Coon, Michael
A icle
Remi ances and child labo in Boli ia
IZA Jou nal o Mig a ion
P o ided in Coope a ion wi h:
IZA – Ins i u e o Labo Economics
Sugges ed Ci a ion: Coon, Michael (2016) : Remi ances and child labo in Boli ia, IZA Jou nal o
Mig a ion, ISSN 2193-9039, Sp inge , Heidelbe g, Vol. 5, Iss. 1, pp. 1-26,
h ps://doi.o g/10.1186/s40176-016-0050-6
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/149448
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/
ORIGINAL ARTICLE Open Access
Remi ances and child labo in Boli ia
Michael Coon
Co espondence: [email p o ec ed]
Depa men o Economics and
Business Adminis a ion, Hood
College, 401 Rosemon A enue,
F ede ick, MD 21701, USA
Abs ac
This a icle explo es he e ec o emi ances on he p e alence and in ensi y
o child labo in Boli ia. Using household su ey da a, we ind ha child en in
emi ance- ecei ing households a e less likely o be ac i ely supplying labo ,
and he numbe o hou s wo ked pe week declines wi h he size o emi ances
ecei ed. The size o he impac a ies be ween u ban and u al households.
Remi ances o u ban households ha e a la ge impac on mo ing child en
comple ely ou o he labo o ce, bu a smalle inc ease in emi ance size is needed
o an equal educ ion in numbe o hou s wo ked in u al a eas.
JEL codes: F24, O15, E26
Keywo ds: Remi ances, Mig a ion, Child labo , Boli ia
1 In oduc ion
Remi ances o Boli ia ha e g own om US$7.4 million in 1995 o US$1.2 billion in
2013 (Wo ld Bank 2015). This emendous g ow h has led o emi ances being an
inc easingly impo an pa o he Boli ian economy. Pe sonal emi ances o Boli ia
ha e o aled as much as he equi alen o 8.04 % o GDP (see Fig. 1). Despi e he
sizable ole emi ances a e now playing in he Boli ian economy, su p isingly li le
esea ch has looked a he impac s emi ances a e ha ing in ha coun y. The ew
s udies ha do exis ha e ound e idence ha some Boli ian households a e e ec i ely
using emi ances as pa o an income di e si ica ion s a egy o o e come insu ance
and capi al ma ke impe ec ions (Alcala e al. 2014) and ha emi ances a e helping
o educe po e y in u ban a eas (Lopez-Videla and Machuca 2014). The p esen s udy
adds o he small, bu impo an , body o esea ch ela ed o Boli ian emi ances by
explo ing he ela ionship be ween emi ance eceip and he use o child labo in
Boli ia. While he ela ionship be ween emi ances and child labo has been s udied
o o he coun ies (see below), his ela ionship is o pa icula impo ance in he
Boli ian con ex , since in July o 2014 lawmake s ook he unp eceden ed s ep o
lowe ing he legal wo king age in Boli ia om he minimum age o 14 es ablished by
in e na ional labo con en ions o 12 yea s old o o mal con ac ed employmen and
10 yea s old o sel -employmen (Wa son 2014). Os ensibly, he go e nmen a gued
ha since hey we e essen ially powe less o p e en he need o child labo among
impo e ished households, he change in he law was necessa y in o de o a o d legal
p o ec ions o child en who would ha e been wo king anyway. Howe e , i is possible
ha he Boli ian go e nmen has ye o exhaus all o i s a ailable esou ces. This
s udy explo es he ela ionship be ween emi ances and child labo , and he po en ial
© 2016 Coon. Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0 In e na ional 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
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 license, and indica e i
changes we e made.
Coon IZA Jou nal o Mig a ion (2016) 5:1
DOI 10.1186/s40176-016-0050-6
o u ilizing emi ance income o educe he p e alence and in ensi y o child labo .
The esul s p esen ed below sugges ha he go e nmen may ha e gi en up oo soon
and ha emi ances migh p o e o be an e ec i e ool in he igh agains child labo
in Boli ia.
2 Backg ound
Boli ia expe ienced a d ama ic inc ease in ou -mig a ion du ing he 1980s, coinciding
wi h he pe iod o hype in la ion ha occu ed om 1984–1985. Be ween 1980 and
1990, he numbe o Boli ians li ing ab oad g ew by 36 % (see Table 1). A s agnan
economy in he 1990s only saw he a e o emig a ion g ow. Be ween 1990 and 2000,
he numbe o Boli ians li ing ab oad g ew by 60 %, hen nea ly doubled again in he
ollowing decade. By 2010 nea ly 700,000 Boli ian ci izens, 6.9 % o he popula ion,
we e li ing ou side o he coun y (Wo ld Bank 2011, 2015). Along wi h he g ow h in
mig a ion came an e en la ge g ow h in emi ance ans e s. Be ween 1995 and 2013,
annual emi ance lows g ew by app oxima ely 16,000 %, gi ing Boli ia one o
he as es emi ance g ow h a es in he wo ld. Remi ance lows g ew as es in
he 5 yea s leading up o he G ea Recession in 2008 (see Fig. 1). Al hough hey
declined in he 2 yea s ollowing he global inancial c isis, by 2013 hey we e
back abo e hei p e- ecession le el.
The pe iod co esponding o he G ea Recession also saw a d ama ic decline in chil-
d en’s schooling in Boli ia. This end, howe e , did no bounce back o p e-c isis
le els. Be ween 2006 and 2013, he numbe o ou -o -school child en inc eased om
40,640 o 169,291. Ne p ima y school en ollmen s ell om 95.57 % o 87.73 %
Fig. 1 Remi ances o Boli ia 1995–2013. Da a Sou ce: Wo ld Bank, Wo ld De elopmen Indica o s 2015
Table 1 Boli ian popula ion li ing ab oad
Yea 1960 1970 1980 1990 2000 2010
Emig an popula ion 120,536 139,100 166,734 226,404 361,475 684,998
G ow h a e (%) –15.4 19.9 35.8 59.7 89.5
Sha e o popula ion (%) 3.3 3.1 3.0 3.3 4.3 6.9
Sou ce: Au ho ’s calcula ions using da a om Global Bila e al Mig a ion Da abase (Wo ld Bank 2011) and Wo ld
De elopmen Indica o s (Wo ld Bank 2015)
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 2 o 26
(UNESCO 2015). While his d op is i sel subs an ial, i is likely ha ce ain g oups
wi hin he coun y we e a ec ed disp opo iona ely. Despi e a 1994 Educa ional Re o m
Law ha p oclaimed he go e nmen was du y-bound o o e ee uni e sal educa ion
and p o ide educa ion o indigenous popula ions in hei na i e languages, dispa i ies
in educa ional a ainmen s ill pe sis be ween ich and poo , and Spanish and
indigenous-language speake s (ILO-IPEC 2012, Reimáo and Taş2015). Fu he mo e,
ne en ollmen a es a e no necessa ily e lec i e o he ue s a e o educa ion, as hey
do no accoun o child en who a e en olled in school, bu may no a end egula ly,
o he quali y o he educa ion hey ecei e. Many child en en olled in school op no
o go because hey mus a el la ge dis ances, and hose who do go o school o en
a i e o ind schools wi h inadequa e supplies (ILO-IPEC 2012).
