remittances and the brain drain revisited: the microdata...
TRANSCRIPT
Remittances and the Brain Drain Revisited: TheMicrodata Show That More Educated Migrants
Remit More
Albert Bollard, David McKenzie, Melanie Morten, andHillel Rapoport
Two of the most salient trends in migration and development over the last two decadesare the large rise in remittances and in the flow of skilled migrants. However, recent lit-erature based on cross-country regressions has claimed that more educated migrantsremit less, leading to concerns that further increases in skilled migration will impederemittance growth. Microdata from surveys of immigrants in 11 major destinationcountries are used to revisit the relationship between education and remitting behavior.The data show a mixed pattern between education and the likelihood of remitting, and astrong positive relationship between education and amount remitted (intensive margin),conditional on remitting at all (extensive margin). Combining these intensive and exten-sive margins yields an overall positive effect of education on the amount remitted for thepooled sample, with heterogeneous results across destinations. The microdata allowinvestigation of why the more educated remit more, showing that the higher incomeearned by migrants, rather than family characteristics, explains much of the higher remit-tances. remittances, migration, brain drain, education JEL codes: O15, F22, J61
Two of the most salient trends in migration and development over the last twodecades are the large rise in remittances and in the flow of skilled migrants.Officially recorded remittances to developing countries have more than tripledover the last decade, rising from $85 billion in 2000 to $305 billion in 2008
Albert Bollard ([email protected]) is a PhD student at Stanford University. David McKenzie
([email protected]; corresponding author) is a senior economist in the Finance and Private
Sector Research Unit of the Development Research Group at the World Bank. Melanie Morten
([email protected]) is a PhD student at Yale University. Hillel Rapoport ([email protected]) is
professor of economics at Bar Ilan University and at EQUIPPE, University of Lille and is currently a
visiting research fellow at the Center for International Development at Harvard University. The authors
are grateful for funding for this project from the Agence Francaise de Developpement (AFD). They
thank all the individuals and organizations that graciously shared their surveys of immigrants, and they
are grateful to Michael Clemens, participants at the 2nd Migration and Development Conference held at
the World Bank in September 2009, three anonymous referees, and the journal editor for helpful
comments. All opinions are those of the authors and do not necessarily represent those of AFD or the
World Bank. A supplemental appendix to this article is available at http://wber.oxfordjournals.org.
THE WORLD BANK ECONOMIC REVIEW, VOL. 25, NO. 1, pp. 132–156 doi:10.1093/wber/lhr013Advance Access Publication May 12, 2011# The Author 2011. Published by Oxford University Press on behalf of the International Bankfor Reconstruction and Development / THE WORLD BANK. All rights reserved. For permissions,please e-mail: [email protected]
132
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
(World Bank 2008, 2009). The number of highly educated migrants fromdeveloping countries residing in Organisation for Economic Co-operation andDevelopment (OECD) countries doubled over 1990–2000 (Docquier andMarfouk 2005) and likely has grown since then as developed countries haveincreasingly pursued skill-selective immigration policies.1
However, despite this positive association at the global level between risingremittances and rising high-skill migration, there are concerns—stemming fromthe belief that more educated individuals may remit less—that increasinglyskill-selective immigration policies may slow or even end the rise in remit-tances. This belief is taken as fact by many; for example, an OECD (2007,p. 11) report says that “low skilled migrants tend to send more money home.”The main empirical evidence to support this assertion across a range ofcountries comes from two recent studies (Faini 2007; Niimi, Ozden, and Schiffforthcoming) whose cross-country macroeconomic analyses find that the highlyskilled (defined as those with tertiary education) remit less.
Yet there are many reasons to question the results of these cross-country esti-mations. Both studies relate the amount of remittances received at a countrylevel to the share of migrants with tertiary education, at best telling us whethercountries that send a larger share of highly skilled migrants receive less or morein remittances than countries that send fewer skilled migrants. The studies donot answer the factual question of whether more educated migrants remit moreor less. There are a host of differences across countries that could cause a spur-ious relationship to appear between remittances and skill level across countries.For example, if poverty is a constraint to both migration and education, richerdeveloping countries might be able to send more migrants (yielding moreremittances) and those migrants might also have more schooling. Faini (2007)treats the share of migrants who are skilled as exogenous. Niimi, Ozden, andSchiff (forthcoming) try to instrument for the education mix of migrants, buttheir instruments seem unlikely to satisfy the exclusion restrictions. Forexample, public spending on education is likely a function of a country’soverall institutional and economic development, which should independentlyaffect the incentive to remit; migrants might send money to overcome poorpublic spending or for investment when complementary infrastructure andinstitutions are in place.
This article revisits the relationship between remittances and education levelusing microdata that permit computing the association between a migrant’seducation level and remitting behavior. The authors assembled the most com-prehensive micro-level database on remitting behavior currently available, com-prising data on 33,000 immigrants from developing countries from 14 surveys
1. In contrast, the number of low-skill migrants (primary education or less) increased only 15
percent over the period. Immigration to OECD countries (as defined by the number of foreign born)
was estimated at 90 million in 2000, about half of total world migration. Of the 90 million immigrants,
60 million were ages 25 or older and were split equally across education categories (primary, secondary,
and tertiary; Docquier and Marfouk 2005).
Bollard, McKenzie, Morten, and Rapoport 133
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
in 11 OECD destination countries. The analysis begins by establishing thefactual relationship between the propensity to remit and education. No attemptis made to estimate the causal impact of education on remittances.2 From apolicy perspective, the concern is whether migration policies that shift the edu-cation composition of migrants affect remittances, not whether education pol-icies that change how much education individuals have affects remittances.Microdata enables asking whether more educated individuals are more or lesslikely to remit (the extensive margin) and whether they send more or less remit-tances if they do remit (the intensive margin). A mixed association is foundbetween education and remittances at the extensive margin, and a strong posi-tive relationship at the intensive margin. Combining both the extensive andintensive margins reveals that, at least in this large sample, more educatedmigrants do remit significantly more—migrants with a university degree remit$300 more yearly than migrants without a university degree, where the meanannual remittance over the entire sample is $730.
The article is organized as follows. Section I summarizes several theories ofremitting behavior and the predictions they give for the relationship betweeneducation and remittances. Section II then describes the dataset of immigrantsurveys with remittances. Section III provides results, and section IV drawssome implications.
I . T H E O R E T I C A L B A C K G R O U N D
Theoretically, there are several reasons to believe that there will be differencesin the remitting patterns of highly skilled and less-skilled emigrants. However,a priori, it is not clear which direction will dominate and thus whether thehighly skilled will remit more or less on average. On the one hand, severalfactors tend to lead highly skilled migrants to be more likely to remit and tosend a larger amount of remittances. First, highly skilled individuals are likelyto earn more as migrants, potentially increasing the amount they can remit.Second, their education may have been funded by family members in the homecountry, with remittances serving as repayment. Third, skilled migrants are lesslikely to be illegal migrants and more likely to have bank accounts, loweringthe financial transaction costs of remitting. On the other hand, several otherfactors might lead highly skilled migrants to be less likely to remit and to remitless. First, highly skilled migrants may be more likely to migrate with theirentire household, so they would not have to send remittances in order to sharetheir earnings with their household. Second, they might come from richerhouseholds, which have less need for remittances to alleviate liquidity con-straints. Third, they might have less intention of returning to their homecountry, reducing the role of remittances as a way of maintaining prestige andties to the home community.
2. Convincing instruments are lacking to identify this impact.
134 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
Before turning to the empirical analysis, it is useful to clarify the theoreticalrelationship between education and remittances and the implied testable predic-tions about education. This will allow identifying the role of several variablesthat, once interacted with education and various possible motivations to remit,have the potential to explain differences in remitting behavior by educationlevel. The discussion is limited to three possible motives for remittances: altru-ism, exchange, and investment. These were selected for general empirical rel-evance and as the motives through which education is most likely to affectremittances.3
Altruism
Altruistic preferences are generally captured by weighting one’s own (themigrant’s) and others’ (relatives) consumption in an individual utility function,with weights reflecting the individual’s degree of altruism, which can itselfdepend on the closeness among the relatives considered (both family and geo-graphic proximity). For given weights and initial distribution of income,altruistic individuals maximize their utility by transferring (remitting) incomeso as to reach the desired distribution between themselves and the beneficiariesof their altruism. Altruistic transfers take place if pretransfer income differencesare sufficiently large or altruism is strong enough and increases with thedonor’s income (the extensive margin) and decreases with the recipients’income (the intensive margin)
What does this basic theoretical framework imply for the comparative remit-ting behavior of highly educated and less well educated migrants? First, edu-cated migrants tend to earn more, which all else equal should induce moreremittances (at both margins). Second, the conventional wisdom is that edu-cated migrants tend to have more family members with them because of ahigher propensity to move with their immediate family, which all else equalshould lower remittances.4 Methodologically, this suggests that the locationand composition of the family (which fraction of the family accompanies themigrant and which fraction stays in the country of origin) is jointly determinedwith remittances. This makes it difficult to estimate the causal impact of familycomposition on remittances. Instead, the analysis simply looks at whetherdifferences in remitting patterns by education level disappear when they areconditioned on family composition. Empirically, the analysis will show thatwhile less educated migrants have more relatives in the home country, they alsohave larger households and more relatives with them in the destinationcountry.
3. See Rapoport and Docquier (2006) for a comprehensive survey of the economic literature on
migrants’ remittances.
4. In this basic framework, education has no impact beyond its effect on the migrants’ income and
family size, composition, and location, and altruistic preferences are independent of education.
