 2025;72:e12405. wileyonlinelib a y.com/jou nal/sjpe
|
1 o 32
h ps://doi.o g/10.1111/sjpe.12405
Recei ed: 29 Decembe 2023
|
Accep ed: 25 Sep embe 2024
DOI: 10.1111/sjpe.12405
SPECIAL ISSUE ARTICLE
Un a eling he gende wage gap: Explo ing ea ly
ca ee pa e ns among uni e si y g adua es
Mal e Sandne 1 | Ipek Yükselen2
1Nu embe g Ins i u e o Technology and
Ins i u e o Employmen Resea ch (IAB),
Nu embe g, Ge many
2Ins i u e o Employmen Resea ch (IAB)
and Uni e si y o Bambe g, Nu embe g,
Ge many
Co espondence
Mal e Sandne , Nu embe g Ins i u e o
Technology and Ins i u ü A bei sma k -
Und Be u s o schung (IAB), Nu embe g,
Ge many.
Email: [email p o ec ed]
Ipek Yükselen, Ins i u e o Employmen
Resea ch (IAB) and Uni e si y o Bambe g,
Nu embe g, Ge many.
Email: ipek[email p o ec ed]om
Abs ac
A la ge body o li e a u e has shown ha he gende wage
gap is small in he i s yea s a e g adua ion and inc eases
g adually wi h age, la gely because o amily decisions,
o en a penal y caused by childbi h. Howe e , he gende
wage gap immedia ely a e g adua ion has ecei ed less a -
en ion. Using a unique da ase ha links 5000 uni e si y
g adua es wi h mas e 's deg ees o equi alen om a la ge
Ge man uni e si y o de ailed employmen eco ds om he
Ge man social secu i y egis e , we speci ically analyze he
gende wage gap a he i s job and i s dynamics du ing he
ini ial yea s o hei ca ee s a e g adua ion. We ind ha
a signi ican gende wage gap al eady exis s in he i s job
a e g adua ion, e en be o e mos young indi iduals make
amily decisions. Howe e , his gende wage gap dec eases
in he i s yea a e en e ing he labo ma ke and hen in-
c eases slowly o e ime. We a ibu e his ini ial dec ease in
he gende wage gap o emale uni e si y g adua es expe i-
encing g ea e e u ns om i m and occupa ional changes
han hei male coun e pa s. This sugges s ha women
may use hese changes o add ess skill misma ches, which
a e mo e common among women han men in hei i s job.
KEYWORDS
ea ly ca ee , gende wage gap, uni e si y g adua es
JEL CLASSIFICATION
I23, J16, J31, J71
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.
© 2025 The Au ho (s). Sco ish Jou nal o Poli ical Economy published by John Wiley & Sons L d on behal o Sco ish Economic
Socie y.
2 o 32
|
SANDNER and YÜKSELEN
1 | INTRODUCTION
Despi e ad ances in women's educa ion and ca ee oppo uni ies in ecen decades, a pe sis en gende wage gap
emains p e alen in economically ad anced na ions (Goldin, 2014; Oli e i & Pe ongolo, 2016), his gap is e en
la ge among indi iduals wi h highe le els o educa ion (Blau & Kahn, 2017; OECD, 2022). Many s udies ha e
examined he gende wage gap among highly educa ed indi iduals and ound ha women's lowe labo supply and
mo e equen ca ee in e up ions (mainly due o child ca e) compa ed o hose o men a e he main easons o
his gende wage gap.1
Less is known abou he exis ence and de elopmen o a gende wage gap a he beginning o a ca ee . This
lack is su p ising gi en he la ge ele ance o s a ing wages and ea ly ca ee wage g ow h on uni e si y g adua es'
u u e labo ma ke ou comes and po en ially on he gende wage gap. Fo example, Oye (2006), Kahn (2010),
and O eopoulos e al. (2012) show ha labo ma ke condi ions a he beginning o he ca ee , such as ecessions,
can ha e an impac on en y wages and, consequen ly, on wages in he long un. Mo eo e , p io wages usually
de e mine wage inc eases due o p omo ions wi hin he same i m (G aham e al., 2000); e en wage inc eases as
a esul o a job change a e usually based on p e ious wages (Hansen & McNichols, 2020). These indings indica e
ha en y wages play a c ucial ole in de e mining u u e wages o e he long un, signi ican ly con ibu ing o he
o igin o he gende wage gap. Knowledge abou he ea ly gende wage gap is also impo an o de eloping new
o adap ing exis ing counseling policies o p o ide e ec i e job sea ch s a egies and o challenge gende no ms
in ca ee choices o g adua es.
Whe he a gende wage gap exis s in he i s yea s a e g adua ion is heo e ically ambiguous. Pa icula ly in
he i s job, some common easons o pay di e ences be ween men and women, such as amily- ela ed decisions
(e.g., childbi h o ma iage), ca ee - ela ed de elopmen s (e.g., p omo ions), wo k expe ience, and i m- speci ic
ne wo ks, may no ye be ele an o less ele an han la e in he ca ee .2 The e o e, we expec no o only a small
gende wage gap in he ini ial job, especially when we accoun o gende di e ences in he ield o s udy and he
cha ac e is ics o he employe in he i s job.
Howe e , pa icula ly in he i s job, bo h he applican s and he i ms ace conside able unce ain y. Fi ms
can assess only he labo ma ke p oduc i i y o candida es wi hou p io wo k expe ience based on hei uni-
e si y g ades and in e iew pe o mance. Gi en ha women cu en ly end o ha e highe GPAs han men
(Becke e al., 2010; F ancesconi & Pa ey, 2018), we may e en expec a gende wage gap ha is condi ional
on di e ences in he ield o s udy choice and employe cha ac e is ics o a o women in he i s job. On he
o he hand, exis ing s udies show ha emale applican s nego ia e less in job in e iews han male applican s do
(Babcock & Lasche e , 2009; Be and, 2011) and may ace s a is ical disc imina ion a labo ma ke en y (Al onji
& Pie e , 2001; Pinks on, 2006). Fu he mo e, di e ences in p e e ences o pe sonali y ai s, such as isk a e -
sion o o e con idence, can be pa icula ly impo an a he beginning o a ca ee . S udies show ha women a e
mo e isk- a e se (e.g., Co és e al., 2023) and less sel - con iden han men (e.g., Adamecz- Völgyi & Shu e, 2022),
which may lead hem o accep lowe - paying job o e s. As a esul , he gende wage gap could be subs an ially in
a o o men, gi en di e ences in ield o s udy and employe cha ac e is ics a he i s job.
This ambigui y abou he gende wage gap may be e en g ea e in he yea s a e labo ma ke en y when
i ms ha e obse ed he p oduc i i y o hei employees o when g adua es change jobs o inc ease hei
wages. I women ea n less han men in hei i s job as a esul o disc imina ion, his gap may na ow o e
ime as women mo e o less disc imina o y i ms o as employe s lea n abou employees' ue p oduc i i y
o e ime (Al onji & Pie e , 2001; Fa be & Gibbons, 1996). Addi ionally, women may co ec ini ial job choices
1Fo example, see Adda e al. (2017), Kuziemko e al. (2018), Kle en e al. (2019a), and Co es and Pan (2020).
2The mean age o Ge man mo he s a i s bi h was 30.5 in 2021 (Fede al S a is ical O ice, 2022), while he a e age age o labo ma ke en y o
women in ou sample is 27. Mo eo e , he a e age age o women a bi h is expec ed o inc ease wi h he le el o educa ion. The e o e, his issue is
no expec ed o be o high magni ude in he case o women wi h a mas e 's deg ee a labo ma ke en y.
|
3 o 32
SANDNER and YÜKSELEN
based on gende no ms a he han p e e ences by changing jobs. Howe e , he gende wage gap may also
inc ease o e ime due o job changes in he ea ly s ages o a ca ee , as he li e a u e shows ha women gen-
e ally ealize lowe e u ns o job mobili y han men (Alb ech e al., 2018; Topel & Wa d, 1992). In addi ion, he
gap may widen o e ime as amily- ela ed decisions become mo e impo an o e ime. O e all, he gende
wage gap a labo ma ke en y and he dynamics o he gende wage gap in he ea ly yea s o a ca ee emain
unclea and hus equi e examina ion.
This s udy examines he gende wage gap immedia ely a e en e ing he labo ma ke and i s e olu ion du ing
he ini ial yea s o a ca ee o mo e han 5000 uni e si y g adua es wi h a mas e 's deg ee o equi alen . We use
unique adminis a i e da a on g adua es o a la ge Ge man uni e si y linked wi h de ailed social secu i y da a om
he In eg a ed Employmen Biog aphies (IEB). This linked adminis a i e da ase p o ides a wide ange o in o -
ma ion om hese wo da a sou ces, including sociodemog aphic cha ac e is ics o he g adua es, he a ained
uni e si y deg ee, ield o s udy, he inal high school and uni e si y g ades, he da e o en ollmen , and he exac
iming o g adua ion, labo ma ke en y, and any occupa ion o i m changes.
Using hese da a, we i s es ima e he gende wage gap a labo ma ke en y among uni e si y g adua es.