Al hough comple e da a a e una ailable o he co esponding pe iod, i is likely ha
he change in school a endance du ing his ime is co ela ed wi h an inc ease in child
labo ac i i y. While he as majo i y o Boli ian child en a e en olled in school, mos o
hem also spli hei ime be ween school and wo k. Despi e a i ying se e al In e na ional
Labo O ganiza ion con en ions on child labo , many child en in Boli ia a e engaged in
some o m o economic ac i i y. A 2012 epo on child labo and educa ion in Boli ia,
which analyzed da a om he 2008 Su ey o Child Labo (Encues a de T abajo In an il),
ound ha only 14.8 % o child en aged 6–13 epo ed school a endance as hei only
ac i i y. Mos child en epo ed some combina ion o schooling and wo k, ei he paid o
unpaid, wi h 1.3 % epo ing de o ing ime o bo h school and paid wo k, 61.5 % epo -
ing going o school and con ibu ing unpaid household labo , and 20.9 % epo ing a
combina ion o he h ee (ILO-IPEC 2012). This a e o labo pa icipa ion is la gely
unchanged om i s 2001 le els (Zapa a e al. 2011). Much like educa ional access, child
labo ac i i y is also disp opo iona ely dis ibu ed among ce ain g oups. Olde child en
a e mo e likely o wo k han younge child en, indigenous child en a e mo e likely o
wo k han non-indigenous child en, and child en in u al households a e mo e likely o
wo k han child en in u ban households (ILO-IPEC/INE 2010).
While many coun ies saw hei economies sh ink du ing he G ea Recession, Boli ia
did no slip in o a ecession i sel , al hough i s g ow h a e did decline somewha
(Wo ld Bank 2015), so changes in he local economy a e unlikely o explain any ob-
se ed changes in school a endance and child labo ac i i y. Bu Boli ia did see a de-
cline in emi ance income. Gi en emi ances’impo an ole in he Boli ian economy,
i is possible ha he decline in emi ance lows ha occu ed du ing his pe iod con-
ibu ed o an inc ease in child labo and, he e o e, indi ec ly o a decline in school
en ollmen . This ela ionship will ul ima ely depend on how indi idual households a e
a ec ed by emi ances, and whe he emi ances con ibu e o a educ ion in child
labo . This s udy explo es he ela ionship be ween emi ances and child labo a he
household le el o de e mine he ex en o which changes in emi ance income
con ibu e o changes in child labo .
Se e al ecen s udies ha e examined he abili y o emi ances o educe child labo .
To he ex en ha child labo is la gely d i en by ex eme po e y (Basu and Van
1998), i ollows ha i emi ances can help li households abo e some minimum
income h eshold, hen hey should help o alle ia e he need o child en o wo k.
Fu he mo e, Eps ein and Kahana (2008) p opose ha ou -mig a ion o wo king-aged
adul s can se e o aise wages in he home coun y, is-à- is a dec ease in he supply
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 3 o 26
o labo , hus educing he need o child labo by inc easing non- emi ance income as
well. Addi ionally, emi ances can mi iga e inancial shocks which migh o he wise
equi e child en o en e he wo k o ce (Beegle e al. 2003).
In gene al, he exis ing empi ical li e a u e suppo s he a gumen ha emi ances
educe he supply o child labo . In a c oss-coun y s udy o 82 coun ies, Ebeke (2010)
inds signi ican e idence ha an inc ease in emi ances as a sha e o GDP is associ-
a ed wi h a educ ion o he sha e o he 10–14-yea -old popula ion ac i ely engaged in
wo k. His s udy also inds ha emi ances ha e a s onge e ec on he educ ion o
child labo in coun ies whe e c edi cons ain s a e highe and GDP g ow h mo e
ola ile, indica ing ha emi ances help insu e agains nega i e income shocks ha
can con ibu e o he need o child en o help supplemen household income. Se e al
coun y-le el s udies also ind e idence ha emi ances can educe he need o child
labo . Acos a (2011) inds ha emi ance eceip signi ican ly educes he likelihood
ha child en will engage in paid labo in El Sal ado . Alca az e al. (2012) ind a signi i-
can inc ease in he p e alence o child labo in Mexico ollowing he decline in emi -
ances esul ing om he 2008 US inancial c isis. Dimo a e al. (2015) ind e idence
ha mig an ans e s educe he supply o child labo in Tanzania. The unde lying as-
sump ion in all o hese s udies is ha emi ances help educe liquidi y cons ain s
and inc ease household income su icien ly o educe he need o child labo .
No all s udies ind uni o mly posi i e esul s, howe e . Fo example, while Acos a
(2011) does ind ha emi ances a e associa ed wi h a dec ease in wage labo , he also
inds ha emi ances end o inc ease he p e alence o domes ic labo , sugges ing ha
emi ances simply lead o a ealloca ion o child labo om paid o unpaid wo k. Simi-
la ly, in he case o El Sal ado , Cale o e al. (2009) ind ha emi ances only educe
he incidence o paid child labo among u al and non-poo households and ha emi -
ances inc ease he incidence o domes ic labo in u ban households. These s udies
sugges ha while inc eased income is-à- is emi ances may bene i he household,
dis up ion caused by he loss o a wo king-aged household membe o mig a ion may
ha e nega i e impac s on he child en le behind. O he s udies ha e ound simila
e ec s wi h espec o in es men in child en’s human capi al. Bansak and Chezum
(2009) examine he e ec s o mig a ion and emi ances on child en’s schooling in
Nepal. Thei indings indica e ha an inc ease in ne emi ance income inc eases he
p obabili y ha young child en a e in school. Howe e , hey also ind ha inc easing
he numbe o absen ee adul s signi ican ly educes he p obabili y o being in school
o child en o all ages. Simila ly, Amuedo-Do an es e al. (2008) examine he di e ing
e ec s o emi ances when hey a e ecei ed by an immedia e membe o he house-
hold e sus when hey a e ecei ed by ex ended amily o iends. They ind ha in
ce ain Hai ian communi ies, emi ances only ha e a posi i e e ec on child en’s
schooling when hey a e ecei ed om ex ended amily o iends, a he han house-
hold membe s who ha e mig a ed. Tha is, he dis up ion caused by losing a household
membe o ou -mig a ion diminishes any bene i s gained om ecei ing emi ances.
In addi ion o con ibu ing o he body o esea ch speci ic o Boli ia, his s udy also
con ibu es o he b oade li e a u e pe aining o emi ances and child labo . Wi h
he excep ion o Dimo a e al. (2015), all o he empi ical s udies men ioned abo e
measu e he e ec s o emi ances on child labo a he ex ensi e ma gins, i.e., whe he
a child wo ks o no . While his is indeed an impo an poin , i may unde es ima e he
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 4 o 26

ull impac o emi ances. Fo ins ance, i we wish o cu ail child labo so ha chil-
d en may a end school, hen a signi ican educ ion in he numbe o hou s wo ked
may ha e a la ge impac han mo ing a child a he ma gins comple ely ou o he
labo o ce. Thus, i is impo an o look beyond he incidence o child labo and exam-
ine emi ance income’s e ec on he in ensi y o child labo as well. Simila ly, Acos a
(2011) and Alca az e al. (2012) bo h es ima e hei models wi h bina y indica o s o
emi ance eceip . This, oo, may be mis ep esen ing he ue impac o emi ances,
as emi ance size can a y widely ac oss households. Fu he mo e, om a policy s and-
poin , i is impo an o di e en ia e be ween he po en ial impac s o inc easing he
sha e o households ecei ing emi ances (i.e., encou aging mo e mig a ion) and
adop ing policies o inc ease he lows o emi ances o households ha al eady ha e a
membe ab oad. This s udy add esses hese issues by analyzing he ela ionship be ween
he size o emi ance income and he numbe o hou s wo ked by child en in he
household, in addi ion o he ela ionship be ween emi ance eceip and he
p e alence o child labo .