Bollard, McKenzie, Morten, and Rapoport 135
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
Exchange and Investment Motives
There are many situations of Pareto-improving exchanges in which remittances“buy” various types of services, such as taking care of the migrant’s assets (landand cattle, for example) or relatives (children, elderly parents) at home. Suchmotivations are generally a sign of temporary migration and signal a migrant’sintention to return. In such exchanges, there is a participation constraint deter-mined by each partner’s external options, with the exact division of the pie(or surplus) to be shared depending on each partner’s bargaining power.
How does education interact with such exchange motives? Two directionsemerge from the short discussion above: one through the effect of education onintentions to return, and another through education’s effect on threat pointsand bargaining powers.
The conventional wisdom is that migrants with higher education have lessintention (and propensity) to return than do migrants with lower education (seeFaini 2007), because they are better integrated or can obtain permanent residentstatus more easily. If that is the case, more educated migrants should transfer lessfor an exchange motive, reflecting their lower propensity to return.5 What aboutbargaining powers? Exchange models allow for different possible contractualarrangements reflecting the parties’ outside options and bargaining powers (see,for example, Cox 1987; Cox, Eser, and Jimenez 1998). This has two complemen-tary implications for education as a determinant of remittances in an exchangemodel. First, to the extent that education is associated with higher income, thisrelationship is likely to increase a migrant’s willingness to pay, leading to higherremittances; and second, to the extent that educated migrants come from moreaffluent families, this relationship is likely to increase the receiving household’sbargaining power, also leading to higher remittances. On the whole, an exchangemotive therefore predicts that education will have an ambiguous effect on remit-tances, with the sign of the effect depending on whether return intentions orbargaining issues matter more to remittance behavior.
The investment motive can be seen as a particular exchange of services in acontext of imperfect credit markets. In such a context, remittances can be seenas part of an implicit migration contract between migrant and family, allowingthe family access to higher income (investment motive) or less volatile income(insurance motive; Stark 1991). Since the insurance motive does not in theorygive rise to clear differences in transfer behavior between highly educated andless educated migrants, the focus here is on the investment motive. The amountof investment financed by the family may include the physical costs (such astransportation) and informational costs of migration, as well as educationexpenditures, and repayment of this implicit loan through remittances isobviously expected to depend on the magnitude of the loan. Thus, the
5. Again, as shown later in the article, this conventional wisdom is not supported by the data;
exchange motives are equally relevant for highly educated and less educated migrants as far as return
intentions are concerned.
136 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
investment motive clearly predicts that, all else equal, more educated migrantsshould remit more to compensate the family for the additional educationexpenditures incurred.
Summary of Predictions
Both the altruistic and the exchange motives for remittances yield uncleartheoretical predictions as to whether more educated migrants remit more orless than do less education migrants. Once migrants’ incomes are controlledfor, their education level should not play a role under the altruistic hypoth-esis (assuming preferences are exogenous to education) except for its effecton the spatial distribution of the family. As already noted, the conventionalwisdom here is that the highly educated tend to move with their immediatefamily, which would lower remittances. Similarly, education is expected tolower remittances under the exchange hypothesis if educated migrants havelower propensities to return; bargaining mechanisms work in the otherdirection and should translate into higher remittances, with the sign of thetotal expected effect being theoretically uncertain. Finally, education islikely to have a clear positive impact on remittances under the investmenthypothesis.
Given these expected mechanisms and the fact that the descriptive statisticsfor the sample do not support the conjecture that more educated migrants havea substantially higher propensity to move with their family or a substantiallylower propensity to return, the other forces at work should be expected todominate, so that migrants with more education would remit more, which isindeed what the analysis shows.
I I . D A T A
The micro-level database on remitting behavior created for this study is the mostcomprehensive available, comprising data on 33,000 immigrants from develop-ing countries derived from 14 surveys in 11 OECD destination countries thatwere the destination for 79 percent of global migrants to OECD countries in2000 (Docquier and Marfouk 2005). The focus on destination country datasources enables looking directly at the relationship between education and remit-tance sending behavior by analyzing the migrants’ decision to remit. It alsopermits capturing the remittance behavior of individuals who emigrate withtheir entire household; using household surveys from the remittance receivingcountries would typically miss such individuals. Since more educated individualsare believed to be more likely to emigrate with their entire household thanless educated individuals (Faini 2007), using surveys from migrant sendingcountries would not be appropriate for examining the relationship betweenremittances and education.
Most of the empirical literature on immigrants uses data from censuses orlabor force surveys, but neither contains information on remittances. That
Bollard, McKenzie, Morten, and Rapoport 137
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
requires special purpose surveys of immigrants. The authors pulled together allpublicly available datasets they were aware of6 and six additional surveys thatare not publicly available but that other researchers generously shared. Table 1provides an overview of the database of migrants, summarizing the datasets,sample population, and survey methodology. Full details of the source of eachdataset are in the supplemental appendix, available online at http://wber.oxfordjournals.org/. The database covers a wide range of populations. Itincludes both nationally representative surveys, such as the New ImmigrantSurvey (NIS) in the United States (drawn from green card recipients) and theSpanish National Survey of Immigrants (ENI), which draws on a neighborhoodsampling frame, as well as surveys focusing on specific migrant communitieswithin the recipient country, such as the Black/Minority Ethnic Survey (BME)in the United Kingdom and the Belgium International Remittance SendersHousehold Survey (IRSHS) of immigrants from the Democratic Republic ofCongo, Nigeria, and Senegal. In all cases, the database includes only migrantswho were born in developing countries.7
For each country dataset, comparable covariates were constructed tomeasure household income, remittance behavior, family composition, anddemographic characteristics. Remittances are typically measured at the house-hold rather than individual level. The level of analysis is therefore the house-hold, and variables are defined at this level whenever possible—for example,by taking the highest level of schooling achieved by any adult migrant in thehousehold. All financial values are reported in 2003 U.S. dollars. In addition,any reported annual remittances that are more than twice annual householdincome are dropped. While remittance data in surveys can be subject tomeasurement error, the use of survey fixed effects will capture any commonsurvey-level effects, and there is no strong reason to believe such measurementerror would be correlated with education status. Mean and median reportedremittances also seem to be of the right order of magnitude when comparedwith other surveys and migrant incomes.
The sample weights provided with the data are always used. Data are pooledby poststratifying by country of birth and by education, so that the combinedweighted observations match the distribution of developing country migrantsto all OECD countries in 2000 (Docquier and Marfouk2005). The supplemen-tal appendix provides further details.
Table 2 presents summary statistics for each country survey and the pooledsamples of all destination countries. Overall, 37 percent of migrants in thedatabase have completed a university degree, ranging from 4 percent in theSpanish Netherlands Interdisciplinary Demographic Institute (NIDI) survey to
6. Exceptions include longitudinal surveys of immigrants from Canada and New Zealand, which
can only be accessed through datalabs in these countries, and so are not included here.
7. High income countries are defined based on the World Bank Country Classification Code, April
2009.
138 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
TA
BL
E1
.M
igra
nt
Dat
ase
ts
Dat
ase
tN
am
eY
ear
Num
ber
of
obse
rvat
ionsa
Popula
tion
Met
hodolo
gy
Aust
ralia
LSIA
Longit
udin
al
Surv
eyof
Imm
igra
nts
toA
ust
rali
a1997
2,5
37
Pri
mary
applica
nt
mig
rant
arr
ivals
Sep
tem
ber
1993
–A
ugust
1995
Sam
ple
of
offi
cial
reco
rds
of
those
livin
gin
citi
esB
elgiu
mIR
SH
SIn
tern
atio
nal
Rem
itta
nce
Sen
der
sH
ouse
hold
Surv
ey2005
377
Imm
igra
nts
from
DR
Congo,
Nig
eria
,and
Sen
egal
Ref
erra
lsth
rough
Em
bass
ies.
Fra
nce
2M
OSurv
eyof
House
hold
s’T
ransf
erof
Funds
toth
eir
Countr
ies
of
Ori
gin
2007
713
Rem
itte
rsto
Alg
eria
,M
oro
cco,
Tunis
ia,
Turk
eyand
the
countr
ies
of
Sub-S
ahar
an
Afr
ica
Inte
rvie
ws
of
rem
itte
rsat
post
offi
ces
inhig
h-m
igra
nt
regio
ns
Fra
nce
DR
EE
SPro
file
and
Tra
ckin
gof
Mig
rants
Surv
ey2006
4,2
78
New
and
regula
rize
dm
igra
nts
wit
h1þ
year
resi
den
ceper
mit
sSam
ple
of
offi
cial
reco
rds
Ger
many
SO
EP
Ger
man
Soci
o-E
conom
icPanel
Stu
dy
2000
854
Res
iden
tpopula
tion
of
the
Fed
eral
Rep
ublic
of
Ger
many
in1984.
Sam
ple
of
offi
cial
reco
rds
Italy
NID
IN
ether
lands
Inte
rdis
cipli
nary
Dem
ogra
phic
Inst
itute
Inte
rnat
ional
Mig
rati
on
Surv
ey
1997
1,0
72
Egypti
ans
and
Ghanaia
ns
who
imm
igra
ted
wit
hin
past
10
years
Inte
rvie
ws
atm
igra
nt
mee
ting
pla
ces
Japan
IDB
Surv
eyof
Bra
zili
ans
and
Per
uvia
ns
inJa
pan
2005
846
Lat
inA
mer
ican
imm
igra
nt
adult
slivin
gin
Japan
Inte
rvie
ws
in15
citi
es
Net
her
lands
CSR
Consu
men
tenbond
Surv
eyof
Rem
itta
nce
s2005
648
Majo
rim
mig
rant
popula
tions:
Moro
ccans,
Turk
s,Suri
nam
ese,
Anti
llea
ns,
Som
alis,
and
Ghanaia
ns
Fac
e-to
-fac
ein
terv
iew
s
Norw
ayL
KI
Liv
ing
Condit
ions
of
Imm
igra
nts
Surv
ey1996
2,4
66
Imm
igra
nts
from
10
countr
ies:
Bosn
iaand
Her
zegovin
a,
Chile,
Iraq,
Iran,
Pakis
tan,
Ser
bia
,Som
alia,
Sri
Lanka,
Turk
ey,
and
Vie
tnam
Rep
rese
nta
tive
surv
eyof
imm
igra
nt
popula
tion
from
thes
eco
untr
ies
Spain
EN
IN
atio
nal
Surv
eyof
Imm
igra
nts
2006
9,2
34
Fore
ign-b
orn
who
(inte
nd
to)
live
inSpain
for
1þ
years
Sam
ple
of
offi
cial
nei
ghborh
ood
rost
ers
(Conti
nued
)
Bollard, McKenzie, Morten, and Rapoport 139
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
TA
BL
E1.
