Health insurance as economic stimulus? Evidence from long-term care jobs

Hackmann, Ma in B.; Heining, Jö g; Klimke, Roman; Polyako a, Ma ia; Seibe ,
Holge
Wo king Pape
Heal h insu ance as economic s imulus? E idence om
long- e m ca e jobs
IAB-Discussion Pape , No. 03/2025
P o ided in Coope a ion wi h:
Ins i u e o Employmen Resea ch (IAB)
Sugges ed Ci a ion: Hackmann, Ma in B.; Heining, Jö g; Klimke, Roman; Polyako a, Ma ia; Seibe ,
Holge (2025) : Heal h insu ance as economic s imulus? E idence om long- e m ca e jobs, IAB-
Discussion Pape , No. 03/2025, Ins i u ü A bei sma k - und Be u s o schung (IAB), Nü nbe g,
h ps://doi.o g/10.48720/IAB.DP.2503
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/315970
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-sa/4.0/deed.de
ISSN 2195-2663
IAB-DISCUSSION PAPER
A icles on labou ma ke issues
03|2025 Heal h Insu ance as Economic S imulus?
E idence om Long-Te m Ca e Jobs
Ma in Hackmann, Jö g Heining, Roman Klimke, Ma ia Polyako a, Holge Seibe
Heal h Insu ance as Economic S imulus?
E idence om Long-Te m Ca e Jobs
Ma in Hackmann (Uni e si y o Cali o nia, Los Angeles)
Jö g Heining (Ins i u ü A bei sma k - und Be u s o schung (IAB))
Roman Klimke (Ha a d Uni e si y)
Ma ia Polyako a (S an o d Uni e si y)
Holge Seibe (Ins i u ü A bei sma k - und Be u s o schung (IAB))
Mi de Reihe „IAB-Discussion Pape “ will das Fo schungsins i u de Bundesagen u ü
A bei den Dialog mi de ex e nen Wissenscha in ensi ie en. Du ch die asche Ve b ei ung
on Fo schungse gebnissen übe das In e ne soll noch o D ucklegung K i ik ange eg
und Quali ä gesiche we den.
The “IAB-Discussion Pape ” is published by he esea ch ins i u e o he Ge man Fede al
Employmen Agency in o de o in ensi y he dialogue wi h he scien i ic communi y. The
p omp publica ion o he la es esea ch esul s ia he in e ne in ends o s imula e
c i icism and o ensu e esea ch quali y a an ea ly s age be o e p in ing.
Con en s
1. In oduc ion............................................................................... 7
2. Ins i u ional Se ing and Da a......................................................... 13
2.1. Long-Te m Ca e Insu ance in Ge many .............................................. 13
2.2. Da a Sou ces and Sample Cons uc ion ............................................. 14
2.2.1. Sampling F ame ............................................................... 14
2.2.2. Cha ac e is ics o Wo ke s .................................................... 15
2.2.3. Mo ali y ........................................................................ 16
2.2.4. Means-Tes ed and Uni e sal LTC Bene icia y Coun s ..................... 16
2.3. Summa y S a is ics ..................................................................... 17
3. Resea ch Design ........................................................................ 18
3.1. Sou ce o Va ia ion ..................................................................... 18
3.1.1. Pa ial Equilib ium ............................................................. 20
3.2. Gene al Equilib ium .................................................................... 21
3.3. Mo ali y ................................................................................. 21
4. Resul s .................................................................................. 22
4.1. Pa ial Equilib ium...................................................................... 22
4.1.1. Indus y Expansion ............................................................ 22
4.1.2. Ana omy o Expansion ........................................................ 24
4.2. Gene al Equilib ium .................................................................... 25
4.3. Mo ali y ................................................................................. 27
5. Wel a e .................................................................................. 28
5.1. App oxima ion App oach .............................................................. 29
5.2. Model-based App oach................................................................. 32
5.2.1. Theo y ........................................................................... 33
5.2.2. Taking Theo y o Da a ........................................................ 35
5.2.3. Wel a e and Coun e ac uals ................................................. 38
6. Conclusion .............................................................................. 40
Re e ences...................................................................................... 42
Appendix .................................................................................... 64
A1. Da a Appendix .......................................................................... 65
A1.1. P ocessing o he Raw Da a .................................................. 65
A1.2. Va iable De ini ions ............................................................ 66
A1.3. Appendix Figu es and Tables ................................................ 68
A2. Model De ails ............................................................................ 90
A2.1. Equilib ium Wages and Queues.............................................. 90
IAB-Discussion Pape 03|2025 3
A2.2. Es ima ion and Model Fi ..................................................... 92
A2.3. Pa ame e Es ima es .......................................................... 96
A2.4. Coun e ac uals ...............................................................100
IAB-Discussion Pape 03|2025 4

Abs ac
We le e age decades o adminis a i e da a and quasi-expe imen al a ia ion in he
in oduc ion o uni e sal long- e m ca e (LTC) insu ance in Ge many in 1995 o examine
whe he heal h insu ance expansions can s imula e local economies. We ind ha he LTC
insu ance ollou led no only o sizeable g ow h o he a ge LTC sec o , bu also o an
agg ega e all in unemploymen and an inc ease in he labo o ce pa icipa ion.
Quan i a i ely, a 10 pe cen age poin inc ease in he sha e o insu ed LTC pa ien s led o 4
mo e nu sing home wo ke s pe 1,000 indi iduals age 65 and olde (12 pe cen inc ease).
Wages did no ise in he LTC sec o o o he sec o s o he economy. The quali y o newly
hi ed nu sing home wo ke s declined, bu his had no nega i e e ec on old-age li e
expec ancy. O e all, he insu ance expansion b ough lowe -skilled wo ke s in o new jobs
a he han ealloca ing wo ke s away om o he p oduc i e sec o s. Ou ma ginal alue o
public unds (MVPF) analysis sugges s ha he e o m paid o i sel when aking he
posi i e iscal ex e nali ies in he labo ma ke in o accoun . To unde s and which ma ke
p imi i es unde pin ou indings and o in o m he ex e nal alidi y o ou esul s, we
de elop and es ima e a gene al model o labo ma ke s wi h p oduc -ma ke subsidies in
he p esence o wedges, such as income axes. Ou model simula ions show ha he
agg ega e wel a e e ec s o insu ance expansions a e heo e ically ambiguous and depend
cen ally on he magni ude o ic ions in inpu ma ke s.
Zusammen assung
Wi nu zen adminis a i e Da en und quasi-expe imen elle Va ia ionen bei de Ein üh ung
de allgemeinen Langzei p lege e siche ung (LTC) in Deu schland im Jah 1995, um zu
un e suchen, ob die Auswei ung de Sozial e siche ung sich posi i au die lokale Wi scha
auswi ken kann. Wi s ellen es , dass die Ein üh ung de Langzei p lege e siche ung nich
nu zu einem be äch lichen Wachs um des Zielsek o s de Langzei p lege üh e, sonde n
auch zu einem Gesam ückgang de A bei slosigkei und einem Ans ieg de
E we bsbe eiligung. Quan i a i üh e ein Ans ieg des An eils de e siche en
Langzei p legepa ien en um 10 P ozen punk e zu 4 meh P legeheimmi a bei e n p o 1.000
Pe sonen im Al e on 65 Jah en und äl e (12 P ozen Ans ieg). Die Löhne s iegen wede im
P legesek o noch in ande en Wi scha szweigen. Die Quali ä de neu einges ell en
P legeheimk ä e nahm ab, was sich jedoch nich nega i au die Lebense wa ung im Al e
auswi k e. Insgesam üh e die Ve siche ungsauswei ung dazu, dass wenige quali izie e
A bei sk ä e neue A bei splä ze bekamen, ans a A bei sk ä e aus ande en p oduk i en
Sek o en zu e lage n. Unse e Analyse des G enze ags ö en liche Mi el (MVPF) leg
nahe, dass sich die Re o m un e Be ücksich igung de posi i en ex e nen iskalischen
IAB-Discussion Pape 03|2025 5
E ek e au dem A bei sma k bezahl gemach ha . Um zu e s ehen, welche
Ma k komponen en unse en E kenn nissen zug unde liegen, und um die ex e ne Validi ä
unse e E gebnisse zu e mi eln, en wickeln und schä zen wi ein allgemeines Modell on
A bei smä k en mi P oduk ma k sub en ionen bei Vo handensein on ökonomischen
Zusa zlas en, wie z. B. Einkommenss eue n. Unse e Modellsimula ionen zeigen, dass die
agg egie en Wohl ah se ek e on Ve siche ungsauswei ungen heo e isch meh deu ig
sind und zen al om Ausmaß de F ik ionen in den Inpu mä k en abhängen.
JEL
D58, H0, H51, I0, I13, I31, I38, J08, J14, J23, J64
Keywo ds
Gene al Equilib ium, Fiscal Ex e nali ies, Heal h Insu ance Expansion, Mo al Haza d,
Ma ginal Value o Public Funds (MVPF)
Acknowledgemen s
We a e g a e ul o Naoki Aizawa, Sydnee Caldwell, Ami abh Chand a, Raj Che y, Rebecca
Diamond, Amy Finkels ein, Ma hew Gen zkow, Ma hias Gnewuch, Joshua Go lieb, Daniel
Haanwinckel, Law ence Ka z, Timo hy Lay on, A ila Lindne , Ad iana Lle as-Muney, Nicole
Maes as, Ma hew No owidigdo, And eas Peichl, Yo am Shem-To , Ma k Shepa d, Isaac
So kin, and many semina pa icipan s o help ul commen s, as well as Ma issa Hindelang
and Jacob Mo is o ou s anding esea ch assis ance. Polyako a g a e ully acknowledges
g an suppo om he W.E. Upjohn Ins i u e o Employmen Resea ch and om he
Na ional Ins i u e on Aging (K01AG05984301).
IAB-Discussion Pape 03|2025 6
1 In oduc ion
A ow (1963) and Felds ein (1971, 1977) ha e a gued ha he p oli e a ion o heal h
insu ance signi ican ly con ibu es o he g ow h o he heal hca e sec o . The ex book
iew o insu ance is ha i mu es he p ice signal inducing ine icien ly high demand o
heal hca e—a phenomenon commonly e e ed o as mo al haza d. Op imal insu ance, in
u n, ades o he wel a e losses om mo al haza d is-a- is he wel a e bene i s o isk
p o ec ion. In his pape , we in es iga e whe he , in p ac ice, publicly unded heal h
insu ance p og ams a e also a o m o an indus ial policy ha channels consume s,
wo ke s, and capi al in o a subsidized indus y.
In he Uni ed S a es alone, mo e han $2 illion in go e nmen expendi u es low o
heal hca e h ough insu ance p og ams each yea , unde sco ing he impo ance o
unde s anding his spending’s b oade economic e ec s. Ye he agg ega e impac s o
heal h insu ance, as en isioned by A ow and Felds ein, ha e been di icul o cap u e
empi ically.1 Two main challenges a ise when ying o measu e he b oad economic
consequences o heal h insu ance expansions. Fi s , many changes in heal h insu ance
p og ams a e inc emen al and a e likely o p oduce only small and di use e ec s on he
heal hca e sec o (Finkels ein, 2007). Second, da a limi a ions o en hampe ou abili y o
p ecisely cap u e how heal hca e wo ke s, i ms, and capi al ealloca e in esponse o
insu ance expansions.
In his pape , we o e come some o hese challenges by exploi ing a majo social insu ance
e o m— he in oduc ion o uni e sal long- e m ca e (LTC) insu ance in Ge many in
1995—and le e aging uniquely comp ehensi e adminis a i e labo ma ke da a.
Go e nmen s commonly jus i y heal h insu ance policy in pa by i s e ec s on heal hca e
employmen (Ku/B an ley, 2021). Howe e , Baicke /Chand a (2012) and Cu le (2018)
1 The e a e a hand ul o excep ions. Finkels ein (2007) has documen ed a signi ican expansion o he U.S.
hospi al sec o ollowing he in oduc ion o Medica e in 1965. He es ima es sugges ha he inc ease in
spending was mo e han six imes la ge han wha he es ima es om he RAND Heal h Insu ance
Expe imen would ha e p edic ed. This was likely a ibu able o he high ixed cos s o in es men s in new
echnologies o capaci y, as well as spillo e e ec s. Kondo/Shigeoka (2013) s udy he 1961 expansion o
uni e sal insu ance in Japan, inding inc ease in u iliza ion, bu no e idence o an inc ease in he numbe o
physicians o nu ses. Dillende (2022) inds ha heal hca e employe s pos mo e job acancies and hi e
addi ional heal hca e wo ke s in esponse o Medicaid expansions in he U.S. Geddes/Schnell (2023)
demons a e ha he impac o insu ance expansions on e ail clinic en y in he U.S. hinges on p ice
egula ions in ou pu ma ke s. Thei analysis e eals ha Medicaid expansion unde he A o dable Ca e Ac
discou ages en y (and he expansion o ca e) when egula ed eimbu semen a es all below he
ma ke -clea ing p ice (consis en wi h e idence om G abowski/G ube (2007) in he case o U.S. LTC).
Con e sely, and in alignmen wi h ou indings, hey show ha when insu ance subsidizes p ices se by
p i a e i ms, insu ance expansions s imula e inc eased e ail clinic en y. Go lieb e al. ( o hcoming) ind
e idence o changes in he income and labo supply o physicians in esponse o he ela i ely la ge
insu ance expansion in he U.S. ollowing he A o dable Ca e Ac .
IAB-Discussion Pape 03|2025 7
cau ion ha iewing heal hca e jobs as d i e s o economic g ow h may be misguided i
such jobs do no imp o e pa ien heal h o i hey simply c owd ou employmen in mo e
p oduc i e sec o s. Thus, an essen ial ques ion— o bo h iscal accoun ing and wel a e
analysis—is whe he insu ance expansions c ea e new jobs on ne o me ely ealloca e
exis ing wo ke s om p oduc i e employmen elsewhe e in he economy (Dus mann e al.,
2021). To in o m his deba e, we examine bo h he pa ial equilib ium e ec s o LTC
insu ance on he LTC ma ke i sel , as well as he gene al equilib ium e ec s on employmen
in he economy as a whole. We ind ha he Ge man LTC insu ance p og am c ea ed jobs
and imp o ed wel a e. Ye we also show ha his s imulus e ec is a ibu able o many
ic ions ha cha ac e ized he Ge man labo ma ke in he mid-1990s and may no
gene alize o o he con ex s.
Ou pape makes h ee main con ibu ions. Fi s , we add ex ensi e new e idence o a small
bu g owing li e a u e ha has been able o ace ou he e ec s o heal h insu ance
expansions on heal hca e sec o employmen (Finkels ein, 2007; Kondo/Shigeoka, 2013;
Dillende , 2022). We do so in an impo an and unique se ing, he ollou o uni e sal
long- e m ca e insu ance in a la ge economy. Second, o he bes o ou knowledge, we
p o ide some o he i s e idence o ac o subs i u ion be ween sec o s o he economy in
esponse o a heal h insu ance expansion. Thi d, we show ha he na u e o ac o
subs i u ion and ma ke ic ions is quali a i ely and quan i a i ely impo an o he
wel a e implica ions o public spending on heal h insu ance.
Ou posi i e analysis p oceeds in wo s eps. We s a by measu ing he e ec o LTC
insu ance on i s p ima y a ge : he long- e m ca e sec o i sel . We ocus speci ically on
employmen in nu sing homes as ou key ou come measu e. Nu sing home ca e is one o
he mos labo -in ensi e pa s o he heal hca e sys em and accoun s o he majo i y o
spending in long- e m ca e. We hen ake a b oade pe spec i e, examining whe he he
new public unds lowing o he long- e m ca e sec o a ec ed he o e all economy. Ou
causal esea ch design is simila in spi i o he app oach in Finkels ein (2007) and
Kondo/Shigeoka (2013). We ake ad an age o he geog aphic a ia ion in pa ial public
assis ance o long- e m ca e se ices ha exis ed p io o he implemen a ion o he
uni e sal insu ance p og am.
The esul s o ou posi i e analysis p o ide se e al key insigh s abou he ela ionship
be ween insu ance and he supply side o ca e.
Fi s , we obse e a d ama ic inc ease in he numbe o nu sing homes and wo ke s in his
labo -in ensi e indus y ollowing LTC insu ance expansion. We es ima e ha a 10
pe cen age poin inc ease in he sha e o insu ed LTC pa ien s led o 0.06 (6 pe cen ) mo e
inpa ien LTC i ms and ou (12 pe cen ) mo e wo ke s pe 1,000 age 65 and olde
indi iduals in Ge many. Scaling his o he agg ega e le el o expansion, which o e ed
IAB-Discussion Pape 03|2025 8
1975 and 2008.13 We use his sample o p o ide a deep-di e in o he e o m’s e ec s on
employmen in LTC, which we e e o as he pa ial equilib ium analysis. We ocus on
wo ke s in nu sing homes, who accoun o he lion’s sha e o LTC employmen , and can be
bes iden i ied in ou da a.14
Ou second da a ex ac comp ises ull labo ma ke biog aphies o a 10 pe cen andom
sample o wo ke s obse ed in 1975 o 2008. We e e o he subse o hese da a ha is
es ic ed o yea s 1985 o 2004 as he “Labo Ma ke Sample (LMS)”. S a ing he analysis in
1985 ensu es a minimum en-yea look-back pe iod o iden i ying wo ke s who e-en e he
labo ma ke . Mo eo e , i mi iga es po en ial conce ns ha geog aphic a ia ion in o e all
employmen in 1970s may co ela e wi h geog aphic a ia ion in he eligibili y o wel a e
p og ams among olde adul s in he 1990s. Limi ing he sample o p e-2005 da a u he
a oids yea s wi h subs an ial changes in he unemploymen bene i sys em in oduced by
he 2005 Ha z IV e o ms (P ice, 2019). These e o ms led o signi ican gaps in
unemploymen bene i his o ies, comp omising he eliabili y o longi udinal
unemploymen analyses o yea s ollowing 2004 (An oni e al., 2019).
2.2.2. Cha ac e is ics o Wo ke s
We obse e he yea o bi h, sex, na ionali y, and educa ional a ainmen o indi iduals in
bo h samples. Fo employmen spells, we u he obse e he (anonymized) employe
iden i ie , employe ’s indus y code, geog aphic loca ion (coun y) o he employe , whe he
employmen was ull- ime o pa - ime, and he employee’s a e age daily wage. See
Appendix A1.2 o addi ional in o ma ion on hese a iables.
We cons uc se e al measu es o labo ma ke expe ience o all indi iduals in ou samples.
13 We de ine nu sing homes as es ablishmen s ca ego ized unde WZ73 indus y codes o p i a e and
o -p o i ins i u ions o “homes” (710), p i a e and no - o -p o i homes (711), and homes in public
owne ship (712). We use ime-consis en indus y codes ollowing he p ocedu e o Ebe le e al. (2011) and
conside only “ egula ” employmen , as de ined by he IAB con en ion—see Appendix A1.2 o de ails.
No - o -p o i ins i u ions, ypically a ilia ed wi h o owned by Ca holic and P o es an chu ches, hold a
signi ican ma ke sha e in he Ge man LTC sec o . While o simplici y o exposi ion we e e o all “homes”
ha all unde hese WZ73 indus y codes as nu sing homes, in p ac ice he indus y classi ica ion is coa se
and includes se e al di e en ypes o ca e and assis ed li ing ins i u ions ha may ha e di e en o mal
ce i ica ions. Some es ablishmen s may unc ion mo e like e i emen homes wi h minimal assis ance o
daily li ing, while o he s may p o ide in ensi e skilled nu sing ca e (ye o he s may se e popula ions
ou side he elde ly, such as adul s and child en wi h disabili ies). All such acili ies could be a ec ed by all
ypes o LTC insu ance bene i s and hence we include hem in ou analysis. Howe e , no all es ablishmen s
ha we classi y as nu sing homes quali y o inpa ien LTC insu ance bene i s, which equi e licensing unde
he Ge man SGB XI. The la e ce i ica ion ypically equi es speci ic skilled nu sing se ices o be o e ed on
si e. LTC insu ance s a is ics epo 6,564 such licensed acili ies (Bundesminis e ium ü Gesundhei (2001),
Anlage 8), while ou sample includes 11,401 acili ies—app ox. 74% mo e. We accoun o his disc epancy
in he wel a e analysis p esen ed in Sec ion 5.
14 IEB indus y codes ha e limi ed p ecision in he yea s we analyze, making i challenging o iden i y o mal
ou pa ien ca e p o ide s.
IAB-Discussion Pape 03|2025 15