3 Theo e ical amewo k
The heo e ical amewo k o his analysis builds o o he model p oposed by Basu
and Van (1998), ex ended o include emi ance income in he household budge
cons ain . Pa en s maximize u ili y wi h espec o consump ion and hei child en’s
leisu e. Child en’s leisu e is conside ed a luxu y good, and only en e s he pa en ’s
u ili y unc ion once some minimum subsis ence le el has been achie ed. The pa en ’s
p e e ences a e exp essed by he S one-Gea y u ili y unc ion
uc;eðÞ¼ c−sðÞ1−eðÞi c≥s
c−si c<s
ð1Þ
whe e cis he pa en ’s consump ion, eis each child’s labo e o le el, and e∈[0, 1].
The minimum subsis ence le el, s, is a pa ame e .
The household’s budge cons ain is
nc þmβc¼mewcþnwaþ ð2Þ
whe e n(≥1) is he numbe o adul s in he household, m(≥1) is he numbe o child en
in he household, each child consumes β(< 1) sha e o wha an adul consumes, w
a
and
w
c
a e he wages
1
paid o adul s and child en, espec i ely, and is he emi ance
income.
Sol ing his model o he case whe e c≥sgi es he op imal alloca ion o child en’s
e o as
ewa;wc;m;n; ðÞ¼
0
1
i s n þmβðÞþmwc≤nwaþ
i s n þmβðÞ−mwc≥nwaþ
snþmβðÞþmwc−nwa−
2mwc
o he wise:
8
>
>
<
>
>
:ð3Þ
Fo he case o he in e io solu ion, compa a i e s a ics a e gi en by Eqs. 4 h ough
8 below.
∂e
∂wa
¼−n
2mwc
ð4Þ
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 5 o 26
∂e
∂ ¼−1
2mwc
ð5Þ
∂e
∂n¼s−wa
ðÞ
2mwc
ð6Þ
∂e
∂m¼nwaþ −ns
2wcm2ð7Þ
∂e
∂wc
¼nw
a−sðÞþ −smβ
2mw2
c
ð8Þ
Equa ions 4 and 5 show child labo dec eases wi h an inc ease in adul wages and e-
mi ance income. Howe e , emi ance eceip is mo e complex han jus a change in
non-labo household income, since mig a ion necessa ily equi es emo ing a leas
one membe om he household o go ab oad, i.e., a dec ease in n. Equa ion 6 gi es
he change in op imal child labo e o wi h espec o he numbe o adul s in he
household. This implies ha when he adul wage is less han he minimum subsis ence
le el, mig a ion will lead o a dec ease in child labo and when he adul wage is g ea e
han he minimum subsis ence le el mig a ion leads o an inc ease in child labo , be-
o e accoun ing o any change in emi ance income. Combining his wi h he ac ha
child e o is dec easing in means ha in he case whe e adul wages a e less han
he subsis ence le el o consump ion, mig a ion and emi ances unambiguously educe
child e o . Fu he mo e, a la ge enough inc ease in emi ance income ha leads o
he case whe e s(n+mβ)+mw
c
≤mw
a
+nw
a
+ yields he co ne solu ion o e=0,
he eby mo ing a child comple ely ou o he labo ma ke . On he o he hand, when
adul wages a e g ea e han he subsis ence le el o consump ion, mig a ion and emi -
ances educe child e o only i >(w
a
−s). Tha is, i emi ances a e insu icien ly
small, such ha hey do no o se he loss o he adul wage due o mig a ion, hen
mig a ion can inc ease child labo . In ui i ely, we would expec ha a a ional, u ili y
maximizing household would no send a membe ab oad in his case unless hey
expec ed emi ance income o o se he loss o domes ic income. Howe e , since he
ealized income may di e om he expec ed income, i is possible ha mig a ion may
indeed lead o an inc ease in child labo e o . Thus, ex an e i is unce ain whe he
emi ances will educe child labo on a e age.
Equa ion 7 gi es he change in child labo e o wi h espec o an inc ease in
he numbe o child en. Child labo is inc easing in he numbe o child en when
sn <nw
a
+ . Tha is, in households whe e he adul income exceeds subsis ence, inc easing
he numbe o child en inc eases child labo e o . While his may seem coun e in ui i e a
i s glance, ecall om Eq. 3, i sn +smβ+wm
c
≤nw
a
+ , hen we each he co ne solu ion
whe e e= 0. As such, he esul o m Eq. 7 only holds i sn <nw
a
+ <sn +smβ+mw
c
.This
would imply ha al hough mo e child en migh necessi a e an inc ease in child labo ,
his need can be o se by an inc ease in income. The e o e, we can po en ially expec
o see bo h child labo and emi ances inc ease wi h espec o he numbe o
child en in he household.
Finally, Eq. 8 indica es ha he e a e bo h income and subs i u ion e ec s de e mining
he child labo supply. When o al child subsis ence, smβ, exceeds adul su plus income,
(w
a
−s)n+ , child labo is dec easing wi h an inc ease in he child wage. Thus, when less
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 6 o 26
labo is needed o mee he subsis ence need, less labo is supplied. Howe e , when adul
su plus income exceeds he child subsis ence le el, child labo is inc easing in child wages.
Thus, child en can be en iced back in o he wo k o ce by inc easing he wage. Howe e , his
esul also only holds o he special case whe e sn +smβ<nw
a
+ <sn +smb +mw
c
,
o he wise we each he co ne solu ion o no child labo e o .
4 Empi ical s a egy
This s udy add esses wo gene al empi ical ques ions. Fi s , a e child en in households
ha ecei e emi ances mo e o less likely o wo k? Second, how does he size o he
emi ances ecei ed by he households a ec he numbe o hou s wo ked by child en?
The i s ques ion is add essed by es ima ing he e ec s o emi ance eceip on child
labo a he ex ensi e ma gins. Tha is, emi ance eceip and labo supply a e mea-
su ed using indica o a iables equal o one i he household ecei es any posi i e
amoun o emi ance income and i he child wo ked a leas 1 h in any labo ac i i y.
This app oach is consis en wi h he p e ious li e a u e which in es iga es how emi -
ance eceip in luences he p opensi y o child en o wo k. To add ess he second
ques ion, he e ec s a e hen es ima ed a he in ensi e ma gins, eplacing he indica o
a iables used in he i s model wi h measu es o he o al numbe o hou s a child
wo ks and he amoun o emi ance income he household ecei es, bo h measu ed on
a weekly basis. Bo h models a e es ima ed i s on he en i e sample. The models a e
also es ima ed sepa a ely o u ban and u al households, as well as o male and emale
child en in o de o explo e any po en ial he e ogenei y be ween hese g oups.