Ou indings show ha males ha e signi ican ly highe wages han emales in hei i s ull- ime job immedia ely
a e g adua ion, despi e ou homogeneous and highly educa ed sample wi h high labo ma ke a achmen . The
es ima ed unadjus ed gende wage gap o app oxima ely 12.5 log poin s co esponds o app oxima ely 10 eu os
pe day o 300 eu os pe mon h. The adjus ed gende wage gap is condi ional on a comp ehensi e se o pe sonal
and p eg adua ion con ols, including g adua ion yea , age, non(Ge man) ci izenship s a us, ield o s udy, he inal
uni e si y g ades, ha ing an app en iceship deg ee, wo ked du ing s udy, and he place o he inal high school
examina ion, is equal o 6.2 log poin s. Including occupa ion ixed e ec s educes he gende wage gap o 4.7 log
poin s. O he pos - g adua ion cha ac e is ics, such as he iming o he i s job, i m ixed e ec s, he sha e o
women in he i m, and he loca ion o he i m, do no subs an ially al e he gende wage gap.
Second, since bo h ca ee pa hs and wages a y widely ac oss ields o s udy (Al onji e al., 2016), we conduc
a subg oup analysis o ou b oad g oups o ields o s udy: economics and business, ma hema ics and na u al
sciences, humani ies and social sciences, and medical s udies. The esul s show ha he unadjus ed ( aw) gende
wage gap in he i s job is p ominen in almos all ield g oups excep medical s udies. The aw gende wage gaps
in each ield g oup a e 8.6 log poin s, 14.1 log poin s, 10.2 log poin s and 1.5 log poin s, espec i ely. Fo ma he-
ma ics and na u al sciences, he gende wage gap disappea s a e con olling o he majo subjec wi hin he ield
o s udy. The adjus ed gende wage gap is he highes in he humani ies and social sciences, a 9.7 log poin s. This
ield g oup also has he lowes a e age daily wage in he i s job, he highes a ia ion in wages, and he highes
sha e o emales.
Thi d, as dynamics a e e y impo an , pa icula ly in he ea ly yea s o a ca ee , and ha e an impac on u u e
wage g ow h, we examine he dynamics o he gende wage gap o e he i s yea s a e labo ma ke en y. Ou
indings e eal a dec ease in he es ima ed gende wage gap in he i s 3 yea s a e labo ma ke en y, ollowed
by an inc ease in subsequen yea s. The la ges educ ion in he wage gap occu s 1 yea a e labo ma ke en y.
Mo eo e , we demons a e ha his dec ease is obse ed only among economics and business g adua es and hu-
mani ies and social sciences g adua es who change bo h i ms and occupa ions wi hin 1 yea o en e ing he labo
ma ke . Howe e , his decline does no occu o g adua es om o he ields o s udy o o hose who emain in
he same i m and occupa ion.
Finally, ou analysis ocuses on wo ield g oups: economics and business and humani ies and social sci-
ences. This analysis shows ha women who change i ms and occupa ions a e hei i s job d i e he decline
in he gende pay gap, as women bene i mo e om hese changes han men. Ou da a e eal ha women
a e mo e likely han men o wo k in a misma ched occupa ion a he i s job. By changing bo h i ms and
occupa ions, women mo e ou o he lowes - paid occupa ions and a e able o co ec his misma ch, leading
o a g ea e inc ease in wages han men. A e compa ing hese empi ical indings wi h se e al heo ies in he
gende wage gap li e a u e, one explana ion o ou inding may be ha women immedia ely a e g adua ion
4 o 32
|
SANDNER and YÜKSELEN
ha e s ong p e e ences o ce ain job and i m ameni ies, such as job meaning and ele ance, o hey ollow
ce ain gende no ms abou i ms and occupa ions leading hem o ini ially accep misma ched jobs. One yea
a e labo ma ke en y, indi iduals' p e e ences o willingness o ollow gende no ms may change, and hey
may co ec his misma ch by changing occupa ion and i m. Howe e , ou da a do no allow o a de ini i e
es o his hypo hesis.
Ou s udy con ibu es o he li e a u e in impo an ways. Fi s , se e al s udies examine he dynamics o he
gende wage gap o e he li e cycle and ind e idence ha he gende wage gap is smalle a younge ages bu
inc eases o e ime, mainly due o amily- ela ed decisions (Alb ech e al., 2018; Be and e al., 2010; Manning
& Swa ield, 2008).3 Al hough hese s udies p o ide aluable insigh s in o he dynamics o he gende wage gap
in gene al, hey do no ocus on he i s job a e g adua ion. The ew pape s ha examine he gende wage gap
a he beginning o a ca ee ely p ima ily on su ey da a. Fo example, Co és e al. (2023) ind in a su ey o
US g adua es ha women ea n 10% less han men in hei i s job. In a ela ed Ge man s udy, F ancesconi and
Pa ey (2018) ound an adjus ed gap o 5–10 log poin s among Ge man college g adua es 12–18 mon hs a e
g adua ion. In con as o his li e a u e, we a e he only s udy o in es iga e he gende wage gap among uni-
e si y g adua es using adminis a i e da a, wi h a ocus on he i s job a e g adua ion.4
Adminis a i e da a help o a oid epo ing bias ha can occu in su ey- based s udies a he beginning o
a ca ee due o equen job changes. Fu he mo e, he adminis a i e na u e o he da a o e comes conce ns
associa ed wi h missing da a, esponse a es, o measu emen e o due o e ospec i e ques ions. Mos o he
o he da a used o s udy he gende wage gap ei he lack comp ehensi e in o ma ion on g adua es' p eg adua ion
cha ac e is ics ( ield o s udy, GPA) o a e unable o ack g adua es as hey ansi ion in o he labo ma ke and
lack in o ma ion on g adua es' occupa ion, indus y, and o he impo an employmen cha ac e is ics. In con as ,
ou linked da a p o ide access o accu a e and comp ehensi e measu es o human capi al de e minan s o p o-
duc i i y, including academic g ades and ield o s udy, as well as de ailed in o ma ion on employmen , wages, and
occupa ions.
Second, ou s udy p o ides unique insigh s in o ea ly ca ee job dynamics and hei impac on he gende
wage gap. A he beginning o ca ee s, a high le el o in o ma ion ic ion can lead o poo job ma ches in he
labo ma ke o ecen g adua es (Ves e lund, 1997). F ed iksson e al. (2018) highligh high sepa a ion a es and
job changes among inexpe ienced employees due o limi ed in o ma ion abou he labo ma ke . The li e a u e
shows ha job changes a e in gene al associa ed wi h wage g ow h bu also ha men a e mo e likely o change
jobs and end o bene i mo e om job mobili y han women, he eby exace ba ing he gende wage gap o e
ime (Alb ech e al., 2018; Del Bono & Vu i, 2011; Manning & Swa ield, 2008; Topel & Wa d, 1992). Howe e ,
hese s udies do no ocus on he i s yea s a e labo ma ke en y because obse ing his c ucial ea ly pe iod
whe e e u ns o job changes may be di e en is di icul wi hou de ailed adminis a i e da a. In con as , we
a e able o ollow all g adua es wi hou a i ion o e ime, which allows us o obse e he exac iming o any
job changes o job sea ch pe iods wi hin he i s ew yea s a e labo ma ke en y. This in o ma ion allows us
o obse e he sha e o emale and male g adua es om each ield o s udy who change jobs and o obse e he
e u ns o hei mobili y, which may ha e long- las ing e ec s on hei u u e labo ma ke ca ee s.
The emainde o his pape is s uc u ed as ollows. Sec ion 2 desc ibes he da ase and i s ad an ages and
sho comings, cha ac e izes he sample o uni e si y g adua es used in he analysis, and p esen s some desc ip-
i e s a is ics. The esul s o he es ima ed gende wage gap a labo ma ke en y and he dynamics o he gende
wage gap o e he i s ew yea s o a ca ee a e p esen ed in Sec ions 3 and 4, espec i ely. We examine gende
di e ences in i m and occupa ional mobili y in Sec ion 5 and he unde lying easons o his mobili y in Sec ion 6
be o e concluding he pape in Sec ion 7.
3The e ec o he child penal y on emales' labo ma ke ou comes is explo ed in se e al s udies, o example, Kle en e al. (2019b) and Dus mann
e al. (2009).
4The s udies by Kunze (2003, 2005) used adminis a i e da a bu ocused on younge g adua es who had comple ed an app en iceship.

|
5 o 32
SANDNER and YÜKSELEN
2 | THE LINKED ADMINISTRATIVE DATASET AND UNIVERSITY
GRADUATES
2.1 | Da a
This s udy is based on a unique adminis a i e da ase o g adua es om a la ge uni e si y linked wi h egis y
da a om he Ge man In eg a ed Employmen Biog aphies (IEB) o he Ins i u e o Employmen Resea ch (IAB).
The linked da ase combines de ailed s udy in o ma ion on each g adua e om he uni e si y egis y wi h
in o ma ion on indi idual employmen eco ds co e ing he whole employmen biog aphy o jobs subjec o social
secu i y con ibu ions om he IEB da ase .
The a ailable da ase om he uni e si y co e s all g adua es o his uni e si y om 1995 un il 2016. Du ing
he obse a ion pe iod, almos all he ields o s udy a e conside ed excep o enginee ing deg ees. The da a a e
highly eliable, as hey a e based on adminis a i e eco ds om he uni e si y egis y. The da ase p o ides in-
o ma ion on he pe sonal cha ac e is ics o each g adua e, such as yea o bi h, gende , na ionali y, dis ic , and
g ade o he ce i ica e o gene al quali ica ion o uni e si y admission (Abi u ), he ea e e e ed o as he inal
high school g ade poin a e age (GPA). The da ase also includes s udy- ela ed cha ac e is ics a he uni e si y,
such as he ield o s udy, he ype o uni e si y deg ee a ained, he inal GPA, and he da es o en ollmen and
g adua ion.
