Block, Joe n; K i ikos, Alexande S.; P iem, Maximilian; S iel, Ca oline
A icle — Accep ed Manusc ip (Pos p in )
Eme gency-aid o sel -employed in he Co id-19
pandemic: A lash in he pan?
Jou nal o Economic Psychology
P o ided in Coope a ion wi h:
Ge man Ins i u e o Economic Resea ch (DIW Be lin)
Sugges ed Ci a ion: Block, Joe n; K i ikos, Alexande S.; P iem, Maximilian; S iel, Ca oline (2022) :
Eme gency-aid o sel -employed in he Co id-19 pandemic: A lash in he pan?, Jou nal o Economic
Psychology, ISSN 1872-7719, Else ie , Ams e dam, Vol. 93, pp. 1-41,
h ps://doi.o g/10.1016/j.joep.2022.102567
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/307397
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-nc-nd/4.0/
Eme gency-Aid o Sel -employed in he Co id-19 Pandemic:
A Flash in he Pan?*
Joe n Block† Alexande S. K i ikos ‡ Maximilian P iem§ Ca oline S iel**
T ie Uni e si y DIW-Be lin DIW-Econ DIW-Be lin
Ok obe 2022
Abs ac :
The sel -employed aced s ong income losses du ing he Co id-19 pandemic. Many
go e nmen s in oduced p og ams o inancially suppo he sel -employed du ing he
pandemic, including Ge many. The Ge man Minis y o Economic A ai s announced a
€50bn eme gency-aid p og am in Ma ch 2020, o e ing one-o lump-sum paymen s o
up o €15,000 o hose acing subs an ial e enue declines. By eassu ing he sel -
employed ha he go e nmen ‘would no le hem down’ du ing he c isis, he p og am
had also he impo an aim o mo i a ing he sel -employed o ge h ough he c isis. We
in es iga e whe he he p og am a ec ed he con idence o he sel -employed o su i e
he c isis using eal- ime online-su ey da a comp ising mo e han 20,000 obse a ions.
We employ p opensi y sco e ma ching, making use o a ich se o a iables ha in luence
he subjec i e su i al p obabili y as main ou come measu e. We obse e ha his
p og am had signi ican e ec s, wi h he subjec i e su i al p obabili y o he sel -
employed being mode a ely inc eased. We e eal impo an e ec he e ogenei ies wi h
espec o educa ion, indus ies, and speed o paymen . No ably, posi i e e ec s only
occu among hose sel -employed whose applica ion was p ocessed quickly. This
sugges s s ess-induced wai ing cos s due o he unce ain y associa ed wi h he
adminis a i e p ocessing and he o e all pandemic si ua ion. Ou indings ha e policy
implica ions o he design o suppo p og ams, while also con ibu ing o he li e a u e
on he ins umen s and e ec s o en ep eneu ship policy in e en ions in c isis si ua ions.
JEL Classi ica ion: C21, H43, L25, L26, J68
PsycINFO Classi ica ion: 2960, 3650
Keywo ds: sel -employmen ; eme gency-aid; ea men e ec s; Co id-19,
en ep eneu ship policy; subjec i e su i al p obabili y
Decla a ions o in e es s: None.
* We g a e ully acknowledge he suppo o Thea Zöllne and he VGSD who p o ided us wi h he da a.
Mo eo e , we hank h ee anonymous e e ees, Ma co Caliendo, Daniel G aebe , S e en Kuenn, and
Johannes Seebaue o aluable commen s. The da a used in his a icle is a ailable: Be schek, I ene;
Block, Joe n; K i ikos, Alexande S.; Lu z, And eas (2022): Su ey among sel -employed pe sons in
Ge many du ing he COVID-19 pandemic. Wa e 2020. [Da ase ]. Zenodo
h ps://doi.o g/10.5281/zenodo.7091989. Supplemen a y ma e ial is a ailable a h ps://po al. dz-
bo.diw.de/s udy/S0021 and in Be schek e al. (2022).
† Joe n Block, T ie -Uni e si y, Ge many; E asmus-Uni e si y Ro e dam, The Ne he lands; and
Wi ene Ins i u ü Familienun e nehmen (WIFU), Uni e si y Wi en/He decke, Ge many.
‡ Alexande K i ikos, (co esponding au ho ), Ge man Ins i u e o Economic Resea ch (DIW-Be lin),
Po sdam Uni e si y, IZA, Bonn and IAB, Nu embe g. Add ess: DIW Be lin, Moh ens . 58, 10117
Be lin, Ge many. e-mail: ak i [email protected]
§ Maximilian P iem, DIW-Econ, Be lin, Ge many.
** Ca oline S iel, DIW-Be lin, Be lin, Ge many.
This is he pos p in o an a icle published in Jou nal o Economic Psychology 93 (2022), 102567, 16 S.
a ailable online a : doi h ps://doi.o g/10.1016/j.joep.2022.102567
© <2024>. This manusc ip e sion is made a ailable unde he CC-BY-NC-ND 4.0 license
h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
1
1. In oduc ion
The Co id-19 pandemic led many coun ies in sp ing 2020 o empo a ily close majo pa s
o hei economies, especially in he se ice and ade indus ies. Sel -employed and mic o-
businesses ( e e ed o as “sel -employed” om now on) a e majo economic ac o s in
hese indus ies. Resea ch shows ha he sel -employed su e ed inancially mo e s ongly
om he dis up ion caused by Co id-19 han o he pa s o he wo king popula ion (Fai lie
and Fossen, 2022b, G aebe e al., 2021). In Ge many, o ins ance, abou 60% o he 4
million sel -employed aced sales and income losses, while only abou 15% o dependen ly
employed indi iduals we e con on ed wi h job o wage losses (see K i ikos e al., 2020).
Howe e , he c isis a ec ed he sel -employed, no jus economically bu also om
a psychological and men al heal h pe spec i e. Fi s e idence (see To ès e al., 2022)
poin s o a wo sening o he men al heal h condi ions among he sel -employed, due in pa
o hei inancial losses (Caliendo e al., 2022b). This nega i ely a ec s hei decision-
making p ocesses, showing ha he economic and psychological condi ions o he sel -
employed a e closely in e connec ed (Wiklund e al., 2019).
Gi en he impo ance o he sel -employed o he Ge man economy and gi en he
need o s eng hen hei con idence in o hei own abili ies o keep hei businesses up and
unning, hei si ua ion was o high conce n o policy make s. Hence, a he end o Ma ch
2020, Ge many in oduced an eme gency-aid p og am (“So o hil e”) o €50 billion
designed o inancially suppo hose sel -employed acing s ong e enue losses due o he
imposed es ic ions. The p og am was a one-o lump-sum g an o up o €15,000 pe sel -
employed and was accessible be ween he end o Ma ch and end o May 2020. The p og am
had no only a inancial aim bu also he aim o mo i a ing he sel -employed o ge h ough
he c isis. Wi h ega d o he la e , he Ge man Minis e o Economic A ai s s a ed a a
p ess con e ence on Ma ch 10, 2020, “ ha we will no le any i m-owne down and ha
no i m should be o ced o lea e he ma ke because o he Co ona pandemic” (DPA 2020).
Thus, So o hil e sough o quell exis en ial ea om inancial ha dship, mo i a e he sel -
employed and p e en massi e exi s om sel -employmen .
In his s udy, we ocus on he aim o he p og am o inc ease he con idence o he
sel -employed in he c isis and in es iga e how he p og am a ec ed he belie s o he sel -
employed ha hei business would su i e he c isis. P e ious esea ch shows ha
subjec i e belie s abou i m su i al and ailu e a e c ucial o hei con inua ion (Khelil,
2016), as he sel -employed will s op in es ing in hei i ms once hey s op belie ing in
2
hei business su i al (Ucbasa an e al., 2010). Thus, i is impo an o in es iga e whe he
policy measu es like So o hil e achie ed hei aim o inc easing subjec i e belie s in he
su i al o hei own business. Secondly, o unde s and how he p og am design a ec s,
and unde wha condi ions he use o such ins umen s inc eases, he con idence o he sel -
employed o su i e a c isis, we also causally examine whe he he speed o paymen
ma e s. Thi d, gi en ha esea ch shows ha educa ion and isk ole ance (see e.g. Van
de Sluis e al., 2008, Caliendo e al., 2010, 2104, 2022a) a e wo impo an pe sonal
cha ac e is ics s ongly a ec ing business de elopmen , we in es iga e how hese ac o s
in luence he impac o So o hil e. These esea ch ques ions a e highly ele an gi en he
huge amoun o axpaye money – €50bn – made a ailable o his p og am. Typical sel -
employmen policy measu es, like Ge many’s a ious s a -up subsidy p og ams, ecei e
yea ly budge s o less han €1bn (see e.g. Caliendo and Kuenn, 2011), cla i ying ha he
amoun made a ailable o his p og am was excep ionally high.
