He ge , Anna; Riphahn, Regina T.
A icle — Published Ve sion
Phasing ou pay oll ax subsidies
In e na ional Tax and Public Finance
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: He ge , Anna; Riphahn, Regina T. (2025) : Phasing ou pay oll ax subsidies,
In e na ional Tax and Public Finance, ISSN 1573-6970, Sp inge US, New Yo k, NY, Vol. 32, Iss. 5, pp.
1501-1531,
h ps://doi.o g/10.1007/s10797-024-09879-0
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/330619
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/4.0/
Vol.:(0123456789)
In e na ional Tax and Public Finance (2025) 32:1501–1531
h ps://doi.o g/10.1007/s10797-024-09879-0
Phasing ou pay oll ax subsidies
AnnaHe ge 1· ReginaT.Riphahn1
Accep ed: 24 Decembe 2024 / Published online: 12 Feb ua y 2025
© The Au ho (s) 2025
Abs ac
Many coun ies subsidize low-income employmen s o small jobs. These subsidies
and hei phasing ou can gene a e labo ma ke ic ions and dis o incen i es.
The Ge man Minijob p og am subsidizes low-income jobs. I gene a es a ‘Mini-
job ap’ wi h subs an ial bunching along he ea nings dis ibu ion. Since 2003, he
newly in oduced Midijob subsidy aims o educe he Minijob-induced no ch in he
ne ea nings dis ibu ion. Midijobs educe pay oll axes o employmen s abo e he
Minijob ea nings ceiling. We in es iga e whe he in oducing Midijobs educed he
Minijob ap. We apply a eg ession discon inui y design using adminis a i e da a
and a di e ence-in-di e ences es ima ion using su ey da a. While in bo h cases ou
esul s show a small posi i e o e all e ec o Midijobs on ansi ions ou o Mini-
jobs, hey a e e ec i e only o a na ow ea men g oup.
Keywo ds Midijobs· Minijobs· Pay oll ax subsidy· Causal e ec s· Di e ence-in-
di e ences· Reg ession discon inui y· SOEP· SIAB
JEL Classi ica ion J21· J38· H24
1 In oduc ion
In ecen decades, many coun ies ha e shi ed minimum income p o ec ion om
adi ional means- es ed ans e s o make wo k pay policies o incen i ize labo
supply.1 Make wo k pay policies condi ion on employmen and ei he p o ide sub-
sidies o employe s o o employees ia nega i e income axes o pay oll ax subsi-
dies. Well-known examples o such policies a e he U.S. Ea ned Income Tax C edi
(EITC) and he U.K. Wo king Family Tax C edi (WFTC). Nume ous con ibu ions
* Regina T. Riphahn
[email p o ec ed]
1 F ied ich-Alexande -Uni e si y E langen-Nü nbe g, Lange Gasse 20, 90403Nu embe g,
Ge many
1 See, e.g., Imme oll e al. (2007) o Ja a e al. (2020).
1502
A.He ge , R.T.Riphahn
e alua e he labo supply e ec s o hese p og ams.2 This pape con ibu es o he
make-wo k pay li e a u e: we s udy he e ec o a new p og am (Midijobs) aimed a
phasing ou a pay oll ax subsidy (Minijobs) in Ge many. The Midijob p og am was
in oduced o incen i ize labo supply and ea nings inc eases o wo ke s wi h e y
low ea nings. This is a challenge in many labo ma ke s.
The Midijob p og am was implemen ed on Ap il 1, 2003. I educes employee social
insu ance con ibu ions o jobs wi hin he Midijob ea nings ange, o iginally 400 o
800 Eu o pe mon h. I was in oduced o add ess a labo supply disincen i e gene a ed
by Minijobs; Minijobs a e low-income employmen s paying up o 400 Eu o pe mon h
a he ime. They cause subs an ial bunching a he Minijob ea nings limi (Gudgeon
& T enkle, 2024; Tazhi dino a, 2020) because Minijob employees pay nei he income
axes no social insu ance con ibu ions. Taxes and con ibu ions all due on o al
ea nings when ea nings exceed he Minijob ea nings h eshold; Tazhi dino a (2020)
poin ed ou ha an a e age woman’s combined ax and social insu ance liabili ies when
ea ning 1 Eu o beyond he p e 2003 Minijob ea nings h eshold eached 45 pe cen o
he o al g oss income. The e o e, wo ke s a ely ex end hei ea nings beyond his limi
which gene a es he so-called ‘Minijob ap’. Figu e1 illus a es he une en numbe o
jobs ac oss he ea nings dis ibu ion. Midijobs we e designed o add ess he Minijob
ap and we in es iga e whe he he Midijob p og am succeeds in his espec .
Bo h p og ams, Minijobs and Midijobs a e used ex ensi ely. A he end o 2021,
4.1 million indi iduals used Minijobs as hei main employmen and abou 3 mil-
lion wo ke s we e employed in Midijobs (see Appendix Figu eA.2).3 Thus, he p o-
g ams join ly co e ed 15.3 pe cen o one-six h o he Ge man labo o ce. Gi en
he magni ude o he Midijob p og am, i s expansions o ea nings limi s o 1,300 (in
2019), 1600 (in 2022), and e en 2000 Eu o in 2023, and he lack o p io e idence,
he Minijob ap add essed in his pape is impo an .
We use wo complemen a y empi ical s a egies. Fi s , we exploi he in oduc-
ion o Midijobs on Ap il 1, 2003, o e alua e hei immedia e e ec s on he p o-
pensi y o ansi ion om a Minijob o highe -ea ning, egula employmen . Fo his
pu pose, we apply a eg ession discon inui y in ime (RDiT) design o adminis a-
i e da a and in es iga e whe he he e is a jump in he ansi ion p obabili y a he
cu o da e. Howe e , his app oach may su e om sho comings, e.g., by cap u -
ing only local e ec s. Ou second app oach exploi s a unique ea u e o income ax
egula ions. Speci ically, Midijobs educed wo ke s’ social insu ance con ibu ion
a es in he ea nings ange jus abo e he Minijob ea nings limi lea ing income
ax a es un ouched. This a enua es he kink in he ne ea nings schedule, pa icu-
la ly o indi iduals who pay low o no income axes (e.g., single indi iduals). In
3 Ano he 3.1 million indi iduals use Minijobs as seconda y employmen . As seconda y job holding
is no in he ocus o he Midijob egula ion we do no discuss i u he ( o a de ailed analysis see,
e.g., Tazhi dino a 2022). − The u iliza ion o Minijobs declined sligh ly in 2015 wi h he in oduc ion o
manda o y minimum wages and again du ing he Co id pandemic. Figu eA.2 clea ly shows ha mo e
employees we e co e ed a e he Midijob ea nings ceiling inc eased on July 1, 2019 om 850 o 1300
Eu o pe mon h.
2 Fo a ecen su ey see Neuma k and Shi ley (2020), also Bas ian and Lochne (2022), Bas ian (2020),
Hoynes and Pa el (2018), Azma (2014), Che y and Saez (2013), Che y e al. (2013), Dahl e al. (2009),
F ancesconi and an de Klaauw (2007), Eissa and Hoynes (2004), and Eissa and Liebman (1996).
1503
Phasing ou pay oll ax subsidies
con as , Midijobs ha dly a ec he ne ea nings schedule o indi iduals who a he
kink addi ionally become subjec o high income ax a es (e.g., seconda y ea ne s
in ma iages) because o hem income axes domina e social insu ance con ibu-
ions. We conside he o me g oup o be ea ed by he e o m whe eas he la e
g oup is ou con ol g oup. We use su ey da a om he Ge man Socioeconomic
Panel (SOEP) which o e s in o ma ion on ma i al s a us and apply a di e ence-in-
di e ences (DID) analysis o s udy he p opensi y o ansi ion om a Minijob o
highe -ea ning, egula employmen . The DID analysis yields an a e age ea men
e ec on he ea ed which cap u es mo e sluggish labo supply esponses han he
discon inui y design.
