A onina, Ma iya; Zaha ie a, Anna
Wo king Pape
How did you ind you job? E ec s o he job sea ch
channels on labou ma ke ou comes in Ge many
Wo king Pape s in Economics and Managemen , No. 1-2025
P o ided in Coope a ion wi h:
Facul y o Business Adminis a ion and Economics, Biele eld Uni e si y
Sugges ed Ci a ion: A onina, Ma iya; Zaha ie a, Anna (2025) : How did you ind you job? E ec s o
he job sea ch channels on labou ma ke ou comes in Ge many, Wo king Pape s in Economics and
Managemen , No. 1-2025, Biele eld Uni e si y, Facul y o Business Adminis a ion and Economics,
Biele eld,
h ps://doi.o g/10.4119/unibi/3001180
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/315015
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 ps://c ea i ecommons.o g/licenses/by/4.0/
Facul y o Business Adminis a ion and Economics
www.wiwi.uni−biele eld.de
33501 Biele eld − Ge many
P.O. Box 10 01 31
Biele eld Uni e si y
ISSN 2196−2723
Wo king Pape s in Economics and Managemen
➔
No. 1-2025
Ma ch 2025
How did you ind you job? E ec s o he job sea ch
channels on labou ma ke ou comes in Ge many
Ma iya A onina, Anna Zaha ie a
How did you find you job? Effec s o he job sea ch
channels on labou ma ke ou comes in Ge many
∗
Ma iya A onina
†
, Anna Zaha ie a
‡
Feb ua y 26, 2025
Abs ac
We s udy he effec o finding a job h ough one’s social con ac on s a ing
wages. Using combined SOEP-INKAR da a o Ge many and p opensi y sco e
analysis - bo h ma ching and weigh ing - we documen ha e e al hi ing is asso-
cia ed wi h a wage penal y o 10%. This penal y is s able o e ime. Sepa a ing
by he ype o he social con ac , we find ha e e als om o me colleagues a e
associa ed wi h a 9% wage p emium compa ed o a di ec o mal applica ion. In
con as , e e als om iends a e associa ed wi h a 7% wage penal y. Ou esul s
highligh pe sis en sel -selec ion o wo ke s on obse able and unobse able cha -
ac e is ics. Using in o ma ion om a sho es o cogni i e abili ies (symbol digi
es ) we documen ha wo ke s ecommended by o me colleagues pe o m bes in
he abili y es , consis en wi h he p edic ions om a so ing model. The lowes
pe o mance is eco ded o hose elying on he help o hei iends. The effec s
a e p ima ily d i en by he sub-sample o women. No significan diffe ences ac oss
sea ch channels a e ound o pe sonali y ai s.
JEL-code: C21, J31, J62, J64
Key wo ds: job sea ch, social ne wo ks, e e als, cogni i e abili ies
∗
We hank He be Dawid, Jan Goebel, Zainab I ikha , Mo i z Kuhn, Simon K¨uhne, Ma cus Pan-
nenbe g and Ka in Rickmeie o hei aluable commen s. The au ho s also would like o hank he
Leibniz Associa ion o p o iding financial suppo o he Leibniz Science Campus “SOEPRegioHub” a
Biele eld Uni e si y.
†
Biele eld G adua e School o Economics and Managemen , Biele eld Uni e si y, 33501 Biele eld,
Ge many. Email: ma iya.a onina@uni-biele eld.de
‡
Chai o Labou Economics, Biele eld Uni e si y, 33501 Biele eld, Ge many. Email:
anna.zaha ie [email p o ec ed]
1 In oduc ion
The e a e mul iple ways o find a job a ying om a o mal applica ion in In e ne ,
in e media ion by he employmen agency o elying on he help om one’s social ne wo k.
The li e a u e sugges s ha finding a job h ough a social con ac is he mos equen
channel o en e ing a job making up 30 −50% o all new ma ches1. Recognizing he
impo ance o his sea ch channel o he labou ma ke , mul iple s udies ied o e alua e
he ela ionship be ween he job sea ch channel, which gene a ed a job, and he s a ing
wage. The empi ical e idence is mixed wi h equally la ge g oups o s udies suppo ing
he idea o wage p emia and wage penal ies om e e al hi ing compa ed o di ec o mal
applica ions2. This aises wo ques ions, fi s , whe he he selec ion o wo ke s o specific
sea ch channels could explain di e se findings in he li e a u e and, second, whe he he
ype o he social con ac ma e s o he s a ing wage.
We answe hese ques ions by using a unique combina ion o a iables con ained in he
Ge man Socio-Economic Panel (SOEP). SOEP is a la ge-scale household su ey which
is ep esen a i e on he na ional le el. I combines in o ma ion abou he ype o he
social con ac p o iding a e e al, dis inguishing be ween amily ies, iends and o me
colleagues, wi h ich in o ma ion on he socio-economic and demog aphic backg ound o
he esponden as well as fi m and occupa ion-specific cha ac e is ics. Mos impo an ,
i includes in o ma ion abou he esul s o a sho es o cogni i e abili ies, called he
symbol-digi es (SDT), as well as pe sonali y ai s3, which allows us o shed ligh on
he ole o obse ed and unobse ed wo ke he e ogenei y in selec ing in o specific sea ch
channels. In addi ion, we exploi a ecen linkage o SOEP wi h he INKAR da abase4
gi ing us a possibili y o accoun o coun y-le el egional economic indica o s.
Ou empi ical analysis is guided by he heo e ical amewo k de eloped in S upny ska
and Zaha ie a (2015). This s udy conside s a con inuum o wo ke s he e ogenous in
abili y/p oduc i i y and sea ching o jobs when unemployed. Mo eo e , he e a e h ee
sea ch channels: o mal applica ions and wo in o mal channels – h ough amily and
p o essional ne wo ks. The model illus a es a s ong selec ion o wo ke s ac oss he
h ee sea ch channels wi h high abili y wo ke s mos ly en e ing jobs ia e e als om
p o essional con ac s, low abili y wo ke s elying on hei amilies and wo ke s in he
middle o he abili y dis ibu ion en e ing jobs in a o mal way. This selec ion pa e n
explains wage p emia associa ed wi h p o essional ecommenda ions bo owing om he
seminal app oach by Mon gome y (1991) and G ano e e (1995). A he same ime, i
1Addison and Po ugal (2002), Kugle (2003), Ma golis and Simonne (2003), Dela e and Saba ie
(2007), Goos and Salomons (2007), Ponzo and Scoppa (2010), G¨u zgen and Pohlan (2024)
2We e iew and discuss bo h g oups o s udies in he nex sec ion.
3SOEP includes in o ma ion abou 10 pe sonali y ai s which can be agg ega ed o he “Big Fi e”
scale: conscien iousness, ex a e sion, ag eeableness, openness, neu o icism. See McC ae and Cos a
(1999), Uysal and Pohlmeie (2011) and Caliendo e al. (2016) o u he de ails.
4Indica o s and Maps o Spa ial and U ban De elopmen
1
p edic s wage penal ies associa ed wi h e e als om amily o close iends i wo ke
abili ies a e no obse able.
We es hese p edic ions by using he app oaches based on in e se-p obabili y weigh -
ing (IPW). These a e cha ac e ized by ha ing wo s ages. In he fi s s age we model
he sel -selec ion o wo ke s in o sea ch channels wi h a (mul inomial) logis ic eg es-
sion. The mul inomial specifica ion is used o accoun o he ou sea ch channels
( amily/ iends/colleagues/ o mal) and gene a es p opensi y sco es which a e used o
cons uc he weigh s in he es ima ion o he ou come a iable. In he second s age we
apply h ee weigh ing me hods: pu e weigh ed mean compa ison (IPW), he augmen ed
in e se p obabili y weigh ing (AIPW) and he IPW - eg ession adjus men (IPWRA).
All he app oaches make he g oups o wo ke s, finding jobs hough diffe en sea ch chan-
nels, compa able o each o he in e ms o obse able cha ac e is ics. Howe e , he la e
wo also model he ou come, which gi es ise o he “doubly- obus ” es ima o s (Imbens
and Woold idge 2009; Glynn and Quinn 2010;Ku z2022)5.
Fi s , ou desc ip i e esul s e eal ha he e is a s ong sel -selec ion o wo ke s
based on o mal qualifica ion and wo king ime, con ibu ing oge he abou 1/3 o he
ow wage penal y associa ed wi h e e al hi ing. Ano he 1/5 o he penal y s ems om
diffe ences in he fi m size and he likelihood o occupa ional misma ch. These findings
suppo he ideas in Galenianos (2013) and Rebien e al. (2020) ha e e als o en lead
o jobs in small fi ms. The obse a ion ha e e ed candida es a e mo e likely o epo
occupa ional misma ch is in line wi h he e idence in Ben olila e al. (2010) pu ing
o wa d he idea ha e e al hi ing gene a es occupa ional misma ch be ween he ini ial
qualifica ion o he wo ke and he job equi emen s. O e all, we conclude ha diffe ences
in obse able cha ac e is ics con ibu e abou 2/3 o he e e al wage penal y. The IPW
me hods e eal a emaining wage penal y o −10%, which is pe sis en o e ime.
Second, sepa a ing by he ype o he social con ac , we find ha e e als om iends
a e associa ed wi h a significan wage penal y equal o −7%. A he same ime, e e als
om o me colleagues show a obus wage p emium be ween 9% and 10%. In o de
o s udy he ole o unobse ed he e ogenei y behind hese findings we compa ed he
pe o mance o esponden s in he symbol digi ex (SDT)6. This es was al eady used
in o he s udies as a p oxy o cogni i e abili ies (e.g. Heineck and Ange (2010), Rich e
e al. (2017)), mo eo e , Lang e al. (2007) show ha he SDT ou come is sufficien ly
5Ano he ad an age o he IPW app oach, in compa ison o he s anda d OLS, is ha i does no
equi e he linea i y assump ion and allows us o pe o m he balancing checks o co a ia es be o e
and a e p opensi y sco e ma ching. Mo eo e , Imbens and Woold idge (2009) highligh ha in case
he no malized diffe ences be ween obse able cha ac e is ics a e conside able, he uncon oundedness
assump ion o he s anda d OLS migh no be ulfilled p ope ly, wi h esul s being ex emely sensi i e
o he specifica ion. Iden ifica ion elying on he es ima ed p opensi y sco es allows o a oid he issue.
6E e y pa icipan was p o ided wi h a mapping be ween pai s o symbols and digi s on he sc een.
The pa icipan s had o ma ch as many symbols o he co esponding numbe s as hey could. The
numbe o co ec , o al and w ong answe s was eco ded a e 30, 60 and 90 seconds.