Conti
nued
Dat
ase
tN
am
eY
ear
Num
ber
of
obse
rvat
ionsa
Popula
tion
Met
hodolo
gy
Spain
NID
IN
ether
lands
Inte
rdis
cipli
nary
Dem
ogra
phic
Inst
itute
Inte
rnat
ional
Mig
rati
on
Surv
ey
1997
1,0
20
Moro
ccans
and
Sen
egale
sew
ho
imm
igra
ted
wit
hin
past
10
years
Geo
gra
phic
al
sam
pling,
&re
fere
nce
sfr
om
sam
ple
d
UK
BM
EB
lack
/Min
ori
tyE
thnic
Rem
itta
nce
Surv
ey2006
993
Mig
rant
min
ori
ties
who
hav
ere
mit
ted
inpast
12
month
sSam
pling
of
geo
gra
phic
al
blo
cks
U.S
.N
ISN
ewIm
mig
rant
Surv
ey2003
7,0
46
Mig
rants
rece
ivin
ggre
enca
rds
May
–N
ove
mber
1993
Sam
ple
of
offi
cial
reco
rds
Pew
Nat
ional
Surv
eyof
Lat
inos
2006
1084
Nat
ionally
repre
senta
tive
sam
ple
of
Lat
ino
resp
onden
tsages
18
and
old
er
Sam
ple
dphone
num
ber
sin
hig
h-L
atin
oare
as
U.S
.Pew
Pew
Nat
ional
Surv
eyof
Lat
inos
2006
1,0
84
Nat
ionally
repre
senta
tive
sam
ple
of
Lat
ino
resp
onden
tsages
18
and
old
er
Sam
ple
dphone
num
ber
sin
hig
h-L
atin
oare
as
Note
:N
um
ber
of
obse
rvat
ions
use
dto
calc
ula
teth
efirs
tre
sult
inea
chco
lum
nof
table
2.
140 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
TA
BL
E2
.Surv
eyM
eans,
by
Educa
tion
Aust
rali
aB
elgiu
mFra
nce
Fra
nce
Ger
many
Italy
Japan
Net
her
lands
Norw
aySpain
Spain
UK
U.S
.U
.S.
Poole
dPoole
dPoole
d
Vari
able
and
educa
tion
leve
lL
SIA
IRSH
S2M
OD
RE
ES
SO
EP
NID
IID
BC
SR
LB
IE
NI
NID
IB
ME
NIS
Pew
Exte
nsi
veIn
tensi
veT
ota
l
Num
ber
of
obse
rvat
ions
2,6
56
451
717
4,2
80
900
1,1
53
1,0
65
836
2,4
66
10,2
82
1,1
13
1,1
52
7,3
52
1,3
04
33,0
22
28,9
81
26,2
76
Fra
ctio
nw
ith
univ
ersi
tyed
uca
tion
0.3
20.5
90.0
70.1
80.2
00.2
10.1
40.1
10.1
20.2
30.0
40.2
00.3
40.0
60.3
60.3
70.3
7
Tota
lre
mit
tance
s($
per
year)
No
univ
ersi
ty286
1,6
81
1,3
80
368
2,7
24
2,6
62
993
988
3,0
99
2,6
91
375
1,5
30
699
793
650
Univ
ersi
ty379
2,4
75*
1,6
52
511
2,2
27
2,9
20
1,4
05*
743**
2,8
35
2,6
29
1,1
45**
671**
868
897
874
Fra
ctio
nw
ho
rem
it
No
univ
ersi
ty0.4
10.9
10.2
30.1
80.6
00.8
00.3
40.4
90.7
80.1
50.5
40.3
10.3
20.3
2
Univ
ersi
ty0.3
70.8
60.2
30.2
00.4
5**
0.9
0**
0.2
90.3
7**
0.4
8**
0.1
70.4
30.2
7**
0.2
7**
0.2
7**
Log
rem
itta
nce
s
No
univ
ersi
ty5.7
86.9
26.6
26.9
77.8
97.7
66.4
97.1
57.9
96.7
77.0
17.3
46.9
66.8
26.9
1
Univ
ersi
ty6.2
3**
7.2
9**
6.9
27.0
18.1
17.7
06.8
1**
7.2
28.4
9*
6.9
27.4
0**
6.9
77.0
26.9
7*
7.0
0
House
hold
inco
me
($per
year)
No
univ
ersi
ty14,4
57
16,9
18
23,1
73
18,6
12
19,5
26
10,9
03
34,0
14
32,4
67
14,0
66
9,0
74
44,6
31
33,2
97
22,4
17
22,6
24
23,5
83
21,9
64
Univ
ersi
ty13,5
56
25,5
34**
31,3
01*
28,6
74**
21,9
84
13,3
02*
43,6
24**
41,9
95**
19,9
14**
10,1
68
50,5
65
61,0
84
34,7
29**
38,9
48**
38,6
69**
39,0
87**
Log
inco
me
No
univ
ersi
ty9.5
9.5
9.8
9.6
9.8
9.3
10.2
10.1
9.4
9.0
10.3
9.2
9.7
9.6
9.5
9.5
Univ
ersi
ty9.8
**
9.8
**
10.0
9.9
**
9.8
9.4
10.4
10.3
**
9.7
**
9.2
10.4
10.0
**
10.2
**
9.9
**
9.9
**
9.9
**
Work
ing
No
univ
ersi
ty0.4
80.7
00.8
70.8
00.6
30.8
20.9
30.4
80.6
80.8
10.8
20.6
60.6
60.6
50.6
60.6
4
Univ
ersi
ty0.6
7**
0.7
40.8
60.8
6**
0.6
70.8
70.9
30.7
0**
0.7
3**
0.6
60.9
0**
0.7
8**
0.7
7*
0.7
5**
0.7
4**
0.7
3**
House
hold
size
No
univ
ersi
ty3.8
11.8
82.5
12.9
01.8
01.5
33.8
21.8
43.3
34.1
03.4
43.7
33.7
6
Univ
ersi
ty3.4
4**
2.5
5**
1.9
0**
2.5
82.1
6**
1.7
6**
3.1
9**
1.9
53.0
4*
3.4
9**
3.1
7**
3.3
5**
3.3
6**
Marr
ied
No
univ
ersi
ty0.7
30.7
20.6
50.6
70.6
10.5
60.4
70.6
40.6
60.5
40.6
30.6
30.6
3
Univ
ersi
ty0.8
0**
0.5
1**
0.7
1*
0.5
90.6
00.4
8*
0.5
6**
0.5
10.8
6**
0.5
60.7
3**
0.7
4**
0.7
4**
Spouse
outs
ide
countr
y
No
univ
ersi
ty0.0
30.2
50.0
50.0
60.4
20.0
50.0
50.0
50.0
6
Univ
ersi
ty0.0
1*
0.1
90.0
1**
0.0
50.1
0**
0.0
3**
0.0
3**
0.0
3**
0.0
3**
Num
ber
of
chil
dre
n
No
univ
ersi
ty1.2
91.1
61.7
81.0
62.5
02.0
61.5
82.2
52.3
71.9
92.0
52.0
3
Univ
ersi
ty1.2
20.8
9**
1.2
7**
1.0
02.1
5**
1.8
5**
0.6
2**
1.3
5**
1.8
1**
1.3
7**
1.3
7**
1.3
7**
Chil
dre
nouts
ide
countr
y
No
univ
ersi
ty0.2
10.1
00.2
50.7
10.2
00.1
60.3
81.1
00.7
30.4
90.4
50.4
80.5
0
Univ
ersi
ty0.0
7**
0.0
60.1
7**
0.4
9*
0.1
50.0
90.2
6**
0.2
1**
0.3
1**
0.3
70.2
4**
0.2
5**
0.2
5**
Num
ber
of
pare
nts
No
univ
ersi
ty1.9
71.1
30.9
51.3
51.4
21.2
72.1
81.8
11.8
41.8
3
(Conti
nued
)
Bollard, McKenzie, Morten, and Rapoport 141
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
TA
BL
E2.
Conti
nued
Aust
rali
aB
elgiu
mFra
nce
Fra
nce
Ger
many
Italy
Japan
Net
her
lands
Norw
aySpain
Spain
UK
U.S
.U
.S.
