The effect of Ukrainian refugees on the local labour markets: the case of Czechia

Pos epska, Agnieszka; Voloshyna, Anas asiia
A icle — Published Ve sion
The e ec o Uk ainian e ugees on he local labou
ma ke s: he case o Czechia
Jou nal o Popula ion Economics
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: Pos epska, Agnieszka; Voloshyna, Anas asiia (2025) : The e ec o Uk ainian
e ugees on he local labou ma ke s: he case o Czechia, Jou nal o Popula ion Economics, ISSN
1432-1475, Sp inge , Be lin, Heidelbe g, Vol. 38, Iss. 1,
h ps://doi.o g/10.1007/s00148-025-01080-9
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/323180
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by/4.0/
Jou nal o Popula ion Economics (2025) 38:30
h ps://doi.o g/10.1007/s00148-025-01080-9
ORIGINAL PAPER
The e ec o Uk ainian e ugees on he local labou
ma ke s: he case o Czechia
Agnieszka Pos epska1·Anas asiia Voloshyna1,2
Recei ed: 4 July 2024 / Accep ed: 6 Janua y 2025 / Published online: 17 Feb ua y 2025
© The Au ho (s) 2025
Abs ac
Following he Russian Fede a ion’s in asion o Uk aine on 24 h Feb ua y 2022, o e
a qua e o he Uk ainian popula ion became displaced. Czechia eme ged as a key
des ina ion, g an ing empo a y p o ec ion o app oxima ely 433,000 Uk ainians by
he end o 2022, hus shel e ing he highes pe capi a numbe o Uk ainian e ugees
wo ldwide. The swi enac men o he ‘Lex Uk aine Ac ’ g an ed he e ugees un e-
s ic ed access o he labou ma ke . This led o a no able inc ease in he numbe o
legally employed Uk ainians and expanded Czechia’s wo k o ce. Using indi idual
mic o-le el da a om 16 wa es o he Labou Fo ce Sample Su ey (LFSS), col-
lec ed be ween he 1s qua e o 2019 and he 4 h qua e o 2022, we examine he
sho - e m impac o he in lux o he Uk ainian e ugees in o he wo k o ce on he
labou ma ke ou comes o locals in Czechia. Inco po a ing se e al empi ical s a e-
gies, including a wo-way ixed e ec s model (TWFE), ex ensions o he canonical
di e ence-in-di e ences (DiD) es ima o , and ma ching on selec i e cha ac e is ics
o indi iduals/dis ic s and p e- ea men ends, we ind consis en e idence ha he
in luxo e ugeeshadnoeconomicallymeaning ul impac on employmen ,unemploy-
men , o inac i i y a es wi hin he local popula ion, ega dless o gende , educa ional
le el, o indus y. Mos impo an ly, we ind consis en e idence o an inc ease in
weekly wo king hou s among local emales in ea ed dis ic s. This inc ease is p i-
ma ily d i en by wo ke s wi h seconda y educa ion employed in he mos a ec ed
sec o s.
Keywo ds Uk ainian e ugees ·Immig an s ·Local labou ma ke ·Labou supply
JEL Classi ica ion F22 ·J15 ·J21
Responsible edi o : Klaus F. Zimme mann
BAgnieszka Pos epska
[email p o ec ed]
Anas asiia Voloshyna
[email p o ec ed]
1Facul y o Economics and Business, Depa men o Economics, Econome ics & Finance,
Uni e si y o G oningen, 9700 AV G oningen, The Ne he lands
2Max Planck Ins i u e o Demog aphic Resea ch, Ros ock, Ge many
123
30 Page 2 o 45 A. Pos epska, A. Voloshyna
1 In oduc ion
Following he in asion o Uk aine by he Russian Fede a ion on 24 h Feb ua y 2022,
o e a qua e o he Uk ainian popula ion became displaced (IOM 2023b; UNHCR
2023b). By Decembe 2022, he Uni ed Na ions High Commissione o Re ugees
epo ed ha nea ly 8 million indi iduals, mainly women o wo king age and child en,
had sough e uge ac oss Eu ope, wi h abou 5 million egis e ing o empo a y p o-
ec ion o equi alen na ional p o ec ion p og ammes. This e ugee c isis is he la ges
in Eu ope since Wo ld Wa II, exceeding he displacemen caused by he Yugosla
Wa s o he 1990s and he Sy ian Ci il Wa .1
Due o hei geog aphicalandcul u alp oximi y, he Viseg adG oup(V4)coun ies
became p ima y des ina ions o Uk ainians leeing he con lic (GLOBSEC 2023).
Czechia, in pa icula , eme ged as a key e uge. By he end o 2022, his mid-sized
Eu opean coun y wi h 10.5 million inhabi an s had g an ed empo a y p o ec ion
o app oxima ely 433,000 indi iduals.2Consequen ly, Czechia hos s he highes pe
capi a numbe o Uk ainian e ugees wo ldwide.
In esponse o he sudden in lux o e ugees, he Czech go e nmen swi ly enac ed
he‘Lex Uk aineAc ’inMa ch 2022(Eu opeanCommission2022b).Alignedwi h he
Eu opean Union Tempo a y P o ec ion Di ec i e (TPD), his legisla ion empo a ily
ex ended bene i s ese ed o pe manen esiden s o Uk ainian ci izens, hei amily
membe s, and o he speci ied indi iduals. Re ugees we e g an ed un es ic ed access
o he labou ma ke , e aining p og ammes, oppo uni ies o sel -employmen ,
unemploymen bene i s, heal hca e, educa ion, and li ing allowances. This app oach
con as s sha ply wi h he usual leng hy p ocedu es in he EU, whe e e ugees o en
wai mon hs o e en yea s o gain such igh s.3
Typically, e ugees ace p olonged p ocedu es o access employmen oppo uni-
ies, esul ing in a g adual and less exogenous in lux o wo ke s in o hos coun ies’
labou ma ke s. Howe e , he immedia e and un es ic ed access g an ed o Uk ainian
e ugees mean ha he demog aphic shock quickly ansla ed in o a signi ican labou
supply shock. By he end o 2022, Uk ainian e ugees legally employed ep esen ed
app oxima ely 1.4% o all employed indi iduals in Czechia.4This in lux was une enly
dis ibu edac ossdis ic sandindus ies.Ce ain egions—suchasCheb,Plzeˇn-Sou h,
Plzeˇn-Ci y, P aha-Eas , Mladá Bolesla , and Tacho —expe ienced pa icula ly high
in lows, wi h Uk ainian e ugees a imes cons i u ing be ween 3 and 32% o hei
o al employmen . Mo eo e , he e ugees a i ed du ing a ou able condi ions in he
1The Yugosla Wa s in he 1990s esul ed in app oxima ely 2 million people leeing Bosnia, 500,000 om
C oa ia, 100,000 om Se bia, and 30,000 om Slo enia (USCRI 1998). The Sy ian Ci il Wa displaced
a ound 6.6 million Sy ians, wi h Eu opean coun ies hos ing jus o e 1 million (UNHCR 2023a).
2This coun includes only hose who secu ed empo a y p o ec ion s a us; he ac ual numbe o e ugees
in Czechia may be highe o lowe .
3In he EU, he ime e ugees wai o ob ain he igh o wo k a ies (ECRE 2024); since Ma ch 2020, o
example Ge many gene ally p ohibi s asylum seeke s in ini ial ecep ion cen es om aking up employ-
men , wi h mos adul s acing a wai o 18 mon hs and up o 24 mon hs in some ede al s a es (ECRE
2023).
4By ‘legal employmen ’, we e e o posi ions o icially egis e ed wi h he Czech Minis y o Labou and
SocialA ai s, includingall ypes o con ac s suchasDPP o sho - e m wo k,bu excludingsel -employed
indi iduals.
123
The e ec o Uk ainian e ugees on he local labou ma ke s… Page 3 o 45 30
Czech labou ma ke , cha ac e ised by he lowes in he EU unemploymen a e o jus
2.22% and a pe sis en su plus o job acancies o e job seeke s (Minis y o Labou
and Social A ai s 2023b; Eu os a , 2023).
In ou pape , we exploi his na u al expe imen o he sudden and o ced in lux
o Uk ainian e ugees, which signi ican ly expanded Czechia’s wo k o ce. We assess
he sho - e m impac —o e 1 yea —o legally employed Uk ainian e ugees on he
labou ma ke ou comes o local wo ke s in Czechia. P e ious esea ch has o en
u ilised such la ge-scale mig a ion wa es igge ed by wa s o poli ical uphea als o
examine hei e ec s on hos coun ies’ labou ma ke s. A p ominen example is he
1980 Ma iel Boa li , du ing which app oxima ely 125,000 Cuban e ugees inc eased
Miami’s labou o ce by abou 7%. While Ca d (1990) ini ially ound minimal e ec s
on local wages and employmen , la e s udies by Ca d (2001) and Bo jas (2003,2017)
con es ed hese indings, e ealing signi ican wage educ ions among low-skilled
na i es. Simila ly, a e Alge ia’s independence in 1962, F ance expe ienced an in lux
o abou 900,000 epa ia es—amoun ing o 1.6% o he o al F ench labou o ce—
wi h es ima es indica ing minimal impac s on unemploymen bu somewha la ge
nega i e impac s on wages (Hun 1992). In Is ael, he a i al o immig an s om he
o me So ie Union be ween 1990 and 1994 inc eased he popula ion by 12%, wi h
s udies iden i ying no o only sho -las ing ad e se e ec s on na i e employmen and
wages (F iedbe g 2001; Cohen-Goldne and Pase man 2011). Mo e ecen ly, be ween
2012 and 2016, he in lux o Sy ian e ugees inc eased Ge many’s popula ion by
~0.7% and Sweden’s by ~1.5%, al hough he e ec s on local labou ma ke s emain
la gely unexamined.
In summa y, empi ical esea ch o en inds li le o no impac o immig a ion on he
popula ion-wide employmen o wages o local wo ke s.5Howe e , when analyses
ocus on speci ic demog aphic g oups—pa icula ly hose who sha e cha ac e is-
ics wi h he immig an s—mo e p onounced and a ied e ec s eme ge; o ins ance,
ad e se e ec s o immig a ion ha e been iden i ied o local low-skilled males and
mino i ies when immig an s compe e di ec ly wi h local wo ke s,6whe eas he in lux
o emale immig an labou p o iding a o dable household se ices has been linked
o inc eased labou o ce pa icipa ion among high-po en ial emale ea ne s, indica -
ing ha immig an s can also complemen he local wo k o ce.7Mo eo e , se e al
s udies—such as hose by Ang is and Kugle (2003); Lemos and Po es (2008); Gli z
(2012); Aydemi and Kı da (2017)—ha e emphasised he c ucial ole o local labou
ma ke condi ions,includingma ke lexibili y,wage igidi y,andp e-exis ingemploy-
men a es, in de e mining how e ec i ely hos economies can abso b immig an s and
he consequen e ec s on na i e wo ke s.
Building on empi ical s a egies commonly employed in his li e a u e, we adop a
egional app oach. Fi s , we cons uc a se o h ee ea men a iables, each designed
5See, o example Ca d (1990); F iedbe g and Hun (1995); Bo jas e al. (1996); Pischke and Velling
(1997); F iedbe g (2001); Ang is and Kugle (2003); Ca d (2009).
6See, o example Hun (1992); Bo jas (1994); Ca ing on and de Lima (1996); Ca d (2001); Bo jas (2003);
Nickell and Saleheen (2008); Dus mann e al. (2005); Bo jas and Ka z (2007); Lemos and Po es (2008);
Mansou (2010); Cohen-Goldne and Pase man (2011); O a iano and Pe i (2011); Gli z (2012); Mays ad
and Ve wimp (2014); Aydemi and Kı da (2017); Bo jas (2017); Ce i o˘glu e al. (2017).
7See, o example Co és and Tessada (2011); Fa e e al. (2011); Co és and Pan (2013).