2
co ela ed wi h es sco es om mo e comp ehensi e in elligence es s. We find ha
wo ke s e e ed by hei colleagues pe o m be e in he es han any o he g oup. The
second bes g oup in e ms o cogni i e sco es a e hose who ound a job in a o mal
way, ollowed by hose elying on he help o hei amilies and iends. These esul s
indica e subs an ial diffe ences in cogni i e abili ies be ween he g oups and suppo he
so ing model in S upny ska and Zaha ie a (2015). In addi ion, we compa ed he non-
cogni i e abili ies based on he “Big Fi e” pe sonali y ai s. We do no find sys ema ic
diffe ences in non-cogni i e abili ies o wo ke s using diffe en sea ch channels. One
mino excep ion is he abili y o o gi e, which is highes among indi iduals e e ed by
hei amily membe s.
Thi d, we pe o m he he e ogenei y analysis and sepa a e he sample by gende and
egion. We find ha he o e all wage penal y om e e als is pe sis en o men and
women and has a simila size. Howe e , he finding ha wage penal ies/p emia a e asso-
cia ed wi h e e als om iends/colleagues is la gely d i en by he subsample o women.
This is inline wi h he obse a ion ha he sp ead in cogni i e sco es be ween hose find-
ing he job wi h a help o iends/colleagues is la ge o women in compa ison o men.
This p o ides ano he indi ec suppo o he impo an ole o cogni i e abili ies in
explaining e e al wage gaps. Wi h espec o egional he e ogenei y ou esul s demon-
s a e ha highe unemploymen is associa ed wi h a s onge use o social ne wo ks.
On he con a y, he e is less eliance on ne wo ks in mo e densely popula ed egions.
A he same ime, wage penal ies om e e als by iends and p emia om e e als by
colleagues a e mo e p onounced in Eas e n Ge many.
The es o he pape is s uc u ed as ollows. Sec ion 2cha ac e izes he cu en s a e
o esea ch and b iefly summa izes he unde lying heo e ical model. Sec ion 3desc ibes
he da a and p esen s desc ip i e s a is ics. I is ollowed by sec ion 4, which ou lines he
empi ical me hods. Nex , sec ion 5p esen s he es ima ion esul s o he o e all sample,
ollowed by he he e ogenei y analysis by gende and Wes /Eas Ge many in sec ion 6.
The obus ness o he esul s is es ed in sec ion 7, while sec ion 8concludes.
2 Li e a u e Re iew and Theo y
2.1 Rela ed li e a u e
The li e a u e on he effec o e e als on wages is ich bu inconclusi e. Whe eas some
s udies find a significan posi i e link, o he s epo wage penal ies associa ed wi h en e -
ing he job ia a e e al (Topa 2011). Pellizza i (2010) highligh s he puzzle by showing
empi ically ha in he Eu opean Union “... p emia and penal ies o finding jobs h ough
pe sonal con ac s a e equally equen and a e o abou he same size” (p. 494). Thus, in
he ollowing, we e iew sequen ially he li e a u e on wage p emiums and wage penal ies
3
associa ed wi h e e als paying pa icula a en ion o he unde lying mechanisms and
o mula ing se e al es able hypo heses o guide ou empi ical analysis.
Wage p emia: The seminal heo e ical con ibu ions da e back o he wo k by
Mon gome y (1991) and G ano e e (1995). Mon gome y (1991) shows ha employee
e e als help fi ms in sc eening he unobse ed abili ies o hei applican s. In pa icula ,
Mon gome y (1991) de elops he idea o homophily by abili y (o he inb eeding bias),
meaning ha high-abili y wo ke s a e likely o ha e high-abili y iends. The e o e, fi ms
expec his ela ionship and offe highe wages o e e ed applican s. Hens ik and Skans
(2016) confi m his esul empi ically and show ha in Sweden en an s a e mo e likely
o be connec ed o high-abili y employees han o low-abili y ones (defined om he es
sco es o wages). Mo eo e , in hei s udy, en e ing wo ke s ecei e highe ini ial wages
i hey ha e a link o he incumben employee.
Ano he explana ion o wage p emia associa ed wi h e e als is p esen ed by Simon
and Wa ne (1992) and u he ex ended by Galenianos (2014) and Dus mann e al.
(2016). Simon and Wa ne (1992) show ha e e als om employees educe unce ain y
abou applican s’ p oduc i i y, which leads o highe ese a ion wages, highe s a ing
wages and lowe wage g ow h a e he fi s pe iod. All h ee s udies p o ide empi ical
suppo o his mechanism based on he US, B i ish and Ge man da a espec i ely,
hough he e idence in Dus mann e al. (2016) is only indi ec (based on e hnic ne wo ks)
and does no include de ailed in o ma ion abou he ype o social con ac who p o ided
a e e al. A ecen s udy by G¨u zgen and Pohlan (2024) ex ends he model o employe -
lea ning by adding o mal sc eening ac i i ies. The au ho s use da a om an employe
su ey o Ge many o es hei model and find a 1% wage p emium. The esul , howe e ,
is d i en solely by he male subsample.
One u he s udy suppo ing he idea o wage p emia is Kugle (2003). In his model,
e e als educe he cos o moni o ing o fi ms, because wo ke s exe pee p essu e on
co-wo ke s, and make i cheape o pay efficiency wages. Also Ioannides and Soe e en
(2006) and Fon aine (2008) show heo e ically ha wo ke s wi h a la ge social ne wo k
ecei e highe wages, whe eas empi ically wage p emia a e suppo ed by Ma golis and
Simonne (2003) o F ance and Goos and Salomons (2007) o heUK.
Wage penal ies: Nex , we conside he li e a u e epo ing wage penal ies associa ed
wi h e e al hi ing. Empi ically wage penal ies a e documen ed by Pis a e i (1999)
o I aly, Addison and Po ugal (2002) o Po ugal and Dela e and Saba ie (2007)
o F ance. Ponzo and Scoppa (2010) a gue ha fi ms may offe low wages and hi e
low-abili y amily ies in he absence o mo e alen ed applican s. This is he idea o
a ou i ism leading o wage penal ies associa ed wi h e e al hi ing. Ben olila e al.
(2010) de elop a model showing ha social con ac s may gene a e a misma ch be ween
he qualifica ion o he wo ke and his wo ke ’s p oduc i e ad an age. Ben olila e al.
(2010) p esen empi ical e idence suppo ing his iew o he US and Eu opean Union,
4
hough he sample is limi ed o young wo ke s (below 35 yea s o age). Ho ´a h (2014)and
Zaha ie a (2018) ex end his se up by in oducing a ne wo k homophily pa ame e , i.e.,
he deg ee o which people o m social connec ions wi h o he s om he same occupa ion.
Galenianos (2013) and Rebien e al. (2020) pu o wa d he idea ha la ge fi ms
ha e mo e financial esou ces o ad e ise hei acancies compa ed o small fi ms, he e-
o e, hey ecei e many applica ions ia he o mal sea ch channel making e e al hi ing
edundan . Bo h s udies p o ide empi ical suppo o his idea.
P emia/penal ies depending on he con ac ype: Labini (2005), S upny ska
and Zaha ie a (2015) and mo e ecen ly Les e e al. (2021) emphasize he poin ha
he ype o social con ac may play a c ucial ole o he sign o he e e al wage gap.
In pa icula , hey show ha e e als om p o essional con ac s and o me colleagues
(weak ies) lead o a wage p emium, whe eas e e als om ela i es and close iends
(s ong ies) gene a e a wage penal y. The unde lying mechanism is based on he so ing
o wo ke s wi h diffe en (unobse ed) abili ies ac oss sea ch channels. Acco ding o
Les e e al. (2021) he so ing amewo k is suppo ed in he US da a, howe e , hei
da ase does no include a di ec measu e o wo ke cogni i e abili ies.
Only a ew o he s udies add essed his issue empi ically, including An oninis (2006)
o Egyp , Meliciani and Radicchia (2011) o I aly and Cappella i and Ta si amos (2015)
o he UK. E en hough hei esul s a e suppo i e o he desc ibed mechanism, he
e idence in he o me s udies is desc ip i e, whe eas he da a in Cappella i and Ta si -
amos (2015) does no include in o ma ion abou he ype o social con ac who p o ided
a e e al. Hence, in his pape we es he heo e ical p edic ions om he so ing model
by using Ge man da a (SOEP), which includes in o ma ion abou wo ke ’s cogni i e abil-
i ies and pe sonali y ai s. We gi e a sho o e iew o he so ing mechanism de eloped
in S upny ska and Zaha ie a (2015) and summa ize ou hypo heses in he nex sec ion.
He e ogenei y by gende and egion: Fu he , we acknowledge ha he e e al
wage gap could diffe be ween he wo gende g oups and ac oss loca ions. Fo example,
Huffman and To es (2002) and W zus e al. (2013) show empi ically o he USA ha
he quali y o e e ed jobs p o ided o a con ac diffe s o men and women. Ma ma os
and Sace do e (2002) s udy he effec s o he pee ne wo ks and also epo diffe ences
by gende and ace. Zhou (2019) emphasizes ha no only he size o he ne wo k is i al
o a success ul job sea ch bu also he willingness o he con ac s o p o ide a e e al.
Ge many p esen s ano he social con ex as a esul o p e ious his o ical e en s. Despi e
he obse able con e gence in many ac o s, Wes Ge many and Eas Ge many a e s ill
diffe en on such indica o s as income le els, he unemploymen a e as well as wo k-
ela ed a i udes and gende oles (Schnabel 2016; Di ksmeie 2015). The la e can
be highly-influen ial wi h ega d o he quali y o he job ansmi ed ia e e als o
men and women. The e o e, in he ollowing analysis we go beyond he o e all effec o
e e als on wages and s udy sepa a e effec s by gende and egion.
5
2.2 Theo e ical conside a ions
In his sec ion we b iefly desc ibe he economic mechanism leading o wage p emia o wage
penal ies om e e als in he model by S upny ska and Zaha ie a (2015). We es i la e
in he main body o he pape . Conside wo ke s wi h he e ogeneous p oduc i i ies yi,
i=1..p,whe epis he numbe o dis inc wo ke g oups. Diffe ences in he p oduc i i y
eflec diffe ences in wo ke ’s abili ies, so ha highe yiis associa ed wi h a highe wage.
Wo ke s’ p oduc i i y/abili y is obse ed by he employe upon he ma ch (in he cou se
o he job in e iew), bu i is no obse able o he econome ician in he da a.