Poole
dPoole
dPoole
d
Vari
able
and
educa
tion
leve
lL
SIA
IRSH
S2M
OD
RE
ES
SO
EP
NID
IID
BC
SR
LB
IE
NI
NID
IB
ME
NIS
Pew
Exte
nsi
veIn
tensi
veT
ota
l
Univ
ersi
ty2.3
2**
1.0
30.7
0**
1.3
21.3
5**
1.3
72.7
4**
2.1
8**
2.2
1**
2.2
3**
Pare
nts
outs
ide
countr
y
No
univ
ersi
ty1.4
80.8
10.4
20.9
41.0
31.0
11.2
30.8
80.9
80.9
81.0
0
Univ
ersi
ty2.0
0**
0.8
80.5
40.6
7**
1.1
7*
1.0
41.3
31.2
6**
1.3
0**
1.3
1**
1.3
1**
Yea
rssp
ent
abro
ad
No
univ
ersi
ty3.7
09.3
217.9
04.0
019.2
06.6
98.3
518.4
610.0
67.2
714.8
97.3
516.4
39.2
011.1
710.2
9
Univ
ersi
ty3.9
1**
12.2
8**
12.7
0**
4.2
113.5
1**
7.0
29.1
819.3
612.4
1**
6.7
414.6
67.0
518.3
48.0
6**
8.7
5**
8.4
0**
Leg
al
imm
igra
nt
No
univ
ersi
ty1.0
01.0
00.8
40.5
10.6
61.0
00.8
70.8
40.8
5
Univ
ersi
ty1.0
01.0
00.8
50.3
9**
0.8
2*
1.0
00.8
5**
0.8
40.8
4
Wil
lre
turn
hom
e
No
univ
ersi
ty0.0
20.4
50.0
60.2
30.3
90.0
10.0
80.3
50.6
30.0
90.1
90.0
90.1
60.1
1
Univ
ersi
ty0.0
40.6
5**
0.1
0*
0.1
70.5
3**
0.0
20.0
80.5
10.7
00.1
3**
0.1
40.0
90.1
2**
0.0
9*
*Sig
nifi
cant
atth
e5
per
cent
leve
l**
signifi
cant
atth
e1
per
cent
leve
l
Note
:Sig
nifi
cant
resu
lts
indic
ate
that
the
mea
nof
the
vari
able
isst
atis
tica
lly
dif
fere
nt
bet
wee
nuniv
ersi
ty-e
duca
ted
and
non-u
niv
ersi
tyed
uca
ted
house
-hold
s.See
table
1fo
rfu
llsu
rvey
nam
es.
Sourc
e:A
uth
ors
’analy
sis
base
don
dat
ades
crib
edin
the
text.
142 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
59 percent in the Belgium IRSHS. The table also summarizes the covariates bythe maximum educational attainment of all adult migrants in the household.Altogether, including both the extensive and intensive margins, more highlyeducated migrants send home an average of $874 annually, compared with$650 for less educated migrants. There are two opposing effects of education:negative on the extensive margin, and positive on the intensive margin. At theextensive margin, migrants with a university degree are less likely to remit any-thing than migrants without a degree: 32 percent of low-skilled migrants sendsome money home, compared with 27 percent of university-educated migrants.However, conditional on remitting (the intensive margin), highly educatedmigrants send about 9 percent more than do less educated migrants.
Characteristics that can affect remittance behavior differ between less andmore educated migrants. First, more skilled migrants are both more likely tolive in a household with working adults and to have a higher householdincome than are low skilled migrants. But contrary to conventional wisdom,household composition does not differ much for migrants by education level:on average, only 6 percent of low skilled migrants have a spouse outside thecountry, compared with 3 percent of high skilled migrants. Low skilledmigrants are significantly less likely to be married (63 percent) than are highskilled migrants (74 percent). Low skilled migrants have more children(an average of 2.03 compared with 1.37 for high skilled migrants), as well asmore children living outside the destination country (0.50) than do high skilledmigrants (0.25). However, low skilled migrants also have more family insidethe destination country than do high skilled migrants: the average householdsize for low skilled migrants is 3.76 people, statistically different from themean household size of 3.36 people for high skilled migrants.8
Another piece of conventional wisdom, that more educated people are lesslikely to return home, is also not supported by the microdata. Indeed, moreeducated migrants have spent less time abroad (mean of 8.4 years) than haveless educated migrants (10.3 years). Reported plans to return home are similarbetween the two groups: 9 percent of high skilled migrants report planning toreturn home, compared with 11 percent of low skilled migrants. While oneshould be cautious with treating both measures as truly reflecting return prob-abilities; at the very least, they do not indicate a strong tendency for the lowskilled to be more likely to return.
The simple comparison of means in table 2 shows differences in remittancebehavior by education status. However, these comparisons show only thatmore educated developing country emigrants remit more than less educateddeveloping country emigrants. This risks confounding differences in remittance
8. In some cases this might reflect households in which poorer, less skilled migrants live with other
immigrants who are not family members. The database can identify the presence of a spouse, child, or
parent in the home country household but cannot identify who migrants live with abroad or the extent
to which they share resources within the household abroad.
Bollard, McKenzie, Morten, and Rapoport 143
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
behavior among migrants from different countries with differences inremittance behavior by education level: the next section aims to separate thesetwo differences.
I I I . R E S U L T S : T H E R E L A T I O N S H I P B E T W E E N E D U C A T I O N A N D
R E M I T T A N C E S
Results are reported for regressions of three remittance measures on education:total remittances (both extensive and intensive margins), an indicator forhaving remitted in the previous year (extensive margin), and log total remit-tances conditional on remitting (intensive margin; table 3). All regressionsinclude country of birth fixed effects and dataset fixed effects.
The key result is that more educated migrants remit more. In the pooledsample, migrants with a university degree remit $298 more per year than non-university educated migrants (row last, last column), with a mean annual remit-tance for all migrants of $734. This overall effect is composed of a negative(statistically insignificant) effect at the extensive margin and a highly significantpositive effect at the intensive margin. The results are consistent when thesecond measure of education, years of schooling, is considered.
Results for individual countries are mixed at the extensive margin, with edu-cation significantly positively associated with the likelihood of remitting in twosurveys (the U.S. NIS and the Survey of Brazilians and Peruvians in Japan), sig-nificantly negatively associated with this likelihood in three surveys (the U.S.Pew survey and both Spanish surveys), and no significant relationship in theother six surveys, with three positive and three negative point estimates. Onegeneral observation is that a more negative relationship appears in surveys thatfocus on sampling migrants through community-sampling methods, such as theNIDI surveys, which take their sample from places where migrants cluster, andthe Pew Hispanic surveys, which randomly dial phone numbers in areas withdense Hispanic populations. One might expect that educated migrants who livein such areas (and who take the time to respond to phone or on-the-streetsurveys) would be less successful than educated migrants who live in more inte-grated neighborhoods and thus who would not be picked up in these surveys.
In contrast, at the intensive margin, 10 of 12 individual surveys show a posi-tive relationship between remittances and education, 5 of them statistically sig-nificant, and 2 show a negative and insignificant relationship. Thus it is notsurprising that when the data are pooled there is a strong positive associationat the intensive margin and that it outweighs the small negative and insignifi-cant relationship at the extensive margin in the total effect.
This point is made graphically on a log scale in figure 1, which plots thenonparametric relationship between total remittances and years of schooling,after linearly controlling for dataset fixed effects using a partial linear model(Robinson 1988), together with a 95 percent confidence interval. The verticallines demarcate the quartiles of years of schooling. Average remittances steadily
144 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
TA
BL
E3
:C
oef
fici
ents
from
Reg
ress
ions
of
Rem
itta
nce
Mea
sure
son
Educa
tion
Aust
rali
aB
elgiu
mFra
nce
Fra
nce
Ger
many
Italy
Japan
Net
her
lands
Norw
aySpain
Spain
UK
U.S
.U
.S.
Poole
dPoole
dPoole
d
Vari
able
LSIA
IRSH
S2M
OD
RE
ES
SO
EP
NID
IID
BC
SR
LB
IE
NI
NID
IB
ME
NIS
Pew
Exte
nsi
veIn
tensi
veT
ota
l
Educa
tion
mea
sure
dby
univ
ersi
tydeg
ree
Tota
lre
mit
tance
s($
per
year)
58.4
922.8
**
291.0
-526.6
237.5
-92.6
-168.8
769.5
**
-554.0
*298.0
*
Num
ber
of
obse
rvat
ions
2,5
37
377
854
1,0
72
846
9,2
34
1,0
20
7,0
46
1,0
84
24,0
33
Exte
nsi
ve:
Rem
its
indic
ator
20.0
19
-0.0
55
0.0
14
0.0
42
-0.0
65
0.0
91**
0.0
12
-0.0
49**
-0.2
32**
0.0
38**
-0.1
40*
-0.0
18
-0.0
10
Num
ber
of
obse
rvat
ions
2,6
54
451
4,2
78
854
1,1
53
1,0
30
2,4
66
10,2
82
1,1
12
7,1
13
1,2
96
32,6
51
25,9
07
Inte
nsi
ve:
Log
rem
itta
nce
s0.3
41*
0.4
33**
0.3
63
0.4
92
0.0
73
-0.0
57
0.3
33**
0.0
93
0.4
30*
0.1
68
0.3
97*
-0.1
99
0.2
49**
0.2
26**
Num
ber
of
obse
rvat
ions
958
317
713
184
545
690
648
3,9
66
761
993
1,1
18
514
11,3
92
9,0
38
Educa
tion
mea
sure
din
years
Tota
lre
mit
tance
s($
per
year)
19.0
8*
86.5
026.3
9-7
.56
-3.0
32.4
0-1
3.6
586.5
364.8
957.8
1
Num
ber
of
obse
rvat
ions
2,5
31
377
854
1,0
72
846
9,1
64
1,0
20
7,0
33
1,0
84
23,9
44
Exte
nsi
ve:
Rem
its
indic
ator
0.0
080
-0.0
042
0.0
018
0.0
145
0.0
010
0.0
024**
0.0
008
-0.0
023
-0.0
072**
0.0
034**
0.0
010
0.0
006
0.0
014
Num
ber
of
obse
rvat
ions
2,6
48
451
5,5
29
854
1,1
53
1,0
30
2,4
50
10,2
01
1,1
12
7,1
00
1,2
96
32,5
35
25,8
07
Inte
nsi
ve:
Log
rem
itta
nce
s0.0
441*
0.0
341
0.0
224*
-0.0
085
-0.0
032
-0.0
040
0.0
247*
0.0
199**
0.0
091
0.0
548*
0.0
329*
0.0
369
0.0
256**
0.0
229**
Num
ber
of
obse
rvat
ions
956
317
713
184
545
690
648
3,9
42
761
993
1,1
16
514
11,3
64
9,0
10
Mea
ns
Tota
lre
mit
tance
s($
per
year)
316
2,1
59
1,3
99
396
2,6
21
2,6
92
1,0
40
932
3,0
89
2,6
79
633
1,4
79
764
2,4
66
734
Fra
ctio
nw
ho
rem
it0.4
00.8
50.2
30.1
90.5
30.7
70.3
40.4
10.7
50.1
50.4
60.3
01.0
00.2
7
Fra
ctio
nw
ith
univ
ersi
ty0.3
20.6
00.0
70.1
80.2
00.2
10.1
20.1
10.1
20.2
30.0
40.2
00.3
30.0
60.3
60.3
10.3
8
Yea
rsof
educa
tion
13.4
14.2
7.7
12.0
11.5
14.1
13.3
10.7
12.2
11.4
7.5
13.4
13.4
9.4
12.9
12.3
13.0
*Sig
nifi
cant
atth
e5
per
cent
leve
l**
signifi
cant
atth
e1
per
cent
leve
lN
ote
:See
table
1fo
rfu
llsu
rvey
nam
es.