123
30 Page 4 o 45 A. Pos epska, A. Voloshyna
o cap u e he exposu e o local labou ma ke s in Czechia o he labou supply shock
esul ing om Uk ainian e ugees en e ing he wo k o ce. Using agg ega ed dis ic -
le el da a on he numbe o legally employed Uk ainians p o ided by he Minis y o
Labou and Social A ai s (2023a), we es ima e he numbe o employed e ugees in
each qua e o 2022 and no malise his by he size o he local labou o ce in each
co esponding dis ic . The esul ing ea men a iables a y in in ensi y ( e e ed o
as ‘ ea men doses’, anging om 0% o 32%) ac oss dis ic s and o e ime, wi h each
‘dose’ ep esen ing a 1% inc ease in he dis ic ’s wo k o ce a ibu able o Uk ainian
e ugees who became legal employees.
Wi h he ea men a iables de ined, we exploi a ia ion in ea men doses ac oss
ime and dis ic s o ela e hem o changes in he labou ma ke ou comes o locals—
p ima ily Czech na ionals and a small subsample o non-Uk ainian mig an s. We
u ilise indi idual qua e ly panel mic oda a om 16 wa es o he Labou Fo ce Sample
Su ey (LFSS) conduc ed in Czechia, spanning om he 1s qua e o 2019 o he 4 h
qua e o 2022. These da a allowus o examinechanges o e ime in bo h he ex ensi e
ma gin (employmen , unemploymen , and inac i i y s a uses) and he in ensi e ma gin
(weekly hou s wo ked). To accoun o po en ial he e ogenei y in ea men e ec s
ac ossdi e en demog aphicg oups,wedisagg ega e hees ima edimpac sbygende ,
educa ional a ainmen , indus y o employmen , ype o employmen con ac , and
coun y o o igin ( o eign-bo n e sus na i e-bo n).
Ou iden i ica ion s a egy un olds in se e al s eps. We begin by implemen ing
a s a ic wo-way ixed e ec s (TWFE) eg ession. Recognising he limi a ions o
he TWFE app oach in ou complex se ing—speci ically, i s po en ial ailu e o
cap u e ea men e ec he e ogenei y ac oss indi iduals and o e ime—we sub-
sequen ly adop he he e ogenei y- obus es ima o p oposed by de Chaisema in and
D’Haul œuille(2024)asou p ima yme hod. Thisex endeddi e ence-in-di e ences
(DiD) me hod enables unbiased es ima ion o ea men e ec s using (non-)bina y,
(non-)s agge ed ea men s and allows o dynamic/in e - empo al ea men e ec
analysis, making i highly sui able o ou con ex .
Gi en he absence o a andomised expe imen , we pay ca e ul a en ion o he
possibili y o sel -selec ion among he e ugees. We no malise ou ea men a iables
by dis ic labou ma ke size o p o ide a meaning ul measu e o ea men in en-
si y and o mi iga e biases om sel -selec ion. This app oach e eals he e ogenei y in
p e- ea men ends in analysed labou ma ke ou comes as well as in p e- ea men
economic cha ac e is ics ac oss bo h ea men and con ol dis ic s. To ensu e he
obus ness o ou es ima es and le e aging he he e ogenei y in p e- ea men ends
ac oss bo h ea ed and con ol dis ic s, we ma ch ea ed and con ol dis ic s on hei
p e-2022 ends, calcula ed sepa a ely o subg oups de ined by gende and educa-
ion le el, as well as gende and indus y o employmen . Addi ionally, we in oduce
ma ching on indi idual cha ac e is ics o ensu e ha ea ed and con ol indi iduals
a e always compa ed wi h simila coun e pa s.
Finally, o ou main es ima o — he ex ended DiD—we igo ously es he pa allel
ends assump ion using placebo es ima o s. Reassu ingly, o he majo i y o ou
analyses ac oss all model speci ica ions and sub-popula ions, hese placebo es s a e
no signi ican . Addi ionally, beyond add essing any po en ial issues wi h he ‘no
an icipa ion’ and ‘no sel -selec ion among locals’ assump ions, we demons a e ha
123

The e ec o Uk ainian e ugees on he local labou ma ke s… Page 5 o 45 30
he e a e no sys ema ic associa ions be ween p e-2022 economic condi ions and 2022
ea men in ensi ies ac oss dis ic s, u he enhancing he c edibili y o ou indings.
Ou esul s a e in line wi h he exis ing li e a u e. We ind consis en e idence ha
he in lux o e ugees had no economically meaning ul impac on employmen , unem-
ploymen , o inac i i y a es o Czech na ionals in he sho un, ega dless o gende ,
educa ional le el, o indus y o employmen . Examining he mos a ec ed indus-
ies, we iden i y mino nega i e e ec s on employmen wi hin sec o s ha saw he
la ges in lux o wo ke s. Addi ionally, he e is some e idence o sugges ha o eign-
bo n indi iduals in Czechia may ha e expe ienced a sligh dec ease in employmen
p obabili y alongside a co esponding inc ease in unemploymen p obabili y. How-
e e , due o he small sample sizes, we ea hese esul s wi h cau ion. We sugges
ha he combina ion o a igh labou ma ke and exis ing sho ages in key indus ies
likely mi iga ed po en ial dis up ions om he in lux o e ugees, enabling he Czech
economy o abso b he new wo k o ce ela i ely smoo hly, a leas in he sho e m.
Mos impo an ly, we ind consis en e idence o an inc ease in weekly wo king hou s
among local women in ea ed dis ic s. This inc ease is p ima ily d i en by wo ke s
wi h seconda y educa ion employed in he mos a ec ed sec o s. The concen a ion o
posi i e e ec s among women wi h seconda y educa ion likely e lec s bo h he na u e
o he jobs Uk ainian e ugees a e aking and he speci ic demands o he Czech labou
ma ke . Indi iduals wi h seconda y educa ion can occupy oles ha complemen hose
illed by he incoming wo k o ce, leading o longe hou s due o inc eased demand o
collabo a i e oppo uni ies.
We acknowledge ha he scope o ou s udy is limi ed o es ima ing he e ec o
Uk ainian e ugees who a e legally employed in Czechia. While inco po a ing da a on
hein o malsec o wouldbeo signi ican in e es ,li leisknownabou henumbe and
loca ion o undocumen ed e ugee wo ke s. Al hough some a emp s ha e been made
o es ima e hese igu es a di e en poin s in ime— o example a su ey conduc ed in
Czechia by Ka ano á e al. (2022a) ound ha 7% o e ugees epo ed using in o mal
labou b oke s, and ano he su ey ac oss se e al EU coun ies by Eu opean Union
Agency o Fundamen al Righ s (2023) indica ed ha 8% o su eyed e ugees had
wo ked wi hou a con ac o wi h a con ac ha did no co e all wo king hou s—
he accu acy o hese pe cen ages and hei a ia ion by dis ic and o e ime emain
unknown. I we accep hese es ima es, i indica es ha mos e ugees a e no employed
in o mally, s essing he ele ance o ou analysis.
Simila ly, o in e he e ec s on local Czech wo ke s, we ely on he highly eli-
able LFSS da a, which e ec i ely cap u es o mal employmen sec o s. Howe e , due
o i s sampling design, i may no ully e lec he expe iences o hose in in o mal
employmen o ce ain mig an g oups. Consequen ly, we can only discuss he pa ial
e ec o he e ugee in low. We in e p e he impac on o he mig an g oups in he
coun y wi h cau ion.
Wi hin he scope o he analysis, we con ibu e o se e al s ands o he li e a u e.
To ou knowledge, we a e he i s o p o ide a ho ough analysis o he impac o he
mos ecen e ugee c isis in Eu ope on he labou ma ke . Since he Uk ainians we e
g an ed access o he labou ma ke almos immedia ely a e en y, we con ibu e
o he b oade li e a u e on he e ec s o immig a ion on he hos coun y’s labou
ma ke . We also documen e ugee se lemen pa e ns consis en wi h he li e a u e
123
30 Page 6 o 45 A. Pos epska, A. Voloshyna
Fig. 1 E olu ion o Uk ainian immig an and e ugee egis a ions in Czechia. No e: This igu e de ails he
coun o Uk ainian immig an s esiding in Czechia, dis inguishing be ween he o al popula ion and hose
o wo king age (18–65 yea s). The no iceable up ick co esponds o he a i al o e ugees. The plo was
c ea ed by he au ho s om da a epo ed by he Minis y o he In e io (2023a,b) o Czechia
on ne wo k e ec s and sel -selec ion o immig an s (Ha on and Williamson 1998;
Wood u and Zen eno 2007; Pa el and Vella 2013; S ua and Taylo 2021).
Las ly, he esul s o his pape a e pa icula ly impo an o policymake s. Fi s ,
ou indings clea ly poin o g oups o wo ke s ha a e ulne able o he in lux o
o eign wo ke s. Second, gi en he inc easingly pola ised public a i udes owa ds
pas and u u e policies on e ugeein eg a ionand accommoda ion (including inancial
assis ance o Uk ainian e ugees) and u u e EU accession, he e is a p essing need
o objec i e, da a-d i en insigh s in o he e ec s o e ugees’ ac i e pa icipa ion in
labou ma ke s.8Theuniqueci cums anceso heun es ic edaccess oemploymen o
Uk ainian e ugees allow us o unde s and hese e ec s al eady in he sho e m. This
is impo an o expanding he body o academic knowledge and in o ming e ec i e
policymaking.
The emainde o his pape is o ganised as ollows. The nex sec ion p o ides back-
g oundin o ma ion on he Uk ainian e ugee in lux in 2022, de ailing hedemog aphic
cha ac e is ics o Uk ainian e ugees, se lemen pa e ns, and wo k o ce in eg a ion.
Sec ion3discusses he da a and desc ip i e s a is ics, while Sec ion 4ou lines he
iden i ica ion s a egy. Baseline esul s, along wi h all ex ensions o he es ima o s
and discussion, a e p esen ed in Sec ion 5, ollowed by auxilia y obus ness checks in
Sec ion 6and conclusions in Sec ion 7.
8In Poland, a Polls e Resea ch Ins i u e su ey shows inc easing opposi ion o aiding Uk ainians, wi h
36% opposed and 26% in suppo (Fo sal 2023). Ano he su ey indica es a di ided s ance, wi h 49.1% in
a ou o aid bu 39.4% iewing Uk ainians nega i ely, some ci ing pe cep ions o a ‘demanding a i ude’ by
e ugees, and 14.5% belie e Uk ainians ha e mo e igh s han Poles (DGP 2023). EU-wide, Eu oba ome e
e eals a sligh decline in suppo o Uk aine: 86% (down om 88%) back humani a ian aid and 77%
(down om 86%) suppo accep ing wa e ugees (Eu opean Commission 2023a,b). Rega ding Uk aine’s
EU accession, 67% o Eu opeans endo se i , bu suppo a ies: high in Denma k (79%) and Po ugal (88%),
lowe in Ge many (60%) and F ance (60%), and e y low in G eece (43%), Hunga y (50%), and Slo akia
(50%) (Eu opean Commission 2023a). Ano he su ey con i ms 63% o e all suppo o Uk aine’s EU
membe ship, wi h less en husiasm in F ance (52%) and Ge many (49%) (GMF 2023).
123
The e ec o Uk ainian e ugees on he local labou ma ke s… Page 7 o 45 30
2 Con ex ual de ails and economic heo y: analysing po en ial labou
ma ke esponses
To unde s and he economic implica ions o he obse ed da a pa e ns, we examine
he in lux o Uk ainian e ugees in o Czechia, ocusing on hei se lemen pa e ns,
demog aphic p o iles, and in eg a ion in o he labou ma ke . We iden i y indus ies
ha expe ienced a no able inc ease in e ugee wo ke s and summa ise he local labou
ma ke condi ions in Czechia. Wi h ele an economic heo y in mind, we also discuss
he po en ial e ec s o he e ugee in lux on he labou ma ke ou comes o local
wo ke s.