The e a e h ee sea ch channels in he labou ma ke . Fi s , unemployed wo ke s can
find a job by sending applica ions o open acancies, his is he o mal channel o job
sea ch wi h a job-finding a e si. This a e is endogenous and depends on he sea ch
effo o he wo ke siand he s ock o acancies . The o mal channel is cos ly o
wo ke s in e ms o effo wi h a cos unc ion C(s)=s2/c. In addi ion, wo ke s can find
a job ia hei p o essional con ac s a a e λi(weak ies) o amily membe s λ0(s ong
ies). This se up gi es ise o he ollowing Bellman equa ion o unemployed wo ke s:
Ui=b+(λi+λ0)Ri+ max
ssRi−s2/c
whe e bdeno es he unemploymen benefi and Riis a wo ke job en inc easing in he
wo ke ype yidue o a highe wage. Maximizing he p esen alue o unemploymen Ui
wi h espec o sleads o he esul ha he o mal sea ch effo o wo ke s is inc easing
in he wo ke ype yi( ia Ri) bu dec easing in he job-finding a e ia p o essional
con ac s λi. In ui i ely, his shows a disincen i e effec : wo ke s wi h a highe p obabili y
o ge ing a p o essional e e al (λi) educe hei o mal sea ch effo si.
Nex , he au ho s de i e λiby ollowing he idea in Mon gome y (1991) and assum-
ing ha wo ke s o m p o essional connec ions wi hin hei abili y g oup. The idea o
homophily by abili y is also suppo ed by Ioannides and Lou y (2004). In gene al, ho-
mophily e e s o he ac ha people a e mo e p one o main ain ela ionships wi h
o he s who a e simila o hem. The e can be homophily by ace, gende , eligion, skill
o abili y and i is gene ally a obus obse a ion in social ne wo ks (see Jackson (2008)
o an o e iew o esea ch on ne wo k homophily). I he ne wo k size o e e y wo ke
is fixed o n, he p obabili y o finding a job ia a p o essional e e al is gi en by:
λi= ai
1−μi
μi
[1 −(1 −μi)n]
whe e aiis he ad e ising effo o fi ms and μiis he unemploymen a e o wo ke s in
he p oduc i i y g oup i. In he abo e equa ion he e m [1 −(1 −μi)n] s ands o he
p obabili y ha an incumben ype iemployee add essed by he fi m has a leas one
unemployed iend who will ake he job. In addi ion, his equa ion shows ha fi ms can
6
e e ed o jobs diffe s significan ly om he compa ison g oup along mul iple obse able
cha ac e is ics poin ing ou o he sel -selec ion bias in he sample. Hence, u he we
discuss ou es ima ion s a egy building on he in e se p obabili y weigh ing app oach.
We ely on he ollowing amily o me hods, in con as o he usual OLS es ima o , as i
allows no only o pe o m he balancing checks o selec ion-d i ing co a ia es and assu e
a common suppo , bu i also does no equi e a linea i y assump ion in he weigh ing
p ocedu e (Glynn and Quinn 2010;Ku z2022).
4 Me hods
The e a e se e al challenges o pe o ming he es ima ion. Fi s , he sel -selec ion bias
is p esen bo h in he se up wi h agg ega ed and disagg ega ed ne wo ks. Second, he
size o he ea ed and compa ison g oups is conside ably diffe en . As we can see om
Table 3 he e a e almos wo imes mo e esponden s, who ound a job wi h he help
o ne wo ks, compa ed o hose who used a o mal indi idual applica ion. Las bu no
leas , is he disagg ega ed ne wo k se up, which calls o a simul aneous compa ison o
mo e han wo sea ch channels.
To ackle hese conce ns we apply he in e se p obabili y weigh ing echniques. These
me hods include he in e se p obabili y weigh ing (IPW), he augmen ed in e se p ob-
abili y weigh ing (AIPW) and he in e se-p obabili y weigh ing - eg ession adjus men
(IPWRA). All o hese me hods ha e wo s ages. In he fi s s age we model he sel -
selec ion in o sea ch channels wi h a logis ic eg ession. I is a bina y logi in he ag-
g ega ed se up wi h wo sea ch channels ( o mal/ne wo ks) and a mul inomial logi in
he disagg ega ed se up ( o mal/ amily/ iends/colleagues). The co esponding p open-
si y sco es p(x)ia e used as weigh s in he es ima ion o he ou come a iable (s a ing
wages) in he second s age. When es ima ing he a e age ea men effec unde he IPW
p ocedu e, he weigh ed ou comes o bo h g oups a e calcula ed in he ollowing way:
ln(Yw i)=ln(Y i)
p(x)i
ln(Ywci)= ln(Yci)
1−p(x)i
,(1)
whe e p(x)iis a p opensi y sco e es ima ed o indi idual iusing a (mul inomial) logi
app oach. Yw i is a weigh ed ou come o obse a ion iassigned o he ea men g oup
, while Yc i is a weigh ed ou come o indi idual iin he compa ison g oup c.Bo h he
ea men and he compa ison g oup weigh s a e no malized wi hin he espec i e g oup.
The ea men effec is calcula ed as a simple diffe ence o weigh ed means:
ATEIPW =N
i=1 ln(Yw i)
N +Nc
−Nc
i=1 ln(Ywci)
N +Nc
,(2)
whe e N and Ncco espond o he numbe o obse a ions in he ea men and he
13
compa ison g oups espec i ely.
AIPW and IPWRA a e ex ensions o he IPW and belong o he class o he “dou-
bly obus ” es ima o s: in bo h cases, sel -selec ion and ou come a e modeled, bu o
he consis ency o he es ima ed coefficien s, i is enough o speci y a leas one o he
models co ec ly (Glynn and Quinn 2010;Ku z2022). The main diffe ence be ween he
wo me hods lies in he second s age when he in e se p obabili y weigh s a e applied.
The AIPW p ocedu e es ima es sepa a e eg ession models o he ou come o each ea -
men g oup and ob ains he ea men -specific p edic ed ou comes. I hen compu es he
weigh ed means o he ob ained alues. In con as , he IPWRA p ocedu e fi s weigh ed
eg ession models o he ou come o each ea men g oup and es ima es he ea men -
specific p edic ed ou comes. I hen compu es he diffe ence. The baseline model o
AIPW/IPWRA in he agg ega ed se up can be w i en in he ollowing way:
ln(Yi)=β0+β1D+X
iβ2+Z
β3+τ +i,(3)
whe e Yiis a dependen a iable, e.g., s a ing wages in he new job; Dis he ea men
indica o , which equals o 1 i a job was ound using ne wo ks; Xiis a ec o o con ol
a iables on indi idual le el; Z is a se o egional indica o s, τ is a ime-fixed effec
and iis an e o e m.
In he disagg ega ed se up wi h mul iple e e al ypes ( amily, iends and colleagues)
we adjus he eg ession equa ion o accoun o he h ee diffe en ea men a ms. The
equa ion becomes:
ln(Yi)=β0+β1D1+β2D2+β3D3+X
iβ4+Z
β5+τ +i,(4)
whe e Yiis again he dependen a iable, while D1 o D3a e dummy a iables co e-
sponding o he h ee e e al channels: iends, amily o colleagues. The in e p e a ion
o o he a iables emains he same.
The se o con ol a iables Xiin he AIPW and he IPWRA ou come models includes
sex, age, age squa ed, yea s o educa ion, occupa ional misma ch/being in s udies/no
aining, a dummy a iable o mig a ion backg ound; a dummy o a long dis ance mo e
as well as mo ing o ano he NUTS3 egion in he p e ious yea ; change in a labou
ma ke s a us om non-pa icipa ion o employmen o job- o-job ansi ion; being in a
ull- ime employmen (compa ed o a pa - ime a angemen ), loga i hm o o al hou s
wo ked du ing he yea ; a dummy a iable o being employed in cons uc ion and i he
cu en ly employing fi m is o a middle size. The se o egional cha ac e is ics Z includes
he sha e o mig an s, he unemploymen a e and he loga i hm o he popula ion densi y.
In o de o co ec o he sample selec ion in he fi s s age o he es ima ion, we
include he ollowing p edic o s: age, sex, yea s o educa ion, ma i al s a us, occupa-
ional misma ch, eloca ion o ano he coun y, li ing in a u al o u ban egion, being
14
employed ull- ime, yea ly wo king hou s a e g ea e han he a e age among he wo k-
ing indi iduals in a gi en yea , changes in he labou ma ke s a us as desc ibed in he
ou come model, wo king in a small, middle o la ge fi m, mig a ion backg ound, numbe
o child en in he household. The se o egional cha ac e is ics Z is he same in bo h
model s ages. Mo eo e , we also include a dummy o a esponden li ing in he Wes o
Eas Ge many a he ime o he su ey.
An impo an condi ion o he alidi y o he IPW echniques is a balance o co-
a ia es in he adjus ed sample. We es i he co a ia es a e balanced by using he χ2
es s a is ic p oposed by Imai and Ra ko ic (2014). This is an adap a ion o GMM-
o e iden ifica ion es o he co a ia es balance. Since all he es ima ion me hods ha e
he same fi s s age, i is enough o calcula e he p- alues o he χ2- es o he IPW
alone. We find ha he null-hypo hesis o co a ia es being balanced canno be ejec ed
(p- alue=0.46)9.
In addi ion o he IPW app oach, we use p opensi y sco e ma ching (PSM) as a
obus ness check in he agg ega ed analysis ( o mal/ne wo ks)10. Mo eo e , we epo
he PSM esul s in he second s age based on a “pu e” weigh ed-mean compa ison as
well as eg ession-based. No e ha PSM can no be applied in he disagg ega ed analysis
wi h mul iple ea men g oups. The ma ching p ocedu e was s a ified be ween he Eas
and he Wes and we used a 1:1 callipe ma ching algo i hm wi h eplacemen 11.The
selec ed callipe equals 0.15 o he SD o he p opensi y sco e. The p opensi y sco e
dis ibu ion is epo ed in Figu e 8in Appendix 9.3. The same ma ching ec o is used
as in he ea men assignmen s models in he PS weighing app oach. A e he PSM,
he median bias is educed om 10.4% o 0.9%, he Rubin’s B pa ame e equals o 7.9,
and he a iance a io R is 1.012.
We used he same se o con ol a iables desc ibed abo e in he fi s s age o he
PSM. Howe e , since he fi s -s age model does no ha e o be sa u a ed, he con ols in
he ou come s age may be ex ended wi h addi ional eg esso s (Guo and F ase 2014).
9The es s we e also pe o med o he subg oups, and he esul s a e epo ed in Sec ion 6.Wi h he
same balancing ec o desc ibed abo e, o all o he subg oups he null-hypo hesis canno be ejec ed.
Thus, he co a ia es a e balanced.