We ind ha while he p og am wo ks on a e age, i does no wo k o all.
The i s s a egy (RDiT) yields small inc eases o a mos 15 pe cen in a e age
mon hly ansi ion a es ou o Minijobs in o highe paying, egula employmen a
he ime o he Midijob in oduc ion. Howe e , hese changes a e concen a ed in
he small g oup o male Minijob holde s. The second s a egy (DID) con i ms hese
pa e ns: he e o m e ec on annual ansi ions ou o Minijobs is signi ican ly
la ge in he ea men han in he con ol g oup. Non-ma ied indi iduals wi h low-
income ax bu dens espond mo e s ongly o he e o m han seconda y ea ne s
in ma iages— ypically emales—who a e subjec o high-income ax a es. The
esul s o bo h iden i ica ion s a egies a e obus o a ious speci ica ion changes
and sample adjus men s. O e all, Midijobs we e in oduced o phase ou Minijob
subsidies and o educe bunching a he Minijob ea nings h eshold. This objec-
i e was missed o mos Minijob employees. The Minijob ap did no disappea
a e he in oduc ion o Midijobs. These indings ha e a clea akeaway: in coun-
ies wi h household axa ion, phase-ou policies om Minijob- ype no ches mus
accoun o he en i e ax domain (including he income ax-induced po ion o he
no ch) o be e ec i e.
These indings complemen he esul s o Tazhi dino a (2020) and Gudgeon and
T enkle (2024). Bo h pape s ocus on Minijob employmen and s udy labo sup-
ply elas ici ies a he Minijob ea nings ceiling. Tazhi dino a (2020) uses bunch-
ing app oaches o de e mine annual labo supply elas ici ies be ween 1999 and
2010. She inds la ge labo supply elas ici ies o men han o women and s ong
inc eases in labo supply elas ici ies o single indi iduals a e 2003. Gudgeon and
T enkle (2024) ocus on a sample o ma ied women only and s udy he esponsi e-
ness o labo supply o shi s in he Minijob ea nings ceilings in 2003 and 2013 o e
ime. The au ho s ind subs an ial delays in ea nings esponses and a gue ha labo
demand ic ions a enua e es ima es o in ensi e ma gin labo supply elas ici ies.
We add o hese con ibu ions by ocusing on he e ec o Midijobs; hey may be
a mechanism o phase ou Minijob subsidies and o acili a e ansi ions o egula
employmen .
Ou esea ch connec s o se e al s ands o he li e a u e. Fi s , we con ibu e o
he in e na ional li e a u e on he e ec i eness o making wo k pay policies such as
he EITC in he Uni ed S a es o he Uni ed Kingdom ax c edi p og ams. Nume -
ous s udies in es iga e whe he he subsidy p og ams enable bene icia ies o g ow
1504
A.He ge , R.T.Riphahn
Fig. 1 Dis ibu ion o g oss mon hly ea nings (Ap il 2003–Decembe 2012). No e: The g aph ep esen s
he bo om pa o he g oss ea nings dis ibu ion o all employed indi iduals egis e ed wi h he unem-
ploymen insu ance. Using all jobs obse ed be ween Ap il 2003 and Decembe 2012 i shows he num-
be o employees pe 50 Eu o bin o mon hly g oss ea nings. Fo depic ions o annual g oss ea nings
dis ibu ions, see Appendix Fig.A.1. Sou ce: SIAB, own calcula ions
Fig. 2 Ne ea nings wi h and wi hou Midijob subsidy by income ax bu den (2003). No e: The g aph
ske ches ne mon hly ea nings along he de elopmen o mon hly g oss ea nings. Up o g oss ea nings
o 400 Eu o pe mon h Minijobs elimina e any di e ence be ween g oss and ne wa nings. Beyond he
Minijob ea nings h eshold he ed (blue) lines indica e he si ua ion be o e (a e ) he in oduc ion o
Midijobs. The dashed lines assume an a e age income ax a e o 30 pe cen whe eas he s aigh lines
assume ze o income axes. Sou ce: Own illus a ion
1505
Phasing ou pay oll ax subsidies
ou o hei need o suppo by expanding hei labo supply.4 Only ew con ibu-
ions add ess he ole o bene i phase-ou . Eissa and Liebman (1996) s udy bene i
phase-ou in he EITC. In e es ingly, hey ind single mo he s o be a he un espon-
si e o inc easing ma ginal ax a es in he phase-ou egion o he p og am.5 Leigh
(2007) in es iga es he ole o phase-ou a es using he 1999 e o m o he Uni ed
Kingdom’s ax c edi . He con i ms ha lowe phase-ou a es ha e posi i e impac s
on labo supply. Compa ing he e ec s o he U.S. EITC and he Ge man Minijob
p og am Be hold and Coban (2013) conclude ha in con as o he EITC, he Ge -
man p og am was ine ec i e in suppo ing low-income ea ne s. Howe e , so a ,
i is s ill un esol ed whe he he phasing ou o he Minijob subsidy by means o
Midijobs succeeds in encou aging highe -ea ning employmen . We a e he i s o
add ess his issue.
Second, we con ibu e o in e na ional esea ch on he employmen e ec s o
pay oll ax subsidies. Mos s udies in es iga e he ex ensi e ma gin o labo supply
and ind no employmen e ec s in esponse o changes in pay oll axes. Saez e al.,
(2019, p.1) a gue ha i is “ ecei ed wisdom” ha he pay oll ax incidence alls on
wo ke s’ ne wages.6 We add o his li e a u e by ocusing on he in ensi e ma gin o
labo supply.
A hi d b anch o s udies in es iga es he s epping s one cha ac e o Minijobs
hemsel es. The 2003 e o m ende ed Minijob employmen mo e a ac i e o
incen i ize he labo ma ke en y o hose p e iously no employed: lawmake s
hoped o Mini- and Midijobs o become s epping s ones in o egula employmen .
Se e al s udies e alua e he p og ams and conclude ha i is unlikely ha he p o-
g ams ac as a s epping s one.7
The li e a u e on Midijobs is limi ed and la gely desc ip i e. Mos s udies explain
he Midijob ins umen and i s u iliza ion.8 Bach e al., (2018a, 2018b) conclude
ha he 2019 expansion o he Midijob ea nings ceiling om 850 o 1,300 Eu o pe
mon h may e en wo sen he pa - ime employmen ap.
Ou analyses con ibu e o he li e a u e in se e al ways. Fi s , while he li e a u e
on pay oll ax subsidies ocuses on he ex ensi e ma gin o labo o ce pa icipa ion,
6 This is b oadly suppo ed in he li e a u e, see e.g., G ube (1997), Ande son and Meye (1997, 2000),
Ko keamäki and Uusi alo (2009), Hu unen e al. (2013), and Bennma ke e al. (2009) who p o ide e i-
dence om Chile, he Uni ed S a es, Finland, and Sweden, espec i ely.
7 See e.g., Fe ig e al. (2005), Fe ig and Klu e (2006), and F eie and S eine (2008) o mo e ecen ly
Caliendo e al. (2016), Lie zmann e al. (2017), and Ca illo-Tudela e al. (2021).