Fo ou analysis, we ely on a su ey answe ed by mo e han 20,000 sel -employed
indi iduals in Ap il and May 2020. Besides in o ma ion on c isis ela ed sales losses,
liquidi y cons ain s, and he willingness o apply o inancial suppo om he eme gency-
aid, he su ey collec ed in o ma ion on mos indi idual- and i m- ela ed cha ac e is ics
ele an o sel -employmen . As ou ou come a iable, we use a measu e ha is based on
he indi idual assessmen abou he p obabili y o “end hei sel -employmen ac i i ies due
o he Co ona-c isis in he nex 12 mon hs.” Resea ch has es ablished ha subjec i e
p obabili y measu es a e an app op ia e way o measu e expec a ions (Manski, 2004),
showing ha hese measu es e lec en ep eneu ial decision-making, hus impac ing i m
su i al (Cassa , 2010, Hyy inen e al., 2014). Mo eo e , beyond ich in o ma ion on he
sel -employed, we make use o he ac ha he da a is su eyed in eal- ime. To causally
analyze whe he he inancial suppo ins umen inc eased he subjec i e su i al
p obabili y, we ely on he condi ional independence assump ion (CIA). The eby, we
compa e sel -employed who al eady ecei ed suppo om he p og am ( he ea men
g oup) wi h hose who planned o apply o he p og am ( he con ol g oup), con olling
o a ich se o a iables ha in luence he applica ion and su i al p obabili y.
We con ibu e o he li e a u e analyzing how he Co id-19 pandemic a ec ed he
sel -employed (Adams-P assl e al., 2020, Block e al., 2022, G aebe e al., 2021) in h ee
ways. This c isis is unique and no exis ing esea ch shows how public policy in e en ions
help he sel -employed deal wi h he psychological consequences o a uly exogenous
c isis like Co id-19. Ou s udy p o ides i s empi ical e idence on he subjec i ely
3
pe cei ed e ec i eness o an eme gency-aid p og am du ing he pandemic and, mo e
b oadly, on he non-mone a y e ec s om policy in e en ions du ing economic c ises,
hus con ibu ing o he li e a u e on he non-mone a y and mo i a ional e ec s o public
policy (S u ze , 2020). Secondly, we in es iga e he impac o a ia ions in he speed o
p ocessing he applica ions and paying ou he eme gency-aid, aking a p ocedu al u ili y
and adminis a i e bu den pe spec i e on public in e en ions (F ey e al., 2004, Block and
Koellinge , 2009). The esul s o ou s udy imply ha p og am impac s du ing a c isis do
no jus depend on i s con en bu also on i s p ocessing speed associa ed, e lec ing s ess-
induced wai ing cos s esul ing om unce ain y (Baekgaa d e al. 2021; G eco and Roge ,
2003; Mona , A e ill, and Laza us, 2003). Thi dly, we analyze e ec he e ogenei ies wi h
espec o a ious indi idual-le el a iables, such as isk- ole ance o educa ional
a ainmen . In ha sense, ou analysis is o high ele ance gi en he ongoing deba e on he
igh design and implemen a ion o such policy ins umen s, in o ming go e nmen s abou
how speci ic a ge g oups pe cei e he public inancial suppo unde he gi en condi ions.
Wi h ou esul s, we con ibu e mo e gene ally o he li e a u e on SME policy in imes o
c ises (Minni i, 2008, Beli ski e al., 2022).
2. Co id-19 and Sel -employmen
2.1. Co id-19 in Ge many and Policy Response
A he ime o da a collec ion in Ap il and May 2020, Ge many was among he coun ies
mos a ec ed by he Co id-19 pandemic. The Ge man go e nmen ied o s op he
sp eading o he i us by implemen ing se e al measu es ha se e ely a ec ed he
economy. Schools, dayca e cen e s, shops, es au an s, and ho els we e closed, excep o
supe ma ke s. A cu ew was imposed, including a ban on public ga he ings wi h mo e han
wo people. E en s, including ade ai s, spo s, and conce s, we e cancelled; a el was
es ic ed. Du ing ha ime, a GDP decline o 9% was p edic ed o 2020 (I W 2020).
To help he economy while a oiding job cu s and a long-las ing ecession, he
Ge man go e nmen in oduced se e al suppo p og ams o mi iga e he consequences o
he pandemic. Ta ge ing es ablished i ms, employe s could send hei employees in o
Ku za bei (sho - ime wo k), whe e he Fede al Employmen O ice co e s a subs an ial
po ion o he wage cos s. Howe e , he sel -employed a e no co e ed by his ins umen .
To add ess his occupa ional g oup, he go e nmen launched he So o hil e eme gency-
aid o €50 billion, accessible om Ma ch 25, 2020, h ough he end o May 2020, o which
€13.7 billion we e ac ually spen . The sel -employed could ecei e immedia e inancial
4
assis ance o up o €9,000 o businesses wi h up o i e employees, and up o €15,000 o
businesses up o en employees - i hey had acu e liquidi y sho alls (Fede al Minis y o
Economic A ai s and Ene gy, 2020). Howe e , suppo p og am unds could only be used
o co e ope a ing cos s; p i a e li ing cos s we e excluded.
2.2. P io Resea ch on Sel -employmen in he Co id-19 Pandemic
The e ec s o Co id-19 on sel -employmen a ac ed empi ical esea ch documen ing ha ,
du ing he c isis, sel -employed in o he coun ies su e ed like hose in Ge many (see
Adams-P assl e al., 2020, G aebe e al. 2021, Beli ski e al., 2022, Kalenkoski and
Pabilonia, 2022), cla i ying ha he pandemic dis up ed sel -employmen globally.
Mo eo e , esea ch poin s o e ec s on sel -employed beyond economic losses:
Desc ip i e (To ès e al. 2022) and causal (Caliendo e al. 2022b) e idence e eals a
wo sening o men al heal h among he sel -employed in he wake o pandemic-d i en
ma ke dis o ions.
Mo eo e , beyond desc ibing how he pandemic a ec ed he sel -employed, esea ch
in es iga es how he sel -employed coped wi h he ea ly s ages o he pandemic and how
go e nmen p og ams esponding o his economic dis up ion a ec ed he sel -employed.
Block e al. (2022) in es iga es how he sel -employed managed he consequences o
Co id-19 by main aining hei liquidi y h ough he use o boo s ap- inancing. Meu e e
al. (2022) demons a e how en ep eneu ial online communi ies o e ed suppo o a ec ed
en ep eneu s. Be schek and E dsiek (2020) show ha sel -employed wi h a highe deg ee
o digi aliza ion we e less a ec ed by he c isis. Wi h espec o go e nmen p og ams,
a ious public policy ins umen s esponding o he economic dis up ion and add essing
he inancing needs o he sel -employed a e iden i ied. Fai lie and Fossen (2022a) p o ide
a disbu semen analysis o he Paycheck P o ec ion P og am and he Economic Inju y
Disas e Loan P og am, bo h in he US, ha aimed o help disad an aged g oups. Fo
China, Liu e al. (2022) show he suppo i e ole o Chinese s a e-owned banks o small
businesses’ lines o c edi , whe e he b oad policy mix comp ised loan gua an ees, di ec
lending o SMEs, g an s, and equi y ins umen s. Belghi a e al. (2022) in es iga e he
e ec s o UK go e nmen al policies o SMEs du ing Co id-19 and examined hei e ec
on he abili y o su i e he pandemic.
5
3. Impac o he Aid P og am on he Subjec i e Su i al P obabili y
3.1. Baseline E ec
Sel -employed ha e expec a ions abou he inancial and non inancial goals o hei
business ac i i ies and base hei in es men decisions on hese expec a ions (Gimeno e
al., 1997). In he con ex o a c isis like he Co id-19 pandemic, hei subjec i e e alua ion
abou he ex en hey will be able o achie e hei own aims (possibly se be o e he c isis)
by u he unning hei businesses is c ucial o he decision be ween con inuing hei
business o closing i (Hyy inen e al., 2014). The assessmen o hese expec a ions abou
u u e p ospec s in luences he e o hey pu in o he en u e, a ec s hei in es men
decisions, and, ul ima ely, he decision o e i m su i al (Ucsbasa an e al., 2013). I
indi iduals belie e ha hey will be able o a ain hei goals, hey will in es in hei
businesses, hus emaining in he ma ke (Koellinge e al., 2007; Ayala and Manzano,
2014; Li e al., 2021). I indi iduals expec ha hey will no longe be able o ealize hei
goals, hey will s op in es ing in hei i ms, leading o i m closu e (Ucsbasa an e al.,
2010; Ayala and Manzano, 2014). Khelil (2016, p. 76) de ines ailu e among sel -employed
as a condi ion when he sel -employed en e “in o a spi al o a psychological s a e o
disappoin men ” and a gues ha “in he absence o economic o psychological suppo ,
en ep eneu s a e o ced o exi om hei en ep eneu ial ac i i ies.” The sel -employed
migh e en close hei business despi e an excellen inancial si ua ion i hey hold nega i e
subjec i e belie s abou he u u e. The e o e, how he sel -employed pe cei e hei u u e
p ospec s is pa icula ly impo an in he con ex o an economic c isis.
Du ing he pandemic, he sel -employed we e among he mos a ec ed occupa ional
g oups, especially hose in he ho el and es au an business, ou ism indus y, e ail,
cul u al, and e en s sec o as well as all indus ies equi ing pe sonal con ac . Fo hem, he
policy measu es o con ain he pandemic mean a de ac o empo a y inabili y o wo k,
whe e hey could no gene a e e enues o co e hei ope a ing expenses and li ing cos s.