10I is impo an o acknowledge ha in he ecen li e a u e he e a e deba es ega ding he expec ed
bias which migh be in oduced wi h he o e co ec ion. The pape by King and Nielsen (2019) c i icized
he applica ion o PSM due o he possibili y o in oducing a highe bias compa ed o he unma ched
case. The au ho s p oposing eso ing o he Mahalanobis dis ance o en opy ma ching ins ead. How-
e e , his pape spa kled deba e caused by he na u e o he simula ed da a used. The simula ion s udy
by Vable e al. (2019) co e ed diffe en ypes o da a and highligh ed ha a p ope common suppo was
essen ial o he ma ching me hods o be unbiased and supe io o OLS in case o dissimila es ima es.
Mo eo e , unlike OLS, PSM allows o ackle he sel -selec ion bias (Ti us 2007).
11Aus in (2014) compa ed 12 diffe en ma ching app oaches and ound ha callipe ma ching has
pe o mance o a leas as good as o he algo i hms and supe io in e ms o bias educ ion.
12Rubin’s B is an absolu e s anda dized diffe ence in means o he p opensi y sco e be ween ea ed
and compa ison g oups. Rubin’s R is he a io o a iances in he p opensi y sco es o he ea ed and
compa ison g oups. In o de o conside sample being balanced, he fi s pa ame e should be below
25%, while he second one is [0.5,2] (Rubin 2001).
15
Consequen ly, he se o addi ional eg esso s includes such a iables as age squa ed,
enu e, o al yea s spen in unemploymen and household size. In o de o accoun
o occupa ional diffe ences we add a se o dummy a iables o being in a pa icula
occupa ion ( ollowing he NACE e . 1.1 classifica ion). Mo eo e , we ex end he se
o egional indica o s wi h he ede al s a e fixed effec s and a bina y indica o o he
egion being u al o u ban. The exac cons uc ion o he main a iables is discussed in
de ail in Appendix 9.2.
5Resul s
In his sec ion we apply he me hodology discussed abo e. We s a by conside ing he
agg ega ed se up wi h a bina y sea ch channel indica o ( o mal s. ne wo k sea ch). Ou
esul s a e summa ized in Table 4whe e columns (1)-(3) co espond o he IPW while
columns (4)-(5) a e based on he PSM app oach. The o e all effec o finding a posi ion
ia ne wo ks on he loga i hm o s a ing wages is nega i e and obus : i is s a is ically
significan acco ding o all es ima ion me hods. Al hough he es ima ed wage penal y
associa ed wi h e e als a ies be ween −8% and −10% a linea hypo hesis abou he
in a-model equali y o he es ima ed coefficien s canno be ejec ed13.Mo eo e ,Table13
in Appendix 9.1 shows ha e e als om iends and ela i es make 84% o all e e al
cases wi h he ecommenda ions om colleagues making up he emaining 16%. This
means ha e e als om s ong ies a e much mo e equen in he da a han p o essional
ecommenda ions. Hence, we find ha hypo hesis H1 is s ongly suppo ed by he da a.
In he second s ep we pe o m he disagg ega ed analysis by dis inguishing be ween he
h ee ypes o ne wo k ies: iends, ela i es, and colleagues. Ou esul s a e summa ized
in columns (6)-(8) o Table 4. I shows ha he coefficien s o ela i es and iends a e
nega i e. Re e als om iends a e associa ed wi h a significan wage penal y equal
o −7%. A he same ime weak ies show a obus posi i e effec on s a ing wages
be ween 9% and 10%. The obse ed wage diffe ences a e in line wi h he heo e ical
model and suppo hypo heses H2 and H3. Howe e , a his s ep he empi ical suppo
o hese hypo heses is s ill incomple e since he unde lying mechanism in he model is
based on abili y/p oduc i i y diffe ences o wo ke g oups using diffe en sea ch channels.
The e o e, in he hi d s ep we use a p oxy o he cogni i e abili ies o wo ke s and
compa e hem ac oss he channels.
In he yea s 2006, 2012 and 2016 SOEP has in oduced a sho nume ical es o
e alua e cogni i e abili ies o he esponden s. This es is e e ed o as a symbol-digi
13As o he sample sizes o he es ima ed g oups, he PSM sample is smalle in e ms o he numbe
o he unique obse a ions bu he equency weigh s compensa e o his diffe ence. The IPW es ima es
do no esul in he sample sh inkage bu a e es ic ed by he collec ed in o ma ion on con ol a iables
in he ea men assignmen and/o ou come models.
16
Table 4: Effec s o he job sea ch channels compa ed o he o mal app oach
Ne wo ks: o e all Ne wo ks: disagg ega ed
(1) (2) (3) (4) (5) (6) (7) (8)
IPW AIPW IPWRA PSM PSM IPW AIPW IPWRA
Ne wo ks -0.074*** -0.106*** -0.104*** -0.131*** -0.082***
(0.013) (0.018) (0.018) (0.025) (0.023)
Ne wo ks: iends -0.071*** -0.071*** -0.072***
(0.018) (0.021) (0.021)
Ne wo ks: ela i es -0.023 -0.001 -0.006
(0.039) (0.035) (0.035)
Ne wo ks: colleagues 0.063 0.087** 0.101***
(0.041) (0.040) (0.039)
Obse a ions 10133 6126 6126 12677 7529 5986 5100 5100
T ea men assignmen model logi logi logi logi logi mlogi mlogi mlogi
Ou come model Weig. A g. linea by ML linea Weig. A g linea Weig. A g. linea by ML linea
Con ol a iables in ou come model No Yes Yes No Yes No Yes Yes
No es: *** p<0.01, ** p<0.05, * p<0.10; S anda d e o s a e clus e ed a NUTS3 le el. Con ol a iables include demog aphic con ols on a
pe sonal/hh le el as well as egional le el con ol a iables. The ou come a iable is loga i hm o he g oss mon hly wages.
es (SDT) and e eals inna e abili ies o he indi idual and he speed o sol ing new
asks. Fo a de ailed desc ip ion see Lang e al. (2007) who show ha SDT ou comes a e
sufficien ly co ela ed wi h es sco es o a mo e comp ehensi e in elligence es . Du ing
he cou se o he es pa icipan s a e gi en 1.5 minu es (90 seconds) o connec he
symbols wi h he digi s, gi en a p e-defined con e sion able. Fo ins ance, digi “1”
co esponds o a symbol “*” and digi “2” o “?”. The ques ion which he esponden s
a e equi ed o answe is “Which numbe co esponds o he symbol “*””? The numbe
o co ec , alse and o al answe s is measu ed a e 30, 60 and 90 seconds in he es .
This es was al eady used as a p oxy o he in-bo n cogni i e abili ies in o he s udies
(Heineck and Ange 2010; Rich e e al. 2017).
Figu e 2shows he dis ibu ion o he numbe o co ec answe s a e 30, 60 and 90
seconds, agg ega ed o e all a ailable es yea s. The mean sco e is inc easing o e ime
om 9.4 co ec answe s in 30 seconds o 29.7 in 90 seconds. Mo eo e , he a iance o
he dis ibu ion inc eased as well, almos doubling om 5.74 o 12.52. Ne e heless, in
all h ee cases app oxima ely 5% o he esponden s ha en’ gi en any co ec answe s.
Un o una ely, he a ailable sample size is low, especially in he disagg ega ed ne wo k
se up s a ing in 2014. Consequen ly, assuming ha he es sco es o adul s a e s able
o e se e al yea s, we ex apola e he esul s o he es pe o med in 2016 o he o he
yea s o in e es in o de o achie e a la ge sample size14.
Table 5 epo s he dis ibu ion o impu ed es sco es by sea ch channels. Fi s h ee
lines o he able show a numbe o co ec answe s in 30, 60 and 90 seconds espec i ely.
14Raw gaps in cogni i e abili ies be o e ex apola ion a e p o ided in able 17 in appendix 9.3.How-
e e , e en a e he ex apola ion he sample size is no sufficien ly la ge o he cogni i e sco es o be
included in he IPW eg essions, so we a e bound o he desc ip i e analysis o es sco es.
17
The nex h ee lines a e dedica ed o he numbe o w ong answe s in he same ime
pe iod, and he las lines demons a e he o al numbe o answe s gi en. The g oups
ha e se e al p o ound s a is ically significan diffe ences. Those who ound a posi ion
wi h he help o colleagues, ga e mo e co ec answe s and made ewe mis akes han any
o he g oup. The second bes g oup in e ms o cogni i e sco es a e hose who ound a
job indi idually. They make significan ly mo e mis akes han hose who loca ed a job
wi h he help o colleagues. A he same ime, hey ga e mo e co ec and o al answe s
and made ewe mis akes han hose who ound a job wi h he help o iends o amily.
These disc epancies e eal ha he a e age cogni i e abili y indeed diffe s be ween he
g oups, wi h ela i ely low-abili y esponden s elying on hei iends o amily o find a
job, middle-abili y ones finding a job indi idually and high-abili y esponden s acqui ing
a posi ion ia e e als om colleagues. We conclude ha his e idence p o ides u he
suppo o hypo heses H2 and H3.
Figu e 2: Dis ibu ions o sco es in he Symbol digi es a diffe en du a ions
Fu he , we find ha diffe ences in he SDT sco es a e obus o con olling o he
yea s o educa ion, mode o he in e iew15, and ID o an in e iewe , despi e a u he
educ ion in he sample size. Fo ins ance, he numbe o w ong answe s gi en in 60
seconds emains s a is ically significan ly diffe en be ween he g oups. Mo e de ail is
a ailable in Table 15 in Appendix 9.3.
Ano he i al pa o indi idual cha ac e is ics includes non-cogni i e abili ies. Fo
example, s onge sel -confidence o be e nego ia ion skills could ha e an affec on he
15This is he way he ques ionnai e was adminis a ed, e.g., a compu e assis ed pe sonal in e iew
(CAPI) o a pape assis ed one (PAPI).
18
Table 5: Nume ic es s (cogni i e abili y), disagg ega ed ne wo k ypes.