In bo h models, endogenei y is likely o lead o biased esul s. I he decision o mi-
g a e and send emi ances is pa o a household’s income di e si ica ion s a egy, hen
many o he ac o s con ibu ing o his decision (bo h obse ed and unobse ed) will
also in luence he decision o e whe he and how much child en in he household will
wo k, leading o p oblems wi h simul anei y bias. To con ol o po en ial endogenei y
issues, he models a e es ima ed using an ins umen al a iables (IV) s a egy. Gi en
he na u e o he model, inding an ins umen ha a ec s he size o emi ances being
sen wi hou di ec ly a ec ing he numbe o hou s wo ked by child en in he house-
hold can be challenging. The ins umen chosen o p edic he incidence and size o
emi ances ecei ed by he household is he exis ence o mig a ion ne wo ks, which is
an ins umen commonly employed in he mig a ion and emi ance li e a u e (Alcala
e al. 2014, Acos a e al. 2008, Acos a 2011, Cale o e al. 2009). In pa icula , he sha e
o households ecei ing emi ances in he household’s geog aphical depa men is used
as a p oxy o mig an ne wo ks. Mig an ne wo ks a e use ul ins umen s because
such ne wo ks can educe he cos s o mig a ion and imp o e employmen oppo un-
i ies in he des ina ion coun ies. Thus, hey can a ec bo h he decision o mig a e
and he size o emi ances being sen , wi hou necessa ily being co ela ed wi h he
unobse ed ac o s in luencing labo supply decisions by he household ecei ing he
emi ances. I is expec ed ha educed cos s o mig a ion, as indica ed by a la ge mi-
g an ne wo k, will lead o an inc ease in bo h he likelihood ha a household ecei es
emi ances and he size o he emi ances ecei ed. Agg ega e measu es, such as sha e
o households ecei ing emi ances, a e also use ul as ins umen s because hey a e
less suscep ible o issues o e e se causali y. Tha is, he agg ega e measu e will impac
he household’s decisions, bu he indi idual household is oo small ela i e o he es
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 7 o 26
o he popula ion o a ec he agg ega e measu e in any meaning ul way. On he o he
hand, agg ega e measu es do no allow o he possibili y ha he p esence o mig an
ne wo ks may a ec households di e en ly, depending on hei abili y o access hem.
To accoun o his, an in e ac ion e m be ween emi ance eceip a he depa men
le el and a household weal h index is also included. Inclusion o household weal h
allows o he ac ha in e na ional mig a ion can be qui e cos ly, so weal hie house-
holds will be in a be e posi ion o ake ad an age o he p esence o mig an ne wo ks
han poo households.
The gene al model is:
Lij ¼β1XþδRjþ∈1ij;ð9Þ
Rj¼β2XþΦZþ∈2ij ð10Þ
whe e L
ij
is a labo supply a iable pe aining o child iin household j,andR
j
is a a i-
able ela ed o emi ances ecei ed by household j. X is a ec o o indi idual and
household cha ac e is ics, and Zis a ec o o ins umen s. Indi idual cha ac e is ics o
he child en include age and gende . Household cha ac e is ics a e pe capi a mon hly
income, a weal h index
2
, an indica o o whe he he household is below he po e y
line, he numbe o child en ac oss a ious age/gende dimensions, age, gende , and
ma i al s a us o he household head, highes educa ion le el in he household, and
whe he Spanish is he i s language o any household membe .
In he i s pa o he analysis, bo h L
ij
and R
j
a e indica o s equal o one i he child
wo ks and he household ecei es emi ances, such ha
Lij ¼1i L
ij >0
Lij ¼0i L
ij ¼0
ð11Þ
and
Rj¼1i R
j>0
Rj¼0i R
j¼0
ð12Þ
whe e L
ij is he obse ed numbe o hou s wo ked by child iin household j,andR
jis
he obse ed alue o emi ances ecei ed by household j.
As Acos a (2011) poin s ou , he e a e se e al me hods a ailable o analyzing models
wi h bina y endogenous and ou come a iables. The e is also some deba e on which o
hese me hods is p e e ed. Ang is (2001) sugges s ha a wo-s age leas squa es
(2SLS) linea p obabili y model is su icien i he in en is o iden i y causali y. 2SLS
has he added bene i o p o iding an easy me hod o assessing ins umen alidi y.
On he o he hand, 2SLS also has a numbe o sho comings wi h bina y endogenous
and ou come a iables. Fi s , p edic ed alues in linea p obabili y models a e no
bound wi hin he uni in e al. This can con ibu e o he e oscedas ici y in single-
equa ion models and lead o “awkwa d”in e p e a ions o condi ional p obabili ies
(Woold idge 2002, p. 455). Addi ionally, as Mo i (2001) and Woold idge (2002) indi-
ca e, neglec ing o con ol o non-linea i y in he i s s age o he linea p obabili y
model can lead o inconsis en es ima es in he second s age. Ano he me hod, which
con ols o nonlinea i ies, is he bi a ia e p obi model de eloped by Heckman (1978).
The bi a ia e p obi model elimina es he p oblems associa ed wi h non-linea i y in he
wo equa ions. Howe e , iden i ica ion in he bi a ia e p obi model equi es s onge
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 8 o 26
abo e, he p e e ed model o his analysis is he bi a ia e p obi model, he esul s
o which a e p esen ed in column 4. Again, he coe icien o emi ance eceip is
nega i e and highly signi ican . In Table 5, i can be seen ha he signs and signi i-
cance o he a iables de e mining emi ance eceip a e consis en wi h hose o he
2SLS model.
Table 5 Fi s -s age/ emi ance equa ion es ima es, child en aged 7–13
(1) (2)
2SLS Bi a ia e p obi
Age 0.0008 0.0101
(0.002) (0.0146)
Male 0.0144 0.0980
(0.0101) (0.0710)
Household income pe capi a −2.25e−6−3.8e−5
(8.18e−6) (4.75e−5)
Po e y indica o −0.0275* −0.1947*
(0.0121) (0.0772)
Ru al 0.0021 0.0191
(0.012) (0.0835)
HH age 0.0029*** 0.0166***
(0.0005) (0.0024)
HH emale 0.073*** 0.4038***
(0.0135) (0.0697)
HH ma ied −0.0168 −0.0870
(0.009) (0.0645)
Max educa ion 0.0003 0.0042
(0.0014) (0.0095)
Child en 0–5 0.009 0.0435
(0.005) (0.0355)
Boys 6–17 −0.0002 0.0026
(0.004) (0.0321)
Gi ls 6–17 0.0108* 0.0804*
(0.0049) (0.0323)
Spanish i s language −0.0074 −0.0321
(0.0122) (0.0943)
Weal h index −0.0167*** −0.1085*
(0.0057) (0.0462)
Sha e o households ecei ing emi ances 0.9064*** 6.3666***
(0.1429) (1.0362)
Sha e*weal h 0.2849*** 1.7559***
(0.0624) (0.4513)
Cons an −0.1388*** −2.9277***
(0.0375) (0.2568)
ρ
12
0.5451**
(0.1960)
Robus s anda d e o s in pa en heses
*p< 0.05, **p< 0.01, ***p< 0.001
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 15 o 26

While he in e p e a ion o coe icien es ima es ob ained by he bi a ia e p obi
model a e no as s aigh o wa d as hose ob ained by 2SLS, es ima es o causal ea -
men e ec s can be eco e ed using he p edic ed alues (Ang is e al. 1996). A e age
ea men e ec s (ATE) and he a e age ea men e ec on he ea ed (ATT) o he
bi a ia e p obi models a e es ima ed using he me hod p oposed by Nichols (2011).