Fo wo ke s in he nu sing home sample we cons uc a measu e o he gene al labo
ma ke expe ience by coun ing he numbe o yea s an indi idual was employed o e a
ixed 15-yea look-back window. Analogously, we cons uc a measu e o nu sing home
expe ience by coun ing he numbe o yea s he indi idual wo ked in a nu sing home
es ablishmen o e he p io 15 yea s. The 15-yea look-back window es ic s he numbe
o yea s ha we can include in eg essions ha use expe ience as an ou come a iable, o
yea s 1990 o 2008, bu helps a oid a censo ing bias in he ea ly yea s o ou sample pe iod.
Fo each nu sing home employmen spell, we addi ionally classi y each wo ke -yea
obse a ion in o a nu sing home incumben o a new hi e. A wo ke is conside ed o be a
new hi e in yea i he pe son was egula ly employed in a nu sing home in yea bu no in
yea − 1.
Fo wo ke s in he labo ma ke sample, we classi y each wo ke -yea obse a ion as being
ei he employed o unemployed (see p ecise de ini ions in Sec ion A1.2). I a wo ke is
unemployed in yea , we also c ea e an indica o o long- e m unemploymen s a us,
which equi es being unemployed in wo consecu i e yea s. We conside a wo ke o be
ejoining he labo o ce in yea i he wo ke was no obse ed in he da a in yea − 1, bu
was obse ed in a leas one yea p io o − 1, and was age 55 o younge (and hence no
close o ea ly e i emen ) in yea .
2.2.3. Mo ali y
We use wo sou ces o mo ali y da a. Fi s , we ob ain coun y-le el mo ali y a es o
indi iduals age 75 and olde om he i al s a is ics o Wes Ge man s a es, co e ing he
yea s 1991 o 2008.15 The second sou ce is he Human Mo ali y Da abase, which allows us
o compu e age g oup-speci ic mo ali y o yea s 1991–2008 o Wes Ge many and 27
o he coun ies.
2.2.4. Means-Tes ed and Uni e sal LTC Bene icia y Coun s
We use his o ical s a is ical epo s o compu e he numbe o indi iduals who ecei ed
means- es ed Hil e zu P lege in 1993. The mos g anula geog aphic esolu ion a which we
we e able o ob ain Hil e zu P lege ecipien coun s consis s o 15 geog aphic egions
co e ing he o me Wes Ge many. These egions include s a e-le el obse a ions o all
s a es (Lände ) and se en sub-s a e egions (Regie ungsbezi ke) wi hin Ba a ia. We sou ced
15 Da a om 1994 onwa ds a e publicly a ailable as online esou ces (S a is ische Äm e des Bundes und de
Lände , 2021a). Mo ali y da a o yea s 1991–1994 we e ob ained h ough w i en eques s om he
s a is ical o ices o he espec i e Ge man ede al s a es.
IAB-Discussion Pape 03|2025 16
s a e-le el ecipien coun s om S a is isches Bundesam (1993: p. 96) and ob ained coun s
o Ba a ia’s Regie ungsbezi ke om S a is ical O ice o Ba a ia (1993: p. 297).
We ob ained LTC claim coun s om ede al LTC insu ance p og am epo s (S a is isches
Bundesam , 1999b: p. 7). We use coun s om 1999, which is he i s yea wi h eliable
s a e-le el LTC insu ance s a is ics. We ex ac ed co esponding da a o Ba a ia’s se en
Regie ungsbezi ke om S a is ical O ice o Ba a ia (1999).
2.3 Summa y S a is ics
Table 1 p o ides desc ip i e s a is ics o he wo analy ic samples. The nu sing home
sample summa ized in column (1) consis s o 25.3 million obse a ions o 1.6 million
unique wo ke s ac oss yea s 1975 o 2008. Indi iduals in he sample a e, on a e age, 36
yea s old. 78 pe cen a e emale, and 94.5 pe cen a e Ge man na ionals. Abou 7 pe cen o
he wo ke s ha e comple ed he uppe ie o high school (Abi u ). 58 pe cen o
obse a ions a e o employmen in he heal hca e sec o , while 8.7 pe cen a e
unemploymen episodes. The employmen spells span nea ly a million unique
es ablishmen s o any kind, o which 22,629 a e nu sing home es ablishmen s.16
Nu sing home spells, summa ized in column (2), accoun o 10.2 million indi idual-yea
obse a ions. Nu sing home spells a e mo e likely o occu a a sligh ly olde age (40 s. 36
yea s old) and be pa - ime (25 pe cen o all employmen spells s. 29 pe cen o nu sing
home spells). The sha e o women is 82%. Wo ke s ea n 2.5 pe cen highe wages du ing
hei nu sing home spells (78.60 EUR/day s. 76.70 EUR/day, in 2020 Eu os). Abou 59
pe cen o nu sing home employmen episodes a e in no - o -p o i (mainly chu ch-owned)
nu sing homes, 27 pe cen in o -p o i , and 14 pe cen in publicly-owned ins i u ions.
The labo ma ke sample, summa ized in column (3) o Table 1, has 44 million obse a ions
o 3.8 million indi iduals in yea s 1985 o 2004. This sample is ep esen a i e o he Ge man
socially insu ed wo k o ce. Wo ke s a e on a e age 38 yea s old and 93 pe cen a e Ge man
na ionals. 41.5 pe cen o wo ke s a e women. 7.7 pe cen o wo ke s a e unemployed, 13
pe cen wo k pa - ime. The a e age daily wage is 96.4 EUR.
In sum, we see ha indi iduals who wo k in nu sing homes a some poin in hei ca ee s
a e p edominan ly women, a e mo e likely o wo k pa - ime, and ea n abou 20 pe cen
lowe wages han he a e age wo ke in Ge many.
16 Recall ha we a e only able o coun es ablishmen s and no i ms in ou da a. Also see oo no e 13 o he
exac de ini ion o which es ablishmen s a e included in ou coun o nu sing homes.
IAB-Discussion Pape 03|2025 17
3 Resea ch Design
3.1 Sou ce o Va ia ion
Ou empi ical s a egy akes ad an age o he geog aphic a ia ion in he p e alence o he
means- es ed Hil e zu P lege p og am p io o he ollou o uni e sal LTC insu ance. A e
he in oduc ion o LTC insu ance in 1995, all indi iduals wi h medically documen ed LTC
needs became eligible o he new insu ance bene i s.17 The new p og am, howe e , had
li le inancial impac on indi iduals al eady ecei ing Hil e zu P lege, as hei ca e needs
we e e ec i ely al eady co e ed by an exis ing public p og am. Consequen ly, egions wi h
a highe p e- e o m sha e o Hil e zu P lege co e age we e likely less a ec ed by he
insu ance expansion.18 The a ia ion in p e-exis ing co e age e lec s di e ences in local
Hil e zu P lege eligibili y de e mina ions as well as he his o ical inancial ci cums ances o
olde indi iduals in 1995.19
To cap u e his geog aphic a ia ion in exposu e o he new insu ance p og am, we
compu e he sha e o indi iduals wi h LTC needs who we e al eady “co e ed” h ough he
Hil e zu P lege p og am p io o 1995 o 15 geog aphic egions o Ge many (de ined in
Sec ion 2.2.4). The denomina o o his sha e is he numbe o claims o LTC insu ance in
1999, adjus ed back o 1993 o accoun o egion-speci ic demog aphic g ow h. This
measu es he la en demand o LTC p io o he expansion unde he assump ion ha a e
he expansion all indi iduals wi h medical needs e ealed hei demand by claiming
bene i s. We de ine he nume a o o be he coun o Hil e zu P lege ecipien s in 1993, jus
be o e he new insu ance p og am was signed in o law. Ou exposu e measu e E is hen:
Hil e zu P lege ,1993
E = 100% − , (1)
g ,1993,1999 × LTC Insu ance Claims ,1999
17 Medical necessi y o LTC is de e mined by independen medical assesso s who ha e no inancial incen i e
o app o e o deny applica ions. Acco ding o Nadash/Do y/ on Schwanen lügel (2018), bene i
de e mina ions, including denials, ha e gene ally been accep ed as eliable and ai , and, i appealed, a e
a ely o e u ned.
18 The e a e wo possible coun e ailing channels ha may complica e his in e p e a ion. Pa ien s quali ying
o Hil e zu P lege can selec cash bene i s o e o mal ca e bene i s in he pos - e o m pe iod. This is
p ima ily ele an o pa ien s wi h ela i ely mino ca e needs, who p e e in o mal o e o mal home ca e,
and a guably less ele an o pa ien s conside ing nu sing home ca e. This po en ial subs i u ion e ec
away om o mal ca e could ha e been mo e p onounced in egions wi h a highe p e- e o m sha e o Hil e
zu P lege co e age, hus a enua ing he e o m’s impac in hese egions (and hus ampli ying ou
es ima es o he causal e ec ). On he o he hand, a eas wi h a highe sha e o Hil e zu P lege paymen s ha
we e now displaced by ede al unds would ha e expe ienced a highe wind all o local budge s, which
could ha e independen ly a ec ed local economies, hus ampli ying he e o m’s impac in hese egions
(and hus a enua ing ou es ima es o he causal e ec ).
19 No ably, ou measu e o a ia ion does no simply e lec di e ences in con empo aneous incomes ac oss
a eas. Figu e A3 shows ha exposu e is unco ela ed wi h a e age a ea-le el income in 1993.
IAB-Discussion Pape 03|2025 18
The de la ion ac o g ,1993,1999 cap u es egion-speci ic aging ends, compu ed as he a io
o he popula ion aged 65 and olde in 1993 o ha in 1999. g ,1993,1999 hus de la es he
1999 coun o LTC insu ance ecipien s by he change in he numbe o olde indi iduals in
each egion be ween 1993 and 1999. In ui i ely, E measu es he sha e o indi iduals wi h
LTC needs who did no ha e insu ance co e age o hese needs p io o he e o m.20 The
geog aphic a ia ion in E is isualized in Figu e A4, which plo s he egion-le el a ia ion in
E and he mapping o coun ies o egions. On a e age, ac oss coun ies,21 68.7 pe cen o
indi iduals wi h LTC medical needs gained insu ance co e age. This exposu e a e exhibi ed
no able egional dispa i ies, anging om less han 60 pe cen in No h Rhine-Wes phalia o
nea ly 80 pe cen in pa s o Ba a ia.
Using his sou ce o a ia ion, we es ima e an e en s udy speci ica ion o examine whe he
a eas wi h g ea e exposu e o insu ance expansion expe ienced di e en ial changes in he
ou comes o in e es . Fo an ou come Y in coun y c wi hin egion in yea , we
c( )
es ima e:
=2008
X
Yc( ) = αc × 1(coun yc) + δ × 1(yea ) + λ × (E ) × 1(yea ) + c( ) (2)
=1975
whe e αc a e coun y ixed e ec s ha cap u e ime-in a ian di e ences in ou comes ac oss
coun ies, and δ a e yea ixed e ec s ha non-pa ame ically cap u e a common ime end
ac oss coun ies. The coe icien s o in e es , λ , mul iply he in e ac ion be ween yea ixed
e ec s and he ( ime-in a ian ) measu e o exposu e E . These coe icien s measu e
whe he a eas wi h g ea e exposu e o he LTC insu ance expansion expe ienced
di e en ial g ow h in ou comes. This lexible speci ica ion does no impose a end b eak in
any speci ic yea , allowing he da a o e eal whe he any changes in ends eme ged a e
he expansion. The iden i ying assump ion o a causal in e p e a ion o λ is ha , absen
he insu ance expansion, ou comes would ha e e ol ed in pa allel ac oss geog aphic a eas
wi h di e en le els o exposu e.
To simpli y he discussion o e ec magni udes, we also epo esul s om a
di e ence-in-di e ences speci ica ion ha pools he coe icien s o he ansi ion yea s
20 We canno dis inguish be ween indi iduals ineligible o Hil e zu P lege and hose no aking i up. While
incomple e ake-up is common and likely p esen in ou con ex , his dis inc ion is no essen ial o ou
analysis, as we a e in e es ed in he e ec i e change in he sha e o subsidized e sus unsubsidized LTC
pa ien s i espec i e o he eason o no ha ing a subsidy.
21 We assign each coun y he le el o E o egion in which he coun y is loca ed. Using coun ies as he uni
o analysis allows o mo e lexible speci ica ions o ime- and place-in a ian ac o s and p o ides a
compa able spa ial uni o analysis ac oss he di e en exposu e egions. I also be e aligns wi h na u al
labo ma ke s. We epo egion-le el speci ica ions in he Appendix; he esul s a e simila .
IAB-Discussion Pape 03|2025 19
(1994 o 1996) and he pos - e o m yea s (1997 o 2008):
=1992
X
Yc( ) = αc × 1(coun yc) + γ × 1(yea ) + δ × (E ) × 1(yea )
=1975 (3)
+ δ94−96 × (E ) × 1(yea 94−96)
+ δ97−08 × (E ) × 1(yea 97−08) + c( )
3.1.1. Pa ial Equilib ium
In ou pa ial equilib ium analysis o he nu sing home ma ke , we conside h ee se s o
ou comes Yc( ) : he numbe o nu sing home wo ke s and es ablishmen s, he wage o new
hi es and incumben wo ke s, and he demog aphic cha ac e is ics and labo ma ke
expe ience o new hi es. We cons uc all ou comes om indi idual-le el da a and
agg ega e o he coun y-yea le el by summing coun s ( o he i s se ) o by aking
a e ages ( o wo ke cha ac e is ics). To accoun o a ying coun y sizes and po en ial
di e ences in aging ends, we scale all coun -based ou comes in o pe capi a e ms using
he coun o he popula ion age 65 and olde . We also use hese popula ion coun s as
weigh s in wage eg essions. Th oughou , we clus e s anda d e o s a he coun y le el.
In he Appendix, we epo he esul s o se e al al e na i e speci ica ions o ou main
ou comes o in e es . We show speci ica ions ha e-es ima e Equa ion (2) a he egion
a he han coun y le el ( op ow o Figu e A6 and columns (1)-(4) o Table A1), o ha keep
he coun y-le el speci ica ion, bu clus e s anda d e o s a he egion le el (bo om ow
o Figu e A6 and columns (5)-(8) o Table A1). We also p esen he esul s o speci ica ions
ha include coun y-speci ic ime ends ( op ow o Figu e A7 and columns (9)-(12) o Table
A1), con ol o he sha e o he popula ion age 65 and olde in he coun y (bo om ow o
Figu e A7 and columns (5)-(8) o Table A2), o con ol o he coun y-yea coun s o people
aged 65 and olde ( op ow o Figu e A8 and columns (1)-(4) o Table A2). Fu he , we show
wo speci ica ions ha use al e na i e ways o de ining exposu e (bo om ow o Figu e A8,
Figu e A9, and Table A3). Figu e A5 shows ha al e na i e de ini ions o exposu e a e highly
co ela ed wi h ou baseline measu e. Ac oss all speci ica ions, ou main akeaways emain
quali a i ely he same.
IAB-Discussion Pape 03|2025 20