(1) (2) (3) (4) T- es
Indi idually NW: iends NW: amily NW: colleagues Diffe ence
Va iable Mean/SE Mean/SE Mean/SE Mean/SE (1)-(2) (1)-(3) (1)-(4) (2)-(3) (2)-(4) (3)-(4)
Co ec asw. (30 sec) 11.148 10.734 11.035 11.428 0.414* 0.113 -0.280 -0.301 -0.694** -0.393
(0.292) (0.176) (0.279) (0.246)
Co ec asw. (60 sec) 23.499 22.587 23.015 23.808 0.911** 0.484 -0.309 -0.427 -1.220* -0.793
(0.489) (0.318) (0.486) (0.482)
Co ec asw. (90 sec) 35.294 34.016 34.435 35.684 1.278** 0.859 -0.391 -0.419 -1.669* -1.249
(0.614) (0.476) (0.677) (0.657)
W ong asw. (30 sec) 0.211 0.255 0.257 0.166 -0.044 -0.046 0.045 -0.003 0.089** 0.091**
(0.018) (0.019) (0.038) (0.025)
W ong asw. (60 sec) 0.361 0.490 0.496 0.260 -0.129*** -0.136** 0.101*** -0.007 0.230*** 0.237***
(0.017) (0.038) (0.046) (0.023)
W ong asw. (90 sec) 0.574 0.732 0.803 0.547 -0.158*** -0.229** 0.028 -0.071 0.186** 0.257**
(0.039) (0.044) (0.089) (0.035)
To al asw. (30 sec) 11.359 10.989 11.293 11.594 0.370* 0.066 -0.235 -0.304 -0.605* -0.301
(0.283) (0.179) (0.274) (0.234)
To al asw. (60 sec) 23.859 23.077 23.511 24.067 0.782* 0.348 -0.208 -0.434 -0.990 -0.556
(0.492) (0.309) (0.452) (0.478)
To al asw. (90 sec) 35.868 34.748 35.238 36.231 1.120** 0.630 -0.363 -0.490 -1.483* -0.993
(0.626) (0.467) (0.646) (0.672)
Numbe o obse a ions 1717 1945 442 484 3662 2159 2201 2387 2429 926
Numbe o clus e s 16 16 16 16 16 16 16 16 16 16
No es: Values a e winso ised a 1 and 99 pe cen iles. Impu ed o all yea s bu 2016, assuming he esul s o he es being cons an o he
same esponden o e 5 yea s. The alue displayed o - es s a e he diffe ences in he means ac oss he g oups. S anda d e o s a e
clus e ed a ede al s a e le el. Obse a ions a e weigh ed using a iable ph as aweigh weigh s.***, **, and * indica e significance a he
1, 5, and 10 pe cen c i ical le el. Table 17 based on o iginal alues is a ailable in Appendix 9.3.
s a ing wages. In o de o es i hese diffe ences a e ele an o no , we use a se o
ques ions on he pe sonali y ai s con ained in SOEP (see McC ae and Cos a (1999) o
u he de ails). These ques ions can be used o cons uc he “Big Fi e” pe sonali y
ai s as shown by Uysal and Pohlmeie (2011) and Caliendo e al. (2016). We use he
ollowing a iables: being a ho ough wo ke and ca ying ou asks efficien ly (p oxies
o conscien iousness), being o iginal and aluing a is ic expe iences (p oxies o he
openness o expe ience), being sociable and communica i e (p oxies o ex a e sion),
being able o o gi e and iendly wi h o he s (p oxies o ag eeableness), wo ying a lo
and being ne ous (p oxies o neu o icism). Howe e , Table 16 in Appendix 9.3 shows no
sys ema ic s a is ically significan diffe ences be ween he g oups. One mino excep ion
is an abili y o o gi e, whe eby mo e o gi ing indi iduals a e mo e o en e e ed by
hei amily membe s compa ed o hose sea ching o mally o ecommended by iends.
In he final s ep we s udy he implica ions o egional cha ac e is ics. We find ha
coun y-specific ac o s ha e p onounced influence on he sel -selec ion o wo ke s in o
o mal job sea ch e sus ne wo ks. Table 6demons a es he coefficien s o he fi s -s age
eg essions o he egional a iables, which ha e been included. In all cases - o he
agg ega ed and disagg ega ed se up - highe egional unemploymen a e is associa ed
wi h a highe p obabili y o u ilizing ne wo ks. On he o he hand, he loga i hm o he
popula ion densi y in a coun y is nega i ely ela ed o he p obabili y o using a ne wo k
ie o e e y g oup bu colleagues. The con ibu ion o he sha e o mig an s is mos
19
p onounced in he case o iendship ies. This is in line wi h he p e ious li e a u e
showing ha immig an wo ke s o en ely on hei ne wo k o iends o ob aining jobs
(see, o example, Dus mann e al. (2016)).
Table 6: Effec s o egional coefficien s on sel -selec ion in job sea ch
PSM IPW
(1)
NW: All
(2)
NW: F iends
(3)
NW: Family
(4)
NW: Colleagues
Unemploymen a e 0.049*** 0.045** 0.055** 0.043*
(0.010) (0.019) (0.024) (0.025)
Log o popula ion densi y -0.131*** -0.225*** -0.303*** -0.008
(0.046) (0.077) (0.094) (0.097)
Sha e o mig an s 0.006 0.035** 0.005 0.002
(0.008) (0.014) (0.018) (0.018)
Obse a ions 10764 2363 564 490
No es: *** p<0.01, ** p<0.05, * p<0.10; S anda d e o s a e clus e ed a NUTS3 le el. Con ol a iables include demog aphic con ols on a
pe sonal/hh le el as well as egional le el con ol a iables. The ou come a iable is a p obabili y o belonging o a g oup based on he job
sea ch me hod. Logi is used o model he bina y choice; mul inomial logi is used o p edic he esul s disagg ega ed by a ne wo k ype.
Numbe o obse a ions is weigh ed in he PSM. In he IPW, numbe o obse a ions p esen s a ailable obse a ions by g oup. The o al
numbe o obse a ions in IPW is 5986.
To sum up, when e alua ing he ne wo k effec on s a ing wages in Ge many, he
coefficien s a e nega i e and obus indica ing an app ox. -9% lowe s a ing wage. How-
e e , when he esul s a e disagg ega ed by he ne wo k ypes, we can see ha jobs ound
ia iends and amily a e associa ed wi h a wage penal y o app oxima ely -7% compa ed
o he o mal indi idual app oach. Con e sely, using he help o colleagues is associa ed
wi h a 9% highe s a ing wage. In line wi h he heo e ical model, he sel -selec ion in o
job sea ch me hods is co ela ed wi h he esul s o he cogni i e abili ies es s: he g oup
o hose, who ound a job wi h he help o colleagues shows he bes ou comes.
6 He e ogenei y Analysis
6.1 Desc ip i e Analysis
This sec ion is dedica ed o he analysis o egional and gende diffe ences in he job
sea ch channels and associa ed wages. Men and women end o ha e diffe en beha iou al
pa e ns in labou ma ke s. Howe e , e en i he beha io is simila ma ke o ces may
gene a e diffe en labou ma ke ou comes (Pe ez 2019), hus, we s udy he implica ions
o e e al hi ing sepa a ely o men and women. In addi ion, we accoun o he long
20
his o ical di ision o Ge many be ween he Eas and he Wes .
Figu e 3p esen s a spa ial dis ibu ion o job sea ch ia ne wo ks (blue) and ia he
o mal indi idual app oach (g een). No e ha he sample is es ic ed o only hese
wo channels. Al hough he e is no ede al s a e, whe e he o mal job sea ch me hod is
mo e p e alen han ne wo king, he spa ial pa e ns a e p esen o bo h. Finding a job
o mally is mo e common in he sou h (e.g. he s a e o Baden-W¨u embe g) and in he
ci y-s a es Be lin and Hambu g, whe eas he use o ne wo ks is mos common in eas e n
ede al s a es and in he wes (Rheinland P alz). Figu e 7in Appendix 9.1 demons a es
an o e lap o egional diffe ences in he job sea ch me hods wi h gende . The ela i e
impo ance o he o mal channel in he sou h is mos ly d i en by he subsample o men.
Whe eas he p e alence o he ne wo k channel in he wes is d i en by he subsample o
women. We find ha up o 61% o women in he wes end o find a posi ion wi h he
help o ne wo ks, while in he o he pa s o he coun y he sha e is lowe .
Figu e 3: Job sea ch channels in he yea s 2002-2019
In e ms o he obse able demog aphic cha ac e is ics o men and women, he mos
p o ound con as is he ollowing. Al hough men and women in he compa ison g oup
a e s a is ically significan ly diffe en only in h ee cha ac e is ics16, he g oups o men
and women, who ound hei posi ions wi h he help o ne wo ks, a e diffe en in all
16The sha e o ansi ions om unemploymen o employmen is highe among men, while he sha e
o esponden s employed pa - ime and he a e age numbe o kids in a household a e highe among
women. The able wi h desc ip i e e idence by gende is a ailable om he au ho s upon a eques .
21
demog aphic cha ac e is ics bu he sha e o mid-sized fi ms. This e idence suppo s ou
mo i a ion o pe o m he analysis o subg oups dis inguishing be ween he wo b oad
egions (Eas s. Wes ) and he wo gende g oups.
6.2 Sepa a ing he effec s by gende
Table 7demons a es he es ima ion esul s in he agg ega ed se up wi h a bina y sea ch
channel indica o . The esul s o men a e con ained in Panel A, while he coefficien s
o women a e epo ed in Panel B. Bo h panels indica e wage penal ies associa ed wi h
e e al hi ing, howe e , o men he esul s a e no obus and a y om s a is ically
insignifican o −9%. Fo women, on he o he hand, all he es ima o s ha e o e lapping
confidence bounds and a mean close o −11% in s a ing wages. This sugges s ha he
penal ies om e e al hi ing a e la ge o women.
Table 7: Effec s o finding a job wi h he help o ne wo ks compa ed o he o mal
app oach, sample spli ing by sex
(1) (2) (3) (4) (5)
IPW AIPW IPWRA PSM PSM
Panel A: Agg ega ed Ne wo ks, Men
Ne wo ks -0.036* -0.096*** -0.088*** -0.077** -0.039
(0.019) (0.029) (0.029) (0.037) (0.034)
Obse a ions 4264 2168 2168 5569 2768
Panel B: Disagg ega ed Ne wo ks, Women
Ne wo ks -0.105*** -0.107*** -0.105*** -0.161*** -0.091***
(0.018) (0.021) (0.021) (0.034) (0.029)
Obse a ions 5869 3958 3958 7108 4761
T ea men assignmen model logi logi logi logi logi
Ou come model Weig. A g linea by ML linea Weig. A g linea
Con ol a iables in ou come model No Yes Yes No Yes
No es: *** p<0.01, ** p<0.05, * p<0.10; S anda d e o s a e clus e ed a NUTS3 le el. Con ol a iables include demog aphic con ols on a
pe sonal/hh le el as well as egional le el con ols and ede al s a e dummies. The ou come a iable is loga i hm o he g oss mon hly
wages.