The IEB is a la ge adminis a i e da ase o indi iduals' employmen biog aphies p o ided by he IAB o he
pe iod 1975–2019. The in o ma ion p o ided by he da ase is highly eliable, as i is a legal equi emen in
Ge many o all employe s o p o ide in o ma ion on hei employees o he Ge man Social Secu i y Adminis a ion.
The IEB da ase includes indi iduals in employmen co e ed by social secu i y, excluding sel - employed indi idu-
als and ci il se an s. Thus, he IEB da ase co e s app oxima ely 80% o he o al labo o ce in Ge many. In ad-
di ion o he p ecise iming o employmen and ou - o - employmen spells, he da ase p o ides in o ma ion on
g oss daily wages, indus y, occupa ion ( h ee- digi ), ull- ime s a us, and o he employmen cha ac e is ics
(Do ne e al., 2010). The da a om he uni e si y a e me ged wi h he IEB da ase using a linkage p ocedu e es-
ablished a he IAB based on an indi idual's ull name, sex, and da e o bi h, wi h a 90% ma ch a e (Mölle &
Rus , 2017).5
2.2 | Sample choice
The ocus o his s udy is on indi iduals wi h a mas e 's deg ee o equi alen wi h a ailable uni e si y GPA da a.
We u he ocus on g adua es who a e wo king ull- ime in hei i s egula job wi h a daily wage o a leas 10
eu os6 and we omi g adua es who a e no ull- ime employed in hei i s job a e g adua ion, ha is, pa - ime,
mini- jobs, in e nships, wo king s uden s, e c., e en i hey subsequen ly swi ch o ull- ime employmen . I an
indi idual has mo e han one wage spell a a gi en ime, we choose he “main” employmen spell as de ined by he
IAB.7
5Please see he s udy by Mölle and Rus (2017) o a mo e de ailed explana ion o he ma ching p ocedu e.
6Wages a e de la ed o he yea 2010 using he consume p ice index.
7Since in o ma ion on wo king hou s is no a ailable in he linked da ase , we ocus on ull- ime jobs in o de o elimina e a po en ial bias in he
gende wage gap induced by di e ences in wo king hou s. Since we ocus on ull- ime employees in ou main analysis, he wo king hou s o men
and women should be easonably compa able. Howe e , e en i employees a e ai ly homogeneous in e ms o ull- ime employmen , males may
s ill wo k mo e hou s han emales, allowing hem o ea n highe wages (e.g., Goldin & Ka z, 2016). The s udy by F ancesconi and Pa ey (2018)
documen s ha di e ences in hou s wo ked among ull- ime employees do no signi ican ly explain he gende wage gap among Ge man g adua es
app oxima ely 12 o 18 mon hs a e g adua ion. The e o e, we expec ha ou esul s a e no d i en by di e ences in wo king hou s be ween
ull- ime employed emale and male g adua es.
6 o 32
|
SANDNER and YÜKSELEN
We ocus on mas e 's o equi alen g adua es o conside he mos policy- ele an g oup wi h g ea e labo
ma ke a achmen , and he esul s a e easie o in e p e o a mo e homogeneous g oup.8 Fu he mo e, e u ns
o mas e 's g adua es expec ed o be highe compa ed o oca ional aining, high school educa ion, o only a
bachelo 's deg ee (Al onji e al., 2016). In addi ion, mas e 's g adua es ha e a highe deg ee o a achmen o he
labo ma ke and o m a ela i ely homogeneous g oup, making i easie o iden i y ac o s ha impac en y- le el
wages. They a e also an ideal g oup o s udy ea ly ca ee gende wage gaps, as child ca e, a key ac o in he wage
gap o highly educa ed indi iduals, has less o an impac a his s age.
Fu he mo e, we exclude indi iduals who a e olde han 35 yea s (1.5% o 104 indi iduals). We also omi g ad-
ua es wi h a gap o mo e han 15 mon hs be ween g adua ion and hei i s employmen spell (14% o 999 indi-
iduals), as hese indi iduals may ha e spen ime ab oad o al eady wo ked on a sel - employed basis (which is no
cap u ed by he da a). Since ou main analysis ocuses on he i s ull- ime job a e g adua ion and subsequen
yea s, we keep g adua es wi h wage spells a he beginning o hei i s job and 1 yea a e hei i s job (8.6%
o 478 indi iduals we e d opped). Finally, a e d opping obse a ions wi h missing alues, he inal sample o he
main wage es ima ions consis s o 5212 indi iduals.
2.3 | Desc ip i e s a is ics
Table 1 p esen s desc ip i e s a is ics on labo ma ke en y o ou p e e ed sample o uni e si y g adua es
in ull- ime employmen , which we use o he wage analysis in he ollowing sec ions. While Panel A o Table 1
documen s p e- g adua ion cha ac e is ics, such as uni e si y and high school GPA, du a ion o s udy, non-
Ge man ci izenship, and o he s, Panel B p esen s pos - g adua ion cha ac e is ics, including cha ac e is ics o
he i s job. G adua es' uni e si y and high school GPAs ange om 1 ( he bes possible g ade) o 4 ( he lowes
passing g ade).
Panel A o Table 1 shows ha bo h men and women comple e hei deg ees in app oxima ely i e and a hal
yea s on a e age, al hough women g adua e a a younge age. The majo i y o g adua es a he uni e si y
(a ound 70%) acqui e some o m o wo k expe ience be o e g adua ion, wi h emales being mo e likely o wo k
du ing hei s udies han males. In addi ion, consis en wi h he li e a u e, he sha e o emale g adua es in-
c eases wi h g adua ion coho , wi h he male– emale a io e e sing e en in he mos ecen coho g oup
(2007–2010). This inding is also in line wi h he o e all popula ion o Ge man g adua es. Consis en wi h he
li e a u e (see, e.g., Becke e al., 2010), emale g adua es en oll a he uni e si y wi h be e inal high school
g ades and lea e he uni e si y wi h sligh ly be e uni e si y g ades (by a ound 11% o he sample s anda d
de ia ion) han males.9 Females a e also mo e han wice as likely o g adua e in he humani ies and social
sciences. G adua es in ma hema ics and na u al sciences accoun o nea ly a qua e o all male g adua es,
compa ed o only 8% o all emale g adua es. Ne e heless, economics and business emain he dominan
ields o choice o bo h gende s. Finally, he sha e o women s udying medicine is app oxima ely 10 pe cen -
age poin s highe han ha o men. Table 1 also shows ha emale g adua es a e mo e likely han male g adu-
a es o ea n a magis e o s a e examina ion deg ee. The as majo i y o g adua es ha e a diploma deg ee,
wi h a g ea e p opo ion o men han women.
Table C2 in he Appendix 3 compa es ou es ima ion sample wi h o icial Ge man egis e da a and o he ep e-
sen a i e s udies. Fo 2010, ou sample shows a sligh ly highe p opo ion o women (50%) han he da a om he
Fede al S a is ical O ice (46%). The sha e o emales by ield o s udy is also compa able. The la ges di e ence is
obse ed in ma hema ics and na u al sciences. Acco ding o ou da a, he p opo ion o emales in his ield is 18%,
8Ano he eason o ocusing on his g oup is ha he Bologna P ocess e o m was implemen ed in Ge many be ween 2005 and 2010, and only a
small p opo ion o g adua es in he sample ha e a mas e 's deg ee. Be o e he Bologna P ocess, bachelo 's and mas e 's deg ees we e combined
in o diploma o magis e deg ees.
9In ou da a, inal high school g ades a e a ailable only o he da a beginning wi h he 2001 g adua ion coho .
|
7 o 32
SANDNER and YÜKSELEN
TABLE 1 Desc ip i e s a is ics.
Male Female
Di .Mean (S d. de .) Mean (S d. de .)
Panel A: P e- g adua ion and pe sonal cha ac e is ics
Final high- school g ade (Abi u )2.245 (0.616) 2.077 (0.594) 0.168***
Indi iduals 1476 1200
Final uni e si y g ade 2.058 (0.604) 1.998 (0.568) 0.060***
Non- Ge man ci izenship 0.015 (0.121) 0.033 (0.179) −0.019***
Age a g adua ion 27.238 (1.864) 26.579 (1.906) 0.660***
Du a ion o s udy 5.592 (1.308) 5.631 (1.372) −0.039
App en iceship 0.065 (0.247) 0.061 (0.239) 0.004
Wo ked du ing s udies 0.673 (0.469) 0.742 (0.438) −0.069***
O igin in he same ede al s a e
as he uni e si y
0.876 (0.329) 0.861 (0.346) 0.016
G adua ion yea
1995–1998 0.283 (0.451) 0.156 (0.363) 0.127***
1999–2002 0.227 (0.419) 0.172 (0.377) 0.055***
2003–2006 0.227 (0.419) 0.269 (0.443) −0.042***
2007–2010 0.263 (0.440) 0.403 (0.491) −0.141***
Field o s udy
Economics and business 0.469 (0.499) 0.328 (0.469) 0.141***
Ma hema ics and na u al
sciences
0.224 (0.417) 0.077 (0.267) 0.147***
Humani ies and social sciences 0.111 (0.314) 0.300 (0.458) −0.189***
Medical s udies 0.196 (0.397) 0.295 (0.456) −0.099***
Type o deg ee
Diploma 0.747 (0.435) 0.576 (0.494) 0.172***
Magis e 0.046 (0.210) 0.114 (0.317) −0.068***
Mas e 0.010 (0.102) 0.015 (0.123) −0.005
S a e examina ion
(S aa sexamen)
0.196 (0.397) 0.295 (0.456) −0.099***
Indi iduals 3258 1954
Panel B: Pos - g adua ion cha ac e is ics
Le he s a e 0.196 (0.397) 0.179 (0.383) 0.118
Le he ci y 0.683 (0.465) 0.655 (0.476) 0.028*
Mean job sea ch du a ion 3.747 (3.128) 3.817 (3.039) −0.070
Du a ion o job sea ch
Less han 1 mon h 0.190 (0.392) 0.161 (0.368) 0.028***
1–3 mon hs 0.326 (0.469) 0.319 (0.466) 0.007
3–5 mon hs 0.214 (0.410) 0.247 (0.431) −0.033***
Mo e han 5 mon hs 0.270 (0.444) 0.272 (0.445) −0.002
(Con inues)
8 o 32
|
SANDNER and YÜKSELEN
while acco ding o he egis e da a, i is 35%. In ou sample, s uden s ha e a sligh ly be e inal high school GPA
han in he su ey da a aken om he s udy Simeane e al. (2014) and uni e si y g ades ha a e simila o hose
o he ep esen a i e sample aken om he su ey da a used by F ancesconi and Pa ey (2018). The g adua es in
ou sample a e, on a e age, abou 5 mon hs younge because we use examina ion a he han ex- ma icula ion
da es. No ably, 11% o ou s uden s a e non- Ge man, compa ed o 22% in he su ey da a used by F ancesconi
and Pa ey (2018), as we canno obse e indi iduals in ou da a i hey mo e o ano he coun y a e g adua ion.