8 See, e.g., Fe ig and Klu e (2006), B and (2005, 2006), He zog-S ein and Sesselmeie (2012),
Be hold and Coban (2013), Fich l (2015), Kelle and Sei e (2015), Sei e (2017), Dundle e al.
(2019), Kelle e al. (2021), and He ge and Riphahn (2022).
5 B owning (1995) calcula es po en ial nega i e income e ec s induced by high ma ginal axes in he
phase-ou egion o he EITC. Howe e , empi ical s udies did no suppo his a ionale (see also Meye
2002). LaLumia (2009) in es iga es he epo ing beha io o sel -employed in esponse o he incen-
i es o he EITC p og am and con i ms ha he phase-ou egion gene a es less o a esponse han he
phase-in egion. T ampe (2007) summa izes he li e a u e which ha dly ound nega i e e ec s o he
EITC phase-ou egion. He inds small nega i e e ec s (howe e , o a discussion see Hoynes 2007 and
T ampe 2008).
4 See, e.g., Ba gain and O sini (2006), Ba gain e al. (2010), Blundell (2000), F ancesconi and an de
Klaauw (2007), o G ogge (2003).
1506
A.He ge , R.T.Riphahn
we s udy he in ensi e ma gin o labo supply. Small jobs es ic o e all labo sup-
ply, limi human capi al in es men s and ca ee p ospec s (Beckmann, 2020), and
inhibi he accumula ion o pension claims. The e o e, i is impo an o unde s and
mechanisms ha suppo ansi ions o egula employmen . Second, he li e a u es
on ‘making wo k pay’ and on he s epping s one cha ac e o small jobs ha e no ye
add essed he ele ance o he phasing ou o pay oll ax subsidy p og ams. I he
Midijob p og am e ec i ely suppo s ansi ions o ex ended labo supply i could
cons i u e a use ul policy o o he (na ional) labo ma ke s, as well.9 Finally, we a e
he i s o s udy he e ec o he in oduc ion o Midijobs on he p opensi y o lea e
Minijob employmen , i.e., he e ec i eness o Midijobs as a labo ma ke policy.
The s uc u e o his pape is as ollows. In sec ion wo, we p o ide ins i u ional
de ail on he Mini- and Midijob p og ams and hei de elopmen o e ime. We
desc ibe ou empi ical analysis based on he egula o y discon inui y in Sec .3 and
ou di e ence-in-di e ences analyses in Sec .4. In Sec .5 we o e a policy discus-
sion and conclude.
2 Backg ound
2.1 Ins i u ions and e o ms
Minijob employees, i.e., hose ea ning no mo e han he Minijob ea nings h eshold
a e exemp om o he wise manda o y social insu ance con ibu ions and income
axes. Ins ead, hei employe s pay a ixed sha e o g oss Minijob ea nings o social
insu ance and ax au ho i ies ( o de ails see, e.g., Collischon e al., 2021).10 This
egula ion has exis ed since he ea ly days o he Ge man social insu ance sys em
(1893) o limi he bu eauc a ic bu den o small jobs (BMAS 2018, p.110). O e
ime, Minijob egula ions we e modi ied wi h a ying objec i es.
A he Minijob ea nings ceiling, he sudden ade-ou o he wage subsidy inhe -
en in Minijob employmen gene a es a la ge, discon inuous change in ax and social
insu ance liabili ies. When hey ea n below he Minijob ea nings ceiling employees
pay nei he social insu ance con ibu ions no income axes. When hey ea n abo e
he Minijob ea nings ceiling, income axes on o al ea nings plus social insu ance
con ibu ions o abou 20 pe cen become payable (see Fig.2). This leads o a (la ge)
educ ion in ne mon hly ea nings. While social insu ance con ibu ions a e ele an
o all employees, income axes di e o join and indi idual ile s; in Sec .4, ou
iden i ica ion s a egy akes ad an age o his he e ogenei y. Those iling indi idu-
ally bene i om a sizable ini ial ax allowance and do no ace income axes imme-
dia ely a e exceeding he ea nings h eshold. In con as , join ile s, i.e., seconda y
9 Di Po o e al. (2022), Dolado e al. (2021), and Sca e (2021) ecen ly s udied ze o-hou s con ac s
and casual wo k in I aly, he Uni ed Kingdom, and Aus alia, espec i ely. Jobs based on hese con ac s
a e simila o Ge man Minijobs.
10 Minijobs also ake he o m o sho - e m employmen ela ionships which do no ex end beyond (cu -
en ly) 70days, independen o ea nings. We dis ega d his second ca ego y o Minijob employmen ,
which is much less p e alen .
1507
Phasing ou pay oll ax subsidies
ea ne s in ma iages, may be a ec ed by sizeable ax a es on hei en i e ea nings
as soon as hey exceed he Minijob ea nings h eshold. Thus, he Minijob ea nings
h eshold gene a es a discon inui y in he le el (a no ch) and he slope (a kink) o he
ne ea nings schedule (Kle en, 2016). This “Minijob ap” ba s inc eases in labo
supply and ea nings and causes subs an ial bunching in he ea nings dis ibu ion
(see Fig.1) (Gudgeon & T enkle, 2024; Tazhi dino a, 2020).
The e o ms implemen ed on Ap il 1, 2003 aised he mon hly Minijob ea n-
ings ceiling om 325 o 400 Eu o, abolished a limi o 15 wo king hou s pe week,
inc eased employe con ibu ion a es om 22 o 25 pe cen o Minijob ea nings,
and–mos in e es ing o us–newly in oduced he Midijob p og am.11 The in en ion
o he 2003 e o m was (a) o educe illici moonligh ing by making legal small jobs
mo e a ac i e and (b) o o e s epping-s one employmen oppo uni ies o he
unemployed and oppo uni ies o upwa d mobili y o hose in ma ginal employ-
men (Eichho s e al., 2012). The in oduc ion o Midijobs did no a ec employe s
who con inued o pay egula social insu ance con ibu ions o abou 20 pe cen on
ea nings beyond he Minijob ea nings ceiling.
Midijobs we e in oduced o incen i ize labo supply beyond he Minijob ea n-
ings ceiling; Midijobs en ail pay oll ax subsidies o employees ea ning be ween
( hen) 400 and 800 Eu o pe mon h. These subsidies phase ou as ea nings inc ease.
Ins ead o ull egula social insu ance con ibu ion a es o 20 pe cen , he Midijob
a es inc eased on a sliding scale s a ing a abou 4 pe cen o mon hly ea nings o
400 Eu o and eaching he unsubsidized le el o 20 pe cen o mon hly ea nings o
800 Eu o. A he same ime, Midijob employees a e ully liable o income axes on
hei o al ea nings (see Fig.2).12
In 2013, he mon hly Minijob and Midijob ea nings limi s we e aised o 450
and 850 Eu o, espec i ely ( o la e adjus men s see Appendix TableA.1). A e
his e o m, social insu ance con ibu ion a es o Midijobs commenced a abou
10 pe cen o mon hly ea nings s a ing a 451 Eu o and inc eased o 20 pe cen a
mon hly ea nings o 850 Eu o. Minijobs and Midijobs ha e been used in ensely.
The in oduc ion o he Midijob subsidy o social insu ance con ibu ion a es
was in ended o a enua e he no ch in he ne ea nings dis ibu ion, phase ou he
Minijob subsidy, and incen i ize labo supply and ea nings beyond he Minijob
ea nings ceiling. We in es iga e whe he Midijobs e ec i ely educed he ba ie s
o exi ing subsidized Minijob employmen and en e ing highe -ea ning employmen .