Sourc
e:A
uth
ors
’analy
sis
base
don
dat
ades
crib
edin
the
text.
Bollard, McKenzie, Morten, and Rapoport 145
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
increase from around $500 in the lowest education quartile to close to $1,000for those with university degrees. Moreover, the positive association increasesmost strongly for migrants with postgraduate education, which shows that notonly do migrants with some university education remit more than thosewithout, but also that migrants with postgraduate degrees remit more thanthose with only a couple of years of university.
Robustness
Although this database on remittances is the most comprehensive available,there are clear limitations, which make it important to see how sensitive theresults are to alternative ways of using these surveys.
First note that the results pertain only to migration in a sample of OECDcountries. The surveys cover a large share of OECD destinations, but they omitother important destinations for developing country migrants such as the Gulfcountries and South Africa. This is a limitation shared by the macro studies(Faini 2007; Niimi and others 2008), which also have data only for migrantsin OECD countries. Nevertheless, the same forces acting on migrants in theOECD countries are likely to apply in these other destinations: more educatedmigrants will earn higher incomes and therefore remit more. Although data arerare, there is some evidence to support this is in a study of Pakistani migrantsin the Gulf countries, which found that conditional on age and duration of
Figure 1. Total Remittances by Years of Schooling
Note: Figure depicts a semiparametric regression line from a partial linear model with datasetdummy variable evaluated at means; 95 percent pointwise confidence intervals shown from 500bootstrap repetitions. Vertical lines separate quartiles.
Source: Authors’ analysis based on data described in the text.
146 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
stay, more educated Pakistani migrants remitted more (Abbasi and Hashmi2000).
Moreover, it is still the case that there are a large number of low skilledmigrants in the OECD. A large majority of migrants in the pooled sample(63 percent) do not have a university education. A reasonable concern iswhether surveys like the U.S. NIS, which capture only legal immigrants, aremissing most of the low-skilled migrants. Comparing the skill distribution ofimmigrants included in the NIS with that of immigrants included in the U.S.Census (which is generally believed to do a good job surveying both legal andillegal migrants), does show a higher skill level in the NIS (12.26 mean years ofeducation) than in the Census (10.84 years). However, once Mexican immi-grants are excluded (the group with the largest number of illegal immigrants),the skill distributions of the NIS (12.96 mean years of education) and theCensus (12.21) are much closer, and16 percent of immigrants in both the NISand the Census have 8 years of education or less. The first two columns oftable 4 then show that the results for the association between remittances andeducation continue to hold in the NIS (and, if anything, are more strongly posi-tive) when Mexican immigrants are excluded (table 4, columns 1 and 2).Columns 3 and 4 show that this is also true for the pooled sample of allsurveys, which suggests that failure to capture illegal migrants in the survey isnot driving the main result.
A second potential concern is whether it is valid to pool so many differentsurveys with different sampling methods and differing degrees of representa-tiveness. Note that survey fixed effects are included in the regression analysis,so that only within-survey variation is used to identify the effect of education;the pooled estimate is thus a consistent estimate for the average associationamong the surveys. Nonetheless, as an alternative, the regressions are run onlyfor the five surveys based on representative sampling from a list of migrants:(Longitudinal Survey of Immigrants to Australia (LSIA), the French Profile andTracking of Migrants Survey (DREES), the German Socioeconomic PanelStudy (SOEP), and the Spanish ENI and the NIS). The results show point esti-mates and levels of statistical significance that are very close to those for thefull pooled sample (see table 4, column 5). This demonstrates that the resultsare not being driven by the specialized surveys of particular migrant groups,such as the Japanese and Belgian surveys.
Finally, one might query whether the results are being driven by students.That could influence the results based on university education if there weremany students studying for undergraduate degrees who do not send remittancesand do not yet have a college degree. There are several reasons to believe thatthis is not the main factor driving results. First, the LSIA and NIS surveys donot include students, which eliminates from the sample students in thecountries that are among the most popular destinations for international study.Second, many international students come for postgraduate education, so theywould be classified as having a college education and remitting little, which
Bollard, McKenzie, Morten, and Rapoport 147
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
TA
BL
E4
.R
obust
nes
sC
hec
ks U
.S.
New
Imm
igra
nt
Surv
ey(N
IS)
sam
ple
Poole
dsa
mple
Vari
able
Full
sam
ple
Excl
udin
gM
exic
ans
Full
sam
ple
Excl
udin
gN
ISM
exic
ans
Nat
ionally
repre
senta
tive
sam
ple
sO
nly
mig
rants
ages
25þ
Only
work
ing
mig
rants
Educa
tion
mea
sure
dby
univ
ersi
tydeg
ree
Tota
lre
mit
tance
s($
per
year)
769.5
**
839.7
**
298.0
*306.8
*318.5
*267.7
308.7
Num
ber
of
obse
rvat
ions
7,0
46
5,9
22
24,0
33
22,9
09
19,6
43
21,3
43
16,6
93
Exte
nsi
ve:
Rem
its
indic
ator
0.0
38**
0.0
46**
-0.0
10
0.0
00
-0.0
11
-0.0
23*
0.0
48**
Num
ber
of
obse
rvat
ions
7,1
13
5,9
84
25,9
07
24,7
78
20,8
75
23,0
43
18,1
47
Inte
nsi
ve:
Log
rem
itta
nce
s0.3
97*
0.4
58**
0.2
26**
0.2
36**
0.2
20**
0.1
92**
0.2
30**
Num
ber
of
obse
rvat
ions
1,1
18
982
9,0
38
8,9
02
6,2
20
8,3
03
7,3
60
Educa
tion
mea
sure
din
years
Tota
lre
mit
tance
s($
per
year)
86.5
3108.0
957.8
168.1
762.5
258.7
878.4
9N
um
ber
of
obse
rvat
ions
7,0
33
5,9
09
23,9
44
22,8
20
19,5
54
21,2
63
16,6
39
Exte
nsi
ve:
Rem
its
indic
ator
0.0
034**
0.0
021
0.0
014
0.0
016
0.0
017
0.0
007
0.0
031*
Num
ber
of
obse
rvat
ions
7,1
00
5,9
71
25,8
07
24,6
78
20,7
75
22,9
54
18,0
83
Inte
nsi
ve:
Log
rem
itta
nce
s0.0
329*
0.0
391*
0.0
229**
0.0
260**
0.0
307**
0.0
235**
0.0
242**
Num
ber
of
obse
rvat
ions
1,1
16
980
9,0
10
8,8
74
6,1
92
8,2
79
7,3
35
Mea
ns
Tota
lre
mit
tance
s($
per
year)
633
719
734
813
614
772
935
Fra
ctio
nw
ho
rem
it0.1
50.1
60.2
70.2
90.2
40.2
80.3
3Fra
ctio
nw
ith
univ
ersi
ty0.3
30.4
00.3
80.3
80.3
80.3
80.3
8Y
ears
of
educa
tion
13.4
14.0
13.0
13.1
13.1
13.0
13.4
*Sig
nifi
cant
atth
e5
per
cent
leve
l**
signifi
cant
atth
e1
per
cent
leve
l.
Note
:N
atio
nall
yre
pre
senta
tive
surv
eys
are
Aust
ralia
LSIA
,Fra
nce
DR
EE
S,
Ger
many
SO
EP,
Spain
EN
Iand
U.S
.N
IS.
See
table
1fo
rfu
llsu
rvey
nam
es.
Sourc
e:A
uth
ors
’analy
sis
base
don
dat
ades
crib
edin
the
text.
148 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
would offset any effect of undergraduates.9 As a final check, the analysis isrestricted to individuals who are working (table 4, last column). Since moreeducated individuals are more likely to be working, this eliminates one channelthrough which the more educated can earn more and thereby remit more.Nevertheless, even with this restriction, there is a significant positive coefficientat both the extensive and intensive margins, and the point estimate for totalremittances is similar in magnitude, although it is not statistically significant.
Taken together, these results indicate that the basic finding of a positiverelationship between total remittances and education appears reasonablyrobust to alternative ways of combining the surveys.
Channels
This section uses these microdata to explore some of the channels throughwhich education might influence remittances. Proxies are added to the modelto control for differences in household income and work status, in householddemographics and the presence of family abroad, in time spent abroad, in legalstatus, and in intentions to return home.
