Busch, Fabian; Fenge, Robe ; Ochsen, Ca s en
A icle — Published Ve sion
The hi ing o olde wo ke s: e idence om Ge many
Empi ical Economics
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: Busch, Fabian; Fenge, Robe ; Ochsen, Ca s en (2024) : The hi ing o olde
wo ke s: e idence om Ge many, Empi ical Economics, ISSN 1435-8921, Sp inge , Be lin,
Heidelbe g, Vol. 68, Iss. 1, pp. 139-163,
h ps://doi.o g/10.1007/s00181-024-02637-5
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/317033
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/
Empi ical Economics (2025) 68:139–163
h ps://doi.o g/10.1007/s00181-024-02637-5
The hi ing o olde wo ke s: e idence om Ge many
Fabian Busch1·Robe Fenge1·Ca s en Ochsen2
Recei ed: 27 Ap il 2023 / Accep ed: 10 June 2024 / Published online: 23 Augus 2024
© The Au ho (s) 2024
Abs ac
This a icle analyses how hi ing olde wo ke s adjus s o demog aphic change in he
labou o ce by using in o ma ion om mo e han 500,000 i ms in Ge many. We ind
obus e idence ha i ms aced wi h an ageing labou ma ke hi e ela i ely mo e
olde wo ke s. Howe e , he pace o his adjus men is ela i ely slow, pa icula ly
when ageing happens ou side he i m. The endency o employ olde people is mo e
conside able in Eas Ge many, whe e he demog aphic change mo es o wa d as e .
Fu he mo e, pa - ime wo king models suppo hi ing olde wo ke s, bu his e ec
becomes less impo an in la ge i ms and Eas Ge many. Finally, while pa ial e i e-
men egula ions enhance lexibili y wi hin he i m, hey, un o una ely, diminish he
employmen oppo uni ies o olde ex e nal job seeke s.
Keywo ds Ageing labou o ce ·Hi ing o olde wo ke s ·Panel da a models
JEL Classi ica ion J11 ·J23 ·C33
1 In oduc ion
The sha e o olde wo ke s in he Ge man labou o ce has inc eased con inuously
in he las wo decades. Compa ed o younge wo ke s, olde wo ke s ypically ha e
lowe unemploymen a es, lowe job sepa a ion a es and lowe job- inding a es (see
Ochsen (2023), o example). In addi ion, on a e age, olde wo ke s ha e longe job
enu e, which means hey o en g ow olde wi hin he i m. The e o e, he i ms’ sha e
o olde wo ke s inc eases when his g oup is la ge (baby boome s). Howe e , less is
known abou he beha iou o i ms in hi ing olde wo ke s in ageing socie ies.
BCa s en Ochsen
[email p o ec ed]
1Depa men o Economics, Uni e si y o Ros ock, Ros ock, Ge many
2Depa men o Labou Economics, Uni e si y o Applied Labou S udies, Schwe in, Ge many
123
140 F. Busch e al.
The li e a u e p o ides empi ical e idence ha a ising sha e o olde pe sonnel
is posi i ely ela ed o he a e age age a hi e o he sha e o hi ed olde wo ke s.1
Conce ning Ge many, Heywood e al. (2010) analyse managemen a i udes owa ds
hi ing olde wo ke s in gene al. Howe e , hey do no measu e he hi ing beha iou
di ec ly. The li e a u e discussed uses c oss-sec ional da a and compa a i ely small
samples. Hence, hey canno cap u e de elopmen s o e ime and canno ep esen
changes in he age s uc u e o he labou o ce and hei ela ionship o he hi ing o
olde wo ke s.
The li e a u e ela ed o he hi ing o olde unemployed is somewha di e en .
Axel ad e al. (2018) conclude ha he p obabili y o an unemployed indi idual inding
a new job dec eases wi h age. Age disc imina ion agains olde unemployed wo ke s
in he hi ing p ocess is epo ed by Neuma k e al. (2019) o he USA and Oesch
(2020) o Swi ze land. In he Ge man con ex , Die z and Walwei (2011) and Ochsen
(2023) calcula e he job- inding a es and conclude ha hese a es dec ease s eeply
wi h age. W igh (2015) concludes ha Ge man es ablishmen s a e s ill eluc an o
hi e olde wo ke s, which os e s long- e m unemploymen .2Consequen ly, i is a
subs an ial p oblem o olde wo ke s o become e-employed a e a job sepa a ion,
especially i hey seek ull- ime employmen (Adams and Heywood 2007 and Daniel
and Heywood 2007).
None o he abo e s udies analyses he join implica ions o he inc easing sha es o
olde employed and unemployed in he labou o ce o he hi ing beha iou o i ms.
Fo his eason, ou esea ch ques ion is how labou ma ke ageing cha ac e is ics a e
ela ed o hi ing olde wo ke s. In pa icula , we a e in e es ed in he ole o he sha es
o ull- ime and pa - ime olde employed, he sha e o pa ial e i emen and he sha e
o olde unemployed in he i ms’ local a ea.
We use he sha e o olde wo ke s (50–64 yea s old) hi ed on all hi es as he a iable
o p ima y in e es . Figu e 1 e eals he e olu ion o his a iable in Ge many om
2000 o 2019. Fi ms hi e inc easingly ela i ely mo e olde wo ke s in he conside ed
pe iod. Howe e , how is his ela ed o he ageing o he labou o ce?
This a icle con ibu es o he li e a u e in wo ways. Fi s ly, we analyse panel da a
using he Es ablishmen His o y Panel and ma ch his panel wi h da a on unemploy-
men a he adminis a i e dis ic le el (Landk eise und k eis eie S äd e). This enables
us o analyse he i m’s hi ing beha iou o e a pe iod o 20 yea s, which con as s
exis ing s udies ha mainly analyse c oss-sec ional da a. Secondly, o conside bo h
he employed and he unemployed pa o he labou o ce, we use a iables a di e en
agg ega ion le els o examine he hi ing beha iou o i ms. We use he age s uc u e
o he employees a he i m le el ( o con ol o he e ec o an ageing wo k o ce)
and he age-speci ic unemploymen sha es a he le el o he adminis a i e dis ic
1Conce ning he a e age age a hi e, see, o example, Adams and Heywood (2007) o Aus alia and
Medei os Ga cia e al. (2017) o Po ugal. Rela ed o he sha e o hi ed olde wo ke s, see, o example,
Sco e al.(1995) and Adle and Hilbe (2009) o he USA, Heywood e al. (1999) o Hong Kong, Daniel
and Heywood (2007) and Kidd e al. (2012) o he UK.