11 The ele an legisla ion (Zwei es Gese z ü Mode ne Diens leis ungen am A bei sma k , Ha z II) was
passed on Decembe 23, 2002 as an ea ly elemen o a bundle o labo ma ke e o ms. Fo a compac
e iew o he e o ms see, e.g., Ca illo-Tudela e al. (2021).
12 Figu eA.3 compa es he implici Minijob and Midijob subsidy o he pa e ns o he US EITC. Fig-
u eA.3.1 shows he si ua ion be o e he 2003 e o m: Minijob subsidies acc ued in he 0-325 Eu o ea n-
ings ange. They inc eased wi h household ax a es and e mina ed ab up ly a he Minijob ea nings
limi ; depending on he ax egime hey amoun o abou 50 pe cen o g oss ea nings. Figu eA.3.2
shows he mon hly c edi a e he 2003 e o m. He e, he Midijob subsidy o ea nings beyond he Mini-
job ea nings limi (now 400 Eu o) is in oduced which is independen o ax a es. I s a s a abou 16
pe cen o g oss ea nings (abou 64 Eu o) and declines along he Midijob ea nings ange. Figu e A.3.3
con as s hese pa e ns wi h he EITC egime o 2003 whe e subsidies a e phased ou wi hou ab up
b eakpoin s as implied by he Minijob ea nings h eshold.
1508
A.He ge , R.T.Riphahn
A p io i, as he Midijob subsidy only co e s social insu ance con ibu ions bu no
income axes, he policy le a subs an ial ax no ch o seconda y ea ne s. The e-
o e, we should no expec he Midijob e o m o a ec ansi ions om Minijobs o
seconda y ea ne s much.
2.2 U iliza ion pa e ns andchanges in heea nings dis ibu ion
While Mini- and Midijobs co e di e en ea nings anges and ea u e di e en u i-
liza ion pa e ns, bo h a e cha ac e ized by a p e alence o emale wo ke s (see,
e.g., Oschmiansky and Be hold (2020), Tazhi dino a (2020), He ge and Riphahn
(2022)). Based on weigh ed su ey da a and in he age g oup ha we s udy (30–59)
mo e han 90 pe cen o Minijobbe s a e emale. Among hese emales, abou 90
pe cen a e ma ied and hus gene ally seconda y ea ne s who ace income axes
when ea ning abo e he Minijob ea nings ceiling.
Typically, Minijobs pay low hou ly wages. Classic employe s o Minijobbe s
a e in he hospi ali y indus y (ba s, es au an s), cleaning and building se ices, o
e ail. Minijob employmen is concen a ed in small es ablishmen s (0–9 employ-
ees) (Collischon e al., 2021). Minijobs a e o en in o mal wi h limi ed du a ion, no
w i en con ac s, i egula wo k hou s, and on-call employmen (B uckmeie e al.,
2018). Bachmann e al. (2012) asked Minijobbe s why hey use a Minijob (wi h
mul iple answe s possible); almos 60 pe cen we e mo i a ed by addi ional ea n-
ings, 15 pe cen by ga he ing wo k expe ience, 14 pe cen indica ed ha his was he
only job hey could ind, and 14 pe cen we e mo i a ed by he possibili y o wo k
lexible hou s. Fo emale Minijobbe s being able o combine wo k and amily li e
as well as lexible hou s we e subs an ially mo e impo an han o male Minijob-
be s (31 pe cen s. 17 pe cen ).
Midijobs a e ypically pa - ime posi ions. Abou 62 pe cen a e held by emales.
Males use Midijobs ypically when hey a e young. 46 (61) pe cen o emale (male)
Midijobbe s a e younge han 35 and 44 (29) pe cen a e aged 35–54 (He ge &
Riphahn, 2022). In Figu eA.4, we desc ibe o he calenda yea s 2002–2017 om
which s a es o o igin indi iduals en e Midijob employmen ; o he pe iod be o e
he Midijob in oduc ion, we use he Midijob ea nings ange. Based on adminis a-
i e da a, no mo e han 20 pe cen o Midijobs o igina e om Minijob employmen .
The e o m o Ap il 1, 2003 no only in oduced he Midijob subsidy o social
insu ance con ibu ions. I also shi ed he h eshold o he Minijob ea nings limi
om 325 o 400 Eu os pe mon h. We p esen he dis ibu ion o mon hly g oss
ea nings a e he e o m (2003–2012) in Fig.1. Figu e3 shows he change in he
dis ibu ion o mon hly g oss ea nings be ween 2002 and 2004. The equency o
jobs ea ning 325–400 Eu o inc eased, ha o jobs ea ning 250–325 Eu o declined.
In e es ingly, he equency o employmen s in he Midijob ea nings ange be ween
400 and 800 Eu o did no inc ease.
1515
Phasing ou pay oll ax subsidies
Figu e2 shows he ela ionship be ween g oss and ne income. The solid lines
desc ibe he si ua ion o a pe son wi h no income ax obliga ion, and he dashed
lines desc ibe he si ua ion o a pe son wi h ela i ely high income ax a es, bo h
be o e and a e he e o m. Fo bo h g oups, ne income alls a he Minijob ea n-
ings ceiling. The decline is la ge o he indi idual wi h income ax obliga ions.
The no ch cha ac e izes he Minijob employmen ap. In bo h cases, he e o m
(blue lines) a enua es he d op in ne ea nings and educes he disincen i es o
expand labo supply. A e he e o m, he no ch almos disappea s o indi iduals
wi hou income ax obliga ions bu no o hose subjec o income axes. The e-
o e and based on he esul s o he discon inui y analyses in Table3 and Fig.5, we
hypo hesize ha indi iduals wi h low income ax a es (e.g., non-ma ied indi idu-
als) espond mo e s ongly o he e o m han hose wi h high income ax a es (e.g.,
seconda y ea ne s in ma iages) because he e o m made a ele an di e ence o
he o me bu no o he la e (Fig.6).
To es his hypo hesis ou ou come o in e es (Y) again is a dicho omous indica-
o o whe he an indi idual i in a Minijob in pe iod lea es he Minijob be ween
pe iod and + 1 and ansi s o employmen wi h ea nings abo e he Minijob
h eshold. We dis inguish he si ua ion be o e and a e he e o m o Ap il 1, 2003
Table 2 RDiT Es ima es o T ansi ions ou o Minijobs–SIAB. Sou ce: SIAB (2017) and own calcula-
ions
No e: Linea eg essions, s anda d e o s clus e ed a he indi idual le el; * p < 0.10, ** p < 0.05,
***p < 0.01. The columns en i led “Coe ” p o ide he es ima e o α2 in Eq.(1), “N” and “P e- e o m
mean Y” p o ide he numbe o obse a ions and p e- e o m mean o he ou come o each sample.