Table 5 shows the results of adding this full set of variables to the pooledmodel, using years of education as the measure of educational attainment. Thesechannels operate as theory would predict. Households with more income andwith adults who work more are more likely to remit: households where a migrantmember is working send $345 more annually, with an extra $38 remittedannually for each 10 percent increase in income. As expected, family compositionvariables are also strongly significant both overall and for the extensive and inten-sive margins: a spouse outside the country is associated with a colossal additional$1,120 remitted each year, approximately one and a half times the mean annualremittance for all migrants. Each child living outside the destination country isassociated with an additional $340 remitted annually and each parent for anadditional $180. Residing in the destination country legally is associated with anadditional $400 annually, providing no evidence that legal migrants lose theirdesire to remain in contact with their home country. Migrants who plan to moveback home also remit significantly more, but this effect is primarily through theextensive margin rather than the intensive margin.
Which channels account for the association between education and remit-tance behavior? Tables 6, 7, and 8 report how the coefficient on education inan ordinary least squares regression changes as controls are added for totalremittances, the extensive margin, and the intensive margin. Each panel of eachtable first shows the baseline education coefficient from regressing remittancesonly on education and country of birth and dataset fixed effects (from table 3).Each succeeding row then shows changes in this coefficient when controls are
9. In the United States, 47 percent of international students are studying for postgraduate degrees,
compared with 12 percent for associate degrees and 32 percent for bachelor degrees (http://opendoors
.iienetwork.org/?p=150827).
Bollard, McKenzie, Morten, and Rapoport 149
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
added for income and work status, family composition, and all controls fromtable 5 (income and family controls, as well as legal status, time spent abroad,and intent to return home).
Remittance behavior is accounted for primarily by income and not by differ-ences in family composition. The baseline result for total remittances fromtable 3, controlling only for country of birth and dataset fixed effects, is thatmigrants with a university degree remit $300 more than migrants without one.Controlling for the full set of covariates (the all row) reduces the coefficient onuniversity degree by two-thirds, and it becomes statistically insignificant. Thethird row adds just the family composition variables to the baseline
TA B L E 5. Remittances on Years of Education for Pooled Sample with AllControls
Total Extensive IntensiveVariable remittances Remits Log remittances
Years of education 37.81 20.002* 0.017**(29.64) (0.001) (0.005)
Log income 384.59** 0.023** 0.364**(105.37) (0.003) (0.034)
Working 345.06** 0.113** 0.514**(90.80) (0.010) (0.065)
Household size 28.14 20.002 0.015(17.67) (0.002) (0.016)
Married 289.77 0.004 20.097(68.78) (0.010) (0.061)
Spouse outside country 1,120.95** 0.145** 0.568**(236.04) (0.020) (0.097)
Number of children 2121.56** 20.006 20.099**(36.44) (0.003) (0.027)
Children outside country 337.78** 0.048** 0.228**(75.14) (0.006) (0.039)
Number of parents 247.07 20.020** 20.125**(53.56) (0.005) (0.045)
Parents outside country 182.58** 0.063** 0.243**(38.02) (0.006) (0.045)
Years spent abroad / 100 2,539.77 0.251** 1.744**(2,533.08) (0.095) (0.656)
Years spent abroad squared / 100 231.43 20.010** 20.033*(27.14) (0.002) (0.015)
Legal immigrant 398.79** 0.096** 0.167**(121.36) (0.018) (0.061)
Will return home 692.30** 0.095** 0.085(201.83) (0.021) (0.072)
Number of observations 23,944 32,535 11,364
*Significant at the 5 percent level ** significant at the 1 percent level.
Note: Includes dummy variables for missing covariates and fixed effects for country of birthand survey. Trimmed remittances greater than twice income. Pooled samples poststratified bycountry and education.
Source: Authors’ analysis based on data described in the text.
150 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
TA
BL
E6
.E
duca
tion
Coef
fici
ent
as
Contr
ols
Are
Added
:T
ota
lA
nnual
Rem
itta
nce
s(U
.S.
dollar
s)
Aust
rali
aB
elgiu
mG
erm
any
Italy
Japan
Spain
Spain
U.S
.U
.S.
Poole
dV
ari
able
LSIA
IRSH
SSO
EP
NID
IID
BE
NI
NID
IN
ISPew
tota
l
Univ
ersi
tyed
uca
tion
Base
line
58.4
922.8
**
291.0
2526.6
237.5
292.6
2168.8
769.5
**
2554.0
*298.0
*(6
1.1
)(3
51.4
)(2
75.6
)(4
11.6
)(3
74.1
)(6
2.8
)(7
49.4
)(2
54.4
)(2
27.2
)(1
37.6
)In
com
e2
10.1
557.0
*238.5
2623.9
166.5
2189.3
**
24.7
396.6
*2
741.5
**
102.3
(62.4
)(2
81.4
)(2
62.2
)(4
07.2
)(3
59.8
)(6
3.5
)(7
29.0
)(1
74.4
)(2
63.8
)(9
2.8
)Fam
ilya
29.8
534.7
237.8
2306.7
317.5
2112.8
26.9
623.6
**
2698.6
**
228.2
*(6
1.4
)(3
10.5
)(2
43.5
)(3
94.7
)(3
80.3
)(5
7.6
)(7
25.9
)(2
04.7
)(2
41.9
)(1
03.1
)A
ll2
16.5
475.8
144.6
2539.6
328.7
2181.7
**
266.2
402.2
**
2835.7
**
99.9
(62.1
)(2
72.7
)(1
79.8
)(3
83.3
)(3
65.3
)(5
8.6
)(6
98.6
)(1
54.3
)(2
69.9
)(7
1.6
)N
um
ber
of
obse
rvat
ions
2,5
37
377
854
1,0
72
846
9,2
34
1,0
20
7,0
46
1,0
84
24,0
33
Yea
rsof
educa
tion
Base
line
19.0
8*
86.5
026.3
92
7.5
62
3.0
32.4
02
13.6
586.5
364.8
957.8
1(9
.01)
(45.1
1)
(29.3
7)
(34.0
5)
(7.9
2)
(7.3
6)
(19.9
5)
(46.5
0)
(44.9
7)
(37.0
8)
Inco
me
7.9
947.8
03.5
12
32.4
42
2.5
92
13.3
92
26.9
544.9
849.1
832.1
2(8
.69)
(38.2
8)
(27.3
3)
(33.3
9)
(11.5
0)
(7.4
1)
(19.6
8)
(40.0
0)
(45.0
9)
(31.9
8)
Fam
ilya
17.0
329.2
825.5
647.3
12
1.8
63.9
310.3
280.7
847.9
555.4
3(8
.98)
(38.4
5)
(27.7
9)
(34.9
3)
(8.6
2)
(6.8
4)
(19.9
8)
(44.7
5)
(46.3
7)
(34.2
4)
All
8.8
633.7
79.6
622.6
41.9
92
7.5
74.5
054.8
127.0
137.8
1(8
.91)
(36.9
4)
(22.8
2)
(32.7
9)
(10.6
3)
(6.8
4)
(19.3
2)
(37.3
2)
(46.3
8)
(29.6
4)
Num
ber
of
obse
rvat
ions
2,5
31
377
854
1,0
72
846
9,1
64
1,0
20
7,0
33
1,0
84
23,9
44
*Sig
nifi
cant
atth
e5
per
cent
leve
l**
signifi
cant
atth
e1
per
cent
leve
l.
Note
:B
ase
line
row
incl
udes
only
countr
yof
bir
thand
dat
ase
tfixed
effe
cts.
Inco
me
row
adds
work
ing
dum
my
and
log
inco
me
tobase
line.
Fam
ily
row
adds
seve
nfa
mil
ym
ember
contr
ols
tobase
line.
All
row
isfu
llsp
ecifi
cati
on
from
table
3.
Tri
mm
edre
mit
tance
sgre
ater
than
twic
ein
com
e.Poole
dsa
mple
spost
stra
tified
by
countr
yand
educa
tion.
See
table
1fo
rfu
llsu
rvey
nam
es.
a.
Incl
udes
house
hold
size
,dum
my
vari
able
ifm
arr
ied,
dum
my
vari
able
ifsp
ouse
isouts
ide
the
countr
y,num
ber
of
childre
n,
num
ber
of
childre
nouts
ide
the
countr
y,num
ber
of
pare
nts
,and
num
ber
of
pare
nts
outs
ide
the
countr
y
Sourc
e:A
uth
ors
’analy
sis
base
don
dat
ades
crib
edin
the
text.
Bollard, McKenzie, Morten, and Rapoport 151
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
TA
BL
E7
.E
duca
tion
Coef
fici
ent
as
Contr
ols
are
Added
:R
emit
sIn
dic
ator
Aust
ralia
Bel
giu
mFra
nce
Ger
many
Italy
Japan
Norw
aySpain
Spain
U.S
.U
.S.