3.2 Gene al Equilib ium
We nex mo e beyond nu sing homes and conside he o e all e ec s o he LTC insu ance
expansion on he local economy.22 We conside se e al coun y-le el employmen
ou comes, including local a es o unemploymen (including long- e m unemploymen ),
labo o ce size, he numbe o wo ke s e-en e ing he labo o ce, a e age wages, and he
o al wage bill. A p ac ical challenge is ha o e all local labo ma ke ends a e hea ily
in luenced by place-speci ic mac oeconomic luc ua ions, making i ha d o sepa a e end
changes ela ed o ea men om he o e all ola ili y. To add ess his, o ou gene al
equilib ium analysis we adop he syn he ic di e ence-in-di e ences me hod, which
weakens he eliance on he pa allel ends assump ion (A khangelsky, Dmi y and A hey,
Susan and Hi shbe g, Da id A. and Imbens, Guido W. and Wage , S e an, 2021). This
da a-d i en p ocedu e op imally eweigh s ea men and con ol uni s, ensu ing ha
ea men uni s a e compa ed o con ols ha had mo e simila ends p io o ea men
(in ou case, his would mean compa ing uni s wi h mo e simila mac o-economic
luc ua ions). To apply his me hod, we con e he con inuous exposu e E o a
dicho omous indica o . Coun ies a e classi ied as ea ed i hei exposu e is abo e he
median, while hose a o below he median a e used as con ol uni s. This bina y measu e
co esponds o a 9.1 pe cen age poin di e ence in exposu e be ween ea ed and con ol
coun ies.
3.3 Mo ali y
A na u al and key ques ion is whe he any o he changes p ecipi a ed by he ollou o
uni e sal LTC insu ance ul ima ely esul ed in imp o ed ca e o he olde popula ion—one
o he s a ed goals o he LTC e o m (Ro hgang, 1997). While we lack di ec measu es o ca e
quali y, we conside mo ali y a olde ages as one way o cap u e he ne e ec o he LTC
e o m on (po en ial) pa ien s. We examine dea hs among he popula ion aged 75 and olde
a he coun y le el. We es ima e Equa ion (2) as ou baseline speci ica ion, bu also epo
esul s om he syn he ic di e ence-in-di e ences speci ica ion in Figu e A10a. In addi ion,
we es ima e a syn he ic con ol speci ica ion ha le e ages a ia ion in mo ali y among
he elde ly (aged 75 and olde ) be ween Ge many and o he high-income coun ies, a he
han a ia ion in exposu e ac oss geog aphic egions wi hin Ge many (Figu e A10b). In his
speci ica ion we conside Wes Ge many o be “ ea ed” a e 1995, and cons uc a
coun e ac ual mo ali y ime end (a “syn he ic” Wes Ge many) by eweighing mo ali y
ends in o he coun ies ollowing Abadie/Diamond/Hainmuelle (2010).
22 This analysis also cap u es he e ec s on he o mal ou pa ien LTC sec o and LTC cash bene i s, which we
canno in es iga e sepa a ely.
IAB-Discussion Pape 03|2025 21
4 Resul s
4.1 Pa ial Equilib ium
4.1.1. Indus y Expansion
Figu e 1 plo s he ime se ies o he numbe o nu sing home wo ke s (Panel 1a) and he
numbe o nu sing home es ablishmen s (Panel 1b) pe 1,000 indi iduals aged 65 and olde
be ween 1975 and 2008. We show he unweigh ed a e age o each ou come sepa a ely o
coun ies in egions wi h abo e-median exposu e (blue solid line) and below-median
exposu e ( ed dashed line). Bo h ime se ies a e no malized o he o e all unweigh ed
mean ac oss all coun ies in 1993 (Figu e A11 shows he aw ime se ies wi hou his
no maliza ion). The g ey shaded a ea ma ks he yea s o he insu ance ollou .
Th ee ac s eme ge. Fi s , inpa ien LTC employmen pe capi a saw pe sis en g ow h o e
he h ee decades ha we s udy, wi h he numbe o nu sing home wo ke s pe 1,000
indi iduals age 65 and olde mo e han ipling om 12 o 40 wo ke s.23 Second, he g ow h
in he nu sing home wo k o ce was accompanied by g ow h in he numbe o nu sing home
acili ies pe capi a un il he la e 1990s, inc easing o mo e han one acili y pe 1,000 olde
adul s in 1990. Howe e , g ow h in es ablishmen s slowed ela i e o popula ion g ow h
a e his pe iod. Thi d, we obse e ha in he pos -expansion yea s, bo h he numbe s o
nu sing home es ablishmen s and o wo ke s pe 1,000 elde ly we e g owing isibly as e
(o declining slowe ) in coun ies ha we e mo e exposed o he insu ance expansion.
In Panels 1c and 1d o Figu e 1, we p esen he esul s o es ima ing he e en s udy in
Equa ion (2), wi h poin es ima es also epo ed in Table 2. The es ima es sugges ha p io
o 1995, he a e o g ow h in he numbe o wo ke s and in he numbe o es ablishmen s
pe 1,000 olde adul s did no di e ac oss geog aphic a eas wi h di e en le els o
means- es ed co e age in 1993. A e he ollou o uni e sal insu ance, howe e , g ow h
was mo e p onounced in a eas ha we e mo e exposed o insu ance expansion.24 The
accele a ion la ened ou a ound eigh yea s a e he new insu ance was signed in o law.
Table 2 p esen s pooled di e ence-in-di e ences es ima es om Equa ion (3). The esul s in
columns (1) and (4) indica e ha , on a e age, a 10 pe cen age poin inc ease in exposu e o
23 This co esponds o an annual g ow h a e o 3.6%, in con as o a 0.2% annual g ow h a e o he o e all
Wes Ge man popula ion and 1.2% o he 65+ popula ion du ing his pe iod (Human Mo ali y Da abase).
24 Figu e A11 sugges s ha places wi h highe exposu e (and hence a lowe p e alence o subsidized ca e p io
o insu ance expansion) had lowe le els o wo ke s and es ablishmen s pe capi a p io o he e o m. These
a eas caugh up wi h less exposed a eas in he numbe o wo ke s pe capi a a e he insu ance expansion.
IAB-Discussion Pape 03|2025 22
he e o m (equi alen o a 15 pe cen inc ease ela i e o he mean) led o an addi ional
ou nu sing home wo ke s pe 1,000 indi iduals age 65 and olde (a 12 pe cen inc ease)
and 0.06 mo e nu sing homes pe 1,000 elde ly (a 6 pe cen inc ease). The inc ease in he
numbe o nu sing home wo ke s is app oxima ely e enly dis ibu ed be ween pa - ime
and ull- ime wo ke s (columns 2 and 3), sugges ing a signi ican ly la ge ela i e e ec o
pa - ime wo ke s. Be o e he e o m, he numbe o pa - ime wo ke s was less han hal
ha o ull- ime wo ke s (nine pa - ime wo ke s pe 1,000 elde ly in 1993 e sus 24 ull- ime
wo ke s). Figu e A12 illus a es he ime se ies and e en s udies o bo h wo ke
ypes.
A 10 pe cen age poin change in exposu e is close o he 9.1 pe cen age poin di e ence in
mean exposu e be ween coun ies wi h abo e- and below-median exposu e. We e e o his
di e ence as “in-sample a ia ion” in he hi d panel o Table 2. Mul iplying hese
“in-sample” pe capi a e ec s by he coun o indi iduals aged 65 and olde , we es ima e
ha o e ing LTC insu ance o an addi ional 9.1 pe cen age poin s o uninsu ed elde ly adds
38,979 nu sing home wo ke s and 542 nu sing home es ablishmen s. We also calcula e an
ou -o -sample es ima e o he agg ega e impac o LTC insu ance. This ex apola ion
assumes linea scaling wi h exposu e, which is inhe en ly mo e specula i e. The
ou -o -sample p edic ion sugges s ha expanding uni e sal co e age by 68.7 pe cen age
poin s (i.e. mo ing om 31.3 pe cen o he po en ial pa ien popula ion being insu ed o
ull co e age; o om 0 pe cen o 68.7 pe cen ) would esul in 0.42 mo e nu sing home
es ablishmen s pe 1,000 olde adul s (a 40 pe cen inc ease ela i e o he mean o 1.04 in
1993) and nea ly a doubling o he nu sing home wo k o ce, adding close o 300,000 nu sing
home jobs.
To con ex ualize hese employmen e ec s, we cons uc a coa se a c elas ici y by di iding
he ex apola ed inc ease in employmen by he implici change in he ou -o -pocke p ice
o an a e age consume :
ΔQ/(Q1 + Q2)
a c = = −0.74 . (4)
ΔP /(P 1 + P 2)
The nume a o conside s changes in employmen pe 1,000 olde adul s (ΔQ=30.5 wo ke s)
ela i e o he a e age employmen be ween he p e- e o m pe iod (Q1=33.1 wo ke s as
seen in column (1) o Table 2), and he pos - e o m pe iod (Q2 = Q1+ΔQ=63.6 wo ke s).
The denomina o conside s he change in p ices o inpa ien ca e. P e- e o m, on a e age
68.7 pe cen o po en ial pa ien s paid he ull p ice o ca e ou -o -pocke , while he
emaining 31.3 pe cen we e ully insu ed h ough means- es ed Hil e zu P lege. Hence, he
a e age ou -o -pocke p ice p e- e o m was P 1 = 0.687Pma ke . Pos - e o m, cos -sha ing
o he newly insu ed d ops o 40 pe cen (c . oo no e 4). The new a e age ou -o -pocke
IAB-Discussion Pape 03|2025 23
p ice hen becomes P 2 = 0.687 × 0.40Pma ke , o 0.28Pma ke .25 Pu oge he , his sugges s
an a c elas ici y o -0.74, which signi ican ly exceeds he elas ici y es ima es in he RAND o
he O egon expe imen s (Newhouse/Insu ance Expe imen G oup, 1993; Finkels ein e al.,
2012). Ou da a on i m en y sugges ha ixed cos s o in es men may be one explana ion
o hese di e ences (Finkels ein, 2007).
4.1.2. Ana omy o Expansion
In his sec ion, we cha ac e ize he na u e o he nu sing home indus y expansion. We
conside he change in wages and he composi ion o newly hi ed wo ke s. Figu e 2
summa izes all poin es ima es.
P ice e ec s
We s a by examining whe he he la ge expansion o he nu sing home wo k o ce was
accompanied by g ow h in wages. We expec ha i ms inc eased hei wages in o de o
a ac mo e new hi es o o imp o e he e en ion o incumben wo ke s. Table 3 displays
he esul s om es ima ing Speci ica ions (2) and (3) o (log) daily wages o new ull- ime
hi es (columns 1 and 2) and o incumben ull- ime wo ke s (columns 3 and 4).26 We ind no
e idence o sys ema ic inc eases in wages on a e age, o ei he he new hi es o he
incumben s. Fo he new hi es, he poin es ima es a e close o ze o when we con ol o
expe ience (column 2), sugges ing ha he wage decline in column 1 is d i en by a shi
owa ds lowe -skilled wo ke s. Fo incumben s, we ind no e idence o a change in wages.
This can be mos clea ly seen in he e en s udy coe icien s o he i s and second yea
pos expansion (λ1997 and λ1998), as hese coe icien s mos ly cap u e he e ec s o wo ke s
who we e hi ed p io o he e o m.
The lack o an inc ease in wages in a apidly expanding sec o is consis en wi h ei he a
pe ec ly elas ic labo supply cu e among lowe -skilled wo ke s, o wi h he exis ence o
labo ma ke ic ions and he p e alence o excess labo supply p io o he insu ance
expansion. Fo example, nu sing home labo ma ke s may no (jus ) clea in p ices in he
p esence o sea ch and ma ching ic ions ha in e ac wi h ax wedges, unemploymen
insu ance bene i s, and collec i e ba gaining ag eemen s, all o which cha ac e ized he
25 I we assume (1) ha he en i e y o he employmen expansion is due o new pa ien s wi hou Hil e zu
P lege, (2) ha he p opo ion o inpa ien pa ien s among all pa ien s unde Hil e zu P lege is he same as
unde ull insu ance in 1999, and (3) ha he s a ing a io o pa ien s o wo ke s is ixed, hen we ob ain an
e en highe elas ici y. In his case, he ela i e inc ease in pa ien s is la ge (conside ing a smalle base o
non-Hil e zu P lege pa ien s). A he same ime, he obse ed inc ease in demand combines a subs i u ion
and an income e ec , which we conside in he s uc u al model ou lined in Sec ion 5.2. Isola ing he
subs i u ion e ec , he model es ima es sugges a demand elas ici y o -1.19.
26 E en s udies a e in Figu e A13. To allow o indi idual-speci ic con ols in coun y-le el eg essions, we i s
esidualize indi idual-le el log wages wi h espec o wo ke ixed e ec s and hen compu e coun y-yea
le el a e ages o hese esiduals.
IAB-Discussion Pape 03|2025 24
sa ing o cos o he go e nmen and on he ma gin (unde simpli ying assump ions
e isi ed in Sec ion 5.2) no addi ional willingness o pay by wo ke s.41 This is he poin
highligh ed by Baicke /Chand a (2012)— ealloca ing wo ke s o subsidized sec o s om
o he sec o s o he economy gene ally does no gene a e wel a e gains on labo ma ke s.
Ou esul s in Sec ion 4, howe e , sugges ha some ma ginal wo ke s en e ed he nu sing
home sec o ou o unemploymen o om ou o he labo o ce. Fo hese wo ke s, wages
ep esen ed a sou ce o su plus, while he go e nmen collec ed addi ional axes and
educed i s unemploymen insu ance expendi u es. To quan i y hese bene i s, we make
se e al assump ions. Fi s , we assume ha o e e y ex a Eu o in wages, he Ge man
go e nmen collec ed 60 cen s in axes and social insu ance con ibu ions.42 Second,
ollowing Mui/Schoe e (2024), we assume ha , on a e age, wo ke s’ ese a ion wages
we e 90 pe cen o he a e - ax (and a e -social insu ance con ibu ion) income.43 This
implies ha a Eu o in g oss wages gene a es 4 cen s (10 pe cen o he pos - ax 40 cen s) in
wo ke su plus o success ul employmen ma ches among wo ke s who we e o me ly
unemploymen o ou o he labo o ce. Adding his wo ke su plus, go e nmen ax
e enue, and unemploymen insu ance sa ings o ou MVPF es ima es abo e, we ob ain a
combined MVPF o he insu ance expansion in he skilled inpa ien nu sing home ma ke o
1.06. In o he wo ds, when we accoun o wo ke s who ound new jobs as a esul o
subsidizing pa ien s’ demand o ca e, we ge posi i e e u ns on a dolla o public
spending.
Gene al Equilib ium
We now u n o he gene al equilib ium analysis, asking: Wha was he o e all economic
e u n on each dolla o public unds spen subsidizing long- e m ca e h ough he uni e sal
long- e m ca e insu ance p og am? Fo his compu a ion, we use he mos conse a i e
e sion o ou gene al equilib ium labo ma ke es ima es— he lowe bound o he 95
pe cen con idence in e als. The gene al equilib ium e ec s a e d i en p ima ily by iscal
ex e nali ies in he labo ma ke . The go e nmen gains addi ional income ax e enue and
sa es on unemploymen bene i s as wo ke s ansi ion in o highe -wage jobs, en e he
labo ma ke , and exi unemploymen . Ou es ima es indica e ha he go e nmen sa es
4.7 billion EUR in unemploymen insu ance paymen s and collec s an addi ional 26 billion
EUR in income axes and con ibu ions o social insu ance p og ams (see column (3) o
41 Wo ke s could ha e expe ienced gains om ha ing a mo e obus labo ma ke , highe job inding a es
om inc eased acancy pos ing, po en ial gains in ameni ies, o o he ac o s, none o which we a e able o
cap u e in his compu a ion. We e isi he ole o acancy pos ing and non-wage compensa ing
di e en ials in Sec ion 5.2.
42 In 1999, a single wo ke wi h no child en paid 18.6% in income ax and 41.6% in social insu ance
con ibu ions, he la e s a u o ily spli equally be ween he employe and he employee (Bundeszen ale
ü Poli ische Bildung, 2023).
43 Mui/Schoe e (2024) es ima e he median ese a ion “ aise” (which aims o cap u e a no ion o ese a ion
wages) o be ci ca 90% o obse ed pos - ax and pos -con ibu ions wages in he Ge man con ex .
IAB-Discussion Pape 03|2025 31

Table 8). These iscal gains a exceed he 8.4 billion EUR in addi ional spending on
long- e m ca e subsidies, o 5.9 billion EUR in pa ien su plus. In summa y, he policy
achie es an in ini e MVPF in gene al equilib ium—i no only pays o i sel bu also
gene a es su plus o he go e nmen .
Summa y
To summa ize, he MVPF analysis o he uni e sal LTC insu ance expansion when ocused
only on pa ial equilib ium wi hou accoun ing o labo ma ke e ec s concludes ha he
MVPF is below 1. This e lec s he amilia ex book case o deadweigh loss om mo al
haza d and is a common inding o many heal hca e p og ams, especially aimed a olde
adul s. Ou poin es ima e is e y simila o o he MVPF es ima es om heal h insu ance
expansions o adul popula ions, as discussed in Hend en/Sp ung-Keyse (2020) and
summa ized in Figu e A16. In con as , when we add he willingness o pay o wo ke s who
we e a ec ed by changes in consume demand, as well as he co esponding iscal
ex e nali ies in axes and UI sa ings, we conclude ha a ma ginal dolla in es ed in o LTC
insu ance gene a es a subs an ial iscal payo . This pu s heal h insu ance expansions o
adul popula ions mo e on pa wi h p og ams o child en which equen ly achie e in ini e
MVPF. Howe e , he mechanism he e is di e en , as i ope a es h ough he indi ec bene i s
o he economic s imulus o local labo ma ke s a he han he di ec bene i s o p og am
ecipien s.
We emphasize ha ou es ima es a e speci ic o he Ge man labo ma ke con ex o he
1990s. The la ge employmen gains and signi ican posi i e iscal ex e nali ies we
documen a e ied o high con empo a y baseline unemploymen a es and low labo o ce
pa icipa ion. These condi ions, in u n, we e a leas pa ially he esul o subs an ial labo
ma ke wedges—high unemploymen bene i s and ax a es (Dus mann e al., 2014). As
such, ou MVPF es ima e he e is no a s uc u al pa ame e o heal h insu ance expansions,
bu is a e lec ion o he economic ci cums ances and exis ing ax and ans e p og ams
unde which he e o m was implemen ed. In he nex sec ion, we o e a o mal model ha
allows us o speak o he ex e nal alidi y o ou es ima es by quan i ying he ole o hese
incumben economic condi ions.
5.2 Model-based App oach
We speci y and es ima e a s uc u al model o he labo ma ke . Building on
Acemoglu/Shime (1999) and W igh e al. (2021), we conside a di ec ed sea ch model, in
which inding a job is a cos ly p ocess and ma ke s do no clea solely in p ices. We
IAB-Discussion Pape 03|2025 32
inco po a e wo ke and i m he e ogenei y, unemploymen bene i s, income axes, and
collec i e ba gaining o e lec he ins i u ional con ex . We pin down he pa ame e s o he
model using ou causal es ima es om Sec ion 4 as a ge momen s. We hen use he model
o e isi he wel a e calcula ion.
5.2.1. Theo y
En i onmen :
The economy is popula ed by a con inuum o po en ial i ms and wo ke s. Wo ke s di e in
hei skill le el φ. We assume ha each i m has a p oduc ion echnology ha equi es one
wo ke . Le j ∈ J index he sec o s o he economy—all i ms in j ha e homogeneous
p oduc ion echnologies. Wo ke s and i ms mee ia sea ch. Fi ms i s decide whe he o
en e a sec o and which skill segmen o en e in. Condi ional on en y, each i m pos s a
φ
acancy wi h a wage w ; he wage will be common among all i ms in j in equilib ium. I j is
j
φ
subjec o binding wage-se ing ic ions, e.g. due o collec i e ba gaining, i ms se w o
j
equal he wage loo . In he nex s age, wo ke s obse e all wage o e s o hei skill ype
and decide whe he o sea ch and apply o one job o e , o whe he o s ay ou o he labo
ma ke . We deno e he a io o he numbe o applican s o he numbe o acancies in each
φ
sec o j and skill segmen φ wi h q > 0 and e e o his as he queue leng h. Each
j
φ φ
applican is hi ed wi h a p obabili y µ(q ). I hi ed, he wo ke ea ns wage w , pays axes a
j j
a e τ, and p oduces φ ou pu uni s. O he wise, he applican emains unemployed and
ob ains unemploymen bene i s as well as bene i s om home p oduc ion. The p obabili y
o being hi ed dec eases in he queue leng h. Con e sely, he p obabili y o illing a
φ φ φ φ
acancy, deno ed wi h η(q ) = µ(q ) × q , inc eases in qj . In ui i ely, i ms ha e highe
j j j
chances o illing acancies in labo ma ke s ha ha e mo e applican s o any ixed numbe
o open posi ions.
Payo s:
The e a e h ee ypes o agen s in he economy— i ms, wo ke s, and consume s o he inal
good. We desc ibe each o hei payo s in u n. A i m’s payo is i s p o i . Fo a i m in
sec o j ha has en e ed skill segmen φ, he p o i is gi en by:
φ φ φ
π = η(q ) × (Pj × φ − w ) − cj
(φ). (5)
j j j
cj
(φ) deno es he cos o acancy pos ing, cap u ing ec ui ing and e en ion cos s. These
cos s a e incu ed wi h ce ain y—i espec i e o whe he he i m manages o ill he
acancy. We allow hese cos s o a y ac oss sec o s and wo ke skill, e lec ing ha i may
IAB-Discussion Pape 03|2025 33
φ
be cos lie o hi e highe -skilled wo ke s. Wi h p obabili y η(q ), he i m ills i s acancy,
j
p oduces ou pu , collec s e enue, and pays wages. This is cap u ed by he i s e m o he
φ
p o i unc ion. The i m pays wage w o ma ched wo ke s, who p oduce φ uni s o ou pu
j
ha he i m can sell a p ice Pj pe uni .
Wo ke i maximizes he expec ed u ili y by di ec ing he sea ch owa d sec o j o s aying
ou o he labo o ce. The expec ed u ili y o wo ke i wi h skill φ is gi en by:
⎧
⎨ φ φ φ u
µ(q ) × ue((1 − τ) × w ) + (1 − µ(q )) × u i i sea ches in j
j j j + ij
Uij = , (6)
⎩ ool
u + i0 i i doesn’ sea ch
whe e ue deno es he payo om employmen when ma ched o a acancy. This payo
φ
depends on he a e - ax wage (1 − τ) × wj . uu is he payo om unemploymen , which
occu s i he wo ke sea ches bu is no ma ched and inc eases in unemploymen bene i s.
uool is he low u ili y om s aying ou o he labo o ce and no sea ching. Wo ke s who do
no sea ch a e assumed o ob ain u ili y om home p oduc ion and leisu e, as well as he
bene i o o egoing he hassle cos o applying o jobs along wi h po en ial s igma e ec s
o no being o mally employed. ij deno es idiosync a ic p e e ence shock ha wo ke i
may ha e o sec o j. P io expe ience in sec o j, o li ing close o i ms in sec o j, o
ha ing a small child a home and needing mo e lexibili y and p oximi y o wo k may lead
he wo ke o ob ain a highe u ili y om choice j, all else equal.
Finally, we conside a ep esen a i e consume who has p e e ences o e ou pu p oduced
in all sec o s o he economy. The consume sol es
whe e (Q) is he u ili y om consuming he ec o o ou pu quan i ies Q and
max (Q)
Q
X
(7)
s. . y ≤ Pj × Qj (8)
j
P
Pj × Qj
j
deno es consume expendi u es on all goods j ∈ J. y deno es income de i ed om
ea nings and ans e s, as de ailed below.
Equilib ium:
Building on p oposi ion 1 in Acemoglu/Shime (1999), we de ine he sea ch equilib ium as a
uple o wages, queue leng hs, ou pu p ices, and ou pu quan i ies ha maximizes he
wo ke ’s expec ed u ili y, Uij
, subjec o he ollowing cons ain s. Fi s , i ms maximize
expec ed p o i s, aking ou pu p ices and any ins i u ional wage-se ing cons ain s as
gi en (Equa ion 10). Second, unde ee en y, i ms’ expec ed p o i s a e ze o (Equa ion 11).
IAB-Discussion Pape 03|2025 34
Thi d, ou pu ma ke s clea in each sec o (Equa ion 12).
max Uij (9)
φ φ
w ,q
φφ
s. . w = w i j is cons ained by wage loo (10)
j
φ φ
η(q ) × (Pj × φ − w ) − cj
(φ) = 0 (11)
j j
QD
j (Pj) = QS
j (Pj
) (12)
In equilib ium wi h di ec ed sea ch, i ms pos e icien wages ha depend on he elas ici y
o he ma ching unc ion (Moen, 1997; Acemoglu/Shime , 1999). In ou se ing, wages and
acancy pos ings a e a ec ed by unemploymen insu ance bene i s, income ax wedges,
and collec i e ba gaining ag eemen s. A2.1 p o ides addi ional model de ails.
5.2.2. Taking Theo y o Da a
Pa ame ic assump ions
We make se e al pa ame ic assump ions o ake his heo e ical model o ou da a. Fi s ,
we assume a single na ional labo and p oduc ma ke , abs ac ing om he geog aphic
a ia ion. We index “ma ke s” by ∈ 0, 1, dis inguishing be ween he obse ed pos - e o m
pe iod and he coun e ac ual wi hou insu ance expansion. Second, o ac abili y, we
con e all ypes o employmen o ull- ime equi alen s. Thi d, we assume ha wo ke
he e ogenei y is a unc ion o he numbe o yea s o heal h ca e expe ience ehc, which
anges om 0 o 18 in ou se ing, de ining 19 di e en ypes o wo ke s. We allow o he
impo ance o heal hca e expe ience o be di e en ac oss sec o s, and we also allow o a
discon inui y in skill a ze o expe ience, as empi ically wo ke s wi h no heal hca e
expe ience a e much less likely o wo k in nu sing homes. We impose he ollowing
ela ionship be ween yea s o heal hca e expe ience and skills:
φj
(ehc) = ζ210
j + ζj × ehc + ζj × 1{ehc > 0}. (13)
Nex , we assume ha he economy has h ee sec o s: o -p o i nu sing homes, public and
no - o -p o i nu sing homes, and all o he sec o s o he economy. We dis inguish be ween
o -p o i and no - o -p o i nu sing homes as he la e may ace di e en wage cons ain s
due o collec i e ba gaining, and may di e in non-wage ma ginal cos s cap u ing
di e ences in i m objec i es (Lakdawalla/Philipson, 1998). The wages in no - o -p o i and
public nu sing homes a e hen de e mined exogenously acco ding o a second-o de
polynomial in he ma ginal e enue p oduc ne o non-wage ma ginal cos s; see Sec ions
IAB-Discussion Pape 03|2025 35
A2.2 and A2.3 o de ails.
Sea ch and ma ching ic ions help econcile he la ge changes in employmen absen
signi ican changes in wages. Following Buchholz (2022), we assume he ollowing o m o
he ma ching unc ion:
 φ