The disagg ega ed se up in Table 8highligh s an e en mo e he e ogeneous pic u e o
he wo g oups. Fo men, none o he coefficien s in any o he me hods is s a is ically
significan ly diffe en om ze o. Mo eo e , he es ima ed ampli ude is app oaching ze o
as well. Fo women, on he o he hand, he coefficien s associa ed wi h e e als om
iends and colleagues a e obus and s a is ically significan in all he h ee es ima ion
me hods. In line wi h he heo e ical p edic ions, hose, who ound a job wi h he help o
22
Figu e 6: Effec s o finding a job ia ne wo ks on wages in consequen yea s
8 Discussion and Conclusions
In his pape we s udy he ela ionship be ween he job sea ch channel and he s a ing
wage. The analysis is ocused on compa ing he wages o e e ed wi h non- e e ed
applican s by using he in e se-p obabili y weigh ing app oach. Ou esul s show a s able
nega i e effec o abou -10% associa ed wi h finding a job wi h he help o a social con ac
in Ge many. This wage gap is obus i es ima ed by diffe en me hods and using a ious
subsamples, mo eo e , i emains s able se e al yea s a e he s a o he employmen
ela ionship. Disagg ega ing he effec by he ype o he social con ac , wo opposi e
pa e ns a e e ealed - while jobs ound ia iends esul in -7% lowe s a ing wages,
he posi ions ound wi h he help o colleagues a e associa ed wi h a wage p emia o 9%.
Combining his analysis wi h in o ma ion on he cogni i e abili y sco e om he sym-
bol digi es (SDT) we find ha applican s ecommended by hei colleagues pe o m
be e in he abili y es han hose finding jobs in he o mal way. In con as , appli-
can s finding jobs wi h a help o a iend pe o m wo se in he es han any o he wo ke
g oup. These findings a e consis en wi h he so ing model o job sea ch as de eloped by
S upny ska and Zaha ie a (2015) and Les e e al. (2021) as well as he seminal wo k by
Mon gome y (1991) and G ano e e (1995). The he e ogenei y analysis shows significan
diffe ences by gende : o men none o he coefficien s is s a is ically significan , while
o women he diffe ences a e mo e p onounced. I is consis en wi h a la ge a ia ion
in cogni i e sco es documen ed o women and indica ing a s onge so ing o women
ac oss sea ch channels compa ed o men.
Concluding he pape , we discuss he limi a ions o ou analysis. The fi s limi a ion
is in he na u e o he da a since SOEP has limi ed in o ma ion abou he employe ,
hus, unobse ed employe he e ogenei y is no accoun ed o . Fu he , ou s udy does
29
no co e he esponden s who ound a job wi h he help o an employmen agency o
e u ned o a p e ious employe . Ano he es ic ion is sel -selec ion in he labou ma ke
pa icipa ion, which esul s in wages o non-pa icipan s no being obse ed. On he one
hand, diffe en me hods allow con olling o his pa o sel -selec ion. On he o he
hand, he usual assump ion is ha he sign o sel -selec ion is known. Machado (2017)
a gues ha o women in he labou ma ke i is no he case, wi h sel -selec ion being
posi i e o he lowe deciles o he abili y dis ibu ion and nega i e o he highe ones.
Thus, one canno claim he exac di ec ion o he bias. Howe e , i he usual posi i e
sel -selec ion is assumed, his pape es ima es he lowe bound o he effec .
Re e ences
Addison, John T and Ped o Po ugal (2002). “Job sea ch me hods and ou comes”. In:
Ox o d Economic Pape s 54.3, pp. 505–533.
An oninis, Manos (2006). “The wage effec s om he use o pe sonal con ac s as hi ing
channels”. In: Jou nal o Economic Beha io & O ganiza ion 59.1, pp. 133–146.
Aus in, Pe e C (2014). “A compa ison o 12 algo i hms o ma ching on he p opensi y
sco e”. In: S a is ics in medicine 33.6, pp. 1057–1069.
Ben olila, Samuel, Claudio Michelacci, and Ja ie Sua ez (2010). “Social Con ac s and
Occupa ional Choice”. In: Economica 77.305, pp. 20–45. issn: 1468-0335.
Caliendo, Ma co, S effen K¨unn, and Ma in Weissenbe ge (2016). “Pe sonali y ai s and
he e alua ion o s a -up subsidies”. In: Eu opean Economic Re iew 86, pp. 87–108.
Cappella i, Lo enzo and Kons an inos Ta si amos (2015). “Wi h a Li le Help om My
F iends? Quali y o Social Ne wo ks, Job Finding and Job Ma ch Quali y
§
”. In:
Eu opean Economic Re iew 78, pp. 55–75.
Dela e, E ic and Ma e a Saba ie (2007). “Social capi al and wages: an econome ic
e alua ion o social ne wo king’s effec s”. In: Labou 21.2, pp. 209–236.
Di ksmeie , Pe e (2015). “The in ica e geog aphies o gende ideologies in Ge many”.
In: Geo o um 64, pp. 12–24.
Dus mann, Ch is ian e al. (2016). “Re e al-based job sea ch ne wo ks”. In: The Re iew
o Economic S udies 83.2, pp. 514–546.
Fon aine, F ancois (2008). “Why a e simila wo ke s paid diffe en ly? The ole o social
ne wo ks”. In: Jou nal o Economic Dynamics and Con ol 32.12, pp. 3960–3977.
Galenianos, Manolis (2013). “Lea ning abou ma ch quali y and he use o e e als”. In:
Re iew o Economic Dynamics 16.4, pp. 668–690.
— (2014). “Hi ing h ough e e als”. In: Jou nal o Economic Theo y 152, pp. 304–323.
Glynn, Adam N and Ke in M Quinn (2010). “An in oduc ion o he augmen ed in e se
p opensi y weigh ed es ima o ”. In: Poli ical analysis 18.1, pp. 36–56.
Goebel, Jan e al. (2018). “The Ge man Socio-Economic Panel (SOEP)”. In: Jah b¨uche
¨u Na ional¨okonomie und S a is ik/ Jou nal o Economics and S a is ics.
Goos, Maa en and Anna Salomons (2007). “Dange ous liaisons: a social ne wo k model
o he gende wage gap”. In: A ailable a SSRN 1021699.
G ano e e , Ma k (1995). Ge ing a Job: A S udy o Con ac s and Ca ee s. Uni e si y
o Chicago P ess.
G ano e e , Ma k S (1973). “The s eng h o weak ies”. In: Ame ican jou nal o soci-
ology 78.6, pp. 1360–1380.
30
Guo, Shenyang and Ma k W F ase (2014). P opensi y sco e analysis: S a is ical me hods
and applica ions. Vol. 11. SAGE publica ions.
G¨u zgen, Nicole and Lau a Pohlan (2024). “Do employe s lea n mo e om e e als han
om o he ec ui men channels?” In: Labou Economics, p. 102574.
Heineck, Guido and Silke Ange (2010). “The e u ns o cogni i e abili ies and pe sonali y
ai s in Ge many”. In: Labou economics 17.3, pp. 535–546.
Hens ik, Lena and Oska No ds ¨om Skans (2016). “Social ne wo ks, employee selec ion,
and labo ma ke ou comes”. In: Jou nal o Labo Economics 34.4, pp. 825–867.
Ho ´a h, Ge gely (2014). “Occupa ional misma ch and social ne wo ks”. In: Jou nal o
Economic Beha io & O ganiza ion 106, pp. 442–468.
Huffman, Ma and Lisa To es (2002). “I ’s no only “who you know” ha ma e s:
Gende , pe sonal con ac s, and job lead quali y”. In: Gende & Socie y 16.6, pp. 793–
813.
Imai, K. and M. Ra ko ic (2014). “Co a ia e balancing p opensi y sco e”. In: Jou nal o
he Royal S a is ical Socie y Se ies B: S a is ical Me hodology 76.1, pp. 243–263.
Imbens, Guido W and Jeff ey M Woold idge (2009). “Recen de elopmen s in he econo-
me ics o p og am e alua ion”. In: Jou nal o economic li e a u e 47.1, pp. 5–86.
Ioannides, Yannis M and Linda Da che Lou y (2004). “Job In o ma ion Ne wo ks, Neigh-
bo hood Effec s, and Inequali y”. In: Jou nal o Economic Li e a u e 42.4, pp. 1056–
1093.
Ioannides, Yannis M and Ad iaan R Soe e en (2006). “Wages and employmen in a
andom social ne wo k wi h a bi a y deg ee dis ibu ion”. In: Ame ican Economic
Re iew 96.2, pp. 270–274.
Jackson, Ma hew O. (2008). Social and Economic Ne wo ks. P ince on Uni e si y P ess.
King, Ga y and Richa d Nielsen (2019). “Why p opensi y sco es should no be used o
ma ching”. In: Poli ical analysis 27.4, pp. 435–454.
Kugle , Ad iana D (2003). “Employee e e als and efficiency wages”. In: Labou eco-
nomics 10.5, pp. 531–556.
Ku z, Ch is oph F (2022). “Augmen ed in e se p obabili y weigh ing and he double
obus ness p ope y”. In: Medical Decision Making 42.2, pp. 156–167.
Labini, Mau o Sylos (2005). Social Ne wo ks and Wages: I is all abou connec ions!
Ci esee .
Lang, F iede R e al. (2007). “Assessing Cogni i e Capaci ies in Compu e -Assis ed
Su ey Resea ch: Two Ul a-Sho Tes s o ln ellec ual Abili y in he Ge man Socio-
Economic Panel (SOEP)”. In: Jou nal o Con ex ual Economics–Schmolle s Jah buch
1, pp. 183–191.
Les e , Benjamin R., Da id Ri e s, and Gio gio Topa (2021). “The he e ogeneous impac
o e e als on labo ma ke ou comes”. In: FRB o New Yo k S aff Repo 987.
Machado, Cecilia (2017). “Unobse ed selec ion he e ogenei y and he gende wage gap”.
In: Jou nal o Applied Econome ics 32.7, pp. 1348–1366.
Ma golis, Da id N and V´e onique Simonne (2003). “Educa ional ack, ne wo ks and
labo ma ke ou comes”. In: Ne wo ks and Labo Ma ke Ou comes (Janua y 2003).
Ma ma os, Da id and B uce Sace do e (2002). “Pee and social ne wo ks in job sea ch”.
In: Eu opean Economic Re iew 46.4-5, pp. 870–879. issn: 0014-2921.
McC ae, Robe R. and Paul T. Cos a (1999). “A fi e- ac o heo y o pe sonali y”. In:
In: Pe in, L.A., John, O.P. (Eds.), Handbook o Pe sonali y: Theo y and Resea ch,
pp. 139–153.
31
Meliciani, Valen ina and Debo a Radicchia (2011). “The in o mal ec ui men channel
and he quali y o job-wo ke ma ches: an analysis on I alian su ey da a”. In: Indus-
ial and co po a e change 20.2, pp. 511–554.