Such condi ions o inancial ha dship may ha e nega i e second o de e ec s. Financial
sca ci y can be linked wi h beha io o inancial a oidance and wi h changing assessmen s
o u u e gains in he o m o an inc ease in discoun ing o u u e gains and losses (Hilbe
e al., 2022a, 2022b). In case he sel -employed decides o mo e away om his
occupa ional o m, i migh also impac he e ec i eness o hei job sea ch (Ge a ds and
Wel e s, 2022), as subsequen inancial ha dship may limi hei cogni i e esou ces,
he eby p e en ing hem om making delibe a e decisions.
6
Fu he , he sel -employed also con on ed a loss o p ocedu al u ili y o he wise
de i ed om sel -employmen (F ey e al., 2004). The sel -employed in hese a ec ed
indus ies we e collec i ely sen in o a “psychological s a e o disappoin men ” wi h
nega i e e ec s on hei subjec i e belie s abou business su i al. This was ue a he
beginning o he pandemic, when i was un o eseeable o how long he pandemic and i s
con ainmen measu es would las . The “s a e o disappoin men ” in combina ion wi h he
expe ience o inancial ha dship is likely o nega i ely in luence he assessmen o hei
business u u e.
Deemed a a high isk o business closu e, he eme gency-aid aimed o p o ide
economic suppo agains insol ency co e ing he ixed business cos s ha con inued o
acc ue despi e no o low e enues. Fu he , gi en he s a emen o he Minis e o
Economic A ai s ha a ac ed a lo o public a en ion and enjoyed a b oad ecep ion
among he sel -employed, he eme gency-aid p o ided mo i a ional suppo encou aging
he sel -employed o emain in business. Public discussion on he eme gency-aid was
guided by one ques ion: did he p og am a ec he subjec i e belie o he sel -employed
o no being “abandoned”? In ha sense, he eme gency-aid p og am aimed a
coun e ac ing nega i e assessmen s o he sel -employed abou he u u e o hei
businesses and a imp o ing hei expec a ions abou he subjec i e su i al p obabili y o
hei businesses by easing po en ial inancial ha dships. We hypo hesize:
H1: Recei ing inancial suppo om he eme gency-aid posi i ely a ec ed he
subjec i e belie o he sel -employed ha hei i ms will su i e he pandemic.
3.2. Mode a ing Fac o s
3.2.1 Se e i y o he C isis by Indus y
As a i s mode a ing ac o , we di e en ia e be ween indus ies acco ding o he deg ee
he c isis a ec ed hem. The eason is ha a p og am ollowing a "wa e ing-can p inciple"
ha does no conside indi idual needs, o en has only limi ed e ec s (Wunsch and
Lechne , 2008; G asho , 2021). We posi ha he e ec o he eme gency-aid p og am
depends on how se e e he c isis hi he espec i e sel -employed and how se e e he
indi idual need was (Caliendo and Kuenn, 2011). Sel -employed who we e only weakly
hi by he c isis and ecei ed he inancial suppo a e no expec ed o ha e a highe
subjec i e su i al p obabili y compa ed o indi iduals who we e weakly hi by he c isis
bu did no ecei e he suppo . In his ange, we expec deadweigh losses among hose
who ob ained he inancial suppo . In con as , among sel -employed who we e s ongly
7
hi by he c isis and ecei ed he inancial suppo , we expec ha hey will assess he
su i al p obabili y o hei businesses highe han indi iduals who we e s ongly hi by
he c isis bu did no ye ecei e suppo . We hypo hesize:
H2a: The posi i e e ec o he eme gency-aid on he subjec i e belie o he sel -
employed ha hei i ms will su i e he pandemic is s onge o hose sel -employed in
s ongly e sus weakly a ec ed indus ies.
3.2.2 Le el o Educa ion
Resea ch shows ha educa ion le els inc ease he business pe o mance o he sel -
employed and i m su i al (e.g. Pa ke and an P aag, 2006; Van de Sluis e al., 2008).
I co ela es wi h an indi idual’s cogni i e abili ies o iden i y and exploi en ep eneu ial
oppo uni ies (Ha og e al., 2010) and wi h an indi idual’s adap abili y o changing
en i onmen s (S asielowicz, 2020). Resea ch inds ha indi iduals wi h highe cogni i e
abili ies achie e be e inancial ou comes (Tang, 2021) and ha e lowe unemploymen
isks (Vélez-Co o e al., 2021). Fo his s udy, we posi ha s ong cogni i e abili ies a e
needed o success ully mas e c isis si ua ions like he pandemic. The eme gency aid may
help o co e he unning cos s o he business on sho - e m basis bu o cope wi h he
long- e m impac s o he c isis, he a ec ed sel -employed mus ha e he capaci y and
willingness o adap hei p oduc s, se ices, and business model. Such changes equi e
s ong cogni i e abili ies. Seeing educa ion le el as a p oxy o cogni i e abili y (Be y,
G uys, and Sacke , 2006), we a gue ha he be e educa ed sel -employed a e able o eac
mo e lexibly o exogenous shocks associa ed wi h high unce ain y, pu ing hem in o a
be e posi ion o bene i om he eme gency-aid. In ha sense we use educa ion le el as a
p oxy o cogni i e abili ies. We hypo hesize:
H2b: The posi i e e ec o he eme gency-aid on he subjec i e belie o he sel -
employed ha hei i ms will su i e he pandemic is s onge o hose sel -employed
wi h a high e sus low le el o educa ion.
3.2.3 Risk Tole ance
The subjec i e belie o whe he one’s own business will su i e a c isis also depends on
one’s isk ole ance. Resea ch shows ha isk ole ance is no only one o he mos
impo an pe sonali y cha ac e is ics a ec ing decision making, beha io , and su i al o
he sel -employed (B ands ä e , 1997; Hansema k, 2003; B own e al., 2011; Caliendo e
al. 2009, 2010, 2012, 2022a; Caliendo and K i ikos, 2012; U big e al., 2012; Willeb ands
14
The ou e expec a ion EX[ . | D = .] con eys ha indi iduals in he compa ison g oup
a e ma ched o ea ed uni s such ha he mean dis ibu ion o he co a ia es in he ma ched
con ol g oup esembles ha o he ea men g oup o he calcula ion o he ATT and ice
e sa o he ATU (Caliendo and Kopeining, 2008). Fu he mo e, we assume o e lap wi h
0 < P (D = 1|X) < 1 o all X, meaning ha indi iduals wi h he same alues o X ha e a
posi i e p obabili y o being ea ed and un ea ed, i.e., he e is no de e minism in ea men
assignmen based on he co a ia es. We apply p opensi y sco e ma ching o educe he
dimensionali y o he co a ia es o a single balancing sco e, P(X), based on which
indi iduals om he con ol g oup a e ma ched o he ea men g oup o he ATT and ice
e sa o he ATU.
5.2. Es ima ion P ocedu e
5.2.1 Ou come Va iable
The aim o he eme gency-aid p og am was o a oid i m closu es by he sel -employed
whose economic su i al was h ea ened by he Co id-19 pandemic. Beyond he inancial
suppo , an impo an aspec was ha he aid p og am in ended o eassu e he sel -
employed ha he go e nmen ‘would no le hem down’ and ha hey could main ain
hei en u e despi e he c isis. Thus, he ques ion is whe he he p og am achie ed his
goal by inc easing hei belie in being able o success ully na iga e he businesses h ough
he c isis. The psychological aspec is pa icula ly impo an in he con ex o he sel -
employed. Mo eo e , he analysis o he subjec i e su i al p obabili y using ma ching
echniques helps o iden i y he pe cei ed u ili y o he p og am by he sel -employed,
wi hou unning in o he p oblem o in en ional mis epo ing.
The e o e, we examine changes in he subjec i e su i al p obabili y o sel -
employed indi iduals in sp ing 2020. Responden s a e asked o assess he likelihood o
qui ing sel -employmen wi hin he coming yea due o he pandemic. We use his
in o ma ion o cons uc ou ou come a iable, cap u ing he subjec i e su i al p obabili y
o he esponden s’ en u es anging om 1 (“ e y unlikely”) o 5 (“ e y likely”).
Appendix Figu e A4 shows he dis ibu ion o he a iable in ou sample, dis inguishing
be ween applican s and non-applican s. Fo ou ea men analysis, we educe i o a bina y
a iable wi h ca ego ies 5 (“ e y likely”) and 4 (“ a he likely”) equaling one; he
emaining ca ego ies equal ze o: 3 (“neu al”), 2 (“ a he unlikely”), and 1 (“ e y
unlikely”). The bina y a iable allows o an in ui i e in e p e a ion o he esul s, since he
ATT coe icien s can be di ec ly in e p e ed as changes in su i al p obabili y. To check
15
he sensi i i y o ou esul s is-à- is he educed explana o y a iable, we conduc
obus ness checks using he o iginal o dinal a iable as dependen a iable; esul s a e e y
simila (Sec ion A.4.2 in he Appendix).
5.2.2 T ea men Va iable
We asked esponden s o indica e whe he hey had applied, o planned o apply, o he
eme gency-aid p og am. Possible answe s a e 1 (“yes, I applied”), 2 (“I am planning o
apply”), 3 (“I am no su e ye ”), and 4 (“I will no apply”). We combine his ques ion wi h
in o ma ion on hei applica ion’s s a us anging om 1 (“app o ed”), o e 2 (“declined”)
o 3 (“I am wai ing o a decision”). We also ha e in o ma ion on he paymen s a us o
hose indi iduals wi h app o ed applica ion, ob aining in o ma ion on he esponden s’
applica ion s a us o he eme gency-aid p og am, illus a ed in Table 3.