The ATE is es ima ed as he mean o he di e ence be ween he p edic ed p obabili y
ha a child wo ks when R
j
= 1, and he p edic ed p obabili y when R
j
= 0, wi h o he
co a ia es held a hei obse ed alues. The ATT is es ima ed as he mean di e ence
in p edic ed p obabili ies condi ional on he child li ing in a household ha ecei es e-
mi ances. The e ec s o emi ance eceip co esponding o he esul s in Tables 4 and 5
a e p esen ed in he i s ow o Table 6. These esul s indica e ha emi ances eceip is
expec ed o educe he likelihood a child wo ks by an a e age o 16.6 % among all house-
holds, and by 25.5 % among he subg oup o emi ance- ecei ing households.
The e a e also a numbe o o he no ewo hy esul s among he con ol a iables in
Tables 4. As was seen in he summa y s a is ics, child en in u al households a e sig-
ni ican ly mo e likely o wo k han hei u ban coun e pa s, and boys a e mo e likely
o wo k han gi ls. The p opensi y o wo k inc eases wi h he child’s age, and child en
in households whe e Spanish is he i s language a e signi ican ly less likely o wo k
han child en in households wi h an indigenous i s language. In addi ion o ac ing as
a p oxy o a household’s e hnici y, he p ima y language spoken also se es as an indi-
ca o o economic oppo uni y. Al hough Boli ia has 35 o icially ecognized languages
(Taylo 2004), Spanish is he adi ional language o he eli e class and also he mos
common sha ed language be ween g oups. As such, poo Spanish skills can lead o eco-
nomic exclusion o he pa en s, hus inc easing he need o child en o help supple-
men income. Fu he mo e, al hough he Educa ion Re o m Bill o 1994 called o
educa ion o be p o ided in indigenous languages, indigenous schools a e ypically o
lowe quali y han Spanish language schools, he eby making he oppo uni y cos o
wo k lowe o indigenous child en. This ela ionship be ween economic inclusion and
educa ion, and he p opensi y o wo k can be seen in o he a iables as well— he likeli-
hood o wo king declines wi h he le el o educa ion in he household, as well as wi h
household weal h. Finally, in Table 5, he es ima e o he co ela ion be ween he e o s
o he wo equa ions o he bi a ia e p obi model, ρ
12
, is posi i e and signi ican ,
Table 6 T ea men e ec es ima es o emi ance eceip on p opensi y o wo k
A e age ea men e ec (ATE) A e age ea men e ec on he ea ed (ATT)
All child en −0.166 −0.255
(0.116) (0.092)
Ru al −0.408 −0.408
(0.103) (0.216)
U ban −0.114 −0.476
(0.077) (0.136)
Male −0.168 −0.251
(0.119) (0.098)
Female −0.184 −0.316
(0.130) (0.010)
No e: S anda d de ia ions in pa en heses
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 16 o 26
indica ing ha unobse ed ac o s con ibu ing o an inc ease in p opensi y o emi
a e also associa ed wi h an inc ease in he p opensi y o child en o wo k and con i m-
ing ha es ima ing he wo equa ions sepa a ely would lead o biased esul s.
Gi en ha he esul s abo e indica e di e en p opensi ies o wo k be ween u al and
u ban households, as well as be ween male and emale child en, he ollowing sec ion ex-
plo es how emi ances a ec hese g oups di e en ly. Table 6 p esen s he es ima ed ea -
men e ec s o emi ance eceip o u ban and u al households and o male and emale
child en ob ained by es ima ing he bi a ia e p obi model sepa a ely o each g oup
(comple e es ima ion esul s a e p esen ed in Tables 7 and 8). Despi e la ge di e ences in
he p opensi y o wo k be ween boys and gi ls, he a e age ea men e ec s appea o be
ai ly simila . Howe e , he ea men e ec on he ea ed is subs an ially la ge o
emales. Thus, i would appea ha al hough boys a e signi ican ly mo e likely o wo k han
gi ls, emi ance income ends o be di ec ed owa d educing he labo e o o gi ls.
Compa ison o he ea men e ec s be ween u ban and u al households yields many
in e es ing esul s. Fi s , he ATT is app oxima ely 7 pe cen age poin s la ge o u ban
households han u al households. I is no en i ely su p ising ha emi ances a e
mo e e ec i e a educing child labo in u ban a eas han u al a eas. Fi s , child en
a e signi ican ly mo e likely o wo k in u al a eas ega dless o whe he hei house-
hold ecei es emi ances. Second, i he child labo decision is a unc ion o al e na i e
uses o he child’s ime, e.g., going o school, he e may be ewe op ions a ailable in
u al a eas. Finally, i child labo is de e mined by household labo demand ( a he han
he need o income), labo ma ke impe ec ions in u al a eas can make hi ing adul
wage labo mo e di icul in u al a eas, hus inc easing he need o compensa e wi h
child labo .
Addi ionally, he ATT o bo h g oups a e qui e la ge ela i e o he obse ed sha e
o child en who epo wo king. This would imply ha al hough he obse ed di e -
ence be ween he sha e o child en wo king in households ha ecei e emi ances and
hose who do no is ai ly small, unde he coun e ac ual o no households ecei ing
emi ances, hese di e ences would be much la ge . Thus, he obse ed di e ences
ound in he summa y s a is ics signi ican ly unde es ima e he causal impac o emi -
ances on educing he p e alence o child labo .
Pe haps he mo e s iking esul is he compa ison be ween ATE and ATT o he
wo g oups. In u ban households, he e ec on he ea ed is ou imes as la ge as he
a e age ea men e ec . This would sugges ha u ban households who op o send
mig an s ab oad bene i much mo e han he a e age household would, i.e., ha he e
is posi i e selec ion in o mig a ion. Howe e , his does no necessa ily appea o be he
case o u al households, as he ATE and ATT a e oughly equi alen , a an app oxi-
ma e 41 % educ ion in he p obabili y ha a child will wo k. One possible explana ion
migh be a kinship e ec ha could exis in he u al a eas. Tha is, i u al illages end
o be made up la gely o ex ended amilies o close kni indigenous popula ions, hen
income and expendi u e decisions migh be made a he communi y le el, a he han
he household, and bene i s om emi ance income migh be sp ead ac oss he en i e
illage. Whe eas in u ban a eas, he popula ion may be mo e segmen ed and emi -
ance income would emain con ained among ecipien households. Un o una ely, he
cu en da a does no con ain in o ma ion on in e -household ela ionships, so his
hypo hesis is no di ec ly es able.