Panel B o Table 1 p esen s pos - g adua ion and employmen cha ac e is ics, such as mobili y, he ime be-
ween g adua ion and he i s ull- ime job,10 es ablishmen size, and he sha e o women in he es ablishmen o
he i s job.11 The able shows ha app oxima ely 70% o he g adua es ind hei i s ull- ime job ou side he
ci y o he uni e si y loca ion, wi h males being sligh ly mo e mobile han emales. On a e age, emale g adua es
ake longe o ind hei i s job han male g adua es, ha is, app oxima ely 3.7 and 3.8 mon hs, espec i ely. A
b eakdown o he du a ion o job sea ches in o di e en ca ego ies shows ha he sha e o male g adua es wi h
a job sea ch du a ion o less han 1 mon h is g ea e . Finally, emale and male g adua es end o wo k in es ablish-
men s o simila size. Howe e , in line wi h he li e a u e, women a e mo e likely o wo k in es ablishmen s wi h a
highe p opo ion o emale employees. A po en ial explana ion o his si ua ion migh be he so ing o uni e si y
g adua es in o speci ic indus ies by gende , esul ing in emale- domina ed indus ies (Helle s ein e al., 2011).
3 | THE GENDER WAGE GAP AT LABOR MARKET ENTRY
We begin ou analysis by examining gende wage di e ences a labo ma ke en y.12 To iden i y gende di e -
ences, we es ima e he ollowing eg ession equa ion:
10He ea e e e ed o as “job sea ch du a ion”, e en hough his ime is no necessa ily spen sea ching o a job.
11 The da a include in o ma ion only on es ablishmen s, no i ms. Howe e , in his pape , we use he e ms “es ablishmen ” and “ i m”
in e changeably.
12 The e m labo ma ke en y e e s o he i s job a e uni e si y g adua ion; we use hese e ms in e changeably.
(1)
Yi
=
𝛼
+
𝛾Femalei
+
𝛽Xi+ϵ
i
Male Female
Di .Mean (S d. de .) Mean (S d. de .)
Fi m size
Less han 25 employees 0.238 (0.426) 0.247 (0.432) −0.009
25–250 employees 0.273 (0.445) 0.266 (0.442) 0.007
250–2000 employees 0.254 (0.436) 0.273 (0.446) −0.018
Mo e han 2000 employees 0.235 (0.424) 0.214 (0.411) 0.020*
Sha e o women in he i m
Less han 40% 0.356 (0.479) 0.201 (0.401) 0.155***
40%–70% 0.405 (0.491) 0.383 (0.486) 0.022
Mo e han 70% 0.239 (0.427) 0.417 (0.493) −0.177***
Indi iduals 3258 1954
No e: This able shows summa y s a is ics o g adua es' p e- g adua ion and pos - g adua ion cha ac e is ics. The sample
consis s o g adua es wi h a mas e 's deg ee o he equi alen , who wo ked ull- ime a hei i s job a e g adua ion
and who ha e a wage spell 1 yea a e hei i s job. ***, ** and * indica e signi icance a he 1%, 5%, and 10% le els.
TABLE 1 (Con inued)
|
15 o 32
SANDNER and YÜKSELEN
P e ious esea ch shows ha i m and/o occupa ional mobili y a ec s wages and con ibu es o wage
g ow h (Ba el & Bo jas, 1981; Topel & Wa d, 1992); mobili y is especially impo an in he ea ly s ages o a
ca ee (Alb ech e al., 2018) and ha men end o bene i mo e om job mobili y han women (Del Bono &
Vu i, 2011; Manning & Swa ield, 2008). To analyze whe he and o which ex end job changes in he ea lies
ca ee s age explain he d op in he gende wage gap, we con inue ou in es iga ion by sepa a ing he sample
in o g adua es who s ay in he same i m and/o occupa ion (Column 2 o Table 4), hose who change occupa-
ions bu emain in he same i ms (Column 3), hose who change i ms bu emain in he same occupa ion
TABLE 4 The gende wage gap by job change s a us.
Dependen a iable: Log daily wage
Pooled S aye s
Only i m
change s
Only occupa ion
change s
Fi m and
occupa ion
change s
(1) (2) (3) (4) (5)
Panel A: Economics, business, humani ies, and social sciences
1 yea a e ×
emale
0.036 *** −0.000 0.038 0.061 0.193 ***
(0.012) (0.008) (0.063) (0.083) (0.070)
1 yea a e 0.130 *** 0.098 *** 0.252 *** 0.248 *** 0.271 ***
(0.007) (0.006) (0.042) (0.063) (0.039)
Female −0.098
***
−0.056 *** −0.156 ** −0.139 * −0.340 ***
(0.015) (0.014) (0.068) (0.074) (0.069)
Sha e o emales 10.737 0.115 0.101 0.131
Sha e o males 10.796 0.091 0.058 0.105
R- squa ed 0.240 0.274 0.306 0.323 0.328
Indi iduals 3114 2407 267 194 312
Panel B: Ma hema ics, na u al sciences, and medical s udies
1 yea a e ×
emale
−0.001 0.006 −0.010 −0.140 −0.088
(0.011) (0.009) (0.042) (0.132) (0.095)
1 yea a e 0.150 *** 0.127 *** 0.182 *** 0.341 *** 0.386 ***
(0.008) (0.007) (0.031) (0.098) (0.054)
Female −0.019 −0.024 0.008 0.074 0.049
(0.016) (0.016) (0.034) (0.171) (0.096)
Sha e o emales 10.783 0.144 0.032 0.081
Sha e o males 10.838 0.086 0.037 0.070
R- squa ed 0.393 0.399 0.624 0.499 0.382
Indi iduals 2098 1718 204 60 137
No e: This able shows he gende wage gap a labo ma ke en y by job change s a us based on he OLS model
speci ied in Equa ion (2). The sample consis s o g adua es wi h a mas e 's deg ee o equi alen who wo k in a ull- ime
job as hei i s job a e g adua ion and who ha e a wage spell 1 yea a e hei i s job. The dependen a iable is he
log g oss daily wage a he i s job. The es ima ions include pe sonal and p e- g adua ion cha ac e is ics as con ols.
The pe sonal cha ac e is ics include age and ha ing Ge man ci izenship. The p e- g adua ion cha ac e is ics include
du a ion o s udy, place o high school inal exam, and wo king du ing s udying. All es ima ions include he beginning
mon h o he i s job as a con ol. Robus s anda d e o s a e in pa en heses. ***, **, and * indica e signi icance a he
1%, 5%, and 10% le els.

16 o 32
|
SANDNER and YÜKSELEN
(Column 4), and hose indi iduals who change i ms and ake up a new occupa ion (Column 5).21 Ou job mo e
ca ego ies ollow Fi zenbe ge e al. (2015) who analyze he e u ns o occupa ion and i m swi che in a sam-
ple o app en ices.
Panel A o Table 4, which ocuses on g adua es in economics, business, humani ies, and social sciences, demon-
s a es ha he gende wage gap does no dec ease signi ican ly o hose indi iduals who ei he s ay in he same
i m and/o occupa ion 1 yea a e s a ing hei i s job (Columns 2–4). In con as , emale i ms and occupa ion
change s inc ease hei wages on a e age by app oxima ely 19 log poin s mo e han hei male coun e pa s
(Column 5). This inc ease mus be conside ed in he con ex ha males who change ei he i m, occupa ion, o
bo h bene i om hese changes by app oxima ely 25–27 log poin s, while he s aye s inc ease hei wages by
only 10 log poin s. In addi ion, he ini ial gende wage gap is la ge o indi iduals who change i ms, occupa ions,
o bo h han o indi iduals who emain in hei occupa ion in he same i m. The g oup ha changes i ms and
occupa ions has he la ges ini ial gap (almos 34 log poin s). In e es ingly, he alloca ion o men and women o he
ou g oups is ela i ely simila ; hus, di e ences in sha es do no seem o explain he di e en e olu ions o he
gende wage gap a e 1 yea .