2A close look a he long- e m unemploymen s a is ics e eals ha olde wo ke s a e s a is ically mo e
p one o be long- e m unemployed han all o he age g oups, which ma ks a c i ical indica o o he chances
o olde people being hi ed om he labou ma ke . Acco ding o da a o 2019, 42 pe cen o all olde (55
o 65 yea s old) unemployed we e long- e m unemployed. Mo eo e , wi h almos one- hi d o all long- e m
unemployed, his age g oup is he la ges among he long- e m unemployed.
123
The hi ing o olde wo ke s: e idence om Ge many 141
Fig. 1 Sha e o hi ed olde wo ke s aged 50–64
whe e he i m is loca ed ( o cap u e he impac o an ageing pool o unemployed). By
conside ing bo h ypes o hi ed olde wo ke s, we can compa e he hi ing beha iou
wi h espec o olde employed and olde unemployed pe sons.
We ind ha he join e ec o an ageing labou o ce on hi ing olde wo ke s is
posi i e bu inelas ic. A highe sha e o olde unemployed inc eases he i ms’ hi ing
o olde wo ke s, bu he elas ici y a ies by a ound 0.07. The in e nal age s uc u e
a he i m le el is mo e impo an because he elas ici y o ull- ime wo ke s is 0.33,
and he elas ici y o pa - ime wo ke s is 0.54, on a e age. Hence, i ms’ adjus men
o he demog aphic end is ela i ely slow, implying i ms eac hesi an ly o he
demog aphic de elopmen . In addi ion, we show ha a ising sha e o employees in
pa ial e i emen schemes is nega i ely ela ed o he sha e o hi ed wo ke s aged 50
and olde . This esul has o be seen in con ex wi h pa icula i ies o he Ge man law
on pa ial e i emen . Al hough he law in ends o alle ia e he employmen si ua ion
o olde wo ke s, we show ha he egula ion impedes hi ing olde wo ke s. Las ly,
we p esen he esul s di e en ia ed by i m size, sec o , and Eas and Wes Ge many.
This pape is o ganised as ollows. The ollowing sec ion p o ides some s ylised
ac s. Sec ion 3desc ibes ou da abase and he empi ical s a egy. Sec ion 4en ails
he discussion o he esul s, and a obus ness check is p o ided in Sec . 5. Sec ion 6
concludes.
2 S ylised ac s
Using da a o he US labou ma ke and he pe iod 1983–1995, Hi sch e al. (2000)
ind ha he sha e o hi ed olde wo ke s (50 yea s and olde ) on all hi es is abou
0.1. In addi ion, hey calcula e he pe cen age o olde wo ke s in he wo k o ce and
ecei e an a e age alue o 0.19. Following Hu chens (1986), hey inally calcula e he
openness o olde hi es ( he p opo ion o hi ed olde wo ke s di ided by he p opo ion
o olde wo ke s in he wo k o ce) and ecei e 0.53. A alue be ween 0 and 1 can be
123
142 F. Busch e al.
Table 1 Di e en labou ma ke measu es
Olde wo ke s Res o he wo ke s
(1)Hi ing sha e 50–64 Hold /H0.139 Hi ing sha e 15–49 H es /H0.861
(2)Sha e wo ke s 50–64 Lold /L0.238 Sha e wo ke s 15–49 L es /L0.762
(3)Hi ing a e H/L0.263 Hi ing a e H/L0.263
(4)Hi ing a e 50–64 Hold /Lold 0.153 Hi ing a e 15–49 H es /L es 0.297
(5)Rela i e hi ing a io (4)/(3)0.583 Rela i e hi ing a io (4)/(3)1.130
(6)Openness (1)/(2)0.583 Openness (1)/(2)1.130
sou ce: EHP-7520 and own calcula ions
in e p e ed as in o ma ion ha hi ing oppo uni ies a e es ic ed o olde wo ke s.
Rea anging his measu e addi ionally allows us o in e p e i as he ela i e hi ing a e
o olde wo ke s ( he p opo ion o hi ed olde wo ke s in all olde wo ke s di ided
by he o e all hi ing a e). F om his, i ollows ha he a e age hi ing a e is abou
wice as la ge as he hi ing a e o olde wo ke s.
We conside he Es ablishmen His o y Panel o Ge many o calcula e he mac oe-
conomic indica o s discussed (Table 1).3Basically, ou measu es a e necessa y o
examine he gene al labou ma ke si ua ion o olde wo ke s in his con ex . Mea-
su e (1), he sha e o hi es aged 50–64 o all hi es (Hold /H), p o ides in o ma ion
on he gene al hi ing beha iou o i ms ela ed o olde wo ke s. Measu e (2), he
sha e o wo ke s in he wo k o ce who a e 50–64 yea s old (Lold /L), o e s he same
idea as he i s measu e, bu o he wo k o ce. Measu e (3), he a io o hi es o he
wo k o ce ( he hi ing a e), indica es how open i ms a e o job seeke s. Measu e (4),
a simila p opo ion, he a io o hi es o olde wo ke s o wo k o ce sha e o olde
wo ke s (Hold /Lold ), indica es he de ia ion o olde wo ke s om he mean.4
The ela i e hi ing a io (5) shows ha he conside ed age g oup is less likely o be
hi ed han he a e age i he numbe is below 1. Openness (6) measu es he openness
o olde hi es ela i e o he p opo ion o olde wo ke s in he i m. As men ioned
abo e, bo h measu es p o ide he same quo ien . The esul s o (1), (2) and (6) a e
simila o he indings o Hi sch e al. (2000). Hence, on a e age, in he las wo
decades, Ge man i ms seem no o be mo e open o olde wo ke s han US i ms in
he 1980s and 1990s. Fu he mo e, ela i e labou ma ke oppo uni ies seem no o
ha e imp o ed o olde wo ke s, e en in ageing imes.