φ j
η(q ) = 1 − exp(−q
) . (14)
j φ
λj
We allow he pa ame e go e ning he ma ching e iciency λ o a y lexibly by skill le el
φ
and sec o . Vacancy pos ing is cos ly, and we allow he acancy cos s c o a y by skill le el
j
and be ween o -p o i and non- o -p o i /public nu sing homes.
Tu ning o he wo ke ’s low payo s, we assume:
e,φ φ
u = (1 − τ) × w + ξj + ν ∗ 1{ehc > 0} ∗ 1{NH} (15)
j j
u
u = b + ξu (16)
ool
u = κ0 + κ1 ∗ 1{ehc > 0} (17)
 
ij = γ × ϑig + (1 − ρ) × ˜ij . (18)
In he low u ili y om employmen , ξj cap u es compensa ing di e en ials o wo king in
sec o j. Pa ame e ν allows compensa ing di e en ials o a y by heal h expe ience o
cap u e di e en ial a achmen o wo king in nu sing homes o wo ke s wi h some heal h
ca e expe ience. We no malize he compensa ing di e en ial in he ou side sec o o ze o.
In he low u ili y om unemploymen , b deno es unemploymen bene i s and ξu cap u es
mone ized non-pecunia y u ili y om unemploymen . Simila ly, uool cap u es mone ized
non-pecunia y u ili y om being ou o he labo o ce which we allow o a y based on
whe he wo ke s ha e any heal h expe ience. Las ly, we impose an assump ion on he
dis ibu ion o idiosync a ic wo ke p e e ences ha leads o a nes ed logi choice p oblem
o he wo ke . We g oup he wo nu sing home sec o s in o one nes g, keeping he ou side
sec o and being ou o he labo o ce as sepa a e nes s. ϑig deno es a as e shock ha
wo ke i may ha e o nes g, whe eas ˜ij is an iden ically and independen ly dis ibu ed
ex eme alue shock o sec o j. The nes ing pa ame e 0 ≤ ρ < 1 go e ns he co ela ion
o wo ke u ili y ac oss di e en ypes o nu sing homes, and γ deno es a scaling
pa ame e .
IAB-Discussion Pape 03|2025 36

Finally, we assume ha he ep esen a i e consume has CES p e e ences, and ha nu sing
homes p oduce homogeneous ou pu s. We hen w i e he consume p oblem as:
 σ−1  σ
σ−1
Qo, σ−1
σ
(Q ) = αo × Jo × ( Jo ) σ + αNH × QNH, (19)
˜
s. . QNH, = QNH, + HzPNH (20)
y + T ≥ Po × Qo + (PNH, − sNH, ) × QNH, . (21)
We assume ha he ou side sec o comp ises Jo homogeneous subsec o s. The α deno e
scaling pa ame e s and σ deno es he elas ici y o subs i u ion. QNH deno es he demand
o nu sing home ca e se ices paid ou o pocke . HZPNH deno es he demand by pa ien s
insu ed ia Hil e zu P lege, which we ea as exogenous. Income y is gi en by he sum o
he wage ea nings o wo ke s, i m p o i s (ze o in equilib ium), acancy cos s (ea ned by
ec ui e s), and o he ma ginal cos s. T deno es a lump-sum ans e om he go e nmen
o he consume , which is income axes ne o go e nmen spending on nu sing home ca e.
The main policy pa ame e o in e es o ou analyses o insu ance expansion is sNH
, which
cap u es he p oduc subsidy ha he go e nmen p o ides o he nu sing home sec o s.
Es ima ion
Ou es ima ion s a egy p oceeds in wo s eps. In he i s s ep, we es ima e he pa ame e s
go e ning wo ke p e e ences, p oduc ion, and he ma ching echnology ia gene alized
me hod o momen s (GMM). In he second s ep, we impose he ma ke clea ing condi ions
in he p oduc ma ke o eco e he p e e ence pa ame e s o he ep esen a i e consume .
We use wo p ima y se s o momen s o es ima ion: i s , he obse ed employmen sha es
and mean wages by sec o and yea s o heal h ca e expe ience in 1999; and second, he
causal es ima es om Sec ion 4.1.1. Figu e 6 illus a es he key empi ical momen s and
model i . Sec ion A2.2 desc ibes he de ails o he es ima ion app oach.
The de ails o all pa ame e es ima es a e epo ed in Sec ion A2.3. We es ima e an elas ici y
o demand o ou -o -pocke p ices o nu sing home ca e o -1.19, sligh ly la ge han ou
es ima e o he a c-elas ici y in Equa ion (4). The model also allows o (and es ima es)
subs an ial income e ec s om he insu ance expansion. The impo ance o income e ec s
in ou con ex is simila o he ole o income e ec s ound in Finkels ein/Hend en/Lu me
(2019) o Medicaid co e age in he U.S.
IAB-Discussion Pape 03|2025 37
5.2.3. Wel a e and Coun e ac uals
Table 9 summa izes ou wel a e analysis. Columns (1) h ough (3) keep he ax and ans e
en i onmen ixed. We epo he equilib ium labo ma ke alloca ion in Panel A and
su plus in Panel B. Mo ing om column (1) o column (2) shows how he alloca ion and
su plus change when we in oduce LTC insu ance in o he economy.44 As in he causal
e idence o Sec ion 4.1.1, he model simula ion o LTC subsidies esul s in an inc ease in
nu sing home employmen and o e all labo o ce pa icipa ion.
We decompose he change in wel a e in o h ee componen s:
ΔW = ΔConsume Wel a e + ΔWo ke Wel a e + ΔGo e nmen Su plus . (22)
We measu e he change in consume wel a e h ough he equi alen a ia ion, conside ing
changes in ou -o -pocke p ices bu holding p e- e o m income y0 + T0 ixed. Consume s
bene i om lowe ou -o -pocke p ices o nu sing home ca e bu a e ha med by he p ice
inc ease o he ou side sec o good (no conside ed in Sec ion 5.1).45 Wo ke wel a e is also
measu ed h ough he equi alen a ia ion, cap u ing changes in acancy pos ings,
a e - ax wages, and hence ea nings.46 Finally, he go e nmen su plus is de ined as ax
e enues minus LTC subsidy spending and UI spending.
We ind ha on ne , consume wel a e wi h LTC insu ance goes down in he model. This
decline in consume wel a e as well as he ex a go e nmen ou lays on LTC subsidies a e,
howe e , o se by wel a e gains in he labo ma ke s as well as by a subs an ial inc ease in
ax e enues collec ed by he go e nmen . As in he model- ee calcula ion, we ge ha in
he gene al equilib ium en i onmen o e all social wel a e goes up when LTC insu ance is
olled ou and ha he o e all go e nmen su plus ( ax e enues minus spending on LTC
insu ance and unemploymen bene i s) inc eases, i.e. he insu ance expansion pays o
i sel . Quan i a i ely, we ind a ne wel a e gain o 0.78 billion EUR pe mon h, o 2,294 EUR
pe ull- ime-equi alen nu sing home wo ke and mon h.
44 We ope a ionalize his by ea ing column (2) as da a—i is he obse ed en i onmen wi h insu ance in
place in 1999 as simula ed by he model. Column (1) hen epo s he esul s o a coun e ac ual simula ion
ha emo es long- e m ca e subsidies.
45 We i s calcula e consume u ili y a p e- and pos - e o m ou -o -pocke p ices, holding income ixed a
p e- e o m le els. We e e o he be o e and a e expansion p ices and u ili y as P0, u0 and P1 − s, u1,
espec i ely. The equi alen a ia ion is hen gi en by e(P0, u1) − e(P0, u0), whe e e() deno es he
expendi u e unc ion.
46 The equi alen a ia ion equals he compensa ing a ia ion gi en he wo ke ’s linea indi ec u ili y
unc ion and is gi en by
nP P P U
˜
1
P P P U
˜
0o
ij ij
EV = γ ×log[ ( exp( ))1−ρ] − log[ ( exp( ))1−ρ] , whe e
i g j∈gγ×(1−ρ) i g j∈gγ×(1−ρ)
U
˜
= U
ij ij − ij
.
IAB-Discussion Pape 03|2025 38
To wha ex en a e hese wel a e gains om a heal h insu ance expansion likely o gene alize
o o he con ex s? We use ou model o help us hink abou his key policy ques ion.
Speci ically, we explo e he sensi i i y o ou ne wel a e numbe s o he gene osi y o
unemploymen insu ance (UI) bene i s, he income ax a e, and he p oduc i i y o he
ou side sec o . We ocus ou discussion on ela i ely la ge changes in hese p imi i es and
p esen analogous esul s on mo e ma ginal changes in Appendix A2.4.
In ou i s coun e ac ual, we e isi he wi h- e sus wi hou -expansion ou comes a e
se ing UI bene i s o ze o.47 We ind a lowe unemploymen a e bu also a lowe labo
o ce pa icipa ion a e sugges ing ha mo e indi iduals e ain om sea ching o jobs
al oge he . Columns (4)-(6) o Table 9 documen hese ac s. We ind ha he insu ance
expansion in his con ex has simila e ec s on social wel a e sugges ing ha he gene osi y
o UI bene i s plays a seconda y ole in d i ing he wel a e esul s.48
We nex conside a 50 pe cen d op in he income ax a e. We es ima e a la ge labo o ce
pa icipa ion in he no-insu ance coun e ac ual and ind ha unde hese ci cums ances
he insu ance expansion is less success ul in inc easing labo o ce pa icipa ion u he ; see
columns (7)-(9). In he con ex wi h lowe ax a es and highe labo - o ce pa icipa ion,
expanding insu ance leads o a signi ican sha e o new nu sing home employees o be
hi ed away om p oduc i e jobs in he ou side sec o . In his en i onmen wi h much lowe
axes, we no longe ind ha he insu ance expansion pays o i sel and es ima e an o e all
wel a e loss o 10 million EUR pe mon h.
Finally, we simula e he in e ac ion be ween p oduc i i y g ow h in he ou side sec o and
he insu ance expansion. As wages inc ease in he ou side sec o , his also d i es up wages
in nu sing home ca e. Wi hou p oduc i i y g ow h in he nu sing home ca e sec o , wage
inc eases will lead o highe ou pu p ices, illus a ing Baumol’s cos disease
(Baumol/Bowen, 1965). We conside a 50 pe cen inc ease in p oduc i i y and again
es ima e subs an ially highe labo o ce pa icipa ion, see columns (10)-(12).49 In con as
o he ax coun e ac ual, we now es ima e subs an ially highe nu sing home wages and, as
a esul , highe nu sing home ou pu p ices. We ind smalle inc eases in nu sing home
employmen and la ge inc eases in p ices, implying ha he insu ance expansion is less
e ec i e in expanding nu sing home jobs. We again es ima e smalle gains in labo o ce
pa icipa ion. O e all we es ima e smalle wel a e gains om he insu ance expansion bo h
47 To ensu e weakly posi i e job inding and acancy illing a es, we g adually inc ease he ma ching
pa ame e λ in 1% inc emen s o he espec i e skill-sec o ype un il he job inding a es o each skill and
sec o a e bounded be ween 0 and 100%.
48 He e we no e, howe e , ha ou model can only accoun o a small sha e o he e o m-induced decline in
unemploymen in ou baseline model, he eby possibly unde s a ing he ole o UI bene i s; see Sec ion
A2.2, Table A8.
49 The calib a ed p oduc i i y inc ease e lec s he cumula i e g ow h in p oduc i i y om he mid-1990s o
he p esen . This se es o assess how di e ences in p oduc i i y be ween hese en i onmen s migh
in luence he wel a e e ec s o he LTC e o m.
IAB-Discussion Pape 03|2025 39
6 Conclusion
in absolu e e ms and e en mo e s ikingly in ela i e e ms—social wel a e in column (10)
exceeds baseline wel a e in column (1) by 77 pe cen .
In summa y, we ind ha ou wel a e es ima es a e sensi i e o he economic en i onmen
in impo an and policy- ele an ways. Lowe income axes o highe p oduc i i y in he
ou side sec o ende policy e o s o s imula e he economy less e ec i e. We also ind
smalle posi i e iscal ex e nali ies and gains in he labo ma ke su plus in economic
en i onmen s wi h highe labo o ce pa icipa ion o begin wi h. When wo ke s de ine he
sho e ma ke side, policy-induced employmen expansions will la gely in ol e he
ealloca ion o wo ke s ac oss sec o s o he economy, mu ing po en ial wel a e gains in he
labo ma ke (Baicke /Chand a, 2012).
A ow (1963) and Felds ein (1971, 1977) hypo hesized ha demand-side mo al haza d
induced by heal h insu ance can igge supply-side expansions in heal hca e ma ke s.
Cap u ing his phenomenon empi ically has p o en challenging. In his pape , we combine
unusually comp ehensi e adminis a i e labo ma ke da a wi h a a ely obse ed
la ge-scale, uni e sal insu ance ollou —Ge many’s 1995 in oduc ion o na ional long- e m
ca e (LTC) insu ance— o examine how a public insu ance expansion can ealloca e wo ke s
ac oss sec o s.
We ind ha he LTC insu ance ollou led o a d ama ic expansion in LTC employmen . While
many mo e wo ke s we e hi ed in o LTC jobs, a e age wages did no inc ease, consis en
wi h conside able slack in he labo ma ke . No ably, his su ge in LTC hi ing d ew in
wo ke s who likely aced obs acles o employmen , such as hose wi h less expe ience,
lowe skill le el, o ex ended gaps in wo k his o ies due o long- e m unemploymen o
non-pa icipa ion in he labo o ce. By p o iding “good” jobs o o lowe -skilled wo ke s,
he e o m aligns wi h an aim o indus ial policy ecen ly highligh ed by Rod ik (2022).
We u he ind ha he e ec o insu ance expansion on he labo ma ke was no limi ed o
LTC wo ke s. Ins ead, we es ima e a educ ion in o e all local unemploymen , and
inc eased labo o ce pa icipa ion. To he bes o ou knowledge, his p o ides some o he
i s e idence o ac o subs i u ion be ween sec o s o he economy ollowing a heal h
insu ance expansion.
Ou no ma i e analysis conside s he ex e nal alidi y o he Ge man expe ience and leads
o wo main policy- ele an akeaways. Fi s , public spending on p og ams o olde adul s
IAB-Discussion Pape 03|2025 40
P oppe , Ca ol; Van Reenen, John (2010): Can Pay Regula ion Kill? Panel Da a E idence on
he E ec o Labo Ma ke s on Hospi al Pe o mance. In: Jou nal o Poli ical Economy, Vol.
118, No. 2, p. 222–273.
Rod ik, Dani (2022): An Indus ial Policy o Good Jobs. The Hamil on P ojec , B ookings
Ins i u ion.
Ro hgang, Heinz (1997): Ziele und Wi kungen de P lege e siche ung: Eine Öekonomische
Analyse. In: Campus Ve lag.
Schmiede , Johannes F.; Von Wach e , Till; Bende , S e an (2012): The E ec s o Ex ended
Unemploymen Insu ance o e he Business Cycle: E idence om Reg ession Discon inui y
Es ima es o e 20 Yea s. In: The Qua e ly Jou nal o Economics, Vol. 127, No. 2, p. 701–752.
Schwand , Hannes; on Wach e , Till M. (2020): Socio-Economic Decline and Dea h: The
Li e-Cycle Impac s o Recessions o Labo Ma ke En an s. NBER Wo king Pape No. 26638.
Sojou ne , Aa on J.; F andsen, B igham R.; Town, Robe J.; G abowski, Da id C.; Chen,
Min M. (2015): Impac s o Unioniza ion on Quali y and P oduc i i y: Reg ession
Discon inui y E idence om Nu sing Homes. In: Indus ial and Labo Rela ions Re iew,
Vol. 68, No. 4, p. 771–806.
S aige , Douglas O.; Spe z, Joanne; Phibbs, Cia an S. (2010): Is he e Monopsony in he
Labo Ma ke ? E idence om a Na u al Expe imen . In: Jou nal o Labo Economics, Vol. 28,
No. 2, p. 211–236.
S a is ical O ice o Ba a ia (1999): P legebe ich Ba a ia 1999. Table: “P lege: K eise,
P legebedü ige nach A des P legeheims, P legegeldemp änge , Jah ”. Las accessed
Augus 2024.
S a is ical O ice o Ba a ia (1993): Regionalbe ich 1993.
S a is ische Äm e des Bundes und de Lände (2021a): S a is ik de S e be älle. Las
accessed Sep embe 2021.
S a is ische Äm e des Bundes und de Lände (2021b): Einkommen de p i a en Haushal e
in den k eis eien S äd en und Landk eisen de Bundes epublik Deu schland 1995 bis 2019,
Reihe 2, Band 3.
S a is isches Bundesam (2013): S a is ik de Sozialhil e - Hil e zu P lege.
S a is isches Bundesam (1999a): P legeheime, Ve gü ung, u.a. nach Region (1999-2015).
Las accessed Augus 2024.
S a is isches Bundesam (1999b): P leges a is ik 1999 - P lege im Rahmen de
P lege e siche ung. 2. Ku zbe ich : Lände e gleich - P legebedü ige.
S a is isches Bundesam (1993): Reihe 2 Sozialhil e, Fachse ie 13, 1993.
IAB-Discussion Pape 03|2025 47