Fo ma hema ics, na u al sciences, and medical s udies, we ind no educ ion in he gende wage gap in he
i s 12 mon hs a e s a ing he i s job, and we obse e no ini ial gap o any o he change g oups (Panel
B o Table 4). Howe e , e en o hese ields, he esul s show ha mo e s ha e he highes wage g ow h,
be ween 18 and 39 log poin s. O e all, he able shows ha women in economics, business, humani ies, and
social science bene i mo e han men om a comple ely new s a a e hei i s job, which includes a change
in i m and occupa ion. This new s a d i es he obse ed educ ion in he gende wage gap wi hin 1 yea
a e he i s job.
As a nex s ep in ou analysis, Figu e 3 shows he dynamics o wages o e 5 yea s a e he i s job o s aye s
and hose g adua es who change bo h occupa ion and i m. Con i ming he esul s shown in Table 4, emale and
male mo e s ini ially ea n lowe wages on a e age han s aye s. Howe e , he wage di e ence be ween s aye s
and mo e s is g ea e o emales han o males, due o he e y low en y wages o hose emales who la e
change bo h occupa ion and i m. In summa y, women wi h low en y wages appea o co ec hei low wages
mo e han men by changing hei i m and occupa ion wi hin 1 yea o labo ma ke en y.
6 | WHY IS SWITCHING FIRM AND OCCUPATION IN THE BEGINNING
OF THE CAREER MORE BENEFICIAL FOR FEMALES THAN FOR MALES?
In his sec ion, we use ou ich adminis a i e da a o in es iga e why women bene i mo e han men om chang-
ing i ms and occupa ions a e hei i s job a e g adua ion. Al hough we a e no aiming o iden i y causal e -
ec s o his highe emale bene i , we a e con iden o elie some in e es ing pa e ns. We use wo es ima ion
app oaches o conduc ou analysis. Fi s , we es ima e whe he women who change i ms and occupa ions di e
om men who change i ms and occupa ions in e ms o demog aphic cha ac e is ics, uni e si y ou comes, and
cha ac e is ics o hei i s job and whe he hese gende di e ences di e om hose o s aye s. Second, we es-
ima e whe he i m and occupa ion cha ac e is ics change di e en ly o males and emales a e job ansi ions.
A he end o he sec ion, he esul s o he wo es ima ion app oaches a e discussed wi h espec o common
heo e ical explana ions o he gende wage gap. Wi h ega d o bina y ou comes, we also applied logi es ima-
ions, which yielded simila esul s.
Table 5 compa es gende di e ences in pe sonal and p e- g adua ion cha ac e is ics (bo h o which a e con-
s an o e ime) be ween i ms and occupa ion change s and s aye s. In Table 6, Panel A compa es gende di -
e ences in he cha ac e is ics o he i s job o indi iduals who change i ms and occupa ions wi h hose who
21We de ine an occupa ion change as when he h ee- digi occupa ion code changes.
|
17 o 32
SANDNER and YÜKSELEN
s ay in he same posi ion. Panel B o Table 6 examines gende di e ences in he cha ac e is ics o he i s and
subsequen jobs 1 yea la e o occupa ion and i m change s. Table C5 in he Appendix 3 shows he mean alues
o all a iables in Tables 5 and 6 by gende and he co esponding mean gende di e ences o s aye s and o
i m and occupa ion change s.
Row (1) o Table 5 epo s he in e ac ion coe icien s be ween he emale a iable and a dummy o i m and
occupa ion change. The coe icien s in Row (1) indica e ha none o he pe sonal cha ac e is ics, p e- g adua ion
cha ac e is ics, o job sea ch cha ac e is ics exhibi g ea e di e ences be ween males and emales who change
i ms and occupa ions han be ween males and emales who s ay in he same i m and occupa ion.
Howe e , Table 6 demons a es ha ou o se e al cha ac e is ics, job- educa ion misma ch and occupa ional
ank (Columns 6–11) a e wo cha ac e is ics ha di e be ween males and emales who swi ch occupa ions and
i ms in hei i s job and de elop di e en ly a e he job change. We sepa a e job- educa ion misma ches in o
wo ypes: ho izon al and e ical misma ches. A ho izon al misma ch is a ield- occupa ion misma ch in which he
employee's ield o s udy does no ma ch he ield equi ed o he job. A e ical misma ch is a skill misma ch
whe e he skill le el o he employee's quali ica ion does no ma ch he equi emen s o he job. Since ou sample
includes highly skilled uni e si y g adua es, only jobs o which uni e si y g adua es a e o e quali ied a e de ined
as e ical misma ches.22 In addi ion, occupa ion ank is a measu e ha anks occupa ions by hei a e age wage
(Column 9 o Table 6).23
22A la ge body o li e a u e has epo ed ha bo h e ical and ho izon al misma ches ha e a nega i e e ec on wages (Bouda ba &
Mon ma que e, 2009; Heijke e al., 2003; Robs , 2007; Wolbe s, 2003).
23A e age wages wi hin h ee- digi occupa ion codes a e calcula ed using he SIAB da a, which ep esen 2% o he IEB da a.
FIGURE 3 The Dynamics o Wages o Job S aye s and Occupa ion and Fi m Change s by Gende . These
igu es plo he dynamics o wages o e 5 yea s a e he i s job o s aye s (le panel) and o i m and
occupa ion change s ( igh panel), and he sample sizes a e 312, and 137 espec i ely. The dependen a iable
is he log g oss daily wage. The g adua ion yea , and pe sonal and p e- g adua ion cha ac e is ics a e added
as con ols. The pe sonal cha ac e is ics include age and ha ing Ge man ci izenship. The p e- g adua ion
cha ac e is ics include du a ion o s udy, place o high school inal exam, and wo king du ing s udy. All
es ima ions include he beginning mon h o he i s job as a con ol. Addi ionally, we con ol o ha ing a child
be ween he yea s.
18 o 32
|
SANDNER and YÜKSELEN
TABLE 5 Pe sonal and p e- g adua ion cha ac e is ics o i m and occupa ion change s.
Dependen a iables
Pe sonal cha ac e is ics P e- g adua ion cha ac e is ics
Finding i s job
cha ac e is ic
Age a he
i s job
Non-
ge man
Du a ion o
s udy
Wo king du ing
s udying App en
O igin same
s a e as he Uni.
Final Uni.
g ade
Du a ion o job
sea ch
(1) (2) (3) (4) (5) (6) (7) (8)
Female × i m and
occupa ion change s
0.058 −0.021 0.226 −0.012 −0.019 −0.031 0.071 −0.342
(0.269) (0.017) (0.171) (0.053) (0.037) (0.041) (0.073) (0.366)
Fi m and occupa ion
change s
0.444 *** 0.003 0.063 0.002 0.054 ** 0.021 0.026 −0.035
(0.156) (0.010) (0.096) (0.037) (0.026) (0.027) (0.045) (0.246)
Female −0.803 *** 0.026 *** −0.053 0.126 *** −0.010 0.034 ** −0.206 *** 0.141
(0.082) (0.007) (0.053) (0.018) (0.010) (0.014) (0.023) (0.131)
Means o dependen
a iable
27.309 0.023 5.515 0.719 0.074 0.858 2.115 3.859
R- squa ed 0.042 0.006 0.003 0.018 0.004 0.002 0.030 0.001
Indi iduals 2719 2719 2719 2719 2719 2719 2719 2719
No e: The able documen s gende di e ences in pe sonal, p e- g adua ion, and i s - job cha ac e is ics, be ween i ms and occupa ion change s and s aye s. The sample size is
2719, including s aye s (Column 2, Table 4) and i m and occupa ion change s (Column 5, Table 4). Each Column is a di e en es ima ion ha is ime in a ian . We use he ollowing
es ima ion equa ion,
whe e he dependen a iable changes in each column. The es ima ions include a emale dummy, a dummy a iable o i m and occupa ion change dummy (
JobChange si
) (= 1 i
an indi idual changes he o his i m and occupa ion wi hin 1 yea a e labo ma ke en y, = 0 i an indi idual s ays a he same i m and occupa ion), and an in e ac ion o hese
dummies. Robus s anda d e o s a e in pa en heses. ***, **, and * indica e signi icance a he 1%, 5%, and 10% le els.
(4)
Yi
=𝛽
0
+𝛽
1Femalei
+𝛽
2JobChange si
+𝛽
3Femalei
⋅
JobChange si
+𝛾
g adyea i
+𝜀
i,
|
19 o 32
SANDNER and YÜKSELEN
TABLE 6 Job cha ac e is ics o i m and occupa ion change s.