Poole
d
Vari
able
LSIA
IRSH
SD
RE
ES
SO
EP
NID
IID
BL
BI
EN
IN
IDI
NIS
Pew
Exte
nsi
veT
ota
l
Univ
ersi
tyed
uca
tion
Base
line
20.0
19
20.0
55
0.0
14
0.0
42
20.0
65
0.0
91**
0.0
12
20.0
49**
20.2
32**
0.0
38**
20.1
40*
20.0
18
20.0
10
(0.0
32)
(0.0
29)
(0.0
27)
(0.0
48)
(0.0
43)
(0.0
34)
(0.0
30)
(0.0
15)
(0.0
81)
(0.0
11)
(0.0
60)
(0.0
10)
(0.0
10)
Inco
me
20.0
52
20.1
12**
20.0
27
0.0
23
20.0
74
0.0
82*
20.0
20
20.0
62**
20.1
85*
20.0
00
20.1
65**
20.0
43**
20.0
33**
(0.0
32)
(0.0
29)
(0.0
27)
(0.0
47)
(0.0
42)
(0.0
35)
(0.0
30)
(0.0
15)
(0.0
78)
(0.0
11)
(0.0
58)
(0.0
10)
(0.0
11)
Fam
ilya
20.0
62
20.0
69*
0.0
15
0.0
39
20.0
46
0.0
88*
20.0
04
20.0
67**
20.2
34**
0.0
22
20.1
48*
20.0
31**
20.0
26*
(0.0
32)
(0.0
32)
(0.0
27)
(0.0
48)
(0.0
41)
(0.0
35)
(0.0
30)
(0.0
13)
(0.0
81)
(0.0
12)
(0.0
60)
(0.0
10)
(0.0
11)
All
20.0
80*
20.1
13**
20.0
27
0.0
28
20.0
65
0.0
83*
20.0
31
20.0
73**
20.1
77*
0.0
06
20.1
61**
20.0
43**
20.0
33**
(0.0
31)
(0.0
30)
(0.0
27)
(0.0
48)
(0.0
38)
(0.0
35)
(0.0
30)
(0.0
14)
(0.0
75)
(0.0
12)
(0.0
59)
(0.0
10)
(0.0
11)
Num
ber
of
obse
rvat
ions
2,6
54
451
4,2
78
854
1,1
53
1,0
30
2,4
66
10,2
82
1,1
12
7,1
13
1,2
96
32,6
51
25,9
07
Yea
rsof
educa
tion
Base
line
0.0
080
20.0
042
0.0
018
0.0
145
0.0
010
0.0
024**
0.0
008
20.0
023
20.0
072**
0.0
034**
0.0
010
0.0
006
0.0
014
(0.0
043)
(0.0
040)
(0.0
025)
(0.0
084)
(0.0
040)
(0.0
005)
(0.0
025)
(0.0
018)
(0.0
021)
(0.0
012)
(0.0
060)
(0.0
009)
(0.0
010)
Inco
me
0.0
014
20.0
117**
20.0
016
0.0
071
20.0
027
0.0
035**
20.0
027
20.0
049**
20.0
074**
20.0
015
20.0
035
20.0
027**
20.0
018
(0.0
042)
(0.0
040)
(0.0
025)
(0.0
084)
(0.0
037)
(0.0
011)
(0.0
026)
(0.0
018)
(0.0
021)
(0.0
012)
(0.0
057)
(0.0
010)
(0.0
010)
Fam
ilya
0.0
018
20.0
060
0.0
050*
0.0
152
0.0
062
0.0
019**
20.0
006
20.0
040*
20.0
054*
0.0
029*
0.0
006
0.0
000
0.0
006
(0.0
044)
(0.0
042)
(0.0
025)
(0.0
087)
(0.0
041)
(0.0
006)
(0.0
026)
(0.0
017)
(0.0
021)
(0.0
012)
(0.0
060)
(0.0
009)
(0.0
010)
All
20.0
025
20.0
115**
0.0
012
0.0
130
0.0
031
0.0
034**
20.0
037
20.0
061**
20.0
046*
0.0
002
20.0
059
20.0
019*
20.0
011
(0.0
041)
(0.0
041)
(0.0
024)
(0.0
087)
(0.0
037)
(0.0
010)
(0.0
026)
(0.0
017)
(0.0
020)
(0.0
012)
(0.0
055)
(0.0
010)
(0.0
010)
Num
ber
of
obse
rvat
ions
2,6
48
451
5,5
29
854
1,1
53
1,0
30
2,4
50
10,2
01
1,1
12
7,1
00
1,2
96
32,5
35
25,8
07
*Sig
nifi
cant
atth
e5
per
cent
leve
l**
signifi
cant
atth
e1
per
cent
leve
l.
Note
:B
ase
line
row
incl
udes
only
countr
yof
bir
thand
dat
ase
tfixed
effe
cts.
Inco
me
row
adds
work
ing
dum
my
and
log
inco
me
tobase
line.
Fam
ily
row
adds
seve
nfa
mil
ym
ember
contr
ols
tobase
line.
All
row
isfu
llsp
ecifi
cati
on
from
table
3.
Tri
mm
edre
mit
tance
sgre
ater
than
twic
ein
com
e.Poole
dsa
mple
spost
stra
tified
by
countr
yand
educa
tion.
a.
Incl
udes
house
hold
size
,dum
my
vari
able
ifm
arr
ied,
dum
my
vari
able
ifsp
ouse
isouts
ide
the
countr
y,num
ber
of
childre
n,
num
ber
of
childre
nouts
ide
the
countr
y,num
ber
of
pare
nts
,and
num
ber
of
pare
nts
outs
ide
the
countr
y
Sourc
e:A
uth
ors
’analy
sis
base
don
dat
ades
crib
edin
the
text.
152 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
specification. The main hypothesis for why less skilled migrants remit more isthat they are more likely to have family members outside the country.Therefore, controlling only for this variable would be expected to increase thecoefficient on education, but the opposite occurs: the coefficient on educationdrops from $300 to $230 and remains statistically significant. This casts doubton the idea that low skilled migrants remit more because of their family com-position. One explanation is the earlier observation that low skilled migrantsare not only likely to have more family abroad, but they are also likely to livein larger households in the host country. The second row of the table adds justthe income variables (a dummy variable for working and log income) to thebaseline specification. The coefficient on university degree falls by more thanhalf and is no longer statistically significant. This suggests that the incomeeffect is a key channel through which education affects remittances—more edu-cated people send more money because they have higher incomes.
Although education becomes insignificant after controlling for income in thepooled sample, this result masks the heterogeneity in the individual surveys. Forexample, the education coefficient remains statistically significant even after con-trolling for all available covariates for three datasets: the Spanish ENI survey,the U.S. Pew dataset, and the U.S. NIS survey. There are several reasons why theeducation coefficient might remain significant in some datasets and not in othersthat cannot be examined with the dataset. One key variable that cannot be con-trolled for is the socioeconomic status of the family in the home country. Moreeducated individuals might come from better-off families and therefore not needto send back as much money. This could explain the negative coefficient in theENI and the Pew datasets.10Or more educated individuals might have fewer tiesto their home country. Time spent away from the home country and desire toreturn home are used to control for this, but they may not fully capture thestrength of the ties. Also lacking are data on whether migrants are using remit-tance to repay family loans—for example, for education. One additional keyissue is that the use of cross-section data does not yield any information abouteconomic shocks that affect the migrant or the migrant’s family.
Table 7 examines the extensive margin. In the baseline specification, moreeducated migrants are less likely to remit anything, but this result is not statisti-cally significant. The negative effect of education on the decision to remit any-thing is strengthened by the inclusion of different sets of covariates. Thecoefficient on education (measured by university degree) is negative and signifi-cant once any covariates are included. The alternative measure of education,years of schooling, is not statistically significant. Table 8 examines the intensivemargin result, which again appears to be driven by the income effect. Adding
10. An alternative explanation may be that the high-earning highly educated migrants are less likely
to respond to surveys. Survey methods that draw a sample from areas known to have a high
concentration of migrants (such as the Pew survey) or from locations where migrants tend to congregate
(such as the NIDI surveys) are especially likely to miss highly educated high-income individuals, who
may be living in areas where there are fewer of their countrymen.
Bollard, McKenzie, Morten, and Rapoport 153
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
TA
BL
E8
.E
duca
tion
Coef
fici
ent
as
Contr
ols
Are
Added
:L
og
Rem
itta
nce
s
Aust
ralia
Bel
giu
mFra
nce
Ger
many
Italy
Japan
Net
her
lands
Spain
Spain
UK
U.S
.U
.S.
