Ede , Ch is oph
A icle
Displacemen and educa ion o he nex gene a ion:
E idence om Bosnia and He zego ina
IZA Jou nal o Labo & De elopmen
P o ided in Coope a ion wi h:
IZA – Ins i u e o Labo Economics
Sugges ed Ci a ion: Ede , Ch is oph (2014) : Displacemen and educa ion o he nex gene a ion:
E idence om Bosnia and He zego ina, IZA Jou nal o Labo & De elopmen , ISSN 2193-9020,
Sp inge , Heidelbe g, Vol. 3, pp. 1-24,
h ps://doi.o g/10.1186/2193-9020-3-12
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/152387
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/
Ede IZA Jou nal o Labo & De elopmen
ORIGINAL ARTICLE Open Access
Displacemen and educa ion o he nex
gene a ion: e idence om Bosnia and
He zego ina
Ch is oph Ede
Co espondence:
[email p o ec ed]
Depa men o Economics,
Johannes Keple Uni e si y Linz,
Al enbe ge S . 69, 4040 Linz,
Aus ia
Abs ac
In his pape , I s udy he e ec o displacemen (in he sense o o ced mig a ion) o
pa en s du ing a iolen con lic on in es men in hei child ens’ educa ion yea s la e .
Using he e hnic di ision du ing he Bosnian Wa as a na u al expe imen , I plausibly
iden i y exogenously displaced households and compa e hem o households who did
no ha e o mo e because o he wa . Displaced pa en s spend be ween 20 and 30%
less on he educa ion o hei child en in p ima y and seconda y school. The esul also
holds o single expendi u e posi ions like ex books, school ma e ials and annual
ui ion in seconda y school. A numbe o obus ness checks and nea es -neighbo
ma ching is pe o med o con i m he inding. A decomposi ion o he causal e ec
shows ha di e ences in income and he s ock o du able goods can a mos explain
one hi d o he inding. Po en ial explana ions o he educed spending o displaced
pa en s on educa ion include al e ed p e e ences h ough he exposu e o iolence,
inc eased unce ain y abou he u u e, and inancial cons ain s.
JEL Classi ica ion: I25; J15; O15
Keywo ds: Displacemen ; Con lic ; Educa ion
1 In oduc ion
Violen con lic is a egula phenomenon in he de eloping wo ld and i s long- e m
consequences can be subs an ial h ough he des uc ion o human and physical capi-
al, damage o in as uc u e, and o ced popula ion mo emen s. Howe e , esea ch on
he economic consequences o wa s and o he iolen e en s ha e only ecen ly gained
momen um as mic o-le el da a om con lic a eas become a ailable. This is an impo -
an line o esea ch as i indica es o policy make s he di ec consequences o iolen
con lic s and which issues need o be deal wi h once a con lic is o e .
The indings in his li e a u e o e in e es ing insigh s in o peoples’ li es in wa - o n
coun ies and on he e ec s o exposu e o con lic . Howe e , o ced mig a ion has seen
ela i ely li le a en ion, e en hough he UNHCR coun s 10.4 million e ugees and 36
million people o conce n in 2009 in hei Global Repo (UNHCR 2009). In his pape I
s udy he e ec s o displacemen o pa en s on educa ional expendi u e on hei child en.1
I use he e hnic di ision du ing he Bosnian Wa be ween 1992 and 1995 as a na u al
expe imen o displacemen . The wa di ided he p e iously e hnically mixed Bosnia and
© Ede ; licensee Sp inge . This is an Open Access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion
License (h p://c ea i ecommons.o g/licenses/by/4.0), which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any
medium, p o ided he o iginal wo k is p ope ly c edi ed.
2014, 3:12
h p://www.izajold.com/con en /3/1/12
2014
Ede IZA Jou nal o Labo & De elopmen Page 2 o 24
He zego ina in o wo pa s and h ough “e hnic cleansing” esul ed in wo a eas o homo-
geneous e hnic make-up. The esul ing displacemen o pa o he popula ion enables
me o unco e i s causal e ec s. The iden i ica ion s a egy ci cum en s he p oblem o
endogenously mig a ed households by using only households who mo ed ac oss he on
line du ing he wa . No signi ican di e ence in ime in a ian household cha ac e is ics
suppo he assump ion ha ea men was exogenous.
The sho ime span be ween he end o he wa and he collec ion o he da a se does
no allow me o e alua e educa ional ou comes o child en o displaced pa en s. Howe e ,
he educa ion sys em o Bosnia and He zego ina equi es pa en s o p o ide ex books,
uni o ms, school ma e ials e c. o hei child en, which gi es me he chance o look a
inpu s in educa ional p oduc ion. In pa icula , I am in e es ed in pa en s’ educa ion
expendi u e on hei child en.
I ind ha displaced pa en s spend signi ican ly less on he educa ion o hei child en
i e yea s a e he end o he wa han compa able households ha we e no displaced.
The es ima es o he educ ion in spending on educa ion ange om 20 o 30%. Con-
side ing ha he a e age household in Bosnia and He zego ina spends mo e han hal a
mon hly household income on he educa ion o a child pe yea , his is quan i a i ely a
la ge di e ence. The inding is obus o a numbe o speci ica ions and nea es -neighbo
ma ching es ima es con i m he magni ude. Displaced pa en s also spend signi ican ly
less on single expendi u e posi ions like annual ui ion in seconda y school, ex books,
and o he school ma e ials.
Recen expe imen al e idence sugges s ha s uden s’ es sco es eac posi i ely o addi-
ional educa ional inpu s wi hin a gi en ins i u ional se ing (Das e al. 2013), con i ming
an in ui ion ha has been challenged in he pas .2As small an in es men as $3 pe
s uden spen mos ly on child’s s a iona y, class oom ma e ials, and p ac ice books can
inc ease es sco es by 0.1 s anda d de ia ions in India (Das e al. 2013). As Hanushek
(2003) poin s ou , he e is e idence ha inc eased school quali y has posi i e e ec s on
labo ma ke ou comes and p oduc i i y.
Explo ing he causal channels h ough which displacemen in luences educa ion expen-
di u e, I show ha a mos one hi d can be explained by di e ences in income and weal h
le els. The employmen s a us o pa en s also ails o explain a majo pa o he di e -
ence, so he e is no e idence ha displaced pa en s subs i u e school inpu s o inc eased
pa en al e o as ound by Hou en ille and Smi h Conway (2008). Po en ial explana ions
o he educed spending o displaced pa en s on educa ion include al e ed p e e ences
h ough exposu e o iolence, inc eased unce ain y abou he u u e, and igh e inancial
cons ain s.
The pape is o ganized as ollows: The nex sec ion e iews he li e a u e, Sec ion 3
discusses he backg ound o he s udy in Bosnia and He zego ina, as well as he da a and
he iden i ica ion s a egy, while Sec ion 4 p esen s he main esul s. Sec ion 5 discusses
some channels h ough which displacemen wo ks and Sec ion 6 concludes.
2 Li e a u e e iew
The a ailabili y o mic o-da a om a la ge numbe o de eloping coun ies has g ea ly
inc eased he possibili ies o esea che s o in es iga e he e ec s o iolen con lic s a he
indi idual and household le el. A la ge numbe o pape s look a he e ec s o exposu e o
iolence on school a ainmen (eg. Ak esh and de Walque 2008; Swee 2009; Shemyakina
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 3 o 24
2011; Chama bagwala and Mo án 2009; Me ouche 2006; Rod iguez and Sanchez 2009).
Mos o hese s udies ind a nega i e e ec o a leas a subg oup o he popula ion
e en a e he end o he iolen spell. Jus ino (2011) su eys his li e a u e. Swee (2009)
analyzes he case o Bosnia and He zego ina and inds lowe comple ion a es in sec-
onda y school h ough exposu e o a iolen con lic . Howe e , he a gues he e ec is
d i en by you h soldie ing and should no ha e long- un consequences. Since displace-
men is a di ec consequence o po en ial exposu e o iolence, my indings con ibu e
o he unde s anding o why educa ional p oduc ion is a ec ed e en a e he end o he
iolence.