Dependen a iables
Median daily Sha e o Sha e o Sha e o Log Ho izon al Ve ical Ho izon al o Occupa ion Occupa ion Occupa ion
Log wage
o ull- ime
employees
Pa - ime
employees
High quali ied
employees
Women
in a i m Fi m size Misma ch Misma ch
Ve ical
misma ch Rank
Rank
<
quan ile 10
Rank
>
quan ile 90
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Panel A: Fi s job cha ac e is ics o i m and occupa ion change s
Female × i m
and occupa ion
change s
−0.086 * −0.059 −0.013 0.044 −0.204 0.008 0.117 ** 0.070 −22.371 * 0.098 ** −0.025
(0.049) (0.041) (0.032) (0.027) (0.252) (0.056) (0.058) (0.053) (11.815) (0.044) (0.034)
Fi m and
occupa ion
change s
−0.120 *** −0.007 −0.063 *** −0.006 −0.341 ** 0.119 *** 0.122 *** 0.150 *** −7.680 0.017 −0.002
(0.028) (0.032) (0.021) (0.017) (0.166) (0.035) (0.039) (0.037) (7.488) (0.024) (0.026)
Female −0.041 *** −0.034 ** −0.016 0.071 *** −0.102 0.122 *** 0.005 0.051 ** −15.238 *** 0.013 −0.036 ***
(0.014) (0.016) (0.011) (0.009) (0.090) (0.017) (0.021) (0.021) (3.602) (0.012) (0.013)
Mean o
dependen
a iables
4.617 0.220 0.372 0.482 5.099 0.219 0.506 0.594 226.031 0.104 0.104
R- squa ed 0.029 0.003 0.009 0.035 0.005 0.032 0.014 0.018 0.016 0.009 0.004
Indi iduals 2719 2719 2719 2719 2719 2719 2719 2719 2719 2719 2719
Panel B: Jobs cha ac e is ics be o e and a e he job change wi hin i m and occupa ion change s
Female × yea
a e
0.046 0.103 * 0.011 −0.007 −20.260 −0.060 −0.131 * −0.153 ** 21.816 −0.115 ** −0.035
(0.053) (0.054) (0.036) (0.032) (483.871) (0.048) (0.079) (0.071) (14.263) (0.049) (0.049)
1 yea a e 0.123 *** −0.030 0.033 −0.022 377.790 −0.028 0.006 0.028 13.199 −0.040 0.057
(0.031) (0.039) (0.024) (0.020) (364.812) (0.034) (0.055) (0.049) (9.661) (0.030) (0.040)
Female −0.161 *** −0.048 −0.051 * 0.119 *** −488.739 ** 0.127 ** 0.139 ** 0.122 ** −38.079 *** 0.102 ** −0.061 *
(0.047) (0.040) (0.030) (0.027) (235.755) (0.055) (0.055) (0.050) (11.476) (0.044) (0.033)
(Con inues)
20 o 32
|
SANDNER and YÜKSELEN
Dependen a iables
Median daily Sha e o Sha e o Sha e o Log Ho izon al Ve ical Ho izon al o Occupa ion Occupa ion Occupa ion
Log wage
o ull- ime
employees
Pa - ime
employees
High quali ied
employees
Women
in a i m Fi m size Misma ch Misma ch
Ve ical
misma ch Rank
Rank
<
quan ile 10
Rank
>
quan ile 90
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Means o
dependen
4.558 0.203 0.338 0.484 4.829 0.296 0.619 0.721 223.341 0.128 0.114
R- squa ed 0.113 0.039 0.059 0.085 0.035 0.065 0.060 0.044 0.054 0.047 0.053
Indi iduals 312 312 312 312 312 312 312 312 312 312 312
No e: Panel A documen s he gende di e ences in he i s job cha ac e is ics be ween i m and occupa ion change s and s aye s based on he OLS model speci ied in Equa ion (4)
The sample size is 2719, including s aye s (Column 2, Table 4) and i m and occupa ion change s (Column 5, Table 4). Panel B uses i m and occupa ion cha ac e is ics as dependen
a iables, which a e ime- a ian ; ha is, hey may a y be o e and a e he job change. This is simila o he OLS model speci ied in Equa ion (2), howe e , he e he ocus is only
on i m and occupa ion change s, and he ime- a ian dependen a iables a e di e en in each column. In addi ion, we con ol only o g adua ion yea . The sample size is 312
and includes only i m and occupa ion change s (Column 5, Table 4). The es ima ions include a emale dummy, a dummy a iable o 1 yea a e he job change (=0 o he i s job,
=1 o he new job 1 yea a e he i s job) and an in e ac ion o hese dummies. Robus s anda d e o s in pa en heses. ***, **, and * indica e signi icance a he 1%, 5%, and 10%
le els.
TABLE 6 (Con inued)

|
21 o 32
SANDNER and YÜKSELEN
Speci ically, Columns (6) and (7), Panel B o Table 6 show ha emale job change s a e mo e likely han male job
change s o wo k in ho izon ally (by 12.7 pe cen age poin s) and e ically (by 13.9 pe cen age poin s) misma ched
i s jobs a e g adua ion. Howe e , compa ed wi h males, emale job change s educe he equency o e ical
misma ch 1 yea a e he i s job by 13.1 pe cen age poin s.
A e co ec ing he e ical misma ch, one migh expec women o ecei e a highe wage as soon as hey
co ec hei misma ch. Al hough Column (9), Panel B o Table 6 shows ha emale job change s wo k in lowe -
anked occupa ions in hei i s job a e g adua ion, hey do no mo e o (signi ican ly) highe - paid occupa ions
on a e age compa ed o men (Row 1, Column 9, Panel B).
Howe e , Figu e B2 in he Appendix 2 shows ha he occupa ional ank dis ibu ions o male and emale
job change s a e qui e di e en a he i s job, as emales a e less likely o wo k in highe - paid occupa ions and
mo e likely o wo k in lowe - paid occupa ions han males a e. A e he job change, he dis ibu ions o males
and emales con e ge, especially in he lowe ail, as emales p edominan ly mo e om lowe - paid occupa ions
o highe - paid occupa ions. In line wi h he con e gence in he lowe ail, Panel B, Column (10) o Table 6 shows
ha a e a i m and occupa ion change, women educe he p obabili y o being in he bo om decile o anked
occupa ions ela i e o men by 11.5 pe cen age poin s.
O e all, he esul s o Tables 5 and 6 demons a e ha emales a e mo e likely han males o s a in an occu-
pa ion ha is in he bo om ail o he occupa ion ank dis ibu ion and change o highe - paid occupa ions i hey
swi ch bo h occupa ion and i m. Mo eo e , hey s a in occupa ions in which hey a e o e quali ied and co ec
his e ical misma ch by changing bo h occupa ion and i m. As ou da a show ha co ec ing he e ical mis-
ma ch a he i s job explains he dec ease in he gende wage gap, he ques ion a ises as o why women need o
change bo h i ms and occupa ions o co ec he misma ch. We now es common hypo heses in he gende wage
gap li e a u e ha may explain ou indings.
Di e en ypes o disc imina ion:
A i s po en ial explana ion o he decline in e ical misma ch and he gende wage gap wi hin 1 yea a e
he i s job is ha i ms disc imina e agains women a he hi ing s age, when employe s do no ha e su i-
cien in o ma ion abou he p oduc i i y o new hi es (Al onji & Blank, 1999). As a i s ype o disc imina ion,
namely, “sc eening disc imina ion”, Pinks on (2003) documen s ha he p oduc i i y signals ha employe s
ecei e om emales a e noisie han om males. The e o e, p oduc i i y signals a he hi ing s age ha e a
smalle o no e ec on women's wages, while hey ha e a la ge e ec on men's wages. In ou case o uni e si y
g adua es, since he employe is able o obse e he cu icula i ae o he applican s, he inal uni e si y g ades
may comp ise he only signal o he employe . We would expec men wi h highe g ades o no change jobs be-
cause hey al eady ha e a good ma ch in hei i s job. In con as , women wi h highe g ades may expe ience a
misma ch compa ed o hei male coun e pa s a he beginning o hei ca ee s and hus change jobs o co ec
he misma ch. Howe e , ou esul s show ha emale mo e s and s aye s ha e be e g ades on a e age han
hei male coun e pa s do, and bo h emale and male mo e s ha e wo se g ades han s aye s. Ne e heless,
he di e ence in he gende gap be ween mo e s and s aye s is insigni ican , as he in e ac ion e m is insigni -
ican (Column 7, Table 5).
The li e a u e on he gende wage gap sugges s ha emales may ace s a is ical disc imina ion in he labo
ma ke . Acco dingly, employe s may expec lowe p oduc i i y om emales and hi e hem o less sui able jobs.
Consequen ly, condi ional on being hi ed, emales wo k in mo e misma ched jobs and ecei e lowe ini ial wages
a he beginning o hei ca ee s. Howe e , o e ime, as employe s lea n abou he ac ual p oduc i i y o new
hi es (“employe lea ning”), such misma ches could be co ec ed, esul ing in highe wages o women (Al onji &
Blank, 1999; Al onji & Pie e , 2001; Pinks on, 2003). I emales ace s a is ical disc imina ion, we would expec
he gende wage gap o likely na ow no only o hose who change i ms and occupa ions bu also o s aye s.
Since we do no ind a signi ican educ ion in he gende wage gap o s aye s, s a is ical disc imina ion is unlikely
o explain he di e en ial e u ns o changing occupa ions and i ms.
22 o 32
|
SANDNER and YÜKSELEN
Ano he o m o disc imina ion sugges ed by he li e a u e is as e- based disc imina ion, whe e employe s
pay women lowe wages o compensa e o hei (o hei cowo ke s') disu ili y.24 The g ea e misma ch and lowe
ini ial wage o emale mo e s ela i e o male mo e s (Table 6) may indica e some o m o as e- based disc imi-
na ion (Becke , 1971). Howe e , i i ms disc imina e agains women, swi ching o nondisc imina o y i ms should
be su icien o women o imp o e hei wages ela i e o hose o men, while an addi ional change in occupa ion
should no be necessa y. As Table 4 shows, his is no he case, as he gende wage gap does no na ow signi i-
can ly o hose who change only i ms. Fu he mo e, i as e- based disc imina ion explains he gende wage gap,
we should obse e ha women who change i ms will mo e o i ms wi h mo e women, as hese i ms ypically
disc imina e less. Con a y o his hypo hesis, women who change hei i m and occupa ion a e mo e likely o
wo k in i ms wi h a g ea e sha e o women in hei i s job (Panel A o Table 6). Mo eo e , ou es ima ion esul s
show ha women do no swi ch o i ms wi h a g ea e emale sha e han men (Column 4, Panel B o Table 6).