Table 3: De ini ion o he ea men and con ol g oup
Su ey
Ques ion
Q30: Did you apply o he
eme gency assis ance
(g an ) om he ede al o
s a e go e nmen ?
Q33: Wha is he s a us o
you applica ion?
Q35: Has he aid al eady
been paid ou …?
Answe ing
Op ions
Yes, I applied
Accep ed
Yes …
No …
Declined
I am wai ing o a decision
I am planning o do so
I am no su e ye
No, I won’
T ea men G oup
Con ol G oup
No e: Table 3 p o ides in o ma ion on he de ini ion o he applied ea men and con ol g oups in he main
ma ching model o his pape . The da k g ey shaded panel indica es he de ini ion o he ea men g oup.
The ligh g ey shaded panel indica es he de ini ion o he con ol g oup.
We a e in e es ed in he subjec i e su i al p obabili y o indi iduals ecei ing he
eme gency-aid. Responden s alling in his ca ego y a e de ined as he ea men g oup
(da k g ey shaded panel in column (3) o Table 3; N=5,743). Indi iduals ‘who did no
apply’ a e no sui able o he con ol g oup as hei easons o no applying a e qui e
16
di e se and hey p obably di e om he ea men g oup along se e al (unobse ed)
dimensions (Table A2 in he Appendix). Ins ead, we ollow Sianesi (2004), and
F ed iksson and Johansson (2008) in using esponden s who a e planning o apply o he
con ol g oup (ligh g ey shaded panel in columns (2) o Table 3). The ad an age is ha
esponden s who a e inclined o apply, sha e impo an cha ac e is ics wi h hose who ha e
al eady applied ega ding hei inancial si ua ion and hei i m’s cha ac e is ics, e c.,
compa ed o indi iduals who did no apply o do no in end o apply.
Responden s who a e planning o apply migh s ill di e om he ea men g oup
in ha hei need o suppo is less u gen . One explana ion could be ha hey we e ei he
(a) inancially less a ec ed by he c isis o (b) had al e na i e sou ces o inance, e.g., own
inancial ese es o suppo h ough al e na i e go e nmen p og ams. Fu he mo e, he e
migh be o he endogenei y issues be ween applican s and hose who we e only planning
o apply in e ms o op imism abou he u u e wi h espec o how quickly he c isis would
end. Among o he a iables in luencing selec ion in o ea men (see Sec ion 5.2.3), we
add ess hese issues in he p opensi y sco e ma ching algo i hm by con olling o e enue
decline, o es ima ed ime o insol ency a e accoun ing o inancial ese es, o
ans e s om he basic income scheme, and, wi h espec o di e ences in op imism, by
con olling o he expec ed du a ion o inancial ha dship due o he c isis as exp essed by
he indi iduals.
2
While hese ac o s should be he main easons o pos poning
applica ions, we canno ule ou ha o he , unobse able ac o s a e p esen . I so, we
would unde es ima e he eme gency p og am’s ea men e ec , i.e. i he con ol g oup’s
decision o pos pone applica ions is associa ed wi h highe su i al p obabili ies han he
coun e ac ual su i al p obabili ies o he ea men g oup.
We u he exclude esponden s whose applica ion was success ul bu o whom he
aid was no paid ou when hey we e su eyed (Table 3, ques ion 35). These indi iduals
canno be easily classi ied as in he ea men o con ol g oup. Knowing how much
inancial suppo hey will ecei e, hey migh be close o he ea men g oup as hey
an icipa e he lump-sum paymen . Howe e , ha ing no ye ecei ed he inancial suppo ,
hey migh be mo e conse a i e in hei expec a ions because i emains unclea whe he
2
O he go e nmen p og ams (e.g. BAFA-subsidy, K W-loans) we e no designed o he sel -employed
bu a ge ed la ge i ms, which explains he e y low esponse a e. The e o e, in he majo i y o he
cases, inancial suppo om al e na i e go e nmen p og ams beyond he eme gency-aid p og am and
he basic-income scheme does no explain he con ol g oup’s decision o pos pone he applica ion. We
con ol o ans e s om he basic income scheme in ou es ima ion. In addi ion, as a obus ness check,
we es ima e an al e na i e model con olling o u he go e nmen suppo p og ams in he p opensi y
sco e ma ching. The esul s a e la gely he same and a ailable upon eques om he au ho s.
17
hey will ecei e he paymen and because he exac da e o he payou is s ill unce ain;
meaning, hey mus b idge he ime inancially. I his e ec domina es, including hem in
he ea men g oup would nega i ely bias he a e age ou come o he ea men g oup.
Fu he mo e, we decided agains using indi iduals wai ing o a decision (ques ion 33 in
Table 3) as a con ol g oup, since he a e age ime ha has elapsed since hei applica ion
(15 days, see Sec ion 4.4, Table 2) exceeds he a e age p ocessing ime (7.5 days), hus
sugges ing hei applica ions somehow di e om a e age (e.g., hei cases a e
complica ed). He e i is unclea how he unce ain y abou he app o al da e a ec s hei
expec a ions abou hei u u e p ospec s.
As we a e in e es ed in he ea men e ec o all sel -employed a ge ed by he
eme gency und, we es ima e bo h he ATT and ATE. I is easonable o belie e ha he
majo i y o he sel -employed who planned o apply a e also eligible o he eme gency-
aid (Table 2 shows a ejec ion a e o 2.4%; 241 o 9,885). I can be assumed ha many o
hem would ha e joined he p og am (Sianesi, 2004).
5.2.3 P opensi y Sco e Ma ching
We apply p opensi y sco e ma ching o ma ch ea ed and un ea ed indi iduals based on a
se o co a ia es ha a e likely o a ec he applica ion o he eme gency-aid and he
esponden s’ expec a ions abou hei i ms’ p ospec s.
Fi s , we con ol o pe sonal cha ac e is ics, including well-known a iables
in luencing en ep eneu ial decision and su i al, like he esponden ’s age (Kau onen e
al., 2014) and gende (Ve heul e al., 2012). Simila ly, we con ol o he esponden s’ sel -
employmen expe ience by accoun ing o he numbe o yea s spen in sel -employmen
(Pa ke , 2018). As discussed in Sec ion 3.2, a ious s udies show ha en ep eneu s’
educa ion and isk ole ance le els in luence hei business pe o mance and su i al. We
include a bina y a iable on educa ion and measu e he sel -employed esponden s’ isk
ole ance on a scale om 1 (low isk- ole ance) o 5 (high isk- ole ance), whe e p e ious
esea ch emphasizes ha indi iduals wi h high isk ole ance a e, a he maximum, isk
neu al (see Dohmen e al., 2011).
We u he con ol o se e al business- ela ed cha ac e is ics ha likely in luence
selec ion in o ea men and he ou come a iable. We include in o ma ion as o whe he
he sel -employed wo k ull- ime o pa - ime in hei i ms and whe he hey ha e
employees. P io esea ch documen s di e en su i al p obabili ies o hese g oups in
compa ison o o he sel -employed pe sons (de V ies e al., 2019). Mo eo e , we expec
18
ull- ime sel -employed (in con as o pa ime sel -employed) and solo sel -employed (in
con as o sel -employed wi h employees) o be mo e ulne able o e enue dec eases
du ing he Co id-19 pandemic and, he e o e, mo e likely o apply o eme gency-aid. We
u he conside he i m’s deg ee o digi aliza ion by ha ing asked he esponden s o
indica e hei en u es’ le el o digi aliza ion be o e he pandemic s a ed on a 5-poin
Like scale. We expec ha mo e digi alized i ms adap hei se ice p o ision o he
equi emen s o he con ainmen measu es mo e easily (Be schek and E dsiek 2020). We
also accoun o imbalances in he indus y s uc u e be ween ea men and con ol g oups
by including a se o indus y ixed e ec s ha indica e he main indus y o he
esponden ’s i m as he impac o he Co id-19 c isis di e s ac oss indus ies.
P io esea ch shows ha he inancial si ua ion, like weal h, li ing cos s, and
household income, is an impo an de e minan o en ep eneu ial beha io and success
(Hu s and Lusa di, 2004; Pa ke and Van P aag, 2006). The e o e, we con ol o he
esponden s’ mon hly p i a e cos o li ing. Second, we measu e whe he hey ecei ed
inancial suppo om he basic-income scheme o accoun o o he sou ces o income ha
migh in luence bo h he likelihood o apply o he p og am and he su i al p obabili y.
Thi d, we use in o ma ion on how hei i ms we e a ec ed by he c isis, as mo e s ongly
a ec ed indi iduals migh be mo e p one o apply o inancial suppo ( hus, in luencing
hei p obabili y o ea men ). No ably, we asked esponden s o indica e how many
mon hs hei en u es would be able o main ain sol ency gi en hei cu en e enue and
cos si ua ions, and accoun o epo ed e enue dec eases due o he Co id-19 pandemic.