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 17 o 26
6.2 In ensi y o wo k
Table 9 p esen s coe icien es ima es o Eq. 9, eplacing he emi ance and labo
supply indica o a iables wi h measu es o he alue o weekly emi ance income and
he numbe o hou s wo ked pe week, espec i ely, as desc ibed by Eqs. 13 and 14.
Column 1 epo s es ima es o a simple Tobi model. The coe icien is nega i e, bu
no signi ican . Columns 2–4 epo he esul s o ins umen al a iable models wi h
a ious le els o con ol o censo ing in he dependen a iables. Columns 2 and 3 a e
Table 7 Bi a ia e p obi es ima es o p opensi y o wo k by gende , u ban and u al households
(1) (2) (3) (4)
Boys Gi ls Ru al U ban
Remi ance indica o −0.9338* −1.0764** −1.4099*** −1.3881**
(0.4032) (0.3446) (0.2174) (0.4722)
Age 0.1440*** 0.1041*** 0.1085*** 0.1332***
(0.0191) (0.0176) (0.0176) (0.0308)
Male 0.1666* 0.0588
(0.0805) (0.0904)
Household income pe capi a 0.0001* 0.0001 0.0000 0.0001*
(0.0001) (0.0001) (0.0001) (0.0001)
Po e y indica o −0.0012 −0.0758 0.2814* −0.4582***
(0.0964) (0.0983) (0.1153) (0.0901)
Ru al 0.9665*** 0.8100***
(0.1039) (0.0909)
HH age 0.0033 0.0025 0.0061 0.0002
(0.0036) (0.0032) (0.0032) (0.0053)
HH emale 0.0407 0.1686 0.0010 0.2568*
(0.1070) (0.1020) (0.0983) (0.1225)
HH ma ied 0.0828 0.1417 0.2497*** −0.1414
(0.0785) (0.0772) (0.0725) (0.0749)
Max educa ion −0.0229 −0.0098 0.0067 −0.0374**
(0.0118) (0.0117) (0.0116) (0.0122)
Child en 0–5 0.0354 −0.0262 −0.0539 0.1333**
(0.0431) (0.0424) (0.0359) (0.0489)
Boys 6–17 0.0162 0.0534 −0.0270 0.1435***
(0.0398) (0.0410) (0.0376) (0.0421)
Gi ls 6–17 0.0701 0.1378*** 0.0278 0.2005***
(0.0400) (0.0360) (0.0375) (0.0384)
Spanish i s language −0.6804*** −0.7856*** −0.6529*** −0.4461
(0.0986) (0.0981) (0.0753) (0.2624)
Weal h index −0.0660** −0.0763** −0.0793*** 0.0098
(0.0249) (0.0272) (0.0228) (0.0344)
Cons an −2.2175*** −1.9548*** −1.5062*** −2.1299***
(0.2873) (0.3022) (0.2636) (0.3416)
Numbe 2758 2593 2036 3315
χ
2
790.9926 766.8504 582.4423 396.2101
Robus s anda d e o s in pa en heses
*p< 0.05, **p< 0.01, ***p< 0.001
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 18 o 26
p esen ed o mainly exposi o y pu poses, while column 4 is he p e e ed speci ica ion.
Column 2 p esen s he esul s o a 2SLS model. The coe icien es ima e o emi ance
income is nega i e and signi ican . Howe e , since nei he equa ion con ols o censo -
ing, hese es ima es a e likely o be biased. The 2SLS model is p esen ed p ima ily o
he pu pose o es ing ins umen alidi y. Resul s o Hansen’sJ es a e p esen ed a
Table 8 Fi s -s age/ emi ance equa ion es ima es by gende , u ban and u al households
(1) (2) (3) (4)
Boys Gi ls Ru al U ban
Age 0.0182 0.0040 −0.0012 0.0143
(0.0197) (0.0215) (0.0255) (0.0179)
Male 0.0044 0.1344
(0.1186) (0.0892)
Household income pe capi a −0.0000 −0.0000 −0.0002 −0.0000
(0.0001) (0.0001) (0.0001) (0.0001)
Po e y indica o −0.1884 −0.2005 −0.0638 −0.2886**
(0.1080) (0.1117) (0.1362) (0.0898)
Ru al −0.0776 0.1609
(0.1116) (0.1229)
HH age 0.0175*** 0.0159*** 0.0231*** 0.0131***
(0.0032) (0.0037) (0.0040) (0.0030)
HH emale 0.3920*** 0.4120*** 0.3838** 0.4154***
(0.1003) (0.0990) (0.1215) (0.0874)
HH ma ied −0.0152 −0.1750 −0.1432 −0.0921
(0.0916) (0.0907) (0.1091) (0.0789)
Max educa ion 0.0179 −0.0096 0.0543*** −0.0114
(0.0138) (0.0128) (0.0156) (0.0114)
Child en 0–5 0.0685 0.0084 0.0164 0.0859
(0.0462) (0.0559) (0.0500) (0.0480)
Boys 6–17 0.0547 −0.0651 −0.0865 0.0679
(0.0413) (0.0522) (0.0503) (0.0422)
Gi ls 6–17 0.0414 0.1145* −0.0392 0.1536***
(0.0456) (0.0458) (0.0583) (0.0395)
Spanish i s language −0.0008 −0.0804 −0.1530 0.4972
(0.1312) (0.1363) (0.1060) (0.3403)
Weal h index −0.1554* −0.0453 −0.1776* −0.1326
(0.0649) (0.0643) (0.0880) (0.0733)
Sha e o households ecei ing emi ances 5.6515*** 7.6036*** 10.6868*** 4.9584**
(1.4282) (1.4798) (2.3359) (1.9128)
Sha e*weal h 1.9761** 1.4407* 2.4640** 2.1800**
(0.6427) (0.6318) (0.8744) (0.7331)
Cons an −3.1419*** −2.7499*** −3.4957*** −3.2247***
(0.3600) (0.3518) (0.4557) (0.4917)
ρ
12
0.5911 0.6164* 1.0518*** 0.9456
(0.3042) (0.2560) (0.2271) (0.7980)
Robus s anda d e o s in pa en heses
*p< 0.05, **p< 0.01, ***p< 0.001
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 19 o 26
he bo om o column 2. The p alue o 0.1238 indica es ha he null hypo hesis can-
no be ejec ed and he ins umen s a e alid. Addi ionally, esul s o he i s s age,
p esen ed in Table 10, show ha ac o s de e mining emi ance size ha e he same
ela ionship, in e ms o sign and signi icance, as hose de e mining he incidence o
emi ance eceip . Column 3 p esen s esul s o he IV Tobi model. The coe icien is
much la ge han hose p esen ed in he 2SLS model, indica ing ha a s ong upwa d
Table 9 Es ima ion esul s o in ensi y o wo k, Ln(hou s pe week + 1), child en age 7–13
(1) (2) (3) (4)
Tobi 2SLS IV Tobi Bi a ia e Tobi
Ln( emi ances + 1) −0.0365 −0.3103*** −2.0792*** −0.3456**
(0.0408) (0.0760) (0.5027) (0.1071)
Age 0.3576*** 0.0874*** 0.3646*** 0.3587***
(0.0336) (0.0093) (0.0428) (0.0340)
Male 0.3445* 0.1226** 0.4839* 0.3680*
(0.1625) (0.0469) (0.2117) (0.1646)
Household income pe capi a 0.0004** 0.0001** 0.0005** 0.0004**
(0.0001) (0.0000) (0.0002) (0.0001)
Po e y indica o −0.0625 −0.0446 −0.5011 −0.1549
(0.1980) (0.0541) (0.2770) (0.2007)
Ru al 2.7294*** 0.6771*** 2.7957*** 2.7262***
(0.1788) (0.0591) (0.2358) (0.1802)