Conce ning measu es (5) and (6), we also can conclude ha when he sha e o he
olde wo k o ce inc eases only due o ageing wi hin he i m, he ela i e hi ing a io
o olde wo ke s mus decline when he a io o hi es aged 50–64 o all hi es emains
unchanged. In con as , when he a io o hi es aged 50–64 o all hi es inc eases,
he ela i e hi ing a io o olde wo ke s mus inc ease when he sha e o he olde
wo k o ce emains cons an . In he i s case, he ela i e hi ing a io o olde wo ke s
declines due o he ageing o he wo k o ce, while in he second case, he ela i e
hi ing a io o olde wo ke s inc eases due o a highe sha e o hi ed olde wo ke s.
3Fo de ailed da a desc ip ion, see Sec . 3.
4Fo a discussion o he i s and second measu e, see, o example, Hi sch e al. (2000).
123
The hi ing o olde wo ke s: e idence om Ge many 143
When bo h causes o ageing happen (wha we obse e), he ela i e hi ing a io o
olde wo ke s may inc ease, dec ease o emain unchanged.
On he igh -hand side o Table 1, we epo he esul s o he age g oup 15 o 49
( es o he wo ke s). The measu es (5) and (6) o he es o he wo ke s a e almos
wice as la ge. This is ela ed o an abo e-a e age hi ing a e o his age g oup (4) and
a hi ing sha e (1) ha is la ge han he sha e o wo ke s o his age g oup (2).
We a gue in he in oduc ion ha we p e e o analyse he sha e o hi ed olde
wo ke s (1) o unde s and he hi ing beha iou o i ms. In p inciple, he hi ing a e o
olde wo ke s (4) seems o be an app op ia e al e na i e. We apply a simple app oach
o mo i a e ha (1) is mo e sui able as he dependen a iable han (4). Gi en ha we
ha e only wo age g oups (old and es ), he e a e di e en ways o w i e down he
hi ing a e:
H
L=H es
L+Hold
L=L es
L
H es
L es
+Lold
L
Hold
Lold
Di iding he inal exp ession by H/Lyields he componen s o he hi ing a e ha
sum up o 1.
L es
L
H es
L es
L
H+Lold
L
Hold
Lold
L
H=H es
H+Hold
H
F om his, i ollows ha he hi ing sha e o olde wo ke s (1) equals he weigh ed
ela i e hi ing a io o olde wo ke s. We decided o use (1) as ou dependen since i
is no di ec ly ela ed o he i m’s wo k o ce ageing.
Figu e 2p o ides he de elopmen o he a iables discussed. The i s alue o
each se ies is scaled o one o ease compa abili y. On he one hand, bo h hi ing a es
ollow a simila end; consequen ly, he a io o bo h is almos unchanged. Hence, he
o e all ela i e labou ma ke oppo uni ies o olde wo ke s do no inc ease du ing
he pe iod conside ed. On he o he hand, he sha e o hi ed olde wo ke s and he
pe cen age o olde in he wo k o ce ha e a simila posi i e pa e n. This means ha
i ms hi e a la ge sha e o olde wo ke s, and ageing wi hin he i m inc eases he
sha e o he olde wo k o ce a he same pace. The da a analyses below examine he
ela ionship be ween he sha e o hi ed olde wo ke s and he sha e o olde employed
a he i m le el. In addi ion, we calcula e he ela i e hi ing a io o each conside ed
subse o da a.
3 Da a and me hod
This s udy uses he weakly anonymous Es ablishmen His o y Panel (EHP) o Ge -
many.5The EHP is a 50 pe cen sample o all es ablishmen s h oughou Ge many
5Da a access was p o ided ia on-si e use a he Resea ch Da a Cen e (FDZ) o he Ge man Fed-
e al Employmen Agency (BA) a he Ins i u e o Employmen Resea ch (IAB) and emo e da a access
(Schmucke e al. 2016).
123
144 F. Busch e al.
Fig. 2 De elopmen o di e en labou ma ke measu es o olde Wo ke s
wi h a leas one employee subjec o social secu i y a he e e ence da e o June 30
o a gi en yea . Hence, he panel s uc u e is yea ly i m-le el da a.
Ou da abase consis s o 9,595,987 obse a ions o 1,893,345 i ms dis ibu ed
ac oss all 401 adminis a i e dis ic s in Ge many om 2000 o 2019.6Ou dependen
a iable is he a io o hi ed olde wo ke s aged 50–64 o all hi es a he i m le el. In
he baseline speci ica ion, we emo e all es ablishmen s wi h less han i e employees
om he panel. Since ou dependen a iable is a ac ion, small i ms a e mo e likely o
hi e ewe people. I a i m hi es exac ly one pe son, he sha e o olde wo ke s en e ing
he i m in ha espec i e yea is ei he 0 o 1. Fu he mo e, we only conside i ms
in he panel ha a e included o a leas i e yea s. This is due o ou gene al in e es
in wi hin- i m changes and he ela ionship be ween he dependen and independen
a iables. Hence, o he baseline model, we use 5,879,369 obse a ions o 530,658
i ms.7
To cap u e he age dis ibu ion in he wo k o ce, we conside he ollowing a iables
as con ols in ou eg essions. We include he you h sha e (15 o 24 yea s) and he
sha e o olde wo ke s (50 o 64 yea s) and conside he emaining age g oup (25
o 49 yea s old) as he e e ence. We di e en ia e be ween ull- ime and pa - ime
employees o accoun o lexible wo king ime models. In addi ion, we include he
sha e o employees wo king in pa ial e i emen schemes. To p oxy he age s uc u e
o he labou o ce u he , we conside he you h sha e and he sha e o olde wo ke s
among he unemployed a he adminis a i e dis ic le el. The Fede al Employmen
Agency p o ides hese da a. Table 2 epo s summa y s a is ics o he a iables o
p ima y in e es . Fo a da a desc ip ion including con ol a iables, see Appendix A.