The des uc ion o schools du ing pe iods o iolence o en ails o explain he educ-
ion in educa ional a ainmen as ha would a ec boys and gi ls in he same way, which
is o en no he case (Shemyakina 2011). Swee (2009) a gues ha you h solde ing has p e-
en ed olde males om a ending school, while Shemyakina (2011) explains ha gi ls
we e kep a home o a oid sexual assaul s and ha assmen on hei way o school. León
(2012) inds e idence ha he dea h o eache s and he heal h s a us o pa en s could be
he causal mechanism.
In a ecen addi ion o he li e a u e Pi o a o a and Swee (2012) sugges ha all
hese indings could be d i en by “selec ion in o ic imiza ion”, ie. wa ime displacemen
changes he composi ion o he popula ion.3Fo he case o Nepal, Pi o a o a and Swee
(2012) show wi h panel da a ha he p oblem is empi ically ele an .
Exposu e o iolence also has impac s on peoples’ p e e ences. Bellows and Miguel
(2009) ind ha i inc eases ci il pa icipa ion, while Voo s e al. (2012) ind ha p e -
e ences owa ds isk, he discoun a e, and al uis ic beha io owa ds neighbo s a e
a ec ed.
The e ec s o displacemen ha e also been s udied in a numbe o pape s. Fiala (2009)
inds ha displaced households ha e in gene al ewe asse s and lowe consump ion qual-
i y. Only he p e iously poo es a e ma ginally be e o a e displacemen . Lowe asse
holdings o displaced households a e also con i med by Rahim e al. (2013). Kondylis
(2010) shows ha displacemen makes men mo e likely o be unemployed and women
mo elikely obeou o helabo o cein hecaseo BosniaandHe zego ina.
In he long- un Sa imäki e al. (2009) ind posi i e e ec s on income o displaced
ag icul u al wo ke s in Finland, which is con i med by Baue e al. (2013) o Ge many.
Howe e , he la e ind lowe income o he es o he displaced e en o he second
gene a ion. The posi i e long- un e ec s o displacemen ound in Sa imäki e al. (2009)
and Baue e al. (2013) a e likely due o inc eased sec o and geog aphic mobili y o he
displaced popula ion. Wi h he excep ion o Baue e al. (2013), he abo e men ioned
con ibu ions ocus on he e ec on he gene a ion o he displaced, howe e he nega i e
consequences o displacemen do no need o end he e.
3 Con ex , da a, and iden i ica ion
3.1 The Bosnian wa
Bosnia and He zego ina became independen in he Fall o 1991 a e he b eakdown
o o me Yugosla ia. The h ee majo e hnic g oups, Bosniaks, Se bs, and C oa s, we e
s uggling o gain powe and e en ually he con lic u ned iolen in Ap il 1992. Ini ially
all h ee e hnic g oups we e igh ing each o he , wi h he Se bs in con ol o he a my
o he o me Yugosla ia (Silbe and Li le 1996, p.222). In Feb ua y 1994, C oa s and
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 4 o 24
Bosniaks eached a peace ag eemen and e en ually joined o ces agains he Se bs. Wi h
ai suppo om he NATO, Bosniaks and C oa s we e able o push he Se bs back and
egain con ol o la ge a eas. The Bosnian Wa ended in Decembe 1995 wi h he Day on
Ag eemen , acco ding o which Bosnia and He zego ina was di ided in o wo en i ies
along he on line a he end o he wa : he Fede a ion o Bosnia and He zego ina and
he Republika S pska. Du ing he wa abou 100,000 - 110,000 people we e killed and an
es ima ed 1.3 - 1.8 mio. we e displaced ( he o al popula ion in 1991, he yea o he las
o icial census, was 4.38 mio.).
The ex ensi e displacemen du ing he wa led o mos ly e hnically homogeneous geo-
g aphic a eas. A e he wa mos Bosniaks and C oa s li ed in he Fede a ion o Bosnia
and He zego ina and mos Se bs in he Republika S pska. Those wo en i ies a e like
sepa a e s a es wi h hei own adminis a ion and hey cu en ly coope a e only in a ew
a eas. As Dahlman and Tua hail (2005) show wi h wo examples, in e hnically mixed
municipali ies ha we e spli be ween he wo en i ies, he e hnic g oups we e almos
comple ely seg ega ed by 1998 along he en i y bo de s, which was no he case be o e
he wa . Howe e , he Day on Ag eemen decla es ha all e ugees and displaced pe sons
ha e he igh o e u n o hei homes o o igin and ge hei p ope y e u ned (Tua hail
and O’Loughlin 2009, p.1047). The e u n mig a ion peaked in he yea s 2002-03,
bu a la ge ac ion o he displaced popula ion does no plan o e u n o hei p e-wa
homes. Also he majo i y o he p ope ies o displaced households in hei home munic-
ipali ies was occupied by o he s, des oyed, o damaged (Tua hail and O’Loughlin 2009,
p.1051).
3.2 The educa ion sys em in Bosnia and He zego ina
The educa ion sys em in Bosnia and He zego ina aces many challenges and changes
hese days. As o 2001, p ima y educa ion las s o 8 yea s, whe e du ing he i s 4
yea s he en i e ma e ial is augh by one eache and in g ades 5-8 each subjec has i s
own eache . Seconda y educa ion is di ided in o oca ional aining and gymnasium
(mo e academically o ien ed), whe e cu icula a e augh in 3-5 yea p og ams (UNESCO
1996; UNESCO 1997). In gene al, p ima y schools ha e no annual ui ion, bu ex books,
school ma e ials, e c. s ill need o be payed o by he pa en s. Only ew municipali ies
ensu e ha ex books o disad an aged a e p o ided (OECD 2006). Low o non-exis en
incomes, mig a ion, and di icul pos -wa condi ions a e common easons why pa en s
a e unable o be “ac i e pa en s” and ail o p o ide school equipmen , supplies, and
ex books (UNESCO 1996).
The pos -wa inancial si ua ion o schools in Bosnia and He zego ina was con-
s ained, o say he leas , as his pa ag aph om a epo abou he educa ion sys em in
he Republika S pska illus a es (UNESCO 1997, p.ii.):
“P ima y educa ion is, in heo y, ee, and is inanced om he go e nmen budge . In
p ac ice he go e nmen is o en unable o pay sala ies, and school epai s ha e o en
become he esponsibili y o he municipal au ho i ies. A he seconda y le el he
cen al go e nmen is expec ed o pay he sala ies o pe sonnel, and he municipali y
all o he cha ges. I was epo ed ha in Decembe 1996 public sec o employees,
including eache s, had no been paid o 4 o 5 mon hs. The educa ion sys em oday
is la gely dependen on inancial sac i ices made by eache s and amilies. Tex books,
2014, 3:12
h p://www.izajold.com/con en /3/1/12

Ede IZA Jou nal o Labo & De elopmen Page 5 o 24
o example, a e ex emely expensi e: an a e age p ima y school ex cos ing DM 1-3.4
[DM = Con e ible Ma k] and a seconda y one as much as DM 7”.
Repo s show ha he cu iculum o p ima y school is designed o a child equipped
wi h 10 ex books pe g ade. Fo mos pa en s ha seems ha dly a o dable, gi en ha a
quali ied eache ea ned in 1996 only 120 Con e ible Ma k pe mon h (UNESCO 1996)
and unemploymen is high.4In he yea s ollowing he publica ion o hese epo s, some
e o ms conce ning he cu iculum ook place and in 2004 p ima y school was ex ended
o 9 yea s (Swee 2010). In e na ional aid has ce ainly imp o ed some issues, bu i is
unclea i his educed he pa en s’ inancial bu den o ha ing child en in school. A p ojec
epo on he educa ion e o m in Bosnia and He zego ina by he Eu opean Union om
2008 obse es (EU 2008): “The educa ion e o m p ocess e ol es a an une en and slow
pace”.