HH age 0.0026 0.0068** 0.0345** 0.0073
(0.0061) (0.0021) (0.0113) (0.0064)
HH emale 0.0730 0.1729** 1.1981** 0.2537
(0.1975) (0.0647) (0.3672) (0.2098)
HH ma ied 0.2672 0.0425 0.0627 0.2314
(0.1526) (0.0415) (0.1964) (0.1531)
Max educa ion −0.0503* −0.0024 −0.0518 −0.0501*
(0.0231) (0.0061) (0.0297) (0.0234)
Child en 0–5−0.0232 0.0149 0.0997 0.0002
(0.0792) (0.0242) (0.1034) (0.0801)
Boys 6–17 0.1231 0.0283 0.1756 0.1337
(0.0771) (0.0239) (0.0984) (0.0780)
Gi ls 6–17 0.2591*** 0.0778*** 0.4345*** 0.2897***
(0.0731) (0.0231) (0.1062) (0.0750)
Spanish i s language −1.6864*** −0.8026*** −1.8707*** −1.7215***
(0.1705) (0.0723) (0.2272) (0.1740)
Weal h index −0.2374*** −0.0418** −0.0856 −0.2143***
(0.0466) (0.0151) (0.0689) (0.0480)
Cons an −6.4294*** −0.2371 −6.9997*** −6.4940***
(0.5557) (0.1611) (0.7126) (0.5609)
Numbe 5351 5351 5351 5351
F173.0204 86.0245
χ
2
1302.5538 3056.4818
Robus s anda d e o s in pa en heses
*p< 0.05, **p< 0.01, ***p< 0.001
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 20 o 26

bias occu s when censo ing in he labo supply equa ion is no con olled o . Howe e ,
he IV Tobi model does no accoun o censo ing in he emi ance eceip equa ion.
Column 4 p esen s esul s om he bi a ia e Tobi model, which con ols o censo ing
Table 10 Fi s -s age/ emi ance equa ion es ima es, child en age 7–13
(3) (2) (4)
2SLS IV Tobi Bi a ia e Tobi
Age 0.0045 0.0047 0.1089
(0.013) (0.0128) (0.1495)
Male 0.0768 0.0780 0.9443
(0.0629) (0.0629) (0.7264)
Household income pe capi a 0.00004 0.00003 −0.0002
(0.00006) (0.0001) (0.0005)
Po e y indica o −0.2270*** −0.2275** −2.2450**
(0.0778) (0.0778) (0.7891)
Ru al 0.0092 0.0116 0.2004
(0.0738) (0.0739) (0.8560)
HH age 0.0158*** 0.0159*** 0.1643***
(0.0158) (0.0030) (0.0235)
HH emale 0.5226*** 0.5240*** 4.3580***
(0.0900) (0.0899) (0.6994)
HH ma ied −0.0853 −0.0841 −0.8552
(0.0568) (0.0568) (0.6570)
Max educa ion 0.0036 0.0044 0.0429
(0.0088) (0.0087) (0.0951)
Child en 0–5 0.0540 0.0533 0.4179
(0.0304) (0.0303) (0.3613)
Boys 6–17 0.0177 0.0160 0.1049
(0.028) (0.0279) (0.3314)
Gi ls 6–17 0.0836*** 0.0830** 0.8643**
(0.031) (0.0309) (0.3315)
Spanish i s language −0.0620 −0.0435 −0.4457
(0.0694) (0.0678) (0.9720)
Weal h index −0.1283*** −0.0996** −1.2015*
(0.0361) (0.0348) (0.4819)
Sha e o households ecei ing emi ances 6.041*** 6.4682*** 66.4583***
(0.9084) (0.8758) (10.9832)
Sha e*weal h 2.102*** 1.7666*** 19.1313***
(0.413) (0.3816) (4.6627)
Cons an −0.8671*** −0.9342*** −30.2639***
(0.234) (0.2300) (2.5959)
ρ
12
0.3381**
(0.1118)
Hansen’sJs a is ic 2.369
p alue 0.1238
Robus s anda d e o s in pa en heses
*p< 0.05, **p< 0.01, ***p< 0.001
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 21 o 26
in bo h equa ions. The coe icien o −0.35 indica es ha a 10 % inc ease in emi ance
income is associa ed wi h a 3.5 % dec ease in he numbe o hou s wo ked by child en.
This esul highligh s he ac ha in addi ion o emi ances being e ec i e a mo ing
some child en comple ely ou o he labo o ce, hey can also help o educe he bu -
den o labo on o he s, hus po en ially eeing mo e ime o de o e owa d human
capi al accumula ion.
Tables 11 and 12 p esen s es ima es o he bi a ia e Tobi model o male and emale
child en, as well as u al and u ban households. The e does no appea o be a la ge
di e ence be ween he e ec s o male and emale child en. The e does, howe e , ap-
pea o be a la ge di e ence be ween coe icien es ima es o u ban and u al house-
holds. Table 11 indica es o u al households a 10 % inc ease in weekly emi ance
income is associa ed wi h a 7 % dec ease in hou s wo ked. Fo u ban households, a
10 % inc ease in emi ance income is associa ed wi h a 10.8 % dec ease in hou s
wo ked. While some o his disc epancy may be due o di e en economic condi ions
in u al a eas (labo ma ke ic ions, lack o al e na i e uses o child’s ime, e c.), i is
wo h no ing ha much o his esul may also be d i en by di e ences in he le els o
emi ance income. Recall ha he median emi ance- ecei ing u ban household e-
cei es 699 boli ianos (US$101), whe eas he median u al household ecei es only 233
boli ianos (US$34). Thus, in absolu e e ms, a 10 % inc ease o an u ban household is
nea ly h ee imes he amoun o a u al household. To pu i ano he way, o a child
wo king he mean numbe o hou s pe week, li ing in a household ecei ing he median
amoun o emi ances, i would ake a 17.9 boli iano inc ease in emi ance income o e-
duce hei weekly employmen by 1 h in a u al a ea, whe eas i would ake 26.16 boli ia-
nos in an u ban household. When iewed in his way, he ma ginal e ec o a emi ed
boli iano is la ge in u al a eas. Fu he mo e, his also sugges s ha emi ances abili y o
educe child labo may be diminishing as emi ance income inc eases.
Many o he o he a iables ound o impac he labo supply decision a he ex en-
si e ma gins also ca y h ough o he in ensi e ma gins. Child en in u al households
wo k mo e hou s pe week han hei u ban coun e pa s. Boys wo k mo e hou s pe
week han gi ls, and olde child en wo k mo e han younge child en. Child en whose
i s language is Spanish wo k signi ican ly less han child en om indigenous-language
households. Household educa ion educes he numbe o hou s wo ked in u ban
households, bu no in u al households. Simila ly, he numbe o child en in he
household ends o inc ease he amoun o wo k in u ban households, bu has no
signi ican e ec in u al households.