Risk a e sion, con idence and job sea ching ime:
The li e a u e shows ha isk a e sion and (o e )con idence may be an impo an componen o ea ly
ca ee job sea ch, wi h women ypically ha ing highe le els o isk a e sion (Co és e al., 2023; Niede le &
Ves e lund, 2007) and lowe le els o (o e )con idence compa ed o men (Adamecz- Völgyi & Shu e, 2022). Mo e
isk- a e se and less sel - con iden women may ha e lowe ese a ion wages a he beginning o hei ca ee s
(Acemoglu & Shime , 1999; Cox & Oaxaca, 1992; Feinbe g, 1977; Pannenbe g, 2010; Pissa ides, 1974) and hus
accep job o e s ea lie , e en i he job pays less and is no a good ma ch (Co és e al., 2023). Howe e , hese
women may no be sa is ied wi h lowe wages and misma ches and change jobs when hey ind a highe - paid and
be e ma ch job. In his case, we would expec women o spend less ime sea ching o a job a e g adua ion han
men would, and women who ind a job mo e quickly would be mo e likely o change jobs. Column (8) o Table 5
shows ha al hough job change s ind hei i s job sligh ly ea lie han s aye s, he e is no signi ican gende
di e ence in job sea ch du a ion o s aye s and job change s in he ields o economics, business, humani ies and
social sciences. In addi ion, wages do no dec ease signi ican ly wi h he du a ion o he job sea ch.
Job ameni ies and gende no ms:
As an al e na i e explana ion, a g owing body o li e a u e has epo ed ha women p e e nonwage job ameni-
ies such as lexibili y o meaning, ele ance, o esponsibili y o he occupa ion mo e han men, who ha e a g ea e
p e e ence o wages (B enøe & Zöli z, 2020; Flabbi & Mo o, 2012; Goldin, 2014; Goldin & Ka z, 2011) Addi ionally,
women may ollow gende no ms when hey make hei ini ial ca ee decisions. Changing p e e ences o ce ain
job a ibu es o shi ing away om gende no ms may also be a mechanism o job change. On he one hand, e-
males may change o mo e lexible jobs in an icipa ion o ha ing child en in he u u e. Howe e , hese changes may
no lead o highe wage gains. On he o he hand, a he beginning o hei ca ee s, women may p e e lowe - paying
jobs wi h a e ical misma ch o compensa e o some job ameni ies and o espond o ce ain gende no ms; how-
e e , o e ime, hey change hei p e e ences and swi ch o highe - paying and less lexible jobs.
Based on he assump ion ha la ge i ms o e mo e lexible wo k a angemen s (Alb ech e al., 2018), we
es whe he women swi ch o smalle i ms. We do no ind ha emales a e mo e likely han men o so in o
la ge o smalle i ms as a esul o a job change (Column 5, Table 6). Howe e , ou da a do no co e o he
p oxies o job ameni ies, such as he meaning o jobs, schedule adap abili y, o elecommu ing (De Schouwe
& Kes e nich, 2022) o whe he a job ul ills gende no ms. The e o e, we belie e ha changing p e e ences o
job ameni ies o shi s away om gende no ms may s ill be impo an explana ions o why women educe hei
e ical misma ches and inc ease hei wages when hey swi ch i ms and occupa ions.
O e all, his sec ion e eals ha women end o make less op imal decisions han men when hey make choices
ega ding hei i s job a e g adua ion, leading o misma ches. Women a emp o co ec such misma ches
by swi ching i ms and occupa ions. Howe e , since hey canno ully close he ini ial wage gap h ough hese
changes, he esul s sugges ha women should aim o make mo e in o med choices o hei i s job o a oid he
24 Becke (1964) shows in he model ha i ms p ac icing as e- based disc imina ion canno su i e in he compe i i e ma ke in he long un.
|
23 o 32
SANDNER and YÜKSELEN
need o la e co ec ions. Enhanced counseling p og ams could p o ide he in o ma ion necessa y o help women
make be e ini ial job choices.
7 | CONCLUSION
Al hough many s udies ha e in es iga ed he gende wage gap, he exis ence and po en ial explana ions o ea ly
ca ee gende wage di e ences emain unclea . This pape analyses he gende wage gap among g adua es o
a Ge man uni e si y wi h a mas e 's deg ee o equi alen a he beginning o hei ca ee s and o e he i s
yea s a e hei labo ma ke en y. We ely on a unique da ase ha links adminis a i e da a on g adua es
o a Ge man uni e si y wi h employmen egis e s o he Ge man social secu i y sys em. This da ase includes
ex ensi e in o ma ion on s uden s' sociodemog aphic cha ac e is ics, educa ional and labo ma ke ou comes, as
well as he exac iming o g adua ion, labo ma ke en y, and any job changes.
We ind a signi ican gende wage gap among uni e si y g adua es in hei i s job, which pe sis s e en a e
we include an ex ensi e se o con ols. The la ges gende wage gap is obse ed among humani ies and social
sciences g adua es, whe e he sha e o emales is highes and he a e age daily wage is lowes . We ind no signi i-
can gende di e ences in he wages o ma hema ics, na u al sciences, o medical g adua es in hei i s job a e
g adua ion. Mo eo e , in con as o p e ious s udies, we ind an immedia e dec ease in he gende wage gap 1
yea a e labo ma ke en y, which emains ela i ely s able he ea e .
Fu he analysis shows ha he decline in he gende wage gap is concen a ed among indi iduals who change
i ms and occupa ions a e hei i s job wi h a deg ee in economics, business, humani ies, o social sciences. To
explain his dec ease in he gende wage gap, we also show ha emale g adua es a e mo e likely o s a hei
ca ee s in jobs o which hey a e o e quali ied and subsequen ly co ec his skill misma ch, leading o an inc ease
in wages. Co ec ing his misma ch is cos ly o emales, which may be an addi ional explana ion o he wage gap
ha eme ges la e in hei ca ee s.
Uni e si ies ha e an impo an oppo uni y o mi iga e he isk o u u e skill misma ches by implemen ing
counseling in e en ions. These in e en ions can p o ide aluable in o ma ion on e ec i e job sea ch s a egies
ha can o e come gende no ms in ca ee choice, and po en ial wage losses esul ing om skill misma ches, pa -
icula ly o emale s uden s. Ou s udy also highligh s signi ican di e ences in labo ma ke en y and ea ly ca ee
pa hs depending on he chosen ield o s udy. Fo his eason, counseling p og ams ha help s uden s unde s and
hei ca ee p ospec s should be ailo ed speci ically o each ield o s udy. By implemen ing such counseling, uni-
e si ies can p o ide g adua es wi h he insigh hey need o na iga e he dynamic labo ma ke success ully.
ACKNOWLEDGEMENTS
We a e hank ul o aluable commen s om Silke Ange , Alex B yson, Be nd Fi zenbe ge , As id Kunze, Ma kus
Nagle , U a Schönbe g, Nikki Shu e, Da id Wilkinson, and all pa icipan s a he Annual LERN Con e ence
2019, ESPE 2019, AASLE 2019, Wo kshop on Labou Economics 2021, Resea ch Semina a he TH Nü nbe g,
In e na ional Con e ence on The Ge man Labo Ma ke in a Globalized Wo ld (ZEW) 2022. Addi ionally, we a e
g a e ul o Joachim Mölle and Ch is ophe Rus o suppo .
ORCID
Mal e Sandne h ps://o cid.o g/0000-0001-8579-4775
Ipek Yükselen h ps://o cid.o g/0009-0008-8885-740X
REFERENCES
Acemoglu, D. & Shime , R. (1999) E icien unemploymen insu ance. Jou nal o Poli ical Economy, 107(5), 893–928.
Adamecz- Völgyi, A. & Shu e, N. (2022) The gende gap in op jobs – he ole o o e con idence. Labou Economics, 79, 102283.
24 o 32
|
SANDNER and YÜKSELEN
Adda, J., Dus mann, C. & S e ens, K. (2017) The ca ee cos s o child en. Jou nal o Poli ical Economy, 125(2), 293–337.
Alb ech , J., B onson, M.A., Thou sie, P.S. & V oman, S. (2018) The ca ee dynamics o high- skilled women and men:
e idence om Sweden. Eu opean Economic Re iew, 105, 83–102.
Al onji, J.G., A cidiacono, P. & Mau el, A. (2016) The analysis o ield choice in college and g adua e school: de e minan s
and wage e ec s. In: Hanushek, E., Machin, S. & Woessmann, L. (Eds.) Handbook o he economics o educa ion, Vol. 5.
Ams e dam, The Ne he lands: Else ie , pp. 305–396.
Al onji, J.G. & Blank, R.M. (1999) Chap e 48: ace and gende in he labo ma ke . In: Ashen el e , O., Laya d, R. & Ca d,
D. (Eds.) Handbook o labo economics, Vol. 3, Pa C. Ams e dam, The Ne he lands: Else ie , pp. 3143–3259.
Al onji, J.G., Blom, E. & Meghi , C. (2012) He e ogenei y in human capi al in es men s: high school cu iculum, college
majo , and ca ee s. Annual Re iew o Economics, 4(1), 185–223.
Al onji, J.G. & Pie e , C.R. (2001) Employe lea ning and s a is ical disc imina ion. The Qua e ly Jou nal o Economics,
116(1), 313–350.
Ang is , J.D. & Pischke, J.- S. (2009) Mos ly ha mless econome ics: an empi icis 's companion. P ince on, NJ: P ince on
Uni e si y P ess.