6We choose he yea 2000 as a s a ing poin o easons o da a a ailabili y.
7We acknowledge ha ou es ic ions educe he sample size conside ably. Fo compa ison easons, we
also es ima e he model on he un es ic ed panel. The esul s a e p o ided in Table 10 in Appendix C.
123
The hi ing o olde wo ke s: e idence om Ge many 145
Table 2 Summa y s a is ics
Model Obs Mean S d. e . Min Max
Fi ms
Hi es 50–64 5,879,369 13.89 23.19 0 100
Full- ime 50–64 5,879,369 22.00 23.09 0 100
Full- ime 15–24 5,879,369 9.60 16.39 0 100
Pa - ime 50–64 5,879,369 25.66 25.42 0 100
Pa - ime 15–24 5,879,369 23.86 27.19 0 100
Pa ial e i emen 5,879,369 0.43 2.14 0 100
Adminis a i e dis ic
Sha e unemploymen 50–64 5,879,369 30.23 5.50 17.70 58.33
Sha e unemploymen 15–24 5,879,369 10.47 2.27 2.70 21.09
We analyse he impac o demog aphic change on he sha e o hi ed olde wo ke s
by using a panel da a model wi h ixed i m-le el e ec s and yea e ec s.8The model
is speci ied as ollows:
yij =β0+
k
βkFkij +
m
λmXmj +αi+δ +ij
Fis a ec o comp ising k=1, ..., K a iables a he i m le el, and Xdesc ibes a
ec o o m=1, ..., M a iables a he adminis a i e dis ic le el. Index i ep esen s
he i m le el, j he adminis a i e dis ic le el and he annual ime dimension.
Finally, depic s he esiduals, α ixed e ec s and δ ime e ec s.
Conce ning he s a is ical ele ance o ou es ima es, we p o ide clus e - obus
s anda d e o s a he dis ic le el o con ol o he e oskedas ici y, se ial co ela ion
and c oss-sec ional dependence in he esiduals.9In addi ion, we conside he False
Posi i e Risk (FPR) o p o ide in o ma ion on he s a is ical e idence o he es ima ed
coe icien s. In con as o he p- alue, he FPR measu es he p obabili y o he null
hypo hesis being ue (Colquhoun 2019 and Colquhoun 2017).10 Fo a discussion o
he misin e p e a ion o p- alue, see, o example, Wasse s ein and Laza (2016). Fo
he compu a ion o he FPR, we e e o Appendix B.
4 Resul s
Fi s , we p o ide he esul s o ou baseline model and Ge many o e all. In he
ollowing subsec ion, we di e en ia e by i m size, sec o and Eas and Wes Ge many.
8We used a Hausman es o decide whe he a andom e ec s model was mo e app op ia e bu ound s ong
e idence o using ixed e ec s.
9We also es clus e - obus s anda d e o s a he i m le el. Howe e , o cap u e he co ela ion pa e n a
he dis ic le el, we decide in a ou o his le el.
10 Fo example, an FPR o 0.05 co esponds o a p- alue o 0.003, and an FPR o 0.01 co esponds o a
p- alue o 0.0005.
123
146 F. Busch e al.
Table 3 Hi ing olde wo ke s:
baseline esul s Dependen a iable: sha e hi es 50–64 Baseline
Fi m le el
Full- ime 50–64 0.2084‡
(0.0024)
Pa - ime 50–64 0.2945‡
(0.0023)
Full- ime 15–24 −0.0394‡
(0.0008)
Pa - ime 15–24 −0.0268‡
(0.0006)
Pa ial e i emen −0.2454‡
(0.0127)
Adminis a i e dis ic le el
Unemployed 50–64 0.0314‡
(0.0056)
Unemployed 15–24 0.0371
(0.0115)
Obse a ions 5,879,369
Fi ms 530,658
R20.1158
Fixed e ec s eg ession. Fo u he con ols, see he da a desc ip ion.
Clus e - obus s anda d e o s a he dis ic le el a e in pa en heses.
†: FPR ≤0.05, ‡: FPR ≤0.01
4.1 Baseline model
Table 3p esen s ou baseline esul s. We ound a posi i e co ela ion be ween he sha e
o olde ull- ime wo ke s ( ull- ime 50–64) and he sha e o hi ed olde wo ke s. A
4.8 pe cen age poin inc ease in he sha e o olde employees co esponds o a one
pe cen age poin inc ease in he sha e o hi ed olde wo ke s. The co esponding
elas ici y o abou 0.33 sugges s a ela i ely slow adjus men o demog aphic change.
This could be due o i ms’ expe iences wi h olde employees and adjus men s in
p oduc ion p ocesses and se ices.