Demog aphic cha ac e is ics o he Uk ainian e ugees By 31 Decembe 2022,
Czechia had welcomed app oxima ely 433,000 Uk ainian e ugees (Fig. 1)—
p edominan lywo king-agewomenandchild en—ademog aphicp o iledis inc om
he coun y’s ypical mig a ion pa e ns.9As shown in Table 1, he age dis ibu ion o
Uk ainians who sough e uge in Czechia la gely mi o s ha o he local popula ion,
wi h a no able di e gence only in he g oup o e 65 yea s old (jus 4% o e ugees
e sus 20% o locals). App oxima ely 64% o e ugees we e o wo king age (18–65
yea s old), 69% o whom we e women. This gende imbalance is mos ly a ibu able
o Uk aine’s wa ime egula ions, which es ic ed many males o comba age om
lea ing he coun y.
The e ugees gene ally had highe educa ional a ainmen le els han he local
Czech popula ion (Table 2). Depending on he sou ce, he pe cen age o hose wi h
e ia y educa ion was es ima ed o be be ween 35 and 49%, no iceably exceeding
he 18% a e age a e among Czech locals. While his gap was somewha na owe
in u ban a eas such as Capi al Ci y P ague and B no-Ci y, wi h local e ia y a es a
34% and 32%, espec i ely, i became mo e p onounced in smalle , mo e pe iphe al
dis ic s like Tacho and Cheb.
Se lemen pa e ns and wo k o ce in eg a ion The in lux o Uk ainian e ugees
led o a signi ican expansion o Czechia’s popula ion and wo k o ce. By he end
o 2022, e ugees cons i u ed a ound 4% o he coun y’s esiden s— he highes pe
capi a numbe o Uk ainian e ugees globally. Due o he immedia e and un es ic ed
access o he local labou ma ke g an ed o Uk ainian e ugees by he ‘Lex Uk aine
Ac ’, he demog aphic shock apidly ansla ed in o a signi ican labou supply shock,
wi h app oxima ely 75,000 secu ing legal employmen by he end o 2022, 79% o
whom we e women (Minis y o Labou and Social A ai s 2023a).10 Acco ding o
su eys conduc ed by Ka ano á e al. (2022b), his accoun ed o oughly hal o he
economically ac i e Uk ainian e ugees.
9F om 2016 o 2021, app oxima ely 57% o immig an s in Czechia we e male, p ima ily om Uk aine,
Slo akia, and Russia (Minis y o he In e io 2023a). Mos o hese immig an s we e labou mig an s
employed in manu ac u ing, as well as in semi-skilled adminis a i e and suppo se ice oles (Minis y o
Labou and Social A ai s 2023a). Un il 2022, Czechia had ecei ed ewe e ugees han mos EU coun ies,
wi h only 1046 by 2021 (Minis y o he In e io 2022). This g oup comp ised la gely younge males om
he o me So ie Bloc and coun ies such as China and Sy ia who we e escaping con lic s and c ises.
10 We app oxima ed his numbe . Fo he de ails, see Sec ion 4.1.
123
30 Page 8 o 45 A. Pos epska, A. Voloshyna
Table 1 Age and gende o Uk ainian e ugees compa ed o he Czech popula ion
Re ugees Locals
O e all P ague B no-més o Tacho Cheb O e all P ague B no-més o Tacho Cheb
Gende
Female 63% 64% 63% 69% 66% 51% 51% 51% 50% 51%
Male 37% 36% 37% 31% 34% 49% 49% 49% 50% 49%
Age
0–5y.o8%8%7% 4%7%5%5%6% 5%5%
6–14 y.o 18% 17% 16% 11% 16% 11% 10% 10% 11% 11%
15–17 y.o 6% 6% 6% 4% 5% 5% 4% 4% 5% 5%
18–64 y.o 64% 65% 67% 79% 67% 59% 62% 61% 61% 58%
65+ y.o 4% 4% 3% 2% 5% 20% 18% 20% 19% 21%
No e: This able compa es he age and gende dis ibu ion be ween Uk ainian e ugees in Czechia as o
31 Decembe 2022 and he Czech na i e popula ion based on he 2021 Census. The able was c ea ed
by he au ho s using da a sou ced om he Minis y o he In e io (2023b) and he 2021 Census (Czech
S a is icalO ice,2024b).Ageca ego iesha ebeenha monised oensu ecompa abili y.Capi al Ci yP ague
and B no-Ci y we e selec ed as economically s onge dis ic s, while Tacho and Cheb we e chosen o
ep esen mo e pe iphe al egions, wi h bo h selec ed andomly o highligh he he e ogenei y o indica o s
ac oss di e en a eas
By ‘legal employmen ’, we e e o posi ions o icially eco ded by he Minis y
o Labou and Social A ai s (2023a), including all ypes o con ac s such as DPP
o sho - e m wo k, bu excluding sel -employed indi iduals. While ano he 5000
Uk ainians ob ained alid ade licences, enabling en ep eneu ial ac i i ies (Minis y
o Indus y and T ade 2023), he loca ions o hese indi iduals by dis ic and hei
a ia ion o e imea eno epo ed.The e o e, hispape ocuses solely on employees.
Addi ionally, e ugees wo king unlaw ully wi hou egis a ion a e na u ally also no
included among he legally employed.
Table 2 Educa ional a ainmen o Uk ainian e ugees compa ed o he Czech popula ion
Re ugees Locals
MoLSA IOM UNHCR O e all P ague B no-més o Tacho Cheb
Educa ion a ainmen
Te ia y 35% 49% 44% 18% 34% 21% 8% 9%
Pos -seconda y 14% 5% 21% 32% 35% 33% 29% 30%
Seconda y 39% 30% 20% 31% 17% 20% 37% 34%
P ima y/basic 7% 15% 3% 13% 8% 9% 17% 17%
No educa ion 5% – 13% 1% 0% 0% 1% 1%
No iden i ied – – 1% 6% 6% 5% 9% 9%
No e: The able was c ea ed by he au ho s using 2021 Census da a o Czechs (Czech S a is ical O ice,
2021) and Uk ainian e ugee educa ion da a om su eys conduc ed by he Czech Minis y o Labou and
Social A ai s (2022) (MoLSA), IOM (2023a), and UNHCR (2022). The la e wo su eys, being non-
ep esen a i e, p o ide only indica i e insigh s. Educa ional ca ego ies we e ha monised o compa abili y.
Fo de ailed in o ma ion on hese changes, including su ey imings and sample sizes, e e o he ex ended
Table B.1
123
The e ec o Uk ainian e ugees on he local labou ma ke s… Page 15 o 45 30
o he o al da ase . This esul s in a sample o 682,757 obse a ions, co esponding
o 179,525 indi iduals.12
The LFS su ey employs a s a i ied wo-s age clus e sampling design and gen-
e a es a na ionally ep esen a i e da ase wi h a la ge sample size, p o iding de ailed
longi udinal in o ma ion on indi iduals’ socio-demog aphic cha ac e is ics (e.g. age,
educa ion, ma i al s a us) and labou ma ke ou comes (e.g. employmen s a us,
employmen his o y, indus y and occupa ion, hou s wo ked, and unemploymen du a-
ion). Gi en ou ocus on he impac o legally employed Uk ainian e ugees on he
employmen ou comes o Czechs, his da ase is highly eliable as i accu a ely cap-
u es he dynamics o he local o mal employmen sec o . Howe e , due o he na u e
o i s sampling design (de ailed in Appendix A), he su ey may no ully cap u e
he expe iences o Czech wo ke s in in o mal employmen o ce ain mig an g oups.
Consequen ly, ou indings a e es ic ed mainly o he o mal employmen sec o , and
we in e p e any e ec s on o he mig an g oups a ec ed by Uk ainian e ugees wi h
cau ion.
4 Iden i ica ion s a egy
In ou analysis, we exploi he na u al expe imen o he sudden and o ced in lux o
Uk ainian e ugees, which signi ican ly expanded Czechia’s wo k o ce. Adop ing a
egional app oach, we assess he sho - e m impac —o e 1 yea om he 1s qua e
up o and including he 4 h qua e o 2022—o legally employed Uk ainian e ugees
on he labou ma ke ou comes o ‘local’ wo ke s in Czechia. Local wo ke s include
p ima ily Czech na ionals and a small subsample o non-Uk ainian mig an s. We
examine bo h he ex ensi e ma gin—analysing s a uses o employmen , unemploy-
men , o inac i i y among he locals—and he in ensi e ma gin, ocusing on weekly
hou s wo ked. Fu he mo e, we di e en ia e he es ima ed e ec s based on gende ,
educa ional a ainmen , indus y o employmen , ype o employmen con ac ( ixed
e m e sus pe manen ), and he coun y o o igin o he Czech esiden s ( o eign-bo n
e sus Czech-bo n). Figu e3illus a es he cen al ocus o he pape .
The iden i ica ion s a egy un olds in se e al s eps, beginning wi h he cons uc-
ion o he ‘ ea men ’ a iables in Sec ion 4.1. Subsequen ly, we implemen a s a ic
wo-way ixed e ec s (TWFE) eg ession, de ailed in Sec ion 4.2. Recognising he
12 Due o a egula o y change implemen ed by he Czech S a is ical O ice (CZSO), unique iden i ie s (IDs)
o indi iduals we e no longe disclosed in he hi d and ou h qua e s o 2022. Howe e , he me hodology
emained consis en , ensu ing ha he subse o indi iduals obse ed in Q3 and Q4 was he same as in Q2
and ea lie . We eco e ed he panel s uc u e o he da a by i s using de e minis ic ma ching o iden i y
unique pai s among indi iduals based solely on a ailable ime-in a ian a iables such as he sequence
numbe o he obse a ion pe iod, gende , yea o bi h, coun y o bi h, and employmen s a us om
all p e iously obse ed pe iods. This app oach success ully ma ched a ound 67% o he obse a ions in
he hi d and ou h qua e s o 2022 o hei co esponding obse a ions om he p e ious qua e . Fo
he emaining 33% o obse a ions, whe e duplica es exis ed due o indi iduals sha ing he same ime-
in a ian cha ac e is ics, we employed p obabilis ic ma ching. A andom o es model, highly sui able o
his classi ica ion ask, was used o calcula e he likelihood o wo indi iduals being a ma ch, allowing he
inco po a ion o ime- a ian a iables such as educa ion le el, ma i al s a us, and o he s. As a esul , we
eliably ma ched he emaining obse a ions, wi h only 0.8% o he obse a ions unma ched in Q3 and
2.2% in Q4 2022. The ma ching p ocess is de ailed in Appendix D.
123

30 Page 16 o 45 A. Pos epska, A. Voloshyna
Fig. 3 Pa hways o e ugee employmen and hei po en ial e ec s on Czechia’s labou ma ke . No e: The
plo was c ea ed by he au ho s using he da a epo ed by he Minis y o he In e io (2023b)and he
Minis y o Labou and Social A ai s (2023a) o Czechia. I illus a es he cen al ocus o he pape :
he impac o Uk ainian e ugees secu ing legal employmen on he labou ma ke ou comes o he local
popula ion (highligh ed by he ed do ed line). The numbe o indi iduals ca ego ised as ‘no seeking
employmen ’ is indica i e and includes people unde 18 o o e 65
limi a ions o he TWFE eg ession in ou complex se ing—speci ically, i s po en ial
ailu e oiden i yacon excombina iono indi idual ea men e ec sandi sdi icul y
in cap u ing ea men e ec he e ogenei y ac oss indi iduals and ime—we adop a
he e ogenei y- obus es ima o in Sec ion 4.3. Acco dingly, we selec as ou p ima y
es ima o he me hod p oposed by de Chaisema in and D’Haul œuille (2024), a a i-
an o he ex ended DiD, which enables he unbiased es ima ion o ea men e ec s
using (non-)bina y, (non-)s agge ed ea men s and allows o dynamic/in e - empo al
ea men e ec analysis, making i highly sui able o ou se ing. Finally, in Sec ion
4.4, we discuss he assump ions and limi a ions o bo h es ima o s, add essing issues
such as sel -selec ion among e ugees, es ing o he pa allel ends, ex ending he
DiD es ima o by ma ching on p e- ends, es ing o sys ema ic associa ions be ween
p e-2022 economic condi ions and 2022 ea men in ensi ies ac oss dis ic s, and
examining he ‘no an icipa ion’ and ‘no sel -selec ion among locals’ assump ions.