3.3 Da a
Fo his s udy I use household su ey da a om he “Li ing S anda d Measu emen
Su ey” (hence o h LSMS) (S a e Agency o S a is ics o BiH e al. 2001) o Bosnia and
He zego ina. The da a collec ion s a ed in 2001 in 25 municipali ies wi h abou 5,400
households. F om 2002-2004 abou hal o he households we e ein e iewed each yea
o o m a 4-yea panel da a se . The LSMS co e s a wide ange o opics. The di e -
en sec ions ask ques ions abou housing, educa ion, heal h, labo , c edi , mig a ion, and
social assis ance. The e a e also sec ions on consump ion, household businesses, and ag i-
cul u al ac i i ies. Fo mos o his pape , I use he c oss-sec ional da a om 2001. I do
his o wo easons: Fi s , he sample size is educed in he panel da a o hal he numbe
o households, and second, he 2002-2004 in e iews co e a limi ed numbe o opics. I
will go in o mo e de ail abou da a issues in he espec i e sec ions.
3.4 Iden i ica ion
In o de o se iously es ima e an e ec o displacemen on household beha io , he ea -
men o displacemen mus be a andom, ie. ex an e, households in bo h, ea men and
con ol g oup, do no di e in hei cha ac e is ics, nei he obse able no unobse able.
Gi en ha displacemen is a o m o mig a ion, his is a s ic equi emen o sa is y. In
his sec ion I a gue ha he Bosnian Wa p o ides a a e, al hough sad, oppo uni y o
s udy his by-p oduc o iolen con lic . I hink o displacemen as a e sion o mig a ion,
whe e he ins inc o sel -p ese a ion domina es all economic conside a ions. The single
mos impo an push- ac o is su i al, while pull- ac o s o ce ain des ina ions do no
pose a p oblem o he es ima ion o a causal e ec , as municipali y ixed e ec s a e always
included in he eg essions.
My iden i ica ion s a egy, in a nu shell, is ha in Bosnia and He zego ina e hnici ies
we e mixed be o e he wa . The Bosnian Wa in oduced a line o di ision along which
wo e hnically homogeneous e i o ies eme ged. The on line was no d awn by eco-
nomic mo i es and people did no so hemsel es in o displacemen bu we e o ced
by hei ins inc o sel -p ese a ion. The andom cou se o he line o di ision and he
absence o sel -selec ion allow me o iden i y he e ec o displacemen on household
beha io by compa ing households ha c ossed he on line o households ha did no
mo e du ing he wa .
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 6 o 24
A he ime when Bosnia and He zego ina was pa o o me Yugosla ia, he pop-
ula ion was a mix o Bosniaks, Se bs, and C oa s in mos municipali ies. The e hnic
con lic in Bosnia and He zego ina be ween 1992 and 1995 caused many people o lea e
hei home and ake e uge on he o he side o he on line. Table 1 desc ibes his
homogeniza ion o he wo en i ies o Bosnia and He zego ina. The sha e o Se bs in
he Fede a ion o Bosnia and He zego ina sh ank o 2.3% om 17.6%, while he sha e o
Bosniaks (C oa s) wen down o 2.2% om 28.1% (1% om 9.2%). Se bs we e lea ing om
he Bosnian/C oa ian side o he on line o he Se b side and Bosniaks/C oa s he o he
way ound.
Du ing he wa , Bosniaks and C oa s in he Se b e i o y we e a isk o being killed,
wha became o be known as “e hnic cleansing”. A main goal o Se b o ces was o c ea e
an e hnically homogeneous e i o y wi hin Bosnia and He zego ina. Se bs beyond he
on line aced a simila a e and we e abandoning whole illages wi hin a ew days (Silbe
and Li le 1996, p.358). E en a e he Day on Ag eemen was signed, displacemen did
no come o a hal . Se e al illages in he Fede a ion o Bosnia and He zego ina and sub-
u bs o Sa aje o a e epo ed o ha e been abandoned a e he local Se bs ealized hey
we e apped in Bosniak e i o y (Silbe and Li le 1996, p.30). Thus displacemen du ing
and a e he wa p oduced o he mos pa e hnically homogeneous egions.
O cou se displacemen s a us is no a ques ion in he LSMS and has o be in e ed
om he da a. The da a include he municipali y o esidence igh be o e he wa and he
loca ion o esidence in 2001. When a pe son did no mo e a all, he/she en e s he con ol
g oups o non-mo e s. I a pe son esided on he o he side o he on line be o e he
wa han he/she does in 2001, he pe son is conside ed displaced. People who changed
hei loca ion o esidence bu did no c oss he on line o people who e u ned o hei
p e-wa esidence a e d opped. A household en e s ei he he con ol o ea men g oup
when bo h, he household head and his/he spouse, did no mo e o go displaced.
No e ha he iden i ica ion s a egy implies he e hnici y o a pe son. A pe son li ing
in he cu en Republika S pska be o e he wa and is now li ing in Fede a ion o Bosnia
and He zego ina, he pe son is conside ed o be a displaced Bosniak. Con e sely, a pe -
son, who has mo ed om he Fede a ion o Bosnia and He zego ina be o e he wa o
Table 1 Main e hnic g oups in Bosnia and He zego ina
Bosnia and He zego ina (O e all)
Bosniaks Se bs C oa s
1991 43.5% 31.2% 17.4%
1996 46.1% 37.9% 14.6%
Fede a ion o Bosnia and He zego ina
Bosniaks Se bs C oa s
1991 52.3% 17.6% 21.9%
1996 72.5% 2.3% 22.8%
Republika S pska
Bosniaks Se bs C oa s
1991 28.1% 55.4% 9.2%
1996 2.2% 96.8% 1.0%
Sou ce: O icial census in 1991 and uno icial census conduc ed by he UN in 1996.
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 7 o 24
he Republika S pska a e he wa , is conside ed o be a displaced Se b. Howe e , in o -
ma ion abou he e hnici y o a pe son is no a ailable in he i s wa e o he da a se ,
bu only in he smalle second wa e. Hence, I can es his p edic ion o he e hnici y by
he mig a ion pa e n only on his smalle sample. Table 2 epo s he e hnici ies o he
g oups o non-mo e s and displaced people. Ou o 1,040 displaced indi iduals, he e is
one C oa and no Se b in he Fede a ion o Bosnia and He zego ina and one Bosniak and
one C oa in he Republika S pska. The e hnici y mix in he sample o non-mo e s is no
as clea -cu , which p obably o igina es in he p esence o encla es in bo h, he Republika
S pska and he Fede a ion o Bosnia and He zego ina. Howe e , hese encla es do no
pose a h ea o my iden i ica ion, because hese people did no go “ ea ed” by o ced
displacemen , nei he did hey endogenously decide o mig a e. The e idence in Table 2
is a s ong a gumen in suppo o my iden i ica ion s a egy.5
The i s main h ea o he iden i ying assump ions would be a sys ema ic cou se o he
on line so ha economically di e en a eas we e di ided. I economically unde eloped
a eas we e a ge ed and conque ed du ing he wa , he displaced households we e poo e
han hei non-mo e coun e pa s al eady be o e he wa . Kondylis (2010, p.241 ) dis-
cusses his possibili y and concludes ha he wa was “de e mined mo e by geo-s a egic
mo i es a he han economic mo i es”. The Se b in asion ollowed he goal o connec
he Se b s onghold a ound Banja Luka wi h he Se b na ion. In addi ion, Kondylis (2010)
p o ides e idence ha p e-wa educa ional a ainmen is unco ela ed wi h wa casual ies
(and hence de eloped a eas we e no mo e con es ed) and he e is no clea connec ion
be ween he p e-wa e hnic mix and wa in ensi y.