“RE” is he ela i e e ec , i.e., he a io o he coe icien es ima e o α2 and he p e- e o m mean o he
dependen a iable and cha ac e izes he ela i e e ec size. The models wi hou added con ols (in ows
1 and 4) con ol o ‘a e ’ indica o , he e-cen e ed ime end (linea o quad a ic), hei in e ac ion(s),
and a cons an e m. The ec o o basic con ol a iables (in ows 2 and 5) addi ionally accoun o
gende , 5 age g oup indica o s, Eas Ge man esidence, and Ge man ci izenship. The ec o o ex ended
con ols (in ows 3 and 6) addi ionally accoun s o 3 indica o s o educa ional a ainmen , 7 indica o s o
enu e, 8 indica o s o occupa ion, 4 indica o s o i m size, and 9 indica o s o indus y. Sample A con-
side s all Minijob-mon hs obse ed be ween Ap il 1, 2002 and Ma ch 31, 2004. Sample B d ops hose
Minijobs ha we e s a ed a e Ap il 1, 2003 and Sample C d ops hose Minijobs ha we e s a ed a e
Dec. 31, 2002. Sample A-2013 eplica es Sample A-2003 a ound he e o m da e o Janua y 1, 2013
Sample A—2003 Sample B—2003 Sample C—2003 Sample A—2013
(1) (2) (3) (4)
Speci ica ion Coe RE Coe RE Coe RE Coe RE
1 Linea no con ols 0.0005 6.5% 0.0001 0.7% −0.0007* -8.2% −0.0003 −1.9%
2 Linea basic con ols 0.0004 4.7% 0.0001 1.1% −0.0006 −7.1% −0.0003 −1.7%
3 Linea ex . con ols 0.0001 1.2% 0.0003 3.5% 0.0001 1.2% −0.0005 −3.3%
4 Quad a ic no con ols 0.0013** 14.8% 0.0011* 13.1% 0.0004 5.3% −0.0014* −8.4%
5 Quad . basic con ols 0.0012* 14.1% 0.0011* 13.0% 0.0005 5.5% −0.0014* −8.4%
6 Quad . ex . con ols 0.0009 10.6% 0.0013** 15.3% 0.0007 8.6% −0.0015** −8.9%
N 853,241 751,217 715,313 1,062,892
P e- e o m mean Y 0.0085 0.0085 0.0084 0.0167
1516
A.He ge , R.T.Riphahn
(a e ) and we di e en ia e g oups ha we e a ec ed o di e en ex en s: second-
a y ea ne s in ma ied couples a e subjec o high income ax a es; we conside
hem as ou con ol g oup obse a ions ( ea = 0).23 Fo hem, he Midijob subsidy
o social insu ance con ibu ions ha dly educes he ele an no ch in he ea nings
dis ibu ion. In con as , non-ma ied indi iduals wi h o wi hou a pa ne in he
household enjoy an indi idual ax allowance ha exemp s annual ea nings o up o
abou 9,000 Eu o om income ax paymen s (G und eibe ag). Fo hese indi id-
uals ma ginal and a e age income ax a es a he Minijob ea nings h eshold a e
low and—depending on o he sou ces o income—may e en be ze o. The e o e,
he in oduc ion o he Midijob subsidy cons i u es a ele an educ ion in he no ch
in hei ne ea nings dis ibu ion. We conside hem as he ea men g oup o he
e o m ( ea = 1).
We use a s anda d DID model and conside co a ia es (X) o educe he esidual
a iance. Le μ be a andom e o . We es ima e he coe icien ec o s α and β in his
model:
Ou e ec o in e es is he es ima e o α3. I indica es whe he indi iduals wi h
low income ax bu dens ( ea = 1) changed hei p opensi y o ansi ion ou o he
Minijob ea nings ange a e he e o m mo e han indi iduals wi h high income ax
bu dens ( ea = 0). I such a di e ence exis s i sugges s ha he e o m educed he
Minijob ap a leas o pa o he popula ion.
The DID es ima e ep esen s a causal e o m e ec i se e al condi ions a e me :
i s , wi hou he e o m, he ime end in he p opensi y o lea e a Minijob o
highe -ea ning employmen should ha e de eloped in pa allel o indi iduals in he
ea men and he con ol g oup. We inspec he e idence on p e- e o m ends in
he nex sec ion. Second, he e o m should no a ec ea men o con ol g oups in
ways o he han h ough he in oduc ion o he Midijob subsidy. This equi emen is
iola ed as he e o m no only in oduced he Midijob pay oll ax subsidy o ea n-
ings abo e he Minijob ea nings h eshold bu also inc eased he Minijob ea nings
h eshold i sel om 325 o 400 Eu o pe mon h. Howe e , as he inc eased ea nings
h eshold a ec ed ea men as well as con ol g oup obse a ions, i will bias he
es ima e o he e ec o he Midijob in oduc ion only i he wo g oups espond di -
e en ly o he h eshold shi . In ha case, α3 pa ly e lec s he e ogeneous esponses
(2)
Y
i +1=𝛼0+𝛼1a e i +𝛼2 ea i +𝛼3
(
a e i ⋅ ea i
)
+𝛽1Xi +𝜇
i
Fig. 4 Mon hly ansi ion a e om Minijob Employmen (Ap il 2002-Ma ch 2004). a Sample A (all
Minijob obse a ions in he obse a ion window). b Sample B (Sample A wi hou Minijobs s a ed a e
Ap il 1, 2003). c Sample C (Sample B wi hou Minijobs s a ed a e Dec. 31, 2002). No e: The g aphs
ep esen he de elopmen o mon hly a e age ansi ion a es om Minijob o egula employmen . The
sample includes all who hold a Minijob as hei main employmen wi hou being egis e ed unemployed
in a gi en mon h. The mon hly ansi ion a es a e de-seasonalized. Fo a ep esen a ion o hese g aphs
by gende see Appendix Figu eA.6. Sou ce: SIAB (2017) and own calcula ions
▸
23 Gudgeon and T enkle (2024) ocus hei analysis o Minijob e o ms on ma ied women exac ly
because hey a e subjec o a la ge no ch in hei po en ial ea nings. Un o una ely, hese au ho s’ admin-
is a i e da a is no a ailable o us.
1517
Phasing ou pay oll ax subsidies
1518
A.He ge , R.T.Riphahn
Table 3 RDiT Es ima es—SIAB—by gende and p edic ed seconda y ea ne s a us. Sou ce: SIAB (2017) and own calcula ions
No e: See Table2. Columns 3 and 4 spli he sample based on he p edic ed p obabili y o being non-ma ied, i.e., single ea ne s. ma ied, i.e., likely seconda y ea ne ,
which was based on he linea eg ession es ima ed on SOEP da a (see Appendix B)
Sample A—2003 Sample A—2003
(1) Men (2) Women (3) P ed. Non-Ma ied (4) P ed. Ma ied
Speci ica ion Coe RE Coe RE Coe RE Coe RE
1 Linea no con ols 0.0083 *** 73.6% − 0.0004 − 5.2% 0.0152 *** 85.9% − 0.0001 − 0.9%
2 Linea basic con ols 0.0078 *** 69.0% − 0.0005 − 6.1% 0.0149 *** 84.2% − 0.0002 − 2.2%
3 Linea ex . con ols 0.0061 *** 54.0% − 0.0007 − 8.5% 0.0129 *** 72.9% − 0.0005 − 5.5%
4 Quad a ic no con ols 0.0063 *** 55.7% 0.0003 3.3% 0.0135 *** 76.3% 0.0005 5.4%
5 Quad . basic con ols 0.0092 *** 81.4% 0.0002 2.4% 0.0136 *** 76.8% 0.0005 5.0%
6 Quad . ex . con ols 0.0071 *** 62.8% 0.0001 1.2% 0.0118 *** 66.7% 0.0002 2.1%
N 95,277 757,964 54,183 890,793
P e- e o m mean Y 0.0113 0.0082 0.0177 0.0092
1519
Phasing ou pay oll ax subsidies
o he change in he h eshold. To gauge he ele ance o his shi in he Minijob
ea nings ceiling o Minijob exi s we again exploi a la e adjus men in he Mini-
job ea nings h eshold: on Janua y 1, 2013, he mon hly Minijob/Midijob ea nings
h esholds inc eased om 400/800 o 450/850. We es his e o m’s e ec on ansi-
ions ou o Minijobs o app oxima e he impac o he 2003 change in he ea nings
ceiling om 325 o 400 Eu os.