4.1 De ining he ea men a iables
To e alua e he impac o he e ugee in lux, we i s iden i y Czech dis ic s ha
expe ienced signi ican inc eases in Uk ainian employmen in 2022 due o he in e-
g a ion o e ugees in o he labou ma ke . The yea 2021 is chosen as he baseline,
ep esen ing he ‘no mal’ employmen le els among he Uk ainian diaspo a be o e he
123
The e ec o Uk ainian e ugees on he local labou ma ke s… Page 17 o 45 30
a i al o e ugees. This baseline p o ides a s able poin o compa ison, as by 2021,
employmen le els o o eign na ionals in Czechia had e u ned o p e-COVID-19
condi ions (Czech S a is ical O ice 2023e). Employmen le els in 2022 a e hen com-
pa ed o hosein2021.T ea men s a usisassigneda hedis ic le el olocal esiden s.
Dis ic s wi h minimal o no inc ease in e ugee employmen a e conside ed con ol
(o ‘no ye ea ed’), while hose wi h signi ican g ow h in Uk ainian employmen
a e classi ied as ea ed. All dis ic s a e ega ded as ‘un ea ed’ be o e 2022 due o
he absence o Uk ainian e ugees. The ea men a iable is de ined a he dis ic
le el, as i is he mos g anula geog aphic uni a ailable in he LFSS o iden i ying
indi iduals’ places o esidence.
We ocus on he employmen su ge in 2022 o isola e he e ec s speci ically
a ibu able o he employmen o Uk ainian e ugees, dis inc om p io in lows o
Uk ainian mig an s in o Czechia. This dis inc ion is essen ial as he demog aphic p o-
ile o he e ugees—p edominan ly highe -educa ed and emale—di e s signi ican ly
om ha o he ypical, less-educa ed, male Uk ainian mig an s who a i ed be o e
2022 (Czech S a is ical O ice 2021). Combining hese g oups would obscu e hese
demog aphic di e ences and complica e he iden i ica ion s a egy. Addi ionally, we
chose o analyse he ealised employmen o Uk ainian e ugees a he han he o e all
demog aphic shock o he Czech labou ma ke , which would include all Uk ainians
o wo king age ega dless o employmen s a us. This app oach is jus i ied as da a on
e ugees’ esidence migh be a ec ed by indi iduals e u ning o Uk aine, eloca ing
o o he coun ies wi hou de egis e ing, o un epo ed s ays. Howe e , he legally
manda ed o icial employmen igu es o e a highe deg ee o accu acy.13
Fo he numbe o employed Uk ainians in Czechia, we ely on agg ega ed dis ic -
le el da a sou ced om he Minis y o Labou and Social A ai s (2023a). Since
o eign employmen da a in Czechia is epo ed by ci izenship wi hou dis inguishing
be ween isa o esidence pe mi ypes, we in e , as a bes app oxima ion, ha he
subs an ial employmen su ges om he i s qua e o 2022 onwa ds p ima ily e lec
he in lux o newly a i ed Uk ainian e ugees. This es ima e may sligh ly o e s a e
e ugee employmen i some diaspo a membe s ( e-)en e ed he wo k o ce. Howe e ,
gi en he high employmen a e (99%) among he Uk ainian diaspo a wi h esidence
pe mi s as o 31 Decembe 2021 (Czech S a is ical O ice 2024b), i is likely ha he
majo i y o he inc ease e lec s e ugee employmen . Con e sely, he es ima e may
unde s a e e ugee employmen i some p e iously employed Uk ainians eloca ed
wi hin Czechia, le he coun y, o exi ed he wo k o ce. Howe e , we obse e no
signi ican in e nal o ex e nal mig a ion wi hin Czechia (see Sec ion 6), sugges ing
ha ou esul s a e obus o his po en ial bias.
To accoun o he ou ine dip in o eign employmen obse ed in he 4 h qua e o
each yea —likely due o seasonal wo ke s lea ing employmen a he end o he ha -
es season—we ely on wo sepa a e benchma ks o ‘usual’ Uk ainian employmen
le els: he a e age 2021 employmen le el o Uk ainians in Czechia by dis ic (d),
as in Eq.1; and he ac ual numbe o he employmen o Uk ainians in Czechia in he
13 Any employe in Czechia is ee o hi e Uk ainian e ugees, bu he/she is obliga ed o epo i o he
local labou au ho i ies.
123
30 Page 18 o 45 A. Pos epska, A. Voloshyna
4 h qua e o 2021 by dis ic (d), as in Eq.2.
Employed Uk ainiansd,a e age in 2021 (1)
Employed Uk ainiansd,4 hqua e o 2021 (2)
Since he impac o an in lux o , say, 10,000 o eign employees may a y ac oss
dis ic s depending on he size o he local labou ma ke , we no malise he ea men
a iable ela i e o each dis ic ’s labou ma ke size, using wo measu es: he numbe
o locals employed in 2021 by dis ic (d), as in Eq.3; and he numbe o wo king-age
locals (18–65 yea s old) by dis ic (d), as in Eq.4,14
Employed Localsd,census 2021 (3)
Locals o Wo king Aged, (4)
The employmen a iable in Eq.3, sou ced om he 2021 census (Czech S a is i-
cal O ice 2021), is ancho ed o he yea 2021, hus emaining s a ic o e ime while
a ying by dis ic . Fixing his alue o 1 yea p io o he labou shock p e en s con-
amina ion o he ea men a iable by subsequen ealisa ions o ou come a iables in
2022, such as local employmen s a us, which could o he wise c ea e a eedback loop.
Employmen le els in Czechia o 2021 we e consis en wi h his o ical no ms. Despi e
a dip in employmen numbe s o 5.235 million in 2020, a ibu ed o he COVID-19
pandemic, he 2021 igu e o 5.29 million aligns wi h p e-pandemic da a om 2019
and 2018, which eco ded 5.303 million and 5.293 million, espec i ely, indica ing a
eco e y in he labou ma ke (Czech S a is ical O ice 2022a,2021). Fo mo e de ails
on local labou ma ke condi ions, see Appendix C.
Since census da a was collec ed in he i s hal o he yea — ypically epo ing
sligh ly lowe employmen le els due o seasonal pa e ns—we include wo king-age
locals (Czech S a is ical O ice 2024b) as a second p oxy o he local labou ma ke
size,asshowninEq.4, oensu e obus ness.UnlikeEq.3, hisp oxy a iesbo hby ime
and ac oss dis ic s, emaining esponsi e o demog aphic and labou ma ke shi s
wi hou being a ec ed by ou come a iables in 2022, unless signi ican mig a ion o
locals o o om e ugee-impac ed dis ic s occu ed. This possibili y was es ed in
Sec ion 6, whe e no suppo ing e idence was ound.
Acco dingly, we employ h ee a ian s o he ea men a iable, de ailed in Eqs.5,
6, and 7:
T ea men I
d, =
⎧
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎩
Employed Uk ainiansd, −Employed Uk ainiansd,a e age in 2021
Employed Localsd,census 2021 i ≥2022
0i <2022
(5)
14 To p e en double-coun ing, he numbe o o icially employed Uk ainians was sub ac ed om he o al
numbe o employed locals o Eq.3and he numbe o wo king-age Uk ainians was sub ac ed om he
o al numbe o wo king-age locals o Eq.4.
123
The e ec o Uk ainian e ugees on he local labou ma ke s… Page 19 o 45 30
and
T ea men II
d, =
⎧
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎩
Employed Uk ainiansd, −Employed Uk ainiansd,4 hqua e o 2021
Employed Localsd,census 2021 i ≥2022
0i <2022
(6)
and
T ea men III
d, =
⎧
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎩
Employed Uk ainiansd, −Employed Uk ainiansd,a e age in 2021
Locals o wo king age 18-65d,
i ≥2022
0i <2022
(7)
whe e dand index dis ic s and ime (yea : qua e ), espec i ely. Each a ian is
designed o cap u e he same phenomenon: signi ican shi s in Uk ainian employmen
le els due o he e ugee in lux du ing any qua e o 2022, ela i e o he ‘usual’
employmen le els o he Uk ainian diaspo a in 2021, no malised by he labou ma ke
size o each dis ic . We ound up he esul ing alues o he nea es in ege , making
hem disc e e, which esul s in he ea men ‘doses’. Each ‘dose’ e lec s a 1% change
in Uk ainian employmen in dis ic da ime , such ha ≥2022, ela i e o he
‘usual’le elin he baselinepe iod ,whe e ∈a e age in 2021,4 h qua e o 2021,
adjus ed o each dis ic ’s labou ma ke size.
The ea men doses de i ed om all h ee ea men speci ica ions ange om –2%
o as high as 32% and a e compa able in e ms o sign, magni ude, and iming o onse .
A subs an ial p opo ion o locals esided in dis ic s ha ecei ed a posi i e ea men
dose, p edominan ly be ween 1 and 4% as documen ed in Fig. 4. Addi ionally, a subse
o locals li ed in dis ic s ha we e i ually una ec ed by he ea men , wi h a ea -
men dose o 0%. Addi ionally, ins ances o nega i e ea men doses we e obse ed
in one o h ee dis ic s, depending on he ea men a iable speci ica ion. These neg-
a i e doses a e p ima ily a ibu able o seasonali y in employmen (as e idenced by
ea men II showing only one such dis ic compa ed o wo o h ee o ea men s I
and III) and possibly o he depa u e o male Uk ainian immig an s.
Dynamic ea men ajec o ies The speci ica ion o he ‘ ea men ’ a iables esul s
in a complex design whe e ea men can u n on o o , luc ua e ac oss ime pe iods,
and commence a di e en imes in a ious dis ic s. Be o e 2022, all dis ic s a e se
o a baseline ‘ ea men ’ dose o ze o. F om 2022 onwa ds, dis ic s exhibi a ying
ea men ajec o ies. Dis ic s may expe ience ea men doses ha a e ei he s ill
ze o, nega i e, posi i e, o bo h. Fo example, as depic ed in Fig. 5 o T ea men I,
B un ál emains a ze o ea men le els, se ing as ou con ol dis ic o all ou
qua e s o 2022. In con as , Blansko consis en ly ecei es a 1% posi i e ea men
dose s a ing om he 1s qua e o 2022. T ea men a ies no only in in ensi y bu
also in iming; o ins ance, P e o ’s ea men begins in he 2nd qua e o 2022,
making i a con ol (no ye ea ed) dis ic o he 1s qua e . T ea men doses can
also change o e ime, possibly e e ing back o ze o a e an ini ial change. P aha’s
123
30 Page 20 o 45 A. Pos epska, A. Voloshyna
Fig. 4 Dis ibu ion o ea men doses I, II, and III in 2022. No e: This his og am, c ea ed by he au ho s
using LFSS da a, displays he coun s o indi iduals ecei ing ea men doses I, II, and III in 2022, co e ing
he 1s o 4 h qua e s, espec i ely
ea men doses inc ease o e ime, eaching 3% by he 2nd qua e o 2022, while
Pelhˇ imo ecei ed a posi i e ea men dose o 2% in he 1s qua e o 2022 be o e
e e ing back o he baseline le el o ze o. The e a e also dis ic s like Pa dubice ha ,
a e an ini ial posi i e dose, expe ience nega i e ea men doses.