An oni, M., Ganze , A., Vom Be ge, P. e al. (2016) Sample o in eg a ed labou ma ke biog aphies (SIAB) 1975- 2014. FDZ-
Da en epo , 04:2016. Ins i u e o Employmen Resea ch (IAB), Nu embe g, Ge many.
Babcock, L. & Lasche e , S. (2009) Women don' ask: nego ia ion and he gende di ide. P ince on, NJ: P ince on Uni e si y
P ess.
Ba el, A.P. & Bo jas, G.J. (1981) Wage g ow h and job u no e : an empi ical analysis. In: Rosen, S. (Ed.) S udies in labo
ma ke s. Camb idge, MA: Na ional Bu eau o Economic Resea ch, Inc, pp. 65–90.
Becke , G. (1971) The economics o disc imina ion. Chicago, IL: Uni e si y o Chicago P ess.
Becke , G.S. (1964) Human capi al: a heo e ical and empi ical analysis, wi h special e e ence o educa ion. New Yo k, NY:
Columbia Uni e si y P ess.
Becke , G.S., Hubba d, W.H. & Mu phy, K.M. (2010) Explaining he wo ldwide boom in highe educa ion o women.
Jou nal o Human Capi al, 4(3), 203–241.
Be and, M. (2011) New pe spec i es on gende . In: Handbook o labo economics, Vol. 4. Ams e dam, The Ne he lands:
Else ie , pp. 1543–1590.
Be and, M., Goldin, C. & Ka z, L.F. (2010) Dynamics o he gende gap o young p o essionals in he inancial and co -
po a e sec o s. Ame ican Economic Jou nal: Applied Economics, 2(3), 228–255.
Black, D.A., Ha iland, A.M., Sande s, S.G. & Taylo , L.J. (2008) Gende wage dispa i ies among he highly educa ed.
Jou nal o Human Resou ces, 43(3), 630–659.
Blau, F.D. & Kahn, L.M. (2017) The gende wage gap: ex en , ends, and explana ions. Jou nal o Economic Li e a u e,
55(3), 789–865.
Bouda ba , B. & Mon ma que e, C. (2009) Choice o ields o s udy o uni e si y Canadian g adua es: he ole o gende
and hei pa en s' educa ion. Educa ion Economics, 17(2), 185–213.
B enøe, A.A. & Zöli z, U. (2020) Exposu e o mo e emale pee s widens he gende gap in STEM pa icipa ion. Jou nal o
Labo Economics, 38(4), 1009–1054.
B uns, B. (2019) Changes in wo kplace he e ogenei y and how hey widen he gende wage gap. Ame ican Economic
Jou nal: Applied Economics, 11(2), 74–113.
Co es, P. & Pan, J. (2020) Child en and he emaining gende gaps in he labo ma ke . Technical epo , Na ional Bu eau o
Economic Resea ch.
Co és, P., Pan, J., Pilossoph, L., Reuben, E. & Za a , B. (2023) Gende di e ences in job sea ch and he ea nings gap:
e idence om he ield and lab. The Qua e ly Jou nal o Economics, 138(4), 2069–2126.
Cox, J.C. & Oaxaca, R.L. (1992) Di ec es s o he ese a ion wage p ope y. The Economic Jou nal, 102(415),
1423–1432.
De Schouwe , T. & Kes e nich, I. (2022) I 's all abou he meaning: wo k meaning, job lexibili y, and he gende wage gap.
Discussion Pape Se ies, DPS 22.04. An we p: Depa men o Economics, KU Leu en.
Del Bono, E. & Vu i, D. (2011) Job mobili y and he gende wage gap in I aly. Labou Economics, 18(1), 130–142.
Deming, D.J. & No ay, K. (2020) Ea nings dynamics, changing job skills, and STEM ca ee s. The Qua e ly Jou nal o
Economics, 135(4), 1965–2005.
Do ne , M., Heining, J., Jacobebbinghaus, P. & Se h, S. (2010) S ichp obe de In eg ie en A bei sma k biog a ien (SIAB)
1975- 2008. FDZ-Me hoden epo , 01:2010. Ins i u e o Employmen Resea ch (IAB), Nu embe g, Ge many.
Dus mann, C., Luds eck, J. & Schönbe g, U. (2009) Re isi ing he Ge man wage s uc u e. The Qua e ly Jou nal o
Economics, 124(2), 843–881.
Fa be , H.S. & Gibbons, R. (1996) Lea ning and wage dynamics. The Qua e ly Jou nal o Economics, 111(4), 1007–1047.
Fede al S a is ical O ice. (2022) Mo he s a 1s bi h and Lände . Wiesbaden, Ge many: Fede al S a is ical O ice.
Feinbe g, R.M. (1977) Risk a e sion, isk, and he du a ion o unemploymen . The Re iew o Economics and S a is ics, 59(3),
264–271.
|
31 o 32
SANDNER and YÜKSELEN
TABLE C5 Desc ip i e s a is ics – s aye s and i m and occupa ion change s.
Dependen a iables
S aye s Fi m and occupa ion change s
Male Female Male– emale Male Female Male– emale
Panel A: Economics, business, humani ies and social sciences
Age a he i s job 27.467 26.674 0.793*** 27.899 26.990 0.908***
Non- Ge man 0.014 0.038 −0.024*** 0.017 0.015 0.003
Du a ion o s udy 5.527 5.463 0.064 5.576 5.716 −0.140
Wo king du ing
s udying
0.659 0.784 −0.124*** 0.676 0.787 −0.110**
App en iceship 0.074 0.063 0.011 0.133 0.096 0.037
O igin in he same
ede al s a e as he
uni e si y
0.843 0.876 −0.033** 0.867 0.875 −0.008
Final Uni. g ade 2.187 1.976 0.210*** 2.223 2.082 0.140**
Du a ion o job
sea ch
3.388 3.518 −0.129 3.110 3.010 0.100
Median daily log
wage o ull- ime
employees in a i m
4.662 4.622 0.041*** 4.543 4.416 0.126***
Sha e o pa - ime
employees in a i m
0.791 0.757 0.035*** 0.809 0.767 0.042*
Sha e o high
quali ied employees
in a i m
0.395 0.379 0.016 0.332 0.302 0.030
Sha e o women in
a i m
0.452 0.524 −0.071*** 0.446 0.561 −0.115***
Log i m size 5.222 5.119 0.102 4.881 4.575 0.306
Ho izon al misma ch
occupa ion
0.148 0.270 −0.122*** 0.267 0.397 −0.130**
Ve ical misma ch 0.469 0.475 −0.005 0.591 0.713 −0.122**
Ho izon al o e ical
misma ch
0.538 0.588 −0.051** 0.688 0.809 −0.121**
Occupa ion ank 236.061 220.823 15.238*** 228.381 190.772 37.609***
Occupa ion ank
<
quan ile 10
0.090 0.118 −0.029** 0.102 0.213 −0.111***
Occupa ion Rank
>
Quan ile 90
0.121 0.085 0.036*** 0.119 0.059 0.060*
Obse a ions 1503 904 176 136
Panel B: Ma hema ics, na u al sciences and medical s udies
Age a he i s job 27.476 26.993 0.482*** 28.120 27.027 1.093***
Non- Ge man 0.014 0.018 −0.004 0.024 0.060 −0.036
Du a ion o s udy 5.692 5.870 −0.178** 5.938 6.008 −0.070
Wo king du ing
s udying
0.660 0.661 −0.002 0.747 0.720 0.027
(Con inues)

32 o 32
|
SANDNER and YÜKSELEN
Dependen a iables
S aye s Fi m and occupa ion change s
Male Female Male– emale Male Female Male– emale
App en iceship 0.043 0.041 0.001 0.133 0.120 0.013
O igin in he same
ede al s a e as he
uni e si y
0.914 0.855 0.059*** 0.880 0.820 0.060
Final Uni. g ade 1.858 2.007 −0.149*** 1.835 1.845 −0.010
Du a ion o job
sea ch
3.330 3.671 −0.341** 2.748 3.462 −0.714
Median daily log
wage o ull- ime
employees in a i m
4.576 4.499 0.077*** 4.306 4.236 0.070
Sha e o pa - ime
employees in a i m
0.689 0.609 0.080*** 0.677 0.625 0.053
Sha e o high
quali ied employees
in a i m
0.317 0.242 0.075*** 0.219 0.198 0.021
Sha e o women in
a i m
0.552 0.714 −0.162*** 0.522 0.755 −0.233***
Log i m size 5.594 5.690 −0.095 3.934 4.046 −0.111
Ho izon al misma ch
occupa ion
0.196 0.084 0.112*** 0.425 0.260 0.165*
Ve ical misma ch 0.121 0.100 0.021 0.402 0.460 −0.058
Ho izon al o e ical
misma ch
0.263 0.139 0.124*** 0.575 0.500 0.075
Occupa ion ank 285.685 290.981 −5.296 227.080 199.800 27.280
Occupa ion ank
<
quan ile 10
0.116 0.098 0.018 0.172 0.180 −0.008
Occupa ion ank
>
quan ile 90
0.372 0.584 −0.212*** 0.172 0.160 0.012
Obse a ions 1148 570 87 50
No e: This able shows summa y s a is ics o g adua es' pe sonal, p e- g adua ion, pos - g adua ion and i s job
cha ac e is ics o s aye s and o i m and occupa ion change s. The sample consis s o g adua es wi h a mas e 's deg ee
o equi alen who wo k in a ull- ime job as hei i s job a e g adua ion and who ha e a wage spell 1 yea a e hei
i s job. ***, **, and * deno e signi icance a he 1%, 5%, and 10% le els.
TABLE C5 (Con inued)