7 Conclusions
The d ama ic ise o emi ances lowing o Boli ia in he pas se e al yea s is undoub -
edly a ec ing he economy in a numbe o ways. This s udy has explo ed how emi -
ances a e a ec ing he p e alence and in ensi y o child labo . The esul s p esen ed
abo e p o ide e idence ha emi ances can po en ially be used o help educe he
p e alence and ex en o child labo in he coun y. Whe eas p e ious esea ch has
ound ha emi ances can play a ole in mo ing child en ou o he labo ma ke , his
s udy also inds ha emi ances can signi ican ly educe he numbe o hou s wo ked
by child en who con inue o engage in economic ac i i y. While i may be p e e able
ha child en no wo k a all, e en a modes educ ion in economic ac i i y can ee up
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 22 o 26
ime o child en o inc ease hei human capi al and p opel hem on o a sus ainable
pa h o li hemsel es and u u e gene a ions ou o po e y and in o a highe s anda d
o li ing.
While he indings sugges ha emi ances can lead o an ac oss he boa d educ ion
in child labo , he di e ence in magni udes o he e ec be ween u ban and u al a eas
indica es ha emi ances a e mo e e ec i e in some a eas han o he s. In pa icula ,
he indings abo e indica e ha emi ances a e mo e e ec i e a mo ing child en
Table 11 Bi a ia e Tobi es ima es o hou s wo ked by gende , u ban and u al households
(1) (2) (3) (4)
Boys Gi ls Ru al U ban
Ln( emi ances + 1) −0.3619* −0.3828* −0.7038*** −1.0878*
(0.1471) (0.1546) (0.0499) (0.4502)
Age 0.4090*** 0.3091*** 0.2605*** 0.7861***
(0.0497) (0.0458) (0.0402) (0.0939)
Male 0.3826* 0.2710
(0.1838) (0.5159)
Household income pe capi a 0.0004* 0.0003 0.0001 0.0008**
(0.0002) (0.0002) (0.0003) (0.0003)
Po e y indica o −0.0209 −0.3483 0.5829* −2.7517***
(0.2763) (0.2942) (0.2941) (0.5588)
Ru al 2.9289*** 2.4878***
(0.2527) (0.2553)
HH age 0.0085 0.0067 0.0228** −0.0047
(0.0092) (0.0087) (0.0077) (0.0251)
HH emale 0.0729 0.5071 0.2346 1.3522
(0.2972) (0.2933) (0.2378) (0.7940)
HH ma ied 0.1473 0.3548 0.4647** −0.8779
(0.2152) (0.2159) (0.1758) (0.4511)
Max educa ion −0.0678* −0.0325 0.0391 −0.2169**
(0.0330) (0.0328) (0.0279) (0.0698)
Child en 0–5 0.0924 −0.0957 −0.0865 0.6800*
(0.1128) (0.1122) (0.0810) (0.3010)
Boys 6–17 0.0903 0.1825 −0.0503 0.8560***
(0.1088) (0.1127) (0.0913) (0.2353)
Gi ls 6–17 0.2075 0.3842*** 0.0505 1.1363***
(0.1099) (0.1018) (0.0886) (0.2580)
Spanish i s language −1.5095*** −1.9637*** −1.4282*** −2.6289*
(0.2391) (0.2538) (0.1703) (1.1791)
Weal h index −0.1980** −0.2239** −0.1761** 0.0293
(0.0653) (0.0711) (0.0537) (0.1738)
Cons an −6.7260*** −5.9101*** −3.7725*** −12.2943***
(0.7629) (0.8138) (0.6351) (2.3495)
Numbe 2758 2593 2036 3315
χ
2
1569.4078 1579.3853 858.4691 411.7604
Robus s anda d e o s in pa en heses
*p< 0.05, **p< 0.01, ***p< 0.001
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 23 o 26
comple ely ou o he labo ma ke in u ban a eas han u al a eas. A he same ime,
a e con olling o he ini ial le el o emi ance income and hou s wo ked, he ma -
ginal impac o a emi ed boli iano on he numbe o hou s wo ked is la ge in u al
a eas han u ban a eas. This sugges s ha mig a ion policy can play an impo an ole
in add essing child labo in Boli ia, and ha policy egimes should di e be ween u al
Table 12 Bi a ia e Tobi es ima es, emi ances equa ion by gende , u ban and u al households
(1) (2) (3) (4)
Boys Gi ls Ru al U ban
Age 0.1885 0.0496 0.0670 0.1701
(0.1981) (0.2216) (0.2226) (0.1823)
Male 0.0489 1.2967
(1.0250) (0.9041)
Household income pe capi a −0.0003 −0.0001 −0.0017 0.0000
(0.0007) (0.0007) (0.0009) (0.0005)
Po e y indica o −2.1848* −2.2761* 0.7878 −3.1085***
(1.0882) (1.1517) (1.3333) (0.9103)
Ru al −0.8053 1.7174
(1.1290) (1.2616)
HH age 0.1728*** 0.1544*** 0.1927*** 0.1267***
(0.0302) (0.0363) (0.0341) (0.0307)
HH emale 4.1826*** 4.4889*** 3.0631** 4.5613***
(0.9948) (0.9963) (1.1255) (0.8542)
HH ma ied −0.1532 −1.7226 −0.5232 −0.8319
(0.9006) (0.9431) (0.9779) (0.8144)
Max educa ion 0.1803 −0.1024 0.5475*** −0.1133
(0.1346) (0.1295) (0.1320) (0.1178)
Child en 0–5 0.6790 0.0531 0.0052 0.8532
(0.4637) (0.5686) (0.4413) (0.4675)
Boys 6–17 0.6575 −0.6051 −0.9621* 0.7760
(0.4226) (0.5321) (0.4505) (0.4281)
Gi ls 6–17 0.4343 1.2362** −0.6108 1.6087***
(0.4625) (0.4714) (0.4899) (0.3886)
Spanish i s language −0.2836 −0.8063 −3.1752*** 5.1546
(1.3428) (1.3997) (0.8889) (3.5112)
Weal h index −1.7588** −0.4324 −2.2230** −1.4578
(0.6521) (0.6794) (0.7024) (0.7888)
Sha e o households 59.1451*** 78.4058*** 123.3242*** 53.6113**
Recei ing emi ances (14.8499) (15.7588) (18.5647) (16.8350)
Sha e*weal h 22.0506*** 14.8830* 26.4385*** 23.5423**
(6.3731) (6.6495) (6.9854) (7.8643)
Cons an −32.0373*** −28.6078*** −35.1037*** −33.9478***
(3.5546) (3.6442) (4.0603) (4.7715)
ρ
12
0.3694* 0.3678* 1.2965*** 0.6923
(0.1633) (0.1577) (0.1113) (0.4437)
Robus s anda d e o s in pa en heses
*p< 0.05, **p< 0.01, ***p< 0.001
Coon IZA Jou nal o Mig a ion (2016) 5:1 Page 24 o 26