To in oduce s uc u e and acili a e iden i ica ion o he empi ical analysis, we
ca ego ise dis ic s d o each qua e as ei he ‘con ol’, ‘swi che s in’, o ‘swi che s
ou ’. We always es ima e he e ec s o ‘swi che s in’ and ‘swi che s ou ’ g oups
sepa a ely o each o he ea men a iables. Table B.2 p o ides an o e iew o
he ea men doses o T ea men I, T ea men II, and T ea men III disagg ega ed by
dis ic and ime.
Dis ic s wi h posi i e ea men doses—con ol The ‘con ol’ g oup e e s o dis-
ic sd ha , a qua e , s illha eale elo ea men equal o hebaseline (consis en ly
ze o in ou case).
Fig. 5 Visualisa ion o ea men I: ea men ajec o ies o selec ed dis ic s. No e: Va iables T ea men II
and T ea men III a e iden ical o T ea men Iby design; hence, we p o ide an example o T ea men I
only
123

The e ec o Uk ainian e ugees on he local labou ma ke s… Page 21 o 45 30
Dis ic s wi h posi i e ea men doses—swi che s in ‘Swi che s in’ e e s o dis ic s
dwhe e, a ime F, he ea men le el ei he inc eases o he i s ime om ze o o
some posi i e alue, o o he same dis ic din any subsequen pe iods , ≥F,
gi en ha he ea men le el emained g ea e han o equal o he baseline. Fo special
cases o dis ic s like Pa dubice, which ini ially expe ienced a ea men dose highe
han he baseline (posi i e) and hen subsequen ly lowe (nega i e), we only include
(d, ) be o e he ea men changes om posi i e o nega i e. Fo Pa dubice, his
means ha when es ima ing e ec s o he ‘swi che s in’, we inco po a e obse a ions
om he 1s qua e o 2022, ca ego ising hem as ‘swi che s in’, and hen exclude
all obse a ions om he ollowing qua e s. The a ionale behind hese exclusions is
ha he in e p e a ion o he weigh ed a e age o he ea men e ec s, esul ing om
bo h posi i e and nega i e ea men doses, becomes ambiguous.
Dis ic s wi h nega i e ea men —swi che s ou ‘Swi che s ou ’ a e he dis ic s
ha ha e e e expe ienced a nega i e ea men dose. Unde his speci ica ion, all
obse a ions o dis ic s like Pa dubice would be conside ed as ‘swi che s ou ’.
4.2 S a ic wo-way ixed e ec s (TWFE)
We s a he analysis by employing he ollowing s a ic wo-way ixed e ec s model:
yi,d, =α+β(T ea men Io II o III)d, +θXi,d, + i+ +i,d, ,(8)
whe e i,d, and index indi iduals, dis ic s, and ime (yea : qua e ), espec-
i ely. The dependen a iable, yi,d, , ep esen s he labou ma ke ou come o in e es
(employmen ,unemploymen ,inac i i y,andweeklyhou s wo ked).The coe icien sβ
on he T ea men I,T ea men II,o T ea men III a iablesa e o p ima yin e es .
The model accoun s o indi idual– iand ime- ixed e ec s , e ec i ely minimis-
ing con ounding isks by con olling o indi idual-speci ic (bu ime-in a ian ) and
ime-speci ic (bu indi idual-in a ian ) unobse ed ac o s, unde he assump ion o
linea addi i e e ec s (Allison 2009; Woold idge 2010).
Le e aging he de ailed indi idual-le el da a in he LFSS, ou analysis inco po a es
acomp ehensi e angeo indi idual-le elcha ac e is ics(X)suchasageca ego y(15–
19, 20–25, …, 60–65); a dummy a iable o being ma ied; a dummy a iable o
ha ingchild enyounge han15 yea s;a ca ego ical a iableindica ingeduca ionle el
(ISCED); a dummy a iable o pension o disabili y s a us; and a dummy a iable
o pa - ime employmen . Addi ionally, a ca ego ical a iable o NACE-1 indus ies
is included, bu only o he weekly hou s wo ked as a dependen a iable. We u he
es ima e he model sepa a ely by gende , educa ion le el, ype o employmen con ac
( ixed e m e sus pe manen ), and he coun y o bi h ( o eign-bo n e sus Czech-
bo n). Appendix A p o ides de ailed desc ip ions o he con ol a iables and how
hey we e cons uc ed.
123
30 Page 22 o 45 A. Pos epska, A. Voloshyna
4.3 Es ima ing he e ogeneous ea men e ec s wi h ex ended
di e ence-in-di e ences (DiD) es ima o
We s a wi h he TWFE eg ession because i is widely used in mic oeconomics
empi ical esea ch—la gely due o i s pe cei ed equi alence o he DiD es ima o . The
canonical DiD model, ea u ing only wo ime pe iods, a bina y ea men a iable,
and dis inc ea men and con ol g oups, allows o he unbiased iden i ica ion o
he a e age ea men e ec on he ea ed (ATT). In such a simple se ing, he ATT
can indeed be es ima ed using a s a ic TWFE eg ession. Howe e , he design o
ou ea men a iable complica es he se ing beyond his canonical model, as ou
ea men can u n on o o , a y ac oss ime pe iods, and commence a di e en
imes in di e en dis ic s.
To ensu e unbiased ATT es ima es in ou TWFE eg ession, one solu ion is o
imposea s ingen assump ion o cons an ea men e ec sac ossindi iduals and o e
ime. This assump ion e ec i ely p ecludes he e ogeneous ea men e ec s, an exclu-
sion which, as indica ed by ecen li e a u e, is seldom ealis ic in applied esea ch.15
Applying he es de eloped by de Chaisema in and d’Haul oeuille (2017), we exam-
ine he po en ial bias in oduced by nega i e weigh s—a signal o ea men e ec
he e ogenei y—in ou ATT es ima es. Ou indings indica e ha nega i e weigh s
indeed likely bias he ATT es ima es, pa icula ly ega ding he a iable weekly hou s
wo ked. Full esul s and a de ailed desc ip ion o he es ing p ocedu e a e p o ided
in Appendix F.
The e o e, as we canno assume cons an ea men e ec s ac oss indi iduals and
o e ime in ou con ex , we adop he he e ogenei y- obus es ima o p oposed by
de Chaisema in and D’Haul œuille (2024) as ou p ima y es ima o . This es ima o
can be seen as an ex ension o he canonical DiD app oach, bu i allows o (non-
)bina y, (non-)s agge ed ea men s and acili a es dynamic/in e - empo al ea men
e ec es ima ion, making i pa icula ly sui able o ou se ing. Unlike he TWFE
eg ession, i g oups indi iduals in a way ha a oids ‘ o bidden compa isons’— ha
is compa isons be ween indi iduals who a e bo h ea ed bu commence ea men
a di e en imes. I es ima es he ac ual- e sus-s a us-quo (AVSQ) e ec o each
ea ed indi idual, a a ian o he ATT.
Fi s ly, we es ima e indi idual e ec s o each ea ed indi idual ac oss all possible
pe iods, compa ing he e olu ion o labou ma ke ou comes be ween ea ed indi id-
uals and a con ol (o no ye ea ed) g oup, p e– and pos - ea men . Following he
app oach sugges ed by de Chaisema in and D’Haul œuille (2024), we no malise he
e ec s o help wi h hei in e p e a ion and make compa ison wi h he TWFE eg es-
sion esul s easie .16 No malisa ion is done h ough di iding he es ima ed indi idual
e ec s by he di e ence be ween he ac ual ea men dose ecei ed and he baseline
ea men le el (ze o in ou con ex ) o each pe iod. The esul is a no malised AVSQ
(nAVSQ) e ec , which is in e p e ed as he a e age o al e ec pe uni o ea men .
15 See, o example de Chaisema in and d’Haul oeuille (2020); Goodman-Bacon (2021); Imai and Kim
(2021); Sun and Ab aham (2021); de Chaisema in and d’Haul oeuille (2022); Bo usyak e al. (2024),
among o he s.
16 This is implemen ed using he S a a command ‘did_mul ipleg _dyn’. Fo de ails, see de Chaisema in
e al. (2023).
123
The e ec o Uk ainian e ugees on he local labou ma ke s… Page 23 o 45 30
We hen agg ega e hese indi idual e ec s o de i e a e age e ec s o all ea ed indi-
iduals, weigh ed by he numbe o indi iduals con ibu ing o each pe iod-speci ic
es ima e. See Appendix F o de ails o he es ima ion p ocedu e.
4.4 Assump ions and limi a ions o he TWFE and ex ended did es ima o s
4.4.1 Re ugees’ sel -selec ion pa e ns
In he absence o a andomised expe imen , mig a ion esea ch equen ly aces he
p oblem o sel -selec ion (Bo jas 1987; Abowd and F eeman 1991; Jaege 2007). In
hiscon ex ,sel -selec ionimplies ha immig an swi hahighe inhe en p obabili yo
employmen —due o speci ic skill se s o a s ong mo i a ion o wo k—may choose o
se le in dis ic s wi h obus economies and high labou demand. Uk ainian e ugees,
who ha e he eedom o se le in any dis ic wi hin Czechia, migh simila ly seek
economically h i ing a eas. The p esence o es ablished Uk ainian diaspo as and
e ugee ecep ion cen es in cen al dis ic s may u he in ensi y his non- andom
se lemen pa e n, making di ec compa isons o labou ma ke ou comes o locals
ac oss dis ic s wi h mo e e sus less employed Uk ainian e ugees po en ially biased.
Wi hou no malising ou ea men a iables by dis ic labou ma ke size, ou da a
indeed indica es clea e idence o sel -selec ion, wi h a signi ican concen a ion o
e ugees in economically h i ing egions cha ac e ised by highe GDP pe capi a,
wages, and le els o educa ional a ainmen (Czech S a is ical O ice 2023a). Fu he -
mo e, e ugee dis ibu ion ac oss dis ic s is posi i ely co ela ed wi h he p esence
o ac i e companies, la ge i ms, labou ma ke igh ness, and subs an ial Uk ainian
diaspo as, while nega i ely co ela ed wi h unemploymen a es (see Table 3).
Howe e , no malising he ea men a iables no only p o ided a mo e meaning ul
measu e o ea men in ensi y bu , mo e impo an ly, allowed us o mi iga e biases
in oduced by sel -selec ion. A e no malisa ion, he highes ea men in ensi ies
(‘doses’)shi ed owa ds smalle , less economicallydominan dis ic s—such as Cheb,
Mladá Bolesla , and Tacho —whe e Uk ainian e ugees a imes comp ised be ween
3 and 32% o he employed popula ions. Addi ionally, pai s o adjacen dis ic s wi h
simila ea men in ensi ies eme ged, ypically in ol ing a cen al dis ic and i s
su ounding a eas. Examples include Plzeˇn-Sou h and Plzeˇn-Ci y, as well as P aha-
Eas and Capi al Ci y P ague. Table B.2 p o ides an o e iew o he ea men doses
o T ea men I, T ea men II, and T ea men III disagg ega ed by dis ic and ime.
Thus, al hough la ge dis ic s like Capi al Ci y P ague and Plzeˇn-Ci y s ill me
he c i e ia o be conside ed ea ed in some qua e s o 2022, no malising ea men
by labou ma ke size unco e ed meaning ul he e ogenei y ac oss dis ic s, e ealing
subs an ial a ia ion in bo h ea ed and con ol dis ic s in (i) p e- ea men ends in
analysed labou ma ke ou comes, such as changes in employmen and unemploymen
a es and (ii) p e- ea men economic cha ac e is ics. In Sec ion 4.4.2, we le e age
poin (i) by ma ching con ol and ea ed dis ic s based on p e- ends ac oss key
a iables, while in Sec ion 4.4.3, we apply poin (ii) o examine whe he a sys ema ic,
s a is ically signi ican associa ion exis s be ween he p e-2022 economic and labou
ma ke condi ions o all 77 Czech dis ic s and hei ea men in ensi ies in 2022.