An inc ease in he sha e o olde pa - ime wo ke s (pa - ime 50 o 64) is also pos-
i i ely ela ed o hi ing olde wo ke s. The co esponding elas ici y is 0.54, indica ing
a sligh ly mo e elas ic ela ionship. This migh highligh he pa icula i ies o Ge -
man law. Usually, pa - ime employees enjoy he same igh s as ull- ime employees
ega ding aca ion ime o dismissal p o ec ion. Howe e , pa ag aph 14, Sec . 3o
he Teilzei - und Be is ungsgese z (TzB G) allows he employe o causelessly limi
he employmen con ac o a maximum o i e yea s i he employee has been unem-
ployed be o e e-employmen o a leas ou mon hs and is a leas 52 yea s old. This
in o ma ion is impo an because i g an s he employe mo e lexibili y. Employe s
migh speci ically a ge hi ing olde wo ke s o exploi his op ion since hey do no
123
The hi ing o olde wo ke s: e idence om Ge many 153
Table 7 Elas ici ies and ela i e hi ing a io
Model Elas ici ies
Full- ime
50–64
Pa - ime
50–64
Pa ial
e i emen
Unemployed
50–64
Rela i e
hi ing a io
Baseline 0.33 0.54 −0.01 0.07 0.58
Fi ms
Small i ms 0.28 0.74 −0.003 0.06 0.68
Medium i ms 0.33 0.47 −0.01 0.09 0.58
La ge i ms 0.45 0.29 0.001 0.08 0.51
Sec o s
P ima y sec o 0.56 0.40 −0.02 −0.06 0.59
Seconda y sec o 0.42 0.43 −0.01 0.11 0.62
Te ia y sec o 0.28 0.64 −0.01 0.01 0.56
Regions
Eas Ge many 0.42 0.40 −0.004 0.13 0.64
Wes Ge many 0.31 0.59 −0.01 0.05 0.57
Cu si e alues a e based upon s a is ically weak e idence
he indings ha i ms in he Eas a e mo e open o olde wo ke s bu less o wo king
ime lexibili y. In addi ion, i ms in bo h egions seem mo e eluc an when ageing
happens ou side he i m, bu he eas e n pa adjus s mo e o he ad anced ageing
p ocess. Finally, we also ind he ad e se e ec s o pa ial e i emen o bo h egions.
Howe e , he economic impac is minimal.
5 Robus ness
In his sec ion, we u he analyse how obus ou esul s a e. Ou speci ica ion po en-
ially su e s om a simul anei y bias. Recognising ha he sha e o hi ed olde wo ke s
( he low) and he sha e o employed olde wo ke s ( he s ock) migh in luence each
o he , we p o ide wo app oaches o analyse how impo an his po en ial p oblem is
o ou es ima es. In addi ion, ou speci ica ion may su e om a po en ial omi ed
a iable bias. The e o e, in he nex sec ion, we a y he se o con ol a iables o see
how ou co e a iables beha e. Subsequen ly, we p o ide ins umen al a iable (IV)
es ima es o de ec a po en ial bias in he pa ame e s o he olde wo ke sha es.
5.1 Speci ica ion
To see how obus he coe icien s o he ull- ime 50–64 and pa - ime 50–64 a i-
ables a e, we ake he baseline es ima es as a e e ence and exclude di e en se s
o con ol a iables. Table 8p o ides an o e iew. O e all, we an nine eg essions.
While he i e co e a iables emain in each speci ica ion, he es a e excluded in
a ious combina ions. This is shown in he lowe pa o he able. Fo compa ison
123
154 F. Busch e al.
Table 8 Hi ing olde wo ke s: speci ica ion es s
Dependen a iable: sha e hi es 50–64 (1) (2) (3) (4) (5) (6) (7) (8) (9)
Fi m le el
Full- ime 50–64 0.2083‡0.2083‡0.2077‡0.2077‡0.2068‡0.2085‡0.2064‡0.2084‡0.1833‡
(0.0024) (0.0024) (0.0024) (0.0024) (0.0024) (0.0024) (0.0024) (0.0024) (0.0026)
Pa - ime 50–64 0.2965‡0.2965‡0.2961‡0.2961‡0.2938‡0.2945‡0.2934‡0.2945‡0.2720‡
(0.0022) (0.0022) (0.0022) (0.0022) (0.0022) (0.0023) (0.0022) (0.0023) (0.0019)
Full- ime 15–24 −0.0432‡−0.0432‡−0.0433‡−0.0433‡−0.0389‡−0.0393‡−0.0390‡−0.0394‡−0.0466‡
(0.0008) (0.0008) (0.0008) (0.0008) (0.0008) (0.0008) (0.0008) (0.0008) (0.0007)
Pa - ime 15–24 −0.0391‡−0.0391‡−0.0389‡−0.0389‡−0.0265‡−0.0268‡−0.0266‡−0.0268‡−0.0398‡
(0.0007) (0.0007) (0.0006) (0.0006) (0.0006) (0.0006) (0.0006) (0.0006) (0.0005)
Pa ial e i emen −0.2641‡−0.2641‡−0.2644‡−0.2645‡−0.2489‡−0.2449‡−0.2449‡−0.2454‡−0.3888‡
(0.0130) (0.0130) (0.0129) (0.0129) (0.0128) (0.0127) (0.0127) (0.0127) (0.0140)
Addi ional i m a iables
Adminis a i e dis ic a iables
Sec o dummies
Yea dummies
Dis ic -yea dummies
R20.1145 0.1145 0.1146 0.1146 0.1157 0.1158 0.1157 0.1158 0.1813
Fixed e ec s eg ession. Numbe o obse a ions 5,879,369. Numbe o Fi ms 530,658. Fo u he con ols, see he da a desc ip ion. Clus e - obus s anda d e o s a he
dis ic le el a e in pa en heses. †: FPR ≤0.05, ‡: FPR ≤0.01
123
The hi ing o olde wo ke s: e idence om Ge many 155
easons, speci ica ion (8) is he baseline speci ica ion al eady shown in Table 3. Ac oss
eg essions (1) o (8), he ull- ime 50–64 coe icien a ies be ween 0.2085 and 0.2064
and he pa - ime 50–64 coe icien is be ween 0.2965 and 0.2934. Bo h di e ences
a e minimal and indica e a obus s a is ical ela ion wi h he dependen a iable. In
eg ession (9), we add mo e han eigh housand dummies a he egional le el. Again,
he coe icien s o in e es only change sligh ly. F om his, we conclude ha ou es i-
ma es and calcula ed elas ici ies in he o me sec ion a e s a is ically eliable. The
emaining h ee a iables also a y only a li le.
5.2 IV es ima ion
A mo e o mal way o handle a po en ial simul anei y bias is o apply IV es ima es.
We ins umen he sha e o olde ull- ime and pa - ime employees o con ol o a
possible simul anei y bias using he ollowing wo ins umen s: Fi s , he one-yea
lagged labou o ce sha es o olde (50–64 yea s) people in he adminis a i e dis ic
(coun y).13 This a iable co e s he comple e egional labou supply o he old in he
o me yea . This ins umen is independen o he i m’s hi ing decisions in he cu en
yea , no only because o he ime lag bu also because his is agg ega ed in o ma ion
unknown o he indi idual i m.