We include wo a iables a ec ing he ou come a iable, i.e. how he sel -employed
assess hei u u e p ospec s. Responden s we e asked abou hei expec a ions ega ding
he du a ion o he pandemic and o he inancial ha dship i will cause. Thus, we ensu e
ha ma ched indi iduals om he ea men and con ol g oups ha e simila expec a ions
abou he u u e and ha di e ences in he subjec i e su i al p obabili y a e no caused
by di e en pe cep ions abou he c isis endu ance. Fu he mo e, we con ol o he
calenda week ha each indi idual was su eyed, since assessmen s o u u e p ospec s
migh depend on p og ession o he c isis and ela ed con ainmen measu es.
3
Finally, he measu es aken by he go e nmen in eac ion o he Co id-19 c isis also
di e ed ac oss he 16 Ge man ede al s a es. To cap u e hese di e ences and o he
3
A he beginning o May 2020, which coincides he end o he su ey (see Figu e 2), he Ge man
go e nmen announced ha i would elax some o he con ainmen measu es by he mid o May; o
ins ance, es au an s would be allowed o eopen and cul u al e en s could ake place in he open ai .
19
egional di e ences in socio-economic s uc u e and i s impac on sel -employmen ac oss
Ge many, we include egion ixed e ec s o he ede al s a e whe e he esponden s’ i m
is loca ed. Table A1 in he Appendix summa izes he co a ia es and compa es hei ealized
alue dis ibu ion be ween he unma ched sample e sus he ea men and con ol g oups
wi hin he ma ched sample.
To ensu e o e lap, we im he ma ching sample o obse a ions wi hin he egion o
common suppo using he max(min{P(X)|D=1, P(X|D=0)} and min(max{P(X)|D=1,
P(X|D=0)} condi ion a he ails o p opensi y sco e dis ibu ion (see Sec ion A.3 in he
Appendix). We use an Epanechniko ke nel o cons uc a weigh ed a e age o he con ol
uni s o he calcula ion o he coun e ac ual ou come, wi h ke nel bandwid h chosen by
c oss- alida ion. The ad an age o he ke nel ma ching es ima o o e o he echniques is
ha we use in o ma ion om a ange o con ol uni s ins ead o elying on a small se o
ma ching pa ne s in he close neighbo hood o he ea ed uni . This is ele an in ou case
as he con ol g oup is smalle han he ea men g oup, hus, equi ing high eplacemen
a es o neighbo hood ma ching, po en ially causing ine icien ATT es ima es (Caliendo
and Kopeining, 2008). As a obus ness check, we e-es ima e ou main esul s wi h
di e en ma ching es ima o s (Sec ion A.4.1 in he Appendix). We boo s ap s anda d
e o s o he a e age ea men e ec s based on B=1,999 eplica ions.
6. Econome ic Resul s
6.1. Main Resul s
Table 4 shows he es ima ed a e age ea men e ec s o he whole ea men g oup, bo h
o a immed model applying he min-max-c i e ion and o a conse a i e imming
model wi h an uppe bound o 0.95. On a e age, he eme gency-aid mode a ely inc eases
he subjec i e su i al p obabili y among hose sel -employed who ecei ed inancial
suppo by 6.5 pe cen age poin s, he e ec is signi ican a he 1%-le el (Table 4, column
1), con i ming hypo hesis 1. Compa ing his e ec o suppo measu es ha sough o
inc ease he su i al o s a -ups (see, e.g. Caliendo and Kuenn, 2011; Caliendo e al.,
2016), hese s udies ind o abou he double e ec size. In his con ex , i mus be
conside ed ha he eme gency-aid consis ed only o a one- ime lump-sum paymen , while
s a -up subsidies comp ised epea ed paymen s o se e al mon hs.
As some obse a ions (n=422) emain unused in he ma ching p ocess, we u he
analyze he obus ness o he e ec size in Sec ion 6.2, also shedding mo e ligh on
he e ogeneous e ec s be ween subg oups.
20
Table 4: ATT o he main sample
T imming app oach
min/max
min / .95
ATT
0.065**
0.058**
SE
(0.023)
(0.021)
p- alue
0.004
0.006
Common suppo
[0.107,0.996]
[0.107,0.950]
N ma ched
6,284
5,174
N unma ched
422
15
N ou o common suppo
50
1,567
N o al
6,756
6,756
No e: Table 4 p o ides in o ma ion on he ATT o he main sample. Column (1) displays he es ima ion
esul o he ma ching model wi h min-max-c i e ion, Column (2) o he ma ching model wi h imming a
p opensi y sco e le el o .95. P opensi y sco es o he ea ed and compa ison g oups a e es ima ed using
p obi eg ession based on he baseline speci ica ion including in o ma ion on esponden s’ socio-
demog aphics, business demog aphics, c isis pe o mance indica o s, and isk a i udes. Ma ching is
pe o med using non-pa ame ic ke nel ma ching wi h an Epanechniko ke nel o es ima e balancing weigh s.
S anda d e o s a e boo s apped wi h B=1,999 eplica ions (*p<.05 **p<.01 *** p<.001).
One migh be conce ned ha he uppe bound is s ill close o uni y and, he e o e,
includes esponden s wi h a nea ly pe ec p edic ion o being ea ed. Excluding pe sons
om he ea men g oup who ha e p opensi y sco es close o 1 does no subs an ially al e
he esul s (Table 4, column 2). Howe e , he conse a i e model disca ds a la ge numbe
o ea ed uni s, ques ioning whe he he es ima ed e ec is s ill ep esen a i e o he ea ed
indi iduals. The e o e, we ocus on he min-max-c i e ion in he subsequen analyses.
I we include he hypo he ical e ec on he con ol g oup, i.e., changes in he
subjec i e su i al p obabili y o he esponden s who a e planning o apply o he
eme gency und (i hey did and ecei ed he paymen ), we ob ain an a e age ea men
e ec o he whole sample popula ion o 6.4%, which is i ually iden ical o he ATT.
Table 5: ATE o he main sample
T imming app oach
min/max
min / .95
ATE
0.064**
0.058**
SE
(0.020)
(0.019)
p- alue
0.002
0.003
Common suppo
[0.107,0.996]
[0.107,0.950]
N ma ched
6,284
5,174
N unma ched
422
15
N ou o common suppo
50
1,567
N o al
6,756
6,756
No e: Table 5 p o ides in o ma ion on he ATE o he main sample. Column (1) displays he es ima ion
esul o he ma ching model wi h min-max-c i e ion, Column (2) o he ma ching model wi h imming a
p opensi y sco e le el o .95. P opensi y sco es o he ea ed and compa ison g oups a e es ima ed in he
same way as in Table 4. S anda d e o s a e boo s apped wi h B=1,999 eplica ions. (*p<.05 **p<.01 ***
p<.001).
21
6.2. E ec He e ogenei ies
The ATT in he main sample measu es he a e age p og am e ec ac oss all indi iduals
who ecei ed inancial suppo om he eme gency-aid und. We a e u he in e es ed in
knowing whe he some indi iduals bene i ed mo e han o he s based on hei exposu e o
he c isis, hei pe sonal cha ac e is ics, o he applica ion p ocess.
6.2.1 E ec by Indus ies
The impac o go e nmen al measu es o con ain he pandemic di e ed ac oss indus ies.
Some indus ies su e ed om e enue declines mo e s ongly han o he s (see Table 1).
The e o e, we explo e he e ogeneous ea men e ec s be ween indus ies and es ima e he
a e age ea men e ec wi hin he pa icula ly a ec ed indus ies – unde which we
subsume ho els and es au an s as well as a s, ec ea ion, and cul u al ac i i ies – agains
less a ec ed indus ies, comp ising manu ac u ing, epai ing o mo o ehicles, ade,
in o ma ion and communica ions, p o essional se ices, educa ion, heal h and social ca e,
and o he se ices.
Table 6: ATT by indus y
Indus ies
se e ely a ec ed by he
c isis
less a ec ed
AATT
0.101**
0.022
SE
(0.034)
(0.036)
p- alue
0.003
0.549
N ma ched
3,235
3,353
N unma ched
15
1
N ou o common suppo
74
78
N o al
3,324
3,432
No e: Table 6 p o ides in o ma ion on he ATT, compa ing esponden s om indus ies pa icula ly a ec ed
by he c isis wi h esponden s om less a ec ed indus ies. Column (1) displays he es ima ion esul o
esponden s om indus ies pa icula ly a ec ed by he c isis, Column (2) o esponden s om less a ec ed
indus ies. P opensi y sco es o he ea ed and compa ison g oups a e es ima ed in he same way as in Table
4. S anda d e o s a e boo s apped wi h B=1,999 eplica ions. (*p<.05 **p<.01 *** p<.001).
On a e age, he eme gency-aid inc eased he subjec i e su i al p obabili y o he
sel -employed in s ongly a ec ed indus ies by 10.1 pe cen age poin s (Table 6), whe eas
he su i al p obabili y in he o he indus ies was – on a e age – una ec ed (hypo hesis
2a). No e ha he e e ence ca ego y is qui e he e ogeneous. The e o e, an insigni ican
o e all e ec does no mean ha single indus ies wi hin his ca ego y did no bene i om
he eme gency aid. Limi s in he sample size p eclude mo e de ailed analysis. F om a policy
pe spec i e, he suppo p og am appea s o ha e p edominan ly imp o ed he subjec i e
su i al p obabili y o sel -employed whose sec o s we e hi ha d by he c isis.