Mon gome y, James D (1991). “Social Ne wo ks and Labo -Ma ke Ou comes: Towa d
an Economic Analysis”. In: 81.5, pp. 1408–1418.
Pellizza i, Michele (2010). “Do F iends and Rela i es Really Help in Ge ing a Good
Job?:” in: IRL Re iew 63.3, pp. 494–510. issn: 00197939.
Pe ez, Ca oline C iado (2019). In isible women: Da a bias in a wo ld designed o men.
Ab ams.
Pis a e i, Luigi (1999). “In o mal ne wo ks in he I alian labo ma ke ”. In: Gio nale
degli economis i e annali di economia, pp. 355–375.
Ponzo, Michela and Vincenzo Scoppa (2010). “The use o in o mal ne wo ks in I aly:
Efficiency o a o i ism?” In: The Jou nal o Socio-Economics 39.1, pp. 89–99.
Rebien, Ma ina, Michael S ops, and Anna Zaha ie a (2020). “Fo mal sea ch and e e als
om a fi m’s pe spec i e”. In: In e na ional Economic Re iew 61.4, pp. 1679–1748.
Rich e , Da id e al. (2017). SOEP scales manual (upda ed o SOEP-Co e 32. 1).Tech.
ep. SOEP su ey pape s.
Rubin, Donald B (2001). “Using p opensi y sco es o help design obse a ional s udies:
applica ion o he obacco li iga ion”. In: Heal h Se ices and Ou comes Resea ch
Me hodology 2, pp. 169–188.
Schnabel, Claus (2016). “Uni ed, ye apa ? A no e on pe sis en labou ma ke diffe -
ences be ween wes e n and eas e n Ge many”. In: Jah b¨uche ¨u Na ional¨okonomie
und S a is ik 236.2, pp. 157–179.
Simon, Cu is J and John T Wa ne (1992). “Ma chmake , ma chmake : The effec o
old boy ne wo ks on job ma ch quali y, ea nings, and enu e”. In: Jou nal o labo
economics 10.3, pp. 306–330.
Sp engholz, Maximilian, Anna Wiebe , and Elke Hols (2022). “Gende iden i y and
wi es’ labo ma ke ou comes in Wes and Eas Ge many be ween 1983 and 2016”.
In: Socio-Economic Re iew 20.1, pp. 257–279.
S upny ska, Yuliia and Anna Zaha ie a (2015). “Explaining U-shape o he e e al hi ing
pa e n in a sea ch model wi h he e ogeneous wo ke s”. In: Jou nal o Economic
Beha io & O ganiza ion 119, pp. 211–233.
Ti us, Ma in A (2007). “De ec ing selec ion bias, using p opensi y sco e ma ching, and
es ima ing ea men effec s: An applica ion o he p i a e e u ns o a mas e ’s de-
g ee”. In: Resea ch in Highe Educa ion 48, pp. 487–521.
Topa, Gio gio (2011). “Labo ma ke s and e e als”. In: Handbook o social economics.
Vol. 1. Else ie , pp. 1193–1221.
Uysal, De ya and Win ied Pohlmeie (2011). “Unemploymen du a ion and pe sonali y”.
In: Jou nal o Economic Psychology 32.6, pp. 980–992.
Vable, Anusha M e al. (2019). “Pe o mance o ma ching me hods as compa ed wi h
unma ched o dina y leas squa es eg ession unde cons an effec s”. In: Ame ican
jou nal o epidemiology 188.7, pp. 1345–1354.
W zus, Co nelia e al. (2013). “Social ne wo k changes and li e e en s ac oss he li e span:
a me a-analysis.” In: Psychological bulle in 139.1, p. 53.
Zaha ie a, Anna (2018). “On he op imal di e sifica ion o social ne wo ks in ic ional
labou ma ke s wi h occupa ional misma ch”. In: Labou Economics 50, pp. 112–127.
Zhou, Min (2019). “Gende Diffe ences in he P o ision o Job-Sea ch Help”. In: Gende
and Socie y 33.5, pp. 746–771. issn: 08912432.
32
9 Appendix
9.1 De ailed demog aphic cha ac e is ics
Table 12: Blinde -Oaxaca decomposi ion
Panel A: o e all decomposi ion
Es ima ed alues
Indi idual 7.149 (0.02)***
Ne wo king: o e all 6.832 (0.02)***
To al diffe ence 0.316 (0.02)***
Endowmen s 0.210 (0.02)***
Coefficien s 0.108 (0.02)***
Panel B: de ailed disagg ega ion
Endowmen s Coefficien s In e ac ion
Age -0.023 (0.02) -0.416 (0.49) 0.006 (0.01)
Age, squa ed 0.029 (0.02)* 0.246 (0.24) -0.008 (0.01)
Fede al S a e 0.001 (0.00) 0.044 (0.04) -0.001 (0.00)
Educa ion: less han high school -0.005 (0.00) -0.005 (0.03) 0.001 (0.01)
Educa ion: mo e han high school 0.007 (0.01) -0.011 (0.02) -0.006 (0.01)
Job- o-job ansi ion 0.000 (0.00) 0.018 (0.02) 0.001 (0.00)
Fi m size by empl: No employees 0.000 (.) 0.000 (.) 0.000 (.)
Fi m size by empl: Small 0.005 (0.01) 0.068 (0.06) -0.020 (0.02)
Fi m size by empl: Middle 0.001 (0.00) 0.054 (0.04) 0.001 (0.00)
Fi m size by empl: La ge 0.036 (0.01)*** 0.049 (0.05) 0.021 (0.02)
Fi s Job Full Time -0.001 (0.00) -0.049 (0.03)* 0.001 (0.00)
# o child en in HH 0.003 (0.00) 0.002 (0.04) -0.000 (0.00)
Numbe o Pe sons in HH -0.003 (0.00) -0.017 (0.05) 0.001 (0.00)
Annual Wo k Hou s o Indi idual 0.056 (0.01)*** -0.088 (0.03)*** -0.010 (0.00)***
The younges child is in ki a age o smalle -0.000 (0.00) 0.015 (0.01) -0.001 (0.00)
Long- e m unemployemen in coun y (INKAR) -0.000 (0.00) -0.018 (0.09) 0.000 (0.00)
Ma ied, li ing oge he -0.000 (0.00) -0.024 (0.02) 0.001 (0.00)
Mig a ion backg ound 0.000 (0.00) 0.016 (0.01) -0.003 (0.00)
NACE code ( e 1.1) 0.001 (0.00) 0.034 (0.04) -0.000 (0.00)
Occupa ion & T aining: misma ch 0.033 (0.01)*** -0.013 (0.02) 0.003 (0.00)
Occupa ion & T aining: in aining 0.001 (0.01) 0.001 (0.01) -0.000 (0.00)
Occupa ion & T aining: no aining 0.015 (0.00)*** -0.000 (0.01) 0.000 (0.00)
Tenu e -0.002 (0.00)*** 0.012 (0.01) -0.003 (0.00)
Popula ion-wo kplace densi y in coun y (INKAR) 0.005 (0.00)** -0.019 (0.01) -0.003 (0.00)
Ru al o u ban egion -0.000 (0.00) 0.008 (0.03) 0.000 (0.00)
Mo ed o ano he coun y 0.005 (0.00)** 0.002 (0.00) 0.002 (0.00)
Sex -0.006 (0.00)* -0.002 (0.01) 0.000 (0.00)
Su ey Yea -0.002 (0.00) -19.524 (9.88)** 0.001 (0.00)
To al # o yea s a esp. spen unempl om 2001 0.004 (0.00)* 0.025 (0.03) -0.001 (0.00)
T ansi ion om unempl. o empl. 0.002 (0.00) 0.007 (0.01) 0.000 (0.00)
Unemploymen a e in coun y (INKAR) 0.005 (0.00)* 0.015 (0.05) -0.001 (0.00)
Yea s o educa ion 0.045 (0.01)*** 0.215 (0.18) 0.015 (0.01)
Cons an 19.461 (9.86)**
No es: S anda d e o s a e clus e ed a NUTS3 le el and epo ed in pa an hesis. *, ** and *** indica e 10-, 5- and 1-pe cen c i ical le el.
The dependen a iable is loga i hm o g oss mon hly wages.
33
Table 13: Demog aphic cha ac e is ics condi ional on he job sea ch me hod, disagg e-
ga ed ne wo k ypes.