Poole
dPoole
d
Vari
able
LSIA
IRSH
S2M
OSO
EP
NID
IID
BC
SR
EN
IN
IDI
BM
EN
ISPew
Inte
nsi
veT
ota
l
Univ
ersi
tyed
uca
tion
Base
line
0.3
41*
0.4
33**
0.3
63
0.4
92
0.0
73
20.0
57
0.3
33**
0.0
93
0.4
30*
0.1
68
0.3
97*
20.1
99
0.2
49**
0.2
26**
(0.1
45)
(0.1
31)
(0.2
11)
(0.4
50)
(0.1
71)
(0.1
46)
(0.1
16)
(0.0
66)
(0.2
02)
(0.1
33)
(0.1
69)
(0.2
16)
(0.0
60)
(0.0
71)
Inco
me
0.2
37
0.2
43*
0.3
06
0.4
08
0.0
21
20.0
86
0.3
33**
0.0
40
0.3
67
0.0
97
0.0
23
20.2
78
0.1
43*
0.1
14
(0.1
38)
(0.1
16)
(0.2
03)
(0.4
45)
(0.1
65)
(0.1
40)
(0.1
16)
(0.0
64)
(0.2
00)
(0.1
23)
(0.1
68)
(0.2
10)
(0.0
58)
(0.0
67)
Fam
ilya
0.2
88*
0.2
58*
0.3
90
0.4
23
0.1
05
20.0
33
0.3
33**
0.0
92
0.4
95**
0.2
06
0.3
64*
20.2
53
0.2
46**
0.2
20**
(0.1
39)
(0.1
28)
(0.2
07)
(0.3
68)
(0.1
78)
(0.1
50)
(0.1
16)
(0.0
61)
(0.1
87)
(0.1
32)
(0.1
66)
(0.2
18)
(0.0
57)
(0.0
66)
All
0.1
79
0.2
25
0.3
18
0.2
93
20.0
15
0.0
03
0.3
23**
0.0
54
0.4
09*
0.1
27
0.0
71
20.3
47
0.1
57**
0.1
18
(0.1
34)
(0.1
18)
(0.2
10)
(0.3
09)
(0.1
76)
(0.1
38)
(0.1
17)
(0.0
59)
(0.1
93)
(0.1
23)
(0.1
65)
(0.2
06)
(0.0
55)
(0.0
63)
Num
ber
of
obse
rvat
ions
958
317
713
184
545
690
648
3,9
66
761
993
1,1
18
514
11,3
92
9,0
38
Yea
rsof
educa
tion
Base
line
0.0
441*
0.0
341
0.0
224*
20.0
085
20.0
032
20.0
040
0.0
247*
0.0
199**
0.0
091
0.0
548*
0.0
329*
0.0
369
0.0
256**
0.0
229**
(0.0
194)
(0.0
174)
(0.0
112)
(0.0
783)
(0.0
163)
(0.0
038)
(0.0
100)
(0.0
076)
(0.0
063)
(0.0
237)
(0.0
146)
(0.0
221)
(0.0
061)
(0.0
071)
Inco
me
0.0
266
0.0
103
0.0
105
20.0
387
20.0
077
20.0
041
0.0
247*
0.0
114
0.0
021
0.0
313
20.0
008
0.0
294
0.0
135*
0.0
112
(0.0
199)
(0.0
164)
(0.0
115)
(0.0
770)
(0.0
164)
(0.0
048)
(0.0
100)
(0.0
075)
(0.0
062)
(0.0
220)
(0.0
126)
(0.0
216)
(0.0
053)
(0.0
062)
Fam
ilya
0.0
383*
0.0
101
0.0
344**
0.0
053
0.0
098
20.0
042
0.0
247*
0.0
247**
0.0
146*
0.0
612**
0.0
392**
0.0
194
0.0
272**
0.0
243**
(0.0
187)
(0.0
167)
(0.0
111)
(0.0
649)
(0.0
168)
(0.0
037)
(0.0
100)
(0.0
072)
(0.0
066)
(0.0
235)
(0.0
144)
(0.0
231)
(0.0
060)
(0.0
070)
All
0.0
227
0.0
060
0.0
268*
20.0
183
0.0
009
20.0
020
0.0
274**
0.0
179*
0.0
086
0.0
319
0.0
172
0.0
128
0.0
169**
0.0
139*
(0.0
193)
(0.0
165)
(0.0
124)
(0.0
511)
(0.0
159)
(0.0
050)
20.0
1(0
.0070)
(0.0
064)
(0.0
224)
(0.0
123)
(0.0
213)
(0.0
052)
(0.0
060)
Num
ber
of
obse
rvat
ions
956
317
713
184
545
690
648
3,9
42
761
993
1,1
16
514
11,3
64
9,0
10
*Sig
nifi
cant
atth
e5
per
cent
leve
l**
signifi
cant
atth
e1
per
cent
leve
l.
Note
:B
ase
line
row
incl
udes
only
countr
yof
bir
thand
dat
ase
tfixed
effe
cts.
Inco
me
row
adds
work
ing
dum
my
and
log
inco
me
tobase
line.
Fam
ily
row
adds
seve
nfa
mil
ym
ember
contr
ols
tobase
line.
All
row
isfu
llsp
ecifi
cati
on
from
table
3.
Tri
mm
edre
mit
tance
sgre
ater
than
twic
ein
com
e.Poole
dsa
mple
spost
stra
tified
by
countr
yand
educa
tion.
a.
Incl
udes
house
hold
size
,dum
my
vari
able
ifm
arr
ied,
dum
my
vari
able
ifsp
ouse
isouts
ide
the
countr
y,num
ber
of
childre
n,
num
ber
of
childre
nouts
ide
the
countr
y,num
ber
of
pare
nts
,and
num
ber
of
pare
nts
outs
ide
the
countr
y
Sourc
e:A
uth
ors
’analy
sis
base
don
dat
ades
crib
edin
the
text.
154 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
only family variables to the baseline specification reduces the coefficient on uni-versity education by approximately 3 percent, but it remains highly significant.However, if only income variables are added to the baseline specification, thecoefficient becomes statistically insignificant, with approximately the samepoint value as the full specification with the full set of covariates.
I V. C O N C L U S I O N S
The key advantage of this analysis over that in other papers in this literature(Faini 2007; Niimi and others 2008) is the ability to link the remittancedecision of migrants with their education level and therefore directly answerthe question of whether more educated migrants remit more. Cross-countrymacroeconomic analyses that relate the amount of remittances received at acountry level to the share of migrants with tertiary education can at best tell uswhether countries that send a larger share of highly skilled migrants receiveless or more remittances than countries that send fewer skilled migrants,without accounting for the other differences between countries that couldunderlie such a relationship.
This new database on migrants allows direct examination of the relationshipbetween education and remittance decisions. Results for the extensive margin(the decision to remit at all) and the intensive margin (the decision on howmuch to remit) combined show that, at least in this combined sample, moreeducated migrants remit significantly more: migrants with a university degreeremit $300 more yearly than migrants without one. Nonetheless, there is someheterogeneity across destination countries, with negative point estimates in afew of the surveys used—mainly in surveys that sample migrants only in areaswhere migrants cluster, thereby missing more educated, higher earningmigrants who may live outside of the immigrant clusters.
The data also allow analysis of several competing theoretical channels thathelp to explain this result. Differences in household composition between highand low skilled migrants do not explain the observed remittance behavior. Oneexplanation may be that although low skilled migrants are more likely to havea spouse and children in the home country, they have larger families onaverage than do high skilled migrants and tend to live in larger households inthe host country. In contrast, there is considerable support for an income effectas the dominant channel through which education operates. More educatedmigrants earn more money and therefore remit more than low skilled migrants.
The article also highlights the clear limitations of existing microdata on remit-tances. While some basic information on migrants can be obtained from censusmicrodata and government immigration records, there are no comparablyreliable sources for remittances. The new database relies on specialized one-offsurveys of migrants. Given the importance of remittances for many developingcountries, it would be beneficial for migrant-receiving countries to include ques-tions on remittances in their regular labor force or household budget surveys.
Bollard, McKenzie, Morten, and Rapoport 155
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from
This would be a first step to being able to analyze how remittance patternschange as countries pursue more skill-selective immigration policies.
Policy debates on migration often raise concerns about the potential negativeeffects of the “brain drain” on developing countries. However, the mainfinding that remittances increase with education illustrates one beneficialdimension of high skilled migration for developing countries. High skilledmigrants work in better jobs and earn more money than low skilled migrantsand in turn send more money back home in remittance flows. This suggeststhat the fear that remittances will fall as the migrant skill level rises is not sup-ported by existing empirical evidence.
RE F E R E N C E S
Abbasi, Saif-Ur-Rehman Saif, and Arshad Hussain Hashmi. 2000. “Migrants Earning at Overseas Job
and Extent of Remittances Transferred to their Families in Pakistan.” International Journal of
Agriculture and Biology 2 (3): 222–25.
Cox, Donald. 1987. “Motives for Private Transfers.” Journal of Political Economy 95 (3): 508–46.
Cox, Donald, Zekeriya Eser, and Emmanuel Jimenez. 1998. “Motives for private transfers over the life
cycle: An analytical framework and evidence for Peru.” Journal of Development Economics 55: 57–80.
Docquier, Frederic, and Abdeslam Marfouk. 2005. “International Migration by Education Attainment,
1990–2000.” In International Migration, Remittances and the Brain Drain, ed. C. Ozden, and M. Schiff,
151–99. New York: Palgrave Macmillan.
Faini, Riccardo. 2007. “Remittances and the Brain Drain: Do more Skilled Migrants Remit More?”
World Bank Economic Review 21 (2): 177–91.
Groenewold, George, and Richard Bilsborrow. 2004. “Design of Samples for International Migration
Surveys: Methodological Considerations, Practical Constraints and Lessons Learned from a
Multi-Country Study in Africa and Europe.” Paper presented at the Population Association of
America 2004 General Conference, Boston, MA, April 1–3.
IDB (Inter-American Development Bank). 2005. “Survey of Brazilians and Peruvians in Japan” Survey
commissioned by the Multilateral Investment Fund.
Miotti, Luis, El Mouhoub Mouhoud, and Joel Oudinet. 2009. “Migrations and Determinants of
Remittances to Southern Mediterranean Countries: When History Matters.” Paper presented at the
2nd Migration and Development Conference, Washington, DC, September 10–11.
Niimi, Yoko, Caglar Ozden, and Maurice Schiff. Forthcoming. “Remittances and the Brain Drain:
Skilled Migrants Do Remit Less.” Annales d’Economie et de Statistique.
OECD (Organisation for Economic Co-operation and Development). 2007. Policy Coherence for
Development 2007: Migration and Developing Countries. Paris: OECD.
Rapoport, Hillel, and Frederic Docquier. 2006. “The Economics of Migrants’ Remittances.” In
Handbook of the Economics of Giving, Altruism and Reciprocity, ed. S.-C. Kolm, and J. Mercier
Ythier, 1135–98. Amsterdan: North-Holland.
Robinson, Peter M. 1988. “Root-N Consistent Semiparametric Regression.” Econometrica 56: 931–54.
Siegel, Melissa. 2007. “Immigrant Integration and Remittance Channel Choice.” Working Paper 2007/
09. Maastricht Graduate School of Governance.
Stark, Oded. 1991. The Migration of Labor. Cambridge, MA: Basil Blackwell.
World Bank. 2008. Migration and Remittances Factbook 2008. Washington, DC: World Bank.
———. 2009. “Migration and Development Brief No. 9.” http://siteresources.worldbank.org/
INTPROSPECTS/Resources/MD_Brief9_Mar2009.pdf. Accessed July 10, 2009.
156 T H E W O R L D B A N K E C O N O M I C R E V I E W
at International Monetary F
und on August 12, 2011
wber.oxfordjournals.org
Dow
nloaded from