The second key assump ion implies ha , ex an e, displaced households we e no di -
e en om non-mo e households. One possibili y is ha households o a ce ain ype
could ha e mo ed in o a eas, whe e hey we e especially exposed o he isk o displace-
men . This includes, o ins ance, a Bosniak amily mo ed o Banja Luka (now he capi al
o he Republika S pska) be o e he wa so he household head could ake a good posi-
ion he e. The da a does no sugges e idence o so ing be o e he wa in Bosnia and
He zego ina, as 70% o he household heads in he da a s ill li ed in hei municipali y o
bi h jus be o e he wa . Conside ing he small size o he a e age municipali y in Bosnia
and He zego ina o 373 km2, his sugges s ha households gene ally do no exhibi high
mobili y. Mo eo e , he esul s o he pape a e no sensi i e o es ic ing he sample o
households who s ill esided in hei bi h municipali y jus be o e he wa . I he esul s
Table 2 Displacemen by e hnic g oups
A e -wa en i y o esidence
Fede a ion o Bosnia
and He zego ina Republika S pska
Displaced Bosniak 285 1
Se b 0 747
C oa ian 1 1
O he 5 0
No mo ed Bosniak 2,847 59
Se b 121 2,306
C oa ian 614 22
O he 108 39
E hnici y da a a e om he second wa e o he he LSMS on Bosnia and He zego ina, which only includes hal o he
household om he i s wa e. The numbe o obse a ions is epo ed.
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 8 o 24
a e pu ely d i en by p e-wa geog aphic so ing, he indings would anish in such a
selec ed sample.
Ano he po en ial p oblem is ha households wi h ce ain cha ac e is ics we e dis-
placed, while o he s a he same loca ion and o he same e hnici y did no mo e,
and hence displacemen was no a li e-sa ing ea men as sugges ed ea lie . In e ms
o obse able cha ac e is ics displaced and non-mo e households do no di e as he
desc ip i e s a is ics in Table 3 show. While Se bs in he sample we e mo e likely o be dis-
placed, he e is no signi ican di e ence in o he household cha ac e is ics. The e is also
no di e ence in he dis ibu ions o he highes le el o educa ion achie ed and he age
o he household head using a Kolmogo o -Smi no equali y o dis ibu ions es .6The
column “Reg ession” epo s he poin es ima e o a eg ession o he household cha ac-
e is ic on he displacemen indica o and municipali y ixed e ec s. Howe e , egional
di e ences a e no a p oblem as municipali y ixed e ec s a e included in all eg essions.
Table 4 es s whe he he p e-wa household cha ac e is ics a e join ly co ela ed wi h
displacemen . In a eg ession o he displacemen indica o on he cha ac e is ics, only he
dummy a iable o he Se b e i o y is a good p edic o o displacemen . The a iables
o household and child cha ac e is ics ( he numbe o child en o he amily in school, a
dummy a iable o he oldes child o he amily, a dummy o emale child en, yea s o
educa ion and age o he household head) a e unable o signi ican ly explain displacemen .
A join es o he i e household/child cha ac e is ics can no ejec he hypo hesis ha
displacemen is a non-selec i e p ocess. The esul s do no change when municipali y
ixed e ec s a e included in column 2.
E en i household cha ac e is ics o displaced and non-mo e s a e no signi ican ly
di e en om each o he , p e-wa expe ience could s ill a ec educa ion decisions.
Al hough, each e hnici y has nominally hei own language, he e a e only mino di -
e ences in he spoken wo d (Mappes-Niediek 2009, p.30). One migh a gue ha he
Table 3 Desc ip i e s a is ics and exogenei y o displacemen
All No mo ed Displaced Di e ence Reg ession
Displaced Family 0.189 0 1
Republika S pska 0.44 0.40 0.61 0.21***
(0.037)
Numbe o Child en 1.81 1.81 1.81 0.00 0.09*
o Family in School (0.056) (0.054)
Oldes Child 0.41 0.41 0.40 -0.01 -0.01
o Family (0.023) (0.025)
Female Child 0.47 0.47 0.48 0.01 0.02
(0.028) (0.030)
Educa ion o 10.30 10.34 10.10 -0.25 -0.22
Household Head (0.279) (0.305)
Age o 45.74 45.80 45.47 -0.33 -0.87
Household Head (0.735) (0.834)
Municipali y FE Yes
Obse a ions 1,952 1,584 368 1,952 1,952
Signi icance le els: * :10% ** : 5% *** : 1%.
S anda d e o s clus e ed a he household le el in pa en hesis.
The column “Reg ession” epo s he poin es ima e o a eg ession o he household cha ac e is ic on he displacemen
indica o and municipali y ixed e ec s.
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 15 o 24
The speci ica ions in Table 9 pe o m some obus ness checks. I someone ou side
he household paid o educa ion expendi u es and ha is he eason why pa en s
spend less, he di e ence in educa ion expendi u e could be inconsequen ial o edu-
ca ional ou pu . Das e al. (2013) show ha an icipa ed public supply o addi ional
school inpu s in India and Zambia is o se by an expendi u e educ ion o pa en s. Fo -
una ely, he LSMS eco ds he expenses paid om someone ou side he household,
howe e only as he o al amoun . Including hese expendi u es and unning he base-
line eg essions wi h he new dependen a iable in column (1) educes he di e ence in
educa ion expendi u e o 23.8% bu emains highly signi ican . Res ic ing he sample o
he households wi hou any ou side unds o educa ion in column (2) shows a sligh ly
inc eased di e ence o 30.6%. These indings imply ha he educed educa ion expen-
di u e is no d i en by displaced households who a e subs i u ing o addi ional ex e nal
unds.
In column (3) dummy a iables o u al and mixed municipali ies a e in e ac ed wi h
he displacemen dummy, while he base g oup a e u ban municipali ies. The in e ac-
ion e ms p oduce posi i e, bu insigni ican poin es ima es, while he di e ence in
he base g oup is es ima ed a 37%. When he sample is es ic ed o households ha
s ill li ed in hei municipali y o bi h jus be o e he wa , one can ule ou ha some
households sel -selec ed o a highe isk o displacemen h ough mig a ion be o e he
wa and now d i e he main inding. This can be uled ou as he es ic ed sample,
ha did no mig a e be o e he wa , also spends 29.1% less on he educa ion o hei
child en.
To sum up, I ind a s ong nega i e ela ionship be ween displacemen and he spending
on he educa ion o hei child en. The esul is obus o a numbe o sample es ic ions
and he inclusion o a ious con ol a iables.
Table 9 Reg ession ou pu : educa ion expendi u e III
Log. o educa ion expendi u e
All unds O iginal O iginal O iginal
(1) (2) (3) (4)
Displaced Family -0.238*** -0.306*** -0.370*** -0.291***
(0.0723) (0.0721) (0.0971) (0.0830)
Displaced Family ×0.151
Ru al Municipali y (0.1646)
Displaced Family ×0.183
Mixed Municipali y (0.1563)
No Funds om Ou side o Household Yes
Ne e Mo ed be o e Wa Yes
Con ols Yes Yes Yes Yes
Municipali y FE Yes Yes Yes Yes
No. Obse a ions 1,947 1,827 1,952 1,454
R-squa ed 0.170 0.202 0.202 0.184
S anda d e o s clus e ed a he household le el in pa en hesis.
Signi icance le els: * : 10% ** : 5% *** : 1%.
Con ol a iables include numbe o child en in HH en olled in school, educa ion o HHH, age o HHH and dummy a iables
o being he oldes child and emales. “All Funds” includes educa ion expendi u es paid by someone om ou side he
household. Column (2) es ic s he sample o households ha do no ecei e unds o educa ion om ou side he
household. Column (4) es ic s he sample o households who s ill li ed in hei municipali y o bi h be o e he wa .
2014, 3:12
h p://www.izajold.com/con en /3/1/12

Ede IZA Jou nal o Labo & De elopmen Page 16 o 24
5 Channels
So a he pape ocused on es ima ing he o al causal e ec o displacemen on educa ion
expendi u e, bu has no o e ed a channel h ough which his e ec migh wo k.
Howe e , i is o g ea in e es o esea che s and policy make s alike o unde s and he
mechanisms h ough which displacemen ope a es. I i s s a wi h some econome ic
conside a ions o guide he analysis.