(1) (2) (3) (4) (2)-(1) (3)-(1) (4)-(1)
Indi idually Ne wo ks: iends Ne wo ks: amily Ne wo ks: colleagues Pai wise - es
Va iable N/Clus e s Mean/(SE) N/Clus e s Mean/(SE) N/Clus e s Mean/(SE) N/Clus e s Mean/(SE) N/Clus e s Mean diffe ence N/Clus e s Mean diffe ence N/Clus e s Mean diffe ence
Na u al loga i hm o g oss wages 4269 7.442 4838 7.147 1220 7.018 1152 7.683 9107 -0.294*** 5489 -0.424*** 5421 0.242***
96 (0.028) 95 (0.031) 91 (0.056) 88 (0.040) 96 96 96
Men 4305 0.435 4890 0.521 1236 0.484 1158 0.546 9195 0.086*** 5541 0.049 5463 0.111***
96 (0.016) 95 (0.012) 91 (0.025) 88 (0.021) 96 96 96
Age, yea s 4305 35.272 4890 36.934 1236 34.252 1158 36.665 9195 1.661*** 5541 -1.020 5463 1.393***
96 (0.262) 95 (0.257) 91 (0.634) 88 (0.420) 96 96 96
Yea s o educa ions 4081 12.894 4566 11.662 1145 11.539 1101 13.163 8647 -1.231*** 5226 -1.354*** 5182 0.270*
96 (0.067) 95 (0.092) 91 (0.125) 88 (0.131) 96 96 96
Ma ied, li ing oge he 4289 0.370 4859 0.420 1231 0.435 1157 0.350 9148 0.050*** 5520 0.065** 5446 -0.020
96 (0.012) 95 (0.015) 91 (0.026) 88 (0.022) 96 96 96
# o HH membe s 4305 2.534 4890 2.695 1236 3.095 1158 2.429 9195 0.161*** 5541 0.561*** 5463 -0.105*
96 (0.039) 95 (0.040) 91 (0.072) 88 (0.054) 96 96 96
# o child en in HH 4305 0.899 4890 1.091 1236 1.055 1158 0.829 9195 0.192*** 5541 0.156** 5463 -0.070
96 (0.030) 95 (0.025) 91 (0.056) 88 (0.038) 96 96 96
Mig a ion backg ound 4305 0.326 4890 0.459 1236 0.412 1158 0.273 9195 0.134*** 5541 0.087** 5463 -0.053**
96 (0.017) 95 (0.028) 91 (0.039) 88 (0.022) 96 96 96
T ansi ion om unempl. o empl. 3566 0.105 3865 0.120 976 0.102 997 0.021 7431 0.015 4542 -0.003 4563 -0.084***
96 (0.010) 95 (0.012) 90 (0.016) 87 (0.006) 96 96 96
Job- o-job ansi ion 4305 0.587 4890 0.533 1236 0.474 1158 0.722 9195 -0.054*** 5541 -0.112*** 5463 0.135***
96 (0.018) 95 (0.022) 91 (0.031) 88 (0.023) 96 96 96
Main job ull- ime 4008 0.728 4556 0.757 1095 0.703 1100 0.767 8564 0.029** 5103 -0.025 5108 0.039
96 (0.013) 95 (0.012) 90 (0.022) 88 (0.023) 96 96 96
Annual wo k hou s 4305 1444.608 4890 1262.939 1236 1167.793 1158 1732.927 9195 -181.669*** 5541 -276.814*** 5463 288.319***
96 (27.660) 95 (30.786) 91 (49.016) 88 (33.418) 96 96 96
Occupa ional misma ch 4264 0.305 4807 0.441 1221 0.439 1155 0.300 9071 0.136*** 5485 0.134*** 5419 -0.005
96 (0.012) 95 (0.012) 91 (0.022) 88 (0.020) 96 96 96
Fi m size: <20 employees 4102 0.191 4499 0.213 1148 0.159 1109 0.170 8601 0.022 5250 -0.032* 5211 -0.021
96 (0.010) 95 (0.012) 91 (0.015) 88 (0.020) 96 96 96
Fi m size: [20,200) employees 4102 0.087 4499 0.078 1148 0.075 1109 0.068 8601 -0.009 5250 -0.012 5211 -0.019
96 (0.007) 95 (0.006) 91 (0.015) 88 (0.012) 96 96 96
Fi m size: ≥200 employees 4102 0.228 4499 0.176 1148 0.170 1109 0.235 8601 -0.052*** 5250 -0.058*** 5211 0.007
96 (0.012) 95 (0.009) 91 (0.020) 88 (0.021) 96 96 96
No es: Significance: ***=.01, **=.05, *=.1. E o s a e clus e ed a ede al s a e-yea le el. Obse a ions a e weigh ed using a iable ph as aweigh weigh s.
34
Figu e 7: Regional dis ibu ion o he job sea ch me hods, sample spli ing by sex.
35
9.2 Main a iables desc ip ion
This sec ion o he appendix desc ibes he cons uc ion o he main a iables which a e
used in he empi ical analysis.
Cu en job ma ke s a us and ansi ions
We use a iable pgjobch o define he cu en job ma ke s a us o a esponden . I is
u he augmen ed by using in o ma ion on he annual wo king ime: i he esponden
epo ed wo king mo e han 52+ hou s, s/he is conside ed o be employed. Based on
his in o ma ion we c ea e he ollowing ca ego ies: job- o-job ansi ion, unemploymen -
o-employmen ansi ion, non-pa icipa ion- o-employmen ansi ion, fi s ime pa ic-
ipan . To cons uc a eliable measu e o unemploymen we addi ionally used a iable
plb0021 a iable. I in he p e ious yea he esponden was egis e ed unemployed ( a i-
able plb0021) and in he cu en yea he s a us is “Employed (wi h/wi hou /unclea )
job change”, we coun ed is as a ansi ion om unemploymen o employmen . Fi s ime
pa icipan s a e excluded om he sample.
Jobsea chme hodcoding
The job sea ch me hod is he cen al a iable o he analysis (“ ea men ”), howe e ,
i is also he one which conside ably es ic s he sample size. Responden s unde aking
a paid ac i i y we e asked i s/he has ound a new posi ion in he pe iod om he Jan,
1 p e ious yea ill he ime o he su ey. The job change includes bo h change o he
posi ion wi hin one company and finding a new employe . I his ques ion was answe ed
affi ma i ely, a esponden was asked abou he way which lead o he cu en posi ion,
i.e., he job sea ch me hod. As SOEP co e s a long ime pe iod, he ways o find a job
ha e changed se e al imes. Fo example, in he yea 2014 ne wo king, as a job sea ch
me hod, was spli by he ne wo k ypes - iend, amily o colleagues. Hence, we use a
ha monized a iable co e ing 18 possible ways o find a posi ion. These 18 ca ego ies
we e agg ega ed o he ou main job sea ch me hods:
1. Fo mal cen alized app oach (agency) includes hose, who ha e ound a posi ion
using he help om an employmen office (A bei sam ), Job-Cen e /ARGE/social
office (Sozialam ), pe sonnel se ice agency (Pe sonalse iceagen u ) o p i a e job
placemen .
2. Fo mal indi idual sea ch co e s he job pos ings in newspape s and he In e ne .
3. In o mal indi idual sea ch includes finding a posi ion ia a social ne wo k o using
social capi al o iends, ela i es, acquiescences o o me colleagues. Thus, i is
mos ly e e ed o as ne wo king.
4. All o he ypes o he job sea ch a e collec ed in o “o he ” ca ego y.
36
Fo he analysis in his pape only hose esponden s who ha e ound hei posi ion
using o mal indi idual app oach ( e e ed o as “indi idually” o “compa ison g oup”)
and wi h he help o hei social capi al (“ne wo king” o “ ea men ”) a e included. All
a iables used o es ima ion and balance a e b iefly desc ibed in Table 14 below. Fi s
column shows a a iable’s name in he da a/ eg essions, while he second column p o ides
a sho desc ip ion o he economic meaning. The nex column indica es, i a a iable
was aken di ec ly om he da ase o was cons uc ed by he au ho s. The o h column
indica es, i a a iable is a dummy coded as 0-1. Finally, he las column p o ides no es
on cons uc ion o he a iables.
Table 14: Desc ip ion o a iables used
age Age, yea s
agesq Age squa ed, yea s Yes
bula Fede al s a e
bula ew Fede al S a e Acco ding o s a is-
ic Yes Eas -Wes Ve sion
educ s age Finished s age o educa ion Yes
Agg ega ion o he ye seduc a i-
able. <12 - less han high school;
[12,13] - high school l l; >13 - mo e
han high school
educ s age 1 Less han high school Yes Yes Based on educ s age
educ s age 2High school Yes Yes Based on educ s age
educ s age 3 Mo e han high school Yes Yes Based on educ s age
empl change A job- o-job ansi ion Yes Yes See a de ailed desc ip ion
fi m size A size o he fi m # o employees, based on pgbe
a iable
fi m size 1Ve y small: (0, 20) employees Yes Yes Based on i m size
fi m size 2 Small: [20, 100) employees Yes Yes Based on i m size
fi m size 3Middle: [100, 200) employees Yes Yes Based on i m size
fi m size 4 La ge: [200, 2000) employees Yes Yes Based on i m size
fi m size 5Ve y la ge: ≥2000 employees Yes Yes Based on i m size
ull ime Wo king ull- ime Yes
hh size #o HHmembe s
hou s wo k yea Annual wo k hou s
ki a less ch The younges child is in ki a age o
smalle Yes Yes
Based on kidgeb01 - kidgeb15 a i-
ables and syea a iable. Fi s , he
cu en age o a child is defined.
Then o each yea he age o he
younges child is sa ed. Finally, a
dummy a iable is c ea ed, indica -
ing i he child is below 6 yea s o
age in a gi en yea .
Va iable Name Sho Desc ip ion Cons . A 0-1
dummy
a iable
No es
Con inued on nex page
37
Table 14: Desc ip ion o a iables used (Con inued)
log labg o Loga i hm o g oss mon ly wages Yes
Basedon helabg o a iable.
Fi s , i was winso ized a 1 and
99 pe cen iles by he ype o he
occupa ion. Secondly, a loga i hm
was aken.
long unempl Long- e m unemploymen , sha e
pe coun y F om INKAR
ma s a us Ma ied, li ing oge he Yes Yes Based on he d11104
nace ull 1d NACE e 1.1, c ea ed, 1 digi Yes
Based on pgnace and pgnace2 a i-
ables. The fi s a iable is a ail-
able om 2002 o 2017 and is coded
acco ding o he fi s e iew o
NACE classifica ion. The second
a iable is a ailable om 2013 o
2019 and based on he e . 2 o
NACE classifica ion. Bo h a i-
ables a e agg ega ed o he majo
g oups and con e ed o e 1.1 ac-
co ding o he NACE guidelines.
nace ull 1d 1 Ag icul u e/fishing Yes Yes As pe NACE e 1.1
nace ull 1d 10 Real es a e/business ca i i ies Yes Yes As pe NACE e 1.1
nace ull 1d 11 Public adminis a ion Yes Yes As pe NACE e 1.1
nace ull 1d 12 Educa ion Yes Yes As pe NACE e 1.1
nace ull 1d 13 Heal h and social wo k Yes Yes As pe NACE e 1.1
nace ull 1d 14 Communi y and social se ices Yes Yes As pe NACE e 1.1
nace ull 1d 15 P i a e HHs as employe s Yes Yes As pe NACE e 1.1
nace ull 1d 16 Ac . o ex a e . o g/bodies Yes Yes As pe NACE e 1.1
nace ull 1d 2 Mining Yes Yes As pe NACE e 1.1
nace ull 1d 3Manu ac u ing Yes Yes As pe NACE e 1.1
nace ull 1d 4Elec /Gas/Wa e Supply/Man-
agemen Yes Yes As pe NACE e 1.1
nace ull 1d 5Cons uc ion Yes Yes As pe NACE e 1.1
nace ull 1d 6 Wholesale/ e ail ade Yes Yes As pe NACE e 1.1
nace ull 1d 7Ho els and es au an s Yes Yes As pe NACE e 1.1
nace ull 1d 8 T anspo and Communica ion Yes Yes As pe NACE e 1.1
nace ull 1d 9Financial In e media ion Yes Yes As pe NACE e 1.1
nonpa empl A shi om non-pa icipa ion in
labou ma ke o employmen Yes Yes See a de ailed desc ip ion
pge ljob Occupa ion ma ches aining
pge ljob 1 Yes, ma ches Yes Yes Based on pge ljob
pge ljob 2No, occ. misma ch Yes Yes Based on pge ljob
pge ljob 3 In aining Yes Yes Based on pge ljob
pge ljob 4No aining Yes Yes Based on pge ljob
pge wzei Tenu e A a p e ious employe
Va iable Name Sho Desc ip ion Cons . A 0-1
dummy
a iable
No es
Con inued on nex page
38