S e ens, Ann H.; Mille , Douglas L.; Page, Ma ianne E.; Filipski, Ma eusz (2015): The Bes o
Times, he Wo s o Times: Unde s anding P o-Cyclical Mo ali y. In: Ame ican Economic
Jou nal: Economic Policy, Vol. 7, No. 4, p. 279–311.
W igh , Randall; Ki che , Philipp; Julien, Benoî ; Gue ie i, Ve onica (2021): Di ec ed Sea ch
and Compe i i e Sea ch Equilib ium: A Guided Tou . In: Jou nal o Economic Li e a u e,
Vol. 59, No. 1, p. 90–148.
IAB-Discussion Pape 03|2025 48
Figu e 1.: In oduc ion o Uni e sal LTC Insu ance and he Nu sing Home Ma ke
High Exposu e
Low Exposu e
10
20
30
40
50
1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Yea
High Exposu e
Low Exposu e
.7
.8
.9
1
1.1
1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Yea
(a) NH Wo ke s pe 1,000 65+ Popula ion (b) NH Es ablishmen s pe 1,000 65+ Popula ion
-20
0
20
40
60
80
1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Yea
Poin es ima es
-.5
0
.5
1
1.5
1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Yea
Poin es ima es
(c) NH Wo ke s, E en S udy (d) NH Es ablishmen s, E en S udy
No es: Panels in he op ow plo he numbe o egula nu sing home wo ke s (A) and nu sing
home es ablishmen s (B) pe 1,000 indi iduals age 65 and o e , on a e age ac oss coun ies
o each yea 1975 o 2008. The a e age in each yea is compu ed sepa a ely o he g oup o
Wes Ge man coun ies wi h ( egion-le el) exposu e a iable E abo e (“high exposu e”) and
below (“low exposu e”) he le el o exposu e o he median coun y. E akes alues om 0
o 1 and measu es he sha e o po en ial long- e m ca e pa ien s who did no ha e public as-
sis ance o long- e m ca e p io o he ollou o uni e sal LTC insu ance (mean and median
o E =0.69) All coun ies wi h exposu e le el a he median a e assigned o he below median
g oup. Fo isual cla i y, bo h ime-se ies a e no malized o he agg ega e mean o he y-axis
a iable ac oss all coun ies in 1993 (see Figu e A11 o aw ime-se ies wi hou he no mal-
iza ion). Appendix A1.2 p o ides he de ini ion o nu sing homes and “ egula ” wo ke s. E
is de i ed in Equa ion 1 and i s geog aphic a ia ion is isualized in Figu e A4. Panels in he
bo om ow display λ coe icien s and 95% con idence in e als om es ima ing he speci i-
ca ion in Equa ion 2 wi h he numbe o egula nu sing home wo ke s (C) and nu sing home
es ablishmen s (D) pe 1,000 indi iduals age 65 and abo e in a coun y as an ou come. λ a e
coe icien s ha mul iply exposu e a iable
E ; λ is no malized o ze o in he p e- e o m yea
= 1993. Mo e eg ession de ails a e epo ed in columns (1) and (4) o Table 2. Reg essions
a e es ima ed on he nu sing home sample; see column (2) o Table 1 o summa y s a is ics.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 49
Figu e 2.: Changes in Nu sing Home Wages and Wo ke So ing
(a) Incumben and S a ing Daily Wage (EUR)
1. Incumben Wages,
No Con ols
P e-Mean: 4.58
2. Incumben Wages,
Indi idual FEs
P e-Mean: 4.58
3. S a ing Wages,
No Con ols
P e-Mean: 4.38
4. S a ing Wages,
Expe ience Con ols
P e-Mean: 4.38
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2
Implied Re o m Impac
(b) Cha ac e is ics, New Hi es (c) O igins, New Hi es
1. Age (Di ided By 10)
2. Ge man Na ional
3. Female
4. Abi u (A-Le el Equi .)
5. NH App en ice a -1
6. Pa -Time a
7. Rolling 15-Yea LM
Expe ience (Di ided By 10)
8. Rolling 15-Yea NH
Expe ience (Di ided By 10)
-0.6 -0.4 -0.2 0.0 0.2
Implied Re o m Impac
1. Employed, in
Heal hca e a -1
2. Employed, No in
Heal hca e a -1
3. Unemployed a -1
4. Tempo a ily
No in LF a -1
5. Ne e in LF
Be o e
-0.2 -0.1 0.0 0.1 0.2 0.3
Implied Re o m Impac
No es: This panel displays agg ega e ea men e ec s δ97−08 and 95% con idence in e als om es i-
ma ing he di e ence-in-di e ences speci ica ion in Equa ion 3. The coe icien measu es he e ec o
uni e sal long- e m ca e insu ance expansion pe uni o exposu e o expansion, E . E akes alues
om 0 o 1 and measu es he sha e o po en ial long- e m ca e pa ien s who did no ha e public assis-
ance o long- e m ca e p io o he ollou o uni e sal LTC insu ance (mean and median o E =0.69).
E is de i ed in Equa ion 1 and i s geog aphic a ia ion is isualized in Figu e A4. The op panel (A)
displays he ea men e ec s on log egula ull- ime daily wages o nu sing home incumben s and
new hi es. A new hi e in yea is de ined as an indi idual no in nu sing home employmen in yea
− 1. Reg ession de ails, including coe icien s o he e en s udy e sion, a e epo ed in Table 3.
In he bo om ow, he ou comes in panel (B) a e he cha ac e is ics and in panel (C) he immedia e
employmen his o y o new egula nu sing home hi es. Tables 4 and 5 epo eg ession de ails o
panels (B) and (C), espec i ely.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 50
Figu e 3.: Changes in O e all Employmen
1. Sha e Unemploymen
2. Sha e Long-Te m
Unemploymen
3. Log Labo Fo ce Size
4. Log Labo Fo ce ha
is Rejoining he LF
5. Log A e age Wage
6. Log To al Wage Bill
-.01 0 .01 .02 .03 .04
Implied Re o m Impac
No es: This panel displays agg ega e ea men e ec s and 95% con idence in e als om es ima ing
he syn he ic di e ence-in-di e ences (SDID) speci ica ion e sion o Equa ion
3, ollowing A khangel-
sky, Dmi y and A hey, Susan and Hi shbe g, Da id A. and Imbens, Guido W. and Wage , S e an (2021)
me hodology. The unde lying da a is he sample o all Ge man labo ma ke biog aphies—Labo Ma -
ke Sample (LMS) summa ized in column (3) o Table 1. The ea men e ec coe icien measu e he
e ec o mo e exposu e o uni e sal long- e m ca e insu ance expansion on ou comes speci ied on
he y-axis. We use a bina y measu e o exposu e o he e o m, which de ines egions wi h an abo e-
median exposu e a iable E as ea ed uni s (i.e. mo e exposed) and coun ies a o below median
exposu e as con ol uni s. E is de i ed in Equa ion 1 and i s geog aphic a ia ion is isualized in
Figu e A4. “Long- e m unemploymen ” is de ined as unemploymen obse ed on in a leas wo con-
secu i e yea s. Table 6 epo s eg ession de ails o all ou comes.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 51
Figu e 4.: In oduc ion o Uni e sal LTC Insu ance and O e all Employmen
(a) Sha e Unemployed (b) Log Labo Fo ce Size
High Exposu e
Low Exposu e
SDID ATT:
-0.005 (0.001)
.04
.06
.08
.1
.12
1985 1988 1991 1994 1997 2000 2003
Yea
High Exposu e
Low Exposu e
SDID ATT:
0.016 (0.005)
8.4
8.45
8.5
8.55
8.6
1985 1988 1991 1994 1997 2000 2003
Yea
(c) Sha e Unemployed, E en S udy (d) Log Labo Fo ce Size, E en S udy
-.015
-.01
-.005
0
.005
1985 1988 1991 1994 1997 2000 2003
Yea
Poin es ima es
-.01
0
.01
.02
.03
.04
1985 1988 1991 1994 1997 2000 2003
Yea
Poin es ima es
No es: Panels in he op ow display ea men and ma ched con ol ime se ies om syn he ic
di e ence-in-di e ence (SDID) speci ica ions, ollowing A khangelsky, Dmi y and A hey, Susan and
Hi shbe g, Da id A. and Imbens, Guido W. and Wage , S e an (2021) me hodology, o he a e o un-
employmen in a coun y (A) and he (log) size o he labo o ce in a coun y (B). The unde lying da a is
he sample o all Ge man labo ma ke biog aphies—Labo Ma ke Sample (LMS) summa ized in col-
umn (3) o Table 1. The sha e unemployed uses he size o he labo o ce in he denomina o , which
is de ined as he coun o all employed and unemployed indi iduals. The ea men g oup a e Wes
Ge man coun ies wi h ( egion-le el) exposu e a iable E abo e (“high exposu e”) and below (“low
exposu e”) he le el o exposu e o he median coun y. E akes alues om 0 o 1 and measu es he
sha e o po en ial long- e m ca e pa ien s who did no ha e public assis ance o long- e m ca e p io
o he ollou o uni e sal LTC insu ance (mean and median o E =0.69) All coun ies wi h exposu e
le el a he median a e assigned o he below median g oup. Fo isual cla i y, bo h ime-se ies a e
no malized o he agg ega e mean o he y-axis a iable ac oss all coun ies in 1993 (see Figu e A11 o
he ime-se ies wi hou he no maliza ion). The bo om panel shows e en s udy cha s o he same
ou comes, using b = 100 boo s ap d aws. Coe icien s a e no malized o ze o in he p e- e o m yea
= 1993. Mo e eg ession de ails a e epo ed in Table 6.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 52

Figu e 5.: In oduc ion o Uni e sal LTC Insu ance and Old-Age Mo ali y
(a) Dea hs pe 100 Age 75+ (b) Dea hs pe 100 Age 75+, E en S udy
High Exposu e
Low Exposu e
7.5
8
8.5
9
9.5
10
1991 1994 1997 2000 2003 2006
Yea
-3
-2
-1
0
1
2
1991 1994 1997 2000 2003 2006
Yea
Poin es ima es
No es: Panel a plo s he aw ime se ies o mo ali y pe 100 age 75 and o e popula ion, on a e age
ac oss coun ies o each yea 1991 o 2008. The a e age in each yea is compu ed sepa a ely o he
g oup o Wes Ge man coun ies wi h ( egion-le el) exposu e a iable E abo e (“high exposu e”) and
below (“low exposu e”) he le el o exposu e o he median coun y. E akes alues om 0 o 1 and
measu es he sha e o po en ial long- e m ca e pa ien s who did no ha e public assis ance o long-
e m ca e p io o he ollou o uni e sal LTC insu ance (mean and median o E =0.69) All coun ies
wi h exposu e le el a he median a e assigned o he below median g oup. Fo isual cla i y, bo h
ime-se ies a e no malized o he agg ega e mean o he y-axis a iable ac oss all coun ies in 1993.
E is de i ed in Equa ion 1 and i s geog aphic a ia ion is isualized in Figu e A4. Panel b display λ
coe icien s and 95% con idence in e als om es ima ing he speci ica ion in Equa ion 2 wi h he
same mo ali y measu e as he ou come. Mo e eg ession de ails and al e na i e speci ica ions a e
epo ed in Table 7.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 53
Figu e 6.: Cos o Employmen and Employmen Ra es by Sec o and Expe ience
(a) Log Mon hly Employmen Cos (EUR) (b) Employmen Type o Popula ion Ra io
Nu sing
Homes
Ou side
Sec o
7.8
8
8.2
8.4
8.6
0 3 6 9 12 15 18
Yea s o Heal hca e Expe ience
Da a
Model Fi
Nu sing
Homes
Ou side
Sec o
0
.2
.4
.6
.8
0 3 6 9 12 15 18
Yea s o Heal hca e Expe ience
Da a
Model Fi
No es: Panel 6a plo s a e age log mon hly cos o employmen ( ull- ime g oss wages and employe
social insu ance con ibu ions, in EUR) o all ull- ime wo ke s age 20–64 in he Labo Ma ke Sample
in 1999 (i.e. a e he ull insu ance expansion). Fo each wo ke , we calcula e yea s o p io employ-
men in he heal hca e sec o be o e 1999. Wage a e ages a e hen compu ed wi hin one-yea bins
o heal hca e expe ience, sepa a ely o nu sing homes employees ( ed lines) and wo ke s in o he
sec o s (black lines). Solid lines epo obse ed da a; dashed lines show model p edic ions (Sec ion
5.2). Panel 6b, cons uc ed analogously, displays he ull- ime equi alen employmen sha es in nu s-
ing homes and o he sec o s in 1999, using all wo ke s age 20–64. Full- ime equi alen s a e calcula ed
by conside ing wo pa - ime jobs as one ull- ime posi ion. Employmen ca ego ies omi ed om he
exhibi a e o mal unemploymen and being ou o he labo o ce. Re e o Sec ion
A1.2 o de ini ions
o he heal hca e sec o and nu sing homes.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 54
Table 1.: Summa y S a is ics
Nu sing Home Sample Labo Ma ke Sample
All Spells Nu sing Home Spells All Spells
1975-2008 1975-2008 1985-2004
(1) (2) (3)
No. o Indi idual-Yea Obse a ions 25,273,408 10,184,624 44,096,416
Indi iduals
No. o Unique Indi iduals 1,639,576 1,639,576 3,787,607
Mean Age 36.2 40.2 38.0
% Female 78.2 81.8 41.5
% Ge man 94.5 95.0 92.9
% High School Educa ion (Abi u ) 7.32 7.01 8.00
% in Heal hca e Sec o 58.2 100.0 9.65
% Unemployed 8.65 0 7.75
Mean 15-Yea Labo Ma ke Expe ience (y s) 8.24 8.69 9.18
Mean 15-Yea Nu sing Home Expe ience (y s) 3.50 5.80 0.14
% Pa -Timea 25.0 29.2 13.3
Mean Daily Wage (EUR)b
All Obse a ions 76.7 78.6 96.4
Nu sing Home Obse a ions 78.6 78.6 83.0
Es ablishmen s
No. o Unique Es ablishmen s
Any 991,814 22,629 1,748,171
Nu sing Homes 22,629 22,629 13,880
O Nu sing Home Employmen Spells, Sha e in
Fo -P o i Nu sing Home 0.27 0.27 0.27
Chu ch-Owned Nu sing Home 0.59 0.59 0.59
Publicly-Owned Nu sing Home 0.14 0.14 0.14
a Condi ional on being employed.
b In cons an 2020 Eu os.
No es: The able epo s summa y s a is ics o wo main analy ic samples “Nu sing Home Sam-
ple (NHS)” and “Labo Ma ke Sample (LMS).” NHS is an ex ac om he annualized uni e se o
he Ge man In eg a ed Employmen Biog aphies o yea s 1975-2008. The ex ac con ains ull la-
bo ma ke biog aphies o indi iduals who we e employed in a nu sing home in a leas one yea
in 1975-2008. Employmen spells a e annualized by aking he spell obse ed on June 30 h o a
gi en yea . Nu sing homes a e de ined as es ablishmen s wi h WZ73 indus y codes o p i a e and
o -p o i ins i u ions o “homes” (710), p i a e and no - o -p o i homes (711), and homes in pub-
lic owne ship (712). LMS is a 10% d aw o he annualized uni e se o labo ma ke biog aphies o
yea s 1985 o 2004. In bo h samples, indi idual-yea obse a ions a e excluded i he place o em-
ploymen is in he ( o me ) Eas Ge many, Be lin, o B emen.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 55
Table 2.: E en S udy Resul s: Agg ega e Response
Ou come (pe 1,000 Age 65+ Popula ion)
Wo ke s Full-Time Pa -Time Fi ms
(1) (2) (3) (4)
Pooled Coe icien s
δ97−08 44.4 21.9 22.5 0.62
(8.35) (4.87) (4.90) (0.17)
E en S udy Coe icien s
1-Yea E ec , λ1997 24.4 12.3 12.1 0.30
(6.74) (4.21) (3.25) (0.14)
3-Yea E ec , λ1999 33.3 18.8 14.6 0.39
(7.91) (4.79) (4.07) (0.17)
5-Yea E ec , λ2001 42.9 24.8 18.1 0.54
(8.69) (5.27) (4.74) (0.17)
10-Yea E ec , λ2006 53.4 22.6 30.7 0.89
(9.57) (5.55) (6.24) (0.23)
Implied Impac
Using In-sample Va ia iona 4.02 1.98 2.03 0.06
Agg ega e Impac , Wes Ge manyb 38,979 19,253 19,726 542.1
Le el o Ou come in 1993
Mean 33.1 23.9 9.19 1.04
S.D. 13.8 10.3 4.57 0.47
Yea s 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008
No. o Obse a ions 10,948 10,948 10,948 10,948
a Mul iplies δ97−08 by 9.1 pe cen age poin di e ence in mean exposu e be ween coun ies wi h
abo e and below he median le el o exposu e
b Scales es ima es by 9,699 housand people age 65+ in Wes Ge many (excluding Be lin and B e-
men) in 1993.
No es: This able epo s he esul s o es ima ing he e en s udy speci ica ion in Equa ion (2) and
he pooled di e ence-in-di e ences speci ica ion in Equa ion (3). The coe icien s measu e he im-
pac o long- e m ca e insu ance expansion—pe uni o exposu e (E )—on he ou come a iables as
speci ied in column i les. Exposu e measu e E is de i ed in Equa ion (1). E akes alues om 0 o
1 and measu es he sha e o po en ial long- e m ca e pa ien s who did no ha e public assis ance
o long- e m ca e p io o he ollou o uni e sal LTC insu ance (mean and median o E =0.69).
Ou come a iables a e: he numbe o egula nu sing home wo ke s, in o al (column 1), and sep-
a a ely by pa - ime and ull- ime s a us (columns 3 and 4), as well as he numbe o nu sing home
es ablishmen s (column 4) pe 1,000 indi iduals age 65 and olde in a coun y. E en s udy esul s
in columns (1) and (4) a e isualized in Figu e 1; he es ima es in columns (2) and (3) a e isualized
in Figu e A12. All speci ica ions include coun y and yea ixed e ec s. S anda d e o s clus e ed a
he coun y-le el a e epo ed in pa en heses.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 56
IAB-Discussion Pape 03|2025
Table 9.: Coun e ac uals and Wel a e
Baseline
100% Reduc ion in
UEB
50% Reduc ion in
Income Tax
50% Inc ease in
P oduc i i y
Insu ance
Di
Insu ance
Di
Insu ance
Di
Insu ance
Di No Yes No Yes No Yes No Yes
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Panel A: Employmen and Ea nings
Nu sing Home Employmen (million FTE) 0.18 0.34 0.16 0.19 0.34 0.16 0.23 0.47 0.24 0.25 0.36 0.11
To al Labo Cos s o Employe s (billion EUR/mon h) 70.6 73.6 2.99 68.8 71.9 3.04 111.5 114.4 2.81 178.7 182.1 3.36
Unemploymen (million FTE) 1.69 1.66 -0.03 0.73 0.72 -0.006 1.85 1.80 -0.05 0.88 0.85 -0.04
Labo Fo ce (million FTE) 19.4 19.9 0.50 18.6 19.1 0.53 30.3 30.5 0.28 30.8 31.0 0.20
Panel B: Wel a e (billion EUR/mon h)
Consume Wel a e 81.0 80.3 -0.69 81.7 81.0 -0.69 130.2 129.1 -1.10 203.0 201.3 -1.72
Wo ke Wel a e 148.7 149.2 0.59 147.7 148.3 0.60 172.1 173.5 1.35 175.4 176.5 1.12
Go e nmen Su plus 33.4 34.3 0.88 33.9 34.8 0.88 25.9 25.7 -0.25 87.8 88.8 0.99
LTC Subsidy Spending -0.42 -1.05 -0.63 -0.43 -1.06 -0.63 -0.42 -1.41 -0.99 -0.57 -1.28 -0.71
UI Spending -1.31 -1.29 0.02 0 0 0 -1.44 -1.40 0.04 -0.68 -0.66 0.03
Tax Re enues 35.2 36.7 1.49 34.3 35.8 1.51 27.8 28.5 0.70 89.1 90.7 1.67
To al Wel a e 263.1 263.9 0.78 263.3 264.1 0.78 328.2 328.2 -0.01 466.2 466.6 0.38
No es: This able summa izes he model-simula ed e ec s o a uni e sal LTC insu ance expansion on a numbe o ou comes in di e en economic en i onmen s.
Fo each en i onmen , we p esen ou comes (i) wi hou uni e sal LTC insu ance, whe e LTC suppo is p o ided ia Hil e zu P lege only and (ii) wi h uni e sal
LTC insu ance co e age. Uni e sal LTC insu ance is modeled as a subsidy o s=1,200 EUR pe mon h ha can only be applied o he cos o nu sing home ca e
by indi iduals who would ha e o he wise had o pay he ull p ice o ca e ou -o -pocke . This way o modeling he subsidy main ains he a ailabili y o Hil e
zu P lege. We also p esen he di e ence be ween (i) and (ii). We conside ou en i onmen s. Columns (1)-(3) p esen esul s o ou baseline economic
en i onmen , columns (4)-(6) e isi he esul s a e educing he le el o unemploymen insu ance bene i s (UEB) om 775 EUR pe mon h o ze o. Columns
(7)-(9) e isi he esul s a e educing he income axes and con ibu ions by 50% ( om abou 50 p.p. o 25 p.p.). Columns (10)-(12) e isi he esul s a e
inc easing he p oduc i i y in he ou side sec o o each skill segmen , φou , (ehc) by 50%. Fo each en i onmen , we epo se e al desc ip i e measu es o
he esul ing alloca ion (Panel A): nu sing home employmen , o al g oss wage paymen s by employe s in all sec o s o he economy, he coun o wo ke s in
unemploymen , and he numbe o indi iduals in he labo o ce (employed o unemployed). Panel B epo s he co esponding no ma i e measu es. We epo
he le el o consume wel a e, wo ke wel a e, and ne go e nmen spending, which cap u es iscal ex e nali ies. Ex a go e nmen ou lay is he spending on LTC
insu ance. The go e nmen can also spend mo e o less on unemploymen insu ance bene i s and can collec mo e o less e enue in axes and social insu ance
con ibu ions. To al wel a e add consume and wo ke wel a e along wi h changes in go e nmen spending. We do no weigh he change in go e nmen
spending by a cos o public unds.
Sou ce: Own calcula ions.
63