A hi d iden i ica ion equi emen is ha he e a e no sys ema ic changes in he
composi ion o he wo k o ce in esponse o he ea men . Fi s , we examined
whe he indi iduals migh swi ch be ween ea men and con ol g oups in esponse
o he e o m. We ound ha indi iduals changed hei ma i al s a us in bo h di ec-
ions in simila magni udes be o e and a e he e o m. Second, we e alua ed
whe he he e we e composi ional changes wi h espec o obse able cha ac e is ics
wi hin he ea men o con ol g oup o e ime. Such changes migh indica e di -
e en ial selec ion in o he wo g oups in esponse o he e o m. TableA.10 p esen s
he p- alues o es s o he hypo hesis ha he mean alues o ea men and con ol
Fig. 5 Mon hly ansi ion a e om Minijob Employmen (Ap il 2002-Ma ch 2004). a By gende (males
le panel, emales igh panel). b By p edic ed single ea ne s a us (single ea ne s le panel, second-
a y ea ne s igh panel). No e: The g aphs ep esen he de elopmen o mon hly a e age ansi ion a es
om Minijob o egula employmen . The sample includes all who hold a Minijob as hei main employ-
men wi hou being egis e ed unemployed in a gi en mon h. The mon hly ansi ion a es a e de-sea-
sonalized. Bo h igu es use subsamples om Sample A. Please no e ha he y-axis o Panel 5b is scaled
di e en om Panel 5a. Sou ce: SIAB (2017) and own calcula ions
1520
A.He ge , R.T.Riphahn
g oup cha ac e is ics changed o e ime. We obse e some mino adjus men s bu no
majo changes in cha ac e is ics.24
The inal iden i ica ion equi emen is he absence o an icipa ion e ec s. I in
esponse o he e o m, Minijobbe s pos poned hei ansi ion o highe -ea ning
employmen o changed he ake up o Minijobs al oge he his biases ou es i-
ma es i i a ec s ea men and con ol g oup obse a ions di e en ly. The e o m
was passed in o law on Decembe 23, 2002, and became e ec i e on Ap il 1,
2003, which does allow o po en ial an icipa o y beha io (see ou discussion in
Sec .3.1). E en hough he e is no a ionale as o why ea men and con ol g oup
obse a ions migh di e in hei esponse, we ne e heless inspec whe he ou es i-
ma es a e sensi i e o he ime window o ou sample.
This iden i ica ion s a egy exploi s po en ial he e ogenei y in ea men e ec s.
Compa ing he beha io al esponses o mo e ( ea men g oup) and less (con ol
g oup) s ongly a ec ed indi iduals does no indica e he o e all a e age e ec o
he Midijob in oduc ion. Howe e , i can o e e idence as o whe he he e is an
e ec a all i hose mos a ec ed espond di e en ly om hose leas a ec ed.
Fig. 6 E en s udy analysis o pa allel ends (SOEP). No e: The igu e depic s es ima ed coe icien s
om linea eg essions and hei con idence in e als. The ou come o ansi ions om Minijobs o eg-
ula employmen is eg essed on yea indica o s and hei in e ac ion wi h he ea men indica o . We
depic 90 pe cen con idence in e als o he in e ac ion e m coe icien s using s anda d e o s ha a e
clus e ed a he indi idual le el. Sou ce: SOEP wa e 35 and own calcula ions
24 In he con ol g oup, he sha e o emale obse a ions d ops om 97 o 95 pe cen and he sha e wi h
e ia y educa ion inc eases om 7 o 9 pe cen . In he ea men g oup, he sha e o obse a ions in Eas
Ge many inc eases and he g ouping o Minijobbe s in he smalle i m size ca ego ies is shi ed some-
wha .
1521
Phasing ou pay oll ax subsidies
4.2 Su ey da a o di e ence‑in‑di e ences analysis (SOEP)
Ou su ey da a a e aken om he Ge man Socio-Economic Panel S udy (SOEP),
an annual household panel su ey collec ed since 1984 (Goebel e al., 2019).25 We
use da a co e ing he yea s 2001–2006 o e alua e he 2003 e o m. Again, we
es ic ou sample o indi iduals aged 30–59 o omi s uden s and e i ees.26
We a e in e es ed in whe he he e o m a ec ed he p opensi y o ansi ion om
Minijob employmen o egula employmen di e en ly o mo e and less a ec ed
indi iduals. Ou sample conside s indi iduals employed in a Minijob as hei main
employmen a he ime o he annual su ey. Since 2001, he su ey asks di ec ly
abou Minijob employmen . We use his sel - epo ed in o ma ion and conside only
hose indi iduals o be in a Minijob who addi ionally indica e o ea n no mo e han
he Minijob ea nings h eshold.27 This lea es us wi h 2,736 pe son-yea obse a-
ions o 1,255 di e en indi iduals in Minijob employmen in he yea s 2001–2006.
The 1,255 indi iduals a e obse ed in 1,604 di e en Minijob employmen ela ion-
ships o e ime. Un o una ely, he sample size is much smalle han in he adminis-
a i e da a. Howe e , as he adminis a i e da a do no p o ide in o ma ion on ma i-
al s a us his analysis equi es su ey da a. Table4 p o ides desc ip i e s a is ics
on ou sample. The as majo i y o ou Minijobbe s is emale (94.4 pe cen ) wi h
he la ges g oup aged 35–44 (mean age is 43). Rela i e o agg ega e popula ion
sha es Minijobs a e used ela i ely mo e in ensely in Wes han in Eas Ge many
and mo e by Ge man ci izens han by non-ci izens. In ou sample, abou 90 pe cen
o he obse a ions a e ma ied and hus in ou con ol g oup. The ea men g oup
comp ises hose who a e single (including hose in cohabi ing couples), di o ced,
o widowed. We obse e 173 (276) and 1,082 (2,460) di e en indi iduals (pe son-
yea obse a ions) in he ea men s. con ol g oups, espec i ely ( o desc ip i e
s a is ics on addi ional con ols see Appendix TableA.11).28
Ou dependen a iable indica es whe he a pe son held a Minijob in pe iod and
in pe iod + 1 ansi ed o egula employmen paying social insu ance con ibu ions
and ea ning abo e he Minijob ea nings limi ; we e alua e ansi ions be ween 2001
and 2006. The a e age annual ansi ion a e is 10.3 pe cen . The las wo columns
o Table4 desc ibe he mean ansi ion a es o di e en g oups. We obse e highe
a es a e han be o e he e o m da e. As expec ed, ma ied indi iduals ( ea = 0)
27 Due o he second es ic ion, we lose 663 o 3,399 obse a ions o 19.5 pe cen o hose who indi-
ca ed o wo k in a Minijob. We also d op wo indi iduals o whom in o ma ion on ma i al s a us, ou
ea men indica o , is missing. We do no use in o ma ion om ea lie su ey yea s because hey applied
a di e en su ey ques ion o collec employmen s a us in o ma ion.
25 We use SOEP 35 (1984–2018), DOI h ps:// doi. o g/ 10. 5684/ soep- co e. 35.
26 We accoun o o e sampling and non- esponse in he da a by applying he c oss-sec ional sample
weigh s p o ided wi h he SOEP da a. In con as o ou analysis o egis e da a we do no omi Minijob-
be s who a e unemployed he e due o he esul ing small sample sizes.
28 We use a ime a ying ea men assignmen whe e indi iduals en e he con ol g oup upon ma iage.
Ideally, he g oup assignmen would be ixed p io o he ea men , e.g., based on ma i al s a us in 2002
o 2001. Howe e , his educes ou sample size by abou hal . The es ima es a e obus in e ms o hei
signs bu ha e la ge s anda d e o s. We conside i implausible ha changes in ma i al s a us a e con-
nec ed o he in oduc ion o Midijobs.