22
6.2.2 E ec by Le el o Educa ion
Since he sel -employed’s le el o educa ion a ec s en ep eneu ial pe o mance and
su i al, we dis inguish be ween pe sons wi h uni e si y deg ee and wi hou . Resul s lis ed
in Table 7 suppo hypo hesis 2b. The eme gency-aid p og am has a s ong and signi ican
e ec , inc easing he subjec i e su i al p obabili y by 10.4 pe cen age poin s among hose
sel -employed wi h a uni e si y deg ee, bu no e ec among pe sons wi hou such deg ee.
Table 7: ATT by educa ion le el
Educa ion
A
uni e si y deg ee
no uni e si y deg ee
ATT
0.104***
0.042
SE
(0.031)
(0.039)
p- alue
0.001
0.291
N ma ched
3,808
2,672
N unma ched
47
41
N ou o common suppo
70
118
N o al
3,925
2,831
No e: Table 7 p o ides in o ma ion on he ATT compa ing esponden s wi h a uni e si y deg ee o
esponden s wi hou one. Column (1) displays he es ima ion esul o he subsample o esponden s wi h
uni e si y deg ee. Column (2) displays he es ima ion esul o he subsample o esponden s wi hou
uni e si y deg ee. P opensi y sco es o he ea ed and compa ison g oup a e es ima ed in he same way as
in Table 4. S anda d e o s a e boo s apped wi h B=1,999 eplica ions (*p<.05 **p<.01 *** p<.001).
6.2.3 E ec by Risk A i ude
Since he sel -employed's willingness o ake isks a ec s hei decision beha io and i m
esul s – including income (H ide and Panos, 2014) and su i al (Caliendo e al., 2010) –
we dis inguish be ween subg oups epo ing di e en le els o isk ole ance.
Table 8: ATT by isk a i ude
Risk a i ude
Low isk
ole ance
Medium isk
ole ance
High isk
ole ance
AATT
-0.005
0.031
0.053
SE
(0.046)
(0.046)
(0.043)
p- alue
0.910
0.509
0.215
common suppo
[0.196,0.980]
[0.220,0.995]
[0.258,0.995]
N ma ched
1,583
2,374
2,288
N unma ched
126
1
140
N ou o common suppo
123
38
83
N o al
1,832
2,413
2511
No e: Table 8 p o ides in o ma ion on he ATT compa ing esponden s wi h a ious le els o isk ole ance.
Column (1) displays he es ima ion esul o esponden s wi h low, Column (2) o esponden s wi h medium,
Column (3) o esponden s wi h high isk- ole ance. P opensi y sco es o he ea ed and compa ison g oup
a e es ima ed in he same way as in Table 4. S anda d e o s a e boo s apped wi h B=1,999 eplica ions
(*p<.05 **p<.01 *** p<.001).
23
As Table 8 shows, we do no ind a signi ican e ec o isk ole ance: he suppo
om he eme gency aid did no measu ably inc ease he subjec i e su i al p obabili y o
he mo e isk ole an sel -employed, hus no con i ming hypo hesis 2c.
6.2.4 E ec by Speed o Paymen
We also in es iga e whe he empo al aspec s in p ocessing and disbu sing he eme gency-
aid a ec he subjec i e su i al p obabili y. We conside how he speed o paymen
in luenced he e ec among he ea ed indi iduals by so ing ea ed indi iduals in o wo
g oups: (i) hose whose applica ions we e p ocessed wi hin 5 days (compa ed o an a e age
o 7.5 days, Sec ion 4.4, Table 2), deno ed as as , and (ii) hose wai ing o mo e han 5
days o hei applica ions o be p ocessed, deno ed as slow.
The esul s, con i ming hypo hesis 2d, a e lis ed in Table 9. Fo he sel -employed
whose applica ions we e p ocessed as , he subjec i e su i al p obabili y inc eases by 6.3
pe cen age poin s on a e age, while we ind no signi ican e ec o indi iduals whose
applica ions we e p ocessed slowly. I appea s ha he speed wi h which he aid was
g an ed and paid ou measu ably a ec s subjec i e su i al p obabili y.
Table 9: ATT by speed o paymen
Speed o paymen
as
(up o 5 days)
slow
(mo e han 5 days)
AATT
0.063*
0.038
SE
(0.032)
(0.024)
p- alue
0.049
0.110
N ma ched
4,457
3,042
N unma ched
2
1
N ou o common suppo
72
19
N o al
4,531
3,062
No e: Table 9 p o ides in o ma ion on he ATT compa ing ea ed esponden s whose applica ions we e
p ocessed wi hin 5 days wi h ea ed esponden s wai ing o mo e han 5 days o hei applica ions o be
p ocessed. Column (1) displays he es ima ion esul o he “ as ” sample, Column (2) o he “slow” sample.
P opensi y sco es o he ea ed and compa ison g oup a e es ima ed in he same way as in Table 4. S anda d
e o s a e boo s apped wi h B=1,999 eplica ions (*p<.05 **p<.01 *** p<.001).
7. Discussion and Conclusions
The Co id-19 pandemic se e ely a ec ed he sel -employed. Many coun ies implemen ed
inancial suppo p og ams designed o help he sel -employed su i e he Co id-19 c isis.
We in es iga e he e ec o he Ge man eme gency-aid p og am, o which €13.7bn was
spen . Launched a he end o Ma ch 2020, sel -employed indi iduals could apply o lump-
30
Meu e , M., Waldki ch, M., Schou, P.K., Buche , E.L., Bu meis e -Lamp, K (2022).
Digi al a o dances: how en ep eneu s access suppo in online communi ies
du ing he Co id-19 pandemic. Small Business Economics 58, 637 – 663.
Minni i, M. (2008). The ole o go e nmen policy on en ep eneu ial ac i i y:
p oduc i e, unp oduc i e, o des uc i e? En ep eneu ship Theo y and P ac ice
32(5), 779-790.
Mona , A., A e ill, J.R., Laza us, R.S. (1972). An icipa o y s ess and coping eac ions
unde a ious condi ions o unce ain y. Jou nal o Pe sonali y and Social
Psychology, 24(2), 237.
Neneh, B. N. (2019). F om en ep eneu ial ale ness o en ep eneu ial beha io : The ole
o ai compe i i eness and p oac i e pe sonali y. Pe sonali y and Indi idual
Di e ences, 138, 273-279.
Nieß, C., Biemann, T. (2014). The ole o isk p opensi y in p edic ing sel -employmen .
Jou nal o Applied Psychology 99(5), 1000-1009.
Pa ke , S.C. (2018). The Economics o En ep eneu ship. New Yo k, Camb idge:
Camb idge Uni e si y P ess, 2nd ed.
Pa ke , S.C., Van P aag, C.M. (2006). Schooling, capi al cons ain s, and en ep eneu ial
pe o mance: he endogenous iangle. Jou nal o Business & Economic S a is ics
24(4), 416-431.
Roy, A. (1951). Some hough s on he dis ibu ion o ea nings. Ox o d Economic Pape s
3(2), 135-145.
Rubin, D. (1974). Es ima ing causal e ec s o ea men s in andomized and
non andomized s udies. Jou nal o Educa ional Psychology 66(5), 688-701.
Sianesi, B. (2004). An e alua ion o he Swedish sys em o ac i e labo ma ke p og ams
in he 1990s The Re iew o Economics and S a is ics 89(1), 133-155.
Simón-Moya, V., Re uel o-Taboada, L., Ribei o-So iano, D. (2016). In luence o
economic c isis on new SME su i al: eali y o ic ion? En ep eneu ship and
Regional De elopmen 28(1), 157–176.
S asielowicz, L. (2020). How impo an is cogni i e abili y when adap ing o changes? A
me a-analysis o he pe o mance adap a ion li e a u e. Pe sonali y and Indi idual
Di e ences 166(1), 110-178.
S u ze , A. (2020). Happiness and public policy: a p ocedu al pe spec i e. Beha iou al
Public Policy 4(2), 210-225.
Tang, N. (2021). Cogni i e abili ies, sel -e icacy, and inancial beha io . Jou nal o
Economic Psychology, 87, 102447.
To ès, O., Benza i, A., Fisch, C., Muke jee, J., Swalhi, A., Thu ik R. (2022). Risk o
bu nou in F ench en ep eneu s du ing he Co id-19 c isis. Small Business
Economics 58(2), 717–739.
Ucbasa an, D., Shephe d, D.A., Locke , A., Lyon, S.J. (2013). Li e a e business ailu e:
he p ocess and consequences o business ailu e o en ep eneu s. Jou nal o
Managemen 39(1), 163-202.
31
Ucbasa an, D., Wes head, P., W igh , M., Flo es, M. (2010). The Na u e o
en ep eneu ial expe ience, business ailu e and compa a i e op imism. Jou nal o
Business Ven u ing 25(6), 541-555.
U big, D., Wei zel, U., Rosenk anz, S., an Wi eloo uijn, A. (2012). Exploi ing
oppo uni ies a all cos ? En ep eneu ial in en and ex e nali ies. Jou nal o
Economic Psychology 33(2):379-393.