Appendix
Online Appendix o:
Heal h Insu ance as Economic S imulus?
E idence om Long-Te m Ca e Jobs
Ma in Hackmann
Jö g Heining
Roman Klimke
Ma ia Polyako a
Holge Seibe
IAB-Discussion Pape 03|2025 64
A1 Da a Appendix
A1.1. P ocessing o he Raw Da a
The p ima y da a sou ce o his pape is he In eg a ed Employmen Biog aphies (IEB)
da abase, which con ains he uni e se o employmen and unemploymen spells o
wo ke s subjec o social secu i y con ibu ions in Ge many. We use da a spanning he
yea s 1975 o 2008. The IEB co e s indi iduals ac oss i e employmen ca ego ies: 1)
employmen subjec o social secu i y (included since 1975), 2) ma ginal pa - ime
employmen (included since 1999), 3) unemploymen bene i eceip unde he Ge man
Social Code Books III (included since 1975) o II (included since 2005), 4) egis e ed
job-seeke s (included since 2000), and 5) pa icipan s in ac i e labo ma ke p og ams
(included since 2000). All spells a e eco ded wi h day-le el p ecision.
Fo each employmen spell, he da abase eco ds he indus y code o he es ablishmen .
To ensu e consis ency o e ime, we ha monize indus y codes o he h ee-digi WZ73
classi ica ion, ollowing he me hodology o Ebe le e al. (2011). Fo spells wi h missing
WZ73 codes, we ex apola e alues wi hin es ablishmen s ac oss ime. When his
ex apola ion is no easible, we impu e missing WZ73 codes using alues om la e
indus y classi ica ions (WZ93 and WZ03), elying on he mos equen ly obse ed WZ73
code du ing o e lapping yea s.
We cons uc wo analy ic samples om he IEB. The “Nu sing Home Sample (NHS)”
comp ises comple e pe son-his o ies o indi iduals employed in nu sing homes o a leas
one day du ing 1975–2008. We de ine a nu sing home as an es ablishmen wi h one o he
h ee ime-consis en WZ73 indus y codes 710, 711, and 712, o p i a e and o -p o i
ins i u ions o homes (710); p i a e, no - o -p o i homes (711); and homes in public
owne ship (712). We also ex ac he “Labo Ma ke Sample (LMS)”, a andom 10% d aw o
comple e pe son-his o ies om he comple e IEB popula ion du ing 1975 o 2008.
We subse he NHS and LMS o include only employmen spells (o igina ing in he IEB’s
“BeH Employee His o y’’ sou ce da a se ) and unemploymen spells ( om “LEH Bene i
Recipien His o y’’ o “LHG Unemploymen Bene i II Recipien His o y’’). We exclude spells
wi h ze o daily wage o bene i a e, ma ginal pa - ime wo ke s (iden i ied ia alues o he
“employmen s a us [e ws a ]” a iable equal o 109 o 209), and no i ica ions o lump-sum
paymen s (“ eason o cancella ion/no i ica ion/ e mina ion [g und]” wi h alue 154). Fo
included spells, we impu e missing alues o he wo kplace loca ion a iable [ao_k eis]
du ing non-employmen pe iods wi h he indi idual’s mos ecen non-missing alue. I
impu a ion is no possible, we d op he co esponding obse a ions om he analy ic
samples.
IAB-Discussion Pape 03|2025 65
We agg ega e he NHS and LMS o he indi idual-yea le el by selec ing he employmen
s a us o he spell ha includes June 30 h o each yea . Fo unemploymen spells ha span
mul iple ins ances o June 30, we spli he spells ollowing he app oach o
Ebe le/Schmucke (2019). I duplica es pe sis , we e ain he spell wi h he highes epo ed
daily bene i amoun . A e coding a iables o in e es , as desc ibed in Sec ion A1.2, we
subse he LMS (bu no he NHS) o he yea s 1985 o 2004.
A1.2. Va iable De ini ions
1. (Un)employmen : Indi iduals a e classi ied as unemployed in yea i hei
employmen s a us [e ws a ] in ha yea is coded as 1 (long- e m unemploymen
assis ance pos -2005 o adul s able o wo k [VeP]), 11 (Unemploymen bene i s [ALG]),
12 (Unemploymen assis ance [ALHI]), o 13 (Main enance allowance [UHG]). All o he
obse a ions wi h a eco d in he NHS o LMS a e classi ied as employed.
2. Regula employmen : Following IAB con en ions, egula employees a e indi iduals
whoseemploymen s a us [e ws a ] is coded as 101 (employees subjec o social
secu i y wi h no special ea u es), 140 (seamen), o 143 (ma i ime pilo s). Non- egula
employees include app en ices, wo ke s in pa - ime p e- e i emen employmen , and
wo king s uden s.
3. Nu sing homes: De ined as es ablishmen s classi ied unde WZ73 indus y codes 710
(p i a e and o -p o i ), 711 (p i a e, no - o -p o i ), and 712 (public owne ship).
4. Heal hca e sec o : Employmen spells a e classi ied as pa o he heal hca e sec o i
hey occu in es ablishmen s wi h one o he WZ73 indus y codes 710, 711, 712, 780,
781, 782, 783, 784, 880, o i he spell co esponds o one o he KldB 1988 occupa ion
codes [be u ] 841, 842, 844, 851, 852, 853, 854, 855, 856, 857, 861, 862.
5. Labo ma ke expe ience: This a iable measu es he numbe o yea s wi h any
employmen du ing a olling 15-yea look-back window o he yea s 1990–2008.
6. Nu sing home expe ience: Cons uc ed analogously o labo ma ke expe ience bu
speci ic o employmen in nu sing homes.
7. Heal hca e expe ience: A 15-yea olling measu e speci ic o employmen in he
heal hca e sec o , as de ined abo e.
8. New hi es: An indi idual is classi ied as a new hi e in yea i hey a e employed in yea
bu no in yea − 1.
9. Wages: Fo employmen spells, wages a e eco ded in he a iable “Daily wage, daily
bene i a e” [ en gel ]. This ep esen s “ he employee’s g oss daily wages […]
calcula ed om he ixed-pe iod wages epo ed by he employe and he du a ion o
he (unspli ) o iginal no i ica ion pe iod in calenda days” (An oni e al., 2019). We
op-code daily g oss wages a he annual uppe ea nings limi se by s a u o y pension
insu ance egula ions.
IAB-Discussion Pape 03|2025 66
10. Unemploymen bene i s: Fo unemploymen spells, [ en gel ] eco ds daily bene i
amoun s. P io o 1998, hese alues e e o wo king days, while om 1998 onwa d,
hey e e o calenda days.
11. Schooling: We cons uc an indica o a iable o indi iduals who ha e passed he
Abi u , coded as 1 i he schooling a iable [schule] equals 8 o 9 (uppe seconda y
school lea ing ce i ica e, equi alen o A-le els). A second dummy a iable is coded as
1 o indi iduals wi h a bachelo ’s deg ee, co esponding o alues o he e ia y
educa ion a iable [ausbildung] equal o 11 (deg ee om a uni e si y o applied
sciences) o 12 (uni e si y deg ee).
IAB-Discussion Pape 03|2025 67
A1.3. Appendix Figu es and Tables
Figu e A1.: Popula ion Sha e Age 65 and O e in Wes Ge many
.1
.12
.14
.16
.18
.2
1956 1966 1976 1986 1996 2006 2016
No es: This igu e plo s he coun o indi iduals age 65 and olde as a sha e o he o al popula ion o
Wes Ge many by yea . The ime se ies includes Be lin and B emen ha a e excluded in ou p ima y
analysis. The da a sou ce is he Human Mo ali y Da abase (2021).
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 68

Figu e A2.: Public P og am Spending on Long-Te m Ca e (in Billion EUR)
Uni e sal LTC Insu ance
Means- es ed Hil e zu P lege
0
5
10
15
20
1991 1995 1999 2003 2007
No es: This igu e plo s agg ega e public spending on means- es ed long- e m ca e bene i s, Hil e zu
P lege, and uni e sal long- e m ca e (LTC) insu ance. The spending s a is ics co e all o Ge many
(Eas and Wes ). Uni e sal LTC insu ance s a ed co e ing ou pa ien se ices in 1995 and inpa ien
se ices in 1996. These ansi ion yea s a e shaded in g ay. The da a sou ce o expendi u es on Hil e
zu P lege is S a is isches Bundesam (2013), Table D5. The sou ce o expendi u es on uni e sal LTC
insu ance is Bundesminis e ium ü A bei und Soziales (2011), Appendix 3.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 69
Figu e A3.: Geog aphic Va ia ion in Baseline Exposu e and Local Income
(a) E and A e age G oss Income (b) E and A e age Disposable Income
SH
HH
NI
NW
HE
RP
BW
SL
OBay
NBay
OP
OF
MF
UF
Schw
p- alue: 0.920
.6
.65
.7
.75
.8
Baseline Exposu e Measu e (E )
14,000 16,000 18,000 20,000 22,000
Income (Pe Capi a EUR, 1995)
SH
HH
NI
NW
HE
RP
BW
SL
OBay
NBay
OP
OF
MF
UF
Schw
p- alue: 0.941
.6
.65
.7
.75
.8
Baseline Exposu e Measu e (E )
13,000 14,000 15,000 16,000 17,000 18,000
Disposable Income (Pe Capi a EUR, 1995)
No es: This igu e es ima es he ela ionship be ween he baseline measu e o exposu e o LTC in-
su ance ollou and measu es o pe capi a in come in 1995, a he le el o 15 exposu e egions. The
baseline exposu e measu e, deno ed wi h E h oughou he ex , is de i ed in Equa ion (1) and mea-
su es he sha e o indi iduals in need o long- e m ca e, who did no ha e means- es ed suppo o
long- e m ca e se ices in 1993, p io o he ollou o uni e sal LTC insu ance (mean o E = 0.69).
This measu e a ies ac oss 15 di e en Wes Ge man geog aphical egions, isualized in Figu e A4.
The co ela ion o he baseline exposu e measu e wi h income pe capi a, isualized in A3a, is s a is-
ically insigni ican a p = 0.920, and he co ela ion wi h disposable income, displayed in in A3b, is
s a is ically insigni ican a p = 0.941. Income da a ha e been ob ained om (S a is ische Äm e des
Bundes und de Lände , 2021b).
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 70
Figu e A4.: Geog aphic Va ia ion in Exposu e E
(.76,.79]
(.72,.76]
(.68,.72]
(.64,.68]
[.59,.64]
Fo me Eas Ge many
Exposu e E ,
% o po en ial demand
wi hou insu ance
No es: This igu e plo s he alue o exposu e o he ollou o uni e sal long- e m ca e insu ance, E ,
o each coun y. Exposu e a ies a he le el o 15 egions; coun ies in he same egion a e assigned
o he same le el o exposu e. E is de ined as he coun o indi iduals who did no ha e means-
es ed suppo o long- e m ca e se ices (Hil e zu P lege) in 1993 di ided by he coun o people
who ecei ed uni e sal long- e m ca e (LTC) bene i s in 1999, adjus ed by coun y-speci ic 1993-1999
popula ion g ow h. See Equa ion (1) o he exac de i a ion o E .
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 71
Figu e A5.: Al e na i e Ways o Compu ing Exposu e o Insu ance Rollou
(a) Baseline and Al e na i e E (I) (b) Baseline and Al e na i e E (II)
SH
HH
NI
NW
HE
RP
BW
OBay
NBay
OP
OF
MF
UF
Schw
SL
Slope: 0.15 (0.02)
R2: 0.86
.93
.94
.95
.96
.97
Al e na i e Exposu e Measu e (I)
.6 .65 .7 .75 .8
Baseline Exposu e Measu e (E )
SH
HH
NI
NW HE RP
BW
OBay
NBay
OP
OF
MF
UF
Schw
SL
Slope: 0.18 (0.04)
R2: 0.61
.08
.09
.1
.11
.12
.13
.14
Al e na i e Exposu e Measu e (II)
.6 .65 .7 .75 .8
Baseline Exposu e Measu e (E )
(c) Al e na i e E (I) and (II)
SH
HH
NI
NW
HE
RP
BW
OBay
NBay
OP
OF
MF
UF
Schw
SL
Slope: 0.73 (0.27)
R2: 0.24
.09
.1
.11
.12
.13
.14
Al e na i e Exposu e Measu e (II)
.93 .94 .95 .96 .97
Al e na i e Exposu e Measu e (I)
No es: This panel displays bi- a ia e ela ionships be ween di e en ways o compu ing exposu e o
long- e m ca e insu ance ollou E . The baseline E is de i ed in Equa ion (1) and measu es he
sha e o indi iduals in need o long- e m ca e (impu ed om 1999 LTC insu ance claims), who did no
ha e means- es ed suppo o long- e m ca e se ices in 1993. The “Al e na i e Exposu e Measu e
I” eplaces denomina o o exposu e wi h he o e all coun o 65 and olde popula ion in 1993; i.e.
i measu es he sha e o 65 and olde popula ion ha was ecei ing means- es ed bene i s. “Al e -
na i e Exposu e Measu e II” is compu ed as he sha e o indi iduals in 1993 who we e no ecei ing
means- es ed bene i s, bu we e in need o ca e (impu ed om 1999 LTC insu ance claims) among
g ,1993,1999
∗LT CClaims ,1999−HzP ,1993
all indi iduals age 65 and olde . In o he wo ds, E = (mean o
65andOlde P opula ion ,1993
E = 0.103). Solid lines plo he line o bes i . The slope coe icien and he R-squa ed a e epo ed
nex o he line.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 72
Figu e A12.: Uni e sal LTC Insu ance and he Nu sing Home Ma ke , by Type o Employmen
(a) Full-Time NH Wo ke s pe 1,000 65+ Popl (b) Pa -Time NH Wo ke s pe 1,000 65+ Popl
High Exposu e
Low Exposu e
10
15
20
25
30
1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Yea
High Exposu e
Low Exposu e
0
5
10
15
20
1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Yea
(c) Full-Time NH Wo ke s, E en S udy (d) Pa -Time NH Wo ke s, E en S udy
0
10
20
30
40
1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Yea
Poin es ima es
-20
0
20
40
60
1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Yea
Poin es ima es
No es: Panels in he op ow plo he numbe o egula ull- ime nu sing home wo ke s (A) and pa -
ime nu sing home wo ke s (B) pe 1,000 indi iduals age 65 and o e , on a e age ac oss coun ies o
each yea 1975 o 2008. The a e age in each yea is compu ed sepa a ely o he g oup o Wes Ge -
man coun ies wi h ( egion-le el) exposu e a iable E abo e (“high exposu e”) and below (“low expo-
su e”) he le el o exposu e o he median coun y. E akes alues om 0 o 1 and measu es he sha e
o po en ial long- e m ca e pa ien s who did no ha e public assis ance o long- e m ca e p io o he
ollou o uni e sal LTC insu ance (mean and median o E =0.69). All coun ies wi h exposu e le el a
he median a e assigned o he below median g oup. Bo h ime-se ies a e no malized o he agg ega e
mean o he y-axis a iable ac oss all coun ies in 1993. See Appendix A1.2 o he de ini ion o nu sing
homes and “ egula ” wo ke s. E is de i ed in Equa ion 1 and i s geog aphic a ia ion is isualized
in Figu e A4. Panels in he bo om ow display λ coe icien s and 95% con idence in e als om es i-
ma ing he speci ica ion in Equa ion 2 wi h he numbe o egula ull- ime nu sing home wo ke s (C)
and pa - ime nu sing home wo ke s (D) pe 1,000 indi iduals age 65 and abo e in a coun y as an ou -
come. λ a e coe icien s ha mul iply exposu e a iable E ; λ is no malized o ze o in he p e- e o m
yea = 1993. Mo e eg ession de ails a e epo ed in columns (2) and (3) o Table 2. Reg essions a e
es ima ed on he nu sing home sample; see column (2) o Table 1 o summa y s a is ics.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 79

Figu e A13.: E en S udy Resul s: Nu sing Home Wages
(a) New Hi es (b) New Hi es, Expe ience Con ols
-1
-.5
0
.5
1990 1994 1998 2002 2006
Yea
Poin es ima es
-.6
-.4
-.2
0
.2
.4
1990 1994 1998 2002 2006
Yea
Poin es ima es
(c) Incumben s (d) Incumben s, Indi idual FE
-.4
-.3
-.2
-.1
0
.1
1976 1980 1984 1988 1992 1996 2000 2004 2008
Yea
Poin es ima es
-.2
-.1
0
.1
1976 1980 1984 1988 1992 1996 2000 2004 2008
Yea
Poin es ima es
No es: This panel display λ coe icien s and 95% con idence in e als om es ima ing he speci ica-
ion in Equa ion 2 wi h he log ull- ime wage o egula new (Panels A and B) o incumben (panels C
and D) nu sing home wo ke s. Wages a e in cons an 2020 EUR. See Appendix 2 o he de ini ion o
nu sing homes and egula wo ke s. Panel A13b displays esul s om a speci ica ion ha con ols o
yea s o p io employmen (o e all and in nu sing homes). Panel A13d displays esul s om a speci-
ica ion ha includes indi idual ixed e ec s. Coe icien s λ a e no malized o ze o in he p e- e o m
yea = 1993. λ mul iply he exposu e a iable E ha akes alues om 0 o 1 and measu es he
sha e o po en ial long- e m ca e pa ien s who did no ha e public assis ance o LTC p io o he oll-
ou o uni e sal LTC insu ance (mean and median o E =0.69). The geog aphic a ia ion in E is isu-
alized in Figu e A4. All speci ica ions include coun y and yea ixed e ec s. The esul s a e summa izes
in Table 3.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 80
Figu e A14.: In oduc ion o Uni e sal LTC Insu ance and Supply o Nu sing Home Ca e: SDID
(a) Sha e Nu sing Home Wo ke s (b) Sha e Nu sing Home Wo ke s, E en S udy
High Exposu e
Low Exposu e
SDID ATT:
0.002 (0.0004)
.01
.015
.02
.025
1985 1988 1991 1994 1997 2000 2003
Yea
0
.001
.002
.003
.004
1985 1988 1991 1994 1997 2000 2003
Yea
Poin es ima es
No es: This exhibi eplica es he analysis in Figu e 1 using he syn he ic di e ence-in-di e ence
(SDID) speci ica ion (A khangelsky, Dmi y and A hey, Susan and Hi shbe g, Da id A. and Imbens,
Guido W. and Wage , S e an, 2021). The ou come is he sha e o egula nu sing home employees
among all employed and unemployed indi iduals. We use a bina y measu e o exposu e o he e-
o m, which de ines egions wi h an abo e-median exposu e a iable E as ea ed uni s and coun ies
a o below median exposu e as con ol uni s. E akes alues om 0 o 1 and measu es he sha e o
po en ial long- e m ca e pa ien s who did no ha e public assis ance o LTC p io o he ollou o uni-
e sal LTC insu ance (mean and median o E =0.69) as a measu e o local exposu e o he e o m. E
is de i ed in 1 and i s geog aphic a ia ion is isualized in Figu e A4. Panel A displays ea men and
ma ched con ol ime se ies. Bo h ime-se ies a e no malized o he agg ega e mean be ween con ol
and syn he ic ea ed uni s ac oss all coun ies in 1993. Panel B shows he co esponding e en s udy
cha . We use b = 100 boo s ap d aws o compu e he 95% con idence in e als. Coe icien s a e
no malized o ze o in he p e- e o m yea = 1993.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 81
Figu e A15.: In oduc ion o Uni e sal LTC Insu ance and O e all Employmen
(a) Log Popula ion Rejoining he Labo Fo ce (b) Log To al Regula Wage Bill
High Exposu e
Low Exposu e
SDID ATT:
0.019 (0.009)
5.3
5.4
5.5
5.6
5.7
5.8
1985 1988 1991 1994 1997 2000 2003
Yea
High Exposu e
Low Exposu e
SDID ATT:
0.022 (0.007)
12.7
12.8
12.9
13
1985 1988 1991 1994 1997 2000 2003
Yea
(c) Log Popula ion Rejoining he Labo Fo ce, E en
S udy (d) Log To al Regula Wage Bill, E en S udy
-.05
0
.05
.1
1985 1988 1991 1994 1997 2000 2003
Yea
Poin es ima es
-.02
0
.02
.04
.06
1985 1988 1991 1994 1997 2000 2003
Yea
Poin es ima es
No es: Panels in he op ow display ea men and ma ched con ol ime se ies om syn he ic
di e ence-in-di e ence (SDID) speci ica ions es ima ed a he coun y-yea le el. The ou come used
in Panel (A) is he log coun o indi iduals ejoining he labo o ce du ing a gi en index yea ; Panel (B)
uses he log o he o al daily wage bill pe aining o egula employmen . Bo h ime-se ies a e no -
malized o he agg ega e mean be ween con ol and syn he ic ea ed uni s ac oss all coun ies in 1993.
The bo om panel shows e en s udy cha s o he same ou comes. We use b = 100 boo s ap d aws
o compu e he 95% con idence in e als. Coe icien s a e no malized o ze o in he p e- e o m yea
= 1993 We use a bina y measu e o exposu e o he e o m, which de ines egions wi h an abo e-
median exposu e a iable E as ea ed uni s and coun ies a o below median exposu e as con ol
uni s. E akes alues om 0 o 1 and measu es he sha e o po en ial long- e m ca e pa ien s who did
no ha e public assis ance o LTC p io o he ollou o uni e sal LTC insu ance (mean and median o
E =0.69) as a measu e o local exposu e o he e o m. E is de i ed in 1 and i s geog aphic a ia ion
is isualized in Figu e A4.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 82
Figu e A16.: MVPF Es ima es in Con ex
No es: This exhibi o e lays ou MVPF es ima es (in ed) o e a simpli ied e sion o Figu e IV.B in Hen-
d en/Sp ung-Keyse (2020). The o iginal igu e epo s a e age MVPFs and 95% con idence in e als
o se e al ca ego ies o public policies, plo ed as a unc ion o he a e age age o each policy’s ben-
e icia ies.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 83
IAB-Discussion Pape 03|2025 84