5.1 Econome ic conside a ions
The es ima ion o he channel h ough which displacemen wo ks aces wo issues: Fi s ,
including a a iable o a channel in oduces an endogenous a iable and hence OLS
p oduces biased es ima es. And second, he p ope coun e ac ual o an e alua ion o a
channel changes he in e p e a ion o he es ima ion esul s.
5.1.1 E alua ing he indi ec e ec
The ollowing wo equa ions desc ibe he e ec o displacemen h ough an in e media y
a iable mi(eg. income) on educa ion expendi u e. Municipali y ixed e ec s a e omi ed
o simplici y.
yi=δ1di+Xiβ1+ρ1mi+θ1midi+ε1i(2)
mi=δ2di+Xiβ2+ε2i(3)
In hese equa ions, displacemen has a di ec e ec δ1on educa ion expendi u e,
changes he in e media y a iable h ough δ2, and migh in luence he e ec o he in e -
media y a iable on educa ion expendi u e h ough θ1. Simply including he in e media y
a iable and i s in e ac ion wi h displacemen in oduces an endogenei y p oblem as
Co (ε1i,ε2i)migh be nonze o. In he case o Co (ε1i,ε2i)= 0, he OLS es ima es in
equa ion (2) will be biased (Ang is and Pischke 2009, p.64). Howe e , he bias can be
signed: Using income as an example, i is mos likely ha Co (ε1i,ε2i)>0, as high
income households a e mo e likely o alue educa ion mo e. This causes he es ima e o ρ1
o be biased upwa ds, ie. he indi ec channel o displacemen -on-income-on-educa ion
expendi u e cap u es oo much o he o al e ec and hence esul s in a downwa d bias
o δ1.
5.1.2 Coun e ac ual
The second p oblem in hese eg essions conce ns he p ope coun e ac ual. The ques-
ion I wan o answe wi h hese eg essions is: “Would he household spend less on
educa ion e en i displacemen would no ha e changed he in e media y a iable?” In
a usual eg ession o educa ion expendi u e on a displacemen indica o , he in e medi-
a y a iable, and i s in e ac ion wi h displacemen , one has o do some calcula ions o ge
a p ecise answe o he ques ion abo e and hypo hesis es ing ge s mo e di icul . How-
e e , when he in e media y a iable is ede ined a single pa ame e deli e s a su icien
answe .
The condi ional expec a ion unc ion is (con ol a iables Xiomi ed)
Eyi|di,mi=α+δdi+ρmi+θmidi.
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 17 o 24
Rede ine he in e media y a iable by sub ac ing he a e age o he in e media y
a iable o he non-displaced households:
˜mi=mi−E[mi|di=0]
Then he expec ed di e ence in educa ion expendi u e be ween displaced and non-
mo e s, gi en no change o he in e media y a iable by displacemen is
Eyi|di=1, ˜mi=E˜mi|di=0 −Eyi|di=0=δ+(ρ +θ) ˜mi−ρ˜mi
=δ(4)
since E˜mi|di=0=0.
The e o e he ans o ma ion o he in e media y a iable allows me o es ima e he di -
e ence in educa ion expendi u e unde he assump ion o no change o he in e media y
a iable h ough displacemen wi h he single pa ame e δ. All o he es ima o s emain
unchanged because a iances and co a iances do no change by sub ac ing a cons an .
5.2 Income, he s ock o du able goods, and he housing si ua ion
One o he mos na u al explana ions o he di e ence in educa ion expendi u e is ha
displaced amilies ha e lowe income and weal h. I is no su p ising ha displaced house-
holds ha e lowe labo income and weal h, especially eal es a e, han households ha did
no ha e o mo e du ing he wa . When households ge displaced, hey usually do no
ha e ime o sell hei eal es a e, which is o en he mos aluable asse o a amily. I dis-
placed amilies en a dwelling while he non-mo e amilies ha e inhe i ed a house o li e
in, i would no be su p ising ha displaced households spend less on educa ion o hei
child en.
The desc ip i e s a is ics in Table 10 documen ha displaced households expe ience
a signi ican educ ion in income, he s ock o du able goods, and a e less likely o own
he dwelling hey li e in. To al income is lowe o displaced households, bu highe non-
labo income (pensions and allowances) pa ly coun e balance he educ ion in labo
income. Also he sha e epo ing ze o labo income o 36.1% is a lo highe han ha o
non-mo e s o 19.1%. A la ge di e ence is in he s ock o du able goods, which can be
conside ed as a p oxy o weal h as da a on inancial asse s is no a ailable.
The las h ee ows show he limi ed in o ma ion on he housing si ua ion a ailable.
The sha e o amilies ha own he dwelling hey li e in is almos eigh imes as high
o he non-mo e s han i is o he displaced. Displaced amilies mos ly en he place
hey li e in, bu a sha e o almos 16% li e in a place wi hou paying o i (ei he on
loan om iends o amilies, as illegal occupan s in abandoned houses, o in eme -
gency lodges). These di e ences a e p ime candida es o explain he la ge educ ion in
educa ion expendi u e ound ea lie .
Income and weal h is con olled o wi h se e al di e en a iables. Household labo
income measu es he sum o labo income epo ed o he las mon h by household
membe s. Household non-labo income measu es he sum o pensions and allowances
pe mon h ecei ed by household membe s, while o al household income is composed
o he sum o he wo a o emen ioned a iables. The a iable du ables is he sum o he
alues o epo ed du able goods in he household, bu no inancial asse s o eal es a e.
Dummy a iables o a epo ed alue o ze o o any o hose a iables a e included in he
eg essions o allow o mo e lexibili y. The housing si ua ion is coded in h ee indica o
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 18 o 24
Table 10 Desc ip i e s a is ics: income, du able goods, and housing
All No mo ed Displaced Di e ence
To al HH Income 481 492 437 -54**
(27.3)
To al HH Income 527 539 473 -66**
(condi ional on >0) (28.2)
Sha e wi h Ze o To al 0.086 0.088 0.075 -0.013
HH Income (0.016)
HH Labo Income 404 426 313 -113***
(27.4)
HH Labo Income 521 527 490 -37
(condi ional on >0) (33.9)
Sha e wi h Ze o 0.224 0.191 0.361 0.170***
HH Labo Income (0.024)
HH Non-Labo Income 77 66 124 58***
(8.9)
HH Non-Labo Income 188 174 229 55***
(condi ional on >0) (15.7)
Sha e wi h Ze o 0.590 0.621 0.458 -0.163***
HH Non-Labo Income (0.29)
S ock o Du able Goods 2,688 2,884 1,852 -1,032***
(244.7)
S ock o Du able Goods 2,795 3,002 1,915 -1,087***
(condi ional on >0) (251.8)
Sha e wi h Ze o 0.038 0.040 0.033 -0.006
S ock o Du able Goods (0.011)
Housing Si ua ion: 0.655 0.790 0.108 -0.682***
Own Dwelling (0.023)
Housing Si ua ion: 0.264 0.148 0.733 0.585***
Ren Dwelling (0.022)
Housing Si ua ion: 0.081 0.062 0.159 0.097***
Li e o F ee (0.016)
Obse a ions 1,901 1,541 360
S anda d e o s in pa en hesis.
Signi icance le els: * : 10% ** : 5% *** : 1%.
Income is ela es o mon hly income. All income and weal h alues a e in Con e ible Ma k.
a iables o owne ship o dwelling, en ing a dwelling, o li ing in he dwelling wi hou
paying o i .
The esul s o he eg essions a e shown in Table 11. Remembe ha he poin es ima es
o he i s ow es ima e he di e ence o he coun e ac ual o no educ ion o income
and/o du ables. Also he unexplained pa o he displacemen e ec is a lowe bound as
discussed abo e. Howe e , he di e ence in educa ion expendi u e is su p isingly obus
o he inclusion o income and du able goods a iables. The di e ence dec eases only
sligh ly and ules ou income and weal h di e ences as he main mechanism. Simila ly
he indica o a iables o he housing si ua ion ail o explain he di e ence in educa ion
expendi u e.