As a second ins umen , we use he a e age employmen sha e o olde wo ke s in
o he i ms in he same coun y. While he i s ins umen co e s he comple e egional
labou o ce, his second ins umen co e s he a e age sha e o olde wo ke s in
neighbou hood i ms. This a iable is aken om he same sample da a as he po en ial
endogenous a iable. This ins umen is independen o he i m’s hi ing decisions
because his in o ma ion is unknown o he indi idual i m. Bo h ins umen s a e
co ela ed wi h he sha e o olde wo ke s in a speci ic i m bu a e no causally
ela ed o he dependen a iable. Howe e , po en ially, ou ins umen al a iables
could be co ela ed wi h o he no -conside ed coun y-le el a iables ha , in u n,
migh in luence he hi ing beha iou o i ms. Consequen ly, he ins umen s would
no be exogenous, and he esul s would be biased. In addi ion, we mus keep in
mind ha he sha e o olde unemployed is la ge han ha o olde wo ke s o olde
wo ke s hi ed. Hence, he e is no e idence o highe ma ke igh ness in his age g oup
compa ed o o he s.
We p o ide h ee IV eg essions (Table 9). In he i s , we ins umen he a iable
ull- ime 50–64; in he second, we ins umen he a iable pa - ime 50–64. In he hi d
eg ession, we ins umen bo h a iables. In he able, we epo he coe icien s o he
ins umen s om he i s s age eg ession and below he second s age coe icien o
he ins umen ed a iable. Taking all h ee esul s oge he , we do no see essen ial
di e ences o he baseline e e ence in Table 3. The addi ional es s a is ics in he lowe
pa o he able gi e us u he in o ma ion on he eg ession quali y. The Kleibe gen–
Paap LM es gi es in o ma ion on unde iden i ica ion, while he Wald es is employed
because o po en ially weak ins umen s. The null hypo hesis o unde iden i ica ion is
ejec ed, and he e is also no e idence o weak ins umen s. The Hansen s a is ic es s
13 We ob ain his ins umen om da a om he Fede al Ins i u e o Resea ch on Building, U ban A ai s
and Spa ial De elopmen .
123
156 F. Busch e al.
Table 9 Hi ing olde wo ke s: baseline IV es ima es
Dependen a iable: sha e hi es 50–64 (1) (2) (3)
Fi s s age
Ins umen ed a iable Full- ime 50–64 Pa - ime 50–64 Full- ime 50–64 Pa - ime 50–64
Sha e o he old in he dis ic labou o ce −10.2619‡0.3163‡0.2655‡0.5389‡
(0.0257) (0.0300) (0.0256) (0.0395)
Sha e o he old in he neighbou i ms 0.3060‡0.2277‡0.3049‡0.1620‡
(0.0233) (0.0211) (0.0233) (0.0298)
Second s age
Full- ime 50–64 0.2710‡0.2424‡
(0.0276) (0.0340)
Pa - ime 50–64 0.3517‡0.3312‡
(0.0278) (0.0362)
Kleibe gen–Paap LM 91.1‡59.3‡73.0‡
Kleibe gen–Paap Wald 261.4‡125.0‡63.7‡
Hansen J-s a is ic χ21.0 1.3 –
Tes o endogenei y χ25.1 4.5 6.2
Obse a ions 5,879,369 5,879,369 5,879,369
Fi ms 530,658 530,658 530,658
R20.1137 0.1135 0.1142
IV = Ins umen al a iable eg ession. Fo u he con ols, see he da a desc ip ion. Clus e - obus s anda d e o s a he dis ic le el a e in pa en heses. †: FPR ≤0.05, ‡:
FPR ≤0.01
123
The hi ing o olde wo ke s: e idence om Ge many 157
he o e iden i ica ion o all ins umen s and s a es ha he ins umen s a e cohe en
wi h each o he in bo h cases. Finally, he endogenei y es p o ides no e idence ha
he po en ial endogenous eg esso s a e, in ac , endogenous. O e all, we conclude
ha a simul anei y bias is small o nonexis en . The esul s in his sec ion suppo he
obus ness o he esul s p o ided in Sec . 4.
6 Conclusions
This a icle analyses how he labou o ce’s in e nal and ex e nal age s uc u e is
ela ed o hi ing olde wo ke s. Unlike p e ious s udies, we combine panel da a om
he Es ablishmen His o y Panel and egional unemploymen da a. Encou agingly,
we ind a posi i e ela ionship be ween hi ing olde wo ke s and ageing among bo h
he employed and unemployed. This sugges s a g adual adjus men o demog aphic
shi s. Howe e , he pace o his adjus men is disappoin ingly slow. Ou measu e o
he ela i e hi ing a io o olde wo ke s u he subs an ia es ou indings. We ind
ad e se labou ma ke oppo uni ies o di e en i m sizes, sec o s, and Eas and Wes
Ge many. Consequen ly, ou measu e shows ha he labou ma ke is less open o
olde wo ke s. The calcula ed elas ici ies u he suppo hese indings. Mo eo e , we
ind ha he speed o adjus men is e en slowe when ageing occu s ou side he i m.
This coincides wi h low job- inding a es o olde wo ke s in Ge many. Su p isingly,
a ising sha e o employees in pa ial e i emen schemes is nega i ely ela ed o he
sha e o hi ed olde wo ke s. Thus, while his policy enhances lexibili y o in e -
nal employees, i un o una ely diminishes he employmen oppo uni ies o olde
ex e nal job seeke s.
Ou esul s should be iewed in he con ex o he ongoing deba e in Ge many
abou he e i emen age, pa icula ly conside ing he policy e o m 2008, which g ad-
ually inc eased he manda o y e i emen age om 65 o 67 yea s, s a ing in 2012.