123
30 Page 24 o 45 A. Pos epska, A. Voloshyna
4.4.2 Pa allel ends assump ion
A c ucial assump ion o bo h ou es ima o s is ha he ends in he s a us-quo ou -
come, condi ional on baseline ea men , a e pa allel. Fo ou main es ima o — he
ex ended DiD—and all i s speci ica ions, we ex ensi ely es his assump ion using
placebo es ima o s as p oposed by de Chaisema in and D’Haul œuille (2024). These
placebo es ima o s eplica e he ac ual es ima o s used in ou empi ical analysis by
compa ing he ou come e olu ion o indi iduals iwho la e become ea ed wi h he
ou come e olu ion o hei espec i e ‘con ol’ indi iduals ac oss p e- ea men pe i-
ods.The a ionale is ha i endsa epa allel, he e shouldbenos a is icallysigni ican
e ec s obse ed, as ea men has no ye s a ed.
Reassu ingly, o he as majo i y o ou ex ensi e se o es ima ions—ac oss all
model speci ica ions and sub-popula ions— he placebo es s we e no signi ican . We
summa ise he esul s o his es in Sec ion 5and epo de ailed esul s o each
a iable and model speci ica ion in Tables B.3–B.5.
Ex ension I: allowing o dis inc ends ac oss ea ed and con ol dis ic s Recognis-
ing he limi a ions o placebo es ima o s in es ing he pa allel ends assump ion (Ro h
2022), as well as he cons ain s imposed by he o a ing s uc u e o ou panel, which
limi s he numbe o p e- ea men qua e s a ailable o es ing, we in oduce an addi-
ional s ep in ou empi ical analysis. No malising he ea men a iables has yielded
a he e ogeneous se o bo h ea ed and con ol dis ic s in e ms o hei p e- ea men
ends ac oss key a iables o in e es —employmen , unemploymen , inac i i y a es,
and weekly hou s wo ked. Le e aging his he e ogenei y, we ex end ou p ima y DiD
es ima o by ma ching ea ed and con ol dis ic s on hei p e-2022 ends o each
a iable sepa a ely.
To cap u e subg oup-speci ic dynamics ha agg ega ed o he dis ic le el ends
migh obscu e, we calcula e p e-2022 ends sepa a ely o he subg oups wi hin
dis ic s de ined by (i) gende and educa ion le el and (ii) gende and indus y o
employmen acco ding o NACE le el 1. Thus, by accoun ing o socio-economic ac-
o s ha may impac speci ic g oups di e en ly and allowing o dis inc ends ac oss
ea ed and con ol dis ic s, we mi iga e po en ial bias.
Using he LFSS da a om he p e- ea men pe iod (2019–2021), we es ima e sea-
sonally adjus ed ends o each subg oup wi hin all o he 77 Czech dis ic s by
eg essing each a iable o in e es on he in e cep , ime, and seasonal dummies. The
es ima ed slope se es as a p oxy o he end in each dis ic . A de ailed accoun o
he end calcula ion and ca ego isa ion p ocess is p o ided in Appendix G. Gi en he
con inuous na u e o he esul ing end a iable, we con e i in o a ca ego ical a i-
able h ough a ‘scaled disc e isa ion’ p ocess. By mul iplying he con inuous end by
a selec ed ac o and ounding o he nea es in ege , his app oach p ese es he da a’s
inhe en a iabili y and imp o es compa ibili y wi h ma ching algo i hms by a oid-
ing a bi a y cu o s and main aining ine dis inc ions wi hin he da a. Finally, using
123
The e ec o Uk ainian e ugees on he local labou ma ke s… Page 31 o 45 30
Fig. 8 Posi i e ea men doses—p obabili yo employmen by gende . No e: The igu edisplayscoe icien
es ima es, s anda d e o s, 95% le el con idence in e als, and he numbe o obse a ions ATT and nAVSQ
ac oss TWFE model in Sec ion 4.2 (columns 1–2) and he expanded DiD model in Sec ion 4.3 (columns
3–8) o ea men I, II, and III. TWFE(2) model con ols o indi idual and ime- ixed e ec s as well as
indi idual-le el cha ac e is ics. DiD(4) model con ols o indi idual and ime- ixed e ec s in addi ion o
ma ching on he (X) indi idual cha ac e is ics. Models DiD(7) and DiD(8) ma ch on bo h he (X) indi idual
cha ac e is icsand dis ic -speci ic DiD(7)o dis ic -, gende -,and educa ion-speci ic DiD(8)p e- ea men
ends o he dependen a iable. Robus s anda d e o s a e clus e ed a he dis ic le el. Fo columns (4–
8), p e- end placebo es s we e conduc ed, and he p- alues we e calcula ed using he s anda d no mal
dis ibu ion. Signi icance le els: * p<0.1; ** p<0.05, *** p<0.01
he es ima ed ac ual- e sus-s a us-quo (AVSQ) e ec s o he DiD in o no malised
ac ual- e sus-s a us-quo (nAVSQ) e ec s epo ed he ein—by di iding he a e age
es ima ed e ec s by he ea men dose in each ea ed pe iod— he nAVSQ alues
u n ou o be la ge .
Fu he examina ion by educa ion le el (shown in Figu es B.5 and B.6) ound no
signi ican e ec s, indica ing ha a 1% ise in o icially employed Uk ainians does
no a ec he unemploymen o inac i i y a es o Czech men and women.
5.2 P obabili y o employmen
Nex , we conside employmen among local wo ke s. Simila ly o unemploymen ,
we ind no consis en ly signi ican e ec s o he in lux o Uk aininan e ugees and
employmen p obabili ies o local wo ke s (Fig. 8). T ea men III showed a weak (bu
posi i e) s a is ically signi ican e ec o emales in he DiD(3-7) models. Howe e ,
wi hou co obo a ion om o he speci ica ions, we e ain om d awing signi ican
conclusions om his esul . Tes ing he pa allel ends assump ion wi h placebo es i-
ma o s yielded a e age p- alues o 0.3 o bo h gende s, de ailed in Tables B.4 and B.5.
Thissuppo s ou DiD assump ions and alida es ou me hodology.Fu he analysis by
educa ion le el ( epo ed in Figu e B.7) did no e eal any signi ican hidden impac s,
123

30 Page 32 o 45 A. Pos epska, A. Voloshyna
sugges ing ha a 1% inc ease in o icially employed Uk ainians has no no iceable
sho - e m e ec on he a e age employmen p obabili y o Czech locals.
When examining sec o s mos impac ed by he e ugee in lux, he end among
emales emains consis en wi h p e ious esul s (Figu e B.9).18 Fo local male wo k-
e s, he DiD(4) es ima o indica ed a po en ial dec ease in employmen p obabili y.
A sec o -speci ic analysis indica es ha his e ec is mainly d i en by a small
(−0.003), bu s a is ically signi ican ( o ea men I only) educ ion in employmen
p obabili y wi hin he manu ac u ing sec o (Figu e B.10). Mo eo e , o emales,
he e is a educ ion in employmen p obabili y in accommoda ion and ood se -
ice ac i i ies (−0.001 o −0.0005) and adminis a i e and suppo se ice ac i i ies
(−0.0005 o −0.0001). Howe e , he esul s a e somewha inconsis en in signi icance
ac oss ea men s and es ima o s. When we examine he da a indus y by indus y, he
numbe o obse a ions who a e employed and ha e epo ed weekly hou s wo ked
wi hin indus ies becomes ela i ely small: ~6.5 housand in he manu ac u ing indus-
y, ~2.7 housand in accommoda ion and ood se ice ac i i ies, and ~1.9 housand
in adminis a i e and suppo se ice ac i i ies. Gi en he smalle sample sizes and
a ia ion in ea men s, his inconsis ency is no su p ising.
No ably, a ound one- hi d o male and one- hi d o emale e ugees ha e ound
employmen in he manu ac u ing sec o . Addi ionally, one- hi d o emale e ugees
ha e en e ed adminis a i e and suppo se ice ac i i ies. This pa e n sugges s ha ,
a leas in he sho un, wo ke s in he mos a ec ed sec o s may ha e aced di ec
compe i ion om Uk ainian e ugees. Howe e , cau ion is wa an ed when d awing
conclusions due o he small sample sizes.
To analyse he mos ulne able pa o he local popula ion, we ocused on males
and emales wi h no- o-basic and seconda y educa ion le els who we e employed
on empo a y con ac s, as hese posi ions a e ypically less secu e. Figu e9 epo s
ha emales wi h no- o-basic educa ion expe ienced small bu signi ican inc eases in
employmen p obabili y acco ding o ou DiD es ima o s. In con as , males wi h no-
o-basic educa ion aced a sligh dec ease in employmen p obabili y unde ea men
II. Fo hose wi h seconda y educa ion, bo h gende s showed minimal changes, wi h
no consis en signi ican e ec s obse ed.
5.3 Weekly hou s wo ked
Ou esul s along he ex ensi e ma gin do no show an economically signi ican e ec
o Uk ainian e ugees on local wo ke s. Nex , we u n o he in ensi e ma gin and
conside weekly wo king hou s. Figu e10 summa ises he esul s and shows a small
bu s a is ically signi ican posi i e co ela ion be ween he ea men doses and he
weekly hou s wo ked by bo h Czech women and men, implying ha a 1% inc ease in
o icially employed Uk ainians ela i e o he local labou ma ke size o each dis ic
has a sho - e m posi i e e ec on he weekly hou s wo ked by locals.
18 The sec o s wi h he highes e ugee employmen a e N-adminis a i e and suppo se ice ac i i ies
(~30%), C-manu ac u ing (~29%), H- anspo a ion and s o age (~7%), I-accommoda ion and ood se ice
ac i i ies (~7%), G-wholesale and e ail ade (~6%), and F-cons uc ion (~6%).
123
The e ec o Uk ainian e ugees on he local labou ma ke s… Page 33 o 45 30
Fig. 9 Posi i e ea men doses—p obabili y o employmen wi h empo a y con ac by gende and educa-
ion le el. No e: The igu e displays coe icien es ima es, s anda d e o s, 95% le el con idence in e als,
and he numbe o obse a ions ATT and nAVSQ ac oss TWFE model in Sec ion 4.2 (columns 1–2) and
he expanded DiD model in Sec ion 4.3 (columns 3–8) o ea men I, II, and III. TWFE(2) model con-
ols o indi idual and ime- ixed e ec s as well as indi idual-le el cha ac e is ics. DiD(4) model con ols
o indi idual and ime- ixed e ec s in addi ion o ma ching on he (X) indi idual cha ac e is ics. Models
DiD(7) and DiD(8) ma ch on bo h he (X) indi idual cha ac e is ics and dis ic -speci ic DiD(7) o dis ic -,
gende -, and educa ion-speci ic DiD(8) p e- ea men ends o he dependen a iable. Robus s anda d
e o s a e clus e ed a he dis ic le el. Fo columns (4–8), p e- end placebo es s we e conduc ed, and
he p- alues we e calcula ed using he s anda d no mal dis ibu ion. Signi icance le els: * p<0.1; **
p<0.05, *** p<0.01
While o emales, he coe icien s a e signi ican ac oss all models, ein o cing
he eliabili y o he obse ed ea men e ec s, o male wo ke s, he signi icance
in DiD models diminishes a e con olling o indi idual cha ac e is ics, dis ic -
speci ic ac o s, and p e- ea men ends speci ic o dis ic , gende , educa ion, o
indus y. This sugges s ha he ini ial ea men e ec s o males may be con ounded
by p e-exis ing ends, cas ing doub on he ea men ’s ac ual impac on hem. E en
hough he ex ensions o he baseline DiD(3) model come a he cos o a loss in
obse a ions— anging om 0.7 o 65.4% o emales and 0.4 o 62.5% o males,
especially no able in he ex ensions whe e ma ching on indi idual cha ac e is ics and
on p e- ea men ends a e applied— hey subs an ially bols e he obus ness o ou
indings. Pa allel end es s wi h placebo es ima o s yield an a e age p- alue o 0.6
o emales and 0.5 o males o all DiD es ima es, as de ailed in Table B.3.