The dummy a iable o displacemen emains signi ican a he one-pe cen le el in
mos speci ica ions and a he i e-pe cen le el in he eg essions wi h he housing a i-
ables. The mos lexible speci ica ion wi h income and du able goods a iables in column
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 19 o 24
Table 11 Reg ession ou pu : income, du able goods, and housing
Log. o educa ion expendi u e
(1) (2) (3) (4) (5) (6) (7)
Displaced Family -0.293*** -0.233*** -0.254*** -0.257*** -0.209*** -0.316** -0.329**
(0.0676) (0.0701) (0.0683) (0.0669) (0.0686) (0.1407) (0.1384)
Log. HH To al Income10.074* 0.051
(0.0401) (0.0422)
Displaced Family ×0.035 0.029
Log. HH To al Income1(0.0788) (0.0799)
Log. HH Labo Income10.042 0.009 0.001
(0.0516) (0.0555) (0.0583)
Displaced Family ×0.076 0.072 0.094
Log. HH Labo Income1(0.1062) (0.1128) (0.1142)
Log. HH Non-Labo Inc.10.026 0.023 0.015
(0.0423) (0.0422) (0.0448)
Displaced Family ×0.006 -0.001 0.006
Log. HH Non-Labo Inc.1(0.0800) (0.0799) (0.0806)
Log. Du able Goods10.065** 0.052* 0.060** 0.052*
(0.0266) (0.0283) (0.0289) (0.0306)
Displaced Family ×0.055 0.046 0.015 0.034
Log. Du able Goods1(0.0559) (0.0568) (0.0589) (0.0604)
Own Dwelling10.161 0.160
(0.1453) (0.1502)
Displaced Family ×0.096 -0.200
Own Dwelling1(0.2789) (0.2820)
Ren Dwelling10.207 0.233
(0.1558) (0.1607)
Displaced Family ×0.093 -0.073
Ren Dwelling1(0.2361) (0.2398)
Ze o Indica o 1Yes Yes Yes Yes Yes Yes
Displaced Family ×Yes Yes Yes Yes Yes Yes
Ze o Indica o 1
F- es 7.51*** 4.37*** 5.92*** 4.18*** 2.99*** 5.61*** 2.12**
Con ols Yes Yes Yes Yes Yes Yes Yes
Municipali y FE Yes Yes Yes Yes Yes Yes Yes
No. Obse a ions 1,901 1,901 1,901 1,901 1,901 1,783 1,783
R-squa ed 0.250 0.253 0.251 0.255 0.258 0.216 0.231
S anda d e o s clus e ed a he household le el in pa en hesis.
Signi icance le els: * : 10% ** : 5% *** : 1%.
Con ol a iables include numbe o child en in HH en olled, educa ion o HHH, age o HHH and dummy a iables o being
he oldes child and emales. “F- es ” esul s om a es o join signi icance o displacemen and all i s in e ac ions.
1Measu ed as he di e ence o he mean o non-mo e s.
(5) shows a di e ence in educa ion expendi u e o 20.9% a e con olling o labo and
non-labo income and he s ock o du able goods. This is s ill wo- hi ds o he o al causal
e ec . When he housing a iables a e included, he e is ac ually a sligh inc ease om
he o al causal e ec es ima ed p e iously.
To conclude, di e ences in income, he s ock o du able goods, and he housing si -
ua ion can explain a mos one hi d o he o al e ec , bu do no seem o be he
main mechanism. The e s ill seems some hing else o be going on ha was induced by
displacemen o he pa en s.
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 20 o 24
5.3 Employmen s a us o pa en s
Ano he mechanism o explo e is whe he di e ences in he employmen s a us o he
pa en s cause he educa ion expendi u e di e ence. I bo h pa en s a e wo king, pa -
en s may no ha e he ime o help hei child en lea n and he e o e spend mo e on
books and school ma e ials o make up o less pe sonal suppo (Hou en ille and Smi h
Conway 2008). The desc ip i e s a is ics in Table 12 show ha displaced pa en s a e less
likely o be employed. This mechanism could he e o e explain he di e ence in educa ion
expendi u e.
I es his hypo hesis by including indica o a iables o employmen s a us o bo h pa -
en s (o a single pa en ), he spouse o he household head, and i no pa en is employed in
Table 13. In e ac ion e ms o he employmen indica o s wi h displacemen a e included
as well. Again he mean o he non-mo e s o each employmen a iable is sub ac ed
om he indica o a iable o in e p e he displacemen dummy as he a o emen ioned
coun e ac ual. In he h ee ollowing columns I add one dummy a iable and he in e -
ac ion e m a a ime. None o hese coe icien s is signi ican a a adi ional le el as can
be seen in Table 13.
As wi h income, he coe icien s show ha only a small po ion o he di e ence in
educa ion expendi u e can be explained by he employmen s a us o he pa en s, a mos
abou one en h in he las eg ession. The main mechanism h ough which displacemen
a ec s educa ion expendi u e is s ill unde ec ed.
5.4 P e e ences, unce ain y, and inancial cons ain s
In summa y, nei he income, he du able goods le el, he housing si ua ion no he
employmen s a us o he pa en s a e able o accoun o he majo i y o he di e -
ence in educa ion expendi u e. The na u al ques ion is hen: How does he e ec come
abou ? Un o una ely, I am no able o ully answe his ques ion. This sec ion discusses
addi ional explana ions.
A possible explana ion is ha displaced households a e able o buy cheape school ma e-
ials and ex books o sha e he supplies wi h o he amilies. Displaced households in
an a ea may build ne wo ks o help each o he . Howe e , i is ha d o imagine ha dis-
placed households ind a way o sa e on educa ion expendi u e ha households who did
no mo e du ing he wa do no ind, especially wi h hei social ne wo k in place. The
non-mo e s in Bosnia and He zego ina a e no exac ly ich o pass up a possibili y o sa e.
Voo s e al. (2012) p esen e idence om ield expe imen s in Bu undi showing ha
exposu e o iolence a ec s p e e ences. In de ail hey epo mo e al uis ic beha io
Table 12 Desc ip i e s a is ics: pa en ’s employmen s a us
All No mo ed Displaced Di e ence
Bo h Pa en s Employed 0.29 0.30 0.23 -0.07***
(0.026)
Spouse o HHH Employed 0.28 0.29 0.23 -0.06**
(0.026)
No Pa en Employed 0.34 0.30 0.48 0.17***
(0.027)
Obse a ions 1,901 1,541 360
S anda d e o s in pa en hesis.
Signi icance le els: * : 10% ** : 5% *** : 1%.
2014, 3:12
h p://www.izajold.com/con en /3/1/12

Ede IZA Jou nal o Labo & De elopmen Page 21 o 24
Table 13 Reg ession ou pu : pa en ’s employmen s a us
Log. o educa ion expendi u e
(1) (2) (3) (4)
Displaced Family -0.281*** -0.290*** -0.276*** -0.274***
(0.0687) (0.0682) (0.0696) (0.0689)
Bo h Pa en s Employed10.097 0.056
(0.0674) (0.0723)
Displaced Family ×0.165 0.203
Bo h Pa en s Employed1(0.1529) (0.1689)
Spouse Employed10.095
(0.0676)
Displaced Family ×0.062
Spouse Employed1(0.1476)
No Pa en Employed1-0.147** -0.129*
(0.0705) (0.0756)
Displaced Family ×0.017 0.099
No Pa en Employed1(0.1340) (0.1470)
F- es 9.74*** 9.34*** 8.08*** 6.14***
Con ols Yes Yes Yes Yes
Municipali y FE Yes Yes Yes Yes
No. Obse a ions 1,901 1,901 1,901 1,901
R-squa ed 0.202 0.200 0.202 0.204
S anda d e o s clus e ed a he household le el in pa en hesis.
Signi icance le els: * : 10% ** : 5% *** : 1%.