Fi ms do espond o demog aphically induced changes in he labou o ce composi-
ion by hi ing mo e olde wo ke s. Howe e , he p opo ion o olde people among
he unemployed ou weighs hei sha e among he hi ed o wi hin he i m, and he
hi ing elas ici y o olde unemployed is e y small. These indings ha e impo an
implica ions o policymake s and p o essionals in he labou ma ke , highligh ing he
need o s a egies o imp o e he employmen p ospec s o olde wo ke s in Ge many.
A p omising agenda o u u e esea ch migh be he conside a ion o gende - ela ed
ques ions. Speci ically, he hi ing beha iou owa ds women in gene al, pa icula ly
olde women, migh yield in e es ing esul s since emale labou is seen as a po en ial
esou ce o labou supply.
123
158 F. Busch e al.
Appendix
Appendix A: Da a desc ip ion
Fi m le el a iables
The dependen a iable is he a io o hi es aged 50–64 o all hi es a he i m le el.
Conce ning he explana o y a iables a he i m le el, we con ol o he exis ing
age s uc u e o ull- ime and pa - ime employees. Fo bo h g oups o employees, We
include he you h sha e (15 o 24 yea s) and he sha e o olde wo ke s (50–64 yea s)
and conside he age g oup 25 o 49 yea s old as he e e ence g oup. We also include
he i m size, measu ed by he numbe o employees.
Fu he mo e, we conside he sha e o pa - ime wo ke s in each i m. In addi ion o
he age s uc u e, we conside he sha e o employees wi h an academic backg ound
as a con ol. We also add he sha e o low-skilled employees and ake he sha e o
medium-skilled employees as a e e ence. Fu he mo e, we conside he sha e o
employees unde going oca ional aining. We also include he sha e o ma ginal
pa - ime wo ke s whose mon hly ea nings a e a mos 450 EUR.
In addi ion, we include he a io o olde wo ke s who le he i m o all sepa a ions
and he sha e o job- o-job mo e s. The la e a iable is he sha e o hi ed wo ke s
wi h an employmen ela ionship wi h ano he i m in he p e ious yea .
Finally, we con ol o indus y-speci ic di e ences in he hi ing beha iou o i ms
by including 13 sec o -speci ic dummy a iables.14
Adminis a i e dis ic le el a iables
To p oxy he size and age s uc u e o he labou o ce, we conside he sha e o he
you h and olde wo ke s among he unemployed and he o al numbe o unemployed a
he adminis a i e dis ic le el. The e e ence is he sha e o unemployed aged 25 o 49
yea s. Also, we con ol o he deg ee o u banisa ion a he adminis a i e dis ic le el.
We ob ained he da a om BBSR (see n. 5) and ollowed hei classi ica ion, which
de ines ou basic ca ego ies. The i s en ails ci ies wi h a leas 100,000 inhabi an s.
Ca ego y 2 comp ises all dis ic s whe e a leas 50 pe cen o all inhabi an s li e in
majo o medium-sized ci ies, and he popula ion densi y is a leas 150 inhabi an s pe
km2. Ca ego y 3 encloses all dis ic s sha ing he a ibu es om Ca ego y 2 bu epo s
a popula ion densi y be ween 100 and 150 inhabi an s pe km2. Finally, Ca ego y 4
comp ises all dis ic s whe e unde 50 pe cen o he popula ion li es in majo o
medium-sized ci ies, and he popula ion densi y is a mos 100 inhabi an s pe km2.
Appendix B: Compu a ion o he alse posi i e isk
The alse posi i e isk (FPR) was in oduced by Colquhoun (2019,2017) and mea-
su es he p obabili y ha he esul occu ed by chance P(H0|da a). The app oach is
14 Fo a de ailed depic ion, see he FDZ da a epo o he Sample o In eg a ed Labou Ma ke Biog aphies
Regional File (SIAB-R) 1975–2021, Table A9 on page 74.
123
The hi ing o olde wo ke s: e idence om Ge many 159
based on he Bayes heo em ha we exp ess in odds:
pos e io odds on H1=Bayes ac o p io odds
This is equal o
P(H1|da a)/P(H0|da a)=P(da a|H1)/P(da a|H0)P(H1)/P(H0)
Following Colquhoun, he Bayes ac o becomes a likelihood a io (LR), and he
p io odds can be exp essed using he p obabili y ha he e is a eal e ec ,
P(H1):P(H1)/(1−P(H1)).
Among o he s, Sellke e al. (2001) p o ide an app oach o calcula e he LR based
on he p- alue: LR =1/(−eplog(p)). Howe e , his measu e can be conside ed only
as long as p<1/e, wi h eas Eule ’s numbe . Taking hings oge he and conside ing
P(H0|da a)=1−P(H1|da a)gi es us he FPR:
FPR =(1/(1+(1/(−eplog(p))(P(H1))/(1−P(H1)))))
Applying he FPR app oach equi es o speci y P(H1) i s . Howe e , speci ying he
p io p obabili y in eg ession analysis is (e en in eplica ion s udies) di icul , and we
should always be ca e ul when de ining his unknown numbe (you ha e o con ince
he eade ). We use
P(H1)/(1−P(H1)) =0.5/(1−0.5)=1
which means ha bo h p obabili ies ha e he same weigh . This is equal o a 50:50
chance o a eal e ec speci ied be o e he da a a e analysed. This seems easonable
when we do no know wha o choose o a e open o he esul s. Hence, he p io
p obabili y o a eal e ec , P(H1),is ixed o0.5. In his case, he FPR is much
la ge han he co esponding p- alue, and o example, p=0.05 is equal o a FPR
o 0.2893.
Appendix C
123
160 F. Busch e al.
Table 10 Hi ing olde wo ke s:
un es ic ed panel Dependen a iable: sha e hi es 50–64 Baseline
i m le el
Full- ime 50–64 0.2702‡
(0.0030)
Pa - ime 50–64 0.4333‡
(0.0023)
Full- ime 15–24 −0.0446‡
(0.0006)
Pa - ime 15–24 −0.0252‡
(0.0004)
Pa ial e i emen −0.3988‡
(0.0128)
Adminis a i e dis ic le el
Unemployed 50–64 0.0139
(0.0055)
Unemployed 15–24 0.0098
(0.0116)
Obse a ions 9,595,987
Fi ms 1,893,345
R20.2003
Fixed e ec s eg ession. Fo u he con ols, see he da a desc ip ion.