IAB-Discussion Pape 03|2025
Table A1.: E en S udy Resul s: Agg ega e Response, Al e na i e Speci ica ions
Ou come (pe 1,000 Age 65+ Popula ion)
A Exposu e Region ( ) Le el S.E. Clus e ed a Region ( ) Le el Coun y-Speci ic Time T end
Wo ke s Full-Time Pa -Time Fi ms Wo ke s Full-Time Pa -Time Fi ms Wo ke s Full-Time Pa -Time Fi ms
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
E en S udy Coe icien s
1-Yea E ec , λ1997 24.7 11.3 13.4 0.36 24.4 12.3 12.1 0.30 25.6 14.9 10.7 0.33
(8.07) (5.77) (2.88) (0.37) (7.04) (4.91) (2.56) (0.24) (6.85) (4.59) (3.16) (0.16)
3-Yea E ec , λ1999 47.2 26.1 21.0 0.40 33.3 18.8 14.6 0.39 35.2 22.6 12.6 0.43
(18.9) (12.6) (6.96) (0.46) (10.3) (6.64) (4.56) (0.31) (8.14) (5.43) (3.97) (0.19)
5-Yea E ec , λ2001 56.5 30.1 26.3 0.65 42.9 24.8 18.1 0.54 45.4 29.9 15.5 0.60
(19.5) (12.2) (8.47) (0.41) (12.0) (6.74) (6.22) (0.35) (9.17) (6.37) (4.64) (0.23)
10-Yea E ec , λ2006 59.6 21.6 38.0 1.14 53.4 22.6 30.7 0.89 57.4 31.0 26.4 0.98
(14.4) (7.67) (9.43) (0.38) (15.3) (6.46) (10.0) (0.44) (12.0) (8.71) (6.29) (0.36)
Le el o Ou come in 1993
Mean 32.5 23.6 8.92 0.99 33.1 23.9 9.19 1.04 33.1 23.9 9.19 1.04
S.D. 5.86 4.25 2.08 0.27 13.8 10.3 4.57 0.47 13.8 10.3 4.57 0.47
Yea s 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008
No. o Obse a ions 510 510 510 510 10,948 10,948 10,948 10,948 10,948 10,948 10,948 10,948
No es: This able summa izes he esul s o al e na i e speci ica ions o he baseline esul s epo ed in 1, A12, and Table 2. Displayed a e λ coe icien s, ob ained
om es ima ing di e en e sions (as speci ied in column i les) o he di e ence-in-di e ences speci ica ion in Equa ion (2). Ou come a iables include he numbe
o egula nu sing home wo ke s, pa - ime and ull- ime nu sing home employmen , and he numbe o nu sing home es ablishmen s, pe 1,000 indi iduals age 65
and olde . The esul s in columns (1)-(8) a e isualized in Figu e A6 and he esul s in columns (9)-(12) in he op ow o Figu e A7. All speci ica ions include coun y
and yea ixed e ec s. Columns (9)-(12) also include coun y-speci ic ime ends. S anda d e o s clus e ed a he coun y le el (columns 1-4, 9-12) and a he egion
le el (columns 5-8) a e included in pa en heses.
Sou ce: Own calcula ions.
85
Table A2.: E en S udy Resul s: Agg ega e Response, Al e na i e Speci ica ions
Ou come (pe 1,000 Age 65+ Popula ion)
Con olling o Coun o Elde ly Con olling o Sha e o Elde ly
Wo ke s Full-Time Pa -Time Fi ms Wo ke s Full-Time Pa -Time Fi ms
(1) (2) (3) (4) (5) (6) (7) (8)
E en S udy Coe icien s
1-Yea E ec , λ1997 21.1 11.1 10.1 0.25 23.5 12.2 11.3 0.30
(6.69) (4.20) (3.20) (0.15) (6.67) (4.20) (3.18) (0.14)
3-Yea E ec , λ1999 28.2 16.8 11.4 0.29 31.4 18.5 12.9 0.36
(7.92) (4.83) (4.04) (0.17) (7.82) (4.80) (3.95) (0.17)
5-Yea E ec , λ2001 34.1 21.5 12.7 0.39 38.8 24.1 14.6 0.50
(8.73) (5.37) (4.70) (0.18) (8.53) (5.32) (4.49) (0.18)
10-Yea E ec , λ2006 34.7 15.6 19.1 0.56 48.8 21.9 26.8 0.84
(9.93) (5.90) (6.29) (0.25) (9.39) (5.63) (5.71) (0.23)
Le el o Ou come in 1993
Mean 33.1 23.9 9.19 1.04 33.1 23.9 9.19 1.04
S.D. 13.8 10.3 4.57 0.47 13.8 10.3 4.57 0.47
Yea s 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008
No. o Obse a ions 10,948 10,948 10,948 10,948 10,948 10,948 10,948 10,948
No es: This able summa izes he esul s o al e na i e speci ica ions o he baseline esul s e-
po ed in 1, A12, and Table 2. Displayed a e λ coe icien s, ob ained om es ima ing di e en
e sions (as speci ied in column i les) o he di e ence-in-di e ences speci ica ion in Equa ion (2).
Ou come a iables include he numbe o egula nu sing home wo ke s, pa - ime and ull- ime
nu sing home employmen , and he numbe o nu sing home es ablishmen s, pe 1,000 indi idu-
als age 65 and olde . Columns (1)-(4) display esul s o speci ica ions addi ionally con olling o
he coun y-yea -le el coun o indi iduals age 65 and abo e, and columns (5)-(8) con ol o he
coun y-yea -le el popula ion sha e o esiden s age 65 and abo e. The esul s in columns (1)-(4)
a e isualized in he bo om ow o Figu e A7 and he esul s in columns (5)-(8) in he op ow o
Figu e A8. All speci ica ions include coun y and yea ixed e ec s. S anda d e o s clus e ed a he
coun y le el a e epo ed in pa en heses.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 86
IAB-Discussion Pape 03|2025 87
Table A3.: E en S udy Resul s: Agg ega e Response, Al e na i e Speci ica ions
Ou come (pe 1,000 Age 65+ Popula ion)
Bina y Exposu e Measu ea Al e na i e Exposu e Measu e (I)b Al e na i e Exposu e Measu e (II)c
Wo ke s Full-Time Pa -Time Fi ms Wo ke s Full-Time Pa -Time Fi ms Wo ke s Full-Time Pa -Time Fi ms
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
E en S udy Coe icien s
1-Yea E ec , λ1997 3.69 2.07 1.62 0.03 112 45.3 66.8 0.70 102 67.2 35.0 2.14
(0.80) (0.50) (0.40) (0.02) (36.5) (22.8) (18.0) (0.89) (34.1) (22.0) (15.8) (0.76)
3-Yea E ec , λ1999 4.88 2.81 2.06 0.05 143 69.2 74.2 1.12 156 103 52.6 2.32
(0.91) (0.56) (0.50) (0.02) (43.7) (26.7) (22.8) (1.00) (37.6) (24.5) (18.7) (0.89)
5-Yea E ec , λ2001 5.87 3.45 2.41 0.07 188 101 87.1 1.81 199 125 74.7 3.22
(1.04) (0.63) (0.58) (0.02) (47.9) (29.3) (26.6) (1.03) (41.3) (27.3) (21.7) (0.97)
10-Yea E ec , λ2006 6.94 3.27 3.67 0.10 242 93.7 148 4.02 246 115 131 4.25
(1.12) (0.69) (0.75) (0.03) (52.4) (31.3) (34.4) (1.39) (44.8) (28.6) (28.2) (1.11)
Le el o Ou come in 1993
Mean 33.1 23.9 9.19 1.04 33.1 23.9 9.19 1.04 33.1 23.9 9.19 1.04
S.D. 13.8 10.3 4.57 0.47 13.8 10.3 4.57 0.47 13.8 10.3 4.57 0.47
Yea s 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008 1975 - 2008
No. o Obse a ions 10,948 10,948 10,948 10,948 10,948 10,948 10,948 10,948 10,948 10,948 10,948 10,948
a The bina y exposu e measu e akes on he alue o one o coun ies abo e he median o he exposu e measu e E de i ed in Equa ion (1), and ze o o he wise.
HzP ,1993
b The al e na i e exposu e measu e (I) is de ined as E = 100% − .
65andOlde P opula ion ,1993
g ,1993,1999
∗LT CClaims ,1999−HzP ,1993
c The al e na i e exposu e measu e (II) is de ined as E = .
65andOlde P opula ion ,1993
No es: This able summa izes he esul s o al e na i e speci ica ions o he baseline esul s epo ed in 1, A12, and Table 2. Displayed a e λ coe icien s,
ob ained om es ima ing di e en e sions (as speci ied in column i les) o he di e ence-in-di e ences speci ica ion in Equa ion (2). Ou come a iables include
he numbe o egula nu sing home wo ke s, pa - ime and ull- ime nu sing home employmen , and he numbe o nu sing home es ablishmen s, pe 1,000
indi iduals age 65 and olde . The esul s in columns (1)-(4) a e isualized in he bo om ow o Figu e A8 and he esul s in columns (5)-(12) in Figu e A9. All
speci ica ions include coun y and yea ixed e ec s. S anda d e o s clus e ed a he coun y le el a e epo ed in pa en heses.
Sou ce: Own calcula ions.
Table A4.: E en S udy Resul s: Coun o New Nu sing Home Hi es By O igin
Ou come (pe 1,000 Age 65+ Popula ion)
Coun New NH Hi esa Among New Nu sing Home Hi esa, Coun o
Employed &
in HC in − 1
Employed &
No in HC in − 1
Unemployed
in − 1
Tempo a ily
No in LF in − 1
Ne e in LF
be o e
(1) (2) (3) (4) (5) (6)
Pooled Coe icien s
δ97−08 9.37 1.32 1.02 1.93 3.91 1.19
(1.82) (0.74) (0.53) (0.42) (0.52) (0.43)
E en S udy Coe icien s
1-Yea E ec , λ1997 8.38 3.10 -0.04 2.35 2.37 0.60
(4.05) (3.14) (0.75) (0.54) (0.53) (0.49)
3-Yea E ec , λ1999 7.52 0.89 1.26 1.69 2.96 0.73
(1.93) (0.72) (0.56) (0.56) (0.66) (0.49)
5-Yea E ec , λ2001 10.4 1.03 2.25 1.96 3.54 1.60
(2.27) (1.03) (0.67) (0.53) (0.68) (0.47)
10-Yea E ec , λ2006 8.81 0.64 0.51 2.03 4.13 1.50
(2.15) (0.84) (0.60) (0.55) (0.68) (0.53)
Le el o Ou come in 1993
Mean 6.85 1.28 1.30 1.13 2.05 1.09
S.D. 2.94 0.84 0.78 0.64 0.93 0.63
Yea s 1977 - 2008 1977 - 2008 1977 - 2008 1977 - 2008 1977 - 2008 1977 - 2008
No. o Obse a ions 10,304 10,304 10,304 10,304 10,304 10,304
a “New Hi es” a e indi iduals who we e no employed in a nu sing home in he yea be o e each
index yea .
No es: The op panel displays pooled δ97−08 coe icien s ob ained om es ima ing he di e -
ence in di e ences speci ica ion in Equa ion (3) a he coun y-yea le el, using E , de i ed in
Equa ion (1), as he measu e o a coun y’s exposu e o he e o m. E akes alues om 0 o 1 and
measu es he sha e o po en ial long- e m ca e pa ien s who did no ha e public assis ance o
LTC p io o he ollou o uni e sal LTC insu ance (mean and median o E =0.69). The geog aphic
a ia ion in E is isualized in Figu e A4. The ou come a iables a e coun s o new egula Nu sing
Home hi es pe 1,000 popula ion age 65 and abo e by employmen s a us be o e each index yea ,
as speci ied in he column i les. The second panel displays λ coe icien s o he e en s udy
in Equa ion (2). Coe icien s we e no malized o ze o in yea = 1993. Resul s om equi alen
speci ica ions, using he sha e o hi es om a espec i e o igin a he han coun s as ou comes,
a e p esen ed in Table 5. All speci ica ions include coun y and yea ixed e ec s. S anda d e o s
clus e ed a he coun y-le el a e epo ed in pa en heses.
Sou ce: Own calcula ions.
IAB-Discussion Pape 03|2025 88
Table A6.: Mon hly Wages by Expe ience and Sec o in 1999
Expe ience Popula ion Da aFP Da aNFP Da aOUT ModelFP ModelNFP ModelOUT
(1) (2) (3) (4) (5) (6) (7)
0 29.6 2.29 2.46 4.10 2.34 2.42 4.09
1 0.81 2.58 2.75 3.27 2.62 3.10 3.24
2 0.47 2.76 3.07 3.15 2.74 3.20 3.32
3 0.39 2.86 3.41 3.41 2.86 3.30 3.40
4 0.31 3.10 3.62 3.55 3.03 3.40 3.52
5 0.26 3.17 3.77 3.61 3.10 3.51 3.55
6 0.24 3.24 3.86 3.72 3.13 3.61 3.56
7 0.22 3.38 3.91 3.74 3.24 3.71 3.64
8 0.19 3.45 3.95 3.82 3.30 3.81 3.66
9 0.17 3.54 4.03 3.86 3.46 3.91 3.78
10 0.14 3.54 4.06 3.88 3.46 4.01 3.75
11 0.12 3.69 4.19 3.96 3.60 4.11 3.85
12 0.11 3.69 4.27 4.05 3.65 4.21 3.87
13 0.10 3.88 4.35 4.13 3.81 4.31 3.98
14 0.08 3.86 4.40 4.16 3.86 4.41 4.00
15 0.07 4.08 4.51 4.21 4.03 4.51 4.12
16 0.07 3.97 4.49 4.31 4.00 4.61 4.08
17 0.06 4.16 4.43 4.28 4.15 4.71 4.18
18 0.41 4.34 4.56 4.49 4.39 4.81 4.36
No es: This able shows ull- ime equi alen mon hly ea nings in 1,000 EUR by yea s o heal h ca e
expe ience and labo ma ke in he pos - e o m yea s (1999). Column (1) deno es he (wo king age)
popula ion in millions. Columns (2)-(4) deno e ull- ime equi alen mon hly wages based on he
Labo Ma ke Sample o o -p o i nu sing homes (column 2), no - o -p o i nu sing homes (column
3), and he ou side sec o (column 4). Columns (5)-(7) p esen co esponding es ima es based on
he quan i a i e model.
Sou ce: Own calcula ions.
Nex we examine changes in unemploymen and gains in labo o ce pa icipa ion
ela i e o he es ima ed changes in nu sing home employmen . We conside he mos
conse a i e es ima es suppo ed by he con idence in e als in column (1) o Table 6. We
conside a 0.00534 − 2 × 0.00177 = 0.00299 p.p. dec ease in he unemploymen sha e.
We benchma k his o he ela i e gain in nu sing home employmen . Using ou pa ial
equilib ium es ima es om Table 2, we es ima e an inc ease o 38,979 wo ke s. Di ided
by he 1993 labo o ce o 22.7 million, we es ima e an inc ease o 0.00172 p.p. This
sugges s ha he d op in UI exceeds he gains in nu sing home employmen by
0.00299/0.00172=1.74, see he six h ow. Ou es ima ed model can only econcile a
smalle educ ion in unemploymen . Likewise, we cons uc conse a i e inc ease in
labo o ce pa icipa ion based on he es ima es in column 3 o Table 2. Di iding he gain
by he gain in nu sing home employmen , we ind a a io o 2.94, see he se en h ow.
We also ensu e ha he job illing and acancy illing a es a e bounded be ween 0 and 1,
see Figu e A17.
IAB-Discussion Pape 03|2025 95