The economic signi icance o he iden i ied e ec s can be be e unde s ood by
examining he ela i e inc eases agains he backd op o he a e age wo king hou s in
2021. Fo males, he analysis based on he TWFE(2) model sugges s a sligh inc ease
in weekly hou s wo ked, anging om 0.07 o 0.14%, compa ed o he a e age o
40.5h wo ked he p e ious yea . This is equi alen o an addi ional 0.03 o 0.06h
(o app oxima ely 1.8 o 3.6min) pe week. Fo emales, he es ima ed inc ease is
ma ginally highe , anging om 0.05 o 0.18% ela i e o hei a e age wo kweek
o 38.0h in 2021, which ansla es o an addi ional 0.02 o 0.07h (o oughly 1.2 o
4.2min) weekly.
123
30 Page 34 o 45 A. Pos epska, A. Voloshyna
Fig. 10 Posi i e ea men doses—weekly hou s wo ked by gende . No e: The igu e displays coe icien
es ima es, s anda d e o s, 95% le el con idence in e als, and he numbe o obse a ions ATT and nAVSQ
ac oss TWFE model in Sec ion 4.2 (columns 1–2) and he expanded DiD model in Sec ion 4.3 (columns
3–10) o ea men I, II, and III. TWFE(2) model con ols o indi idual and ime- ixed e ec s as well as
indi idual-le el cha ac e is ics. DiD(4) model con ols o indi idual and ime- ixed e ec s in addi ion o
ma ching on he (X) indi idual cha ac e is ics. Models DiD(7), DiD(8), and DiD(10) ma ch on bo h he (X)
indi idual cha ac e is ics and dis ic -speci ic DiD(7), o dis ic -, gende -, and educa ion-speci ic DiD(8),
o dis ic - and indus y-speci ic DiD(10) p e- ea men ends o he dependen a iable. Robus s anda d
e o s a e clus e ed a he dis ic le el. Fo columns (4–10), p e- end placebo es s we e conduc ed, and
he p- alues we e calcula ed using he s anda d no mal dis ibu ion. Signi icance le els: * p<0.1; **
p<0.05, *** p<0.01
The p e e ed DiD(4) model speci ica ion es ima es e eal a mo e sizable e ec o
men and women. Men expe ience an inc ease in weekly wo k hou s by 0.22 o 0.54%,
co esponding o an inc ease o 0.09 o 0.22 h (o 5.4 o 13.2min). Fo emales, he
impac is a 0.37 o 0.45% ise ela i e o hei usual wo k hou s in 2021, leading o
0.14 o 0.17 addi ional hou s (o 8.4 o 10.2min) pe week. In e es ingly, in oducing
he ex ended DiD(7, 8, 10) models u he magni ies he e ec es ima ed o emales,
indica ing a 0.32% up o a 0.62% inc ease in weekly hou s wo ked.19
Indi idually, hese inc eases amoun o a ela i ely modes change in weekly wo k-
ing hou s. The la ge e ec s obse ed in he DiD model, compa ed o he TWFE,
can be a ibu ed o he DiD model’s abili y o cap u e dynamic ea men e ec s o e
ime. In agg ega e e ms, hough, e en small pe cen age inc eases in a e age weekly
hou s wo ked can accumula e o a subs an ial impac ac oss he wo k o ce. These
19 I is in e es ing o no e ha TWFE consis en ly epo s la ge coe icien s han DiD. This di e ence
a ises because TWFE inco po a es all he da a a ailable be o e he ea men om 2019 o 2021. This
pe iod includes he yea 2020 and some imes also he i s qua e o 2021, when he e was a sligh dec ease
in hou s wo ked, employmen a e, and pa icipa ion a e. By including his da a, he esul ing coe icien s
o ea men e ec s a e lowe han hose de i ed using DiD, which only conside s he single pe iod be o e
he ea men begins. I migh also esul om he po en ial bias in ou ATT es ima es due o he in luence
o nega i e weigh s on he ea men e ec s (non-con ex combina ion o he e ec s) ha we es ed o in
Appendix F. Ou es s e ealed ha , in ou case, he TWFE model indeed appea s o be a ec ed by his
issue.
123
The e ec o Uk ainian e ugees on he local labou ma ke s… Page 35 o 45 30
Fig. 11 Posi i e ea men doses—weekly hou s wo ked by gende and educa ion le el. No e: The igu e
displays coe icien es ima es, s anda d e o s, 95% le el con idence in e als, and he numbe o obse a-
ions ATT and nAVSQ ac oss TWFE model in Sec ion 4.2 (columns 1–2) and he expanded DiD model in
Sec ion 4.3 (columns 3–10) o ea men I, II, and III. TWFE(2) model con ols o indi idual and ime-
ixed e ec s as well as indi idual-le el cha ac e is ics. DiD(4) model con ols o indi idual and ime- ixed
e ec s in addi ion o ma ching on he (X) indi idual cha ac e is ics. Models DiD(7), DiD(8), and DiD(10)
ma ch on bo h he (X) indi idual cha ac e is ics and dis ic -speci ic DiD(7), o dis ic -, gende -, and
educa ion-speci ic DiD(8), o dis ic - and indus y-speci ic DiD(10) p e- ea men ends o he dependen
a iable. Robus s anda d e o s a e clus e ed a he dis ic le el. Fo columns (4–10), p e- end placebo
es s we e conduc ed, and he p- alues we e calcula ed using he s anda d no mal dis ibu ion. Signi icance
le els: * p<0.1; ** p<0.05, *** p<0.01
inc emen al changes a he indi idual le el may sugges a non- i ial enhancemen
in o e all labou supply (along he in ensi e ma gin), po en ially e lec ing shi s in
labou ma ke dynamics and p oduc i i y.
Fu he analysis, disagg ega ed by educa ion le el as depic ed in Fig. 11, e eals
ha locals wi h seconda y educa ion a e p ima ily d i ing he gains in weekly hou s
wo ked.Consis en wi h hep e iousdiscussion, esul s ac ossallmodel speci ica ions
a e mo e obus o emales, wi h coe icien s emaining signi ican in bo h TWFE
and ex ended DiD models and inc easing in magni ude compa ed o p io esul s. The
es ima ed inc ease in hou s wo ked anges om 0.03 o 0.08h (1.8 o 4.8min weekly)
o TWFE(2) o 0.15 o 0.19h (9 o 11.4min pe week) o DiD(4). Wi h he DiD(7,
8, 10) ex ensions, hese igu es g ow u he o 0.15 o 0.26h o 9 o 15.6min weekly.
The esul s a e less conclusi e o male wo ke s, echoing p e ious indings.
The concen a ion o posi i e e ec s wi hin he seconda y educa ion b acke likely
e lec s he na u e o he jobs Uk ainian e ugees a e aking o he speci ic demands
o he Czech labou ma ke . Indi iduals wi h seconda y educa ion may occupy oles
ha complemen he posi ions illed by he incoming wo k o ce, leading o an inc ease
in hei hou s due o ei he inc eased demand o collabou a i e oppo uni ies. These
123
30 Page 36 o 45 A. Pos epska, A. Voloshyna
obse ed e ec s can also be di ec ly linked o he p o ile o he Uk ainian e ugee
popula ion. Mos e ugees a e emale, many o whom a e highly educa ed bu ace
language ba ie s and un amilia i y wi h he Czech labou ma ke . These challenges
may lead hem o accep jobs equi ing seconda y o lowe le els o educa ion.
Las ly, o asce ain whe he he obse ed inc ease in wo king hou s ac oss he gen-
e alpopula ionwasdi ec lya ibu able o e ugeesen e ing he wo k o ce,we ocused
ou analysis on Czech locals employed in sec o s mos impac ed by he e ugee in lux.
Consis en ly wi h p e ious esul s, we ind consis en pa e ns among local emale
wo ke s and no o local male wo ke s. Ou indings, epo ed in Figu e B.11, indi-
ca e ha hese sec o s expe ienced signi ican inc eases in wo king hou s, pa icula ly
among emales, ac oss all models. This end unde sco es he sec o -speci ic impac
o he e ugee in lux, closely aligned wi h indus ies adi ionally domina ed by, o
mo e adap able o, emale employmen . The p edominance o emale e ugees, cou-
pled wi h hei linguis ic challenges and un amilia i y wi h he Czech labou ma ke ,
has likely d i en his end, as hese indi iduals ypically ind employmen in sec o s
acing labou sho ages o hose mo e open o new wo k o ce en an s.
5.4 Fo eign-bo n indi iduals in Czechia
So a , we ha e ocused on he e ec s on all local wo ke s in he Czech Republic.
Howe e , ou analysis sugges s (in line wi h he li e a u e) ha he in lux o e ugees
has a he e ogenous e ec on he local labou o ce, sugges ing ha ce ain wo ke s
migh be mo e ulne able o he in lux (such as wo ke s in mos a ec ed sec o s o
ha ma ch he demog aphics o he incoming wo ke s). One po en ially ulne able
g oup is o eign-bo n wo ke s al eady esiding in he hos coun y. We es ima e he
models o o eign-bo n wo ke s only, and we ind some e idence suppo ing his
conjec u e. As summa ised in Figu e B.12, while he TWFE(2) model iden i ies no
signi ican e ec s, he DiD (3, 4) models indica e a sligh dec ease in employmen
p obabili y o o eign-bo n indi iduals, om −0.009 o −0.014, and an inc ease in
unemploymen p obabili y, om 0.011 o 0.016, ollowing a 1% inc ease in o icially
employed Uk ainians.
These esul s sugges ha he in lux o Uk ainian e ugees nega i ely a ec ed he
employmen a es and posi i ely he unemploymen a es o o eign-bo n esiden s
in he sho e m. Howe e , gi en ha o eign-bo n indi iduals ep esen only abou
0.04% o ou da ase , hese indings mus be app oached wi h cau ion as he limi ed
sample size es ic s he s eng h o ou conclusions and unde sco es he impo ance
o conduc ing u he esea ch wi h a mo e ex ensi e da ase .
6 Robus ness checks
6.1 Seconda y e ec s: local popula ion mo emen s
As is o en obse ed in he li e a u e co e ing simila e ugee o mig a ion e en s,
he e ec s on locals can be dis o ed by seconda y ac o s, pa icula ly due o he
123

The e ec o Uk ainian e ugees on he local labou ma ke s… Page 37 o 45 30
Table 6 Analysis o ne mig a ion and popula ion s abili y in (un-) ea ed dis ic s
Va iables Ne mig a ion Ne mig a ion, emales
(1) (2) (3) (4) (5) (6)
A e age ea men I 30.59 11.74
(54.97) (24.27)
A e age ea men II 33.71 14.91
(66.59) (29.38)
A e age ea men III 69.49 21.93
(109.73) (48.48)
No. o obse a ions 77 77 77 77 77 77
No. o dis ic s 77 77 77 77 77 77
No e: The able p esen s coe icien es ima es and he co esponding s anda d e o s in pa en heses. The
dependen a iables a e ne mig a ion in Czechia epo ed in columns (1–3) and ne mig a ion among
emales in Czechia epo ed in columns (4–6). Da a sou ced om he Czech S a is ical O ice (2023d).
Signi icance le els: * p<0.1; ** p<0.05, *** p<0.01
po en ial mo emen o locals away om he mos a ec ed dis ic s. Al hough he
sho ime ame o ou analysis limi s he ex en o his issue, we es o his conce n
by eg essing ne mig a ion igu es (change be ween 2021 and 2022) by dis ic on
he A e age T ea men I,II,o III
d. As shown in Table 6, all coe icien s a e posi i e
and no s a is ically signi ican , sugges ing ha dis ic s ha e expe ienced s able ne
mig a ion simila o ha obse ed in p e ious yea s, including he 2021–2022 pe iod.