Con ol a iables include numbe o child en in HH en olled, educa ion o HHH, age o HHH and dummy a iables o being
he oldes child and emales. “F- es ” esul s om a es o join signi icance o displacemen and all i s in e ac ions.
1Measu ed as he di e ence o he mean o non-mo e s.
owa ds neighbo s, mo e isk aking, and a highe discoun a e. Fo my pu pose he
highe discoun a e is o special in e es . I displaced households we e mo e exposed o
iolence han non-mo e s, as can be expec ed, hen displaced pa en s could ha e a highe
discoun a e han pa en s who we e no displaced. Indi iduals wi h a highe discoun
a e would in es less in p ojec s ha gene a e a payo in he u u e - such as educa ion.
Un o una ely, wi h he c ude consump ion da a a ailable, such a hypo hesis can no be
es ed se iously.
Two o he po en ial explana ions come o mind. Fi s , displaced households ace a lo o
unce ain y abou he u u e and y o p epa e hemsel es by cu ing down spending on
e e y non- i al posi ion, which includes educa ion expendi u e. In a simple wo pe iod
model, an agen wi h con ex ma ginal u ili y educes consump ion in he i s pe iod i
he isk o pe iod 2 income inc eases. Kimball (1990) calls his phenomenon p udence and
de ines i as “ he p opensi y o p epa e and o ea m onesel in he ace o unce ain y”.
In 2001 he es i u ion o p ope y o in e nally displaced households and he possibili y
o e u n o hei homes om be o e he wa was s ill an issue in Bosnia and He zego ina.
Many displaced households p obably aced a highly unce ain u u e. On he o he hand
one could a gue ha inc eased unce ain y h ough displacemen p omp s households o
in es mo e in highly mobile asse s, such as human capi al, which hey could ake wi h
hem i hey e e had o change loca ion again.
The second po en ial explana ion a e inancial cons ain s. Displaced households could
ace igh e inancial cons ain s due o he loss o hei social ne wo k o c edi o
he lack o p ope y hey could use as colla e al. Un o una ely, inancial asse s a e no
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 22 o 24
included in he su ey o in e abou inancial cons ain s. E en i inancial cons ain s a e
no binding now, he expec a ion ha hey will bind in he u u e would al eady induce
households cu back on expendi u es oday (Dea on 1991).
Howe e , I am no able o gi e a ull explana ion how he di e ence in educa ion expen-
di u e comes abou . A change in he discoun a e o displaced households, inc eased isk,
and inancial cons ain s a e consis en wi h he indings o his pape , bu o he channels
h ough which displacemen wo ks can no be uled ou in gene al.
6Conclusion
This pape con ibu es o he li e a u e on he consequences o iolen con lic s by look-
ing a he e ec o displacemen , in he sense o o ced mig a ion, on in es men in
educa ion. I ind a obus s a is ical ela ionship be ween he displacemen s a us o pa -
en s and he amoun spen on he educa ion o hei child en. The educ ion o educa ion
expendi u e h ough displacemen is in he ange o 20 o 30% compa ed o pa en s who
did no mo e du ing he wa . The es ima ed di e ence in educa ion expendi u e is obus
o many speci ica ions and a se ies o es s indica e ha selec ion bias is no he sou ce o
he esul . Nea es -neighbo ma ching on co a ia es inds quan i a i ely simila e ec s.
A numbe o channels o how displacemen migh a ec educa ion expendi u e a e dis-
cussed. Di e ences in income and du able goods le els can explain a mos one hi d o
he baseline esul . A lowe owne ship a e o housing among displaced households also
ails o explain he main inding. The employmen o pa en s and suppo om ou side
he household can also be uled ou as he main mechanisms.
I u he discuss ecen expe imen al indings on he e ec o exposu e o iolence on
p e e ences, he hypo hesis ha he displaced households ace mo e unce ain y abou
he u u e, o mo e igid inancial cons ain s han non-mo e s. Tha would lead hem o
cu back spending on e e y non- i al posi ion, including he educa ion o hei child en.
Mo e wo k needs o be done o ully unde s and how iolen con lic in luences peoples
li es. Resea ch has shown nega i e consequences o exposu e o iolence and displace-
men , bu how exac ly he changes in economic ou comes come abou is no ully
unde s ood ye .
Endno es
1The e m “displacemen ” e e s o o ced mig a ion and is di e en o job loss, which
he labo li e a u e o en e e s o as displacemen .
2As Das e al. (2013) show, when pa en s expec o ecei e addi ional esou ces, hey
subs i u e ou o hei own educa ional expendi u e and he e o e he o al educa ional
inpu s a ailable o s uden s do no inc ease much. This subs i u ion is likely esponsible
o he ailu e o ind posi i e e ec s in p e ious expe imen al s udies.
3A p oblem o he es ima ion o causal e ec s eme ges as he o en used
di e ence-in-di e ence app oach be ween bi h coho s o ea men and con ol
egions does no accoun o non- andom wa ime displacemen . This gene a es a
sample co ela ion be ween ou comes and con lic in ensi y ha is no he causal e ec
o exposu e o con lic .
4De ailed epo s abou condi ions a schools du ing he academic yea 2000-01 a e, o
my knowledge, no a ailable.
5C oa s a e ha dly ound as displaced people in he da a, because households o C oa s
who ound hem sel in Se b e i o y likely mo ed o he - hen newly o med - Republic
o C oa ia (IDMC 2009). I will he e o e ocus on Bosniaks and Se bs om now on.
2014, 3:12
h p://www.izajold.com/con en /3/1/12
Ede IZA Jou nal o Labo & De elopmen Page 23 o 24
6This is also ue o he la ge sample o all household heads and hei spouses in he
LSMS da a se , no jus he ones wi h child en in school ha is used he e.
Compe ing in e es s
The IZA Jou nal o Labo & De elopmen is commi ed o he IZA Guiding P inciples o Resea ch In eg i y. The au ho
decla es ha he has obse ed hese p inciples.
Acknowledgemen s
The esea ch p esen ed in his a icle was conduc ed as pa o my PhD disse a ion a Simon F ase Uni e si y. My senio
supe iso , Anke Kessle , con ibu ed by p o iding commen s on esea ch design and exposi ion, while he junio
supe iso s, Simon Woodcock and Ch is Mu is, help wi h commen s on he exposi ion o he esea ch. I also hank
Gus a o Bobonis, Alex Ka ai ano , K ishna Pendaku , and con e ence pa icipan s a he A lan ic Summe Con e ence on
Public Policy Resea ch, he De elopmen Con e ence o he Ge man Economic Associa ion, he con e ence o he
Eu opean Associa ion o Labou Economis s, semina pa icipan s a Johannes Keple Uni e si y, he edi o (Ha mu
Lehmann), and ou anonymous e e ees o help ul commen s. All emaining e o s a e mine.
Recei ed: 31 Ma ch 2014 Accep ed: 11 June 2014
Published:
Re e ences
Ak esh R, de Walque D (2008) A med Con lic and Schooling: E idence om he 1994 Rwandan Genocide. HiCN Wo king
Pape 47. h p://www.hicn.o g/wo dp ess/wp-con en /uploads/2012/06/wp47.pd . Accessed 23 Jun 2014
Ang is J, Pischke J (2009) Mos ly Ha mless Econome ics. P ince on Uni e si y P ess, P ince on NJ
Baue T, B aun S, K asnicka M (2013) The Economic In eg a ion o Fo ced Mig an s: E idence om Pos -Wa Ge many.
Econ J 123: 998–1024
Bellows J, Miguel E (2009) Wa and Local Collec i e Ac ion in Sie a Leone. J Public Econ 93: 1144–1157
Chama bagwala R, Mo án H (2009) The Human Capi al Consequences o Ci il Wa : E idence om Gua emala. HiCN
Wo king Pape : 59. h p://www.hicn.o g/wo dp ess/wp-con en /uploads/2012/06/wp59.pd . Accessed 23 Jun 2014
Das J, De con S, Habya imana J, K ishnan P, Mu alidha an K, Sunda a aman V (2013) School Inpu s, Household
Subs i u ion, and Tes Sco es. Am Econ J Appl Econ 5: 29–57
Dahlman C, Tua hail G (2005) B oken Bosnia: The localized geopoli ics o displacemen and e u n in wo Bosnian places.