Table A7.: Employmen Sha e by Expe ience and Sec o in 1999
Expe ience Popula ion Da aFP Da aNFP Da aOUT Da aOOL ModelFP ModelNFP ModelOUT ModelOOL
(1) (2) (3) (4) (5) (6) (7) (8) (9)
0 29.6 0.0004 0.0007 0.53 0.41 0.0004 0.0007 0.54 0.41
1 0.81 0.01 0.02 0.39 0.49 0.01 0.02 0.39 0.50
2 0.47 0.02 0.03 0.42 0.45 0.02 0.03 0.42 0.47
3 0.39 0.02 0.04 0.43 0.44 0.02 0.04 0.44 0.44
4 0.31 0.03 0.05 0.44 0.43 0.03 0.05 0.43 0.42
5 0.26 0.03 0.05 0.45 0.43 0.03 0.05 0.44 0.41
6 0.24 0.03 0.06 0.45 0.42 0.03 0.06 0.43 0.41
7 0.22 0.03 0.06 0.47 0.40 0.03 0.06 0.46 0.38
8 0.19 0.03 0.07 0.48 0.39 0.03 0.07 0.46 0.38
9 0.17 0.03 0.07 0.49 0.38 0.03 0.07 0.48 0.35
10 0.14 0.03 0.07 0.48 0.39 0.03 0.07 0.47 0.36
11 0.12 0.03 0.07 0.51 0.37 0.03 0.07 0.49 0.34
12 0.11 0.02 0.07 0.52 0.36 0.02 0.07 0.51 0.33
13 0.10 0.03 0.06 0.55 0.34 0.03 0.06 0.54 0.31
14 0.08 0.02 0.07 0.57 0.31 0.02 0.08 0.56 0.29
15 0.07 0.02 0.07 0.58 0.30 0.02 0.07 0.57 0.27
16 0.07 0.03 0.07 0.59 0.29 0.03 0.07 0.59 0.27
17 0.06 0.03 0.08 0.60 0.26 0.03 0.08 0.60 0.25
18 0.41 0.02 0.07 0.67 0.21 0.02 0.07 0.67 0.21
No es: This able shows ull- ime equi alen employmen sha es by yea s o heal h ca e expe ience
and labo ma ke in he pos - e o m yea s (1999). Column (1) deno es he (wo king age) popula-
ion in millions. Columns (2)-(5) deno e employmen sha es based on he Labo Ma ke Sample o
o -p o i nu sing homes (column 2), no - o -p o i nu sing homes (column 3), he ou side sec o
(column 4), and he sha e ha is ou o he labo o ce (column 5). Columns (6)-(9) p esen co e-
sponding es ima es based on he quan i a i e model.
Sou ce: Own calcula ions.
A2.3. Pa ame e Es ima es
In his sec ion, we p esen u he de ails on he calib a ed and he es ima ed
pa ame e s.
Calib a ed pa ame e s:
Table A9 summa izes he calib a ed pa ame e s. Fi s , we calib a e he ac ion o
pa ien s insu ed ia Hil e zu P lege in he pos - e o m o 31%. Second, we se he p ice
subsidy o he a e age inpa ien bene i s, weigh ed by he ca e le el dis ibu ion o LTC
bene icia ies. We calcula e a p ice subsidy o 1,200 EUR pe mon h. Thi d, we calcula e
use he a e age ax a e and social secu i y con ibu ions o singles wi hou child en in
1999. We use he ax a e o 18.6% and social secu i y con ibu ion o 20.6% paid by
employee and also by he employe .52 This implies a wedge be ween o al cos s o he
employe and he ne akehome pay o wo ke s o in 1999 o 18.6%+2×20.6%
τ = = 50%.
1+20.6%
52 See .
IAB-Discussion Pape 03|2025 96
Table A8.: O he Momen s
Momen Da a Model
(1) (2)
Mon hly NH P ice Pos in 1k Eu o 2.03 2.00
Mon hly NH Re enues Pos in bn Eu o 1.14 1.46
% Change in NH Wage -0.12 0.02
NH Employmen Pos / NH Employmen P e 1.92 1.92
% Change in Expe ience among NH Wo ke s -0.30 -0.13
D op in UI Recipien s/ Change in NH Employmen 1.74 0.19
Change in LF/ Change in NH Employmen 2.94 3.04
No es: This able p esen s he model i on a ious o he a ge ed momen s. Column (1) deno es
he da a es ima e and column (2) p esen s he model-simula ed coun e pa . The i s wo ows
p esen a e age nu sing home p ices and o e all nu sing home e enues in 1999. The hi d ow
conside s he e o m-induced ela i e change in ull- ime nu sing home wages. The ou h ow
conside s he e o m-induced ela i e change in nu sing home employmen . The i h ow consid-
e s he e o m-induced ela i e change in he a e age numbe o yea s o heal h ca e expe ience
among nu sing home wo ke s. The six h ow conside s he e o m-induced ela i e change in he
he numbe o indi iduals claiming unemploymen bene i s, ela i e o he e o m-induced change
in nu sing home employmen . The se en h ow conside s he e o m-induced ela i e change in la-
bo o ce pa icipa ion, ela i e o he e o m-induced change in nu sing home employmen . Mo e
de ails a e p o ided in he ex .
Sou ce: Own calcula ions.
Finally, we se he unemploymen bene i s in 1999 o 775 EUR pe mon h.
Table A9.: Calib a ed Pa ame e s
Momen Da a
(1)
Sha e Hil e-zu P lege Pos 0.31
LTC P ice Subsidy in 1,000 Eu o pe mon h: s 1.21
% Tax and Social Secu i y Con ibu ions: τ 0.50
Unemploymen bene i s in 1,000 Eu o pe mon h: b 0.78
No es: This able p esen s calib a ed pa ame e s. The i s ow p esen s he ac ion o nu sing
home pa ien s insu ed ia Hil e zu P lege. The second ow p esen s he p ice subsidy o inpa ien
nu sing home ca e in he pos - e o m pe iod. The hi d ow p esen s income axes and social se-
cu i y con ibu ions as a sha e o he o al labo cos s o he employe . The las ow p esen s he
unemploymen bene i s pe mon h in 1,000 EUR.
Sou ce: Own calcula ions.
Pa ame e Es ima es:
Nex , we u n o he es ima ed pa ame e s. We p esen he es ima ed λ pa ame e s
go e ning he ma ching echnology and he acancy pos ing cos s c, by yea s o heal h
ca e expe ience and sec o , in columns (1)-(6) in Table A10. The λ pa ame e s a e
IAB-Discussion Pape 03|2025 97
Figu e A17.: Job Finding and Vacancy Filling P obabili y
(a) P e-Re o m Vacancy Filling Ra e in Nu sing (b) P e-Re o m Job Finding P obabili y in Nu sing
Homes
Homes
0
.2
.4
.6
.8
1
0 3 6 9 12 15 18
Yea s o Heal hca e Expe ience
Fo -P o i
No -Fo -P o i
Ou side
(c) Pos -Re o m Vacancy Filling Ra e in Nu sing
Homes
0
.2
.4
.6
.8
1
0 3 6 9 12 15 18
Yea s o Heal hca e Expe ience
Fo -P o i
No -Fo -P o i
Ou side
(d) Pos -Re o m Job Finding P obabili y in Nu s-
ing Homes
.2
.4
.6
.8
1
0 3 6 9 12 15 18
Yea s o Heal hca e Expe ience
Fo -P o i
No -Fo -P o i
Ou side
.5
.6
.7
.8
.9
1
0 3 6 9 12 15 18
Yea s o Heal hca e Expe ience
Fo -P o i
No -Fo -P o i
Ou side
No es: This igu e p esen s he es ima ed acancy and job inding p obabili y by sec o and yea s
o heal h ca e expe ience. Fo -p o i nu sing homes and no - o -p o i nu sing homes a e deno ed
by ‘Fo -P o i ’ and ‘No - o -P o i ’, espec i ely. Figu es A17a and A17c display he equilib ium a-
φ
cancy illing a e, η(q ), o he p e- and he pos - e o m pe iod. Figu es A17b and A17d display he
j
φ
equilib ium job inding p obabili ies, µ(q ), o he p e- and he pos - e o m pe iod.
j
Sou ce: Own calcula ions.
p esen ed o o -p o i nu sing homes (column 1), no - o -p o i nu sing homes (column
2) and i ms in he ou side sec o (column 3). Likewise, he acancy pos ing cos s c a e
p esen ed o o -p o i nu sing homes (column 4), no - o -p o i nu sing homes (column
5) and i ms in he ou side sec o (column 6). The las wo columns p esen he wo ke
skills φ, as de ailed in Equa ion (13), in nu sing homes (column 7) and he ou side sec o
(column 8).
The emaining pa ame e a e summa ized in Table A11. The i s h ee pa ame e s go e n
he demand o he ep esen a i e agen , see Equa ions 19-21. The nex eigh pa ame e s
go e n he wo ke ’s u ili y unc ion, see Equa ions 15-18. Ou empi ical speci ica ion
adds non-wage ma ginal cos s o he i m p o i Equa ion (5):
φ φ φ
π = η(q ) × ((Pj − mcj
) × φ − w ) − cj
(φ). (A8)
j j j
IAB-Discussion Pape 03|2025 98
Table A10.: Ma ching Technology and Hi ing Cos s Pa ame e s
Expe ience LambdaFP LambdaNFP LambdaOu cFP cNFP cOu SkillsNH SkillOUT
(1) (2) (3) (4) (5) (6) (7) (8)
0 0.50 0.82 0.90 0.16 0.06 0.16 1.65 2.19
1 0.96 0.81 0.81 0.34 0.21 0.34 1.72 1.85
2 0.83 0.91 0.79 0.35 0.12 0.35 1.79 1.90
3 0.81 0.84 0.79 0.35 0.16 0.35 1.86 1.95
4 0.82 0.77 0.84 0.32 0.20 0.32 1.93 1.99
5 0.80 0.77 0.82 0.36 0.23 0.36 2.00 2.04
6 0.78 0.87 0.80 0.43 0.14 0.43 2.07 2.09
7 0.80 0.88 0.79 0.44 0.14 0.44 2.14 2.13
8 0.79 0.89 0.77 0.49 0.14 0.49 2.21 2.18
9 0.83 0.76 0.79 0.46 0.30 0.46 2.28 2.22
10 0.84 0.76 0.77 0.56 0.34 0.56 2.35 2.27
11 0.86 0.76 0.78 0.55 0.37 0.55 2.42 2.32
12 0.87 0.78 0.75 0.61 0.37 0.61 2.49 2.36
13 0.88 0.82 0.76 0.58 0.37 0.58 2.56 2.41
14 0.88 0.79 0.73 0.64 0.39 0.64 2.63 2.46
15 0.92 0.79 0.75 0.60 0.43 0.60 2.70 2.50
16 0.87 0.85 0.70 0.73 0.39 0.73 2.77 2.55
17 0.89 0.84 0.72 0.71 0.41 0.71 2.85 2.60
18 0.95 0.87 0.72 0.61 0.37 0.61 2.92 2.64
No es: This able p esen s es ima ed ma ching echnology pa ame e s, λ, hi ing cos s pa ame e s,
c, and wo ke skills, φ, by sec o and yea s o heal h ca e expe ience. Fo -p o i nu sing homes a e
abb e ia ed by ‘FP’, no - o -p o i nu sing homes a e abb e ia ed by ‘NFP’, and he ou side sec o
is abb e ia e by ‘Ou ’. We assume ha wo ke s skills a e homogeneous be ween o -p o i and
no - o -p o i nu sing homes, condi ional on expe ience, which we hence deno e by ‘SkillsNH’, see
Equa ion 13. Mo e de ails a e p o ided in he ex .
Sou ce: Own calcula ions.
and he nex h ee ows o Table A11 p esen he co esponding ma ginal cos
es ima es.
Building on he ma ginal cos es ima es, we can exp ess he wages paid by no - o -p o i
and public nu sing homes, which a e de e mined ia collec i e ba gaining. As discussed
in he main ex , we model wages, de e mined ia collec i e ba gaining, as a second
o de polynomial in he ma ginal e enue p oduc ne o non-wage ma ginal cos s:
 2
φ
w = β0 + β1 × φ × (Pj − mc) + β2 × φ × (Pj − mc) . (A9)
j
The nex h ee pa ame e s in Table A11 show he co esponding pa ame e es ima es.
Finally, we p esen he es ima ed equilib ium ou pu p ices in he las ou ows. ‘Pos ’
e e s o he pos - e o m p ices and ‘P e’ e e s o p e- e o m en i onmen wi hou
IAB-Discussion Pape 03|2025 99
uni e sal LTC insu ance, whe e LTC suppo is only p o ided ia Hil e zu P lege. P ices in
nu sing homes and he ou side sec o a e deno ed by ‘NH’ and ‘OUT’, espec i ely.
Table A11.: Pa ame e Es ima es
Pa ame e Es ima e
(1)
Demand: αNH 2.71
Demand: αOUT 5.85
Demand: σ 1.19
Wo ke s: Flow Payo Unemploymen (b + ξu) 1.01
Wo ke s: Nes ing Pa am e γ 0.49
Wo ke s: S anda d De ia ion Shocks ρ 0.03
Wo ke s: ξF P -0.15
Wo ke s: ξNF P -0.09
Wo ke s: κ 7.64
Wo ke s: κ1 -0.47
Wo ke s: ν 1.14
Wage: β0 2.42
Wage: β1 0.71
Wage: β2 0
Fi ms: FP 0.002
Fi ms: NFP 0
Fi ms: OUT 0.05
NH P ice P e 1.87
NH P ice Pos 2.00
OUT P ice P e 2.11
OUT P ice Pos 2.14
No es: This able p esen s he pa ame e es ima es go e ning he quan i a i e model. The i s
h ee pa ame e s αNH
, αOu , and σ demand o he ep esen a i e agen , see Equa ions 19-21. The
nex 8 pa ame e s go e n wo ke ’s u ili y unc ion, see Equa ions 15-18. b + ξu deno es he low-
payo om unemploymen , γ is a scaling pa ame e , ρ go e ns he co ela ion be ween u ili ies
o wo king ac oss long- e m ca e sec o s. ξF P and ξNF P deno e compensa ing di e en ials o
wo king in o -p o i and no - o -p o i nu sing homes ( ela i e o wo king in he ou side sec o ). ν
cap u es an ex a compensa ing di e en ial o wo ke s wi h some heal h ca e expe ience o wo k
in nu sing homes, see Equa ion (15). The nex h ee ows o Table esen es ima es o he non-wage
ma ginal cos s by sec o , mcj
, see Equa ion (A8). They a e ollowed by he pa ame e s go e ning
he collec i e ba gaining wage p o ile desc ibed in Equa ion (A9). Finally, we p esen he es ima ed
equilib ium ou pu p ices in he las ou ows. ‘Pos ’ e e s o he pos - e o m p ices and ‘P e’ e e s
o p e- e o m en i onmen wi hou uni e sal LTC insu ance, whe e LTC suppo is only p o ided
ia Hil e zu P lege. P ices in nu sing homes and he ou side sec o a e deno ed by ‘NH’ and ‘OUT’,
espec i ely.
Sou ce: Own calcula ions.
A2.4. Coun e ac uals
In his sec ion, we discuss he coun e ac ual analysis in mo e de ail. We s a wi h a
discussion o baseline ou comes be o e u ning o a discussion o addi ional
coun e ac ual exe cises.
IAB-Discussion Pape 03|2025 100

Baseline esul s:
We s a wi h a mo e de ailed discussion o ou baseline en i onmen and p esen he
model-simula ed ou comes wi h and wi hou uni e sal LTC insu ance in he i s wo
columns o Table A12 (copied om Table 9). We es ima e ha inc easing long- e m ca e
insu ance co e age om 31% o 100% c ea es an addi ional 164,000 nu sing home
jobs.
Abou 32,000 indi iduals a e emo ed om unemploymen and labo o ce pa icipa ion
inc eases by 497,000 indi iduals. Pu oge he , o e all employmen inc eases by 529,000
indi iduals implying subs an i e employmen gains in o he sec o s o he economy. This
is because he e o m boos ed incomes among new nu sing home hi es who hen spend
hei incomes on all goods in he economy gene a ing posi i e spillo e e ec s o o he
sec o s.
Figu e A18a plo s he e o m-induced employmen expansion in nu sing homes by
expe ience. We ind ha he inc ease in nu sing home employmen was concen a ed
among lowe expe ienced wo ke s, sugges ing ha he baseline esul is pa ially d i en
by he conside able slack in he labo ma ke among people who we e p e iously
ou -o - he labo o ce and hus held li le heal h ca e expe ience.
Figu es A18b-A18d illus a e ha labo ma ke s do no ‘jus ’ clea in wages and ins ead
depend on job pos ing decisions by i ms. Figu e A18b illus a es p e- e o m and
pos - e o m nu sing home employmen bu also coun e ac ual nu sing home
employmen had wages no adjus ed. We ind ha mo e han hal o he inc ease in
nu sing home employmen can be a ibu ed o changes in acancy pos ings.
Unemploymen would ha e dec eased e en u he , had wages emained he same,
Figu e A18c. In con as , a la ge ac ion o he inc eases in labo o ce pa icipa ion can
be a ibu ed o changes in wages, see Figu e A18d, sugges ing ha changes in acancy
pos ing played less o a ole o he employmen gains in o he sec o s.
Ma ginal coun e ac uals:
We complemen he discussion o he coun e ac ual analysis in he main ex wi h added
esul s o mo e ma ginal policy in e en ions. Table A12 is s uc u ed as Table 9 in he
main ex bu conside s addi ional in e en ions. S a ing wi h he esul s o UI bene i s,
we now conside a 1% educ ion in UI bene i s (columns 4-6) in addi ion o he 100%
educ ion in UI bene i s (columns 7-9) discussed in he main ex . In ega ds o he ole o
income axes, we now conside a 1% educ ion in income axes (columns 10-12) in
addi ion o he 50% educ ion in income axes (columns 13-15) discussed in he main
ex . Finally, and in ega ds o he ole o p oduc i i y shocks, we now conside a 10%
IAB-Discussion Pape 03|2025 101
Figu e A18.: Mechanisms
(a) Nu sing Home Employmen (Millions) (b) Nu sing Home Employmen (Millions)
0
.01
.02
.03
.04
0 3 6 9 12 15 18
Yea s o Heal hca e Expe ience
1993 (P e-Re o m) 1999 (Pos -Re o m)
.15
.2
.25
.3
.35
P e Pos wi h P e-Wages Pos
En i onmen
(c) Unemploymen (Millions) (d) Labo Fo ce (Millions)
1.6
1.62
1.64
1.66
1.68
P e Pos wi h P e-Wages Pos
En i onmen
19
19.2
19.4
19.6
19.8
P e Pos wi h P e-Wages Pos
En i onmen
No es: Figu es A18a -A18d p o ide mo e de ails on composi ional changes in nu sing home em-
ploymen and he ole o sea ch and ma ching ic ions, based on es ima es om he quan i a-
i e model. Figu e A18a p esen s ull- ime equi alen nu sing home employmen in millions wi h
and wi hou uni e sal LTC insu ance by yea s o heal hca e expe ience. Figu es A18b-A18d display
ull- ime equi alen nu sing home employmen , unemploymen , and labo o ce pa icipa ion in
millions o h ee economic en i onmen s. The i s ba deno es equilib ium ou comes in he p e-
e o m pe iod (wi hou uni e sal LTC insu ance) and he hi d column shows equilib ium ou comes
in he pos - e o m pe iod (wi h uni e sal LTC insu ance). The cen e column p esen s ou comes us-
ing pos - e o m equilib ium queue leng hs bu p e- e o m wages.
Sou ce: Own calcula ions.
inc ease in he p oduc i i y o he ou side sec o (columns 16-18) in addi ion o he 50%
inc ease in p oduc i i y (columns 19-21) discussed in he main ex . The e o m e ec s o
he mo e ma ginal in e en ions he e a e di ec ionally consis en wi h he indings
p esen ed in he main ex , bu smalle in absolu e magni ude.
The las h ee columns conside one addi ional coun e ac ual combining he ole o
p oduc i i y shocks and income axes. Speci ically, we lowe income axes by 50% and
also inc ease he p oduc i i y in he ou side sec o by 50%. We now ind almos no
expansion in labo o ce ollowing he e o m sugges ing ha he e o m is mos ly
eloca ing wo ke s om he ou side sec o o he nu sing home sec o . This educes
alloca i e e iciency and in u n esul s in a wel a e loss.
IAB-Discussion Pape 03|2025 102
IAB-Discussion Pape 03|2025 103
Table A12.: Coun e ac uals and Wel a e
Baseline
1% Reduc ion in
UEB
100% Reduc ion in
UEB
1% Reduc ion in
Income Tax
50% Reduc ion in
Income Tax
10% Inc ease in
P oduc i i y
50% Inc ease in
P oduc i i y
50% Inc . in P od.,
50% Dec . in Inc. Tax
Insu ance
Di
Insu ance
Di
Insu ance
Di
Insu ance
Di
Insu ance
Di
Insu ance
Di
Insu ance
Di
Insu ance
Di No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24)
Panel A: Employmen and Ea nings
Nu sing Home Employmen (million FTE) 0.18 0.34 0.16 0.18 0.34 0.16 0.19 0.34 0.16 0.18 0.35 0.17 0.23 0.47 0.24 0.20 0.38 0.18 0.25 0.36 0.11 0.26 0.37 0.12
To al Labo Cos s o Employe s (billion EUR/mon h) 70.6 73.6 2.99 70.6 73.6 2.99 68.8 71.9 3.04 71.7 74.7 3.02 111.5 114.4 2.81 92.0 95.4 3.43 178.7 182.1 3.36 195.1 197.6 2.55
Unemploymen (million FTE) 1.69 1.66 -0.03 1.69 1.66 -0.03 0.73 0.72 -0.006 1.70 1.67 -0.04 1.85 1.80 -0.05 1.60 1.52 -0.08 0.88 0.85 -0.04 0.62 0.59 -0.03
Labo Fo ce (million FTE) 19.4 19.9 0.50 19.4 19.9 0.50 18.6 19.1 0.53 19.7 20.2 0.50 30.3 30.5 0.28 22.6 23.0 0.46 30.8 31.0 0.20 33.6 33.6 0.02
Panel B: Wel a e (billion EUR/mon h)
Consume Wel a e 81.0 80.3 -0.69 81.0 80.3 -0.69 81.7 81.0 -0.69 82.4 81.7 -0.70 130.2 129.1 -1.10 105.5 104.6 -0.89 203.0 201.3 -1.72 223.5 221.6 -1.90
Wo ke Wel a e 148.7 149.2 0.59 148.6 149.2 0.59 147.7 148.3 0.60 149.0 149.6 0.60 172.1 173.5 1.35 152.9 153.6 0.70 175.4 176.5 1.12 222.7 224.6 1.82
Go e nmen Su plus 33.4 34.3 0.88 33.4 34.3 0.88 33.9 34.8 0.88 33.6 34.5 0.88 25.9 25.7 -0.25 44.2 45.2 1.06 87.8 88.8 0.99 47.5 47.5 -0.08
LTC Subsidy Spending -0.42 -1.05 -0.63 -0.42 -1.05 -0.63 -0.43 -1.06 -0.63 -0.42 -1.07 -0.64 -0.42 -1.41 -0.99 -0.44 -1.15 -0.71 -0.57 -1.28 -0.71 -0.59 -1.34 -0.74
UI Spending -1.31 -1.29 0.02 -1.31 -1.28 0.02 0 0 0 -1.32 -1.29 0.03 -1.44 -1.40 0.04 -1.24 -1.18 0.06 -0.68 -0.66 0.03 -0.48 -0.45 0.03
Tax Re enues 35.2 36.7 1.49 35.2 36.7 1.49 34.3 35.8 1.51 35.4 36.9 1.49 27.8 28.5 0.70 45.8 47.6 1.71 89.1 90.7 1.67 48.6 49.2 0.64
To al Wel a e 263.1 263.9 0.78 263.1 263.9 0.78 263.3 264.1 0.78 265.0 265.8 0.78 328.2 328.2 -0.01 302.5 303.3 0.86 466.2 466.6 0.38 493.8 493.6 -0.16
No es: This able summa izes he model-simula ed e ec s o a uni e sal LTC expansion o on a numbe o ou comes in di e en economic en i onmen s. Fo each
en i onmen , we p esen ou comes (i) wi hou uni e sal LTC insu ance, whe e LTC suppo is p o ided ia Hil e zu P lege only and (ii) wi h uni e sal LTC insu ance
co e age cap u ed by he subsidy s=1,200 EUR pe mon h o he cos o nu sing home ca e o indi iduals who would o he wise pay ou -o -pocke . This exe cise
main ains suppo ia Hil e zu P lege. We also p esen he di e ence be ween hese ou comes. We conside eigh en i onmen s. Columns (1)-(3) p esen esul s
o ou baseline economic en i onmen , columns (4)-(6) e isi he esul s a e educing he UI bene i s by 1% and columns (7)-(9) p esen esul s a e educing
UI bene i s om 775 EUR pe mon h o ze o. Columns (10)-(12) (13-15) e isi he esul s a e educing he income axes by 1% (50%, om abou 50 p.p. o 25
p.p.). Columns (16)-(18) (19-21) e isi he esul s a e inc easing he p oduc i i y in he ou side sec o o each skill segmen , φou , (ehc) by 10% (50%). Finally,
columns (22)-(24) e isi he esul s a e inc easing he p oduc i i y in he ou side sec o by 50% and educing income axes by 50%. Fo each en i onmen , we
p esen nu sing home demand in 1m pa ien s, nu sing home employmen in 1m ull- ime equi alen employees, o al g oss wage ea nings in bn EUR pe mon h,
numbe o employmen indi iduals in 1m, and labo o ce pa icipa ion in 1m indi iduals. Nex , we p esen he pa ial equilib ium wel a e e ec in bn pe mon h.
Tu ning o wel a e in gene al equilib ium, we deno e he equi alen a ia ion (consume su plus) in bn pe mon h, holding income ixed a he p e- e o m le el.
We hen p esen he wo ke su plus in he labo ma ke in bn pe mon h. Tu ning he iscal ex e nali ies, we p esen go e nmen LTC subsidy spending in bn pe
mon h, go e nmen spending on UI bene i s in bn pe mon h, income ax e enues in bn pe mon h, and o e all ne spending go e nmen spending in bn pe
mon h. Finally, we combine he equi alen a ia ion, he labo ma ke su plus, and he ne impac o go e nmen spending which comp ises he ne wel a e e ec
in gene al equilib ium in bn pe mon h.
Sou ce: Own calcula ions.
Imp in
IAB-Discussion Pape 03|2025
Publica ion Da e
Ma ch 7, 2025
Publishe
Ins i u e o Employmen Resea ch
o he Fede al Employmen Agency
Regensbu ge S aße 104
90478 Nü nbe g
Ge many
All igh s ese ed
This publica ion is published unde he ollowing C ea i e Commons licence: A ibu ion –
Sha eAlike 4.0 In e na ional (CC BY-SA 4.0)
h ps://c ea i ecommons.o g/licenses/by-sa/4.0/deed.de
Download
h ps://doku.iab.de/discussionpape s/2025/dp0325.pd
All publica ions in he se ies “IAB-Discusssion Pape ” can be downloaded om
h ps://iab.de/en/publica ions/iab-publica ions/iab-discussion-pape -en/
Websi e
h ps://www.iab.de/en
ISSN
2195-2663
DOI
10.48720/IAB.DP.2503
Co esponding au ho
Jö g Heining
Tele on ++49-(0)911-178-5392
E-Mail joe [email protected]