Van de Sluis, J., Van P aag, M., Vij e be g, W. (2008). Educa ion and en ep eneu ship
selec ion and pe o mance: a e iew o he empi ical li e a u e. Jou nal o
Economic Su eys 22(5), 795-841.
Vélez-Co o, M., Ru e-Pé ez, S., Pé ez-Ga cía, M., Ca acuel, A. (2021). Unemploymen
and gene al cogni i e abili y: A e iew and me a-analysis. Jou nal o Economic
Psychology, 87, 102430.
Ve heul, I., Thu ik, R., G ilo, I., Van de Zwan, P. (2012). Explaining p e e ences and
ac ual in ol emen in sel -employmen : gende and he en ep eneu ial
pe sonali y. Jou nal o Economic Psychology, 33(2), 325-341.
Wiklund, J., Nikolae , B., Shi , N., Foo, M.D., B adley, S. (2019). En ep eneu ship and
well-being: pas , p esen , and u u e. Jou nal o Business Ven u ing 34(4), 579-
588.
Willeb ands, D., Lamme s, J., Ha og, J. (2012). A success ul businessman is no a
gamble . Risk a i ude and business pe o mance among small en e p ises in
Nige ia. Jou nal o Economic Psychology 33(2), 342-354.
Wunsch, C., Lechne , M. (2008). Wha did all he money do? On he gene al
ine ec i eness o ecen Wes Ge man labou ma ke p og ammes. Kyklos, 61(1),
134-174.
32
Online Appendix
1 Tables
Table A1: Summa y s a is ics
Whole
Sample
Ma ched
sample
T ea men
Sample
Con ol
Sample
Va iables and ca ego ies
%
N
%
N
%
N
%
N
Risk ole ance
Low isk ole ance
29%
4,894
27%
1,832
26%
1,512
32%
320
Medium isk ole ance
36%
5,993
36%
2,413
36%
2,050
36%
363
High isk ole ance
35%
5,972
37%
2,511
38%
2,181
33%
330
Mon hly li ing cos s (in €
Up o 500
2%
413
1%
85
1%
58
3%
27
501 o 1,000
16%
2,632
14%
932
13%
773
16%
159
1,001 o 1,500
26%
4,304
26%
1,730
26%
1,487
24%
243
1,501 o 2,000
22%
3,790
24%
1,618
25%
1,411
20%
207
2,001 o 2,500
14%
2,352
15%
997
15%
855
14%
142
2,501 o 3,000
8%
1,423
9%
604
9%
511
9%
93
3,001 o 3,500
5%
771
5%
327
5%
268
6%
59
3,501 o 4,000
3%
451
3%
182
3%
144
4%
38
4,001 o 4,500
1%
211
1%
100
2%
89
1%
11
4,501 o 5,000
2%
279
1%
100
1%
81
2%
19
Mo e han 5,000
1%
233
1%
81
1%
66
1%
15
Sales decline due o pandemic
No decline o inc ease
2%
296
0%
31
0%
15
2%
16
Up o 25%
6%
933
3%
200
3%
153
5%
47
26% o 50%
13%
2,134
10%
680
9%
520
16%
160
51% o 75%
17%
2,862
17%
1,142
17%
968
17%
174
76% o 99%
25%
4,163
28%
1,921
29%
1,660
26%
261
100%
38%
6,471
41%
2,782
42%
2,427
35%
355
Es ima ed ime o insol ency
No sepa a e business accoun
9%
1,601
9%
621
10%
554
7%
67
Al eady insol en
26%
4,366
25%
1,668
24%
1,402
26%
266
One mon h
16%
2,742
19%
1,273
19%
1,094
18%
179
Two mon hs
16%
2,627
17%
1,164
17%
995
17%
169
Th ee mon hs
15%
2,487
16%
1,060
16%
901
16%
159
Fou o six mon hs
12%
1,997
11%
731
10%
603
13%
128
Mo e han 6 mon hs
6%
1,039
4%
239
3%
194
4%
45
Gende
Male
48%
8,121
52%
3,532
53%
3,017
51%
515
Female
51%
8,665
47%
3,201
47%
2,711
48%
490
Di e se
0%
73
0%
23
0%
15
1%
8
Age
Up o 39 yea s
22%
3,759
24%
1,591
23%
1,348
24%
243
40 o 49 yea s
28%
4,714
28%
1,923
29%
1,647
27%
276
50 o 59 yea s
37%
6,211
36%
2,443
36%
2,089
35%
354
60 yea s and mo e
13%
2,175
12%
799
11%
659
14%
140
Educa ion
O he
21%
3,524
22%
1,512
23%
1,324
19%
188
P o essional educa ion
18%
3,042
20%
1,319
20%
1,150
17%
169
Uni e si y deg ee
61%
10,293
58%
3,925
57%
3,269
65%
656
Fede al s a e
Baden-Wü embe g
10%
1,625
11%
710
10%
572
14%
138
Ba a ia
17%
2,789
9%
621
8%
463
16%
158
Be lin
11%
1,828
19%
1,298
22%
1,241
6%
57
33
B andenbu g
3%
430
2%
154
2%
137
2%
17
B emen
1%
143
0%
31
0%
23
1%
8
Hambu g
5%
903
6%
430
6%
352
8%
78
Hesse
8%
1,329
6%
405
5%
298
11%
107
Mecklenbu g-Wes e n Pome ania
1%
199
1%
48
1%
41
1%
7
Lowe Saxony
8%
1,279
6%
374
5%
296
8%
78
No h Rhine-Wes phalia
21%
3,502
29%
1,957
31%
1,761
19%
196
Rhineland Pala ina e
4%
725
2%
148
2%
89
6%
59
Saa land
1%
103
0%
26
0%
20
1%
6
Saxony
5%
881
4%
276
4%
232
4%
44
Saxony-Anhal
1%
247
1%
53
1%
42
1%
11
Schleswig-Hols ein
3%
584
2%
142
2%
107
3%
35
Thu ingia
2%
292
1%
83
1%
69
1%
14
Du a ion o sel -employmen
0 o 4 yea s
19%
3,271
17%
1,138
17%
954
18%
184
5 o 10 yea s
25%
4,193
24%
1,651
25%
1,416
23%
235
11 o 20 yea s
33%
5,569
34%
2,313
34%
1,960
35%
353
21 o 30 yea s
17%
2,889
19%
1,281
19%
1,093
19%
188
Mo e han 30 yea s
6%
937
6%
373
6%
320
5%
53
Indus y ca ego y
Manu ac u ing
6%
963
6%
390
6%
330
6%
60
T ade; epai o mo o ehicles
2%
419
2%
161
3%
151
1%
10
Accommoda ion and ood se ice
2%
328
3%
172
3%
162
1%
10
In o ma ion and communica ion
12%
2,032
10%
654
9%
532
12%
122
P o essional se ices
8%
1,266
6%
401
6%
322
8%
79
O he se ice ac i i ies
5%
923
4%
245
3%
191
5%
54
Educa ion
12%
2,026
11%
774
11%
643
13%
131
Human heal h and social wo k ac .
7%
1,255
6%
435
6%
351
8%
84
A s, en e ainmen and ec ea ion
41%
6,921
48%
3,260
50%
2,845
41%
415
O he
4%
726
4%
264
4%
216
5%
48
Le el o digi iza ion
(con inuous scale om 1 o 5)
Mean (s d. de )
2.89
(1.17)
2.86
(1.14)
2.87
(1.14)
2.85
(1.17)
Pa - ime/ ull- ime sel -employed
Pa - ime
11%
1,806
5%
312
4%
204
11%
108
Full ime
89%
15,053
95%
6,444
96%
5,539
89%
905
Solo sel -employed
No
21%
3,564
24%
1,629
25%
1,408
22%
221
Yes
79%
13,295
76%
5,127
75%
4,335
78%
792
Applica ion o basic secu i y
(“Ha z IV”)
I will no apply
83%
13963
80%
5398
80%
4599
79%
799
I applied
8%
1312
8%
553
9%
519
3%
34
I plan o apply
9%
1584
12%
805
11%
625
18%
180
Expec ed du a ion o inancial
ha dship
no ha dship
1%
252
1%
40
1%
32
1%
8
1 o 3 mon hs
20%
3,343
17%
1,156
17%
956
20%
200
4 o 6 mon hs
40%
6,751
41%
2,744
40%
2,304
43%
440
7 o 9 mon hs
15%
2,498
16%
1,109
17%
961
15%
148
10 o12 mon hs
15%
2,606
17%
1,124
17%
981
14%
143
Mo e han one yea
8%
1,409
9%
583
9%
509
7%
74
Su ey week
Ap il 6 – Ap il 12, 2020
20%
3,310
17%
1,156
16%
920
23%
236
Ap il 13 – Ap il 19, 2020
26%
4,335
25%
1,674
24%
1,406
26%
268
Ap il 20 –Ap il 26, 2020
41%
6,831
43%
2,916
44%
2,513
40%
403
Ap il 27 – May 4, 2020
14%
2,383
15%
1,010
16%
904
10%
106
34
Table A2: Reasons o no applying o he eme gency aid p og am
Reasons
sha e o hose
who did no apply
O he s
0.10
I hink I am no eligible
0.23
I would need u he in o ma ion
0.04
An applica ion was no possible ye
0.02
Se e o e loaded
0.01
I am wai ing un il condi ions become mo e cla i ied
0.14
No enough ime
0.01
Re enue decline will occu la e
0.14
Unse led by h ea ened consequences o p o iding inco ec
in o ma ion
0.13
I am no in inancial di icul ies ye
0.18
2 Figu es
Figu eA1: Mon hly inancial loss du ing he c isis
35
Figu eA2: Du a ion o sol ency wi hou go e nmen suppo
Figu e A3: Expec ed du a ion o inancial ha dship due o he Co id-19 pandemic
36
Figu e A4: Subjec i e p obabili y o occupa ional su i al du ing he nex 12
mon hs
Figu e A5: Timing o he applica ion p ocess
A.3 Ma ching Quali y
We ope a ionalize he ca ego ical a iables X by a se o dummy a iables esul ing in a
o al o 66 a iables in he p opensi y sco e ma ching. To e i y he ma ching quali y, we
calcula e he s anda dized bias acco ding o Rosenbaum and Rubin (1985), inding ha he
numbe o a iables wi h absolu e s anda dized biases abo e 5% and he mean absolu e
s anda dized bias a e subs an ially educed a e ma ching (Table A3, ows 1-5 and 7), wi h
a mean alue below 5% being gene ally conside ed a success ul bias educ ion (Caliendo
and Kopeining, 2008). One can also ely on wo measu es de eloped by Rubin (2001) o
sepa a ely analyze he ma ching e ec on bias educ ion and a iance. To analyze bias
educ ion, Rubin sugges s compa ing he numbe o s anda d de ia ions be ween he means
37
o he co a ia e dis ibu ions o he ea men and con ol g oups -- usually e e ed o as
Rubin’s B – a guing ha s anda d de ia ions should be less han hal a s anda d de ia ion
apa a e ma ching, p e e ably close o one-qua e . We ob ain a alue o 0.26, showing
ha we success ully educed he bias be ween ea men and con ol g oups (Table A3, ow
8). Since he e is a well-known ade-o be ween bias educ ion and a iance (Caliendo
and Kopeining, 2008), we analyze Rubin’s R, he a io be ween he p opensi y sco e’s
a iances in bo h g oups, be o e and a e he ma ching.