Ann Assoc Am Geog aph 95: 644–662
Dea on A (1991) Sa ings and Liquidi y Cons ain s. Econome ica 59: 1221–1248
EU (2008) Ins i u ional and Capaci y Building o Bosnia and He zego ina Educa ion Sys em. h p://www.cep.edu. s/si es/
de aul / iles/Educa ion_S a egy_2008_-_2015.pd . Accessed 23 Jun 2014
Fiala N (2009) The Consequences o Fo ced Displacemen in No he n Uganda. HiCN Wo king Pape : 65. h p://www.
hicn.o g/wo dp ess/wp-con en /uploads/2012/06/wp65.pd . Accessed 23 Jun 2014
Hanushek E (2003) The Failu e o Inpu -Based Schooling Policies. Econ J 113: 64–98
Hou en ille A, Smi h Conway K (2008) Pa en al E o , School Resou ces, and S uden Achie emen . J Hum Resou 43:
437–453
IDMC (2009) A p o ile o he in e nal displacemen si ua ion: C oa ia. h p://unpan1.un.o g/in adoc/g oups/public/
documen s/un c/unpan016823.pd . Accessed 23 Jun 2014
Jus ino P (2011) Violen Con lic and Human Capi al Accumula ion. HiCN Wo king Pape : 99. h p://www.hicn.o g/
wo dp ess/wp-con en /uploads/2012/06/wp99.pd . Accessed 23 Jun 2014
Kennedy P (1981) Es ima ion wi h co ec ly in e p e ed dummy a iables in semiloga i hmic equa ions. Am Econ Re 71:
801
Kimball M (1990) P ecau iona y Sa ings in he Small and in he La ge. Econome ica 58: 53–73
Kondylis F (2010) Con lic Displacemen and Labou Ma ke Ou comes in Pos -Wa Bosnia and He zego ina. J De Econ
93: 235–248
León G (2012) Ci il con lic and Human Capi al Accumula ion: The Long Te m Consequences o Poli ical Violence in Pe ú.
J Hum Resou 47: 991–1022
Mappes-Niediek N (2009) Die E hno-Falle. Ch is oph Links Ve lag, Be lin
Me ouche O (2006) The Human Capi al Cos o Landmine Con amina ion in Cambodia. HiCN Wo king Pape : 25. h p://
www.hicn.o g/wo dp ess/wp-con en /uploads/2012/06/wp25.pd . Accessed 23 Jun 2014
MHHR (Minis y o Human Righ s and Re ugees) (2003) Compa a i e Indica o s on Re ugees, Displaced Pe sons and
Re u nees, P ope y Law Implemen a ion, and Recons uc ion in BiH om 1991 o 30 June 2003. h p://www.mh .
go .ba/PDF/LjudskaP a a/common%20co e%20documen %20BH- e -ENGLISH.pd . Accessed 23 Jun 2014
Mooney E (2005) Access o Educa ion o In e nally Displaced Child en. The B ookings Ins i u ion-Uni e si y o Be n.
h p://www.b ookings.edu/ p/p ojec s/idp/20050111_mooney.pd . Accessed 23 Jun 2014
OECD, Mission o Bosnia and He zego ina (2006) Ac ion Plan on School En olmen and Comple ion in Bosnia and
He zego ina. h p://www.oscebih.o g/documen s/osce_bih_doc_2006092510260492eng.pd . Accessed 23 Jun 2014
Pi o a o a M, Swee E (2012) Quan i ying The Mic oeconomic E ec s o Wa : How Much Can Panel Da a Help? HiCN
Wo king Pape 116. h p://www.hicn.o g/wo dp ess/wp-con en /uploads/2012/06/HiCN-WP-116.pd . Accessed 23
Jun 2014
Rahim A, Jaimo ich D, Ylönen A (2013) Fo ced displacemen and beha io al change: An empi ical s udy o e u nee
households in he Nuba Moun ains. HiCNWo king Pape 157. h p://www.hicn.o g/wo dp ess/wp-con en /uploads/
2012/06/HiCN-WP-157_1.pd . Accessed 23 Jun 2014
Rod iguez C, Sanchez F (2009) A med Con lic Exposu e, Human Capi al In es men s and Child Labo : E idence om
Colombia. HiCNWo king Pape 68. h p://www.hicn.o g/wo dp ess/wp-con en /uploads/2012/06/wp68.pd .
Accessed 23 Jun 2014
2014, 3:12
h p://www.izajold.com/con en /3/1/12
22 Jul 2014
Ede IZA Jou nal o Labo & De elopmen Page 24 o 24
Sa imäki M, Uusi alo R, Jän i M (2009) Long-Te m E ec s o Fo ced Mig a ion. IZA Discussion Pape 4003. h p:// p.iza.
o g/dp4003.pd . Accessed 23 Jun 2014
Silbe L, Li le A (1996) The Dea h o Yugosla ia. Penguin Books, London
S a e Agency o S a is ics o BiH, S a is ical Ins i u e o he Fede a ion o BiH, Republika S pska S a is ical Ins i u e and The
Wo ld Bank (2001) Li ing S anda d Measu emen Su ey - Bosnia and He zego ina. h p:// inyu l.com/bde zn .
Accessed 23 Jun 2014
Shemyakina O (2011) Educa ion and A med Con lic in Tajikis an. HiCN Wo king Pape 106. h p://www.hicn.o g/
wo dp ess/wp-con en /uploads/2012/06/wp1061.pd . Accessed 23 Jun 2014
Swee E (2009) On Wa and Schooling A ainmen : The Case o Bosnia and He zego ina. HiCN Wo king Pape 57. h p://
www.hicn.o g/wo dp ess/wp-con en /uploads/2012/06/wp57.pd . Accessed 23 Jun 2014
Swee E (2010) Toge he o Sepa a e? Pos -con lic Pa i ion, E hnic Homogeniza ion, and he P o ision o Public
Schooling. h p://indi idual.u o on o.ca/swee/e hnic11.pd . Accessed 23 Jun 2014
Tua hail G, O’Loughlin J (2009) A e E hnic Cleansing: Re u n Ou comes in Bosnia-He zego ina a Decade Beyond Wa .
Ann Assoc Am Geog aph 99: 1045–1053
UNESCO (1996) Re iew o he educa ion sys em in he Fede a ion o Bosnia and He zego ina. h p://unesdoc.unesco.
o g/images/0011/001144/114461eo.pd . Accessed 23 Jun 2013
UNESCO (1997) Re iew o he educa ion sys em in he Republika S pska - Bosnia and He zego ina. h p://unesdoc.
unesco.o g/images/0010/001092/109215eo.pd . Accessed 23 Jun 2014
UNHCR (2009) Global Repo 2009. h p://www.unhc .o g/g 09. Accessed 23 Jun 2014
an Ga de en K, Shah C (2002) Exac in e p e a ion o dummy a iables in semiloga i hmic equa ions. Econome ics J 5:
149–159
Voo s M, Nillesen E, Ve wimp P, Bul e E, Lensink R, an Soes D (2012) Violen Con lic and Beha io : A Field Expe imen in
Bu undi. Ame ican Economic Re iew 102: 941–964
Ci e his a icle as: Ede : Displacemen and educa ion o he nex gene a ion: e idence om Bosnia and
He zego ina. IZA Jou nal o Labo & De elopmen
Submi you manusc ip o a
jou nal and benefi om:
7 Con enien online submission
7 Rigo ous pee e iew
7 Immedia e publica ion on accep ance
7 Open access: a icles eely a ailable online
7 High isibili y wi hin he fi eld
7 Re aining he copy igh o you a icle
Submi you nex manusc ip a 7 sp inge open.com
2014, 3:12
h p://www.izajold.com/con en /3/1/12
10.1186/2193-9020-3-12
2014, 3:12