The e o e, we ind no conclusi e e idence o abno mal popula ion mo emen in o o
ou o he ea ed dis ic s.
Rega ding he subs an ialmo emen o he Uk ainiandiaspo a in andou o Czechia
in 2022, e idence indica es ha people did no mo e away. Re ugees unde empo a y
p o ec ion a e, by de ini ion, empo a y esiden s, and i is qui e di icul o hem o
ob ain pe manen isas wi hin less han a yea . The e o e, i is unlikely ha many
ha e swi ched o pe manen s a us du ing his pe iod. The numbe o Uk ainians wi h
pe manen esidence in Czechia ac ually inc eased sligh ly om 90,776 on Decembe
31, 2021, o 93,545 on Decembe 31, 2022, sugges ing ha hose al eady esiding
pe manen ly likely s ayed.
Meanwhile, he numbe o Uk ainians holding empo a y isas ose d ama ically
om 106,099 in 2021 o 542,737 in 2022—an inc ease o 436,638 (Minis y o he
In e io 2023a). As o Decembe 31, 2022, he e we e 433,071 egis e ed e ugees
(Minis y o he In e io 2023b), u he indica ing ha he local Uk ainian popula ion
did no mo e away bu was joined by a signi ican numbe o new a i als. This sug-
ges s ha ei he a small numbe o Uk ainian e ugees managed o swi ch o pe manen
esidence (app oxima ely 2769 indi iduals), o he e was an in lux o Uk ainians who
a i ed in 2022 bu did no egis e as e ugees and ins ead egis e ed as pe manen
o empo a y esiden s (app oxima ely 6336 indi iduals)—a ela i ely small pe cen -
age compa ed o he 433,071 e ugees. Impo an ly, hese indi iduals should also be
conside ed simila o he e ugees, as hey all a i ed du ing he c isis and sha ed he
123
30 Page 38 o 45 A. Pos epska, A. Voloshyna
same legal igh s ega ding access o educa ion, employmen , heal hca e, and o he
se ices.
6.2 Al e na i e ea men a iable: wo king-age Uk ainians
To ensu e he obus ness o ou indings, we conduc an addi ional analysis using an
al e na i e ea men a iable: he numbe o wo king-age Uk ainians, ega dless o
employmen s a us, ac oss egions and ime. We compa e dis ic s wi h ea lie a i als
o Uk ainians o hose wi h la e a i als.
We de ine he al e na i e ea men a iable as
Al e na i e T ea men d, =⎧
⎨
⎩
Re ugees o wo king age 18-65d,
Locals o wo king age 18-65d,
i ≥2022
0i <2022
(10)
whe e dand index dis ic s and ime (yea : qua e ), espec i ely.
Implemen ing a TWFE model based on his al e na i e ea men helps e i y
whe he heobse ede ec s emainconsis en when he ea men a iableisexpanded
beyond employmen o include he b oade wo king-age popula ion. Un o una ely,
we canno inco po a e ou ex ended DiD es ima o , as he demog aphic shock was
so signi ican ha no dis ic could easonably be conside ed un ea ed. In oducing
a andom cu -o o signi y una ec ed a eas would be a bi a y. Ins ead o ma ching
dis ic s on hei p e- ea men ends, as done in ou main speci ica ion using he
ex ended DiD model, we con ol o local labou ma ke cha ac e is ics (unemploy-
men a e and labou ma ke igh ness) a he dis ic le el, lagged by 1 yea . The
esul ing TWFE eg ession model is speci ied as
yi,d, =α+β(Al e na i e T ea men )d, +θXi,d, +Zd, −4+ i+ +i,d, ,(11)
whe e i,d, and index indi iduals, dis ic s, and ime (yea : qua e ), espec-
i ely. The dependen a iable, yi,d, , ep esen s he labou ma ke ou come o in e es
(employmen , unemploymen , inac i i y, and weekly hou s wo ked). The coe icien β
on he Al e nam i eT ea men a iable is o p ima y in e es . The model accoun s
o indi idual– iand ime- ixed e ec s , indi idual-le el cha ac e is ics (X)used
in p e ious es ima ions, as well as labou ma ke condi ions (Z), p oxied by he unem-
ploymen a e and labou ma ke igh ness lagged by ou pe iods (1 yea ).
Tables B.6 and B.7 summa ise he esul s o local males and emales ac oss di e -
en educa ion le els sepa a ely. The esul s align wi h ou main indings: he e appea s
o be a co esponding inc ease in weekly hou s wo ked, wi h no consis en s a is ically
signi ican nega i e e ec s iden i ied o he local popula ion.
6.3 DiD: non-no malised ea men e ec analysis by dose
Fo ou p ima y es ima o — he ex ended DiD—we add ess po en ial issues om
he no malisa ion o ea men e ec s. We ca ego ise ea ed dis ic s based on
123
The e ec o Uk ainian e ugees on he local labou ma ke s… Page 39 o 45 30
hei a e age ea men dose, using he same h esholds as hose in model 9
(A e age T ea men I,II,o III
d). We hen ecalcula e he AVSQ (non-no malised)
e ec sepa a ely o each A e age T ea men Dose—1%, 2%, 3%, and ≥4%—
wi hou no malisa ion.
Addi ionally, as he ea men pe iod p og essed om he 1s o he 4 h qua e o
2022, ewe dis ic s emained as con ols. This educ ion migh ha e hinde ed ou
abili y o iden i y signi ican e ec s, gi en he insu icien numbe o obse a ions o
se e as con ols la e in he yea . To add ess his, we designa ed dis ic s expe iencing
0 o 1% ea men as ‘con ols’, ecalcula ed he A e age T ea men Dose labelling
hem ‘adjus ed’ and e-es ima ed he esul s.
As depic ed in Figu es B.14–B.21, consis en ly wi h he esul s in he main analysis,
among emale and male wo ke s, he coe icien s o he p obabili y o employmen
emain insigni ican and he coe icien s o weekly wo king hou s ollow he same
pa e n, in e ms o sign and signi icance, al hough he magni ude is somewha smalle .
We ind some e idence o s a is ically signi ican inc ease in he p obabili ies o unem-
ploymen and inac i i y;howe e , he esul sa eno consis en h oughou hedi e en
model speci ica ions, and he economic magni ude is e y small.
7 Conclusions
In his pape , we explo e he na u al expe imen o he sudden and o ced in lux o
Uk ainian e ugees o igo ously assess he sho - un impac on he locals’ labou
ma ke ou comes in he Czech Republic. On a e age, we ind no (consis en ly) signi -
ican e ec s on employmen , unemploymen , and inac i i y p obabili ies o male and
emale wo ke s. Mo eo e , we ind ha , condi ional on employmen , local wo ke s
inc eased hei wo king hou s. Indi idually, he magni ude o hese e ec s is small.
Howe e , he o e all e ec on labou supply (along he in ensi e ma gin) is ce ainly
no negligible.
Ou empi ical e idence is aluable no only because i is he i s o documen he
e ec s o he mos ecen e ugee c isis in Eu ope bu also because he e a e clea
policy implica ions. We iden i y wo g oups o wo ke s, pa icula ly ulne able o he
la ge and sudden in lux o wo ke s in o he labou ma ke : wo ke s in indus ies mos ly
a ec ed by he in lux o and o eign-bo n indi iduals. We ind e idence o a dec ease
in he p obabili y o employmen and an inc ease in he p obabili y o unemploymen
o hese wo g oups. Howe e , hese esul s a e o e y small magni ude. They a e also
based on ela i ely small sample sizes, in i ing u he esea ch ocusing on EU-wide
analysis o be e cap u e he impac on mos a ec ed g oups. Fu he mo e, we belie e
ha his pape ’s esul s shed ligh on he po en ial ou comes o policies ex ending he
igh s oimmedia eaccess o helabou ma ke o e ugeewo ke s.Bydoingso, hough
s ill in he sho un, we con ibu e o he objec i e and da a-d i en body o knowledge,
p o iding insigh s in o he e ec s o e ugees’ ac i e pa icipa ion in labou ma ke s
on he local wo ke s.
The eissugges i ee idence ha he con ex o he Czech labou ma ke con ibu ed
o he null e ec s iden i ied. A he ime o he e ugees’ a i al, he Czech labou
ma ke was excep ionally igh —i had he lowes unemploymen a e in he Eu opean
123
30 Page 40 o 45 A. Pos epska, A. Voloshyna
Union. Jus a mon h be o e he e ugee in low began, he e we e 266,783 open job
acancies epo ed in he coun y—la gely ou numbe ing egis e ed job seeke s—wi h
he majo i y equi ing only basic educa ion (73%). Employmen oppo uni ies we e
abundan in sec o slike e ail ade, cons uc ion,public adminis a ion,andeduca ion,
which we e al eady acing labou sho ages. These sec o s became he main employe s
o mos Uk ainian e ugees, who o en ook on low-skilled oles such as p oduc and
equipmen assemble s, helpe s in cons uc ion, p oduc ion, and anspo , o s a iona y
machine ope a o s, he eby helping o alle ia e he wo k o ce gap.
The combina ion o a igh labou ma ke and exis ing sho ages in key indus ies
likely mi iga ed po en ial dis up ions om he in lux o e ugees, enabling he Czech
economy o abso b he new wo k o ce ela i ely smoo hly, a leas in he sho e m.
This sugges s ha he Czech expe ience may no be di ec ly applicable o coun ies
wi h di e en labou ma ke condi ions. This con ex aises impo an conside a ions
o policy design. The la gely neu al ou comes in Czechia could, he e o e, suppo
a gumen s o policies ha acili a e e ugees o immig an s in illing exis ing labou
sho ages, pe haps h ough a ge ed ma ching o hei skills wi h ma ke needs, such
as h ough skilled mig a ion isas, a he han endo sing un es ic ed labou access in
all con ex s.
Supplemen a y In o ma ion The online e sion con ains supplemen a y ma e ial a ailable a h ps://doi.
o g/10.1007/s00148-025-01080-9.
Acknowledgemen s We a e pa icula ly g a e ul o Rob Alessie, Viola Angelini, Ge a d an den Be g,
edi o Klaus F. Zimme mann, and h ee anonymous e e ees o hei sugges ions on his p ojec . We
would also like o hank he a endees o he Eu opean Associa ion o Labou Economis s Con e ence, he
IZA Wo kshop on Labou Ma ke s and Inno a ion du ing Times o Wa and Recons uc ion, he IUSSP
Wo kshop on Popula ion and Con lic , he IFO Wo kshop on Re u n and In eg a ion P ospec s o Uk ainian
Re ugees, Da id Ma golis and he pa icipan s o he Mic oeconome ics and Policy E alua ion cou se
a he Pa is School o Economics, Jan S uhle and he pa icipan s o he Labou Ma ke Adjus men o
Economic Shocks cou se a he Uni e si y o Be gen, and he pa icipan s o he G oningen Uni e si y PhD
semina se ies o hei aluable inpu . All e o s and omissions a e he sole esponsibili y o he au ho s.
Da a A ailabili y The p ima y indi idual-le el da a used in his s udy (LFSS) a e no publicly a ailable bu
can be accessed h ough app op ia e p ocedu es om he Czech S a is ical O ice.
Decla a ions
Con lic o In e es The au ho s decla e no compe ing in e es s.
Open Access Thisa icleislicensedunde aC ea i eCommonsA ibu ion4.0In e na ionalLicense,which
pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as you gi e
app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons licence,
and indica e i changes we e made. The images o o he hi d pa y ma e ial in his a icle a e included
in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line o he ma e ial. I
ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended use is no pe mi ed
by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he
copy igh holde . To iew a copy o his licence, isi h p://c ea i ecommons.o g/licenses/by/4.0/.
123