TableA3: Ma ching quali y
Be o e ma ching
A e ma ching
Numbe o a iables
….wi h absolu e s anda dized bias o
0 o less han 1%
4
14
1 o less han 3%
12
29
3 o less han 5%
11
15
5 o less han 10%
17
7
mo e han 10%
22
1
… in o al
66
66
Mean absolu e s anda dized bias in %
6.8
2.1
Rubin’s B
1.01
0.26
Rubin’s R
0.94
1.28
(Re-)es ima ion o he p opensi y sco e: Pseudo- R2
0.14
0.01
Ideally, he a io should be close o one and no exceed [0.5; 2] (Rubin 2001). Table A3
ow 9 illus a es ha he bias educ ion is indeed accompanied by an inc ease in a iance
om 0.94 o 1.28; howe e , he ob ained a io in a iances is s ill close o 1. Finally, e-
es ima ing he p opensi y sco e a e ma ching ob ains a Pseudo-R2 o 0.01, meaning ha
he emaining a ia ion in he ea men pa icipa ion a e ma ching canno be explained
wi h he co a ia es, i.e., he e a e p ac ically no sys ema ic di e ences in he dis ibu ion
o co a ia es be ween he ea ed and con ols a e ma ching (Table A3, ow 10). To sum
up, he a ious measu es indica e ha he ma ched sample is balanced and, condi ional on
he co a ia es, po en ial ou comes a e independen o ea men .
In addi ion o condi ional independence, we equi e ha he p opensi y sco e dis ibu ions
o he ea ed and un ea ed o e lap; i.e. he e is no p opensi y sco e P(X) pe ec ly
p edic ing ea men o non- ea men . Figu e A6 shows he p opensi y sco e dis ibu ion
o bo h g oups. As expec ed, he dis ibu ion o he ea men g oup is le -skewed wi h
ea ed indi iduals ha ing a highe p obabili y o being ea ed han he un ea ed.
Howe e , we ind su icien common suppo o he app oxima e in e al o [0.11;0.99]
and – impo an ly -- he e a e no holes; i.e., we do no obse e a eas ou o common suppo
wi hin he in e al [0.11;0.99], which o he wise would in alida e ou imming app oach
based on he min-max-c i e ion.
38
Figu e A6: Common suppo
A.4 Robus ness Checks
A.4.1 Nea es -Neighbo -Ma ching
To e i y whe he ou esul s a e sensi i e o he choice o he ma ching algo i hm, we
epea he analysis wi h a p opensi y sco e based on nea es -neighbo -ma ching wi h wo
neighbo s and eplacemen . Resul s a e lis ed in Table A4, columns (1) and (2), and a e
qui e simila o hose ob ained unde he Epanechniko ke nel es ima o bo h in e ms o
size e ec and e iciency.
4
The a e age ea men e ec amoun s o 6.9 pe cen age poin s
agains 6.5 pe cen age poin s in he main analysis, and he a e age ea men e ec o he
ea ed is 6.8 pe cen age poin s wi h nea es -neighbo ma ching agains 6.7 pe cen age
poin s in he main analysis. Appa en ly, using mo e obse a ions om he con ol g oup as
ma ching pa ne s unde he ke nel es ima o ma ginally inc eases e iciency wi hou
biasing he esul s. Imposing a calipe o 0.05 does no subs an ially al e he esul (Table
A4, columns (3) and (4)).
Table A4: Nea es -Neighbo -Ma ching wi h wo neighbo s and eplacemen
NN2-Ma ching
wi hou imming
Calipe 0.05
ATE
ATT
ATE
ATT
ea men e ec
0.069
0.068
0.072
0.074
SE
(0.022)
(0.024)
(0.022)
(0.025)
p- alue
0.002
0.005
0.001
0.003
N ma ched
6756
6756
6738
6738
N ou o common suppo
0
0
18
18
N o al
6756
6756
6756
6756
No es: Robus s anda d e o s we e es ima ed ollowing Abadie and Imbens (2016).
4
We calcula e analy ical s anda d e o s ollowing Abadie and Imbens (2016), since Abadie and Imbens
(2008) show ha boo s apping does no p o ide consis en s anda d e o s in he case o nea es -neighbo
ma ching wi h a ixed numbe o neighbo s and eplacemen . No e ha imming is less ele an wi h
nea es -neighbo ma ching since only he wo closes obse a ions a e used, whe eas ke nel ma ching also
uses in o ma ion om a away con ol uni s, depending on he bandwid h chosen.
39
A.4.2 O dinal ou come a iable
The o iginal ou come a iable ha we use o measu e he subjec i e su i al
p obabili y is an o dinal a iable anging om 1 (“ e y unlikely”) o 5 ( e y likely”). In
he main analysis, we ecode he a iable o ob ain a bina y a iable ha can be di ec ly
in e p e ed as p obabili y by se ing ca ego ies 5 (“ e y likely”) and 4 (“ a he likely”)
equal o one, and he emaining ca ego ies 3 (“neu al”), 2 (“ a he unlikely”), and 1 (“ e y
unlikely”) equal o ze o. To e i y whe he he esul s a e sensi i e o he de ini ion o he
bina y a iable, we e-es ima e he ea men e ec s wi h he o iginal a iable. The esul s,
lis ed in Table A5, show a obus posi i e e ec bo h o he ATE and he ATT. Howe e ,
he in e p e a ion o he magni udes is less in ui i e, as ecei ing inancial suppo om he
eme gency und inc eases he su i al pe cep ion by 0.192 uni s on a e age on a scale om
1 o 5. Since he o dinal a iable con ains mo e a ia ion ac oss indi iduals, he ea men
e ec s a e mo e e icien ly es ima ed, suppo ing ou conclusion ha he eme gency
p og am had a measu able e ec on he sel -employed pe sons’ occupa ional su i al
p obabili y, e en hough he magni ude o he e ec is mode a e.
Table A5: O dinal ou come a iable
O dinal ou come a iable
ATE
ATT
ea men e ec
0.188
0.192
SE
(0.052)
(0.058)
p- alue
0.000
0.001
common suppo
[0.107,0.996]
[0.107,0.996]
N ma ched
6284
6284
N unma ched
422
422
N ou o common suppo
50
50
N o al
6756
6756
No es: S anda d e o s a e boo s apped wi h B=1,999 eplica ions. The
p opensi y sco e is es ima ed wi h he Epanechniko ke nel ma ching algo i hm
applying he min-max imming c i e ion. The ou come a iable is he subjec i e
p obabili y o s ay sel -employed o e he nex 12 mon hs coded as 1 (“ e y
unlikely”), 2 (“likely”), 3 (“neu al”), 4 (“likely”), and 5 (“ e y likely”).
Table A6 lis s he esul s o he he e ogenei y analysis, which a e in line wi h he
bina y model wi h an excep ion o he le el o educa ion. The in ui ion o he di e ging
esul be ween o dina y and bina y ou come a iable is he ollowing: While he suppo
p og am inc eased he su i al a e o bo h g oups, i only shi ed he pe cei ed su i al
o people wi h a uni e si y deg ee a e o ca ego y 4 (“likely”) and 5 (“ e y likely”); see
also ca ego ies 1 o 3, Figu e A7).