Clus e - obus s anda d e o s a he dis ic le el a e in pa en heses.
†: FPR ≤0.05, ‡: FPR ≤0.01
123
The hi ing o olde wo ke s: e idence om Ge many 161
Table 11 Summa y s a is ics o subsec ions
Obs Mean S d. e Obs Mean S d. e Obs Mean S d. e
Small i ms Medium i ms La ge i ms
Hi es 50–64 2,839,260 13.78 29.44 2,905,010 14.04 22.50 750,970 14.29 14.68
Full- ime 50–64 2,839,260 14.17 26.71 2,905,010 23.26 22.07 750,970 28.24 16.08
Full- ime 15–24 2,839,260 10.97 23.65 2,905,010 9.29 14.34 750,970 7.29 8.05
Pa - ime 50–64 2,839,260 21.30 29.60 2,905,010 26.23 24.92 750,970 28.99 20.62
Pa - ime 15–24 2,839,260 21.48 31.10 2,905,010 24.45 26.93 750,970 25.75 24.63
Pa ial e i emen 2,839,260 0.09 1.41 2,905,010 0.38 2.07 750,970 1.07 2.83
Sha e unemploymen 50–64 2,839,260 30.03 5.46 2,905,010 30.33 5.52 750,970 30.14 5.48
Sha e unemploymen 15–24 2,839,260 10.52 2.27 2,905,010 10.48 2.28 750,970 10.38 2.27
P ima y sec o Seconda y sec o Te ia y sec o
Hi es 50–64 141,683 16.47 25.01 2,058,404 14.55 23.37 3,663,985 13.41 22.98
Full- ime 50–64 141,683 26.47 21.29 2,058,404 22.98 21.67 3,663,985 21.26 23.87
Full- ime 15–24 141,683 8.60 14.26 2,058,404 8.86 14.15 3,663,985 10.07 17.59
Pa - ime 50–64 141,683 30.48 30.28 2,058,404 24.38 26.54 3,663,985 26.19 24.49
Pa - ime 15–24 141,683 23.01 28.66 2,058,404 28.80 29.93 3,663,985 21.14 25.04
Pa ial e i emen 141,683 0.90 3.22 2,058,404 0.22 1.30 3,663,985 0.53 2.44
Sha e unemploymen 50–64 141,683 31.40 5.95 2,058,404 30.52 5.63 3,663,985 30.01 5.39
Sha e unemploymen 15–24 141,683 10.70 2.28 2,058,404 10.65 2.28 3,663,985 10.37 2.26
123
162 F. Busch e al.
Table 12 Summa y s a is ics o subsec ions
Obs Mean S d. e Obs Mean S d. e
Eas Ge many Wes Ge many
Hi es 50–64 1,023,460 17,09 25.69 4,855,909 13.22 22.57
Full- ime 50–64 1,023,460 25,99 23.56 4,855,909 21.15 22.90
Full- ime 15–24 1,023,460 6,98 13.01 4,855,909 10.15 16.97
Pa - ime 50–64 1,023,460 26,53 29.04 4,855,909 25.48 24.59
Pa - ime 15–24 1,023,460 23,53 30.80 4,855,909 23.93 26.36
Pa ial e i emen 1,023,460 0,63 2.99 4,855,909 0.39 1.92
Sha e unemploymen 50–64 1,023,460 31,87 6.94 4,855,909 29.88 5.07
Sha e unemploymen 15–24 1,023,460 9,80 2.26 4,855,909 10.62 2.25
Acknowledgemen s We would like o hank Axel Bö sch-Supan, Be nha d Boockmann, Ul ich Walwei,
and wo anonymous e e ees o help ul commen s.
Funding Open Access unding enabled and o ganized by P ojek DEAL.
Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License, which
pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as you gi e
app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons licence,
and indica e i changes we e made. The images o o he hi d pa y ma e ial in his a icle a e included
in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line o he ma e ial. I
ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended use is no pe mi ed
by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he
copy igh holde . To iew a copy o his licence, isi h p://c ea i ecommons.o g/licenses/by/4.0/.
Re e ences
Adams SJ, Heywood JS (2007) The age o hi ing and de e ed compensa ion: e idence om Aus alia.
Econ Rec 83(261):174–190
Adle G, Hilbe D (2009) Indus y hi ing pa e ns o olde wo ke s. Res Aging 31(1):69–88
Axel ad H, Malul M, Luski I (2018) Unemploymen among younge and olde indi iduals: does con en ional
da a abou unemploymen ell us he whole s o y?. J Labou Ma k Res 52(3)
Cai L, Law V, Ba hga e M (2014) Is pa - ime employmen a s epping s one o ull- ime employmen ? Econ
Rec 90(291):462–485
Colquhoun D (2017) The ep oducibili y o esea ch and he misin e p e a ion o P alues. R Soc Open Sci
4:171085
Colquhoun D (2019) The alse posi i e isk: a p oposal conce ning wha o do abou p- alues. Am S a
73:192–201
Daniel K, Heywood JS (2007) The de e minan s o hi ing olde wo ke s: UK e idence. Labou Econ
14:35–51
Die z M, Walwei U (2011) Ge many - no coun y o old wo ke s? J Labou Ma k Res 44(4):363–376
Heywood JS, Ji jahn U, Tse s a dze G (2010) Hi ing olde wo ke s and employing olde wo ke s: Ge mans
e idence. J Popul Econ 23(2):595–615
Heywood JS, Ho L-S, Wei X (1999) The De e minan s o Hi ing Olde Wo ke s: E idence om Hong-Kong.
ILR Re 52(3):444–459
Hi sch BT, Macphe son DA, Ha dy MA (2000) Occupa ional age s uc u e and access o olde wo ke s.
ILR Re 53(3):401–418
123