1522
A.He ge , R.T.Riphahn
ha e a much lowe a e age ansi ion a e han non-ma ied indi iduals ( ea = 1).
The ansi ion a e o men is highe han ha o women (16 s. 11 pe cen ). As in
he SIAB da a, he likelihood o lea ing Minijobs declines wi h age.
Figu eA.7 shows he de elopmen o ansi ion a es sepa a ely o ou ea men
and con ol g oups using weigh ed da a. Un o una ely, ou da a p o ide only wo
annual obse a ions p io o he e o m, i.e., 2001 and 2002. Howe e , in hese yea s
he de elopmen o ansi ion a es ou o Minijobs is simila o ea men and con-
ol g oups which sugges s pa allel pa hs p io o he e o m.29
4.3 Resul s: di e ence‑in‑di e ences analyses
Table5 shows ou i s se o DID esul s. Column 1 o e s esul s wi hou con ol
a iables and shows ha he es ima e o α3 is s a is ically signi ican and posi i e.
I sugges s ha a e he e o m he ea men g oup o non-ma ied indi iduals
inc eased hei p opensi y o lea e a Minijob o egula employmen by abou 14
pe cen age poin s mo e han he con ol g oup o ma ied pe sons; ela i e o a mean
ansi ion a e o abou 10 pe cen his is a a he la ge uncondi ional e ec . The
esul is con i med in columns 2 and 3, whe e we i s conside con ols o basic
demog aphics (gende , age g oup, Eas Ge man esidence, and o eign ci izenship)
and hen add an ex ended se o con ols (see able no es and Appendix TableA.11
o desc ip i e s a is ics). Column 4 shows ha he es ima ion esul s a e obus
when we eplace he o e all ‘pos ’ e ec wi h a se o calenda yea ixed e ec s.30
TableA.12 p esen s he esul s o addi ional es s and desc ibes e ec he e oge-
nei ies. Column 1 shows he esul s a e omi ing hose obse a ions o which we
canno be su e whe he a po en ial ansi ion happened be o e o a e he e o m on
Ap il 1, 2003.31 The es ima es on he hus educed sample con i m he signi ican
posi i e ea men e ec . In columns 2 and 3 we e alua e he sensi i i y o he esul s
o he conside ed ime window o obse a ions. Fi s , we omi wo yea s o pos -
e o m obse a ions (column 2) and hen we add an addi ional pos - e o m obse -
a ion yea (column 3): ou main esul ha dly changes. In column 4 we omi male
obse a ions; he esul shows ha women espond subs an ially less o he e o m
han men, con i ming he esul s o ou RDiT analyses. Omi ing obse a ions wi h
29 In a sepa a e analysis, we es ima ed an e en s udy o desc ibe he p e- and pos - ea men di e ences
be ween ea men and con ol g oups. We eg essed he ou come on yea indica o s and hei in e ac-
ions wi h he ea men indica o bu wi hou addi ional con ols. Figu e6 shows he di e ences be ween
ea men and con ol g oups. Again, ou come di e ences p io o he e o m a e insigni ican , and ansi-
ion a es o he non-ma ied inc ease as e a e he e o m han hose o ma ied Minijobbe s.
30 We use sampling weigh s in he analyses o SOEP da a. The esul s in Table5 a e sensi i e o his
choice.
31 The unce ain y is due o he annual in e iew which in o ms only abou he s a us a he ime o he
in e iew bu no abou when a s a us change occu ed. We omi wo g oups o obse a ions: hose o
whom we know he Minijob s a us in 2002 bu do no know whe he he ansi ion o he 2003 s a us ook
place be o e o a e he e o m da e o Ap il 1, 2003 and hose o whom we know hei Minijob s a us
p io o Ap il 1, 2003 bu do no know whe he hei ansi ion o he 2004 s a us ook place be o e o
a e he e o m da e o Ap il 1, 2003.
1523
Phasing ou pay oll ax subsidies
Eas Ge man esidence as in column 5 e eals somewha la ge e ec s in Wes han
in Eas Ge many.32
Finally, we need o accoun o he ac ha he e o m o Ap il 1, 2003 shi ed
he Minijob ea nings ceiling om 325 o 400 Eu o pe mon h. I ea men ( he
non-ma ied) and con ol ( he ma ied) g oups esponded di e en ly o his change
his may bias ou inding. In o de o gauge he o e all ele ance o he ceiling shi
we conside he e o m o Janua y 1, 2013 when he Minijob ea nings ceiling was
inc eased again, his ime om 400 o 450 Eu o pe mon h. We e alua e he impac
o his e o m on changes in ansi ion p opensi ies. We use he same sample and
ea men de ini ions as be o e jus shi ing he obse a ion pe iod o 10yea s la e
(see Tables A.11 and A.13 o desc ip i e s a is ics).33 Table A.14 shows he esul s
on Minijob ansi ions o he pe iod 2011 o 2016. We e alua e he e o m he e o-
genei y o ea men and con ol g oups a ound he inc ease in he Minijob ea nings
ceiling om 400 o 450 Eu os on Janua y 1, 2013. In his case, he esul s yield
a nega i e and s a is ically insigni ican es ima e o α3. O e all, he ansi ion a es
Table 4 Desc ip i e S a is ics—Basic Con ols: 2003 Re o m Sample (SOEP). Sou ce: SOEP wa e 35
and own calcula ions
No e: The desc ip i e s a is ics desc ibe he sample o 2,736 pe son-yea obse a ions. The da a a e no
weigh ed
Desc ip i es Mean ansi ion a e when a iable has
Va iable Mean S d. De Value 0 Value 1
T ansi ion (0/1) 0.1031 0.3041 0.0000 1.0000
Pos (0/1) 0.5808 0.4935 0.0850 0.1179
T ea (0/1) 0.1009 0.3012 0.0995 0.1809
Female (0/1) 0.9441 0.2298 0.0952 0.2353
Age: 30–34 (0/1) 0.1648 0.3711 – 0.1537
Age: 35–39 (0/1) 0.2288 0.4201 – 0.1347
Age: 40–44 (0/1) 0.1988 0.3992 – 0.1116
Age: 45–49 (0/1) 0.1659 0.3721 – 0.1037
Age: 50–54 (0/1) 0.1411 0.3482 – 0.1108
Age: 55–59 (0/1) 0.1005 0.3007 – 0.0678
Eas Ge many (0/1) 0.0757 0.2645 0.1085 0.1667
Fo eign Na ionali y (0/1) 0.1137 0.3175 0.1148 0.1068
32 Ma i al s a us may be a weak p oxy o he income ax bu den. The e o e, we in es iga ed whe he i
migh ep esen al e na i e mechanisms, ins ead. We eplaced ou ea men indica o o no ma ied (T)
s. ma ied (C) indi iduals by se e al al e na i es. Fi s , o he sample o ma ied pe sons (N = 2,460)
we used an indica o o whe he a pe son has child en (C) o no (T). Second, we conside ed non-ma ied
indi iduals wi h (C) s. wi hou s able pa ne s (T). In nei he case did we ob ain s a is ically signi ican
ea men e ec s. Thi d, we compa ed non-ma ied (T) indi iduals only o ma ied indi iduals wi hou
child en (C) o sa egua d agains e ec s o childca e. He e, we con inue o ind signi ican posi i e ea -
men e ec s, suppo ing ou main esul s.
33 We d op 32 indi iduals o whom in o ma ion on ma i al s a us, ou ea men indica o , is missing in
he da a.
1524
A.He ge , R.T.Riphahn
inc eased sligh ly o he con ol g oup o ma ied indi iduals a e he e o m (see
ow 1) and ansi ion a es o he ea men g oup a e gene ally signi ican ly highe
( ow 2). Impo an ly, he e o m did no a ec he ela i e ansi ion a es o he wo
g oups. I hese pa e ns simila ly held in 2003 ou indings o a signi ican inc ease
in ansi ion a es a e he 2003 e o m as epo ed in Tables5 and A.12 a e no
likely o be biased by he change in Minijob ea nings ceiling ha happened simul-
aneously wi h he in oduc ion o he Midijob subsidy. The indings in Table A.14
co obo a e ou inding o a signi ican and la ge inc ease in ansi ions ou o Mini-
jobs a e he 2003 e o m o ou ea men g oup, he non-ma ied.
4.4 Di e ence indi e ences analyses using adminis a i e da a
Compa ed o he su ey da a, ou adminis a i e da a o e la ge samples, mo e p e-
cise in o ma ion, and a e obse ed a mon hly ins ead o annual equency. E en
hough he adminis a i e da a lack in o ma ion on ma i al s a us, we can epli-
ca e he di e ence-in-di e ences analyses using his la ge da abase. In his sec-
ion, we show wo app oaches o app oxima e he su ey-based analyses. In he i s
app oach, we eplace he non-ma ied ea men indica o om Sec .4.3 wi h an
indica o o male gende ; in he su ey da a, he sha e o males among he non-
ma ied (22.1 pe cen ) is six imes as high as he sha e o males among he ma -
ied (3.7 pe cen ). The e o e, his migh be a easonable i s app oxima ion. In he
second app oach, we es ima e a p edic ion equa ion o being non-ma ied wi h he
su ey da a and use i o p edic non-ma ied s a us in he adminis a i e da a. The
Table 5 DID Es ima es—2003 e o m baseline esul s (SOEP). Sou ce: SOEP wa e 35 and own calcula-
ions
No e: All es ima ions use 2,736 pe son-yea obse a ions. Linea eg essions, s anda d e o s clus e ed a
he indi idual le el a e in pa en heses. The ec o o basic con ols accoun s o an indica o o gende , 5
indica o s o age g oup, an indica o o Eas Ge man esidence, and an indica o o non-Ge man ci izen-
ship. The ec o o ex ended con ols accoun s o 5 educa ion indica o s, 7 enu e indica o s, 4 i m size
indica o s, 9 indus y indica o s, and 8 occupa ion indica o s. Column 4 eplaces he pos indica o wi h a
se o calenda yea ixed e ec s. The es ima ions use c oss-sec ional sample weigh s o accoun o non-
esponse and o e sampling. * p < 0.10, ** p < 0.05, ***p < 0.01
Full sample Full sample Full sample Full sample
(1) (2) (3) (4)
Pos 0.004 0.004 −0.000 –
(0.015) (0.015) (0.015)
T ea 0.002 −0.022 −0.038 -0.0041
(0.029) (0.031) (0.032) (0.033)
Pos * ea 0.139*** 0.123** 0.131*** 0.134***
(0.052) (0.051) (0.050) (0.050)
Con ols—basic No Yes Yes Yes
Con ols—ex ended No No Yes Yes
Yea ixed e ec s No No No Yes
1531
Phasing ou pay oll ax subsidies
Hu unen, K., Pi ilä, J., & Uusi alo, R. (2013). The employmen e ec s o low-wage subsidies. Jou nal
o Public Economics, 97, 49–60.
Imme oll, H., Kle en, H. J., K eine , C. T., & Saez, E. (2007). Wel a e e o m in Eu opean Coun ies: A
mic osimula ion analysis. Economic Jou nal, 117, 1–44.
Ja a, H. X., Gasio , K., & Mako ec, M. (2020). Wo k incen i es a he ex ensi e and in ensi e ma gin in
Eu ope: The ole o axes bene i s and popula ion cha ac e is ics. Social Indica o s Resea ch, 152,
705–778.
Kelle , Be nd and Ha mu Sei e (2015). A ypical o ms o employmen in he public sec o - a e he e
any? WSI-Diskussionspapie No. 199, Düsseldo : Wi scha s- und Sozialwissenscha liches Ins i-
u in de Hans-Böckle -S i ung.
Kelle , B., Pusch, T., & Sei e , H. (2021). Midijobs: Die unbekann e Va ian e de a ypischen Beschä i-
gung. WSI Mi eilungen, 14(2), 160–170.
Kle en, H. (2016). Bunching. Annual Re iew o Economics, 8, 435–464.
Ko keamäki, O., & Uusi alo, R. (2009). Employmen and wage e ec s o a pay oll- ax cu - E idence
om a egional expe imen . In e na ional Tax and Public Finance, 16, 753–772.
LaLumia, S. (2009). The ea ned income ax c edi and epo ed sel -employmen income. Na ional Tax
Jou nal, 62(2), 191–217.
Leigh, A. (2007). Ea ned income ax c edi s and labo supply: New e idence om a B i ish na u al
expe imen . Na ional Tax Jou nal, 60(2), 205–224.
Lie zmann, T., Schmelze , P., & Wieme s, J. (2017). Ma ginal employmen o wel a e ecipien s: S ep-
ping s one o obs acle? Labou , 31(4), 394–414.
Meye , B. D. (2002). Labo supply a he ex ensi e and in ensi e ma gins: The EITC, wel a e, and hou s
wo ked. Ame ican Economic Re iew, 92(2), 373–379.
Neuma k, D. & Shi ley, P. (2020). The long- un e ec s o he Ea ned Income Tax C edi on Women’s
Ea nings, NBER Wo king Pape No. 24114, Camb idge, Mass.
Oschmiansky, F ank and Julia Be hold (2020), Minijobs und Midijobs, a ailable a h ps:// www. bpb.
de/ hemen/ a bei / a bei sma k pol i ik/ 317249/ minij obs- und- midij obs/ (las access: Augus 9, 2022).
Saez, E., Schoe e , B., & Seim, D. (2019). Pay oll axes, i m beha io , and en sha ing: E idence om a
young wo ke s’ ax cu in Sweden. Ame ican Economic Re iew, 109(5), 1717–1763.
Sca e, Rachel (2021). Flexibili y o ce ain y? The agg ega e e ec s o casual jobs on labou ma ke s,
Edinbu gh School o Economics Discussion Pape Se ies No. 294, Uni e si y o Edinbu gh.
Sei e , H. (2017). Wie lassen sich En wicklung und S uk u en a ypische Beschä igungs e häl nisse
e klä en? WSI-Mi eilungen, 2017–01, 5–15.
Tazhi dino a, A. (2020). Do only ax incen i es ma e ? Labo supply and demand esponses o an unusu-
ally la ge and salien ax b eak. Jou nal o Public Economics, 184(Ap il), 104162.
Tazhi dino a, A. (2022). Inc easing Hou s Wo ked: Moonligh ing Responses o a La ge Tax Re o m.
Ame ican Economic Jou nal: Economic Policy, 14(1), 473–500.
T ampe, P. (2007). The EITC disincen i e: The e ec s on hou s wo ked om he phase-ou o he ea ned
income ax c edi . Econ Jou nal Wa ch, 4(3), 308–320.
T ampe, P. (2008). The EITC disincen i e: A eply o D Hila y Hoynes. Econ Jou nal Wa ch, 5(1),
59–65.
Publishe ’s No e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in published maps
and ins i u ional a ilia ions.
